Cellular, Acellular, and Matrix‐Like Placental Products for Diabetic Foot Ulcers: A 12‐Week Randomised‐Trial Meta‐Analysis

ABSTRACT

Placental‐derived cellular, acellular, and matrix‐like products (CAMPs) are widely used as adjuncts to standard care for diabetic foot ulcers (DFUs), but trials differ in product class, comparator, and healing window. We estimated the effect of adjunctive placental‐derived CAMPs on complete wound closure at a single, clinically meaningful 12‐week endpoint. The review was registered in the International Prospective Register of Systematic Reviews (PROSPERO; CRD420261399700) and reported per Preferred Reporting Items for Systematic Reviews and Meta‐Analyses (PRISMA) 2020, with search reporting guided by PRISMA‐S. PubMed/MEDLINE, Scopus, and the Cochrane Central Register of Controlled Trials (CENTRAL) were searched through 21 June 2026 for randomised or quasi‐randomised trials comparing a placental, amniotic, chorionic, amnion‐chorion, placental membrane, or umbilical cord‐derived product plus standard care against standard care or a non‐placental control in adults with active DFUs. The primary outcome was complete wound closure at 12 weeks. Risk ratios (RRs) were pooled in a random‐effects model (inverse‐variance, restricted maximum likelihood [REML], Hartung‐Knapp). Risk of bias was assessed with the Cochrane Risk of Bias 2 tool (RoB 2) for the 12‐week outcome and certainty with the Grading of Recommendations Assessment, Development and Evaluation (GRADE) approach. Prespecified sensitivity analyses included odds ratios, leave‐one‐out, product‐class subgroups, and exclusion of high‐risk‐of‐bias studies. Of 474 records, 14 randomised articles were retained; 9 standard‐care‐controlled comparisons (944 participants) provided extractable 12‐week complete‐closure data for the primary analysis. Complete closure occurred in 298 of 505 product‐treated participants (59.0%) versus 157 of 439 controls (35.8%). Adjunctive CAMPs increased the likelihood of closure (RR 1.70, 95% confidence interval [CI] 1.31–2.20). Heterogeneity was moderate (I² = 51.9%) and the prediction interval crossed the null (0.93–3.11). The effect was robust to leave‐one‐out analysis (RR 1.59–1.81), an odds‐ratio model (odds ratio [OR] 3.36, 95% CI 1.68–6.71), and exclusion of the single high‐risk‐of‐bias trial (RR 1.73, 95% CI 1.29–2.33). Certainty was moderate, downgraded once for inconsistency. Adjunctive placental‐derived CAMPs improve 12‐week complete closure of DFUs compared with standard care, with moderate certainty. The wide prediction interval indicates that the average benefit may not be reproduced by every product, population, or setting, underscoring the need for product‐specific, standard‐care‐anchored trials and for coverage policy grounded in endpoint‐specific randomised evidence.

Trial Registration: Registration number: CRD420261399700.

Key Points

1. This review focused on complete diabetic foot ulcer closure at a single 12‐week endpoint across placental‐derived product classes.

2. Nine standard‐care‐controlled randomised comparisons included 944 participants; closure occurred in 59.0% of product‐treated participants versus 35.8% of controls (pooled RR 1.70, 95% CI 1.31–2.20).

3. Evidence certainty was moderate. The prediction interval (0.93–3.11) crossed the null, so the pooled average should not be assumed for every product or setting.

4. Product‐specific, standard‐care‐controlled trials are needed to clarify which patients and products benefit most.

Abbreviations

CAMP

cellular, acellular and matrix‐like product

CAMPs

cellular, acellular, and matrix‐like products

CENTRAL

Cochrane Central Register of Controlled Trials

CI

confidence interval

cUC

cryopreserved umbilical cord

dACM

dehydrated amnion chorion membrane

DFU

diabetic foot ulcer

DFUs

diabetic foot ulcers

dHACA/DHACA

dehydrated human amnion and chorion allograft

dHACM

dehydrated human amnion/chorion membrane

dHAICM

dehydrated human amnion‐intermediate layer‐chorion membrane

dHAM

dried human amniotic membrane

FT‐DPM

full‐thickness decellularised placental membrane

GRADE

Grading of Recommendations Assessment, Development and Evaluation

HACM

human amnion/chorion membrane

HAM

human amniotic membrane

HSAM

hypothermically stored amniotic membrane

I²

I‐squared

ITT

intention‐to‐treat

k

number of studies or comparisons

MEDLINE

Medical Literature Analysis and Retrieval System Online

mITT

modified intention‐to‐treat

n

number

NHS

National Health Service

OR

odds ratio

p

p value

PICO

population, intervention, comparator, and outcome

PRISMA

Preferred Reporting Items for Systematic Reviews and Meta‐Analyses

PROSPERO

International Prospective Register of Systematic Reviews

RCT

randomised controlled trial

RCTs

randomised controlled trials

REML

restricted maximum likelihood

RISMA‐S

Preferred Reporting Items for Systematic Reviews and Meta‐Analyses literature search extension

RoB 2

Cochrane Risk of Bias 2 tool

RR

risk ratio

SOC

standard of care

SWC

standard wound care

tau²

tau‐squared

UK

United Kingdom

USA

United States of America

1. Introduction

Diabetic foot ulcers (DFUs) precede most nontraumatic lower‐limb amputations and carry five‐year mortality that rivals or exceeds many common cancers [1, 2, 3]. Standard care, comprising offloading, debridement, infection management, and restoration of perfusion, heals many ulcers, but a substantial fraction stall, and these are the wounds that drive infection, hospitalisation, and limb loss [4, 5]. Placental‐derived tissue products, encompassing amniotic, chorionic, amnion‐chorion, placental membrane, and umbilical cord‐derived allografts, have been developed to restart healing in such wounds by providing extracellular matrix scaffolding, growth factors, and, in some products, viable cells [6]. In current United States wound‐care nomenclature, these are grouped, alongside non‐placental scaffolds, as cellular, acellular, and matrix‐like products (CAMPs) [7]. Multiple randomised controlled trials (RCTs) [8, 9] and several meta‐analyses [10, 11, 12, 13, 14, 15, 16] suggest that placental‐derived products improve DFU healing, but existing syntheses have pooled heterogeneous healing windows of 4, 6, 12, and 16 weeks, mixed complete with partial healing, and often focused on narrow product subsets, limiting their usefulness at the bedside and in coverage decisions. We therefore undertook a systematic review and meta‐analysis restricted to a single, clinically anchored endpoint, complete wound closure at 12 weeks, across the full breadth of placental‐derived CAMPs studied in randomised standard‐care‐controlled trials, with outcome‐specific risk‐of‐bias assessment using the Cochrane Risk of Bias 2 tool (RoB 2) [17] and certainty assessment using the Grading of Recommendations Assessment, Development and Evaluation (GRADE) [18] approach.

