Intent to Treat Analysis of the Primary and Secondary Outcomes for the ODINN Intact Fish Skin Graft for Deep Diabetic Foot Wounds Trial

ABSTRACT

There is a significant need for trials that evaluate the treatment of University of Texas (UT) grade 2 and 3 diabetic foot ulcers (bone, joint, or tendon exposed wounds). We undertook a trial looking at the effect of intact fish skin graft (IFSG) on these deep and difficult‐to‐heal ulcers. 262 patients Intent to Treat (ITT) patients with UT grade 2 and 3 DFUs were randomised to receive intact fish skin graft (IFSG) or a standardised treatment (SOC) that adhered to the International Working Group on the Diabetic Foot (IWGDF) guidelines. The secondary endpoints that were measured included wound area reduction (WAR), healing rates at 20 and 24 weeks; closure rates by UT grade, perfusion, quality of life, pain reduction and IFSG safety. We report ITT (all randomised) (mITT previosly reported) The (WAR) at 12 weeks was 65.53% for IFSG versus 30.82% for SOC (p = 0.007). UT 2 wounds (60% of total) exhibited a closure rate of 47% versus 23% at 16 weeks for IFSG versus SOC (p = 0.0033). Target wound infections were comparable (39 vs. 37) and major outcomes were comparable during the 24 week period (target‐limb amputations 8% vs. 7%). Time‐to‐heal favoured IFSG (restricted mean to 24 weeks 17.31 vs. 19.37 weeks; KM/log‐rank significant; Cox HR 1.59). The in the treatment of deep complex diabetic foot wounds the addition of IFSG significantly improved the number of patients with total wound closure as well as the time to wound closure without increased risk of complications. This improvement in total wound closure and time to wound closure was noted across prior amputation status, quality of perfusion, and UT grade.

Key Points

• Intact fish skin graft.

• University of Texas 2 and 3 diabetic foot wounds.

• Deep complex diabetic foot wounds.

• Multi national DFU Trial.

1. Introduction

All investigators in this space are aware that most diabetic foot ulcer trials encompass Wagner 1 and 2 or University of Texas grade 1 ulcers. It is well recognised that deep diabetic foot ulcers (DFUs), University of Texas (UT) grade 2 and 3 are not well studied. Therefore we decided to evaluate if wild‐caught North Atlantic cod—intact fish skin graft (IFSG) (Kerecis, Marigen, Isafjodur, IS) which reconstitutes a dermal‐like wound bed could significantly impact deeper and more difficult‐to‐heal (UT 2 and 3) DFUs.

Therefore, a protocol was designed to treat a population of deep (joint capsule, tendon or bone) DFUs with IFSG versus a regional standard of care. This was submitted for an Innovation Horizon 2020 European Commission Fast Track grant. Upon the successful grant application, we initiated a prospective randomised controlled trial in May 2020 to be conducted in 4 European Union countries. Trial enrollment began in France (180 patients) with the remaining patients enrolled in Italy (41), Germany (40), and Sweden [1].

In October 2024, primary outcomes of the study were published. This previously published manuscript [1] reported on the primary outcome of the modified intent to treat (mITT) analysis of 255 patients (7 patients excluded). The coherts were randomised in a 1:1 ratio to receive IFSG or a treatment that adhered to the International Working Group on the Diabetic Foot guidelines [2] (standard of care/SOC) during 14 weeks of assigned treatment applications. Patients were followed over a total of 24 weeks. The results of the primary endpoint revealed that healing was achieved in 44% of IFSG patients, in comparison to 26% in the SOC group patients (p < 0.001, unadjusted). This represented a 66% relative improvement in the closure rate of deep diabetic foot wounds at 16 weeks [1].

At the inception of the study in 2020 the secondary objectives originally included: “change in grade of ulcer that alters the prognosis and makes the ulcer ready for skin graft or standard‐of‐care treatment; quality of life; pain reduction; healing trajectories; decrease in wound surface; time to ulcer healing; percentage of healed ulcers at 20 weeks (checkpoint only); healed ulcers at 24 weeks (checkpoint only), and the percentage of ulcers healed 50% or more at 12 weeks”. Cost analysis was to be performed in which global cost of wound treatment of the two arms and projected costs for non‐healed ulcers were calculated. These costs we are going to be country specific. In addition, two other secondary endpoints were proposed: time to start standard‐of‐care treatment as judged by the treating clinician and a blinded panel of clinicians (a hypothetical determination, since treatment was set to continue with IFSG); time to perform a split thickness graft or other skin covering operation as judged by a blinded panel of clinicians (again, a hypothetical determination, since treatment was still set to continue with IFSG). Neither of these last two secondary endpoints turned out to be applicable and were quickly removed from subsequent versions of the protocol. It should be noted that by January 2021 the secondary endpoints in version 18 of the protocol were narrowed down to: “time to healing or to autologous skin grafting; quality of life; and wound area reduction at 4, 8 and 12 weeks”. In addition, cost‐effectiveness as calculated by global cost of wound treatment of the two arms and projected costs for non‐healed ulcers by country, and of course safety outcomes continued to be evaluated as a portion of standard trial protocol.

