Risk of tuberculosis, serious infection and lymphoma with disease-modifying biologic drugs in rheumatoid arthritis patients in Taiwan

Transcription

1 International Journal of Rheumatic Diseases 2014; 17 (Suppl. 3): 9 19 ORIGINAL ARTICLE Risk of tuberculosis, serious infection and lymphoma with disease-modifying biologic drugs in rheumatoid arthritis patients in Taiwan Ying-Ming CHIU, 1,2 Hui-Chu LANG, 3 Hsiao-Yi LIN, 4 Ming-Ta YANG, 5,6 Chi-Hui FANG, 7,8 Ya-Wen YANG, 7,8 Vernon F. SCHABERT 9 and Boxiong TANG 7,8 1 Division of Allergy Immunology & Rheumatology, Changhua Christian Hospital, Changhua, Taiwan, 2 Department of Nursing, College of Medicine & Nursing, HungKuang University, Taichung, 3 Institute of Hospital and Health Care Administration, National Yang-Ming University, Taipei, Taiwan, 4 Division of Allergy, Immunology and Rheumatology, Department of Medicine, Veterans General Hospital, Taipei, Taiwan, 5 IMS Health, Taipei, Taiwan, 6 IMS Health, New York, New York, USA, 7 Pfizer Inc., Taipei, Taiwan, 8 Pfizer Inc., New York, New York, USA, and 9 IMS Health, Woodland Hills, California, USA Abstract Aim: To determine the risk of adverse events in rheumatoid arthritis (RA) patients treated with biological disease-modifying anti-rheumatic drugs (bdmard) versus traditional DMARDs (tdmard). Method: This retrospective study used Taiwan s National Health Insurance Research Database to capture data for adult patients diagnosed with RA between 1 January 1999 and 31 December 2009 and treated with tdmard or bdmard. The endpoints were patients with cases of an inpatient serious bacterial infection (SBI), diagnosis of tuberculosis (TB) or lymphoma. Within the bdmard cohort, individual bdmards with adequate data were also compared (adalimumab and etanercept). Propensity-score matching was used to adjust for significant (P 0.05) patient characteristics. Incidence rate ratios (IRR) of SBI/TB/lymphoma cases versus non-cases were adjusted for exposure time (rate per patient-years) and 95% confidence intervals were constructed to assess whether IRRs differed from 1.0. Results: Of potential patients, 7888 tdmard, 3459 bdmard (including 1492 etanercept and 746 adalimumab) patients were matched for analysis. A total of 2150 cases were identified and of these 1711 were SBI, 406 as TB and 33 as lymphoma. For all cases except SBI, the IRR was higher for bdmard versus tdmard (SBI 1.04 [ ]; TB 2.67 [ ]; lymphoma 3.24 [ ]). Excepting lymphoma, IRR was higher for adalimumab versus etanercept (SBI 1.83 [ ]; TB 2.35 [ ]; lymphoma 1.49 [ ]). Conclusions: There was a higher risk for specified infections and lymphoma with bdmard versus tdmard and adalimumab versus etanercept. Key words: epidemiology, health services and health care economics, rheumatoid arthritis. INTRODUCTION Correspondence: Professor Hui-Chu Lang, Institute of Hospital and Health Care Administration, National Yang-Ming University, No.155, Sec. 2, Li-Nong St., Taipei 112, Taiwan. hclang@ym.edu.tw Disease-modifying antirheumatic drugs (DMARDs) are widely used as first-line treatment for the management of moderate to severe rheumatoid arthritis (RA). The primary goal of RA pharmacotherapy is to improve clinical symptoms and halt or deter progression to structural joint damage. 1 Treatment guidelines for RA patients with active disease recommend a traditional DMARD (tdmard), such as methotrexate, as a first step. 2 4 In the absence of adequate response with one or more tdmards, and depending on 2014 Asia Pacific League of Associations for Rheumatology and Wiley Publishing Asia Pty Ltd

2 Y. M. Chiu et al. prognostic factors, the introduction of a biologic anti-tumor necrosis factor (anti-tnf) agent, or biological DMARDs (bdmard), is typically the next recommended treatment option. 2 4 The bdmards target TNF-a, a key proinflammatory cytokine, and an important target due to its role in both joint inflammation and bone mass degradation. The introduction of these drugs has signaled a major advance in RA therapy. 5,6 The efficacy of three anti-tnf agents, adalimumab, etanercept and infliximab, has been well established in randomized controlled trials. 5,7,8 Although direct comparisons of available anti-tnf agents in randomized controlled settings are not available, improvements in symptom control appear to be similar across agents. 5,7,8 Patients with RA are known to be at high risk of infection 9 and lymphoma. 10 It is likely that this results from multiple factors, including the disease itself (through altered immunologic function), as well as due to comorbidities and pharmacotherapy. 9,11 Although it is hypothesized that RA itself is a risk factor for increased infection, it is currently unknown how much RA may increase infection risk independent of related factors, such as treatment with DMARDs. One study by Smitten et al. 12 adjusted for confounders including comorbid conditions and prescription medication use and found an elevated hazard ratio for infection requiring hospitalization among patients with RA (2.03; 95% CI: ). Both the tdmards and anti-tnf bdmards interrupt RA pathophysiology by targeting the inflammatory process. 13 Anti-TNF agents target TNF, a key proinflammatory cytokine, by direct interference with receptor binding. 1 However, TNF has a beneficial role in the immune system and in tumor surveillance. 