2. Methods

2.1. Protocol Registration and Reporting

The protocol for this systematic review and meta‐analysis was registered in the International Prospective Register of Systematic Reviews (PROSPERO) on 19 May 2026 under the title Placental and amniotic products for diabetic foot ulcers: a 12‐week randomised‐trial meta‐analysis, registration number CRD420261399700 [19]. The review was reported in line with Preferred Reporting Items for Systematic Reviews and Meta‐Analyses (PRISMA) 2020. Search reporting followed Preferred Reporting Items for Systematic Reviews and Meta‐Analyses literature search extension (PRISMA‐S) principles [20]. For each database, the date of search, search strategy, number of records retrieved, and duplicate handling were recorded. Full search strategies are provided in (Supplementary Table S1).

2.2. Review Question and PICO

This review examined whether placental, amniotic, chorionic, amnion‐chorion, placental membrane, or umbilical cord‐derived products improve healing in adults with active diabetic foot ulcers (DFUs) when used as an adjunct to standard care. The population, intervention, comparator, and outcome (PICO) was defined before study selection. The population was adults with active diabetic foot ulcers. The intervention was a topical placental, amniotic, chorionic, amnion‐chorion, placental membrane, or umbilical cord‐derived product. The comparator was standard care, usual care, placebo or sham dressing, or another control treatment not involving a placental, amniotic, chorionic, or umbilical cord‐derived product. The primary outcome was complete wound closure at 12 weeks. The 12‐week endpoint was used as the main rule for the quantitative analysis. Studies were included in the primary pooled analysis when complete wound closure at 12 weeks was reported or could be extracted in usable binary form from a randomised or quasi‐randomised comparison against standard care or another non‐placental control. Where more than one analysis population was reported, intention‐to‐treat or modified intention‐to‐treat data were preferred over per‐protocol data. Per‐protocol data were retained for sensitivity analysis or narrative comparison when needed.

2.3. Eligibility Criteria

• −Randomised and quasi‐randomised clinical trials were eligible if they included adults with active diabetic foot ulcers and compared an eligible placental, amniotic, chorionic, amnion‐chorion, placental membrane, or umbilical cord‐derived product against standard care, usual care, placebo or sham dressing, or another non‐placental control treatment.
• −Eligible products included dehydrated human amnion/chorion membrane, amniotic membrane products, chorionic membrane products, placental membrane allografts, cryopreserved placental membrane, lyopreserved or dehydrated placental/amniotic products, full‐thickness decellularised placental membrane products, and umbilical cord‐derived tissue products.
• −Studies were excluded from the primary meta‐analysis if they were not focused on active diabetic foot ulcers, evaluated only non‐diabetic wounds, venous leg ulcers, pressure ulcers, surgical wounds, or recurrence prevention without active ulcer treatment, or if diabetic foot ulcer data could not be separated from a mixed wound population. Animal studies, laboratory studies, preclinical studies, in vitro studies, case reports, case series, retrospective studies, prospective single‐arm studies, narrative reviews, editorials, letters, protocols, and conference abstracts without sufficient usable outcome data were excluded from the quantitative synthesis.
• −Studies evaluating non‐placental advanced wound products alone were excluded from the main synthesis. Trials comparing two active placental or amniotic products, two preservation methods, or two application schedules without a standard‐care or non‐placental control arm were not included in the primary meta‐analysis. If relevant, these studies were retained for descriptive reporting.

2.4. Search Strategy

The final database searches were completed on 21 June 2026, after PROSPERO registration. Three databases were searched: PubMed/MEDLINE, Scopus, and the Cochrane Central Register of Controlled Trials (CENTRAL). The search was built around three concepts: diabetic foot ulcer, placental/amniotic/chorionic/umbilical cord‐derived products, and randomised or controlled clinical evidence. The exact syntax was adapted to each database. Web of Science was listed in the protocol as a source to be searched if available. It was not included in the final database search because no functional search interface was available on the platform at the time of searching. This protocol deviation was documented. Backward and forward citation checking of included trials and relevant prior reviews was used as an additional safety check.

2.5. Search Results and Study Selection

Across PubMed/MEDLINE, Scopus, and CENTRAL, 474 records were retrieved before manual duplicate checking and eligibility assessment. The PubMed search retrieved 100 records, the Scopus search retrieved 258 records, and the CENTRAL search retrieved 116 records. After title and abstract screening, full‐text assessment, and manual deduplication, 14 unique randomised articles were retained for this review. Of these, 9 studies provided standard‐care‐controlled, extractable complete wound closure data at 12 weeks, or sufficiently close review‐relevant 12‐week binary data, and were included in the primary meta‐analysis. The remaining 5 studies were retained for sensitivity or descriptive analysis because they did not meet all criteria for the primary 12‐week pooled analysis. Records were compared manually by title, authors, year, journal, product, comparator, and study identity. PubMed was screened first and formed the initial working set. Scopus was then checked against PubMed. CENTRAL was checked last against both Scopus and PubMed. Titles, abstracts, and full texts were screened against the prespecified eligibility criteria. GT performed the initial screening and eligibility assessment. DGA checked uncertain records, proposed exclusions, and final inclusion decisions. Any discrepancy was discussed between GT and DGA and resolved by consensus.

