Introduction

The importance of classifying disease has been recognised since ancient times, where treatises in the Hippocratic Corpus described the categorisation of diseases. In these texts, diseases were described in the ad capite ad calcem style, meaning from the top to the bottom of the body, in anatomical order [1]. The importance of the classification of diseases has long since been a foundation of modern medicine, and the ability to categorise with greater precision has allowed clinicians to select tailored treatments and improve patient outcomes.

Although Hippocrates began his classifications of diseases at the top of the body, it is also at the ‘bottom’ that classification is particularly important. Perianal manifestations of Crohn’s disease are common, with 26% of patients with Crohn’s disease developing a fistula within the first 20 years from diagnosis [2]. This particular phenotype can result in complex disease and encompasses marked variation in severity, anatomy, and responsiveness to medical and surgical treatments [3].

Classifying patients with pCD for treatment selection and inclusion in clinical trials is difficult and has taken different forms ever since it was first described by Penner and Crohn [4]. Perhaps the most widely used include Park’s classification [5] and the American Gastroenterological Association (AGA) definitions of simple and complex fistulising disease [6]. These classification systems focus predominantly on anatomical features. Classification systems to categorise non-fistulising features of perianal Crohn’s disease such as strictures, fissures, ulcers and skin tags have also been described. The most ubiquitous of these is the Cardiff–Hughes classification [7].

Recently, the Treatment Optimisation and CLASSification of perianal Crohn’s disease (TOpClass) consortium of experts in pCD developed a novel classification system for fistulising pCD, designed to focus less on anatomical and morphological elements. Instead it classifies disease according to distinct stages of severity requiring different treatment approaches and is based around patient goals [8].

This review article describes both historical and recent advances in the classification of fistulising pCD. We re-evaluate recent clinical trials relating to the treatment of pCD and relate these to the TOpClass classification. Finally, this article discusses the classification of non-fistulising pCD.

The classification of perianal Crohn’s disease

Historical classifications of fistulising perianal Crohn’s disease

The need for a novel clinically relevant classification system was identified in guidelines developed by an expert consensus process in 2014 [9]. A systematic review, later updated by Geldof et al., identified 18 classification systems relating to fistulising pCD [8]. The majority of these systems describe fistulae on the basis of their anatomy or disease activity. The most commonly used anatomical or morphological classifications are the Parks, Cardiff–Hughes and American Gastroenterological Association (AGA) classifications. The seminal Park’s classification, published in 1976, classifies fistulae into intersphincteric, transsphincteric, suprasphincteric and extrasphincteric anatomical positions [5]. This classification system, which was developed from analysis of a large cohort of patients treated surgically, was modified in 2001 to include submucosal fistulae [10]. This terminology remains ubiquitous in both clinical and research settings (Fig. 1).

Fig. 1
figure 1

Parks classification (from Parks et al.) [5]. Type 1 is intersphincteric, type 2 is transphincteric, type 3 is suprasphincteric, and type 4 is extrasphincteric

Full size image

The Cardiff–Hughes classification (Table 1), developed in 1978, classifies pCD according to three categories of disease morphology: ulceration, fistula/abscess and stricture. Fistulising disease is graded on a scale of 0–2 on the basis of the anatomical location (high/low) and anatomy (superficial and complex) [7]. This approach is analogous to the Montreal classification for luminal inflammatory bowel disease (IBD), defining the anatomical extent and nature of the disease [11]. The 2003 AGA classification describes fistulae as simple or complex on the basis of the anatomical level at which the sphincter is involved, number of external openings, associated abscesses or proctitis [6].

Table 1 Classification systems
Full size table

Whilst these classification systems focus on anatomical features, others such as the Fistula Drainage Assessment (FDA) and Perianal Disease Activity Index (PDAI) focus on measures to quantify disease activity. The FDA was developed as an outcome measure for the ACCENT study investigating the use of infliximab to treat pCD and relies on a simple examination of a fistula to determine whether it is active, or closed [12]. The PDAI is also widely used in research, measuring disease activity according to five features: discharge, pain/restriction of activities, restriction of sexual activity, type of perianal disease and degree of induration [13]. Although this index was validated in a cohort of 37 patients during its development, it lacks adequate psychometric measurement properties as identified in a systematic review and COnsensus based Standards for the selection of health Measurement INstruments (COSMIN) appraisal by Joshi et al. (manuscript in progress). Additionally, some features such as the degree of induration remain subjective and may require cross-cultural validation.

