Pooled historical MRI data as a basis for research in multiple sclerosis--a statistical evaluation.

Area of Science:

• Neuroimaging
• Clinical Trials
• Biostatistics

Background:

• Pooled placebo data are valuable for historical controls and power calculations in therapeutic studies.
• Investigating heterogeneity in placebo arm data is crucial for assessing database utility.
• The Sylvia Lawry Centre for Multiple Sclerosis Research maintains a database of controlled studies.

Purpose of the Study:

• To assess the usefulness of pooled placebo arm data from multiple sclerosis (MS) trials as historical controls.
• To investigate the degree of heterogeneity in magnetic resonance imaging (MRI) measures within the pooled placebo data.
• To determine if adjustments for population differences can mitigate heterogeneity.

Main Methods:

• Analysis of pooled placebo arm data from 14 controlled studies.
• Adjustment of magnetic resonance imaging (MRI) measures for differences in study populations.
• Stratification analysis to account for heterogeneity.

Main Results:

• Significant heterogeneity was observed in MRI measures across pooled placebo groups, even after population adjustments.
• Heterogeneity was substantially reduced when comparing studies from the same image analysis center.
• The pooled database remains useful for analyzing inter-variable relationships when heterogeneity is explicitly managed.

Conclusions:

• Pooled placebo data offer utility for statistical analyses but require careful handling due to inherent heterogeneity.
• Explicitly accounting for heterogeneity as a stratification factor is essential for valid comparisons.
• Comparing new treatment groups with pooled data necessitates caution due to potential variations in MRI measures.