Introduction

Diabetic foot ulcers (DFUs) have become a global health concern, with an estimated incidence of 19 to 35% in patients with diabetes mellitus [1]. It has been reported by the International Diabetes Federation that there will be 9.1–26.1 million patients develop DFUs annually [1]. Patients with DFUs are related with decreased quality of life (QoF) and increased risk of depression [2, 3]. Furthermore, diabetic foot infection (DFI) is more frequent in the DFU patients due to the incomplete skin and exposed bone, which may result in increased amputation risk to as high as 92% [4, 5]. It was reported that approximately 1 in 6 DFU patients will suffer from amputation, causing a mortality rate of about 47% within 5 years and a recurrence risk as high as 66% [6, 7].

DFUs treatment is associated with about 1/3 of the total diabetic care cost [8]. The primary goal of DFU treatment is to promote the re-epithelialisation of wound to reduce the complications risk associated with ulceration and to improve the patient’s QoF to a ‘pre-ulceration’ status. Besides glycemic control and revascularization, standard of care (SOC) treatment has been commonly selected as the conventional application for DFU wound management, which usually consists of the surgical sharp debridement, wound moist dressing, application of removable or irremovable off-loading device, and infection control [7, 9, 10]. The review of Everett et al. [11] summarized a total of 7 critical SOC practices, including surgical debridement, dressings promoting a moist wound environment, wound off-loading, vascular assessment, treatment of active infection, glycemic control, and ultidisciplinary care. Although these SOC practices are considered the “gold standard” for DFU care, the 20-week healing rate of DFU after SOC was less than 30% [12], and 40 and 65% of healed DFUs will recur within 1 year and 5 years, respectively [1]. Therefore, current SOC alone may not be sufficient to heal all DFUs and prevent their recurrence [13].

In recent years, a broad spectrum of novel treatments have been developed to improve diabetic wound healing. In areview by Snyder et al. [14], they identified a total if 76 commercially available skin substitutes used to treat chronic wounds. The majority of these substitutes do not contain cells and are derived from human placental membrane (the placenta’s inner layer), animal tissue, or donated human dermis allograft. These skin substitutes, whether allogeneic or xenogeneic graft, could provide the essential structure of extracellular matrix, signals for cellular migration, proliferation, angiogenesis, and endogenous matrix production and biochemical functions for enhancing wound healing [15, 16]. Many studies have demonstrated that these dermal matrices are effective when applied as adjuvant treatment to enhance DFUs healing [17,18,19]. However, high-level evidence to comprehensively illustrate the effectiveness and safety of SOC plus dermal matrix over SOC alone is still scarce.

Thus, the current systematic review will be conducted with the aim of evaluating the effectiveness and safety of dermal matrix application as an adjuvant treatment of SOC, basing on the available evidence from randomized controlled trials (RCTs).

Materials and methods

This study was carried out in accordance with the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guideline [20], and the checklist is presented in Supplementary Appendix 1.

Data sources

The following three databases were independently searched by two authors: PubMed, Embase and CENTRAL. The searching was completed using a method of combination of subject and free terms, with the following key terms: “diabetic foot ulcer”, “acellular dermal matrix”, “cellular dermal matrix”, “wound healing”, and so on. No restriction on the publication countries/ regions and publication date, while the publication language was restricted on English. Additionally, the references lists of the included studies were reviewed, and the potential related studies were hand searched and screened for eligibility.

Inclusion and exclusion criteria

The retrieved records from the three databases were screened according to the following inclusion criteria: (1) patients: diagnosed with DFU; (2) intervention: biogenic skin substitutes for enhancing DFU healing, whether allogeneic or xenogeneic dermal matrix graft; (3) comparison: between SOC and dermal matrix; (4) outcomes: treatment outcomes of DFU, including ulcer healing rate, healing time, wound area, ulcer recurrence, amputation risk and complication risk; (5) studies: only prospectively designed RCTs were eligible.

Studies were excluded according to the following criteria: (1) patients with ulcers on foot caused by reasons other than diabetes, or patients with ulcers caused by diabetes on the lower leg; (2) patients treated with methods other than dermal matrix or SOC; (3) no available data on the effectiveness and safety outcomes; (4) studies designed as case series, cohort study, retrospective case-control study, systematic review/ meta analysis, literature review, and so on.

