Introduction
Ulcerative colitis (UC) is a mucosal inflammatory disease of the colon with heterogeneous clinical courses. Patients may have chronic active disease, a relapsing–remitting course, early high disease activity followed by remission, or progressive worsening after an initially mild course. Some patients may also present with acute severe ulcerative colitis (ASUC),1 which remains a clinically important and not uncommon presentation of UC, characterized by bloody diarrhea accompanied by systemic toxicity and increased inflammatory markers up to alarming levels.
The ASUC form of UC may occur in up to one-fourth of patients, often requiring potent anti-inflammatory treatment, and up to 20% of patients may require colectomy during the first admission.2,3 Moreover, the likelihood of colectomy increases with repeated ASUC hospitalizations.4
Intravenous (IV) corticosteroids (IV-CS) are the first-line medical therapy for ASUC, with a clinical response achieved in approximately 65% of patients. Early identification of steroid nonresponse is essential; according to the Oxford criteria, lack of response by days 3-5, a stool frequency of > 8/d, and a C-reactive protein (CRP) level of > 45 mg/L are associated with an 85% risk of colectomy.5 The introduction of biologic rescue therapy, particularly IV infliximab (IV-IFX), represented a major advance in ASUC management.6 Current guidelines recommend rescue therapy with IV cyclosporine (CYC) or IV-IFX in patients who are unresponsive to IV-CS by day 3.7 Nevertheless, despite optimized rescue strategies, a substantial proportion of patients (approximately 20%-30%) still progress to colectomy following IFX, highlighting an ongoing unmet need for fast-acting and effective alternatives.8
Upadacitinib (UPA), an oral, selective Janus kinase-1 (JAK1) inhibitor with a rapid onset of action, is approved for the treatment of moderate-to-severe UC and Crohn disease (CD). Recent systematic reviews and network meta-analyses suggest that UPA is among the most effective agents for inducing clinical remission in moderate-to-severe UC, including in patients with prior biologic exposure.9,10 Given its high efficacy and rapid clinical activity, UPA may be a potential rescue therapy for IV-CS and IFX refractory and/or intolerant patients with ASUC. However, the current evidence base in ASUC remains limited, largely consisting of case reports and small series with short follow-up.11
Therefore, we aimed to evaluate the short- and long-term (week 48) outcomes, colectomy rates, drug persistence, and safety profile of UPA as rescue therapy in CS- and IFX-refractory patients with ASUC.
Materials and methods
Study design
In this retrospective multicenter cohort study, we evaluated patients from 11 centers across the country who presented with an ASUC flare and received UPA after IV-CS and/or IFX from May 2024, the date of public reimbursement coverage, to August 2025.
Study population and eligibility criteria
The study population included patients with a diagnosis of UC for at least 3 months who were hospitalized for ASUC according to the Truelove and Witts criteria.12 Before the index flare, patients had either been in clinical remission or had mild-to-moderate clinical activity. Eligible patients had received at least 3 days of IV-CS therapy, equivalent to 1 mg/kg methylprednisolone, and had no or only a partial response. These patients were subsequently treated with IV-IFX at a dose of 5-10 mg/kg using standard induction at weeks 0 and 2, but the patients continued to have no or only partial response while still meeting the Truelove and Witts criteria for ASUC.
Patients unable to receive IV-CS (due to prior documented steroid unresponsiveness and uncontrolled diabetes mellitus or steroid-related osteonecrosis) were included if they had received IV-IFX for ASUC; likewise, patients with IFX allergy or those who developed ASUC while on IFX were included if they received IV-CS therapy for ASUC.
Upadacitinib treatment protocol
Upon UPA induction, azathioprine and IFX were stopped in all patients. Meanwhile, CS treatment was tapered if still in use. All patients received UPA treatment at 45 mg/d for 8 weeks.
Patients who continued treatment subsequently received maintenance therapy of 15-45 mg/d at the treating physician’s discretion.
