Infliximab induction regimens in steroid‐refractory acute severe colitis: a multicentre retrospective cohort study with propensity score analysis

Accelerated induction regimens of infliximab have been proposed to improve response rates in patients with steroid‐refractory acute severe colitis.


| INTRODUC TI ON
Acute severe ulcerative colitis (ASUC) is a medical emergency with up to 30% of patients requiring colectomy during their index admission 1,2 and is associated with a mortality of up to 2.9% in peripheral centres and about 1% in specialist IBD units. 3 ASUC is traditionally defined by the Truelove and Witt's criteria, 4 which combine frequency of bloody stools (≥6 per day) with at least one marker of systemic toxicity: pulse rate > 90 bpm, temperature > 37.8°C, haemoglobin < 105 g/L and/or an ESR > 30 mm/h. ASUC requiring hospitalisation occurs in 10%-25% at diagnosis and in 20%-30% during the disease course of ulcerative colitis. 5,6 Intravenous corticosteroids remain the cornerstone of first-line therapy for ASUC. A meta-analysis of cohort studies and randomised trials, published in 2007, examined the response to corticosteroids in ASUC. The authors reported a pooled response rate to intravenous steroids of 67%, indicating that up to 40% of patients fail to respond. 7 Over the last decade, in patients failing corticosteroids, rescue therapies, including ciclosporin and infliximab, [1][2][3] have been used as an option to avoid colectomy. While there is no difference in response rates between infliximab and ciclosporin, [8][9][10] a majority of clinicians now appear to favour infliximab mainly citing convenience and safety. 11 Despite the use of rescue therapies, a significant proportion of patients still undergo colectomy. 12 The data on rescue therapies indicate that the rates of nonresponse to infliximab rescue vary from 40%-55%. [8][9][10] Reasons for nonresponse may include patient and disease factors or treatment factors such as timing of rescue and dosing schedules. A key and unanswered question remains the optimal dosing strategy of infliximab in steroid-refractory ASUC. Current regimens have extrapolated dosing schedules for management of moderate-to-severe disease in an out-patient clinic setting to the hospitalised in-patient, and use a standard induction regimen of 5 mg/kg intravenously at week 0, 2 and 6. However, there are multiple reasons why ASUC may be associated with increased clearance of infliximab. These include hypoalbuminaemia, leakage of infliximab itself into the stool, activation of the reticuloendothelial system and higher circulating TNF levels. [13][14][15] This enhanced clearance of infliximab may also be associated with worse clinical outcomes.
This has led to the concept of 'accelerated induction rescue therapy,' where higher dosages or increased frequency of induction dosing have been proposed. 15 There are no published randomised controlled trials on the efficacy and safety of accelerated induction.
Although there is increasing use of accelerated induction in clinical practice, the data from the published small cohort studies are conflicting. [16][17][18] Our recent meta-analysis of the available cohort studies showed no conclusive evidence for benefit of accelerated induction in reducing colectomy rates in steroid-refractory disease. 19 However, the majority of existing studies are single-centre cohorts with significant limitations including small sample sizes. Furthermore, such studies did not take into account provider bias, which could be an important determinant in selection of the type of rescue therapy. 20 We have now performed a multicentre retrospective cohort study in 11 centres in the United Kingdom to compare the outcomes of using accelerated induction to standard induction regimens for ASUC in the real-world setting. We have used propensity score matching method to reduce the impact of provider bias in treatment selection.

| MATERIAL S AND ME THODS
This was a multicentre retrospective cohort study. We included patients with ASUC meeting modified Truelove and Witt's criteria admitted between May 2016 and May 2018 for intravenous corticosteroids in 11 acute hospitals in the UK (Six University Teaching Hospitals and Five Peripheral Secondary Care Hospitals).

| Inclusion and exclusion criteria
We included consecutive hospitalised patients needing intravenous steroids who received at least three doses of intravenous steroids.
Physicians completing the case reports assessed them as meeting modified Truelove and Witt's criteria. We excluded patients with: a diagnosis of IBD unclassified, Crohn's colitis, infective colitis; coexistent cytomegalovirus; admission for elective surgery; and prior therapy with anti-TNF.

| Study design
Patients who received infliximab following failure of intravenous rescue therapy were stratified into two groups. The standard induction rescue group comprised of patients who received a dose of infliximab 5 mg/kg at week 0 and no further doses until 2 weeks after first dose. The accelerated induction group included patients who received at least two doses of 5 mg/kg with a second dose received on or before 7 days after the first dose and/or those who received 10 mg/kg for their first dose with a further dose within 2 weeks. We recorded available data on clinical and laboratory data at baseline, at commencement of rescue therapy, 30 days, 90 days, 6 months and 12 months.
Our primary outcome measure was colectomy rate at 30 days.
Secondary outcome measures were index admission colectomy rates, colectomy rates at 90 days, 6 months and 12 months, the length of hospital admission and adverse events including post-operative complications and mortality.

| Statistical analysis
Continuous variables were summarised using mean and standard deviation and compared using Student's t test or the Mann-Whitney U test. Categorical variables were expressed as proportions and analysed by Fisher's exact test or chi-squared test as appropriate.
Propensity score-adjusted matching was used to minimise the possibility of provider bias in the choice of rescue treatment. Baseline clinical and demographic variables were matched in a 1:1 fashion to create a matched cohort with baseline variables which are independent of the initial infliximab dose. We ascribed a priori determined factors considered to affect the choice of rescue therapy including C-reactive Protein (CRP), serum albumin, CRP/albumin ratio, haemoglobin and the presence of pancolitis in the propensity score matching.
Logistic regression was used to generate bivariate propensity scores using these variables. We used the greedy matching algorithm with the nearest calliper matching neighbour (random order) within a 0.01 propensity score selected for the best match in the matched cohort. We confirmed balanced covariates distribution after matching.
Kaplan-Meier survival curves were plotted for the primary outcome of 30-day colectomy rates in both unmatched and matched cohorts between those receiving standard induction compared to accelerated induction and the rates compared by log-rank statistic.
All tests were two-sided and a P value of <.05 was considered significant. We used spss statistics version 25 (IBM Corp.) for analysis.
As this was a retrospective data collection, in accordance with the UK Health Research Authority guidance, no central ethical committee submission was made. Individual institutions sought permissions to conduct a local service evaluation as appropriate.

