Exosomes and Extracellular Vesicles in Wound Care and Wound Healing: An Evidence-Based Critical Review

Executive Summary

The principal finding of this review is a sharp asymmetry with consequences: the preclinical literature on extracellular vesicles (EVs) in wound healing is large, remarkably consistent and strongly positive, whereas the human literature is small, methodologically weak and bipolar. Every controlled study in which a purified EV product was tested in an acute, standardised wound has returned a null (no difference) result. In chronic ulcers, two controlled positive results — one a small single-centre randomised trial, the other a non-randomised pilot of 4 patients — carry a high or very high risk of bias; to these has been added a third positive signal, posted to a registry as of September 2026 but not yet peer-reviewed (see below and Section 6). Nowhere in the world is there an EV or exosome product approved as a drug for any indication [1–3]. Nor is there a licensed EV/exosome medicinal product in Türkiye (see Section 11). There is nothing in this field today that can be presented as "validated clinical practice"; every human instance of an EV product being used on a wound is either within a clinical trial, within a single-patient/access programme, or the use of an unapproved commercial product outside the licensing pathway.

What the biology supports. EVs are particles released from cells, bounded by a lipid bilayer and incapable of self-replication. The current position of the International Society for Extracellular Vesicles (ISEV) (MISEV2023) is that the correct generic term is "extracellular vesicle" and that the word "exosome" should not be used unless endosomal origin has actually been demonstrated — which almost no wound-healing study does [4]. At the mechanistic level, EV preparations act at every stage of repair: they reprogramme macrophages towards a reparative phenotype via TLR4/NF-κB-related cargo, drive angiogenesis through PI3K/AKT (± eNOS) and DLL4/Notch signalling, increase keratinocyte and fibroblast proliferation via Wnt/β-catenin and YAP, and limit myofibroblast differentiation through suppression of TGF-β1/Smad2/3. This mechanistic framework has been built almost entirely from in vitro studies and rodent wounds [5–8].

What the animal data show, and why the effect sizes are not reliable as estimates of efficacy. Eight or more systematic reviews and meta-analyses of animal cutaneous-wound EV studies report large pooled benefits: standardised mean difference (SMD) for wound closure approximately 3.2–8.4; SMD for angiogenesis up to 9.3; SMD for re-epithelialisation ~4.7–5.1; SMD for scar width between −5.8 and −8.1; flap survival +35.5 percentage points [9–15]. One of these syntheses [15] carries a data-source caveat explained in Sections 5.1 and 6.7 and should not be counted as an independent analysis. Yet the same analyses report that SYRCLE risk of bias was "unclear for most studies owing to inadequate reporting" [9,12], that allocation concealment remained unclear in 24 of 24 studies in one flap meta-analysis [14], that heterogeneity reached I² 91–97% [11,14], and that publication bias was statistically detectable (Egger P=0.000 in type 2 diabetic wounds [12]; Begg Z=2.47, P=0.013 and Egger coefficient 5.86, t=4.52, P=0.004 in diabetic foot ulcer models [11]; Egger P<0.001 for both flap survival and angiogenesis [14]). Of more than 80 published animal studies, only two reported an effect in the adverse direction [16]. An 83-study systematic review contained only mice and rats — no EV efficacy study could be identified in a porcine cutaneous wound model [6,17]; rodent excisional wounds close largely by contraction, whereas human chronic wounds close by re-epithelialisation [18,19].

What the human data show. Purified EV products have been applied to the wounds of roughly 150–250 patients worldwide in controlled or quasi-controlled studies; to this can be added a few hundred patients in uncontrolled commercial series. Two independent first-in-human platelet-derived EV (pEV) wound studies — different sponsors, different continents, different products — found no difference in healing compared with control: Exopharm's Plexoval II study (ACTRN12620000944932, 11 healthy volunteers, paired 4 mm punch-biopsy wounds, 100 µg pEV subcutaneously) gave a mean time to healing of 22.8 ± 8.7 days in treated wounds and 22.8 ± 8.7 days in placebo wounds [20]; Rion's phase 1b study of its purified exosome product (PEP) at split-thickness skin-graft donor sites (NCT04664738, n=7) found median re-epithelialisation of 18.5 days versus 19.25 days with standard care, statistically non-significant [21]. The only positive randomised controlled trial — Wharton's jelly MSC exosome gel in diabetic foot ulcer (DFU) (NCT06812637, Egypt, 110 enrolled / 85 analysed) — reported complete healing at a mean of 6 weeks in the treatment arm versus 20 weeks in the control arm; however, there was no significant difference at weeks 2 and 4, there was a three-fold imbalance in baseline ulcer area (treatment 6 cm², control 2 cm²), attrition was 23%, and the exosome dose administered was never measured [22]. A non-randomised venous ulcer pilot of 4 patients using a GMP-compliant process and a release potency assay found an area reduction at day 30 of 385 mm² versus 106 mm² (P=0.004) [23]. A single burn patient treated with allogeneic bone-marrow MSC-EVs at a dose of 1×10⁷ particles/cm² achieved >99% closure of a 192 cm² deep second-degree burn within one week, sustained over 52 weeks, with no seroconversion and no development of panel-reactive antibodies [24].

Data added to the picture in September 2026. The results of the Rion Phase 2a trial (NCT06319287, PEP-TISSEEL + standard care vs standard care, DFU), described as "the most important missing data" at the time the source literature was compiled, were posted to ClinicalTrials.gov on 20 July 2026 [25,26]. According to the registry data, 59 patients were randomised (28/31), and complete closure at week 12 in the intention-to-treat (ITT) population was reported as 13/28 (46.4%) versus 7/31 (22.6%); no p value is given in the registry. The absolute difference calculated in this review from the registry counts is 23.8 points (95% CI 0.2–47.5; two-sided Fisher's exact test P=0.062). The trial is open-label, single-centre despite its title, approximately 35% of the arms did not complete the protocol, and there is no peer-reviewed publication. This signal — positive in direction, and in magnitude in the same class as two adjunctive therapies with moderate-certainty evidence (sucrose octasulfate dressing, leucocyte-platelet-fibrin patch) — is low-certainty evidence and requires multicentre replication with blinded assessment (see Sections 6.3 and 16.3).

Safety. In every human study identified there is no serious adverse event attributed to an EV wound product — this is the most consistent finding in the literature — but total exposure is a few hundred patients, follow-up is mostly ≤6 months, and animal studies have almost never assessed harm [9,27]. By contrast, the documented human harms come almost entirely from unapproved commercial products: bacteraemia after intravenous unapproved exosome products in the 2019 Nebraska cluster [1,28], foreign-body granuloma and skin necrosis after cosmetic intradermal injection [29,30], and hepatitis B, Escherichia coli, Klebsiella pneumoniae and Streptococcus contamination detected in FDA inspections [31].

What it means for the clinician. Standard care — sharp debridement, non-removable knee-high offloading for plantar DFU, compression for venous leg ulcers (VLU), infection control and revascularisation — remains the comparator that every EV product must beat; under good standard care, control-arm closure is approximately 24% at week 12 and 31% at week 20 [32], or approximately 33% over a 12–24-week window in modern trials [33]. The two adjunctive therapies that have reached moderate-certainty evidence (sucrose octasulfate dressing; autologous leucocyte-platelet-fibrin patch), each on the strength of a single RCT at low risk of bias, add roughly 12–18 percentage points to closure [34–36]. Whole-cell MSC therapy, the closest biological comparator, is supported in DFU by 32 randomised trials and 2,059 patients, with a pooled healing OR of 4.64 (95% CI 3.11–6.90) — roughly two orders of magnitude more human data than the entire EV wound field, which has no pooled human healing estimate at all [37]. The honest position is this: EV wound therapy is clinically about 10–15 years behind cell therapy; it retains genuine theoretical advantages in manufacturing, storage and the avoidance of live-cell risks.

A caution about the secondary literature. This review documents two verified instances in which review articles dated 2026 turned null or non-existent primary data into positive claims. A review published in Frontiers in Medicine (Arkoubi 2026) writes that "an exploratory phase II trial (NCT05125562)" with hypoxic MSC-derived exosomes achieved "71% complete closure … versus 33% with standard care" in Wagner grade 2 DFUs at week 12; NCT05125562 is in fact a withdrawn phase 2 trial testing intravenous ExoFlo infusion for mild-to-moderate COVID-19 that never enrolled a patient, belongs to a different company, and the quoted outcome figures cannot be traced to any registered trial [38,39]. A review dated 2026 in Frontiers in Immunology summarises the pEV skin-graft donor-site study as "100% re-epithelialisation, shortened healing time", directly contradicting the null result of the primary publication [5,21]. Readers evaluating this field should treat narrative reviews dated 2026 as unreliable secondary sources and return to the primary reports and registry records.

Introduction to Exosomes and Extracellular Vesicles

Definitions, and why the word "exosome" is problematic in this literature

MISEV2023, the current ISEV consensus, defines EVs as "particles that are released from cells, are delimited by a lipid bilayer and cannot replicate on their own (i.e. do not contain a functional nucleus)". The word "naturally" in the 2018 definition was removed so that engineered and culture-conditioned vesicles are not excluded [4]. ISEV recommends "EV" as the generic term plus operational qualifiers, and explicitly discourages "the use of biogenesis-based terms unless such an EV population has been specifically separated and characterised"; it states that "unless subcellular origin can be demonstrated, what is being studied is likely a broad EV population rather than specifically exosomes" [4].

This directly affects how the wound-healing literature is read. Almost no wound paper demonstrates endosomal origin. The published "exosome" wound products should therefore be understood as small EV (sEV) preparations of mixed biogenesis that also contain non-vesicular extracellular particles (NVEPs). MISEV2023 defines NVEPs as non-EV particles composed of cell-derived components that, even if they contain lipids, do not form a delimiting bilayer, and notes that "NVEPs and EVs may have overlapping physicochemical properties and NVEPs may vastly outnumber EVs in biological matrices", naming lipoproteins (HDL, LDL, IDL, VLDL) as "another major confounding class" [4]. A skin-specific review dated 2026 acknowledges that "current methods often fall short of effectively separating exosomes from other EV subtypes" [5].

Operationally, MISEV2023 prefers size-based terms — small EVs "often <200 nm in diameter", large EVs ">200 nm" — and notes that "the measured diameter depends on the characterisation method used" and that "there is no strict consensus on upper and lower size limits"; differential ultracentrifugation "yields EV populations with overlapping size profiles" [4]. The widely quoted numerical ranges (exosomes ~30–150 nm, microvesicles ~100–1,000 nm, apoptotic bodies ~1–5 µm) come from the older literature and are not supported by the MISEV2023 text; this review therefore uses the small/large EV convention. MISEV2018 remains the source of the widely used three-category marker scheme (transmembrane markers such as CD9/CD63/CD81; cytosolic markers such as TSG101 and ALIX/syntenin-1; non-EV co-isolate markers such as albumin and apolipoproteins) [40].

Biogenesis

Exosomes in the strict sense originate from the intraluminal vesicles of secretory multivesicular endosomes and are released upon fusion of these endosomes with the plasma membrane; ESCRT-0 (Hrs), ESCRT-I/II/III and the accessory proteins TSG101 and ALIX control both the quantity and the molecular composition of secreted exosomes [41]. ESCRT-independent pathways also exist: ceramide generated by neutral sphingomyelinase-2 "triggers the budding of exosome vesicles into multivesicular endosomes"; the tetraspanins CD9, CD63 and CD81 organise membrane microdomains as "sorting machinery for exosomes"; a syndecan–syntenin–ALIX axis constitutes "a dedicated machinery required for exosome biogenesis" [41]. Microvesicles/ectosomes bud directly from the plasma membrane under the control of ARF6 and RhoA; hypoxia acts via RAB22A; apoptotic bodies form a third category [41]. Cargo sorting encompasses ubiquitinated and lipid-raft proteins, hsc70/ALIX motif recognition, and RNA loading via sumoylated hnRNPA2B1, YBX1, KRAS–MEK-controlled Ago2 and MVP; the membrane is rich in ceramide, sphingosine-1-phosphate and cholesterol [41].

How EVs differ from neighbouring therapeutic classes

EVs sit between four neighbouring product classes, and conflation of these classes is the most common source of hype in this field.

Versus whole-cell therapy. MSC-EVs "do not self-replicate and largely avoid the risk of tumourigenicity", carry "biocompatibility and low immunogenicity", "can be preserved for long periods at −80 °C and are not inactivated even after repeated freeze–thaw", and "eliminate the risks associated with cell homing, differentiation and senescence in prolonged culture" [42]. The freeze–thaw claim is a review-level statement of plausibility and conflicts with the primary storage literature: a 23–36% particle loss and approximately 70% degradation of EV miRNA in a single freeze–thaw cycle, and declining bioactivity after five cycles, have been documented (Section 4.7) [43]; it is quoted here as the rationale for EV products, not as a proven property. The core premise is the paracrine hypothesis — "the healing effect is produced not by cell engraftment but by paracrine interactions between MSCs and neighbouring cells" [44]. These are plausibility arguments: in no wound type is there a human head-to-head study comparing EVs with their parent cells, and no dedicated systematic comparison in animal wound models was identified either.

Versus conditioned medium / secretome. EVs are a fraction of the secretome. The whole secretome "contains additional bioactive factors", whereas EVs "offer targeted delivery and protection from degradation" [44]. The critical point is this: conditioned medium is clinically active on its own, which blurs what the effect belongs to — human studies with adipose-derived stem cell (ADSC) conditioned medium have reported skin rejuvenation via suppression of MMP-1/MMP-2 and an increase in type I collagen; umbilical-cord MSC conditioned medium improved erythema, texture and post-laser recovery [5]. Several products presented in the literature as "exosome" therapy are in fact unfractionated conditioned medium — including the Jordanian DFU study, the Indonesian secretome gel and the "human exosome" comparator in the Korean post-surgical scar randomised trial (Section 6).

Versus platelet-rich plasma (PRP). PRP acts via free growth factors (PDGF, VEGF, TGF-β); PRP-derived exosomes package the cargo with "improved stability compared with free growth factors", are better retained in hydrogel, and enhance re-epithelialisation and collagen synthesis via Hippo/YAP, JAK2/STAT3 and sphingosine-1-phosphate/S1PR1/AKT/FN1 signalling — the same review also acknowledges that "variability in PRP preparation and exosome isolation limits reproducibility" [45]. PRP-EVs are an interesting intermediate case: their comparator is not cells but the growth factors in plasma.

Versus recombinant growth factors. EV preparations deliver a multi-component, redundant signal rather than a single ligand; whether this is an advantage in practice is unresolved, and the clinical history of single growth factors in wounds is cautionary (Section 12).

A reporting-quality problem that must be stated at the outset

Across the whole EV field, only 13.6% of the 5,093 open-access articles analysed (2012–2020) used all four MISEV characterisation categories; in the 2015–2020 subset, only 701 of 4,579 articles (15.3%) cited MISEV [46]. In the wound literature specifically, 44.9% of 83 preclinical cutaneous EV studies did not report assessment of at least one positive cytosolic marker, and 62.8% did not report major components of non-EV co-isolated structures [6]. The EV-TRACK platform exists precisely so that methodological metadata can be deposited and compared [47]. The reader should keep these figures in mind: the mechanistic and efficacy claims summarised in Sections 3 and 5 rest on a literature that frequently does not document what was actually administered.

Biology and Mechanisms of Action in Wound Healing

Normal wound repair proceeds through four overlapping phases: haemostasis, inflammation, proliferation (angiogenesis, granulation, re-epithelialisation) and remodelling [48,49]. Chronic wounds become trapped in dysregulated inflammation, hypoxia, impaired angiogenesis, cellular senescence, protease imbalance, infection/biofilm and abnormal extracellular matrix (ECM) turnover. The proposed phase-by-phase effects of EV preparations can be summarised didactically as follows: in the inflammatory phase, a shift of macrophages from the pro-inflammatory (M1-like) to the reparative (M2-like) phenotype and a reduction in oxidative stress; in the proliferative phase, transfer of pro-angiogenic signals to endothelial cells and stimulation of keratinocyte and fibroblast migration/proliferation; in the remodelling phase, a more organised tissue architecture through regulation of the collagen type III/type I ratio and the matrix metalloproteinase (MMP)/TIMP balance [48,49]. This scheme, however, is a summary framework; the level of evidence varies sharply by mechanism, and this section separates them explicitly.

In the literature on which this review is based, no human wound biopsy or mechanistic sub-study — before/after biopsy paired with transcriptomics or immunohistochemistry after EV treatment — was identified. Two independent reviews dated 2026 describe the human literature solely as small studies, case series and case reports [5,7]. The only mechanistic readouts at the level of human tissue (increased type I collagen and glycosaminoglycans) come not from wounds but from cosmetic studies of photoageing and acne scarring [5].

Inflammatory phase: macrophage reprogramming is the best-supported mechanism

The dominant documented mechanism is a shift from the M1-like to the M2-like (reparative) macrophage phenotype. Bone-marrow MSC-EVs transfer miR-223, which targets Pknox1 in a mouse wound model, lowering TNF-α and raising IL-10, RELM-α and arginase-1, and induces M2 polarisation [8,50]. LPS-preconditioned MSC-EVs carry let-7b in a rat burn model, suppressing TLR4/NF-κB and activating STAT3/AKT; miR-181c from umbilical-cord MSC exosomes suppresses TLR4/NF-κB/p65 in mouse burns, reducing TNF-α and IL-1β and increasing IL-10; miR-146a acts on the same axis [8,44]. Other axes are miR-21-5p (NF-κB suppression, pressure ulcer model), miR-203a-3p (SOCS3/JAK2/STAT3, diabetic wounds), miR-1246 (AKT/ERK1/2/STAT3 plus macrophage autophagy) and keratinocyte-EV MALAT1, which sponges miR-1914-3p to upregulate MFG-E8 [7]. Adaptive immune effects are documented but subtler: PD-L1 on the EV surface inhibited CD8+ T-cell cytokine production and reduced CD8+ T-cell numbers while "accelerating skin cell migration and wound healing in vivo" [44]; induction of regulatory T cells, with increased IL-4/IL-10/TGF-β and decreased IL-17/IFN-γ, has been shown in a non-skin (type 1 diabetes) model [42].

Pathogenic EV biology is equally real and is the strongest argument against treating "EV" as a label for the direction of effect: miR-93-5p in diabetic ulcer fibroblast EVs targets ATG16L1 in macrophages, impairing autophagic clearance of reactive oxygen species and increasing NLRP3 activity [7]; EVs from diabetic ADSCs increased monocyte TGF-β1 secretion — the opposite of healthy or engineered ADSC-EVs [7]; Staphylococcus aureus vesicles impair cutaneous repair by inhibiting macrophage efferocytosis via p38 MAPK–MerTK cleavage [51].

Certainty: low for the mechanism class (consistent across many independent in vitro and rodent systems, with depletion/rescue experiments in some studies), very low for the claim that any single miRNA is the functional agent (see Section 3.6).

Proliferative phase: angiogenesis is the most consistently measured effect

Angiogenesis is the dominant measured outcome in roughly half of preclinical studies [16]. Named axes with in vitro plus rodent support are: miR-125a, which suppresses the Notch ligand DLL4 to increase endothelial tip-cell formation [8,52]; miR-126/miR-126-3p, which targets PTEN and PIK3R2 to activate PI3K/AKT (deferoxamine-preconditioned bone-marrow MSC-EVs, diabetic rodents) [8]; miR-21, acting on PTEN/SPRY1 in endothelial cells and NOTCH1/DLL4 in endothelial progenitors [8]; miR-423-5p, which suppresses Sufu in the Hedgehog pathway [8]; miR-221-3p, acting via SPRED1/Ras/ERK [5]; Wnt4 protein cargo, which activates β-catenin in endothelium [53]; and — a rare lipid-mediated mechanism — sphingosine-1-phosphate signalling via S1PR1/AKT/FN1 in diabetic wounds from PRP-derived exosomes [45]. Upstream conditioning changes the angiogenic cargo: hypoxia upregulates miR-31-5p, miR-126-5p and miR-21-3p; atorvastatin and melatonin act on PTEN/AKT/eNOS via miR-211-3p; blue light increases miR-135b-5p and miR-499a-3p; the long non-coding RNAs KLF3-AS1 and HOTAIR, together with circ-ITCH, have also been proposed — circ-ITCH appears only as a review-level claim with no verifiable primary wound study behind it and should not be cited as evidence [5,44].

Re-epithelialisation and dermal repair operate largely through Wnt/β-catenin, AKT/HIF-1α and YAP. The strongest single mechanistic study in this literature remains the Wnt4 study: human umbilical-cord MSC exosomes increased β-catenin nuclear translocation and PCNA in HaCaT keratinocytes and rat dermal fibroblasts, Wnt4 shRNA knockdown abolished both the in vitro signalling and the in vivo healing effect, the AKT inhibitor LY294002 reversed a separate anti-apoptotic arm of increased Bcl-2/decreased Bax, and in rats with deep second-degree burns 6 of 6 exosome-treated wounds re-epithelialised completely within two weeks versus 1 of 6 controls; CK19 and PCNA increased, the collagen I:III ratio rose, and TUNEL-positive cells decreased [54]. The authors attributed part of the AKT activation to co-delivered cytokines (PDGF-BB, G-CSF, VEGF, IL-6, IL-8) — an early and honest acknowledgement of multi-component action. A complementary mechanism is 14-3-3ζ protein cargo, which governs the "self-control" of the Wnt response via YAP [55]. Barrier-level effects include increased epidermal stratification, repair of the permeability barrier through ADMSC-EV-derived de novo ceramide synthesis, and hair-follicle regeneration and telogen→anagen transition via Wnt3a/Wnt11-containing EVs and miR-218-5p [44].

Certainty: low. The effects are reproducible across laboratories and species; one study meets a genuine causality bar (knockdown plus pharmacological rescue); but all in vivo data are rodent and most outcomes are morphometric.

Remodelling phase: antifibrotic activity is mechanistically consistent but morphological

Anti-scarring activity converges on TGF-β/Smad. Umbilical-cord MSC exosomal miR-21, miR-23a, miR-125b and miR-145 suppress TGF-β/SMAD2 in mice, reducing myofibroblast differentiation, collagen deposition and scar formation [56]; receptor-level blockade occurs via miR-21-5p targeting TGFBR2 and miR-125b-5p targeting TGFBR1 [7]; miR-192-5p acts via IL-17RA/Smad, miR-29a via TGF-β2/Smad3 and miR-204-5p via TGF-β/Smad, each reducing α-SMA and collagen I/III [7,8]. Other axes include increased TGF-β3 and BM-MSC-EV miR-214 with IL-33/ST2 blockade [5]. Optimal remodelling is framed as preserved type I collagen predominance with increased MMP activity relative to TIMP [44].

Context dependence is decisive here too: M2-like macrophage EVs transferred the lncRNA ASLNCS5088, which relieves the suppression of glutaminase by miR-200c-3p, increasing α-SMA and collagen; CXCL2-containing macrophage EVs activated fibroblast CXCR7/PI3K/mTOR [7]. A review dated 2026 states explicitly that "whether a given EV axis switches from reparative to fibrogenic as healing progresses … remains largely untested within the same experimental system" [7].

Certainty: very low for a clinical anti-scar effect. These are histological and immunoblot outcomes in rodents; the only human scar data are in acne scarring and post-surgical facial/chest scars (Section 6); in one study, plant-derived exosomes performed the same as human ADSC-derived ones.

Metabolic and ageing-related mechanisms relevant to diabetic and aged wounds

ADSC-EVs restored fibroblast viability under high glucose with protection linked to the Keap1/Nrf2 axis [7,57]; Nrf2-overexpressing ADSC exosomes increased vascularisation and closure in a DFU rat model [58]; pharmacological Nrf2 activation combined with bone-marrow MSC-EVs accelerated diabetic wound healing. Conversely, patient-derived circulating exosomal miR-181b-5p "worsened diabetic foot ulcer by increasing cellular senescence and suppressing angiogenesis via the Nrf2/HO-1 pathway" — a reminder that endogenous EV biology in diabetes is part of the pathology [59]. In aged mice, human embryonic stem cell exosomes rejuvenated senescent endothelial cells via miR-200a/Keap1/Nrf2; they reduced SA-β-gal-positive HUVECs, lowered P16 and P21, reduced malondialdehyde and raised SOD/CAT/GSH-Px (all P<0.001), and the effect was abolished by the Nrf2 inhibitor brusatol and by miR-200a knockdown [60].

Delivery-related pharmacology: retention is the practical constraint

When given intravenously, EVs are rapidly cleared by the liver and spleen; "only about 1% remains after 24 hours"; locally applied EVs "are rapidly eliminated after being transported to surrounding tissues and capillaries" [61]. An estimate circulates in the review literature that "approximately 10–20% of exosomes administered by conventional routes (intravenous or topical) reach the target" [62]. No wound-specific pharmacokinetic data — EV half-life in the wound bed, percentage retained 24 or 72 hours after topical versus intradermal application — could be identified; one review states explicitly that there are no quantitative retention data for either route [63]. The only verified large-animal biodistribution study is the porcine pulmonary distribution of platelet-derived PEP: "nebulisation achieved effective uptake in airway epithelium and tracheal tissue, and catheter-guided delivery was able to localise exosomes to a single region within the lung; no significant off-target accumulation was observed" [64,65]. The documented rationale for every carrier strategy in Section 8 is this retention problem, not a demonstrated superiority of any biomaterial.

Three reasons to read the standard "EV miRNA X acts on pathway Y" narrative with scepticism

Stoichiometry. Quantitative analysis found that "on average, 100 exosomes correspond to only one copy of a given miRNA", with the ratio of miRNA molecules per exosome "significantly less than one" [66,67]. Independent absolute quantification showed that a single EV contains 0.00006 ± 0.000037 to 0.0034 ± 0.0018 copies of specific viral miRNAs — "a single mature miRNA is found in 300 EVs … or, more often, in thousands of EVs" — and that even after forced fusion with VSV-G, EVs carrying miR-BHRF1-2 or miR-BART1 "could not regulate Renilla luciferase activity in recipient cells" at up to 1×10⁵ EVs per cell; the authors concluded that EV-derived miRNAs "do not act as agents of intercellular communication", noting that their experiments were entirely in vitro [68]. Inference, stated as inference: at these ratios, delivering 10⁴ particles per cell yields on the order of 1–30 copies of a given miRNA per cell — orders of magnitude below what target suppression normally requires.

Non-vesicular material that co-isolates. MISEV2023's NVEP language is the authoritative basis for the contaminant critique, and 62.8% of wound studies did not report non-EV co-isolates [4,6]. Three dedicated methodological papers address precisely this failure mode — misattribution of soluble MSC paracrine bioactivity to EVs [69], immunomodulatory bioactivity emerging only after soluble cytokines are removed [70], and some reported EV functions being explained by transfection-reagent artefacts [71] — the last being directly relevant because most "EV miR-X" wound studies use transfected donor cells. The full texts of these three papers could not be accessed; they are cited for their documented titles and claims.

The field's own standard. A 2026 review states that "enrichment of a single miRNA should not be equated with functional dominance unless necessity is shown by depletion or rescue experiments" and notes that EV efficacy "may also depend on molecules associated with the vesicle surface" [7].

The defensible mechanistic synthesis is therefore multimodal: surface ligands and EV proteins (Wnt4, 14-3-3ζ, PD-L1), co-delivered cytokines, lipids (ceramide, sphingosine-1-phosphate) and bulk RNA act together; single-miRNA reports are treated as pathway hypotheses unless depletion and rescue have been performed.

Note: "from review" = the primary study could not be verified in this review; data are taken from a secondary review.

Exosome Sources and Manufacturing

The source landscape

Therapeutic EVs for wound healing have been produced from almost every accessible cell type and from several non-cellular sources. Across 83 preclinical cutaneous studies published from 2015 to June 2024, the distribution of parent cell sources was as follows: adipose tissue 32 (38.6%), umbilical cord 19 (22.9%), bone marrow 16 (19.3%), amniotic tissue 2 (2.4%), oral mucosa 3 (3.6%), synovium 2 (2.4%), other 9; in 11 studies (13.3%) the source was not reported at all. The MSC donor was human in 58 studies (69.9%) [6]. A separate landscape review tabulates bone marrow, adipose tissue, umbilical cord, macrophage, endothelial progenitor, fibroblast, keratinocyte, periodontal tissue, cord blood, platelet-rich plasma (PRP), urine-derived stem cell, saliva, milk, amniotic fluid and plant sources one by one in terms of cargo and advantages–disadvantages [72].

The clinically decisive point is this: human wound data exist for only three product classes. Platelet-derived EVs (two independent first-in-human wound studies and a completed Phase 2a programme with results posted to the registry; see Section 6), umbilical cord / Wharton's jelly MSC exosomes (one RCT and case reports) and bone marrow MSC-EVs (a single published burn patient and an epidermolysis bullosa study that is still recruiting). Everything else — adipose tissue, induced pluripotent stem cells (iPSC), endothelial progenitors, urine-derived, milk, plant, bacterial — is at the preclinical stage in wounds; the only partial exception is adipose-derived material used in aesthetic scar studies.

Does source selection matter? The honest answer: "yes, but not in an established direction"

The three direct head-to-head studies contradict one another.

In the same diabetic mouse model, with 1×10⁹ EVs in carboxymethylcellulose applied topically every three days, ADSC-EVs significantly accelerated closure (P<0.0001 at day 14), reduced scar width (P<0.005) and increased epithelial thickness, re-epithelialisation and vessel count (all P<0.0001); BM-MSC-EVs, by contrast, were found "ineffective" and did not improve closure over vehicle [73]. The cargo basis was mapped: 99 shared miRNAs versus 70 unique to ADSC and 14 unique to BMSC; 38 shared proteins versus 41 ADSC-specific and 24 BMSC-specific proteins; ADSC cargo was enriched in angiogenesis pathways (Wnt, FGF, EGFR, PDGF, TGF-β, HIF-1α) [73].

In serum-free, xeno-free culture with the same isolation method, growth factor content per 10⁶ cells differed by source: VEGF-A 12.15 ± 5.28 (bone marrow), 12.65 ± 8.88 (adipose tissue), 4.79 ± 5.63 (umbilical cord) pg; FGF-2 13.66 ± 7.71 / 4.72 ± 1.15 / 1.02 ± 0.7; HGF 2.84 ± 1.6 / 7.19 ± 1.01 / 1.47 ± 0.49; PDGF-BB 10.43 ± 11.55 / 1.56 ± 1.30 / 1.64 ± 0.39; TGF-β was detectable only in umbilical cord (3.61 ± 0.69). Functionally, BM-MSC exosomes were best for dermal fibroblast proliferation and migration (10–20 µg, P<0.001), and UC-MSC exosomes were best for keratinocyte migration; superiority depended on source, dose and target cell. There was no in vivo arm [74].

A 2026 comparison of UC-MSC versus AD-MSC exosomes (same differential ultracentrifugation, 48-hour serum-free conditioned medium; peak diameter ~140 ± 18 nm and ~105 ± 12 nm respectively) found UC-MSC-Exo richer in TGF-β1 (>10,000 pg/mL), IL-10, GM-CSF and HGF, and AD-MSC-Exo richer in PDGF-BB, VEGF and FGF. UC was superior on senescence and inflammation endpoints (IL-6 suppression P<0.001; P<0.0001 in UV-aged human skin explants), whereas AD was superior for collagen deposition in explants (P<0.05) and melanin reduction (111.2 ± 5.7 versus 124.0 ± 8.2 µg/mg protein, P<0.001). The authors concluded that the source should be matched to the indication [75].

