Comparative Effectiveness of Abatacept Versus Tumor Necrosis Factor Inhibitors in Patients with Rheumatoid Arthritis Who Are Anti-CCP Positive in the United States Corrona Registry

The Corrona RA registry is an independent, prospective, national, observational cohort. Patients are recruited from 175 private and academic practice sites across 41 states in the US, with 725 rheumatologists participating. As of March 2018, the Corrona database included information on 48,535 patients with RA. Data on 367,457 patient visits and 169,968 patient-years of follow-up observation time have been collected, with a mean patient follow-up of 4.3 (median 3.4) years.

To augment the number of patients with laboratory data available, we included patients who were included in the Corrona Comparative Effectiveness Registry to Study Therapies for Arthritis and Inflammatory Conditions (CERTAIN; ClinicalTrials.gov identifier: NCT01625650) study, which was a substudy of the Corrona RA registry that mandated regular data collection at 3-month intervals, with uniform collection of anti-CCP status for all patients initiating biologic treatment [24]. The Corrona CERTAIN substudy included a subset of patients with RA who were ≥ 18 years of age, had initiated biologic therapy, had moderate or high disease activity [e.g., Clinical Disease Activity Index (CDAI) score > 10] at enrolment, and were either biologic-naïve or TNFi-experienced. Patients were recruited from 43 sites in the US, with 117 rheumatologists participating. Enrolment was from November 2010 to May 2015 and included 1 year of follow-up with regular visits at 3-month intervals. A total of 2795 patients completed enrolment.

This study was carried out in accordance with the Declaration of Helsinki. All participating investigators were required to obtain full board approval for conducting noninterventional research involving human subjects with a limited dataset. Sponsor approval and continuing review was obtained through a central Institutional Review Board (IRB), the New England Independent Review Board (NEIRB; no. 120160610). For academic investigative sites that did not receive a waiver to use the central IRB, full board approval was obtained from the respective governing IRBs and documentation of approval was submitted to Corrona, LLC prior to the initiation of any study procedures. All patients in the registry were required to provide written informed consent and authorization prior to participating.

This analysis included patients with RA who initiated abatacept or a TNFi (adalimumab, certolizumab pegol, etanercept, golimumab, or infliximab) in the Corrona RA registry or the CERTAIN comparative effectiveness substudy between 1 December 2005 and 31 August 2016 (Fig. 1). Eligible patients were aged ≥ 18 years, had a follow-up visit approximately 6 months after initiation of abatacept or TNFi treatment (visit window: 3–9 months using visit closest to 6 months if there were > 1 visit), had CDAI score measured at or prior to initiation of treatment and within the appropriate timeframe prior to the visit (for initiations between visits, CDAI score measured within 4 months prior to initiation was used) and had anti-CCP+ status at or prior to initiation of treatment. Follow-up CDAI score was measured at the 6-month visit or at the time of switch if patients switched medications prior to the 6-month visit. In addition, patients were included if they had no prior exposure to other non-TNFi biologics or tsDMARDs and if they were not in remission (CDAI ≤ 2.8) at the time of initiation. TNFi initiators with prior use of abatacept were excluded.

Data were collected during the study period from physician assessment and patient questionnaires completed during the clinical encounters. These forms were used to gather information on disease severity and activity [including serologic markers (anti-CCP) and components of ACR response criteria]; comorbidities; use of medications including steroids, csDMARDs, tsDMARDs and bDMARDs; and adverse events. As a strictly observational registry that reflects typical clinical practice, the Corrona registry does not mandate that laboratory data, including serologic markers and acute-phase reactants, be collected. In the CERTAIN substudy, laboratory data were a requirement, with a centralized laboratory performing all assays. Data elements collected in both the overall Corrona RA registry and the CERTAIN substudy included CDAI (swollen joint count in 28 joints, tender joint count in 28 joints, Physician Global Assessment and Patient Global Assessment), modified ACR 20, 50, and 70% response (mACR20, mACR50, and mACR70) criteria (mACR is based on two out of four measures; it does not include erythrocyte sedimentation rate or C-reactive protein), the modified Health Assessment Questionnaire (mHAQ) assessing physical function and five-dimension EuroQol questionnaire (EQ-5D). Data on demographics, insurance status, comorbid conditions, RA disease characteristics, and RA medication were available for > 98% of patients.

