The Comparative Effectiveness of a Human Fibroblast Dermal Substitute versus a Dehydrated Human Amnion/Chorion Membrane Allograft for the Treatment of Diabetic Foot Ulcers in a Real-world Setting

Background. Impaired wound healing is associated with serious complications in patients with diabetes. Diabetic foot ulcers (DFUs) can lead to costly complications and an increased mortality rate. Standard treatments for DFUs often need to be augmented with adjunctive therapies designed to stimulate healing in recalcitrant wounds.Objective. This analysis was conducted to evaluate the comparative effectiveness of a human fibroblast-derived dermal substitute (HFDS) and a dehydrated human amnion/chorion membrane allograft (dHACM) for the treatment of DFUs. Materials and Methods. Using a wound care-specific electronic health record database, real-world outcomes from 122 patients with 122 DFUs receiving treatment in 2014 in 72 wound care facilities across the United States were evaluated. Key criteria for entry into the analysis included ulcer size ≥ 1 cm² to < 25 cm², ulcer duration ≤ 1 year, and ulcer area reduction ≤ 20% in the 14 days prior to the first treatment. Key exclusion criteria included lack of follow-up visits and lack of baseline wound measurements. The frequency of wound closure by weeks 12 and 24, median time to wound closure, hazard ratio with 95% confidence interval, and P value were estimated from a Cox model with terms for treatment, baseline wound area, baseline wound duration, baseline wound depth, and wound location. Results. The results show the incidence of wound closure for HFDS compared with dHACM was significantly improved by weeks 12 (55% vs. 32%) and 24 (76% vs. 50%). The HFDS treatment significantly increased the probability of wound closure by 107%, with a median time to closure of 7.4 weeks (38%) less than that of dHACM treatment (P = .02).

A quarter of the 22.3 million patients with diabetes in the United States are expected to develop a diabetic foot ulcer (DFU) at some point in their lives.^1,2 Diabetic foot ulcers have been estimated to affect 1% to 8% of patients with diabetes annually.^3,4 These ulcers are associated with prolonged hospitalization, increased risk of infection, and costly amputations.^5,6 The 5-year mortality rate for amputee patients is close to 50%. Patients who have undergone an amputation have the same 5-year mortality rate as those suffering from colon cancer.⁷ In addition, DFUs pose a significant financial burden on US payers at an estimated annual direct medical cost of up to $13 billion.⁸ 

Previous studies have shown that wounds that penetrate to the bone, recurrent wounds, wounds of long duration (> 30 days), and wounds associated with peripheral vascular disease are risk factors for infection and subsequent amputation or bone resection. In a clinical trial conducted in 2006, Lavery et al⁹ integrated patient education and regular podiatric care for patients determined to be high risk for foot infections. Despite these efforts, 9.1% of patients enrolled in the study developed foot infections during follow-up, with 1 out of 5 developing infections that extended to the bone. The risk of amputation for patients with foot infections was 155 times greater than those who did not have one.⁹ It has been demonstrated that early detection and intervention of ulcers exhibiting delayed healing are essential to reducing the risk of a serious outcome.^5,10 Standard treatment for DFUs includes moist dressings, debridement, wound offloading, and infection control. Even with the best care, these wounds are often slow to heal, requiring clinicians to seek out advanced therapies to reduce the patient’s risk of amputation.¹¹ There are currently more than 60 cellular-based and/or tissue-based products for wounds listed by Medicare, which creates a difficult decision for clinicians when choosing an appropriate, safe, and effective adjunctive therapy. Few of the products have been extensively studied in randomized controlled trials (RCTs), and only 3 cellular and/or tissue-based products are approved by the US Food and Drug Administration (FDA) for wound healing.

There are several products intended to provide a wound covering that have come to market as Section 361 Human Cellular and Tissue-Based Products (361 HCT/P’s) that do not require FDA premarket review of clinical efficacy or safety data. One such 361 HCT/P is a dehydrated human amnion/chorion membrane (dHACM; Epifix, MiMedx Group Inc, Marietta, GA), which is a collagenous allograft wound covering derived from the submucosa of donated human placenta and contains no living cells.

