The heart-brain geometry paradox: adaptive growth vs maladaptive remodelling

This commentary refers to ‘Left ventricular geometry, brain architecture, and cognition: an observational study’, by P. G. Masci et al., https://doi.org/10.1093/eurheartj/ehaf722 and the discussion piece ‘Left ventricular geometry and brain health: interpreting a complex association’, by P. G. Masci, https://doi.org/10.1093/eurheartj/ehag376. Masci et al. used the UK Biobank imaging cohort to ascertain relationships between the heart and brain using a novel latent index of left ventricular geometry (gGeom), shifting focus from haemodynamics to heart structure. A notable finding is the observed correlation between gGeom and increased volumes of the hippocampus and subcortical structures. However, a stronger correlation was found between gGeom and the presence of white matter hyperintensity (WMH) and microstructural damage. This paradox is intriguing but likely stems from a pervasive methodological shortcoming: confounding two biologically distinct processes, physiological adaptation and pathological remodelling.1 A central limitation of their analysis resides in omitting the distinction between various forms of heart remodelling. In optimal conditions, physiological remodelling, as observed in athletes, may serve as a mechanism for preserving hippocampal volume.2 Conversely, small vessel disease and cognitive decline have been linked to pathological remodelling, a consequence of prolonged chronic haemodynamic stress.3 This finding aligns with the observed association between gGeom and WMH. The interpretation of causality is complicated by the absence of differentiation. Subgroup analyses of measures such as left ventricular mass index or specific geometric configurations (e.g. concentric and eccentric) might provide more thorough clarification. Furthermore, the potential of biomarkers (e.g. NT-proBNP or galectin-3) to refine the focal point merits consideration.4 Additionally, the genetic basis of the interaction could be assessed through Mendelian randomization.