Pyoderma gangrenosum: From historical perspectives to emerging investigations

Abstract

Pyoderma gangrenosum (PG) is a rare disease of unknown aetiology, first described over a century ago. Initially thought to have an infectious cause, and now primarily considered an autoinflammatory condition, PG continues to be poorly understood, commonly misdiagnosed, and difficult to treat. In this review, we discuss the journey of our understanding of PG to date, including first descriptions, challenges with diagnosis, presumed pathogenesis, and treatments used. We highlight major historical landmarks and their importance, explain the rationale behind current investigations, note outstanding gaps in knowledge, and explore the future directions of PG research. We summarise what we have known, what we are working on knowing, and what we have yet to explore about PG, illustrating overall trends to invigorate future research.

Abbreviations

CCR

CC chemokine receptor

FDA

Food and Drug Administration

IL

interleukin

MMP

matrix metalloproteinase

MPO

myeloperoxidase

PAPA

pyogenic arthritis, pyoderma gangrenous, and acne

PAPASH

pyogenic arthritis, pyoderma gangrenous, acne, and hidradenitis suppurative

PASH

pyoderma gangrenous, acne, and hidradenitis suppurative

PG

pyoderma gangrenosum

PSTPIP1

proline‐serine–threonine‐phosphatase interactive protein 1

RCT

randomised controlled trial

TNF

tumour necrosis factor

UC

ulcerative colitis

VEGF

vascular endothelial growth factor

1. INTRODUCTION

Pyoderma gangrenosum (PG) is an uncommon prototypical neutrophilic dermatosis that was first described over 100 years ago,1, 2 but continues to be poorly understood. Misdiagnosis is not uncommon in clinical practice, likely due to the lack of specific markers. ³ Once the diagnosis of PG has been established, systemic immunosuppression is generally the standard of care; however, there are no FDA‐approved treatments for this condition. Since PG's initial description, many researchers have been trying to understand its pathogenesis and association with other comorbidities, and the search for the best approach to diagnosis and treatment is ongoing.4, 5 More recently, there has been a push for improvement in measuring disease severity. ⁶ As these investigations continue, understanding what has been carried out in the past might shape the future direction of PG.

2. FIRST DESCRIPTIONS AND DIAGNOSIS

PG was first described as its own clinical entity in 1908 and again in 1916 by Louis Brocq, a French physician (Figure 1).1, 2 Brocq reported a series of patients with distinct ulcers of which he discerned three major components: (a) the ridge, (b) the external slope of the border; and (c) the internal slope of the border. He described the ridge as “featuring a regular, geometric, circular, or elliptic pattern;” the external slope as erythematous, infiltrated, and sometimes painful; and the internal slope as undermined, “sharp as a cliff,” and “dimpled by purulent cavities.”2, 8 Histopathology showed neutrophilic infiltration.1, 2, 9 Brocq chose the name geometric phagedenism to capture both the geometrical pattern of the ulcer and its rapidly extensive, necrotic nature (phageton [Greek], meaning consumption).2, 8

FIGURE 1.

Louis‐Anne‐Jean Brocq, MD (1856‐1928). Reproduced from Wikimedia Commons ⁷

The term pyoderma gangrenosum was not introduced until 1930, when Brunsting et al ¹⁰ described five cases of PG (Figures 2 and 3), of which four were associated with ulcerative colitis (UC) and one with idiopathic chronic purulent pleurisy. The cause of PG at the time was thought to be infectious, and the term pyoderma was used by dermatologists to denote a purulent infection of the skin due to pyogenic organisms. ¹⁰ The term gangrenosum referred again to the condition's rapidly extensive, necrotic nature.8, 10 Brunsting et al also noted similarities between their cases and others that had been previously described with different names. ¹⁰ For example, a similar syndrome of ulceration of the skin in association with diarrhoea was reported before, and many cases were attributed to cutaneous infection with Entamoeba histolytica (amebiasis cutis).12, 13 In some of these cases, however, amoebic organisms were not demonstrated in neither the bowel nor the skin, including that of Cullen ¹⁴ following appendectomy in 1924. This case and others, such as those reported by Meleney15, 16, 17 (as chronic undermining nongangrenous burrowing ulcers, in 1935), instead isolated haemolytic streptococci, which were thought to be causative.18, 19 Ultimately, PG has had many names, although pyoderma gangrenosum was adopted rapidly following Brunsting et al's description (Table 1).9, 21 By the 1940s, it was accepted that PG most often occurred in association with UC, but was also thought to occur with trauma, infectious arthritis, upper respiratory infection, pleuritis, pericarditis, cholecystitis, genitourinary infections, abscessed tooth, or (rarely) independently.20, 21 The first example of PG with extracutaneous involvement was reported in 1985 as steroid‐responsive pneumonitis, ²² and pulmonary involvement has been the most common extracutaneous manifestation to date. ²³

