Clinical usefulness of genetic information for predicting radiographic damage in rheumatoid arthritis

To determine whether knowledge of genetic information aids the prediction of radiographic damage for patients with rheumatoid arthritis (RA) in whom extensive sociodemographic, family history, clinical, and immunologic information is available.

Subjects included 146 Caucasian women who were participants in a community based longitudinal study of RA. Our primary outcome measure was the severity of erosive disease. Nongenetic covariates included age at RA onset, disease duration, family history of RA, education level, family income, baseline values of function, painful and swollen joint groups and pain rating, and rheumatoid factor positivity. All women were genotyped for the HLA-DRB1 shared epitope (SE) and the tumor necrosis factor a (TNFa) microsatellite. Likelihood ratio tests (LRT) were performed to evaluate the usefulness of genetic information for predicting radiographic damage in RA, after adjusting for nongenetic covariates. Receiver operating characteristic (ROC) curves displaying the sensitivity and specificity of combinations of nongenetic and genetic information were derived, and the areas under the curves (AUC) were compared.

Genetic information contributed significantly to the prediction of radiographic damage in RA even after adjusting for all nongenetic covariates (p value for LRT = 0.0019). The odds ratio describing the risk of severe erosive disease among individuals who had inherited both the SE and TNFa allele 11 (TNFa11) was 7.6 compared to individuals who were SE and TNFa11 negative. Analysis of ROC curves confirmed the usefulness of genetic information.

Genetic information is useful for predicting radiographic damage in RA even for patients in whom extensive sociodemographic, family history, clinical, and immunologic information is available.