2.6. Data Extraction

For each included study, the extracted data included author and year, country, study design, sample size, product name, product class, comparator, number of wounds healed at 12 weeks in the intervention group, number analysed in the intervention group, number of wounds healed at 12 weeks in the control group, number analysed in the control group, analysis population, funding or sponsorship, and notes on whether the data were exact, reconstructed, pilot, per‐protocol, derived, or descriptive only. GT extracted the study data. DGA checked the extracted data for consistency and clinical plausibility. Any unclear point was discussed and resolved by consensus before analysis. Where a study reported percentages without explicit event counts, were reconstructed only when the denominator was clear. Reconstructed or derived data were kept identifiable and were not used in the primary analysis unless considered sufficiently robust for the prespecified 12‐week binary outcome. For studies reporting more than one analysis population, intention‐to‐treat or modified intention‐to‐treat data were preferred. Reyzelman et al. [21], was included in the primary analysis using the modified intention‐to‐treat population. Per‐protocol data from that study were retained for interpretation but were not used as the primary extraction.

2.7. Risk of Bias Assessment

Risk of bias was assessed for the primary outcome of complete wound closure at 12 weeks. The Cochrane Risk of Bias 2 tool (RoB 2) for randomised trials was used [17]. The assessment considered bias arising from the randomisation process, bias due to deviations from the intended interventions, bias due to missing outcome data, bias in measurement of the outcome, and bias in selection of the reported result. Each study was judged for the 12‐week closure outcome, not for the paper as a whole. This was important because some studies reported several outcomes and time points, while the main question of this review was complete wound closure at 12 weeks. GT performed the initial RoB 2 assessments. DGA reviewed the risk‐of‐bias judgements for consistency and clinical plausibility. Any disagreement or uncertainty was resolved by discussion and consensus. The final judgement for each study was recorded as low risk of bias, some concerns, or high risk of bias. Studies judged to be at high risk of bias were not automatically removed from the main analysis, but their influence was examined in sensitivity analysis where possible.

2.8. Certainty of Evidence

The certainty of evidence was assessed using Grading of Recommendations Assessment, Development and Evaluation (GRADE) for the primary outcome of complete wound closure at 12 weeks [18]. The certainty assessment considered risk of bias, inconsistency, indirectness, imprecision, and publication bias. Evidence from randomised trials started as high certainty and was rated down when there were important concerns. Risk of bias was informed by the RoB 2 assessment. Although one primary‐analysis study was judged to be at high risk of bias, its influence was examined using leave‐one‐out analysis and a high‐risk‐of‐bias‐excluded sensitivity analysis. Inconsistency was judged from the direction and size of study effects, I‐squared, tau‐squared, the prediction interval, and visual inspection of the forest plot. Indirectness was considered when the population, intervention, comparator, or outcome did not closely match the review question. Imprecision was judged from the confidence interval and the amount of available information. Publication bias was considered from the pattern of available studies, study size, sponsorship pattern, and funnel plot inspection where appropriate. Because fewer than 10 trials were included in the primary analysis, formal small‐study‐effect testing was not performed. The final certainty of evidence was reported as high, moderate, low, or very low.

2.9. Statistical Analysis

The primary effect measure was the risk ratio for complete wound closure at 12 weeks. Risk ratios with 95% confidence intervals were calculated for each study. The primary meta‐analysis included randomised or quasi‐randomised comparisons with a standard‐care or non‐placental control arm and extractable complete wound closure counts at 12 weeks. Where more than one analysis population was available, intention‐to‐treat or modified intention‐to‐treat data were preferred over per‐protocol data.

The primary analysis was performed in R using Posit Cloud and the meta package [22]. Random‐effects meta‐analysis was performed using the inverse‐variance method on log risk ratios, with restricted maximum‐likelihood estimation of tau‐squared and Hartung‐Knapp adjustment for the random‐effects confidence interval. Heterogeneity was assessed using I‐squared, tau‐squared, Cochran Q, the prediction interval, and visual inspection of the forest plot. The primary pooled analysis included 9 studies [8, 9, 21, 23, 24, 25, 26, 27, 28]. Reyzelman et al. [21] was included in the primary analysis using the modified intention‐to‐treat population, but was judged to be at high risk of bias because of the allocation process. Its influence was therefore examined using leave‐one‐out analysis and a high‐risk‐of‐bias‐excluded sensitivity analysis. Five additional randomised studies [29, 30, 31, 32, 33] were retained outside the primary pooled analysis for sensitivity or descriptive reporting because they had earlier primary healing assessment but follow‐up to 12 weeks, active‐comparator designs without a standard‐care‐only arm, time‐to‐event or model‐estimated results without raw 12‐week event counts, derived subgroup data, or no clean 12‐week binary endpoint. Prespecified sensitivity and exploratory analyses included use of odds ratios instead of risk ratios, leave‐one‐out analysis, product‐class subgroup analysis, and sensitivity analyses excluding studies judged to be at high risk of bias. Because Game et al. [27] reported 16 usual‐care participants at randomisation but elsewhere described an analysed usual‐care population of 15, an additional denominator sensitivity analysis repeated the primary model using 1/15 rather than 1/16 for the control arm. Product‐class subgroups included amnion‐chorion products, cryopreserved or viable placental/amniotic membrane products, amnion‐only products, umbilical cord‐derived products, and full‐thickness decellularised placental membrane products. Sponsorship sensitivity analysis was planned because many trials were industry‐funded or industry‐linked. Where too few non‐industry studies were available for a meaningful pooled sensitivity analysis, this was reported descriptively. Publication bias was considered by examining study size, direction of effect, sponsorship pattern, and funnel plot appearance where appropriate. Formal small‐study‐effect testing, including Egger or Begg tests, was not performed because the primary analysis included fewer than 10 studies.

