This analysis encompasses cumulative (controlled and open-label extension) OI data from 16 abatacept clinical trials with mean (SD) abatacept exposure of 36.9 (26.2) months and 21,274 total patient-years. Types of OIs occurring during abatacept treatment and incidence rates of OIs were similar in the controlled periods with placebo and abatacept. IRs of OIs remained stable with long-term abatacept usage when considering open-label extension data.
In this analysis, the OI IR was relatively low in patients treated with abatacept. A similar IR of OIs, including TB, was observed with abatacept versus placebo in the controlled periods of the clinical trials. In the cumulative periods, the IR of OIs seen with abatacept was low and consistent with that observed in the controlled periods with the exception of herpes simplex and herpes zoster, for which lower IRs were reported in the cumulative periods versus controlled periods. Furthermore, the IR range of OIs observed for abatacept in this analysis (0.17–0.21 per 100 patient-years, across both periods) was generally consistent with the range reported in literature for other bDMARDs; in particular, TNF inhibitors (0.15–0.30 per 100 patient-years from a US study; notably,were included but shingles (herpes zoster) was excluded from the study by Baddley et al) . In some published post-marketing observational studies, the rates (95% CI) of certain OIs for other biologic therapies varied considerably by biologic treatment and geographic area: infliximab, 1.08 (0.37–3.22; Japan; ), 8.0 (2.0–50.0; Spain;  infliximab versus etanercept, 17.6 (4.3–72.9), adalimumab versus etanercept, 10.28 (2.35–44.94; France; ). There is some evidence to suggest a similar risk for OIs between some biologics, but there may be differences for other OI outcomes (e.g. herpes zoster) ; hence, information is still lacking.
Although patients receiving an active comparator (ATTEST: infliximab; AMPLE: adalimumab) were excluded from this analysis, previously published data from these studies indicate that OIs were less frequently reported with abatacept versus an active comparator. In the ATTEST study, no OIs were reported with abatacept, but two OIs occurred in the infliximab group by Day 197 (controlled period, primary time point): a pseudomonal lung infection and a Pneumocysitis jiroveci pneumonia . The IR (95% CI) of OIs did not increase during abatacept treatment in the cumulative versus controlled period for patients who were treated with infliximab and switched to abatacept at Day 365: 0.0 (0.0–0.0) versus 2.6 (0.7–6.6) . In the AMPLE study, two cases of OI (1 with abatacept and 1 with adalimumab) occurred during the first year of treatment (controlled period); both were AEs of mucocutaneous oral candidiasis, and neither patient discontinued treatment . In the 2-year cumulative period, 8 cases of OI occurred, 4 per treatment group: 1 case of histoplasmosis (AE) and 3 cases of oral candidiasis (1 SAE, 2 AE) with abatacept; 1 case of disseminated histoplasmosis (SAE), 2 cases of TB (miliary, pulmonary; both SAEs) and 1 oral candidiasis (AE) with adalimumab. None of the OIs in the abatacept group led to discontinuation, but both patients with reported TB in the adalimumab group discontinued the study .
The IRs for TB were low and comparable in both the abatacept and placebo groups and across the study periods in the analysis. The risk of TB in patients with RA has been shown to be elevated due to both the disease and the mechanism of action of many RA therapies, including steroids. However, risk of TB is largely driven by the country where such studies are conducted (i.e. endemic areas versus non-endemic areas) [26, 27]. While we observed a rate of 80/100,000 within the cumulative period, nearly all such cases occurred in countries with general population rates of TB between 20-80/100,000 person years. As such, one cannot exclude the possibility that some TB cases might be newly acquired in these endemic countries; reactivation of latent TB is also possible. In addition, although TB screening was applied at study entry, the possibility of some false negative results cannot be excluded. In a systematic literature review, seven observational studies addressing TB were identified; notably, most of them had a moderate or high risk of bias as per the Hayden’s tool . The review showed an increased risk of TB in patients receiving TNF inhibitors, compared with the general population and with patients receiving csDMARDs (adjusted hazard ratio: 2.7 to 12.5 per study) . Whether abatacept substantially increases the risk of TB and its relative risk to other RA therapies, such as TNF antagonists, is unknown. To date, no such comparisons have been made in real-world data in areas endemic for TB. There have been limited direct comparisons in the context of randomized clinical trials, and in each of these small trials fewer cases of TB were observed in patients treated with abatacept versus infliximab . Regardless, given the potential risk of TB in patients with RA, it is important to screen for TB prior to starting any biologic therapy .
Herpes zoster IRs reported in the literature for patients with RA who received biologic agents range from 1.61 to 2.71 per 100 patient-years [1, 3]. Notably, the data from both controlled (IR 1.9 per 100 patient-years) and cumulative (IR 1.53 per 100 patient-years) periods for herpes zoster lie within this reported range, supporting the currently known safety profile of abatacept. A systematic literature review used to inform the EULAR guidelines for the treatment of RA highlighted the lack of comparative data for risk of OIs with abatacept versus other bDMARDs and targeted synthetic (ts)DMARDs , however, a recent study using data from 3 large US healthcare claims databases found that the risk of overall OIs and TB was similar with abatacept versus other b/tsDMARDs . In the sensitivity analysis, an elevated risk of OIs for abatacept was observed in one of the three databases, which might be due to imbalances between the groups, such as greater co-medication differences in abatacept versus other b/tsDMARDs initiators, or due to channeling bias .
In a previous analysis of tofacitinib in the real-world setting, the herpes zoster risk was significantly higher for tofacitinib versus abatacept, with adjusted hazard ratio of 2.01 (95% CI 1.40–2.88) .
Overall, there is inconsistency in defining OIs across clinical trial research programs. Some efforts have been made to categorize the OIs, including a consensus review and recommendations for reporting of OIs from clinical trials . In the published review, the definition of an OI (an indicator infection) was formulated as “the presence, or specific presentation, of a pathogen that suggests a higher likelihood of an alteration in host immunity” . In the present analysis, the indicator species and indicator infection definitions were employed to define OIs and adjudicate the OI cases, to enhance clinical relevance, and to improve generalizability of the findings. However, our adjudication efforts were limited as discussed below.
Certain strengths and limitations of this study should be considered when interpreting the results. Importantly, the analysis includes details not previously published on OI events reported in the cumulative periods of the trials, which allows for better generalizability of results, as the original controlled trial groups were maintained in the analysis . The methodologic approach of retrospectively adjudicating the OI cases resulted in difficulties in adjudicating some of the non-serious outcomes (herpes zoster) due to lack of clinical details collected at the time of the event. In this analysis, the IRs were computed for all reported infections as defined by the standard MedDRA terms, thus including the microorganisms and presentation that may not typically be considered as opportunistic (e.g., Pseudomonas aeruginosa-caused pneumonia) ; this limitation may have led to an overestimation of IRs. Lastly, each of the trials used in this analysis may be associated with general limitations common to all clinical trial studies. These include stringent patient eligibility criteria, which may lead to a specific patient population with fewer comorbidities but more severe RA at the start of the trial, compared with routine clinical practice. Some specific limitations of the clinical trials, such as sample size and short follow-up period (e.g., ACCOMPANY trial) [33, 34] should also be taken into account.