The outcomes and analysis of the secondary endpoints of this very large prospective study have not been published. Many large prospective randomised trials tend not to publish their secondary endpoints, especially if their primary endpoints are significant, but we believe that the natural history of the patients with the deep diabetic foot wound will be interesting to the wound care practitioner. In addition, the reasons some of these endpoints were not analysed will be helpful to the design of future similar studies.

2. Materials and Methods

The ODINN trial was a multicenter, open‐label, randomised, parallel group clinical trial conducted in 15 tertiary care centers with diabetic foot units across France, Italy, Germany, and Sweden. The primary endpoint was the percentage of wounds healed at 16 weeks with follow‐up continuing through 24 weeks. The trial protocol was approved by the relevant ethics committees at each center and followed the tenets of the Declaration of Helsinki. All patients enrolled had the opportunity to undergo informed consent and ask appropriate questions in their native language as appropriate to the country in which the patient was enrolled in the trial. The trial was registered on clinialtrials.gov (NCT04257370) and all patients were provided with written informed consent in their native language.

Blinding the patients and the primary researchers in such a study is fundamentally impossible. The process of applying a tissue to wound requires a very distinct procedure from standard moist wound care. In addition, moist wound care requires a cadence of a minimum of 3 times a week dressing changes, if not daily as practiced in the study, whereas the instructions for use for IFSG make the dressing change cadence a once a week event. Both of these realities of care make blinding the patient fundamentally impossible. The fact that many of these visits occurred in the community made having a blinded evaluator separate from the care provider completely impractical. In addition, during the COVID 19 pandemic, exposing patients to more research personnel than absolutely necessary would have been ethically wrong.

Therefore, an external blinded panel was the only way to blind the primary endpoint evaluation. In order to provide the best blinding possible, a secondary blinded adjudication committee was utilised. This 3 clinician blinded committee received all patient pictures without any indication of treatment received. Photographs were identified by site and patient identifier. No treatment identifying data was provided to the committee. The committee reviewed the photographs separately, and when consensus was obtained, the wound was considered closed. If consensus was not obtained, the wound was considered not healed by the committee and treatment was continued.

Inclusion was the presence of a diabetic foot ulcer (DFU) penetrating to bone, joint or tendon (University of Texas grade 2 or 3). Randomization was done per country in a randomised block size between 4 and 10. With a stratification instrument, treatment with IFSG or standard of care was evenly divided in patients with ABPI from 0.6 to 0.9 and in patients with perfusion above 0.9; stratification also took place by wound associated with amputation or not. The randomization took place after inclusion assessment. The definition of non‐healing was < 20% reduction in ulcer area in the previous month. The patients had initial treatment applications of IFSG versus SOC in an environment similar to, or actually, an operating room. However, not all initial applications occurred in the operating room setting. IFSG was applied on week 1 to 6, then on week 8, 10, 12 and 14. In the standard of care (SOC) arm, the dressing was changed daily (once per day) according to the local wound management protocol. The application of a suitable dressing chosen by the investigator included hydrocolloid, alginate, hydrocellular, charcoal, silver, petrolatum gauze, or pressure dressings; the dressing regime as well as the offloading conformed to the International Working Group on the Diabetic Foot's Guidelines (IWGDF). The quality and concordance of off‐loading was confirmed by wound care nurses on a weekly basis.

Primary on‐site clinical visits occurred at week 0, 7, 16, and 24. One hundred and eighty of the 262 patients followed this schedule, with the other visits occurring in the community. The remaining 82 patients were seen weekly at the primary clinical site on week's 1–16, 20, and 24. Therefore, this study represented a community care, decentralised wound model, especially in France, where a much larger proportion of therapy occurred in the community.