Meta-analytic subgroup data split the difference: across 83 preclinical studies, ADSC-derived EVs showed the best efficacy for closure and collagen deposition, whereas BM-MSC-derived EVs performed better on revascularisation; apoptotic small EVs outperformed apoptotic bodies and conventional sEVs for closure and collagen, while conventional sEVs were better on revascularisation (only five apoptotic EV studies); subcutaneous injection was superior to dressing/topical coverage on all three outcomes [6]. An ADSC-specific preclinical diabetic wound meta-analysis (20 studies qualitative, 12 pooled) reported SMD 4.22 (95% CI 3.07–5.36) for closure, SMD 9.27 (4.70–13.83) for neovascularisation and SMD 2.19 (0.94–3.44) for collagen deposition; risk of bias was unclear in all included studies [76]. In one DFU meta-analysis the ADSC subgroup was significant (Z=3.29, P=0.001) while the BM-MSC subgroup was not (Z=1.70, P=0.09); yet there was no statistical difference between subgroups (χ²=0.00, P=0.96) — that is, the apparent source ranking fails the formal subgroup test [11].

Certainty of evidence for source superiority: very low. Source effects are outcome- and context-specific and are confounded with culture medium, isolation method and dose metric. No study has placed adipose, bone marrow and umbilical cord EVs side by side in the same in vivo wound model at a matched particle dose.

Non-cellular and unconventional sources

Platelet-derived material is the only non-MSC source with human wound data and has a genuine manufacturing rationale: pooled, banked, non-proliferating blood components eliminate from the outset the problems of cell banking, passage number and donor cell-line comparability. Mayo Clinic/Rion's PEP product is manufactured from pooled human platelets by a patented multi-step ultrafiltration; it is stored as a lyophilised powder in a sealed glass vial at room temperature, with a reported shelf life of up to 24 months; approximately 200 billion exosomes/mL (up to 2 trillion per vial), ~100 nm, CD63/CD81 positive; nanoparticle tracking analysis (NTA) and transmission electron microscopy are used as release tests; the cargo contains PDGF, VEGF and TGF-β [65]. Exopharm's Plexaris platform, by contrast, uses LEAP (ligand-based exosome affinity purification) chromatography on platelet EVs [20].

Other sources documented in rodent wounds are: endothelial progenitor cells (Erk1/2-mediated angiogenesis) [77]; urine-derived stem cells (exosomal DMBT1 supporting angiogenesis in diabetic repair) [78]; macrophages (exosome-driven M1→M2 transition) [79]; bovine colostrum/milk EVs (including engineered miR-31-5p targeting HIF1AN; ~60% of miRNA intact after 5 days at 37 °C); probiotic bacterial EVs (Lactobacillus rhamnosus GG, miR-21-5p-like; L. reuteri, miR-21a-5p); keratinocytes, dermal fibroblasts, cord blood, amniotic fluid, dental/oral MSCs, saliva (UBE2O mRNA) and plant-derived exosome-like nanovesicles from wheat, ginseng, aloe vera, grapefruit, ginger and strawberry [72]. Plant-derived vesicles have now also entered the human wound case literature: a 75-year-old man with a 5 × 2 cm partial avulsion injury on the nasal dorsum with exposed cartilage received Rosa damascena stem cell-derived exosomes topically on days 6, 11 and 16 after injury; re-epithelialisation was reported at day 20 and healing at day 30 [80].

Preconditioning and engineering as potency levers

Some form of modification was used in 60 of the 83 preclinical wound studies (72.3%): 33 (42.3%) modified the parent cell — 21 by genetic modification (20 lentiviral, 1 knockout), 12 by culture conditions (hypoxia 2, 3D spheroid 2, selenium/LPS/IFN-γ/melatonin 6, bioreactor 1); 25 (30.1%) modified the isolated EV; of these, 22 added a biomaterial and only one performed cargo loading by electroporation (miR-146a) [6]. The bioreactor study reported a 40–80-fold increase in EV yield [6]. Hypoxic preconditioning of ADSCs (1% O₂, 24 hours) produced exosomes differing by 215 upregulated and 369 downregulated miRNAs (upregulated: miR-21-3p, miR-126-5p, miR-31-5p); in STZ-induced diabetic mice, wounds treated with hypoxic exosomes had closed almost completely by day 14, whereas normoxic exosomes produced partial healing [81]. Pioglitazone, atorvastatin, melatonin, coenzyme Q10, deferoxamine and low-intensity ultrasound are described as cargo-shifting levers that "integrate relatively smoothly into upstream bioprocess design" [82].

Because preconditioning changes the cargo, a preconditioned EV is a different product requiring its own characterisation and potency package. One review notes this explicitly: "although pretreatment may increase EV yield … it is unclear whether the differences it creates in EVs will affect safety" [44].

Donor status is a live and under-studied risk for autologous products: the only direct evidence is that endogenous keratinocyte-derived EVs in the wound fluid of patients with diabetes are "dysfunctional, quantitatively altered or compositionally impaired", and that PRP-EV miRNA signatures differ between healing and non-healing DFUs [82]. There are no data on the effect of donor age; nor was any study found comparing diabetic-donor MSC-EVs with healthy-donor MSC-EVs in a matched wound model. This calls either for allogeneic pooled starting material or for donor qualification to be defined as a release criterion.

Isolation is not a technical detail but a product-defining variable

As a didactic introduction, the three common methods can be summarised as follows [83]: Ultracentrifugation (UC) is the historical gold standard; it is low-yield, equipment-intensive and carries a risk of vesicle aggregation. Size-exclusion chromatography (SEC) preserves vesicle integrity, gives high purity and is increasingly preferred in clinical-grade manufacturing; however, non-vesicular material may co-elute. Tangential flow filtration (TFF) offers high scalability and is ideal for commercial production; it requires optimisation to prevent membrane fouling. These methods are not interchangeable: each yields a different vesicle population and a different contaminant profile.

The isolation distribution across the 83 preclinical wound studies is as follows: ultracentrifugation 81 (97.6%), ultrafiltration 46 (55.4%), precipitation kits 7, size-exclusion chromatography 1, combined techniques 54 (65.1%) [6]. The trade-offs are well defined: ultracentrifugation is accessible but co-isolates contaminants and carries a risk of vesicle damage; SEC preserves bioactivity but may co-elute non-vesicular material; immunoaffinity increases specificity but at the cost of recovery and scalability; polymer precipitation is simple, high-yield and "mostly impure"; microfluidic methods are not mature for manufacturing. The choice of isolation is "not a technical detail but a product-defining variable" and "the absence of a standard workflow leads to serious batch-to-batch variability" [5,82].

This creates a specific and rarely acknowledged translational risk: approximately 98% of the preclinical wound evidence rests on ultracentrifugation, whereas clinical-grade products use tangential flow filtration or ultrafiltration. The transition from preclinical to clinical therefore usually means a change in the EV population — a transition that requires a comparability study that is almost never reported.

Upstream, in the same 83 studies, collection conditions were serum-free medium 42 (50.6%), EV-depleted fetal bovine serum 22 (26.5%), chemically defined medium 2 (2.4%) and not reported 20 (24.4%); the most common conditioning period was 48 hours [6]. "EV depletion" of fetal bovine serum is incomplete: 18 hours of ultracentrifugation removed only ~70% of EVs in the 75–165 nm range and left 7.90 ± 0.34 × 10⁹ EV/mL; a commercial exosome-depleted serum removed ~75% and left 6.41 ± 0.14 × 10⁹ EV/mL — hence many published cell-derived EV preparations are contaminated with residual bovine EVs [84]. Human platelet lysate, including its EV-depleted and calcium chloride-defibrinated variants, is the established xeno-free route for GMP MSC-EV manufacturing [85].

Characterisation, dose metrics and potency

Characterisation practice in the 83-study wound sample: size distribution reported in 82 studies (98.8%) (NTA 57, dynamic light scattering 17, AFM/TRPS 3); morphology in 82 (TEM 75, SEM 7, cryo-TEM 1); surface markers in 83 (100%) but characterisation adequate to MISEV2023 in only 46 (55.4%); positive cytosolic markers missing in 37 (44.9%), purity/non-EV contaminant assessment absent in 52 (62.8%); protein measured in 45 (54.2%), RNA in 2 (2.5%); protein yield "rarely reported" (100 µg from 20 mL of ADSC conditioned medium in one study) [6].

Dose is the field's most consequential documentation failure. Reported doses are distributed as protein mass from 2 µg to 5 mg (66.2% of studies), particle number from 2×10¹⁰ to 2×10¹² (10.3%) and no dose reported at all (22.1%); only one of 68 studies normalised to body weight, and dose–response was formally tested in only one [9]. Other studies express dose as µg/cm² (83.76 ± 126.4 µg/cm² across nine meta-analysed studies) [16], µg/mL, RNA µg/mL, nmol miRNA or "1% v/v" in hydrogel. MISEV2023 adds the deeper problem: even a numerically precise particle count is not a precise EV dose — "quantification of yield and specificity for total EVs will most likely be an estimate, because particle number quantification is not always EV-specific" [4]. Because protein mass is the dominant metric and is precisely the quantity inflated by co-isolated soluble protein, dose reporting and purity reporting are not independent problems. The proposed solution is dual-metric release (particle number + protein mass) tied to a functional potency readout; because particle number alone "is misleading when formulations differ in cargo richness or contaminant load", and protein alone "may capture non-vesicular material" [82,86].

Potency assays should be phenotype-matched to the claim: endothelial migration/tube-formation rescue and perfusion for ischaemic wounds; macrophage polarisation and reduction of inflammatory mediators for inflamed wounds; reduction of bacterial load and protease tolerance for infected wounds; fibroblast migration, keratinocyte proliferation and collagen organisation for senescent wound edges. "Without a claim-linked potency assay, manufacturing scale-up will remain unstable" [82]. The ISCT 2025/2026 signature series frames the requirement thus: "manufacturing is not making a product — it is making the same product every time"; critical quality attributes must be "not only analytically reproducible but biologically meaningful", potency assays should be integrated early rather than validated retrospectively, and it must be clearly understood that increased particle yield does not guarantee biological activity [87]. The multi-donor mixed lymphocyte reaction has emerged as the leading functional assay for immunomodulatory potency — but there is no equivalent consensus assay for a wound-healing claim whose mechanism is irreducibly multifactorial. This is likely to be a rate-limiting step for licensure.

The only human wound study to apply a release potency assay is the Italian venous ulcer pilot: autologous serum EVs were released against a BrdU incorporation and tubulogenesis assay at 5×10⁴ s-EVs per target cell, all five lots yielded 57.75–70.9% of the positive control, and Bact/Alert sterility testing accompanied it [23]. This remains the most methodologically rigorous manufacturing description in the human EV wound literature.

Storage and stability

Consensus practice for liquid EV products is −80 °C, rapid freezing, the minimum number of freeze–thaw cycles and a stabilising buffer [43]. A single freeze–thaw cycle lost 23–36% of EVs from blood cell sources and degraded approximately 70% of EV miRNA; three cycles reduced EV quantity by 37–43% and increased size and polydispersity; bioactivity declined after five cycles [43]. Storage at −20 °C caused marked aggregation of umbilical cord MSC EVs within one month; 4 °C is acceptable for about one week; room temperature caused size growth and membrane degradation, EVs from one source showed "almost no biological activity" after 12 months at ambient temperature, and the potential for complete loss of bioactivity after ~4 days at 37 °C has been reported [43]. Phosphate-buffered saline (PBS), "the most commonly used buffer", itself causes aggregation and cargo loss, whereas protease inhibitors, trehalose and human serum albumin prevent particle loss; trehalose preserves the function of freeze-dried EVs [43]. Yet in the 83-study preclinical wound sample, storage was at −80 °C in 47 studies, the storage medium was PBS in 46, and freeze–thaw cycles or pre-isolation storage were reported in none of the 83 — so an unknown share of the between-study variability in reported potency may stem from storage rather than biology [6]. Source B's statement "stable at −20 °C" contradicts this storage literature and has not been used in this review.

Lyophilisation with a disaccharide lyoprotectant is the only documented route to a room-temperature-storable wound care product; PEP is the current proof point [65]. Regulators do not treat this as a convenience: lyophilisation "creates a distinct product presentation with its own formulation, manufacturing process, container and reconstitution requirements", and room-temperature shelf-life claims "must be based on long-term data in the intended container" [88].

Scale-up and batch consistency

Published GMP routes exist — human platelet lysate-based MSC-EV protocols, closed automated bioreactors with tangential flow filtration, large-scale bioreactor manufacturing and the platelet-derived PEP process [65,85,89,90]. Reported concentrations are 8.1 × 10¹⁰ particles/mL from a hollow-fibre (Quantum) run with 250–500 mL of circulating conditioned medium and 4–7 × 10⁹ particles/mL from microcarrier 3D systems; 3D culture with tangential flow filtration improves both yield and activity [89,91]. Bone marrow MSCs secrete approximately 1–4 µg of exosomal protein per million cells per day in standard 2D culture; hypoxia or genetic modification increases this 2–3-fold; umbilical cord MSCs show 3–4-fold higher yield in 3D systems; a single hollow-fibre bioreactor unit "produces exosomes equivalent to 70 T225 flasks per week" [92]. Batch-to-batch variability is acknowledged only qualitatively — "very large differences in EV markers and concentration were observed depending on the day of collection" — and no accessible source provided quantitative coefficient-of-variation data for particle number or potency [89].

Cross-process yield benchmarking is currently impossible from the published literature, because figures are given as particles/mL without volume, protein µg without particle count, and are never reported as particles per cell. The figures above should be read as illustrative, not comparable. (The citation details for Ref [92] were captured incompletely in Source A; the yield and cost-driver figures should not be quoted without re-verification against the primary article.)

Preclinical Evidence

What do the pooled animal data say?

Between 2022 and 2026, at least eight systematic reviews and meta-analyses of animal cutaneous wound EV studies have been published. All report large, statistically significant pooled benefits. The principal estimates are given below.

Pooled preclinical effect estimates (meta-analyses)

† Data-provenance caveat for ref [15]: Every pooled estimate attributed to this review should be read with the following caveat and should not be counted as an independent synthesis: the study quality score reported by the review (5.08 ± 0.752 on the modified STAIR tool) and the "peak efficacy at day 7" odds ratio (OR 1.82; 95% CI 0.69–2.95) are numerically identical to the mean quality score and day-7 pooled OR published four years earlier in ref [16]; it is, moreover, a single-author supplement-issue article that also presents the same author's uncontrolled clinical series of 738 patients (see Section 6.6). The identity of these two statistics raises a data-provenance question that the published article does not address. The review-specific pooled SMDs (closure 5.42, angiogenesis 5.48, re-epithelialisation 5.06, collagen 4.78, scar width −8.10) are therefore given only for completeness, not as corroboration of the other analyses [15,16].

Note: Table numbering follows the specification in Source A; Table 2 (mechanisms) is in Section 3 and Table 1 (sources) is in Section 4.

Three observations matter more than the numbers themselves.

First, the pooled effect varies roughly threefold (from SMD 3.16 to 8.41) across analyses drawing on overlapping pools of primary studies, and the smallest, most recent analysis reports both the largest effect and the clearest publication bias — exactly the pattern expected in a field dominated by small, unblinded studies [11,12].

Second, the only pooled analysis whose confidence intervals are compatible with "no benefit" is the one that uses odds ratios rather than standardised mean differences (day 7 OR 1.82; 95% CI 0.69–2.95; day 14 OR 2.29; 95% CI 0.01–4.56) [16]. The apparent certainty is determined by the choice of effect metric, not by the data. This is a useful corrective: SMDs of 5–9 imply that treatment and control distributions barely overlap at all, which is not plausible for any wound-healing adjuvant ever tested in humans.

Third, in absolute terms, where individual studies give proper numbers, the reproducible signal is faster healing rather than higher final closure. Representative values: in one burn/microneedle study, 51.67% versus 28.83% closure at day 5 and roughly 77% versus 43% at day 15, reaching 98.5% at day 20 [99]; roughly 50%, 75% and 90% closure at days 7, 14 and 21 with adipose-tissue MSC exosomes [100]; in an aged-mouse pressure ulcer model, EV-treated wounds healed like young controls at days 7 and 14 — but "all groups reached similar closure rates by day 21" [60]. The defensible summary is: a gain of roughly 20–30 absolute percentage points at days 5–15, with convergence by day 20–21.

Risk of bias, publication bias and the near-absence of negative studies

Every meta-analysis that assessed risk of bias reached the same conclusion. The SYRCLE assessment is "unclear for most studies owing to inadequate reporting" [9]; "all studies were rated at unclear risk of bias in most domains", with only one of 21 studies at low risk for blinding [12]; in the flap/graft meta-analysis, allocation concealment was unclear in 24 of 24 studies (100%), random housing in 20 of 24 (83%), caregiver/investigator blinding in 18 of 24 (75%) and outcome-assessor blinding in 18 of 24 (75%), and no study was rated at high risk in any domain — that is, the tool could not discriminate, because the papers did not report [14,101]. Beyond the bias domains: no study calculated a sample size or reported exclusions, and only 11 of 51 studies described temperature control [16]. Sex reporting was 57% male, 10% female and 33% not reported; the authors note that oestrogen affects wound healing but that stratification was impossible [12]. Follow-up was 8–28 days, most often 13–14 days — "insufficient to assess long-term matrix remodelling" [9,12,16]. Geographic concentration is extreme: 84% of the studies in the largest sample (57 of 68) come from a single country (China) [9].

Publication bias tests give a divided but weighted result. Positive findings: Egger P=0.000 and imputation of nine studies by trim-and-fill (type 2 diabetic wounds; direction of effect preserved after imputation) [12]; an asymmetric funnel plot with Begg Z=2.47, P=0.013 and Egger bias coefficient 5.86, t=4.52, P=0.004 (DFU models) [11]; Egger P<0.001 for both flap survival (6.735) and angiogenesis (9.088) — the authors concluded that "true effects are probably smaller than observed" and that the findings are "exploratory and hypothesis-generating" [14]. Negative findings: funnel plots showing "no significant asymmetry" at days 7 and 14 (the same authors also warned that "a culture of publishing only effective studies may have led to substantial bias") [16]; "no clear indication of publication bias" in any outcome in one diabetic wound analysis [13].

Against more than 80 positive studies, only two primary studies reported a negative direction of effect: ADSC exosomes overexpressing miR-20b, which suppressed Wnt9b/β-catenin and reduced closure (inhibition restored healing), and endothelial-cell exosomes preconditioned with advanced glycation end products, which reduced closure via ERK1/2-mediated autophagy [16]. A review tabulating roughly 34 in vivo diabetic wound studies found "no explicit negative/null findings" and observed that "quantitative outcome reporting: limited; most studies report qualitative or semi-quantitative ('increased', 'promoted') results rather than exact closure percentages" [17]. The subgroup nulls within the positive analyses are also instructive: in the DFU analysis, the BM-MSC subgroup (Z=1.70, P=0.09) and the medium-dose subgroup (P=0.11) were not significant; the dose subgroup difference (χ²=14.29, P=0.0008) was interpreted as a U-shaped, non-linear dose–response, the low-dose stratum was found to be the most stable (I²=42%), and the authors rated GRADE certainty as "limited" [11].

The species gap: there is no porcine cutaneous wound EV efficacy study

This is the single most important point of evidence hierarchy in the field. Among 68 animal studies there are 36 mouse, 30 rat, one macaque and one rabbit — no pig [9]. Among 83 studies there are 61 mouse and 22 rat, with no large animal at all [6]. Among the 24 flap/graft studies, "23 of 24 studies used small rodent models and only one used a miniature pig"; the identity and individual results of this single porcine study could not be recovered from the meta-analysis [14]. A 2025 diabetic wound review tabulating roughly 34 in vivo studies states "no porcine study presented" [17]. One meta-analysis author states the limitation directly: "reliance on rodent models remains a limitation; progress towards large-scale testing and more advanced in vivo models is slow" [13]. The line in Source B reading "UC-MSC-Exo hydrogel in a porcine full-thickness burn model" contradicts this finding and, as it could not be verified, was not included in this review.

The existing porcine study in the clinically most advanced EV wound programme is a pulmonary biodistribution and safety study, not a wound efficacy study [64,65]. The available large-animal cutaneous data are limited to the rabbit ischaemic ear wound: "PEP-Tisseel achieved complete wound closure with functional regenerated skin at day 28, whereas persistent wound defects were seen in the PEP-alone and Tisseel-alone groups" — a result taken from a review rather than the primary paper, with no dose stated [65].

The quantitative significance of this is as follows: roughly 90–93% of preclinical studies used full-thickness excisional wounds in rodents; these wounds close with a substantial contraction component, whereas human chronic wounds close largely by re-epithelialisation [6,9]. Splinted mouse models were developed precisely to force healing to proceed by granulation and re-epithelialisation, as in humans [18]; the magnitude of the contraction component in unsplinted mouse excisional wounds has been formally measured [19]. None of the accessible reviews and meta-analyses reported whether the included wounds were splinted; the share of "closure" in this literature attributable to contraction therefore cannot be estimated — it can only be identified as a structural bias. An aggravating point: in the type 2 diabetic meta-analysis, "re-epithelialisation" was pooled from only two studies, whereas "closure" came from 19 and more [12]; the literature preferentially measures the outcome most exposed to the contraction artefact and least measures the outcome that matters clinically.

Representative primary studies

What is absent from the preclinical literature

No verified EV study in a radiation-damaged skin model could be found — this matters because an ongoing human trial targets chronic radiation ulcers [103]. There is no study in the Zucker diabetic fatty rat; the verifiable diabetic literature is confined to the STZ rat/mouse and the db/db mouse. No registered report, multi-laboratory replication or independent confirmatory study of any EV wound finding could be found. There is no pooled estimate for time to closure in days, wound infection rate, or functional skin outcomes such as tensile strength or elasticity — only morphometric surrogates. And, critically for safety: across 68 studies "no study reported harmful events", but "assessment of possible adverse effects has received only minimal attention" [9]. That is not evidence of safety; it is the absence of safety assessment.

Human Clinical Evidence

The shape of the human evidence base

As of September 2026, the published human evidence consists of: two first-in-human platelet-derived EV (pEV) wound trials (both null for efficacy), a 110-patient randomised controlled trial (RCT) in diabetic foot ulcer (DFU) from Egypt, a completed 59-patient Phase 2a trial in DFU from the United States with results posted to the registry, a 4-patient venous leg ulcer (VLU) pilot from Italy, two Phase I perianal fistula trials from Iran, a Phase 1b skin-graft donor-site trial from the USA, three randomised scar/post-laser trials from South Korea, a handful of single-patient case reports and one large uncontrolled commercial case series. Alongside these stand several unfractionated conditioned-medium/secretome studies that are frequently — and wrongly — counted as EV evidence.

Total exposure to purified EV products in controlled or semi-controlled human wound trials is roughly 200–250 patients worldwide, with a few hundred more patients in uncontrolled commercial series. There is no Phase 3 trial, no approved product, and no peer-reviewed meta-analysis giving a pooled effect size for human EV wound-healing outcomes. The pooled human EV effect sizes that do exist belong to aesthetic outcomes: in a review including 39 aesthetic studies, wrinkle reduction 20.2% (95% CI 15.3–25.2), improvements in pigmentation/elasticity/texture/erythema 14.7–23.4%, hair density +23.6% (18.1–29.0) — of which only the acne scar and post-laser components are adjacent to wounds [104]. An unpublished SSRN preprint proposing a meta-analysis of exosome therapy in DFU exists; it is not peer-reviewed and is not cited here for any effect size [105].

The two first-in-human platelet-EV trials and what their convergence means

Exopharm Plexoval II (ACTRN12620000944932, Australia). First-in-human, randomised, double-blind, placebo-controlled Phase I trial in eleven healthy adult volunteers (11/11 completed; mean age 29.0; 8 male; 10/11 white); sponsor Exopharm Ltd; product platelet EVs purified by LEAP chromatography. Each participant received two 4 mm skin punch-biopsy wounds (one on each upper inner arm); one wound was randomised to pEV and the contralateral one to placebo — a within-subject paired design. The dose was 100 µg pEV in 340 µL, subcutaneous (300 µg/mL product), as a single injection on day 0 immediately after wound creation; placebo was 340 µL formulation buffer. Safety: "no study deaths, serious adverse events or adverse events leading to withdrawal were reported"; there were no clinically significant laboratory abnormalities. Efficacy was null: mean time to healing was 22.8 ± 8.7 days in treated wounds and 22.8 ± 8.7 days in placebo wounds; 11/11 participants had closed by day 30. The authors attributed the lack of effect to the wound beds of healthy volunteers not being deficient in the cargo the product carries [20].

Rion PEP in split-thickness skin graft donor sites (NCT04664738, USA). Phase 1b, prospective, open-label, dose-randomised, within-subject controlled trial; 7 subjects completed (3 low dose, 4 high dose), mean age 49.3; topical platelet-derived purified exosome product (PEP), highest dose tested 20% (2 vials); the contralateral donor site received standard of care. Primary outcomes were safety plus the Vancouver Scar Scale and photographic wound assessment. There were no "treatment-related adverse events". Re-epithelialisation took a median of 18.5 days with PEP and 19.25 days with standard of care — no statistically significant difference. The authors recommended larger efficacy trials [21]. The registry record shows the trial as completed (2 February 2024), the sponsor as Rion Inc. (not Mayo Clinic), an enrolment of 8, three interventions listed (10% PEP, 20% PEP, Tisseel fibrin sealant) and no results in the registry [106]. Several 2026 review articles still list this trial as "ongoing"; this is out of date.

These are two entirely independent trials — different sponsors (Exopharm / Rion), countries (Australia / USA), products (LEAP-purified pEV / PEP), registries (ANZCTR / ClinicalTrials.gov), populations (healthy volunteers / surgical patients) and wound models (4 mm punch biopsy / split-thickness donor site). That two unrelated sponsors, with two different platelet-EV products, in two different acute standardised wound models, each with within-subject control, found no difference in healing versus control is the most important convergent finding in this field. It is also the finding least reflected in the review literature.

An earlier Exopharm trial (PLEXOVAL, first dosing announced on 29 January 2020, "up to 20 participants" under the Australian Clinical Trial Notification scheme) is described in company and press sources; whether it is distinct from Plexoval II and whether its results were published could not be determined; it is therefore not counted as evidence here.

The single positive randomised controlled trial and its problems

Wharton's jelly MSC-exosome gel, diabetic foot ulcer (NCT06812637, Egypt). A 20-week (4 weeks treatment + 16 weeks follow-up) double-blind (triple-blind in the registry) randomised trial at Kafrelsheikh University; 110 enrolled, 85 analysed (exosome + standard of care n=30; standard of care alone n=24; vehicle placebo + standard of care n=31). Patients had type 2 diabetes, mean age 52 ± 8 (42–62), ulcers <30 cm², neuropathic/ischaemic/mixed, unresponsive to standard of care. Wharton's jelly MSCs were expanded in DMEM/F12 with 15% fetal bovine serum; exosomes were obtained from 48-hour serum-free supernatant by ultracentrifugation at 13,000 g, 45,000 g and 110,000 g for 5 hours; transmission electron microscopy reported 32–34 nm vesicles; flow cytometry CD9 86.7%, CD63 89.3%, CD81 94.7%, HSP70 94.6%. Exosome concentration and dose were never quantified. The gel was applied topically after debridement; once weekly for 4 weeks, then every 3 days. The primary outcome was percentage reduction in ulcer size at week 16. Reported results: complete healing in the exosome arm at a mean of 6 weeks (4–8) vs a mean of 20 weeks (12–28) in control; 53/85 (62%) complete healing overall; median ulcer area in the treatment arm fell from 6 cm² at baseline to 4 cm² at week 2 and 2.7 cm² at week 4, with between-group P=0.001 reported. Wound infection 2/30 (6%) in the exosome arm vs 7/24 (29%) in control; osteomyelitis 1 (3%) vs 2 (8%); in the exosome arm, fever 2 (6%) and bullae 3 (10%); one death in the placebo arm judged unrelated. Clinicians ruled that there were no adverse events attributable to the gel [22].

The methodological problems are serious and must be weighed together with the headline result. Differences in ulcer size reduction were not significant at weeks 2 and 4; separation emerged only from week 6 onwards, and the paper's reporting of these comparisons ("p>0.01", "p>0.05") is internally inconsistent; no exact p value is given for the headline complete-healing comparison. Attrition was 25/110 (23%), unevenly distributed across arms (10/11/4), and the registry arm sizes (40/35/35) differ from the published completers (30/24/31). Most importantly, baseline ulcer area differed markedly between arms — roughly 6 cm² in the treatment arm, 2 cm² in the standard-of-care arm and 2.7 cm² in the placebo arm. This threefold imbalance works against the treatment arm (larger ulcers should heal more slowly), and some may read it as strengthening the result; but a threefold baseline imbalance in the primary prognostic variable is also a randomisation-failure signal, undermines confidence in the entire allocation process, and makes "healing at 20 weeks" in a control arm of 2 cm² ulcers implausibly slow against published standard-of-care benchmarks [32,33]. Blinding is described inconsistently between the paper and the registry, the exosome dose is not quantified, no funding or conflict-of-interest statement could be found, and the trial is single-centre. The product description also contains internal inconsistencies: 32–34 nm on transmission electron microscopy is below the conventional small-EV range, and serum-containing medium was used for expansion.

Certainty for this claim (topical Wharton's jelly MSC exosome gel accelerates DFU healing): low, arguably very low. It is a randomised, placebo-controlled trial — which would normally start high — but it is downgraded twice for risk of bias (baseline imbalance signalling randomisation failure, inconsistent blinding description, 23% uneven attrition, arm differences from the registry, absence of exact p values, unquantified dose) and once for imprecision/indirectness (single centre, single country, 85 analysed patients, no independent replication).

Rion Phase 2a PEP-TISSEEL trial (NCT06319287): results posted to the registry

At the time this review was prepared, the results of the trial referred to as the field's most important awaited data have been posted to ClinicalTrials.gov (results submitted 20 July 2026; registry verified 22 September 2026) [25,26]. This information overrides secondary texts that describe the same record as having "no results" during 2026.

Design. Phase 2a, prospective, randomised, open-label, standard-of-care-controlled trial; although the official title says "multicentre", only one site is listed in the registry (Professional Education & Research Institute, Blue Ash, Ohio). Sponsor Rion Inc. Registered status: completed; primary completion 23 April 2025, study completion 1 November 2025; last update 22 June 2026. Arms: Arm A — 2 vials of lyophilised PEP in 10 mL TISSEEL fibrin sealant (15 mg/mL PEP-TISSEEL) applied topically once weekly for 12 weeks + standard wound care (dressings and offloading devices); Arm B — Fibracol (collagen-alginate dressing) + the same dressing and offloading protocol, 12 weeks. A 3-month post-treatment safety follow-up in both groups.

Participant flow. Actual enrolment 59 patients. Arm A: 28 started, 18 completed (10 did not complete). Arm B: 31 started, 20 completed (11 did not complete). The non-completion rate was approximately 35% in both arms.

Baseline characteristics. Baseline ulcer area averaged 2.7 cm² in both arms (Arm A ±1.79; Arm B ±2.30) — unlike the Egyptian trial, the arms are balanced. Ulcer duration 8.0 ± 1.37 vs 7.5 ± 1.48 weeks; plantar location 21/28 vs 21/31; hypertension 22/28 vs 25/31; peripheral neuropathy 21/28 vs 25/31; previous lower-extremity amputation 17/28 vs 14/31; age ≥65 13/28 vs 6/31 (Arm A older); male 24/28 vs 24/31; BMI 30.2 vs 30.7 kg/m².