FIGURE 2.

Louis A. Brunsting, MD (1900‐1980). Reproduced with permission from Perry ¹¹

FIGURE 3.

Landmark 1930 case series describing and naming pyoderma gangrenosum (PG) ¹⁰

TABLE 1.

The many early denominations of pyoderma gangrenosum (PG) ²⁰

Brocq	Geometric phagedenism	1908
Cullen	Postoperative progressive gangrene (denomination by Meleney)	1924
Zurhelle, Klein	Exulcerating papillary pyoderma	1928
Bolog	Chronic ulcerative pyoderma	1928
Tischnenko, Kroiczic	Chronic serpiginous ulcerative pyoderma	1928
Smith	Chronic pyoderma	1929
Brunsting, Goekerman, O'Leary	Pyoderma gangrenosum	1930
Meleney	Gangrenous ecthyma or impetigo	1933
Holman	Phagedenic ulceration	1935
Fox, Maloney	Chronic serpiginous ulcerative pyoderma	1935
Meleney, Johnson	Chronic undermining burrowing ulcer	1935
Zeisler	Pyogenic lesions of the skin associated with chronic ulcerative colitis	1938
Ormsby	Dermatitis gangraenosa	1938

Today, PG is considered an inflammatory disease which may present in association with autoimmune disorders and cancers.24, 25 It could be also be a skin manifestation of autoinflammatory syndromes, and rare familial forms have been reported.26, 27, 28 Interestingly, nonulcerative forms have too been described, and extracutaneous manifestations may also occur preceding, during, or following the appearance of skin lesions.23, 24 Differential diagnosis is broad, divided into six categories by Weenig et al: vasculopathy, vasculitis, cancer, infection, exogenous tissue injury, and other inflammatory disorders. ³ Histopathologic findings are considered nonspecific, although suppurative inflammation is characteristic, and secondary leukocytoclastic and lymphocyte‐mediated vasculitis may be seen. ⁹

Traditionally considered a diagnosis of exclusion, the first suggested diagnostic criteria for PG were described in 2004 by Su et al, ²⁹ and in 2018, two more sets of diagnostic criteria were proposed (Table 2).4, 5 These newer criteria dissect those prior into component parts, and include a validation process. Further refinement in different clinical settings of these criteria are needed to support their applicability in medical practice.

TABLE 2.

Diagnostic criteria for ulcerative pyoderma gangrenosum (PG)

2004 Su et al 30 (Mayo Clinic, Rochester) a
Major criteria	Rapid progression of a painful, necrolytic cutaneous ulcer with an irregular, violaceous, and undermined border
	Other causes of cutaneous ulceration have been excluded
Minor criteria	History suggestive of pathergy or clinical finding of cribriform scarring
	Systemic diseases associated with PG
	Histopathologic findings (sterile dermal neutrophilia, +/− mixed inflammation, +/− lymphocytic vasculitis
	Treatment response (rapid response to systemic steroid treatment)