Wound area was measured as length × width. Although additional measurement methods were used in the trial, the primary and consistent method across all sites was length × width. Accordingly, a post hoc analysis by UT grade was conducted, while perfusion status was included in the predefined statistical analysis plan. Each ulcer was photographed before debridement, after debridement, and after placement of the IFSG (if applicable), with a centimetre scale and a de‐identified patient ID visible in each image. An acetate tracing (transparent outline of the ulcer) was also made and photographed using the same scale and identifier. The largest diameter and its perpendicular axis were measured at each visit. If complete closure occurred before week 16, it was confirmed at a clinical visit 2 weeks later. When closure occurred at week 16, 20, or 24, confirmation was obtained through a follow‐up phone call 4 weeks after the initial documentation of closure.

Quality of life was assessed at 3 timepoints in the study, (week 0, 16 and 24). Quality of Life (QOL) was evaluated using the EQ‐5D‐5L questionnaire in France, while in Germany and Italy QOL was evaluated using the Wound‐QoL‐17 questionnaire. Pain was assessed at every visit using a Visual Analog Scale (VAS) of 0–100.

2.1. Patients

The study participants were 18 and older with diabetes mellitus and a lower limb wound below the malleolus penetrating to the tendon or capsule (UT grade 2) or to the bone or joint (UT grade 3) that had been present for at least 1 month. Acute amputation wounds that occurred as part of the treatment of an ulcer that had been open for greater than 30 days or amputation wounds that had not closed in 30 days could be included. Non‐ischemic to moderately ischemic wounds were also included and assessed by an ankle to brachial systolic pressure index (ABPI) of at least 0.6. Notably, patients with active but treated osteomyelitis could be included within the study as well.

Patients were excluded if they had active, unmanaged osteomyelitis, immune deficiency, autoimmune disease, or had undergone arterial reconstruction within the previous month. Additionally, those who had known fish allergies were excluded. In addition, these patients receiving treatment with systemic glucocorticoids or other treatment known to delay wound healing met the exclusion criteria. Finally, pregnant or breastfeeding women and women who had future pregnancy plans were excluded from the study.

There were no statistically significant differences between the IFSG and SOC groups I (see Table 1). Of note 39% of the IFSG and 36% of the SOC groups had ankle‐brachial indices < 0.9. Both groups had approximately 59% of their respective cohort with a UT grade 2 ulcer; and ~50% of the wounds in each group were post‐amputation.

TABLE 1.

Patient characteristics.

Demographics	IFSG	SOC
Sex
Male	104 (81%)	95 (75%)
Female	24 (19%)	31 (25%)
Age, years	67.3 (12.2)	66.8 (11.0)
Clinical characteristics
Weight, kg	84.8 (20.3)	89.2 (21.0)
BMI	27.7 (6.0)	29.3 (6.3)
History of myocardial infarction	30 (23%)	31 (25%)
History of renal or kidney disease	47 (37%)	51 (40%)
Type 2 diabetes	109 (86%)	118 (94%)
Diabetes duration, years	21.1 (13.2)	20.4 (12.9)
Haemoglobin A1c	8.2 (SD 4.5)	8.5 (SD 7.0)
Ankle brachial pressure index (ABPI)	0.95, 0.92 (0.2)	0.96, 0.94 (0.2)
ABPI category
Normal	54 (42%)	60 (48%)
0.91 to 0.99	25 (20%)	21 (17%)
0.60 to 0.90	50 (39%)	45 (36%)
Wound duration, days	30.7 (SD 46.6)	27.3 (SD 34.5)
Amputation wound	64 (50%)	66 (52%)
UT Grade
Grade 2	75 (58%)	75 (59%)
Grade 3	54 (42%)	51 (41%)

Initial wound size is interesting; in the entire 4 country cohort the initial wound size for the IFSG median was 7.18 cm² versus 5.71 cm² (p = 0.059). Further if one adjusts for outlayers the median becomes 5.20 cm² versus 4.95 cm² (p = 0.87). In the French cohort, which encompassed 180 of the 262 total patients the mean wound area for the IFSG cohort was 10.48 cm² versus 9.76 cm² in the SOC arm. This was a near statistical difference in the mean and median in various cohorts, but statistically they did start the same size wound.

The drop out (defined as no data points available) rate in the IFSG cohort at 16 weeks was 0.75% (1:133), while at 24 weeks 2.25% (3:133); while the dropout rate in the SOC arm was 0.77% (1:129) at 16 weeks, and 2.33% (3:129) at 24 weeks. Therefore there was absolutely no difference in the dropout rate. In the intent to treat (ITT) analysis there were 262 patients, who signed informed consent. Eighty five percent (223) completed the treatment period the entire study period. Of those that discontinued the study prematurely 16 (9%) were in the IFSG group and 23 (13%) in the SOC arm (Table 2).

TABLE 2.

The reason for study withdrawal or drop out.