Primary outcomes. Complete wound closure at week 12 (ITT population): Arm A 13/28 (46.4%) vs Arm B 7/31 (22.6%). Dose-limiting toxicity: 0/28 vs 0/31. No p value or confidence interval is given in the registry.

Secondary outcomes. Percentage area reduction at week 12 (mITT set 1): Arm A 78.9% ± 33.2 (n=23) vs Arm B 66.2% ± 47.0 (n=24). Mean change on the visual analogue pain scale 1.5 ± 2.15 (n=23) vs 0.6 ± 1.65 (n=22). Semmes-Weinstein score 1.9 ± 2.90 (n=14) vs 5.7 ± 4.31 (n=16). Wound-Q subscales (n=14/arm) physical function 47.9 vs 62.1, lower extremity 59.6 vs 72.4, sleep 67.4 vs 83.2 — these quality-of-life values numerically favour control; being small subsets, they should not be interpreted.

Safety (6-month follow-up). Serious adverse events Arm A 2, Arm B 6; other adverse events 8 vs 5; deaths 0 vs 1 (Arm B). Serious infections in Arm A: sepsis (1), cellulitis of a left plantar midfoot ulcer (1), left lateral foot abscess (1). In Arm B: severe sepsis (1), right foot abscess/infected DFU (1), COPD exacerbation with pneumonia (1). Non-serious foot-related events 13 in Arm A and 9 in Arm B (including new wounds, infection and reopening of the index ulcer).

This review's own calculation from the registry counts (not the sponsor's analysis; performed for certainty grading): absolute difference 23.8 points (95% CI 0.2–47.5), risk ratio 2.06, two-sided Fisher's exact test P=0.062. With n=59 the difference does not reach the conventional significance threshold; the confidence interval is wide enough to touch zero.

Other controlled and quasi-controlled human studies

Autologous serum-derived EVs in chronic venous leg ulcer (CS2/1095/0090491; registered programme NCT04652531 SER-VES-HEAL, Italy). Single-centre pilot case-control study in 4 patients with 5 lesion pairs (one patient enrolled twice); median age 77, 75% female, median ulcer duration 11 months, all unresponsive for ≥3 months. Twenty-two mL of serum was obtained from 50 mL of whole blood; EVs were isolated by charge-based precipitation with PEG 400 + protamine hydrochloride (1:4) under GMP-compliant conditions with Bact/Alert sterility testing; they were characterised by transmission electron microscopy, Western blot (CD63/CD9/CD81), MACSPlex flow cytometry and ELISA, and a release potency assay was applied (BrdU incorporation + tubulogenesis at 5×10⁴ EVs per target cell; all five lots at 57.75–70.9% of the positive control). Approximately 3 mL was injected along the wound margins with a 25 G needle three times weekly for two weeks (6 doses). The control was within-patient: the larger lesion was actively treated, the smaller received standard care — a deliberately worst-case, non-randomised allocation. Results at day 30: surface area reduction 385 mm² with EVs, 106 mm² with sham, P=0.004; 75–100% granulation in 3 of 5 EV-treated lesions, none with sham; increased CD31+ microvascular proliferation on histology. "No local or systemic adverse reaction or adverse event was recorded." Funding came from a grant that included Unicyte, and one author was on the Unicyte scientific advisory board [23]. The registered programme (SER-VES-HEAL, University of Turin), with an estimated 10 patients, a once-weekly peri-wound injection schedule for 3 weeks and a contralateral-ulcer control, appears as "recruiting" as of May 2023, with no results posted [107].

This study's potency assay, sterility testing and histological confirmation make it, methodologically, the most rigorous manufacturing description in the human EV wound literature — but n=4 without randomisation and 30 days of follow-up are hypothesis-generating only. Certainty: very low.

Placental and umbilical cord MSC exosomes in complex perianal fistula (Iran). There are two related Phase I studies. The first (Pak et al. 2023, PMID 36640153) treated complex perianal fistula in non-Crohn's patients with human placental MSC-derived exosomes; as the full text could not be accessed, sample size, dose and results could not be extracted from the primary source; a secondary summary describes the study as safe, with significant healing and no adverse effects [96,108]. The second is fully verified: an open-label Phase I in 5 patients (3 men, 2 women, median age 35, range 31–47) with refractory perianal Crohn's fistula unresponsive to anti-TNF-α for ≥6 months; IRCT20200413047063N3, dual registration NCT05499156; umbilical cord MSC-derived exosomes, 5 mL of a 50 µg/mL solution (250 µg in total), a single injection into the fistula tract, no control group; 4 of 5 patients (80%) responded — 3 complete fistula closures, 1 partial healing; skin inflammation regressed within one month in all symptomatic patients; "five patients (100%) reported neither systemic nor local adverse effects" and there were no laboratory abnormalities at 6-month follow-up [95]. The linked registration (NCT05499156) lists placental MSC exosomes and n=80, whereas the publication used umbilical cord exosomes in n=5, and there is a third, overlapping registration (NCT05402748, placental MSC exosomes, non-Crohn's complex anal fistula, planned n=80, Tehran) — the same group holds at least three overlapping fistula-exosome registrations with inconsistent product and sample-size fields [109,110].

Randomised scar and post-laser trials (South Korea). These are adjacent to wounds, not chronic wound evidence; but they are the most methodologically robust randomised EV skin trials in existence.

Kwon et al. 2020: a 12-week prospective, double-blind, randomised, split-face trial in 25 completers with atrophic acne scars (18 men, 7 women, aged 19–54, all Korean, Fitzpatrick III–IV, ECCA ≥50); block randomisation, sealed envelopes. The product was an ADSC conditioned-medium-derived exosome (ExoSCRT, ExoCoBio): mode size 117.4 nm, stock 3.26×10¹¹ particles/mL, CD9/CD63/CD81 positive, calnexin and cytochrome-C negative. Dose 9.78×10¹⁰ particles/mL in gel on the treatment day, thereafter 1.63×10¹⁰ particles/mL; 1 mL to one half of the face, twice daily for 2 days after each session; following three fractional CO₂ laser sessions at 3-week intervals (10,600 nm, 7.2–9.0 mJ, 240–300 µs, 20–25% coverage, 2 passes); the opposite half of the face received control gel. Primary outcome: ECCA reduction 32.5% on the exosome side (95% CI 24.8–40.2), 19.9% on the control side (12.2–27.6), P<0.01; ≥2-grade improvement on the Investigator Global Assessment 16/25 vs 12/25, P=0.02; post-laser erythema significantly lower on the exosome side (P=0.03); healing time 4.1 vs 4.3 days (P=0.03); mild hyperpigmentation 1 vs 2 (P=0.32); no scarring or permanent events. The trial was not registered, follow-up was 6 weeks after the last session, the population was of a single ethnicity, and although the authors declared no conflict of interest, the manufacturer produced and supplied both the active and the control gel [111]. An independent systematic review applying RoB 2 rated this and the other scar RCTs as "some concerns", mainly because of unclear randomisation and lack of blinding [112].

Park and Park 2026: randomised 75 patients (25/arm) 1:1:1 with 3–6-week-old postoperative facial scars (hypertrophic and keloid excluded); blinded outcome assessors (three plastic surgeons) and blinded image analysts, treating clinician not blinded. Arms: (I) 1064 nm Nd:YAG fractional non-ablative laser alone; (II) laser + human ADSC conditioned-medium exosome (Exomide); (III) laser + plant-derived exosome (Exome). Five sessions at 2-week intervals; follow-up only 2 weeks after the last session. Greyscale intensity reduction 2.9 vs 11.3 vs 13.1 (P=0.01), modified Vancouver Scar Scale improvement 0.7 vs 2.5 vs 2.8 (P=0.03), observer POSAS 3.1 vs 5.4 vs 5.3 (P=0.03) — but patient-reported POSAS 2.5 vs 2.8 vs 2.7 (P=0.26, not significant), and there was no difference between human- and plant-derived exosomes on any outcome (all P>0.05). Transient erythema or burning in two participants in Arm II. The authors state their limitations explicitly: no power calculation, no prospective registration, 2-week follow-up, baseline scar dimensions not systematically recorded; the study is explicitly exploratory and "should not be regarded as definitive evidence of efficacy or superiority" [97].

Jeong, Park and Jung 2026: a prospective, double-blind, randomised, split-scar trial in 10 patients (3 men, 7 women, mean age 30.6 ± 3.3; mean scar age 4.3 months, range 1–11 months) with a linear scar on the anterior chest wall after costal cartilage harvest for revision rhinoplasty. Each scar was divided into medial and lateral halves; following microneedle radiofrequency (SCARLET, two passes), 1 cc of an ADSC exosome formulation (ASCE+ SRLV, ExoCoBio; ExoSCRT technology) was applied topically to one half and 1 cc of hyaluronic acid (HA) to the other; once weekly for 8 consecutive weeks. Quantitative assessment at 8 weeks; 1-year follow-up by qualitative photography only. Results: Vancouver Scar Scale pliability 1.4 ± 0.52 vs 2.3 ± 0.67 at week 4 (P<0.05; significant at weeks 4–6), no difference in height and vascularity; Observer Scar Assessment Scale pigmentation 2.6 ± 1.51 vs 4.7 ± 1.83 at week 3, pliability and relief 2.0 ± 1.15 vs 3.6 ± 1.58 at week 8 (P<0.05), no difference in thickness and vascularity; Patient Scar Assessment Scale colour 2.5 vs 4.1 at week 4, stiffness 3.3 vs 4.3 at week 3, thickness 3.4 vs 4.4 at week 2 (P<0.05), no difference in pain/itch. No adverse events reported, no wound dehiscence. The product was supplied by ExoCoBio and the company contributed to the scientific discussions; the trial was not registered. Limitations stated by the authors themselves: HA's own wound-healing effect (which may make the exosome effect appear smaller than it is), the absence of an RF-only control arm (RF itself remodels scars), the subjectivity of the scales, n=10 and histology limited to a single case [113].

The plant-versus-human null finding is analytically important. If a Rosa or other plant-derived preparation performs identically to a human ADSC-derived preparation on every outcome, a substantial part of the observed post-laser benefit may not be specific to EV cargo but may arise from vehicle, occlusion and moisturising effects. In the Jeong study, the fact that the control was a biologically active substance (HA) raises the same question from the opposite direction. The appearance of Rosa damascena stem cell exosomes in a published human wound case report reinforces this concern rather than resolving it [80].

Conditioned-medium and secretome studies that should not be counted as EV evidence

hUC-MSC derivatives in DFU (NCT06825884, Jordan). Phase I/II, single-centre, open-label, single-arm, uncontrolled, n=10 (7 men, mean age 52.2, mean diabetes duration 13.3 years, mean HbA1c 8.9%, University of Texas stage II–III, mean ulcer duration 10.5 months, mean baseline area 5.58 cm²). The product is explicitly "unfractionated conditioned medium derived from hUC-MSCs, the complete secretome encompassing soluble growth factors, cytokines and naturally released extracellular vesicles (including exosomes)", and the authors acknowledge that the EV/exosome fraction was not separately isolated or analysed. Manufacturing used the explant method, α-MEM with 5% platelet lysate, harvest at passage 4, 0.22 µm filtration and storage at −80 °C; measured content EGF 89.1 pg/mL, CXCL12 1,159.4 pg/mL, TGF-β1 3,151.3 pg/mL. Dose once-weekly 5 mL perilesional injection, up to 10 sessions, 24-month follow-up. All 10 patients reached complete closure at a mean of 4.2 weeks (P<0.00001) — stage II at a mean of 20.1 days, stage III at 35.0 days; area fell from 5.58 ± 5.3 to 0.00 cm² by day 35; no recurrence at 24 months. Adverse events were mild and transient: injection-site pain (7), erythema (4), oedema (4), itching (3), rash (1), fever (1); no amputation or death [114].

The 24-month recurrence-free follow-up is a genuine strength and the safety profile is reassuring; but a single-arm design cannot attribute 100% closure to the intervention — chronic DFUs also heal with intensified standard care — and the product is not a purified EV preparation. Certainty for efficacy: very low (the design cannot support an efficacy claim). A Phase 2 trial of 10% secretome-hUC-MSC gel in chronic diabetic and trophic ulcers was published in Indonesia in 2023; its design, sample and results could not be extracted [115].

Case reports and commercial series

Four wound case reports merit attention. In a 38-year-old man with cerebral palsy, a recurrent 5×4×5 cm right ischial pressure ulcer, colonised with MRSA and resistant to 9 months of wound care and antibiotics after a failed posterior thigh flap, received ExoFlo (MSC-derived EVs), 1 cc in 4 cc of saline, delivered into the wound base and walls by six weekly injections over 8 weeks; complete healing at week 8, durability for 2 years until recurrence due to prolonged pressure; no adverse events, regulatory pathway not stated [116]. In a man in his 60s with a 0.7×1.1 cm scalp defect that had not healed for 7 months after angiosarcoma resection and chemoradiotherapy, PEP — first in collagen (which leaked), then from week 26 in Tisseel fibrin sealant — achieved complete epithelialisation [117]. A 48-year-old man with a 192 cm² deep second-degree burn of the dorsum of the foot from a petrol fire received allogeneic bone marrow MSC-EVs (AGLE-102) within 48 hours of injury as a single topical dose of approximately 1×10⁷ particles/cm², in sterile isotonic solution, under an occlusive film dressing; the report describes >99% closure within one week (from 192 cm² to 1 cm²), closure sustained for 52 weeks without additional doses, POSAS falling from 8 at week 1 to 2 at week 52, pain regressing by week 4 and itching by week 8, prevention of burn wound conversion with perfusion normalisation documented by laser Doppler within one week, and no adverse events, seroconversion or development of panel-reactive antibodies [24]. Finally, a 46-year-old woman with a deep second-degree scald burn of the thigh (1% total body surface area) received once-daily topical umbilical cord MSC exosomes (ExoVia, BIONET Therapeutics) for nine consecutive days after one week of standard care had produced no healing; complete closure at day 18 and "excellent skin quality with minimal scarring" at one month; dose not stated [93].

Double-counting caution. The burn case report above is almost certainly the publication of the single patient enrolled in Aegle Therapeutics' burn trial NCT05078385. The registry record shows actual enrolment of 1 (estimated target 10), status Completed, start 10 August 2023, study completion 19 November 2024, a protocol permitting up to 3 applications (within 48 hours, days 5–7, week 2) and no results posted; the sites include Ryder Trauma Center, Miami — and the dose (1×10⁷ particles/cm²), the single dose within 48 hours of injury, the laser Doppler burn-conversion outcome, the POSAS outcome and the Miami author group match this protocol exactly [24,118]. The Aegle burn "trial" and this "case report" should be counted once, not twice. A company press release describing the same patient states "99% epithelialisation within 7 days" and a POSAS fall "from 40 at baseline to 10 at day 90"; these figures differ from the peer-reviewed report's POSAS values falling from 8 at week 1 to 2 at week 52 — they are different instruments or subscales and should not be conflated [119].

The largest human dataset in this field is also the weakest. A retrospective, uncontrolled commercial case series of human placental MSC exosomes (XoGlo, Kimera Labs) reports a total of 738 patients; of these, 175 received wound management (131 women, 44 men, aged 19–72, mean 48) and 118 scar treatment (78 women, 40 men, aged 21–56, mean 32); administration was topical and intralesional, combined with nitric oxide supplementation, dermoelectroporation, ultrasound, acoustic shock wave and amniotic membrane scaffolds; in a keloid subset of 21 patients, 18 (85.7%) were recurrence-free at 2 years (intralesional approximately 1×10⁹ exosomes/mL), increased neovascularisation without skin grafting was reported in burn cases, and it is stated that there was "no allergic reaction, hypersensitivity response or adverse event attributable to exosome therapy" [15]. There is no control group, randomisation or blinding; there are heterogeneous confounding adjunctive procedures, non-standardised outcome definitions and a commercial product — and the author is simultaneously the treating clinician and the systematic reviewer who included his own series in his own meta-analysis. The 2-year keloid follow-up is the only genuine strength. Certainty: very low; this is clinical experience, not evidence. Three small aesthetic case reports/series (n=3 atrophic scars and post-procedure wounds; n=1 nodulocystic acne scarring, ECCA from 180 to 90) are in the same tier [112].

The conclusion reached by every systematic review of the human data

Independent systematic reviews converge. A review restricted to human placental MSC-exosome studies found only 4–5 human studies meeting inclusion criteria against 21 animal studies; it called this "a significant translational barrier", noted that "reliance on rodent models and methodological heterogeneity limit direct clinical extrapolation" and that "the available human evidence rests predominantly on early-phase studies, case series and observational studies", and assessed risk of bias for none of the included human studies [15]. This review requires an additional caution: the study quality score and the day-7 "peak efficacy" odds ratio it reports are numerically identical to figures published four years earlier in an independent systematic review; it is a single-author supplement-issue article and the largest human dataset it contains is the author's own uncontrolled 738-patient series. Its pooled estimates should not be treated anywhere in this field as an independent synthesis [15,16]. A review examining human clinical outcomes of MSC-exosome treatments for scarring, ageing and hyperpigmentation found six human studies, rated all randomised trials as "some concerns" on RoB 2 and concluded that "the current evidence remains preliminary owing to small samples, heterogeneous designs and limited follow-up durations" [112]. A 2025 review examining EV clinical applications found that translation "remains largely focused on early-phase trials and case studies, lacking the support of large-scale randomised controlled trials", confirmed that no EV-based wound or skin product has FDA approval, and noted that the only skin/wound registered trial it detailed (NCT02565264) had not published results [120]. A 2026 review examining EV therapy in diabetic skin lesions names exactly one human study — the Egyptian randomised trial — and gives no registration number [82]. A 2024 review on exosomes in diabetic wound healing, despite promising to summarise clinical advances, presents no human study with an identifier, enrolment count or result; its evidence is entirely rodent [17].

The review literature both outnumbers and runs ahead of the primary human evidence; several reviews recycle the same four to six studies and, through citation, create the illusion of a broader evidence base.

Documented unreliable secondary reporting

Two verified examples deserve to be named explicitly, because they directly affect how the reader should evaluate this literature.

First: Arkoubi 2026 and NCT05125562. A single-author review in Frontiers in Medicine (Arkoubi AY, "Role of exosome therapy in reconstructive surgery: a review", 18 March 2026) states that "an exploratory Phase II trial (NCT05125562)" of hypoxic MSC-derived exosomes achieved "71% complete closure in Wagner grade 2 diabetic foot ulcers … versus 33% with standard care" at week 12 [38]. The identifier resolves — but to nothing resembling this description. NCT05125562 is "Extracellular Vesicle Infusion Treatment for Mild-to-Moderate COVID-19", sponsored by Direct Biologics LLC; Phase 2; the intervention is a single intravenous ExoFlo infusion on day 1 in three randomised dose groups between 7×10¹¹ and 10.5×10¹¹ extracellular vesicles; status WITHDRAWN; dates 7 December 2022 – 7 March 2023 [39]. It is a withdrawn (never enrolled) intravenous COVID-19 trial of a different company's product — not a Phase II DFU trial, not hypoxic MSC-derived exosomes, not topical. The cited 71% versus 33% closure figures cannot be traced to any registered trial. The same review's claim that "a search of ClinicalTrials.gov revealed more than 20 registered exosome clinical trials for wound healing and scar reduction" is made without a list; an independent count found roughly ten to fifteen wound/scar EV interventional registrations worldwide. The review's evidence table also cites a "Thakur et al." hypertrophic scar pilot (n=12, 2.3-point gain on the Vancouver Scar Scale) and an "Allenson et al." split-face rejuvenation randomised trial (n=30, 31% wrinkle reduction at 6 months); neither author name could be matched to an indexed exosome scar or rejuvenation study, and the review's reference list could not be accessed — these remain unverified and should not be repeated.

Second: a 2026 Frontiers in Immunology review that turns a null result into a positive one. Li Y et al., "Application of stem cell-derived exosomes in skin wound healing: mechanisms, prospects, and challenges", tabulates approximately 19 human "clinical applications" of exosomes in skin — DFU, deep second-degree burn, skin graft donor sites, non-healing scalp wound, ischaemic nasal and auricular flaps — without giving a registration identifier for any, and summarises the platelet-EV skin graft donor site trial as having achieved "100% re-epithelialisation, shortened healing time" [5]. This directly contradicts the primary publication of NCT04664738, which reported median re-epithelialisation of 18.5 versus 19.25 days without a statistically significant difference [21]. The same review, to its credit, also reminds readers that a frequently misattributed 53-patient DFU dataset reporting increased neovascular formation and complete or progressive healing was hUC-MSC cell transplantation, not EV therapy [5].

Taken together, these two cases show that 2026 narrative reviews in this field systematically convert null or untraceable primary results into positive claims. Another variety of the same pattern was seen in the preparation of this review: although the results of NCT06319287 were posted to the registry in July 2026, a detailed evidence review prepared within the same year described this trial as "no results"; it was corrected by a direct re-reading of the registry record on 22 September 2026 (see Section 6.4). Clinicians and guideline developers should work from primary reports and registry records, not from review tables.

Aggregate human safety signal

Across every verified human study — 11 healthy volunteers with paired wounds, 85 analysed DFU patients (Egypt), 59 randomised DFU patients (Rion Phase 2a), 7 graft donor site patients, 5 Iranian fistula patients, 10 Jordanian DFU patients, 25, 75 and 10 patients in the Korean scar trials, one burn patient with no seroconversion or development of panel-reactive antibodies at 52-week follow-up, and several hundred patients in uncontrolled commercial series — no human study has attributed a serious adverse event to an EV wound product. In the Rion Phase 2a, serious adverse events were fewer in the EV arm than in control (2 vs 6) and the single death was in the control arm; the three serious infections in the EV arm (sepsis, cellulitis, abscess) lie within the natural history of the DFU population itself and were not linked to the product in the registry [26]. Reported events were injection-site pain, erythema, oedema and itching (Jordan), transient erythema and burning (Korea), mild hyperpigmentation (Kwon) and fever and blistering in a minority (Egypt, considered not attributable). This is the most consistent finding in the human literature. It is nevertheless a weak safety dataset: total exposure is a few hundred patients, follow-up is mostly ≤6 months, event detection is not standardised, and there is no immunogenicity dataset apart from the single burn patient.

Note: Registry fields were taken from server-side-generated ClinicalTrials.gov records and primary publications; the NCT06319287, NCT04173650, NCT05078385, NCT06793748, NCT06429033, NCT06812637, NCT02565264, NCT05475418, NCT04652531 and NCT05243368 records were verified directly from the ClinicalTrials.gov data interface on 22 September 2026. Fields that were not visible are marked "not reported". The "n=20 VLU platelet-EV pilot RCT", "n=5 refractory DFU hydrogel case series" and "n=30 donor-site ADSC-exosome spray RCT" mentioned in one of the source texts could not be matched to any registration or publication and were not included in the table.

Evidence by Wound Type

The distribution of evidence across wound types is highly uneven, and the mismatch between where animal studies are concentrated and where human studies exist is informative in itself. Diabetic wounds dominate the preclinical literature (39 of 83 studies used a diabetic model; 90.4% were full-thickness excisional) [6] and are also the sole indication of the two controlled human studies — the Egyptian RCT and the Rion Phase 2a. Burns have substantial rodent evidence and two single-patient human reports. Pressure ulcers have one aged-mouse model and one human case report. Venous leg ulcers have essentially no animal model; there is one 4-patient human pilot and its registered programme. Epidermolysis bullosa has no animal evidence but has the field's only matched-wound controlled trial. Radiation ulcers have one registered human trial but no verified animal model study.

A structural caveat applies to all wound types. The preclinical models showing the largest effects are those in which perfusion is the limiting factor — diabetic, ischaemic flap and ischaemia-reperfusion pressure ulcer models — because angiogenesis is the mechanism EVs most reliably affect [14,16]. These same models are the ones most likely to overstate benefit in human wounds where the limiting factor is not capillary density but infection, biofilm, unrelieved pressure or inadequate offloading. EVs are not antimicrobial and do not oxygenate tissue; the rodent evidence should not be read as predicting benefit in an infected, poorly offloaded ulcer. The fact that the standard-of-care arm of the Rion Phase 2a included dressings and offloading devices is instructive in showing that EV is being developed only as an adjuvant added on top of correctly delivered standard care [25,26].

Therapeutic Delivery Methods

The routes actually used and what is known about them

The dominant route of administration in rodent wound studies is subcutaneous or intradermal injection around the wound; topical application in a carrier comes second, and intravenous (IV) administration appears in only a small fraction of studies [131]. In pooled diabetic wound analyses, doses ranged from 1 µg to 2 mg of EV protein, follow-up lasted 12–21 days, and topical use was described as "limited" [13]. When no carrier is used, repeated dosing is the rule; in a burn study with an engineered vesicle, 1.0 µg/µL was injected on days 0, 3, 6 and 10 [132].

In humans the picture is different and is converging on two formats. The first is topical application in a carrier: the weekly Wharton's jelly exosome gel in the Egyptian study [22]; PEP reconstituted in TISSEEL fibrin sealant in Rion's trials — the graft donor-site study [21,106], the completed Phase 2a DFU trial with results posted to the registry (NCT06319287; 2 vials PEP + 10 mL TISSEEL, 15 mg/mL, once weekly for 12 applications; see Section 6.4) [25,26] and the chronic radiation ulcer protocol [103]; topical AGLE-102 under an occlusive film dressing in the burn patient [24]. The second is perilesional or intralesional injection: 3 mL along the wound margins in the Italian venous ulcer pilot [23]; 5 mL perilesional in the Jordanian secretome study [114]; 1 cc intralesional ExoFlo in the pressure ulcer case [116]; 250 µg into the fistula tract in the Iranian Phase I trial [95]; 100 µg subcutaneous in Plexoval II [20]. Systemic administration is not used in wound indications.

There is an unresolved tension between the preclinical and clinical directions of travel. The 83-study meta-analysis found subcutaneous injection superior to dressing/covering application for closure, collagen and revascularisation [6]; yet clinical development has moved towards topical gels and lyophilised powders in sealant. One primary study, meanwhile, found IV administration significantly faster than local injection (P<0.05) — the exact opposite of the meta-analytic subgroup [100]. The most likely explanation is that the operative variable is not the route but the delivered dose; this is also consistent with the recurring criticism that hydrogel studies rarely define the delivered dose [82]. No study in this literature could be found that compared injection, topical and hydrogel delivery head-to-head in a single model.

Why carriers were adopted: not proven superiority, but a retention problem

The rationale for every biomaterial strategy is the documented retention problem described in Section 3.5: roughly 1% of IV-administered EVs remain at 24 hours, locally administered EVs are rapidly cleared into the surrounding tissue and capillaries, and repeated dosing itself "may delay the natural healing process" [61]. There is also a didactically plain fact: EVs dripped onto a wound in free suspension are easy to apply, but proteases in the exudate and dilution rapidly inactivate these vesicles; "topical free EV" is therefore in practice the easiest and biologically least efficient format. There are no quantitative wound-specific retention data at all — the EV half-life in the wound bed, and the percentage remaining at 24 or 72 hours after topical or intradermal application, are unknown [63].

Evidence that carriers are superior to free EVs

Only two studies with cleanly matched arms support a quantitative claim.

A thermosensitive, self-healing, pH-responsive hydrogel (FHE: Pluronic F127 + oxidised hyaluronic acid + ε-poly-L-lysine) gels in about 10 seconds at 37 °C and degrades faster under acidic wound conditions (about 20% mass remaining at pH 7.4 and about 1% at pH 5.5 on day 13). In diabetic mouse full-thickness wounds, day-14 closure was 88.67 ± 6.9% for FHE@exosome, 76.3 ± 3.2% for free exosome, 64.3 ± 9.8% for hydrogel alone and 36.3 ± 10.4% for saline; about 45 versus about 20 new vessels per field were counted, and formation of skin appendages with less scarring was seen only in the loaded hydrogel group [133].

A sprayable, photo-crosslinked methacrylated decellularised dermal matrix loaded with hUC-MSC exosome and β-cyclodextrin–borneol gave release of 85.57 ± 4.04% at 48 hours, 86.67 ± 3.51% at 120 hours or 87.67 ± 2.08% at 168 hours depending on crosslinking time. In 10 mm wounds in STZ-diabetic mice, the remaining wound area on day 14 was 1.07 ± 1.27% with the loaded hydrogel, 8.25 ± 1.64% with free exosome, 13.83 ± 1.90% with hydrogel alone, 14.33 ± 3.51% with a commercial chitosan bio-dressing and 21.04 ± 2.90% in the control [134].

In both studies the free EV arm outperformed the empty carrier and the loaded carrier outperformed both; the carrier effect and the EV effect thus appear to be additive rather than interchangeable. But these are two rodent studies. The only meta-analysis designed specifically to pool hydrogel-delivered and free EVs is behind a paywall; only its conclusion that it "holds clinical translation potential" and that "further research and clinical trials are needed" was accessible, and the pooled effect sizes could not be obtained [135]. The accessible diabetic wound meta-analysis included six studies using biomaterial carriers but did not perform a carrier-versus-free subgroup comparison [13]. The 2025 meta-analysis pooling animal studies of platelet-derived EVs in diabetic wounds (9 studies, 128 animals) found an overall healing SMD of 4.43 (95% CI 2.85–6.01) and reported an SMD of 7.96 (5.05–10.87) in the hydrogel-combination subgroup — same direction, but this is a subgroup comparison and does not come from matched arms [98]. Certainty that carrier delivery is superior to free EV in humans: no evidence. In animals: very low; supported by individual controlled studies, not by pooled analysis.

Release kinetics are tunable — and cluster around the duration of a rodent experiment

Release can be tuned from hours to about 27 days: PEG hydrogels give profiles adjustable between 6 and 27 days; Pluronic F127 gels in about 17 seconds at 37 °C; transglutaminase-crosslinked collagen prolongs release; polysaccharide hydrogels release EVs in a pH-responsive manner; MXene–M2 exosome nanohybrids provide sustained release for up to 7 days [61]. Release durations cluster at 7–14 days, which coincides with typical rodent closure endpoints and makes it difficult to distinguish the claim of "sustained release" from the statement "a single application was sufficient". The most defensible practical advantage is dose sparing and fewer applications.

Microneedles, sprays and printed scaffolds

The best-documented EV microneedle study used a detachable GelMA/PEGDA array (16 × 16 needles, 600 µm) bearing β-cyclodextrin-acryloyl for loading; 100 µg/mL HUVEC exosome was co-loaded with tazarotene, achieving >80% cumulative exosome protein release over 10 days in vitro and fluorescence-verified release up to day 8 in vivo. In STZ-diabetic mice the combination patch outperformed the patch alone and patch + drug in closure, collagen, re-epithelialisation, tensile strength and CD31/α-SMA vessel density — but because there was no exosome-alone arm, the superiority of the microneedle over topical EV has not been shown even in the rodent [136]. Other formats include detachable bilayer patches with M2-macrophage exosome in the tips and polydopamine behind, Fe-MSC nanovesicle/polydopamine core-shell hyaluronic acid microneedles and PVA hydrogel-forming microneedles. A dedicated 2025 review states explicitly that microneedle wound technology is "still in the development stage", lacks a unified standard and has "no clinical trial validating its efficacy" [137]. Sprayable bioadhesive microcarriers loaded with thymosin-β4-engineered ADSC exosome, cryogenic 3D-printed scaffolds and stereolithographic EV-GelMA constructs exist as primary studies but without verified release or comparison data.