Data collection and follow-up
Demographic characteristics, prior clinical- and treatment-related features, and clinical and laboratory findings at UPA initiation and at weeks 1, 4, 8, 12, 24, and 48 were recorded. Each patient who started UPA was evaluated using an intention-to-treat (ITT) approach for UPA persistence or progression to colectomy. Based on UPA start dates, patients were considered eligible for ITT analyses at each timepoint only if they had reached the corresponding week of follow-up, irrespective of treatment continuation.
Because follow-up colonoscopies and fecal calprotectin measurements were performed at irregular, center-specific intervals and driven by individual clinical needs, these data were not included in the analysis as endpoints. Endoscopic remission rates considering last available endoscopic assessment under UPA treatment are presented in Table 1.
Clinical activity and outcome definitions
Clinical activity under UPA therapy was evaluated using the partial Mayo (pMayo) score because it captures and quantifies core clinical subparameters, including stool frequency and rectal bleeding. Clinical remission was defined as a pMayo score ≤ 2 with all subscores < 2, and clinical response as a decrease in pMayo score of ≥ 3 points or a reduction of ≥ 30%. Adverse events related to UPA were graded according to the Common Terminology Criteria for Adverse Events (CTCAE) on a severity scale from grade 1 to 5.13
Literature search
To contextualize our findings, we performed a targeted, nonsystematic PubMed search for studies evaluating rescue treatment strategies in ASUC. Search terms included combinations of “acute severe ulcerative colitis,” “ASUC,” “rescue therapy,” “infliximab,” “cyclosporine,” “tofacitinib,” “upadacitinib,” “JAK inhibitor,” and “colectomy.” Original studies, systematic reviews, meta-analyses, and relevant guideline documents published in English were selected according to their relevance to rescue therapy and colectomy outcomes in ASUC.
Statistical analysis
Categorical variables were presented as frequency (%), and continuous variables as mean ± SD or median (IQR) according to distribution. Follow-up comparisons were performed using the chi-square test or Fisher’s exact test for categorical variables, and the paired t-test or Wilcoxon signed-rank test for continuous variables. Time-to-event analyses were conducted using the Kaplan-Meier method and univariate Cox analysis.
Receiver operating characteristic (ROC) curve analysis was performed to determine the area under the curve (AUC) value and optimal baseline CRP cutoff associated with colectomy risk, with the optimal threshold identified using the Youden index. A 2-sided P value < .05 was considered statistically significant. All statistical analyses were performed using IBM SPSS Statistics, Version 31.0 (IBM Corp., Armonk, NY, United States).
Ethics approval was obtained from the local ethics committee, and the study was conducted in accordance with the Declaration of Helsinki.
Results
Patient demographics and baseline treatment characteristics
The study cohort consisted of 50 patients. Of these, 54% (n = 27) were male, and the median age at UC diagnosis was 27 years (IQR, 18; range, 12-58 years). Table 1 summarizes the demographic and treatment characteristics.
The majority of patients had a history of IFX exposure, and more than half had been treated with ≥2 biologic agents; additional treatment details are presented in Table 1. The majority (76%; n = 38) of patients had UC >1 year when they presented with ASUC, whereas only 12% (n = 6) had a disease course <6 months.
Upadacitinib treatment persistence
Considering the UPA initiation dates, 86% of patients (n = 43) were eligible for ITT analysis at week 48, and 92% (n = 46) at week 24. The overall UPA continuation rate in the entire cohort was 52% (Table 1), with week-specific continuation rates of 90% at week 4, 70% at weeks 8 and 16, 61% at week 24, and 51% at week 48. Discontinuation of UPA was due to primary nonresponse in 42% (n = 10), secondary loss of response in 46% (n = 11), and other causes—including pregnancy, reimbursement issues, or adverse events—in 13% (n = 3).
Safety and adverse events
Adverse events related to UPA were observed in 24% of patients (n = 12); in all but 1patient these events were nonsevere. The only severe adverse event occurred in 1 patient who developed marked leukopenia and herpes zoster. The most frequently observed adverse events were upper respiratory tract and urinary tract infections.
Baseline disease activity and clinical response
At UPA initiation, all patients had a Mayo endoscopic subscore of 3, and the median pMayo scores was 8 (IQR, 2) (Table 1). Treatment with UPA resulted in a significant reduction in stool frequency, rectal bleeding, pMayo scores, and CRP levels as early as week 1, and these improvements were sustained through weeks 24-48 in patients who continued therapy (Figure 1, Table 2).