3. Results

3.1. Study Selection

The search identified 474 records across PubMed/MEDLINE, Scopus, and CENTRAL, including 100 records from PubMed/MEDLINE, 258 from Scopus, and 116 from CENTRAL. Records were screened manually using the planned source‐by‐source workflow: PubMed/MEDLINE was screened first, Scopus was then checked against PubMed/MEDLINE, and CENTRAL was checked last against both previous sources. After title and abstract screening, 448 records were excluded. A total of 26 full‐text articles were assessed or checked for eligibility, and 12 were excluded. The main reasons for exclusion were absence of usable 12‐week or review‐relevant binary closure data, active‐comparator or dosing‐comparison designs that did not answer the primary standard‐care‐controlled question, duplicate or overlapping reporting, and inability to retrieve the full text. A total of 14 randomised articles were retained in the review. Of these, 9 studies contributed data to the primary 12‐week meta‐analysis and 5 were retained for sensitivity or descriptive evidence. The full selection process is shown in Figure 1.

FIGURE 1.

PRISMA.

3.2. Characteristics of Included Studies

The included randomised studies evaluated a range of placental, amniotic, chorionic, amnion‐chorion, placental membrane, and umbilical cord‐derived products used as adjuncts to standard diabetic foot ulcer care [8, 9, 21, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33]. The primary meta‐analysis was deliberately restricted to studies with a standard‐care or non‐placental control arm and extractable complete wound closure data at 12 weeks [8, 9, 21, 23, 24, 25, 26, 27, 28]. This produced a clinically focused dataset rather than a broad pooling of all placental or amniotic wound product trials. The evidence base was not evenly distributed across product classes. Several studies evaluated dehydrated human amnion/chorion or related amnion‐chorion products [9, 23, 24, 31, 32], while fewer studies assessed amnion‐only products [27, 29, 30, 33], cryopreserved or viable placental membrane [8], hypothermically stored amniotic membrane [26], full‐thickness decellularised placental membrane [21], or umbilical cord‐derived products [25, 28]. The remaining randomised studies were informative but did not fit the primary 12‐week standard‐care‐controlled extraction rule and were therefore kept outside the main pooled estimate [29, 30, 31, 32, 33]. Study characteristics, product types, comparators, and analysis notes are summarised in Table 1.

TABLE 1.

Characteristics and main findings of included randomised studies.

