Deep Ulcers in Acute Severe Ulcerative Colitis at Index Endoscopy Predict Corticosteroid Resistance Open Access

Ulcerative colitis (UC) is a chronic inflammatory disease of unclear etiology that affects the colonic and rectal mucosa. It typically follows a relapsing and remitting course, with about 25% of patients developing acute severe ulcerative colitis (ASUC) during their lifetime, a disease exacerbation that requires hospitalization with significant morbidity [1, 2]. The classification system introduced by Truelove and Witts, which considers the frequency of bloody bowel movements and markers of systemic inflammation, remains widely used to identify patients with ASUC [2].

Initial management relies on high-dose intravenous (IV) corticosteroids, yet approximately 30% of patients exhibit steroid-refractory disease which necessitates escalation to second-line therapies and in more severe cases, colectomy [1, 3]. To determine the response to IV corticosteroids in ASUC, there are indices based on clinical, laboratory, and imaging parameters that have been shown to predict the need for colectomy [4‒6]. However, these scores are employed on day 3 of hospitalization, and the majority were developed prior to the widespread adoption of biologic agents.

Recent studies have identified elevated C-reactive protein, lower albumin levels, and greater luminal colonic damage as predictors of corticosteroid non-response at admission [7, 8]. These findings highlight the role of endoscopic evaluation in assessing rectal and colonic injury, as measured by the Mayo Endoscopic Score (MES) and the Ulcerative Colitis Endoscopic Index of Severity (UCEIS), which correlate with an increased need for rescue therapy and colectomy [9‒13]. Specifically, the presence of deep and extensive ulcerations on admission endoscopy in ASUC patients has been strongly linked to a higher risk of colectomy [11, 14]. This study aimed to evaluate the predictive value of endoscopic features observed during admission endoscopy for IV steroid response in ASUC patients treated at our institution.

This study included patients admitted to a tertiary academic hospital with a diagnosis of ASUC between January 2015 and July 2023. Patients were identified using the Truelove and Witts criteria (≥6 bloody bowel movements per day accompanied by at least one of the following: body temperature ≥37.8°C, heart rate ≥90 beats per minute, hemoglobin ≤10.5 g/dL, or an erythrocyte sedimentation rate of ≥30 mm/h/C-reactive protein ≥30/mg/L). Upon admission, patients received a 3- to 5-day course of IV hydrocortisone at a dose of 100 mg every 6 h.

Patients were excluded if they lacked endoscopic evaluation at admission or during hospitalization. Additionally, fecal samples were systematically analyzed for pathogenic bacteria, including Clostridioides difficile toxin-associated diarrhea. Patients with a bacteria-triggered disease flare were excluded from the study.

Medical records were reviewed to gather the relevant sociodemographic, clinical, and laboratory data at admission. Endoscopic data were retrospectively retrieved from procedures documented through both written reports and photographic evidence. In cases where the MES, UCEIS, or ulcer type (superficial vs. deep) description was missing, two experienced IBD endoscopists (M.J.M., F.D.C.) reached a consensus on the most appropriate classification based on the available reports and endoscopic images. Specifically, MES, UCEIS, and ulcer type were missing in 1 patient, while UCEIS and ulcer type were not available in 2 additional patients. Deep ulcers were defined as excavated defects in the mucosa, with a slightly raised edge as reported previously [10].

Non-response to IV steroids was defined as the need for second-line therapies, such as infliximab or ciclosporin, due to insufficient or unsustained clinical improvement. This determination was based on the persistence of fever, abdominal pain, and/or frequent bloody bowel movements, particularly in the presence of elevated inflammatory markers, while on IV corticosteroids. Clinical judgment determined the need for rescue therapies, and the Oxford criteria [4] were not rigidly followed.

