Scientists at the National Institute of Arthritis and Musculoskeletal and Skin Diseases (NIAMS), working with researchers at sites across the country, have employed a new statistical method to identify predictors of spine damage severity in people with ankylosing spondylitis (AS). These predictors include a number of genetic markers, as well as being older at disease onset, being male, and being a smoker. The findings were published in the July issue of Arthritis & Rheumatism.

Ankylosing spondylitis is an inflammatory arthritis of the spine. It is characterized by joint stiffness, pain and extra bone growth that can result in partial or complete spinal fusion. Spine damage, as seen on x-ray, is one of the measurable features of AS that most greatly impacts quality of life and daily activities.

To reduce a field of 155 candidate variables that the research team thought might be predictors into a more promising subset for further study, Michael Ward, M.D., an investigator in the NIAMS Intramural Research Program, and his colleagues used a complex statistical technique called the random forest. This divides a sample population into subgroups based on random combinations of the candidate variables, and compares the resulting classification schemes — or "trees" — against real data, which reveals which variables, if any, truly have predictive value.

The results of the random forest approach revealed a number of highly ranked factors, both clinical and genetic, that seemed predictive of spine damage as seen on x-ray. Ward's team then used a more conventional statistical approach, logistic regression, to determine the strength of the association between a predictor and spine damage severity. They found that the likelihood of having severe damage increased with age of disease onset, and men were twice as likely as women to be in the group. Current smokers were more than four times as likely to have severe damage as nonsmokers.

Of the genetic markers found in the AS patient population, many associated with immune system genes were predictive of disease severity. Of note, HLA-B27, which is a genetic marker strongly associated with susceptibility to AS, was not associated with disease severity. Interestingly, the random forest approach identified a genetic marker, DRB1*0801, that seems protective. Study participants with this marker were dramatically less likely to have the most severe spine damage.

Because changes to the spine progress slowly, over years, the study was limited to participants whose disease had been active for at least 20 years. It included more than 400 participants from the National Institutes of Health in Bethesda, Maryland, from Cedars-Sinai in Los Angeles, and from the University of Texas at Houston.

The researchers hope that giving people with AS a better understanding of their risk for severe spine damage will help them take advantage of the factors that can be controlled, such as not smoking and following their treatment plans.

The mission of the National Institute of Arthritis and Musculoskeletal and Skin Diseases (NIAMS), a part of the Department of Health and Human Services' National Institutes of Health (NIH), is to support research into the causes, treatment, and prevention of arthritis and musculoskeletal and skin diseases; the training of basic and clinical scientists to carry out this research; and the dissemination of information on research progress in these diseases. For more information about NIAMS, call the information clearinghouse at 301-495-4484 or 877-22-NIAMS (free call) or visit the NIAMS Web site at

# # #

Ward MM, Hendrey MR, Malley JD, Learch TJ, Davis JC Jr, Reveille JD, Weisman MH. Clinical and immunogenetic prognostic factors for radiographic severity in ankylosing spondylitis. Arth & Rheum. 2009 July 15;61(7):859-66.

Last Reviewed: