Cognitively Impaired Subjects with Normal Total Cholesterol Using Lipophilic Statins Undergo Accelerated Decline to Dementia and Loss of Cognition and General Function Correlating with Loss of Regional Brain Metabolism in a Multi-Center Longitudinal Study

Introduction: While lipophilic statins have been associated with accelerated cerebral metabolic decline in subjects with early Mild Cognitive Impairment (eMCI) having below-median serum cholesterol levels (J Nucl Med. 2022;63:Supp2), the potential clinical significance of this more rapid loss of brain metabolism requires further elaboration.

Methods: Of a consecutive series of 392 eMCI subjects prospectively enrolled in the Alzheimer’s Disease Neuroimaging Initiative database, 299 had details of statin use sufficient to be categorized as none (nonS), lipophilic (LS) or hydrophilic (HS), along with recorded baseline serum cholesterol values; 171 fell below the median nonS cholesterol levels (206 mg/dl) and had documented continuous nonS, LS or HS use from baseline to at least 8 years thereafter or to the end of their study participation if less than 8 years, including 70 who also had FDG-PET at both baseline and 2 years later. Differential rates of conversion to dementia were assessed with Kaplan-Meier statistics, and a Cox multivariable model controlled for all identified factors with univariate hazard ratios with p<0.10 significance. Statistical parametric mapping (spm) was used to measure the extent of gray matter (2mm)³ voxels meeting height threshold of p<0.01 in regions of declining metabolism from baseline to 2 years and assess their correlations with indices of decline in general function (FAQ), global cognition (ADAS13 and MoCA), and memory domain tests. Standardized Volume of Interest (sVOI) methods were used to corroborate results from spm analyses and quantify rates of regional metabolic decline.

Results: There were no significant differences between nonS and LS subjects, nor between PET and non-PET subgroups, in age, sex, education, APOE genotype, serum glucose, or BMI. Over 8 years follow-up, nonS and LS had dementia conversion rates of 13.1% and 38.5%, respectively, and the multivariate hazard ratio associated with LS use was 3.25 (95% CI 1.19-8.91). At 4 years, the last time point at which the majority of subjects in each group were still participating, the rate of conversion for the HS group (9.0%) was similar to that for the nonS group (11.6%), and less than half of that for the LS group (20.4%).

In difference-of-differences spm analyses, the greatest difference in metabolic decline between LS and nonS groups occurred in posterior cingulate cortex (PCC, 1499 contiguous voxels), as further corroborated by sVOI analyses in which LS declined twice as quickly, and their functional decline significantly correlated with metabolic decline centered in PCC and extensively involving surrounding cortex (4,848 voxels, p<0.0005 after multiple comparison correction). Metabolic decline in PCC also correlated with global cognitive decline in LS (377 voxels), but not in nonS (0 voxels) subjects. Decline in other regions affected by Alzheimer’s Disease correlating with decreased performance in global cognition of LS subjects included posterior precuneus, left medial temporal, and right lateral cortex centered in inferior temporal and extending into middle temporal cortex (peak t=5.16, p<0.0005 after multiple comparison correction for extent), and right lateral temporal decline also significantly correlated with decreasing memory (peak t=5.90), which was further corroborated by sVOI analyses.

Conclusions: Use of lipophilic, but not hydrophilic, statins by subjects with early cognitive impairment and normal cholesterol levels at baseline was associated with nearly double the risk of becoming demented within 4 years, and risk of becoming demented within 8 years was triple that for non-statin users. Moreover, this was preceded during the first 2 years after baseline by metabolism in brain regions associated with early Alzheimer's demonstrating significantly greater decline, and correlating with magnitudes of clinically measurable loss of cognitive and general function.