Effectiveness of a Purified Type I Collagen Matrix Plus the Antimicrobial Polyhexamethylene Biguanide for Use in Cutaneous Wounds: Analysis of a Population of Three Combined Registries

Topical collagens and antimicrobials for application in the management of wounds have been well studied.^1-3 A combination product comprising collagen plus a broad-spectrum antimicrobial was developed in the past decade. PCMP (PuraPly AM; Organogenesis Inc) is a dry, cross-linked bilayer of scaffold porcine small intestine submucosal collagen saturated with the topical antimicrobial PHMB.^4-6 PCMP is a class II medical device that received US Food and Drug Administration 510(k) clearance (K051647) as a wound dressing for wound management.⁷ PCMP is intended for use in partial- and full-thickness wounds; pressure, venous, diabetic, and chronic vascular ulcers; and tunneled/undermined, surgical, draining, and traumatic wounds, with the exception of third-degree burns. 

Preclinical and clinical trials of collagen and PHMB individually have been performed. PHMB is a highly positively charged molecule that interacts with negatively charged phospholipids in bacterial membranes, leading to cellular disruption and death. PHMB has broad antimicrobial activity against gram-positive, gram-negative, and intracellular bacteria.⁸ Use of PHMB dressings has been reported to accelerate the time to significant reductions in bacterial counts when compared with commonly used antimicrobials, even in organisms resistant to multiple antimicrobials.^9,10 Purified type I collagen ECM binds elastase and metalloproteinases present in high concentrations in nonhealing wounds. ECM thereby acts to promote tissue collagen deposition.¹¹ In chronic wounds, ECM aids in restoring a favorable enzyme balance, supports granulation tissue formation, and supports epithelialization.¹² Data show that collagen wound dressings with the highest percentage of in vivo sourced collagen are the most efficient in inactivating proteases in wounds.¹³ 

Limited published data on PCMP exist. Five post-marketing studies have been performed. The study designs were noninterventional, observational cohort case series. 

A prospective case series of 41 patients treated for various wound types reported healing rates, mean percentage wound closure, and time to healing.¹⁴ At 12 weeks, most wounds (73.2%) had reduced in area from baseline, and many (63.4%) showed greater than or equal to 70% reduction in area compared with baseline. The time to CWC was 6.7 weeks; in contrast, the mean wound duration prior to treatment with PCMP was 103 weeks. In a case series of 5 patients, the average time to CWC after application of PCMP was 6.8 weeks.¹⁵ A different case series assessed 9 wounds managed with PCMP and reported healing in 6 wounds (66.7%) at an average of 10 weeks after application.¹⁶ In an interim analysis of 63 wounds, the median baseline wound area was 6.5 cm², and the mean wound duration at baseline was 4 months.⁵ CWC was achieved in 43 wounds (68.3%). The mean time to closure of wounds that healed after application of PCMP was 5 weeks. 

The industry-sponsored (Organogenesis, Inc) RESPOND Registry study was a prospective, noninterventional, observational cohort PCMP registry of 307 patients (67 VLUs, 62 DFUs, 45 PIs, 54 PSWs, and 79 other wounds).¹⁷ The Cox-derived wound closure rate was 73% at week 32. According to K-M analysis, the median time to wound closure was 17 weeks. In PCMP-treated wounds, the incidence of wounds with greater than 60% reduction in area and depth was 81% and 71%, respectively, and the incidence of wounds with greater than 75% reduction in volume was 85%.¹⁷ In all 5 noninterventional studies, PCMP reportedly was well tolerated, with no related systemic or localized AEs.

The primary objective of the current study was to assess the effectiveness of PCMP as measured by frequency of wound closure and median time to wound closure through 48 weeks. Secondary objectives were to assess changes in wound area, depth, and volume from baseline. 

