NRN1 and CAT Gene Polymorphisms, Complex Noise, and Lifestyles Interactively Affect the Risk of Noise-Induced Hearing Loss

Shuang-yan LIU Hangzhou Normal University https://orcid.org/0000-0002-9261-2025 Jia-Rui XIN Hangzhou Normal University Zheng LI Hangzhou Normal University Song LEI Ningbo center for disease control and prevention Ying-Qi CHEN Hangzhou Normal University Tian-Yu ZHAO Central people' hospital of zhanjiang Hai-Yan WANG Hangzhou Normal University Liang-Wen XU Hangzhou Normal University Mei-Bian ZHANG Zhejiang Provincial CDC: Zhejiang Provincial Center for Disease Control and Prevention Lei YANG (  yanglei62@hznu.edu.cn ) Hangzhou Normal University https://orcid.org/0000-0001-7618-9936


Introduction
Noise-induced hearing loss (NIHL) is a slowly progressive sensorineural hearing loss caused by long-term exposure to harmful levels of noise.The World Health Organization (WHO) estimates that approximately 22% of the hearing loss in adults is attributable to occupational and environmental noise exposure, and by 2030, almost 1 billion people will be at the risk of NIHL [1] .
As a major occupational health risk, NIHL has become the second-largest occupational disease in China [2].NIHL is thought to be a complex disease caused by genetic and environmental factors.
The main factors include exposure to high levels of noise and individual susceptibility, such as age, gender, education level, smoking frequency, alcohol consumption, and usage practice of hearing protection devices [3][4][5][6].Therefore, the single-locus method may not be appropriate to study common complex disorders such as NIHL.
Noise is the most common environmental factor leading to occupational hearing loss.The noise exposure metrics used in most previously published studies mainly concentrates on equivalent continuous sound level (Leq) and cumulative noise exposure (CNE).These metrics have been established based on the study of Gaussian noise and the equal-energy hypothesis (EEH), which assumes that the damage to the auditory system caused by noise exposure is proportional to the duration of exposure multiplied by the noise intensity.However, the EEH has been found unsuitable for "complex noise or non-Gaussian (non-G)" noise.Complex noise is ubiquitous in industrial and military environments.It is composed of a transient high-energy impulsive noise superimposed on stationary (Gaussian) background noise [7].Both animal experiments and epidemiological studies have shown that the EEH underestimates the cochlear impact of complex-noise exposure.The impact of a complex-noise-induced impulse on the auditory system was assessed using kurtosis first by Erdreich [8].This method has simplified the time-domain variables of noise that affect hearing (e.g., pulse peak value, duration, and inter-pulse distribution) into one easy-to-calculate parameter (i.e., kurtosis), which is convenient for classifying the noise type.A high kurtosis indicates that the impulse of the complex noise was high [9].To date, the efficacy of kurtosis in assessing complex noise has been preliminarily verified in human studies [10,11].
However, few studies have analyzed the interaction between genetic variants, noise exposure (especially kurtosis), and lifestyle factors on modulating NIHL.Previous studies have never reported multidimensional interactions involving multiple (> 7) genes [especially the Neuritin1 (NRN1) gene] and kurtosis.Therefore, in a case-control study with 307 NIHL patients and 307 ageand gender-matched healthy controls, a total of 18 variants in these 7 susceptibility genes (CAT, HSP70, CDH23, CASP3, CASP7, NOX3, and NRN1), three noise metrics (noise kurtosis, CNE, adj-CNE), and four lifestyle factors (smoking, video volume, physical exercise, working pressure) were included to explore the associations of gene-gene, gene-noise-metric, gene-lifestyle-factor, and noise-metric-lifestyle-factor interactions with the risk of NIHL.Our results lay the foundation for a comprehensive prevention program against NIHL.

Subjects
Subjects were continuously recruited between October 2017 and December 2018 from five manufacturing factories with high noise levels in the Zhejiang Province of East China.Inclusion criteria for the subjects were as follows: (1) individuals who had never worked in high noise-level environments from different enterprises, (2) the binaural hearing threshold difference was < 30 dB per frequency; (3) no history of military service; (4) no family history of hearing loss; (5) no history of an ear disease; (6) no history of ototoxic drugs; and (7) no history of diabetes.NIHL was diagnosed based on binaural high-frequency (3000, 4000, and 6000 Hz) average hearing threshold > 25dB.The subjects were divided into two groups-patients with NIHL (n = 307) and controls with normal hearing (n = 307)-who were matched for gender and age ( 3 years).

