High Transepidermal Water Loss at the Site of Wound Closure Is Associated With Increased Recurrence of Diabetic Foot Ulcers: The NIDDK Diabetic Foot Consortium TEWL Study Free

Diabetic foot ulcers (DFUs) are the leading cause of nontraumatic amputations and lead to significant patient morbidity and mortality (1–4). Currently, the lifetime risk of DFUs is estimated to be 19–34%, with a 20% incidence of lower-extremity amputation (5). Although 59.3% of DFUs will have healed at 12 weeks, the subsequent 12-week incidence of wound recurrence, defined as the reopening of a previously healed wound at the same location, is estimated at 40% (6). Prevention of wound recurrence relies on an understanding of the factors that predict the reopening of a wound.

Functionally, a wound represents a breach of the body’s primary defensive barrier against external threats. The U.S. Food and Drug Administration (FDA) defines complete wound closure as complete wound surface reepithelialization without discernible exudate, drainage, or dressing requirements, confirmed at consecutive study/clinical visits 2 weeks apart (7). This current definition of wound closure emphasizes resurfacing of the wound but not restoration of skin barrier functionality. In preclinical studies of wound biofilm infection, we observed that wound surface reepithelialization was achieved but without restoration of the skin barrier function. In such cases, wound closure was associated with high transepidermal water loss (TEWL) readings at the site of closure (8,9). TEWL is a widely accepted objective measure of skin barrier function, measuring the amount of evaporative water lost from the body by noneccrine mechanisms diffusing across the stratum corneum (10). We hypothesized that poor wound-site skin barrier function as measured by high TEWL is associated with increased recurrence risk and that TEWL can be used as a predictive biomarker for wound recurrence in individuals with a newly closed DFU.

The primary objective of this study was to test whether TEWL measurement at the site of wound closure predicted recurrence of the index DFU by 16 weeks. Secondary objectives included 1) testing whether TEWL measurement at the site of wound closure predicted time to index DFU wound recurrence within 16 weeks and 2) assessing whether participant self-report of index DFU wound recurrence was concordant with clinician assessment of index DFU wound recurrence by 16 weeks. Additional exploratory analyses were conducted to understand whether TEWL measurements from different locations on the closed wound or normalization of TEWL using measurements from a contralateral site (i.e., with no wound) could be more predictive of DFU wound recurrence by 16 weeks.

This multicenter noninterventional study of participants with a recently healed DFU and enrolled in the National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK) Diabetic Foot Consortium (DFC) tested whether TEWL measurements could predict DFU recurrence (ClinicalTrials.gov identifier NCT04558775). The sites involved were Indiana University, University of Michigan, Stanford University, University of Arizona Tucson, University of California San Francisco, University of Pittsburgh, and University of Miami and the Data Coordinating Center at the University of Michigan. Enrollment occurred between June 2020 and December 2022. The study was conducted over a period of 2.5 years. This study protocol was approved by the DFC Steering Committee, Western Institutional Review Board, and NIDDK Observation and Safety Monitoring Board. Written informed consent was required from all participants.

The study design is shown in Supplementary Fig. 1. Briefly, adults with diabetes and a recently healed DFU were eligible for this study. Details of eligibility criteria are available at NCT04558775 and in Supplementary Fig. 1. Only one healed DFU was considered per participant. Visit 1 (baseline) TEWL measurements were obtained within 2 weeks of initial documentation of clinical wound closure. Confirmation of wound closure occurred 2 weeks later (visit 2). Participants whose wound had not remained closed were excused from further study participation. TEWL measurements were obtained from participants with confirmation of wound closure at visit 2. These participants were then contacted weekly by phone for up to 16 weeks to assess for wound recurrence. Clinical assessment of wound status occurred after week 16 for participants reporting no wound recurrence during the follow-up period (visit 3a) or within 4 weeks of the first report of wound recurrence occurring before 16 weeks of follow-up (visit 3b) (Supplementary Fig. 1).

Sex, ethnicity, race, age, and social factors (education level, occupation, tobacco use history, and social support for wound dressing changes and transportation) were collected at visit 1. Baseline wound characteristics and related medical conditions were also collected for each participant, including diabetes type and duration, duration of study ulcer, occurrence of previous DFUs, current DFU, therapies to treat the study ulcer (including offloading), wound location, previous amputation, ankle brachial index, absolute ankle pressure, toe pressure, transcutaneous oxygen pressure, HbA_1c, skin disease, peripheral vascular disease, and any foot deformities.