IBM^® SPSS^® Statistics 26.0 was used for statistical analysis. Categorical variables were presented as frequencies, while continuous variables were expressed as means (standard deviation) or medians (interquartile range). Statistical significance was defined as two-tailed p < 0.05. Normality of data distribution was assessed to determine the appropriate statistical tests. Receiver operating characteristic curves were used to evaluate the predictive performance of endoscopic variables, with the area under the curve (AUC) serving as the metric for predictive power.

This study was approved by the Local Ethics Committee (Reference No. 114/2023) on November 3, 2023. Given its retrospective design, the requirement for informed consent was waived.

We included a total of 62 patients who were admitted with ASUC and were started on IV steroids (Table 1). Among them, 16 patients (25.8%) did not respond and required additional medical interventions, such as infliximab or ciclosporin. Specifically, 15 patients were treated with infliximab, 1 patient received cyclosporine, and 1 patient received both therapies. The median time to treatment escalation was 6 days. In this cohort, no patient required surgical intervention during hospitalization; however, one individual ultimately underwent total colectomy due to refractory disease 2 years after the initial admission.

Our data revealed no significant differences in corticosteroid response based on sex, age at UC diagnosis, or age at hospital admission (Table 1). No significant difference in treatment outcomes was found between patients with ASUC as disease presentation and those with known UC. Similarly, we observed no significant relationship between disease duration prior to admission and corticosteroid response. Moreover, in our population, the presence of extraintestinal manifestations or a familial history of IBD did not show an association with the outcome. Also, smoking status (current or past) was not associated with corticosteroid response.

In patients with previously known and staged UC, we collected data on disease extent, using the Montreal Classification system, and prior medical treatments (Table 1). We found that the proportion of IV corticosteroid nonresponders and responders were similar across the different disease extent classifications, including ulcerative proctitis (E1), left-sided UC (E2), and extensive UC (E3). Although our study primarily focused on the initial episode of ASUC, we identified 5 individuals who had undergone IV steroid treatment for previous admissions, but for whom complete records were unavailable. Importantly, there was no evidence of worse outcomes in this subgroup. Furthermore, patients receiving oral corticosteroids at admission (median duration of 12 days) showed no significant difference in response to IV corticosteroid therapy. Regarding the maintenance medical therapy, including oral mesalazine, rectal mesalazine, immunosuppressants, and biologics, our analysis did not reveal any significant associations with lack of response to IV steroids. Specifically, of the 7 patients in our cohort on azathioprine at admission, only 1 patient did not respond to IV steroids. Regarding biologic-experienced patients, 6 out of 8 responded to corticosteroids. Of the 2 patients who did not respond, one, previously treated with adalimumab, responded to infliximab rescue therapy, while the other, who had received vedolizumab and infliximab, responded to cyclosporine.

We collected various analytical parameters at admission for each patient, including hemoglobin, total leukocyte count, neutrophil count, platelet count, and erythrocyte sedimentation rate (Table 1). Our analysis did not reveal any significant differences in these parameters between nonresponders and responders to corticosteroid treatment. Similarly, the levels of albumin and C-reactive protein were comparable between the two groups. We also collected laboratory parameters on day 3 of IV steroid therapy, where CRP levels were significantly lower in responders compared to nonresponders (p = 0.004). In contrast, albumin levels at day 3, similarly to admission, did not differ between the two groups.

In our cohort, 88.7% of patients underwent CMV colitis testing, with no evidence of CMV disease. Among those with colonic ulcers, 31 out of 33 (93.9%) tested negative, while the two untested patients demonstrated an adequate clinical response to IV corticosteroids.

Most patients (88.5%) underwent rectosigmoidoscopy within the first 24 h of hospitalization. Our results showed a significant difference in the UCEIS between the two groups (p = 0.011; non-response: median 7, IQR 5–8; response: median 5, IQR 5–6). Similarly, MES was significantly associated with corticosteroid non-response (p = 0.001), with only 6.9% of patients with MES = 2 being nonresponders, compared to 42.4% of those with MES = 3. In our cohort, ulcer type (superficial vs. deep) was significantly associated with corticosteroid response (p = 0.002). Deep ulcers (Fig. 1) were more prevalent in nonresponders (66.7%) compared to responders (33.3%), whereas superficial ulcers were more frequently observed in responders (86.7%) than in nonresponders (13.3%).