There are 2 bioengineered living cellular therapies approved by the FDA for the treatment of DFUs. One is a bioengineered living cellular construct (BLCC; Apligraf, Organogenesis Inc, Canton, MA) that is comprised of human neonatal keratinocytes and fibroblasts in an extracellular matrix (ECM) of bovine and human collagen and other ECM proteins. A small, randomized controlled clinical study¹² comparing dHACM to BLCC reported that dHACM had better rates of complete healing than BLCC (95% at 6 weeks with dHACM compared with 45% at 6 weeks with BLCC) with a median time to healing of 13 days for dHACM versus 49 days for BLCC. In order to assess these findings in a larger cohort of patients with similar DFU profiles, comparative-effectiveness research analyses were conducted using real-world data from an electronic health record (EHR) database. The real-world results differed substantially from those reported in the dHACM RCT. In fact, the comparative-effectiveness research analysis showed the frequency of wound closure by week 12 of 48% for BLCC and 28% for dHACM and median times to wound closure of 13.3 weeks for BLCC versus 26 weeks for dHACM.¹³ The BLCC versus dHACM comparative-effectiveness analysis was undertaken to further understand the effectiveness of cellular-based versus noncellular-based products as well as to assess the reproducibility of findings from studies that included few subjects to studies of larger patient populations in real-world practice settings.

The other FDA-approved cellular-based therapy is a human fibroblast-derived dermal substitute (HFDS; Dermagraft, Organogenesis Inc, Canton, MA) comprised of human neonatal dermal fibroblasts cultured in vitro onto a bioabsorbable mesh to create a three-dimensional human dermal substitute. A HFDS contains metabolically active living cells that produce human collagen, matrix proteins, growth factors, and cytokines.¹⁴

Although the efficacy of HFDS has been established through FDA-reviewed and FDA-approved large, prospective, RCTs, the comparative effectiveness of HFDS and dHACM for treating DFUs in real-world settings has not been investigated yet. Given the similarities between HFDS and BLCC and the discordant findings previously reported for BLCC and dHACM, the investigators sought to further evaluate the real-world outcomes of dHACM by comparing it to the other commercially available bioengineered cellular technology (HFDS). Effectiveness is the extent to which an intervention achieves its intended effect in routine care conditions (real-world situations).^15,16 While randomized clinical trials are required to gain FDA approval by determining a product’s efficacy and safety, the results of these studies may not accurately reflect results seen in the real world. The EHRs used in the daily care of patients contain millions of observations on health outcomes from using various therapies, which allow researchers to compare the effectiveness of various treatments using real-world data.^17,18 The purpose of this analysis is to evaluate real-world outcomes of treatment with HFDS compared with dHACM for the treatment of DFUs. 

Study design and data collection. The authors used deidentified data consistent with the terms and conditions of the Health Insurance Portability and Accountability Act of 1996 (HIPAA) from the WoundExpert EHR database (Net Health, Pittsburgh, PA) to perform a retrospective analysis comparing the effectiveness of HFDS versus dHACM for the treatment of DFUs. This wound care-specific EHR database is utilized in 90% of wound care clinics across the United States.¹⁹ Treatment records included patient baseline demographics, wound location, wound size and duration, and wound-specific information recorded at each visit, including area measurements and treatments. An analysis was conducted on DFUs that received their first HFDS or dHACM treatment in 2014 from 72 wound care facilities across the United States.

The primary analyses were the frequency of DFUs achieving wound closure by weeks 12 and 24 and the median time to wound closure. Wound measurements of length and width were used to calculate wound area in cm². As patients with complete wound closure do not always follow-up, wound closure was defined as an ulcer achieving an area ≤ 0.25 cm².

Study population. Patients who received at least 1 treatment with HFDS or dHACM on a DFU with a location coded as foot, toe, heel, metatarsal head, toe web space, toe amputation site, or transmetatarsal amputation site were eligible for inclusion in the analysis. Wound size, duration, and healing trajectory criteria for inclusion in these analyses were based on criteria employed by Zelen et al¹² in a previous study evaluating outcomes of dHACM for DFUs. Accordingly, criteria for entry into the analysis included ulcer size ≥ 1 to < 25 cm², ulcer duration ≤ 1 year, and ulcer area reduction ≤ 20% in the 14 days prior to the first treatment with HFDS or dHACM. Exclusion criteria included lack of follow-up visits and lack of baseline wound measurements.

Wounds that received alternate cellular-based and/or tissue-based products up to 28 days prior to or concurrent with the first HFDS or dHACM treatment were also excluded. Censored observations included last visit with an area measurement for nonhealed wounds, visit where an alternate product was applied, and visit where the subsequent application of the same product occurred if > 183 days had passed since the initial application. 

Statistical analyses. The treatment period started with the first use of HFDS or dHACM. Cox proportional hazards regression analysis was used to estimate the percentage of DFUs with wound closure by weeks 12 and 24 and the median time to wound closure. Statistical significance was defined as P < .05. Baseline characteristics were compared using two-sample t-tests for continuous variables and Fisher’s exact tests (two-tailed) for difference in proportions between treatments. The frequency of wounds closed at weeks 12 and 24, median time to wound closure, hazard ratio with 95% confidence interval (CI), and P value were estimated from the Cox model with terms for treatment, baseline wound area, baseline wound duration, baseline wound depth, and wound location.

Relevant patient information identified in the EHR included in the study is shown in Table 1. A total of 59 wounds (59 patients) and 63 wounds (63 patients) treated with HFDS and dHACM, respectively, met the eligibility requirements for inclusion in the analysis. There were no statistically significant differences in baseline patient characteristics or baseline wound characteristics between the 2 treatment groups (Tables 1, 2). The average age at the start of treatment for patients in both groups was about 61 years (P = .702). There were more women than men in both groups, with 46 women versus 13 men in the HFDS group and 48 women versus 15 men in the dHACM group (P = .833). Body mass index was about the same for both groups (48 for HFDS vs. 49 for dHACM), and each patient selected had only 1 wound treated (P = .622). The average wound size was 4.8 cm² in the HFDS group and 5.2 cm² in the dHACM group (P = .661). The average wound duration prior to treatment was 4.2 months in the HFDS group and 4.6 months in the dHACM group (P = .412).

The average number of treatment applications was 4.6 in the HFDS treatment group and 3.5 in the dHACM treatment group (P = .012). For patients receiving multiple applications, the average interval between applications was 11.2 days in the HFDS group and 17 days in the dHACM group (P = .058) (Table 3). 

Cox proportional hazards analysis after adjusting for ulcer area, duration, depth, and location showed HFDS treatment significantly improved the median time to DFU wound closure by 38%, achieving the endpoint 7.4 weeks sooner than dHACM-treated patients (Figure 1). 

Cox regression model estimated frequency of wound closure for HFDS compared with dHACM was significantly improved by weeks 12 (55% vs. 32%) and 24 (76% vs. 50%) (Figure 2). Human fibroblast-derived dermal substitute treatment significantly increased the probability of wound closure by 107% compared with dHACM treatment by the method of the Cox proportional hazards analysis after adjusting for key covariates including ulcer size and duration (Hazard Ratio = 2.07 [95% CI, 1.12–3.82]; P = .02). 

Building on previous research and using the same real-world comparative effectiveness analysis methods to show improved treatment outcomes with a bioengineered living cellular construct,¹³ HFDS was shown to increase the probability of wound closure by 107% compared to treatment with a nonviable dehydrated amniotic membrane. 

The increased use of EHRs provide a detailed record of individual treatment regimes, failures, and successes that are not limited in the ways that RCTs are known to be.^20,21 The inclusion/exclusion criteria defined in RCTs often prevents the data from being replicated in real-world settings, especially for at-risk patients with common comorbidities. A RCT may also be geographically limited, creating a demographic and clinical center bias. In contrast, observational data generated from EHRs is not limited in patient number, location, or other burdens of feasibility. 

The use of EHRs allows researchers to provide additional clinical evidence unable to be obtained in prospective RCTs, and to do so in more cost-effective ways. Where RCTs provide the data on efficacy in a controlled setting, comparative-effectiveness research using EHRs evaluates outcomes in real-world clinical practice. 

The scientific basis of chronic DFUs has been studied at length. It has been shown that fibroblasts from chronic ulcers have a low proliferative capacity and a senescent phenotype in vitro.²² Human fibroblast-derived dermal substitute, comprised of living and metabolically active human fibroblasts, secretes growth factors and cytokines while providing human collagen and ECM proteins to the wound, which are thought to modify the inflammatory state of the ulcer and stimulate healing. 

The safety and efficacy of HFDS were studied as part of a large, multicenter, pivotal RCT conducted for FDA approval by Marston et al,¹⁴ where conventional treatment of debridement, saline-moistened gauze, and pressure relieving orthotics was enhanced with the addition of HFDS for the treatment group. The results showed the HFDS group healed a significantly higher percentage of ulcers compared with the control group by week 12 (30% vs. 18.3%; P = .023). The HFDS group also had a significantly faster rate of healing, and by week 12 the median percent wound closure for the HFDS group was 91% compared with 78% for the control group (P = .044).

A follow-up analysis of the adverse events reported in the multicenter RCT¹⁴ was done to investigate the incidence of lower-extremity amputation and bone resection in patients treated with HFDS compared with the conventional care control group.¹⁰ The analysis showed that while 8.9% of all patients enrolled in the study underwent amputations or bone resections, the number of patients that required these procedures was significantly less in the HFDS-treated group (5.5%) compared with the control group (12.6%; P = .031).

