Peripheral Biomarker Signatures in Rheumatoid Arthritis-Associated Interstitial Lung Disease

Abstract

Objective: Rheumatoid arthritis-associated interstitial lung disease (RA-ILD) is a significant cause of morbidity and mortality among patients with RA, yet effective risk stratification for RA-ILD is lacking. We sought to characterize unique peripheral blood biomarker signatures in RA that could improve RA-ILD discrimination beyond clinical and genetic risk factors.

Methods: We performed a cross-sectional study of participants in the Veterans Affairs Rheumatoid Arthritis Registry. ILD was validated through systematic record review. Autoantibodies (n = 14), proinflammatory cytokines/chemokines (n = 36), adipokines (n = 3), alarmins (n = 3), and matrix metalloproteinases (n = 9) were measured from banked serum or plasma. MUC5B rs35705950 genotyping was obtained via microarray. Principal component (PC) analysis was performed to derive unique biomarker signatures. Logistic regression was used to compare models predicting RA-ILD presence using combinations of clinical, peripheral biomarker, and genetic variables.

Results: Among 2,001 participants with RA (88.7% male, mean age 63.7 years), 121 (6.4%) had RA-ILD. A total of 15 PCs were identified, with 8 significantly associated with RA-ILD after adjusting for clinical factors. These eight included biomarker themes of innate and allergic responses, autoantibodies (anti-cyclic citrullinated peptide, rheumatoid factor, anti-malondialdehyde-acetaldehyde adduct), adipokines, alarmins, tissue remodeling, and neutrophil chemotaxis. Models with PCs (area under the receiver operating characteristic curve [AUC] 0.739) and PCs with MUC5B (AUC 0.751) outperformed clinical risk factors alone (AUC 0.630; P < 0.001).

Conclusion: Peripheral biomarker signatures are associated with RA-ILD and improve RA-ILD identification beyond clinical risk factors. In addition to demonstrating the potential for peripheral biomarkers to aid RA-ILD risk stratification, these findings suggest diverse pathways are involved in RA-ILD pathogenesis.

Figure 1

Receiver operating characteristic curves for logistic regression models for rheumatoid arthritis–associated interstitial lung disease including clinical risk factors, principal components, and MUC5B promoter variant. Receiver operating characteristic curves, along with their respective area under the receiver operating characteristic curve (AUC), are shown in each panel for clinical model alone, clinical model with principal components analysis (clinical + PCA), and clinical model with PCA and MUC5B promoter variant (clinical+PCA+MUC5B). (A) Clinical risk factors include age, sex, race and ethnicity, and smoking history. (B) Expanded (Exp.) clinical risk factors include the above plus body mass index, Disease Activity Score With 28‐Joint Count, prednisone use, conventional disease‐modifying antirheumatic drug (DMARD) use, and biologic or targeted synthetic DMARD use. All model comparisons were significantly different by nested likelihood ratio test (P < 0.001).