Selecting High-Risk Patients With Pediatric Crohn’s Disease for Top-Down Anti-TNF as per the 2021 ECCO-ESPGHAN Guidelines: A 5-Year Nationwide Retrospective Analysis From Scotland (2016-2020)

Abstract

Background

The 2021 ECCO-ESPGHAN guideline on the medical management of pediatric Crohn’s disease promotes early risk stratification and top-down anti-TNF for patients deemed high risk of severe disease course.

Aims

We aimed to objectively assess the risk-benefit profile of the guideline’s risk stratification policy and guidance on top-down anti-TNF in a nationwide population-based cohort study.

Methods

Using a prospectively identified nationwide cohort of all new pediatric patients (<17 years) diagnosed with Crohn’s disease in Scotland between January 1, 2016 and December 31, 2020 and retrospectively applying the current management algorithm, we explored the guideline’s ability to accurately risk stratify patients. Phenotypic and treatment data were retrospectively collected from electronic medical records with a maximum follow-up of 18 months post-diagnosis.

Results

Four hundred and eighteen children (258/418 [62%] male; median [interquartile range {IQR}] age at diagnosis: 13.2 [11.2-14.8] years) were included. High-risk phenotype was present in 224/418 (54%) with 53/224 (24%) of high-risk patients not requiring anti-TNF therapy within 18 months of diagnosis. Conversely, 78/194 (40%) of the low-risk group received anti-TNF within 18 months. High-risk patients were more likely to require anti-TNF (171/224 [76%] vs 78/194 [40%], P < .001) and had shorter median (IQR) time to anti-TNF start (5.0 [1.0-8.0] months vs 6.5 [3.3-13.0] months, P = .01).

Conclusions

Our data support the guideline’s ability to identify patients more likely to require early treatment escalation. However, this approach would have led to potential over- and under-treatment in a substantial proportion of patients. This underscores the importance of frequent and comprehensive monitoring, along with flexible treatment strategies that adapt to changes in disease status.

Introduction

Crohn’s disease (CD) is a chronic inflammatory disorder that can affect the length of the gastrointestinal tract and is the most common form of inflammatory bowel disease (IBD) in children, who have more extensive disease than adults.^1,2 Precision medicine where patients can accurately be stratified using their individual disease and biological characteristics allowing better prediction of disease course and treatment response is the aim of future pediatric IBD (PIBD) management.³ However, as a complex disease with multiple etiologies and pathogenic mechanisms resulting in heterogenous phenotypes, there are enumerable challenges to achieving this goal.

In this context, the European Crohn’s and Colitis Organisation (ECCO) and the European Society for Paediatric Gastroenterology, Hepatology and Nutrition (ESPGHAN) have played a central role in establishing evidence-based guidelines for the management of pediatric CD. The 2014 ECCO-ESPGHAN guidelines⁴ encompassed a comprehensive set of recommendations that reflected the most current understanding of pediatric CD management. These guidelines were shown to positively impact disease outcomes, with a lower risk of relapse and surgery, though no impact on disease-related complications.⁵ The 2021 iteration built on this and promoted a move to early risk stratification, based on clinical and radiological indices, and a “top-down” (anti-TNFα within 4 weeks of diagnosis) approach for patients who were deemed high risk.⁶ High risk was defined by extensive panenteric or severe disease, perianal, stricturing, and/or penetrating behavior. Those with low-moderate risk receive accelerated step-up treatment beginning with exclusive enteral nutrition (EEN) or steroids with early assessment and treatment escalation if treatment targets are not met by 12 weeks.

No nationwide population-based data exist that have objectively assessed the risk-benefit profile of this new approach. Given the profound impact on clinical decision-making, it is crucial to assess how the introduction of the 2021 guidelines might influence real-world practice compared to preceding years. By retrospectively applying the current management algorithm to a nationwide cohort managed between 2016 and 2020, we aimed to elucidate the extent of the guideline’s ability to accurately risk stratify patients and correctly identify those who require early escalation to anti-TNF therapy.

