Anti-TNF therapy for ulcerative colitis in Brazil: a comparative real-world national retrospective multicentric study from the Brazilian study group of IBD (GEDIIB)

Background Anti-TNF therapy represented a landmark in medical treatment of ulcerative colitis (UC). There is lack of data on the efficacy and safety of these agents in Brazilian patients. The present study aimed to analyze rates of clinical and endoscopic remission comparatively, between adalimumab (ADA) and infliximab (IFX), in Brazilian patients with UC, and evaluate factors associated with clinical and endoscopic remission after 1 year of treatment. Methods A national retrospective multicenter study (24 centers) was performed including patients with UC treated with anti-TNF therapy. Outcomes as clinical response and remission, endoscopic remission and secondary loss of response were measured in different time points of the follow-up. Baseline predictive factors of clinical and endoscopic remission at week 52 were evaluated using logistic regression model. Indirect comparisons among groups (ADA and IFX) were performed using Student's t, Pearson χ2 or Fisher's exact test when appropriated, and Kaplan Meier analysis. Results Overall, 393 patients were included (ADA, n = 111; IFX, n = 282). The mean age was 41.86 ± 13.60 years, 61.58% were female, most patients had extensive colitis (62.40%) and 19.39% had previous exposure to a biological agent. Overall, clinical remission rate was 66.78%, 71.62% and 82.82% at weeks 8, 26 and 52, respectively. Remission rates were higher in the IFX group at weeks 26 (75.12% vs. 62.65%, p < 0.0001) and 52 (65.24% vs. 51.35%, p < 0.0001) when compared to ADA. According to Kaplan–Meier survival curve loss of response was less frequent in the Infliximab compared to Adalimumab group (p = 0.001). Overall, endoscopic remission was observed in 50% of patients at week 26 and in 65.98% at week 52, with no difference between the groups (p = 0.114). Colectomy was performed in 23 patients (5.99%). Age, non-prior exposure to biological therapy, use of IFX and endoscopic remission at week 26 were associated with clinical remission after 52 weeks. Variables associated with endoscopic remission were non-prior exposure to biological therapy, and clinical and endoscopic remission at week 26. Conclusions IFX was associated with higher rates of clinical remission after 1 year in comparison to ADA. Non-prior exposure to biological therapy and early response to anti-TNF treatment were associated with higher rates of clinical and endoscopic remission.


Background
Ulcerative colitis (UC) is a chronic and progressive immune-mediated inflammatory bowel disease (IBD) which can affect a variable extension of the large bowel and predominantly impacts individuals in the third and fourth decades of life [1]. The course of the disease is characterized by relapsing and remitting mucosal inflammation and up to 15% of patients may develop severe disease at diagnosis [2]. The aim of treatment is to induce and maintain clinical and endoscopic remission, but the adoption of evolving treatment targets such as histological healing has recently been considered [3]. Biological agents are indicated for patients with moderate to severely active disease, refractory to conventional treatment with aminosalicylates and immunomodulators, as well as for those who are steroid-dependent or refractory [4].
The most commonly used tumor necrosis factor (TNF) alpha inhibitors in UC are infliximab (IFX) and adalimumab (ADA). Golimumab is also approved in UC, but experience with this agent in Brazil is limited. The efficacy of IFX in inducing and maintaining remission in moderate to severe UC, in addition to reducing the need for colectomy is supported by the pivotal ACT 1 and ACT 2 studies [5]. Accordingly, data from the pivotal Phase 3 ULTRA I and II studies demonstrated that treatment with ADA was able to induce and maintain remission in patients with moderate to severe UC and was well tolerated [6].
Although pivotal studies represent an important source of data on the efficacy and safety of a new treatment, their selective designs, with multiple inclusion and exclusion criteria, prevent the generalization of findings for routine clinical practice. Therefore, real world studies are warranted. Even though several European and North American studies have demonstrated real-world experience on the efficacy and safety of anti-TNF agents in the management of UC [7][8][9][10][11], the phenotype of disease can differ in various ethnic groups, which may be associated to different genetic backgrounds [12]. Thus, it is crucial to have data of specific drug efficacy in diverse populations with different demographic and socioeconomic background, such as Latin America. Possible differences in comparison with North American or European populations may limit extrapolation of available data, what emphasizes the need for specific local studies.
In this scenario, there is lack of data on the efficacy and safety of anti-TNF agents in Brazilian patients in UC, as public and private access to these agents is relatively recent [13]. Therefore, the present study aimed to analyze rates of clinical and endoscopic remission comparatively, between ADA and IFX, in Brazilian UC patients, and evaluate possible factors associated with clinical and endoscopic remission after 1 year of treatment.