Study design

This post-marketing, open-label, prospective, noninterventional, observational cohort study of 434 patients from 3 combined registries assessed the effectiveness of PCMP on various types of cutaneous wounds (VLUs, DFUs, PIs, PSWs, and wounds of other etiologies). The combined registry population analyses used data from the 3 individual PCMP registries listed in the next section. Baseline patient characteristics (age, sex, and number of wounds), wound characteristics (duration, area, depth, and volume), and treatment characteristics (number of PCMP applications, interval of time between applications) were collected by all wound care facilities that participated in each of the individual registries. Statistical testing (log-rank test for equality over strata of the 3 registries) was performed to show that data from the 3 registries were comparable and could be combined (Table). Additionally, Cox analyses were used to adjust for patient, wound, and treatment characteristics to ensure that patients and wounds across the registries were well matched. Eligible wounds were evaluated weekly for up to 48 weeks for wound size, healing, and change from baseline area, depth, and volume. 

Pooling of data for the combined registry population: adherence to ITT principles

The combined registry database was constructed by pooling raw patient data listings of patients treated with PCMP who participated in any 1 of the 3 independent PCMP registry studies. The 3 registry studies that were combined are the RESPOND Registry study that enrolled 307 patients at 28 sites,¹⁷ the NYU Langone Hospital PCMP registry that enrolled 86 patients at a single site (Long Island, NY), and the Northwell Health PCMP registry that enrolled 41 patients at a single site (Long Island, NY).¹⁴ No patient data were excluded from analyses, and ITT principles were strictly followed.¹⁸ The 3 registries were independent of each other. Industry (Organogenesis Inc [hereafter, Sponsor]) provided all raw patient data listing files from registry data transfers. The Sponsor had access to the patient data on the ITT population of 434 patients treated with PCMP. The 3 complete data sets for each individual registry were transferred from the Sponsor to Virtu Stat, Ltd, a third-party, independent statistical group that wrote unique coding and programming to extract the raw patient data (comma-separated value files) to generate new tables, listings, and files for the combined registry. ITT analyses were done on all patient data. These data were not from a federal agency or other clinical outcomes databases. Data were collected on every patient treated in the 3 registries (N = 434).

Primary and secondary analyses

The primary analyses were frequency and time to wound closure by or at 48 weeks. The secondary analyses were the incidence of wounds with greater than 60% reduction in wound area and depth from baseline, and greater than 75% reduction in wound volume from baseline.

Treatment

No study procedures were performed prior to patients signing an informed consent form. At the initial study visit, baseline wound measurements were recorded and PCMP was applied. At subsequent visits, PCMP was applied at the discretion of the investigator.

PCMP was applied, and standard of care appropriate to the wound type was used (eg, compression for VLU, offloading for DFU and PI, moist wound care for PSW). Standard of care treatments administered were defined as wound therapy regimens consistent with recognized wound care guidance documents and investigators’ institutional treatment practices.^18-22 

Patients

Patients with 1 wound of at least 0.2 cm² but less than 200 cm² were eligible for inclusion. Partial- and full-thickness wounds of various etiologies, including PIs, VLUs, DFUs, chronic vascular ulcers, open surgical wounds, or traumatic wounds with tissue loss were eligible for inclusion. Individuals were excluded if they were receiving concurrent treatment with other topical antimicrobials or skin substitute products, if they had received previous PCMP treatment for the study wound, or if they had a third-degree burn or a known sensitivity to any of the PCMP materials.

Ethics approvals

Written informed consent to participate in the registry in each of the 3 registries was obtained for all patients before any study procedures were performed. The registries comprising the combined registries were approved by the New York University Langone (Winthrop) Hospital Institutional Review Board, NYU Langone Hospital PCMP Registry Study, Northwell Health PCMP Registry Study, and Biomedical Research Alliance of New York. The RESPOND Registry Study was approved by Sterling Institutional Review Board (central IRB) (28 clinical sites). The 2 single-center studies and the PCMP registry were in compliance with current International Council for Harmonisation of Technical Requirements for Pharmaceuticals for Human Use Good Clinical Practice guidelines; US Food and Drug Administration Title 21 Code of Federal Regulations parts 11, 50, and 56; and applicable regulations.^23,24

Data collection

Recording data. Study personnel were responsible for collecting and recording the data in standardized eCRFs. Wounds were assessed for both wound area reduction from baseline and healing. Investigators (A.O., S.A.G.) assessed patients for untoward medical events and recorded all AEs and SAEs assessed by the principal investigator as being related to PCMP. These events were entered into the PCMP electronic data capture system.