2018 Delphi Consensus b
Criteria	Designation
Biopsy with neutrophilic infiltrate	Major
Exclusion of infection on histology	Minor
Pathergy	Minor
Personal history of inflammatory bowel disease or inflammatory arthritis	Minor
Papule, pustule or vesicle that rapidly ulcerates	Minor
Peripheral erythema, undermining border, and tenderness at site of ulceration	Minor
Multiple ulcerations (at least one occurring on anterior lower leg)	Minor
Cribriform or wrinkled paper scares at healed ulcer sites	Minor
Decrease in ulcer size after immunosuppressive treatment	Minor

2018 PARACELSUS score c
Criteria	Designation	Value (points)
Progressing disease (ulcer developing within 6 weeks)	Major	3
Assessment of relevant differential diagnoses	Major	3
Reddish‐violaceous wound margin	Major	3
Amelioration (Alleviation) by immunosuppressant drugs	Minor	2
Characteristically irregular (bizarre) ulcer shape	Minor	2
Extreme pain >4/10 on visual analogue scale	Minor	2
Localization of lesion at site of trauma (pathergy)	Minor	2
Suppurative inflammation in histopathology	Additional	1
Undermined wound border	Additional	1
Systemic disease associated	Additional	1

^aBased on essential elements of PG as proposed by the authors. Diagnosis requires presence of both major and at least two minor criteria.

^bBased on a retrospective study of patients with ulcerative PG and its mimickers published in PubMed (venous leg ulcers, vasculitis, and calciphylaxis). Presence of 1 major criteria plus 4 minor criteria is the threshold for diagnosis.

^cBased on a retrospective study of patients with ulcerative PG and chronic venous stasis ulcers. If total points are ≥10, PG is highly likely. If <10, PG is unlikely.

3. PATHOGENESIS

In his first description, Brocq suggested that geometric phagedenism was a complication of ulcerations, with a probable infectious cause.1, 2 Brunsting et al ¹⁰ suggested a generalised infectious syndrome through symbiosis between staphylococci and streptococci, but also proposed a role of “reactivity or allergy of the individual host.” The notion was that an infection with these invading organisms caused a “marked lowering of bodily resistance” and therefore skin ulcerations in susceptible, “reactive” hosts. ¹⁰ The idea of an “allergic” pathogenesis involving sensitisation of the skin to bacteria was also suggested by Michelson and Bruning ³⁰ in Germany at the time, although it is unclear the specific definition of “allergy” in this context. The first well‐documented, exclusively noninfectious explanation was proposed by Cohen ³¹ in 1936, who suggested that previously isolated organisms such as streptococci, staphylococci, and Pseudomonas aeruginosa, were “secondary invaders.” He instead posited the cause to be a vitamin deficiency, partially influenced by the knowledge of vitamin deficiency as a common occurrence in UC. ³¹ The cause of UC itself had several different considerations then, notably including infection, and the idea of a primarily infectious, bacterial aetiology for PG continued to prevail through the 1950s.20, 21, 32, 33

In 1953, Rostenburg ²⁰ suggested that PG was an example of the Shwartzman phenomenon, which at the time referred to thrombosis and subsequent tissue necrosis following exposure to endotoxin, and is now generally described in the context of disseminated intravascular coagulation. In the localised form of the phenomenon, necrosis occurs to the first intradermal injection site following re‐exposure to endotoxins through a second intravenous dose 24 hours later.20, 34, 35 It is now believed that the Shwartzman phenomenon connotes an innate hypersensitivity response accompanied by an acute small vessel vasculopathy, and intradermal injection with the pro‐inflammatory cytokines tumour necrosis factor (TNF)‐α and interleukin (IL)‐1 are effective substitutes for endotoxin.34, 36 While details of cellular immunology at the time did not clearly distinguish innate versus acquired immunity, the Shwartzman phenomenon was thought to be distinct from a type 1 or type 2 antibody‐mediated as well as a type 3 immune complex‐mediated (i.e. Arthus) hypersensitivity reaction. ²⁰ In 1956, Wright and Greco ³³ reported the first case of PG successfully treated with systemic corticosteroids. At the time, it was understood that the use of corticosteroids in the setting of active infection was detrimental, so Wright and Greco argued that PG was not, in fact, caused primarily by infection. They instead suggested a “violent antigen‐antibody reaction” for which cortisone was beneficial through its anti‐inflammatory actions, specifically depression of antibody formation. ³³ By the 1960s, given the overall ineffectiveness of antibiotic therapy in contrast to treatment successes with corticosteroids, immunological considerations for PG became popular.

