Skip to main content
NCBI home page
eClinicalMedicine logoLink to eClinicalMedicine
. 2026 Jul 20;98:104086. doi: 10.1016/j.eclinm.2026.104086

Recurrence after healing in pyoderma gangrenosum: a prospective registry-based cohort study

Jordan Gillespie a,b, Jaclyn Roland-McGowan a,b, Katelyn Downey a,b, Emile Latour c, Alex G Ortega-Loayza b,∗
PMCID: PMC13393349  PMID: 42495104

Summary

Background

Pyoderma gangrenosum is a rare neutrophilic dermatosis with potential for relapse after healing, yet recurrence in real-world settings remains unknown. We aimed to evaluate recurrence patterns to inform clinical practice and future research.

Methods

Data were abstracted from a uniquely, prospectively maintained pyoderma gangrenosum registry. A cohort of patients with pyoderma gangrenosum seen between December 1st, 2019 and March 1st, 2026 who healed within this timeframe was identified. Two subgroups were then created: those who remained healed, and those who had a recurrence. Descriptive statistics summarised recurrence characteristics; Kaplan–Meier methods estimated a cumulative incidence of recurrence, and Cox proportional hazards modelling identified predictors of recurrence.

Findings

Of 116 patients with healed pyoderma gangrenosum, 41 (35%) experienced a recurrence, generating a recurrence proportion of 35·3% (95% CI 26·7–44·8). Nominal associations with recurrence on univariable analysis included history of diabetes mellitus (hazard ratio 2·1, 95% CI 1·1–4·0, p = 0·024) and ever-smoking (1·8, 1·0–3·4, p = 0·057). Baseline characteristics differing between recurrence and non-recurrence groups included age (p = 0·033), history of diabetes (p = 0·0060), and ulcer age at presentation (p = 0·046). Cumulative recurrence risk was calculated as 27% (95% CI 18·1–35·0) at 1 year and 41·3% (95% CI 30·4–50·5) at 5 years.

Interpretation

Pyoderma gangrenosum recurred in approximately a third of patients. A history of diabetes showed a borderline association with recurrence, while the univariable signal for smoking was attenuated after adjustment. Recurrence risk remains elevated five years after initial healing, and peristomal pyoderma gangrenosum is associated with higher recurrence rates, suggesting that maintenance therapy may be necessary for certain patient populations after achieving ulcer closure. Our data highlight the need for prospective evaluation of maintenance approaches. Future work will focus on individual recurrence predictors.

Funding

This is an unfunded study.

Keywords: Pyoderma gangrenosum, Wounds, Recurrence, Real-world evidence

Research in context.

Evidence before this study

To characterise what is known about recurrence after healing in pyoderma gangrenosum, we reviewed the existing literature for recurrence rates, risk factors for recurrence, and clinical outcomes. All years were searched and the search term used was “pyoderma gangrenosum AND recurrence.” Previous studies report highly variable recurrence rates (approximately 15%–61%), and much of the available evidence was derived from small cohorts where recurrence, and its predictors, was not the focus of the study. While clinical factors such as systemic disease (e.g., inflammatory bowel disease) have been associated with recurrence, there remains limited prospective, real-world data quantifying recurrence risk over an extended period or identifying predictors.

Added value of this study

This study provides prospective, real-world evidence from a uniquely maintained 5-year pyoderma gangrenosum registry, offering the most comprehensive evaluations of recurrence after healing in pyoderma gangrenosum to date. Among 116 patients with healed pyoderma gangrenosum, 41 (35%) had recurrence, with cumulative pyoderma gangrenosum recurrence risk reaching 41·3% at 5 years. This study characterises recurrence timing, morphology, and distribution, demonstrating that most recurrences are local, and occur within the first-year post-healing. It also identifies clinically relevant associations, including diabetes and smoking. This work provides longitudinal recurrence-free survival estimates and emphasises gaps in follow-up care during the post-healing period.

Implications of all the available evidence

These findings reinforce that pyoderma gangrenosum is a chronic, relapsing condition requiring long-term management beyond initial ulcer healing. The substantial recurrence risk supports the need for maintenance of medical therapy, close clinical follow-up after healing, and proactive management of modifiable comorbidities such as diabetes. Overall, this study advances the evidence base by providing prospective data to inform clinical decision-making, guide patient counselling, inform clinical trial design, and support future research on identification of individual risk factors of recurrence and long-term disease control in pyoderma gangrenosum.