Instruments. Wound measurements were obtained at each patient visit using a ruler. The maximum length, width, and depth of the index wound was measured before use of PCMP at each study visit through to the EOS. Investigators (A.O., S.A.G.) were instructed to measure the greatest length of the open wound end to end perpendicular (90° angle) to width. The width was measured at the widest width of the open wound side to side perpendicular to the length. Maximum depth was measured by moistening a cotton-tipped applicator with normal saline and placing the applicator tip in the deepest aspect of the wound, then using a ruler to measure the distance to the skin level. Cross-sectional area (cm²) was calculated by computerized algorithms using length and width measurements in the eCRFs. Wound volume (cm³) was calculated by computerized algorithms using length, width, and depth measurements in the eCRFs. Patient demographics were collected by site investigators (A.O., S.A.G.) using methods consistent with their respective institution’s approved standards and entered into the eCRFs. Digital cameras were used to take photographs before debridement, after debridement but before PCMP application, after PCMP treatment application, before removal of the PCMP (if applicable), after subsequent applications of PCMP, and upon healing (if it occurred). These images were entered into the eCRFs.

Statistical analysis

General. All statistical analyses were performed according to normal theory statistics and ITT principles. The 3 individual PCMP registries all shared a prospective, noninterventional study design wherein no treatment randomization was involved. Continuous data were summarized using descriptive statistics: count (n), mean, standard deviation, and median. Categorical data were described using frequencies and percentages. Analyses were performed using SAS version 9.4 (SAS Institute). To assess the effectiveness of PCMP, the following variables were measured: actual and percent change in wound area from baseline through to the EOS using wound measurements (wound length [cm], wound width [cm], maximum wound depth [mm], cross-sectional area [cm²], and wound volume [cm³]), and time to wound closure for each individual patient (if wound closure occurred) using dates and wound variables (baseline wound dates, wound characteristics, follow-up patient visit dates). 

Primary and secondary end points. The primary efficacy end point of the study was time to and frequency of CWC (ie, 100% wound closure) by or at 48 weeks. CWC was defined as 100% wound closure with epithelium. The secondary efficacy end points were the proportion of patients who achieved greater than 60% reduction in wound area from baseline (ie, >60% epithelialization [ie, a layer of epithelium visible on the wound surface]), the proportion of patients who achieved greater than 60% reduction in wound depth from baseline, and the proportion of patients who achieved greater than 75% reduction in wound volume from baseline. Frequency of wound closure (ie, wound closure rates) as a function of time in weeks was determined using the K-M time-to-event analysis (SAS version 9.4) through week 48 for all wounds, as well as subgroups of wounds treated, such as VLUs, DFUs, PIs, PSWs, and other (ie, wounds not meeting the diagnostic criteria for any of the identified wound types occurring in <5 patients). K-M time-to-event analysis was used to determine the median time to wound closure.

Cox proportional hazards regression. A stepwise Cox model that adjusted for patient demographics (ie, age, race, sex, body mass index) and wound characteristics (ie, area, depth, volume, ulcer duration at baseline) was performed to compute the adjusted wound closure rates for all wounds, VLUs, DFUs, PIs, PSWs, and other wounds. Identifying and including the race or ethnicity of patients in the Cox analyses may provide information about participants included in a study and the potential generalizability of the results of a study. The Cox analyses were used to estimate the frequency of wound closure at every time point at which patients were evaluated. All analyses were ITT and included data from all patients who received PCMP (Figures 1-4). Forward and backward selection for Cox modeling were used to analyze effects on healing in the combined registry ITT population. Covariates included the 3 individual registries, wound care facility, wound type, patient demographics, and wound characteristics (eg, baseline wound measurements). Computations of HRs and P values were done to identify statistically significant positive and negative risk factors for healing. All HRs were computed on the ITT population using all data collected from day 1 through week 48. 

