Abstract
Objectives
This study aimed to evaluate if and how the incidence of serious infection (SI) and active tuberculosis (TB) differ among seven biologic DMARDs (bDMARDs) in patients with RA considering the line of therapy.
Methods
Patients with RA from the British Society for Rheumatology Biologics Register for Rheumatoid Arthritis (BSRBR-RA) cohort who initiated etanercept, certolizumab, infliximab, adalimumab, abatacept, rituximab or tocilizumab from the first to fifth line of therapy were included. Follow-up extended up to 3 years. The primary outcome was SI and the secondary outcome was TB. Event rates were calculated and compared using Cox proportional hazards models, controlling for confounding with inverse probability of treatment weights. Comparisons were made overall and stratified by line of therapy. Sensitivity analysis was restricted to all treatment courses from 2009 (tocilizumab availability) until the end of the study (2018).
Results
Among 33 897 treatment courses (62 513 patient-years) the incidence of SI was 4.4/100 patient-years (95% CI 4.2, 4.5). After adjustment, hazards ratios (HRs) of SI were slightly higher with adalimumab and infliximab compared with etanercept. However, no clear pattern was observed when stratifying by line of therapy in terms of incidence rate or HR. Sensitivity analyses showed similar HRs among these treatments. Regarding TB, all 49 cases occurred during the first three lines of treatment and rarely since 2009.
Conclusion
The risk of serious infections does not appear to be influenced by the line of therapy in patients with RA. However, the risk of TB seems to be more frequent during the initial lines of treatment or prior to 2009.
Keywords: rheumatology, epidemiology, adverse events, biologic, safety, outcomes, observational, cohort, longitudinal, comparative
Rheumatology key messages.
Line of therapy does not significantly affect the risk of serious infections in RA patients.
Tuberculosis risk comparisons are confounded by treatment start year and line of therapy.
Introduction
RA is a chronic systemic autoimmune disease that primarily affects joints, but can also involve other organs, leading to joint damage and disability [1]. International recommendations advocate the use of a step-up strategy, aimed at treating early and targeting disease remission or low disease activity (treat to target) [2, 3]. Adding biologic DMARDs (bDMARDs) or targeted synthetic DMARDs (tsDMARDs) to conventional synthetic DMARDs (csDMARDs) is recommended if the aim is not reached with csDMARDs alone [3, 4]. However effective these treatments are, they also can lead to adverse events, such as infections. The risk of serious infections, such as tuberculosis (TB) is increased with TNF inhibitors (TNFis), and probably also other bDMARDs, although it seems now to occur at rates similar to that of the general population as adequate prevention is carried out [5].
When evaluating adverse effects of RA treatments, including serious infections, conclusions from randomised controlled trials may be limited because they are generally underpowered for such outcomes, with small sample sizes and short study duration. Additionally, the population is very selected and often does not represent the patients treated in routine clinical care [6]. Most information about the risk of serious infections therefore comes from observational studies. This is also not without limitations, as in observational studies, patients are not randomised and baseline characteristics of patients usually differ between treatment groups. For example, patients with non-TNFi bDMARDs are generally older, with longer disease duration, more previous bDMARD exposures and worse functional status [7], which may add a number of unknown confounding factors when looking at the risk of infection. The risk of infection is also not constant with time and seems to be higher during the first months of therapy [8–11]. It is also unknown if sequential biologic therapy exposure leads to lasting immune defects that could lead to a cumulative change in the immune system. In contrast, observational studies come to different conclusions about the risk of infections with different bDMARDs, with some finding an increased risk with certolizumab and others a lower risk compared with other bDMARDs [12–14]. In a study from the BSR Biologics Register for Rheumatoid Arthritis (BSRBR-RA), the rate of serious infections was lower with certolizumab pegol than etanercept [adjusted hazard ratio (aHR) 0.75 (95% CI 0.58, 0.97)] [14]. Yet this difference in infection rate did not persist when restricted to patients who were bio-experienced, suggesting unmeasured confounders may in part explain these observations.
In this study, we aimed to evaluate if line of therapy should be taken into account when evaluating the risk of serious infections, including TB. Our hypothesis is that comparing patients within the same line of therapy should reduce these unmeasured confounders and allow us to compare the risk of infections more adequately between different bDMARDs. The aim of our study was thus to evaluate if and how the risk of serious infections overall and active TB specifically differ between different bDMARDs (TNFi, tocilizumab, abatacept and rituximab) by line of therapy in a real-world population of RA patients.
Methods
Study population
Patients with a rheumatologist’s diagnosis of RA and registered in the BSRBR-RA cohort, starting either etanercept, certolizumab pegol, infliximab, adalimumab, abatacept, rituximab or tocilizumab from first line to fifth line of therapy between October 2001 and March 2019 were included. Patients treated with Janus kinase (JAK) inhibitors were not included in this study, as these were only recently introduced in the UK. The BSRBR-RA register is a prospective observational study aiming to evaluate safety events with biologics in the UK and has been described elsewhere [15]. Patients do not have to be bionaïve at the point of registration.
This study complies with the Declaration of Helsinki. Ethical approval was obtained from the UK North West Multicentre Research Ethics Committee (MREC 00/8/53) and all participants provided written informed consent.