TEWL measurements (g · m⁻² · h⁻¹) were taken using a commercial device (DermaLab; Cortex Technology, Aalborg, Denmark) as previously published (8,11). Briefly, a portable, handheld, noninvasive probe was placed on the participant’s closed wound, and the readings were recorded on the attached console. Probes were sent to the manufacturer for calibration once per year and when subject to mechanical shock (e.g., accidentally dropped). Documentation of calibration was maintained at each study site. Five measurements were obtained (at center and four corners following a clock-face pattern [3, 6, 9, and 12 o’clock]) over the closed wound site. Each measurement took ∼10–30 s. A reference (control) TEWL measurement was taken from intact skin at an anatomically matched site on the participant’s contralateral foot (three consecutive measurements at a central point). The TEWL value at the closed DFU site was a continuous measurement for evaluation as a potential biomarker to predict recurrence. The primary TEWL measurement for analysis was the center value measured on the index DFU at visit 1.

Study staff at all sites were trained in the use of the TEWL measurement device at the start of the study using a standard protocol. The TEWL Biomarker Analysis Unit (BAU; Indiana University and University of Pittsburgh) developed training videos in English and Spanish that were disseminated to the sites for implementation. The BAU was responsible for oversight of the training and quality control checks on data generated from individual sites with the assistance of the Data Coordinating Center.

Wound closure was defined using the FDA guidance (7) as complete wound closure where there was skin reepithelialization without drainage or dressing requirements at two consecutive study visits 2 weeks apart. The primary end point of participant self-reported DFU wound recurrence by 16 weeks was defined for the index DFU by participant report during weekly phone calls with research staff. DFU wound recurrence was defined as affirmative responses to both “Do you see any discharge from the closed wound?” and the follow-up question “Has the wound reopened?” The weekly participant assessment of DFU wound recurrence was assessed primarily by phone call but also was assessed during in-clinic visits before clinician assessment if the participant had a clinic visit for another cause. Because of concerns about the validity of participant assessment of wound recurrence, all participants viewed a training video (Supplementary Video 1) that reviewed signs of wound recurrence and demonstrated how to assess their wound site for recurrence.

The planned statistical approach to develop and validate a model to examine the predictive ability of TEWL as a biomarker of DFU wound recurrence used area under the curve (AUC) from the receiver operating characteristic (ROC) analysis as the measure of model performance. We planned to develop the model using a training set and test it in a validation set. Sample size estimation was based on comparing the additional contribution of TEWL relative to a base model of known predictors using the AUC. Previous literature (12–15) suggested that wound size, patient-reported duration of the wound, and anatomic depth correctly predicted wound recurrence with an AUC of 0.68. We hypothesized that adding TEWL to the base model would increase the AUC to 0.80, assuming 30% of study participants would achieve DFU wound recurrence in 16 weeks and 30% would drop out. A sample of 268 participants for the training data set would achieve 80% power under these assumptions using a two-sided z test at a 5% type I error rate. An additional 134 participants (half the size of the training set) were required for a conservative validation data set. Therefore, the total enrollment target was 402 participants.

Participant characteristics, including visit 1 TEWL measurements, were summarized overall and by wound recurrence by 16 weeks using means and SDs for continuous variables and counts and percentages for categorical variables. The planned statistical approach for the primary objective split participants into training and validation cohorts by a simple random sampling algorithm, with each cohort containing two-thirds and one-third of the participants, respectively. A logistic regression for DFU wound recurrence was fit on the training cohort, controlling for self-reported wound duration of the index DFU in months. This model was then extended to include the visit 1 TEWL biomarker. A comparison of these models was made using a z test based on ROC analysis with AUC as the primary accuracy index.