Introduction

Pyoderma gangrenosum is a rare, inflammatory, neutrophilic dermatosis characterised by painful, ulcerative lesions with significant morbidity and a high propensity for recurrence. Recurrence estimates vary widely in the literature, with reported rates ranging from 15·0% to 61·0% based on one clinical trial, two reviews, and two cohort studies.1, 2, 3, 4, 5 The risk of recurrence appears to be heavily influenced by several factors, including uncontrolled underlying systemic comorbidities classically associated with pyoderma gangrenosum (e.g. inflammatory bowel disease [IBD]).6, 7, 8 One study reported that patients with haematologic disorders have the strongest association with pyoderma gangrenosum recurrence and poor outcomes.9 Moreover, the classic ulcerative subtype has been associated with greater disease severity and the highest recurrence risk.10 However, it is unclear whether this association reflects the inherent nature of the ulcerative subtype or simply its disproportionate representation in the literature.10

Pyoderma gangrenosum has been associated with immune dysfunction, both innate and adaptive, and patients frequently require immunosuppressive treatments.11 This can prove difficult as previous case reports have reported recurrent pyoderma gangrenosum caused by vaccines and immunotherapy.12,13 Notably, pathergy, where minor trauma incites new lesions, can act as a potent trigger for post-healing ulcers, and surgical interventions have been recognised as triggers for pyoderma gangrenosum recurrence.11 Furthermore, rare auto-inflammatory syndromes such as PAPA (pyogenic arthritis, pyoderma gangrenosum, and acne) and PASH (pyoderma gangrenosum, acne, and suppurative hidradenitis) have been shown to have high rates of ulcer recurrence.14 Taken together, the current evidence shows that several factors contribute to the propensity for recurrence, though few studies have quantified which factors have the most influence or which patients may be at higher risk. As such, this study aims to determine the frequency of pyoderma gangrenosum recurrence, identify clinical risk factors, evaluate predictors of recurrence episodes, explore the outcomes between recurrent and non-recurrent patients, and compare recurrence patterns between classic and peristomal pyoderma gangrenosum using a well-characterised registry database, PyGaS, which is a prospectively maintained pyoderma gangrenosum registry at Oregon Health & Science University.

Methods

Ethics

This study was conducted using patient data from PyGaS. The study was approved by the Institutional Review Board (#00027619). Informed consent was obtained from all patients included in the registry.

From PyGaS, we isolated a cohort of patients who were seen between December 1st, 2019 and March 1st, 2026 with clinician-confirmed healing of their pyoderma gangrenosum target ulcer, which was defined as the largest ulcer being treated. Initial pyoderma gangrenosum diagnosis, regardless of subtype, was made using histopathology when available and was clinically informed by PARACELSUS scoring.15 The target ulcer was determined by the physician at the baseline visit and data regarding it was abstracted from clinical documentation. Patients were then categorised into those who remained healed by March 1st, 2026, and those who had recurred. Recurrence was defined as the development of new ulcerative pyoderma gangrenosum lesions after complete resolution of the initial ulcer, confirmed by clinician evaluation. Distal recurrence was defined as the development of a new pyoderma gangrenosum lesion at a separate anatomical site after clinician-confirmed healing of the initial target ulcer, while local recurrence was defined as the reappearance of pyoderma gangrenosum at the same anatomical site as the initial target ulcer, including within previous scar, following clinician-confirmed complete healing.

Data were abstracted from our registry and electronic medical records and included demographics, underlying comorbidities, ulcer characteristics, current and previous treatments, clinical outcomes, and recurrence episodes. Ulcers were graded using the physician global assessment (PGA), which uses the following scale: 0 = Total resolution of target ulcer with no signs of active pyoderma gangrenosum (clear); 1 = Almost completely healed target ulcer with only minimal signs of active pyoderma gangrenosum (minimal disease); 2 = Evidence of target ulcer healing which involves at least 50% of ulcer/ulcer margin (mild); 3 = Evidence of target ulcer healing which involves less than 50% of ulcer/margin (moderate); 4 = No evidence of target ulcer healing (severe).

Statistics

Descriptive statistics were used to summarise patient, ulcer, and recurrence characteristics. Descriptive statistics were computed on available cases, with the extent of missingness reported in the tables. Uncontrolled diabetes was not considered as a candidate covariate because it was missing in more than half of patients, with missingness unlikely to be at random. Among the candidate covariates considered for selection, missingness was 0·3%, and selection was performed on complete cases. The covariates retained in the final Cox model were fully observed, so the final model included the entire sample with no exclusions for missing data.

Continuous variables were compared using t-tests (or Wilcoxon rank-sum where appropriate); categorical variables were compared using Chi-square or Fisher's exact tests, as appropriate. Time to recurrence was analysed using Kaplan–Meier methods. Cumulative incidence of recurrence was estimated from the Kaplan–Meier recurrence-free survival curve as 1-S(t), where S(t) is the Kaplan–Meier estimate of recurrence-free survival. Deaths were treated as censoring events. Given the low rate of death before recurrence (n = 3), the 1–S(t) estimate is expected to closely approximate the cumulative incidence function. A competing-risks sensitivity analysis confirmed that censoring them did not materially affect the recurrence estimates. Associations between baseline patient- and ulcer-level characteristics and time to recurrence were analysed using Cox proportional hazards regression. The proportional hazards assumption was assessed using scaled Schoenfeld residuals; no violations were detected.