There is no human study of EV-loaded microneedles, cryogels, nanofibres, decellularised matrix scaffolds or 3D-printed constructs. Specifically for electrospun nanofibre EV dressings, no primary study that could be verified as an EV-loaded electrospun wound study was found — this is a category asserted in reviews but not demonstrated.

Combination Therapies

The combinations with the strongest rodent support are those in which the co-modality corrects a defect of the wound environment that EVs cannot remedy — hypoxia, bacterial burden or oxidative stress. This is consistent with the mechanistic fact that EVs support regeneration but are neither antimicrobial nor oxygenating. None of these combinations has human evidence, and for several important pairings there is no evidence at all.

EV + platelet-rich plasma, with a protein-matched comparator. In STZ-diabetic rats, a single topical application of 1% v/v PRP-derived exosomes within an alginate hydrogel disc was compared with untreated wounds, alginate alone and protein-matched PRP in hydrogel. PRP-exosomes closed faster than all comparators including PRP at every time point ("almost closed by day 14"), with greater vessel area and number and longer neo-epithelium; the mechanism is YAP dephosphorylation and nuclear translocation in fibroblasts driving CTGF, and in endothelium PI3K/Akt and Erk1/2 via exosome-carried VEGF, bFGF, PDGF-BB and TGF-β [138]. This remains the single best-designed study in which an EV product is compared with its non-vesicular parent preparation.

EV + small-molecule drug. GelMA/PEGDA microneedles loaded with HUVEC exosomes and tazarotene outperformed patch alone and patch + drug in closure, collagen, vessels and tensile strength in diabetic mice — but there is no exosome-only arm [136].

EV + oxygen-generating material. An antioxidant polyurethane/calcium peroxide cryogel loaded with ADSC exosomes, releasing oxygen for more than 10 days, healed both diabetic and infected diabetic rat wounds within two weeks; faster closure and re-epithelialisation, more collagen and neovascularisation, reduced oxidative stress and mature epithelium containing hair follicles were seen [129].

EV + hyperbaric oxygen — animal only. The combination of hyperbaric oxygen + M2-macrophage-derived exosomes was compared with hyperbaric oxygen alone, exosomes alone and untreated arms in STZ type 2 diabetic rats: the combination gave the highest wound contraction rate at days 7 and 14, the most angiogenesis, fibroblast density and collagen organisation, the strongest increase in TGF-β and VEGF, a decrease in TNF-α and IL-1β, the highest catalase, superoxide dismutase and glutathione with reduced malondialdehyde, and superior tensile strength [139]. The review devoted to this combination states that "clinical studies of the combination of hyperbaric oxygen with exosomes have not yet been conducted" and points to a specific mechanistic hazard: in a hyperoxic environment "the lipid membrane of exosomes may undergo oxidation and degradation" [62]. Most of the animal literature it cites is not cutaneous (spinal cord injury, spinal ischaemia-reperfusion, sciatic nerve).

EV + antibacterial agent. In the FHE hydrogel the antibacterial function is provided by ε-poly-L-lysine [133]; vancomycin-loaded ginger-derived vesicles within a Pluronic F127 thermosensitive hydrogel target S. aureus; tannic acid/chitosan-dopamine hydrogels with 3D-cultured MSC-EVs exploit tannic acid's interference with bacterial cell wall synthesis; polysaccharide hydrogels with MSC-EVs have been reported with "adequate antibacterial capacity against multidrug-resistant bacteria" [61]. The most concrete verified example of making the vesicle itself antibacterial is the GL13K antimicrobial peptide conjugated to watermelon-derived EVs via mussel-inspired DOPA chemistry, delivered in a Pluronic F127/chitosan hydrogel that gels at 37 °C: >95% inhibition of S. aureus and E. coli in vitro, approximately 60% closure at day 5 and near-complete closure at day 14 in infected rat wounds, with increased collagen and M2 polarisation [130]. No in vivo study against established mature biofilm could be verified; biofilm efficacy should therefore not be claimed — the infected models used are acute inoculation models.

EV + phototherapy. There are polydopamine photothermal backings on M2-exosome microneedle tips and iron-MSC nanovesicle/polydopamine core-shell microneedles [137]; an NIR-II photothermal telluride-selenide exosome construct for diabetic pressure ulcers has been reported at title level only.

EV + conventional advanced dressing. The sprayable dECM-exosome hydrogel was benchmarked against a commercial chitosan bio-dressing and outperformed it (residual area at day 14 1.07% vs 14.33%) — one of very few examples in which an experimental EV construct was compared with a commercial product rather than saline [134].

Combinations with no evidence at all, and those that are "clinically sensible but without direct evidence". Combining Source C's clinical reasoning with Source A's evidence scan, the picture is as follows:

The NPWT gap is probably a practical rather than an incidental one: suction and sharp debridement will physically remove the EV load, and no published study appears to have tested an EV instillation protocol (e.g. NPWT with instillation). This is a genuine open question, not a closed negative result.

Human combination evidence exists only in the aesthetic and scar setting — fractional CO₂ laser + topical ADSC exosomes [111], Nd:YAG laser + exosome "booster" [97] and microneedle radiofrequency + ADSC exosomes [113]; the second of these found plant-derived and human-derived products indistinguishable (see Section 6.4).

Certainty for all EV combination therapies in human wounds: no evidence.

Safety and Adverse Effects

The structural problem with EV safety data

Because no EV product holds marketing authorisation anywhere in the world, there is no post-marketing pharmacovigilance dataset for EV therapeutics. Almost all documented human harm comes from unapproved products used outside trials; this skews the visible safety signal towards contamination and procedural events rather than intrinsic EV pharmacology. At the same time, usable safety data in the animal literature is almost non-existent: across 68 studies "no study reported harmful events" but "only minimal attention was paid to the assessment of possible adverse effects" [9]. This is an absence of assessment, not evidence of safety.

Safety data from clinical trials

The only pooled safety estimate for human EV therapy across all indications comes from a systematic review and meta-analysis in which 21 reports entered the safety analysis: serious adverse events 0.7% (95% CI 0.1–5.2) and any adverse event 4.4% (95% CI 0.7–22.2); serious event rates were similar for autologous and allogeneic administration, the overall adverse event frequency was significantly higher with autologous products but of unclear clinical meaning; the authors concluded that "EV-based therapy is safe and offers a promising opportunity in treatment" [27]. This conclusion must be read against its own confidence intervals: the upper bounds of 22% for any adverse event and 5.2% for serious events are derived from small, largely open-label early-phase studies with non-uniform detection and cannot exclude clinically meaningful harm; none of the pooled studies is a wound indication.

Specifically for wounds, the picture summarised in Section 6.9 applies: in no verified human study has a serious adverse event been attributed to an EV wound product; total exposure is a few hundred patients and follow-up is mostly ≤6 months. The largest controlled dataset added to this picture in 2026 is the PEP-TISSEEL Phase 2a study (see Section 6.4): over 6 months of safety follow-up, serious adverse events were 2 in the PEP-TISSEEL arm and 6 in the standard-care arm; other adverse events 8 versus 5; deaths 0 versus 1 (standard-care arm); dose-limiting toxicity 0 in both arms. The serious infections in the PEP arm (sepsis 1, cellulitis 1, abscess 1) are consistent with the expected course of a DFU population and were not attributed to the product in the registry; however, this is open-label, single-centre registry data without a peer-reviewed publication, and adverse event attribution was made by the sponsor [26].

Documented harms from unapproved products

The 2019 Nebraska cluster is the defining safety event of this field. The FDA received "multiple recent reports of serious adverse events" in patients in Nebraska who received unapproved exosome products; the reports were flagged by the Centers for Disease Control and Prevention (CDC) and the Nebraska Department of Health and Human Services; the FDA issued a Public Safety Notification on 6 December 2019 and a Public Safety Alert on 9 December 2019 [1,140]. The Nebraska alert documented isolates of E. coli, Enterobacter cloacae and polymicrobial agents; described presentations with bacteraemia, fever, chills and injection-site reactions; and listed product contamination and "possible tumour growth" among the risks [28]. A public health presentation later quantified the cluster as five cases of E. coli sepsis developing within hours of intravenous administration of a contaminated placenta-derived product (bacterial isolates matching clinical specimens), and added that FDA inspections since 2020 had identified 50 violations at facilities, including contamination with hepatitis B, E. coli, Klebsiella pneumoniae and Streptococcus species [31]. The figure of five cases comes from this secondary presentation, not from a primary FDA or CDC count; the archived CDC page describes "serious adverse events including bacterial infections" without giving a case count [141].

Dermatology-specific harms cluster in two mechanisms. The first is *infection and contamination*; disruption of the skin barrier by microneedling or intradermal injection amplifies this — the public health assessment explicitly states that risk is increased with invasive compared with topical application [31]. The second is *foreign body or granulomatous reaction to co-injected non-EV material*: the first reported case is a 50-year-old woman who developed skin-coloured papules and nodules on both cheeks 7 weeks after intradermal injection of an unapproved exosome product marketed for wrinkle reduction and skin whitening; histopathology showed "granulomatous reaction with histiocytic infiltration around amorphous basophilic material", stains for fungi, mycobacteria and common filler agents were negative, and there was minimal response to doxycycline, intralesional steroid and hyaluronidase [29]. The amorphous basophilic material is the diagnostic clue: this was not an identifiable filler but excipient or unpurified residue; in other words, the factors that determine risk in cosmetic use are the impurity profile and product identity. Other published reports include a case series of adverse reactions after intradermal exosome formulations [142], a case of necrosis after dermal injection of lyophilised exosomes [30], a review of complications after exosome therapy for aesthetic rejuvenation [143] and a report on high-frequency ultrasonography in the management of these reactions [144]. The public health assessment tabulated eight individuals with injection-site papules/nodules, one foreign body granuloma and one case of skin necrosis three days after intradermal injection [31].

Under-detection is substantial. A commentary relays, alongside the Nebraska sepsis cases, cases of rapid cancer progression after exosome injection in Japan and unverified patient deaths that were not officially reported, and documents the scale of the market: 669 medical institutions in Japan (mostly cosmetic) offer exosome therapy, and about 60 businesses linked to roughly 38 companies market these interventions in the United States [145]. Because these products are sold outside any approval pathway, adverse event reporting is voluntary or absent; published case counts should be read as a floor. An AI-assisted multi-domain analysis evaluating 18 manufacturers and 70 product formulations, 2.7 million social media posts, 4,350 non-scientific articles and 37,437 consumer reviews found only 18% of manufacturers to be highly transparent, showed that growth factor concentrations differed significantly by source (H=18.73; P<0.01) and classified 27% of claims as misleading; positive sentiment (54%) is driven by provider-influencer endorsements [146].

Mechanistic hazards with experimental support

Thrombogenicity is the strongest quantitative safety signal in the EV field, and it is route-specific. Intravenous administration of human umbilical cord MSC-derived large EVs caused rapid death from pulmonary thromboembolism in mice at doses ≥1 µg/g body weight within the 0.125–4 µg/g range; "UC-EVs activated coagulation in a dose- and tissue factor-dependent manner", tissue factor pathway inhibitor blocked the procoagulant activity, and the authors proposed slower infusion or heparin prophylaxis as precautions [147]. Because EV-derived tissue factor is an established clinical biomarker of thrombosis in cancer and acute leukaemia, the mechanism is relevant to humans. This finding concerns intravenous or systemic dosing and platelet- or large-EV-rich preparations; the ≥1 µg/g lethal threshold should not be generalised to topical wound dressings — but it is directly important for any future systemic EV strategy and for high-dose intralesional injection into highly vascularised tissue.

Tumour promotion is model-dependent and bidirectional. Rat bone marrow MSC-derived exosomes increased the proliferation, invasion and metastasis of colorectal cancer stem cells and reduced their apoptosis [148]; reviews document both tumour-promoting and anti-tumour effects of the same MSC-EV preparations [149,150]. The defensible statement is that a theoretical risk exists with animal and in vitro support but no human evidence. Two clinical corollaries that follow from reasoning rather than citation, and are flagged as such: chronic wounds with malignant transformation potential (Marjolin's ulcer) and patients with occult malignancy are the populations in which this risk would matter; no source identified in this review addresses this specific clinical link.

The human dataset on immunogenicity is almost non-existent. It is listed as a risk category in the Japanese professional guideline [151]; immune cell-derived EVs are noted to require particular attention because they are "rich in a wide variety of immune function-related molecules" [152]; the pooled clinical trial analysis found a higher, but uninterpretable, overall adverse event rate with autologous than with allogeneic products [27]. The single most informative human datum is the burn case report's finding of no seroconversion and no development of panel-reactive antibodies 52 weeks after a single allogeneic dose — encouraging, but n=1 and a single dose [24].

Aberrant angiogenesis and fibrosis/hypertrophic scarring are mechanistically plausible but evidentially empty. EV cargo is pro-angiogenic and pro-proliferative, and the same literature documents pro-fibrotic EV activity (Section 3.3). In none of the sources underpinning this review was a direct citation found attributing aberrant angiogenesis or excessive scarring to an EV wound product. These should be stated not as observed events but as theoretical hazards to be captured prospectively in trials.

Biodistribution is discussed only qualitatively. "Adverse biodistribution" is a named risk category [151], and interpretation is methodologically confounded because a signal "may represent intact product, released cargo, free label or label-derived material" [88]. No source has quantified hepatic or splenic accumulation of applied EVs, and there is no wound-specific biodistribution data; the only verified large-animal study is the porcine pulmonary distribution of a platelet-derived product [64].

Donor pathogen transmission follows existing biological product frameworks: transmission of infectious disease (viruses, bacteria, fungi) is risk category number 1 in the Japanese guideline [151], hepatitis B is among the contaminants found on FDA inspection [31], and xeno-free media free of animal-derived components are recommended to eliminate contaminants derived from fetal bovine serum [151]. Prion risk is a logical consequence of animal-derived raw materials, but no EV-specific prion guidance or event has been identified.

Manufacturing, consistency, sterility and storage as a safety issue

The authoritative framework is this: "safety control begins with the source cells and the manufacturing process and cannot rely on final vial testing alone"; the safety-relevant quality domain covers microbial control, viral safety, endotoxin, residual DNA and protein, and source-specific risks [88]. The Japanese guideline defines eight risk categories: transmission of infectious disease; undesirable immune reactions; impurity contamination; undesirable changes in EV properties; adverse biodistribution; organ and developmental toxicities; tumour formation or abnormal tissue development; failure to demonstrate the expected efficacy [151].

Batch consistency is a safety issue, not merely a commercial one: upstream parameters (MSC source, culture density, hypoxic preconditioning) substantially alter EV composition and potency, and increased particle yield does not guarantee biological activity [87]; "very large differences in EV markers and concentration were observed depending on the day of collection" [89]. No accessible source provided quantitative data on batch-to-batch variability (coefficient of variation for particle count or potency).

What manufacturing failure actually looks like in this sector is documented in an FDA warning letter sent to a company producing amniotic products (not exosomes, but the same regulatory and technical space): inadequate media fill validation, missing environmental monitoring, unvalidated cleanroom disinfection, two sterility failures within four months without root-cause investigation, expiry dating without stability data, and unvalidated stopper and cap sterilisation [153]. An exosome-specific warning letter sent to a clinic in February 2026 applied the 21 CFR 1271.10(a) minimal manipulation and homologous use criteria to classify an umbilical cord product and exosomes as unlicensed biological products and unapproved new drugs, and explicitly cited the FDA's exosome safety notification and "multiple serious adverse event reports" [154].

Storage and stability failures have been quantified: a single freeze-thaw cycle costs 23–36% of particles and approximately 70% of EV miRNA; three cycles lose 37–43% of particles and increase polydispersity; bioactivity declines after five cycles; PBS itself promotes aggregation and loss [43]. The regulatory expectation is this: long-term data at −80 °C should form the primary basis of shelf life, permitted freeze-thaw cycles should be established experimentally, lyophilisation constitutes a separate product presentation with its own formulation, process, container and reconstitution requirements, and room-temperature claims should rest on long-term data in the intended container; in-use stability should cover dilution, transfer, light exposure and adsorption losses [88]. Source B's statement "stable at −20 °C" contradicts this literature: marked aggregation within one month at −20 °C has been documented for umbilical cord MSC EVs [43].

Finally, a definitional safety problem: the product may not be what it claims to be. ISEV warns that clinics claiming "purified" EVs may not be delivering purified material and are exposing patients to "dangerous products containing impurities and pathogens such as bacteria or viruses" [2]. An aesthetic medicine review concludes that complications "typically arise from non-sterile or mischaracterised products, cGMP non-compliance, or injection of minimally processed materials mislabelled as purified exosomes" [155].

Regulatory Status

The single most important regulatory fact

No regulatory authority in the world has approved an EV or exosome medicinal product for any indication. This is independently confirmed by the FDA ("There are currently no FDA-approved exosome products", 2019) [1], ISEV ("as of June 2020, there are no approved extracellular vesicle- or exosome-based therapies worldwide") [2], the Japanese professional guideline ("pharmaceutical products containing EVs as the main active ingredient have not yet received regulatory approval", 2025) [151], an aesthetic medicine review ("no regulatory authority in the world has yet approved an exosome product for any medical indication", 2026) [155], a global regulatory analysis [3], a Chinese government trade portal for China [156] and a public health assessment for Canada [31]. No source asserting any approval was found; however, a formal negative check of every national register was not possible.

Four categories must be kept strictly separate, because commercial and clinical communication routinely conflates them:

1. Marketing authorisation for a medicinal product — does not exist for any EV product anywhere in the world.
2. Investigational clearance — exists: FDA IND clearances for AGLE-102 in burns (15 May 2018) and in dystrophic epidermolysis bullosa (20 May 2019); Fast Track (1 October 2020) and Rare Pediatric Disease (6 October 2020) designations for DEB; no orphan drug or RMAT designation was found for an EV wound product [119,127].
3. Manufacturing site GMP authorisation — exists in at least two countries (ExoCoBio in South Korea; Erciyes University GENKÖK in Türkiye) and is repeatedly presented in the trade press as if it were product approval. It is not. [157,158]
4. Cosmetic notification or ingredient listing — exists in several countries and confers no right to any therapeutic claim.

Country-by-country status

United States. Exosomes intended to treat disease "are regulated as drugs and biological products under the Public Health Service Act and the Federal Food, Drug, and Cosmetic Act" [1]. The 361 HCT/P exemption does not apply: FDA warning letters apply the minimal manipulation and homologous use criteria of 21 CFR 1271.10(a) to classify these products as unlicensed 351 biological products and unapproved new drugs; hence an IND is required for clinical use and a BLA for marketing [153,154]. The peer-reviewed rationale is that exosomes "are not a tissue in the conventional sense" and that their intended effects are "to modulate inflammation, promote regeneration" [159]. The FDA has no EV-specific guidance; sponsors are expected to "apply existing biological product guidance according to the characteristics of the final product", and early consultation is recommended if the active constituent or potency strategy is undefined [88]; the FDA is also running a regulatory science activity on quality control in EV biomanufacturing. Enforcement is active: a warning letter sent to a Nevada clinic in February 2026 covered an umbilical cord product together with exosomes and explicitly referred to the exosome safety notification and to "numerous reports of serious adverse events" [154]. Exosome cosmetic ingredients are marketed with PCPC/INCI listing, and the US has the largest exosome cosmetics market, predominantly plant- and MSC-derived (consultancy source) [160].

European Union. The critical and frequently misreported point is this: "EVs and cell fragments of human cell origin do not by themselves meet the current ATMP definition"; classification depends on the characteristics of the final product, and the Committee for Advanced Therapies (CAT) classification procedure is optional and non-binding [88]. Most human-cell-derived EV products will therefore be handled as biological medicinal products rather than ATMPs unless they contain cells or genetic material. Several secondary reviews assert that EVs are ATMPs in the EU; this claim contradicts the EMA position as relayed in the near-primary regulatory literature, and this review flags the contradiction rather than resolving it — no EMA or CAT primary document was opened in the research underlying this review, and the EMA reflection paper on ATMP classification (EMA/CAT/600280/2010 rev.1, 21 May 2015) remains the document to which a sponsor must refer [161]. The governing framework is Regulation (EC) No 1394/2007, which entered into force on 30 December 2008 [162]. If a product is classified as an ATMP, the EMA's 2025 guideline governing quality, non-clinical and clinical requirements for investigational advanced therapy medicinal products applies [163]. In cosmetics, the Cosmetics Regulation (EC) No 1223/2009 prohibits ingredients of human origin; human-derived exosomes are therefore prohibited in EU cosmetics (Annex II, "cells, tissues or products of human origin"; the specific entry number is taken from secondary sources and has not been verified against the Official Journal) [160,164]. A 2026 peer-reviewed analysis of European and global EV regulatory challenges exists but could not be accessed [165].

United Kingdom. The MHRA is the competent authority under the Human Medicines Regulations 2012 and the Human Tissue (Quality and Safety) Regulations 2007; the EU-derived ATMP regulation was retained after Brexit [162]. An aesthetic medicine review states that "the MHRA is aligned with EU standards and treats exosome formulations as biological substances subject to batch reproducibility and defined potency testing" [155]. No MHRA exosome-specific guidance document, safety alert or approval was found; claims about illegal clinical use of human-derived exosomes in the United Kingdom appear only in industry and advocacy sources. This is the most weakly sourced regulatory line in the review and should be verified directly with the MHRA before being relied upon.

Japan. Japan has the widest legal gap, and this gap is being exploited at scale. EV and culture supernatant treatments that physicians prepare in their own practices and administer on site fall outside the Act on the Safety of Regenerative Medicine (ASRM), which covers only cell-based and gene-based therapies, and are therefore not reviewed by a Certified Special Committee for Regenerative Medicine [151]. Professional societies have directly expressed concern about "the global trend of unapproved products and preparations claimed to contain EVs and exosomes being prescribed to patients by physicians in private clinics" [151]; a society official stated that "some manufacturers circumvent the ASRM criteria by offering cell-free formulations such as lyophilised culture supernatant products, thereby entirely avoiding accredited committee review and government oversight" [166]. The result is 669 Japanese medical institutions offering exosome therapy, mostly for cosmetic purposes, with minimal restriction where patients pay privately [145]. The regulatory response is ongoing: the PMDA Science Board report (2023) addressed EV-specific considerations, the PMDA has published early quality considerations for first clinical trial notifications under the heading "new modality: extracellular vesicle products" [167], and draft quality guidelines for natural and engineered EV products are in preparation, with MHLW publication planned for fiscal year 2026; until then requirements remain case by case [88]. The Japanese professional guideline itself is now the most detailed risk taxonomy in the field (Section 10.5) [151]. In cosmetics, stem cell culture supernatant is conditionally permitted under the MHLW Guideline on Biologically Derived Materials (Notification No. 210, 2003) with strict advertising control [160].

South Korea. Korea has the world's only dedicated EV therapeutics guideline: the MFDS "Guideline on Quality and Non-clinical Evaluation of Extracellular Vesicle Therapy Products" (안내서-0917-03), current version 26 September 2024 (previous version 17 July 2023), issued by the Cell and Gene Therapy Products Division; it covers product quality, non-clinical safety and clinical evaluation [157]. An independent assessment notes that the guideline "regulates EVs as biological products (not cell therapy)" and "provides an EV-specific development framework but does not set universal specifications or numerical limits" [88]. The legal basis is the Act on Advanced Regenerative Medicine and Advanced Biopharmaceuticals, enacted in 2019 and implemented in 2020; oversight is shared by MOHW and MFDS [162]. There is no MFDS-approved exosome drug; the reported MFDS GMP authorisation for ExoCoBio's exosome biopharmaceuticals is a facility authorisation [157]. Cosmetic human-derived culture fluids are regulated under Article 8 of the Cosmetics Act, Annex 3 "Safety Standards for Human Cell and Tissue Culture Fluids", and the MFDS has taken enforcement action on the labelling and claims of human stem cell culture fluid cosmetics [160].

China. The NMPA Center for Drug Evaluation (CDE) published the "Scope, Classification and Interpretation of Advanced Therapy Medicinal Products" in a June 2025 draft, formally defining extracellular vesicles including exosomes as advanced therapy drugs; "no exosome pharmaceutical product has currently received market approval in China" [156]. The device route is closed: the October 2024 classification guidance ruled that stem-cell-derived exosomes should not be regulated as medical devices, and the December 2025 guidance classified a "sodium hyaluronate exosome membrane liquid dressing" as a product requiring evaluation under combination product procedures — directly relevant to any exosome wound dressing [156]. In cosmetics, human cells, tissues and human-derived products are prohibited as ingredients under the Safety and Technical Standards for Cosmetics (2015); the NMPA reaffirmed this in September 2024, and use of the term "exosome" or related efficacy claims is non-compliant [156,160]. China hosts the largest share of registered EV clinical trials (approximately 61% by one count; 34.6% of human-derived EV trials on ClinicalTrials.gov) [3,168]. A Chinese regulatory science paper on quality evaluation of stem-cell-derived EV therapeutics exists but could not be accessed [169].

Türkiye. In Türkiye, clinical trials of medicinal products are regulated by the Turkish Medicines and Medical Devices Agency (TİTCK) and clinical research ethics committees; licensing of advanced therapy medicinal products is governed by the Regulation on Licensing of Advanced Therapy Medicinal Products (May 2023; preceded by the draft ATMP guideline of 8 November 2018) [170]. No exosome-specific TİTCK classification document, guideline or enforcement action was found; Turkish-language searches returned clinic marketing pages rather than agency documents. The notable development is that on 24 July 2026 TİTCK granted the Erciyes University GENKÖK centre a manufacturing site authorisation certificate for the production of umbilical cord MSC-derived exosomes for clinical use; this was announced as the first such authorisation granted to a public institution in Türkiye and was reported again in the press in September 2026 [158]. This authorisation is a permission to manufacture under GMP conditions; it is not a product licence, and since it rests on a single news source it should be confirmed with the agency. As of today there is no licensed EV or exosome medicinal product in Türkiye; clinical use of an exosome product for a wound is therefore possible only within the framework of a clinical trial approved by an ethics committee and TİTCK, with the volunteer's informed consent. The Regulation on Promotion and Information Activities in Health Services (12 November 2025) prohibits the promotion of methods that are not scientifically proven or not defined by the Ministry; this review is likewise for information purposes only within that framework. Together with the United Kingdom, this line is the most weakly sourced regulatory line in the review.

Contextual countries. Canada: as of 2 April 2025, "there are no approved exosomes of any kind given by any route in Canada"; unauthorised products circulate in aesthetic clinics, and Section 13 cease-use orders have been recommended [31]. Taiwan: the TFDA published case-review regulations for exosomes on 21 March 2024 and revised documentation requirements on 2 February 2026; the Regenerative Medicine Act was promulgated on 19 June 2024 [160,162].

Three regulatory contradictions the reader should know

First, the claim in a 2025 review that "no drug regulatory authority has published specific technical evaluation guidance for extracellular vesicle-based drugs" is an over-generalisation: the Korean MFDS guideline exists and Japan's MHLW guidelines are planned for fiscal year 2026 [88,152,157]. Second is the EU ATMP classification contradiction described above [88,155,159]. Third, the claim in a 2026 aesthetic medicine review that Japan and South Korea operate "accelerated review pathways for exosome-containing cosmetics and regenerative injectables" is not supported by the Japanese professional guideline or the Korean MFDS documents and should be treated with scepticism [151,155,157].

Practical conclusion for the reader

A physician who applies an exosome product to a wound outside a clinical trial in the US, EU, United Kingdom, China or Türkiye is administering an unapproved drug. In Japan the same act may be legally possible outside ASRM review; that is precisely why the professional societies have issued guidance. Everywhere in the world today, the appropriate route is enrolment in a clinical trial, not clinical use.

Comparison with Existing Wound Care Treatments

The comparator is not a single product but a bundle of care — and that bundle is better evidenced than most adjuvants

The answer to whether an EV product "works" depends on what the product is compared against. In diabetic foot ulcer (DFU), the comparator is the standard care bundle defined by the International Working Group on the Diabetic Foot (IWGDF): "local debridement, offloading, revascularisation, treatment of infection where required" and "basic wound dressings" that absorb exudate and maintain a moist healing environment [34]. The practical guideline makes this concrete: sharp debridement repeated as needed; dressing selection to control exudate; a non-removable knee-high offloading device as the preferred treatment for neuropathic plantar ulcers; prompt treatment of infection; consideration of revascularisation if ankle pressure is below 50 mm Hg or the ankle-brachial index below 0.4; perfusion assessment if the ulcer does not heal within 4–6 weeks [171]. The United Kingdom's NICE NG19 guideline lists the same components — offloading, infection control, assessment and control of ischaemia, debridement, appropriate dressings — as standard DFU treatment [172]. In venous leg ulcer (VLU), the comparator is compression [126].

Only two components of this bundle rest on moderate-certainty evidence. Non-removable knee-high offloading devices versus removable devices: ulcer healing RR 1.24 (95% CI 1.09–1.41; moderate certainty); infection RR 0.58 (0.34–0.99; low); non-adherence RR 0.07 (0.01–0.79; very low); offloading device versus therapeutic footwear, healing rate RR 1.39 (0.89–2.18; low) and plantar pressure 239 kPa lower [173]. In VLU, compression versus care without compression: complete healing RR 1.77 (95% CI 1.41–2.21; 10 trials, 1,215 participants) and HR 2.17 (1.52–3.10; 5 trials, 733 participants) for time to healing; moderate certainty, downgraded one level for risk of bias, most trials small [126].

Two conclusions follow. First, no EV trial that fails to standardise and document offloading or compression is interpretable; the IWGDF criticises the existing adjuvant literature for exactly this reason: there is a "serious lack of clear descriptions of standard care delivery, including the type and quality of offloading provided" [34]. Second, part of the "unmet need" that advanced therapies promise to meet is not a biological ceiling but an implementation failure: the guideline itself cites "numerous published surveys showing low use of non-removable offloading devices, particularly total contact casts, in clinical practice" [173].

The benchmark an EV product must exceed is the control-arm closure rate under good standard care: approximately 24.2% (95% CI 19.5–28.8) at week 12 and 30.9% (26.6–35.1) at week 20 [32]; or, across 32 modern randomised trials, a pooled 33.15% (95% CI 31.18–35.11) in the 12–24-week window, with a mean healing time of approximately 50 days and an infection rate of 17.4% [33]. A formal correction to this 2025 meta-analysis has been published and its content could not be accessed; the 33.15% figure should be checked against the corrected version before being used in any decision document. For VLU, the pooled absolute control-arm healing rate could not be verified; only the relative Cochrane effect is available for this wound type.

Where do EVs sit in the treatment hierarchy?

The IWGDF positions every adjuvant therapy — the class into which an EV product would fall — as usable only "in non-infected ulcers that fail to heal within 4–6 weeks despite optimal clinical care, and where resources exist to support these interventions"; sucrose octasulfate dressings, autologous cell patches, placental membrane allografts, topical oxygen and hyperbaric oxygen are on that list [171]. The realistic regulatory and clinical niche for an EV product is therefore "adjuvant to optimised standard care in hard-to-heal, non-infected ulcers" — a niche already occupied by four to five marketed alternatives.

Within this niche, only two interventions reach moderate certainty, and each rests on a single randomised trial at low risk of bias.

Sucrose octasulfate (TLC-NOSF) dressing. The EXPLORER trial randomised 240 patients with neuroischaemic DFU (126 vs 114) in an international, multicentre, double-blind design: closure at week 20 was 60/126 (48%) vs 34/114 (30%), adjusted OR 2.60 (95% CI 1.43–4.73), P=0.002; monthly follow-up for up to 20 weeks [35]. The arithmetic absolute difference is 18 points, but since no source has published a confidence interval for this difference, no such interval should be quoted. IWGDF grading Conditional/Moderate; downgraded because it is a single trial and the optimal starting time is uncertain [34].