The training and validation cohorts were pooled for additional post hoc assessments of the primary objective and assessments of the secondary and exploratory objectives. Differences between those with and without wound recurrence were compared using a two-sample t test and box plots to visually aid in the comparison. Low-high TEWL cutoffs were determined by ROC analysis with the Youden index (16) from a simple logistic regression model for DFU wound recurrence as a function of visit 1 TEWL. The percentages of low and high values for visit 1 TEWL were summarized descriptively and with a bar diagram for those with and without wound recurrence by 16 weeks postclosure. The odds of DFU wound recurrence for those with high TEWL compared with those with low TEWL was assessed with this simple logistic regression model. The simple logistic regression model for DFU wound recurrence as a function of visit 1 TEWL was expanded to include potential confounding characteristics based on literature review (6,17–19). These included self-reported wound duration, DFU history, skin disease, peripheral vascular disease, most recent HbA_1c within 90 days of screening, years of education, tobacco use history, social support for dressing changes and transportation, and structural deformities of the affected foot.

The secondary objective of time to DFU wound recurrence was summarized by calculating mean time to recurrence and using a Kaplan-Meier curve by low-high TEWL status. Time to recurrence was defined as the time from confirmed healing to recurrence for those whose wound recurred or last follow-up date for those who dropped out or whose wound remained healed, respectively. A Cox proportional hazards model was used to investigate whether visit 1 low-high TEWL was associated with time to DFU wound recurrence within 16 weeks. Yule’s Q was used to assess the secondary objective of concordance between participant self-report of DFU wound recurrence and clinician assessment. Pearson correlation coefficients were used to assess the relationship between each of the four periphery TEWL measurement locations and the center TEWL. Model assumptions were investigated where appropriate for all statistical analyses.

All statistical analyses were conducted in SAS (version 9.4 for Windows; SAS Institute, Cary, NC) and R (version 4.3.1). Complete case analyses were performed, with no imputation for missing data except when week-16 participant-reported healing status was missing. In this case, we used the visit 3 clinician assessment if it occurred within the week-16 protocol-prescribed visit window. No multiplicity adjustments were made. We provide the magnitude of effects with measures of variability (e.g., SDs or 95% CIs), and P values for secondary and exploratory end points should be interpreted with caution. Statistical testing was conducted at the 0.05 significance level using two-tailed tests; two-sided P values are reported.

Data collected for the TEWL study are available to the public through the NIDDK Central Repository at https://repository.niddk.nih.gov/home/.

From 17 June 2020 to 16 December 2022, 476 patients consented to study participation, with 418 participants enrolled with a confirmed closed DFU (Supplementary Figs. 1 and 2). Of these, 368 (88%) provided wound recurrence outcome data. Fifty participants (12%) discontinued early and did not provide wound recurrence outcome data, a majority of whom were lost to follow-up (n = 44 [11%]). Characteristics of participants who discontinued early were comparable to those of the main study population.

Selected baseline characteristics of the enrolled participants, overall and stratified by DFU recurrence and dropout status, are listed in Table 1. Most of the enrolled participants had type 2 diabetes; were male, non-Hispanic or -Latino, and White; and were between 35 and 75 years of age. Approximately half reported a history of tobacco use. Additional baseline characteristics are listed in Supplementary Table 1.

Wound recurrence before or up to week 16 occurred in 79 participants (21.5%). The mean of TEWL values from the center of the healed DFU at visit 1 was higher for those with recurrence compared with those without (27.71 g · m⁻² · h⁻¹ [SD 14.75] vs. 22.72 g · m⁻² · h⁻¹ [SD 13.93], respectively; P = 0.006) (Fig. 1A). A sensitivity of 0.43, specificity of 0.78, and AUC of 0.60 were observed from the logistic regression for DFU wound recurrence controlling for the visit 1 TEWL value (Fig. 1B). The Youden index of 0.21 from this model was used to identify low (TEWL <30.05) and high (TEWL ≥30.05) TEWL cutoffs. The percentage of participants with high TEWL reporting wound recurrence by 16 weeks was 35% (34 of 98) versus 17% (45 of 270) of those with low TEWL (Fig. 1C). The odds of recurrence for participants with high TEWL was 2.66× (95% CI 1.57, 4.49; P < 0.001) that for patients with low TEWL. Exploratory analyses that adjusted the visit 1 TEWL measurement with the contralateral limb TEWL measurement did not yield stronger associations between TEWL and DFU recurrence despite accounting for within-patient variability (Supplementary Table 2). TEWL measurements at specific locations on the wound were highly correlated with the TEWL measurement at the center of the closed wound, supporting our decision to use the center of the closed wound for all analyses (Supplementary Table 3). We tested the association of TEWL (as a continuous variable) with wound recurrence in a multivariable logistic regression, controlling for other potential explanatory factors. The adjusted odds of recurrence was increased by 2% for each 1-unit increase in TEWL (odds ratio 1.02; 95% CI 1.00, 1.04; P = 0.05) (Supplementary Table 4). We explored other potential predictors of wound recurrence in the base model, in addition to wound duration, and tested the addition of TEWL to this model using the pool cohort of participants; similar results were obtained (Supplementary Table 4). The lack of social support for dressing changes was associated (odds ratio 2.42; 95% CI 1.03, 5.69; P < 0.05) with recurrence in this cohort (n = 340).

