Data from a MSbase registry on patients with multiple sclerosis show in its findings that the incorporation of the Multiple Sclerosis Severity Score improved the prediction accuracy of relapses in MS.
Findings from a data registry on 5866 patients with multiple sclerosis (MS) showed that the addition of the Multiple Sclerosis Severity Score (MSSS) with the MSBase prediction model improved the accuracy of individualized prognostics in MS.1 Previously, the MSBase prediction model did not include MSSS, even though the model itself had multiple demographic and clinical characteristics to estimate hazards of relapses with MS.
The data showed that when MSSS was included in the model, the accuracy of individual prediction of relapses improved by 31%. Additionally, the inclusion on the MSSS improved the confirmed disability accumulation (CDA) by 23%, and of confirmed disability improvement (CDI) by 24% (Harrell C). Notably, there was an increase in the amount of variance explained for relapses by 49%, for CDI by 11%, and for CDA by 10% as compared with the MSBase model alone.
The correspondence for relapse events had a 31% improvement in prediction accuracy. The improvement for relapse events went from 56% (95% CI, 54%-57%) in a baseline model, MSBASE, without MSSS to 87% (95% CI, 86%-88%) in the model that included MSSS. The correspondence index for CDA had a 23% improvement in prediction accuracy. The improvement for CDA went from 63% (95% CI, 61%-66%) for the original model to 85% (95% CI, 83%-88%) in the model version that incorporated MSSS. The correspondence for CDI events showed a 24% improvement, increasing from 67% (95% CI, 63%-71%) to 91% (95% CI, 87%-95%).
The data was pulled from a MSBase registry on January 18, 2019, and the criteria for the study composed of patients with the diagnosis of MS or clinically isolated syndrome based on the 2005 or 2010 revised McDonald criteria. In addition, the patients with MS needed to have a commencement of a disease-modifying therapy (DMT) during the prospectively recorded follow-up, have a minimum pre-DMT follow-up of 6 months, and a minimum on-treatment prospective follow-up of 6 months.
Also, the records needed to display the disability score (EDSS) between 6 months before and 1 month after the index DMT commencement, not within 30 days from a past relapse. Any patients in the registry were excluded if they had an inactive primary progressive of MS. The baseline was defined as the date of a start of a new disease-modifying therapy along with having a recorded EDSS score and also followed by at least 6 months of recorded clinical follow-up. Using the dataset, the MSSS was calculated at all eligible baselines.
In a recent phase 2 study, 6-months of ketogenic diet for people with relapsing multiple sclerosis resulted in a significant reduction in weight, fatigue, and depression, and improved quality of life.
The marginal proportional hazards models were used to evaluate the associations between MSSS and relapse, CDA, and CDI, which were adjusted for three principal components representative of patients’ demographic and clinical characteristics. The models in comparison from with and without MSSS was assessed with penalized r2 and Harrell C. The patients with MS that were started on DMT during prospective follow-up showed as the age 38.4 (±10.6) years, 72% women, and with a disease duration of 8.5 (±7.7) years.
The results confirmed the value for MSSS for individual-level prognostics and expanded the research for understanding MSSS as a group-level predictor of disability outcomes.2-3 Kalincik and colleagues wrote, “We show that—on a group level—hazard of relapses, CDI and, to a lesser extent, CDA are higher in patients with higher MSSS. The increased probability of disability improvement as well as the risk of disability worsening with higher MSSS may appear counter-intuitive.”
Future implications for using the model include reducing the number of input variables, eliminating any extensive data. Research studies in the future should try to find ways to improve prediction over a longer time, as the predictive accuracy of the existing model deteriorates after a longer follow-up.
Furthermore, there should be the inclusion of prespecified analyses of drug efficacy stratified on baseline MSSS in future randomized clinical trials in MS. Kalincik and colleagues noted, “We advocate that disability rank scores be incorporated into future prognostic models of MS.”
