Analysis of Infections and All‐Cause Mortality in Phase II, Phase III, and Long‐Term Extension Studies of Tofacitinib in Patients With Rheumatoid Arthritis

To determine the rate of infection and all‐cause mortality across tofacitinib phase II, phase III, and long‐term extension (LTE) studies in patients with moderately to severely active rheumatoid arthritis (RA).

receptors for interleukin-2 (IL-2), IL-4, IL-7, IL-9, IL-15, and IL-21. These cytokines are integral to the activation, proliferation, and function of lymphocytes; inhibition of their signaling may result in modulation of multiple aspects of the immune response (1). In addition, inhibition of JAK-1 results in attenuation of signaling by additional proinflammatory cytokines, such as IL-6 and interferon-␥ (IFN␥) (1).
In a large, global clinical development program, tofacitinib was evaluated as monotherapy or in combination with methotrexate (MTX) or other nonbiologic disease-modifying antirheumatic drugs (DMARDs). The program included 2 ongoing long-term extension (LTE) studies that enrolled patients from prior qualifying index phase II or III studies (Table 1). Patients with RA already have a higher risk of death (2) and infection (3) compared with the general population, and the risk of infection is further increased in patients treated with DMARDs (4,5). Here, we describe pooled data on infections, including serious infections, and mortality across this tofacitinib RA clinical development program.

PATIENTS AND METHODS
Three collections of pooled safety data were used for the analyses (see Supplementary Appendix 2, available on the Arthritis & Rheumatology web site at http://onlinelibrary. wiley.com/doi/10.1002/art.38779/abstract).
Pooled phase II, phase III, and LTE studies in tofacitinib-treated patients (P2P3LTE). This data set includes any patient in the phase II or phase III studies described in Table 1 or in LTE studies (patients entering from the phase II/III studies shown in Table 1) who received tofacitinib at any time (all doses combined).
Pooled phase III studies (P3All). This data set consists of pooled data from 5 phase III studies in patients with moderate-to-severe RA and an inadequate response to nonbiologic or biologic DMARDs (Table 1). Patients received tofacitinib (5 mg twice daily or 10 mg twice daily) or placebo as monotherapy or in combination with nonbiologic DMARDs (mainly MTX). Patients randomized to receive placebo were advanced in a double-blind manner to receive tofacitinib at a dosage of 5 mg twice daily or 10 mg twice daily, at 3 months or 6 months. Data from patients receiving placebo were classified within the placebo group until the patients advanced to the tofacitinib group and then were classified within the tofacitinib group. One phase III study included an active-treatment control group (40 mg of adalimumab, administered subcutaneously every 2 weeks, in patients receiving background MTX).
Pooled LTE studies. This data set consists of pooled data from 2 ongoing, open-label LTE studies (Table 1). RA patients from qualifying index studies were allowed to enroll in LTE studies (Table 1). In LTE studies, patients from phase II studies (Table 1) initiated treatment with tofacitinib at a dosage of 5 mg twice daily. In patients from phase III studies (Table 1) and additional ongoing phase I, phase II, and phase III studies (see Supplementary Appendix 2, available on the Arthritis & Rheumatology web site at http://onlinelibrary. wiley.com/doi/10.1002/art.38779/abstract), treatment with tofacitinib was initiated at a dosage of 10 mg twice daily, with the exception of patients in phase III studies from China and Japan, in whom treatment was initiated at a dosage of 5 mg twice daily, as required by the protocols.
In LTE studies, the dosage of tofacitinib could be reduced from 10 mg twice daily to 5 mg twice daily, tofacitinib could be temporarily discontinued for up to 28 days, or the dosage of tofacitinib could be increased from 5 mg twice daily to 10 mg twice daily at the discretion of the investigator. Dosage adjustment or discontinuation of permitted concomitant RA medications (including MTX, leflunomide, sulfasalazine, antimalarials, auranofin, injectable gold preparations, nonsteroidal antiinflammatory drugs, and/or glucocorticoids [up to 10 mg prednisone/day or equivalent]) was allowed. Patients were classified according to a dosage of 5 mg twice daily or 10 mg twice daily, based on the highest tofacitinib dosage administered during the first 135 days of treatment in LTE studies.