While the safety and efficacy of HFDS are well established, the investigators performed this current comparative-effectiveness analysis to compare the effectiveness of HFDS treatment to a donated human placental allograft when applied in routine clinical practice. While the possibility of bias in patient selection and treatment patterns for HFDS and dHACM may exist, the current analyses did adjust for wound characteristics that are known to affect or act as predictors of healing: ulcer area, duration, depth, and location. The results showed HFDS treatment significantly reduced the median time to DFU wound closure by 7.4 weeks (38% improvement) compared with dHACM-treated patients. The statistical incidence of wound closure significantly improved for HFDS compared with dHACM by weeks 12 (55% vs. 32%) and 24 (76% vs. 50%). Human fibroblast-derived dermal substitute also clearly demonstrated a clinically meaningful and significant increase in the probability of wound closure by 107% when compared with dHACM treatment.

Although the real-world results of dHACM observed in this analysis remain substantially lower than those reported in the RCT by Zelen et al,¹² the proportion of DFUs achieving complete closure was higher for HFDS when compared with their pivotal RCT results. Interestingly, the 72% relative increase in wound closure by week 12 observed for HFDS compared with dHACM in this real-world analysis is comparable to the 64% relative increase in wound healing over standard care that was reported in the HFDS pivotal RCTs.¹⁴ Comparing the results of this comparative-effectiveness analysis of HFDS versus dHACM to the previously published BLCC versus dHACM comparative-effectiveness analysis findings shows comparable relative improvements in frequency of wound closure by week 12 for both bioengineered cellular therapies versus dHACM. The relative increases in wound closure of BLCC compared with dHACM was 71% (48% vs. 28%, respectively) and 72% for HFDS compared with dHACM (55% vs. 32%, respectively).

Comprehensive methods were used in this comparative-effectiveness research analysis; however, as with all retrospective studies, the use of retrospective data has inherent limitations. Electronic health record databases are typically not developed for research purposes. The quantity of data obtained in EHRs is not rigidly controlled and standardization of these data is challenging. The database used in this study was not continuously monitored in real time for data entry as in a prospective RCT. As such, there may have been incomplete data for some patient clinic visits. Aspects of the database, such as prior surgeries or concurrent medications, may have been recorded with fewer details than in prospectively conducted trials. Adverse events were spontaneously reported rather than actively collected to help ensure more complete reporting as in RCTs. The types of secondary dressings employed, specific techniques for offloading, or the particular method of debridement administered were all data permitted to be entered in open text fields making consistent comparisons between treatment groups difficult. However, the most relevant sections pertaining to wound assessment, such as wound characteristics and measurements used in these analyses, are detailed and reliable, allowing for comprehensive and accurate assessment of wound closure outcomes.

Utilizing EHRs to capture real-world outcomes offers many advantages including the ability to conduct longitudinal analyses of larger numbers of patients over longer periods of time. A cost analysis on the use of HFDS was not included in these analyses because the EHR used does not capture all costs related to wound care and outcomes. However, the significant differences in the percentage of patients who achieve wound closure and their time to wound closure in this study suggest cost savings with HFDS may be realized. The cost effectiveness of wound care technology should be evaluated by analyses that include cost of the product, time to wound closure, and the cost incurred over time for patients not achieving wound closure. 

This comparative-effectiveness research analysis of EHR data from US wound care facilities showed the use of HFDS was significantly more effective than dHACM for the treatment of DFUs in real-world settings. The HFDS-treated wounds were found to have a significantly higher frequency of wound closure compared with dHACM-treated wounds. When using HFDS as compared with dHACM, the incidence of wound closure was significantly higher by weeks 12 (55% vs. 32%, respectively) and 24 (75% vs. 50%, respectively). The median time to achieve wound closure for wounds treated with HFDS was 7.4 weeks (38%) sooner compared with dHACM-treated wounds. The results from this real-world analysis further support HFDS efficacy data and provide valuable information to guide both physicians and payers for how to best treat patients with DFUs.

The authors thank Biostatistical Consulting Inc for statistical analyses and Jasmin Hunter for writing assistance and editorial support in manuscript preparation. Deidentified patient data released to Organogenesis were consistent with the terms and conditions of Net Health’s client contracts and the requirements of HIPAA. Net Health was not involved in the analysis, interpretation, or reporting of the data.

From the Advanced Footcare Centers, Chattanooga, TN; Sabolinski LLC, Franklin, MA; and Organogenesis Inc, Canton, MA

Address correspondence to:
Michael Sabolinski, MD
55 Jefferson Road
Franklin, MA 02038
sabolinski@gmail.com

Disclosure: This study was funded by Organogenesis, Inc (Canton, MA). Dr. Kraus is a member of the Organogenesis speaker’s bureau. Dr. Sabolinski serves as a consultant for Organogenesis, AOBiome (Cambridge, MA), Neumedicines Inc (Pasadena, CA), and Allergan (Parsippany-Troy Hills, NJ). Mr. Parsons is an employee of Organogenesis; and Ms. Skornicki is a former employee of Organogenesis.