Colectomy rate and week-specific outcomes
The overall colectomy rate during follow-up was 22% (n = 11) in the entire cohort, with nearly half of the colectomies (46%, n = 5) performed by the end of first week. The flow of patients and their treatment responses by treatment week is summarized in Figures 2-3. At week 1, the partial clinical response rate was 60% (n = 30), and the clinical remission rate was 30% (n = 15), yielding an overall week-1 UPA-associated clinical efficacy of 90% (n = 45).
The clinical remission rate increased to 46% (n = 23) by week 8 and remained at 42% at week 48 (n = 18) (Figure 3). In this retrospective real-world cohort, and using an intention-to-treat approach, 23% of patients who reached week 48 experienced loss of response, whereas 51% maintained clinical response.
Predictors of colectomy and upadacitinib treatment persistence
In time-dependent analyses, disease duration at the time of UPA initiation (p: 0.947), history of multiple biologic therapies (p: 0.161), use of IFX rescue therapy (p: 0.320), IV-CS induction (p: 0.526), and response to IV-CS induction (p: 0.333) were not significantly associated with the need for surgery, whereas higher baseline CRP and lower baseline albumin at UPA initiation was significantly associated with an increased risk of colectomy (p: 0.006, p: 0.026 respectively). A cutoff initial CRP of 32 mg/L predicted colectomy (sensitivity, 82%; specificity, 65%) in ROC analysis (AUC, 0.732). With respect to UPA treatment persistence, the presence of UPA-related adverse events (p: 0.813), history of multiple biologics (p: 0.708), and prior IFX rescue therapy (p: 0.515) had no significant impact on continuation rates.
Discussion
In this retrospective multicenter cohort of 50 patients with ASUC who were resistant or intolerant to previous IV-CS and/or anti–tumor necrosis factor (anti-TNF) rescue therapies, UPA was associated with clinical remission and avoidance of colectomy as a third-line rescue treatment. These outcomes did not appear to be substantially influenced by prior treatment history or disease duration. To our knowledge, this patient cohort is the largest to date for which the efficacy of UPA was evaluated as a third-line rescue therapy with a 1-year follow-up in ASUC.
Given the substantial risk of colectomy in ASUC, treatment with IFX or CYC is recommended as rescue therapy after IV-CS failure. For both IFX and CYC, colectomy rates can reach approximately 35%-45% at 1-year follow-up.14 Among anti-TNF agents, IV IFX is the only rescue agent in IV-CS–refractory ASUC. Although low albumin, low hemoglobin, and obesity have been shown as negative predictors of anti-TNF effectiveness in ASUC, the dominant limitation may be rapid IFX loss through extensively inflamed colonic mucosa, leading to inadequate drug exposure and failure to achieve effective peak levels for remission.15 Epidemiological studies, including 1 from our country, have shown that colectomy rates have decreased or that colectomy has been delayed in the biologic era.16,17 However, these epidemiological data do not clearly indicate whether colectomy rates among patients with ASUC also decreased during the same period.
Furthermore, in recently published cohorts, the long-term success of IV anti-TNF agents, even following the high-dose intense induction treatment, is not very promising and frequently ends with colectomy during the follow-up period.18 Adalimumab, certolizumab, and golimumab are subcutaneous anti-TNF agents that are generally less effective options after IFX exposure in UC anddo not have an established role in the rescue treatment of ASUC.
Despite these established rescue options, many patients with ASUC still require colectomy, underscoring the need for fast, effective therapies. Accordingly, rapid-onset small molecules with predictable pharmacokinetics are gaining interest as alternatives to conventional rescue strategies. The potential role of JAK inhibitors as second- or third-line therapy in patients with ASUC has previously been reported in the literature, notably with tofacitinib.19–21 In our cohort, within a week following the first dose of UPA introduction, clinical remission or response was achieved in 90% of patients and nearly half of the patients were in clinical remission at week 48.
