Background

Genome-wide association studies (GWAS) have become a key tool for identifying the genetic factors of complex diseases. However, over 80% of modern genomic research has been biased toward European ancestry data. Latin American populations, which feature a complex mixture of Indigenous American, European, and African ancestries, possess immense genetic diversity but have remained relatively understudied.

Analyzing complex traits in admixed populations is prone to severe statistical bias. Population stratification, socioeconomic environments, and differences in residential areas based on ancestral proportions can mix with genetic variants to create false-positive signals. Previous studies comparing unrelated individuals faced limitations in strictly separating environmental factors from pure genetic factors. It was difficult to determine whether specific disease risk variants were the result of actual biological mechanisms or illusions arising from accumulated social disparities along ancestral lines.

To address these confounding factors, within-family designs—comparing parents with children or siblings with each other—have gained attention. Comparing siblings who inherit genes from the same parents and share the same home environment effectively eliminates environmental noise. Instances of large-scale precision analysis using family-unit genome-wide data in large admixed populations have been extremely rare until now.

Key Findings

An international research team overcame this challenge using large-scale family data from a prospective cohort in Mexico City. The researchers extracted genomic information from tens of thousands of individuals with confirmed kinship in Mexico City and tracked the randomly distributed proportions of ancestral DNA among siblings, following Mendel's law of segregation. This approach leverages the fact that due to chromosomal recombination during meiosis, the proportions of inherited Indigenous and European ancestries differ even among siblings.

As a result of the analysis, the researchers captured distinct phenotypic differences based on ancestry proportions in major complex traits, including height and type 2 diabetes (T2D). Even among siblings raised by the same parents, it was reconfirmed that a higher proportion of Indigenous American ancestry is associated with a statistically significant reduction in adult height. The observation of a height reduction effect even under complete control of environmental confounding factors clearly demonstrates that this is a direct genetic influence attributable to ancestral lineage, rather than an environmental difference.

A more meaningful pattern emerged in the T2D analysis. While previous population comparison studies reported that higher proportions of Indigenous ancestry were associated with a sharp increase in diabetes prevalence, this within-family analysis model showed that a substantial portion of that risk was attenuated. Even after adjusting for shared environments, the finding that specific genetic loci of Indigenous ancestry are directly involved in glucose metabolism and insulin secretion pathways remained significant. In particular, it was identified that genetic variants regulating lipid metabolism and energy homeostasis exhibit different genetic background effects depending on ancestry.

Significance and Outlook

This achievement demonstrates that family-based analysis is the most powerful validation tool for removing environmental bias in admixed population genomic research. It provides empirical evidence that much of the susceptibility to complex traits and diseases, previously attributed to racial and ethnic differences, may have been due to environmental confounding. The methodology presented by the researchers is expected to become a standard research template for rigorously verifying the biological causality of genetic variants found in multi-ethnic cohorts.

It also provides an opportunity to improve the predictive accuracy of Polygenic Risk Scores (PRS) for non-European populations. Existing PRS algorithms trained on European data often show a fatal limitation where predictive power drops by more than half when applied to Latino patients. If models are recalibrated based on pure genetic effect sizes with environmental factors removed, the reliability of precision medicine algorithms customized for admixed populations is expected to improve significantly.

However, caution is required when generalizing these results to other Latino or admixed populations in different regions. This is due to the specific characteristics of the urban environment in Mexico City and potential differences in the detailed sub-lineages of local Indigenous populations. Expanding large-scale multi-cohort cohorts covering wider geographic ranges and conducting functional genomic validation remain essential follow-up tasks.