Questionnaire survey
A questionnaire was designed for each subject based on the needs of the investigation.
Collected information included the following: (1) general information (age, sex, etc.); (2) noise exposure factors (factory, work situation, duration of daily noise exposure, etc.); (3) lifestyle factors (smoking, video volume, physical exercise and working pressure).Variables were defined as follows: (1) smoking: average daily cigarette use for ≥ 1 year; (2) video volume: high video volume ≥ 40% of the maximum volume; low video volume < 40% of the maximum volume; (3) regular physical exercise: average physical exercise for once a month or more often for ≥ 1 year.Participants completed study questionnaires and met with trained investigators in a face-to-face interview.All the participants signed the informed consent form, and the study was approved by the Science Ethics Committee of Hangzhou Normal University (2017LL107).

Noise waveform recording and analysis
A digital noise dosimeter (ASV5910-R, Hangzhou Aihua Instrument Co., Ltd.) that can operate continuously at a sampling rate of 48 kHz was used to record the noise for each subject for the entire shift duration.Eight-hour, continuous equivalent A-weighted sound levels (LAeq,8h) can be measured with a noise dosimeter, which was attached to the clothing of the participant at the shoulder by clips, with the microphone pointed up (Supplementary material 1).The measurement time was 8 hours per shift.A sound level calibrator (Hangzhou Aihua Instrument, AWA6221B) was used to calibrate the noise dosimeter before and after each sampling cycle.MATLAB software (Natick, MA) was used to calculate the sampling kurtosis in a continuous 40-s window of the noise signals during the entire shift.The equation used to calculate kurtosis is shown in Formula 1.
) 2   (1) where, x i is the i th value, ̅ is the sample mean, and β is noise kurtosis.Theoretically, the kurtosis value of Gaussian noise is 3 (β = 3) and that of complex non-Gaussian noise is greater than 3.The larger the kurtosis value, the higher the impulse of the complex noise.The selection of a 40-s window is acceptable for kurtosis measurement based on a 48-kHz sampling rate, as observed from previous animal data [22,23].The median kurtosis calculated in a 40-s window was used as the kurtosis value of the entire shift time.In this study, a median kurtosis of 4 was used as the boundary value between Gaussian and complex non-Gaussian noise.
Both noise level and noise exposure time should be used to assess NIHL.Therefore, a comprehensive noise exposure metric (CNE) was used to quantify noise energy for each worker according to Formula 2 [10]: where, LAeq,8h is the equivalent continuous A-weighted noise exposure level normalized to an 8-h working day and T is the time of noise exposure in years.CNE is measured in dB (A) per year.
To incorporate kurtosis (β) into the evaluation of complex noise environments and unify CNE calculations for epidemiologic data, including both Gaussian and complex noise, the kurtosisadjusted CNE (adj-CNE) was calculated according to Formula 3 [10]: When Gaussian noise has a kurtosis of β = 3, the term [ ln(β)+1.9 log (2) ] becomes equal to 10.Thus, for Gaussian noise, the adj-CNE equals the unadjusted CNE.Equation (3) shows that when LAeq,8h is fixed, the adj-CNE will be larger for complex noise (β > 3) than that for Gaussian noise (β = 3).

Hearing testing and hearing loss diagnosis
Experienced otolaryngologists performed pure-tone audiometry for the left and right ears of each participant at 500, 1000, 2000, 3000, 4000, 6000, and 8000 Hz in a sound-insulated room with background noise < 25 dB (A) (Supplementary material 2).All the subjects were required to be outside of their daily noise environment for at least 16 h before the test.The results of the pure-tone audiometry were adjusted according to gender and age by following the ISO 1999-2013 standard.
The detailed information about the screened SNPs is shown in Table 1.We performed the genotyping analysis using the Kompetitive allele-specific polymerase chain reaction (KASP) method as previously described [17].The primer and probe sequences are shown in Tables S1 and    S2.To control the quality, we randomly selected 10% of the samples and re-classified the genes; the concordance of the 18 SNPs was > 95%.

Statistical analysis
Normally distributed continuous variables are expressed as mean ± standard deviation (SD), and categorical variables are presented as percentages.Student's t-test and the Chi-square test were used to compare the continuous variables and categorical variables, respectively, between the cases and controls.Non-normally distributed continuous variables were expressed as median (with the lower and upper quartiles) [M (P25, P75)] and analyzed using the Mann-Whitney U test.The cutoff values for the CNE and adj-CNE were determined to be 97.1420dB(A) and 96.9939 dB(A), respectively, based on the receiver operating characteristic (ROC) curve between CNE, adj-CNE, and NIHL.The Hardy-Weinberg equilibrium (HWE) was tested using the Chi-square test.The generalized multifactor dimensionality reduction (GMDR Software Beta 0.9, www.ssg.uab.edu/gmdr/)method [24] was used to examine the effects of all possible interactions.
The sign test of cross-validation consistency (CVC), testing balanced accuracy (TEBA), and trained balanced accuracy (TRBA) were calculated.A multivariate logistic regression model was used for the stratified analysis of the significant interactions obtained from the GMDR.Multiple comparisons were corrected using the Benjamini-Hochberg procedure.P < 0.05 indicated that the differences were statistically significant (shown in bold in the following tables).

Results
A total of 614 participants (474 males and 140 females), including 307 NIHL patients and 307 controls, were selected (Table 2).The median age of the subjects was 35 years.The median kurtosis in the NIHL group was 7.25 (4.63-14.30),which was significantly higher than that in the control group [5.85 (4.06-12.51);P = 0.006].The proportion of the subjects exposed to complex noise (β ≥ 4) was significantly greater in the NIHL group than that in the control group (P = 0.038).The median HTL346 in the NIHL group was 36.83(29.83-49.83)dB, which was significantly higher than that in the control group [17.83(14.17-21.00)dB; P < 0.001).The proportion of the subjects with high CNE (≥ 97.1420), high adj-CNE (≥ 96.9939), smoking habit, propensity to watch videos at high volume, or sedentary lifestyle was significantly greater in the NIHL group than in the control group (P < 0.05).However, there was no significant difference in education level or working pressure between the two groups (P > 0.05).
The genotype frequencies among the cases and controls did not deviate from the HWE for any of the 18 SNPs (P > 0.05, Table 3).We assessed for NIHL-related interactions among the 18 genetic variants by using GMDR.Consequently, 18 models were generated from the 18 SNPs (Table 4).
After adjusting age, gender, education level, years of noise exposure, kurtosis, CNE, adj-CNE, smoking, video volume, physical exercise, and working pressure, a significant two-locus model (P = 0.0107) involving NRN1 rs3805789 and CAT rs7943316 was found (Table 4, Figure S1).The CVC of this two-locus model was 10/10, and the TEBA was 0.5768.We then conducted a stratified analysis for the significant models by using logistic regression.When compared with the subjects  S3).
We next asked whether there were any multidimensional interactions between the genes and noise metrics by using the GMDR method.After adjustments were made for age, gender, education, smoking, video volume, physical exercise, and working pressure, the best model for the risk of NIHL was found to be the interaction between NRN1 rs3805789, CAT rs7943316, and kurtosis.This interaction had the score of 10/10 for CVC and 10 for the sign test (P = 0.0010; Table 5A, Figure S2).S4).In the GMDR model, a two-locus model including NRN1 rs3805789 and CAT rs7943316 was found to be significant.This observation is consistent with the results of gene-gene interactions.A four-locus model including NRN1 rs3805789, CAT rs7943316, kurtosis, and adj-CNE was found to be the interaction, in which the CVC was 10/10, and the TEBA was 0.5856 (P = 0.0107; Table 5A).In addition, a five-locus model was also identified for the risk of NIHL.In this model, the CVC was 10/10, and TEBA was 0.5856 (P = 0.0107; Table 5A).
The GMDR model was used to screen for the best gene-lifestyle-factor combinations.After adjusting age, gender, education level, years of noise exposure, kurtosis, CNE, and adj-CNE, the best six-locus model involving NRN1 rs3805789, CAT rs7943316, smoking, video volume, physical exercise, and working pressure for the risk of NIHL was found to be the interaction, which scored 10/10 for CVC and 9 for the sign test (P = 0.0010; Table 5B).A four-locus model involving NRN1 rs3805789, CAT rs7943316, smoking, and physical exercise was found to be the interaction, which scored 9/10 for CVC and 9 for sign test (P = 0.0107; Table 5B).A three-locus model involving NRN1 rs3805789, CAT rs7943316, and working pressure was found to be the interaction, which scored 5/10 for CVC and 9 for the sign test (P = 0.0107; Table 5B).Moreover, a five-locus model was also identified for the risk of NIHL.The corresponding CVC and TEBA were 7/10 and 0.5570, respectively (P = 0.0107; Table 5B).
We next evaluated the interaction combinations between noise metrics and lifestyle factors via the GMDR model.The results revealed that, after adjusting age, gender, and education level, a fivelocus model involving smoking habit, video volume, physical exercise, working pressure, and kurtosis was found to be the interaction, which scored 10/10 for CVC and 9 for the sign test (P = 0.0107; Table 5C).A four-locus model involving smoking, video volume, physical exercise, and working pressure was found to be the interaction (P = 0.0107; Table 5C), in which the CVC was 10/10, and the TEBA was 0.5509.A seven-locus model involving smoking, video volume, physical exercise, working pressure, kurtosis, CNE, and adj-CNE was found to be the interaction, which scored 10/10 for CVC and 9 for the sign test (P = 0.0107; Table 5C).Furthermore, a sixlocus model was also identified for the risk of NIHL.The corresponding CVC and TEBA were 7/10 and 0.5437, respectively (P = 0.0107; Table 5C).

