Microscopic Assessment in Inflammatory Bowel Disease

See “Addition of Histology to the Paris Classification of Pediatric Crohn Disease Alters Classification of Disease Location” by Verstraete et al on page 242 and “Endoscopic Versus Histological Disease Extent at Presentation of Paediatric Inflammatory Bowel Disease” by Ashton et al on page 246. In this issue, 2 separate retrospective single-center studies showed that inflammatory bowel diseases (IBDs), both Crohn disease (CD) and ulcerative colitis (UC), are more extensive if judged by histology compared with macroscopic assessment. Ashton et al (1) showed on 172 children that the rate of panenteric disease increased when using histologic findings from 33% to 57% in CD and pancolitis from 66% to 83% in UC. Fernandes et al (2) reported a 57% discordance rate between microscopic and mascroscopic classification, when applied to 60 children with CD. Moreover, the agreement on disease extent between different raters was better when histologic findings were incorporated in the classification schemes, but not in the upper gastrointestinal (UGI) tract. The latter finding should come as little surprise because histologic UGI may be seen in 50% to 70% of patients with CD, though the clinical relevance of these mostly mild findings is highly questionable (3). Remarkably, nearly 90% of patients with IBD in the first study had microscopic involvement of the UGI tract (1). A classification system that includes the vast majority into 1 group cannot be a robust discriminant tool of the UGI tract. It is well accepted that the traditional IBD categorization into CD, UC, and IBD-unclassified is not discriminative enough to dictate management decisions. Physicians try, therefore, to further phenotype the disease based on the parameters that can predict important outcomes, hence the Montreal classification in adults and the Paris modification for children (4). Similarly, the choice between the 2 contending classification themes (ie, macroscopic vs microscopic) should be based on clinical relevance. The 2 studies in this issue elegantly highlight the discrepancy in the themes but do not necessarily prove which is superior. The discussion on how to classify disease extent goes hand in hand with a broader and even more important dispute—whether histology should be used to monitor response to treatment and quantifying disease severity. If we are not prepared to use histology as a practical goal of treatment, then there is little justification to use it for classifying the disease at the outset. Achieving histologic remission is a high hurdle to achieve, likely requiring multiple combined aggressive immunosuppressive drugs. In a study, only 13% of patients treated with long-term anti-tumor necrosis factor regimens had histologic remission (5). Similarly, only one-third of those with mucosal healing (MH) in the ACT trials had complete histologic remission (6). More than 80% of the patients with CD who had macroscopic healing with anti-tumor necrosis factor had residual microscopic inflammation in at least 1 biopsy (7); other studies showed approximately 30% to 40% histologic activity in those with deep remission (8). Although the exact utility of histologic remission in predicting long-term disease outcomes is yet to be determined, it is still plausible to assume that the deeper the remission, the better the control of the disease with all its long-term benefits (8). Indeed, histologic remission has been shown to be superior to macroscopic MH in predicting long-term remission (8–12) and for cancer prevention (13). Simply showing superiority is, however, insufficient to make practice recommendations. It is the effect size (ie, the degree of added benefit of histology over clinical and biochemical remission or macroscopic remission) that is of importance. The number needed to treat to achieve 1 histologic remission over macroscopic remission is still unknown. The number needed to treat must then be balanced against the cost and risks of the required added therapies to achieve this extended goal. In fact, there is still a controversy even in the definition of the desired MH associated with “macroscopic” remission. In UC, the evidence is contradicting whether complete MH (ie, Mayo endoscopic subscore 0) is clinically superior to near MH (ie, subscore 1), and in CD it is customary to define MH as lack of ulcers, allowing for some residual minimal macroscopic inflammation. Moreover, it is mandatory that all outcome measures show sufficient reliability, validity, and, if used to detect change over time, responsiveness. Although several multi-item measures of histologic activity exist in IBD, such as the Geboes or Riley scores in UC and the Global Histologic Disease Activity Score in CD, none were developed and validated following accepted psychometric criteria (8,14). Indeed, the 2 studies in this issue used inventory definitions of histologic remission. A large adult study showed concordance between macroscopic and microscopic degree of inflammation in UC only for the extreme groups of those in remission and those with severe disease, eluding to poor interobserver reliability (15). In this issue, Fernandes et al (2) showed that the agreement between gastroenterologists was higher when histology was incorporated in the Paris classification. This is an important and first step toward a more comprehensive psychometric evaluation but the reliability of 2 independent pathologists must now be also determined. The importance of thorough psychometric development and evaluation of our measurement tools should not be underestimated. Accepted cut off values that define lack of activity and categorization into mild, moderate, and severe activity must be determined before incorporating into clinical practice. It is still unclear how to address borderline histologic findings (eg, chronicity with minimal active inflammation, predominant eosinophilic infiltrates, gastritis, duodenitis in the presence of Helicobacter pylori or nonsteroidal anti-inflammatory drugs use). Furthermore, histologic assessment is highly dependent on the number of biopsies obtained. If histology is to be used, data are needed to support standardization of the required number of biopsies per segment, the recommended locations, and how to approach contradicting results from adjacent biopsies. During the past 15 years, we have gradually moved from measuring symptoms only to ensuring normalization of serum inflammatory markers, and now to calprotectin and macroscopic MH. Measurements in IBD are certainly a moving target. Although we are still struggling with optimizing our present measurement tools, the 2 articles in this issue set the stage to the next discussion whether our classification scheme and disease management in IBD should be taken to the next level. The discussion is timely and of great importance; however, there are still many open questions needed to be addressed before histology can be incorporated in the clinical management of our patients; primarily understanding whether the costs of such action worth the benefits. Regardless, an accepted valid, reliable, and responsive measure of histology with a clear practical user's guide is badly needed to advance also this field.