6 Therefore, interruption of the inflammatory cascade with anti- TNFs may also suppress immunologic response. Following the 1998 introduction of two anti-tnf agents (infliximab and etanercept), reports from the US Food and Drug Administration s Adverse Event Reporting System suggested a higher incidence of tuberculosis (TB) 14 and lymphoma 10 in patients treated with these drugs. The close proximity of these events to anti-tnf therapy initiation, and the known immunosuppressive actions of anti-tnf agents, suggested a potential causal link. However, available data were limited and inadequate to make a clear association. The development of registries in several countries for patients treated with biologic agents, as well as the publication of a number of claims-based studies, has provided a larger database and longer timeframe from which to evaluate these safety endpoints. Despite differences in methodology, registry and health claims database studies conducted in the US and Western Europe have found a significantly higher risk for serious bacterial infection (SBI) with bdmards compared with tdmards. 6,15 17 Estimates of risk have been highly variable, ranging from a 20% to a 400% increase, and appear to be greatest during the first 6 months of treatment. 6,15,16 Compared with patients who have not received anti-tnf treatment, a higher incidence of TB has also been reported with anti-tnf agents in Korea, Spain, Sweden and the US The potential for negative safety endpoints among anti-tnf agents has also been explored. As adalimumab, etanercept and infliximab were the first anti-tnf agents to be approved for RA treatment, the largest body of evidence has been generated for these agents. Studies have compared individual agents, as well as monoclonal antibody therapy as a group (adalimumab, infliximab) versus a soluble receptor fusion protein (etanercept). The mode of TNF neutralization differs between the monoclonal antibodies and the soluble receptor fusion protein, and a biologic basis has been noted for the risk of reactivation of latent TB with monoclonal antibodies. 22 In a French registry study, a higher risk for non-tb infections was associated with adalimumab and infliximab relative to etanercept treatment. Odds for infection were times greater for the monoclonal antibodies versus etanercept. 23 Use of steroids was also implicated as a risk factor for infection. However, other studies based on UK 16,24 and Italian 11 registry data have not distinguished a significant difference between these agents. A higher rate of TB with infliximab and adalimumab relative to etanercept was reported in registry studies conducted in Great Britain 25 and France. 26,27 Greater age and being born in a TB-endemic area posed a higher risk for patients treated with adalimumab or infliximab versus etanercept. 27 A higher risk for lymphoma has also been reported for patients treated with adalimumab or infliximab compared to etanercept in a French study. 27 However, in a US study, no significant differences in lymphoma rates were noted between anti-tnf agents. 28 However, all of these adverse events are relatively rare, and most studies to date have been based on data captured during a 6-month to 5-year interval. Estimates of risk have varied considerably among studies, and not all studies have reported multiple safety endpoints. The objective of the current study was to evaluate the 10 International Journal of Rheumatic Diseases 2014; 17 (Suppl. 3): 9 19

3 Risk of adverse events in RA incidence rate of SBI, TB and lymphoma over a 10-year period using the National Health Insurance Research Database (NHIRD) in Taiwan. Studying these outcomes in a TB endemic area such as Taiwan 29 makes it more likely to capture an association, compared with data obtained from a low-tb prevalence area (where events may be too rare to reach statistical significance). Specifically, the incidence of these events was compared between tdmards and bdmards, and between individual bdmards. It was hypothesized a higher incidence of SBI, TB and lymphoma would be observed in RA patients using bdmards compared with tdmards. It was additionally hypothesized that, among the bdm- ARDs, etanercept would be associated with the lowest number of events. METHODS This retrospective, longitudinal study used data collected by the Bureau of National Health Insurance (BNHI) of Taiwan, a single government payer that covers 99.5% of individuals in Taiwan. 30 The NHIRD is a longitudinal database of BNHI medical claims that houses up to 15 years of electronic medical records data for more than 23 million patients. These data include ICD-9-CM medical diagnoses linked to physician prescriptions, as well as demographic information. Data accessed for this study were collected between January 1, 1999 and December 31, Participants Patients included in this study were required to have more than one diagnosis with RA (ICD-9-CM 714.0x) during the study period, to be 18 years of age on the date of first diagnosis, and to hold a catastrophic illness card. RA is one of 30 illnesses currently covered by catastrophic illness cards, which, once issued, are valid for life. To obtain a catastrophic illness card due to RA, an adult patient must be diagnosed with RA two or more times, each time meeting the 1987 American College of Rheumatology diagnostic criteria. 