Methods

Case Ascertainment

Scotland had a population of 5 479 900 at mid-2021,⁷ and the point prevalence of PIBD is one of the highest reported worldwide (106/100 000 in those aged <17 years).⁸ Specialist pediatric gastroenterology and IBD care are provided through managed clinical networks, which include all district general hospitals, coordinated by 3 tertiary centers (West—based in Glasgow, South-East—based in Edinburgh, and North—based in Aberdeen). Each pediatric center maintains a prospective registry of patients with IBD within their network along with detailed biologic prescription data for the Scottish PIBD Biologics Registry,⁹ allowing complete nationwide case accrual.

Using our ongoing Scottish PIBD Biologics Registry, we analyzed a prospectively identified nationwide cohort of all new patients diagnosed with pediatric CD (<17 years) in Scotland in PIBD services over a 5-year period, before the introduction of the ECCO-ESPGHAN guidelines, between January 1, 2016 and December 31, 2020. Each patient in Scotland carries a unique Community Health Index (CHI) number, avoiding duplication of patients. Each patient’s diagnosis was reviewed and confirmed based on the revised Porto criteria.¹⁰ Data were retrospectively collected from electronic medical records with a maximum follow-up of 18 months post-diagnosis. The data included basic demographics, anthropometrics, disease phenotype, medication history, disease course, and surgical history. Paris location¹¹ and behavior at diagnosis were determined, allowing stratification of this historic cohort into high risk or low-medium risk based on the current criteria as defined by the ECCO-ESPGHAN guidelines.⁶ This allowed analysis of a cohort diagnosed and managed prior to the introduction of the guidelines to allow us to assess the accuracy of early risk stratification in predicting the subsequent need for anti-TNF. A high-risk phenotype was defined as deep colonic ulceration, extensive small bowel disease, fistulizing perianal disease, growth failure/pubertal delay, stricturing, and/or penetrating luminal disease. Outcome data were collected regarding any escalation to anti-TNF therapy or surgical intervention within 18 months of diagnosis. Data on anti-TNF therapy included type of anti-TNF therapy, indication, and time from diagnosis to commencing anti-TNF therapy. Surgical intervention was defined as either perianal intervention (Seton suture or drainage of abscess), IBD-related bowel resection, or IBD-related bowel resection with stoma formation.

Statistical Analysis

The study was reported as per the STROBE statement for observational cohort studies. Categorical variables were expressed as counts and frequencies (%). Non-normally distributed variables were described as median and interquartile range (IQR), while normally distributed variables were described as mean and SD. Differences in continuous variables were analyzed using the Mann-Whitney U test or 2-sample t-test as appropriate, whereas differences in categorical data were explored using Fisher’s exact test. The standard Kaplan-Meier method was used to assess the proportion of patients in both high- and low-risk groups that required escalation to anti-TNF within 18 months of diagnosis. Multiple group comparisons were analyzed using 1-way analysis of variance. Differences in the statistical significance was set at a P value of <.05. Statistics were performed using R-Studio (version 4.1, Posit Software, 2022) and GraphPad Prism (version 10.3.0 for Mac, GraphPad Software, 2024). Anthropometric z-scores were calculated using LMSGrowth (Harlow Healthcare, London, UK) based on the Medical Research Council British 1990 growth reference.¹²

Ethical Considerations

In line with national practice, ethics approval was not required as this was a review of service delivery and clinical practice.¹³

