The incidence of remission and indicators of inadequate response to advanced therapy in patients with ulcerative colitis: results from medical charts in the United Kingdom

Abstract Objective A considerable proportion of patients with moderate-to-severe ulcerative colitis (UC) treated with advanced therapies do not achieve remission, even after 1 year of treatment, and suboptimal response to advanced therapies is frequently observed in clinical practice. This study aimed to analyze clinical practice data in the United Kingdom (UK) and assess the rates of clinical remission and inadequate response with advanced therapies among patients with UC. Methods This retrospective chart review included patients with UC who initiated a new advanced therapy (i.e. adalimumab, infliximab, golimumab, tofacitinib, or vedolizumab) between January 2017 and September 2019 from eight clinics across the UK. At least 12 months of data before and after starting an advanced therapy were required. Remission was assessed using components of the Mayo score. Inadequate response was defined by therapeutic adjustment or emergency treatment. Results Among 238 patients included (female: 46.6%; median age: 42.0 years; median follow-up: 28.8 months), 178 patients (74.8%) were biologic-naïve. At 12 months, 87 patients (53.9%) had achieved remission (median time to remission: 7.6 months), although 29 (33.3%) among them had required therapeutic modifications to achieve remission. At 12 months, 105 patients (44.3%) had at least one indicator of an inadequate response (median time to the first indicator of inadequate response: 18.0 months). Conclusions Nearly half of the patients did not achieve remission, and almost half of the included patients had an inadequate response within 1 year after treatment initiation. More effective therapies are needed to effectively treat UC. PLAIN LANGUAGE SUMMARY Treatment of ulcerative colitis (UC) follows a stepwise approach that considers disease severity and disease activity. It has the goal of achieving and maintaining steroid-free remission and healing of the gut lining. Treatment options for UC include several conventional and advanced therapies. However, suboptimal response to treatment has been reported in previous observational studies. This results in treatment adjustments, such as therapy discontinuation, dose intensification, and addition of conventional therapies, as well as lengthy concurrent use of corticosteroids. This retrospective chart review evaluated clinical practice data from eight clinics across the United Kingdom. This was done to assess rates of clinical remission and indicators of suboptimal response to advanced therapies among patients with UC. The analysis included data from January 2017 to September 2019. Nearly half of the patients did not have clinical remission within 1 year after starting advanced therapies. Optimization of advanced therapies was often seen, even in patients in remission. The most common indicators of suboptimal therapy were therapy discontinuation, dose escalation of advanced therapy, and the addition of conventional therapies. Our findings suggest that the efficacy of advanced therapies to treat UC remains insufficient in clinical practice. Thus, there is a need for more effective treatment alternatives to achieve better outcomes for patients with UC.

Current guidelines for UC recommend a stepwise approach depending on disease activity, with the goal of achieving and maintaining steroid-free remission and mucosal healing [4][5][6] . Conventional therapeutic options include 5-aminosalicylic acid (5-ASA), corticosteroids (CS), and immunosuppressants. Patients with a lack of response to conventional therapies are treated with advanced therapies such as anti-tumor necrosis factor alpha (TNFa) inhibitors, anti-integrins, anti-interleukin agents, and Janus kinase (JAK) inhibitors 7,8 . In addition, the novel class of oral sphingosine-1-phosphate receptor modulators was recently approved in Europe as an advanced therapy for patients with UC 9 .
In recent real-world studies, suboptimal response to advanced treatment for UC has been reported frequently. For example, approximately one-third of patients could not achieve remission within 1 year of treatment initiation 10,11 . In another study, nearly half of the patients received a dose escalation of advanced therapy during maintenance 12 . Additionally, more than two-thirds of patients received combination therapy with immunosuppressants while being treated with advanced therapies, with an associated increased risk of adverse effects 13,14 . Furthermore, prolonged use of steroids is common in patients with UC, despite the fact that achieving CS-free remission is an important therapeutic goal 15,16 .
There is insufficient evidence on predictors of sustained efficacy of advanced therapies in UC in real-world practice. Therefore, the primary aim of this study was to evaluate clinical remission and the occurrence of indicators of inadequate response to advanced therapies among patients with UC in the UK in the 12 months after initiation. For subgroup analysis, the associations with prior biologic exposure, concurrent administration of CS at therapy start, and achievement of remission within 12 months were evaluated. Secondary objectives included the exploration of factors associated with time to remission and time to inadequate response, as well as discontinuation, dose escalation, CS dependency, and concomitant use of CS during maintenance.

