In 2001, the draft sequence of the first human genome was reported, followed by a near-complete sequence published in 2004. One promise of the trove of genomes was to allow scientists to uncover the genetic basis of human diseases, and thereby to facilitate the design of rational diagnostics and therapeutics. Two decades later, more than 420,000 loci have been associated with hundreds of phenotypes by genome-wide association studies (GWASs),[1] a comprehensive, unbiased, and hypothesis-free approach. For any given phenotype, there were a large number of underlying genetic loci, each exhibiting only small effect, but in combination, they could explain a significant proportion of the variance of phenotypes in the general population. Thus, by aggregating the effects of relevant genetic loci into a single indicator, polygenic scores (PGSs) (also referred to polygenic risk scores [PRSs], or genetic or genomic risk scores [GRSs] for diseases) have emerged as a quantitative measure to predict an individual's genetic predisposition to a phenotype. Although different methods have been proposed to construct PGSs, some common principles and reporting standards have been described to a less extent in recent reviews.[2,3] To date, as documented by the PGS Catalog (www.PGSCatalog.org), over 2600 PGSs have been developed and demonstrated to be predictive of phenotype variations throughout the whole life course, which may facilitate the translation from initial GWAS discoveries to clinical practices. These PGSs can generate informative predictions for heritable traits and diseases, and the potential clinical utility of PGS has been reported in several chronic non-communicable diseases, including cancers, diabetes, chronic respiratory disease, and psychiatric conditions.[4–7] For example, Dai et al[8] and Jin et al[9] constructed PRS-19 and PRS-112 for lung cancer and gastric cancer, respectively, based on a meta-analysis of GWASs with a large sample size in the Chinese population, which were further proved to be effective in identifying high-risk individuals in an independent large prospective cohort from the Chinese population. These efforts could be useful to stratify a population into categories with distinct risks for lifestyle interventions or modified screening approaches. A growing number of clinical trials are under way to explore the additional benefits from individualized risk stratification screening based on PGS, particularly for breast and colorectal cancers. Other potential clinical uses of PGSs include prediction of prognosis, stratification of patients according to therapeutic benefit, and identification of drug targets. For example, a recent study showed that a PGS based on 27 genetic variants could not only identify individuals at increased risk for incident and recurrent coronary heart disease events, but also indicate that statin therapy has a greater absolute risk reduction of primary coronary heart events in high-risk individuals defined by genetic risk.[10] In addition to adult traits and diseases in later life, early-life growth traits, such as birth weight, have also been associated with numerous genetic variants. However, by using GWASs of unrelated individuals, it is hard to decipher the contributions of fetal genome (direct effect) and maternal genome through the intrauterine environment (indirect effect) to fetal traits.[11] Family-based GWASs can separate direct from indirect genetic effects by using parental genotypes as controls. Recently, Juliusdottir et al[12] performed GWASs on birth weight in parent–offspring trios to analyze the relationships between fetal growth traits and alleles transmitted from either the mother or the father. The authors found no evidence of paternal-specific loci affecting offspring birth weight. By using haplotype-specific PRSs, they found that the maternal genome contributed to increased birth weight through blood glucose-raising alleles, while blood pressure-raising alleles reduce birth weight largely through the fetal genome.[12] By comparing germline de novo mutations between offspring in families conceived spontaneously and assisted reproductive technology (ART) based on whole-genome sequencing trios, Wang et al[13] demonstrated the increased mutations in offspring conceived by ART were primarily originated from fathers. Therefore, such a trio approach enables accurate phasing and the assignment of the parental origin of alleles to show parent-of-origin effects. Recently, PGSs have been used to test and select embryos for patients considering in vitro fertilization (IVF) by some companies in the USA.[14] Unlike preimplantation genetic diagnosis (PGD) to avoid genetic disorders (eg, cystic fibrosis or hereditary breast and ovarian cancers), embryo selection based on polygenic scores (ESPS) was advocated to screen common complex diseases (eg, type 2 diabetes, prostate cancer, or hypertension) or non-clinical phenotypes (eg, cognitive ability, education, household income, or subjective well-being). However, concerns have been raised.[14] First, the expected gain from ESPS in a certain family is smaller than that observed from the general population. Second, there may be potential unintended consequences of ESPS, such as selecting for adverse traits as a result of pleiotropy. Therefore, ESPS is not ready for clinical uses, although disease risk for individuals with extreme PRSs equivalent to monogenic mutations makes meaningful contribution to genetic risk prediction and aforementioned medical applications. In future, for a specified disease, ESPS may be cautiously used when it is based on gene panel of rare variants with high penetrance, rather than common low-penetrant variants, particularly for those genes with well-known biological function to exclude potentially adverse health outcomes. However, legal and ethical factors need to be extensively considered in ESPS practices, such as altering population demographics, exacerbating inequalities in society, and devaluing certain traits. Nevertheless, the capacity of PGSs to quantify genetic predisposition to phenotypes has just begun to be established. Additional efforts are required in future studies to facilitate the application of PGSs in clinical practices. First, rare and low-frequency variants as well as common variants with weaker effects may substantially contribute to the underlying genetic architecture of phenotypes, but the majority of such variants are not yet known. GWASs based on whole-genome sequencing technology would enable a comprehensive investigation of rare and low-frequency predisposing variants. For instance, in the most recent study, Wang et al[15] determined hundreds of rare pathogenic variants with large effects in the coding and regulatory regions of cancer predisposition genes, and then built a high-quality haplotype reference panel to aid in the discovery of numerous new common/low-frequency susceptibility variants. The findings of these variants will definitely help improve the predictive ability of PGSs, but studies with very large sample sizes will be required for further identification of large numbers of such variants.[3] Second, most GWASs have been conducted in populations of European ancestries, which results in much lower predictive power of PGSs in non-European populations. To eliminate such a disparity, besides an investment in GWASs across all major ancestries, future studies with diverse populations are required to perform systematic and thorough evaluations of the utility of PGSs across populations. Third, to address the clinical validity of PGSs, evaluations are required in the clinical or public health context, such as prospective cohorts considering other non-genetic factors for absolute risk estimation, randomized trials with clinically meaningful outcomes, and health economic evaluations or feasibility studies. Fourth, in a family-based GWAS, parental genotypes can be used as controls to obtain unbiased estimates of direct genetic effects.[11] Therefore, future studies combining information from standard GWASs and from analyses of families may be critical to determine the parent-of-origin effects and to properly investigate the relationships between early-life phenotypes and the future risk of later-life diseases. Funding The work presented in this article was supported by a grant from the National Natural Science Foundation of China (No. 82125033). Conflicts of interest None.
Search from 250 M+ research papersSearch from 250 M+ research papers
Open Access
Mar 1, 2024
Carlos Iribarren + 7
