Abstract
Cognition is one of the most widely used measures of brain health. Poor cognitive function reflects both the progression of neurodegenerative disorders and the decline of cardiometabolic conditions. Moreover, the genetic determinants of cognitive function are pleiotropic with various aspects of physical health and mortality. To date, little research representing the diverse Asian populations has been done. To understand the epidemiological role of cognition and its biological underpinnings in this population, we performed PheWAS on 542 unique research variables collected in the HELIOS Study (Singapore), with a dataset comprising 8207 individuals (Chinese: Indian: Malay = 69:18:13) aged 30-84 years in whom ‘g’, general factor for cognition, was derived. We used ‘g’ as a predictor and adjusted for age, sex, ethnicity, the interaction of age and sex and genetically determined ancestries. We then performed 2-sample Mendelian Randomization and downstream sensitivity analyses using the GWAS summary statistics from the UK Biobank to ascertain causality.
We found that ‘g’ was associated with a broad range of phenotypes (Bonferroni p < 8.1x10⁻⁵, 148 [27.3%] phenotypes), ranging from greater familiarity with digital devices (β=0.21, p=5.5x10⁻¹³⁸), to smaller waist-hip-ratio (β=-0.06, p=2.5x10⁻¹⁵) and lower carbohydrate intake (β=-0.07, p=3.5x10⁻¹¹). After further hierarchical clustering to reduce redundancies and ensure representativeness across domains, we shortlisted 23 (4.2%) representative phenotypes. The Inverse Variance Weighted 2-sample MR demonstrated causal linkage of genetically predicted ‘g’ with 14 demographic and physiological phenotypes, including less body fat percentage, lower alcohol consumptions, higher polyunsaturated/monounsaturated fatty acids (PUFA/MUFA) consumptions, household income and total year of education, with significant horizontal pleiotropy. From the initial 147 SNPs in the instrument-exposure ‘g’, we selected 134 SNPs which uniquely predict ‘g’ and performed further annotations. The SNPs are linked with genes involved in neural development, neurotransmission, neuroprotection, and chromatin modification and transcriptional regulation. Our study demonstrates broad implications of cognitive function in lifestyle choices and cardiometabolic health indices as potential targets for early health maintenance at population levels.
Introduction
Most of what we know about the genetics of cognition comes from studies of European populations. However, the global burden of dementia is shifting toward the Asia-Pacific region, which is expected to contribute approximately 40% of incident cases globally by 2025. There is a significant "diversity gap" in cognitive research, as genetic determinants and lifestyle risk factors (like diet or ancestry-specific variants) vary across ethnicities. This study seeks to understand the epidemiological role and biological underpinnings of cognition specifically within the diverse Asian context, moving beyond simple correlation to identify which health outcomes are actually caused by cognitive levels.
Methods and Datasets
The study utilized a two-stage approach:
- PheWAS (Phenome-wide Association Study): Researchers scanned 542 unique variables from the HELIOS study (Health for Life in Singapore) to find associations with ‘g’ (a general factor for cognition derived from multiple tests). The cohort consists of 8,207 individuals (69% Chinese, 18% Indian, and 13% Malay).
- Mendelian Randomization (MR): To move from correlation to causation, 2-sample MR was performed using summary statistics from the UK Biobank (UKB) to ascertain causality for shortlisted phenotypes.
- Clustering and Annotation: Hierarchical clustering was used to group redundant variables, and functional annotation was performed on 134 unique SNPs to identify biological pathways.
Key Results
The PheWAS revealed that 148 out of 542 phenotypes (27.3%) were significantly associated with ‘g’. The strongest associations were found in the socio-demographic domain, particularly regarding digital literacy and education.
MR confirmed that higher genetically predicted ‘g’ is causally linked to 14 phenotypes, including:
- Socioeconomic: Higher household income and more years of education.
- Lifestyle: Greater computer usage and lower alcohol consumption.
- Physiological: Greater height, better lung function (PEF, FEV1, FVC), and lower body fat percentage.
Interestingly, the "Digital Divide" appears to be cognitively driven; the strongest association was between higher ‘g’ and familiarity with digital devices. Furthermore, higher cognitive function was causally linked to lower body fat percentage and smaller waist-to-hip ratios, suggesting cognition influences lifestyle choices affecting subcutaneous fat more than metabolic-specific fat storage.
Discussion and Conclusion
The study highlights that cognitive health is a central pillar of overall cardiometabolic and socioeconomic well-being. A key finding was the causal link between higher ‘g’ and healthier dietary choices (higher PUFA/MUFA, lower alcohol), implying that cognitive ability leads to better-informed nutritional decisions.
While the study is limited by the use of European-based summary statistics for MR outcomes due to a lack of large-scale Asian GWAS data, it provides critical insights into the Asian context. Future work should focus on Asian-specific genetic resources to validate these findings. Cognitive health maintenance should be a primary target for public health strategies to combat adiposity and cardiometabolic decline in multiethnic populations.
Summary updated