Predicting Aldehyde Dehydrogenase 2 (ALDH2) Polymorphism: ALDH2-genotype and Facial-FlushingReaction Mismatching Cases

Abstract

Backgrounds: Aldehyde dehydrogenase 2(ALDH2) gene encodes a critical metabolic enzyme which plays an important role in breakdown of acetaldehyde. In addition, the risk of various alcohol-related diseases and cancers can be altered according to the ALDH2 genotype. Currently ALDH2 genotype is predicted by facial flushing reaction questionnaire since it is not possible to genotype every individual. However, some individuals, who consume relatively large amount of alcohol than average, tend to be adapted to facial flushing reaction and hence express opposite phenotype from their ALDH2 genotype.

Objectives: This study aims to develop ALDH2 polymorphism prediction model with alcohol related variables to increase predictive performance with current screening method, facial flushing questionnaire and also to explore genes associated with facial flushing reaction others than ALDH2 and score polygenic effect of associated SNPs.

Methods: 1807 study participants from Korean Healthy Twin study cohort were involved in prediction modelling of ALDH2 genotype. 2/3 of study participants were included in training dataset and significant alcohol related variables were selected as predictors with no multicollinearity by backward stepwise regression. Candidate prediction models were validated with 1/3 of study participants and predictive performance of models were measured with AUC with ROC curve, Hosmer-Lemeshow test, reclassification. In addition, ALDH2(rs671) conditional GWAS was performed using family-based score test for association (FASTA) from GenABEL in R software and GWAS summary statistics were used to calculate polygenic score. Results: Selected predictors apart from facial flushing reaction were 1) amount of alcohol consumption (g/week), Hazard domain of AUDIT, AUDIT question No.8. With these variables 5 candidate models were generated and Models with facial flushing reaction, Hazard domain of AUDIT and AUDIT question No.8 as predictors were selected as the best prediction model to discriminate active and inactive ALDH2 carriers with AUC of 0.93 which is 0.013 higher than facial flushing reaction questionnaire alone. Results from ALDH2(rs671) conditional GWAS had no significant SNPs yet clumped 17 SNPs at suggestive significant level including rs6480460 in PRF1 were used to generate polygenic score. Regardless of ALDH2 genotype, individuals with no flushing reaction expressed stronger polygenic effect on metabolic ability to breakdown alcohol.

Conclusion: Facial flushing questionnaire is an excellent screening tool for ALDH2 polymorphism. However, prediction model with additional alcohol related variables as predictors can improve the predictive performance. Also, inactive ALDH2 carriers with no facial flushing reaction had stronger polygenic effect of SNPs related to metabolic ability for alcohol intake.