We tested the hypothesis that adding TEWL to the base model (unverified self-reported wound duration) would significantly improve the performance of the prediction of wound recurrence. A logistic regression for DFU wound recurrence was fit, controlling for self-reported wound duration of the index DFU in months, and conducted on the training cohort composed of 244 randomly selected participants (Supplementary Table 5). This model was then extended to include the visit 1 TEWL biomarker. An AUC of 0.63 was observed for the logistic regression for DFU wound recurrence, controlling for self-reported wound duration of the index DFU, and an AUC of 0.65 was observed when the model was extended to include the visit 1 TEWL biomarker (ΔAUC 0.02; 95% CI −0.04, 0.08; P = 0.26) (Supplementary Table 5).

The Kaplan-Meier plot of time to wound recurrence by low or high TEWL status is presented in Fig. 2. The mean time to wound recurrence was 104.8 (SE 1.5) days for those in the low TEWL group and 91.4 (SE 3.7) days for the high TEWL group (Fig. 2). The Cox model indicated the hazard of wound recurrence for those with high TEWL was 2.43× the hazard for those with low TEWL (hazard ratio 2.43; 95% CI 1.56, 3.79; P < 0.001).

Participant self-reported wound recurrence was highly concordant with clinician assessment of wound recurrence (Yule’s Q 0.99; SE 0.003) (Table 2). Of the 75 participants who reported wound recurrence by 16 weeks postclosure, 70 (93%) received a clinician assessment of wound recurrence. Similarly, of the 269 participants who did not report wound recurrence by 16 weeks postclosure, 258 (96%) received a clinician assessment of no wound recurrence.

DFUs are challenging complications for individuals with diabetes, given their long disease course with slow healing and high rates of recurrence (6). Literature reports a large range of 20–40% or higher recurrence rates within 1 year that increase with time in the DFU population (6,20). These increased recurrence rates reflect the race/ethnicity of the population studied, geographical location, outpatient versus inpatient population studied, and diverse treatment modalities and could be influenced by environmental, societal, and other factors. Individuals with a closed DFU are viewed as being in a state of remission rather than healed (6). For patients whose DFU recurs, prognosis is poor (21). DFU recurrence often leads to amputation and death (20). Understanding the factors that underlie DFU recurrence and identifying patients at high risk of recurrence are central to developing novel strategies to prevent DFU recurrence and increase remission days. This study translated mechanistic preclinical results to the first test of skin barrier function, measured by TEWL, as being associated with DFU recurrence in humans. A TEWL value >30 g · m⁻² · h⁻¹ was significantly associated with DFU recurrence in this study. At a wound site, clinically determined to be closed by current standards, an observation of TEWL >30 g · m⁻² · h⁻¹ should be considered a red flag, justifying additional clinical attention such that the wound may remain closed. A closed wound site with high TEWL should be managed to minimize the risks of infection and inflammation, safeguarded from mechanical pressure, and kept moist to allow tissue repair to progress (22,23). This work, highlighting the clinical significance of functional wound healing, provides the rationale for prospective studies identifying effective management of the closed DFU to extend ulcer-free days in remission, including the study of patient-focused measures (22,23).