By the 1970s, changes in humoral factors, cell‐mediated immunity, and/or neutrophil function were thought to be the cause of PG. ³⁷ Both hyper‐ and hypogammaglobulinemia, sometimes in the setting of monoclonal gammopathy, had been reported with PG, although these findings were not reproducible or causative.37, 38, 39 Impaired cellular immunity was reported in a few cases, represented by cutaneous anergy and T‐cell lymphopenia, although these findings were again inconsistent.37, 40, 41 Abnormalities in neutrophil function had too been reported: in 1975, Jacobs and Goetzl ⁴² identified a “streaking leukocyte factor,” a serum factor which enhanced neutrophil random migration. In 1980, Holt et al ⁴³ described decreased neutrophil migration, reduced chemotaxis, and abnormal phagocytosis in PG patients. By the early 1980s, abnormalities of cell‐mediated immunity versus neutrophil dysfunction had fallen in favour. ⁴⁴

In 1997, Lindor et al ²⁷ described PAPA (pyogenic arthritis, PG, and acne) syndrome, which is caused by an autosomal dominant mutation in the PSTPIP1 (proline‐serine–threonine‐phosphatase interactive protein) gene. This genetic mutation was discovered as the cause of PAPA syndrome in 2002 by Wise et al ⁴⁵ In 2003, Shoham et al ⁴⁶ demonstrated that PSTPIP1 is a ligand of pyrin, an important modulator of innate immunity that inhibits the activation of the inflammasome and subsequent expression of IL‐1beta. The recognition of monogenic autoinflammatory diseases such as PAPA, PAPASH (pyogenic arthritis, PG, acne, and hidradenitis suppurativa), and PASH (PG, acne, and hidradenitis suppurativa) syndromes has contributed significantly to our current understanding of the mechanisms of polygenic non‐Mendelian inherited autoinflammatory skin diseases such as PG.9, 47 It is currently thought that a genetically susceptible background is involved, and several genes have been implicated. Still, no triggers have been identified.24, 47, 48

The first study to demonstrate IL‐8 overexpression in tissue was performed by Oka et al ⁴⁹ in 2000, who demonstrated that human fibroblasts overexpressing IL‐8 induced ulcers resembling PG in mice. In 2007, Oka ⁵⁰ re‐demonstrated this overexpression in a patient with PG. In 2007, Bister et al ⁵¹ performed a study using human skin biopsies of PG which showed that abundant expression of matrix metalloproteinase (MMP) 9, MMP‐10, and TNF‐α, and lack of epithelial MMP‐1 and MMP‐26 characterised PG. This marked the first multipatient molecular‐level study of PG. In 2010, a landmark study by Marzano et al ⁵² elucidated the role of cytokines, MMPs, vascular endothelial growth factor (VEGF), and inflammatory cell markers in PG. They showed that immunoreactivities of TNF‐alpha, IL‐8, IL‐17, MMP‐2, MMP‐9, myeloperoxidase (MPO), and VEGF were significantly higher in PG‐affected tissue; MPO, IL‐8, and MMP‐9 expression were especially increased in certain types of PG (ulcerative and bullous) and compared to another neutrophilic dermatosis, Sweet's syndrome. ⁵² In 2012, Fischer‐Stabauer et al ⁵³ demonstrated IL‐17 was also abundant in tissue affected with PG. There has also been evidence the adaptive immune system plays a role in the pathogenesis of PG. In 2015, Caproni et al ⁵⁴ investigated the proportions of regulatory T cells (Tregs) to T helper (Th)17 cells in lesional skin biopsies of patients with PG, and found that the Treg/T17 cell ratio was reduced. In 2016, Quaglino et al ⁵⁵ were the first to analyse cytokine/chemokine expression in the peripheral blood of patients with PG. They demonstrated an overexpression of CC chemokine receptor (CCR)5 and CCR6, and downregulation of CCR4, suggesting a Th1/Th17‐mediated inflammatory pattern (with downregulation of Th2). ⁵⁵

