Prediction of Annualized Relapse Rate at First Clinic Visit Among Patients with Multiple Sclerosis
Evan Madill1, Brian Healy2, Mariann Polgar-Turcsanyi1, Tanuja Chitnis1
1Neurology, Brigham and Women's Hospital, 2Biostatistics Center, Massachusetts General Hospital
Objective:
To model relapse likelihood over the next three years at the time of a patient’s first clinic visit (FV) for multiple sclerosis (MS).
Background:
Annualized relapse rate (ARR) and treatment considerations vary greatly amongst individuals with MS. Although there are known prognostic factors at a population-level, predicting the risk of relapse for individuals remains difficult, particularly at initial presentation.
Design/Methods:
We queried records from CLIMB, a longitudinal study of MS at Brigham and Women’s Hospital. Demographic and clinical characteristics were evaluated. Adult patients with relapsing-onset MS and at least three years of follow-up who were not treated with high-efficacy therapy prior to FV were included. We modeled the three-year relapse risk for those who started glatiramer or interferon after FV between 1997 and 2010 using multiple regression in Stata.
Results:
1086 patients were included (1017 with relapsing-remitting and 69 with secondary progressive MS [SPMS]). 74.9% were female and 91.5% were white. Mean age was 40.0 (SD 10.2) and median disease duration was 5.1 years (IQR 1.4 to 11.6). Median EDSS was 1.5 (IQR 1 to 2.5). ARR during the first three years after FV was 0.27. A multiple regression model predicted 10.3% of the variance in relapses. Younger age, non-SPMS diagnosis, and relapses in the year before FV were significantly predictive of relapses after FV.
Conclusions:
We model the three-year ARR for a real-world sample of relapsing-onset MS patients at the time of their first clinic visit. The sample was limited to patients on low-efficacy therapy prior to 2010 to reduce confounding from treatment. While age and progressive disease were inversely correlated with relapses, as expected, most of the variance in relapses was unexplained. Relapses in the year prior to FV, but not total relapses, were correlated with future relapses. Whether relapses ‘cluster’ is an area for future study.