Autologous leucocyte-platelet-fibrin patch (LeucoPatch). An observer-masked, multicentre randomised trial in 32 specialist clinics in the United Kingdom, Denmark and Sweden; patients whose ulcer area shrank by less than 50% during a 4-week run-in period were randomised (269 patients): healing within 20 weeks 45/132 (34%) vs 29/134 (22%), OR 1.58 (CI 1.04–2.40), P=0.0235; median time to healing 72 (IQR 56–103) vs 84 (64–98) days; HR up to week 12 1.709 (95% CI 1.071–2.728), P=0.0246; no device-related adverse events, no difference in amputation/infection/antibiotic use, and no increase in anaemia despite repeated blood draws [36]. Two caveats must be kept: because the confidence interval is printed as 96% in one source and 95% in another, the coverage level should not be asserted; and the randomised (137) and analysed (134) denominators differ. The factor limiting practicality: preparing the patch at the bedside requires 18–36 mL of venous blood every week [34].

Everything else in the adjuvant class is of Low certainty, and for several classes there is a recommendation against routine use:

• Placental and amniotic allografts: Conditional/Low from 10 trials, only 3 of which were at low risk of bias; cost per healed ulcer "over $2,000 for dHAM and over $3,000 for dehydrated umbilical cord"; no patient/carer blinding in the pivotal trials [34].
• Hyperbaric oxygen (HBO): Conditional/Low from 18 randomised trials, only 3 of which were double-blind, with the evidence "conflicting". The positive double-blind trial (HODFU) reported complete healing at one year in 90 patients of 52% with HBO vs 29% with hyperbaric air (P=0.03) — however, the arm denominators could not be verified and the control rate appears as 27% in some secondary sources; it should be confirmed before use [174]. The arm-level results of the negative double-blind trial could not be reached through eight independent access routes; the "conflicting trials" claim should be stated qualitatively [175]. The current Cochrane review covers 12 trials and 577 participants (10 of them DFU): healing at week 6 RR 2.35 (95% CI 1.19–4.62; 5 trials, 205 participants), "this benefit was not seen at one-year long-term follow-up"; major amputation RR 0.36 (0.11–1.18; 5 trials, 312 participants), not significant; the review predates the negative trial of 2016 and has not been updated [176].
• Topical oxygen: Conditional/Low. The TWO2 trial randomised 73 patients after excluding rapid healers during a 2-week run-in period: closure at week 12 15/36 (41.7%) vs 5/37 (13.5%), OR 4.57 (97.8% CI 1.19–17.57), P=0.010; adjusted OR 6.00 (1.44–24.93), P=0.004; at month 12 20/36 (56%) vs 10/37 (27%), P=0.013. The 97.8% interval reflects alpha spending for interim analyses, and the large OR rests on 15 versus 5 events [177]. A second low-risk trial found no difference [34].
• Negative pressure wound therapy (NPWT): Conditional/Low in favour for post-surgical diabetic foot wounds; Strong against in non-surgical ulcers. The pivotal trial randomised 162 patients after partial foot amputation: healing 43/77 (56%) vs 33/85 (39%), P=0.040; faster time to closure (P=0.005) and to 76–100% granulation (P=0.002); similar frequency and severity of adverse events [178]. Cochrane's review of 11 randomised trials and 972 adults reports RR 1.44 (1.03–2.01) for post-operative healing — however, this comes from a single trial of 162 participants, i.e. the same trial whose arm-level results were just given, and is not independent evidence; foot ulcer healing RR 1.40 (1.14–1.72; 5 trials, 486 participants), amputation RR 0.33 (0.15–0.70; 441 participants) (post-operative amputation RR 0.38; 0.14–1.02; 292 participants). Cochrane rated the certainty low or very low for all primary comparisons and concluded that "the reliability of the evidence provided by the trials is too low for us to be certain of the benefits and harms" [179]. This produces a real and instructive tension: while Cochrane's pooled estimate for non-surgical ulcers is nominally in favour of NPWT, the IWGDF gives a strong recommendation against it — both are defensible because the same evidence is rated as low certainty; a clean example of the same data producing opposite practical recommendations depending on how certainty is weighted.
• Cellular skin equivalents: Conditional/Low; the guideline recommends against routine use; 10 randomised trials, "all at high risk of bias due to lack of blinding, high dropout and per-protocol analysis" [34]. The pivotal Apligraf/Graftskin DFU trial randomised 208 patients (112 vs 96): complete closure at week 12 63/112 (56%) vs 36/96 (38%), P=0.0042, median time to closure 65 vs 90 days (P=0.0026), significantly lower rates of osteomyelitis and amputation (numbers not verified) [180]. Note: this trial is in Diabetes Care 2001;24(2):290–295, not in Archives of Surgery; the frequently made "Veves, Arch Surg 2001" citation belongs to a different trial (collagen/ORC dressing) and should not be used.
• Acellular matrices: Conditional/Low, routine use not recommended; 13 randomised trials, all at high risk of bias, most unblinded; the IWGDF states that "the absence of negative trials … may indicate publication bias, and most trials were industry-sponsored" [34]. The pivotal Dermagraft trial shows how the effect size shifts with the choice of analysis: while the journal report of 314 patients at 35 US centres gives complete closure at week 12 of 30.0% (39/130) vs 18.3% (21/115), P=0.023 [181], the FDA PMA Summary of Safety and Effectiveness reports 163 vs 151 randomised patients and, in the primary population of ulcers older than 6 weeks (n=245), closure at week 12 of 39% vs 32%; Bayesian probability that the device increases closure 98.4%; median percentage closure 91% vs 78% (P=0.044), faster closure (P=0.040), study wound infection 10.4% vs 17.9% [182]. The regulator's primary analysis (a 7-point difference) is markedly less impressive than the journal's headline (12 points) — a warning that will apply directly to how EV trial results are presented in the future.
• Recombinant growth factors: Conditional/Low, the guideline recommends against use. Of the seven PDGF trials, the low-risk-of-bias one of the two double-blind trials "reported no difference in healing between the two groups"; of the four EGF trials, the low-risk one "reported an improvement in healing at week 12, but the effect size was only moderate"; in the three G-CSF trials "none showed benefit"; no growth factor trial reported the outcomes of "sustained healing, amputation, quality of life, new infection, resource use or mortality" [34]. The becaplermin history is the most useful precedent for the EV field: the FDA label prints complete closure rates of 48% vs 25% placebo (study 1); 50% vs 36% vs 35% (study 2, the pivotal three-arm trial); 44% vs 36% vs 22% (study 3); 36% vs 32%, "not statistically different" (study 4) [183]; the EMA's assessment, almost certainly of the same pivotal trial, gives 395 participants, healing of 49.6% vs 34.6% and median time to healing of 71 vs 79 days — different numbers in different analysis populations for the same programme; they should be cited separately without harmonisation [184]. In 2008 a boxed warning was added for increased cancer mortality (retrospective insurance database study: cancer mortality 3.9 vs 0.9 per 1,000 person-years in patients given three or more tubes); the EU opened an Article 20 procedure in March 2009 and the CHMP in September 2010 requested a contraindication in pre-existing malignancy and a mandatory pharmacoepidemiological study; the boxed warning was removed in November 2018 — a three-year extension had weakened the signal and a Veterans Affairs database study comparing 6,429 users with 6,429 non-users over 11 years had shown no increase in cancer incidence or mortality — but the malignancy statement was retained in the Warnings and Precautions section [183–185]. A topical product with a single growth factor reached the market with a closure gain of about 15 points, received a boxed warning from observational pharmacovigilance, and lost that warning a decade later — and the guideline panel still recommends against its use.
• Explicit recommendations against: autologous skin grafts (Strong/Low), platelet therapies other than the leucocyte-platelet-fibrin patch (Conditional/Low), cold atmospheric plasma, ozone, nitric oxide and CO₂ (Strong/Low), antiseptic and antimicrobial dressings (Strong/Moderate), honey, collagen-alginate, phenytoin and herbal products (Strong/Low), and most alternative debridement methods (Strong/Low) [34].

EVs versus whole-cell stem cell therapy: the defining evidence asymmetry

In DFU, whole-cell MSC therapy is the closest biological comparator, and the contrast is striking. A 2026 meta-analysis reports, in 32 randomised trials and 2,059 patients, ulcer healing OR 4.64 (95% CI 3.11–6.90), I²=53%; amputation OR 0.29 (0.18–0.49), I²=0%; time to healing mean difference −16.83 days (−27.93 to −5.74); ankle-brachial index +0.14 (0.05–0.22); transcutaneous oxygen +11.58 mmHg (5.36–17.80); rest pain −1.04 (−1.49 to −0.59); ulcer area −2.15 cm²; neovascularisation OR 15.36 (4.62–51.09); the autologous bone marrow MSC subgroup OR 8.33 (P<0.01), and the suggested dose range is 1×10⁷–1.2×10⁹ cells intramuscularly [37]. An older meta-analysis of six randomised trials had found a mean difference for healing of 0.52 (0.38–0.65; P<0.00001), larger in ulcers ≥5 cm² (0.76; 0.55–0.97), and no increase in the risk of adverse events [186]. Umbilical cord MSCs in particular, in 6 randomised trials and 380 patients with DFU and peripheral arterial disease, gave ulcer healing OR 2.88, transcutaneous oxygen SMD 1.39, ankle-brachial index SMD 1.22, angiographically confirmed neovascularisation, and reduced pain and ulcer area — but intermittent claudication did not improve (SMD 0.83), only two trials recorded adverse events (all "transient, minor and local"), and the authors called for larger multicentre trials [187].

Two qualifications keep this comparison honest. First, the largest pooled MSC dataset used RoB 2 and reported 28 of 32 trials at low risk for sequence generation, but only 3 at low risk for allocation concealment and 15 at high risk for blinding; the overall judgement was "moderate risk of bias", with no GRADE rating and no synthesis of adverse events [37]. Second, and decisively, the IWGDF assessed the same literature with GRADE and recommended against cell therapy (Conditional; Low certainty); the rationale: weak evidence, high heterogeneity, the "high resource burden" of requiring cell culture and harvesting, and equity concerns "particularly in the health systems of low-income countries" [34]. A striking pooled OR and a guideline recommendation against use coexist — this is exactly the outcome an EV field with the same methodological profile should expect.

As regards regulatory precedent: stempeucel/REGENACIP (allogeneic pooled bone marrow MSCs) is approved in India by the CDSCO for "no-option" critical limb ischaemia in atherosclerotic peripheral arterial disease and Buerger's disease at Rutherford stage III-5 or III-6, in patients unsuitable for revascularisation or in whom revascularisation has failed, with rest pain and/or ulcer; the supporting programme includes a single-arm phase III, and there is no finding of any MSC product approved for wounds or critical limb ischaemia in Japan, Korea, the EU or the USA. The exact year and legal nature (conditional or full) of the Indian approval could not be verified [188].

The framing of EVs as the cell-free substitute for mesenchymal stromal cell therapy is now explicit in the 2026 literature [189], and a parallel DFU-specific translational assessment frames the engineered exosome programme as "lessons learned, ongoing challenges" [190]. However, there is no head-to-head human randomised trial comparing MSC-EVs with whole MSCs in any wound type. The superiority of EVs over cells is a pharmaceutical argument — no living cells, no risk of embolism or engraftment, off-the-shelf storage, lower point-of-care resource burden — not a proven clinical superiority.

Intra-EV comparisons the evidence can and cannot support

Different sources. Context-dependent, not established (see Section 4.2). Meta-analytic subgroups favour adipose-derived EVs for closure and collagen and bone marrow-derived EVs for revascularisation, but the formal subgroup test showed no significant difference and the bone marrow subgroup was non-significant in one DFU analysis [6,11]. Certainty: very low.

Different routes. A preclinical subgroup favours subcutaneous injection over dressing/covering [6], one primary study found intravenous faster than local [100], and clinical development has turned to topical. There is no head-to-head route comparison in a single model, and in humans only topical and perilesional/intralesional routes have been used. Certainty: very low; the dose delivered is a more likely functional variable than the route.

Free versus hydrogel/scaffold-delivered EVs. Two rodent studies with matched arms show loaded carrier > free EVs > empty carrier: in one, day-14 closure 88.67% vs 76.3% vs 64.3%; in the other, remaining area 1.07% vs 8.25% vs 13.83% [133,134]. In the animal meta-analysis of platelet-derived EVs in diabetic wounds, the hydrogel combination subgroup (SMD 7.96; 95% CI 5.05–10.87) showed a larger effect than EVs alone (overall SMD 4.43), but this is a between-study subgroup comparison, not a matched arm [98]. The meta-analysis made for this purpose is paywalled, and the accessible one did not run such a subgroup [13,135]. Certainty: very low in animals; no evidence in humans — apart from a simple gel or fibrin sealant, no EV-loaded hydrogel has been used in a human wound trial.

Native versus engineered EVs. Engineering is the norm in preclinical work (72.3% of studies used a modification) [6], and specific constructs report striking results — near-complete closure at day 14 with hypoxic preconditioning versus partial healing with normoxic exosomes [81]; with an extrusion mimetic carrying 28 miRNAs, a reduction in remaining wound area at day 9 from 23.01 ± 10.61% to 9.47 ± 6.39%, at roughly 1/120 of the preparation cost of MSC exosomes [132]. However, engineering can reverse the direction of effect (miR-20b overexpression and advanced glycation end-product preconditioning reduced closure) [16], every engineered variant is a new product requiring its own characterisation and potency package, and no engineered EV product has entered a human wound trial. Certainty: no human evidence; engineering should be understood as a research programme, not as a therapeutic advance.

Across wound types. See Table 5. The only wound type with randomised human evidence is DFU; the only wound types with within-patient controlled human evidence are skin graft donor sites, punch biopsy wounds and (unreported) epidermolysis bullosa; the two controlled acute wound results are null.

Efficacy, safety, practicality, scalability and manufacturing: an honest side-by-side comparison

Source B's didactic framework is useful for showing the key differences between the three treatment classes at a glance; however, some statements there (for example that EVs are "stable at −20 °C" or "have no tumourigenicity") have been corrected against Source A's primary literature: the storage literature shows aggregation at −20 °C [43], and tumour promotion is a theoretical but, depending on the model, experimentally supported hazard (see Section 10.4) [148–150].

In terms of efficacy, EVs have less human evidence than any marketed adjuvant; the two clean controlled results in acute wounds are null (Section 6.2) and the two positive results in chronic ulcers are at high risk of bias (Sections 6.3, 6.4). One more data point was added to this picture in 2026: the results of Rion's PEP-TISSEEL Phase 2a trial (NCT06319287), posted on ClinicalTrials.gov on 20 July 2026, give complete closure at week 12 as 13/28 (46.4%) in the treatment arm and 7/31 (22.6%) in the standard-of-care arm (ITT; n=59; details in Section 6.4) [26]. This 23.8-point absolute difference is in the same class in magnitude as sucrose octasulfate (48% vs 30%, week 20) and LeucoPatch (34% vs 22%) — but not in the same class in certainty: it is registry data from an open-label, single-centre trial with a non-completion rate of about 35% and no peer-reviewed publication, and no P value is given in the registry; the Fisher's exact test calculated in this review from the registry numbers gives two-sided P=0.062 (this calculation is not a sponsor analysis). The moderate certainty of EXPLORER and LeucoPatch comes from double-blind or observer-masked, multicentre, low-risk-of-bias designs; the PEP-TISSEEL result meets none of these conditions, and the certainty of evidence should be rated as low.

In terms of safety, EVs compare favourably in the short term with becaplermin's malignancy history and with the theoretical embolism and tumourigenicity concerns of whole-cell therapy — but with a small fraction of the exposure and without long-term follow-up. In terms of practicality, a lyophilised room-temperature powder reconstituted into fibrin sealant may count as the most clinic-friendly advanced therapy format designed to date; it is advantageous compared with the total contact cast (requires a trained casting team, documented low real-world use), the hyperbaric chamber (many sessions, hard to scale), the weekly 18–36 mL venous blood draw for the leucocyte-platelet patch, or cellular skin equivalents dependent on the cold chain and with limited shelf life. In terms of scalability, closed bioreactors with tangential flow filtration and pooled non-proliferating starting material (platelets) are genuinely scalable; EV products derived from autologous serum or PRP are not. In terms of manufacturing, the field's gaps are distinct and heavy: no universal identity marker, no accepted purity threshold, no consensus potency assay for the wound healing claim, no numerical release specification in any regulatory framework (the Korean guideline explicitly "sets no universal specification or numerical limit"), and batch release criteria are negotiated product by product — a major barrier for a small wound care sponsor, and the explanation for why clinic-grade "exosome" products have no meaningful quality control [88,157].

Cost and practicality: the known and the unknown

No peer-reviewed manufacturing cost estimate, price analysis or health economic evaluation of EV or exosome therapy — in wound care or in any indication — could be found. This should be stated explicitly rather than filled with plausible-looking numbers. Clinic and vendor price pages exist, but they are marketing material and have been deliberately excluded. Consequently, nowhere in this review is a cost per dose, cost per healed ulcer or cost-effectiveness ratio given for EV wound therapy.

What can be documented is the cost structure:

• Upstream yield is the dominant determinant. Bone marrow MSCs secrete roughly 1–4 µg of exosomal protein per million cells per day in standard 2D culture, increasing 2–3-fold with hypoxia or genetic modification; adipose tissue yield roughly doubles under hypoxia; umbilical cord yield increases 3–4-fold in 3D systems; a single hollow-fibre bioreactor unit "produces exosomes equivalent to 70 T225 flasks per week"; exosome-mimetic production can be roughly tenfold higher than native small EV isolation [92]. One wound study reported a 40–80-fold yield increase from a 3D perfusion bioreactor [6]. The citation details of source [92] are incomplete; the yield figures should not be relied upon without re-verification against the primary article.
• Purification losses and equipment capital are named barriers: "high initial equipment cost" for magnetic separation and high-throughput isolation platforms, and "high production costs limit accessibility, particularly in resource-constrained settings" — without any figures being given [92].
• The quality control burden is unusually high, because the release, stability, comparability and biodistribution packages must be established product-specifically, with functional potency assays and without a universal specification [88].
• Scale-up does not directly lower the cost per effective dose, because higher particle yield does not mean higher potency; closed-system bioreactors with in-line monitoring reduce batch failures and deviations — that is where the real economic gain lies [87].
• Cold chain and lyophilisation. The baseline for shelf life is −80 °C by default; a lyophilised room-temperature presentation eliminates the cold chain but requires its own formulation, container, reconstitution instructions and long-term stability data [88]. PEP's claim of 24 months at room temperature is the existing evidence point for a wound-relevant format [65].
• Treatment frequency and clinical workflow. Human wound protocols to date have required weekly application for 4–12 weeks (Egyptian gel: weekly × 4, then every 3 days; Rion Phase 2a: 12 weeks of application; radiation ulcer protocol: weekly × 20 weeks) or repeated injections (venous ulcer pilot: 6 injections over 2 weeks; Jordanian secretome: up to 10 weekly sessions; pressure injury case: 6 weekly injections). This is comparable to the weekly burden of the leucocyte-platelet patch and exceeds that of the single-application skin equivalent. No data on consumables per session, staff time or total episode cost could be found.
• The economics driving the grey market. Legitimate development requires "controlled facilities, validated isolation and purification systems, sterility and stability studies" — unregulated cash-pay clinics skip these costs by operating outside compliance frameworks [159]. It is this asymmetry, not scientific disagreement, that explains how a field with no approved product coexists with an unapproved commercial market of roughly 60 US businesses and 669 Japanese institutions [145].

Flagged as inference: with MSC output at the level of 1–4 µg of exosomal protein per million cells per day and preclinical dosing in the microgram-per-gram range, a clinically meaningful systemic dose would require a very large cell culture footprint; topical or local wound application at lower absolute doses is the more economically attainable route. This reasoning is partly sourced but is not a costed analysis.

Unless otherwise stated, certainty ratings and qualitative statements are taken from IWGDF 2023 [34,173].

Current Experimental Treatment Strategies

This section defines what is actually being done with EVs in human wounds today, and under what authority. Nothing here is established clinical practice. Three types of activity must be distinguished from one another.

Registered research use — the legitimate route

Between ten and thirteen interventional registry records worldwide constitute the entire formal wound EV pipeline (see Section 16). Active or recently completed programmes are as follows:

Platelet-derived purified exosome product (Rion, USA). Four records: skin graft donor sites (NCT04664738; phase 1; completed February 2024; null result published) [21,106]; diabetic foot ulcer (NCT06319287; phase 2a; PEP-TISSEEL 15 mg/mL, 12 weeks) [25]; chronic radiation ulcer (NCT06793748; phase 1/2; 2 vials PEP + 10 mL TISSEEL weekly, 20 weeks; estimated n=184; "not yet recruiting" despite all of its dates having passed; last update 6 April 2025) [103]; and a mechanistic intradermal study in healthy skin (NCT06429033; 75 or 150 mg; completed 18 December 2025, actual enrolment 8, results not yet posted to the registry) [125]. At this review's evidence cut-off date the status of the DFU trial differs from when Source A was written: NCT06319287 has been completed (study completion 1 November 2025) and its results were posted to ClinicalTrials.gov on 20 July 2026 — actual enrolment 59 patients (28 PEP-TISSEEL + standard of care; 31 standard of care), single-centre, open-label; complete closure at week 12 13/28 (46.4%) vs 7/31 (22.6%), ITT; serious adverse events 2 vs 6; no peer-reviewed publication yet (details in Section 6.4) [26]. The product's clinical appeal is its format: a lyophilised room-temperature powder reconstituted in fibrin sealant, applied the way a surgeon already applies a sealant [65].

AGLE-102 (Aegle Therapeutics, USA). Allogeneic bone marrow MSC-EVs in recessive dystrophic epidermolysis bullosa (NCT04173650; phase 1/2A; within-patient matched wound control, up to six topical doses two weeks apart over 10 weeks, estimated n=8; registry still "Recruiting", last update 25 June 2025, results due under FDAAA on 30 September 2026, not yet reported) [94,121] and in deep second-degree burns (NCT05078385; phase 1; actual enrolment 1; completed; single patient published) [24,118]. The RDEB trial is the best-designed EV wound trial in existence — matched wounds in the same patient, a 22-week safety window, closure and time-to-closure secondary endpoints, POSAS scar quality, Wong-Baker FACES pain and quality of life — and its unreported results are, after peer-reviewed publication of the now-posted Rion phase 2a result, the field's most important gap.

Academic single-country studies. The completed and published Egyptian Wharton's jelly randomised trial (NCT06812637) [22]; the Iranian fistula programme (NCT05402748, NCT05499156, IRCT20200413047063N3) — one phase I published, registry records internally inconsistent [95,109,110]; the Jordanian conditioned-medium study (NCT06825884) [114]; the Japanese autologous plasma-derived exosome study (NCT02565264) — in Unknown status since 8 September 2020, no results and no publication [122]. To these are added three further records verified from ClinicalTrials.gov for this review: NCT05475418 (Shanghai Ninth People's Hospital; autologous adipose tissue exosomes + sterile hydrogel dressing; pilot, actual n=5; completed 15 October 2023; results not posted) [123]; NCT04652531 (SER-VES-HEAL) (University of Turin; autologous serum EVs in VLU, peri-wound injection once weekly for 3 weeks; estimated n=10; non-randomised, contralateral ulcer as control; last verified "Recruiting" in May 2023; the registry programme of the Italian pilot [23]) [107]; NCT05243368 (Córdoba, IMIBIC; personalised nutrition + MSC exosomes in DFU with peripheral arterial disease; estimated n=30; "Recruiting" in October 2023; no results) [124].

Off-trial individual-patient and compassionate use

Four published wound case reports document the use of commercial or institutional EV products outside a registered trial: ExoFlo injected into a refractory ischial pressure wound [116]; PEP in collagen, then in fibrin sealant, in a scalp wound after chemoradiation [117]; ExoVia umbilical cord exosomes in a scald burn [93]; plant-derived Rosa damascena exosomes in an ischaemic nasal degloving injury [80]. None states the regulatory pathway. These reports have a legitimate function — a safety signal and mechanistic plausibility in wounds that have exhausted standard options — but they are successes selected by publication bias, several involved concurrent debridement or surgery, and they cannot support any efficacy claim.

The unapproved commercial market — the dominant real-world activity

Quantitatively, most human exposure to "exosome" products occurs here: approximately 60 businesses belonging to 38 companies in the USA and 669 medical institutions in Japan, predominantly cosmetic (skin and hair), not wound care [145]. Transparency is poor (only 18% of manufacturers highly transparent; 27% of claims in the misleading category) [146]; products marketed as purified exosomes are frequently unfractionated conditioned medium or minimally processed material [2,155]; the documented harms in Section 10 arise almost exclusively from here. The field's largest published wound dataset — 175 wound management and 118 scar treatment patients within a series of 738 patients — comes from this sector, is retrospective and uncontrolled, uses numerous confounding adjunct modalities, and was included by its own author in his own systematic review [15].

Emerging and Future Treatments

Every approach in this section is preclinical, with the two exceptions noted in the table. The most useful way to read it is to ask which strategies solve a problem that actually blocks translation.

Strategies targeting a documented bottleneck. Cell preconditioning (hypoxia, melatonin, pioglitazone, atorvastatin, deferoxamine, LPS, static magnetic field + nanoparticles) is the engineering strategy closest to translation, because it requires no genetic construct and no post-isolation manipulation; one review describes it as "technically simpler, safer and more universal than genetically engineered exosomes" [63,131]. Retention engineering — stimulus-responsive hydrogels, sprayable matrices, microneedles — targets the only properly documented pharmacological deficiency (rapid clearance) [61]. EV mimetics and hybrids target cost and reproducibility: a cocktail of 28 miRNAs encapsulated in DC-cholesterol/DOPE liposomes, hybridised with watermelon-derived EVs and extruded through 800→100 nm membranes achieved 94.94 ± 2.58% miRNA encapsulation and, with subcutaneous 1.0 µg/µL on days 0, 3, 6 and 10 in mouse deep second-degree burns, reduced the residual wound area on day 9 to 9.47 ± 6.39% versus 23.01 ± 10.61%; denser and more organised collagen, lower α-SMA; efficacy "comparable to natural MSC-Ex" and preparation cost approximately 1/120 [132]. Antimicrobial-peptide-engineered vesicles make the vesicle itself bactericidal while preserving the healing cargo, removing the need for a separate antibiotic [130].

Strategies with a weaker rationale for wounds. Surface engineering for targeting — Lamp2b-peptide fusions, lactadherin C1C2 "exosome display", lipid insertion, bio-orthogonal click chemistry, aptamers — was developed for systemic administration [63,191]. Because EVs are applied directly to the target in an open wound, the rationale for retention engineering is far stronger than that for targeting engineering.

Strategies that introduce new risk. Parent-cell genetic modification (predominantly lentiviral in the 21 studies using it) [6] adds vector and genetically modified organism regulatory burden; bacterial and probiotic EVs raise questions of lipopolysaccharide and GMO release; plant-derived vesicles have no standard active ingredient and no defined identity.

What has not yet been tried. There is no verified wound study of engineered apoptotic vesicles (apoV) as a product class, of circRNA-engineered EVs (circ-ITCH) or of lncRNA H19-engineered EVs — HOTAIR is the only long non-coding RNA with verified wound evidence. (This is distinct from the conventional apoptotic small EV and apoptotic body preparations in the five rodent studies pooled in the 83-study systematic review and reported in Table 1 and Table 10; those are non-engineered apoptotic EV fractions [6].) There is no verified wound study of curcumin- or metformin-loaded EVs. The nanozyme + EV wound system was not verified. There is no engineered-EV wound study in a large animal; the only non-rodent data in the engineered literature are a rabbit arm in the bilayer hydrogel study. Apart from autologous PRP and plasma-derived preparations, no personalised or autologous engineered EV approach has been published.

Principal Limitations and Unresolved Questions

This section gathers in one place the methodological problems raised piecemeal in the preceding sections. The aim is to state plainly which structural gaps stand behind the field's "promising" language, and what is not known when the next clinical trial, guideline or patient conversation is being designed. None of the problems is specific to a single study; all are recurring patterns spread across the literature as a whole.

Definition and reporting failures

The field cannot reliably say what it is studying. MISEV2023 ties the term "exosome" to demonstrated endosomal origin; wound studies almost never demonstrate this origin. Published "exosome" products are therefore small EV (sEV) preparations of mixed biogenesis plus co-isolated non-vesicular particles [4]. Across the field as a whole, only 13.6% of EV papers use all four MISEV characterisation categories and only 15.3% cite MISEV [46]. Looking specifically at the wound literature, 46 of 83 preclinical studies (55.4%) provided adequate characterisation to MISEV2023, 44.9% lacked a positive cytosolic marker, and 62.8% never assessed non-EV co-isolates at all [6].

Isolation methods are not comparable and are product-defining: 97.6% of preclinical studies used ultracentrifugation, whereas clinical manufacturing uses tangential flow filtration (TFF) — a population change that is almost never bridged by a comparability study [6,82]. There is no accepted numerical threshold for purity; the particle/protein ratio "is not a universal purity specification" [88,193].

The clinical implication is this: the "product" found effective in the preclinical literature may have become a physically different preparation by the time it reaches a clinical trial; the extent to which the animal data represent the human product is unknown in most programmes.

In this literature, dose is not a quantity

Doses are reported as µg protein (from 2 µg to 5 mg), particle number (from 10⁵ to 10¹²), µg/cm², µg/mL, RNA µg/mL, nmol miRNA, "1% v/v" or not stated at all (22.1% of preclinical studies) [9,10]. Only one of 68 studies normalised dose to body weight, and dose–response was formally tested in only one [9].

In human studies only six records specify a measurable dose: 1×10⁷ particles/cm² (AGLE-102, burn), 9.78×10¹⁰ particles/mL (Kwon, acne scar), 100 µg in 340 µL (Plexoval II), 250 µg (Iran, fistula), 75 or 150 mg (PEP, intradermal) and 15 mg/mL (PEP, in fibrin sealant) [20,24,25,95,111,125]. The field's only positive randomised controlled trial never measured its dose [22]. Among registered EV trials across all indications, only 35.2% specified a dosing strategy, only 12.1% reported the isolation methodology, and characterisation methods were described in 36.1% [194]. MISEV2023 adds that even a precise particle count is not a precise EV dose [4].

The consequence: no preclinical dose can be translated into a human dose, no two studies can be pooled on a common scale, and dose–response optimisation has not yet begun. Because protein mass, the dominant metric, is precisely the quantity inflated by soluble co-isolates, the dose and purity problems are not independent of each other.

The animal-to-human gap is structural, not a matter of scale

Four separate features make the preclinical base non-transferable.

Species: mice and rats only; across the field's largest study denominators one macaque, one to two rabbits and a single unidentifiable miniature pig; there is no porcine cutaneous wound EV efficacy study at all — yet the pig is the skin model closest to humans [6,9,14,17].

Model: 90–93% of studies used full-thickness excisional wounds that close substantially by contraction; human chronic wounds close by re-epithelialisation. Splinted models developed to correct this problem exist, but no review has reported whether the included wounds were splinted [6,18,19].

Endpoint: "closure", the contraction-sensitive metric, is pooled from 19 or more studies, while re-epithelialisation comes from only two to four studies [12,13].

Comorbidity: rodent wounds lack the ischaemia, biofilm, repetitive trauma, neuropathy and multimorbidity burden of the human diabetic foot ulcer; infection models are acute inoculation models, not biofilm models; the verified diabetes models are only STZ and db/db.