Results

Cohort Characteristics and Phenotype

Four hundred and eighteen pediatric patients (258/418 [62%] male; median [IQR] age at diagnosis 13.2 [11.2-14.8] years) were diagnosed with CD in Scotland between January 1, 2016 and December 31, 2020: 230/418 (55%) in the West, 105/418 (25%) in the South-East, and 83/418 (20%) in the North. The median (IQR) weight z-score at diagnosis was −0.3 (−1.32 to 0.6), and the height z-score at diagnosis was −0.03 (−0.76 to 0.91). An overview of the Paris location and behavior for the total cohort is provided in Tables 1 and 2, respectively. Patients were considered to have a high-risk phenotype at diagnosis in 224/418 (54%). As a proportion of the total cohort, 166/418 (40%) had extensive disease or deep colonic ulcers, 39/418 (9%) had fistulizing perianal disease, 49/418 (12%) had stricturing disease, and 5/418 (1%) had penetrating disease; 33 patients had multiple high-risk disease behaviors (breakdown of patients with multiple high-risk behaviors available in Supplementary Table 1). The remaining 194/418 (46%) had an inflammatory phenotype with no features of severe disease and were considered as having a low-medium-risk phenotype (Table 3). Follow-up data to 18 months post-diagnosis were available for all 418 patients, with no patients lost to follow-up.

Patient Management

Therapy was non-mutually exclusive as prescriptions per patient often overlapped within the first 18 months post-diagnosis. Examining the non-biologic management of the total cohort, 333/418 (80%) received EEN, 201/418 (48%) received steroids, 341/418 (82%) received a thiopurine either as monotherapy or co-immunosuppression, and 53/418 (13%) received methotrexate.

In the total cohort, anti-TNF was commenced within 18 months in 249/418 (60%). Of those in the high-risk group, 171/224 (76%) received an anti-TNF: 26/171 (15%) within 4 weeks, 110/171 (65%) within 1-12 months, and 35/171 (20%) within 12-18 months of diagnosis. Additionally, 53/224 (24%) of patients with a high-risk phenotype at diagnosis received no anti-TNF therapy within 18 months. High-risk patients who did not receive anti-TNF had a higher proportion of isolated colonic disease compared to those who received anti-TNF, though no other significant differences in Paris location or behavior were observed (Table 4). In the low-risk group, 78/194 (40%) in total received an anti-TNF: 9/78 (12%) within 4 weeks, 44/78 (56%) within 1-12 months, and 25/78 (32%) within 12-18 months (Figure 1). Patients were more likely to commence an anti-TNF within 18 months of diagnosis in the high-risk group (high risk: 171/224 [76%] vs low risk: 78/194 [40%], P < .001) and had a shorter median time to commencing the first anti-TNF (high risk: 5.0 [1.0-8.0] months vs low risk: 6.5 [3.3-13.0] months, P = .01) (Supplementary Figure 1). When examining the impact of each high-risk phenotypic feature, this significance was maintained, with no one high-risk phenotype driving the difference, and each combination having comparable rates of anti-TNF use (Supplementary Figure 2).

Regarding the type of anti-TNF, 190/249 (76%) received infliximab, and 59/249 (24%) received adalimumab. Patients moved within class from either infliximab to adalimumab or vice versa for immunogenic loss of response in 20/249 (8%) (15 initially commencing infliximab, 5 adalimumab). There was no statistical difference in the rate of immunogenic loss of response between high- and low-risk groups (high risk: 13/190 [7%] vs low risk: 7/59 [12%], P = .54). Second-line biologic therapy (ustekinumab) was commenced in 1 patient at 11 months post-diagnosis due to primary non-response to infliximab. This patient had small bowel and perianal disease.

The indication for anti-TNF varied between groups. In high-risk patients, 77/171 (45%) were commenced due to a severe phenotype at diagnosis, 60/171 (35%) failed immunomodulator monotherapy, 32/171 (19%) had primary non-response to induction, and 2/171 (1%) escalated for other reasons. In the low-risk group, 13/78 (17%) were commenced due to evolution of their disease to a severe phenotype after diagnostic investigations were completed (7 developed fistulizing perianal disease and 6 developed extensive small bowel disease), 13/78 (17%) developed acute severe colitis unresponsive to steroids, 33/78 (42%) failed immunomodulator monotherapy, 11/78 (14%) were steroid-dependent, and 8/78 (10%) escalated for other reasons including management of extra-intestinal manifestations.

