Interactive Effects of Occupational Hearing Loss and Glutathion S-Transferase M1 Genotype on Tinnitus Among Noise-exposed Steelworkers

Abstract

Background

Tinnitus is a known long-term health effect of noise exposure. Whether susceptible genotypes predispose noise-exposed workers to tinnitus remains unclear. This study investigated the interactive effects of noise exposure, auditory function, and glutathione S-transferase (GST) gene polymorphisms on tinnitus risk.

Methods

We conducted a cross-sectional study among steelworkers. Participants underwent audiological testing to assess hearing levels and were surveyed for tinnitus. Blood samples were genotyped for GSTM1 and GSTT1 deletion polymorphisms using polymerase chain reaction. Multivariable logistic regression was used to evaluate associations between tinnitus and pure-tone audiometry (PTA), distortion product otoacoustic emission (DPOAE), and GST genotypes while controlling for age and other covariates.

Results

Among 239 male steelworkers (mean age ± SD: 48.29 ± 7.6 years), 27% reported tinnitus. Cumulative noise exposure and thresholds at low- and high-frequency PTA were significantly associated with tinnitus (adjusted odds ratio (OR) = 1.03, 1.06, and 1.04, 95% confidence interval (CI) = [1.00,1.06], [1.02,1.11], and [1.02,1.06], p = 0.045, 0.002, <0.001, respectively). Although neither DPOAE nor GSTM1 nor GSTT1 genotypes showed a significant effect on tinnitus, we observed a significant interaction between low-frequency PTA and the GSTM1 genotype (p = 0.03). Specifically, compared with workers with non-null GSTM1 genes, workers with the GSTM1-null genotype were significantly more susceptible to low-frequency PTA-associated tinnitus (OR = 1.02 and 1.13, 95% CI = [0.96,1.08] and [1.06,1.22], respectively).

Conclusion

The interactive effect of auditory functional status and antioxidant gene polymorphisms suggests that hair cell damage and oxidative stress in the cochlea both contribute to tinnitus development.

1. Introduction

Excessive noise exposure in the steel industry causes irreversible inner ear damage and represents a major occupational health hazard. Prolonged noise can lead to noise-induced hearing loss (NIHL) and tinnitus. The NIHL is a form of sensorineural hearing loss involving cochlear hair cell damage, loss of synaptic connections between hair cells and auditory nerve fibers, and subsequent neural degeneration [1]. On the other hand, how tinnitus is associated with noise exposure is less well known.

Tinnitus is the perception of sounds that do not come from an outside source [2,3]. It can significantly impact the quality of life for millions worldwide [4,5], partly through its associations with depressive symptoms [6,7]. The prevalence of tinnitus increases with age [5,8]. While hearing loss is a well-established risk factor [[9], [10], [11]], tinnitus may also be associated with a vascular etiology [12,13]. However, the underlying mechanisms of tinnitus remain poorly understood.

Noise exposure is a key contributor to tinnitus [[14], [15], [16]], but not all exposed individuals develop tinnitus, suggesting a complex interplay of genetic and environmental factors. To investigate the genetic factors associated with tinnitus, recent studies have explored genetic susceptibility to tinnitus. In particular, one study examined rare coding variants and noncoding common variants in the Han Chinese population and indicated that genetic predispositions may play a role in tinnitus development [[17], [18], [19], [20], [21], [22], [23], [24], [25], [26]].

Among candidate genes, glutathione S-transferases (GST proteins) are of particular interest due to their role as detoxifying enzymes involved in conjugation reactions with toxic compounds. Genetic polymorphisms in GSTs have been linked to susceptibility to noise-induced temporary and permanent threshold shifts in pure-tone audiometry (PTA) (as shown in our previous study [27,28]) and cisplatin-induced ototoxicity [[29], [30], [31]]. Nevertheless, prior research has focused either on genetic predispositions other than GST genes in relation to tinnitus, or on the effects of GST genes on hearing loss.

To address the potential involvement of oxidative stress in auditory function [27,28,32] and tinnitus development [33], we conducted an epidemiological study of noise-exposed steelworkers in Taiwan. Specifically, we investigated the association between tinnitus, auditory function (including PTA and distortion product otoacoustic emission (DPOAE)), and GSTM1 and GSTT1 genotypes. We hypothesized that GSTM1- and GSTT1-null genotypes would exacerbate tinnitus susceptibility in those with noise-related hearing impairment due to diminished antioxidant protection.

