Pyoderma gangrenosum among patients with inflammatory bowel disease: a descriptive cohort study

Abstract

BACKGROUND

Pyoderma gangrenosum (PG) is a severe extra-intestinal manifestation of inflammatory bowel disease (IBD).

OBJECTIVE

To better characterize PG features and management among an IBD cohort.

METHODS

Subjects with PG were identified using a large database at a tertiary center. Patient demographics and clinical characteristics were summarized using descriptive statistics.

RESULTS

80 patients with an episode(s) of PG were identified, yielding an overall prevalence of 1.9%. Overall, 93% of patients with PG had some degree of colonic involvement. Thirty-one (39%) patients required hospitalization for PG. Underlying bowel disease was active at the time of PG episode(s) in 52 (65%) patients. PG location was variable, with the lower extremity being the most common. Most patients (71.3%) required multiple therapies to achieve PG healing.

CONCLUSIONS

We have described one of the largest case series of PG among patients with IBD. The variety of treatment strategies used highlights the lack of clear guidelines in managing this complex group of patients.

INTRODUCTION

Pyoderma gangrenosum (PG), a neutrophilic dermatosis affecting the skin, is one of the more common extra-intestinal manifestations of inflammatory bowel disease (IBD)^1–4. The lesion of PG usually begins as a papule or pustule at a site of trauma with a surrounding violaceous and undermined border⁵ with subsequent necrosis of the dermis resulting in deep ulcers as the lesion progresses. PG is characterized by pathergy and most commonly occurs on the legs, but lesions can occur anywhere (e.g. adjacent to stoma) ^6,7.

PG occurs in 0.5–5% of patients with IBD and has previously been shown to be more common in women, IBD patients with other extra-intestinal manifestations and has also been particularly associated with ulcerative colitis (UC) and among patients with colonic Crohn’s disease (CD)^7–11. The development of PG is one of the more severe extra-intestinal manifestations of IBD with the potential for both significant morbidity and tremendous impact on activities of daily living and quality of life^5–6. Severe disease can be cosmetically disfiguring and may lead to permanent scarring, significant pain, bacterial super-infection, and skin grafting. Moreover, potent immunosuppression with therapeutic agents such as intravenous cyclosporine or anti-tumor necrosis factor (anti-TNF) agents may be needed, independent of underlying intestinal disease activity^7,12,13.

Due to the relative rarity of this condition, there are a limited number of published large, IBD cohorts with PG. In this report, we review a large, well-phenotyped IBD cohort with the aim of better characterizing the clinical features, underlying disease characteristics, and management strategies of PG.

MATERIALS AND METHODS

Subjects

Subjects with PG were identified using our longitudinal IBD database, which contains detailed clinical (demographics, disease phenotype, disease history, extra-intestinal manifestations, treatment record), serologic, and genetic data on over 4000 patients with IBD. Medical records of patients identified as having one or more episodes of PG between the years 1997–2012 were reviewed to confirm and better characterize the PG episode(s) and underlying IBD. Clinical data collected included demographics (age, gender, ethnicity, race), family history, disease duration, disease phenotype (IBD subtype and disease location/behavior according to the Montreal Classification¹⁴), surgical history, treatment record, smoking status, and presence or absence of other extra-intestinal manifestations. IBD disease activity at the time of a PG episode was determined retrospectively by the presence of symptoms and, when available, elevated inflammatory markers, fecal calprotectin, and endoscopic correlation.

Case Definition

An episode of PG was defined based on either dermatology consultant impression or a dermatologic lesion with three or more of the following typical clinical features¹⁵: (i) leg or peristomal location (ii) pathergy (iii) initial pustular lesion (iv) purulent discharge (v) violacious or undermined borders (vi) crater-like holes/cribiform scarring. Histology reports were reviewed, when available, to exclude alternative diagnoses.

Statistical Analysis

Patient demographics and clinical characteristics were summarized using descriptive statistics.

Ethical Considerations

The study was approved by our institutional research board, and all subjects provided written, informed consent when initially enrolled into the IRB-approved database.

RESULTS

Demographics

We identified 104 patients from the database with at least one episode of PG. Among these, 24 did not meet our above criteria for defining PG and it was determined they did not have PG and were therefore excluded. Thus, 80 patients with a confirmed episode(s) of PG were included in this case series. At the time of the study, the database contained information on 4137 patients, yielding an overall prevalence of PG among this tertiary care population of 1.9%.

