Personalized Treatment Response in Progressive MS: Can the Patient's Profile Influence the Outcome?

TL;DR

This study shows how patient profiles can help predict treatment outcomes in multiple sclerosis, supporting personalized therapy approaches.

Contribution

A novel statistical method was developed to predict individual treatment responses in MS using baseline patient characteristics.

Findings

An individualized response score successfully categorized SPMS patients as responders or non-responders to Siponimod.

Responders showed significantly better outcomes on EDSS and SDMT compared to non-responders.

The method revealed variability in treatment effects across different endpoints in MS patients.

Abstract

Evidence from clinical trials providing average effects in populations is often used to forecast individualized patient outcomes similar to the trial patients. Multiple sclerosis (MS), known for notable heterogeneity in outcomes, makes the evaluation of potential heterogeneity of treatment effect (HTE) significant. Identifying factors that predict individual treatment response is crucial for optimizing patient care, and this study aimed to demonstrate the feasibility (proof of concept) of applying a statistical method to predict individual treatment response in MS trials.

We developed an individualized response score (RS) to predict treatment response in patients with active secondary progressive MS (SPMS). The RS was a continuous combination of baseline clinical characteristics, including age, sex, previous relapses, EDSS, and disease duration. We used data from the EXPAND trial to train and validate the RS. A training dataset (70% of the data) was used to identify optimal response thresholds for four key outcomes: Expanded Disability Status Scale (EDSS), Timed 25 Foot Walk (T25FW), 9‐Hole Peg Test (9HP), and the Symbol Digit Modalities Test (SDMT). The remaining 30% of the data served as a validation set to assess the RS's predictive performance. The continuous RS was binarized (into responder and non‐responder) based on the threshold representing the top 25% versus the bottom 75% of the continuous score distribution.

Using baseline profiles, SPMS patients exhibiting varying benefits from Siponimod across different outcomes were successfully categorized as responders or non‐responders. The overall effect of Siponimod on the EDSS was HR = 0.79 (95% CI: 0.65‐0.95), while responders’ demonstrated a HR = 0.64 (95% CI: 0.49‐0.84) versus a HR = 0.97 (95% CI: 0.74‐1.27) for non‐responders’, interaction p = 0.027. Siponimod's overall effect on SDMT progression was HR = 0.75 (95% CI: 0.63‐0.88). Responders' demonstrated a HR = 0.59 (95% CI: 0.43‐0.80) vs a HR = 1.00 (95% CI: 0.69‐1.44) for non‐responders, interaction p = 0.031. On the entire dataset, Siponimod exhibited a non‐significant effect on 9HPT (HR = 0.86, 95% CI: 0.66‐1.10) and on T25FW (HR = 0.95, 95% CI: 0.81‐1.12), whereas responders’ demonstrated a HR = 0.68 (95% CI: 0.47‐0.97) on 9HPT and a HR = 0.77 (95% CI: 0.60‐0.98) for T25FW.

This analysis demonstrated the ability to define responders to a therapy based on their baseline profile and evaluate the treatment effect on multiple endpoints, showing that the benefit on different outcomes can vary across patients.

Multiple sclerosis (MS) is a highly heterogeneous disease, making it challenging to predict individual treatment responses. This study explored the feasibility of a statistical approach to identify patients most likely to benefit from treatment in clinical trials. Using data from the EXPAND trial, we developed an individualized response score based on baseline characteristics to predict Siponimod's effect in secondary progressive MS. Patients were categorized as responders or non‐responders, revealing significant variations in treatment outcomes across key measures. These findings highlight the potential for personalized treatment strategies, allowing for more tailored and effective therapy selection in MS.

Figures