MRI validation of the NINDS-CSN proposed neuropsychological assessment protocol for vascular cognitive impairment

In 2005, the NINDS-Canadian Stroke Network sponsored a workshop to develop epidemiological, pathological, experimental, neuropsychological and imaging standards for the assessment of vascular cognitive impairment (VCI). From this workshop, 60, 30 and 5-minute neuropsychological protocols were developed to assess cognitive function. The current study presents preliminary results on the validation of the 60-minute protocol using MRI measures of macro- and micro-structural integrity of white matter (WM) in patients with recent ischemic stroke. Patients 6 to 18 months post-ischemic stroke were tested on the cognitive assessment protocols and had MRIs of the brain. Macro-structural integrity of WM was assessed using a FLAIR pulse sequence with semi-automated assessment of WM hyperintensities using SABRE. Micro-structural integrity of WM was assessed using diffusion tensor imaging (DTI) determined fractional anisotropy (FA) of radiologically defined normal appearing white matter (NAWM). Relationship of imaging markers to neuropsychological performance measures was assessed in a series of regressions controlling for estimates of pre-morbid IQ. 21 ischemic stroke patients (mean age 69.7 + 11.9 year; mean Barthel score = 87.6 + 19.5; mean MMSE = 27.3 + 3.2) were tested and imaged. The average standardized performance on the 60 minute neuropsychological screening test for the patients was z = -1.5 (range -4.9 to 0.2). The sample's average volume of SABRE determined WM hyperintensities was 11.7 + 21.8 cm3 and NAWM FA was 0.32 + 0.04. Overall, pre-morbid IQ, WM hyperintensities and NAWM integrity significantly predicted performance on the 60-minute protocol (F[3,18] = 8.02, p = .002), accounting for approximately 62% of the shared variance. Higher volumes of WM hyperintensities and lower NAWM FA values were associated with poorer cognitive performance. NAWM integrity and WM hyperintensities independently improved the predictive model over the use of estimated pre-morbid IQ alone (R2 change 0.26 and 0.13 respectively, both p's < .05). These results provide preliminary support for the NINDS-CSN workshop proposed neuropsychological protocol for VCI. Both the volume of FLAIR determined WM hyperintensities and NAWM integrity significantly and independently predicted cognitive performance even while controlling for pre-morbid IQ. This ongoing project is supported by the NIH (NS057514).