The present study is based on the Taiwanese steelworker cohort previously reported by Lin et al [34]. Both studies analyzed 239 workers and used a job-exposure matrix to evaluate cumulative noise exposure. The prior study related cumulative noise exposure to DPOAE. The current study examined tinnitus as the primary outcome. In contrast to the prior work, it incorporates new variables: namely, tinnitus status and GSTM1 and GSTT1 deletion genotypes analyses. The design of the present study allows the first examination of gene–environment interactions specifically for tinnitus. Thus, whereas the prior study characterized the association between DPOAE and cumulative noise exposure, the present work provides insight into tinnitus etiology.

2. Materials and methods

• A.Participants

This study was designed as a cross-sectional investigation of noise-exposed steelworkers in Taiwan. Participants were recruited from the largest steel manufacturing company (CSC) in Taiwan. Initially, we recruited 347 male employees on the day shift. Most participants started working at the factory after graduating from college and continued until retirement. They were excluded from further analysis if they had: (1) a history (acute or chronic) of head injuries, otolaryngological diseases, other diseases affecting hearing, or a family history of congenital deafness (n = 15), (2) abnormal results on a tympanometry test (n = 24), (3) a history of noise exposure before working at this factory (n = 51), (4) a history of leisure noise exposure (n = 4), (5) use of ototoxic medications in the recent 6 months (n = 11), and (6) missing DPOAE measurements (n = 3). Following these exclusions, data from 239 participants were analyzed (Fig. 1). Participants with abnormal tympanometric findings were excluded to minimize confounding effects of middle ear pathology as this study aimed to investigate the impact of environmental exposure.

Fig. 1.

Diagram of cohort recruitment for this study. It shows the flow of participants from initial recruitment through final analysis including exclusion criteria and sample sizes at each stage. “Other diseases” in the figure refers to a history of head injuries, otolaryngological diseases, other diseases affecting hearing, or a family history of congenital deafness. DPOAE, distortion product otoacoustic emission.

This study was performed in accordance with the guidelines contained in the Declaration of Helsinki, and it was approved by the Institutional Review Board of Taipei Medical University (N201906017, issued on 2019/6/18). Informed consent was obtained from all participants.

• B.Exposure Assessment and Auditory Examination

All assessments were performed on a regular workday. The assessments included structured interviews, otoscopic examinations, tympanometry, PTA, and DPOAE measurements. Personal noise exposure measurements were conducted separately.

• (1)Questionnaire

We used a structured questionnaire to gather information on demographics (age, education, etc.), work history, health habits (cigarette smoking and alcohol consumption), medical history (including tinnitus), noise exposure (both work-related and leisure activities), medication use (including specific ototoxic medications), dietary supplements, and hearing protection usage.

Participants who smoked cigarettes provided details on duration and daily quantity to calculate total tobacco use. Those reporting regular alcohol consumption were classified as alcohol users. Because artillery exposure during military service was considered a significant source of pre-employment noise, participants with such a history were excluded. Participants who reported taking Aspirin, Lasix (the brand name of the commonly prescribed furosemide medications in Taiwan), gentamycin, or chemotherapy within the past 6 months were excluded due to potential hearing damage from these medications. In addition to asking whether they had tinnitus, we also asked about the severity of their tinnitus. Level 0 means no tinnitus, Level 1 means tinnitus is only present when I pay attention to it, Level 2 means tinnitus is bothering sometimes, Level 3 means tinnitus is usually present and causes problems, Level 4 means tinnitus is always present and causes a lot of problems, Level 5 means tinnitus causes extreme problems and cannot be tolerated. Trained interviewers conducted interviews with each participant to clarify questions and ensure questionnaire comprehension.

• (2)Auditory functions

To ensure accurate hearing test results, eligible participants were instructed to avoid exposure to loud noise for at least 48 hours before the assessment. Pre-shift auditory assessments were conducted on the morning of the test between 6:30 AM and 7:30 AM. An otolaryngologist (C.Y. Lin) used a handheld otoscope to examine participants' ears for occluding cerumen and assess the status of the outer and middle ear.

An audiologist performed tympanometry, PTA (air conduction but not bone conduction), and DPOAE measurements for all participants in a sound-attenuating chamber meeting the International Organization for Standardization (ISO) 8253-1:1989 standards [35]. This ensures minimal background noise (≤25 dBA) for accurate testing. Bone-conduction audiometry was not performed regularly in this study.