Effect sizes at SMD 3–8 should be read as a consistent direction, not as a transferable magnitude. The field's history with growth factors and skin equivalents is the natural quantitative prior for how much attenuation to expect (see Section 12.2).

Publication bias and the missing negative literature

Three of the most recent meta-analyses detected statistically significant publication bias (Egger P=0.000 with imputation of nine studies; Begg Z=2.47 P=0.013 and Egger t=4.52 P=0.004; Egger P<0.001 for two separate endpoints), two did not, and of more than 80 primary animal studies only two reported an effect in the negative direction [11–14,16]. A diabetic wound review tabulating approximately 34 in vivo studies states "no explicit negative/null findings" [17].

In humans the pattern is worse in a different way: the dominant failure mode is not negative publication but non-publication. Of 471 EV trials registered by the end of 2023, only 15 had posted results to ClinicalTrials.gov [194]. For wound trials specifically, the position as of 22 September 2026 is as follows:

• NCT06319287 (Rion, PEP-TISSEEL, DFU, Phase 2a): results were posted to the registry on 20 July 2026 [26]; but there is still no peer-reviewed publication. This means the field's most important missing dataset has been partly closed, but has not passed independent peer review (details in Sections 6.3 and 16.3).
• NCT04173650 (Aegle, AGLE-102, epidermolysis bullosa): results are due under FDAAA on 30 September 2026; not yet reported [121].
• NCT05078385 (Aegle, burn): completed November 2024, results not posted; published as a single-patient case presentation [24,118].
• NCT06429033 (PEP, intradermal, healthy skin): completed December 2025; estimated results submission was May 2026, and the registry still has no results [125].
• NCT02565264 (Kumamoto, plasma-derived exosomes): "status unknown" for six years, no publication [122].

Of EV trials across all indications, 23.4% are in unknown status; 1.7% terminated, 2.0% withdrawn, 2.0% suspended — none of these attributable to a wound indication [168].

The net effect is a literature in which the visible evidence is dominated by uncontrolled positive case series, while the available controlled data are largely neutral or of low certainty.

The secondary literature is actively misleading

Section 6.8 documents two verified cases: a review that presented a withdrawn intravenous COVID-19 trial identifier as a Phase II diabetic foot ulcer exosome trial, with untraceable outcome figures [38,39], and a review that reported a null-result donor-site trial as "100% re-epithelialisation" [5,21]. Additional patterns are: reviews stating that no human wound exosome data exist standing side by side with reviews declaring a single product "the only clinical-grade exosome therapeutic reported to have been applied in wound management in humans" — both demonstrably false [65,72]; reviews listing trials completed more than two years earlier as ongoing [106]; reviews tabulating 19 human "clinical applications" without a single registry identifier [5]; reviews recirculating the same four to six studies to create an impression of breadth [15,112].

Mechanistic attribution remains unresolved

The single-miRNA narrative is weakened by stoichiometry: a given miRNA is present at approximately one copy per 100–10,000 vesicles, and even in experiments using forced fusogens, target suppression could not be demonstrated with 10⁵ EVs per cell [66,68]. Co-isolated soluble factors and transfection reagent artefacts [69–71], together with the field's own emerging standard of "a depletion or rescue experiment requirement" [7], complete the picture. What the active principle of a therapeutic EV preparation is — and therefore what must be measured for potency and release — is genuinely unknown. Only one of the wound studies in this review met the full causality bar (knockdown plus pharmacological rescue) [54].

Specific unresolved questions

• Which dose, and by which metric? There is no dose-ranging study in any human wound. The only preclinical dose–response signal is U-shaped; the mid-dose subgroup is non-significant [11].
• Which potency assay corresponds to a wound-healing claim? The multi-donor mixed lymphocyte reaction serves the immunomodulation claim; there is no equivalent consensus assay for the multifactorial wound mechanism [87].
• Does source matter, and in which direction? Head-to-head studies conflict, and formal subgroup testing shows no difference [11,73–75].
• Do carriers add clinical benefit? Two rodent studies say yes; the meta-analysis conducted for this purpose is behind a paywall; there is no human evidence [133–135].
• How long do EVs remain in the wound bed? Unknown. There are no wound-specific pharmacokinetic data [63].
• Does donor age or donor diabetes reduce potency? Unknown; the only relevant evidence is that endogenous keratinocyte EVs in diabetic wound fluid are dysfunctional [82].
• What is the immunogenicity of repeated allogeneic dosing? Unknown; there are 52-week negative seroconversion and panel-reactive antibody data in n=1 patient [24].
• Is there long-term risk? Unknown. In the wound indication, the longest human follow-up is 24 months (a single-arm secretome study), 52 weeks (n=1 burn) and 2 years (an uncontrolled keloid series and one case report) [15,24,114,116]. The 3-month post-treatment safety follow-up of NCT06319287 does not change this picture [26].
• What does it cost? Unknown — see 15.8.
• Does an EV product add anything on top of properly delivered standard care? Because none of the existing studies documents offloading or compression quality to the standard required by FDA guidance and the IWGDF, this question cannot be answered with today's data [34,195]. NCT06319287 is a step ahead in having written offloading into the protocol; however, the quality of its delivery is not reported in the registry [26].

Cost and health economics: an open gap

No peer-reviewed cost-of-production model, price comparison, cost-effectiveness analysis or health economic evaluation of EV/exosome therapy — in wound care or in any indication — could be found. The only numerical prices available are clinic and vendor marketing pages; these have been excluded as sources. Cost drivers are documented qualitatively (a yield of 1–4 µg of exosomal protein per million cells per day, purification losses, capital equipment, quality-control burden, cold chain, batch variability, once-weekly to 12-week treatment regimens), but there are no figures for cost per dose, cost per healed ulcer or incremental cost-effectiveness [87,88,92]. By contrast, the comparator literature has cost anchors — for placental allografts, "over $2,000 for dHAM and over $3,000 for dehydrated umbilical cord" per healed ulcer — and the IWGDF repeatedly stresses the lack of cost-effectiveness data even for marketed adjunctive therapies [34]. Today, every economic claim about EV wound therapy is speculation.

Limitations of this review's research scope

Some questions remained open because the sources could not be accessed; these should be treated as gaps of this review, not as findings about the world:

• Registries other than ClinicalTrials.gov could not be counted. ChiCTR, the Iranian Registry of Clinical Trials (through its own interface), CRIS (Korea), jRCT (Japan), ISRCTN, CTRI (India) and the WHO ICTRP were closed to robot access, returned HTTP 405 or were refused by policy. The two non-ClinicalTrials.gov identifiers reported here (ACTRN12620000944932 and IRCT20200413047063N3) were taken from journal articles, not from registries. Since China hosts the largest share of EV studies, the true global number of wound-related EV trials is probably higher than what could be counted here. The best available counter-evidence is that the only scoping review that explicitly searched ClinicalTrials.gov, the WHO ICTRP and ChiCTR found 591 records, included 73 and reported no wound/ulcer/burn/scar category [196]. No global trial count in this review should be taken as complete.
• The absence of a porcine wound study is an inference drawn from three review denominators and repeated searches, not a proven negative.
• Two purpose-specific meta-analyses were behind a paywall or inaccessible: the preclinical meta-analysis of hydrogel-delivered vs free EVs [135] and the meta-analysis of clinical trials evaluating the safety and efficacy of human EV therapy; the pooled safety data were partly relayed from a secondary summary [27].
• Several primary sources could not be accessed: the full text of a perianal fistula Phase I study [96], an Indonesian secretome study [115], the reference list of the review containing the misattributed trial identifier [38], the arm-level results of an important negative hyperbaric oxygen trial [175] and the content of the official correction to the modern standard-of-care benchmark meta-analysis [33].
• There is no peer-reviewed publication for the posted results of NCT06319287; the statistical calculations in this review (absolute difference, RR, Fisher's exact test) were made from the raw numbers in the registry table and do not represent the sponsor's own analysis [26].
• Some bibliographic details are incompletely verified; where the author list, volume or full title could not be confirmed, Section 19 states what was verified rather than offering a plausible completion.

The Clinical Trial Landscape

The countable wound EV pipeline consists of thirteen to fourteen registrations worldwide

An independent count identifies approximately thirteen to fourteen interventional registrations for EV products in wound, ulcer, burn, scar or fistula indications; total planned enrolment across all of them is in the range of 300–400 participants (approximately 500 if the 184-participant radiation ulcer target is counted). The pipeline is dominated by two US sponsors — Rion (platelet-derived PEP; four registrations) and Aegle Therapeutics (bone marrow MSC-EV AGLE-102; two registrations) — plus single academic studies in Egypt, Jordan, Iran, Japan, China, Italy and Spain. The claim of "more than 20" registered wound EV trials is not supported by countable identifiers [38]. An independent review examining the diabetic foot ulcer exosome pipeline also lists the same small cluster — the Australian Phase I, the PEP intradermal safety study and the PEP Phase 2a — and concludes that no completed human DFU trial exists [197]; that conclusion, correct at the time the review was written, has no longer been valid since July 2026.

Status breakdown as of 22 September 2026 (details and sources in Table 4):

• Completed and published in peer review: NCT04664738 (PEP, donor site, null) [21]; NCT06812637 (WJ-MSC gel, DFU) [22]; NCT06825884 (Jordan, conditioned medium — should not count as EV evidence) [114]; the Iranian Phase I fistula study [95]; the unregistered Korean scar studies [97,111,113].
• Completed, results posted to the registry, no peer-reviewed publication: NCT06319287 (Rion, PEP-TISSEEL, DFU, Phase 2a) — 59 patients, results posted on 20 July 2026; this is the first and only wound EV trial to post results to a registry [26].
• Completed, published only as a single-patient case report: NCT05078385 (actual enrolment 1) [24,118].
• Completed, no results in the registry, no publication: NCT05475418 (Shanghai, autologous adipose exosome + hydrogel, n=5, October 2023) [123]; NCT06429033 (PEP intradermal, healthy skin, n=8, December 2025; estimated results submission of May 2026 has passed) [125].
• Recruiting, but with a stale registry record: NCT04173650 (Aegle, EB; last updated June 2025, estimated completion of March 2026 has passed; FDAAA results obligation 30 September 2026) [94,121]; NCT04652531 (SER-VES-HEAL, Turin, autologous serum EV, VLU, estimated n=10; last verified May 2023) [107]; NCT05243368 (Córdoba, nutrition + MSC exosome, DFU with PAD, estimated n=30; October 2023) [124]; NCT05402748 (Iran, placental exosome, fistula; last updated November 2022) [109].
• Not yet recruiting, all dates passed: NCT06793748 (PEP-TISSEEL, chronic radiation ulcer, estimated n=184; last updated April 2025) [103].
• "Unknown status" for six years, no publication: NCT02565264 (Kumamoto) [122].

No wound, ulcer, burn or scar EV trial could be identified as terminated or withdrawn; the 6 terminated (1.7%), 7 withdrawn (2.0%) and 7 suspended (2.0%) records in the global EV dataset are not broken down by indication, and none of them should be attributed to a wound trial [168].

Field-wide context

Three independent bibliometric analyses place the wound pipeline within the broader EV field. Of 355 human-derived EV studies registered on ClinicalTrials.gov between 2010 and 2025 (543 screened), 207 (58.3%) were observational and 148 (41.7%) interventional; the status distribution was recruiting 119 (33.5%), unknown 83 (23.4%), completed 56 (16.1%), suspended 7, terminated 6, withdrawn 7; the countries were China 123 (34.6%), USA 83 (23.4%), Italy 31 (8.7%), with Africa represented only by Egypt; the interventional EV source was MSC in 66.2% (predominantly bone marrow); the phases were I 46 (31.1%), I/II 36 (24.3%) and only 3 Phase III (androgenetic alopecia, ARDS, skin rejuvenation); the routes were intravenous 46 (31.1%), subcutaneous 24 (16.2%), nebulised 14 (9.5%) and "external liquid dressing", the closest proxy to topical wound use, 14 (9.5%) [168]. Of 471 EV trials registered up to 31 December 2023, 80.7% were diagnostic and only 19.3% therapeutic; among therapeutic trials, respiratory indications accounted for 28.6%, cancer 11.0% and autoimmune 6.6%, while "ulcer" and "perianal fistula" were each represented by at least three registrations; the therapeutic EV source was MSC in 59.3% (bone marrow 27.8%, umbilical cord 13.0%) [194]. Of 90 exosome clinical trials with publications indexed in PubMed up to 28 March 2025, only 5 concerned dermatology or wounds [108].

The two readouts that matter most

The first is NCT06319287 — Rion's Phase 2a trial comparing topical PEP-TISSEEL with standard care in 59 diabetic foot ulcer patients, with weekly application for 12 weeks. The main source on which this review rests (evidence cut-off 22 September 2026) described this trial as "past its completion date without posting results, the field's most important missing data"; however, according to the ClinicalTrials.gov record, the results were posted on 20 July 2026 [26]. The registry data show the following: complete closure at week 12 (ITT) in 13/28 (46.4%) in the PEP-TISSEEL arm and 7/31 (22.6%) in the standard care arm; mean baseline ulcer area 2.7 cm² in both arms; dose-limiting toxicity 0/0; serious adverse events 2 vs 6 and deaths 0 vs 1 (standard care arm) during the 6-month safety follow-up. No p value is given in the registry; the absolute difference calculated in this review from the registry numbers is 23.8 points (95% CI 0.2–47.5), RR 2.06, two-sided Fisher's exact test P=0.062. The non-completion rate is high (Arm A 10/28, Arm B 11/31), the design is open-label, and despite the word "multicentre" in its title the trial is single-centre. This is the only randomised Phase 2 trial in which a purified EV product has been tested against a standard care comparator in a chronic wound, and its result is in the favourable direction; however, the result is not statistically conclusive, has not undergone peer review, and its magnitude (approximately 24 points) is in the same class as moderate-certainty adjunctive therapies such as sucrose octasulfate and the leukocyte-platelet-fibrin patch (Section 12.2). Certainty of evidence: low. The peer-reviewed publication is expected to disclose details that do not appear in the registry table — randomisation method, allocation concealment, offloading adherence, reasons for non-completion, and the agreement between per-protocol and ITT analyses.

The second is NCT04173650 — Aegle's Phase 1/2A trial in recessive dystrophic epidermolysis bullosa, with the field's only paired within-patient wound-controlled design and results expected on 30 September 2026 under the FDAAA [94,121]. If the paired wound outcome is positive in a genetic blistering disease where the mechanism (COL7A1-related) is defined and the unmet need is absolute, this will be the strongest efficacy signal the field has produced. The sponsor's most recent public communication predates 15 July 2025 [198].

The field today stands at a threshold at which one readout awaits peer-reviewed publication and the other is still overdue. More than the content of any review, the full text of these two readouts will determine the direction of the field.

Research Priorities

Adopt the regulator's endpoint rules prospectively

The FDA's endpoint requirement has been clear since 2006: complete closure is "skin re-epithelialisation without drainage or dressing requirements confirmed at two consecutive study visits two weeks apart"; incidence of closure or time to closure is acceptable; "partial healing is not adequate as a primary endpoint in Phase 3 trials, because the clinical benefit of incremental changes in wound size has not been demonstrated"; both arms should receive the same, prospectively defined standard care; blinded third-party assessment is recommended when blinding is impractical; and follow-up of at least 3 months after closure should be performed [195]. The Wound Care Collaborative Community's formal proposal to update this guidance and an endpoint validation programme encompassing clinician, literature and patient surveys have been published [199,200].

Measured against these rules, many common EV trial designs are disqualified: a primary endpoint of wound area reduction (the Egyptian trial's primary endpoint was percentage reduction in ulcer size), unblinded assessment, absence of post-closure follow-up and non-standardised standard care. NCT06319287 is closer to these rules in making complete closure the primary endpoint and offloading a protocol component; however, it is open-label and its 3-month post-closure follow-up is for safety only [26]. Priority 1: in every EV wound trial, prospectively, a primary endpoint of closure at two consecutive visits, ≥12 weeks of post-closure follow-up and central blinded photographic adjudication.

Fix dose and potency before the next efficacy trial

Priority 2: report dose as a dual metric — particle count and protein mass — anchored to a claim-specific functional potency assay; publish the isolation method and characterisation panel in full [82,86,88]. The reference frameworks for therapeutic EV dose, "cell-equivalent" metrics and trial readiness are the ISEV clinical translation position paper and the ISEV/ISCT definition statement on MSC-derived small EVs for therapeutic purposes; since both were inaccessible in the research on which this review rests, they are cited not for quoted content but as the documents a sponsor must consult [201,202]. Precedent already exists within the field: the Italian venous ulcer pilot released batches against a BrdU incorporation plus tubulogenesis assay, at a defined particle-to-target-cell ratio and with each batch measured against a positive control [23]. Priority 3: develop and validate a consensus-based wound-healing potency assay. Phenotype-matched candidate readouts are described in the literature — rescue of endothelial migration and tube formation plus perfusion for ischaemic wounds, macrophage polarisation for inflamed wounds, bacterial load reduction and protease tolerance for infected wounds, fibroblast migration and keratinocyte proliferation for senescent wound edges — but none is standardised, and "without a claim-specific potency assay, manufacturing scale-up will remain undecidable" [82]. The ISCT's position is that potency assays should be integrated early rather than validated retrospectively and must be biologically meaningful, not merely analytically reproducible [87].

Close the species and model gap before scaling human trials

Priority 4: conduct a porcine cutaneous wound EV efficacy study. This is the most conspicuous gap in the preclinical package supporting products that have reached Phase 2. Priority 5: use splinted rodent models and report splinting, so that "closure" measures re-epithelialisation rather than contraction; report re-epithelialisation as a primary endpoint rather than an occasional secondary one [18,19]. Priority 6: build harm assessment into preclinical protocols — no adverse events were reported in 68 studies because none were ever looked for [9].

Standardise reporting and make it auditable

Priority 7: make MISEV2023-compliant reporting and EV-TRACK data deposition mandatory at submission. The infrastructure exists — MISEV2023 as the reference standard [4], EV-TRACK for transparent methodological metadata [47], and ISEV's ongoing work on the MISEV update, AI-assisted rapid documentation (EV-Checklist) and translation of ISEV standards into regulatory science [203–205]. The isolation strategy should be reported as a product attribute, because "isolation defines identity" [206]. Wound care and regenerative medicine journals are the effective enforcement point: when voluntary, compliance is 15.3% [46].

Design the single trial that would change the answer

Since the certainty of the two best-evidenced adjunctive therapies rests on a single low-risk-of-bias randomised trial for each, a single well-conducted, adequately powered, blinded EV randomised trial with documented offloading could move EV therapy from very low to moderate certainty. This is not an unattainable standard but a concrete and specific target. Its specification follows from everything above: multicentre; a single stratified wound type; minimal exclusions; rigorous randomisation with concealment; a prospectively defined and audited standard care package in both arms; blinded central endpoint adjudication; an a priori sample size calculation; intention-to-treat analysis; reported reasons for withdrawal; complete closure at two consecutive visits as the primary endpoint; ≥12 weeks of post-closure durability follow-up; prospective adverse event and immunogenicity capture; a dual-metric dose with a released potency value for every batch used; and optimised standard care as the comparator — with, if the field is confident enough, a second arm against a moderate-certainty adjunctive therapy (sucrose octasulfate or the leukocyte-platelet-fibrin patch) [34,195,207]. The result of NCT06319287 provides the first real human effect-size estimate for the power calculation of such a Phase 3 (approximately 24 points, with a wide confidence interval); this is a significant advance in design terms.

Publish what exists

Priority 8: report the pending trials in peer review. The Rion Phase 2a diabetic foot ulcer trial now carries results in the registry, but there is no peer-reviewed publication and the registry table does not include the randomisation method, adherence or reasons for withdrawal [26]. The FDAAA results obligation for the Aegle epidermolysis bullosa trial is 30 September 2026 [121]. The Shanghai pilot (NCT05475418) and the PEP intradermal study (NCT06429033) are completed but without results [123,125]. The Japanese plasma exosome trial has been silent for six years [122]. Exopharm's initial PLEXOVAL data have never been published. Among all EV trials, 15 of 471 have posted results [194]. There is no new preclinical study that would add as much value to this field as the peer-reviewed publication of trials that have already been conducted.

Answer the comparative questions currently being assumed

Priority 9: conduct the head-to-head studies the field has skipped — EVs vs their parent cells at matched dose in the same wound model; EVs vs matched conditioned medium vs PRP; adipose vs bone marrow vs umbilical cord EVs at matched particle dose in a single in vivo model; carrier-delivered vs free EVs with pooled analysis; and EVs against a marketed advanced dressing rather than saline. The last of these has been done exactly once, in rodents [134]. That the comparator of NCT06319287 was Fibracol (a collagen-alginate dressing) is the first human application of the "real dressing rather than saline" principle [26]. Priority 10: generate wound-bed pharmacokinetics and biodistribution with a defined measured entity; current data cannot distinguish intact product from released cargo or free label [88].

Generate the safety and economic data that do not exist

Priority 11: prospective long-term safety cohorts covering immunogenicity under repeated allogeneic dosing, malignancy surveillance in populations at risk of wound-associated malignant transformation, and fibrosis/scar outcomes with validated scales at ≥6 months. Priority 12: publish a cost-of-production model and a cost-effectiveness analysis. The absence of any health economic evidence is a commercial and policy liability as much as a scientific one for a therapy that will enter a cost-constrained adjunctive market in which products costing thousands of dollars per healed ulcer already exist [34].

Evidence-Based Conclusions

What is realistically possible in clinical practice today?

Outside a clinical trial: nothing. No EV or exosome product has been licensed for any indication in the United States, the European Union, the United Kingdom, Japan, South Korea, China, Turkey, Canada or anywhere else in the world [1–3,31,151,155,156]. In the US, the EU, the UK, China and Turkey, applying an exosome product to a wound outside a clinical trial amounts to administering an unlicensed medicine. The legal gap in Japan allows cell-free preparations to be used in the clinic outside regenerative medicine review; for this reason the country's own professional societies have issued guidance against the practice [151,166].

The situation specific to Turkey is clear: there is no EV/exosome medicinal product licensed by TİTCK; the permit granted to Erciyes University GENKÖK on 24 July 2026 is a manufacturing site authorisation for umbilical cord MSC exosomes intended for clinical use, not a product licence [158]. Therefore, the only legitimate way to apply exosomes to a wound in Turkey is to take part as a volunteer, with informed consent, within a clinical trial protocol approved by an ethics committee and TİTCK; every application outside this is the use of an unlicensed product.

The three defensible clinical actions today are:

1. Referring a suitable patient to a registered clinical trial.
2. For a wound that has genuinely exhausted the evidence-based options, in countries where national legislation allows, considering single-patient use of a GMP-manufactured investigational product through a formal early access/compassionate use pathway — with explicit documentation that the product is unlicensed, that the two controlled human results in acute standard wounds were null, that the three controlled results in chronic ulcers (the Egyptian RCT, the Italian pilot and the Rion Phase 2a) are in the favourable direction but of low or very low certainty, and that long-term safety is unknown. Since Turkey has no defined procedure for this pathway, in our country this option in practice reduces to participation in a clinical trial.
3. Optimising standard care first; because in most chronic wounds the largest recoverable gain is proper offloading or compression, which surveys show is frequently not delivered [126,172,173].

If an EV product is licensed in the future, its likely place is within an ordinary wound care workflow rather than a complex cell therapy procedure: wound assessment → debridement, offloading, infection and vascular management → product selection → topical application within a standard dressing → serial wound measurement → safety follow-up. This is a projection; it is not a practice supported by today's evidence.

What the evidence actually supports — graded

The three conclusions that matter most

First, the two null platelet-EV trials should reset expectations more than the single positive randomised trial raises them — and the Rion Phase 2a has not changed this balance but sharpened it. The null results came from unrelated sponsors, in different countries, with different products, in clean standard acute wounds, with within-patient controls that eliminate between-patient confounders and with objective closure endpoints — and they agreed with each other. The positive Egyptian trial, by contrast, came from a single centre, with a threefold baseline imbalance in the primary prognostic variable, 23% unbalanced withdrawal, arm sizes that do not match its own registry record, no exact p value for its main endpoint and no dose specified. When internal validity is weighed consistently, the null results carry more information. The strongest available defence of the positive result — and a balanced review should present it in its strengthened form — is that healthy volunteers and surgical donor sites are wound beds that do not lack what EVs supply; this is exactly the explanation the Plexoval II authors themselves proposed. On this reading, a null in acute wounds says nothing about chronic ulcers, and the trials that test the hypothesis in the right population are the Egyptian RCT, the Italian pilot and the Rion Phase 2a. The favourable but statistically inconclusive result of the Rion Phase 2a (46.4% vs 22.6%; P=0.062) reinforces the "null in acute wounds / favourable signal in chronic ulcers" pattern; but whether this pattern is real or the known tendency of small open-label trials can only be distinguished by peer-reviewed publication and a multicentre, blinded replication. This means the field now has a genuinely testable clinical hypothesis; it does not mean the hypothesis has been confirmed.

Second, the methodological problems of this field are not new — they are the same problems that produced a generation of marketed-but-low-certainty wound therapies. Cellular and acellular skin equivalents are regulator-approved and widely reimbursed; yet they carry guideline recommendations against routine use, and the IWGDF cites lack of blinding, per-protocol analyses, suspected publication bias and industry sponsorship as the reasons [34]. Becaplermin reached the market with a closure gain of approximately 15 points, received a boxed warning for cancer through observational pharmacovigilance, lost it a decade later — and is still not recommended [183,185]. The pooled healing OR of whole-cell MSC therapy is 4.64, and the guideline recommends against its use [34,37]. The EV field is on the same trajectory, with a single difference in its favour: each of the two interventions that earned moderate certainty achieved this with a single well-designed, endpoint-blinded, multicentre randomised trial. This is an attainable target. The effect size of NCT06319287 is in exactly this class; what is missing is not magnitude but design quality and replication.

Third, the gulf between what this literature says and what it shows has become not only an epistemic problem but a patient safety problem. A field with no approved product anywhere sustains approximately 60 US businesses and 669 Japanese institutions selling "exosome" therapy, in a market where 27% of claims were found to be misleading and only 18% of manufacturers are transparent [145,146]; meanwhile, 2026 review articles in indexed journals convert a retracted intravenous COVID-19 trial into a Phase II diabetic foot ulcer success and a null donor-site result into "100% re-epithelialisation" [5,38]. The documented harms — bacteraemia from contaminated product, granuloma from unpurified material, necrosis after dermal injection — are manufacturing and identity failures; that is, precisely the failures a regulated pathway prevents [1,29–31]. The most useful thing the scientific community can do for patients in the near term is not one more rodent study; it is to publish the already completed trials in peer review, to report dose and potency in an auditable form, and to stop citing results that cannot be traced to a primary source. For the physician in Turkey, the corollary is equally clear: offering a patient exosome "therapy" when no licensed product exists is following the market, not the evidence; the correct sentence to say to the patient is that this is a promising but as yet unproven approach that can only be tried within a clinical trial.

References

Numbers follow the order of first appearance in the text (Vancouver). Citations are left in their original language; clinical trial registrations were checked on 22 September 2026.