31 Additionally, to be included, patients must have been prescribed a tdmard or bdmard at least once during the study period. Qualifying tdmards included azathioprine, cyclosporine, gold sodium thiomalate, hydroxychloroquine, leflunomide, methotrexate, minocycline, penicillamine D or sulfasalazine. Qualifying bdmards included etanercept, adalimumab or rituximab, as these were the three bdmards available in Taiwan during the study interval. It should be noted that these medications were not available during the entirety of the study period; etanercept and adalimumab were approved for reimbursement for RA treatment in March 2003 and September 2004, respectively. Rituximab, now approved as a second-line treatment for RA, was not approved for reimbursement in Taiwan for RA until November BHNI treatment provisions allow a patient to receive bdmard treatment for RA only after having failed at least two tdmards with a 6-month interval for each therapy. All patients who received etanercept, adalimumab or rituximab as first-, second- or third-line treatments were included in the analysis that compared tdmard and bdmard outcomes. However, in the analysis, comparing the bdmards outcomes were included only if they occurred during use of the first prescribed bdmard (i.e., before drug switching or the end of the study). Subsequent bdmard use was excluded from the analysis. Because it was anticipated that the rituximab sample size would be inadequate for bdmard-specific analysis, rituximab was not included for comparison in this study segment. Also excluded from the study were patients diagnosed with RA only once during the study interval, patients < 18 years of age when first diagnosed with RA, and patients first diagnosed with RA after July 1, The study also excluded patients who did not hold an RA catastrophic illness card, who were never prescribed a tdmard or bdmard, and who experienced an adverse event before ever receiving treatment with a tdmard or bdmard. Propensity score matching Patients were divided into cohorts based on the index treatment type administered (bdmard or tdmard). As tdmards have been used for RA treatment longer than bdmards, patients in the bdmard cohort were matched at a 1 : 2 ratio with patients in the tdmard cohort, based on propensity score. Additionally, a higher proportion of patients who received bdmards as first-line treatment received etanercept as compared to adalimumab; therefore, patients in the etanercept cohort were propensity score-matched with adalimumab patients using a 2 : 1 ratio. Propensity score matching was performed using logistic regression analyses, with the index treatment as the dependent variable and all measured baseline characteristics as independent variables. Covariates included demographics, indicators of disease severity and comorbidities; those that reached significance at the P 0.05 level were used to create the propensity score. For International Journal of Rheumatic Diseases 2014; 17 (Suppl. 3):

4 Y. M. Chiu et al. bdmard compared to tdmard use, propensity score covariates included age, chronic obstructive pulmonary disease (COPD)/asthma, diabetes, disease duration, number of tdmards, sex and steroid exposure. For within-bdmard use comparisons (etanercept vs. adalimumab), propensity score covariates included disease duration, number of tdmards and steroid exposure. Baseline variables and case identification Baseline characteristics included in this study were age (standardized to the end of the study period), sex, duration of disease (from first observed RA diagnosis until the end of the study period), number of different tdm- ARDs prescribed, patient exposure to steroids (including betamethasone, cortisone, dexamethasone, fluocortolone, hydrocortisone, methylprednisolone, paramethasone, prednisolone, prednisone, prednylidene and triamcinolone), and comorbidities present in the 180- day period prior to initial RA diagnosis date, defined by ICD-9-CM codes. Comorbidities included diabetes mellitus, excluding type 1 (250.xx, excluding 250.x1 and 250.x3), COPD/asthma (493.xx), hypertension (401.xx) and hyperlipidemia (272.0, 272.1, and 272.4). Cases were identified as any patient with the presence of SBI requiring hospitalization or one or more ICD-9- CM codes for TB (010.xx 018.xx) or lymphoma (202.8) during the interval between the first RA diagnosis and study end. SBI ICD-9-CM codes included those for encephalitis (323.x, 054.3), endocarditis (421.x), meningitis (320.x, 049.x), osteomyelitis (730.0x, 730.1x, 730.2x), pneumonia (481.x, 482.x), pyelonephritis (590.x), septic arthritis (711.0x, excluding ), and septicemia or bacteremia (038.x, 790.7). Treatment exposure and incidence rate Exposure to DMARD treatment was calculated in, starting on the date of first RA diagnosis. For case patients, this included the number of years between the initiation date for tdmard or bdmard and the occurrence of the safety endpoint (SBI, TB or lymphoma). For non-case patients, this included the number of years between the first tdmard or bdm- ARD initiation and the end of the observation period (31 December 2009). Only adverse events that occurred during the period of drug use or within 90 days following the last prescription administered were considered valid. In cases where multiple events occurred for one patient, all events were recorded. The incidence rate and incidence rate ratio (IRR) were computed for the propensity score-matched cohorts. The exposure-adjusted incidence was determined by dividing the number of cases by the total exposure time of cases and non-cases. The IRR for treatment outcomes with bdmards compared to tdmards was calculated by dividing the average number of