An audiologist used impedance tympanometry (Grason-Stadler GSI-37 Auto Temp, Eden Prairie, MN, USA) to assess middle-ear function. If tympanometry was classified as type B or type C in any ear, the participant was considered to have an abnormal tympanogram (i.e., the participant might have abnormalities in the middle ear, eardrum, or Eustachian tube). Participants with abnormal tympanograms were excluded. To measure PTA, the audiologist used a Grason-Stadler GSI-68 audiometer (Eden Prairie, MN, USA), which was calibrated in accordance with ISO 389-2:1994 standards [36]. The audiometric procedures were conducted in accordance with ISO 8253-1:1989 standards [35]. Telephonics TDH-39P earphones (Telephonics Corp., Santa Ana, CA, USA) were the main audiometric earphones used in this study. An ascending method with 5 dB steps was used to obtain the hearing threshold levels of both ears (i.e., the Hughso–Westlake method). The test frequencies for each ear were 500, 1000, 2000, 3000, 4000, 6000, and 8000 Hz.

The DPOAE were measured using a GSI Corti (Grason-Stadler, Eden Prairie, MN, USA). The stimulus intensities of f₁ and f₂ were 65 and 55 dB sound pressure level, respectively. The f₂-to-f₁ ratio was set at 1.22, and the 2 f₁ − f₂ DPOAE was recorded at six points over a frequency range for f₂ (i.e., 1000, 2000, 3000, 4000, 6000, and 8000 Hz) [37].

• (3)Cumulative noise exposure

During a work shift (7 AM to 3 PM), we asked the participants to wear TES-1355 Noise Dose Meters (TES Electrical Electronic Corporation, Taipei, Taiwan) to measure personal noise exposure for each participant. The time-weighted average (TWA) of the A-weighted sound pressure level (L_eq-8h) was calculated to present their daily noise exposure. Since most workers in this steel factory had consistent job locations and tasks, past daily noise levels were assumed to represent historical noise exposure. Regarding those participants who had changed their job locations and tasks, we used questionnaires and historical noise monitoring data from the factory (described in [28]) to estimate their former cumulative noise exposure.

We also measured the A-weighted sound levels (dBA) through a frequency-dependent filter to mimic the effects of human hearing by using a sound level meter (TES-52, TES Electrical Electronic Corporation, Taipei, Taiwan) to measure their work areas. The meter met American National Standard Institute specifications. The meter was placed at a height equivalent to that of the average worker's ears (155 cm).

If a worker used hearing protection in a specific area, the corresponding noise level was adjusted based on the device's noise reduction rating. It is important to note that the noise reduction rating might overestimate real-world noise.

To calculate the total noise exposure level over a worker's job history by using historical TWA exposure over the years (L_eq-total, dBA-year), we used information about the participants' employment history including successive job titles, departments, and the corresponding dates. The job titles were grouped into 99 groups representing similar levels of noise exposure, based on the workplace location and assigned tasks. The job-exposure matrix is one of many age–exposure matrices, and it was developed to compare the effects of exposure among industries in which the age distributions differ. Here, we adapted the “equal energy rule” to convert noise exposure levels, which vary over the course of a lifetime, into continuous noise exposure levels of equivalent impact on the auditory system [38]. Specifically, the “equal energy rule” states that approximately equal energies (with the A-weighting of different frequencies) have approximately equal effects on hearing impairment. L_eq-total was determined using the logarithmic sum of the TWA noise levels, as follows:

$$L_{eq-total}=10\times \log \left[\sum T_i\times {10}^{\frac{L_{eq-8h_i}}{10}}\right]\text{,}$$

where T_i is the duration of each job i (years) and L_eq-8hi is the predicted TWA noise exposure for the job held during i.

• (4)Genotyping of GSTM1 and GSTT1 genes

Venous blood samples were collected in heparinized tubes at the factory and placed immediately in a refrigerated container. Samples were labeled with a numerical identifier and transported on dry ice to the laboratory on the same day. Genomic DNA was extracted from all blood samples using standard phenol/chloroform extraction techniques [39].