Non‐missing	16 (IFSG)	23 (SOC)	39 (15%)
Unexpected amputation [n (%)]	6 (37.5)	7 (30.4)	13 (33.3)
Serious adverse event other than amputation [n (%)]	3 (18.8)	3 (13.0)	6 (15.4)
Withdrawal of consent [n (%)]	1 (6.3)	1 (4.3)	2 (5.1)
Wound area increased by 25% over the study period [n (%)]	3 (18.8)	0 (0.0)	3 (7.7)
Lost to follow‐up [n (%)]	0 (0.0)	5 (21.7)	5 (12.8)
Investigator/sponsor/decision data safety/Monitoring Board [n (%)]	3 (18.8)	1 (4.3)	4 (10.3)
Other [n (%)]	0 (0.0)	6 (26.1)	6 (15)

See Consort diagram

2.2. Statistical Methods

2.2.1. Healed Ulcers a t16, 20 and 24 Weeks

Previously reported healing time at 20 and 24 weeks was documented in the primary endpoint manuscript [1]. We analysed the number of healed ulcers at 16, 20, and 24 weeks using Fischer's Exact test looking at completely epithelialized wounds. For patients that did not complete the trial, in the ITT week healing endpoints, subjects without documented healing at the relevant timepoint were conservatively counted as not healed (including early discontinuations/missing assessments). For time‐to‐closure analyses, subjects without confirmed healing were censored at last available assessment (or at the administrative censoring timepoint) (Table 3).

TABLE 3.

Healed Ulcers at 20 and 24 weeks (ITT).

Endpoint	IFSG	SOC	Unadjusted OR (95% CI)	Unadjusted p‐value
Proportion healed at 20 weeks	43.6% (58/133)	28.7% (37/129)	1.92 (1.18–3.11)	0.012
Proportion healed at 24 weeks	51.1% (68/133)	34.1% (44/129)	2.03 (1.27–3.24)	0.005

2.2.2. Wound Area Reduction

Utilising the week 2 starting area, wound area reduction was calculated by comparing the wound area in 12,16, 20 and 24‐week timepoints. If the wound area reached 0 before the set timepoint, the wound was considered to have been reduced by 100% in area. Subjects who dropped out of the study before the time points were not recorded as 100% reduced in area; therefore, they were not included for this analysis. Likewise, subjects who remained and were missing baseline data were excluded. The Wound Percent Area Reduction (WAR) at 12,16, 20 and 24 weeks for SOC versus FSG‐patients were tested using the Mann–Whitney Test.

2.2.3. Co Factor Effect on Wound Healing

Welch's t‐test, which estimates a mean difference allowing unequal variances and is generally more reliable than the equal‐variance t‐test.

2.2.4. Pain

Differences in pain were measured using the Wilcoxon rank‐sum test.

2.2.5. Quality of Life

Analysis was conducted utilising the Mann–Whitney U Test and Cronbach's Alpha.

2.2.6. Healing Trajectory

The time to healing was analysed using Kaplan–Meier survival curves, a log‐rank test, and the Cox proportional hazards model, adjusting for sex, country, and age at baseline.

3. Results

3.1. Intent to Treat Analysis Primary Endpoint; Percentage Completely Closed at 16 Weeks in Each Cohort—Including All Randomised Patients

The intent to treat analysis including all 262 patients who were randomised; showed a complete closure rate in the intact fish skin group of 40.6% (54/133) versus 23.3% (30/129) in the standard of care arm. (p = 0.003) Further analysis of the intent to treat population, at 20 weeks; showed a complete closure rate in the intact fish skin group of 43.6% (58/133) versus 28.7% (37/129) in the standard of care arm. (p = 0.012) and 24 weeks; showed a complete closure rate in the intact fish skin group of 51.1% (68/133) versus 34.1% (44/129) in the standard of care arm. (p = 0.005) (Figure 1) This presents a 78% relative improvement in the healing rate at 16 weeks for the group who was treated with intact fish skin graft. This relative improvement is actually greater than the relative improvement in the mITT analysis.

FIGURE 1.

Intent to Treat proportion healed over time.

3.2. Primary Endpoints Previously Published

The previous publication reported a modified intent to treat analysis. It must be noted that there is no universal definition for what constitutes a modified intention‐to‐treat group. Different studies use different criteria for exclusion, leading to varied definitions across trials.

The standard ITT principle maintains prognostic balance from randomization by including all randomised participants regardless of subsequent protocol deviations. An mITT analysis deviates from this by excluding specific subgroups.

Common reasons for exclusion in an mITT analysis include patients who do not receive at least one dose of the assigned therapy or patients who violate the protocol in a major way, such as lacking a necessary baseline assessment or failing to meet eligibility criteria at baseline.