Table 1 outlines the demographic characteristics of the 80 patients with at least one episode of PG. Fifty-eight (73%) had Crohn’s disease, 19 (24%) had ulcerative colitis, and 3 had IBD-unclassified. Overall, 46 (58%) were female and 34 (42%) were male. Sixty-nine (86%) were of Northern-European origin. The mean age of IBD onset was 31 + 14.5 years. Twenty-four (30%) patients reported a positive family history of IBD. Thirty-one patients (39%) were former or current smokers at the time of IBD diagnosis. Fifty-three (66%) had experienced at least one other extra-intestinal manifestation. The majority of patients had previous exposure to corticosteroids (93%), 5ASA or sulfasalazine (84%), immunomodulators (96%), or a biologic (85%).

Table 1.

Clinical and Demographic Characteristics of IBD-Pyoderma Gangrenosum (PG) Cohort

Clinical/Demographic Feature	PG subjects n (% of overall cohort)
Female	46 (58)
Male	34 (42)
Crohn’s Disease	58 (73)
Ulcerative Colitis	19 (24)
IBD-U	3 (3)
Disease Reclassification	21 (26)
Mean Age of IBD onset (years) ± S.D.	31 ± 14.5
Smoking History	31 (39)
Family History IBD	24 (30)
Previous IBD related surgery	51 (64)
Previous Medication Use
Corticosteroids	74 (93)
5-ASA	67 (84)
Immunomodulator*	77 (96)
Cyclosporine	16 (20)
Biologic+	68 (85)
Other++	3 (4)
Other EIM
Oral Ulcers	53 (66)
Ophthalmologic	12 (15)
Arthritis	12 (15)
Sacroilitis/Ankylosing	24 (30)
Spondylitis	3 (<1)
Erythema Nodosum	11(14)
PSC	1(<1)

^*6-mercaptopurine, azathioprine, methotrexate

⁺infliximab, adalimumab, certoliziamab, natilizamab

⁺⁺others include thalidomide

Colonic and ileocolonic disease were most common among those with Crohn’s disease (43% and 45%, respectively) and extensive colitis was most common among those with ulcerative colitis (85%). Overall, 93% of patients with PG had some degree of colonic involvement.

Pyoderma Characteristics

Dermatology consultation was obtained in 31 (39%) of cases with 21 (26%) patients undergoing biopsy of the lesion, and the remaining cases diagnosed clinically. The mean age of onset of the first episode PG was 39.3 years. The majority of patients had only one episode of PG (61 [76%]). Twelve (15%) patients had 2–3 episodes and 7 (9%) had more than three episodes. Underlying IBD was active at the time of the PG episode(s) in 52 (65%) patients. Four patients developed PG several years before the onset and diagnosis of IBD.

PG location was variable (Table 2) with the most common sites being the lower extremity (53% with unilateral leg involvement and 19% bilateral leg involvement). Peri-stomal PG was noted in 23% (18) of patients. Less common sites included trunk, upper extremity, face (figure 1), scalp, and perianal/pubic region (figure 2).

Table 2.

Features of Pyoderma Gangrenosum

PG Feature	PG Subjects n (% of PG Cohort)
Age at first PG episode	39.8
IBD Active at onset	52 (65)
Number of PG episodes
1	61(76)
2–3	12 (15)
>3	7 (9)
Location*
Unilateral Lower Extremity	42 (53)
Bilateral Lower Extremity	15 (19)
Peristomal	18 (23)
Trunk	5 (6)
Back	4 (5)
Upper Extremity	5 (6)
Face	3 (4)
Perianal	1 (1)
Pubic	1 (1)
Scalp	1 (1)

^*patient may have had PG at ≥1 site

Figure 1.

Severe facial pyoderma gangrenosum. Note full-thickness ulceration with visualization of underlying dentition and characteristic “heaped-up” edges with a violaceous border. This lesion healed with a combination of cyclosporine, intravenous immunoglobulin, and hyperbaric oxygen treatments.

Figure 2.

Perianal pyoderma gangrenosum in a patient with ileal pouch-anal anastomosis. These lesions healed with intravenous corticosteroids and cyclosporine.

Therapy

A variety of therapeutic approaches were used for treating PG in this cohort. Most patients (71.3%) required multiple therapies to achieve PG healing, while the remaining patients were managed with a single agent. Six patients (8%) received no treatment. Thirty-one (39%) patients required hospitalization for a PG episode. At the time of PG onset, 10 (13%) were actively on an anti-TNF agent for the treatment of their underlying IBD.

Systemic corticosteroids was the most commonly prescribed therapy, with 47 (59%) of patients receiving either oral or intravenous steroids at some point during the PG episode. Intra-lesional injection of corticosteroids were used in 9 (11%) of patients. Anti-TNF agents and cyclosporine were used in the treatment of PG with equal frequency (29%). Azathioprine was used as an adjuvant treatment in 6 (8%) patients. Hyperbaric oxygen was used in 3 (4%) patients and intravenous immunoglobulin was used in one patient. Surgical intervention was part of the management of 16 (20%) patients. This included skin grafting, tissue debridement, and/or re-siting of stoma. Four (5%) patients were managed exclusively with surgical intervention and did not receive any specific medical therapy for PG.