Exposure of interest
Treatment exposure was defined as the time from the beginning of the treatment until 3 half-lives (see Supplementary Methods, available at Rheumatology online) after treatment stop, loss of follow-up, death or the end of the study period (defined as 3 years after treatment start or March 2019), whichever came first. The exception was rituximab, where exposure was defined as the time from the first infusion to 360 days after the last infusion, death, loss of follow-up or the end of the study period. The inclusion of a 3-year end-of-study period aimed to mitigate bias arising from varying infection rates in treatments administered over longer durations, such as older treatments when fewer options were available. Since infection incidence is more common at treatment initiation, a 3-year study period should adequately assess any significant differences [8, 9]. Treatments were classified by biosimilarity, i.e. bio-originator and their biosimilar therapies were considered the same. If a patient was under exposure of several biologics at the same time (i.e. switching biologics during the 3 half-lives following drug discontinuation), these were each considered as different treatment exposures.
Outcomes
The primary outcome was the occurrence of serious infections, defined as any first infection requiring hospitalization and/or intravenous antibiotics and/or resulting in death and/or active TB infection (regardless of other ‘serious’ features) during treatment exposure. The secondary outcome was the occurrence of active TB infection only during treatment exposure. The outcomes included the first infection that emerged during the use of the bDMARDs, but not recurring infections that might have occurred within the same bDMARD treatment course. However, we did include infections that happened in the same patient but while using a different bDMARD treatment.
Covariates of interest
Apart from bDMARD treatment, the major covariate of interest was line of therapy (from first to fifth line of therapy). Every change to a new bDMARD or tsDMARD was considered a new line of therapy, even if the participant returned to a treatment they had before. Although switching to a tsDMARD was regarded as discontinuation of the previous therapy and was added as a new line of treatment, the follow-up period under tsDMARDs was not taken into account due to the limited number of patients (see Exposure of interest).
A direct switch to a biosimilar with the same mechanism of action was not considered as a change of therapy. In the UK, abatacept is rarely prescribed as a first bDMARD and no bionaïve abatacept participants were included in BSRBR-RA. Other covariates of interest were baseline age, gender, disease duration, seropositivity, use of concomitant treatment with csDMARDs (yes/no), use of oral glucocorticoids (GC) (yes/no), presence of comorbidities as defined by the Rheumatic Disease Comorbidity Index [RDCI, ranging from 0 (no comorbidity) to 8] [16], 28-joint DAS (DAS-28), HAQ, calendar year of treatment initiation and time since study entry to the start of each line of bDMARD (categorized from 0 to 4, with 0 starting just before or at the moment of entering the study, 1 starting during the first year, 2 starting during the second year, 3 starting during the third year and 4 for the fourth year or more). Baseline was defined as the start date of each treatment course. The time since study entry was included as a confounding factor for serious infections, to adjust for potential lower reporting for patients registered for a long time in the cohort.
Statistical analysis
The number of events and incidence rates were reported. To adjust for potential confounding and selection bias, a multinominal logistic regression was used to create a propensity score for each treatment course, including all covariates of interest at baseline (date of start of each treatment course) except calendar year of treatment, time since study entry and line of therapy. Calendar year of treatment was not included as the model would no longer converge, which was thought to be because specific treatment cohorts were included at different time periods in BSRBR-RA. Covariates with missing values (Supplementary Table S1, available at Rheumatology online) were imputed using multiple imputations with chained equations with 20 imputation sets (supplementary methods, available at Rheumatology online). The propensity score was created separately for each set of imputations [17]. HRs of serious infection were then estimated using Cox regression with inverse probability weighting, based on the propensity scores, and adjusting for time since study entry. Treatment exposure was analysed without stratification and stratified by line of therapy (first line, second line, third line, fourth line, fifth line). We did not include later lines of therapy, as the number of patients in each treatment course was very small.
As a sensitivity analysis, only participants prescribed treatment since tocilizumab was available (2009) were included, as tocilizumab is one of the last treatments of a new class that was introduced in the BSRBR-RA, prior to JAK inhibitors.
We did not adjust for multiple hypothesis testing, as this was a pharmacovigilance study and we were more concerned about a type 2 error when studying safety outcomes.
Analyses were done using Stata version 14.0 (StataCorp, College Station, TX, USA) and R version 3.60 (R Foundation for Statistical Computing, Vienna, Austria) with packages haven, ggplot and tableone.
Results
A total of 33 897 treatment courses were included in this analysis, 10 643 with etanercept, 7835 with adalimumab, 4430 with infliximab, 1614 with certolizumab, 5556 with rituximab, 2633 with tocilizumab and 1186 with abatacept, contributing to 62 513 person-years of follow-up. The mean observation time per drug ranged from 1.64 (tocilizumab) to 2.16 (rituximab) by treatment and did not vary substantially among treatments.
Participants starting abatacept, tocilizumab and rituximab were older, had more previous bDMARDs, longer disease duration and more comorbidities compared with patients with TNFis (Fig. 1, Table 1). Rituximab and tocilizumab were rarely given as a first bDMARD treatment (7.3% and 11.4%, respectively). Compared with first-line bDMARDs, with later lines of therapy, patients were older, more often prescribed oral GC and had longer disease duration and higher HAQ scores (Supplementary Table S2, available at Rheumatology online).
Figure 1.
Percentage of treatment course by line of therapy and bDMARDs. ABA: abatacept, ADA: adalimumab, CERT: certolizumab pegol; ETN: etanercept; IFX: infliximab; RTX: rituximab; TCZ: tocilizumab
Table 1.
Baseline characteristics at the start of each treatment
| Characteristics | ETN | ADA | IFX | CERT | RTX | TCZ | ABA |
|---|---|---|---|---|---|---|---|
| Patients, n | 10 643 | 7835 | 4430 | 1614 | 5556 | 2633 | 1186 |
| Patient-years | 19 117 | 14 501 | 8135 | 2725 | 12 009 | 4341 | 1686 |
| Line of therapy, n (%) | |||||||
| 1 | 6075 (57.1) | 4886 (62.4) | 3332 (75.2) | 1151 (71.3) | 406 (7.3) | 301 (11.4) | 0 (0.0) |
| 2 | 3837 (36.1) | 2206 (28.2) | 727 (16.4) | 210 (13.0) | 2592 (46.7) | 584 (22.2) | 201 (16.9) |
| 3 | 536 (5.0) | 596 (7.6) | 267 (6.0) | 129 (8.0) | 1846 (33.2) | 923 (35.1) | 367 (30.9) |
| 4 | 147 (1.4) | 107 (1.4) | 82 (1.9) | 83 (5.1) | 568 (10.2) | 596 (22.6) | 379 (32.0) |
| 5 | 48 (0.5) | 40 (0.5) | 22 (0.5) | 41 (2.5) | 144 (2.6) | 229 (8.7) | 239 (20.2) |
| Age, years, mean (s.d.) | 57.2 (12.1) | 56.9 (12.0) | 56.3 (12.2) | 55.6 (12.6) | 60.2 (11.6) | 58.7 (12.0) | 60.3 (11.5) |
| Female, n (%) | 8173 (76.8) | 6061 (77.4) | 3403 (76.8) | 1230 (76.2) | 4342 (78.1) | 2091 (79.4) | 945 (79.7) |
| Disease duration, years, mean (s.d.) | 13.1 (9.5) | 13.4 (9.7) | 13.7 (9.4) | 10.5 (9.3) | 16.5 (10.1) | 15.5 (10.4) | 17.9 (10.7) |
| Seropositivity, n (%) | 6441 (65.0) | 4805 (62.6) | 2904 (66.3) | 824 (58.2) | 3506 (67.0) | 1350 (62.3) | 638 (61.6) |
| Concurrent csDMARDs, n (%) | 7805 (73.3) | 6156 (78.6) | 4189 (94.6) | 1368 (84.8) | 4755 (85.6) | 2094 (79.5) | 1011 (85.2) |
| Concurrent oral GC, n (%) | 4319 (40.6) | 3232 (41.3) | 2027 (45.8) | 470 (29.1) | 2642 (47.6) | 1221 (46.4) | 701 (59.1) |
| DAS-28, mean (s.d.) | 5.8 (1.5) | 6.1 (1.3) | 6.4 (1.2) | 5.7 (1.2) | 5.7 (1.4) | 5.3 (1.6) | 5.3 (1.4) |
| HAQ, mean (s.d.) | 1.8 (0.7) | 1.9 (0.6) | 2.1 (0.6) | 1.5 (0.8) | 1.9 (0.6) | 1.7 (0.7) | 1.9 (0.7) |
| Smoking, n (%) | |||||||
| Never smoker | 4290 (41.4) | 3076 (39.9) | 1766 (40.3) | 680 (45.2) | 1875 (38.0) | 929 (42.0) | 426 (41.0) |
| Ex-smoker | 3982 (38.5) | 2908 (37.7) | 1662 (37.9) | 526 (34.9) | 1913 (38.7) | 795 (35.9) | 373 (35.9) |
| Current smoker | 2081 (20.1) | 1726 (22.4) | 954 (21.8) | 300 (19.9) | 1150 (23.3) | 490 (22.1) | 240 (23.1) |
| RDCI, mean (s.d.) | 1.2 (1.3) | 1.1 (1.3) | 1.1 (1.2) | 1.0 (1.2) | 1.4 (1.4) | 1.4 (1.4) | 1.5 (1.5) |
| Year of treatment initiation, n (%) | |||||||
| ≤2004 | 4518 (42.5) | 1901 (24.3) | 2954 (66.7) | 0 (0.0) | 20 (0.4) | 0 (0.0) | 1 (0.1) |
| 2005–2008 | 1999 (18.8) | 4313 (55.0) | 1128 (25.5) | 0 (0.0) | 1210 (21.8) | 23 (0.9) | 43 (3.6) |
| 2009–2013 | 1173 (11.0) | 869 (11.1) | 179 (4.0) | 904 (56.0) | 3128 (56.3) | 1131 (43.0) | 303 (25.5) |
| ≥2014 | 2953 (27.7) | 752 (9.6) | 169 (3.8) | 710 (44.0) | 1198 (21.6) | 1479 (56.2) | 839 (70.7) |
ABA: abatacept; ADA: adalimumab; CERT: certolizumab; ETN: etanercept; IFX: infliximab; RTX: rituximab; TCZ: tocilizumab.
Serious infections
In total, there were 2732 serious infections (any first infection with the treatment of interest requiring hospitalization and/or intravenous antibiotics and/or resulting in death and/or active TB infection) with a crude rate of 4.4 (95% CI 4.2, 4 5) per 100 patient-years of bDMARD exposure. Participants treated with adalimumab and infliximab had higher crude HRs of infection than etanercept. Crude HRs of infection were lower with certolizumab and rituximab and not significantly different with tocilizumab and abatacept compared with etanercept (Fig. 2, Table 2). After adjustment, HRs of serious infections were still higher with infliximab and adalimumab than etanercept, but not significantly different for the other groups, although there was a higher HR of infection with tocilizumab [Fig. 2; aHR 1.19 (95% CI 0.98, 1.44)]. There was no clear pattern when stratifying by line of therapy, whether in the incidence rate or HR.
Figure 2.
HRs of serious infection by treatment compared with etanercept: (A) unadjusted; (B) adjusted for age, gender, disease duration, seropositivity, use of concomitant treatment with csDMARDs, use of oral GC, presence of comorbidities as defined by the RDCI, DAS-28 and HAQ using propensity score weighting and time since study entry; (C) adjusted for age, gender, disease duration, seropositivity, use of concomitant treatment with csDMARDs, use of oral GC, presence of comorbidities as defined by the RDCI, DAS-28 and HAQ using propensity score weighting and time since study entry and stratified by line of therapy. ABA: abatacept; ADA: adalimumab; CERT: certolizumab pegol; ETN: etanercept; IFX: infliximab; RTX: rituximab; TCZ: tocilizumab
Table 2.
Incidence of serious infections per drug
| Characteristics | Total | ETN | ADA | IFX | CERT | RTX | TCZ | ABA |
|---|---|---|---|---|---|---|---|---|
| Patients, n | 33 897 | 10 643 | 7835 | 4430 | 1614 | 5556 | 2633 | 1186 |
| Patient-years | 62 513 | 19 117 | 14 501 | 8135 | 2725 | 12 009 | 4341 | 1686 |
| Serious infections, n | ||||||||
| All lines of therapy | 806 | 680 | 481 | 73 | 433 | 191 | 68 | |
| 1st line | 503 | 466 | 389 | 56 | 56 | 17 | – | |
| 2nd line | 260 | 159 | 68 | 7 | 203 | 46 | 11 | |
| 3rd line | 32 | 44 | 15 | 5 | 115 | 73 | 23 | |
| 4th line | 8 | 7 | 7 | 1 | 46 | 37 | 21 | |
| 5th line | 3 | 4 | 2 | 4 | 13 | 18 | 13 | |
| Incidence per 100 patient-years (95% CI) | ||||||||
| All lines of therapy | 4.2 (3.9, 4.5) | 4.7 (4.3, 5.1) | 5.9 (5.4, 6.5) | 2.7 (2.1, 3.4) | 3.6 (3.3, 4.0) | 4.4 (3.8, 5.1) | 4.0 (3.2, 5.1) | |
| 1st line | 4.6 (4.2, 5.0) | 4.9 (4.5, 5.4) | 6.3 (5.7, 6.9) | 2.7 (2.0, 3.5) | 7.0 (5.4, 9.0) | 3.0 (1.9, 4.8) | – | |
| 2nd line | 3.7 (3.3, 4.2) | 4.1 (3.5, 4.8) | 5.2 (4.1, 6.5) | 2.4 (1.1, 5.0) | 3.6 (3.1, 4.1) | 5.0 (3.7, 6.6) | 3.8 (2.1, 6.8) | |
| 3rd line | 3.3 (2.3, 4.7) | 4.5 (3.4, 6.1) | 3.4 (2.0, 5.6) | 3.1 (1.3, 7.3) | 2.8 (2.3, 3.4) | 4.9 (3.9, 6.2) | 4.6 (3.1, 6.9) | |
| 4th line | 3.5 (1.7, 7.1) | 4.2 (2.0, 8.8) | 5.4 (2.6, 11.3) | 0.8 (0.1, 6.0) | 3.8 (2.9, 5.1) | 3.7 (2.7, 5.1) | 3.7 (2.4, 5.7) | |
| 5th line | 4.4 (1.4, 13.8) | 9.8 (3.7, 26.1) | 5.8 (1.4, 23.0) | 9.2 (3.5, 24.8) | 4.6 (2.7, 7.9) | 5.1 (3.2, 8.1) | 3.9 (2.3, 6.8) | |
| Unadjusted HR (95% CI) | 33 897 | Ref. | 1.1 (1.0, 1.2)* | 1.4 (1.2, 1.6)* | 0.6 (0.5, 0.8)* | 0.9 (0.8, 1.0)* | 1.0 (0.9, 1.2) | 0.9 (0.7, 1.2) |
| Adjusted HR (95% CI) | ||||||||
| All lines of therapy | 33 897 | Ref. | 1.1 (1.0, 1.3)* | 1.3 (1.1, 1.5)* | 0.8 (0.5, 1.1) | 1.0 (0.8, 1.1) | 1.2 (1.0, 1.4) | 1.2 (0.8, 1.7) |
| 1st line | 16 151 | Ref. | 1.1 (1.0, 1.3) | 1.3 (1.1, 1.6)* | 0.7 (0.5, 0.9)* | 1.6 (1.2, 2.2)* | 0.8 (0.5, 1.4) | – |
| 2nd line | 10 357 | Ref. | 1.1 (0.9, 1.4) | 1.1 (0.7, 1.6) | 0.8 (0.3, 2.3) | 1.0 (0.8, 1.2) | 1.4 (1.0, 2.0) | 1.0 (0.5, 1.9) |
| 3rd line | 4664 | Ref. | 1.3 (0.8, 2.1) | 0.8 (0.4, 1.8) | 0.9 (0.3, 3.0) | 0.7 (0.5, 1.2) | 1.4 (0.9, 2.3) | 1.5 (0.7, 2.9) |
| 4th line | 1962 | Ref. | 1.3 (0.5, 3.8) | 1.4 (0.4, 4.4) | 0.2 (0.0, 1.9) | 1.1 (0.5, 2.3) | 1.0 (0.5, 2.4) | 1.0 (0.4, 2.4) |
| 5th line | 763 | Ref. | 2.4 (0.5, 11.7) | 0.7 (0.1, 7.3) | 2.8 (0.5, 14.5) | 1.0 (0.2, 4.1) | 0.7 (0.2, 3.1) | 0.8 (0.2, 3.4) |
ABA: abatacept; ADA: adalimumab; CERT: certolizumab; ETN: etanercept; IFX: infliximab; RTX: rituximab; TCZ: tocilizumab.
