Abstract
Background/Objectives: Acute severe ulcerative colitis (ASUC) is a high-risk condition associated with intense inflammation and a substantial risk of early colectomy. Intravenous steroids (IVS) remain the first-line therapy; however, steroid nonresponse is common. While the C-reactive protein-to-albumin ratio (CAR) has been proposed as a prognostic marker, data on its dynamic changes and late-phase performance remain limited. This study aimed to evaluate the serial performance of albumin and CAR at days 3 and 7 and to assess their association with early and delayed steroid response in a real-world cohort of biologic-naive ASUC patients. Methods: Biologic-naive patients hospitalized with ASUC between 2010 and 2023 were retrospectively analyzed. Disease severity and treatment response were defined using Truelove-Witts and Oxford criteria. Clinical and biochemical parameters, including albumin, CAR, and neutrophil-to-lymphocyte ratio (NLR), were assessed at baseline and on days 3 and 7. Associations with steroid response were evaluated using ROC analysis and multivariable logistic regression. Results: Ninety-eight patients were included. The IVS response rate was 11.2% on day 3 and 56.1% by day 7. Among nonresponders, 25% required infliximab rescue therapy, and colectomy occurred in 12.2%. Although day 3 CAR showed a trend toward discrimination, this did not reach statistical significance (p = 0.098) and should be considered exploratory. By day 7, responders had significantly higher albumin levels and lower CAR values (p < 0.05). Albumin (AUC = 0.702) and CAR (AUC = 0.713) demonstrated comparable performance. In multivariable analyses, day 7 albumin and CAR were significantly associated with steroid response, whereas NLR was not associated. Conclusions: Albumin and CAR are clinically relevant biomarkers associated with steroid response in ASUC. Early changes, particularly in CAR, may provide preliminary signals of nonresponse but require cautious interpretation. In contrast, day 7 measurements appear to reflect ongoing inflammatory dynamics and treatment evolution rather than true baseline predictors. Serial assessment of these biomarkers may support treatment monitoring and help optimize the timing of rescue therapy.