Study design
A national retrospective multicenter study (24 centers) was carried out initially including 424 patients with UC treated with anti-TNF therapy (ADA or IFX). No patients with golimumab were included. Data were collected electronically from patient records. Inclusion criteria were clinical, endoscopic and/or histological criteria for UC diagnosis; adult patients (age over 18 years old); use of ADA or IFX due to active UC in any phase of treatment. Patients with indeterminate colitis or Crohn's disease (CD), hospitalized patients, or those who had undergone previous colorectal surgery, or with lack of essential data were excluded. Patients were allocated into 2 groups (ADA or IFX) and a comparative study was performed.

Included variables
We analyzed sex, age, age at diagnosis, body mass index, disease duration, active smoking, associated comorbidities, concomitant or previous extraintestinal manifestations (EIM), use of corticosteroids at the treatment initiation (as co-induction), concomitant use of azathioprine, and previous exposure to biological therapy. The extent of the disease was classified according to the Montreal classification (E1: proctitis; E2: left-sided colitis and E3: extensive colitis) [14]. Disease activity was assessed using the partial Mayo score [15] at baseline, weeks 8, 26 and 52, or at the last visit. In addition, the need for colectomy during follow-up, loss of response until week 52 of treatment, and presence of adverse events were evaluated. Biochemical data such as hematocrit (%), hemoglobin (g/dl), albumin (g/dl), C-reactive protein (mg/dl), and fecal calprotectin (μg/g) were additionally evaluated, when available.

Definitions and outcomes
The primary outcome was the proportion of patients achieving clinical remission at weeks 8, 26 and 52. Secondary outcomes were clinical response, endoscopic remission, and rates of secondary loss of response. Clinical remission was defined as partial Mayo score ≤ 2. Clinical response was defined as a reduction on partial Mayo subscore ≥ 2 points between baseline and weeks 8, 26, and 52. Endoscopic remission was defined as endoscopic Mayo subscore ≤ 1. Secondary loss of response was defined as a need for one of the following outcomes during follow-up: colectomy, dose optimization, need for corticosteroids as rescue therapy or a switch to another biological agent.

Statistical analysis
Data were reported using "as-observed" analysis, with the denominator being the total number of patients with available data in the pre-established time points. For quantitative variables with normal distribution, mean and standard deviation (SD) were presented, and Student's t test was used to compare two independent samples. Categorical data were presented as percentages, and Pearson χ 2 or Fisher's exact test were used to compare two proportions (from independent samples). Univariate logistic regression was used to identify predictors on categorical outcomes, such as presence or absence of remission at week 52. Survival analysis was performed using Kaplan-Meier curves and log-rank test; the considered outcomes were loss of response and colectomy. Values of p < 0.05 were considered statistically significant. The statistical analyses were performed using IBM SPSS v. 22.0 (UNICOM Global, Mission Hills, United States).

Ethical considerations
This study was approved by the Local Research Ethics Committee, Botucatu Medical School (CAAE: 13,973,519.0.1001.5411) and by all respective boards from participating centers (listed in Declarations).