The GSTM1 genes (chr 1: 109687814-109709039 on GRch38) and GSTT1 (chr 22: 270308-278486 on GRch38) genes were simultaneously genotyped using a multiplex polymerase chain reaction (PCR) approach as described by [28]. The DNA sample was amplified with three pairs of primers as shown in Table 1. The PCR was carried out in a total volume of 50 μl, containing 0.5 μg genomic DNA, 0.4 μM of each of the above primers, 1 U Taq DNA polymerase (AmpliTag; Perkin-Elmer Inc., Woodstock, NY, USA), and 0.15 μM of dNTP (Boehringer Mannheim GmbH, Mannheim, Germany). The PCR conditions were as follows: 95°C for 3 minutes, followed by 30 cycles of 95°C for 30 seconds, 58°C for 30 seconds, and 72°C for 30 seconds, with a final extension at 72°C for 10 minutes. A 10 μl PCR aliquot was run on 2% agarose gel stained with ethidium bromide. The PCR produced three DNA fragments of 480 bp (GSTT1), 350 bp (genes that produce albumin), and 215 bp (GSTM1). The albumin gene was used as an internal control to assure successful amplification in the PCR processes. Lack of such PCR product was considered poor quality and was excluded. In both GSTT1 and GSTM1 genes, gene deletions are known to be responsible for the existence of null alleles. Individuals homozygous for a given null allele lack the respective PCR-amplified DNA fragment.

• (5) Statistical analysis

Table 1.

Polymerase chain reaction (PCR) assay primers and product sizes. The sequence for albumin was chosen as the internal control to ensure the quality of PCR processes

Gene	Primer (5′-3′)	Sequence	Product size (bp)
GSTT1 (Deletion)	Forward	TTC CTT ACT GGT CCT CAC ATC TC	480
	Reverse	TCA CCG GAT CAT GGC CAG CA
GSTM1 (Deletion)	Forward	GAA CTC CCT GAA AAG CTA AAG C	215
	Reverse	GTT GGG CTC AAA TAT ACG GTG G
Genes that produce albumin	Forward Reverse	GCC CTC TGC TAA CAA GTC CTA C GCC CTA AAA AGA AAA TCG CCA ATC	350

R software (R version 4.0.3, October 10, 2020) was used for all analyses. Data distribution was assessed using the Shapiro–Wilk test. Nonparametric tests were used for variables that did not meet the assumption of normality. In our initial analysis, we compared age, cumulated noise exposure, PTA, and DPOAE using the Wilcoxon signed-rank test. The chi-square test was used to compare categorical variables. A two-tailed p value <0.05 was considered statistically significant. Subsequently, we employed multivariate logistic regression models to assess the effects of age, cumulated noise exposure, hypertension, dyslipidemia, cigarette amount, alcohol consumption, and GSTM1 and GSTT1 genotypes on the odds of developing tinnitus. These covariates chosen a priori based on the literature: The variables such as age, hypertension, dyslipidemia, and cigarette amount were chosen due to the vascular origins of some cases of tinnitus [12,13] (there were only 7 participants reported diabetes, so we did not include diabetes as a covariate), and alcohol consumption was included because previous studies found associations between tinnitus and alcohol consumptions [9]. All these factors were included in the multivariate logistic regression as adjustment factors. We performed sensitivity analyses to explore potential interactions between cumulative noise exposure and auditory functions and other factors such as GSTM1 and GSTT1 genotypes. In addition to logistic regression, we also performed ordinal logistic regression analysis using tinnitus severity levels and compared the results with those using logistic regression analysis.

3. Results

Fig. 2 illustrates the distributions of age and cumulative noise exposure for the 239 participants. The average age was 48.29 years (with a standard deviation of 7.6 years), with a median age of 50.5 years. The average cumulative noise exposure was 90.26 dB-year (with a standard deviation of 11.14 dB-year), with a median of 87.86 dB-year. Tinnitus was reported by 65 participants (27%). The Wilcoxon test revealed no significant difference in age between those with and without tinnitus (p = 0.16) (Table 2). On the other hand, the tinnitus group showed significantly higher cumulative noise exposure than the no-tinnitus group (p = 0.04) (Table 2).

Fig. 2.

Distribution of age and cumulative noise exposure and tinnitus severity among the participants.

Table 2.

