A18 DIETARY PREDICTORS OF BIOLOGICAL ACTIVITY IN CROHN’S DISEASE: A RETROSPECTIVE COHORT STUDY

Abstract Background Patients with Crohn’s disease (CD) often seek advice on optimizing their diet to reduce gut inflammation. The relationship between dietary patterns, major food groups and individual nutrients, with disease activity in Crohn’s disease (CD) is incompletely understood and warrants further investigation. Aims 1.To determine whether a diversified (DD) or nondiversified (NDD) dietary pattern is related to biological activity in CD (BACD) in long-term follow up. 2.To determine if specific foods or nutrients are associated with increased BACD. Methods In this retrospective cohort study, forty-six CD patients (52% male) in remission completed 3-day food records between 2015–2017 for a 3-month intervention study and were classified as DD or NDD. Remission was defined by a Harvey Bradshaw Index <5 and no endoscopic ulcerations within 6 months of baseline data collection. Patients were classified as NDD if dietary fibre was ≤15 g/day or total fruit/vegetable servings ≤3/week, and if they consumed ≥3 servings/week of red and processed meat. Patients were otherwise defined as DD. A retrospective chart review captured BACD data. BACD was defined as one of either fecal calprotectin (FCP) ≥250 ug/g, hospitalization for CD flare, bowel resection for active CD, biologic dose escalation/switch due to non-response (not therapeutic drug monitoring), corticosteroid use, endoscopic evidence of apthous or large ulcers, or active disease on contrast enhanced ultrasound or magnetic resonance enterography. Machine learning methods with random forest prediction models assessed if diet composition was associated with BACD followed by univariate Mann-Whitney tests to compare differences between high and low disease activity. Results Sixteen patients (35%) had BACD during the mean 42 month follow up (31–54 months,SD ± 6.6). See Table 1 for additional demographics. Based on the random forest prediction model, both vitamins and minerals, food groups and Mediterranean diet cut-points could predict disease activity responses (ROC-AUC = 0.68 and 0.75, respectively). For these models, baseline intake of vitamins E, D, B1, and C and leafy greens, and fruit intake were the most important predictors of BACD. For the univariate analysis, the high disease group had lower intakes of fiber, vitamin E, and C (p = 0.047, 0.066, and 0.09, respectively). A higher proportion of patients consumed a NDD with BACD compared to those without BACD (50% vs. 23.3%, p=0.07). Conclusions To our knowledge, this is the first study to assess if dietary patterns, foods and nutrients are able to predict disease activity over a mean 42 month follow up. Further research into the dietary determinants of BACD in CD is warranted. With higher baseline FCP observed in the BACD, multivariate analyses to assess the independent effect of diet to predict BACD is required. Funding Agencies Litwin IBD Pioneers Foundation, Alberta’s Collaboration of Excellence for Nutrition in Digestive Diseases (Ascend)

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.