cases per total in the bdmard cohort by the average number of cases per total in the tdmard cohort. The IRR for treatment outcomes between etanercept versus adalimumab was calculated by dividing the average number of cases per total among etanercept users by the average number of cases per total among adalimumab users. Exposure-adjusted incidence rates were calculated separately for SBI, TB and lymphoma. All analyses were performed using SAS 9.1 (SAS Institute Inc., Cary, NC, USA). Algorithm for event attribution SBI, TB and lymphoma cases were ascribed to the tdm- ARD or bdmard cohort using a predetermined algorithm. Events were ascribed to tdmard if the event occurred while the patient was receiving a tdmard (i.e., before receiving bdmard) until 31 December 2009, or to bdmard if the event occurred while the patient was receiving their first bdmard until 31 December Events that occurred while the patient was receiving etanercept or adalimumab as first-line therapy (until drug switching or until 31 December 2009) were ascribed to the medication received (etanercept or adalimumab) at the time of the event. Events that occurred on the first date of a new drug prescription were ascribed to a previous drug exposure. The analysis included only patients with a first event that occurred during treatment with a tdmard or bdmard; patients who experienced events before receiving any DMARDs were excluded. Statistical analyses Cohorts matched by propensity scores were analyzed using the Pearson chi-square test for categorical variables and Wilcoxon rank-sum test for continuous and count variables. Analyses were performed to obtain IRRs for bdmard as compared to tdmard outcomes, as well as etanercept versus adalimumab outcomes. RESULTS Patients A total of RA patients met the inclusion criteria (Fig. 1). Among the patients, 4033 had been treated with bdmards and had been treated only with tdmards. Before propensity score matching, baseline 12 International Journal of Rheumatic Diseases 2014; 17 (Suppl. 3): 9 19

5 Risk of adverse events in RA The baseline characteristics after 1 : 2 (bdm- ARD : tdmard) propensity score-matching of patients treated with either etanercept or adalimumab as firstline therapy are shown in Table 2. Most patients were female (> 80%), had been treated with approximately four different tdmards and had similar comorbidity profiles. Figure 1 Patient selection criteria. characteristics differed significantly between patients in the bdmard and tdmard cohorts. Patients in the bdmard cohort were younger (57.8 vs years), more likely to be women (82.3% vs. 78.8%), had a longer average RA duration (8.0 vs. 7.5 years), had been treated with more tdmards (4.2 vs. 2.9), and had higher rates of steroid exposure (83.3% vs. 72.4%) (P < for all). Additionally, patients in the bdm- ARD cohort had a lower incidence of comorbid diabetes (P < ) and hypertension (P < 0.05) compared with the tdmard cohort. Patient baseline characteristics after 1 : 2 (bdm- ARD : tdmard) propensity score-matching are shown in Table 1. The measured baseline characteristics between matched bdmard (n = 4033) and tdmard (n = 8066) patients showed no differences, with the exception of a statistically significant, slightly higher number of tdmards ever used in the bdmard cohort versus the tdmard cohort (4.2 vs. 4.1, P < 0.05). However, this difference represents a standardized effect (d) of 0.08 standard deviations, less than half of the conventional threshold d = 0.20 for a small effect. 32 Incidence of adverse events: bdmard versus tdmard The number of cases and the adverse event incidence rate per for the bdmard and tdmard cohorts are shown in Table 3. An adverse event endpoint occurred in 2721 patients. After applying the event attribution algorithm, 1972 out of (7888 tdmard, 3459 bdmard) patients had events that occurred in time periods that were eligible for analysis. SBIs were the most common adverse events across all cohorts (1711 events), followed by TB (406) and lymphoma (33). A nominally higher incidence rate per occurred in the bdmard versus tdmard cohort for each adverse endpoint: SBI, 3068 ( ) versus 2956 ( ); TB, 1458 ( ) versus 546 ( ); lymphoma, 133 (64 245) versus 41 (26 62). The IRR estimates showed elevated risk among bdmard users of 2.67 ( ) for TB and 3.24 ( ) for lymphoma. The IRR for SBI did not reach significance (1.04, 95% CI ). Incidence of adverse events: adalimumab versus etanercept A total of 164 patients out of the 2238 matched etanercept and adalimumab patients had at least one adverse event. All events were eligible for analysis after applying the attribution algorithm. The number of events, incidence rate per , and the IRR comparisons of adalimumab versus etanercept are shown in Table 4. Among the matched adalimumab and etanercept patients, there were a total of 116 SBIs, 58 TB events, and four lymphoma events. For SBI, the incidence rate for the adalimumab group was 4967 ( ), higher than incidence rate of 2708 ( ) in the etanercept group. The SBI IRR for adalimumabtreated patients was 1.83 ( ). The incidence rate of TB for the adalimumab group was 2888 ( ); this was higher than the incidence rate of 1228 ( ) in the etanercept group. The IRR for TB in the adalimumab group was 2.35 International Journal of Rheumatic Diseases 2014; 17 (Suppl. 3):

