Clinical risk stratification: Development and validation of the DAAE score, a tool for estimating patient risk of transition to secondary progressive multiple sclerosis

Tom A Fuchs,Robert Zivadinov,Tetyana Pryshchepova,Bianca Weinstock-Guttman,Michael G Dwyer,Ralph H B Benedict,Niels Bergsland,Dejan Jakimovski,Tomas Uher,Julia R Jelgerhuis,Frederik Barkhof,Bernard M J Uitdehaag,Joep Killestein,Eva M M Strijbis,Menno M Schoonheim

DOI: https://doi.org/10.1016/j.msard.2024.105755

Abstract:Background: Because secondary progressive multiple sclerosis (SPMS) is associated with worse prognosis, early predictive tools are needed. We aimed to use systematic literature review and advanced methods to create and validate a clinical tool for estimating individual patient risk of transition to SPMS over five years. Methods: Data from the Jacobs Multiple Sclerosis Center (JMSC) and the Multiple Sclerosis Center Amsterdam (MSCA) was collected between 1994 and 2022. Participants were relapsing-remitting adult patients at initial evaluation. We created the tool in four stages: (1) identification of candidate predictors from systematic literature review, (2) ordinal cutoff determination, (3) feature selection, (4) feature weighting. Results: Patients in the development/internal-validation/external-validation datasets respectively (n = 787/n = 522/n = 877) had a median age of 44.1/42.4/36.6 and disease duration of 7.7/6.2/4.4 years. From these, 12.6 %/10.2 %/15.4 % converted to SPMS (median=4.9/5.2/5.0 years). The DAAE Score was named from included predictors: Disease duration, Age at disease onset, Age, EDSS. It ranges from 0 to 12 points, with risk groups of very-low=0-2, low=3-7, medium=8-9, and high≥10. Risk of transition to SPMS increased proportionally across these groups in development (2.7 %/7.4 %/18.8 %/40.2 %), internal-validation (2.9 %/6.8 %/26.8 %/36.5 %), and external-validation (7.5 %/9.6 %/22.4 %/37.5 %). Conclusion: The DAAE Score estimates individual patient risk of transition to SPMS consistently across datasets internationally using clinically-accessible data. With further validation, this tool could be used for clinical risk estimation.