Transcatheter aortic valve replacement (TAVR) can produce iatrogenic embolic ischemia in the brain, and diffusion MRI lesions of varying size are seen in most TAVR patients. Lesion volumetry is used as a safety outcome and as an endpoint in trials of embolic protection devices. However, unlike large lesions, very small TAVR-related lesions are more likely to reverse at later timepoints, and have unclear clinical impact. Therefore, inclusion of particularly small lesions in endpoints may skew results and obscure assessment of device efficacy. Unfortunately, there is no established minimum clinically meaningful lesion volume.

We performed a post-hoc analysis of data from all subjects with MRI in Phases 1 and 2 of the REFLECT study (n=211 intervention, n=106 control). For each subject, total supra-threshold cerebral ischemic lesion (SCIL) volume was calculated at each possible threshold from 0 to 1000mm³ in 100mm³ increments, by summing the volume of all supra-threshold individual lesions (excluding lesions smaller than the given threshold). At each threshold, the relationship between SCIL volume and pre/post-procedure NIH stroke scale/score (NIHSS) change was assessed and compared across and between treatment arms.

SCIL volume vs. NIHSS correlations ranged from r=0.3-0.378 in the combined arms, r=0.256-0.557 in controls, and r=0.151-0.388 in treatment. In both combined and treatment groups, maximum correlation was at a 300mm³ threshold. In controls, maximum correlation was more than twice as high at a 1000mm³ threshold compared to the inclusion of all lesions (r=0.557 vs. r=0.256, p=0.004). 

Whole-brain lesion volumetry on DWI related more strongly to clinical outcome  when very small lesions are eliminated from analyses. In particular, inclusion of lesions below 300mm³ in quantitative endpoints may be counterproductive and may actually obscure clinical relevance.