Study	Design and setting	Participants	Intervention and comparator	Outcome window and main findings	Funding and conflicts reported
Zelen et al. 2013 [29] USA	Prospective randomised single‐centre trial	25 patients: EpiFix, n = 13; standard care, n = 12	EpiFix dehydrated human amniotic membrane allograft plus standard care compared with standard care alone	The prespecified healing assessments were at 4 and 6 weeks, but patients were followed weekly for up to 12 weeks or until complete healing. At 6 weeks, 12 of 13 ulcers healed with EpiFix compared with 1 of 12 with standard care.	No funding statement was identified in the article text reviewed.
Lavery et al. 2014 [8] USA	Multicentre controlled randomised blinded trial	97 patients: Grafix, n = 50; control, n = 47	Grafix human viable wound matrix compared with standard wound care	Complete wound closure by 12 weeks occurred in 31 of 50 patients with Grafix and 10 of 47 controls. Median time to healing was 42 days with Grafix and 69.5 days with control.	Funded by Osiris Therapeutics Inc. The authors received research funding from Osiris for participation in the study.
Zelen et al. 2016 USA [23]	Prospective randomised controlled multicentre comparative study	100 patients: Apligraf, n = 33; EpiFix, n = 32; standard wound care, n = 35	EpiFix dehydrated human amnion/chorion membrane and Apligraf were compared with standard wound care	Earlier interim findings had focused on 4 to 6 weeks, but the larger completed cohort reported 12‐week outcomes. Within 12 weeks, complete closure occurred in 31 of 32 EpiFix‐treated ulcers, 24 of 33 Apligraf‐treated ulcers, and 18 of 35 standard‐care ulcers.	A clear funding statement was not identified in the article text reviewed.
DiDomenico et al. 2018 [24] USA	Prospective randomised multicentre clinical trial	80 patients: dHACA, n = 40; standard care, n = 40	Aseptically processed dehydrated human amnion and chorion allograft plus standard care compared with standard care alone	Complete wound closure at 12 weeks occurred in 34 of 40 ulcers with dHACA and 13 of 40 ulcers with standard care. Mean time to healing was 37 days and 67 days, respectively.	Funded by Musculoskeletal Transplant Foundation, grant 002.
Tettelbach et al. 2019 dHACM [9] USA	Prospective multicentre randomised controlled trial in 14 wound clinics	110 patients in the intent‐to‐treat analysis: dHACM, n = 54; standard care, n = 56	Weekly dehydrated human amnion/chorion membrane plus standard care compared with standard care using alginate dressings and offloading	Complete healing within 12 weeks occurred in 70% of the dHACM group and 50% of the standard‐care group in the intent‐to‐treat analysis.	Funded by MiMedx Group Inc. The article states that the study was sponsored by MiMedx and that investigators and adjudicators received research funding.
Tettelbach et al. 2019, EpiCord [25] USA	Prospective multicentre randomised controlled trial in 11 US centres	155 patients in the intent‐to‐treat analysis: EpiCord, n = 101; alginate control, n = 54	EpiCord dehydrated human umbilical cord allograft plus standard care compared with standardised alginate dressing care	Complete healing within 12 weeks occurred in 71 of 101 EpiCord‐treated ulcers and 26 of 54 control ulcers.	Funded by MiMedx Group Inc.
Serena et al. 2020 [26] USA	Multicentre prospective randomised controlled trial in 14 centres	76 patients: HSAM, n = 38; standard care, n = 38	Hypothermically stored amniotic membrane plus standard care compared with standard care alone	The study included a 12‐week treatment phase and a 4‐week follow‐up phase. Unadjusted closure at 12 weeks was 55% with HSAM and 29% with standard care. Cox‐adjusted closure at 12 weeks was 60% and 38%, respectively.	Funded by Organogenesis Inc.
Game et al. 2021 [27] UK	UK multicentre patient‐ and assessor‐blinded randomised pilot trial	31 randomised patients: dHAM, n = 15; usual care, n = 16	Omnigen dried human amniotic membrane added to standard care every 2 weeks compared with standard care alone	Healing was assessed within 12 weeks. Four of 15 ulcers healed with dHAM compared with 1 of 16 in the usual‐care group. Wound area reduction was greater in the dHAM group.	Sponsored by University Hospitals of Derby and Burton NHS Foundation Trust. Funding was reported from East Midlands Academic Health Science Network, Medilink East Midlands Ltd., and NuVision Biotherapies Ltd.
Thompson et al. 2019 [30] USA	Prospective randomised study in a Midwestern wound care clinic	13 randomised patients	Total contact cast plus human amniotic allograft compared with total contact cast plus standard wound care	The study reported time to closure and 90‐day recurrence rather than a clean 12‐week closure endpoint. Most participants healed during the study. Mean time to closure was 29.50 days with allograft and 26.20 days with standard wound care. 90‐day recurrence was 14.29% and 83.33%, respectively.	Supported in part by Derma Sciences Inc., which provided allografts and some instant total contact casts. The authors disclosed no other financial relationships.
Cazzell et al. 2024 [31] USA	Prospective multicentre randomised controlled trial	218 patients: dACM plus standard care, n = 109; standard care, n = 109	NuShield dehydrated amnion chorion membrane plus standard care compared with standard care alone	The primary endpoint was time to and frequency of complete wound closure by or at 12 weeks. Cox‐estimated wound closure at week 12 was 50% with dACM plus standard care and 35% with standard care alone. Median time to closure was 84 days with dACM and was not achieved in the standard‐care group.	Funded by Organogenesis Inc. Several authors reported consultant or speaker roles for Organogenesis.
Reyzelman et al. 2025 [21] USA	Prospective multicentre randomised controlled trial	108 patients in the modified intention‐to‐treat analysis: FT‐DPM, n = 57; standard care, n = 51	Full‐thickness decellularised placental membrane plus standard care compared with standard care alone	The study assessed closure and wound area reduction through 12 weeks, with additional recurrence follow‐up among healed patients. In the modified intention‐to‐treat analysis used in the present meta‐analysis, complete closure at 12 weeks occurred in 22 of 57 ulcers (38.6%) with FT‐DPM and 13 of 51 (25.5%) with standard care. In the per‐protocol analysis, 12‐week closure was 48% with FT‐DPM and 27% with standard care. Percentage area reduction was 79% and 56%, respectively.	Partially funded by LifeNet Health. One author served as a key opinion leader for LifeNet Health.
Caporusso et al. 2025 [28] USA	Multicentre randomised controlled trial in complex diabetic foot ulcers	220 patients: cUC plus standard care, n = 118; standard care, n = 102	Cryopreserved ultra‐thick human amniotic membrane derived from umbilical cord plus standard care compared with standard care alone	At 12 weeks, complete closure was 39.0% with cUC, 46 patients, and 36.3% with SOC, 37 patients, p = 0.68. Healing increased to 49.2% versus 47.1% at 16 weeks, 58 versus 48 patients, and to 77.1% versus 71.6% at 50 weeks, 91 versus 73 patients. No significant difference was found at any time point.	Supported by Bio Tissue Holdings Inc. The authors declared no conflicts of interest.
Hall et al. 2025 [32] USA	Interim analysis from two prospective multicentre randomised studies	53 analysed patients: dHAICM plus standard care, n = 28; standard care, n = 25	AmchoPlast dehydrated human amnion‐intermediate layer‐chorion membrane plus standard care compared with standard care alone	Complete closure at 12 weeks occurred in 2 of 11 patients, 18.2%, with dHAICM plus SOC and 1 of 12 patients, 8.3%, with SOC alone; risk ratio 2.18, 95% CI 0.23 to 20.8; Fisher exact p = 0.59.	The studies were sponsored by Cellution Biologics Inc.
Lavery et al. 2025 [33] USA	Randomised clinical trial comparing treatment frequency	40 patients: weekly AmnioExcel, n = 20; biweekly AmnioExcel, n = 20	AmnioExcel human amniotic allograft applied weekly compared with AmnioExcel applied every other week	Patients were followed over a 12‐week evaluation period. Healing did not differ significantly between groups. The reported healing in 30.0% of the weekly group and 50.0% of the biweekly group, p = 0.20; Reported healed ulcers as 5 of 20 and 10 of 20, respectively. Time to healing was 69.3 days with weekly application and 45.8 days with biweekly application, p = 0.15. Infection occurred in 35.0% and 25.0%, respectively, p = 0.49.	Funded by a grant from Integra Life Science. The authors declared no potential conflicts of interest.

Abbreviations: cUC, cryopreserved umbilical cord; dACM, dehydrated amnion chorion membrane; dHACA, dehydrated human amnion and chorion allograft; dHACM, dehydrated human amnion/chorion membrane; dHAICM, dehydrated human amnion‐intermediate layer‐chorion membrane; dHAM, dried human amniotic membrane; FT‐DPM, full‐thickness decellularised placental membrane; HSAM, hypothermically stored amniotic membrane.

3.3. Primary 12‐Week Meta‐Analysis

Nine randomised standard‐care‐controlled comparisons contributed extractable complete wound closure data at 12 weeks to the primary risk‐ratio meta‐analysis [8, 9, 21, 23, 24, 25, 26, 27, 28] (Table 2). These studies included 505 participants in the placental‐derived CAMP plus standard‐care arms and 439 participants in the standard‐care or non‐placental control arms. Across the 9 studies, complete closure occurred in 298 of 505 product‐treated participants (59.0%) and 157 of 439 control participants (35.8%).

TABLE 2.

Studies entered into the primary 12‐week risk‐ratio meta‐analysis.

