A randomized feasibility trial of the modified Atkins diet in older adults with mild cognitive impairment due to Alzheimer’s disease

Background: Alzheimer’s disease (AD) is increasing in prevalence, but effective treatments for its cognitive impairment remain severely limited. This study investigates the impact of ketone body production through dietary manipulation on memory in persons with mild cognitive impairment due to early AD and explores potential mechanisms of action. Methods: We conducted a 12-week, parallel-group, controlled feasibility trial of a ketogenic diet, the modified Atkins diet (MAD), compared to a control diet in patients with cognitive impairments attributed to AD. We administered neuropsychological assessments, including memory tests, and collected blood samples at baseline and after 12 weeks of intervention. We performed untargeted lipidomic and targeted metabolomic analyses on plasma samples to detect changes over time. Results: A total of 839 individuals were screened to yield 38 randomized participants, with 20 assigned to receive MAD and 18 assigned to receive a control diet. Due to attrition, only 13 in the MAD arm and nine in the control arm were assessed for the primary endpoint, with two participants meeting ketosis levels used to define MAD adherence criteria. The average change from baseline in the Memory Composite Score was 1.37 (95% CI: −0.87, 4.90) points higher in the MAD group compared to the control group. The effect size of the intervention on baseline MAD change was moderate (Cohen’s D = 0.57, 95% CI: −0.67, 1.33). In the 15 participants (nine MAD, six control) assessed for lipidomic and metabolomic-lipidomics and metabolomics, 13 metabolites and 10 lipids showed significant changes from baseline to 12 weeks, including triacylglycerols (TAGs, 50:5, 52:5, and 52:6), sphingomyelins (SM, 44:3, 46:0, 46:3, and 48:1), acetoacetate, fatty acylcarnitines, glycerol-3-phosphate, and hydroxy fatty acids. Conclusions: Attrition was greatest between baseline and week 6. All participants retained at week 6 completed the study. Despite low rates of adherence by criteria defined a priori, lipidomic and metabolomic analyses indicate significant changes from baseline in circulating lipids and metabolites between MAD and control participants at 12-week postrandomization, and MAD participants showed greater, albeit nonsignificant, improvement in memory.