Demographic characteristics of participants with and without tinnitus

Variable	Tinnitus group (n = 65)	No tinnitus group (n = 174)	p value
Age (years)	49.74 (6.02)	47.74 (8.06)	0.16 ‡
Cumulative noise exposure Leq-total (dBA-years)	92.86 (10.26)	89.29 (11.33)	0.04 ‡
Low-frequency PTA (dB)	22.79 (8.48)	18.78 (8.10)	<0.001 ‡
High-frequency PTA (dB)	40.63 (18.19)	29.74 (16.59)	<0.001 ‡
Low-frequency DPOAE (dB)	2.02 (5.29)	2.91 (4.46)	0.22 ‡
High-frequency DPOAE (dB)	-12.24 (5.19)	-10.45 (5.68)	0.02 ‡
Hypertension	11 (17%)	34 (20%)	0.62†
Dyslipidemia	13 (20%)	35 (20%)	1†
Cigarette amount (cigarettes-years)	6.95 (13.14)	7.78 (12.73)	0.53 ‡
Alcohol consumption	4 (6%)	40 (23%)	0.002†
GSTM1 null genotype	28 (43%)	86 (49%)	0.56†
GSTT1 null genotype	41 (63%)	95 (55%)	0.38†

Footnote: Values are presented as mean (standard deviation) for continuous variables and number (percentage) for categorical variables. Wilcoxon tests (presented with ‡) were used for continuous variables, whereas chi-square tests (presented with †) were used for categorical variables. A two-tailed p value <0.05 was considered statistically significant. “Low-frequency” is defined as tones below 2 kHz and “high-frequency” as tones above than 2 kHz. DPOAE, distortion-product otoacoustic emissions; PTA, pure-tone audiogram.

Fig. 3 depicts the mean and standard errors of PTA thresholds and DPOAE levels across all tested frequencies for participants with and without tinnitus. The data demonstrated a general trend that participants with tinnitus tended to have poorer hearing as indicated by higher PTA thresholds and lower DPOAE levels compared to those without tinnitus. Statistical analysis using the Wilcoxon test confirmed these observations, revealing significantly worse performance in low- and high-frequency PTA and high-frequency DPOAE in the tinnitus group than the no-tinnitus group (Table 2).

Fig. 3.

PTA thresholds and DPOAE levels in participants with tinnitus (solid lines) and no tinnitus group (dashed lines). The error bars represent standard error (SE) of the mean. The dotted line indicates background sound level. DPOAE, distortion-product otoacoustic emissions; PTA, pure-tone audiogram.

In addition to the significant differences in cumulative noise exposure, low- and high-frequency PTA, high-frequency DPOAE between the tinnitus and no tinnitus group, Table 2 also shows significant between-group differences in alcohol consumption (p = 0.002) but not in hypertension, dyslipidemia, cigarette amount, GSTM1, or GSTT1.

We conducted a series of multivariate logistic regression analyses to examine the association between each target variable of interest (e.g., cumulative noise exposure, PTA, or DPOAE) and the presence of tinnitus. In each model, one target variable of interest was entered as the main predictor, while adjusting for the same set of potential confounders including age, hypertension, dyslipidemia, cigarette amount, alcohol consumption, GSTM1, and GSTT1.

Results revealed associations between tinnitus and cumulative noise exposure and low- and high-frequency PTA, but not low- or high-frequency DPOAE or GSTM1 or GSTT1 (Table 3). Specifically, greater cumulative noise exposure and worse PTA thresholds at both low- (0.5, 1, and 2 kHz) and high-frequency PTA (3, 4, 6, and 8 kHz) were significantly correlated with an increased risk of tinnitus. Collinearity diagnostics based on variance inflation factors revealed no substantial multicollinearity among the independent variables, with all variance inflation factors <5.

Table 3.

Multivariate logistic regression analysis of factors associated with tinnitus

Variable	Adjusted OR (95% CI)	p value
Cumulative noise exposure Leq-total (dBA-years)	1.03 (1.00,1.06)	0.045
Low-frequency PTA (dB)	1.06 (1.02,1.11)	0.002
High-frequency PTA (dB)	1.04 (1.02,1.06)	<0.001
Low-frequency DPOAE (dB)	0.98 (0.92,1.05)	0.57
High-frequency DPOAE (dB)	0.95 (0.89,1.01)	0.12
GSTM1 (non-null vs. null)	1.54 (0.85, 2.83)	0.16
GSTT1 (non-null vs. null)	0.74 (0.40,1.34)	0.32

Footnote: Values are presented as odds ratios (ORs) with 95% confidence intervals (CIs). The logistic regression models were adjusted by age, hypertension, dyslipidemia, cigarette amount, and alcohol consumption. DPOAE, distortion-product otoacoustic emissions; PTA, pure-tone audiogram.