Study/comparison	Product class	Product	Analysis set	Product + SOC healed/total (%)	SOC healed/total (%)	RR (95% CI)	Weight
Lavery 2014 [8]	Cryopreserved/viable placental membrane	Grafix	ITT	31/50 (62.0%)	10/47 (21.3%)	2.91 (1.61–5.26)	8.6%
Zelen 2016 [23]	Amnion‐chorion	EpiFix/dHACM	EpiFix arm versus SWC	31/32 (96.9%)	18/35 (51.4%)	1.88 (1.36–2.61)	15.1%
DiDomenico 2018 [24]	Amnion‐chorion	dHACA	ITT	34/40 (85.0%)	13/40 (32.5%)	2.62 (1.64–4.16)	11.3%
Tettelbach 2019 dHACM [9]	Amnion‐chorion	dHACM	ITT	38/54 (70.4%)	28/56 (50.0%)	1.41 (1.03–1.93)	15.5%
Tettelbach 2019 EpiCord [25]	Umbilical cord‐derived	EpiCord	ITT	71/101 (70.3%)	26/54 (48.1%)	1.46 (1.08–1.98)	15.8%
Serena 2020 [26]	Hypothermically stored amniotic membrane	HSAM	ITT/unadjusted 12‐week data	21/38 (55.3%)	11/38 (28.9%)	1.91 (1.07–3.39)	8.9%
Game 2021 [27]	Amnion‐only	dHAM/Omnigen	Trial‐reported 12‐week outcome	4/15 (26.7%)	1/16 (6.3%)	4.27 (0.54–33.98)	1.1%
Caporusso 2025 [28]	Umbilical cord‐derived	cUC	ITT	46/118 (39.0%)	37/102 (36.3%)	1.07 (0.76–1.51)	14.7%
Reyzelman 2025 [21]	Full‐thickness decellularised placental membrane	FT‐DPM/Matrion	mITT	22/57 (38.6%)	13/51 (25.5%)	1.51 (0.85–2.68)	9.0%
Overall	—	—	Random‐effects model	298/505 (59.0%)	157/439 (35.8%)	1.70 (1.31–2.20)	100.0%

In the primary random‐effects model, adjunctive placental‐derived CAMPs increased the likelihood of complete wound closure at 12 weeks compared with standard care or non‐placental control. The pooled risk ratio was 1.70, with a 95% confidence interval from 1.31 to 2.20 (Figure 2). The point estimates generally favoured the product plus standard‐care groups, although the size of effect varied across studies. Between‐study heterogeneity was moderate (I² = 51.9%, tau² = 0.0560, p = 0.034), and the prediction interval crossed the null (0.93 to 3.11), indicating that the magnitude of benefit may differ across settings, products, and trial populations.

FIGURE 2.

Primary meta‐analysis: Risk Ratio, k = 9.

The primary forest plot also shows why the result should not be interpreted as a uniform effect of every product. Some studies showed large relative increases in closure, particularly among earlier smaller trials, whereas larger and more recent studies tended to produce more modest estimates. Caporusso 2025, for example, contributed a large number of participants and had an effect estimate close to the null [28], while several amnion‐chorion studies showed larger relative effects [9, 23, 24]. This pattern informed the interpretation of heterogeneity and the certainty assessment. A graphical summary of the primary 12‐week result is shown in Figure 3.

FIGURE 3.

Graphical summary of the primary 12‐week randomised‐trial meta‐analysis. CAMPs plus standard care improved complete wound closure at 12 weeks compared with standard care or non‐placental control, although the prediction interval suggests that effects may vary across products and settings.

3.4. Sensitivity Analyses

Sensitivity and exploratory analyses supported the main finding while clarifying that the pooled effect should be interpreted as an average across heterogeneous products rather than as a uniform effect for every product class. In the product‐class subgroup analysis, the overall random‐effects estimate remained RR 1.70 (95% CI 1.31 to 2.20), but the amount and strength of evidence differed across product categories (Figure 4). Amnion‐chorion products contributed the largest subgroup of studies and favoured adjunctive treatment, although heterogeneity was present within this subgroup [9, 23, 24]. Umbilical cord‐derived products showed a more variable pattern, with one study favouring adjunctive treatment and one larger study having an effect estimate close to the null [25, 28]. Single‐study product classes, including cryopreserved/viable placental membrane, hypothermically stored amniotic membrane, amnion‐only products, and full‐thickness decellularised placental membrane, were informative but too sparse for firm class‐specific inference [8, 21, 26, 27]. The test for subgroup differences was not statistically significant, so the subgroup findings should be considered exploratory.

FIGURE 4.

Product‐class subgroup analysis, k = 9.

The primary finding was preserved when the analysis was repeated using odds ratios rather than risk ratios. In the odds‐ratio sensitivity analysis, the pooled effect remained in favour of adjunctive placental‐derived CAMPs (OR 3.36, 95% CI 1.68 to 6.71; Figure 5). Heterogeneity was greater in the odds‐ratio model (I² = 69.3%, tau² = 0.4762), which was expected given the higher sensitivity of odds ratios when baseline risks and event rates vary. The direction and statistical significance of the effect were nevertheless consistent with the primary risk‐ratio analysis.

FIGURE 5.

Sensitivity analysis: Odds ratio, k = 9.

Leave‐one‐out analysis showed that the main inference was not dependent on any single study (Figure 6). After omitting one study at a time, the pooled risk ratio ranged from 1.59 to 1.81, and all leave‐one‐out pooled estimates remained statistically significant. The lowest estimate was observed when DiDomenico 2018 was omitted (RR 1.59, 95% CI 1.23 to 2.05) [24]. The highest estimate was observed when Caporusso 2025 was omitted (RR 1.81, 95% CI 1.43 to 2.28) [28]. Omission of Reyzelman 2025 produced a pooled risk ratio of 1.73 (95% CI 1.29 to 2.33), essentially matching the high‐risk‐of‐bias‐excluded sensitivity analysis [21].

FIGURE 6.

Leave‐one‐out sensitivity analysis.