Clinical characteristics
A total of 424 patients were initially evaluated, and 31 were excluded for lack of essential data. A total of 393 patients were included in the full analysis. Baseline characteristics are described in detail in Table 1. The mean age was 41.86 ± 13.60 years, 61.58% were female and 39.13% had associated comorbidities, the most frequent being hypertension (12.79%). Regarding the type of anti-TNF agent, 111 patients received ADA and 282 received IFX. Most patients had extensive colitis (62.40%) and 19.39% were previously exposed to a biological agent. The frequency of EIMs was higher in the ADA group. Co-induction with corticosteroids and the use of azathioprine in combination therapy were more frequent in the IFX group. In the comparative analysis between treatments, a higher frequency of moderate to severe endoscopic activity was observed in the IFX group at baseline.
Efficacy data from the whole sample is illustrated in detail in Table 2. Overall clinical remission rates were 66.78% at week 8, 71.62% at week 26 and 82.82% at week 52, respectively. Overall clinical response rates were 61.25% at week 8, 83.85% at week 26 and 87.46% at week 52, respectively. Endoscopic remission was observed in 50% of patients at week 26 and 65.98% at week 52. Additionally, there was a decrease in the Mayo score throughout the study period and an improvement in biochemical parameters.
Comparative data is illustrated in Fig. 1. Clinical remission rates were higher in the IFX group at weeks 26 (IFX: 75.12% vs. ADA: 62.65%, p < 0.0001) and 52 (IFX: 65.24% vs. ADA: 51.35%, p < 0.0001). There was no significant difference in endoscopic remission rates between the groups after 26 and 52 weeks. Table 3 describes in detail a comparative analysis in efficacy, clinical and biochemical parameters between the groups. After induction, clinical remission rate was higher in the ADA group at week 8 (IFX: 66.14% vs. ADA: 68.32%, p < 0.0001). C-reactive protein values were lower in the IFX group at weeks 26 (p = 0.0181) and 52 (p = 0.0008) and fecal calprotectin levels were also lower in the IFX group at week 52 (p = 0.0047). Table 4 describes additional efficacy and safety parameters. The mean time of treatment duration was longer with IFX as compared to ADA (41.23 ± 33.14 vs. 28.93 ± 23.36 months, p < 0.001). Secondary loss of response rates and need for anti-TNF dose optimization were more frequent in the ADA group (p < 0.001). Moreover, more patients in the ADA group needed to switch biological therapy as compared to IFX (p = 0.015). Colectomy rates were higher in the IFX group (p = 0.007). Adverse events were reported in 13 (11.71%) patients in the ADA group and in 44 (16.67%) patients in the IFX group (p < 0.0001). There were no differences between the groups in infectious adverse events. Overall, 4 patients died during anti-TNF treatment (1 in ADA group and 3 in IFX group), because of severe colitis or infection.
According to the Kaplan-Meier survival curve, loss of response was less frequently observed in Infliximab as compared to the Adalimumab group (p = 0.001), Fig. 2A. There was no difference regarding colectomy rates between the groups (p = 0.651), Fig. 2B.

Factors associated with clinical and endoscopic remission at week 52
According to logistic regression analysis (   p < 0.0001). The type of anti-TNF used was not associated with endoscopic remission at week 52.
According to the Kaplan-Meier survival curve, clinical remission at week 52 was not different between biologic naïve or biologic exposed patients (p = 0.783), Fig. 3A. On the other hand, biologic naïve patients showed a lower probability of loss of response as compared to biologic exposed patients (p = 0.003), Fig. 3B. Colectomy rates were not different between the groups, Fig. 3C.

