Genetic analysis of 1.158 million people identifies rare variants in 209 genes associated with blood lipids
Background: Assessing rare-variant effects in population data
A study analyzing 1,158,017 people with diverse genetic backgrounds identified rare coding variants in 209 genes associated with blood lipid levels. Published in Nature Genetics, the study estimated the effects of individual variants on lipid levels and provided evidence to support genetic diagnosis of familial hypercholesterolemia.
Family studies have helped identify variants that cause lipid abnormalities, but whether the effects estimated in those studies are of similar magnitude in the general population requires separate assessment. Even variants in the same pathogenicity category can differ in their effects on lipid levels and disease risk. The European ancestry bias in existing data is another limitation. Variants relatively common in non-European populations may be missed in diagnosis because evidence for interpreting them is limited.
Key findings: 800 associations and similar effects across populations
The researchers combined genotype imputation data from the US Million Veteran Program, exome sequencing data from UK Biobank, and whole-genome sequencing data from the US All of Us program. Genotype imputation uses reference genome data to estimate the genotypes of variants that were not directly measured. Of all participants, 238,243 (20.57%) were genetically similar to non-European reference populations. The traits analyzed were total cholesterol, low-density lipoprotein cholesterol (LDL-C), high-density lipoprotein cholesterol (HDL-C), and triglycerides.
After analyzing associations for about 3 million rare coding variants by population, the researchers combined results from 14 populations using fixed-effects meta-analysis. Under an additive model, which assumes effects proportional to allele count, they detected 800 variant–lipid trait associations across 184 genomic loci. These do not represent 800 distinct variants. The findings included 176 associations involving predicted loss-of-function variants and 624 involving missense variants that change amino acids; 209 genes were implicated in total. Not all of these genes were newly discovered. Ten gene–trait pairs did not overlap previously known lipid-associated loci, and a separate analysis under a recessive model identified 109 associations.
There were 130 lipid-associated alleles that reached statistical significance only in non-European populations. This should not be interpreted to mean they have no effect in other populations. Variant frequencies differed substantially across populations, but effect sizes for significant associations that could be compared were very similar across populations.
Reassessment of variants related to familial hypercholesterolemia also yielded additional evidence. Seven variants previously classified as benign, likely benign, or of uncertain clinical significance showed LDL-C-raising effects similar in magnitude to those of known pathogenic or likely pathogenic variants. Here, “benign” refers to a classification indicating that a variant does not cause disease. These results provide grounds to reconsider pathogenicity classifications; they do not mean that official classifications in clinical databases have already changed.
Implications and outlook: Evidence for diagnostic interpretation and target discovery
In an analysis of associations with coronary artery disease, five genes, including RORC, met the significance threshold after accounting for multiple testing. Based on human genetic data and experimental results from previous studies cited in the article, the researchers proposed RORC inhibition as a potential therapeutic strategy for lowering LDL-C. This study was not a clinical trial in which inhibitors were administered to patients to evaluate efficacy or safety.
Population-scale effect estimates may inform clinical interpretation of rare variants and follow-up functional studies. However, the effects of rare noncoding variants were not sufficiently assessed, and the scope of the recessive analysis was limited by the scarcity of rare homozygous variants. Gaps in findings for non-European populations also remain, so it would be premature to conclude that broadly applicable risk scores or prevention guidelines have been fully developed.
The study design and results are available in the publicly available full text.
Nature Genetics, Published online: 25 May 2026. DOI: 10.1038/s41588-026-02613-y
Summary: Bypassing the ancestry-biased bottlenecks of historical genome-wide studies, this milestone exome-wide association study (EWAS) synthesizes deep genomic profiles from 1,158,017 individuals across the Million Veteran Program, UK Biobank, and All of Us cohorts. By deploying a high-throughput multi-ancestry analytical matrix, the framework successfully uncovers highly penetrant, rare coding variants across 209 genes fundamentally linked to systemic blood lipid traits. Beyond validating classical benchmarks like APOB and PCSK9, the multi-omic metadata isolates over 30 novel candidate drug targets showing robust allelic effect sizes across non-European sub-populations. This diverse population genetics repository eliminates confounding missing heritability, providing a programmable, low-noise computational baseline for universal cardiovascular multi-genic risk stratification and targeted nucleotide therapeutic design.
If genetic testing for familial hypercholesterolemia identifies a variant whose significance is uncertain, the LDL-C effect estimates from this study can be considered as supporting evidence when reassessing its pathogenicity. They may be particularly useful for interpreting variants in non-European populations, for which evidence has been limited. Statistical association alone cannot establish an individual’s diagnosis or determine treatment; clinical findings and additional evidence must also be assessed.
In drug research, candidates such as RORC, for which genetic evidence connects lipid levels and coronary artery disease, could be prioritized for follow-up functional and safety studies. This is a potential use at the target-discovery stage. The findings do not demonstrate drug efficacy, reduced toxicity, or shorter clinical-development timelines.