Antemortem‐postmortem correlates of 4D‐Flow MRI in cerebral vessels

Abstract

AbstractBackgroundCerebrovascular disease (CVD) is associated with increased dementia risk and often co‐occurs with Alzheimer’s disease (AD). Non‐invasive quantitative 4D‐Flow MRI can be used to characterize cerebrovascular hemodynamics and study CVD/AD interactions during life, but has not been validated against postmortem pathological assessment. This study compares antemortem 4D‐Flow metrics with postmortem pathological assessment in cerebral arteries to inform the interpretation of 4D‐Flow MRI.MethodsThree participants from the Wisconsin ADRC underwent antemortem 4D‐Flow and postmortem pathological examination. Subjects had MCI or dementia before death with suspected AD etiology (Table 1). Antemortem 4D‐Flow was acquired on a 3.0T system (GE Healthcare). Vessel cross‐sections from 4D‐Flow angiograms of cerebral arteries were used to estimate blood flow, pulsatility index (PI), and luminal area. Upon death, brain extraction and neuropathological examination were performed according to NIA‐AA guidelines. Sampled cerebral arteries included left anterior (A1) and middle (M1) cerebral, and left internal carotid arteries, basilar artery, and left and right posterior cerebral (P1 and P2) and vertebral arteries. Mason’s trichrome stained sections of the arteries were captured using an Aperio Digital Slide Scanner. QuPath software was used to delineate the intima, media and functional luminal area of the vessels. A pixel classifier measured the area of smooth muscle cells (SMC) and fibrous tissue (Figure 1). The percentage of stenosis (%stenosis) was calculated as the area of atherosclerotic plaque divided by the area encompassed by the elastica. The percentage of media fibrosis (%fibrosis) was calculated as the area of fibrotic tissue divided by the sum of the SMC and fibrotic areas within the media. Spearman correlations were used to compare antemortem and postmortem vascular measures without correction for multiple comparisons.ResultsImaging‐postmortem correlates were highest for PI vs. %fibrosis (ρSpearman=0.42), mean flow vs. %stenosis (ρSpearman=0.40), and comparisons of luminal area (ρSpearman=0.43) (Table 2, Figure 2). Correlations between other imaging and pathology measures were low (|ρSpearman|<0.21).ConclusionThese data provide preliminary evidence that 4D‐Flow metrics correlate with underlying arterial vessel stenosis and fibrosis. Future work is ongoing to investigate antemortem‐postmortem associations in more cases.