Study design and data collection
This multicenter, retrospective chart review aimed to describe the incidence of remission, indicators of inadequate response to advanced therapy, and requirement for treatment adjustment in patients with a diagnosis of UC in a reallife setting in the UK. Gastroenterology departments treating patients with UC in the National Health Service (NHS) were randomly invited to participate in this study through the Clinical Research Network of the National Institute for Health Research. Study site selection was based on available capacity in terms of personnel and time as well as willingness to participate in this study. Data collection was carried out by gastroenterology departments from three teaching hospitals and five district general hospitals in the UK via an electronic case report form.
A web-based data entry form was used to capture data in a standardized format. Before the study started, all site investigators received detailed information on the study background, its objectives, and related regulatory obligations. Regular contact with study sites was maintained throughout the data collection process to resolve data entry discrepancies and query missing data fields. Data collection and management were carried out using MySQL version 8.0 and statistical analyses were done in Stata version 17.0 (StataCorp LLC, College Station, TX, USA).
Inclusion criteria comprised a physician-confirmed diagnosis of UC, an age of 18 years or older, and initiation of a new advanced therapy between January 2017 and September 2019. The agents considered as advanced therapies for UC management were adalimumab, golimumab, infliximab, tofacitinib, and vedolizumab. All eligible patients had to be included in an initial screening list. In selecting patients from this list, the aim was to achieve an even distribution of patients with respect to the class of index therapies received (i.e. anti-TNF, anti-integrin, or JAK inhibitor). Up to 40 eligible patients were selected from each participating study site.
Patients who received biological/JAK inhibitor therapy for non-UC conditions (e.g. Crohn's disease, indeterminate colitis, or rheumatoid arthritis) during the whole study period were excluded from the study. Individuals who underwent a total colectomy before the start of an advanced treatment were also excluded.
According to the inclusion criteria, data were extracted for patients with a minimum of 12 months of available data before (baseline period) and after (follow-up period) the index date. The index date was defined as the date on which patients were prescribed a new advanced therapy and was considered the start of the follow-up period. The study endpoints were investigated for individual patient follow-up periods, ending at death, discontinuation of index therapy, or end of follow-up at the study site, whichever came first.

