Objective:

To extend 2-stage models of heterogenous treatment effect (HTE) to brain atrophy outcomes utilizing the MS PATHS research network.

Background:

2-stage HTE models derived from real-world datasets have the potential to advance personalized medicine in MS. Brain atrophy provides an objective measure benefiting from highly standardized MRI capture.

Design/Methods:

Analyses included patients in MS PATHS who initiated DMT, had complete baseline data, and ≥1 year follow-up without switching DMT. DMT groups were based on efficacy: high (natalizumab, anti-CD20 monoclonal antibodies, alemtuzumab), moderate (fumarates, S1PR modulators, and cladribine) and low (teriflunomide, interferons, glatiramer acetate). Patients were split into training (70%) and test (30%) sets, stratified by group. In stage one, baseline brain parenchymal fraction (BPF) change risk models were derived by linear mixed effect models with baseline covariates as inputs; performance was assessed by predicted versus actual BPF change R-squared in the test set. In stage 2, propensity score (PS) weighting was performed, using multinomial logistic regression with baseline BPF change risk score as one of the covariates. Average treatment effect (ATE) and HTE were calculated with low-efficacy DMT as the reference.

Results:

Analyses included 594 high-, 184 moderate-, and 108 low-efficacy DMT-treated patients. Risk model performance R-squared=0.56. After PS adjustment, covariate SMDs were <10%. In the ATE model, moderate- and high-efficacy groups had less brain atrophy than the low-efficacy group (p<0.001). In the HTE model, BPF change was similar across groups for patients with low risk of brain atrophy. In patients with high risk of brain atrophy, high- and moderate-efficacy groups remained similar; the low-efficacy group had 2-fold worse BPF change. 

Conclusions:

HTE models suggest individuals with low baseline risk of brain atrophy may achieve similar outcomes regardless of DMT efficacy choice. The worst relative outcomes occur when patients with high brain atrophy risk are treated with low-efficacy DMTs.