Background/Purpose: Dysregulations in androgenic-anabolic (A-A) steroids and cytokines are recognized in RA and pre-RA subjects (Rheum Dis Clin N Am 2005; 31: 131-60). However, deviations in correlations of these interacting systems have not been reported. This study analyzed baseline serum levels of the IL-1 profile (IL-1β and IL-1ra), A-A steroids, and cigarette smoking in a cohort of pre-RA and CN subjects.

Methods: The community-based cohort enrolled 21,061 adults (12,381 F, 8,680 M) in 1974. Over 3-20 (median 12) yrs, 54 (36 F, 18 M) cases developed RA by 1988 ACR criteria. Four CN who did not develop RA were matched to each case on age (±2 yrs), sex, and race. Baseline stored (-70˚C) sera were assayed in pre-RA-CN sets at national referral laboratories without knowledge of status. IL-1β and IL-1ra were assayed using ELISA immunoplates (R&D Systems Inc., Minneapolis, MN) and A-A steroids by RIA. Multiple imputation and aggregate methods were used to enter values for a minority of randomly missed test results. Biomarkers were normalized using z-scores within sexes to adjust for any dimorphism. Logistic regression searched for independent predictors of dependent lower vs higher IL-1β z-score subgroups. Predictors of A-A levels were identified by linear regression. Frequency distributions were determined for the dichotomous IL-1β subsets and the pre-RA vs CN subgroups with A-A steroids, IL-1ra, and baseline cigarette smoking (7-scale), and evaluated by Fisher’s exact test.

Results: In 257 total subjects (54 pre-RA, 203 CN), lower (n=165) vs higher (n=92) IL-1β z-score subgroups strongly (p < 0.0001) associated with combined RA and CN lower vs higher IL-1ra frequencies (Table 1). Logistic regression confirmed that only IL-1ra levels predicted the IL-1β dichotomy, including 6 other variables in the model (Table 2). In 165 lower IL-1β subjects (36 pre-RA, 129 CN), pre-RA had significant deficits of higher Δ4A (p < 0.0001) and T (p < 0.001) vs CN (Table 1). Linear regression confirmed that the dependent A-A steroid levels were predicted by the pre-RA vs CN status at strengths equivalent to known negative correlations with age. In the 92 higher IL-1β subgroup, 9 (50%) of 18 pre-RA vs 8 (10.8%) of 74 CN had higher cigarette usage (p < 0.001), consistent with an inflammatory association of smoking and RA risk.

Conclusion: Androstenedione and testosterone levels were significantly lower in pre-RA than CN subjects who had lower IL-1β levels, whereas cigarette usage was respectively greater in the RA subjects who had higher IL-1β levels. The new findings support neuroendocrine immune interactions in the risk of developing RA.