P <0.05.
In the sensitivity analysis (patients from 2009 onwards), there were 973 serious infections with a crude rate of 3.6 (95% CI 3.4, 3.9). After adjustment, HRs of serious infections compared with etanercept were higher with rituximab, but not the other treatments (Supplementary Table S3, available at Rheumatology online). Stratifying by line of therapy did not change the HRs overall.
TB
There were 49 cases of active TB during the study, with an incidence rate of 8.5 (95% CI 5.7, 9.9) per 10 000 patient-years (Table 3). There were no cases observed in the tocilizumab group. The crude incidence rates of active TB were higher with adalimumab and infliximab than etanercept. There was no significant difference between the other treatment groups and etanercept, with very wide CIs. All cases occurred during the first three lines of treatment: 35 during the first line, 12 during the second line and 2 cases during the third line. Of these 49 cases, only 3 had a previous recorded history of TB infection. All patients with TB were reported to have had chest X-ray screening prior to starting their bDMARD, except three who had their treatment started before 2006. Most of the cases occurred before 2008 (Supplementary Table S4, available at Rheumatology online). It is not recorded whether patients in the BSRBR-RA had received treatment for latent TB prior to starting bDMARDs.
Table 3.
Incidence of TB per drug
| Characteristics | All | ETN | ADA | IFX | CERT | RTX | TCZ | ABA |
|---|---|---|---|---|---|---|---|---|
| Patients, n | 10 643 | 7835 | 4430 | 1614 | 5556 | 2633 | 1186 | |
| Patient-years | 19 117 | 14 501 | 8135 | 2725 | 12 009 | 4341 | 1686 | |
| Active TB, n | ||||||||
| All lines of therapy | 49 | 9 | 23 | 13 | 2 | 1 | 0 | 1 |
| 1st line | 35 | 6 | 15 | 12 | 2 | 0 | 0 | – |
| 2nd line | 12 | 3 | 8 | 1 | 0 | 0 | 0 | 0 |
| 3rd line | 2 | 0 | 0 | 0 | 0 | 1 | 0 | 1 |
| 4th line | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
| 5th line | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
| Incidence per 10 000 patient-years (95% CI) | ||||||||
| All lines of therapy | 7.5 (5.6, 9.9) | 4.5 (2.4, 8.7) | 15.2 (10.1, 22.9) | 15.2 (8.8, 26.2) | 7.2 (1.8, 28.7) | 0.8 (0.1, 5.7) | – | 5.7 (0.8, 40.5) |
| 1st line | 11.1 (8.0, 15.5) | 5.3 (2.4, 11.7) | 15.2 (9.1, 25.1) | 18.3 (10.4, 32.3) | 9.3 (2.3, 37.0) | – | – | – |
| 2nd line | 6.0 (3.4, 10.5) | 4.2 (1.3, 12.9) | 19.9 (10.0, 39.8) | 7.3 (1.0, 51.6) | – | – | – | – |
| 3rd line | 2.2 (0.6, 9.0) | – | – | – | – | 2.4 (0.3, 16.8) | – | 19.3 (2.7, 136.7) |
| 4th line | – | – | – | – | – | – | – | – |
| 5th line | – | – | – | – | – | – | – | – |
| Unadjusted HR (95% CI) | Ref. | 3.3 (1.4, 7.8) | 3.4 (1.6, 7.3) | 1.6 (0.3, 7.2) | 0.2 (0.0, 1.4) | – | 1.2 (0.1, 9.4) |
ABA: abatacept; ADA: adalimumab; CERT: certolizumab; ETN: etanercept; IFX: infliximab; RTX: rituximab; TCZ: tocilizumab.