The MES and UCEIS demonstrated significant discriminatory power in predicting IV corticosteroid non-response (Table 2 and Fig. 2), with the UCEIS showing a prediction capability (AUC = 0.717; p = 0.005, 95% CI: 0.566–0.867) comparable to MES (AUC = 0.731; p = 0.001, 95% CI: 0.597–0.865). A MES = 2 had a high negative predictive value (93.1%), emphasizing its utility in identifying patients who will respond to IV corticosteroids (Table 3). Notably, the ulcer type (superficial vs. deep) demonstrated the highest discriminatory power, with an AUC of 0.771 (p = 0.001, 95% CI: 0.604–0.937), and the presence of deep ulcers had the highest positive predictive value (66.7%) for IV corticosteroid non-response. These findings underscore the utility of endoscopic indices and ulcer characteristics in identifying ASUC patients at admission with a higher risk of corticosteroid non-response.

The current study aimed to identify endoscopic features at hospital admission that could predict the response to IV corticosteroids in patients with ASUC. Our findings demonstrate that both the UCEIS and MES are valuable tools in predicting corticosteroid response, with the presence of deep ulcers emerging as a pivotal predictive feature.

Our results indicate that 25.8% of patients admitted for ASUC did not respond to IV steroids. To minimize the influence of infectious triggers, patients with documented fecal pathogenic bacteria were excluded. Additionally, CMV colitis, which contributes to steroid refractoriness and poor outcomes in ASUC, often presenting with deep colonic ulceration, was ruled out in this population [15]. These data suggest a low likelihood of infectious causes contributing to steroid refractoriness in our cohort.

There is evidence suggesting that prior oral corticosteroid exposure, particularly for more than 1 week, may reduce responsiveness to IV steroids in patients admitted for ASUC [16]. We did not observe this trend in our cohort, as patients with prior exposure to corticosteroids (oral or IV) exhibited a response to IV steroids comparable to corticosteroid-naïve patients, which also aligns with the findings of other studies [8, 17]. Similarly, previous exposure to immunosuppressants or biologics also did not affect treatment outcomes in our cohort. It has been suggested a reduced IV corticosteroid response in patients exposed to azathioprine [17], but we did not verify this trend. Nonetheless, we acknowledge that the limited number of patients in these treatment subgroups may have reduced the statistical power to detect significant differences in corticosteroid response.

Multiple indices using clinical, radiologic, and/or biologic parameters have been developed to predict outcomes in ASUC, but with limitations, particularly that the most widely employed are calculated 48–72 h hours after hospitalization [4‒6, 18]. More recently, the Albumin, CRP and Endoscopy (ACE) Index [8] was introduced to predict outcomes at admission, incorporating an albumin level ≤30 g/L, a CRP ≥50 mg/L, and a MES of 3, with each parameter scoring 1 point. Another recent study was developed and externally validated, the Admission Model for Intensification of Therapy in Acute Severe Colitis (ADMIT-ASC) [7], which includes the sum of CRP ≥100 mg/L (1 point), albumin ≤25 g/L (1 point), UCEIS ≥7 (2 points) or ≥4 (1 point). In both scores, there was an association of increased severity and corticosteroid non-response.

