Logic regression analysis of association of gene polymorphisms with low HDL: Tehran Lipid and Glucose Study

Introduction

Logic regression is a generalized regression method that can recognize complex Boolean interactions of binary variables. It has been successfully applied to single-nucleotide polymorphism (SNP) data because of the importance of interactions in SNP association studies. The objective of this study is to assess the association between high-density lipoprotein (HDL) function and some related gene polymorphisms after adjusting for potential confounders using logic regression.

Methods

Subjects in this cross-sectional study were randomly selected from among participants of the Tehran Lipid and Glucose Study (TLGS). A total of 436 subjects (172 men and 264 women), aged ≥ 20 years, were selected to be included in the current study. Logic regression analysis was used in order to recognize the combination of genetic main effects and possible interactions associated with HDL-C level. Cross validation and randomization test were carried out to avoid over fitting of the models.

Results

The cross validation test suggested three Boolean combinations with four predictors for a fully-adjusted logic model. The fully adjusted model showed that those who carry an Apo E gene E3 allele or have high TG level have an odds ratio of 2.35 (95% CI:1.3–4.25) for having low HDL compared to other subjects. In addition, subjects with high TG level have an odds ratio of 2.73 (95% CI: 1.65, 4.53) for having low HDL.

Conclusion

The results showed that logic regression is a powerful method to find the interaction between high TG level and Apo E polymorphism associated with low HDL.