In complex perianal fistulising Crohn’s disease, cross-sectional imaging, in particular magnetic resonance imaging, is a key component of a thorough assessment. A number of radiological classifications have been developed, including the St James’s University Hospital Classification (see Table 2) [14]. This characterises fistulae on the basis of their location, complexity and involvement of the anal sphincter, elaborating on the Park’s classification by incorporating features such as abscesses and additional tracts that are visible on MR imaging [14]. More recent advances include MRI-based disease activity indices, such as the Van Assche Index (VAI) and the Magnetic Resonance Novel Index for Fistula Imaging in Crohn’s Disease (MAGNIFI-CD), which aim to provide objective assessments of disease activity on the basis of MRI-derived anatomical features. However, their clinical utility remains limited owing to the complexity of scoring, the requirement for gadolinium contrast and their reliance on largely static anatomical parameters, which may reduce sensitivity to subtle changes in fistula activity.

An ideal classification would be biological in nature, identifying with certainty different versions or stages of the disease process according to biological markers. No such classification nor the biological data to produce one currently exist, so pragmatic, phenotype-based classifications remain necessary but require improvement over historical versions.

Table 2  St James’s University Hospital radiological classification
Full size table

TOpClass classification

The TOpCLASS consortium recently introduced a novel classification system for fistulising perianal Crohn’s disease (pfCD), designed to align patient and clinician goals more effectively by allowing for dynamic movement between classes as disease progression or remission occurs [8]. This classification system, developed by expert consensus and informed by systematic review, heralded an innovative approach categorising patients with fistulising pCD at therapeutically distinct stages through the natural history of the disease [8]. Other key features include the centrality of patient goals, and ability to move between classes, reflecting the chronic nature of the disease and the changing aims of patients as they live with the condition [8]. This represents a pragmatic approach, with clinical descriptors providing greater use in a modern multidisciplinary team (MDT) setting than anatomical descriptors alone. Additionally, the classes allow easier stratification into groups for clinical trials assessing differences in pathogenesis and response to treatment, and recognising the different patient populations and interventions within pfCD but also the different outcomes relevant at different stages of the disease (Fig. 2).

Fig. 2
figure 2

Reproduced with permission from Geldof et al. [8]

The TOpCLASS classification system for perianal fistulising Crohn's disease.

Full size image

The TOpClass consortium recently published treatment guidance relating to each class, based on background evidence for medical and surgical therapies alongside the practices of MDT members at eight IBD centres in Europe, the USA and Australia [15]. The expert panel voted on a series of new statements, and each centre reviewed a series of case vignettes relating to the classification groups. Position statements for surgical treatment in each class are summarised below (Table 3) [15]:

Table 3 Consensus statements on treatment optimisation in perianal Crohn’s disease
Full size table

The TOpClass classification has not yet been validated in a large prospective cohort. However, the classification system has been retrospectively applied to 96 patients with fistulising pCD [16]. The majority of patients in this cohort were initially classified as class 2b, but around 52.1% of patients changed to a different class. Retrospective classification, particularly in terms of patient goals, presents challenges, and further validation in large prospective cohorts is necessary to improve understanding of how patients may transition between classes. The prospective arm of cohort studies such as Goals, Needs and Determinants Of Multimodal therapy in perianal cRohn’s Fistula (GONDOMAR) may provide insights [17].

Non-fistulising perianal Crohn’s disease

Although fistulae are the most common manifestation of pCD, population studies indicate that non-fistulizing manifestations are also frequently observed [18]. ‘Fissures’ and ulcers may occur in up to 1/3 patients, abscesses in around half, strictures in 7% and skin tags up to 10% [18, 19] of patients. Haemorrhoids, anal cancer and regional cutaneous manifestations are associated with pCD but probably do not represent actual pCD manifestations, and data around them are limited.