Postepithelialization ulcer recurrence in individuals with diabetes points toward the possibility of defective wound closure. TEWL is widely used to assess the barrier function integrity of the skin (24,25). Physiological or low TEWL values indicate an intact and healthy skin barrier function, whereas any defect in barrier integrity manifests as high TEWL. Compromised barrier function of the skin may let external threats such as allergens and pathogens into the body, thus elevating systemic health risk factors (26). TEWL readings have been found to be useful in a variety of clinical applications. In humans, there is evidence of burn wound closure without complete reestablishment of barrier function, as manifested by high TEWL of the repaired skin (27). Atopic dermatitis is characterized by an impaired epidermal barrier, manifested as high TEWL, which could be associated with sensitization to allergens (28–30). In atopic dermatitis as well as in psoriasis, temperature and TEWL values may help clinicians determine disease severity and select patients who need intensive treatment (31). TEWL measurement detects food anaphylaxis in real time before clinical symptoms (32). The observation in this work that high TEWL was associated with DFU wound recurrence is consistent with reports that poor quality of wound closure is likely to increase the risk of wound recurrence (33).

In all DFUs, nonhealing wounds are prone to infections, a process facilitated by impaired immune responses in such populations. It has been proposed that such infections may be complicated by biofilm formation that worsens wound prognosis (34). Preclinical experimental studies testing the biological factors underlying wound recurrence have revealed that faulty wound reepithelialization, such as complications caused by biofilm infection, may cause tissue repair to be functionally compromised such that restoration of barrier function does not take place even in a fully reepithelialized wound (8,9). For example, biofilm infection induces miR-146a and miR-106b in the wound-edge tissue, which in turn silence tight junction proteins ZO-1 and ZO-2, causing leaky reepithelialization (8). Other examples include depletion of essential skin lipids and disintegration of the extracellular matrix by biofilm pathogen–dependent factors (9,35). Clinically, among the risk factors for the recurrence of DFUs, multidrug-resistant bacterial infection has been recognized as a significant component (21). The quality of tissue repair dictates the biomechanical properties of the healed tissue, influencing its vulnerability to wound recurrence. Based on the findings of this work, barrier integrity of the repaired skin, as quantified through TEWL, emerges as one objective noninvasive point-of-care measure of the quality of DFU closure.

Key findings of the current study include the observations that the center TEWL of closed wounds was significantly higher in wounds that recurred; time to wound recurrence was significantly shorter, and such wounds had a 2.34× higher risk of recurrence compared with wounds with low TEWL. The a priori planned statistical approach indicated that TEWL measured at visit 1 at the center of the closed DFU did not provide sufficient additional predictive benefit over unverified patient-reported wound duration or patient-reported wound duration with other potential predictors (12–14). It is not uncommon for patients with chronic wounds to be unable to recall when the wound first occurred (36,37). Patient-reported wound duration cannot be verified. However, wound duration is one of several risk factors that were identified in a meta-analysis of the current literature on the recurrence of DFUs. Others include diabetic peripheral neuropathy, peripheral vascular disease, poor glucose control, plantar location of an ulcer, osteomyelitis, smoking, history of amputation, multidrug-resistant bacterial infection, and Wagner grade 3 or 4 (21). Based on the multifactor analysis in this study, we found that lacking support for dressing changes was also associated with a high risk of recurrence. An objective TEWL measurement as a predictor of recurrence risk has an advantage over patient-reported duration. TEWL is actual observed data. In contrast, wound duration relies on patient recall and may be viewed as proxy data (38). Compared with subjective patient-recall parameters, objective measurements are superior because of higher reliability and validity and reduced susceptibility to bias and error. Therefore, although the validation component of this first study testing the association of high TEWL with DFU recurrence had some limitations, results of the observational component were decisive and consistent with preclinical mechanistic studies that provide a scientific rationale (8,9).

An interesting outcome of the study was the observation that participant self-reported recurrence was highly concordant with clinician assessment of wound recurrence. Although the literature is varied as it relates to concordance between participant and clinically reported outcomes, this gives us confidence in the use of patient self-reported assessment of recurrence. The use of participant training in a clinical setting to assess wounds could meaningfully impact getting people to care early in recurrence.

A primary limitation of the study was related to the training and auditing/quality check of study staff performing the TEWL measurement. All sites were given a one-time training at the launch of the study at the site. Any new staff who were onboarded after this were not trained/certified by the BAU but rather trained by staff who had completed the initial training by the BAU. It is possible that the accuracy of the data could have been affected by this limitation. A recent pilot study conducted at a single center on chronic wounds of multiple etiology had two trained personnel conduct the TEWL measurements through the entire course of the study for all enrolled participants and showed a tight correlation with recurrence outcome (39). Wound area measurements were not performed, because the study started with closed wounds.