Discussion
Many patients with RA will cycle through multiple types of targeted therapy during the course of their lifetime [18]. In this study, we provide the first comprehensive analysis of whether multiple sequencing of biologics in RA leads to any cumulative increase in infection risks over time. The findings from this large, national prospective pharmacovigilance study show no signal that cycling through biologics associates with changing infection rates. We also found that adjustment or not for confounding factors when evaluating infection risk can have important effects on the estimated risk of serious infections. Overall, there is no clear trend in the risk of infections over subsequent lines of therapy after differences in patient characteristics are considered. This study also showed that TB infection is rare but is most likely to occur with the first line of bDMARD therapy.
This increase in the risk of TB for the first line of treatment might be due to the fact that most cases happened prior the widespread use of the IFN-γ assay for screening. Another possible reason is that treatment for active TB, usually a reactivation, would happen in the first line of treatment and clear the infection for future lines of therapy. It could also be that the risk is higher with TNFis than other bDMARDs and subsequent lines are mostly non-TNFis. Recommendations for the screening of TB in the UK were first published in 2005 by the British Thoracic Society [19], with the consideration of IFN-γ immunological testing first discussed in 2006 in the National Institute for Health and Care Excellence (NICE) guidance [20] for those with positive TB skin testing (TST) or when TST might be less reliable. However, it is only since 2016 that IFN-γ immunological testing is now recommended (with or without TST) in the NICE guidance [21] prior to immunosuppressive therapy, and since 2022 that the first international rheumatologic recommendations on screening of opportunistic infections in patients with autoimmune inflammatory rheumatic diseases are available from the EULAR proposing the use of IFN-γ immunological testing over TST when available [22]. This could explain why active TB cases occurred mainly in patients starting treatment in 2008 or before, as screening may have been less effective if IFN-γ immunological testing was rare. However, there were still cases after the introduction of IFN-γ screening, which may more likely represent primary infections and not reactivation of latent TB. Information about IFN-γ testing was added only after 2016 in the BSRBR-RA register. Hence it is not possible to know how many patients underwent IFN-γ testing prior to treatment start in our study. In any case, this study suggests that when comparing the risk of TB between treatments, both the year of start of treatment and the line of therapy are likely to be major confounding factors and therefore it would be unwise to draw inferences regarding TB risk across agents.
For serious infections, we did not see associations with the line of therapy, nor were there significant differences between classes after adjusting for confounders. Although the CIs do not exclude small differences, it seems unlikely that there are clinically meaningful differences in overall serious infection rates either by line of therapy or class of biologic used when considering the drugs for which the BSRBR-RA has data available. These results are in line with a meta-analysis that did not find any pattern of increasing serious infections for a particular bDMARD [23]. It is also possible that patients with a serious infection are less likely to change their line of therapy and many may discontinue biologics, so there may be some healthy user effect with respect to infections over time [24], but confounders were adjusted for in our analysis with each subsequent line. However, it was not possible to adjust for the year of the start of treatment in our study, as cohorts of patients with each treatment in the BSRBR-RA could start and end at different times with few different treatments included in some years. However, in the sensitivity analysis, we restricted the sample to patients starting treatment after 2009 to avoid confounding by time trend. We found, as for the incidence of TB, that the incidence of serious infections was lower in more recent years. Other cohorts have also shown this decrease in serious infections with time [9, 25]. This supports year of treatment initiation as an important confounding factor when comparing serious infections between bDMARDs. We also did not have an adequate history of prior infection on all patients to include in our analysis, even though this information is associated with an elevated risk of subsequent infections [26], so the people selected for all of these drugs may be the ones that are less likely to have infections over time. However, this is probably not different among drugs and should thus not influence comparability. A potential bias in favour of TNFis in later lines could arise if patients with serious infections during their first therapies (mostly TNFis) were switched to non-TNFi therapy. However, most patients tend to stay on the same drug after serious infections [24], making this bias less likely to affect the results.
The strengths of our study are the large cohort comparing seven different bDMARDs, with some patients followed-up for almost 2 decades, and the availability of a wide range of disease- and treatment-specific variables, allowing adjustment for several confounding factors for serious infections. However, when looking at later lines of therapy or TB, there was limited statistical power to permit any definite conclusions on infection risk, with some groups experiencing very few events. Despite the low number of events, we included the incidence rate data in the analysis for these groups. CIs were provided, representing the important uncertainty around the estimates. In conclusion, we did not find major differences in the risk of serious infection among bDMARDs after taking into account potential confounding factors, including line of therapy. We did not find patterns by line of therapy for overall serious infections. However, we found that the incidence of TB was higher in the first lines of therapy and that some cases happened even after the widespread use of screening with IFN-γ assay.
Supplementary Material
Acknowledgements
The authors acknowledge the enthusiastic collaboration of all consultant rheumatologists and their specialist nurses in the UK in providing the data (visit www.bsrbr.org for a full list of contributors to the BSRBR-RA Contributors Group). The authors would like to gratefully acknowledge the support of the National Institute for Health Research (NIHR), through the Comprehensive Local Research Networks at participating centres. In addition, the authors acknowledge support from the BSR Executive, the members of the BSRBR Registers Committee and the BSRBR Project Team in London for their active role in enabling the register to undertake its tasks. The authors also acknowledge the seminal role of the BSR Clinical Affairs Committee for establishing national biological guidelines and recommendations for such a register. The authors would like to acknowledge the Centre for Epidemiology Versus Arthritis (Arthritis Research UK grant 21755), who provided the infrastructure support for the study. K.L.H. is supported by the NIHR Manchester Biomedical Research Centre. The BSRBR-RA study team has an active relationship with the UK National Rheumatoid Arthritis Society. The effectiveness and safety of multiple lines of biologic therapy are key research priorities for this organization.
