Reductions in brainstem volume as a key macrostructural indicator in at-risk populations for Alzheimer's disease.

Background: Alterations to brain macrostructure, assessed via T1-weighted magnetic resonance imaging are observed in preclinical models of Alzheimer's disease (AD), reflecting susceptibility, prodromal stages of AD or correlates of early AD pathophysiology. While changes in cingulate and medial temporal lobe structures may be functionally implicated in cognitive decline, little is known about the viability of brain-based biomarkers that support autonomic functions implicated in preclinical AD risk such as the brainstem.

Methods: In a series of multiple linear regressions, we assess the volume of the brainstem in two asymptomatic at-AD-risk samples, assessed via the presence of either mild cognitive impairment (MCI, N = 148), or extremely high polygenic risk (N = 13) with matched demographics (mean age = 67 [range 58-76], in both cases). We further determine the strength of the association, compared to 150 other structural MRI features.

Results: We observed brainstem volume reductions (MCI: b = -0.29, P = 0.018; Genetic risk: b = -1.29, P = 0.002) in both samples. The magnitude of each preclinical AD marker (MCI / AD-polygenic risk)- brainstem association was empirically larger (Z > 2.3, P < 0.05, in both cases) than 150 frequently segmented MRI features. We further replicate the negative AD-polygenic risk score- brainstem association in UK Biobank (N = 31968; b = -0.002, P = 0.03), with weaker evidence that the association was larger than all other MRI features (Z = 1.622; P = 0.052).

Conclusions: These observations suggest that AD risk, assessed via the presence of MCI or extremely high AD-polygenic risk score is linked to reduced brainstem volume before most typically observed morphological brain alterations. This conforms with evidence implicating the brainstem as one of the earliest sites of morphological neurodegeneration and provides a plausible biological mechanism linking prodromal autonomic symptoms to AD risk in later life. These observations warrant future investigation into the molecular correlates of AD-linked brainstem dysfunction, assessment as a candidate biomarker, and the exploration of brainstem mediated treatment strategies in AD prevention.