To Investigate The Value Of Genetic Risk Score In The Risk Prediction Of Noise-Induced Hearing Loss Based On A Cohort

Background and purposeNoise-induced hearing loss(NIHL)is a progressive hearing loss caused by longterm and repeated exposure to high intensity noise.Individuals exposed to the same intensity of environmental noise have different susceptibility to NIHL.NIHL is a complex disease with the interaction of environmental and genetic factors.Previous studies on its genetic susceptibility mostly remained at the level of a single gene and a few loci,and there were few reports on the combined effects of multiple genes.Based on the NIHL related SNPs screened by the research group in the early stage,a genetic risk scoring(GRS)model is planned to be established to further analyze the contribution of the combined effect of SNPs to the risk assessment of NIHL,to analyze the risk prediction strength of different genetic risk scoring methods for NIHL and to screen the optimal genetic risk scoring method.To provide a reference for the prevention and control of NIHL.Materials and Methods1.object of studySamples participating in NIHL disease risk GRS model construction were employees of a steel plant who participated in the cohort from 2006 to 2015.A total of286 cases were screened and 500 cases were controlled.The samples participating in the NIHL disease risk GRS model verification were the workers of another steel plant,and a total of 107 cases were screened and 231 cases were controlled.2.Gene locus screening and model selectionAccording to the differences in the distribution of each genotype in the case control group,nine SNPs related to NIHL susceptibility were screened out,respectively:rs2289273,rs3813346,rs1738254,rs3735715,rs1131824,rs11004085,rs612969,rs2763979,rs35253356.OR >1 and P < 0.05 were used for model selection.Respectively,codominant model,dominant model and ercessive model were constructed for 9 loci.3.Construction of genetic risk scoring model(1)Genetic risk scoring model was constructed at 9 lociThe GRS model was constructed according to SC-GRS,DL-GRS,PG-GRS and EV-GRS score calculation method of 9 SNPs that were screened and related to NIHL susceptibility.(2)Genetic risk scoring model was constructed at 6 lociBased on the screened 9 loci,6 loci with effect value OR≥1.5 were selected to construct GRS model according to SC-GRS,DL-GRS,PG-GRS and EV-GRS scoring calculation method.Logistics regression was adopted for 9-site model and 6-site model respectively to analyze the disease risk of each rating group(OR and 95%CI),and ROC curve was drawn for the four GRS models to evaluate the advantages and disadvantages of the model construction.4.Validation of genetic risk scoring modelDuring model validation,4 kinds of GRS scores should be calculated for 9 sites and 6 sites respectively according to their models.Also,each score should be divided into groups to obtain OR and 95%CI.The ROC curve of the 4 scoring models should be drawn,and the prior and posterior probabilities of the 4 models should be calculated.These three indicators were used to comprehensively analyze which genetic risk scoring model had higher validity in predicting NIHL risk.result1.Basic information of the research objectIn the model construction group,the binaural high frequency hearing threshold,age and smoking were higher in the case group than in the control group,and other factors were evenly distributed in the two groups.In the model validation group,the mean hearing threshold of both ears in the case group was higher than that in the control group,and other factors were evenly distributed in both groups.2.Selection of gene locus modelrs3813346 and rs612969 were codominant models.rs1738254,rs3735715,rs11004085,rs2763979 and rs35253356 were dominant models.The loci rs2289273 and rs1131824 were recessive models.3.Construction of genetic risk scoring model(1)Genetic risk scoring model was constructed at 9 lociThe SC-GRS,DL-GRS,PG-GRS and EV-GRS models constructed at 9 loci could all predict the risk of HIHL to a certain extent.DL-GRS and PG-GRS showed an obvious trend in the quantity-response relationship between GRS score and NIHL disease.The effect value OR increased with the increase of GRS score,and AUC≥0.650,P < 0.05.(2)Genetic risk scoring model was constructed at 6 loci6 loci with OR≥1.5 were modeled,and the results showed that DL-GRS model could best reflect the quantity-response relationship between the score and NIHL,and the OR value increased with the increase of the score(P < 0.05).Although other models have this metrological-response relationship,some P> 0.05.The AUC of DL-GRS model was 0.642,smaller than that of DL-GRS(0.658)and PG-GRS(0.651)at 9 loci.4.Validation of genetic risk scoring model(1)Validation of 9 locus genetic risk scoring modelIn the validation results of the model,the effect value OR of PG-GRS increased with the increase of the score.In the group with higher score(2～3,≥3),P < 0.05.Compared with the AUC of each validation group,the maximum value of PG-GRS was 0.643,indicating the highest predictive validity.With PG-GRS score of 1.5 as the limit,the risk of disease in the high group was 2.588 times higher than that in the low group(OR=2.588;95%CI=1.353-4.952),and the posterior probability of the highly rated group was 51.16%,which was 19.5% higher than the prior probability.(2)Validation of 6 locus genetic risk scoring modelAt the time of model validation,the quantity-response relationship still existed in DL-GRS,but P > 0.05.In PG-GRS,except the OR value of the 3-4 rating group was lower than that of the 2-3 rating group,the effect value of OR of the other groups increased with the increase of the rating value,and P < 0.05.The maximum AUC of the PG-GRS model was 0.637,which was lower than that of the PG-GRS model constructed at site 9(0.643).Taking score 2 as the boundary,the risk of disease in the high-level group was 2.096 times that in the low-level group(OR=2.096,95%CI=1.238-3.549),and the posterior probability of the high-level group was 44.74%,13.18% higher than the prior probability.ConclusionGene loci rs2289273,rs3813346,rs1738254,rs3735715,rs1131824,rs11004085,rs612969,rs2763979 and rs35253356 are associated with NIHL genetic susceptibility.The PG-GRS calculation method of 9-locus genetic risk scoring model has strong predictive power for NIHL.