1. US Food and Drug Administration. Public Safety Notification on Exosome Products. 6 December 2019. https://www.fda.gov/vaccines-blood-biologics/safety-availability-biologics/public-safety-notification-exosome-products
2. International Society for Extracellular Vesicles. Patient information and safety notice: extracellular vesicles/exosomes and unproven therapies. 8 August 2020. https://www.isev.org/patient-information-and-safety-notice--extracellular-vesicles-exosomes-and-unproven-therapies
3. Verma N, Arora S. [Global regulatory landscape for exosome-based products.] Pharmaceutics. 2025;17(8):990. doi:10.3390/pharmaceutics17080990. https://www.mdpi.com/1999-4923/17/8/990
4. Welsh JA, Goberdhan DCI, O'Driscoll L, et al. Minimal information for studies of extracellular vesicles (MISEV2023): from basic to advanced approaches. J Extracell Vesicles. 2024;13(2):e12404. doi:10.1002/jev2.12404. PMID 38326288. https://isevjournals.onlinelibrary.wiley.com/doi/10.1002/jev2.12404
5. Li Y, Zheng F, Bai Y, Yu Y, Zhang R, Feng Q, Yu Y, Zhu L, Wang D, Liu H. Application of stem cell-derived exosomes in skin wound healing: mechanisms, prospects, and challenges. Front Immunol. 2026;17:1846916. doi:10.3389/fimmu.2026.1846916. https://www.frontiersin.org/journals/immunology/articles/10.3389/fimmu.2026.1846916/full
6. Zhu Y, Yang H, Xue Z, Tang H, Chen X, Liao Y. Mesenchymal stem cells-derived small extracellular vesicles and apoptotic extracellular vesicles for wound healing and skin regeneration: a systematic review and meta-analysis of preclinical studies. J Transl Med. 2025;23(1):364. doi:10.1186/s12967-024-05744-0. https://link.springer.com/article/10.1186/s12967-024-05744-0
7. Zheng D, Liu T, Liu Y, et al. Extracellular vesicles in wound healing and scar formation: molecular regulation of macrophages, fibroblasts, and their crosstalk. Front Cell Dev Biol. 2026;14:1917543. doi:10.3389/fcell.2026.1917543. https://www.frontiersin.org/journals/cell-and-developmental-biology/articles/10.3389/fcell.2026.1917543/full
8. Ju C, Liu D. Exosomal microRNAs from mesenchymal stem cells: novel therapeutic effect in wound healing. Tissue Eng Regen Med. 2023;20(5):647–660. doi:10.1007/s13770-023-00542-z. https://pmc.ncbi.nlm.nih.gov/articles/PMC10352215/
9. Al-Masawa ME, Alshawsh MA, Ng CY, Ng AMH, Foo JB, Vijakumaran U, Subramaniam R, Abdul Ghani NA, Witwer KW, Law JX. Efficacy and safety of small extracellular vesicle interventions in wound healing and skin regeneration: a systematic review and meta-analysis of animal studies. Theranostics. 2022;12(15):6455–6508. doi:10.7150/thno.73436. PMID 36185607. https://pmc.ncbi.nlm.nih.gov/articles/PMC9516230/
10. Mussin NM, Zhilisbayeva KR, Baspakova A, Kurmanalina MA, Tamadon A. Umbrella review of mesenchymal stem cell-derived extracellular vesicles in preclinical models: therapeutic efficacy across diverse conditions. Front Cell Dev Biol. 2025;13:1655623. doi:10.3389/fcell.2025.1655623. PMID 41158310. https://www.frontiersin.org/journals/cell-and-developmental-biology/articles/10.3389/fcell.2025.1655623/full
11. Wang X, Pan L, Shen H. Efficacy of stem cell-derived exosomes in promoting diabetic foot ulcer healing: a meta-analysis of preclinical animal studies. J Diabetes Res. 2026;2026:2578542. doi:10.1155/jdr/2578542. PMID 41846240. https://pmc.ncbi.nlm.nih.gov/articles/PMC13140873/
12. Wang X, et al. Efficacy of MSC-derived small extracellular vesicles in treating type II diabetic cutaneous wounds: a systematic review and meta-analysis of animal models. Front Endocrinol. 2024;15:1375632. doi:10.3389/fendo.2024.1375632. https://www.frontiersin.org/journals/endocrinology/articles/10.3389/fendo.2024.1375632/full
13. Luo C, Chen L, Liu J, He Y, Zeng C, Liao W. A systematic review and meta-analysis to investigate the effectiveness of exosome for diabetic wounds. J Tissue Viability. 2025;34(3):100917. doi:10.1016/j.jtv.2025.100917. (Author order differs between the source records consulted.) https://www.sciencedirect.com/science/article/pii/S0965206X25000658
14. Perkasa-Hendropriyono AAW, Perdanakusuma DS, Utomo DN, Budiman B, Lajevardi SS, Hussain G, Deva AK. Exosome therapy for flap and skin graft survival: a systematic review and meta-analysis of preclinical evidence. JPRAS Open. 2026;50:665–684. doi:10.1016/j.jpra.2026.05.042. https://www.sciencedirect.com/science/article/pii/S2352587826002093
15. Chernoff G. Human placental mesenchymal stem cell-derived exosomes in wound healing and scar therapy: a systematic review and meta-analysis. Aesthet Surg J. 2026;46(Suppl 1):S54–S61. doi:10.1093/asj/sjaf163. PMID 41800724. https://academic.oup.com/asj/article/46/Supplement_1/S54/8512050
16. Prasai A, et al. Role of exosomes in dermal wound healing: a systematic review. J Invest Dermatol. 2022;142(3):662–678.e8. doi:10.1016/j.jid.2021.07.167. PMID 34461128. https://www.sciencedirect.com/science/article/pii/S0022202X21020881
17. Jin W, Li Y, Yu M, Ren D, Han C, Guo S. Advances of exosomes in diabetic wound healing. Burns Trauma. 2025;13:tkae078. doi:10.1093/burnst/tkae078. https://academic.oup.com/burnstrauma/article/doi/10.1093/burnst/tkae078/8021121
18. Galiano RD, Michaels J, Dobryansky M, Levine JP, Gurtner GC. Quantitative and reproducible murine model of excisional wound healing. Wound Repair Regen. 2004. doi:10.1111/j.1067-1927.2004.12404.x. PMID 15260814. (Volume and pages not verified.)
19. Chen L, Mirza R, et al. The murine excisional wound model: contraction revisited. Wound Repair Regen. 2015. doi:10.1111/wrr.12338. PMID 26136050. PMCID PMC5094847. (Cited for the point, not for a numerical share of closure.)
20. Johnson J, Law SQ, Shojaee M, Hall AS, Bhuiyan S, Lim MBL, Silva A, Kong KJW, Schoppet M, Blyth C, Ranasinghe HN, Sejic N, Chuei MJ, Tatford OC, Cifuentes-Rius A, James PF, Tester A, Dixon I, Lichtfuss G. First-in-human clinical trial of allogeneic, platelet-derived extracellular vesicles as a potential therapeutic for delayed wound healing. J Extracell Vesicles. 2023;12(7):e12332. doi:10.1002/jev2.12332. PMID 37353884. PMCID PMC10290200. Registration ACTRN12620000944932 (ANZCTR). https://pmc.ncbi.nlm.nih.gov/articles/PMC10290200/pdf/JEV2-12-12332.pdf
21. Wan R, Wyles SP, Zhao C, Behfar A, Moran SL. Clinical safety and regenerative potential of a purified exosome product in treating skin graft donor sites: a phase 1b trial. J Plast Reconstr Aesthet Surg. 2026;118:9–18. PMID 42155421. NCT04664738. https://www.sciencedirect.com/science/article/abs/pii/S1748681526002494
22. Kishta MS, Hafez AM, Hydara T, et al. The transforming role of Wharton's jelly mesenchymal stem cell-derived exosomes for diabetic foot ulcer healing: a randomized controlled clinical trial. Stem Cell Res Ther. 2025;16(1):559. doi:10.1186/s13287-025-04690-y. NCT06812637. https://link.springer.com/article/10.1186/s13287-025-04690-y
23. Gibello L, D'Antico S, Salafia M, Senetta R, et al. Autologous serum-derived extracellular vesicles for chronic venous leg ulcers: a first pilot case-control study. Pharmacol Res. 2023;190:106718. Trial number CS2/1095/0090491. https://www.sciencedirect.com/science/article/pii/S1043661823000749
24. Schulman CI, Namias N, Pizano L, et al. Topical MSC-derived extracellular vesicles accelerate healing of second-degree burns: a case report. J Burn Care Res. Published online 20 August 2026. doi:10.1093/jbcr/irag148. PMID 42623276. https://academic.oup.com/jbcr/advance-article/doi/10.1093/jbcr/irag148/8767230
25. NCT06319287. Phase 2a multi-center prospective, randomized trial to evaluate the safety and efficacy of topical PEP-TISSEEL for diabetic foot ulcers. Sponsor Rion Inc.; collaborator Professional Education & Research Institute. Status Active, not recruiting; enrolment 40; study completion 12 October 2025; no results posted. Checked 22 September 2026. https://biotechhunter.com/trials/NCT06319287
26. NCT06319287. A phase 2a multi-center, prospective, randomized controlled study to evaluate the safety and efficacy of topically applied PEP-TISSEEL in subjects with diabetic foot ulcers — posted results section. Sponsor: Rion Inc. Status Completed; actual enrolment 59; study completion 1 November 2025; results first submitted 20 July 2026; last update 22 June 2026. ClinicalTrials.gov, accessed 22 September 2026. https://clinicaltrials.gov/study/NCT06319287
27. Van Delen M, Derdelinckx J, Wouters K, Nelissen I, Cools N. A systematic review and meta-analysis of clinical trials assessing safety and efficacy of human extracellular vesicle-based therapy. J Extracell Vesicles. 2024. doi:10.1002/jev2.12458. PMID 38958077. (Pooled adverse-event figures taken from a summary reproducing the article; publisher full text not retrievable.) https://isevjournals.onlinelibrary.wiley.com/doi/10.1002/jev2.12458
28. Nebraska Department of Health and Human Services. Health Advisory, 6 December 2019. https://dhhs.ne.gov/han%20Documents/ADVISORY12062019.pdf
29. Choi H, Kwak JH, Shin BS, Na CH, Kim MS. Foreign body granuloma after exosome injection. JAAD Case Rep. 2025;64:19–21. doi:10.1016/j.jdcr.2024.03.026. (The DOI year and the printed volume/year differ; verify before citing.) https://www.sciencedirect.com/science/article/pii/S2352512624001565
30. AlBargawi, et al. Necrosis following dermal injection of lyophilized exosomes. J Cosmet Dermatol. 2025. doi:10.1111/jocd.70387. (Title and citation only.)
31. Cividino M, Ryding D. Novel aesthetic treatments: an introduction to exosomes. Public Health Ontario / Ontario Ministry of Health presentation, 8 April 2025. (Government presentation; the "five *E. coli* sepsis cases" and "50 infractions" figures are from this secondary source, not from a primary FDA or CDC tally.) https://www.publichealthontario.ca/-/media/Event-Presentations/25/05/novel-aesthetic-treatments-introduction-exosomes.pdf
32. Margolis DJ, Kantor J, Berlin JA. Healing of diabetic neuropathic foot ulcers receiving standard treatment. A meta-analysis. Diabetes Care. 1999;22(5):692–695. doi:10.2337/diacare.22.5.692. PMID 10332667. (Verified from an institutional bibliographic record reproducing the abstract.)
33. Coye TL, Bargas Ochoa M, Zulbaran-Rojas A, Martinez Leal B, Quattas A, Tarricone A, Chung J, Najafi B, Lavery LA. Healing of diabetic neuropathic foot ulcers receiving standard treatment in randomised controlled trials: a random effects meta-analysis. Wound Repair Regen. 2025;33(1):e13237. doi:10.1111/wrr.13237. PMID 39654287. (Verified from an institutional research-portal record reproducing the abstract; the publisher page was not accessible.) **A formal correction exists (doi:10.1111/wrr.70010, PMID 40165304) whose content could not be retrieved; the 33.15% figure should be re-checked against it.**
34. Chen P, Vilorio NC, Dhatariya K, et al. Guidelines on interventions to enhance healing of foot ulcers in people with diabetes (IWGDF 2023 update). Diabetes Metab Res Rev. 2024. doi:10.1002/dmrr.3644. PMID 37232034. https://iwgdfguidelines.org/wp-content/uploads/2023/07/IWGDF-2023-07-Wound-Healing-Guideline.pdf
35. Edmonds M, Lázaro-Martínez JL, Alfayate-García JM, et al. Sucrose octasulfate dressing versus control dressing in patients with neuroischaemic diabetic foot ulcers (Explorer): an international, multicentre, double-blind, randomised, controlled trial. Lancet Diabetes Endocrinol. 2018;6(3):186–196. doi:10.1016/S2213-8587(17)30438-2. PMID 29275068. (Arm-level figures verified from two independent secondary reproductions of the published abstract; the publisher page returned HTTP 403.)
36. Game F, Jeffcoate W, Tarnow L, Jacobsen JL, Whitham DJ, Harrison EF, Ellender SJ, Fitzsimmons D, Löndahl M. LeucoPatch system for the management of hard-to-heal diabetic foot ulcers in the UK, Denmark, and Sweden: an observer-masked, randomised controlled trial. Lancet Diabetes Endocrinol. 2018;6(11):870–878. doi:10.1016/S2213-8587(18)30240-7. PMID 30243803. (Verified from an institutional reproduction of the abstract; the confidence-interval coverage level is rendered as 96% in one reproduction and 95% in another.)
37. Li W, Liu G, You Y, Lu H, Fan W, Zhang X. Bone marrow-derived mesenchymal stem cells combined with intramuscular injection promote the healing of diabetic foot ulcers: a systematic review and meta-analysis. Front Endocrinol. 2026;17:1763071. doi:10.3389/fendo.2026.1763071. PMID 41884216. https://www.frontiersin.org/journals/endocrinology/articles/10.3389/fendo.2026.1763071/full
38. Arkoubi AY. Role of exosome therapy in reconstructive surgery: a review. Front Med (Lausanne). 2026;13:1697228. doi:10.3389/fmed.2026.1697228. Published 18 March 2026. https://www.frontiersin.org/journals/medicine/articles/10.3389/fmed.2026.1697228/full
39. NCT05125562. Extracellular vesicles infusion treatment for mild-to-moderate COVID-19. Sponsor Direct Biologics, LLC. Phase 2; single intravenous ExoFlo infusion in three dose groups from 7×10¹¹ to 10.5×10¹¹ extracellular vesicles; **status WITHDRAWN**; 7 December 2022 to 7 March 2023. Checked 22 September 2026. https://biotechhunter.com/trials/NCT05125562
40. Théry C, Witwer KW, Aikawa E, et al. Minimal information for studies of extracellular vesicles 2018 (MISEV2018). J Extracell Vesicles. 2018;7(1):1535750. doi:10.1080/20013078.2018.1535750. (PMID not independently verified; cite the DOI.) https://www.tandfonline.com/doi/abs/10.1080/20013078.2018.1535750
41. van Niel G, D'Angelo G, Raposo G. Shedding light on the cell biology of extracellular vesicles. Nat Rev Mol Cell Biol. 2018;19(4):213–228. doi:10.1038/nrm.2017.125. PMID 29339798. https://www.nature.com/articles/nrm.2017.125
42. Kou M, Huang L, Yang J, et al. Mesenchymal stem cell-derived extracellular vesicles for immunomodulation and regeneration: a next generation therapeutic tool? Cell Death Dis. 2022;13(7):580. doi:10.1038/s41419-022-05034-x. https://www.nature.com/articles/s41419-022-05034-x
43. Ahmadian S, Jafari N, Tamadon A, Ghaffarzadeh A, Rahbarghazi R, Mahdipour M. Different storage and freezing protocols for extracellular vesicles: a systematic review. Stem Cell Res Ther. 2024;15:453. doi:10.1186/s13287-024-04005-7. https://link.springer.com/article/10.1186/s13287-024-04005-7
44. Mesenchymal stem cell-derived extracellular vesicles in skin wound healing: roles, opportunities and challenges. Mil Med Res. 2023;10:36. doi:10.1186/s40779-023-00472-w. PMID 37587531. (Author list could not be confirmed in the underlying research.) https://link.springer.com/article/10.1186/s40779-023-00472-w
45. Saberian M, Davoudi M, Ghafourian A, Moradi-Sardareh H, Taslim Bakhsh F, Afrisham R. Biogenesis, isolation, and mechanistic applications of platelet-rich plasma and PRP-derived exosomes in diabetic and chronic wound healing: a comprehensive narrative review. Regen Eng Transl Med. 2025. doi:10.1007/s40883-025-00468-z. https://link.springer.com/article/10.1007/s40883-025-00468-z
46. Poupardin R, Wolf M, Strunk D. Adherence to minimal experimental requirements for defining extracellular vesicles and their functions: a systematic review. Adv Drug Deliv Rev. 2021. doi:10.1016/j.addr.2021.113872. https://www.biorxiv.org/content/10.1101/2021.04.23.441160v1.full
47. Van Deun J, Mestdagh P, Agostinis P, et al. EV-TRACK: transparent reporting and centralizing knowledge in extracellular vesicle research. Nat Methods. 2017. doi:10.1038/nmeth.4185. PMID 28245209. https://www.evtrack.org/about.php
48. Hu P, Yang Q, Wang Q, Shi C, Wang D, Armato U, et al. Mesenchymal stromal cells-exosomes: a promising cell-free therapeutic tool for wound healing and tissue regeneration. Burns Trauma. 2019;7:38. doi:10.1186/s41038-019-0178-8.
49. An Y, Lin S, Tan X, Zhu S, Nie F, Zhen Y, et al. Exosomes from adipose-derived stem cells and application to skin wound healing. Cell Prolif. 2021;54(3):e12993. doi:10.1111/cpr.12993.
50. MSC-derived exosome promotes M2 polarization and enhances cutaneous wound healing. PMID 31582986. (Stem Cells Int 2019; authors and volume not verified.) https://pubmed.ncbi.nlm.nih.gov/31582986/
51. Staphylococcus aureus vesicles impair cutaneous wound healing through p38 MAPK-MerTK cleavage-mediated inhibition of macrophage efferocytosis. PMCID PMC11708000. https://pmc.ncbi.nlm.nih.gov/articles/PMC11708000/
52. Exosomes secreted by mesenchymal stem cells promote endothelial cell angiogenesis by transferring miR-125a. PMID 27252357. (J Cell Sci 2016; authors and pages not verified.) https://pubmed.ncbi.nlm.nih.gov/27252357/
53. Human umbilical cord mesenchymal stem cell exosomes enhance angiogenesis through the Wnt4/β-catenin pathway. Stem Cells Transl Med. 2015;4(5):513–522. PMID 25824139. https://academic.oup.com/stcltm/article/4/5/513/6397350
54. Zhang B, Wang M, Gong A, Zhang X, Wu X, Zhu Y, Shi H, Wu L, Zhu W, Qian H, Xu W. HucMSC-exosome mediated-Wnt4 signaling is required for cutaneous wound healing. Stem Cells. 2015;33(7):2158–2168. doi:10.1002/stem.1771. PMID 24964196 (not 25964196, a common miscitation). https://academic.oup.com/stmcls/article/33/7/2158/6407330
55. HucMSC exosome-delivered 14-3-3ζ orchestrates self-control of the Wnt response via modulation of YAP during cutaneous regeneration. Stem Cells. 2016;34(10):2485–2500. PMID 27334574. https://academic.oup.com/stmcls/article/34/10/2485/6408372
56. Umbilical cord-derived mesenchymal stem cell-derived exosomal microRNAs suppress myofibroblast differentiation by inhibiting the transforming growth factor-β/SMAD2 pathway during wound healing. PMID 27388239. (Fang S et al., Stem Cells Transl Med 2016; author list, volume and pages not verified; content taken from a retrieved secondary review.) https://pubmed.ncbi.nlm.nih.gov/27388239/
57. ADSC exosomes improve high glucose induced fibroblast oxidative stress injury and accelerate DFU wound healing via regulating Keap1/Nrf2 axis. PMID 40505843. (Journal and year not retrieved.) https://pubmed.ncbi.nlm.nih.gov/40505843/
58. Exosomes from adipose-derived stem cells overexpressing Nrf2 accelerate cutaneous wound healing by promoting vascularization in a diabetic foot ulcer rat model. PMID 29651102. https://pubmed.ncbi.nlm.nih.gov/29651102/
59. Circulating exosomal miR-181b-5p promoted cell senescence and inhibited angiogenesis to impair diabetic foot ulcer via the Nrf2/HO-1 pathway. PMCID PMC9067436. (Journal and authors not retrieved.) https://pmc.ncbi.nlm.nih.gov/articles/PMC9067436/
60. Chen B, Sun Y, Zhang J, et al. Human embryonic stem cell-derived exosomes promote pressure ulcer healing in aged mice by rejuvenating senescent endothelial cells. Stem Cell Res Ther. 2019;10:142. doi:10.1186/s13287-019-1253-6. https://link.springer.com/article/10.1186/s13287-019-1253-6
61. Li Z, Liu J, Song J, et al. Multifunctional hydrogel-based engineered extracellular vesicles delivery for complicated wound healing. Theranostics. 2024;14(11):4198–4217. doi:10.7150/thno.97317. https://www.thno.org/v14p4198.htm
62. Zhang S, Zhou J, Liu J, et al. Hyperbaric oxygen in combination with exosomes: a new strategy to promote tissue repair. Front Bioeng Biotechnol. 2025;13:1639060. doi:10.3389/fbioe.2025.1639060. PMID 40994671 (a correction exists, PMID 41280640). https://www.frontiersin.org/journals/bioengineering-and-biotechnology/articles/10.3389/fbioe.2025.1639060/full
63. Yang G, Waheed S, Wang C, Shekh M, Li Z, Wu J. Exosomes and their bioengineering strategies in the cutaneous wound healing and related complications: current knowledge and future perspectives. Int J Biol Sci. 2023;19(5):1430–1454. doi:10.7150/ijbs.80430. https://www.ijbs.com/v19p1430.htm
64. Pulmonary biodistribution of platelet-derived regenerative exosomes in a porcine model. Int J Mol Sci. 2024. doi:10.3390/ijms25052642. PMID 38473889.
65. Wan R, Nishimura K, Behfar A, Zhao C, Moran SL. Platelet-derived purified exosome product: manufacturing, delivery and clinical translation. Pharmaceutics. 2026;18(2):222. doi:10.3390/pharmaceutics18020222. https://www.mdpi.com/1999-4923/18/2/222
66. Chevillet JR, Kang Q, Ruf IK, Briggs HA, Vojtech LN, Hughes SM, et al. Quantitative and stoichiometric analysis of the microRNA content of exosomes. Proc Natl Acad Sci U S A. 2014;111:14888–14893. doi:10.1073/pnas.1408301111. PMID 25267620. https://pubmed.ncbi.nlm.nih.gov/25267620/
67. Turchinovich A, Tonevitsky AG, Cho WC, Burwinkel B. Check and mate to exosomal extracellular miRNA: new lesson from a new approach. Front Mol Biosci. 2015;2:11. doi:10.3389/fmolb.2015.00011. https://www.frontiersin.org/journals/molecular-biosciences/articles/10.3389/fmolb.2015.00011/full
68. Albanese M, Chen YFA, Hüls C, Gärtner K, Tagawa T, Mejias-Perez E, et al. MicroRNAs are minor constituents of extracellular vesicles that are rarely delivered to target cells. PLoS Genet. 2021;17(12):e1009951. doi:10.1371/journal.pgen.1009951. PMID 34871319. https://journals.plos.org/plosgenetics/article?id=10.1371/journal.pgen.1009951
69. Whittaker TE, et al. Experimental artefacts can lead to misattribution of bioactivity from soluble mesenchymal stem cell paracrine factors to extracellular vesicles. J Extracell Vesicles. 2020;9(1):1807674. doi:10.1080/20013078.2020.1807674. (Title-level; full text not retrievable.)
70. Roux Q, et al. Depletion of soluble cytokines unlocks the immunomodulatory bioactivity of extracellular vesicles. J Extracell Vesicles. 2023;12:e12339. doi:10.1002/jev2.12339. (Title-level.)
71. McConnell RE, et al. Transfection reagent artefact likely accounts for some reports of extracellular vesicle function. J Extracell Vesicles. 2022;11:e12253. doi:10.1002/jev2.12253. (Title-level.)
72. Wang F, Yao J, Zuo H, Jiao Y, Wu J, Meng Z. Diverse-origin exosomes therapeutic strategies for diabetic wound healing. Int J Nanomed. 2025;20:7375–7402. doi:10.2147/IJN.S519379. https://www.dovepress.com/diverse-origin-exosomes-therapeutic-strategies-for-diabetic-wound-heal-peer-reviewed-fulltext-article-IJN
73. Pomatto M, Gai C, Negro F, et al. Differential therapeutic effect of extracellular vesicles derived by bone marrow and adipose mesenchymal stem cells on wound healing of diabetic ulcers. Int J Mol Sci. 2021;22(8):3851. doi:10.3390/ijms22083851. https://www.mdpi.com/1422-0067/22/8/3851
74. Than UT, Hoang DH, Nguyen TD, et al. Induced mesenchymal stem cell-derived exosomes and their effects on skin cells. Front Mol Biosci. 2020;7:119. doi:10.3389/fmolb.2020.00119. https://www.frontiersin.org/journals/molecular-biosciences/articles/10.3389/fmolb.2020.00119/full
75. Ponnikorn S, et al. Comparative characterisation of umbilical cord versus adipose mesenchymal stem cell exosomes. Aesthet Surg J. 2026;46(Suppl 1):S38–S53. doi:10.1093/asj/sjaf254. https://academic.oup.com/asj/article/46/Supplement_1/S38/8512054
76. Soltani S, Zahedi A, Vergara AJS, et al. Effect of adipose-derived stem cell exosomes on diabetic wound healing: a systematic review and meta-analysis of preclinical studies. Stem Cell Rev Rep. 2024;20(8):2016–2031. doi:10.1007/s12015-024-10753-z. (Bibliographic details verified from the publisher record; not independently PMID-confirmed.) https://link.springer.com/article/10.1007/s12015-024-10753-z
77. Zhang J, Chen CY, Hu B, et al. Exosomes derived from human endothelial progenitor cells accelerate cutaneous wound healing by promoting angiogenesis through Erk1/2 signaling. Int J Biol Sci. 2016;12(12):1472–1487. doi:10.7150/ijbs.15514. PMID 27994512.
78. Chen CY, Rao SS, Ren L, et al. Exosomal DMBT1 from human urine-derived stem cells facilitates diabetic wound repair by promoting angiogenesis. Theranostics. 2018;8(6):1607–1623. doi:10.7150/thno.22958. PMID 29556344.
79. Kim HS, Wang SY, Kwak G, et al. Exosome-guided phenotypic switch of M1 to M2 macrophages for cutaneous wound healing. Adv Sci. 2019;6(20):1900513. doi:10.1002/advs.201900513. PMID 31637157.
80. Carvajal J, Carvajal M. Could exosomes be a promising therapy to accelerate wound healing and skin regeneration in ischemic nasal flaps? Aesthet Surg J Open Forum. 2025;7:ojaf060. doi:10.1093/asjof/ojaf060. PMID 40688485. https://academic.oup.com/asjopenforum/article/doi/10.1093/asjof/ojaf060/8160111
81. [Hypoxia-preconditioned adipose-derived stem cell exosomes accelerate diabetic wound healing.] J Nanobiotechnology. 2021;19:202. doi:10.1186/s12951-021-00942-0. PMID 34233694. (Author list and exact title not captured in the underlying research.) https://link.springer.com/article/10.1186/s12951-021-00942-0
82. Zhang Y, Di T, Wu H, et al. Extracellular vesicle therapies for diabetic skin lesions: mechanisms, engineering, and clinical translation. Int J Nanomedicine. 2026;21:626719. doi:10.2147/IJN.S626719. https://www.dovepress.com/extracellular-vesicle-therapies-for-diabetic-skin-lesions-mechanisms-e-peer-reviewed-fulltext-article-IJN
83. Gurunathan S, Kang MH, Jeyaraj M, Qasim M, Kim JH. Review of the isolation, characterization, biological function, and multifarious therapeutic approaches of exosomes. Cells. 2019;8(4):307. doi:10.3390/cells8040307.
84. Lehrich BM, Liang Y, Khosravi P, Federoff HJ, Fiandaca MS. Fetal bovine serum-derived extracellular vesicles persist within vesicle-depleted culture media. Int J Mol Sci. 2018;19(11):3538. doi:10.3390/ijms19113538. PMID 30423996. https://www.mdpi.com/1422-0067/19/11/3538
85. Pachler K, et al. A Good Manufacturing Practice-grade standard protocol for exclusively human mesenchymal stromal cell-derived extracellular vesicles. Cytotherapy. 2017. PMID 28188071. https://pubmed.ncbi.nlm.nih.gov/28188071/
86. Gupta D, Zickler AM, El Andaloussi S. Dosing extracellular vesicles. Adv Drug Deliv Rev. 2021. PMID 34481030. https://pubmed.ncbi.nlm.nih.gov/34481030/
87. Tertel T, Giebel B, Lee RH, Muraca M, Ortiz LA, Parolini O, Perruche S, Robbins PD, Rohde E, Strader SR, Terai S, Toh WS, Yarani R, Weiss DJ, Wiest EF, Lim SK. ISCT exosomes signature series: manufacturing and quality considerations for MSC-EV therapeutics. Cytotherapy. 2026;28(3):102016. doi:10.1016/j.jcyt.2025.102016. https://www.sciencedirect.com/science/article/pii/S1465324925009120
88. Moon H, Jung SH, Bae S, et al. A product control framework for extracellular vesicle therapeutics: quality attributes, stability, analytical procedures, and biodistribution. Int J Pharm X. 2026;12:100650. doi:10.1016/j.ijpx.2026.100650. (The fetched record gave a publication date of 1 December 2026, which post-dates the compilation of the underlying research; the volume and date require verification.) https://www.sciencedirect.com/science/article/pii/S2590156726001702
89. Wiest EF, Zubair AC. Generation of cGMP-grade MSC-derived extracellular vesicles using automated bioreactors. Biology. 2025;14(3):313. doi:10.3390/biology14030313. PMID 40136569. https://www.mdpi.com/2079-7737/14/3/313
90. Kink JA, et al. Large-scale bioreactor production of extracellular vesicles from mesenchymal stromal cells for treatment of acute radiation syndrome. Stem Cell Res Ther. 2024. doi:10.1186/s13287-024-03688-2. https://link.springer.com/article/10.1186/s13287-024-03688-2
91. Haraszti RA, Miller R, Stoppato M, et al. Exosomes produced from 3D cultures of MSCs by tangential flow filtration show higher yield and improved activity. Mol Ther. 2018;26(12):2838–2847. doi:10.1016/j.ymthe.2018.09.015. PMID 30341012.
92. [Review of scalable exosome production for therapeutic applications.] ScienceDirect article S2096691125000196; accessed 22 September 2026. **This citation is incomplete: the underlying research captured no journal title, volume, DOI, PMID or author list, and attempts to resolve them from the notes found no further detail. It is load-bearing for the yield and cost-driver figures in Sections 4.8 and 12.6, which should therefore be re-verified against the primary article before being relied upon.** https://www.sciencedirect.com/science/article/pii/S2096691125000196
93. Lin WH. Topical application of human umbilical cord-derived mesenchymal stem cell exosomes for a deep second-degree scald burn wound: a case report. Cureus. 2026;18(4):e107198. doi:10.7759/cureus.107198. PMID 42153042. https://www.cureus.com/articles/455510-topical-application-of-human-umbilical-cord-derived-mesenchymal-stem-cell-exosomes-for-a-deep-second-degree-scald-burn-wound-a-case-report
94. NCT04173650. Mesenchymal stem cell extracellular vesicles for the treatment of recessive dystrophic epidermolysis bullosa wounds (AGLE-102). Sponsor Aegle Therapeutics. Registry record read via server-rendered mirrors, checked 22 September 2026. https://ichgcp.net/clinical-trials-registry/NCT04173650
95. Nazari H, Alborzi F, Heirani-Tabasi A, et al. Evaluating the safety and efficacy of mesenchymal stem cell-derived exosomes for treatment of refractory perianal fistula in IBD patients: clinical trial phase I. Gastroenterol Rep (Oxf). 2022;10:goac075. doi:10.1093/gastro/goac075. PMID 36518984. IRCT20200413047063N3; dual-registered NCT05499156. https://academic.oup.com/gastro/article/doi/10.1093/gastro/goac075/6886496
96. Pak H, et al. Safety and efficacy of injection of human placenta mesenchymal stem cells derived exosomes for treatment of complex perianal fistula in non-Crohn's cases: clinical trial phase I. J Gastroenterol Hepatol. 2023. PMID 36640153. (Full text not retrievable; sample size and outcomes not extracted.) https://pubmed.ncbi.nlm.nih.gov/36640153/
97. Park JY, Park JH. Laser-assisted delivery of exosome boosters for postoperative facial scars: a randomized comparative study. Life. 2026;16(2):217. Published 28 January 2026. (IRB 20202524; not prospectively registered.) https://www.mdpi.com/2075-1729/16/2/217
98. Cai Z, et al. The efficacy of platelet-derived extracellular vesicles in the treatment of diabetic wounds: a systematic review and meta-analysis of animal studies. Arch Dermatol Res. 2025;317:244. doi:10.1007/s00403-024-03781-9. https://link.springer.com/article/10.1007/s00403-024-03781-9
99. Elakkawi MMJ, Zhu Y, Xiong Y, et al. Study on the mechanism of hUCMSCs-derived 3D exosome combined with HAMA microneedles in the treatment of deep second-degree burns. Bioact Mater. 2026;56:468–482. doi:10.1016/j.bioactmat.2025.09.031. https://www.sciencedirect.com/science/article/pii/S2452199X25004311
100. Hu L, Wang J, Zhou X, Xiong Z, Zhao J, Yu R, Huang F, Zhang H, Chen L. Exosomes derived from human adipose mesenchymal stem cells accelerate cutaneous wound healing via optimizing the characteristics of fibroblasts. Sci Rep. 2016;6:32993. doi:10.1038/srep32993. https://www.nature.com/articles/srep32993
101. Hooijmans CR, Rovers MM, de Vries RBM, Leenaars M, Ritskes-Hoitinga M, Langendam MW. SYRCLE's risk of bias tool for animal studies. BMC Med Res Methodol. 2014;14:43. doi:10.1186/1471-2288-14-43. PMID 24667063.