Discussion
This study reports the indirect retrospective comparison between IFX and ADA in a multicenter Brazilian cohort of UC patients. This population of predominantly biological-naive patients (80.6%) showed an overall clinical response rate at week 8 of 61.25%, and the proportion of patients in clinical remission at weeks 8, 26 and 52 was 66.78%, 71.62 and 82.82%, respectively. Clinical remission rates were significantly higher in patients treated with IFX at weeks 26 and 52. Overall, endoscopic remission was observed in 65.98% at week 52, with no differences between the two groups. Older age, no prior exposure to biological therapy, treatment with IFX, clinical response at week 26 and endoscopic remission at week 26 were associated with higher rates of clinical remission at week 52.
Since the approval of IFX [5], the first biological agent for the management of UC, the landscape of medical treatment of the disease has drastically improved, especially for individuals with moderate to severe disease and those refractory to conventional therapies. Biological therapy, mainly TNF-alpha inhibitors, has been associated with clinical remission, endoscopic healing, reduction in the need for hospitalizations and colectomy [16]. Data supported by randomized controlled trials have demonstrated that both IFX and ADA are effective in  inducing and maintaining clinical remission in patients with moderate to severely active UC [5,6], although clinical remission rates after 1 year were not higher than 20.5% for both agents.
The results of the present study are in tune with the reported efficacy of these agents in the real-world scenario, where the proportion of responders after 52 weeks of exposure varies from 30 to 65% for ADA [7-10, 13, 17, 18] and from 39 to 70% for IFX [11,[19][20][21][22], respectively. The discrepancy in the performance of biologics between the real-world scenario and pivotal studies is usually attributable to the restriction of concomitant treatments that could favor a response and the required washout period between one drug and another in the design of randomized clinical trials. However, it is important to emphasize that there is significant heterogeneity in methodology, patient populations and clinical scoring systems among the available real-world studies in UC, which limits extensive comparisons with our results.  In the present study, there was a statistically significant difference in the proportions of remitters in the IFX group vs. the ADA group favoring ADA at week 8 and IFX in both weeks 26 and 52. The higher efficacy with ADA after 8 weeks in our study is probably a consequence of unadjusted confounding factors indicating more severe disease at baseline in patients treated with IFX, what could have reduced efficacy numbers in the IFX group. In a recent systematic review and network meta-analysis of important trials in biologic-naïve UC patients, IFX, ADA and vedolizumab were superior to placebo, and vedolizumab was superior to ADA for maintenance of clinical remission and endoscopic improvement in patients who responded to induction therapy [23]. Several reallife studies comparing the efficacy of different anti-TNF agents demonstrated controversial findings. A recent observational retrospective Italian study comparing IFX, ADA and golimumab observed a better treatment effectiveness in patients treated with IFX as compared to other treatments with the lowest percentages of response rates in all outcomes in patients treated with golimumab (p < 0.01). However, after applying a propensity analysis, no statistically difference in each outcome evaluated was identified [8]. Accordingly, a recent Korean retrospective study reported no significant differences between IFX and ADA treatment in the rate of clinical remission or clinical response at 8 or 52 weeks [11]. Conversely, the U.S. cohort study using a large administrative claims database showed that the risk of corticosteroid use was significantly lower in IFX-treated patients, as compared to ADA (HR, 0.82; 95% CI, 0.68-0.99), although the risk of hospitalization and serious infections were comparable [24]. These results from different countries demonstrate the variation in findings in the real-world setting.
Despite advances in medical management of UC with the introduction of biologics and small molecules, the potential for disease modification in terms of colectomy rates in the biological era remains unclear. Our study showed a colectomy rate of 4.5% in patients who used ADA and 6.59% with IFX, in a follow-up of 28.93 ± 23.36 and 41.23 ± 33.14 months, respectively. This number can be compared to other similar studies. A recent Swiss population-based study assessing colectomy rates in UC patients demonstrated a significantly decrease in colectomy rates for UC over time after 2005 [25]. Accordingly, a Canadian population-based study showed a significant decrease in the temporal trends of elective colectomy rates from 1997 to 2009, along with a marked increase in the prescriptions of infliximab after 2005 [26]. Although this time-trend decrease might also be related to improvements in care, including earlier diagnosis and adoption of guideline recommendations in clinical practice, the decreasing trend suggests a potential role of biologics as disease-modifying agents in the natural course of UC. It may take time to observe this trend in cohort studies around the globe. More population-based data from Brazil regarding reduction of colectomy rates over time in UC are awaited.