Surgery

Surgery was required in 43/418 (10%) of the total cohort. Patients with high-risk disease were more likely to require surgery than those with low-risk disease (37/224 [17%] vs 6/194 [3%], P < .001). Considering the type of surgery, resection with primary anastomosis was required in 17/37 (46%) and 4/6 (66%), resection and stoma formation in 3/37 (8%) and 1/6 (17%), and surgical management of perianal disease in 22/37 (60%) and 2/6 (33%) in high- and low-risk patients, respectively. There was no significant difference in the median (IQR) time to surgery (high-risk: 1 [0-3] months vs low risk: 1.5 [0-14] months, P = .503). A breakdown of the timing of anti-TNF commencement in relation to surgery is outlined in Table 5. A greater proportion of high-risk patients required surgery before commencing anti-TNF compared to low-risk patients though this was not statistically significant (22/37, 59% vs 2/6, 33%, P = .38). The proportion of high-risk patients who required early surgery (prior to commencing anti-TNF) constituted 10% of the total high-risk cohort (22/224). Within the high-risk group, more patients required surgery before commencing anti-TNF than in those who were established on anti-TNF (pre-anti-TNF: 22/37 [59%] vs post-anti-TNF: 11/37 [30%], P = .011) or never required anti-TNF (pre-anti-TNF: 22/37 [59%] vs no anti-TNF: 4/37 [11%], P < .001). Additionally, 3 high-risk patients required surgery under 4 weeks after starting anti-TNF (2 right hemicolectomy and 1 small bowel resection).

Discussion

The findings of our nationwide study exploring the impact of the 2021 ECCO-ESPGHAN guidelines on a historic cohort of patients with CD offer valuable insights into the complexities of clinical decision-making and underscore the need for ongoing refinement of treatment algorithms. The period between 2016 and 2020 witnessed a change in the management of pediatric CD in Scotland, with an inexorable rise in biologic use from 20% to 44%¹⁴ in this high-PIBD-incidence country.¹⁵ This was facilitated by increased biosimilar penetrance from 3% to 91% over the same period, improving overall cost-effectiveness.¹⁴ Hence, our cohort already had a comparatively high degree of biological utilization prior to the implementation of the guidelines.

Effective early disease management may modulate subsequent disease course and prevent more severe disease phenotypes from emerging.¹⁶ The best approach to this remains a topic of debate. The ECCO-ESPGHAN guidelines recommend top-down anti-TNF for high-risk patients and an accelerated step-up approach for low-medium-risk patients.⁶ Top-down anti-TNF has been shown to be both clinically efficacious and cost-effective in adult¹⁷ and pediatric^18-20 cohorts in the short-to-medium term; however, long-term follow-up data have failed to show meaningful differences in important endpoints including clinical, biochemical, and endoscopic remission, surgery rates, or hospitalizations.^21,22 While there are clearly patients who need up-front biological therapy, there are many patients who will achieve disease control without biologics. The ability to predict disease course early after diagnosis and stratify patients to their appropriate therapy is the ultimate goal. Our data largely support the ability of the guidelines to differentiate between groups; however, there was discordance between risk stratification and treatment decisions with a substantial proportion of high-risk patients (24%) not requiring anti-TNF therapy within the initial 18 months post-diagnosis. Conversely, 40% of low-risk patients received anti-TNF therapy, potentially indicating initial under-treatment. The ability of our low-risk patients to transition into severe phenotypes (34% in this cohort) during the early phases of the disease highlights the dynamic and unpredictable nature of pediatric CD progression. As such, the importance of frequent and comprehensive monitoring is underscored, as well as the need for flexible treatment strategies that can adapt to changes in disease status. The observed transitions also emphasize the potential limitations of risk stratification based solely on initial presentation and highlight the importance of the guideline’s recommendation for accelerated step-up early in the disease course in those not responding to initial management strategies by 12 weeks.⁶ It is essential that this message is understood and implemented and the emphasis is not placed solely on immediate biologics for high-risk disease. Based on their results in the PIBD SETQuality inception cohort, the authors highlighted the need to broaden the scope of indications for early anti-TNF. They also noted that patients without any predictors of poor outcome, as well as those with no or mild disease, should still follow a step-up approach.²⁰ Recent adult data are increasingly supportive of top-down therapy and there may well be a shift towards this in clinical practice.²³ As we have shown, this approach may well lead to over-treatment of a substantial proportion of patients. How this is accommodated into clinical practice in the next iteration of guidelines will be of great interest.