In some studies, data collection and analyses were still ongoing at the time of this analysis (i.e., the databases are not locked; some values may change for the final, locked databases) ( In general, phase III and LTE studies excluded patients with a significant infection within the past 6 months; white blood cell counts of Ͻ3.0 ϫ 10 3 /mm 3 ; absolute neutrophil counts of Ͻ1.2 ϫ 10 3 /mm 3 ; untreated or inadequately treated Mycobacterium tuberculosis (TB) infection (as determined by chest radiography and QuantiFeron testing or, if QuantiFeron tests were unavailable, by purified protein derivative skin testing); recurrent herpes zoster infection or disseminated (single episode) herpes simplex virus infection; or infection with human immunodeficiency virus, hepatitis B virus (HBV), or hepatitis C virus (HCV). Study treatment was discontinued and patients were withdrawn from the studies if a serious infection developed or if they had 2 sequential absolute neutrophil counts of Ͻ0.5 ϫ 10 3 /mm 3 .
Safety evaluations. Safety monitoring included the collection of all adverse events (AEs) and serious AEs (SAEs), regardless of the presumed causality, graded in terms of severity and relationship to study treatment and coded using Medical Dictionary for Regulatory Activities version 13.1 (for phase II and phase III studies) and version 14.1 (for LTE studies). The severity of neutropenia and lymphopenia was defined according to the Outcome Measures in Rheumatology (OMERACT) criteria (6).
SAEs were defined as follows: fatal or life-threatening, requiring hospitalization or extension of existing hospitalization, resulting in persistent or significant disability/incapacity or congenital abnormality/birth defect, or considered to be an important medical event. Serious infections were defined as those requiring hospitalization for treatment or parenteral antimicrobial therapy or otherwise meeting the SAE criteria.
The investigators assessed cause of death and its relationship to the study medication. Deaths occurring in phase III and LTE studies after February 25, 2009 were also adjudicated by an independent external cardiovascular safety end point adjudication committee whose members were blinded with regard to the treatment assignments.
Statistical analysis. Overall mortality and infection data pooled from the P2P3LTE population (cutoff date: April 19, 2012) are reported. Exposure-estimated incidence rates and exposure-adjusted event rates were calculated as the
In the P3All population, the exposure-adjusted event rates for infection of any severity were similar among patients who received tofacitinib 5 mg twice daily, patients who received tofacitinib 10 mg twice daily, and those who received placebo (84. 7  In the P2P3LTE population, 259 patients had serious infections (overall incidence rate 3.09 events per 100 patient-years; 95% CI 2.73-3.49). The most common serious infection was pneumonia; other commonly reported serious infections included skin and soft tissue infection. The incidence rates of serious infection were stable over time, with a decrease in the point estimates during the 2 latest time intervals (36-42 months and Ͼ42 months) ( Figure 1A).
In the LTE population but not the P3All population, the incidence rates of serious infection were numerically higher in the group receiving tofacitinib at a dosage of 10 mg twice daily compared with the group receiving tofacitinib at a dosage of 5 mg twice daily, but the 95% CIs overlapped (3.6 events per 100 patientyears [95% CI 2.96-4.38] and 2.62 events per 100 patient-years [95% CI 2.11-3.24], respectively) ( Table  2). In the P3All and LTE populations, the incidence rates were numerically higher in patients receiving background DMARDs (compared with those receiving tofacitinib monotherapy) and in patients receiving glucocorticoids, although the 95% CIs overlapped ( Table 2).
In a Cox proportional hazards model in the P2P3LTE population, age (Ն65 years), corticosteroid dose Ն7.5 mg, diabetes, and tofacitinib dose were independent factors associated with an increased risk of serious infection, because the lower limits of the 95% CIs were Ͼ1 ( Figure 1B) Serious herpes zoster infections requiring hospitalization or parenteral antiviral therapy were infrequent, with 3 cases in phase II studies (1 each in patients receiving tofacitinib at dosages of 5, 10, and 15 mg twice daily), 5 cases in phase III studies (4 in patients receiving tofacitinib 5 mg twice daily and 1 in a patient receiving tofacitinib at a dosage of 10 mg twice daily), and 13 cases in LTE studies (11 in patients receiving tofacitinib 5 mg twice daily and 2 in patients receiving tofacitinib 10 mg twice daily). The 3 cases in phase II studies, 3 of 5 cases in phase III studies, and 9 of 13 cases in LTE studies occurred in Asian patients. Overall, 4 patients had herpes zoster ophthalmicus, and 2 patients had multidermatomal herpes zoster. More details regarding herpes zoster infection in this clinical development program for tofacitinib in RA have been reported elsewhere (8).