Discussion
In the current study, the association between gene polymorphisms, noise metrics, lifestyle factors, and NIHL was preliminarily explored using univariate analysis.The GMDR method was used to detect the association of the interaction among multiple factors with the risk of NIHL.The GMDR method explores interactions by collapsing the high-dimensional interactions of multiple factors into a single dimension.This method not only avoids biases associated with disease risk by adjusting confounding covariates, but also explores complex multi-locus interactions between genetic and environmental factors.Over the past ten years, the GMDR method has been widely applied to analyze the associations of gene-gene and gene-environment interactions with many complex diseases [25][26][27].
Increasing evidence has shown that multiple genes are closely associated with susceptibility to NIHL.Given that multiple genetic loci with moderate effects fail to reach genome-wide significance due to the limited power in most genetic studies [28], the present study focused on the associations of multi-locus interactions with NIHL risk by analyzing 18 variants in 7 susceptibility genes via the GMDR method.These genes were CAT, HSP70, CDH23, CASP3, CASP7, NOX3, and NRN1.These risk genes play significant roles in apoptosis, cell adhesion, and oxidative stress during the development of NIHL.We identified for the first time that the interaction between NRN1 rs3805789 and CAT rs7943316 increased susceptibility to NIHL.We further validated this genetic interaction via stratified analysis.The results illustrated that subjects carrying NRN1 rs3805789-CC and CAT rs7943316-AT, NRN1 rs3805789-CT and CAT rs7943316-AA, NRN1 rs3805789-CT and CAT rs7943316-TT, NRN1 rs3805789-CT/TT and CAT rs7943316-AA, or NRN1 rs3805789-CC and CAT rs7943316-AT/TT had higher risks of NIHL than those with NRN1 rs3805789-CC and CAT rs7943316-AA.Yang et al. [13] found that CAT rs208679 and rs769217 were significantly associated with the risk of NIHL.A study by Wang et al. [12] have studied the association of CAT rs7943316 with NIHL susceptibility.Their results indicated that carriers of T allele (AT+TT) of rs7943316 have significantly higher risks of NIHL than those with AA genotype (P < 0.05), and observed that a significant interaction model involving GJB2 rs4880, SOD2 rs137852540, and CAT rs769214 might associated with NIHL.These results are similar to our results presented here.CAT is an oxidative-stress gene.Its mutation weakens the anti-oxidant system in the cochlea, thereby hampering the elimination of the reactive oxygen species generated by noise exposure.
Consequently, the structure and function of the cochlea are impaired, ultimately causing hearing loss.Furthermore, noise exposure can damage cochlear hair cells and ribbon synapses between hair cells and nerve fibres [29][30][31].NRN1 is a small polypeptide closely related to the plasticity of neurites in the human central nervous system.As a neurotrophic factor, NRN1 has multiple effects in the nervous system.It can significantly promote the growth and branch formation of neurites [32] and establishment of functional synapses [33].Additionally, it is necessary for the survival of neurons [34].A previous study by our group has shown that a recombinant NRN1 induced extensive neuritogenesis from PC12 cells [35].Picard et al. have observed that knocking out NRN1 impairs the development and plasticity of excitatory visual cortical networks in mice [36].Taken together, these studies reveal that NRN1 may play an important role in NIHL by promoting neurodevelopment and neural plasticity.
Complex noise is ubiquitous in industrial environments.Complex noise with impact and pulse damages the auditory system more than steady-state noise at the equivalent level [10,11,37].
Previous studies on noise have considered only the effect of noise energy on the auditory system, ignored the effect of noise temporal structure, and underestimated the degree of hearing loss associated with complex noise.In this study, we focused on noise kurtosis.This factor was used to describe the characteristics of noise pulse, distinguish between steady-state and complex noises, and assess the effect of complex noise on hearing loss.We found that the NIHL group had a higher median noise kurtosis and a larger proportion of workers exposed to complex noise than the control group, consistent with our previous report [17].In a previous study, mean kurtosis was used to describe the temporal structure of noise, and a mean kurtosis of 10 was used as the boundary value between Gaussian and complex noises.In