Background/Purpose:  Mavrilimumab is an anti-GM-CSF receptor Ab that has recently demonstrated clinical efficacy in Phase 2 studies in rheumatoid arthritis (RA) patients. Predictive biomarkers to identify patients most likely to benefit from new RA therapies such as mavrilimumab would be of great utility. Recent work has demonstrated that a subset of RA patients has autoantibodies to peptidyl arginine deiminase 4 (PAD4), a citrullinating enzyme involved in NETosis and the formation of autoantigens targeted by anti-citrullinated protein antibodies (ACPA). Since PAD4 is expressed by neutrophils and macrophages, key cell types targeted by mavrilimumab, we explored whether anti-PAD4 autoantibodies may be associated with clinical response to mavrilimumab.

Methods:  Novel assays to detect autoantibodies reactive with PAD4, PAD3 or PAD2 were developed and used to measure anti-PAD4, PAD3 and PAD2 levels in 288 subjects enrolled in the phase 2a study CP219 and anti-PAD4 and PAD3 levels in 323 subjects enrolled in the phase 2b study 1071. All subjects in the study met the ACR criteria for a diagnosis of RA.

Results:  Overall, 35% of the subjects tested positive for anti-PAD4 antibodies and 20% tested positive for anti-PAD3 antibodies. In the phase 2b study 1071 in DMARD-IR patients, the subjects who tested positive for anti-PAD4 were enriched for the presence of ACPA (OR = 54.7, 95% CI = 7.5, 400.0), had higher baseline joint erosion (mean difference = 9.9, 95% CI = 2.6, 17.2) and higher modified total Sharp scores (mean difference = 16.2, 95% C.I. = 3.9, 28.5) than subjects who tested negative. There was a significant treatment-biomarker effect across clinical endpoints ACR20, ACR50, ACR70 and change in DAS28-CRP. Subjects treated with the 150 mg dose of mavrilimumab, the highest dose evaluated, and who tested negative for anti-PAD4 (68% of subjects), had significantly greater benefit from mavrilimumab compared to active placebo (36.7% difference from placebo; odds ratio for ACR50 response = 17.61, 95% CI =3.86, 80.29) relative to subjects who tested positive for anti-PAD4 (8.7% difference from placebo; odds ratio for ACR50 response = 1.46, 95% CI = 0.47, 4.56; P = 0.01 for treatment-biomarker interaction). Similarly, the DAS28-CRP response was enhanced in autoantibody negative (mean difference = -1.44, 95% CI = -1.91, -0.98) compared to autoantibody positive subjects (mean difference = -0.79, 95% CI = -1.45, -0.13; P = 0.022 for treatment-biomarker interaction). Similar trends were observed for an association between greater clinical responses to mavrilimumab relative to placebo in anti-PAD4 negative subjects in the earlier phase 2a trial CP219. Finally, this effect is specific to anti-PAD4 antibodies as no association between response and presence of antibodies to PAD3 was observed, and we only detected very low levels of anti-PAD2 antibodies in the vast majority of RA patients at levels comparable to healthy controls.

Conclusion:  Autoantibodies to PAD4 are a promising novel predictive biomarker for the targeted GM-CSF receptor inhibitor mavrilimumab, and further validation of this biomarker in future clinical studies is warranted.

« Back to ACR/ARHP Annual Meeting 2016

ACR Meeting Abstracts - https://acrabstracts.org/abstract/lack-of-autoantibodies-to-peptidyl-arginine-deiminase-4-predict-increased-efficacy-of-mavrilimumab-in-rheumatoid-arthritis/