Candidate predictors were selected from the baseline demographic, clinical, comorbidity, ulcer, and treatment variables. To reduce redundancy, alternate versions of the same construct were not entered simultaneously. Variables with sparse counts were flagged for caution. Univariable Cox proportional hazards models were fit for each candidate predictor. Variables meeting the screening criterion of p < 0·25 at the variable level (likelihood ratio test) were carried forward for consideration in multivariable modelling. Variables retained after univariable screening were entered into an initial multivariable Cox model. Purposeful selection was then applied by iteratively removing the variable with the highest variable-level likelihood ratio test p-value, while checking whether removal meaningfully changed the estimated coefficients of the remaining variables. Variables for the final multivariable model were retained if they remained statistically important or acted as potential confounders. All analyses were performed using R version 4·5·4 (R Foundation for Statistical Computing, Vienna, Austria).16 Cox proportional hazards models and Kaplan–Meier estimates were fit using the survival package (v3·8–6), with survival curves produced using ggsurvfit (v1·2·0). Purposeful variable selection followed the approach described by Hosmer, Lemeshow, and Sturdivant.17 Statistical significance was set at p < 0·050.

Role of funding source

AGO-L is supported by National Institutes of Health, National Institute of Arthritis and Musculoskeletal and Skin Diseases, R01 AR083110. KD and JRM are supported by the National Center for Advancing Translational Sciences of the National Institutes of Health under award number UL1TR002369. The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH. None of the funders had any role in study design, data collection, data analyses, interpretation, or writing of report.

Results

116 patients in the PyGaS registry were found to have achieved complete target ulcer healing during follow-up. Of these, 89 (77%) had classic pyoderma gangrenosum, 21 (18%) had peristomal pyoderma gangrenosum, four (3%) had vegetative pyoderma gangrenosum, one (<1%) had bullous pyoderma gangrenosum, and one (<1%) had subcutaneous pyoderma gangrenosum. Recurrence occurred in 41 patients (35%); of these, 30 (73%) had classic pyoderma gangrenosum, nine (22%) had peristomal pyoderma gangrenosum, and two (5%) had vegetative pyoderma gangrenosum. One patient had recurrence of both classic and peristomal pyoderma gangrenosum at different times, but they were allocated to the peristomal group as this was their initial target ulcer type. The remaining 75 patients had no recurrence during the follow-up period, yielding a recurrence proportion of 35·3% (95% CI 26·7–44·8). Demographic and clinical characteristics of the overall cohort, the recurrence cohort, and the non-recurrence cohort are reported in Table S1. In the overall cohort, the mean age was 55·3 years, 99 (85%) of 116 were White, 88 (76%) were female, and 47 (41%) received a systemic biologic before healing. Statistically significant differences between the recurrence and healed cohort included age category (<65 or ≥65 years, p = 0·033), patient history of diabetes (p = 0·0060), uncontrolled diabetes (p = 0·0090; substantial missing values), and age of target ulcer at baseline visit (stratified into <3 months, 3–<6 months, 6–12 months, and >12 months, p = 0·046). Among the 116 patients included in the study, 27 (23%) were receiving <20 mg/day of prednisone (or equivalent), seven (6%) were receiving ≥20 mg/day, and 82 (71%) were not receiving oral corticosteroids at the time of healing. Of these 82 patients, 75 were not receiving any systemic immunosuppressive therapy, and seven were receiving a non-steroidal systemic agent. There were no missing data for this variable. Oral corticosteroid use at the time of healing was not associated with recurrence (Fisher's exact test, p = 0·58).

Among the 41 patients whose pyoderma gangrenosum recurred, 32 (78%) were local recurrences and nine (22%) were distal recurrences. The most common location for recurrence was the lower extremities (n = 25/41, 61%) followed by the trunk (n = 7/41, 17%), peristomal area (n = 5/41, 12%), upper extremity (n = 3/41, 7%), and breast (n = 1/41, 2%). Among patients with available data, 30 (26%) of 114 had one or more additional active pyoderma gangrenosum lesions at the time the target ulcer was documented as healed. The proportion of patients with additional active lesions was similar between those who subsequently experienced recurrence and those who did not (11/40 [28%] vs 19/74 [26%]; Fisher's exact test, p = 0·83). Data regarding the presence of additional lesions at healing were missing for two patients.

After initial healing, patients were followed quarterly. At the first clinical visit following recurrence, the ulcers had a mean target ulcer wound age of 1·4 months (SD 1·6, range 0·2–6·0 months) and mean size of 4·8 cm2 (SD 7·9, range 0·2–35·0). 27 (66%) of 41 patients had a single ulcer and 14 (34%) had multifocal disease. On assessment, 24 (59%) of 41 patients had severe disease, six (15%) had moderate disease, five (12%) had mild disease, and six (15%) had minimal disease. The mean time from healing to recurrence was 8·9 months (SD 8·0), and most patients had a single recurrence (n = 30/41, 73%; range 1·0–4·0).