our study presented here, the median kurtosis of 4 was considered as the boundary.Noise kurtosis damages the auditory system via direct mechanical force and by disrupting the metabolism [10].Considering that complex noise is more harmful to the auditory system than steady-state noise, researchers have begun to adjust the energy parameters or exposure time by using kurtosis.For example, Zhao et al. [10] and Goley et al. [38] have proposed correction methods for exposure time and noise energy, respectively.In this study, the correction method for the exposure time was used to adjust the CNE.We observed an association between CNE, adj-CNE, and NIHL through univariate analysis.The multidimensional interactions between genes and noise metrics were analyzed using the GMDR method.The best model for the risk of NIHL was found to be the interaction among NRN1 rs3805789, CAT rs7943316, and kurtosis.To date, studies have never reported the associations of such interactions with the risk of NIHL.The further stratified analysis revealed that the subjects exposed to complex noise who carried NRN1 rs3805789-CT and CAT rs7943316-TT or NRN1 rs3805789-CT/TT and CAT rs7943316-AA were at a higher risk of NIHL than those exposed to steady noise who carried both NRN1 rs3805789-CC and CAT rs7943316-AA.This observation illustrates that complex-noise exposure increases the effect of the interaction between NRN1 rs3805789 and CAT rs7943316 on NIHL risk.In addition, interaction among NRN1 rs3805789, CAT rs7943316, kurtosis, CNE, and adj-CNE was also identified as a risk factor for NIHL.This result indicates that kurtosis, CNE, and adj-CNE may affect the development of NIHL not only through direct effects but also through interactions with genes.
The risk of NIHL was also affected by lifestyle factors.In this study, we found a significant difference in smoking, video volume, and physical exercise between the two groups.We further investigated gene-lifestyle-factor interactions while investigating the effects of noise-metriclifestyle-factor interactions on NIHL.We observed a cross-reaction involving NRN1 rs3805789, CAT rs7943316, smoking, video volume, physical exercise, and working pressure for the risk of NIHL.Furthermore, we also found a potential five-locus noise-metric-lifestyle-factor interaction model involving smoking, video volume, physical exercise, working pressure, and kurtosis, as well as a seven-locus model including smoking, video volume, physical exercise, working pressure, kurtosis, CNE, and adj-CNE.These results are similar to the previous results of our group [17].
Previous results showed that there were positive interactions between noise kurtosis with smoking, video volume and physical exercise.However, previous studies analyzed only the interactions between two-category variables via crossover analysis and failed to analyze the effects of CNE and adj-CNE on the risk of NIHL.Many studies have shown that smoking-induced hearing loss is likely due to vascular changes, including capillary contraction, increased blood viscosity, and cochlear anoxia [39,40].High-volume noise exposure may lead to hearing loss via a mechanism involving reduced cochlear oxygen tension during and after noise exposure [41].Moreover, lack of exercise affects blood, oxygen, and nutrient flow to the cochlea, leading to the degradation of the stria vascularis (SV).Blood vessels in the SV are essential for transporting necessary factors, such as oxygen and glucose, to the cochlea [42].
This study is superior to previous studies in multiple aspects.First, we firstly focused on the effects of multidimensional interactions on NIHL risk by analyzing 18 variants, three noise metrics, and four lifestyle factors.Second, we identified for the first time that interaction between NRN1 rs3805789 and CAT rs7943316 increases NIHL susceptibility.Third, the associations of the interactions among NRN1 rs3805789, CAT rs7943316, and kurtosis with the risk of NIHL was detected for the first time.However, this study had some limitations as well.First, we could not obtain data regarding other important confounding factors, such as hypertension and diabetes, due to technical reasons.Second, the analyses of lifestyle factors depended on the recollection of the subjects, which can be unreliable.Third, because the sample size is not large enough, the results obtained from this study should be verified by studies involving larger sample sizes.Finally, this study is an association study, the mechanisms of the gene-gene or gene-environmental-factor interactions should be investigated in future laboratory and clinical studies.