Study endpoints
Baseline patient demographics recorded included age at the start of index therapy and sex. Baseline clinical characteristics comprised the following: disease duration; use of conventional and advanced therapies during the 12-month baseline period; frequency of stools per day; rectal bleeding; endoscopic findings; physician's global assessment (PGA) of disease severity; and C-reactive protein and fecal calprotectin levels in the 4 weeks before the index date, if available.
The components of the Mayo score (i.e. stool frequency, rectal bleeding, endoscopic findings, and PGA) were used to characterize disease activity and define the achievement of clinical remission. Depending on the availability of data at the time of assessment, achievement of remission was defined based on a stepwise approach as either: a complete Mayo score of not more than 2; a partial Mayo score of not more than 1; scores for both patient-reported outcomes (stool frequency and rectal bleeding) of not more than 1; or a PGA score of not more than 1 17,18 . The assessments of disease activity were performed at 3, 6, 12, and 24 months after initiation of index therapy.
Indicators of inadequate response were defined as at least one of the following events: discontinuation of index therapy due to lack of efficacy; dose escalation of index therapy; augmentation with at least one newly prescribed non-advanced therapy (5-ASA, azathioprine, mercaptopurine, methotrexate, ciclosporin, or tacrolimus); CS dependency (use for 12 weeks); CS use during the maintenance phase (12 weeks after the index date until the end of the follow-up period); or UC-related hospitalization, surgery, or emergency department visit. The earliest identified of one of these proxies defined the time of first observed inadequate response.
Secondary endpoints comprised the estimation of factors associated with time to achievement of remission and indicators of inadequate response. Additionally, rates of discontinuation and dose escalation of index therapy were analyzed, as well as the leading reasons for therapy discontinuation based on pre-set response categories including: "worsening or unsatisfactory control of UC-related symptoms"; "switch to another advanced therapy"; "acute drug-related reaction or adverse events"; "development of neutralizing anti-drug antibodies"; "surgical intervention"; or "patient's decision to terminate the treatment". The predefined answer categories for reasons for dose escalation of the index treatment included: "worsening or unsatisfactory control of disease symptoms"; "low drug concentration"; or "dose adjustment (increase) due to weight change". The initial dose of index therapies at the start of the maintenance phase was assessed separately for each agent. Furthermore, the proportion of patients with concurrent use of CS at the index date and during the maintenance phase was described, as well as the number of patients with CS dependency for at least 12 weeks.

Subgroup analysis
Patients were stratified according to prior use of biological therapy, active concurrent use of CS at the index date, and achievement of the first remission in the first 12 months after the index date; the latter was used to analyze the inadequate response and therapeutic adoptions, such as augmentation with conventional therapies, dose escalation, and prolonged CS use, prior to achieving primary remission.

Statistical analysis
Patient characteristics during baseline were analyzed using descriptive statistics for the overall population and the predefined subgroups. Absolute and relative frequency tabulations described categorical variables. Mean, standard deviation, median, and interquartile range (IQR) were reported for continuous variables. Furthermore, time-to-event analyses were performed using the Kaplan-Meier method accounting for individual follow-up periods, censoring patients at the end of the observed period. The median time to the event and 95% confidence interval (CI) were reported for the overall population and the predefined subgroups. Patients were not included in the time-to-event analyses if the required data to assess the respective event at a specific time point were not available. Therefore, the number of patients in the analyzed populations varies by the endpoint.
Subgroup comparisons were performed using the Chisquared test for categorical variables, unpaired t-test and Wilcoxon rank-sum test for continuous variables, and log-rank test for time-to-event estimates. Factors associated with time to remission and time to inadequate response were estimated with multivariate Cox proportional hazards models. Estimated hazard ratios (HRs) and related 95% CIs were reported for each model. Covariates included in the model were age at index, sex, index therapy, experience with advanced therapies before the index date, active concurrent use of CS at index, and concurrent use of any of the aforementioned non-advanced therapies at index.

Ethical considerations
The study protocol was reviewed and approved by a scientific steering committee consisting of the study sponsor, the research organization conducting the analysis, and one independent gastroenterologist. The ethical and regulatory approval for this retrospective chart review in the UK (IRAS Project ID 288918) was granted by the Health Research Authority and Health and Care Research Wales. Given that the study addressed an anonymized data set, no informed consent of patients was required.

Patient baseline characteristics
In total, 238 patients with UC were included in this study (Table 1). Over half of patients were male (n ¼ 127, 53.4%) with a median age of 42 years (IQR: 29-58). The median disease duration amounted to 6.1 years (IQR: 2.1-11.1). Based on data documented in the 4 weeks prior to the start of index treatment, the median C-reactive protein level was 6 mg/L (n ¼ 154, IQR: 3-18) and the median fecal calprotectin level was 510 mg/g (n ¼ 63, IQR: 110-600). Before the index date, 60 patients (25.2%) had been treated with anti-TNFa agents, whereas other advanced therapies had not been prescribed previously. One-third of patients (n ¼ 84, 35.3%) received CS concomitantly with the index therapy at the start of the study period. Among the included patients, 45 patients (18.9%) were treated with adalimumab as the index therapy, 4 (1.7%) with golimumab, 79 (33.2%) with infliximab, 5 (2.1%) with tofacitinib, and 105 (44.12%) with vedolizumab. Available data from included patients were retrospectively analyzed for a median follow-up of 28.8 months (IQR: 20.3-38.2) post-index date.