Patient and wound characteristics

Each patient had 1 ulcer identified as the index ulcer for treatment with PCMP. Of the 434 wounds managed with PCMP, 95 were VLUs, 78 were DFUs, 90 were PIs, 73 were PSWs, and 98 were other wounds. Baseline patient demographics, wound characteristics, and treatment characteristics for all PCMP-treated patients are shown in the Table. The percentage of male patients was 53.2%, and the percentage of female patients was 46.8%. The median age was 69 years, and 66% of the PCMP-treated patients (n = 287) were 65 years or older. At the first treatment application, the mean wound area was 15.1 cm², the mean wound depth was 4.9 mm, and the mean wound volume was 7.2 cm³ (Table).

Treatment characteristics

The average number of treatment applications used in PCMP-treated wounds was 6.2. In total, 29% of PCMP patients received 3 or fewer applications. For patients who received multiple applications, the median interval between PCMP applications was 7.7 days (Table).

Primary and secondary end points

The primary end point (determined using Cox) of wound closure for all wounds (N = 434) was 51% at week 20 (n = 221), 56% at week 24 (n = 243), 62% at week 28 (n = 269), and 72% at week 48 (n = 312) (Figure 1). The incidence of achieving greater than a 60% reduction in baseline area and depth was 78% and 66%, respectively. The incidence of wounds with greater than a 75% reduction in baseline volume was 80% (Figure 2). The K-M median time to wound closure for all 434 wounds was 19 weeks (Figure 4).

Research question 1: Will combined registry population analyses show significant differences or comparability to prior PCMP analyses?

The combined registry population analyses showed results similar to those of prior PCMP registry studies. Although previous noninterventional PCMP studies differed with respect to sample size, proportion of wound types, follow-up times, and statistical methods, effectiveness results were consistent with the combined registry population analyses.

Subgroup analyses by wound type

By or at week 48 the wound closure frequency as determined using Cox analysis was 72% for all wounds, 72% for VLUs, 52% for DFUs, 63% for PIs, 95% for PSWs, and 67% for other wounds (Figure 3). The K-M median time to wound closure was 19 weeks for all wounds, 22 weeks for VLUs, 24 weeks for DFUs, 23 weeks for PIs, 12 weeks for PSWs, and 14 weeks for other wounds (Figure 4).

Research question 2: Will low variability and a large sample size allow for meaningful subgroup analyses?

Results of continuous data showed low variability consistent with the large sample size studied (N = 434). Meaningful results were demonstrated for wound type subgroup frequency (percentage) and median time (weeks) to healing (Figures 3, 4). In addition, Cox analyses used to test multiple patient and wound variables defining patient and wound subgroups showed results that were significant (P < .0004) with extremely low variability (standard error of the mean = 0.00523).

Cox analysis of risk factors for healing (predictors of healing)

Cox proportional hazards stepwise modeling of time and frequency of healing for the combined registry population (N = 434) showed that only baseline area (cm²) was a statistically significant risk factor for healing, with an HR of 0.982 (standard error of the mean = 0.00523; P < .0004). The HR of less than 1.0 shows that wounds with larger baseline areas heal poorer than wounds with smaller baseline areas. As the baseline area increases by 1 cm², the hazard (ie, probability of healing) decreases by 1.8% (HR = 0.982). Patient age, race, sex, and body mass index were not statistically significant predictors of healing (all P > .05). Baseline wound depth and volume were not statistically significant predictors of wound healing (P > .05 for both). Neither wound type nor treatment by wound care facility was significant (ie, P > .05 for all HRs). 

Research question 3: Will the use of Cox analysis show statistically significant risk factors (positive and negative predictors) for healing in the combined registry population?

