Background/Purpose: While some studies have linked air pollution to risk for rheumatoid arthritis (RA), U.S.-based studies utilizing nationwide exposure data remain limited and often exclude the combined effect of genetic factors. Given the rising pollution level, it is essential to integrate environmental and genetic factors to better identify individuals at risk for RA and to inform prevention strategies.

Methods: This case-control study used data from NIH All of Us v8 (AoU). We included participants with short read whole genome sequencing (srWGS) data. A case of RA was defined by three criteria: a) 3 or more encounters with RA-related ICD-9/10 codes, b) prescription of RA-specific medications, c) exclusion of patients with other concurrent autoimmune diseases. Incident RA cases required at least 1 year of electronic health record (EHR) preceding their earliest RA diagnosis. Control subjects did not meet any of the three criteria and had at least 1 year of EHR before their last recorded ICD codes. Cases and controls were 1:10 frequency-matched on age, sex, race, and the duration from the earliest to the latest recorded encounter. Genetic risk was quantified by a polygenic risk score (PRS) based on 92 RA-related genetic variants. Environmental exposures were assessed using annual average air pollutant concentrations from Environmental Protection Agency AirData (1980–2024), aggregated at the 3-digit ZIP code level. Only exposure data preceding RA diagnosis for cases and prior to the last condition entry for controls were included. Variables selected by the LASSO and Elastic Net Cox models using 5-fold cross-validation were used to fit the final Cox model, adjusting for demographics and smoking history. Integrated Area Under the Curve (iAUC) was computed by the weighted average of time-dependent AUCs with 95% bootstrap confidence intervals (CI) in R v4.4.0.

Results: 283 cases of incident RA and 2,830 matched controls were included. PRS was significantly associated with increased risk of incident RA (HR = 1.25, CI: 1.12-1.40). Among air pollutants, lead showed the strongest association (HR = 6.76, CI: 2.00-22.82), followed by particulate matter (PM2.5) (HR = 1.38, CI: 1.30-1.46), SO2 (HR = 1.10, CI: 1.06-1.15) and ozone (HR = 1.1, CI: 1.03-1.17). Conversely, NO2 exposure was associated with a reduced RA risk (HR = 0.93, CI: 0.90-0.96). Higher income levels and more advanced education were linked to reduced risk of RA incidence. Smoking history and ambient pressure showed no significant association. The combined model of PRS and air pollutants achieved the highest integrated AUC (iAUC = 0.72, CI: 0.70-0.76), outperforming models with PRS only (iAUC = 0.66, CI: 0.64-0.70).

Conclusion: Integrating PRS with air pollutants enhances the prediction of RA incidence. Notably, lead and PM2.5 exhibited the strongest associations with increased risk for RA. These findings underscore the importance of combined genetic and environmental assessments in identifying individuals at elevated risk for RA.

Figure 1. Time-Dependent AUC curves comparing the predictive performance of PRS-only, air pollutants-only, and the combined models for RA risk, adjusted for demographics and smoking history. All models exhibited a general decline in AUC over time. The combined model and air pollutants-only model showed notable decreases around years 10 and 20, whereas the PRS-only model maintained a relatively stable AUC throughout the follow-up period. *PRS: Polygenic Risk Score; AUC: Area Under Curve.

« Back to ACR Convergence 2025

ACR Meeting Abstracts - https://acrabstracts.org/abstract/genetic-and-environmental-risk-factors-and-incident-rheumatoid-arthritis/