Because Reyzelman 2025 was judged to be at high risk of bias, its influence was examined separately [21]. Excluding that study left 8 studies in the high‐risk‐of‐bias‐excluded sensitivity analysis. The pooled estimate remained materially unchanged (RR 1.73, 95% CI 1.29 to 2.33; Figure 7). The prediction interval was wider and crossed the null (0.87 to 3.46), and heterogeneity remained moderate (I² = 57.8%, tau² = 0.0695, p = 0.020). This analysis supported the decision to retain Reyzelman 2025 in the primary modified intention‐to‐treat analysis while making its risk‐of‐bias status explicit. Game et al. reported 31 randomised participants (15 dHAM and 16 usual care) and 4/15 versus 1/16 healed at 12 weeks. Because the article also describes one usual‐care participant who received no treatment and an analysed population of 15 versus 15, we repeated the primary analysis using 1/15 for the control denominator. The pooled estimate was essentially unchanged (RR 1.70, 95% CI 1.32–2.19; I² = 51.6%; τ² = 0.0558).

FIGURE 7.

Sensitivity analysis excluding the high‐risk‐of‐bias study, k = 8.

3.5. Risk of Bias

Risk‐of‐bias judgements for the primary outcome are shown in Figure 8. Risk of bias was assessed using the Cochrane Risk of Bias 2 tool (RoB 2) for randomised trials, with judgements made for complete wound closure at 12 weeks rather than for each article as a whole [17]. Among the 9 studies in the primary meta‐analysis [8, 9, 21, 23, 24, 25, 26, 27, 28], 2 studies were judged to be at low overall risk of bias, 6 were judged to have some concerns, and 1 study, Reyzelman 2025, was judged to be at high risk of bias [21]. The high‐risk judgement for Reyzelman 2025 was driven by concerns about the allocation process. The study was retained in the primary analysis using the modified intention‐to‐treat population, but its effect on the pooled estimate was checked directly in leave‐one‐out and high‐risk‐of‐bias‐excluded sensitivity analyses.

FIGURE 8.

Risk of Bias assessment using Cochrane RoB 2.

The most frequent concerns across the primary studies related to deviations from intended interventions, largely reflecting the practical difficulty of blinding wound‐product trials. Missing outcome data contributed to some concerns in selected studies, while measurement of wound closure was generally more secure when blinded or independent assessment methods were used. Among the 5 studies retained for sensitivity or descriptive evidence [29, 30, 31, 32, 33], all were judged to have some concerns overall. These studies were useful for contextual interpretation but were not treated as part of the primary pooled effect.

3.6. Certainty of Evidence

The GRADE summary for the primary outcome is shown in Figure 9. The certainty of evidence for complete wound closure at 12 weeks was judged to be moderate using the Grading of Recommendations Assessment, Development and Evaluation (GRADE) approach [18]. The evidence started as high certainty because all studies contributing to the primary analysis were randomised or quasi‐randomised trials. It was downgraded by one level for inconsistency because heterogeneity was moderate and the prediction interval crossed the null. The certainty was not further downgraded for risk of bias. Although one primary‐analysis study, Reyzelman 2025, was judged to be at high risk of bias [21], both the leave‐one‐out analysis and the high‐risk‐of‐bias‐excluded sensitivity analysis showed that the pooled effect was not driven by that study.

FIGURE 9.

GRADE summary of certainty of evidence.

Indirectness was not considered serious because the included studies addressed adults with active diabetic foot ulcers treated with adjunctive placental, amniotic, chorionic, placental membrane, or umbilical cord‐derived products compared with standard care or a non‐placental control [8, 9, 21, 23, 24, 25, 26, 27, 28]. Imprecision was also not considered serious because the confidence interval around the primary pooled effect did not cross the null and remained compatible with a clinically meaningful benefit. Publication bias was considered possible but was not formally tested because fewer than 10 studies were included in the primary meta‐analysis. Several trials were small or product‐sponsored, and this was considered when interpreting the overall certainty. However, the available evidence consistently favoured adjunctive product treatment, and the robustness analyses did not suggest that the primary conclusion depended on a single small or high‐risk study.

4. Discussion

This systematic review and meta‐analysis restricted its primary inference to a single, clinically anchored question: whether adjunctive placental‐derived cellular, acellular, and matrix‐like products (CAMPs) increase complete wound closure of diabetic foot ulcers at 12 weeks compared with standard care [7]. Across 9 randomised standard‐care‐controlled comparisons and 944 participants [8, 9, 21, 23, 24, 25, 26, 27, 28], the pooled risk ratio (RR) favoured adjunctive product treatment (RR 1.70, 95% confidence interval [CI] 1.31 to 2.20), corresponding to complete closure in 298 of 505 product‐treated ulcers (59.0%) versus 157 of 439 controls (35.8%). The direction of effect was consistent across nearly all studies, and the finding was stable under product‐class subgroup analysis, leave‐one‐out testing, an odds‐ratio model, and exclusion of the single trial judged to be at high risk of bias [21].

Two features temper this headline. First, between‐study heterogeneity was moderate (I² = 51.9%), with larger and more recent trials generally producing smaller relative effects than earlier and smaller trials. Second, and most importantly, the prediction interval crossed the null (0.93 to 3.11). The confidence interval describes the average pooled effect, whereas the prediction interval describes the range of effects that might plausibly be observed in a future trial, product, or clinical setting. The contrast between a statistically significant pooled estimate and a null‐spanning prediction interval is the central message of this review: on average, adjunctive placental‐derived CAMPs improve 12‐week complete closure, but the magnitude of benefit is not uniform, and not every product class, product, population, or setting should be expected to reproduce the mean effect.

4.1. How This Review Extends Prior Evidence

Several systematic reviews and meta‐analyses have already suggested that placental and amniotic‐derived products can improve healing in diabetic foot ulcers [10, 11, 12, 13, 14, 15, 16]. However, many prior syntheses pooled different follow‐up windows, combined complete and partial healing outcomes, or focused on narrower product groups. The present review adds a more specific and cautious synthesis by restricting the primary analysis to randomised evidence with extractable complete wound closure data at 12 weeks, while keeping trials with earlier, non‐matching, model‐estimated, active‐comparator, or dosing‐comparison endpoints outside the main pool [29, 30, 31, 32, 33]. This approach provides a cleaner estimate for a clinically common decision point.