| Study cohort
We included data on 131 patients from 11 centres across the UK receiving rescue therapy for steroid-refractory ASUC, of which 102 received standard induction regime and 29 received accelerated induction regimen. The baseline characteristics are recorded in Table 1. There were differences in blood parameters between the patients receiving standard induction and accelerated induction rescue ( Table 2). Patients receiving accelerated regimen were more likely to have higher CRP levels, higher CRP/albumin ratio and lower albumin levels at day 1 and day 3 and there were no differences between the two groups in terms of haemoglobin on day 1 or day 3.

| Colectomy rates: entire cohort
The overall colectomy rate among the 131 patients who received rescue therapy was 29%. Table 3 reports the colectomy rates at 30 days, 90 days, 6 months and 12 months in patients receiving rescue therapy. There was no significant difference in overall colectomy rates between in-patients receiving standard induction vs accelerated induction group (P = .996) ( Table 3 and Figure 1).

| Colectomy rates: propensity scorematched cohort
Using propensity score matching, we included 52 matched patients receiving rescue therapy for comparison. The baseline characteristics and blood markers in the cohort are detailed in Table 4.

| Duration of hospital stay and complications
The mean duration of hospital stay in patients treated with standard induction was 4.4 days (SD 1.6) less than patients given accelerated induction rescue therapy (P < .01) in the unmatched cohort. In the propensity score-matched cohort, there was no significant difference in the length of stay between standard induction and accelerated induction groups (23.6 ± 4.3 vs 19.2 ± 7.1 days, P = .09). There was no difference in complication rates between the two groups (18.6% vs 20.7%, P = .8) but there was one death in the accelerated induction group ( Table 5).

| D ISCUSS I ON
Despite the increasing use of infliximab rescue therapy in patients failing intravenous steroids, a significant proportion of ASUC patients do not respond adequately to standard induction dosing.
Pharmacokinetic data have led to increasing use of intensified or accelerated dosing schedules in rescue therapy for ASUC patients.
Our large multicentre retrospective study showed no difference in colectomy rates in the overall cohort of patients receiving standard vs accelerated dosing schedules but when provider bias was accounted for in the propensity-matched cohort, we found a reduction in short-term colectomy rates in patients receiving accelerated induction.
The first study to report the potential benefit of more frequent infliximab infusions in ASUC patients was from Gibson et al in Ireland, 16 who in their cohort of 50 hospitalised patients with ASUC showed a reduction in short-term colectomy rates in the 15 patients who received three doses of 5 mg/kg within 24 days when compared to those receiving standard induction regimen (6.7% vs 40%, P = 0.039). This study also suggested shortened time to colectomy in those receiving standard regime although the long-term colectomy rates were similar. Notably, 38% of these patients had lower endoscopic disease severity (Mayo 2), and the authors did not correct for provider bias in the choice of regimen.
Furthermore, the definition of accelerated dosing in this study did not include the need for a further dose 7 days after the first dose or increased front-loading dose. Subsequent studies examining the use of increased frequency of infliximab at 5 mg/kg 17,18,21,22 and a recent meta-analysis 19 have not confirmed the benefit as re- A number of patient-and disease-related variables have been suggested as high risk for needing colectomy in patients with ASUC. 24,25 These indices were developed in the pre-infliximab rescue therapy but also those who may potentially benefit from different dosing strategies. This lack of consensus inevitably leads to variations in the management and dosing regimens 27 as seen in our study. The blood parameters at first and second doses of rescue therapy indicate lack of improvement or indeed worsening which along with clinical symptom may prompt a second dose as accelerated induction (Data S1).
One of the strengths of our study is the attempt to compare the outcomes between the different dosing regimens after accounting for the potential bias of baseline clinical and demographic variables and the potential impact of these in clinicians' choice using a propensity scorematched method. Our model incorporated established disease severity markers such as CRP, serum albumin, CRP/albumin ratio and haemoglobin levels at induction and endoscopic disease severity. This is the first study to report a benefit of accelerated induction regimes when taking into the potential for provider bias based on differing disease severity.
Nalagatla et al 23 adjusted for the propensity score in their multivariable model and found no difference in in-hospital colectomy rates (OR 0.70, 95% CI 0.16-3.01). However, the overall colectomy rates in both groups in this study (8%-9%) were substantially lower than our study (17%-21%).
This may be related to overall lower disease severity in all parameters finding not seen in our unmatched cohort. However similar to our results in that study, the overall complication rate including infectious and/or non-infectious complications was not significantly higher in the high-dose group compared to standard-dose group in the propensity-matched cohort. Thus, overall accelerated dosing regimens did not seem to increase the risk of complications. There was one death in the accelerated induction group as a result of post-operative rectal stump leak and sepsis resulting in multiorgan failure.
We acknowledge that our study has a number of limitations. Due to the retrospective nature of the study, we were unable to collect every variable each day following admission with ASUC and were also unable to record the objective assessment of response and remission. We also had no data on serum infliximab levels or biomarkers such as faecal cal-