Within the 41-patient recurrence cohort, 56 discrete recurrence episodes were identified. Notably, 31 (55%) of 56 episodes had no documented clinical follow-up visit between initial healing and recurrence, and there were 25 episodes (45%) with a follow-up visit between initial healing and recurrence. The mean time from initial healing to follow-up one was 122·6 days (median 51·0 days). There were 14 recurrence episodes (25%) with a second follow-up visit after initial healing, and mean time from follow-up one to follow-up two was 72·5 days (median 83·0 days). At follow-up one, among the 25 visits with documented data, 22 (88%) noted that the patient was taking maintenance therapy and three (12%) noted no maintenance medications. Of the 22 visits where the patient was noted to be on maintenance therapy, three (14%) visits noted only topical interventions, 13 (59%) of 22 noted both topical and systemic interventions, and six (27%) noted only systemic medications. Among the initial follow-ups where the patient was taking maintenance therapy, there were seven (32%) of 22 instances where the patient was noted to be on a different maintenance therapy compared to the visit where they were noted to be healed. Contrary to this, there were 15 (68%) instances noting identical therapy.

Time to recurrence was analysed using Kaplan–Meier methods (Figure S1). The median recurrence-free survival was not reached during the follow-up period, but recurrence-free survival probabilities over time were estimated at specific time points (Table S2). At 6 months after healing, recurrence-free survival was estimated to be 82·9% recurrence-free (95% CI 76·1–90·2). It decreased at 12 months to 73·0% (65·0–81·9) and 64·3% (55·6–74·3) at 24 months, with the curve plateauing at 58·7% (49·5–69·6) at 36 months and beyond. A cumulative incidence curve was created from the Kaplan–Meier recurrence-free survival curve and is shown in Fig. 1. The cumulative incidence of pyoderma gangrenosum recurrence was found to be 27·0% (95% CI 18·1–35·0) at 1 year and 41·3% (30·4–50·5) at 5 years.

Fig. 1.

Fig. 1

Open in a new tab

Cumulative incidence of recurrence. Estimated cumulative incidence of pyoderma gangrenosum recurrence over time following healing estimated from the Kaplan–Meier recurrence-free survival curve following healing of the target ulcer. The shaded region represents the 95% CI.

Associations between baseline patient- and ulcer-level characteristics and time to recurrence were analysed using Cox proportional hazards regression. In the univariable analyses, shown in Fig. 2, several variables showed nominal associations with time to recurrence, including history of diabetes mellitus (hazard ratio [HR] 2·1, 95% CI 1·1–4·0, p = 0·024), age category ≥65 (0·4, 0·2–1·1, p = 0·045), and ever-smoking (1·8, 1·0–3·4, p = 0·057). After purposeful selection, ever-smoking and history of diabetes mellitus were retained in the final multivariable model (Table S4). The adjusted estimate for ever-smoking was attenuated and no longer nominally significant (adjusted HR 1·6, 95% CI 0·8–3·0, p = 0·16), and the same pattern was observed for diabetes mellitus (1·9, 1·0–3·6, p = 0·061).

Fig. 2.

Fig. 2

Open in a new tab

Univariable forest plot showing associations between selected baseline factors and time to pyoderma gangrenosum recurrence. Only covariates meeting the univariable screening criterion (p < 0·25) are shown; see Table S3 for the full univariable results, including all treatment drugs analysed. Medication dosing, and its effect, was not included in the analysis. HR, hazard ratio.

Hospitalisation and mortality outcomes were then summarised by recurrence status. During post-healing follow-up, eight (20%) of 41 patients in the recurrence cohort had a history of hospitalisation, compared with 10 (13%) of 75 patients in the non-recurrence cohort (p = 0·43). Three patients (7%) in the recurrence cohort had been hospitalised for a prolonged period of time compared with five (7%) in the non-recurrence cohort (p > 0·99). One patient (2%) in the recurrence cohort died during follow-up, compared with three patients (4%) in the non-recurrence cohort (p > 0·99). Deaths were treated as censoring events. Only three deaths occurred before recurrence, and a competing-risks sensitivity analysis confirmed that censoring them did not materially affect the recurrence estimates.

A sub-analysis using similar methodology was performed to assess the difference in recurrence risk between patients with classic and peristomal pyoderma gangrenosum (Figure S2). When analysing recurrence patterns of patients with classic pyoderma gangrenosum, 24 (80%) of 30 had local recurrences, whereas six (20%) had distal recurrences. Among patients with peristomal pyoderma gangrenosum, six (67%) of nine had local recurrences and three (33%) had distal. Mean target ulcer wound age at recurrence baseline visit was 1·7 months (SD 1·7 months, range 0·2–6·0 months) for patients with classic pyoderma gangrenosum and 0·5 months (SD 0·2, range 0·2–1·0 months) for patients with peristomal pyoderma gangrenosum. Mean size at recurrence baseline visit for classic pyoderma gangrenosum ulcers was found to be 5·2 cm2 (SD 8·3 cm2, range 0·2–35·0 cm2) and 4·4 cm2 (SD 7·8 cm2, range 0·6–25·0 cm2) for peristomal pyoderma gangrenosum ulcers. For patients with classic pyoderma gangrenosum, 17 (57%) of 30 had a single ulcer at recurrence, compared with eight (89%) of nine patients with peristomal pyoderma gangrenosum. 13 (43%) patients with classic pyoderma gangrenosum had multifocal disease compared with one (11%) patient with peristomal pyoderma gangrenosum. At recurrence baseline assessment, for classic patients with pyoderma gangrenosum, 16 (53%) of 30 had a PGA of four (severe), four (13%) had a PGA of three (moderate), four (13%) had a PGA of two (mild), and 6/30 (20·0%) had a PGA of one (minimal disease activity). For peristomal patients with pyoderma gangrenosum, seven (78%) of nine had a PGA of four and two (22%) had a PGA of three. The mean time from healing to recurrence was 9·8 months (SD 8·7 months) for patients with classic pyoderma gangrenosum and 6·7 months (SD 5·5 months) for patients with peristomal pyoderma gangrenosum. Of the patients with peristomal disease who recurred, seven (78%) of nine had IBD compared with one (3%) of 30 with classic pyoderma gangrenosum. The mean number of recurrences was 1·2 (SD 0·5) for patients with classic pyoderma gangrenosum and 2·0 (SD 1·1) for patients with peristomal pyoderma gangrenosum.