6 Y. M. Chiu et al. Table 1 Baseline Characteristics of bdmard and tdmard cohorts after propensity-score matching bdmard (n = 4033) tdmard (n = 8066) P-value Age (years) Mean Gender (n, %) Female % % Male % % Disease duration (years) Mean Number of tdmards Mean Steroid exposure (n, %) No % % Yes % % Comorbidities (n, %) Diabetes % % COPD/asthma % % Hypertension % % Hyperlipidemia % % bdmard, biological disease-modifying anti-rheumatic drug; COPD, chronic obstructive pulmonary disease, tdmard, traditional disease-modifying anti-rheumatic drug. Patients in bdmard cohort were ever treated with slightly higher number of tdmards compared to patients in tdmard cohort. Table 2 Baseline characteristics of etanercept and adilmumab groups after propensity-score matching First anti-tnf drug Etanercept (n = 1492) Adalimumab (n = 746) P-value Age (years) Mean Gender (n, %) Female % % Male % % Disease duration (years) Mean Number of tdmards Mean Steroid exposure (n, %) No % % Yes % % Comorbidities (n, %) Diabetes % % COPD/asthma 9 0.6% 6 0.8% Hypertension % % Hyperlipidemia % 5 0.7% TNF, tumor necrosis factor; bdmard, biological disease-modifying anti-rheumatic drug; COPD, chronic obstructive pulmonary disease, tdmard, traditional disease-modifying anti-rheumatic drug. ( ). The incidence rate of lymphoma was 144 (4 800) and 96 (20 280) in the adlimumbab and etanercept groups, respectively. The IRR for lymphoma in the adalimumab group was 1.49 ( ), suggesting no difference in risk compared with etanercept. DISCUSSION Results from this study showed a higher risk for tuberculosis and lymphoma in patients receiving bdmards compared with patients receiving tdmards, but not for SBI. Among bdmard patients, those receiving 14 International Journal of Rheumatic Diseases 2014; 17 (Suppl. 3): 9 19

7 Risk of adverse events in RA Table 3 Risk of TB, lymphoma and SBI for bdmard and tdmard groups tdmard (n = 7888) bdmard (n = 3459) TB Number of event Patient years Rate/ ( ) 1458 ( ) IRR Referent 2.67 ( ) Lymphoma Number of event Patient-years Rate/ (26 62) 133 (64 245) IRR Referent 3.24 ( ) SBI Number of event Patient-years Rate/ ( ) 3068 ( ) IRR Referent 1.04 ( ) bdmard, biological disease-modifying anti-rheumatic drug; CI, confidence interval; COPD, chronic obstructive pulmonary disease; IRR, incidence rate ratio; SBI, serious bacterial infection; TB, tuberculosis; tdmard, traditional disease-modifying anti-rheumatic drug. Patients in bdmard cohort showed significantly higher risks of TB and lymphoma as compared to patients in tdmard cohort. adalimumab experienced higher risk for tuberculosis and serious bacterial infections than etanercept patients, but not for lymphoma. These results partly support existing literature indicating an increased risk of adverse events with the use of bdmards compared to tdmards, 6,15 17 and they provide evidence of elevated risk for patients who use adalimumab versus etanercept among bdmards. Other studies have similarly reported higher bdm- ARD risks for TB infection, but they have also reported higher risk for SBI, which was not confirmed here. These findings also support an association between bdmard use and lymphoma risk previously supported largely by adverse event reports. Although the relative risk for TB and lymphoma events was higher than for SBI, these events were uncommon. Only 406 TB events occurred in of exposure, and 33 lymphoma events occurred in of exposure. The increased lymphoma risk in bdmard compared to tdmard cohorts observed in this study could also Table 4 Risk of TB, lymphoma and SBI for etanercept and adalimumab groups Etanercept (n = 1492) Adalimumab (n = 746) TB Patient years Case Rate/ ( ) 2888 ( ) IRR Referent 2.35 ( ) Lymphoma Patient-years Case 3 1 Rate/ (20 280) 144 (4 800) IRR Referent 1.49 ( ) SBI Patient years Case Rate/ ( ) 4967 ( ) IRR Referent 1.83 ( ) CI, confidence interval; IRR, incidence rate ratio; SBI, serious bacterial infection; TB, tuberculosis. Patients on adalimumab were associated with a 2-fold increase in risk of TB, and a 1.8-fold increase in risk of SBI, compared to etanercept. be the result of residual unbalanced disease activity between the two cohorts, despite propensity score matching. Several studies have found a strong relationship between RA inflammatory activity and lymphoma which would account for the increase in risk for lymphoma found in this study. Specifically, the observed higher risk of lymphoma could be the result of common genetic risk factors for RA malignancy, predisposition and severity. 33,37 As an example, the human leukocyte antigen (HLA)-DRB1 shared-epitope genotype is affiliated with death related to malignancy in RA. 38 Additionally, there is a level of skepticism concerning the potential impact of bdmard or other treatments on site-specific risk of cancer in RA 33,36 which further bolsters the theory of the influence of residual disease activity on increased lymphoma risk in these patients. As noted, previous studies have shown increased bacterial infection risk associated with bdmard use. 15 However, other studies have applied a broader definition of SBI, most commonly as any infection that led to hospitalization or death, or required intravenous International Journal of Rheumatic Diseases 2014; 17 (Suppl. 3):