DISCUSSION

Pyoderma gangrenosum is a severe extra-intestinal manifestation of IBD with lesions having significant pain and scarring and patients often requiring hospitalization. Moreover, PG management often requires aggressive immunosuppressant therapy, even in the absence of active underlying IBD. The current study aimed to characterize clinical and dermatologic features of a large IBD patient cohort followed at a tertiary, referral center with at least one episode of PG.

We noted 80 cases of PG among the cohort yielding a prevalence of 1.9%. This is agreement with other published series in which prevalence of PG has ranged from 0.5–5%^8–11,16. Among our cases, 73% had Crohn’s disease although previous studies have not consistently demonstrated a clear association between a particular IBD subtype with PG. PG was more commonly found in patients with UC in a large population based Swiss cohort¹⁶, whereas in a population-based cohort from Manitoba, PG was more prevalent in CD¹⁰ and this finding subsequently been supported by others¹⁷. Our results may be, in part, a reflection of a higher incidence of CD in the patients seen at our center and its important to note that this cohort is a tertiary referral center ascertained cohort in contrast to the Swiss and Manitoba-based studies above. While data on IBD subtype is conflicting, there is more consistency among studies evaluating IBD disease location and association with PG. Colonic disease has previously been associated with a several EIM’s including PG, erythema nodosum, and IBD-associated arthritis^7–16–18. We similarly noted that colonic and ileocolonic disease was most common among those with Crohn’s disease (43% and 45%, respectively) and extensive colitis was most common among those with ulcerative colitis (85%). Overall, 93% of patients had some degree of colonic involvement. The predominance of colonic disease in subjects with EIM suggests underlying mechanisms intrinsic to colonic inflammation, which may include genetic, immunologic, and/or microbiologic interactions such as cross-reactivity of skin, joint, and eye and gut antigens¹⁹. This intriguing overlap and potential link is further supported by the association between perinuclear (p) antineutrophil cytoplasmic antibodies (ANCAs), a serologic marker often detected in patients with colonic IBD including UC and UC-like CD phenotypes²⁰, with a number of EIM such as uveitis²¹. In addition genetic variation at the MHC has previously been associated with both extensive colonic disease and the presence of EIMs²².

We noted a slight female predominance (58%) among our PG cohort. Several studies have previously noted an association between EIMs and female gender including a Swedish study that found that erythema nodosum was more than 3 times more likely to occur in women¹⁸. In addition, Vavricka et al. noted an EIM prevalence of 50% in females compared to only 34% in males (p<0.001) ¹⁶. A female predilection for PG is less clear from the literature. A recent systematic review identified 60 cases of PG, 55% of whom were female⁷. The findings of female predominance suggest a possible role of hormonal influences and some have suggested that estrogen may have a modulating effect on EIM’s²³. Clearly, more study is needed to explain these observed gender differences.

The literature suggests that patients with one EIM are more likely to develop additional EIMs^16–18. For example, Vavricka et al. observed that patients with IBD-associated arthritis had an OR of 2.9 for subsequent PG development¹⁶. We noted that two-thirds of patients with PG had either a prior history of EIM, or subsequently developed another EIM. IBD-associated arthritis was the most common other EIM noted in our cohort. These findings point to a common underlying pathophysiology for EIMs in IBD, suggesting a need for further study to better define the apparent common etiology in CD.

We noted a mean age of first episode PG onset of 39 years. This age is somewhat younger compared to other series in which the age of PG diagnosis has ranged from 45 to 62^23–27. However, these published cases were not necessarily limited to patients with IBD and so included patients with PG associated with hematologic disorders, for example, more commonly seen in older individuals. An IBD-specific cohort found an age of PG onset of 36 years, similar to our results¹⁷.

The relationship between PG onset or recurrence and underlying IBD disease activity is unclear. Traditionally, PG behavior and IBD disease activity were thought to follow independent courses, with approximately half of patients having active disease at the time of PG onset⁶. A more recent study noted 75% of patients with PG onset had active disease²⁸. We found that approximately two-thirds of patients had active IBD at the time of PG onset. These findings suggest that the association between disease activity and PG is stronger than previously reported. More prospective studies objectively evaluating disease activity at the time of PG development are needed to better characterize the influence of active intestinal inflammation on PG, as has been shown with other EIMs such as erythema nodosum²⁸.