Contributor Information
Kim Lauper, Centre for Epidemiology Versus Arthritis, University of Manchester, Manchester Academic Health Science Centre, Manchester, UK; Division of Rheumatology, Geneva University Hospitals and Geneva Centre for Inflammation Research, Faculty of Medicine, University of Geneva, Geneva, Switzerland.
Lianne Kearsley-Fleet, Centre for Epidemiology Versus Arthritis, University of Manchester, Manchester Academic Health Science Centre, Manchester, UK.
James B Galloway, Centre of Rheumatic Disease, King’s College London, London, UK.
Kath D Watson, Centre for Epidemiology Versus Arthritis, University of Manchester, Manchester Academic Health Science Centre, Manchester, UK.
BSRBR-RA Contributors Group, Centre for Epidemiology Versus Arthritis, University of Manchester, Manchester Academic Health Science Centre, Manchester, UK.
Kimme L Hyrich, Centre for Epidemiology Versus Arthritis, University of Manchester, Manchester Academic Health Science Centre, Manchester, UK; National Institute of Health Research Manchester Biomedical Research Centre, Manchester NHS Foundation Trust, Manchester Academic Health Science Centre, Manchester, UK.
Mark Lunt, Centre for Epidemiology Versus Arthritis, University of Manchester, Manchester Academic Health Science Centre, Manchester, UK.
Supplementary material
Supplementary material is available at Rheumatology online.
Data availability
The data that support the findings of this study are available from the BSR. Restrictions apply to the availability of these data (see https://www.rheumatology.org.uk/practice-quality/registers/requesting-registers-data).
Authors’ contributions
All the authors provided substantial contributions to the conception or design of the work and the acquisition and interpretation of data. K.L. and M.L. performed the statistical analysis and all authors contributed to the analysis interpretation. K.L. wrote the first draft. All the other authors participated in the final drafting of the work or revising it critically for important intellectual content and approved the version to be published.
Funding
The BSRBR-RA is funded by a grant from the BSR. The BSR currently receives funding from AbbVie, Amgen, Celltrion, Eli Lilly, Pfizer, Samsung Bioepis, Sanofi and Sandoz and in the past from Hospira, MSD, Roche, SOBI and UCB. This income finances a wholly separate contract between the BSR and the University of Manchester to host the BSRBR-RA. All decisions concerning study design, data capture, analyses, interpretation and publication are made autonomously of any industrial contribution.
Disclosure statement: K.L. reports honoraria from Celltrion, Viatris and Pfizer outside the submitted work. J.G. reports speaker fees/honoraria from AbbVie, Galapagos, Gilead, Janssen, Eli Lilly, Novartis, Pfizer and UCB and research funding from AbbVie, AstraZeneca, Galapagos, Gilead, Gritstone, Janssen, Moderna, Novovax and Pfizer outside the submitted work. K.L.H. reports honoraria from AbbVie and grants from Pfizer and Bristol Myers Squibb outside the submitted work. All other authors report no conflicts of interest.
References
- 1. Aletaha D, Smolen JS.. Diagnosis and management of rheumatoid arthritis: a review. JAMA 2018;320:1360–72. [DOI] [PubMed] [Google Scholar]
- 2. Fraenkel L, Bathon JM, England BR. et al. 2021 American college of rheumatology guideline for the treatment of rheumatoid arthritis. Arthritis Care Res 2021;73:924–39. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Smolen JS, Landewé RBM, Bijlsma JWJ. et al. EULAR recommendations for the management of rheumatoid arthritis with synthetic and biological disease-modifying antirheumatic drugs: 2019 update. Ann Rheum Dis 2020;79:685–99. [DOI] [PubMed] [Google Scholar]
- 4. Singh JA, Saag KG, Bridges SL. et al. ; American College of Rheumatology. 2015 American college of rheumatology guideline for the treatment of rheumatoid arthritis. Arthritis Care Res (Hoboken) 2016;68:1–25. [DOI] [PubMed] [Google Scholar]
- 5. Gómez-Reino JJ, Carmona L, Descalzo MÁ; Biobadaser Group. Risk of tuberculosis in patients treated with tumor necrosis factor antagonists due to incomplete prevention of reactivation of latent infection. Arthritis Care Res 2007;57:756–61. [DOI] [PubMed] [Google Scholar]
- 6. Kilcher G, Hummel N, Didden EM, Egger M, Reichenbach S; GetReal Work Package 4 Rheumatoid arthritis patients treated in trial and real world settings: comparison of randomized trials with registries. Rheumatol (United Kingdom) 2018;57:354–69. [DOI] [PubMed] [Google Scholar]
- 7. Courvoisier DS, Chatzidionysiou K, Mongin D. et al. The impact of seropositivity on the effectiveness of biologic anti-rheumatic agents: results from a collaboration of 16 registries. Rheumatology (Oxford) 2021;60:820–8. [DOI] [PubMed] [Google Scholar]