Another difference is the product frame. Earlier reviews mainly assessed amniotic membrane, human amnion/chorion membrane, or dehydrated amnion/chorion allografts [10, 11, 12, 13, 14, 15, 16]. In contrast, this review used a broader placental‐derived CAMP framework that included amnion‐chorion products, amnion‐only products, viable or cryopreserved placental membrane, hypothermically stored amniotic membrane, umbilical cord‐derived products, and full‐thickness decellularised placental membrane [8, 9, 21, 23, 24, 25, 26, 27, 28]. This broader scope is useful for contemporary wound‐care terminology and coverage discussions, but it also means that the pooled estimate should not be interpreted as the effect of any single product.

The review also adds a more structured assessment of certainty. Risk of bias was judged for the 12‐week closure outcome itself, rather than for each article as a whole, using the Cochrane Risk of Bias 2 tool (RoB 2) [17], and certainty of evidence was assessed using the Grading of Recommendations Assessment, Development and Evaluation (GRADE) approach [18]. The product‐class analysis, prediction interval, leave‐one‐out analysis, odds‐ratio sensitivity analysis, and high‐risk‐of‐bias‐excluded sensitivity analysis all supported the same overall interpretation: adjunctive placental‐derived CAMPs appear to improve 12‐week complete closure compared with standard care, but the size of benefit varies across products and settings.

4.2. Limitations

Several limitations should temper interpretation. First, the evidence base is dominated by industry‐funded or industry‐linked trials, and too few independent studies were available to support a meaningful sponsorship sensitivity analysis; reporting and publication bias cannot be excluded, and because fewer than 10 trials entered the primary analysis, formal small‐study‐effect testing (Egger or Begg) was not performed. In addition, Web of Science was listed in the protocol as a source to be searched if available, but it was not included in the final database search because no functional search interface was available on the platform at the time of searching; this protocol deviation was documented. Second, screening and extraction were performed by one reviewer with adjudication of uncertain records and final decisions by a second reviewer, rather than fully independent duplicate screening; disagreements were resolved by consensus, but this workflow is less protective against selection error. Third, the products pooled under the CAMP framework are biologically and manufacturing‐diverse, spanning amnion‐only, amnion‐chorion, viable or cryopreserved placental membrane, hypothermically stored amniotic membrane, umbilical cord‐derived, and full‐thickness decellularised placental membrane products, so the pooled estimate is a class‐level average rather than the effect of any single product, and product‐class subgroups were limited by small numbers per class. Fourth, blinding is intrinsically difficult in wound‐product trials, and deviations from intended interventions were the most frequent risk‐of‐bias concern. Fifth, several randomised studies reported healing at endpoints other than 12 weeks, or as model‐estimated rather than raw event counts, and were retained outside the primary pool; this conservative rule strengthens internal consistency but sets aside otherwise informative data. Finally, complete closure at 12 weeks, although a standard and decision‐relevant endpoint, does not capture durability of healing, recurrence, ulcer‐free survival, amputation, or cost, outcomes increasingly central to both clinical and coverage decisions.

4.3. Implications for Research and Practice

For practice, these findings support adjunctive placental‐derived CAMPs as a reasonable option for diabetic foot ulcers that fail to progress on well‐delivered standard care, while making clear that the product is an adjunct to, not a substitute for, the fundamentals of offloading, debridement, infection control, and perfusion. The variability captured by the prediction interval argues for a response‐based approach: favouring products with the strongest standard‐care‐controlled randomised evidence, monitoring the early healing trajectory, and reconsidering the intervention when an ulcer does not respond, rather than applying a fixed number of grafts irrespective of progress.

For policy, the timing is consequential. In the United States, CAMPs are the subject of an unusually active Medicare coverage and payment debate. After repeated delays, the Medicare Administrative Contractor Local Coverage Determinations for skin substitute grafts in diabetic foot and venous leg ulcers were withdrawn in late 2025, while the Calendar Year 2026 Physician Fee Schedule moved to a single per‐square‐centimetre payment rate, a response to Part B spending on these products that rose from roughly $250 million in 2019 to more than $10 billion in 2024 [34, 35]. The agency has explicitly invited high‐quality, peer‐reviewed evidence to inform future coverage. An endpoint‐specific, standard‐care‐anchored synthesis reporting risk ratios, prediction intervals, RoB 2, and GRADE is precisely the kind of analysis on which coverage should rest, and it cuts against both uncritical adoption and blunt, one‐size‐fits‐all restriction.

For research, the priorities are clear: adequately powered, standard‐care‐controlled randomised trials with prespecified 12‐week complete‐closure endpoints; head‐to‐head comparisons across product classes; independent, non‐industry funding; and longer follow‐up capturing recurrence, ulcer‐free and hospital‐free survival, amputation, and cost. Greater consistency in endpoint definition and analysis population would also let future syntheses pool a larger share of the available randomised evidence.

5. Conclusion

Adjunctive placental‐derived cellular, acellular, and matrix‐like products improve complete closure of diabetic foot ulcers at 12 weeks compared with standard care, with moderate‐certainty evidence and a stable, internally consistent average effect. The average benefit is real but not uniform: a significant pooled estimate sits alongside a prediction interval that crosses the null, indicating that the mean effect should not be assumed for every product, patient, or setting. These products are best understood as evidence‐supported adjuncts to, not replacements for, sound standard care, and both clinical use and coverage policy are best served by endpoint‐specific randomised evidence, product‐level data, and outcomes that extend beyond closure to durability, limb salvage, and cost.

Funding

The authors have nothing to report.

Ethics Statement

Ethical approval was not required because this study was a systematic review and meta‐analysis of previously published studies.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Data S1: Supporting Information.

Supplementary File 1 Search strategies and PRISMA‐S reporting.

Table S1: Sources searched and record counts.

Table S2: Exact database search strategies.

Table S3: Record handling and study‐selection accounting.

Table S4: PRISMA‐S reporting checklist for the search.

Table S5: PRISMA 2020 checklist with reporting locations.

Data Availability Statement

All data used in this review and meta‐analysis were extracted from published randomized studies. Extracted data and analysis materials are available from the corresponding author upon reasonable request.