Abstract Background Infliximab (IFX) is the primary rescue therapy for steroid-refractory acute ulcerative colitis (UC), but up to 40% of patients do not respond. The Medical Rescue Therapy (MRT) score is one of the few tools available to predict early IFX response, incorporating admission UCEIS and day three post-rescue C-reactive protein (CRP).1 However, our local protocol employs an accelerated IFX dose at 48 hours, making a day two score which forecasts response to the next dose clinically valuable, as it may help expedite decisions on subsequent IFX, colectomy or JAK-inhibitor therapy.2 We aimed to (1) externally validate the day three MRT score; and (2) to develop and internally validate a novel, earlier predictive model using day two clinical and biochemical markers. Methods We conducted a single-centre retrospective study of 161 admissions for steroid-refractory acute UC (2012-2025) treated with IFX rescue at a tertiary IBD referral centre. The primary outcome was 14-day treatment failure, a composite of colectomy or switch to a JAK inhibitor. We externally validated the MRT score and developed a new predictive model using logistic regression with repeated 10-fold cross-validation. Results IFX failure at 14 days occurred in 25/161 admissions (15.5%). Static admission parameters did not predict outcomes, but dynamic, on-treatment changes in CRP and stool frequency emerged as key discriminators by day two post-IFX (p 0.001). The day three MRT score showed moderate discrimination (AUC 0.716) with a high NPV (91.2%) but low PPV (35.0%). We developed the Likelihood of Early Infliximab Failure (LEIF) score, a day two model incorporating absolute day two CRP, day two stool frequency, and the Day 2:Day 0 CRP ratio. The LEIF score demonstrated strong discriminative ability (cross-validated AUC of 0.851), effectively identifying true responders with high specificity (91.2%) and NPV (93.9%), while reliably flagging likely non-responders with a sensitivity of 68% and PPV of 58.6%. This allows for high-confidence risk stratification 24 hours earlier than existing tools. Conclusion The LEIF score leverages early, dynamic clinical and biochemical signals to accurately predict IFX failure at day two, aligning with the critical decision-making window in modern accelerated protocols. By integrating this score with point-of-care therapeutic drug monitoring, clinicians can create a mechanistically-informed framework to distinguish pharmacokinetic from pharmacodynamic failure, thereby personalising rescue therapy to either optimise dosing or expedite a class-switch. Further work is required to externally validate this tool. References: 1. Ramaswamy PK, Subhaharan D, Edwards J, White L, Patrick D, Shahzad A, et al. P223 Early prediction and objective criteria to define failure of rescue medical therapy in acute severe ulcerative colitis. J Crohns Colitis. 2024 Feb;18(Suppl 1):i557-9. doi: 10.1093/ecco-jcc/jjad212.0353. 2. Etchegaray A, Tambakis G, Kumar R, Croft A, Radford-Smith G, Walker GJ. Sequential rescue therapy with JAK inhibitors in corticosteroid and infliximab-refractory acute severe ulcerative colitis: a case series. Therapeutic Advances in Gastroenterology. 2025 Mar;18:17562848251323511. Conflict of interest: Dr. Goetz, Naeman: No conflict of interest Etchegaray, Amirah: No conflict of interest Tambakis, George: No conflict of interest Phillips, Jennifer: No conflict of interest Kumar, Rina: No conflict of interest Brown, Alison: No conflict of interest Radford-Smith, Graham: No conflict of interest Walker, Gareth: No conflict of interest Croft, Anthony: No conflict of interest
Goetz et al. (Thu,) studied this question.