Arterial elastance is a heritable and causal vascular endotype of hippocampal atrophy and cognitive decline

Background - Cerebral amyloid-β (Aβ) is a hallmark of Alzheimer’s disease (AD). Post-mortem evidence suggests myocardial Aβ accumulation in AD, raising the hypothesis of an amyloid-mediated heart–brain axis. Increased arterial load has also been implicated in neurodegeneration; yet the vascular endotype most strongly linked to AD-sensitive hippocampal atrophy and cognitive decline remains uncertain. Methods We conducted a two-stage imaging–genetic investigation. Sixteen patients with mild cognitive impairment underwent simultaneous brain–heart 18F-florbetaben PET-MRI to quantify cerebral and myocardial Aβ and assess associations with cardiovascular magnetic resonance (CMR) phenotypes, including effective arterial elastance (Ea) and aortic distensibility (AoD). In 21,673 UK-Biobank participants, multivariable linear regression tested associations of Ea and AoD with brain volumes and fluid intelligence. Mediation analysis evaluated hippocampal volume as a mediator of the Ea–cognition relationship. Genome-wide association analysis (GWAS) of Ea was conducted in European-ancestry individuals followed by functional mapping (FUMA). Two-sample Mendelian randomisation tested the causal effect of genetically predicted Ea on hippocampal atrophy (ENIGMA). Results Cortical and myocardial Aβ uptake were unrelated, and cortical Aβ showed no association with CMR-derived myocardial extracellular volume or diastolic function. Instead, higher Ea correlated with greater cortical Aβ uptake (ρ=0.577, P<0.05), smaller hippocampal volume and worse cognition. In UK-Biobank, elevated Ea was independently associated with reduced hippocampal and subcortical nuclei volumes and lower fluid intelligence; hippocampal atrophy mediated ~20% of Ea-cognition relationship. By contrast, AoD demonstrated weaker or paradoxical associations with brain architecture and no relationship with fluid intelligence (Figure 1). GWAS identified six independent loci for Ea, led by SH2B3 (P=3.3×10⁻¹⁴), a regulator of blood pressure and vascular inflammation. Functional mapping prioritised cardiometabolic and immune genes, with enrichment of inflammatory and MAPK-related signalling pathways and adipose tissue expression, supporting cardiometabolic inflammatory modulation of arterial load (Figure 2). Mendelian randomisation showed a significant negative causal effect of genetically predicted Ea on hippocampal volume (IVW β=−1.15, P=0.001). Conclusion Cerebral Aβ accumulation in MCI is dissociated from myocardial amyloid deposition, challenging the concept of amyloid-mediated heart–brain axis. Instead, effective arterial elastance emerges as a heritable and genetically regulated vascular endotype, driven by cardiometabolic inflammatory pathways and causally linked to Alzheimer-sensitive hippocampal atrophy and cognitive impairment. Targeting arterial load therefore represents a scalable, mechanistically grounded strategy to modify the trajectory of neurodegeneration.