102. Hooijmans CR, et al. Facilitating healthcare decisions by assessing the certainty in the evidence from preclinical animal studies. PLoS One. 2018. doi:10.1371/journal.pone.0187271. PMID 29324741. https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0187271
103. NCT06793748. Study to evaluate the safety, tolerability, and efficacy of PEP-TISSEEL in subjects with chronic radiation ulcer. Sponsor Rion Inc.; collaborators Walter Reed National Military Medical Center and the Henry M. Jackson Foundation. Phase 1/2; estimated enrolment 184; status Not yet recruiting. Checked 22 September 2026. https://ichgcp.net/clinical-trials-registry/NCT06793748
104. [Systematic review and meta-analysis of exosome therapies in aesthetic medicine, 39 human clinical trials to 2 February 2025.] Aesthet Surg J. 2026;46(Suppl 1):S13. doi:10.1093/asj/sjaf178. https://academic.oup.com/asj/article/46/Supplement_1/S13/8512052
105. Kuang S, Dong D, Xu J, Zhang S. Efficacy and safety of exosome-based therapy for diabetic foot ulcer: a systematic review and meta-analysis of clinical studies. SSRN working paper 7427549. **Preprint, not peer-reviewed; cited only to record its existence.** https://papers.ssrn.com/sol3/papers.cfm?abstract_id=7427549
106. NCT04664738. PEP on a skin graft donor site wound. Sponsor Rion Inc.; collaborator ProPharma Group. Status Completed, study completion 2 February 2024; enrolment 8; no results posted. Checked 22 September 2026. https://biotechhunter.com/trials/NCT04664738
107. NCT04652531 (SER-VES-HEAL). Autologous serum-derived extracellular vesicles to treat venous trophic lesions not responsive to conventional treatments. Sponsor: University of Turin. Estimated enrolment 10; status Recruiting (last verified May 2023); no results posted. ClinicalTrials.gov, accessed 22 September 2026. https://clinicaltrials.gov/study/NCT04652531
108. [Analysis of 90 published exosome clinical trials indexed in PubMed to 28 March 2025.] J Pharmacol Pharmacother. 2026. doi:10.1177/0976500X251361201. https://journals.sagepub.com/doi/10.1177/0976500X251361201
109. NCT05402748. Safety and efficacy of injection of human placenta mesenchymal stem cells derived exosomes for treatment of complex anal fistula. Sponsor Tehran University of Medical Sciences. Phase 1/2; enrolment 80; status Active–Recruiting with last update 1 November 2022. Checked 22 September 2026. https://www.centerwatch.com/clinical-trials/listings/NCT05402748/
110. NCT05499156. Safety of injection of placental mesenchymal stem cell derived exosomes for treatment of resistant perianal fistula in Crohn's patients. Sponsor Tehran University of Medical Sciences. Phase 1/2; enrolment 80. (Product and sample-size fields conflict with the publication that cites this registration, reference 82.) Checked 22 September 2026. https://biotechhunter.com/trials/NCT05499156
111. Kwon HH, Yang SH, Lee J, Park BC, Park KY, Jung JY, Bae Y, Park GH. Combination treatment with human adipose tissue stem cell-derived exosomes and fractional CO2 laser for acne scars: a 12-week prospective, double-blind, randomized, split-face study. Acta Derm Venereol. 2020;100:adv00310. doi:10.2340/00015555-3666. PMID 33073298. (No trial registration.) https://www.medicaljournals.se/acta/content/html/10.2340/00015555-3666
112. The efficacy of MSC-derived exosome-based therapies in treating scars, aging and hyperpigmentation: a systematic review of human clinical outcomes. Cosmetics (MDPI). 2025;8(4):268. https://www.mdpi.com/2571-841X/8/4/268
113. Jeong KH, Park KY, Jung DW. A split-scar study: study of surgical scar using exosomes. J Cosmet Dermatol. 2026;25(3):e70756. doi:10.1111/jocd.70756. PMCID PMC12945872. https://onlinelibrary.wiley.com/doi/10.1111/jocd.70756
114. [Phase I/II single-arm trial of human umbilical cord MSC-derived conditioned medium in diabetic foot ulcer; NCT06825884, Jordan.] Stem Cell Res Ther. 2025. doi:10.1186/s13287-025-04736-1. PMID 41287013. (Exact title and author list not captured.) https://link.springer.com/article/10.1186/s13287-025-04736-1
115. [10% secretome-hUCMSC gel for chronic wounds (diabetic and trophic ulcer), phase 2, Indonesia.] J Multidiscip Healthc. 2023. doi:10.2147/JMDH.S408162. (Design, sample size and outcomes not extractable.)
116. Messa GE, Tiongco RP, Lau FH. Recurrent ischial pressure ulcer treated with injected exosomes. J Surg Case Rep. 2022;2022(6):rjac271. PMID 35774473. https://academic.oup.com/jscr/article/2022/6/rjac271/6618692
117. Pumford, et al. [Non-healing scalp wound treated with purified exosome product.] Mayo Clin Proc. 2024. Article S0025-6196(24)00207-6. (Author list beyond the first author, exact title, volume and pages were not captured; details of the case as reported in reference 49, the journal page itself not being retrievable.) https://www.mayoclinicproceedings.org/article/S0025-6196(24)00207-6/fulltext
118. NCT05078385. Safety of extracellular vesicles for burn wounds (AGLE-102). Sponsor Aegle Therapeutics; collaborator Congressionally Directed Medical Research Programs. Actual enrolment 1; status Completed; no results posted. Checked 22 September 2026. https://biotechhunter.com/trials/NCT05078385
119. Aegle Therapeutics Corp. announces positive data for the first patient in a phase 1/2a clinical trial dosed with AGLE-102. PR Newswire, 8 January 2024. (Company source; its POSAS figures differ from the peer-reviewed case report, reference 9, and must not be mixed.) https://www.prnewswire.com/news-releases/302027568.html
120. [Clinical applications of extracellular vesicles.] Front Bioeng Biotechnol. 2025;13:1671963. doi:10.3389/fbioe.2025.1671963. https://www.frontiersin.org/journals/bioengineering-and-biotechnology/articles/10.3389/fbioe.2025.1671963/full
121. FDAAA TrialsTracker record for NCT04173650 (results due 30 September 2026, not yet reported; tracker data check 21 September 2026). https://fdaaa.trialstracker.net/trial/NCT04173650/
122. NCT02565264. Effect of plasma derived exosomes on intractable cutaneous wound healing: prospective trial. Sponsor Kumamoto University, Japan. Early Phase 1; enrolment 5 anticipated; status Unknown; last update 8 September 2020; no results posted. Checked 22 September 2026. https://ichgcp.net/clinical-trials-registry/NCT02565264
123. NCT05475418. Pilot study of human adipose tissue derived exosomes promoting wound healing. Sponsor: Shanghai Ninth People's Hospital, Shanghai Jiao Tong University. Actual enrolment 5; status Completed (15 October 2023); no results posted. ClinicalTrials.gov, accessed 22 September 2026. https://clinicaltrials.gov/study/NCT05475418
124. NCT05243368. Evaluation of personalized nutritional intervention together with the application of MSC-derived exosomes on the regenerative capacity and wound healing of cutaneous ulcers in diabetics. Sponsor: Maimónides Biomedical Research Institute of Córdoba (IMIBIC). Estimated enrolment 30; status Recruiting (October 2023); no results posted. ClinicalTrials.gov, accessed 22 September 2026. https://clinicaltrials.gov/study/NCT05243368
125. NCT06429033. Purified exosome product (PEP) injected into the hypodermis. Sponsor Clinical Testing of Beverly Hills. Phase 1; enrolment 9; recruitment status disputed between registry mirrors. Checked 22 September 2026. https://biotechhunter.com/trials/NCT06429033
126. Shi C, Dumville JC, Cullum N, Connaughton E, Norman G. Compression bandages or stockings versus no compression for treating venous leg ulcers. Cochrane Database Syst Rev. 2021;(7):CD013397. doi:10.1002/14651858.CD013397.pub2. https://www.cochrane.org/evidence/CD013397_compression-bandages-or-stockings-versus-no-compression-treating-venous-leg-ulcers
127. FDA grants Fast Track status to Aegle Therapeutics' AGLE-102 for the treatment of dystrophic epidermolysis bullosa. PR Newswire, 1 October 2020. (Company source.) https://www.prnewswire.com/news-releases/fda-grants-fast-track-status-to-aegle-therapeutics-agle-102-for-the-treatment-of-dystrophic-epidermolysis-bullosa-301140426.html
128. Extracellular vesicles derived from human adipose-derived stem cell prevent the formation of hypertrophic scar in a rabbit model. Ann Plast Surg. 2020. doi:10.1097/SAP.0000000000002357. PMID 32282497. (Numerical outcomes not retrievable.) https://pubmed.ncbi.nlm.nih.gov/32282497/
129. Shiekh PA, Singh A, Kumar A. Exosome laden oxygen releasing antioxidant and antibacterial cryogel wound dressing OxOBand alleviate diabetic and infectious wound healing. Biomaterials. 2020;249:120020. doi:10.1016/j.biomaterials.2020.120020. PMID 32305816. https://www.sciencedirect.com/science/article/abs/pii/S0142961220302660
130. Bai Z, Zhao Y, Gong Y, et al. Temperature-responsive hydrogel delivery of antimicrobial peptide engineered watermelon-derived extracellular vesicles enables sequential infection control and wound healing. Colloids Surf B Biointerfaces. 2026;261:115438. doi:10.1016/j.colsurfb.2026.115438. https://www.sciencedirect.com/science/article/abs/pii/S0927776526000263
131. Lu, et al. Native and engineered extracellular vesicles for wound healing. Front Bioeng Biotechnol. 2022;10:1053217. doi:10.3389/fbioe.2022.1053217. (Full author list not captured.) https://www.frontiersin.org/articles/10.3389/fbioe.2022.1053217/full
132. Lei Z, Chen X, Chen K, et al. Exosome-like vesicles encapsulated with specific microRNAs accelerate burn wound healing and ameliorate scarring. J Nanobiotechnology. 2025;23:264. doi:10.1186/s12951-025-03337-7. https://link.springer.com/article/10.1186/s12951-025-03337-7
133. Wang C, Wang M, Xu T, et al. Engineering bioactive self-healing antibacterial exosomes hydrogel for promoting chronic diabetic wound healing and complete skin regeneration. Theranostics. 2019;9(1):65–76. doi:10.7150/thno.29766. https://www.thno.org/v09p0065.htm
134. Liu B, Chen L, Huang C, et al. A sprayable exosome-loaded hydrogel with controlled release and multifunctional synergistic effects for diabetic wound healing. Mater Today Bio. 2025;34:102159. doi:10.1016/j.mtbio.2025.102159. PMID 40822934. https://www.sciencedirect.com/science/article/pii/S259000642500729X
135. Yang Y, Cui J, Kong Y, Hou Y, Zhang H, Ma C. The role of hydrogel biomaterials in the intervention of wound healing and skin regeneration via exosomes: a systematic review and meta-analysis of preclinical animal studies. Adv Wound Care. 2025;14(12):669–684. doi:10.1089/wound.2024.0058. **Pooled effect sizes paywalled and not obtained.** https://journals.sagepub.com/doi/abs/10.1089/wound.2024.0058
136. Yuan M, Liu K, Jiang T, et al. GelMA/PEGDA microneedles patch loaded with HUVECs-derived exosomes and tazarotene promote diabetic wound healing. J Nanobiotechnology. 2022;20:147. doi:10.1186/s12951-022-01354-4. PMID 35305648. https://link.springer.com/article/10.1186/s12951-022-01354-4
137. Jian X, Deng Y, Xiao S, Qi F, Deng C. Microneedles in diabetic wound care: multifunctional solutions for enhanced healing. Burns Trauma. 2025;13:tkae076. doi:10.1093/burnst/tkae076. https://academic.oup.com/burnstrauma/article/doi/10.1093/burnst/tkae076/8015689
138. Guo SC, Tao SC, Yin WJ, Qi X, Yuan T, Zhang CQ. Exosomes derived from platelet-rich plasma promote the re-epithelization of chronic cutaneous wounds via activation of YAP in a diabetic rat model. Theranostics. 2017;7(1):81–96. doi:10.7150/thno.16803. https://www.thno.org/v07p0081.htm
139. Qahl SH, Bakheet RS, Ghouth NM, et al. Enhanced repair of type 2 diabetic wounds using combined hyperbaric oxygen and M2 macrophage-derived exosomes: an experimental rat model. Tissue Cell. 2026;103:103701. doi:10.1016/j.tice.2026.103701. https://www.sciencedirect.com/science/article/abs/pii/S0040816626003952
140. US Food and Drug Administration. Public Safety Alert due to Marketing of Unapproved Stem Cell and Exosome Products. Posted 9 December 2019. https://www.fda.gov/safety/medical-product-safety-information/public-safety-alert-due-marketing-unapproved-stem-cell-and-exosome-products
141. Centers for Disease Control and Prevention. Stem cell and exosome products (archived page). https://archive.cdc.gov/www_cdc_gov/hai/outbreaks/stem-cell-products.html
142. Park, et al. Adverse reactions following intradermal injection of exosome-based formulations: a case series. J Cosmet Dermatol. 2025. doi:10.1111/jocd.70520. PMID 41097876. (Title and citation only; full text not retrievable.)
143. Nahm WJ, et al. Complications after exosome treatment for aesthetic skin rejuvenation. Dermatol Rev. 2024. doi:10.1002/der2.242 (correction 2025, doi:10.1002/der2.70017). (Title and citation only.)
144. Davarcıoğlu, et al. The role of high-frequency ultrasonography in managing adverse reactions to exosome-based rejuvenation. J Cosmet Dermatol. 2026. doi:10.1111/jocd.71023. (Title and citation only.)
145. Fujita M, Hatta T, Ikka T, Onishi T. [Commentary on unapproved exosome interventions and the regulatory gap.] Stem Cell Reports. 2024;19(11):1517–1519. doi:10.1016/j.stemcr.2024.09.008. https://www.sciencedirect.com/science/article/pii/S2213671124002704
146. Rahman E, Sayed K, Rao P, et al. Exosome revolution or marketing mirage? AI-based multi-domain evaluation of claims, scientific evidence, transparency, public sentiment, and media narratives. Aesthetic Plast Surg. 2025;49(12):3454–3479. doi:10.1007/s00266-025-04712-3. https://link.springer.com/article/10.1007/s00266-025-04712-3
147. Yang BL, Long YY, Lei Q, et al. Intravenously administered mesenchymal stromal cell-derived large extracellular vesicles cause tissue-factor-dependent pulmonary thromboembolism in mice. Acta Pharmacol Sin. 2024;45:2300–2312. doi:10.1038/s41401-024-01327-3. https://www.nature.com/articles/s41401-024-01327-3
148. Rat bone marrow mesenchymal stem cells-derived exosomes promote the proliferation, invasion, and metastasis and inhibit apoptosis of colorectal cancer stem cells. PMID 41140708. https://pubmed.ncbi.nlm.nih.gov/41140708/
149. Dual impacts of mesenchymal stem cell-derived exosomes on cancer cells. PMID 37973719. https://pubmed.ncbi.nlm.nih.gov/37973719/
150. Mesenchymal stem cell derived exosomes in cancer progression, metastasis and drug delivery: a comprehensive review. J Cancer. 2018;9:3129. https://www.jcancer.org/v09p3129.htm
151. Terai S, Asonuma M, Hoshino A, et al. Guidance on the clinical application of extracellular vesicles (JSRM/JSEV). Regen Ther. 2025;29:43–50. doi:10.1016/j.reth.2025.02.008. https://www.sciencedirect.com/science/article/pii/S2352320425000343
152. [Review of the extracellular vesicle drug landscape.] Signal Transduct Target Ther. 2025. doi:10.1038/s41392-025-02312-w. https://www.nature.com/articles/s41392-025-02312-w
153. US Food and Drug Administration. Warning letter, Frontier Biologics LLC (Waco, TX), 1 November 2024. (Amniotic products, not exosomes; cited for the documented sterility and aseptic-processing failures.) https://www.fda.gov/inspections-compliance-enforcement-and-criminal-investigations/warning-letters/frontier-biologics-llc-686059-11012024
154. US Food and Drug Administration. Warning letter, Dynamic Stem Cell Therapy (Henderson, NV), 11 February 2026. https://www.fda.gov/inspections-compliance-enforcement-and-criminal-investigations/warning-letters/dynamic-stem-cell-therapy-712579-02112026
155. [Overview of exosomes in aesthetic medicine.] Aesthet Surg J. 2026;46(Suppl 1):S1. **Journal, year, volume, supplement and first page are verified from the publisher page; the article title, author list and DOI were not captured and could not be resolved from the notes.** It is load-bearing for the "no regulatory agency worldwide has yet approved an exosome product" statement and for the UK, Japan and South Korea regulatory characterisations in Section 11.2 — the last of which this report already flags as unsupported by the Japanese and Korean primary sources. https://academic.oup.com/asj/article/46/Supplement_1/S1/8512048
156. Ministry of Commerce of the People's Republic of China, China WTO/FTA Consultation Network. [Analysis of Chinese regulatory status of exosome products.] March 2026. http://chinawto.mofcom.gov.cn/article/jsbl/zszc/202603/20260303621897.shtml
157. Republic of Korea Ministry of Food and Drug Safety. Guideline on quality, non-clinical assessment of extracellular vesicles therapy products. Guidance 안내서-0917-03, current version 26 September 2024 (previous version 17 July 2023), Cellular & Gene Therapy Products Division. https://www.mfds.go.kr/brd/m_1060/view.do?seq=15514
158. Erciyes University Genome and Stem Cell Center (GENKÖK). "A first in Türkiye": TİTCK manufacturing-site licence for clinical-grade umbilical-cord MSC-derived exosome production. University announcement, 24 July 2026; press coverage September 2026. (Manufacturing-site licence; not a product marketing authorisation.)
159. Regulatory, ethical, and clinical barriers to exosome use in interventional pain medicine. 2026. ScienceDirect article S2772594426000117. **Citation incomplete: the title and year are as given in the underlying research, but no journal name, volume, DOI, PMID or author list was captured and none could be resolved from the notes. It is load-bearing for the US section 351 versus 361 HCT/P rationale in Section 11.2 and for the grey-market economics in Section 12.6.** https://www.sciencedirect.com/science/article/pii/S2772594426000117
160. CIRS Group. Panoramic analysis: current regulatory status and compliance challenges of exosome cosmetics worldwide. (Regulatory consultancy source, flagged; used only for jurisdiction-level cosmetic rules.) https://www.cirs-group.com/en/cosmetics/panoramic-analysis-current-regulatory-status-and-compliance-challenges-of-exosome-cosmetics-worldwide
161. European Medicines Agency, Committee for Advanced Therapies. Reflection paper on classification of advanced therapy medicinal products. EMA/CAT/600280/2010 rev.1, 21 May 2015. (Identified; not opened in the underlying research.) https://www.ema.europa.eu/en/documents/scientific-guideline/reflection-paper-classification-advanced-therapy-medicinal-products_en.pdf-0
162. Yoon J, Lee S, Kim MJ, Kim J-H. [Comparative regulation of advanced therapy medicinal products across jurisdictions.] Front Pharmacol. 2025;15:1486812. doi:10.3389/fphar.2024.1486812. https://www.frontiersin.org/journals/pharmacology/articles/10.3389/fphar.2024.1486812/full
163. European Medicines Agency. Guideline on quality, non-clinical and clinical requirements for investigational advanced therapy medicinal products in clinical trials. EMA/CAT; 2025. https://www.ema.europa.eu/
164. Regulation (EC) No 1223/2009 on cosmetic products, Annex II (prohibited substances; cells, tissues or products of human origin). (Entry number taken from secondary sources; not verified against the Official Journal.) https://www.legislation.gov.uk/eur/2009/1223/annex/II
165. Limongi, et al. Regulatory challenges and opportunities for cell-derived extracellular vesicles in pharmaceutical development: a European and global perspective. J Extracell Vesicles. 2026. doi:10.1002/jev2.70332. PMID 42373536. (Not retrievable; listed as a priority source.)
166. The Regen Report. Inside Japan's regenerative medicine laws: JSRM explains. Interview, 17 October 2025. (Trade source, flagged.) https://theregenreport.com/2025/10/17/inside-japans-regenerative-medicine-laws-jsrm-explains/
167. Pharmaceuticals and Medical Devices Agency (PMDA), Japan. New modality: extracellular vesicle products — quality considerations for first-in-human clinical trial notifications. PMDA web page, 2026. https://www.pmda.go.jp/english/
168. [Bibliometric analysis of 355 human-derived extracellular vesicle clinical trials registered on ClinicalTrials.gov, 2010–2025.] Clin Exp Med. 2026. doi:10.1007/s10238-026-02124-4. https://link.springer.com/article/10.1007/s10238-026-02124-4
169. Na JH, et al. Quality evaluation considerations for stem cell-derived extracellular vesicles-based therapeutic products in China. J Extracell Vesicles. 2026. doi:10.1002/jev2.70297. (Not retrievable; listed as a priority source.)
170. Türkiye İlaç ve Tıbbi Cihaz Kurumu (TİTCK). İleri Tedavi Tıbbi Ürünler Ruhsatlandırma Yönetmeliği (Advanced Therapy Medicinal Products Licensing Regulation), May 2023; earlier ATMP draft guideline 8 November 2018. (The TİTCK page could not be opened; text consulted via a secondary reproduction.) https://www.titck.gov.tr/mevzuat/ileri-tedavi-tibbi-urunler-ruhsatlandirma-yonetmeligi-29052023112150
171. Schaper NC, van Netten JJ, Apelqvist J, et al. IWGDF Practical Guidelines on the prevention and management of diabetes-related foot disease (IWGDF 2023 update). https://iwgdfguidelines.org/wp-content/uploads/2023/07/IWGDF-2023-01-Practical-Guidelines.pdf
172. National Institute for Health and Care Excellence (NICE). Diabetic foot problems: prevention and management. NICE guideline NG19. London: NICE; 2015 (last updated 2019). https://www.nice.org.uk/guidance/ng19
173. Bus SA, Armstrong DG, Crews RT, et al. Guidelines on offloading foot ulcers in persons with diabetes (IWGDF 2023 update). https://iwgdfguidelines.org/wp-content/uploads/2023/07/IWGDF-2023-06-Offloading-Guideline.pdf
174. Löndahl M, Katzman P, Nilsson A, Hammarlund C. Hyperbaric oxygen therapy facilitates healing of chronic foot ulcers in patients with diabetes. Diabetes Care. 2010;33(5):998–1003. PMID 20427683. (Figures verified from a secondary source; per-arm denominators not verified and the control percentage is quoted as 27% elsewhere versus 29% in the source used.)
175. Fedorko L, Bowen JM, Jones W, et al. Hyperbaric oxygen therapy does not reduce indications for amputation in patients with diabetes with nonhealing ulcers of the lower limb: a prospective, double-blind, randomized controlled clinical trial. Diabetes Care. 2016;39(3):392–399. PMID 26740639. **Arm-level results could not be retrieved through eight independent access routes; cited qualitatively only.**
176. Kranke P, Bennett MH, Martyn-St James M, Schnabel A, Debus SE, Weibel S. Hyperbaric oxygen therapy for chronic wounds. Cochrane Database Syst Rev. 2015;(6):CD004123. doi:10.1002/14651858.CD004123.pub4. (Still the current Cochrane position as of September 2026; it predates reference 177.) https://www.cochrane.org/evidence/CD004123_hyperbaric-oxygen-therapy-treating-chronic-wounds
177. Frykberg RG, Franks PJ, Edmonds M, et al. A multinational, multicenter, randomized, double-blinded, placebo-controlled trial to evaluate the efficacy of cyclical topical wound oxygen (TWO2) therapy in the treatment of chronic diabetic foot ulcers: the TWO2 study. Diabetes Care. 2020;43(3):616–624. PMID 31619393. (Figures verified from two secondary sources reproducing the abstract, which agree.)
178. Armstrong DG, Lavery LA; Diabetic Foot Study Consortium. Negative pressure wound therapy after partial diabetic foot amputation: a multicentre, randomised controlled trial. Lancet. 2005;366(9498):1704–1710. doi:10.1016/S0140-6736(05)67695-7. PMID 16291063. NCT00224796. (Abstract verified verbatim from the authors' institutional repository.)
179. Liu Z, Dumville JC, Hinchliffe RJ, Cullum N, Game F, Stubbs N, Sweeting M, Peinemann F. Negative pressure wound therapy for treating foot wounds in people with diabetes mellitus. Cochrane Database Syst Rev. 2018;(10):CD010318. doi:10.1002/14651858.CD010318.pub3. PMID 30328611. https://www.cochrane.org/evidence/CD010318_negative-pressure-wound-therapy-treating-foot-wounds-people-diabetes-mellitus
180. Veves A, Falanga V, Armstrong DG, Sabolinski ML; Apligraf Diabetic Foot Ulcer Study. Graftskin, a human skin equivalent, is effective in the management of noninfected neuropathic diabetic foot ulcers: a prospective randomized multicenter clinical trial. Diabetes Care. 2001;24(2):290–295. (Note: this trial is in Diabetes Care, not Archives of Surgery.) https://diabetesjournals.org/care/article/24/2/290/24044/Graftskin-a-Human-Skin-Equivalent-Is-Effective-in
181. Marston WA, Hanft J, Norwood P, Pollak R; Dermagraft Diabetic Foot Ulcer Study Group. The efficacy and safety of Dermagraft in improving the healing of chronic diabetic foot ulcers: results of a prospective randomized trial. Diabetes Care. 2003;26(6):1701–1705. doi:10.2337/diacare.26.6.1701. PMID 12766097. (Verified from an institutional reproduction of the abstract.)
182. US Food and Drug Administration. Summary of Safety and Effectiveness Data, PMA P000036 (DERMAGRAFT). https://www.accessdata.fda.gov/cdrh_docs/pdf/p000036b.pdf
183. US Food and Drug Administration. REGRANEX (becaplermin) gel 0.01% prescribing information, revision 08/2019, reference ID 4481700, application 103691s5138. https://www.accessdata.fda.gov/drugsatfda_docs/label/2019/103691s5138lbl.pdf
184. European Medicines Agency. Regranex Article-20 EPAR assessment report. https://www.ema.europa.eu/en/documents/variation-report/regranex-h-c-212-a20-33-epar-assessment-report-variation_en.pdf
185. Smith & Nephew. FDA approves removal of boxed warning from REGRANEX gel. 5 December 2018. (Company statement; used for the 2008 addition and November 2018 removal dates and the supporting database-study descriptions.) https://www.smith-nephew.com/en/news/2018/12/05/20181205-fda-approves-removal-of-boxed-warning-from-regranex-gel
186. Guo J, Dardik A, Fang K, Huang R, Gu Y. Meta-analysis on the treatment of diabetic foot ulcers with autologous stem cells. Stem Cell Res Ther. 2017;8(1):228. doi:10.1186/s13287-017-0683-2. https://link.springer.com/article/10.1186/s13287-017-0683-2
187. [Umbilical-cord mesenchymal stem cells in diabetic foot ulcer and peripheral artery disease: 6 randomised trials, 380 patients.] Int J Low Extrem Wounds. 2026. doi:10.1177/15347346261426552. https://journals.sagepub.com/doi/10.1177/15347346261426552
188. [Review of mesenchymal stromal cell product regulatory status, including CDSCO approval of stempeucel/REGENACIP for no-option critical limb ischaemia in India.] Stem Cell Res Ther. 2024. doi:10.1186/s13287-024-03957-0. (The exact year and legal character of the Indian approval could not be verified.) https://link.springer.com/article/10.1186/s13287-024-03957-0
189. Al-Jeabory M, et al. Exosomes and small extracellular vesicles as an alternative to mesenchymal stromal cell therapy. Int J Mol Sci. 2026;27(9):3737. doi:10.3390/ijms27093737. PMID 42123323.
190. Jung HD, et al. Engineered exosomes for diabetic foot ulcers: lessons learned, challenges remain. Mil Med Res. 2026;13(1):100022. doi:10.1016/j.mmr.2026.100022. PMID 42063467.
191. Zhang Y, Wu D, Zhou C, et al. Engineered extracellular vesicles for tissue repair and regeneration. Burns Trauma. 2024;12:tkae062. doi:10.1093/burnst/tkae062. https://academic.oup.com/burnstrauma/article/doi/10.1093/burnst/tkae062/7831355
192. Ye H, Wang F, Xu G, Shu F, Fan K, Wang D. Advancements in engineered exosomes for wound repair. Front Bioeng Biotechnol. 2023;11:1301362. doi:10.3389/fbioe.2023.1301362. https://www.frontiersin.org/journals/bioengineering-and-biotechnology/articles/10.3389/fbioe.2023.1301362/full
193. Webber J, Clayton A. How pure are your vesicles? J Extracell Vesicles. 2013;2. doi:10.3402/jev.v2i0.19861. PMID 24009896. (Origin of the particle-to-protein purity metric.) https://pubmed.ncbi.nlm.nih.gov/24009896/
194. Mizenko, et al. Critical systematic review of registered extracellular vesicle clinical trials. J Extracell Vesicles. 2024. doi:10.1002/jev2.12510. (Full author list and exact title not verified in the underlying research.) https://escholarship.org/content/qt3kv8w2gb/qt3kv8w2gb.pdf
195. US Food and Drug Administration (CDER/CBER/CDRH). Guidance for Industry: Chronic Cutaneous Ulcer and Burn Wounds — Developing Products for Treatment. June 2006. https://www.fda.gov/media/71278/download
196. Duong A, Parmar G, Kirkham AM, Burger D, Allan DS. Registered clinical trials investigating treatment with cell-derived extracellular vesicles: a scoping review. Cytotherapy. 2023;25(9):939–945. doi:10.1016/j.jcyt.2023.04.007. (Searched ClinicalTrials.gov, the WHO ICTRP and ChiCTR on 10 June 2022; 591 registrations retrieved, 73 included; no wound category and no non-ClinicalTrials.gov identifiers reported.) https://www.sciencedirect.com/science/article/abs/pii/S1465324923001020
197. Yang Z, Yang M, Rui S, et al. Exosome-based cell therapy for diabetic foot ulcers: present and prospect. Heliyon. 2024;10(20):e39251. doi:10.1016/j.heliyon.2024.e39251. https://www.sciencedirect.com/science/article/pii/S2405844024152826
198. Aegle Therapeutics company news index (designations, financings and milestones; most recent item 15 July 2025). Accessed 22 September 2026. (Company source.) https://aegletherapeutics.com/news/
199. Proposed addendum to the 2006 US Food and Drug Administration guidance on chronic cutaneous ulcers and burn wounds: recommendations from the Wound Care Collaborative Community. PMID 42308122. (Content not retrieved; the single most valuable un-fetched source for endpoint recommendations.) https://pubmed.ncbi.nlm.nih.gov/42308122/
200. Driver VR, et al. Identification and content validation of wound therapy clinical endpoints relevant to clinical practice and patient values for FDA approval. Part 1. Survey of the wound care community. Wound Repair Regen. 2017. doi:10.1111/wrr.12533. Part 2 (literature survey): doi:10.1111/wrr.12676, PMID 30315716. Part 3 (patient survey), Gould LJ, et al.: doi:10.1111/wrr.12872.
201. Lener T, Gimona M, Aigner L, et al. Applying extracellular vesicles based therapeutics in clinical trials — an ISEV position paper. J Extracell Vesicles. 2015;4:30087. doi:10.3402/jev.v4.30087. PMID 26725829. (Content not accessed.)
202. Witwer KW, van Balkom BWM, Bruno S, et al. Defining mesenchymal stromal cell (MSC)-derived small extracellular vesicles for therapeutic applications. J Extracell Vesicles. 2019;8(1):1609206. doi:10.1080/20013078.2019.1609206. PMID 31069028. (Content not accessed.)
203. Akbar N, et al. Updating MISEV: when, what, how, and why? J Extracell Vesicles. 2026;15(2):e70230. PMID 41608852.
204. Poupardin R, et al. EV-Checklist: AI-powered rapid documentation for enhancing transparency and accessibility of extracellular vesicle research data. J Extracell Vesicles. 2026;15(7):e70343. PMID 42442940.
205. Yoshioka Y, et al. Regulatory science of extracellular vesicles: from ISEV standards to translational implementation. J Extracell Vesicles. 2026;15(8):e70358. PMID 42605958.
206. Preußer C, et al. Isolation defines identity: functional consequences of extracellular vesicle purification strategies. Adv Healthc Mater. 2026;15(13):e04684. PMID 41555698.
207. Greer N, Foman N, Dorrian J, et al. Advanced wound care therapies for non-healing diabetic, venous, and arterial ulcers: a systematic review. Washington (DC): Department of Veterans Affairs (US); 2012. NCBI Bookshelf NBK132247. https://www.ncbi.nlm.nih.gov/books/NBK132247/