We created a cohort of patients with peristomal pyoderma gangrenosum and concurrent IBD. Within this cohort, when examining those whose pyoderma gangrenosum recurred, five (71%) of seven patients had Crohn's and one (14%) had ulcerative colitis. One patient (14%) had unspecified IBD. At the gastroenterology visit closest to the recurrence episode, two (29%) of seven patients were noted to have active gastrointestinal symptoms, whereas two (29%) were noted to be symptom free (three patients were without available data). Three patients (43%) had calprotectin lab results available, and of these, two (29%, 2/7) were elevated while one (14%, 2/7) was within normal range. Similarly, six (86%) of seven patients had C-reactive protein lab results available. Of these, one (14%, 1/7) was elevated, and five (71%, 5/7) were normal. Five patients (71%) had endoscopic test results: four patients (57%, 4/7) had active disease, and one (14%, 1/7) had normal findings.

Conversely, there were eight patients with peristomal pyoderma gangrenosum and IBD that did not have a recurrence of their pyoderma gangrenosum. Six (75%) of eight patients had Crohn's, one (13%) had ulcerative colitis, and one (13%) was unspecified. Four patients (50%) had gastrointestinal symptoms at the most recent gastrointestinal follow-up, and two (25%) were symptom free. Two patients (25%) had previous abnormal calprotectin lab results, and three (38%) had normal results. Eight patients (100%) had C-reactive protein values available, and these were within normal range. Three patients (38%) had active disease on endoscopy, four (50%) had no signs of disease.

Estimated recurrence-free survival in patients with classic and peristomal pyoderma gangrenosum are shown in Table S5. At 6 months after healing, estimated recurrence-free survival was 83·6% (95% CI 76·1–91·9) for patients with classic pyoderma gangrenosum and 80·2% (95% CI 64·6–99·6) for patients with peristomal pyoderma gangrenosum. 12-month recurrence-free survival was 76·1% (95% CI 67·4–85·8) for patients with classic pyoderma gangrenosum and 60·2% (95% CI 40·6–89·0) for patients with peristomal pyoderma gangrenosum. 24-month recurrence-free survival was 67·7% (95% CI 58·1–78·8) for patients with classic pyoderma gangrenosum and 46·8% (95% CI 27·7–79·1) for patients with peristomal pyoderma gangrenosum. 36-month recurrence-free survival was 60·4% (95% CI 49·9–73·0) for patients with classic pyoderma gangrenosum and 46·8% (95% CI 27·7–79·1) for patients with peristomal pyoderma gangrenosum. Additionally, a Kaplan–Meier curve showing recurrence-free survival following healing of the index ulcer suggested a higher recurrence rate among patients with peristomal pyoderma gangrenosum (Figure S3). A cumulative incidence curve was created from the Kaplan–Meier recurrence-free survival curve for both peristomal and classic pyoderma gangrenosum (Fig. 3). Cumulative incidence of recurrence for classic pyoderma gangrenosum was found to be 23·9% (95% CI 14·2–32·6) at 1 year and 39·6% (95% CI 27·0–50·1) and 5 year compared with 39·8% (95% CI 11·0–59·4) and 53·2% (95% CI 20·9–72·3) for peristomal pyoderma gangrenosum.

Fig. 3.

Fig. 3

Open in a new tab

Cumulative incidence of recurrence by pyoderma gangrenosum subtype. Estimated probability of pyoderma gangrenosum recurrence over time following healing estimated from the Kaplan–Meier recurrence-free survival curve following healing of the index ulcer stratified by pyoderma gangrenosum subtype. Shaded region represents the 95% CI. PG, pyoderma gangrenosum.

A Cox proportional hazards regression was performed for patients with classic and peristomal pyoderma gangrenosum (Fig. 4). In univariable analyses, history of diabetes mellitus showed the strongest association with time to recurrence (HR 2·1, 95% CI 1·1–4·1, p = 0·026). After purposeful selection, diabetes mellitus, systemic immunomodulator or antibiotic use before healing, and pyoderma gangrenosum type were retained in the final multivariable model (Table S7). Diabetes mellitus remained nominally significant after adjustment (adjusted HR 2·2, 95% CI: 1·1–4·1, p = 0·020). Peristomal pyoderma gangrenosum was not associated with a higher hazard of recurrence relative to classic pyoderma gangrenosum (adjusted HR 2·1, 95% CI: 1·0–4·7, p = 0·069), nor was systemic immunomodulator or antibiotic use (adjusted HR 1·8, 95% CI: 0·9–3·5, p = 0·089).