Non-fistulising disease is also included in a number of the classification systems that are in widespread use for fistulising pCD, such as the Cardiff–Hughes classification (33) or the Buchmann classification [20], which classify largely on the basis of anatomical or morphological features. Other indices that we use frequently to assess disease activity, such as the Perianal Disease Activity Index [13], were validated on patients with both fistulising and non-fistulising pCD, and include elements such as fissures and skin tags.

Several classification systems focussing on specific features of non-fistulising pCD have been described. In 1975, Greenstein categorised anorectal strictures in Crohn’s by length, subdividing into annular < 2 cm strictures, tubular > 2 cm strictures and strictures due to post-ileostomy atrophy [21]. More recently, for a small paediatric case series, strictures were defined as severe or non-severe, with the former being unable to tolerate a colonoscope or digital rectal examination [22]. In terms of ulcerating disease, Horaist et al. performed an a expert consensus process to define and classify common pCD lesions [23], classifying ulcers by depth, extension and location. Fissuring and ulcerating disease has been classified as simple or complicated, with complicated disease defined by the involvement of both sphincters and requiring operative management [24]. The AGA technical review classified skin tags broadly into two types, with type 1 being large, hard and cyanotic, and type two being soft, flat and painless [6].

The wide variety of features that are included in classification systems for non-fistulising pCD emphasises the lack of consensus around which features are related to pCD and which are common proctologic conditions that may present in the CD population. Key considerations for any future classification system would include characteristic features, the interaction between true pCD manifestations and common proctological problems, how to define severity, boundaries and movement between classes, and how to capture the relationship between fistulising and non-fistulising disease. This may enable future trials to capture these manifestations of pCD more effectively and provide further insight about treatment outcomes and prognosis.

Retrospective classification of previous clinical trials

Clinical trials exploring the medical and surgical management of pCD often involve diverse patient cohorts, but they typically provide limited detail on individual patient goals. The PISA II trial was a landmark study that sought to address this balance by incorporating a patient-preference treatment arm into its methodology [25]. The TOpClass classification enables more effective organisation of patients with pCD into homogeneous groups for inclusion into future studies [8]. To date, the TOpClass classification has not been used in the context of a clinical trial evaluating treatments for pCD.

A scoping review of clinical trials in pCD was performed to understand how the TOpClass classification relates to existing evidence. The National Institutes of Health database was searched via ClinicalTrials.gov using the search term “Perianal Crohn’s disease”. The search was limited to trials first posted from 1 January 2000 to 18 June 2024. Only phase 2, 3 and 4 trials were included. The search was conducted on 18 June 2024. Trials that were not yet recruiting, were terminated without results or involved children (0–17 years) were excluded. Twenty-four key studies investigating the management of pCD were identified, with a further three studies identified via citation search. Eight studies were excluded in total because they had been terminated without publishing, were not yet recruiting, did not relate to Crohn’s or did not relate to perianal fistulae. Details of the patient cohorts reported in the included trials were re-assessed by two reviewers, including one senior author, to determine how and whether they could be classified using the TOpClass classification.

Of the 19 studies, 4 studies described cohorts consistent with class 2a. Seven were classified as class 2a or 2b, but were unable to subdivide further. Seven were consistent with class 2a, b, or c but were unable to be classified any further. One study was classified as class 2a or 2c. Overall, eight studies did not specify whether patients with diverting stomas were excluded from the patient cohorts.

Four studies described patient cohorts consistent with the TOpClass 2a group. These were, PISA II (Anti-Tumour Necrosis Factor vs Surgical Closure following Anti-TNF), the ADMIRE trials (adipose-derived mesenchymal stem cells) and a trial investigating darvadstrocel (expanded adipose stem cells – eASC) [25, 26]. Of these, ACCENT II and the BM–MSC trial did not specify whether patients with diverting ostomies were excluded so there may have been patients in class 2c in these cohorts. Stomp II did not exclude patients with stoma, and therefore patients in this cohort with stoma would be grouped into class 2c (NCT04847739). In all these cohorts, it was assumed that, by giving informed consent to take part in these clinical trials, they were indicating a preference for (demonstrating that their goals aligned with) the interventions being investigated. PISA I is a good example of a trial in which the interventions did not all align to a single patient goal [27], which was probably a factor in difficulty in recruitment.

