Multicenter registry of pediatric inflammatory bowel disease from a developing country

This multicenter national study reports several points in PIBD that may differ from the existing literature. Among the 72 included patients, we found that the mean age at diagnosis was 7.9 years with 38% being with VEO-IBD. The age at diagnosis in our cohort was younger when compared to previous studies from EUROKIDS (12.5 years), Korea (13.6 years), and Singapore (10.5 years) [5, 14, 15]. Epidemiologic studies from Korea demonstrated an increasing trend in incident cases with decreasing age at diagnosis [16, 17]. A recent multinational Asian PIBD cohort in 2022 reported a younger age at diagnosis (mean age of 9 years) with 29.3% VEO-IBD [4] The differences in the age at diagnosis may be caused by the following: (1) varied genetic predisposition either with ethnic difference or potentially monogenic IBD, and (2) early life environmental triggers such as childhood acute infectious gastroenteritis and/or antibiotics use [18]. We noted a male predominance in the overall PIBD and more significance in CD, similar to the previous reports [15, 16].

With regards to the presenting characteristics, only 44% exhibited abdominal pain, which may be due to the fact that these children were younger at diagnosis. A study from India that included 197 patients aged ≤ 14 years also reported that children with VEO-UC (< 6 years) had a lower proportion of expressed abdominal pain when compared to the older UC ones [19]. UC children were more commonly presented with bloody stools when compared to CD, while CD had several presenting features that define more pronounced systemic inflammation [20, 21]. Almost one-third of CD children in our cohort presented with perianal lesions. Anemia for age, thrombocytosis, elevated systemic inflammatory markers and hypoalbuminemia are the characteristic blood markers for PIBD [22]. A meta-analysis of individual patient data revealed that the aforementioned laboratory tests provided an area under the curve that can differentiate children with vs. without PIBD with an acceptable to excellent discriminative capacity (area under the curve of 0.76–0.84). However, the single best marker was fecal calprotectin (with an area under the curve of 0.95), [23]. Unfortunately, the test was not commonly performed in our country due to its unavailability in the past. Interestingly, we found that only a small proportion (9.7%) of PIBD children had normal ESR and CRP. A study from the UK revealed that among 256 PIBD children, 21.9% had normal ESR and CRP and 9% had all normal blood tests [21]. Therefore, even with normal inflammatory markers, clinicians should be aware of PIBD in young children presented with suggestive signs and symptoms and consider referral for GI endoscopy and biopsy to confirm the diagnosis. We could not observe the differences in the initial disease burden or severity when stratifying by the VEO-IBD status (Table 2). Conflicting with our result, a recent report from Canada demonstrated that the infantile-onset IBD had worse outcomes than the older children [24].

A large proportion (65%) of children in our cohort were given at least one form of systemic corticosteroids as an initial treatment with the aim to induce disease remission, which would likely be due to the fact that most children presented with moderate to severe disease both in CD (69%) and UC (56%). Huang et al. reported that Asian PIBD children initially presented with moderate to severe disease in 54% of CD and 49% of UC cases [4]. Another potential explanation was that exclusive enteral nutrition was not widely used during the study period therefore most CD children with mild to moderate disease severity likely received systemic corticosteroids early on after the PIBD diagnosis. Aminosalicylates and azathioprine were the mainstay maintenance medications for UC and CD children (91.7% and 82.6%), respectively (Table 3). A small proportion of children received infliximab during the first 3 months after diagnosis (17.1% in CD and 5.6% in UC) and at 1 year after diagnosis (17.1% in CD and 13.9% in UC). It was not until 2022 that the Thai government pays for the anti-TNF therapy via the universal health coverage program, therefore treatment with anti-TNF was limited during the study period due to its high cost. Recently, ECCO-ESPGHAN statement suggested an ‘up-front’ anti-TNF therapy in CD children with predictor(s) of poor outcome such as inflammatory (B1) phenotype with extensive disease involving small bowel and colon, B1 with perianal disease, B2 and/or B3 phenotypes [25]. This strategy would result in a higher rate of remission and better long-term outcome with less surgical requirement. However, the number of patients in our cohort who received infliximab was relatively small which limited the robust statistical analysis.

Among 62 children (86% follow-up) with available data at 1 year after diagnosis, almost one-third failed to enter clinical remission with a non-significant difference between CD and UC. Among those who failed to enter clinical remission and underwent a follow-up endoscopy, only 18.8% had documented MH when compared to 40.9% among children who successfully entered clinical remission at 1 year after diagnosis. We found several variables in the univariable analyses associated with failure to enter remission: diarrhea, weight issues, hypoalbuminemia, and moderate-severe disease. However, diarrhea (OR 9.3) and weight issues (OR 4.9) at the initial presentation remained significant independent predictors in the multivariable analyses. Several adult studies in UC have shown that frequent bowel movements and low albumin level predicted treatment failure, colectomy [26], and greater usage of immunosuppression [27]. Growth failure is considered a poor prognostic outcome in PIBD especially among CD children therefore these patients may have a lower capacity to achieve clinical remission.

Almost half of the patients remained on corticosteroid at 1-year after diagnosis. Two predictors showed a trend of significance in the univariable analysis but became significant in the multivariable analysis (females [OR 3.1] and CD [OR 3.0]) (Table 5). Gender difference in the use of steroids is difficult to explain by biological plausibility. In contrast, an adult study found that UC females had a higher remission rate and lower colectomy rate at 1 year when compared to males [28], the results may imply some differences as compared to our study. CD initially had a higher proportion of children who receive corticosteroids from the beginning when compared to UC which may explain the finding of significance at a 1-year follow up. However, a 19% drop in the use of 5-ASA was noted in the UC children. We noted that 8 patients had 5-ASA discontinued and 5/8 possibly had intolerance. We therefore hypothesized that some proportion of these children may suffer from 5-ASA intolerance which has been previously reported in 13.8% [29]. As previously mentioned that only 15.5% received infliximab at 1 year after diagnosis, these children may not have their disease under good control with highly efficacious maintenance medications such as anti-TNF therapy hence leading to the need of systemic corticosteroids.

We are aware of some limitations in our study. Most data were retrospectively collected and may subject to missing information. We did not routinely perform genetic and/or immunologic testing in all VEO-IBD cases during the earlier years. The information on small bowel evaluation in our cohort were limited therefore the finding of disease location for CD may be different when compared to other studies. There was a limited number of patients received biologic agents and the ones with follow up fecal calprotectin and endoscopy to define MH (i.e., endoscopic remission) which is now considered as an optimal target of IBD management both in children and adults [30]. Both multivariable analyses also had limited number of patients to perform robust analyses especially during the forward stepwise selection of interested variables to make stable models, therefore only few strong predictors could be added into the final multivariable models.