Population-specific reference panels are crucial for genetic analyses: an example of the CREBRF locus in Native Hawaiians

Abstract

Statistical imputation applied to genome-wide array data is the most cost-effective approach to complete the catalog of genetic variation in a study population. However, imputed genotypes in underrepresented populations incur greater inaccuracies due to ascertainment bias and a lack of representation among reference individuals, further contributing to the obstacles to study these populations. Here we examined the consequences due to the lack of representation by genotyping in a large number of self-reported Native Hawaiians (N = 3693) a functionally important, Polynesian-specific variant in the CREBRF gene, rs373863828. We found the derived allele was significantly associated with several adiposity traits with large effects (e.g. ~ 1.28 kg/m2 per allele in body mass index as the most significant; P = 7.5 × 10−5), consistent with the original findings in Samoans. Due to the current absence of Polynesian representation in publicly accessible reference sequences, rs373863828 or its proxies could not be tested through imputation using these existing resources. Moreover, the association signals at the entire CREBRF locus could not be captured by alternative approaches, such as admixture mapping. In contrast, highly accurate imputation can be achieved even if a small number (<200) of internally constructed Polynesian reference individuals were available; this would increase sample size and improve the statistical evidence of associations. Taken together, our results suggest the alarming possibility that lack of representation in reference panels could inhibit discovery of functionally important loci such as CREBRF. Yet, they could be easily detected and prioritized with improved representation of diverse populations in sequencing studies.

A murine model of the human CREBRFR457Qobesity-risk variant does not influence energy or glucose homeostasis in response to nutritional stress

“…This relationship between the CREBRF R457Q variant and BMI was highly significant (β 1.356, p 1.12 x 10 -13 ), as were the associations with obesity risk (OR 1.305, p 1.12 x 10 -5 ) and diabetes risk (OR 0.586, p 6.68 x 10 -9 after adjusting for BMI) (6). The relationship between CREBRF R457Q and obesity/diabetes risk has now been replicated by numerous groups in other populations across Oceania (9,(11)(12)(13)(14)(15). These data provide strong support for the observed associations, but are not sufficient to prove causality.…”

The CREBRF diabetes-protective rs373863828-A allele is associated with enhanced early insulin release in men of Māori and Pacific ancestry

“…While this strategy is important to understanding whether CREBRF p.Arg457Gln alters glucose homeostasis prior to disease development, it is possible that the impact of the A allele changes with age or with development of age-related metabolic diseases and related medications. Indeed, when body composition was assessed by DXA scan, the A allele of rs373863828 associated with increased fat mass in old age [9], but, in contrast, also increased lean mass in infants [20]. As alluded to earlier it may be under conditions where there is increased pressure on the β-cells to produce insulin (such as in prediabetes) the effect of CREBRF p.Arg457Gln on β-cells function becomes more apparent.…”

“…The A allele of rs373863828, a Arg457Gln missense variant in the CREBRF gene, has been associated with a 30-40% reduction in risk of T2D in people of Polynesian (Aotearoa NZ Māori and Pacific peoples) [4][5][6] and Oceanic (Marianas and Micronesian) [7] ancestries, as well as reduced risk of gestational diabetes in Māori and Pacific women with obesity [8]. CREBRF rs373863828 has a minor allele frequency of 5-27% in people of Polynesian and Oceanic ancestry but is rare (<0.01%) in non-Pacific populations [4,5,9,10]. The mechanism by which this genetic variant protects from diabetes is not known.…”

“…The Gln (A) allele of CREBRF p.Arg457Gln promotes fat cell development and adiposity [4, 9], suggesting that the protective effect of the A allele on T2D may relate to preferential lipid storage in subcutaneous fat depots over metabolically more harmful visceral or ectopic fat deposition [15]. While genetically-driven metabolically-favourable fat distribution phenotypes have been reported [18, 19], there is little evidence of CREBRF Arg457Gln being a factorable adiposity allele.…”

Using population-specific add-on polymorphisms to improve genotype imputation in underrepresented populations

“…tens of millions of SNPs). The quality of imputation is dependent on the suitability of the tag SNPs and the similarity of haplotype structure between the reference panel and the study population[2, 3, 4, 5].…”