Achievement of clinical remission
Among patients with available Mayo score or non-invasive components of the Mayo score (n ¼ 177) at baseline, 61 (46.1%) did not achieve first remission in the 12 months after initiation of the index therapy (median time to first remission: 7.6 months; 95% CI: 6.6-14.5; Figure 1). No differences were observed for rates of patients achieving remission between biologic-naïve and biologic-experienced patients or between patients with and without the use of CS at index (Supplemental Figure 1). One-third of patients (n ¼ 29, 33.3%) required at least one treatment adjustment to achieve remission within 12 months; the most frequently observed adjustments were augmentation with non-advanced therapies (n ¼ 12, 13.8%) and escalation of the index therapy (n ¼ 9, 10.3%).
.899 Median (IQR) 6 (4-7) 6 (4-7) 5 (5-6) .243 6 (4-7) 5 (4-7) .668 CRP level a (mg/L) .414 FC level a (lg/g)  Figure 2). Patients who did not achieve remission in the first 12 months after the index date experienced higher rates of inadequate response than patients in remission (at 12 months, 56.3% vs 34.4%; log-rank p < .001; Figure 3). Higher rates of inadequate response over time were observed among biologic-naïve patients than biologic-experienced patients (at 12 months, 50.3% vs 26.7%; log-rank p < .001; Supplemental Figure 2(A)). Patients with active use of CS at the index experienced higher rates of inadequate response than patients without the use of CS at the index (at 12 months, 52.4% vs 39.8%; log-rank p ¼ .039; Supplemental Figure 2(B)).   Table 2). The adjusted model for estimating the association with time to inadequate response showed that using vedolizumab as the index therapy correlated with a lower risk of experiencing an indicator of inadequate response than when using anti-TNFa agents (HR: 0.47; 95% CI: 0.31-0.71; p < .001; Table 3). No statistically significant relationship was found for prior exposure to advanced therapies, active use of CS at the index date, or any of the remaining covariates.

Discontinuation of advanced therapies
Overall, 50 patients (21.1%) discontinued the index therapy in the first 12 months after treatment initiation. During the entire follow-up period, 88 patients (37.0%) discontinued their index therapy, and 62 of these patients (70.5%) switched to another advanced treatment. Among biologic-naïve patients, higher rates of discontinuation of the index therapy (at 12 months, 24.4% vs 8.3%; log-rank p < .001) and higher rates of switching to other advanced therapy (at 12 months, 18.6% vs 3.4%; log-rank p ¼ .004) were observed than among biologic-experienced patients. Discontinuation rates were not significantly different between patients with and without active use of CS at the index, but patients without concomitant use of CS at the index experienced higher rates of switching to another advanced therapy (at 12 months, 20.0% vs 11.7%; log-rank p ¼ .022).
Among the patients who discontinued the index therapy, the most frequent primary reason for discontinuation (Table 4) was the worsening or unsatisfactory control of UC signs and symptoms (n ¼ 34; 38.6%), followed by switching to another advanced therapy (n ¼ 21; 23.9%) and the   occurrence of an adverse event (n ¼ 10; 11.4%). In total, 95 adverse events were reported among patients (Supplemental Table 1), the most frequent of which was anemia (n ¼ 14; 14.7%), followed by upper respiratory tract infection (n ¼ 9; 9.5%) and skin rash (n ¼ 6; 6.3%).