To balance for predisposing factors that may increase a patient’s likelihood of receiving either abatacept or TNFis, a propensity score—or the probability of treatment selection—was calculated for each eligible patient using baseline (at the time of drug initiation) patient demographics and disease characteristics [25]. Propensity score-matched treatment groups were created for abatacept and TNFis. Patients within each treatment group were matched 1:1 without replacement by prior TNF exposures of 0, 1, and ≥ 2 using the caliper method maximizing the number of patients including in the analysis. Separate propensity score models were fit, by prior biologic use stratum, to enable different covariates that were imbalanced within the stratum to be included (online supplementary table S1). Effectiveness at 6 months after treatment initiation was evaluated in both treatment groups.

The primary outcome was mean change in CDAI score over 6 months following initiation. Secondary outcomes at 6 months included achievement of remission (CDAI ≤ 2.8), low disease activity or remission (CDAI ≤ 10) in those with moderate or high disease activity at initiation, mACR20, mACR50, and mACR70 responses, and change from baseline in mHAQ score. Switching status among anti-CCP+ initiators of abatacept versus TNFis after propensity score matching was also assessed. Subgroup analyses were conducted by biologic-naïve and TNFi-experienced status at initiation.

A formal statistical analysis plan was developed prior to conducting the study. Anti-CCP positivity was defined as anti-CCP ≥ 20 U/ml. Baseline demographics and characteristics were compared between the treatment cohorts, and standardized differences were estimated. Standardized differences provide a measure of the imbalance in treatment groups with regards to the variable of interest, even if there are no statistically significant differences. The absolute value of the standardized difference of ≤ 0.1 for the overall population [25] and ≤ 0.2 within stratum (biologic naïve and TNFi experienced) was taken to indicate a negligible difference in the mean or prevalence of a covariate between treatment groups [25]. p values were calculated using t tests for normally distributed continuous variables and Chi-square tests for categorical variables. Propensity score models were fitted for each prior biologic category (0, 1, and ≥ 2) and patients were matched 1:1 within each stratum; the results of the matching using standardized differences are presented.

A total of 525 abatacept initiators and 1595 TNFi initiators met the inclusion criteria (Fig. 1). The baseline characteristics prior to propensity score matching (online supplementary table S2) showed that the two treatment groups differed (absolute value of the standardized differences > 0.1) in terms of age, duration of RA, prior number of csDMARDs, prior biologic use, current MTX use, and current prednisone use. After propensity score matching, the overall population included 330 pairs of abatacept and TNFi initiators and, of these, 97 were biologic naïve and 233 were TNFi experienced (one prior TNFi, n = 164; ≥ 2 prior TNFis, n = 69).

After propensity score matching, the baseline characteristics between abatacept initiators and TNFi initiators were well balanced with absolute value standardized differences of ≤ 0.1 in the overall population and of ≤ 0.2 in the biologic-naïve and TNFi-experienced cohorts (Table 1). The exception to this was erosive disease in the overall population, which was present in a higher proportion of TNFi initiators than abatacept initiators. In the overall population, most patients were female and in their late 50s, with approximately a decade of disease duration. Approximately 30% of patients had been enrolled in the CERTAIN substudy.

In the overall population, there was no significant difference in mean change in CDAI score at 6 months between abatacept initiators and TNFi initiators (abatacept initiators, − 9.85 [95% confidence interval (95% CI) − 11.40 to − 8.30]; TNFi initiators, − 8.53 [95% CI − 10.09 to − 6.98]; p = 0.24; Fig. 2). Similar results were shown for biologic-naïve patients (abatacept initiators, − 9.81 [95% CI − 12.47 to − 7.15]; TNFi initiators, − 12.31 [95% CI − 14.97 to − 9.66]; p = 0.19). TNFi-experienced patients who were anti-CCP+ and who initiated abatacept had a significantly greater improvement in CDAI score than those initiating a TNFi (abatacept initiators, − 9.87 [95% CI − 11.75 to − 7.98]; TNFi initiators, − 6.96 [95% CI − 8.85 to − 5.08]; p = 0.033).

In both the overall population and the biologic-naïve and TNFi-experienced cohorts, mHAQ score improvement (Fig. 3) and binary outcomes (low disease activity, remission, mACR20 and mACR70; Fig. 4) at 6 months were similar between treatments. Among TNFi-experienced patients, abatacept initiators were more likely to achieve a mACR50 response than TNFi initiators [1.79 (95% CI 1.12–2.86); p = 0.014].

Switching status was similar between the abatacept and TNFi initiators in the overall population and in the biologic-naïve cohort (p = 0.059 and p = 0.87, respectively; online supplementary table S3). However, TNFi-experienced abatacept initiators were more likely to remain on treatment at 6 months than TNFi initiators (85.0 vs. 73.8%; p = 0.010).