This success rate is even comparable to the clinical remission rates reported with first-line IFX in moderate-to-severe UC both in phase III trials and in real-world cohorts, which may be partly related to a prior chronic active disease background in some patients and to treatment class switching.22,23 In the recent meta-analysis by Jena et al., the 1-month colectomy rates were reported as 11.5% for tofacitinib and 11.2% for UPA, which are broadly comparable to the week 1 and week 4 colectomy rates observed in our study, 10% and 16%, respectively.21 In the same meta-analysis, the 3- to 12-month colectomy rates for JAK inhibitors were reported as 22.6% for tofacitinib and 23.4% for UPA, which almost exactly correspond to the 48-week colectomy rate of 22% observed in our study.21 Overall, half of the patients were still on UPA at the end of follow-up, a rate comparable to real-world data of biologic treatments. Furthermore, 4 out of 5 patients avoided colectomy by week 48. This outcome compares favorably with the reported 6-month colectomy rate of 37% after third-line treatment in patients with ASUC who had failed IFX or CYC.24 Since all of our patients were IV-CS and IV–anti-TNF nonresponsive or partially responsive (frequently initially), and more than half had experienced dual biologic treatment, this result may deserve further attention.
While our analysis captured patient responses at a minimum of 1 week, some patient responses even appeared within 3 days. Although as a limitation our study does not provide information on very short-term (<1 week) clinical and laboratory responses in patients with ASUC receiving UPA rescue therapy, the week 1 outcomes reported in our study may be expected to largely reflect these very short-term responses. The short half-life and rapid action of UPA highlight it as a favorable alternative to available rescue agents in ASUC.
Because these patients were at high risk of colectomy, further delays in effective treatment could have compromised perioperative safety. The rapid clinical response observed in this cohort therefore supports the potential role of UPA as a rescue option in steroid- and IFX-nonresponsive or partially responsive patients. On the other hand, half of the colectomies in our cohort occurred within the first week and in their review, Daminos et al. reported 56% of colectomies being at the index hospitalization.25 Both results further support the finding that the effects of UPA, including lack of efficacy in nonresponders, tend to manifest early.
Regardless of prior treatment history and age of disease onset, the strongest factor associated with colectomy was a higher baseline CRP level. This finding suggests that, beyond other variables, the intensity of inflammation is a key determinant of treatment response in ASUC. Therefore, in patients presenting with ASUC who are considered for third-line UPA, a baseline CRP > 32 mg/L should be viewed as a warning signal.
By the end of the first month, risk of colectomy dropped by nearly 75% in our cohort. This finding may reflect the rapid suppression of proinflammatory cytokine production by JAK inhibitors, particularly UPA. Unlike anti-TNF agents, UPA may also be less affected by fecal drug loss in ASUC.15 Notably, given the near 90% rate of prior IFX exposure in our cohort and the fact that more than half of patients had been exposed to 2 or more biologics, prior treatment history did not appear to determine the clinical success of UPA in ASUC, nor did it influence UPA continuation rates.
Loss of response still occurred in approximately one-fourth of patients by 1 year. Gradual flares mainly developed in patients who had initially responded without achieving clear clinical remission, and most post–week- 1 colectomies also occurred in this subgroup. Dose interruption or de-escalation was sometimes followed by mild-to-moderate flares, particularly in patients without endoscopic or biochemical remission, although these observations should be interpreted cautiously.
It remains to be determined whether the short half-life of UPA underlies both rapid clinical improvement and flares after missed doses.
Most patients in our cohort had established and treatment-experienced UC rather than newly diagnosed disease. Only approximately 10% of patients presented with their first ASUC episode within 6 months of UC diagnosis, whereas more than 75% developed ASUC ≥1 year after diagnosis. In addition, the median disease duration exceeded 4 years, and more than half of the patients were dual-biologic experienced at the time of the index ASUC flare.
Therefore, this cohort largely represents a difficult-to-treat real-world ASUC population. Not all patients received IV-CS or IV-IFX during the index flare because of prior contraindications, allergy, or documented nonresponse. Nevertheless, these results reflect real-world clinical practice and support the relevance of an intention-to-treat analysis in this setting.