Conclusion
In

Figure 2
Figure 2 Joint effects of the NRN1 rs3805789 and CAT rs7943316 with kurtosis on NIHL risk.The reference group was defined as subjects exposed steady-state noise who carry NRN1 rs3805789 CC and CAT rs7943316 AA.Ref: reference group.The odds ratios (ORs) were calculated by the logistic regression analysis after adjustment for age, gender, education, years of exposure, smoking, video volume, physical exercise, and working pressure.

FDR (false discovery
rate) was obtained with the use of Benjamin-Hochberg procedure for controlling the false positive rate in multiple comparison.

Figure S1
Figure S1 Interactions of NRN1 rs3805789 and CAT rs7943316 associated with NIHL risk identified by GMDR analysis.In each cell, the left bar represents a positive score and the right bar a negative score.High-risk cells were indicated by dark shading, low-risk cells by light shading.

Figure S2
Figure S2 Interactions of NRN1 rs3805789, CAT rs7943316 and kurtosis associated with NIHL risk identified by GMDR.0: Steady-state noise; 1: Complex noise.In each cell, the left bar represents a positive score and the right bar a negative score.High-risk cells were indicated by dark shading, low-risk cells by light shading.

Figures Figure 1
Figures

Figure 2 Joint
Figure 2

Figure 2 Joint
Figure 2 The joint effects of the individual interactions of NRN1 rs3805789 and CAT rs7943316 with kurtosis on NIHL risk were analyzed via logistic regression analysis.The results showed that, after adjusting age, gender, education level, years of noise exposure, smoking, video volume, physical exercise, and working pressure, the subjects exposed to complex noise who carried NRN1 rs3805789-CT and CAT rs7943316-TT or NRN1 rs3805789-CT/TT and CAT rs7943316-AA had higher risks of NIHL than those exposed to steady noise who carried NRN1 rs3805789-CC and CAT rs7943316-AA (OR: 5.961, 95% CI: 1.219-29.155;OR: 1.607, 95% CI: 1.035-2.494;P < 0.05) (Figure2, Table

Table 2
General characteristics of subjects between NIHL and control groups conclusion, complex noise, high CNE, high adj-CNE, smoking, high video volume, and sedentary lifestyle are environmental risk factors for NIHL.Concurrence of NRN1 rs3805789 and CAT rs7943316 constitutes a genetic risk factor for NIHL.Complex noise exposure significantly increases the risk of NIHL in subjects with a high genetic risk score.Interactions between genes and lifestyle as well as noise metrics and lifestyle affect the risk of NIHL.These results provide a theoretical basis for screening genetic and environmental risk factors to prevent NIHL.Abbreviations NIHL: Noise-induced hearing loss; CNE: Cumulative noise exposure; adj-CNE: kurtosis-adjusted cumulative noise exposure; EEH: equal-energy hypothesis; ISO: International Standardization Organization; SV: stria vascularis; ROC: receiver operating characteristic; KASP: kompetitive allele specific polymerase chain reaction; Neuritin1: NRN1; Catalase: CAT; Caspase3: CASP3; Caspase7: CASP7; NADPH Oxidase3: NOX3; Cadherin-23: CDH23; Heat shock protein: HSP; CV: coefficient of variation; OR: Odds ratio; CI: Confidence interval; SD: Standard deviation.

Table 3
The genotype and allele frequencies of 18 SNPs in NIHL cases and control subjects

Table 4
Association of Multidimensional Gene-Gene Interactions of 18 SNPs with NIHL Risk

Table 5
Associations of interactions among genes, noise metrics and lifestyle factors with the risk of NIHL

Table S1
TableS3Jointed effects of NRN1 and CAT on NIHL by using the logistic regression models ORs (95%CI) and P value were obtained with the use of multivariate logistic regression analysis after adjustment for age, gender, education, years of exposure, kurtosis, CNE, adj-CNE, smoking, video volume, physical exercise, and working pressure.