The number of reports of positive test results for HBV infection or HCV infection was small. There were 3 cases of HBV infection in Asian patients: 1 possible reactivation of HBV with an elevated transaminase level, 1 possible false-positive result of a hepatitis surface antigen test (or transient reactivation of HBV without an elevation in the transaminase level and with a negative HBV DNA test result), and 1 case of new HBV infection, which was mild in severity with a transaminase level 2-3 times the upper limit of normal. Antiviral agents were not given in 2 cases (it is unknown whether antiviral agents were given in the third case), and at the time of this report, 1 case of hepatitis was resolved (2 were ongoing).
TB and opportunistic infections. In the P2P3LTE population, TB was reported in 16 patients (4 from the Philippines, 3 from Korea, 2 each from India and China, and 1 each from Japan, Thailand, Mexico, the US, and Spain). There were no cases of TB in phase II studies. In the P3All population, 6 cases of TB (all in patients treated with tofacitinib at a dosage of 10 mg twice daily) were reported; there were no cases in the placebo group or the adalimumab group. In the LTE population, TB was reported in 10 patients (5 in the group receiving tofacitinib at a dosage of 5 mg twice daily and 5 in the group receiving tofacitinib at a dosage of 10 mg twice daily).
Excluding cases of TB, 25 patients within the P2P3LTE population experienced an opportunistic infection (0.30 events per 100 patient-years [95% CI 0.20-0.44]). In addition to the 2 cases of multidermatomal herpes zoster described above, 8 patients had esophageal candidiasis (several cases represented incidental findings on endoscopies performed for other reasons), 6 patients had cytomegalovirus infection/viremia, 3 patients had cryptococcal infection (2 cases of pneumonia and 1 case of meningitis), 3 patients had Pneumocystis jiroveci pneumonia, 2 patients had nontuberculous mycobacteria in the lung, and 1 patient had BK virus-associated encephalitis, which improved with discontinuation of study drug and appropriate medical therapy. Among patients with opportunistic infections, 1 death occurred (due to P jiroveci pneumonia).
Baseline glucocorticoid treatment did not appear to be associated with an increased risk of opportunistic infection, including TB, in the P3All and LTE study populations. In the P3All cohort but not the LTE cohort, male patients and patients ages Ն65 years were at slightly increased risk compared with female patients and those ages Ͻ65 years (the 95% CIs overlapped).
Neutrophil and lymphocyte levels. In the P3All population, dose-dependent decreases in neutrophil counts in the tofacitinib and adalimumab groups were of a similar magnitude. Neutrophil counts remained largely within reference ranges, and the rates of neutropenia of any severity (as defined by the OMERACT criteria) were low during tofacitinib treatment (moderate-to-severe neutropenia at month 12 in Ͻ1% of patients receiving tofacitinib 5 mg twice daily and 1.9% of patients receiving tofacitinib 10 mg twice daily, versus Ͻ1% of patients receiving adalimumab). In the overall LTE population, no further decreases in the mean neutrophil count were observed (Ͻ1% of patients had moderate-to-severe neutropenia). No patients had life-threatening neutropenia. There was no observed relationship between the presence of neutropenia and the occurrence of either treated or serious infections in the phase III or LTE study populations. In the LTE study population, 2.1% of patients with confirmed neutropenia (i.e., identified at 2 sequential visits) and 4.6% of those without neutropenia (95% CI 3.98-5.32) developed a serious infection.
At baseline, the mean lymphocyte counts were Ͻ2 ϫ 10 3 /mm 3 (lymphopenia, as defined by the OMERACT criteria) in all P3All study groups, and 25% of these patients had OMERACT criteria-defined moderate-to-severe lymphopenia (lymphocyte counts of 0.5-1.5 ϫ 10 3 /mm 3 ). Among the tofacitinib-treated patients in the P3All studies, the mean lymphocyte count decreased ϳ10% between baseline and month 12. Lymphopenia did not appear to be dose-related, and the frequency of lymphopenia in the P3All studies was similar between tofacitinib-treated patients and placebotreated patients at month 3 and month 6. Moderate-tosevere lymphopenia developed in a larger number of patients receiving background DMARDs compared with patients receiving tofacitinib monotherapy.