Cox analysis showed that baseline wound area was a negative risk factor for healing that predicted a lower probability of wound closure as wound size increased. As a covariate, baseline wound area demonstrated an HR of 0.982 at a level of significance (P = .0004). While demonstrating statistical significance (P = .0004), wounds with larger baseline areas were not associated with appreciably slower healing rates. The computed HR of 0.982 indicates that the hazard (ie, healing) decreases by just 0.018 (1.8%) as the baseline area increases by 1 cm² (Cox standard unit calculation). Baseline wound area was identified as the only statistically significant covariate or factor in the combined registry population that affected frequency and median time to wound closure through to EOS. Risk factors that previously have been reported to affect healing did not reach statistical significance at an alpha level of .05.

Safety profile

Safety data were collected by active reporting through 48 weeks and entered into eCRFs. Safety assessments were made by the principal investigators (A.O., S.A.G.; their respective site) by evaluating patients’ signs and symptoms. Investigators were asked to record any untoward medical event attributable to PCMP. Only AEs and SAEs related to PCMP were reported in the electronic databases of each of the 3 individual registries. The combined registry patient safety data listings showed that there were no AEs or SAEs attributable to PCMP. Overall, AE and SAE reporting by attribution was consistent with demonstration of a favorable safety profile.

Research Question 4: Will longer duration patient follow-up result in previously unreported treatment-emergent AEs or SAEs attributable to PCMP?

Through 48 weeks, no treatment-emergent AEs or SAEs attributable to PCMP were reported. The safety profile was assessed as favorable.

This study presents new, unpublished data from de novo analyses performed on a combined registry data set of 434 patients through 48-week follow-up. Results were robust (low variability) and showed an overall healing rate of greater than 70%, with a time to healing of approximately 5 months. Subgroup analyses of chronic and acute wound types performed on sample sizes of 73 to 98 patients demonstrated healing percentages that ranged from 52% to 95%. This report expanded on previously published articles on PCMP in 3 primary areas: 40% more patients were analyzed (434 in the current study vs 307 patients in the RESPOND Registry), ITT K-M and adjusted Cox analyses of 48-week follow-up data were presented, and Cox-derived risk factors for healing were identified. Data from the combined registry analyses may prove useful to better characterize the clinical effectiveness of PCMP and inform future interventional study designs.

The effects on time to healing and frequency of healing tested by the Cox stepwise modeling method included the covariates of age (<65 years, ≥65 years), sex, registry (ie, RESPOND, NYU Langone, Northwell), baseline wound measurements (area, depth, volume), and wound type (VLU, DFU, PI, PSW, other). Only the baseline area showed statistically significant results. That baseline area was a statistically significant predictor of wound healing (HR = 0.982; P = .0004) was not a new or surprising finding. The unexpected results of the full stepwise Cox modelling conducted on combined registry data were which covariates were not statistically significant. Wound type and registry (ie, wound care facilities), age, and sex were neither positive nor negative risk factors for healing. These data appear to demonstrate consistency of PCMP treatment effects across multiple wound types and wound care facilities. The large sample size available for analysis allowed for meaningful numbers in subgroup analyses and robust results.

Randomized controlled trials are needed to add to the existing post-marketing data on PCMP. Comparative-effectiveness assessment research using real-world data that includes hundreds of centers and thousands of patients would provide useful, complementary evidence. Considering the expected increase in the number of older patients with chronic wounds, the potential for PCMP to aid in optimizing health care resources utilization in the elderly should be evaluated.^22-26 Future focused health economics and outcomes research studies are strongly encouraged. 

This study has limitations. All 3 registries were open-label, observational, noninterventional, cohort study designs. The primary limitation of interpreting data from noninterventional studies is the absence of a control group. Without a comparator, the possibility exists that similar clinical outcomes would have been observed with other skin substitutes or wound care therapies. Patient-reported outcomes data on quality of life, activities of daily living, and ability to work are warranted. The current study did not capture data on social isolation and depression, 2 factors that reportedly are possible determinants of wound healing.

This combined registry analysis of noninterventional, observational, cohort data of 434 patients treated with PCMP showed a 72% frequency of wound closure and a median time to healing of 19 weeks. Consistent results were demonstrated across all wound types. PCMP may be a useful adjunct in the management of various types of skin wounds. Future prospective, randomized controlled trials and controlled comparative-effectiveness research studies are warranted.