One caveat is that, high-frequency DPOAE showed a significant difference between the tinnitus and no-tinnitus groups in Wilcoxon test (Table 2), but high-frequency DPOAE showed no significant effect in logistic regression analysis after adjustment (Table 3). In the same logistic regression model, while covariates were selected using a backward-elimination procedures, the covariate “age” made high-frequency DPOAE from having a significant effect on tinnitus to no significant effect on tinnitus. Another caveat is thtat multivariate ordinal logistic regression analysis with tinnitus severity levels showed that the binary classification of tinnitus presence was robust in logistic regression analysis (Table 4).

Table 4.

Multivariate ordinal logistic regression analysis of tinnitus severity

Variable	Adjusted OR (95% CI)	p value
Cumulative noise exposure Leq-total (dBA-years)	1.03 (1.00,1.06)	0.03
Low-frequency PTA (dB)	1.06 (1.02,1.10)	0.001
High-frequency PTA (dB)	1.04 (1.02,1.06)	<0.001
Low-frequency DPOAE (dB)	0.98 (0.92,1.05)	0.63
High-frequency DPOAE (dB)	0.95 (0.90,1.01)	0.10
GSTM1 (non-null vs. null)	1.48 (0.82,2.67)	0.19
GSTT1 (non-null vs. null)	0.73 (0.40,1.32)	0.30

Footnote: Values are presented as odds ratios (ORs) with 95% confidence intervals (CIs). The ordinal logistic regression models were adjusted by age, hypertension, dyslipidemia, cigarette amount, and alcohol consumption. DPOAE, distortion-product otoacoustic emissions; PTA, pure-tone audiogram.

To evaluate the robustness of our findings and explore the effect of GST genotypes on tinnitus, we conducted sensitivity analyses using different datasets (age <50 years vs. age ≥50 years, with and without hypertension, with and without dyslipidemia, cigarette consumption vs. no cigarette consumption, alcohol consumption vs. no alcohol consumption, GSTM1 non-null vs. null genotype, and GSTT1 non-null vs. null genotype. Fig. 4 showed a significant interaction between low-frequency PTA and GSTM1 genotype (p = 0.03), with the finding that higher thresholds at low-frequency PTA were significantly associated with increased odds of tinnitus among workers lacking the GSTM1 genes (OR = 1.13, 95% CI = [1.06,1.22]), whereas the corresponding estimate was close to null among workers with the GSTM1 genes (OR = 1.02, 95% CI = [0.96,1.08]). In addition, there was a significant interaction between cumulative noise exposure and dyslipidemia (p = 0.04), with the finding that higher cumulative noise exposure was associated with increased odds of tinnitus among workers without dyslipidemia (OR = 1.05, 95% CI = [1.02,1.09]), whereas the corresponding estimate was close to null among workers with dyslipidemia (OR = 0.95, 95% CI = [0.88,1.04]).

Fig. 4.

Sensitivity analyses of GST genotypes and clinical subgroups in relation to tinnitus risk. Stratified analyses were conducted by age (<50 years [open circles] and ≥50 years [solid diamonds]), hypertension, dyslipidemia, cigarette consumption, alcohol consumption, and GSTM1/GSTT1 genotypes (non-null [open circles] and null [solid diamonds]). Significant interactions were indicated by asterisks: interaction p values: low-frequency PTA × GSTM1 = 0.03; cumulative noise × dyslipidemia = 0.04. GST, glutathione S-transferase; PTA, pure-tone audiogram; smoke, cigarette amount.

4. Discussion

In this study of noise-exposed Taiwanese steelworkers, we found that tinnitus was associated with higher cumulative noise exposure and higher thresholds at low- and high-frequency PTA, while neither DPOAE measures nor GSTM1/GSTT1 genotypes had main effects. Specifically, workers with greater cumulative noise exposure and worse low- and high-frequency PTA thresholds were more likely to report tinnitus. Importantly, there was a significant interaction between tinnitus and low-frequency PTA: only the workers lacking GSTM1 genes showed significantly higher tinnitus odds per unit increase in low-frequency PTA thresholds than those with intact GSTM1 genes. To our knowledge, this is the first study investigating the role of GSTM1/GSTT1 polymorphisms in tinnitus risk. Our findings revealed a gene-environment interaction: GSTM1 and GSTT1 polymorphisms by themselves do not trigger tinnitus, but when combined with NIHL, the presence or absence of detoxifying enzymes can modify the pathological processes leading to tinnitus.