Increasing anti-TNF use has been paralleled with decreased surgical resection rates in pediatric CD.²⁴ In our cohort, high-risk patients had an increased likelihood of requiring surgery with a proportion (10%) requiring surgery at diagnosis, before anti-TNF was commenced. This study is not able to comment on the modulation of risk of surgery with “top-down” anti-TNF but gives further evidence of the utility of current risk stratification criteria. While a minority of low-risk patients required surgical intervention, some patients still extended their inflammatory phenotype or developed steroid-refractory acute severe colitis or perianal disease within the first year. This again highlights the necessity for a high level of clinical vigilance early in the disease course to identify those with rapidly evolving disease.²⁵

We note with interest that 13 of the low-risk patients in our cohort required escalation to anti-TNF therapy due to acute severe colitis which did not respond to steroid therapy. Pediatric acute severe colitis is a clinical diagnosis of severe colitic symptoms, defined by a Paediatric Ulcerative Colitis Activity Index (PUCAI) score of 65 or higher.²⁶ It is not dependent on an underlying diagnosis of ulcerative colitis, and it is well recognized that pediatric CD can present in this manner, as demonstrated in our cohort. Conventional management for any child presenting with acute severe colitis, outlined in the 2018 ECCO-ESPGHAN Ulcerative Colitis Consensus Guideline, recommends early escalation to second-line medical therapy (usually anti-TNF therapy in the form of infliximab) in patients not responding to intravenous steroids.²⁶ The 2021 ECCO-ESPGHAN CD guidelines make no specific mention of acute severe colitis. We do wonder whether future iterations of pediatric CD guidelines should incorporate acute severe colitis as a marker of high-risk disease (and need for early anti-TNF therapy) if present at diagnosis, or as a clear marker of a change to high-risk phenotype if developing later, and requiring escalation to anti-TNF therapy.

It is important to acknowledge the limitations of our study, including the potential impact of the SARS-CoV-2 pandemic on standard models of PIBD care, and the retrospective nature of the analysis. We also recognize the potential limitation of using requirement for treatment escalation as an outcome measure when this is non-standardized and at the discretion of practicing clinicians. However, as the 2014 ECCO-ESPGHAN guidelines were embedded in practice in all Scottish centers by the beginning of this study period, we would expect this to have reduced variations in practice. We also acknowledge the potential that the 2014 ECCO-ESPGHAN guidelines may have influenced the selection of high-risk patients for anti-TNF therapy in this cohort of patients. However, although the 2014 guidelines identified potential predictors of poor outcome, including extensive disease, stricturing or penetrating disease, and severe perianal disease at diagnosis, first-line anti-TNF therapy was only specifically recommended for patients with active perianal fistulizing disease.⁴ We note that in the current study, even when these patients (those with perianal or penetrating disease) are removed from the analysis, the rate of escalation to anti-TNF therapy in the remaining high-risk patients is unchanged.

In conclusion, our study provides insights into the utility of the risk stratification system recommended in the 2021 ECCO-ESPGHAN guidelines. It will be some time before an individual’s genetic and molecular “multiomic” profile can be used for prospective risk stratification in routine clinical practice, and the guidelines remain the current clinical standard for stratifying therapeutic approaches to disease severity. The observed degrees of potential over-treatment and under-treatment, the potential for rapid disease evolution, and the association between risk stratification and surgical outcomes all highlight the need for frequent and proactive disease monitoring to achieve the best outcomes for young patients with this complex disease.

Funding

D.I.F.W. has been supported by an Edinburgh Children’s Hospital Charity (ECHC) research fellowship.

Conflicts of Interest

G.S., D.C.W., R.H., and I.C. have no conflicts of interest to declare. D.I.F.W. received an honorarium from Nutricia-Danone.

References

Author notes