AbstractBackgroundBrain glucose hypometabolism is among the earliest pathogenic changes in Alzheimer’s disease (AD). This metabolic dysfunction points to the personal bioenergetic capacity, defined as the ability to maintain energy homeostasis under all circumstances including deregulated glucose uptake, as a potential source of resilience to the disease. Fasting blood acylcarnitine profiles are a central readout for this capacity in the absence of dietary glucose and capture the activity and efficiency of glucose‐independent routes of mitochondrial energy metabolism.MethodWe used fasting serum acylcarnitine profiles of 1,531 participants (465 with normal cognition, 762 with mild cognitive impairment, and 304 with clinical AD) from the AD Neuroimaging Initiative to perform unsupervised subgroup identification using hierarchical clustering. Identified subgroups were investigated for differences in A/T/N biomarker profiles and cognitive status. The contributions of genetic and potentially modifiable factors defining the subgroups were quantified using analysis of explained variance. The influence of the strongest determining factors on longitudinal cognitive trajectories was estimated using linear mixed‐effects models and gene‐by‐environment interaction analysis.ResultWe found several bioenergetically distinct subgroups with significant differences in AD biomarker profiles and cognitive function. The strongest genetic contribution to this bioenergetic endophenotype seems to be specifically linked to succinylcarnitine metabolism and significantly modulates the rate of future cognitive decline. In contrast, potentially modifiable sustainment of beta‐oxidation efficiency seems to decelerate bioenergetic aging, thus creating a bioenergetic reserve that delays progression of cognitive decline. Using gene‐by‐environment interaction analysis, we demonstrate that this molecular framework identifies a subgroup of individuals that is likely to benefit significantly from personalized therapeutic, dietary or lifestyle interventions tailored to increase resilience against bioenergetic disturbances in AD.ConclusionOur study reports on a set of genetic and metabolic markers that define bioenergetically distinct subgroups with significant differences on the AD biomarker and cognitive level. Longitudinal data suggest that targeting the modifiable fraction of this endophenotype might be a promising strategy to slow down disease progression in individuals with specific allelic configurations.