Background: Abatacept (ABA) can induce decrease in gamma-globulins, but the long-term safety of such reduction is unknown. Moreover, it is suggested that such decrease is dissociated from disease activity response.

Objectives: To evaluate ABA-induced gamma-globulins reduction and its correlation with disease activity control and the risk of infection in rheumatoid arthritis(RA) patients.

Methods: This is a retrospective inception cohort of RA patients undergoing ABA for the first time in a large single tertiary cohort (2007 to 2019). Patients were evaluated regarding clinical and inflammatory data, total and specific (IgG, IgM, IgA) gamma-globulins assessed before, at 3 and 6 months, and then every 6 months up to discontinuation/censoring. The occurrence of severe or recurrent infections as cause of discontinuation of treatment was recorded. All patients were submitted to a systematic infectious screening protocol before and during treatment.

Results: One hundred seventy-nine RA patients were included. They were predominantly female (93%;n=167) and had a positive rheumatoid factor (RF, 84%; n=151). Median(range) age and disease duration were 55.1(17-81.3) and 14(1.6-39.8) years, respectively. ABA was used in median as the 3 ^rd (1-8) bDMARD. Most patients (74.3%, n=134) had already used a TNFi previously and 34.1% (n=61) had failed to rituximab. Baseline DAS28 [median(range)] was 4.9 (2-8.62), CDAI 27(5-66), HAQ 1.5 (0-3), ESR 18(2-111), and CRP 8.1(0.1 -135.7). Levels of total gamma-globulins(TGG) were 1.2 (0.5-2.8g/dL), IgG 1,094(422-2,785mg/dL), IgM104(19-405 mg/dL), and IgA 324 (100-739mg/dL). Forty-five patients (25.1%) had low IgG before ABA, but only 8(4.5%) had IgG<600mg/dL and low TGG. At 3 months, remarkable decreases were observed in total gamma-globulin(1.2 vs. 1.1g/dL), IgG(1,094 vs. 992mg/dL), IgM(104 vs. 93mg/dL), and IgA(324 vs. 302mg/dL), in comparison to baseline(p<0.05). Longitudinal measurements demonstrated that levels of gamma-globulins remained stable for subsequent evaluations(p>0.05). At baseline, TGG, IgG, IgM and IgA correlated positively to all disease activity parameters: DAS28(r=+0.33; r=+0.31;r=+0.24 and r=+0.37, respectively; p<0.02 for all), SDAI (r=+0.35; r=+0.37; r=+0.43; r=+0.34, respectively; p<0.04 for all), CDAI (r=+0.32; r=+0.29; r=+0.39; r=+0.31, respectively; p<0.03 for all, except IgA), swollen joints (r=+0.24; r=+0.23; p<0.01 only for TGG and IgG, respectively); tender joints (r=+0.18; r=+0.21; p<0.04 only for IgG and IgM respectively), ESR (r=+0.35; r=+0.24; r=+0.41; p<0.02 for TGG, IgG and IgA, respectively) and CRP (r=+0.20; r=+0.16; r=+0.27; p<0.05 for TGG, IgG and IgA, respectively). Longitudinally, most of these correlations were lost(p>0.05). Moreover, at 6 months, 25 patients (14%) achieved DAS28 ≤ 2.6 and 49 (27.4%) had low disease activity (DAS28 ≤ 3.2). Baseline or longitudinal measures of TTG and subtypes were similar among patients who responded and those who did not, regardless of the cut-off used(p>0.05. Only 14 patients (9.4%) had the medication discontinued due to infections: 3 due to recurrent infections and 11 to severe infections. One patient died. The frequencies of low IgG (under the normal limit of normality) at baseline and all time points(p>0.05) were similar among patients with and without severe/recurrent infections. None of these patients had low TGG or very low IgG(<600mg/dL) at baseline or during treatment and the levels of TGG and all subtypes were comparable at all time points (p>0.05).

Conclusion: This cohort of real-world RA patients shows that reduction in gamma-globulin levels induced by ABA is safe, non-progressive and not associated to a higher risk of infection, even in patients with low IgG or TGG. Additionally, it is not associated to clinical response.