Dose escalation of advanced therapies
The initial dosage at the start of the maintenance phase was 40 mg every 2 weeks (n ¼ 42) or 160 mg every 2 weeks (n ¼ 1) for adalimumab, 300 mg every 8 weeks (n ¼ 88) or 300 mg every 4 weeks (n ¼ 9) for vedolizumab, and 5 mg twice a day (n ¼ 5) for tofacitinib. The median initial dosage was 350 mg every 8 weeks (mean ¼ 315.3 mg; n ¼ 73) for infliximab and 50 mg every 4 weeks (mean ¼ 66.7 mg; n ¼ 3) for golimumab. The prescribed dosing regimens at the start of the maintenance phase across the different therapies are reported in Supplemental Table 2. At 12 months, escalation of index treatment was reported for 34 patients (16.5%). Among biologic-naïve patients, higher rates of treatment escalation were observed than among biologic-experienced patients; no differences were observed for rates of escalation between patients with and without CS use at the index. During the entire follow-up period, 49 patients received treatment escalation; 36 patients (73.5%) owing to worsening of UC-related signs or symptoms, and nine patients (18.4%) owing to low drug concentration. No differences in rates of dose escalation were observed within the predefined subgroups.

Concurrent use of CS and assessment of the proportion of patients with CS dependency
At the start of the index therapy, 142 patients (59.7%) were treated with systemic CS, 11 (4.6%) with rectal budesonide, 15 (6.3%) with locally acting budesonide/CS, and 14 (5.9%) received both systemic and locally acting CS. At 12 months, 17 patients (8.7%) had received CS during the maintenance phase of the advanced therapy and nine patients (4.2%) were CS dependent.