8 Y. M. Chiu et al. (i.v.) antibiotics. 6,15,16,24 Other research has recorded any infection that fell under general adverse event guidelines, 17 while other studies have evaluated only TB events. 18,25 Additionally, this study followed a population for a total of 10 years, capturing data on all patients who initiated DMARD use in that time period from the time of treatment initiation. To increase the precision of this study, results were based on person years and adjusted to account for the time patients were persistent on DMARDs. Additionally, propensity score matching was used to help determine the extent of events attributable to medication. Patients receiving bdmards showed several differences in baseline characteristics than did patients on tdm- ARDs, which might have confounded infection risk estimates without the use of propensity matching as performed in this study. The current study also found that adalimumab use was associated with a 83% increase in SBI risk (IRR: 1.83, 95% CI: ) and more than a two-fold increase in TB (IRR: 2.35, 95% CI: ) relative to etanercept. Although lymphoma risk was nominally higher in adalimumab versus etanercept (144 vs. 96 cases per person years, respectively), results did not reach statistical significance. However, the cohort only contained four total lymphoma events (three in 3132 person years for etanercept, one in 697 person years for adalimumab) and was not powered to detect differences with such low incidence rates. The available literature evaluating risk associated with adalimumab versus etanercept varies. Some studies examining the effects of various bdmards have found a lower safety risk with etanercept versus other bdm- ARDs. 23,25 27 However, other studies comparing individual bdmard outcomes did not find statistically significant differences between etanercept and adalimumab for SBIs. 16,24 These latter studies had a much broader definition of events, including any illness that led to hospitalization or death, or that required i.v. antibiotics. Two other studies, using data from the British Society for Rheumatology Biologics Register (BSRBR), have shown a 4- to 14-fold increased risk of TB infection for RA patients receiving adalimumab as compared to etanercept. 25,26 Subjects in both of these studies had much lower TB incidence than the current study (each data set contained only 40 cases of TB). This may be because the studies were conducted in France and the UK, both countries with low TB prevalence. 39,40 However, the BSRBR collects its information via semiannual questionnaires administered to patients and providers, as well as from the British death registry. The current study used NHIRD data, which includes all registration files and claims data for reimbursement from patients in Taiwan. Therefore, this research did not rely on information return from individual patients and providers and may represent a more comprehensive dataset for estimation of TB incidence. 41 A French study found an increased risk of lymphoma with use of adalimumab, as compared to etanercept, with standardized incidence ratios (SIRs) of 4.1 and 0.9, respectively. The SIRs compared the risk of lymphoma in patients who received bdmards to the general French population over a 3-year period. 27 However, this study included all patients who received bdmards and was not limited to patients with RA. Although infection risk may be increased with bdm- ARD use, the consequence of managing increased infections must be balanced against the benefit obtained from bdmard therapy. Clinicians should be aware of risk potential and take relevant precautions when prescribing bdmard treatment to certain patients. The current results also indicate the importance of careful patient observation and the need to institute appropriate, timely management in case infection occurs. 13 Additional research is necessary to assess the cost of increased safety risk with adalimumab or other monoclonal antibody anti-tnf agents, as compared to etanercept. LIMITATIONS This was an observational study based on claims data, leading to potential confounds from the lack of control over treatment selection. Participants were matched using propensity scoring to reduce the impact of such confounds, but unmeasured patient characteristics may still have influenced results. The study period ended in 2009, which necessitated the exclusion of biologics not approved in Taiwan market at the time or thosenewer to the market (infliximab, abatacept). Furthermore, as information on the effectiveness of RA treatments cannot be readily obtained from health insurance claims data, no data on treatment effectiveness were available for analysis. Therefore, this study s outcomes show adverse events independent of treatment effectiveness and patient satisfaction. However, prior literature suggests similar efficacy for all anti-tnf agents. 6 8 Although there seems to be a naturally elevated risk of infection with RA, the extent of risk attributable to RA itself versus risk caused by comorbidities, medications or other potential contributing factors is unknown 16 International Journal of Rheumatic Diseases 2014; 17 (Suppl. 3): 9 19