- 8. Galloway JB, Hyrich KL, Mercer LK. et al. ; British Society for Rheumatology Biologics Register. Anti-TNF therapy is associated with an increased risk of serious infections in patients with rheumatoid arthritis especially in the first 6 months of treatment: Updated results from the British Society for Rheumatology Biologics Register with special emph. Rheumatology 2011;50:124–31. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Smolen JS, Landewé RBM, Bijlsma JWJ. et al. Treatment benefit or survival of the fittest: What drives the time-dependent decrease in serious infection rates under TNF inhibition and what does this imply for the individual patient? Ann Rheum Dis 2020;79:685–99. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Askling J, Fored CM, Brandt L. et al. Time-dependent increase in risk of hospitalisation with infection among Swedish RA patients treated with TNF antagonists. Ann Rheum Dis 2007;66:1339–44. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Curtis JR, Xi J, Patkar N. et al. Drug-specific and time-dependent risks of bacterial infection among patients with rheumatoid arthritis who were exposed to tumor necrosis factor α antagonists. Arthritis Rheum 2007;56:4226–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Singh JA, Wells GA, Christensen R. et al. Adverse effects of biologics: A network meta-analysis and Cochrane overview. Cochrane Database Syst Rev 2011;2011:CD008794. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Van Dartel SAA, Fransen J, Kievit W. et al. Difference in the risk of serious infections in patients with rheumatoid arthritis treated with adalimumab, infliximab and etanercept: results from the Dutch Rheumatoid Arthritis Monitoring (DREAM) registry. Ann Rheum Dis 2013;72:895–900. [DOI] [PubMed] [Google Scholar]
- 14. Rutherford AI, Subesinghe S, Hyrich KL. et al. Serious infection across biologic-treated patients with rheumatoid arthritis: results from the British Society for Rheumatology Biologics Register for Rheumatoid Arthritis. Ann Rheum Dis 2018;77:annrheumdis-2017-212825. [DOI] [PubMed] [Google Scholar]
- 15. Watson K, Symmons D, Griffiths I, Silman A.. The British Society for Rheumatology biologics register. Ann Rheum Dis 2005;64(Suppl 4):iv42–3. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. England BR, Sayles H, Mikuls TR, Johnson DS, Michaud K.. Validation of the rheumatic disease comorbidity index. Arthritis Care Res 2015;67:865–72. [DOI] [PubMed] [Google Scholar]
- 17. Granger E, Sergeant JC, Lunt M.. Avoiding pitfalls when combining multiple imputation and propensity scores. Stat Med 2019;38:5120–32. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Kearsley-Fleet L, Davies R, De Cock D. et al. ; BSRBR-RA Contributors Group. Biologic refractory disease in rheumatoid arthritis: results from the British Society for Rheumatology Biologics Register for Rheumatoid Arthritis. Ann Rheum Dis 2018;77:1405–12. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Ormerod LP, Milburn HJ, Gillespie S, Ledingham J, Rampton D.. BTS recommendations for assessing risk and for managing Mycobacterium tuberculosis infection and disease in patients due to start anti-TNF-α treatment. Thorax 2005;60:800–5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. National Collaborating Centre for Chronic Conditions. Tuberculosis: clinical diagnosis and management of tuberculosis, and measures for its prevention and control. London: Royal College of Physicians, 2006. doi: 10.1016/j.jhin.2006.06.003. [DOI] [PubMed]
- 21. NICE. NICE guideline: Tuberculosis (NG33). Natl Inst Heal Care Excell 2016;178. [Google Scholar]
- 22. Fragoulis GE, Nikiphorou E, Dey M. et al. 2022 EULAR recommendations for screening and prophylaxis of chronic and opportunistic infections in adults with autoimmune inflammatory rheumatic diseases. Ann Rheum Dis 2023;82:742–53. [DOI] [PubMed] [Google Scholar]
- 23. Sepriano A, Kerschbaumer A, Smolen JS. et al. Safety of synthetic and biological DMARDs: A systematic literature review informing the 2019 update of the EULAR recommendations for the management of rheumatoid arthritis. Ann Rheum Dis 2020;79:760–70. [DOI] [PubMed] [Google Scholar]
- 24. Subesinghe S, Rutherford AI, Byng-Maddick R, Hyrich KL, Galloway JB.. Biologic prescribing decisions following serious infection: results from the British Society for Rheumatology Biologics Register — rheumatoid arthritis. Rheumatol (United Kingdom) 2018;57:2096–100. [DOI] [PubMed] [Google Scholar]
- 25. Jinno S, Lu N, Jafarzadeh SR, Dubreuil M.. Trends in hospitalizations for serious infections in patients with rheumatoid arthritis in the US between 1993 and 2013. Arthritis Care Res (Hoboken) 2018;70:652–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Zink A, Manger B, Kaufmann J. et al. Evaluation of the RABBIT risk score for serious infections. Ann Rheum Dis 2014;73:1673–6. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
The data that support the findings of this study are available from the BSR. Restrictions apply to the availability of these data (see https://www.rheumatology.org.uk/practice-quality/registers/requesting-registers-data).