DoliClock Uses Brain Lipids to Measure Biological Age and Detect Accelerated Aging
A new lipid-based aging clock built from prefrontal cortex lipidomics detects accelerated aging in autism, schizophrenia, and Down syndrome.
Summary
Researchers at the National University of Singapore developed DoliClock, an Elastic Net machine learning model trained on post-mortem prefrontal cortex lipidomic data from 242 human brain samples. Using dolichol lipids as primary predictors, the clock achieved a median absolute error of 8.96 years in age prediction. Critically, individuals with autism spectrum disorder, schizophrenia, and Down syndrome all showed significantly accelerated biological aging compared to neurologically healthy controls. The study also found that Shannon entropy of the lipidome increases sharply around age 40–50, implicating dysregulation of the mevalonate pathway and dolichol accumulation as early markers of brain aging. These findings establish lipid-based clocks as a complementary approach to DNA methylation clocks for studying neurological aging.
Detailed Summary
Aging is the primary risk factor for most neurological disorders, yet the molecular mechanisms linking lipid changes in the brain to biological aging remain poorly understood. This study addresses that gap by introducing DoliClock, the first lipid-based biological aging clock built from human prefrontal cortex tissue, offering a novel molecular lens on brain aging distinct from existing DNA methylation or transcriptomic approaches.
The researchers used a publicly available post-mortem lipidomic dataset (Yu et al.) comprising 242 brain samples representing 163 lipid species. Samples included 195 neurologically healthy controls and individuals with schizophrenia (n=27), autism spectrum disorder (n=15), and Down syndrome (n=5). Twenty-six machine learning models were benchmarked across 100 bootstrap iterations; linear models, specifically Elastic Net with PCA-reduced lipid features, performed best. The final DoliClock model was trained exclusively on dolichol lipid species using 10,000 bootstrapped iterations stratified by age, sex, and ethnicity, with sex, ethnicity, and post-mortem interval included as covariates.
Dolichol lipids — particularly dolichol-19 and dolichol-20 variants — emerged as the strongest age-associated features. Their summed concentration showed a near-linear increase with chronological age, and their Shannon entropy correlated with age at r=0.92 (P<0.001) when calculated in isolation. Shannon entropy of the full lipidome increased significantly around age 40–50, suggesting a threshold shift in lipid homeostasis at midlife, potentially reflecting dysregulation of the mevalonate pathway, which governs dolichol biosynthesis. Ethnicity significantly influenced lipid principal components, while sex showed no significant effect in this dataset.
Applied to neurological disorder groups, DoliClock revealed significantly elevated age acceleration in autism (P=0.047), schizophrenia (P=0.008), and Down syndrome (P=0.015) after multiple-testing correction. This aligns with epidemiological data showing reduced life expectancy in these populations and supports the hypothesis that these conditions involve premature biological aging. Notably, one dolichol-20 variant (C100H164ONa) showed a distinct correlation pattern from other dolichols, hinting at unique regulatory or functional divergence potentially relevant to disease mechanisms.
The study is limited by small sample sizes for the disorder groups, particularly Down syndrome (n=5), and relies on post-mortem tissue which restricts longitudinal or interventional applications. Nevertheless, DoliClock demonstrates that lipidomics can provide biologically meaningful aging estimates and may complement or extend epigenetic clock approaches, particularly in brain tissue where methylation clocks have shown mixed results for schizophrenia.
Key Findings
- DoliClock predicted prefrontal cortex biological age with a median absolute error of 8.96 years using only dolichol lipids.
- Shannon entropy of brain lipids surges around age 40–50, suggesting midlife dysregulation of the mevalonate pathway.
- Autism, schizophrenia, and Down syndrome all showed statistically significant accelerated biological aging vs. healthy controls.
- Dolichol-19 and dolichol-20 concentrations correlated with chronological age at r=0.92, qualifying them as aging biomarkers.
- Ethnicity significantly influenced lipid principal components; sex did not show significant lipid variance in this dataset.
Methodology
Post-mortem human prefrontal cortex lipidomic data (242 samples, 163 lipid species) were analyzed using PCA-reduced Elastic Net regression benchmarked against 26 machine learning models over 100 bootstrap iterations. DoliClock was trained exclusively on dolichol species with 10,000 stratified bootstrap iterations, incorporating sex, ethnicity, and post-mortem interval as covariates.
Study Limitations
Disorder subgroups were small (Down syndrome n=5, autism n=15), limiting statistical power for slope and interaction analyses. The post-mortem tissue source precludes longitudinal validation or clinical translation without equivalent in-vivo lipidomic methods.
Enjoyed this summary?
Get the latest longevity research delivered to your inbox every week.