Study screening and data extracting

Two authors independently screened all electronic records retrieved from the databases, according to the inclusion and exclusion criteria to select eligible studies. At the beginning, the records were imported into the EndNote software version X9 to eliminate the duplicates. Then, the two authors reviewed the titles/ abstracts of the remaining non-duplicates, to remove clearly irrelevant studies. After then, the full-text of the remained studies was downloaded and reviewed to evaluate the eligibility for inclusion.

The data extraction process was also completed by two authors independently, to obtain the following items: (1) study characteristics: the first author’s name, publication year, corresponding country/ region, study period, and follow-up time; (2) subjects characteristics: patients number, dropped patients number, male percentage, mean age, mean BMI, diabetes type (type I or type II), length of diabetes history, glycosylated hemoglobin percentage (HbA1c%), ankle-brachial index (ABI), patients with diabetic polyneuropathy (DPN), DFU grade according to Wagner or other classifications, DFU site (planta, dorsal, or other sites), DFU size, and DFU age; (3) treatment details: treatment regimens in screening phase and treatment phase, screening criterion for randomization in screening phase, and dermal matrix product; (4) outcome evaluations: ulcer healing rate, healing time, wound area, ulcer recurrence, amputation risk and complication risk.

Quality assessment

The quality of the included RCTs was assessed using the Cochrane Collaboration tool for risk of bias assessment [21], which evaluates a total of 7 kinds of biases for RCTs as follows: (1) randomization sequence generation, (2) allocation concealment, (3) blinding of participants and personnel, (4) blinding of outcome assessment, (5) incomplete outcome data, (6) selective reporting and (7) other bias.

Statistical analysis

When results of per-protocol (PP) and intention-to-treat (ITT) analyses were both reported, the ITT principle was followed in the analyses process. Comparisons of continuous outcomes (including time to complete heal and wound area, at final follow-up) between SOC and dermal matrix groups, were expressed as mean difference (MD) and 95% confidence interval (95% CI). Single-rate meta analysis was performed to calculate the pooled healing rate and ulceration recurrence rate of dermal matrix group at the final follow-up. Comparisons of dichotomous outcomes between SOC and dermal matrix groups, were expressed as odds ratio (OR, complete healing rate) and risk ratio (RR, ulcer recurrence rate, amputation risk and complication risk) as well as their 95% CIs. Heterogeneity among studies was estimated by I2 statistics. If I2 ≥ 50%, it indicates significant heterogeneity and random-effect model was applied. Otherwise, a fixed-effect model was used in case of non-significant heterogeneity.

If significant heterogeneity was detected, sensitivity analysis was performed by omitting each individual study sequentially to assess the impact of each study on the results. For outcomes reported in more than 5 studies, funnel plot was plotted, and publication bias was assessed using Egger’s and Begg’s tests (p < 0.100 and p < 0.050 were considered to indicate significant publication bias respectively). If significant publication bias was detected, non-parametric trim-and-filling method was used to adjust the publication bias. Data analyses were performed using the R language version 4.2.1 (R Foundation for Statistical Computing, Vienna, Austria). All statistical tests were two-sided and P value of less than 0.05 was considered significant.

Results

Study selecting

The flow chart of studies screening is shown in Fig. 1. From the initial search, 520 studies were identified, of which 132 were duplicates that wereimmediately excluded. A futher 334 records were excluded after screening the titles/ abstracts, leaving 54 articles for full-text review. As a result, a total of 15 [22,23,24,25,26,27,28,29,30,31,32,33,34,35,36] and 14 [22, 23, 25,26,27,28,29,30,31,32,33,34,35,36] RCTs were included in the qualitative and quantitative synthesis (meta-analysis), respectively.

Fig. 1
figure 1

Flowchart of study searching and screening

Full size image

Summary of the included studies

Table 1 summarizes the characteristics of the included studies. A total of 15 trials involving 1524 subjects were included in the analysis. The studies randomized a total of 689 patients to receive SOC alone and 835 patients to receive SOC plus dermal matrix. The male percentages were reported in 13 of the studies, ranging from 34.6 to 100.0%. The mean age was reported in 14 studies, ranging from 55.2 to 66.6 years. The mean BMI was available in 10 studies, ranging from 28.5 to 36.5 kg/m2. The follow-up periods were 4, 6, 12, 16, 21, 24, 28, and 42 weeks in 1 [22], 1 [36], 5 [25, 29,30,31, 34], 4 [23, 24, 27, 35], 1 [32], 1 [26], 1 [28] and 1 [33] studies, respectively. At the final follow-up, a total of 147 and 155 patients dropped out for follow-up.