Seven patients were excluded in the modified intent to treat analysis secondary to 5 patients the wound did not qualify, 1 patient was randomised after closure of the study, and in 1 case the site did not provide care.

When the 7 patients who should not be included in the overall analysis were excluded, the primary endpoint was the proportion of DFUs healed at 16 weeks; the results only changed minimally. The mITT results revealed that healing was achieved in 44% of intact fish skin graft patients in comparison to 26% in the standard of care group patients (p < 0.001, unadjusted). This represented a relative 66% improvement in wound closure at 16 weeks [1]. Whereas, the ITT analysis reveals a 78% relative improvement in closure rate.

3.3. The ITT Proportion of Wounds Closed at 20 and 24 Weeks

Healed ulcers at 20 and 24 weeks were previously reported in the primary manuscript. More detailed results are analysed here for completeness. At 20 and 24 weeks, similar to 16 weeks, there was a statistically significant difference, favouring IFSG over SOC. The relative improvement in healing at 20 weeks was 41%, and that 6‐month relative improvement in healing was 46%, despite not having any product applied after week 14.

3.4. Time to Healing

The restricted mean time to healing was 17.31 weeks (95% CI 15.5, 18.67) compared to SOC which was 19.37 weeks (95% CI, 18.09, 20.66). The log‐rank test p‐value was 0.032. In the intent to treat analysis, the Kaplan Meier curve seen here (Figure 2) demonstrates a log‐rank p‐value of 0.00448, meaning that there is a statistically significant difference between IFSG and SOC for their healing‐time distribution, with the results favouring IFSG. The IFSG median was reached (meaning > 50% healed) at 24 weeks, while the SOC median was not reached (less than half of the patients healed).

FIGURE 2.

Time to heal intent to treat analysis.

It also must be noted that in patients who did not undergo any bone resection, the impact of IFSG at 16 was significant, with 50% of that cohort closed at 16 weeks, in comparison to 24.4% in the SOC arm that did not have bone resection.

Change in UT grade, percentage of ulcers healed 50% or more at 12 weeks, and wounds with a WAR of 80% at 16 weeks and WAR effect on the QoL‐17.

Upon review of the data, it became apparent that the UT grades were not collected by all sites at all timepoints. This data was collected more routinely in the French cohort. Therefore, this a priori secondary endpoint was an interesting attempt to collect a data set which is not routinely collected. In addition, this is not a clinically useful data point. In addition, “ready for skin grafting” was not a practical endpoint.

The percentage of wound area reduction (WAR) at 12 weeks (62.53%) was statistically significant (p = 0.007) between IFSG and SOC (30.82%) (Figure 3). 50.4% of wounds treated with intact fish skin had reduced greater than 80% in wound area at 16 weeks versus 35.1% of wounds in the standard of care arm (p = 0.031) (Table 4).

FIGURE 3.

Wound area reduction over time ITT (LOCF, winsorized means).

TABLE 4.

IFSG versus SOC safety data.

	IFSG (133)	SOC (129)
Deaths	3 (2%)	6 (5%)
Minor amputation	10 (8%)	8 (6%)
Major amputation	0 (0%)	1 (1%)
Number of patients with SAE	56 (42%)	49 (38%)
Number of patients with wound related re hospitalizations	21 (16%)	21 (16%)

Of note in the German and Italian subset in which the Wound QoL‐17 was collected, an adjusted analysis was fit into linear regression models of QoL‐17 with predictors of WAR > 50%, and WAR > 80%, gender, UT score (3 vs. 2), and age; only complete cases (255) were used. Reaching WAR > 50% or WAR > 80% seems to be associated with higher QoL‐1. Therefore, since IFSG increases the chance and decreases the time to having both > 50% and > 80% WAR, hitting those WAR thresholds, treatment with IFSG improves the patients' QOL in these settings.

Of note, there were 75 UT grade 2 DFUs in each group, with a closure rate of 47% versus 23% at 16 weeks for IFSG versus SOC (p = 0.0033). While there were 54 UT grade 3 ulcers in the IFSG group with a closure rate of 33.3% at 16 weeks and 51 in the SOC arm with a closure rate of 19.6%. This represented a relative improvement in healing in the UT 2 group of 104%, and in the UT 3 group a 70% improvement in healing. It should also be noted that the UT grade had the greatest effect on likelihood of the wound to heal. With UT 3 predicting non healing compared to UT 2.

