Plasma metabolomic profiles of dementia: a prospective study of 110,655 participants in the UK Biobank

Abstract

Background: Plasma metabolomic profile is disturbed in dementia patients, but previous studies have discordant conclusions.

Methods: Circulating metabolomic data of 110,655 people in the UK Biobank study were measured with nuclear magnetic resonance technique, and incident dementia records were obtained from national health registers. The associations between plasma metabolites and dementia were estimated using Cox proportional hazard models. The 10-fold cross-validation elastic net regression models selected metabolites that predicted incident dementia, and a 10-year prediction model for dementia was constructed by multivariable logistic regression. The predictive values of the conventional risk model, the metabolites model, and the combined model were discriminated by comparison of area under the receiver operating characteristic curves (AUCs). Net reclassification improvement (NRI) was used to estimate the change of reclassification ability when adding metabolites into the conventional prediction model.

Results: Amongst 110,655 participants, the mean (standard deviation) age was 56.5 (8.1) years, and 51 186 (46.3%) were male. A total of 1439 (13.0%) developed dementia during a median follow-up of 12.2 years (interquartile range: 11.5-12.9 years). A total of 38 metabolites, including lipids and lipoproteins, ketone bodies, glycolysis-related metabolites, and amino acids, were found to be significantly associated with incident dementia. Adding selected metabolites (n=24) to the conventional dementia risk prediction model significantly improved the prediction for incident dementia (AUC: 0.824 versus 0.817, p =0.042) and reclassification ability (NRI = 4.97%, P = 0.009) for identifying high risk groups.

Conclusions: Our analysis identified various metabolomic biomarkers which were significantly associated with incident dementia. Metabolomic profiles also provided opportunities for dementia risk reclassification. These findings may help explain the biological mechanisms underlying dementia and improve dementia prediction.

Keywords: Dementia; Metabolites; UK Biobank.

Fig. 1

Data processing and analyses flow diagram of this study. Thirty-eight metabolites were significant following multiple testing in multi-variable cox proportional hazards models. For the development of a prediction model, participants were randomly assigned to the training and testing group for model development. After a 10-fold cross-validation test, 24 metabolites were assigned a nonzero coefficient in the elastic net regression model amongst the 249 included metabolites. Receiver operating characteristic (ROC) curve was created and area under curve (AUC) was calculated for predictive value comparison. Categorical net reclassification improvement (NRI) was calculated to investigate the reclassification ability

Fig. 2

Adjusted HR (95% CI) of incident dementia for metabolites after multiple testing. Hazard ratios (HR) are per 1 standard deviation (SD) higher of Z-transformed metabolic marker and are adjusted for age, gender, education level, systolic pressure, anti-hypertension treatment, diabetes mellitus, smoking status, history of stroke, history of coronary heart disease, and APOE ε4 allele. CI, confidence interval; LDL, low-density lipoprotein; HDL, high-density lipoprotein; VLDL, very-low-density lipoprotein; IDL, intermediate-density lipoprotein

Fig. 3

ROC and AUC analysis of incident dementia prediction model development and predictive value comparison. An elastic net regression model based on lasso penalty was used for dementia prediction. After 10-fold cross-validation, 24 of 249 metabolites were selected for the dementia prediction model. Xb1 curve used conventional risk factors as input signals, while the Xb2 curve was for 24 selected metabolites and Xb3 was for conventional risk factors and 24 selected metabolites. There was no clinically significant difference (P = 0.042) found between the AUC of Xb1 and Xb3. Conventional risk factors included age, gender, education level, systolic pressure, anti-hypertension treatment, diabetes mellitus, smoking status, history of stroke, history of coronary heart disease, and APOE ε4 allele. ROC, receiver operating characteristic; AUC, area under curve