In a recent pre-print study posted to the medRxiv* server, researchers conducted a comprehensive genome-wide association study (GWAS) to elucidate the genetic architecture of circulating retinol, identify its potential causal relationships with various clinical phenotypes, and evaluate its therapeutic or nutritional implications.
Study: Genetic influences on circulating retinol and its relationship to human health. Image Credit: SciePro/Shutterstock.com
*Important notice: medRxiv publishes preliminary scientific reports that are not peer-reviewed and, therefore, should not be regarded as conclusive, guide clinical practice/health-related behavior, or treated as established information.
Background
Vitamin A, essential for vision and immunity, comprises compounds like retinol (found in animal products), retinoids, and carotenoids (plant-based precursors). Retinoic acid, a potent signaling molecule derived from these, controls gene expression.
The liver is the primary storage for dietary retinol, transported via retinol-binding protein 4 (RBP4) and stabilized in circulation by transthyretin (TTR). Synthetic retinoids, similar to retinol, treat skin conditions and some cancers.
While high vitamin A diets have potential health benefits, research results vary. Some genetic studies have highlighted genes influencing retinol abundance, providing an opportunity for further research.
About the study
The present study analyzed datasets for genome-wide meta-analysis of circulating retinol levels. The largest dataset came from the INTERVAL study, which measured retinol in the blood of UK donors.
Another study was the METSIM, which focused on middle-aged Finnish men and used a high-throughput platform for measurement. A 2011 GWAS of circulating retinol incorporated data from the alpha-tocopherol, beta-carotene cancer prevention (ATBC), and the Prostate, Lung, Colorectal, and Ovarian (PLCO) studies.
The TwinsUK cohort served as a replication dataset, mainly involving female participants. Genome-wide meta-analysis of these datasets aimed to annotate and define specific variants, assess the genetic architecture of retinol, and prioritize potential genes linked to retinol levels. The study used various statistical and bioinformatics tools for analysis and interpretation.
The researchers integrated retinol GWAS with genetic effects on messenger ribonucleic acid (mRNA) and protein expression to identify causal genes not achieving genome-wide significance. They used the FUSION method for a transcriptome and proteome-wide association study of circulating retinol.
FUSION uses models of genetically regulated expression with significant non-zero heritability. Variant weights from these models were combined with the effects of the same variants on retinol to estimate the direction of regulated expression associated with increasing retinol levels.
Tissue selection for the study was based on known retinol biology and statistical significance. They applied corrections for multiple testing and then used Mendelian randomization to identify potential causal effects of retinol on human traits.
The pipeline also explored drugs impacting circulating retinol and used TwinsUK data to analyze retinol polygenic scores.
In the TwinsUK study, a normative model of retinol versus age was created using a generalized additive model (GAMLSS) in R v4.4.1. This semi-parametric method accounts for non-standard data traits like heteroskedasticity.
The model's efficacy was determined through criteria like AIC and BIC, with results considering factors like body mass index (BMI) and age-related retinol levels. Analyses ran on various systems, mainly R versions 4.1.1 and 4.0.3 and Python 2.7.17 or 3.6.9.
Study results
In the present study involving 22,274 individuals of European ancestry, researchers probed the genetic underpinnings of circulating retinol. This involved collating data from INTERVAL, METSIM, ATBC, and PLCO studies.
The primary meta-analysis (METSIM+INTERVAL) identified eight significant genetic loci linked to circulating retinol. Notably, six of these loci were newly discovered, contrasting earlier research.
The effects these loci exerted ranged between 0.066 to 0.172 SD per effect allele. The reliability of these findings was further bolstered by replication attempts in the TwinsUK cohort, which confirmed seven out of the eight identified single nucleotide polymorphisms (SNPs).
Interestingly, the genetic markers pinpointed did not correlate with dietary retinol intake. A more expansive meta-analysis, which included data from all four studies, reinforced most of these loci's significance.
Moreover, a rare genetic variant on chromosome five linked to decreased plasma retinol was spotted. Other noteworthy rare variants were spotted in genes such as FRAS1-related extracellular matrix protein 2 (FREM2), NAD(P)HX dehydratase (NAXD), and chromodomain helicase DNA binding protein 1 (CHD1). Despite these findings, post-correction, no significant gene-level associations with retinol were observed.
The study methodically prioritized genes. Four primary loci, viz. glucokinase regulator (GCKR), forkhead box p2 (FOXP2), RBP4, and TTR, emerged with consistent evidence suggesting their causal roles in retinol levels. The GCKR gene, which plays pivotal roles in metabolism, stood out.
The FOXP2 gene on chromosome seven also showed strong causal indications. While its contributions to brain function, especially language, are well-documented, its role in peripheral tissues remains largely uncharted. Exploration into this gene demonstrated its influence over several pathways, including extracellular matrix biology and interleukin signaling.
An expansive phenome-wide analysis with over 17,000 traits hinted at possible limitations, potentially blurring retinol's effects on certain diseases. In-depth scrutiny of retinol's instrumental variables across 1,141 binary outcomes revealed connections to eight disease phenotypes.
Notably, retinol might increase heart malformation risks but offers protective effects against type 2 diabetes with coma and inflammatory liver disease.
While cancer risk overall did not seem to correlate with retinol levels, a particular lung cancer type showed a potential protective relationship with retinol, echoing earlier research.
Furthermore, the study found connections between retinol and factors like lipids, kidney function, and certain agents like lithium.
Notably, lipid species such as triglycerides and cholesteryl ester traits appear to affect retinol levels. Also, creatinine, reflecting kidney health, may causally influence retinol levels.
The study performed a polygenic score (PGS) analysis for retinol, emphasizing the complex relationship between retinol and age and the importance of genetic factors in determining retinol levels.
*Important notice: medRxiv publishes preliminary scientific reports that are not peer-reviewed and, therefore, should not be regarded as conclusive, guide clinical practice/health-related behavior, or treated as established information.
Evo model set to transform synthetic biology and disease diagnosis