Genetic association studies showed that Hb F is under the influence of major quantitative trait loci (QTL) in β-thalassemia (β-thal) carriers. Single nucleotide polymorphisms (SNPs) at three major QTLs, BCL11A, HBS1L-MYB intergenic region and XmnI-HBG2 were individually validated in univariate models. However, their relative effect sizes on Hb F regulation are unknown. We genotyped 99 Chinese β-thal carriers for the three major QTLs and performed genetic association studies using three different statistical models, including mass univariate analysis, multivariate linear regression and partial least square regression structural equation modeling (PLS-SEM). Performances of the three models were compared and effect sizes of the three QTLs in a multivariate model were assessed. Traditional mass univariate analysis and multivariate linear regression showed limited statistical power in our small cohort and the latter was constrained by multicollinearity. Partial least structural equation modeling showed significant positive associations of each QTL (p <0.05) with Hb F regulation, together explained 34.4% of variance. The HBS1L-MYB intergenic region polymorphism (HMIP) demonstrated the highest effect on Hb F prediction with effect size f2 0.294. PLS-SEM offered a statistically powerful multivariate model for multi-locus genetic association studies. We reproduced findings of previous studies with a much smaller cohort and demonstrated HMIP as the strongest regulator of Hb F in Chinese β-thal carriers.
Keywords: fetal hemoglobin (Hb F); partial least square (PLS); quantitative trait loci (QTL); structural equation modeling (SEM); β-Thalassemia (β-thal).