Comorbidities for Predicting Progression Independent of Relapse Activity in Multiple Sclerosis Treated With B-Cell Depletion.

BACKGROUND

Progression independent of relapse activity (PIRA) has been shown to account for a majority of the disability accumulation in relapsing-remitting multiple sclerosis (RRMS) patients treated with disease-modifying therapies. While comorbidities are common in MS and linked to disability progression, their role in PIRA remains unclear. We investigated whether comorbidities at treatment initiation could predict PIRA in RRMS patients receiving B-cell depleting therapy.

METHODS

This population-based cohort study included RRMS patients initiating rituximab between August 2010 and April 2019, without relapses during six-year follow-up. PIRA was defined as confirmed disability worsening (Expanded Disability Status Scale). We performed hypothesis-driven analysis of pre-specified comorbidities and data-driven analysis of all ICD-10 codes from secondary care. Comorbidity-PIRA associations were assessed using adjusted logistic regression. Predictive performance for elastic net, random forest, XGBoost, and neural networks was evaluated as the area under the curve (AUC) and Brier score through nested cross-validation.

RESULTS

Among 2837 patients, 563 (20%) experienced PIRA within 6 years. The most prevalent comorbidities were depression/anxiety (36%), hypertension (15%), and headache (8%). No pre-specified comorbidities were statistically significant predictors of PIRA. Among all diagnosis codes, only neuromuscular bladder dysfunction reached significance after multiple comparison correction. Predictive models demonstrated modest performance (AUC 0.563-0.652) and adding comorbidities did not improve prediction.

CONCLUSIONS

In this rituximab-treated RRMS cohort, comorbidities at therapy start were not associated with higher PIRA risk when accounting for MS-associated disability. The association with bladder dysfunction likely reflects spinal tract engagement rather than an independent comorbidity effect.