Table 1 Characteristics of the included studies
Full size table

Table 2 summarizes the basic information about the status of diabetes mellitus and DFUs at randomization. HbA1c% levels were reported in 12 studies, with a range of 7.11 to 10.2%. The mean ABI was reported in 4 studies, with a range of 0.7 to 1.2. A total of 12 studies reported the mean DFU size, with a range of 1.3 to 32.1 cm2. The treatment details of the included studies are listed in Table 3. Before the treatment phase, 8 studies [23,24,25, 27, 28, 31, 32, 34] included screening phase and applied SOC treatment (surgical debridement, wound dressings, wound off-loading, and infection management) to the enrolled patients lasting for 1 or 2 weeks. At the end of the screening phase, the including criterion for randomization was set as less than 20% [25, 27], 30% [23, 31, 32] or 40% [28] reduction of wound area in the ulcer site. Many different dermal matrix products were applied for wound repair, with the GraftJacket matrix (Wright Medical Technology, Inc., Arlington, TN, USA) being the most commonly researched product in 4 RCTs [22, 26, 29, 35].

Table 2 Baseline information about the statuses of diabetes mellitus and diabetic foot ulceration (DFU)
Full size table
Table 3 Treatment details of the included studies
Full size table

Figure 2 shows the results of the quality assessment of the included RCTs. Due to the obviously different treatment process, the blinding of the patients was difficult. As a result, there is a high risk of bias in the item of “blinding of participants and personnel”. The item “blinding of outcome assessment” was also presented with high risk of bias in 4 of the studies. The other items were of relatively low risk of bias.

Fig. 2
figure 2

The light bulb diagram for quality assessment of the included studies

Full size image

Effectiveness of the dermal matrix in DFU

The comparison of mean time to complete healing between SOC and dermal matrix is presented in Fig. 3. Six primary trials were pooled with random-effect model (I2 = 97%), and significantly shorter time was required to achieve complete healing in dermal matrix group compared to the SOC group (see the forest plot in Fig. 3A, MD = 2.84, 95%CI: 1.37 ~ 4.32, p < 0.001***). The funnel plot (Fig. 3B), Egger’s test (p = 0.143) and Begg’s test (p = 0.573) indicate that there is no significant publication bias. The forest plot of sensitivity analysis (Fig. 3C) showed there was no single study that significantly influenced the pooling result.

Fig. 3
figure 3

The pooling result for complete healing time compared between SOC and dermal matrix groups at final follow-up. A forest plot; B funnel plot; C forest plot for sensitivity analysis

Full size image

The pooling result for complete healing rate of dermal matrix group at final follow-up is shown in Fig. 4. Thirteen studies and 16 arms were pooled with random effects model (I2 = 86%), and the pooled healing rate of dermal matrix group was 0.70 (95CI: 0.61 ~ 0.78) (see the forest plot in Fig. 4A). The forest plot of sensitivity analysis (Fig. 4D) revealed that none of the studies had a significant impact on the pooled result.. However, the funnel plot (Fig. 4B), Egger’s test (p = 0.012) and Begg’s test (p = 0.529) indicated the presence of significant publication bias. Therefore, a trim and filling funnel plot was generated, which resulted in an adjusted healing rate of 0.56 (95CI: 0.47 ~ 0.66) for the dermal matrix group (Fig. 4C).

Fig. 4
figure 4

The pooling result for complete healing rate of dermal matrix group at final follow-up. A forest plot; B funnel plot; C trim and filling funnel plot; D forest plot for sensitivity analysis

Full size image

The comparison of the complete healing rate between SOC and dermal matrix is presented in Fig. 5. Thirteen studies and 16 arms were pooled with fixed-effect model (I2 = 33%), resulting in a significantly higher complete healing rate in dermal matrix group compared to the SOC group (see the forest plot in Fig. 5A, OR = 0.40, 95%CI: 0.33 ~ 0.49, p < 0.001***). The funnel plot (Fig. 5B), Egger’s test (p = 0.224) and Begg’s test (p = 0.242) did not show any significant publication bias. Sensitivity analysis was not conducted as there was no significant heterogeneity.

Fig. 5
figure 5

The pooling result for complete healing rate compared between SOC and dermal matrix groups at final follow-up. A forest plot; B funnel plot

Full size image

Figure 6 shows the comparison of wound area between SOC and dermal matrix, which pooled the data from three studies and four arms using random effects model (I2 = 98%). No significant difference between two groups was found (Fig. 6A, MD = 0.29, 95%CI: − 0.32 ~ 0.91, p = 0.352). The forest plot of sensitivity analysis (Fig. 6B) showed there was no arm caused significant influence on the pooling result.

Fig. 6
figure 6

The pooling result for wound area compared between SOC and dermal matrix groups at final follow-up. A forest plot; B forest plot for sensitivity analysis

Full size image

Safety of the dermal matrix in DFU

The ulcer recurrence rate comparison between SOC and dermal matrix is presented in Fig. 7A-B. Five studies were pooled with fixed-effect model (I2 = 32%), and no significant difference in ulcer recurrence rate was observed between the two groups (Fig. 7A, RR = 1.32, 95%CI: 0.92 ~ 1.89, p = 0.138). The funnel plot (Fig. 7B), Egger’s test (p = 0.827) and Begg’s test (p = 1.000) did not show any significant publication bias. Sensitivity analysis was not conducted as there was no significant heterogeneity.

Fig. 7
figure 7

The pooling results for ulcer recurrence rate compared between SOC and dermal matrix groups (A and B), and for ulcer recurrence rate of dermal matrix group (C-E), at final follow-up. A & C. forest plot; B & D. funnel plot; E forest plot for sensitivity analysis

Full size image

The pooling result for ulcer recurrence rate of dermal matrix group at final follow-up is shown in Fig. 7C-E. Five studies were pooled with random-effect model (I2 = 64%), resulting in a pooled ulcer recurrence rate of 0.11 (95CI: 0.05 ~ 0.17) for the dermal matrix group (see the forest plot in Fig. 7C). The forest plot of sensitivity analysis (Fig. 7E) showed that no study had a significant influence on the pooled result. The funnel plot (Fig. 7D), Egger’s test (p = 0.738) and Begg’s test (p = 1.000) did not indicate the presence of significant publication bias.

Figure 8A shows the forest plot comparing the overall amputation risk between SOC and dermal matrix groups using fixed-effect model (I2 = 0%), demonstrating that dermal matrix application could significantly lower the overall amputation risk (RR = 1.83, 95%CI: 1.15 ~ 2.93, p = 0.011*). After then, subgroup analyses were conducted for both major (Fig. 8B) and minor (Fig. 8C) amputation risks, showing that dermal matrix application could significantly lower the major amputation risk (RR = 2.64, 95%CI: 1.30 ~ 5.36, p = 0.007**), but had no significant impact on the minor amputation risk (RR = 1.02, 95%CI: 0.49 ~ 2.12, p = 0.959). Publication bias test and sensitivity analysis were not performed.

Fig. 8
figure 8

The pooling results for overall amputation risk (A) major amputation risk (B) and minor amputation risk (C) compared between SOC and dermal matrix groups, at final follow-up

Full size image

The comparison of complication rate between SOC and dermal matrix is presented in Fig. 9. Thirteen studies and 15 arms were pooled with fixed-effect model (I2 = 0%), and no significantly different complication rate was observed between two groups (Fig. 9A, RR = 1.06, 95%CI: 0.93 ~ 1.20, p = 0.409). The funnel plot (Fig. 9B), Egger’s test (p = 0.494) and Begg’s test (p = 0.622) did not indicate the presence of significant publication bias. Sensitivity analysis was not conducted as there was no significant heterogeneity.

Fig. 9
figure 9

The pooling result for complication risk compared between SOC and dermal matrix groups, at final follow-up. A forest plot; B funnel plot

Full size image

Discussion

In this study, we performed a systematic review and meta-analysis based on high-level evidence from RCTs, and found that application of dermal matrix was associated with significantly shorter time to complete healing, increased healing rate, and reduced amputation risk, compared to SOC alone. However, there was no significant difference in wound area, ulcer recurrence and complication risk, between the two groups.