While PG location was variable, the leg was the most common site, with 57 episodes of PG involving the lower extremity either unilaterally or bilaterally. The propensity for the lesion to form on the lower extremity is consistent with most other reports evaluating PG characteristics ^1,5–7,15. Pathergy, the development or worsening of a lesion after trauma, is a classic characteristic of PG and may be responsible for the high incidence of lesions on the leg, as this is a high risk area for minor injuries¹⁵. This phenomenon may also explain the finding of peristomal pyoderma, as minor trauma to the area with appliance changes are common. We noted 15 episodes of peri-stomal pyoderma, making it the second most common site for PG development, similar to previous reports^15,17,25. Several patients had multiple concurrent lesions and while most had only 1 episode of PG, nearly one quarter of patients had multiple episodes over time.

Some have recommended a more aggressive management approach for patients with multi-lesional PG citing a greater tendency for lesions to progress when present in more than one location²⁹. A number of less commonly encountered sites were also observed including the face (figure 1) and perianal area (figure 2). Whether multifocal PG or the occurrence of a lesion on an atypical site carries a worse prognosis needs to be further elucidated.

A variety of therapies were used in the management of PG among our cohort. The finding that over 70% of patients required multiple treatments to achieve PG healing and that approximately 40% of patients required hospital admission specifically for the management of PG underscores the severe nature of many of these PG episodes.

Systemic corticosteroids were the most commonly used treatment, with nearly 60% of patients receiving intravenous or oral corticosteroids. Only 11% were treated with topical or intra-lesional steroids, a lower figure than most other reports^26–27, suggesting cases among our cohort were considered too severe for local therapy. While no controlled studies for corticosteroids are available, many reports have shown their efficacy and they are therefore considered by many to be first line therapy for PG^29–31. A notable exception was a report by Vidal et al., in which cyclosporine (CSA) was more commonly used and more effective than corticosteroids, leading the authors to conclude that CSA should be first line treatment for the induction of rapid wound healing²⁷. Four percent of patient received hyperbaric oxygen for treatment, usually as an adjuvant therapy for severe cases. Just under one-third of our cohort received intravenous CSA, usually together with systemic steroids. Most of these patients started CSA after failing to improve after several days of intravenous steroids. This practice is supported by a small prospective cohort of 11 patients that were treated with CSA after failing steroid mono-therapy, whereby 100% achieved complete wound healing with a mean time to healing of 1.4 months¹². Moreover, the authors noted that patients transitioned to azathioprine had no recurrence, a practice that was uncommon in our cohort with only 6 patients started on azathioprine as PG therapy. Twenty-nine percent of patients were started on anti-TNF therapy (29%) in the management of PG. As with patients started on CSA, these patients were largely steroid mono-therapy failures. Interestingly, 13% of patients were already on an anti-TNF at the time of PG onset. This may suggest a unique mechanism of PG development in this subset of patients given that anti-TNF has been demonstrated to heal PG in a small randomized control trial and a number of small case series^13–32. A similar parallel has been noted with psoriasis, a dermatologic lesion also treated with anti-TNF therapy. Cases of lesions erupting paradoxically following anti-TNF therapy has been described in a subset of IBD patients^33–34. Uncontrolled interferon-alpha levels in the setting of TNF inhibition is one hypothesis for a mechanism underlying this paradoxical finding³⁵. More study on the underlying pathophysiology and immunologic underpinnings of PG are needed to better clarify optimal treatment regimen(s).

A number of limitations in the current study are noted. First, the retrospective nature of this study limited the ability to collect standardized information. The quality of the description of PG lesions and associated therapy were variable, a well-recognized shortcoming of retrospective reviews of this relatively rare condition. Important treatment details including efficacy and time to healing were not consistently documented and therefore not reported. A standardized approach for monitoring the clinical course of PG is therefore needed. Moreover, assessing IBD disease activity retrospectively relies on objective assessments available in the medical record at that time (e.g. endoscopic reports, fecal calprotectin, inflammatory markers, etc.). Assessment of disease activity was not always formally assessed at the time of new PG diagnosis. Second, our case definition was limited by the fact that no validated definition of PG is available. However, we used descriptive, clinical criteria cited by most authors and therefore our definition is well aligned with other reports in the literature. Third, our cohort consisted of severe cases of PG followed at a tertiary IBD referral center and therefore likely reflect more severe cases and limit the generalizability of our findings to other settings.

However, despite these limitations, we describe one of the largest case series of PG in a well- characterized cohort of patients with IBD. Hospitalization was not uncommon, underscoring the morbidity and cost associated with this severe extra-intestinal manifestations. Moreover, a variety of treatment strategies were used highlighting the lack of clear guidelines as how to effectively manage this complex group of patients. More study on the clinical characteristics of PG complicating IBD is needed as well as on the underlying genetic and immunologic underpinnings of this debilitating dermatologic manifestation.