Background Studies have reported that poor prognostic factors (PPF) in RA, such as high anti-citrullinated protein antibodies (ACPA), are associated with erosions, rapid radiographic progression and/or extra-articular manifestations.1,2 Evidence from clinical trials and clinical practice indicate differences in treatment effects of biologic (b)DMARDs by ACPA status and/or level.3,4 PPF also play an important role in clinical management of patients (pts) with RA, including inpatient admissions.
Objectives To compare hospitalization rates of ACPA-positive (+) with ACPA-negative (–) pts managed with abatacept (ABA), non-ABA bDMARDs or conventional (c)DMARDs.
Methods This is a retrospective cohort analysis of Clinformatics Data Mart, a database of administrative health claims including results for outpatient laboratory tests, processed by national laboratory vendors under contract with the managed care organization, for a total of ∼53 million unique lives over 13 years. This analysis was restricted to adult pts (aged ≥18 years) who had at least two ICD-9-CM diagnosis codes for RA between Jan 2007 and Dec 2014 (identification period) and 12 months (M) of membership/drug benefit. Pts with ankylosing spondylitis, Crohn's disease, lupus, psoriasis or ulcerative colitis at or before the index date were excluded. ACPA+ was based on >19 or >5 U/mL, depending on the test utlilized. Follow-up for pts initiating ABA was from first day of treatment to first hospitalization, end of enrolment or end of 12M follow-up. The primary outcome was all-cause hospitalization at 12M. Descriptive statistics such as Wilcoxon rank-sum test for continuous variables or Pearson's chi-square test for categorical variables were used. Cox proportional hazard model was used to examine all-cause hospitalization adjusted for age, sex, region and past hospitalization. Additional covariates included co-morbidities that were different between ACPA+/– pts. Similar analyses were performed for pts treated with cDMARDs and non-ABA bDMARDs.
Results A total of 496 ABA, 7438 cDMARD and 2118 non-ABA bDMARD pts were included, with an overall past hospitalization rate of 35.5, 25.4 and 23.2%, respectively. Overall, 59.5, 41.5 and 51.7% of pts were ACPA+ in the ABA, cDMARD and non-ABA bDMARD cohorts, respectively. ACPA+ pts were older and less likely to have obstructive sleep apnoea, depression and myalgia/myositis across all cohorts. Unadjusted 12M hospitalization rate in ACPA+ vs ACPA– pts was 10.8 vs 16.4%, 13.7 vs 14.3% and 8.5 vs 10.8% in the ABA, cDMARD and non-ABA bDMARD cohorts, respectively. After controlling for baseline covariates, the hazard ratio for hospitalization in the ACPA+ (vs ACPA–) group was 0.57 (95% CI: 0.34, 0.98; p=0.04), 0.95 (95% CI: 0.84, 1.08; p=0.47) and 0.88 (95% CI; 0.66, 1.17; p=0.38) in the ABA, cDMARD and non-ABA bDMARD cohorts, respectively (Figure).
Conclusions ACPA+ pts with RA treated with abatacept have a lower rate of hospitalization than ACPA– pts. This pattern was not observed with cDMARDs or non-abatacept bDMARDs. Further efforts including matching and subgroup analysis should be explored for direct comparisons between the cohorts.
Visser K, et al. Ann Rheum Dis 2010;69:1333–7.
Combe B, et al Ann Rheum Dis 2016; doi: 10.1136/annrheumdis-2016–210602. [Epub ahead of print].
Gottenberg JE, et al. Ann Rheum Dis 2012;71:1815–9.
Sokolove J, et al. Ann Rheum Dis 2016;75:709–14.
Disclosure of Interest E. Alemao Shareholder of: Bristol-Myers Squibb, Employee of: Bristol-Myers Squibb, L. Burns Shareholder of: Bristol-Myers Squibb, Employee of: Bristol-Myers Squibb, Z. Guo Shareholder of: Bristol-Myers Squibb, Employee of: Bristol-Myers Squibb