Prediction model for the efficacy of folic acid therapy on hyperhomocysteinaemia based on genetic risk score methods

Abstract

No risk assessment tools for the efficacy of folic acid treatment for hyperhomocysteinaemia (HHcy) have been developed. We aimed to use two common genetic risk score (GRS) methods to construct prediction models for the efficacy of folic acid therapy on HHcy, and the best gene–environment prediction model was screened out. A prospective cohort study enrolling 638 HHcy patients was performed. We used a logistic regression model to estimate the associations of two GRS methods with the efficacy. Performances were compared using area under the receiver operating characteristic curve (AUC). The simple count genetic risk score (SC-GRS) and weighted genetic risk score (wGRS) were found to be independently associated with the efficacy of folic acid treatment for HHcy. Using the SC-GRS, per risk allele increased with a 1·46-fold increased failure risk (P < 0·001) after adjustment for traditional risk factors, including age, sex, BMI, smoking, alcohol consumption, history of diabetes, history of hypertension, history of hyperlipidaemia, history of stroke and history of CHD. When used the wGRS, the association was strengthened (OR = 2·08, P < 0·001). Addition of the SC-GRS and wGRS to the traditional risk model significantly improved the predictive ability by AUC (0·859). A precise gene–environment predictive model with good performance was developed for predicting the treatment failure rate of folic acid therapy for HHcy.

Using the optimal method—explained variance weighted genetic risk score to predict the efficacy of folic acid therapy to hyperhomocysteinemia

“…Our study explored whether adding GRS to traditional risk factors can improve the predictive ability. First of all, we undertook four different GRS methods (SC-GRS, DL-GRS, PG-GRS, and EV-GRS) [18][19][20][21][22][23][24]34]. SC-GRS just added the number of the risk alleles across every SNP at the selected loci.…”

“…However, the approach based on the single SNP analysis is limited by low predictive power and weak effect, which is insufficient [17]. Several previous studies have shown that applying multiple genetic loci with relatively small effects into the global genetic risk score (GRS) might improve the identification and capture a better complex relationship between genetics and efficacy of folic acid [18].…”

“…GRS is used to evaluate the effects of multiple genetic susceptible factors in the risk predictive models [18][19][20][21]. Four approaches are commonly used for GRS to determine genetic risk which is based on risk SNPs: i.e.…”

Combining genetic risk score with artificial neural network to predict the efficacy of folic acid therapy to hyperhomocysteinemia

“…To combine numbers of SNPs’ effect, we’d like to undertake the genetic risk score (GRS) method. There are four common kinds of GRS, (1) simple count genetic risk score (SC-GRS), (2) direct logistic regression genetic risk score (DL-GRS), (3) polygenic genetic risk score (PG-GRS), and (4) explained variance weighted genetic risk score (EV-GRS) 21 , 35 – 37 . The SC–GRS just calculated the number of risk alleles across every SNP at the chosen loci.…”

“…First of all, our study was conducted in a single center. Secondly, the risk factors (age, diabetics, and the methylation level at some promoter regions) which have been previously reported to be associated with HHcy had not been enrolled in the establishment of our ANN model 35 . These factors were turned out to be not associated with our HHcy patients or were not tested in our research.…”