In the present study, prior exposure to another biologic was associated with a lower chance of long-term clinical remission, which may reflect disease severity at baseline. This is a common practice in real setting, what is captured in observational studies as ours. Data evaluating the influence of previous exposure to biologics in response to anti-TNF treatment is conflicting. The randomized, double-blind, placebo-controlled ULTRA 2 trial demonstrated that among patients who had previously received anti-TNF agents, rates of remission at week 8 were 9.2% on ADA patients and 6.9% on placebo (p = 0.559); corresponding values for week 52 were 10.2% and 3% (p = 0.039) [6]. Conversely, in a retrospective cohort study from the Spanish ENEIDA registry, response to prior treatment with IFX was the only predictive factor of response to ADA at week 12, which was observed in 90% of IFX remitters, 53.8% of responders and 33.3% of primary non-responders (p = 0.01). These observations should be interpreted with caution given that most of studies do not assess whether previous treatment was discontinued due to pharmacokinetic failure, immunogenicity, or mechanistic failure.
Early response to anti-TNF treatment has consistently been reported as a predictive factor of higher long-term remission rates. Data from an Italian cohort of ADAtreated patients showed that clinical remission and low C-reactive protein at week 12 predicted clinical remission at week 54 (OR 4.17, 95% CI 2.36-19.44; OR 2.63, 95% CI 2.32-14.94, respectively). Likewise, a Swedish retrospective multi-center study of IFX treatment in UC patients identified non-response at 3 months as an independent risk factor for poor outcome, predicting subsequent colectomy [22]. We observed similar results in the present study since clinical response at week 8 was associated with endoscopic remission and clinical response at week 26 was associated either with clinical and endoscopic remission at week 52.
Data derived from two meta-analysis and a pooled analysis of IBD trials have demonstrated no increased risk of serious infections in antitumor necrosis factors-treated patients compared with placebo [27][28][29]. In our study, anti-TNF treatment was well tolerated, and no new safety signals were observed, with no difference between the groups. Apart from that, the percentage of adverse events in each group was low in comparison to pivotal trials and other retrospective studies. Comparative safety analyses between different anti-TNFs are limited in the literature. Through the analysis of a health insurance database, it was demonstrated that subcutaneously administered anti-TNFs exhibited a higher risk of serious infections (HR, 1.34; 95% CI, 1.18-1.53) than intravenous anti-TNF [30]. On the other hand, a retrospective study in pediatric IBD patients observed a higher overall incidence of infections in infliximab compared with adalimumab-treated patients [31]. Additionally, a Brazilian retrospective single-center study demonstrated no differences between ADA and IFX patients in CD management (63.2% with IFX and 64.5% with ADA, p = 0.879), with no differences in infections or treatment interruption [32]. The lower numbers of adverse events in our UC multicentric national study are probably associated to limitations in data capturing by different physicians. The infusion site reaction rate was in line with previous reports. Data from TREAT registry reported an incidence of 2.8% with IFX in terms of infusion reactions, most common being headache and arthritis [33] while injection-site reactions were reported in an incidence of 0.1/100 patient-years in the adalimumab safety PYRAMID registry [34]. These numbers were comparable to the findings of our study.
Our study is associated with some limitations which need to be considered before the final analysis of the results. Firstly, the sample size was limited considering the increasing incidence and high estimated prevalence of UC in Brazil (estimated prevalence of 66.45 per 100,000 in 2020) [35]. However, this was unavoidable considering that biologics were just recently reimbursed for UC management in Brazil and that penetration of anti-TNF agents in Latin America is lower in UC as comparable to the rest of the world [36]. Another important limitation was a natural selection bias, a common feature of the retrospective and observational study design, where more severe patients could be directed to IFX use. However, these findings highlight the that physicians may confront in the management of IBD patients in our country [37]. The observational nature of this study carries the inherent biases associated with the retrospective study design. In addition, not all information regarding clinical scores were readily available in all timepoints of interest in medical charts, which limited the assessment of treatment response. No patients with golimumab were included, demonstrating the lack of experience with this agent in our country. Safety analysis was probably underestimated, due to bias in data collection. Another important point is that a systematic evaluation of Mayo endoscopic subscores was not available for the entire population of patients. Lastly, not all patients included in the study presented moderate-to-severe activity at induction, which may have interfered with the final results. As this is the first multicenter study in the country with the aiming to evaluate the use of anti-TNF in patients with UC, we found it interesting to evaluate the scenario and epidemiological profile of all patients with indication for the use of the medication and, therefore, the study inclusion criteria were more comprehensive. We are aware that other study designs would be more appropriate for the study, such as propensity score matching, or the inclusion of biologic naïve patients, exclusively. Further studies can clarify these issues in the future. Despite these limitations, our study provides insightful information as being the first multi-center Brazilian study to report on long-term outcomes of anti-TNF treatment in UC patients, from both private and public settings.

Conclusions
In summary, in this national Brazilian retrospective study, anti-TNF therapy with IFX and ADA were effective in the management of UC. Clinical remission rates were higher with ADA at week 8. On the other hand, IFX was associated with higher rates of clinical remission at weeks 26 and 52 in comparison to ADA. Patients naive to biological therapy presented higher rates of clinical and endoscopic remission.