PSMD, a novel marker of small vessel disease, and its association with cognitive function in the community: Neuroimaging / Optimal neuroimaging measures for early detection

Background Peak width of skeletonized mean diffusivity (PSMD) has been proposed as a novel neuroimaging marker for small vessel disease (SVD), which contributes to vascular cognitive impairment and dementia. Although this marker has been related to lower cognitive function in clinical samples with inherited and sporadic SVD, additional research is needed to assess its utility for risk prediction in large, well‐characterized population‐based cohorts. Methods We included 929 stroke‐ and dementia‐free participants from the Offspring cohort of the Framingham Heart Study. PSMD was derived from DTI using an automated algorithm, and further log‐transformed to normalize its distribution. To evaluate general cognitive function, we created a composite cognitive score including Logical Memory – immediate and delayed recall, Digit Span Forward and Backward, Category Fluency – animals, Phonemic Fluency – FAS, Trail Making Test A and Test B, and the Boston Naming Test. We used linear regression models to relate PSMD to general cognitive function adjusting for age, sex and educational level. In a secondary model, we additionally adjusted for diabetes, smoking and hypertension. Finally, we included white matter hyperintensity (WMH) volume as percentage of total intracranial volume in the primary and secondary models to assess the contribution of PSMD beyond WMH. Results Higher PSMD values were associated with lower general cognitive function (Beta (SE), ‐0.61 (0.10), p<0.001). Results remained virtually unchanged after additional adjustment for vascular risk factors in the secondary model (‐0.60 (0.10), p<0.001). After the inclusion of WMH, associations remained significant in the primary (‐0.58 (0.11), p<0.001) and secondary (‐0.57 (0.11), p<0.001) models, with PSMD explaining an additional 2% of the variance in cognition beyond WMH. Conclusion PSMD was strongly associated with general cognitive function in our sample, suggesting that this novel marker may be suitable for risk prediction for SVD in community‐based samples, and is potentially a better predictor than well‐established markers such as WMH. Additional validation and longitudinal analyses are currently underway in population‐based cohorts and clinical samples in the context of the MarkVCID consortium.