Currently, PG is thought to be primarily a neutrophil‐mediated autoinflammatory disease, which involves aberrant activation of the inflammasome.52, 56 This is supported by recent major studies which also provide novel insight to innate immune system mutations56, 57 and upregulation of the Janus kinase (JAK) signalling pathway in lesional skin. ⁵⁸ Other factors include involvement of the adaptive immune system, external triggers (e.g., pathergy), and genetic predisposition (Figure 4), although it is poorly understood how these factors interact to influence PG's presentation, natural course, prognosis, or outcomes. ²⁴

FIGURE 4.

Proposed schema of PG pathogenesis as is currently understood. APC, antigen presenting cell; CCR, CC chemokine receptor, CXCR, CXC chemokine receptor; DAMP, damage‐associated molecular pattern; IL, interleukin; MMP, matrix metalloproteinase; PAMP, pattern‐associated molecular pattern; T, thymus; Th, T helper; TLR, toll‐like receptor; TNF, tumour necrosis factor; VEGF, vascular endothelial growth factor

4. TREATMENT

The first proposed treatment for PG by Brocq was surgical excision, which continued to be the mainstay of therapy well beyond his death in 1928.1, 2, 8 By the 1960s, UC was generally thought to occur through autoimmune mechanisms, and an estimated 30% to 40% of PG was associated with UC.30, 59 The strategy up until this point was to administer antibiotics and treat the associated disorder (before proceeding with surgical excision) given the lack of knowledge regarding PG‐specific mechanisms. Indeed, the first randomised controlled trial (RCT) for corticosteroids as a treatment for UC was published in 1955, ⁶⁰ 1 year prior to Wright and Greco's report of PG treated with corticosteroids. ³³ The first clinical trial for azathioprine as a treatment for UC was published in 1966, ⁶¹ 3 years before its first successful treatment for PG. ^30b These first cases of successful treatment of PG either with corticosteroids or azathioprine were novel in that they did not appear to be associated with an underlying comorbidity. Thus, immunosuppressive drugs appeared to be beneficial for PG itself, and the first successful treatment of PG with cyclophosphamide was reported in 1967. ⁶² Dapsone was described as an adjuvant therapy in 1966. ⁶³ Various other successful immunosuppressive or anti‐inflammatory medications were then described, including cyclosporine in 1985, ⁶⁴ oral tacrolimus in 1993, ⁶⁵ topical tacrolimus in 1998, ⁶⁶ intravenous immunoglobulin in 1995, ⁶⁷ and mycophenolate mofetil in 1998. ⁶⁸

The first TNF‐alpha inhibitor to resolve PG was infliximab in 2002. ⁶⁹ Successful treatment with etanercept was described in 2004, ⁷⁰ adalimumab in 2006, ⁷¹ the IL‐inhibitors anakinra ⁷² and canakinumab ⁷³ in 2009 and 2015, respectively, and the IL‐12/IL‐23 antagonist ustekinumab in 2011. ⁷⁴ All these treatments are still considered options depending on previous responses to other available treatments.

