Tumor necrosis factor markers show sex-influenced association with rheumatoid arthritis.

Abstract

The observation that not all shared-epitope genotypes confer the same risk suggests that a second HLA-region locus may confer risk. Tumor necrosis factor alpha (TNFgamma) is a possible candidate. We examined TNFalpha for sex influences on HLA-associated risk for rheumatoid arthritis (RA). DRB1 and TNF microsatellite typing of 297 Caucasian RA patients (132 men, 165 women) and 267 Caucasian controls was performed. The TNFab microsatellite haplotype distribution differed among the male RA, female RA, and control groups (P < 0.01); the difference was largely an excess of TNFa2b1 haplotypes in the male RA group. However, this did not simply reflect an excess of shared-epitope haplotypes bearing TNFa2b1. In RA, not all shared-epitope-bearing haplotypes had the same TNFab. The *0401-bearing haplotypes commonly had TNFa6b5, TNFa2b1, TNFa10b4, and TNFa11b4, while the *0404-bearing haplotypes had TNFa11b4. In the female RA group, TNFa2b1 was most often on *0401-bearing haplotypes. In the male RA group, there was a surprise: TNFa2b1 was often on HLA haplotypes without shared-epitope DRB1 alleles. To estimate the relative strength of associated HLA markers, we performed logistic regression analyses stratified by sex and controlling for a potential confounder, age at disease onset. Among women, TNFa2b3 favored RA (odds ratio 1.932, P < 0.05) while TNFa6b5 was protective (odds ratio 0.522, P < 0.05). Among males, TNFa2b1 and TNFa11b4 conferred elevated odds ratios (2.58 and 1.681, respectively, P < 0.05). However, the odds ratios for TNFa2b1 in men and TNFa2b3 in women were generally well below those for RA-associated DRB1 markers (for example, DRB1*0401 3.553 in male RA patients and 6.991 in female RA patients). Certain TNFab-bearing HLA haplotypes modify RA risk in a manner influenced by sex but independent of DRB1, particularly TNFa2b1 in men.