Estimating individual treatment effect on disability progression in multiple sclerosis using deep learning.

Disability progression in multiple sclerosis remains resistant to treatment. The absence of a suitable biomarker to allow for phase 2 clinical trials presents a high barrier for drug development. We propose to enable short proof-of-concept trials by increasing statistical power using a deep-learning predictive enrichment strategy. Specifically, a multi-headed multilayer perceptron is used to estimate the conditional average treatment effect (CATE) using baseline clinical and imaging features, and patients predicted to be most responsive are preferentially randomized into a trial. Leveraging data from six randomized clinical trials (n = 3,830), we first pre-trained the model on the subset of relapsing-remitting MS patients (n = 2,520), then fine-tuned it on a subset of primary progressive MS (PPMS) patients (n = 695). In a separate held-out test set of PPMS patients randomized to anti-CD20 antibodies or placebo (n = 297), the average treatment effect was larger for the 50% (HR, 0.492; 95% CI, 0.266-0.912; p = 0.0218) and 30% (HR, 0.361; 95% CI, 0.165-0.79; p = 0.008) predicted to be most responsive, compared to 0.743 (95% CI, 0.482-1.15; p = 0.179) for the entire group. The same model could also identify responders to laquinimod in another held-out test set of PPMS patients (n = 318). Finally, we show that using this model for predictive enrichment results in important increases in power.

Large, nonplateauing relationship between clinical disability and cerebral white matter lesion load in patients with multiple sclerosis.

OBJECTIVE: To better characterize the relationship between cerebral white matter lesion load (CWM-LL) and clinical disability by (1) covering the entire range of the Kurtzke Expanded Disability Status Scale (EDSS), (2) minimizing nonbiological sources of variability, and (3) increasing pathologic specificity by studying CWM lesions that are hypointense on T1-weighted magnetic resonance imaging. DESIGN: Cross-sectional, retrospective study. SETTING: Hospital-based multiple sclerosis (MS) clinic. Patients  A total of 110 patients with untreated MS were recruited and studied from June 1, 1997, through June 30, 2003. MAIN OUTCOME MEASURES: Cube-rooted CWM-LL and EDSS-measured clinical disability scores. RESULTS: We found a large, nonplateauing relationship between cube-rooted CWM-LL and concurrent EDSS scores, more so for T1-hypointense than T2-hyperintense lesions (r = 0.619 vs 0.548). Correlations between the EDSS scores and CWM-LL diminished when, as typically done in clinical trials, only those patients with EDSS scores of 0 to 6.0 were studied (n = 92; r = 0.523 for T1-hypointense lesions and r = 0.457 for T2-hyperintense lesions); more important, a series of boot-strapped correlations suggested that this decrease was not simply due to smaller sample size, and these relationships remained even after correcting for disease duration. CONCLUSION: A large, nonplateauing relationship exists between CWM-LL and EDSS-measured clinical disability when patients with MS are studied to examine the entire range of disability, minimize nonbiological sources of variability, and increase pathologic specificity.