DFUs are always characterized by chronicity and recurrence, making them difficult to fully heal and potentially leading to minor or major limb amputations. The clinical challenges related with DFUs treatment have spawned multiple adjuvant techniques to improve the wound healing. Usually, an ulcer continued for more than 4 weeks is qualified as chronic wound, which can present additional challenges to complete the wound healing because of infection, biofilm formation, and underlying tissue desiccation that cause exacerbated conditions and disturbed healing process. Multiple biologic dressings have been applied in clinical researches in the setting of DFUs, showing promise treatment outcomes. Martin et al. [37] evaluated the outcomes of 17 consecutive patients with neuropathic diabetic foot wounds treated with an acellular matrix, which showed a 20-week healing rate of 82.4% with an average healing time of 8.9 ± 2.7 weeks. Lee et al. [38] compared the efficacy of applying a paste formulation of acellular dermal matrix (ADM) with conventional foam dressing in treating DFUs, reporting an increased healing rate (56.52% vs. 23.08%), increased ratio of healed area (74.17% ± 30.84% vs. 51.87% ± 32.81%) and deceased length of time to heal (13.54 ± 9.18 vs. 21.5 ± 11.98 days) in ADM group, at the 60-day primary outcome mark. Zelen et al. [25] compared the clinical outcomes of human reticular acellular dermis matrix (HR-ADM) versus SOC to facilitate wound closure in non-healing DFUs. At the final follow-up (12 weeks), DFUs of HR-ADM and SOC groups healed in 80 and 20% of the patients, with a mean healing time of 40 days and 77 days, respectively. There was no significantly increased adverse or serious adverse events between the two groups or any adverse events related to the graft. In another RCT by Hahn et al. [33], the clinical outcomes of a micronized dermal matrix (MDM) was compared with conventional negative-pressure wound therapy (NPWT) in the treatment of DFUs. As a result, all wounds treated with MDM showed healthy granulation tissue without noticeable complications during follow-up. The MDM group showed a higher healing rate compared to NPWT group, at 42 and 120 days, while similar healing rates were achieved between two groups at 6-month follow-up period. In 2017, a systematic review and meta-analysis conducted by Guo et al. [39] compared the efficacy and safety of ADM in DFU treatment, which showed that compared with the SOC alone, the ADM group was associated with higher complete healing rates at 12 and 16 weeks, and shorter mean time to complete wound healing. The adverse event rates in both groups were similar, indicating that the use of ADM did not increase the risk of adverse events.

In the current study, we included 15 RCTs involving 1524 patients, and demonstrated that dermal matrix is an effective and safe treatment option for enhancing DFU healing. The results of this study support previous studies documenting the successful application of dermal matrix therapy. Dermal matrix acts as a sterile tissue graft which can be applied directly to wound beds of DFUs and integrate with the surrounding host tissues to actively stimulate cell migration, angiogenesis, and epithelialization, resulting in accelerated wound healing [40]. Although the final follow-up wound area was similar between two treatment groups, the percentage area reduction (PAR) was demonstrated to be significantly increased in several studies [23, 26, 27, 31, 34]. However, the PAR was not pooled by meta-analysis due to the non-availability of the primary data (mean value and standard deviation of PAR).

This study, nevertheless, has several limitations that should be pointed out. Firstly, the dermal matrix products used in the studies varied among different manufacturers, which may introduce potential risk of bias. Secondly, due to the inconcealability of the treatment process with dermal matrix in the primary trials, an additional risk of bias may be caused by the unblinded application of dermal matrix to patients. Additionally, the studies reported outcomes at different follow-up times, making it difficult to pool the data. In this study, we selected the data at the final follow-up to conduct the analyses. Finally, most of the current available RCTs have relatively small sample sizes and short-term follow-up periods, which indicates that some more trials with larger sample size and longer follow-up period are required to provide some more convincing evidence.

Conclusions

The results of the current meta-analysis demonstrated that the application of dermal matrix as an adjuvant therapy to SOC can effectively enhance the healing process of DFUs and reduce the amputation risk when compared to SOC alone. Additionally, dermal matrix application was well tolerated by the subjects without added complication risk. However, some further well-designed prospective trials with larger sample sizes and longer follow-up periods are required to provide more convincing evidence.