Fig. 4.

Fig. 4

Open in a new tab

Univariable forest plot showing associations between selected baseline factors and time to pyoderma gangrenosum recurrence for patients with classic and peristomal pyoderma gangrenosum. Only covariates meeting the univariable screening criterion (p < 0·25) are shown; see Table S6 for the full univariable results, including all treatment drugs analysed. Medication dosing, and its effect, was not included in the analysis. HR, hazard ratio; PG, pyoderma gangrenosum.

Discussion

This unique cohort study from the first five-year prospective pyoderma gangrenosum registry demonstrated that recurrence is a common and clinically significant outcome in a real-world setting. Our estimated overall recurrence proportion of 35% (41/116) aligns with previous studies; however, the reported range is quite wide (15–61%),1, 2, 3, 4, 5 and due to different study designs, it is difficult to make conclusions. As our study presents results from a multi-year patient registry of a rare disease with over 100 patients, our results reflect the most accurate information regarding pyoderma gangrenosum recurrence data.

On univariable analysis, history of diabetes mellitus and history of smoking were associated with time to recurrence, though this did not reach statistical significance. This finding is consistent with the broader wound healing literature, in which diabetes has been associated with impaired wound healing,18 increased age can predict lower rates of wound healing,19 and increased wound age decreases the likelihood of healing.20 Age and baseline ulcer wound age differed between recurrence and non-recurrence groups on descriptive analysis but were not retained in the final multivariable model after purposeful selection. Our group has previously described that wound healing comorbidities have a higher impact on pyoderma gangrenosum healing outcomes.21 Furthermore, these results suggest that controlling comorbidities (e.g. diabetes) or lifestyle modifications (e.g. smoking) are crucial to not only successful healing outcomes but also to recurrence prevention, though larger multivariable analyses are needed in order to confirm this.

The majority of recurrent ulcers were local, unilesional recurrences located on the lower extremities with an average time from healing to recurrence of 8·9 months. Our findings confirmed previous case reports noting that recurrent ulcers are most likely to be local, located on the lower extremities, and in patients with unilesional disease.13 Therefore, close follow-up and maintenance therapy after initial ulcer healing are indicated, particularly for patients with ulcerations/lesions on the lower extremity, as early intervention minimises the impact of the recurrence. Additionally, our registry captures data systematically while patients have active disease; however, when examining the period after initial healing before the first ulcer recurrence, over half of the recurrence episodes had no follow-ups (31/56, 55%), and where there was a follow-up, the mean time from healing to follow-up was 122·6 days (roughly four months; median 51·0 days). This shows the importance of early close follow-up to ensure appropriate maintenance therapy, with a greater emphasis for patients with pre-existing diabetes or who smoke. Our results suggest that follow-up within 1–3 months of initial healing seems to be vital for early intervention and/or recurrence prevention.

Recurrence-free survival remained at or above 58·7% at 6, 12, 24, and 36 months. Additionally, the cumulative incidence of pyoderma gangrenosum recurrence was found to be 27·0% and 41·3% at 1 and 5 years, respectively. This suggests that even if a patient's wound has remained healed for one year, close follow-up is indicated as the incidence of recurrence notably increases from one to five years. Taken further, our results suggest that long term maintenance treatment of pyoderma gangrenosum in high-risk populations may be indicated for months to years after target ulcer healing in order to prevent recurrence, though future research is needed to support guidelines on optimal maintenance strategies.

A sub analysis was performed to compare recurrence patterns between classic and peristomal pyoderma gangrenosum. Similar results to the larger, pooled analysis were observed for ulcer characteristics: most recurrences in both subgroups were unilesional and local. More patients with peristomal pyoderma gangrenosum presented with a high PGA score at recurrence compared to their classic pyoderma gangrenosum counterparts. Additionally, patients with peristomal pyoderma gangrenosum had a higher recurrence rate and decreased recurrence-free survival probabilities compared with patients with classic pyoderma gangrenosum, though this was an exploratory analysis and should be further studied to confirm our findings. The cumulative incidence rates of peristomal pyoderma gangrenosum recurrence at one and five years were also notably greater than classic pyoderma gangrenosum. This is most likely due to local ulcer trauma from frequent stoma changes, which can impair healing. Additionally, pathergy may contribute to the increased recurrence rates observed in patients with peristomal pyoderma gangrenosum.22 Among patients with peristomal pyoderma gangrenosum who experienced recurrence in our cohort, seven (78%) of nine had concurrent IBD, compared with one (3%) of 30 patient with classic pyoderma gangrenosum who recurred. Taken together, these results suggest that patients with peristomal pyoderma gangrenosum require closer follow-up, maintenance therapy for the pyoderma gangrenosum (not only for underlying comorbidities), or longer maintenance therapy as this population is at higher risk of recurrence following ulcer healing compared to other pyoderma gangrenosum subtypes.