9 Risk of adverse events in RA and cannot be explained by these data. A study on predictors of infection in RA patients found a variety of factors that increased risk for infection requiring hospitalization, including the presence of comorbidities, treatment with corticosteroids, age, and disease severity. 42 It has been recommended that other potential explanations for increased infection risk in RA patients should be investigated, such as increased infection rates resulting from complications due to joint damage, increased surgeries or skin defects related to RA. 42 However, it remains noteworthy that RA severity is associated with increased infection, despite the lack of evidence to prove a causal link between RA and infection. Another caution is that the interpretation of these outcomes may not be generalizable to all regions, because areas with higher rates of TB infection are likely to have increased TB rates due to the risk of infection endemic to the region. These data represent TB risk in RA patients receiving DMARDs in Taiwan, which is an endemic area. 29 Although the relative risk for TB infection based on treatment exposure should in theory be constant across regions regardless of local risk, it is challenging to precisely estimate relative risk in settings where baseline risk is low. In such cases, very small differences in observed cases will have an exaggerated influence on the estimated relative risk. From 2004 to 2008, TB incidence in Taiwan ranged from 62 to 74 per people; in comparison, in 2010, TB incidence was 13.6 per people in the UK and 3.6 per in the US. 41,43 It is therefore unlikely that these outcomes could be generalized to low-incidence regions such as the UK and the US. Likewise, the somewhat high incidence of TB observed in this study may be explained by the classification of infections (including TB) based only on a single diagnosis or hospitalization. Using this approach, some patients may have been diagnosed with TB without ever receiving confirmation and/or treatment, and the actual study population with TB may be lower than estimated by this study. CONCLUSIONS This research confirms that treatment with bdmards in patients with RA is associated with a higher risk of TB, as well as with risk for incident lymphoma, compared with tdmards. Additionally, risk of adverse events (in particular, SBI and TB) vary based on bdm- ARD type, with a higher risk associated with the monoclonal antibody therapy adalimumab, as compared to etanercept, a soluble receptor fusion protein. This study expands the evidence base for differential risk of infection posed by specific bdmards. ACKNOWLEDGEMENTS This study was based in part on data from the Taiwan National Health Insurance Research Database provided by the Bureau of National Health Insurance, Department of Health and managed by National Health Research Institutes. The interpretation and conclusions contained herein do not represent those of the Bureau of National Health Insurance, Department of Health or National Health Research Institutes. CONFLICT OF INTEREST Ming-Ta Yang is an employee of IMS Health who was a paid consultant to Pfizer in connection with the development of this manuscript. Vernon F. Schabert was an employee of IMS Health who was a paid consultant to Pfizer during the development of the study and manuscript. This study was funded by Pfizer Inc. Ya-Wen Yang is an employee of Pfizer Taiwan. Chi-Hui Fang and Boxiong Tang were employees of Pfizer during the development of the study and manuscript. REFERENCES 1 Smolen JS, Aletaha D, Koeller M, Weisman MH, Emery P (2007) New therapies for treatment of rheumatoid arthritis. Lancet 370, National Institute for Health and Clinical Excellence (2010) Rheumatoid Arthritis. (Clinical Guideline CG79). Available from URL: 3 Saag KG, Teng GG, Patkar NM et al. (2008) American College of Rheumatology 2008 recommendations for the use of nonbiologic and biologic disease-modifying antirheumatic drugs in rheumatoid arthritis. Arthritis Rheum 59, Smolen JS, Landewe R, Breedveld FC et al. (2010) EU- LAR recommendations for the management of rheumatoid arthritis with synthetic and biological diseasemodifying antirheumatic drugs. Ann Rheum Dis 69, Keystone EC, Kavanaugh AF, Sharp JT et al. (2004) Radiographic, clinical, and functional outcomes of treatment with adalimumab (a human anti-tumor necrosis factor monoclonal antibody) in patients with active rheumatoid arthritis receiving concomitant methotrexate therapy: a randomized, placebo-controlled, 52-week trial. Arthritis Rheum 50, International Journal of Rheumatic Diseases 2014; 17 (Suppl. 3):

10 Y. M. Chiu et al. 6 Dixon WG, Symmons DP, Lunt M et al. (2007) Serious infection following anti-tumor necrosis factor alpha therapy in patients with rheumatoid arthritis: lessons from interpreting data from observational studies. Arthritis Rheum 56, Lipsky PE, van der Heijde DM, St Clair EW et al. (2000) Infliximab and methotrexate in the treatment of rheumatoid arthritis. Anti-Tumor Necrosis Factor Trial in Rheumatoid Arthritis with Concomitant Therapy Study Group. N Engl J Med 343, Weinblatt ME, Kremer JM, Bankhurst AD et al. (1999) A Trial of Etanercept, a recombinant tumor necrosis factor receptor:fc fusion protein, in patients with rheumatoid arthritis receiving methotrexate. N Engl J Med 340, Doran MF, Crowson CS, Pond GR, O Fallon WM, Gabriel SE (2002) Frequency of infection in patients with rheumatoid arthritis compared with controls: a population-based study. Arthritis Rheum 46, Brown SL, Greene MH, Gershon SK, Edwards ET, Braun MM (2002) Tumor necrosis factor antagonist therapy and lymphoma development: twenty-six cases reported to the Food and Drug Administration. Arthritis Rheum 46, Favalli EG, Desiati F, Atzeni F et al. (2009) Serious infections during anti-tnfalpha treatment in rheumatoid arthritis patients. Autoimmun Rev 8, Smitten AL, Choi HK, Hochberg MC et al. (2008) The risk of hospitalized infection in patients with rheumatoid arthritis. J Rheumatol 35, Tugwell P, Singh JA, Wells GA (2011) Biologicals for rheumatoid arthritis. BMJ 