Seven studies described patient cohorts that could be classified as either class 2a or 2b, but it was not possible to subdivide further. These included CHARM, DIVERGENCE 2, Fuzion CD, ExoFlo, USTAP, ENTERPRISE and a study investigating the use of therapeutic dose monitoring in infliximab [28, 29]. These studies included a patient cohort of active fistulae, where patients with stomas were excluded, but the range of fistulae complexity reported would include both fistulae suitable and not suitable for repair.

A total of seven studies described patient cohorts with active fistulae but did not provide adequate anatomical information or information regarding patient goals to classify any further. These trials included ACCENT, ADAFI, CHOICE and studies investigating the use of Fibrin Glue, Topical metronidazole, Bone Marrow Derived Stem Cells (BM-MSC), and the use of endoscopic ultrasound to administer Humira [30,31,32,33,34,35,36]. None of these studies specified whether patients with stomas were included or excluded from their cohorts.

Within the limitations of a retrospective analysis of registered studies, these findings demonstrate the heterogeneity of patient cohorts included in previous clinical trials investigating treatments for fistulising pCD. In particular, this indicates that, whilst the I and C of Patient/Population, Intervention, Comparison, Outcome (PICO) have been chosen carefully and represent the focus of a trial’s design, the P and O have often not been considered in adequate detail. Patients with substantially different disease phenotypes in terms of complexity and suitability for different treatments are frequently grouped together.

For example, the precise impact of defunctioning ostomies on the microbiome and subsequent inflammation in the distant bowel remains unclear. However, it is reasonable to suggest that patients with stomas may respond differently to novel treatments compared with those without, just as might be the case in those with fistulae too complex to repair, or active anorectal disease which precludes repair. Reporting results separately for these groups, which the TOpClass classification would facilitate, could provide more meaningful insights.

Class 2a patients (fistulae suitable for repair) who are willing to undergo surgery are ideal for a trial of surgical repair versus seton removal alone, for example, or surgical repair with or without therapeutic drug monitoring – supported combination medical therapy. Class 2b patients might be included in trials in which the outcomes of interest are quality of life improvement and downstaging to class 2a. Those class 2b patients with high complexity fistulae, or uncontrolled proctitis, might enter trials which compare open versus VAAFT-assisted fistula rationalisation (downgrading a complex fistula not amenable to repair), or different advanced medical therapies, to bring about anatomical or biological rationalisation, respectively.

Researchers can then consider the various options for trial inclusion criteria, interventions and outcomes of interest according to class, facilitating a much wider and richer landscape for studying pfCD at every stage of the disease (Fig. 3; Table 4).

Fig. 3
figure 3

PRISMA diagram for scoping review

Full size image
Table 4 Summary of scoping review
Full size table

Conclusions

This narrative review provides an overview of the classification systems that have been described in the literature to classify fistulising pCD and the described recent advances such as the TOpClass classification system. The re-categorisation of existing clinical trials using the TOpClass classification highlights the practical benefits such a system can deliver in both clinical and research settings. Finally, this article has outlined several existing classification systems in use for non-fistulising pCD and highlighted the need for a novel classification. The many advances in the treatment of pCD in terms of advanced medical therapies, novel procedures such as VAAFT, fistula laser closure (FiLaC) and regenerative medicine are encouraging for the future. There is a need to ensure that there is consistency in how we define, classify and report outcomes in clinical trials.

To return to Hippocrates, it is true that “that which is used, develops… that which is not used, wastes away”. Ultimately, the test of any new clinical measure or classification system is whether healthcare professionals in busy wards and clinics around the world find it to be a useful adjunct to their daily practice, and whether researchers adopt it as they design the next iteration of clinical trials in pfCD. The benefits of the TOpClass classification system will lie in its integration into shared decision-making processes within joint surgical–medical clinics and IBD MDTs, and its role in re-classifying and stratifying patients in clinical trials. It is expected that the TOpClass classification will continue to develop and evolve through active use and improvement in the future, until a true biological classification can replace it.