There are limitations to this study. As the patient cohort comes from a single center, the conclusions might not be generalisable. Further efforts are now being implemented for multicenter and international studies that would help to expand and corroborate our results.23 Additionally, heterogeneity in post-healing follow-up, missing follow-up data, and the potential for surveillance or ascertainment bias may have affected our results, which can be expected in projects using real-world data. Lastly, sample size is moderate, and there was no sample size calculation performed for the study, which might limit statistical power and the ability to detect smaller associations. Despite these limitations, our results show a high burden of recurrence in patients with pyoderma gangrenosum using a unique 5-year prospective registry. The findings will inform changes to standard of care supporting maintenance therapy for pyoderma gangrenosum, support the evidence for payers to cover the necessary medications after healing, and aid in the design of clinical trials.

In conclusion, pyoderma gangrenosum recurrence occurs in one in three patients, with higher rates in the peristomal pyoderma gangrenosum subtype, and is nominally associated with history of diabetes and smoking. For all patients with pyoderma gangrenosum, cumulative recurrence risk reaches 41·3% by five years post-healing, showing that maintenance therapy following initial healing may be a key target for recurrence prevention in certain populations. These results support the consideration of maintenance strategies and close follow-up to prevent or treat recurrences promptly in patients with pyoderma gangrenosum. Future research should focus on establishing guidelines for maintenance therapy and treatment duration in patients at high risk of recurrence, understanding the relationship (if any) between recurrences and underlying pyoderma gangrenosum associated comorbidities, understanding the potential impact of multi-lesional disease in recurrences, and clarifying individual genetic predisposition.

Contributors

Study conception and design were performed by Jordan Gillespie, Jaclyn Roland-McGowan, and Alex Ortega-Loayza. The majority of the data collection was performed by Katelyn Downey while Jordan Gillespie and Jaclyn Roland-McGowan assisted. All authors had access to the study data. Data verification was performed by Katelyn Downey, Jordan Gillespie, and Jaclyn Roland-McGowan. The analysis was performed by Emile Latour. The manuscript was written by Jordan Gillespie, Jaclyn Roland-McGowan, and Katelyn Downey. Critical edits were provided by Alex Ortega-Loayza. All authors commented on previous versions of the manuscript. All authors read and approved the final manuscript.

Data sharing statement

Deidentified data (not including individual participant data or a data dictionary) collected from this study may be shared upon reasonable request to the corresponding author after publication. No other documents such as a protocol or statistical analysis plan will be made available.

Declaration of interests

AGO-L is former President of the Pacific Dermatology Association, serves as an associate editor for Dermatology (Karger), and is an editorial board member of the American Journal of Clinical Dermatology. In addition, he is a consultant for Genentech, GuidePoint, InflaRx, Corvus Pharmaceuticals, Castle Biosciences, Clarivate, TFS HealthScience, Otsuka, and Leo Pharma. He is an advisor to Bristol Meyer Squibb, Boehringer Ingelheim, Janssen, and Sanofi. AGO-L has received research grants from Lilly, Janssen, and Pfizer. All other authors have no conflicts of interest to report.

Acknowledgements

During the preparation of this work, the authors used NotebookLM in order to create a figure in the manuscript. After using this tool/service, the authors reviewed and edited the content as needed and take full responsibility for the content of the published article.

Footnotes

Appendix A

Supplementary data related to this article can be found at https://doi.org/10.1016/j.eclinm.2026.104086.

Appendix A. Supplementary data

Supplementary Material
mmc1.docx (4.9MB, docx)