343, d Keane J, Gershon S, Wise RP et al. (2001) Tuberculosis associated with infliximab, a tumor necrosis factor alphaneutralizing agent. N Engl J Med 345, Curtis JR, Patkar N, Xie A et al. (2007) Risk of serious bacterial infections among rheumatoid arthritis patients exposed to tumor necrosis factor alpha antagonists. Arthritis Rheum 56, Galloway JB, Hyrich KL, Mercer LK et al. (2011) 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 emphasis on risks in the elderly. Rheumatology (Oxford) 50 (1), Listing J, Strangfeld A, Kary S et al. (2005) Infections in patients with rheumatoid arthritis treated with biologic agents. Arthritis Rheum 52, Askling J, Fored CM, Brandt L et al. (2005) Risk and case characteristics of tuberculosis in rheumatoid arthritis associated with tumor necrosis factor antagonists in Sweden. Arthritis Rheum 52, Gomez-Reino JJ, Carmona L, Valverde VR, Mola EM, Montero MD, BIOBADASER Group (2003) Treatment of rheumatoid arthritis with tumor necrosis factor inhibitors may predispose to significant increase in tuberculosis risk: a multicenter active-surveillance report. Arthritis Rheum 48, Seong SS, Choi CB, Woo JH et al. (2007) Incidence of tuberculosis in Korean patients with rheumatoid arthritis (RA): effects of RA itself and of tumor necrosis factor blockers. J Rheumatol 34, Wolfe F, Michaud K, Anderson J, Urbansky K (2004) Tuberculosis infection in patients with rheumatoid arthritis and the effect of infliximab therapy. Arthritis Rheum 50, Wallis RS (2008) Tumour necrosis factor antagonists: structure, function, and tuberculosis risks. Lancet Infect Dis 8, Salmon-Ceron D, Tubach F, Lortholary O et al. (2011) Drug-specific risk of non-tuberculosis opportunistic infections in patients receiving anti-tnf therapy reported to the 3-year prospective French RATIO registry. Ann Rheum Dis 70, Dixon WG, Watson K, Lunt M et al. (2006) Rates of serious infection, including site-specific and bacterial intracellular infection, in rheumatoid arthritis patients receiving anti-tumor necrosis factor therapy: results from the British Society for Rheumatology Biologics Register. Arthritis Rheum 54, Dixon WG, Hyrich KL, Watson KD et al. (2010) Drug-specific risk of tuberculosis in patients with rheumatoid arthritis treated with anti-tnf therapy: results from the British Society for Rheumatology Biologics Register (BSRBR). Ann Rheum Dis 69, Tubach F, Salmon D, Ravaud P et al. (2009) Risk of tuberculosis is higher with anti-tumor necrosis factor monoclonal antibody therapy than with soluble tumor necrosis factor receptor therapy: the three-year prospective French Research Axed on Tolerance of Biotherapies registry. Arthritis Rheum 60, Mariette X, Gottenberg JE, Ravaud P, Combe B (2011) Registries in rheumatoid arthritis and autoimmune diseases: data from the French registries. Rheumatology (Oxford) 50 (1), Wolfe F, Michaud K (2004) Lymphoma in rheumatoid arthritis: the effect of methotrexate and anti-tumor necrosis factor therapy in 18,572 patients. Arthritis Rheum 50, Ruan SY, Chuang YC, Wang JY et al. (2012) Revisiting tuberculous pleurisy: pleural fluid characteristics and diagnostic yield of mycobacterial culture in an endemic area. Thorax 67, Bureau of National Health Insurance, Department of Health, Executive Yuan (2011) National Health Insurance in Taiwan 2011 Annual Report. Available from URL: IN TAIWAN 2011 ANNUAL REPORT.pdf. Accessed May 16, International Journal of Rheumatic Diseases 2014; 17 (Suppl. 3): 9 19

11 Risk of adverse events in RA 31 Arnett FC, Edworthy SM, Bloch DA et al. (1988) The American Rheumatism Association 1987 revised criteria for the classification of rheumatoid arthritis. Arthritis Rheum 31, Cohen J (1988) Statistical Power Analysis for the Behavioral Sciences. 2nd edn. Lawrence Erlbaum, Hillsdale, NJ. 33 Mercer L, Davies R, Galloway J et al. (2013) Risk of cancer in patients receiving non-biologic disease-modifying therapy for rheumatoid arthritis compared with the UK general population. Rheumatology 52, Hellgren K, Smedby K, Feltelius N, Baecklund E, Askling J (2010) Do rheumatoid arthritis and lymphoma share risk factors? Arthritis Rheum 62, Ekstrom K, Hjalgrim H, Brandt L et al. (2003) Risk of malignant lymphomas in patients with rheumatoid arthritis and in their first degree relatives. Arthritis Rheum 48, Baecklund E, Akbom A, Sparen P, Fektekius N, Klareskog L (1998) Disease activity and risk of lymphoma in patients with rheumatoid arthritis: nested case-control study. BMJ 317, Symmons D (2007) Lymphoma and rheumatoid arthritis again. Rheumatology 46, Mattey D, Thomson W, Ollier W et al. (2007) Association of DRB1 shared epitope genotypes with early mortality in rheumatoid arthritis: results of eighteen years of follow-up from the early rheumatoid arthritis study. Arthritis Rheum 56, United Kingdom Health Protection Agency. Geography TB data by country. Available from URL: org.uk. 40 United Kingdom Health Protection Agency (2010) World Health Organization (WHO) estimates of tuberculosis incidence by country, Available from URL: National Health Insurance Research Database, Taiwan. Available from URL: index.htm. Accessed May 18, Doran MF, Crowson CS, Pond GR, O Fallon WM, Gabriel SE (2002) Predictors of infection in rheumatoid arthritis. Arthritis Rheumatol 46, U.S. Centers for Disease Control and Prevention (2010) Trends in Tuberculosis, Available from URL: Accessed May 16, International Journal of Rheumatic Diseases 2014; 17 (Suppl. 3):