“Ulcer-free, hospital-free, activity-rich days” are a major patient-centered goal in the management of DFUs. This work bridges the gap between clinical outcomes (e.g., wound closure) and patient-centered outcomes (e.g., remission days). It highlights the critical need to revisit current wound healing clinical end points to include restoration of barrier function at the site of closure.

Acknowledgments. The authors acknowledge the efforts of Catherine Martin, Dr. Aristidis Veves, and Dr. Kellen Chen in the review of the entire manuscript and project officers Dr. Henry Burch and Dr. Yan Li in study guidance. All TEWL devices used for this study were purchased from Cortex Technologies. Results from preclinical studies by C.K.S. et al. (PMIDs 24771509 and 37395583) served as a basis for a pilot clinical study; this pilot clinical study (principal investigators C.K.S. and G.C.G.), funded by DiaComp (collaborative funding program), guided the current DFC study. The authors acknowledge all DFC members and the Data Coordinating Center for contributions to the development of the study design and implementation.

R.P.-B. is an editor of Diabetes Care but was not involved in any of the decisions regarding review of the manuscript or its acceptance.

Funding. This project was made possible by NIDDK grants U01DK119099 (C.K.S., G.M.G., S.R., M.S., L.T., and S.S.M.-S.), U24DK122927 (J.J., G.K., and C.S.), U01DK119100 (M.S.C.), U01DK119083 (C.H. and R.P.-B.), U01DK119094 (G.C.G.), U01DK119085, and U01DK119102.

Any opinions, findings, conclusions, or recommendations expressed in this article are those of the authors and do not necessarily reflect the views of the funding agencies.

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. C.K.S., G.M.G., S.R., M.S., L.T., S.S.M.-S., M.S.C., C.H., T.L.Z.J., R.P.-B., G.K., C.S., and G.C.G. substantially collaborated on the development of the manuscript. J.J., G.K., and C.S. analyzed the data as part of the Data Coordinating Center. All authors provided final approval of this submitted version to be published. C.S. and C.K.S. are the guarantors of this work and, as such, had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Prior Presentation. The findings of this study were presented at the 83rd Scientific Sessions of the American Diabetes Association, San Diego, CA, 23–26 June 2023; Gordon Research Conference on Skin Barrier Function, Waterville, NH, 6–11 August 2023; C2Ship Think Tank Virtual Meeting, 23 August 2023; National Assembly of the American Orthotic and Prosthetic Association, Indianapolis, IN, 6–9 September 2023; and Diabetic Lower Extremity Symposium hosted by Harvard University, Boston, MA, 7–8 November 2023.

Handling Editors. The journal editor responsible for overseeing the review of the manuscript was M. Sue Kirkman.

This study was conducted by the NIDDK DFC. Leads on the study include the following: Indiana University (Chandan K. Sen, PhD, FNAI; Sashwati Roy, PhD; Gayle Gordillo, MD, FACS [currently at University of Pittsburgh]), Indiana University Biomarker Analysis Unit (Chandan K. Sen, PhD, FNAI; Mithun Sinha, PhD; Lava Timsina, PhD), University of Arizona (Geoffrey C. Gurtner, MD, FACS; Kellen Chen, PhD), University of Michigan (Rodica Pop-Busui, MD, PhD; Crystal Holmes, DPM; Katherine Gallagher, MD; Kayvan Najarian, PhD; Brian Schmidt, DPM), University of California San Francisco (Michael S. Conte, MD; Alexander Reyzelman, DPM; Shant Vartanian, MD; Monara Dini DPM; Dean Schillinger, MD), University of Miami (Robert S. Kirsner, MD, PhD; Hadar Lev-Tov, MD, MS; Rivka C. Stone, MD, PhD; Marjana Tomic-Canic, PhD), and University of Pittsburgh (Peter Rubin, MD, FACS). Data Coordinating Center: University of Michigan (Cathie Spino, ScD). Members of the Observation and Safety Monitoring Board include the following: Robert Frykberg, DPM, MPH; Cynthia Chang, PhD; Raul Guzman, MD; and L.J. Wei, PhD.