References

  • 1.Thomas K., Ormerod A., Craig F., et al. Clinical outcomes and response of patients applying topical therapy for pyoderma gangrenosum: a prospective cohort study. J Am Acad Dermatol. 2016;75(5):940–949. doi: 10.1016/j.jaad.2016.06.016. [DOI] [PubMed] [Google Scholar]
  • 2.Becker S.L., Velasco R., Ortega-Loayza A.G. Recalcitrant pyoderma gangrenosum: clinical burden and unmet needs. Clin Cosmet Investig Dermatol. 2023;16:2143–2152. doi: 10.2147/CCID.S381490. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 3.Vidal D., Puig L., Gilaberte M., Alomar A. Review of 26 cases of classical pyoderma gangrenosum: clinical and therapeutic features. J Dermatolog Treat. 2004;15(3):146–152. doi: 10.1080/09546630410031909. [DOI] [PubMed] [Google Scholar]
  • 4.Bar D., Baum S., Druyan A., Mansour R., Barzilai A., Lidar M. Clinical course and prognostic disparities of pyoderma gangrenosum based on underlying disease: a long-term comparative study in 124 patients. Ann Dermatol Venereol. 2025;152(2) doi: 10.1016/j.annder.2025.103364. [DOI] [PubMed] [Google Scholar]
  • 5.Ormerod A.D., Thomas K.S., Craig F.E., et al. Comparison of the two most commonly used treatments for pyoderma gangrenosum: results of the STOP GAP randomised controlled trial. BMJ. 2015;350 doi: 10.1136/bmj.h2958. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6.Saklecha A., Tom W., Sprague J. Resolution of chronic recurrent pyoderma gangrenosum with latent tuberculosis treatment. Pediatr Dermatol. 2021;38(1):181–183. doi: 10.1111/pde.14341. [DOI] [PubMed] [Google Scholar]
  • 7.Syed Z., Syed S., Burmaster K. A case report of localized vegetative pyoderma gangrenosum of the foot in a patient with Myasthenia gravis in an outpatient setting. Cureus. 2023;15(6) doi: 10.7759/cureus.41181. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8.Dasanu C., Bockorny B., Alexandrescu D. Pyoderma gangrenosum due to lenalidomide use for multiple myeloma. J Oncol Pharm Pract. 2015;21(6):471–473. doi: 10.1177/1078155214541975. [DOI] [PubMed] [Google Scholar]
  • 9.Kim Y., Lee K., Won C., et al. Clinicopathological features and prognosis of pyoderma gangrenosum in Korea: a single centre, retrospective, observational study over 20 years. Indian J Dermatol Venereol Leprol. 2023;89(1):25–34. doi: 10.25259/IJDVL_968_20. [DOI] [PubMed] [Google Scholar]
  • 10.Flora A., Kozera E., Frew J.W. Pyoderma gangrenosum: a systematic review of the molecular characteristics of disease. Exp Dermatol. 2022;31(4):498–515. doi: 10.1111/exd.14534. [DOI] [PubMed] [Google Scholar]
  • 11.Xia F., Liu K., Lockwood S., et al. Risk of developing pyoderma gangrenosum after procedures in patients with a known history of pyoderma gangrenosum-A retrospective analysis. J Am Acad Dermatol. 2018;78(2):310–314.e1. doi: 10.1016/j.jaad.2017.09.040. [DOI] [PubMed] [Google Scholar]
  • 12.Clark A., Williams B. Recurrence of pyoderma Gangrenosum potentially triggered by COVID-19 vaccination. Cureus. 2022;14(2) doi: 10.7759/cureus.22625. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13.Nassim D., Alajmi A., Jfri A., AlHalees Z., Kanou M., Litvinov I.V. A case of recurrent pyoderma gangrenosum injection site reaction triggered by Maveropepimut-S (previously known as DPX-Survivac) in a patient undergoing immunotherapy for ovarian cancer: a case report. SAGE Open Med Case Rep. 2026;14 doi: 10.1177/2050313x251411501. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14.Cugno M., Borghi A., Marzano A.V. PAPA, PASH and PAPASH syndromes: pathophysiology, presentation and treatment. Am J Clin Dermatol. 2017;18(4):555–562. doi: 10.1007/s40257-017-0265-1. [DOI] [PubMed] [Google Scholar]
  • 15.Jockenhöfer F., Wollina U., Salva K.A., Benson S., Dissemond J. The PARACELSUS score: a novel diagnostic tool for pyoderma gangrenosum. Br J Dermatol. 2019;180(3):615–620. doi: 10.1111/bjd.16401. [DOI] [PubMed] [Google Scholar]
  • 16.R Core Team . R Foundation for Statistical Computing; Vienna, Austria: 2026. _R: a Language and Environment for Statistical Computing_.https://www.R-project.org/ [Google Scholar]
  • 17.Hosmer D.W., Lemeshow S. Wiley; New York: 2000. Applied Logistic Regression. [Google Scholar]
  • 18.Swoboda L., Held J. Impaired wound healing in diabetes. J Wound Care. 2022;31(10):882–885. doi: 10.12968/jowc.2022.31.10.882. [DOI] [PubMed] [Google Scholar]
  • 19.Brocklehurst J. Ageing: an independent risk factor in the management of lower extremity wounds. Int Wound J. 2024;21(11) doi: 10.1111/iwj.70105. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20.Bosanquet D.C., Harding K.G. Wound duration and healing rates: cause or effect? Wound Repair Regen. 2014;22(2):143–150. doi: 10.1111/wrr.12149. [DOI] [PubMed] [Google Scholar]
  • 21.Zhao H., Sengupta S.K., Haddadin O., Sisley J.M., Downey K., Ortega-Loayza A.G. Effects of comorbidities on delayed wound healing in patients with pyoderma gangrenosum. J Eur Acad Dermatol Venereol. 2025;39(5):e436–e438. doi: 10.1111/jdv.20457. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22.Patel D.K., Locke M., Jarrett P. Pyoderma gangrenosum with pathergy: a potentially significant complication following breast reconstruction. J Plast Reconstr Aesthet Surg. 2017;70(7):884–892. doi: 10.1016/j.bjps.2017.03.013. [DOI] [PubMed] [Google Scholar]
  • 23.Haddadin O.M., Jacobson M.E., Becker S.L., et al. Minimum data set for treatment effectiveness in pyoderma gangrenosum for an international registry: an international multidisciplinary eDelphi consensus. Br J Dermatol. 2026 doi: 10.1093/bjd/ljag084. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplementary Material
mmc1.docx (4.9MB, docx)

Articles from eClinicalMedicine are provided here courtesy of Elsevier

RESOURCES