In our population, we observed that increased endoscopic severity was associated with failure of IV steroids, consistent with the findings of the ACE Index and ADMIT-ASC cohorts. However, we did not observe significant differences in albumin or CRP levels between IV steroid responders and nonresponders. Additionally, we investigated whether CRP levels exceeding 50 mg/dL were associated with a reduced response to corticosteroids as previously reported [8], but our findings did not support this hypothesis. One possible explanation for this discrepancy is that the ACE Index cohort included both ASUC and non-ASUC patients, with 17.5% of individuals in the analysis having a MES of 1 or less. Additionally, while the ACE Index showed promise in its generation cohort, it performed less effectively in the validation set [19]. Similarly, although almost all patients scoring 3 or 4 points on the ADMIT-ASC required rescue treatment strategies, most patients (78.8%) scored 1 or 2 points, with IV response rates of 70% and 33%, respectively, making the ADMIT-ASC score less useful in this low severity context.

Endoscopic evaluation was the most useful parameter in predicting IV steroid response in our cohort, as increasing severity of luminal damage assessed by UCEIS or MES was associated with treatment failure. Consistent with this, previous studies have also demonstrated that both the UCEIS and MES can predict the need for colectomy in patients with ASUC, with UCEIS performing slightly better in this regard [9, 20]. In our work, we further investigated specific endoscopic features associated with steroid resistance. While both MES and UCEIS incorporate the presence of colonic ulcers as indicators of severity, UCEIS provides an additional layer of stratification by distinguishing between superficial and deep ulcers. Our analysis focused on this distinction in predicting treatment outcomes, as bleeding scores in UCEIS may be more subjective. Moreover, the presence of deep ulcers in ASUC has previously been linked to refractory disease [3, 11, 21, 22]. Our findings indicate that the presence of deep ulcers was the most significant predictor of non-response to corticosteroids, with non-response rates of 66.7% and an AUC of 0.771, which was superior to both UCEIS and MES in predicting treatment failure.

Our findings raise the question of whether early initiation of rescue therapies should be considered for patients with deep ulcers, as they are unlikely to respond to IV steroids, rather than adhering to the standard stepwise approach [23]. To our knowledge, no studies have directly compared high-dose IV corticosteroids vs. second-line therapies (infliximab or cyclosporine), with or without IV steroids, at day 0 for ASUC [24]. However, identifying poor prognostic factors at admission, such as deep ulcers, allows for risk stratification and supports the development of prospective studies to address this relevant clinical question [24, 25].

This study has limitations, as it was conducted at a single center with a retrospective design. The decision to initiate advanced therapies was based on clinical judgment, and the Oxford criteria on day 3 were not strictly applied, which may limit the generalizability of the findings. In 3 patients, endoscopic scores were assessed retrospectively using available images. Additionally, endoscopic evaluation was limited to the left colon to minimize the risk of procedure-related complications. As a result, disease severity may have been underestimated in some cases, and prior topical therapy may have influenced endoscopic evaluation.

In conclusion, endoscopic assessment of patients with ASUC should be performed at admission as it allows for risk stratification. Individuals with deep colonic ulceration should be closely monitored for complications, and the decision threshold for initiating rescue therapies should be lowered, as IV steroids are less effective in these cases.

This study was approved by the Local Ethics Committee (Reference No. 114/2023) on November 3, 2023. Given its retrospective design, the requirement for informed consent was waived.

José Cotter is a member of the Editorial Board of GE – Portuguese Journal of Gastroenterology. The other authors have no other conflicts of interest to declare.

The current study did not receive any funding from external sources. The data utilized in this study were generated as part of routine clinical practice and collected from patient records.

João Carlos Gonçalves: conception and design of the study, data collection, statistical analysis, data interpretation, and manuscript drafting. Vítor Macedo Silva, Cláudia Macedo, Cátia Arieira, Tiago Cúrdia Gonçalves, Francisca Dias de Castro, and Maria João Moreira: data collection, data interpretation, and critically reviewed and approved the manuscript for submission. Joana Magalhães: study supervision, data interpretation, and manuscript drafting. José Cotter: conception and design of the study, study supervision, data interpretation, and manuscript drafting.

Requests for sharing of de-identified data by third parties will, after written request to the corresponding author, be considered. If the request is approved and a data access agreement is signed, only de-identified data will be shared.