AI-Driven Benefit-Risk Analysis for Ulcerative Colitis

The ORAL Surveillance boxed warning was derived from RA patients aged ≥50 with at least one cardiovascular risk factor. UC patients are typically younger (median age at diagnosis 30–40 years) with lower baseline cardiovascular risk. ArcaScience models the absolute risk difference by adjusting for age distribution, baseline CV risk factor prevalence, inflammatory burden (which itself contributes to thrombosis in IBD), and JAK selectivity. We integrate OCTAVE long-term extension data, U-ACCOMPLISH cardiovascular event rates, and real-world IBD registries to provide UC-specific absolute risk estimates rather than extrapolated relative risk from the RA population.

Switching from biologics (anti-TNF, vedolizumab) to small molecules (tofacitinib, upadacitinib, ozanimod) is increasingly common in UC as patients cycle through therapies. ArcaScience models the washout period safety considerations including overlapping immunosuppression risk, disease flare during transition, and the immunogenicity implications of prior biologic exposure on subsequent therapy response. For patients switching from anti-TNF to JAK inhibitors, the platform accounts for residual anti-TNF drug levels and their interaction with JAK inhibitor immune modulation. This is particularly relevant for acute switches where washout periods are compressed due to active disease.

Yes. Acute severe UC (ASUC) is a medical emergency where BRA must balance rapid rescue therapy against surgical timing. ArcaScience models the benefit-risk of IV corticosteroid rescue, cyclosporine versus infliximab second-line rescue, and the emerging role of tofacitinib as rapid-acting rescue therapy. The platform integrates Travis criteria, Oxford ASUC scoring, and surgical complication rates for delayed versus early colectomy. Critically, the BRA accounts for the risk of medical over-treatment delaying necessary surgery, which increases post-operative morbidity and mortality.

Pregnancy planning is a critical BRA consideration in UC given the young patient population. ArcaScience models the reproductive safety profiles across all UC therapy classes. Anti-TNF agents (infliximab, golimumab) have extensive pregnancy safety data and are generally continued through pregnancy. In contrast, tofacitinib and upadacitinib are teratogenic (FDA pregnancy category X equivalent) requiring washout before conception. Ozanimod requires a 3-month washout due to prolonged lymphocyte recovery. Our platform models the disease flare risk during mandatory washout periods, optimal therapy transition strategies for conception planning, and the benefit-risk of continuing vedolizumab (pregnancy data reassuring but limited) versus switching to anti-TNF before conception.

The traditional step-up approach (5-ASA to immunomodulators to biologics) versus early aggressive therapy (top-down) represents a fundamental BRA question in UC. ArcaScience models cumulative safety exposure across both strategies, including the long-term immunomodulator risks (lymphoma with thiopurines, hepatosplenic T-cell lymphoma with combination therapy), progressive disease damage from delayed effective therapy, and the potential to de-escalate after achieving deep remission. The platform integrates CALM trial methodology with UC-specific data to quantify whether early intensive therapy's short-term safety burden is offset by reduced long-term colectomy rates and cumulative steroid exposure.