Mortality. The all-cause mortality rate among patients in the P2P3LTE population receiving tofacitinib, including deaths occurring within 30 days of the last dose, was 0.30 events per 100 patient-years (95% CI 0.20-0.44). The mortality rate including deaths occurring at any time after the last dose was 0.53 events per 100 patient-years (95% CI 0.40-0.71). Mortality rates in the LTE population were generally consistent with those in the P3All population (Table 4) and, within the P3All population, the rates were similar in the groups receiving tofacitinib 5 mg twice daily, tofacitinib 10 mg twice daily, adalimumab, and placebo. Narratives for deaths are included in Supplementary Appendix 3 (available on the Arthritis & Rheumatology web site at http://onlinelibrary.wiley.com/doi/ 10.1002/art.38779/abstract). Two deaths were reported among patients receiving tofacitinib in phase II studies. In the phase III studies, 12 deaths occurred in the tofacitinib group (10 of which occurred up to 30 days after administration of the last dose of study treatment), and 1 death each occurred in the placebo and adalimumab groups. In the LTE studies, 31 deaths occurred (14 of which occurred within 30 days after administration of the last tofacitinib dose).
All 14 deaths that occurred during the phase III studies and 20 of 31 deaths that occurred during the LTE studies were adjudicated. Among the deaths that occurred during the phase III studies, 6 were attributable to infection, 3 were attributable to noncardiovascular causes classified as "other" (1 in a patient receiving placebo), 2 were attributable to cardiac causes (1 in a patient receiving adalimumab), 1 was due to cancer, 1 was due to trauma, and 1 death was of unknown cause. Among the 20 adjudicated deaths in the LTE studies, 6 were due to infection, 6 were attributable to cancer, 3 were attributable to cardiac causes, 2 were attributable to noncardiovascular causes classified as "other," 2 were due to suicide, and 1 death was of unknown cause.

DISCUSSION
In this analysis of infection and mortality data from the global tofacitinib RA clinical development program, the overall rates during tofacitinib therapy appear to be similar to those observed during treatment with biologic agents used in RA (9)(10)(11)(12)(13)(14)(15)(16). The P2P3LTE data set represents the largest population of RA patients treated with tofacitinib for the longest duration of time and is particularly useful for assessing overall rates in the program and over time. The limitation of this data set is that it is not categorized according to dosage, because patients in several of the phase II studies were treated with tofacitinib dosages other than 5 mg twice daily and 10 mg twice daily, and patients may have changed dosages when they transitioned from the qualifying index study to the LTE study. Additionally, because by study design the duration and extent of exposure in LTE studies were longer and greater with tofacitinib 5 mg twice daily than with tofacitinib 10 mg twice daily, comparisons between dosages in those studies should be interpreted with some caution. Finally, an inherent limitation of all clinical trial data sets is the relatively limited placebo-controlled phase with regard to duration as well as patient numbers; this restricts the strength of any direct comparison between the active-treatment and placebo groups. Therefore, for additional contextualization, the following discussion refers to data from external compilations of clinical trial safety analyses and observational databases.
Infection AEs were the most common cause of permanent discontinuation of tofacitinib in both the P3All and LTE populations; however, this was attributable in part to protocol-mandated discontinuation criteria requiring study withdrawal in the case of serious infection. The overall rates of serious infection reported in patients treated with tofacitinib (3.1 events per 100 patient-years) were consistent with the rates previously reported in clinical trial safety analyses of adalimumab (3.9-5.1 events per 100 patient-years [17]), rituximab (3.9-4.3 events per 100 patient-years [15]), tocilizumab (3.8-5.1 events per 100 patientyears [16]), etanercept (3.8 events per 100 patient-years [18]), abatacept (2.0-3.1 events per 100 patient-years [19,20]), and golimumab (5.09 events per 100 patientyears [21]). Furthermore, a meta-analysis of published data (22) showed that the rates of serious infection in patients treated with tofacitinib and in those treated with approved biologic DMARDs were similar. The rates of serious infection associated with tofacitinib treatment are also within the ranges demonstrated in various large observational databases of patients receiving tumor necrosis factor (TNF) inhibitor therapies (10)(11)(12)(13)(14).