The first RCT for PG was not published until 2006, when Brooklyn et al ⁷⁵ assessed the efficacy of infliximab at 5 mg/kg versus placebo. The study enrolled 30 patients, and primary outcome was improvement based on reduction in ulcer size, depth, and degree of undermining over 6 weeks. At week 2, 46% (6/13) of patients in the infliximab group had improved compared with 6% (1/17) of the placebo group (P = .025), and the 23 patients without improvement were offered and accepted infliximab treatment. This study concluded that infliximab was superior to placebo in the treatment of PG, with an overall response rate to infliximab of 69% over 6 weeks. Adverse effects were reported in four patients (one in the infliximab group), including one serious adverse event in the infliximab group and one in the placebo group. ⁷⁴ Only one additional RCT has been performed since then, the STOP GAP (Study of Treatments fOr Pyoderma GAngrenosum Patients) trial in 2015 by Ormerod et al, ⁷⁶ which compared oral cyclosporine (4 mg/kg/day) to oral prednisolone (0.75 mg/kg/day). This study enrolled 112 patients, four of which (3.6%) were lost to follow‐up. The primary outcome was speed of healing over 6 weeks, of which there was no between group difference (adjusted mean difference 0.003 cm²/day, 95% confidence interval −0.20 to 0.21; P = .97). By 6 months, ulcers had healed in 47% of participants in both the cyclosporine group (28/59) compared with the prednisolone group (23/53). Overall, 40 participants in the cyclosporine group (68%) and 35 in the prednisolone group (66%) experienced at least one adverse reaction, including nine serious adverse events: two in the cyclosporine group. This study concluded that prednisolone and cyclosporine did not differ across a range of objective and patient‐reported outcomes. ⁷⁶

Currently, clinical trials for TNF‐alpha ⁷⁷ and IL‐17 inhibitors are ongoing, ⁷⁸ yet the present lack of PG‐specific diagnostic markers, severity indices, and outcomes measures limits the study of different treatments' effectiveness (Table 3). It is the authors' belief that the key to improving the therapeutic approach to PG lies in developing such tools, in addition to understanding its complicated pathogenesis.

TABLE 3.

Evidence for treatments used in pyoderma gangrenosum (PG)

Medication			Evidence level	Selected references
Topical		Corticosteroids	4a	[79, 80]
		Dapsone	4b	[81]
		Phenytoin	4a	[82]
		Tacrolimus	4a	[83, 84]
		Timolol	5	[82b]
		Pimecrolimus	5	[85]
Intralesional		Corticosteroids	5	[86]
Systemic		Azathioprine	5	[87, 88]
		Chlorambucil	4a	[89]
		Clofazimine	4b	[90, 91]
		Corticosteroids	1	[76]
		Cyclophosphamide	4a	[92]
		Cyclosporine	1	[76]
		Dapsone	4b	[93]
		Intravenous immunoglobulin	4b	[94, 95]
		Methotrexate	4b	[96, 97]
		Mycophenolate mofetil	4b	[98]
		Tacrolimus	5	[99, 100]
		Thalidomide	5	[101, 102]
Biologic	TNF‐α inhibitors	Adalimumab	5	[103, 104]
		Etancercept	4b	[105, 106]
		Infliximab	1	[75]
	IL‐1 inhibitors	Anakinra	5	[72, 107]
		Canakinumab	4a	[73]
	IL‐12/23 inhibitor	Ustekinumab	5	[74, 108]

Note: Evidence level: (1) randomised controlled trial; (2) prospective cohort; (3) retrospective cohort, case control; (4) case series (n ≥ 4), (4a) prospective uncontrolled (observational) trial, (4b) retrospective chart review; (5) case report.

Abbreviations: n, number of patients in study.

5. CONCLUSIONS AND FUTURE DIRECTIONS

PG is a complex disease with a rich history (Figure 5). ¹⁰⁹ Its pathogenesis has been perplexing since its discovery, which has posed challenges to diagnosis and treatment. Further molecular characterisation of patients with PG is necessary to clarify PG's pathogenesis and its interrelated factors. The future discovery of diagnostic markers in addition to recently proposed diagnostic criteria for PG will improve diagnosis. Outcomes measures and their validation will also allow for measurement of severity and thus assessment of treatment effectiveness. Notably, there is a striking lack of RCTs of treatments for PG, and more RCTs are likely necessary for a favoured treatment to arise with FDA approval. This review serves to capture the journey to our current understanding of PG, as appreciation of these major discoveries paves the way toward the future.

FIGURE 5.

Timeline summary of major events in PG history ⁷⁵

CONFLICT OF INTEREST

The authors declare no conflicts of interest.

Hobbs MM, Ortega‐Loayza AG. Pyoderma gangrenosum: From historical perspectives to emerging investigations. Int Wound J. 2020;17:1255–1265. 10.1111/iwj.13389