Rescue options for ASUC remain limited and are often confined to the narrow window between IV-CS and IFX, while alternatives such as ustekinumab are only used sporadically, and agents with slower onset like vedolizumab and newer IL-23 receptor blockers have limited evidence in the acute setting. Along with responders, all patients, including those in whom treatment with rescue UPA failured, were prospectively followed to their endpoints, which are frequently obscure or unmentioned in the literature. Unintended overlap between UPA and preceding IV anti-TNF therapy, potentially at residual or therapeutic serum levels, may have contributed to the initial response.
However, this finding is unlikely to explain the sustained long-term outcomes. Because success was mainly established on clinical remission with long-term results, this point may have a negligible effect on the long-term treatment successand is an unavoidable fact in any rescue study design of ASUC.
Patients with ASUC whose UPA rescue treatment failed (including loss of response) either went to colectomy or transitioned to a chronic active stage and started on alternative treatment modalities including some available new-era biologic/small molecules phase studies. Whether the higher colectomy rates reported after anti-TNF rescue in the literature—compared with the ∼20% rate observed in our cohort—reflect differences in prior treatment exposure or instead relate to the distinct mechanism of action, favoring JAK inhibitors, may deserve consideration.
One important aspect of UPA therapy is its potential adverse event profile. Safety signals in our cohort were broadly consistent with real-world expectations for JAK inhibition in a high-risk hospitalized population. Patients with ASUC are particularly prone to thrombotic complications due to the high inflammatory burden and recent exposure to CS, and venous thromboembolism has been reported in association with JAK inhibitors.
In line with European Crohn’s and Colitis Organization (ECCO) recommendations, our patients with ASUC received prophylactic low–molecular-weight heparin (data not shown), and no venous thromboembolism was observed in our cohort.26 Although approximately one-fourth of patients experienced adverse events, predominantly nonsevere, with infections comprising the most common events, only 1 patient had a severe event. In addition, UPA-related adverse events did not adversely affect treatment persistence. Collectively, these findings support a favorable safety profile of UPA in patients with ASUC.
Finally, in clinically appropriate patients, rapid disease control with UPA may help avoid or delay colectomy and create a therapeutic window for alternative treatment strategies. However, this potential benefit should be balanced against the need for early recognition of nonresponse, as surgery should not be inappropriately delayed in patients who are unlikely to benefit from continued medical therapy.
This study has limitations inherent to its retrospective, multicenter design. Treatment decisions, including colectomy timing, concomitant therapies, and dose adjustments, were not standardized. Laboratory parameters such as CRP and albumin may also have influenced colectomy decisions, potentially confounding their predictive value.
Endoscopic outcomes and fecal calprotectin were not uniformly captured, and mild adverse events may have been under-reported. In the limited number of patients who underwent colectomy, the exact interval between the last UPA dose and surgery was not systematically recorded; therefore, the effect of recent UPA exposure on postoperative outcomes could not be formally assessed, particularly considering the very short plasma half-life of UPA. Nevertheless, the multicenter design, relatively large number of patients with ASUC, inclusion of highly treatment-experienced patients, and availability of 48-week outcomes strengthen the clinical relevance of the findings and support the rationale for prospective controlled studies.
In patients who underwent colectomy, UPA was generally discontinued once surgery was planned or at the time of colectomy; however, the exact interval between the last UPA dose and surgery was not systematically captured. Therefore, its impact on postoperative outcomes could not be formally assessed. No clear center-reported signal suggested an adverse effect of recent UPA exposure, but this finding should be interpreted cautiously given the retrospective design and limited number of colectomy events.
In conclusion, in this multicenter cohort of highly treatment-experienced patients with ASUC in whom IV-CS and IFX rescue therapies had failed, third-line UPA was associated with a low colectomy rate and sustained response for up to 1 year with an acceptable safety profile. Although UPA remains a novel option in this setting, it appears to be an effective treatment option for patients with ASUC, regardless of prior treatment history or disease duration. This treatment may provide sufficient disease control to avoid or postpone colectomy, while giving patients additional time to move on to other therapies or participate in clinical trials before surgery becomes inevitable.