3.5. Quality of Life

In France the EQ‐5D‐5L questionnaire was utilised, this questionnaire measures five dimensions of health: mobility, self‐care, usual activities, pain/discomfort, and anxiety/depression, with answers being rated on a scale from 1 (no problems) to 5 (extreme problems). In Italy Germany in Sweden Wound QoL‐17 was collected. This questionnaire contains 17 items with a 5‐point Likert scale ranging from 0 (not at all) to 4 (very much). If a minimum of 75% of items are answered by the patient, a total score (arithmetic mean) can be calculated from all items. In the same way, three subscales (body: items 1–5; psyche: items 6–10; everyday life: items 11–16) can be calculated. Currently, the data within both of these subsets are being further analysed. In such a large longitudinal study, also going on during the time of the COVID pandemic, enhanced statistical rigour is being added to this analysis. It should be noted that during the course of the study fewer participants answered the quality of life questionnaires which is relatively standard per studies of this nature.

3.6. Pain

Pain was reported by study participants in both groups. There were no significant differences in the distribution of pain at 0 (p = 0.8795), 16 (p = 0.784), or 24 weeks (p = 0.414). Pain assessed was not expected to be different due to diabetic‐related neuropathy being a complication for patients with diabetic foot ulcers. The large range of the pain assessment tool (0–100) may not be representative of differences in pain scores.

3.7. Cost

This is country specific and beyond the scope of this article.

3.8. Post Hoc Subgroup Analysis

The effect of IFSG application on patients with suboptimal perfusion and in patients that had undergone previous amputation are interesting areas of exploration. This is especially additive to the body of literature when the patients in these categories treated with SOC are included. The effect of either therapy as affected by body mass index (BMI) was also analysed.

3.9. Status Post Amputation

A total of 130 patients were enrolled after bone resection (amputation). All of these amputations/bone resections (including partial calcanectomy) were performed to treat an ulcer which had been present for greater than 30 days (93), or a bone resection wound which had not healed for 30 days (37). Of the ninety‐three (93) (72%), of these were after acute bone resection, 46 patients were UT 2 (21 SOC, 25 IFSG), and 47 patients were UT 3 (27 SOC, 20 IFSG), prior to the acute bone resection.

At week 16, those UT 2/UT 3 resections demonstrated 31.1% (n = 14) wound healing for IFSG treated wounds in comparison to 16.7% (n = 8) wound healing in the SOC patients. This trend continued during week 20 with 37.8% (n = 17) wound healing in the IFSG group versus 23.4% (n = 11) wound healing in the SOC group. Similarly, this pattern continued at week 24 with 44.2% (n = 19) wound healing in the IFSG group in comparison to 29.8% (n = 14) wound healing in the SOC group. Statistical analysis was not utilised in this setting as the trial was not designed to answer this question.

3.10. Effect of Limited Perfusion

Another relevant subgroup analysis is based upon ABPI at baseline. ABPI was stratified as > 0.6, < 0.9, and > 0.9 among the various treatment groups (IFSG vs. SOC). Although this study was not powered for this level of stratification, we present descriptive statistics. At 16 weeks those patients with a normal APBI (> 0.9) that were treated with IFSG demonstrated a closure rate of 52% versus those treated with SOC 24%. While those with an ABPI < 0.9 but > 0.6 that were treated with IFSG was 24% at 16 weeks, versus SOC which was 16%. Therefore, as one would expect, recognised patients with DFU that are well vascularized or have been revascularized do better than those who are not. Statistical analysis was not utilised in this setting as the trial was not designed to answer this question.

3.11. Effect of BMI

The mean BMI of the patient in the intact fish skin group was 27.7 (26.8 median) while the standard of care arm was 29.3 (28.9 median). BMI was not associated with healing by 24 weeks in either arm after adjustment for age, sex, and baseline wound area (IFSG p = 0.821; SOC p = 0.160). The BMI × Arm interaction was not significant (p = 0.330). Similar results were observed in the adjusted Cox model. Comparisons of healed proportions across BMI categories (< 21 vs. ≥ 21; < 30 vs. ≥ 30; < 40 vs. ≥ 40) within each arm were non‐significant (Fisher's exact p > 0.33 across contrasts). Therefore, in the entire 4 country cohort, BMI did not have an effect on wound healing.

3.12. Safety

The study took place during the COVID‐19 pandemic, which increased the numbers of overall infections in both groups.

There was no statistical difference in adverse event (AE's) nor serious adverse event (SAE) between the cohort across the 4 countries. In the IFSG group (n = 133), 79 patients (59.8%) experienced at least one adverse event (AE), and 56 patients (42.4%) had at least one SAE. Seven patients (5.3%) developed osteitis within 4 weeks of initiating treatment.