The risk of serious infection associated with tofacitinib treatment was increased in older patients and was numerically higher in those receiving tofacitinib with background DMARDs and/or glucocorticoids, compared with patients receiving tofacitinib monotherapy. In the LTE but not the P3All populations, the tofacitinib dosage was also an independent risk factor for serious infection. This was confirmed by a Cox proportional hazards model that identified tofacitinib dosage, age, diabetes, and corticosteroid dose as independent risk factors and is consistent with reports derived from multiple databases of RA patients receiving biologic DMARDs (16,(23)(24)(25). However, the results of interaction tests suggest that the other risk factors are independent of tofacitinib treatment and reflect the underlying characteristics of the population of RA patients treated with DMARDs. A separate analysis of infection of any . Each of these rates was also higher than the rates reported in the literature for RA patients treated with biologic agents (0.89-1.6 events per 100 patientyears) (15)(16)(17)(18)(19)(20)(21). In the tofacitinib RA clinical development program, approximately one-quarter of patients were Asian, and a higher incidence of herpes zoster was observed in Asian countries compared with other countries. Most cases of herpes zoster infection were mild to moderate, cases of serious herpes zoster infection were rare, and patients recovered after receiving appropriate medical management.
The higher rates of herpes zoster infection among patients treated with tofacitinib may be related to the mechanism of action of tofacitinib, which involves a decrease in lymphocyte activation and proliferation. The human antiviral defense is also associated with intact responses to type I IFN (IFN␣ and IFN␤) and type II IFN (IFN␥) (27), the receptors of which signal via JAK-1. Because tofacitinib inhibits signaling through JAK-1, it is possible that such a mechanism is related to an increased risk of herpes zoster infection. The higher rates of herpes zoster infection observed in control groups may reflect the increasing absolute or reporting rate of herpes zoster infection in the overall population as well as the possibility of an increased background prevalence in certain Asian countries. The American College of Rheumatology recommends that RA patients who are Ն60 years of age should receive a herpes zoster vaccination prior to beginning treatment with a biologic agent (28).
The incidence of TB in the tofacitinib program was consistent with the incidence reported for TNF inhibitors (0.02 [global] to 2.56 [South Korea]) and nonbiologic DMARDs (0.01 [global] to 0.28 [South Korea]) (9,(29)(30)(31)(32)(33)(34)(35)(36)(37)(38)(39). The majority of cases were reported in countries where TB is endemic. Opportunistic infections other than TB have also been reported in association with DMARD therapy (5,16,40). There is no established definition of opportunistic infection, which makes comparisons between studies and agents difficult. In addition, differences in endemic rates of certain types of opportunistic infection make comparisons across global studies uncertain. The opportunistic infections observed in the tofacitinib program are consistent with the types of opportunistic infection reported in association with biologic therapies (5,16,(41)(42)(43).
A higher rate of serious infection was observed in patients with absolute lymphocyte counts of Ͻ0.5 ϫ 10 3 /mm 3 . Although it was unclear whether the occurrence of a serious infection was the result of lymphopenia or whether lymphopenia was the result of infection, lymphocyte counts should be determined prior to initiation of tofacitinib therapy and should be monitored regularly (e.g., every 3 months) during therapy, to inform decisions regarding treatment discontinuation.
Mortality rates in patients with RA are generally 1.5-1.6-fold higher than those in the general population (2). The reported mortality rates in RA patients treated with tofacitinib are similar to those reported in RA patients treated with biologic DMARDs: 0.51 deaths per 100 patient-years with rituximab (15), 0.53 events per 100 patient-years with tocilizumab (16), and 1.1 events per 100 patient-years with etanercept, infliximab, or adalimumab (44) compared with tofacitinib, with events per 100 patient-years of 0.30 (including deaths occurring within 30 days of the last dose) and 0.53 (including deaths occurring at any time after the last dose).
This analysis provides important safety information regarding treatment with tofacitinib in patients with RA. Overall, the risk of infection, including serious infection, appears to be stable over time and similar to that observed in RA patients treated with biologic agents. Mortality rates are consistent with published rates in patients with active RA, including those receiving other DMARDs. It will be important to be vigilant for the emergence of any new safety signals or changes in the trends of AE profiles.