Causal models show simvastatin works in progressive MS without lowering cholesterol

Clinical trials tell you whether a drug works. They are much less good at telling you how — and without that, it is hard to know which patients to give it to, or what to measure next.

This PNAS paper, first-authored by Dr Arman Eshaghi, applied structural equation models to the MS-STAT trial, in which 140 people with secondary progressive MS were randomised to simvastatin or placebo and followed for two years with brain MRI, cognitive and physical disability assessment, and serum cholesterol measurement. The question was whether the drug's effect on brain atrophy and disability was mediated by lowering cholesterol, or ran through some other route.

Two competing models were tested head to head, and the cholesterol-independent model was the more likely. Decomposing the treatment effect showed simvastatin acting directly on disability — accounting for 69% of the overall effect on EDSS — and separately on brain atrophy, which in turn was responsible for a further 31%. In other words, the benefit does not travel through serum cholesterol at all, and brain atrophy is a genuine intermediate on the causal path rather than an incidental correlate. That has direct consequences for how successor trials should be designed and which outcome measures are worth powering for.

Read the paper: Applying causal models to explore the mechanism of action of simvastatin in progressive multiple sclerosis.