Discussion
To our knowledge, this is one of the largest chart reviews investigating the efficacy of advanced therapies for UC in the UK in a real-life setting. Of note, almost half of the patients did not achieve remission in the 12 months after the index date, and one-third of patients in remission had treatment adjusted at least once previously. In comparison, a previous multicenter retrospective study from the UK on patients with UC reported that one-third of patients did not show remission, defined as a partial Mayo score of not more than 1 without any bleeding, at 52 weeks 11 . However, because this other study only included patients who started treatment with vedolizumab, it is not representative of the overall use of advanced therapies in the UK.
Moreover, our findings suggest that previous experience with advanced therapies is correlated with later achievement of remission, whereas in another multicenter retrospective study that assessed the real-world effectiveness of tofacitinib in the UK, the authors concluded that prior biologic exposure had no impact on the short-term rates of response or remission 19 . However, their study mainly included much younger patients and only a small proportion of these were biologic-naïve, which limits the validity of the comparison with biologic-experienced patients.
Furthermore, our study showed that almost half of the patients starting advanced therapies experienced at least one indicator of inadequate response within 12 months. In a recent multinational chart review among patients with UC treated with anti-TNFa agents in Europe and Canada, the estimated rates of suboptimal therapy, including dose escalation, augmentation, discontinuation of therapy, switch to another anti-TNFa agent, or emergency procedures, yielded similar results 20 . Events of inadequate response in our study, mainly driven by discontinuation of the index therapy, were more frequent and observed earlier among biologic-naïve patients than biologic-experienced patients, but no statistically significant association was found between the likelihood of experiencing an indicator of inadequate response and prior exposure to biologics. This may indicate a possible bias, as biologic-naïve patients may be monitored better when they are exposed to advanced therapies for the first time and treatment may be more urgent. We also observed that biologic-naïve patients were more likely to have severe rectal bleeding at baseline. Further subgroup analysis revealed that the indicators of inadequate response defined in this study were more common in patients who did not achieve remission within 12 months. Thus, we could validate the assessed proxies as meaningful quality measures and provide a benchmark for these performance indicators in our retrospective cohort study.
Discontinuation of index treatment occurred in approximately 20% of patients after 12 months in our study sample; this is consistent with a previous multicenter retrospective chart review, which found that two-thirds of patients treated with second-line adalimumab or golimumab in the UK remained on the same treatment, and that one-fifth switched to another advanced therapy within 1 year 21 . Interestingly, discontinuation rates in our study were higher among biologic-naïve patients than among those who had previous experience with biologics, and indicators of inadequate response to advanced therapy were mainly driven by discontinuation rates. One explanation of this finding might be that with the recent advent of a variety of advanced therapies, patients and treating physicians have the option to switch therapy, whereas previously, treatment had to be maintained owing to a lack of other medical alternatives.
During the entire follow-up period in our study, approximately 20% of patients received a dose escalation; most of them did so owing to worsening of disease signs or symptoms. This finding is within the range of results from previous observational studies, which revealed varying rates of dose escalation depending on the observed advanced therapies [19][20][21] . In a retrospective study in the USA and Europe, dose escalation during maintenance of biologic treatment was observed in one-fifth to one-third of patients with UC in Europe. The reasons for switching therapy were most frequently related to inadequate efficacy of the treatment 22 .
Concomitant use of CS at the start of index therapy was observed in one-third of our study sample. After 1 year of follow-up, dependence on CS was less frequent in this study than in previous research. A multicenter prospective study in the UK reported that approximately 15% of IBD patients had steroid excess or dependency, which was similar to the proportion reported in a registry study in the USA 15,23 . However, both studies addressed a broader population of patients with UC and were not limited to users of advanced therapies.
Several limitations must be considered when interpreting the results of this study. First, data were collected retrospectively in gastroenterology departments of clinics in England and Wales, which might limit extrapolation to other regions or countries. Moreover, despite the wide-ranging recruitment of sites, data from sites that had sufficient capacity and expressed interest in participating in the study were extracted, which could have led to selection bias. Additionally, assessments of disease activity and inadequate response to advanced therapies were constrained to the available data. To maximize the number of patients with any disease activity measure, a modification of the Mayo score was used to follow up disease activity over time, e.g. using the non-invasive components of the Mayo score for patients in whom endoscopic findings were not documented. However, one out of four patients had to be excluded from the analysis on achievement of clinical remission owing to missing data, resulting in a possible unspecifiable selection bias for this outcome. Furthermore, the evaluation of disease status, during the follow-up period, based on serum and fecal biomarkers, i.e. C-reactive protein and fecal calprotectin, was not feasible. Given that these values were not regularly documented in the study sites, an analysis of the evolution of these biomarkers after the index therapy could not be performed retrospectively. Finally, although data on efficacy and treatment discontinuation were analyzed using multivariate regression models adjusted for several patient and disease characteristics, unknown influencing factors may have been omitted.
A major strength of this study is the multicentric design and the inclusion of a sufficiently large patient sample based on real-world data of routine clinical decision-making with advanced therapies. The analyzed data set is representative of UC care in the UK, although allocation bias cannot be ruled out owing to the small number of participating study sites. Data extraction was performed for all eligible patients at the study sites who met the inclusion and exclusion criteria. In this way, we avoided potential selection biases associated with consent-based research studies.

Conclusions
In the UK, nearly half of the patients with UC starting advanced therapy did not achieve remission within the first year of treatment. The delay in achieving clinical remission was significantly related to prior exposure to biologics. In patients who had not previously received biologics, indicators of inadequate response occurred earlier than in patients who had previous experience with biologics. Furthermore, optimization of advanced therapies in terms of dosing, treatment switching, and use of concomitantly prescribed agents were frequently observed, even in patients who achieved clinical remission in the 12 months after therapy initiation. Finally, this study validates a set of indicators, based on changes in treatment patterns and healthcare resource utilization, as a meaningful quality measure for real-world efficacy in a retrospective analysis of medical records. In conclusion, this retrospective chart review found that the efficacy of advanced therapies remains insufficient in clinical practice in the UK. Given the emergence of novel therapeutic approaches to UC in recent years, further studies are needed to assess the impact of the changing treatment landscape in real-world settings.

Transparency
Declaration of funding