In the SOC group (n = 129), 67 patients (51.9%) experienced at least one AE, and 49 patients (38.0%) had at least one SAE, and osteitis occurred in 6 patients (4.7%) within 4 weeks of treatment initiation.

There were a total of 18 patients who underwent a minor amputation during the study. Ten patients underwent minor amputation in the IFSG versus 8 in the SOC arm. While 1 patient underwent a major amputation in the standard of care arm, no patient has had a major amputation in the IFSG arm.

Wound related infections were documented 39 versus 37 times in IFSG versus SOC. However, these were noted in 30 patients in the IFSG while they were noted in 27 patients in the SOC arm. Wound related re‐hospitalisation occurred 26 times in 21 patients in the IFSG arm and 28 times in 21 patients in the SOC arm. None of these differences were statistically significant. Of note in patients that followed the protocol (PP) the number of patients that had wound related SAEs diminished to 9 in the IFSG versus 23 in the SOC group.

There were a total of 9 deaths in the study. All recorded deaths occurred in France. There were 3 deaths in the IFSG group: one patient died of diabetic foot ischemia, and two patients died from unknown causes reported as “faintness” and “patient death” by the investigators. There were 6 deaths (6.6%) in the SOC group: one patient died from COVID‐19 infection, another from bronchial superinfection, one following cardiac decompensation on rhythmic and ischemic heart disease with urinary superinfection, one from acute exacerbation of chronic kidney failure leading to heart failure, one due to a fall resulting in a per trochanteric fracture in the context of sepsis with bacteremia, and one from an unspecified cause reported as “death” by the investigator.

Therefore, of serious adverse events, there were twice as many deaths in the standard of care arm as in the intact fish skin arm of the study. There were a very similar number of amputations in the two cohorts, with 1 major amputation in the standard of care arm and none in the intact fish skin group arm. Wound‐related hospitalizations occurred in the same number of patients in both arms and serious adverse events occurred in the same number of patients in the 2 arms, excluding the deaths. If patients and caregivers followed the protocol, the group treated with IFSG had a significantly lower number of infections than the SOC arm.

4. Discussion

In the secondary endpoint analysis, multiple descriptors of wound healing have been provided. Obtaining the variety of data, beside the primary outcome, such as proportion healed at week 20 and 24, percentage wound area reduction, and time to healing all require research resources and infrastructure. In the lessons learned of this wound type study, it would appear that the investigators would keep percentage closure at 4 months as a primary end‐point, as these deeper wounds have meaningful endpoints at this timeframe.

However, they would probably approve a design that excluded further application from anyone who had not reduced in wound area by 40% at 8 weeks, including a cross‐over model for standard of care. In addition, the new protocol would allow for continued application of active therapy (IFSG) to those who were responding out to 22 weeks. In addition, in the current era, the inclusion of digital imaging at all visits would potentially enhance the data quality and ease the analysis. Finally, even if handled as a second follow‐up study, 1 year follow‐up to check for freedom from major or minor amputation and death would have enhanced this study. In addition, this added data would enhance the understanding of IFSG application's impact on the long‐term health and well‐being of patients with deep diabetic foot ulcers.

There is much information to be gleaned from the SOC arm in such a large prospective randomised controlled trial. This is the largest prospective data collection set of diabetic patients with such deep ulcers. We are now able to say that UT grade 2 ulcers receiving IWGDF guideline adherent therapy have a 28% chance of closure at 6 months; remarkably, UT grade 3 ulcers receiving similar care closed at 35%. This counterintuitive finding may have to do with more extensive surgical intervention and uncovering or exposing medullary bone in the UT 3 lesions with aggressive debridement. In the standard of care arm in patients who had undergone bone resection, 16.7% were closed at 16 weeks, and only 29.8% at 6 months.

While we are not presenting much in the way of quality of life scores in this paper, it is important to note that the tools we used may not be ideal. Others have shown that the Wound QoL‐17 has significant variables and the questions may be skewed at times, but more useful towards a United States population [3]. As to quality of life measures it was interesting to see that having such deep diabetic foot wounds, and even possibly undergoing surgery, did not impact the Wound QOL −17 scores much at baseline, with the majority of patients reporting that their disease impacted them “not at all” or “moderately.” While changes in QoL mostly favoured IFSG, with some areas of QoL achieving or nearing statistical significance, it is likely that they do not represent a clinically relevant change. In these neuropathic ulcers, it is not particularly surprising to the clinician not to see changes in pain as the patients do not have much pain at baseline. Of course, a less biased and more recent quality of life measure such as the NeuroQoL tool may well be better for patients suffering from diabetic foot wounds and may have helped the investigators in this study show a greater benefit from wound closure. Of note the QoL tools were chosen to meet European regulatory recognition in 2017 [4].

The healing trajectory was notably different between groups in that, of those patients who healed, those treated with IFSG healed on average 2 weeks earlier than those treated with SOC. This faster healing in the IFSG‐group led to a total of 46 additional weeks of closed wounds for the 262 patient cohort at 16 weeks, in addition to 42 more weeks of closed wounds for the cohort at 6 months. In this UT 2 and 3 wound population one must note that the SOC (no active therapy) results in 38/100 wounds being closed at half a year, while those treated with IFSG have a greater than one in two chance of being closed at 6 months. There is a 69% relative improvement in those patients treated with IFSG. In disciplines other than wound care, such relative improvements would be considered profound. The wound healing trajectory recognises that no patient received IFSG following the 10th application during week 14 of their care. While the study question is yet to be determined, careful examination of the healing trajectory stimulates further postulations regarding IFSG applications and whether additional applications would have further enhanced wound closure rates at 5 and 6 months.

For patients who had undergone amputation for an ulcer that had been present for > 30 days, the healing effects were significant in those treated with the IFSG in comparison to the SOC treatment group. Although ABPI measures and wound healing outcomes were not designed to demonstrate statistically powered relevance, patients in groups receiving IFSG continued to notice advantages in wound closure in comparison to those patients within the SOC group.

It was also interesting to see the reduction in secondary endpoints that were expected to be analysed from the inception of the study at 11 to 6 by the end of the study. Some of these were dropped as being impractical. For example, no one at any of the eventual sites would employ skin grafting for DFU closure. Therefore, this was an artificial endpoint for the study investigators. While we could have attempted to extrapolate when this would have been appropriate by photography, this would have turned into a retrospective analysis. In addition, downgrading ulcers from UT 3 to UT 2, and so forth, while technically possible is not clinically utilised, and while included in the first version of the protocol, it was soon abandoned as it was noted to be clinically naïve, although documented to a large extent in France. Neither of these endpoints were practical in the long term, and we would not recommend including them in the future. Future studies will benefit by analysing: closure at 4 months (16 weeks), freedom from amputation and death at 1 year, and freedom from recurrence. Other endpoints should include re‐admission from hospitalisation for DFU. (Which we do include in this safety data but was not actually an analysed endpoint).

In the final protocol, secondary endpoints collected were: time to healing, quality of life, wound area reduction at 4, 8 and 12 weeks, safety evaluated by the number of reported adverse events, and a hypothetical point of a time to switch over to SOC. This speaks to an evolution in the protocol over time in the reality left collecting all secondary endpoints throughout the duration of a trial.

As to a plan to report cost, the cost calculation requires different models in each of the four countries in which the trial occurred. At the time of the study, none of the countries had IFSG commercially available in the outpatient setting. In addition, the cost of an unhealed DFU is calculated differently in all 4 countries. One has to note the cost of community care in France in comparison to the other participating countries where care was delivered primarily in the clinic. Based upon these variables, a total cost analysis of IFSG for the treatment of these deep DFUs is currently undergoing country‐by‐country portfolio development.

Safety data is notable for a significant difference in the death rate between the 2 groups, but due to the fact that this study went on during the COVID‐19 pandemic, the authors do not feel there is much that can be extrapolated from this. There did not appear to be a significant advantage in freedom from further operation or re‐hospitalisation in either group.

5. Conclusion

This large multi country, multicenter prospective randomised trial was effective in assessing most of its potential secondary endpoints while providing significant epidemiology data in regards to how University of Texas grade 2 and 3 ulcers behave under International Working Group on Diabetic Foot guidelines. The addition of IFSG to the patient's care plan not only significantly improved the number of total patients' wounds closed but decreased the time to closure and had an impact on patients with amputations and with poor perfusion. At this point in time, intact fish skin graft has a very large UT 2 and 3 DFU trial and a well‐recognised UT 1 DFU trial [5] therefore, at this point in time there is not a significant reason to pursue further prospective randomised clinical research in UT 1–3 diabetic foot wounds.

Funding

This work was supported by European Union's Horizon 2020 research and innovation program, 878896. Kerecis LLC, Reykjavik IS.

Conflicts of Interest

Dr John C. Lantis is the director of the scientific advisory Council for Kerecis LLC the maker of intact fish skin graft.

Data Availability Statement

The data that support the findings of this study are available from Kerecis LLC. Restrictions apply to the availability of these data, which were used under licence for this study. Data are available from the author(s) with the permission of Kerecis LLC.