Noise exposure and NIHL have been linked to tinnitus [40]. Tinnitus is thought to originate from aberrant neural activity triggered by cochlear injury. Noise exposure initially damages outer hair cells (OHC), subsequently affecting the reticular lamina and stria vascularis, and ultimately impairing inner hair cells, nerve fibers, and the fibrocytes in the spiral ligament, resulting in a permanent threshold shift in PTA [[41], [42], [43], [44]]. While DPOAE primarily estimates OHC status, PTA reflects the combined function of inner hair cells and OHC [45].

Although cochlea damage is known to be a critical factor in tinnitus, the relationship between hearing loss and tinnitus is complex as not all individuals with hearing loss experience tinnitus, and vice versa. Previous studies have shown that extended high-frequency PTA or precision PTA (with fine frequency resolution) can detect hearing loss in individuals who have tinnitus and normal PTA results [46,47]. Additionally, previous studies have found that tinnitus is associated with reduced wave I amplitude in auditory brainstem response [48] and increased intensity discrimination thresholds [49], suggesting decreased cochlea outputs in individuals with tinnitus despite normal PTA. These subtle cochlear deficits may also contribute to functional impairments as reflected in elevated speech audiometry thresholds observed in patients with tinnitus [50].

The gene-environment interaction can explain at least part of the complexity of the relationship between hearing loss and tinnitus. The GST proteins, specifically mu class (including GSTM1, GSTM2, GSTM3, etc.), have been found in animal hair cells and the stria vascularis [51,52]. Previous studies investigating the gene-environment interaction in hearing loss found that GST proteins demonstrate a protective effect on hair cells, partially mediated through the stria vascularis after noise exposure [[53], [54], [55], [56]]. GSTM1, GSTT1, and GSTP1 polymorphisms have been associated with noise-induced temporary threshold shift [27], and GSTM1 genes are linked to DPOAE in noise-exposed workers [57] and show a protective effect against NIHL [58].

The gene-environment interaction revealed by our study supports a dual-mechanism model in which hair cell damage and impaired antioxidant capacity jointly raise tinnitus risk. In our steelworkers, higher PTA thresholds (indicating hair cell loss) were strongly predictive of tinnitus, in line with the notion that tinnitus often begins in the damaged cochlea. The interactive effect with GSTM1 suggests that oxidative damage is a key component of tinnitus: noise exposure generates reactive oxygen species in the inner ear, and GST enzymes are vital for neutralizing these radicals. Workers with the GSTM1-null genotype lack one pathway of glutathione conjugation, potentially allowing more free-radical damage. In contrast, workers with functional GSTM1 may better scavenge free radicals, mitigating some cochlear injury or neural hyperactivity.

In this study, participants with tinnitus exhibited lower high-frequency DPOAE levels compared to those without tinnitus. However, this association between tinnitus and high-frequency DPOAE did not remain significant after adjustment for age, hypertension, dyslipidemia, cigarette amount, and alcohol consumption. This finding is broadly in line with the inconsistent prior literature: while there are studies showing decreased DPOAE among individuals with tinnitus [46,59,60], some studies have observed increased DPOAE in individuals with tinnitus and hyperacusis [61] or those with tinnitus and normal hearing [62].

The strengths of our study includes the well-defined occupational cohort with relatively homogeneous noise exposure and ethnic background, which minimizes confounding by population stratification.

However, we acknowledge several limitations. First, tinnitus was assessed by self-report, which may be subject to reporting bias. Second, our sample size, although decent for a single-site study, was modest for detecting genetic effects. Therefore, this sample size might be underpowered to detect smaller effect sizes for GSTT1 genes or other potential interactions, and replication in larger cohorts is warranted. On the other hand, if GSTT1 influences multiple auditory phenotypes (e.g., hearing loss and tinnitus), we cannot rule out pleiotropy in our analysis. Third, this cross-sectional study could not provide causal relationships between tinnitus, hearing loss, and genotypes. Fourth, our study included only male participants, which may limit the generalizability of our findings to female populations. Fifth, although we included PTA and DPOAE in our measurements, we did not incorporate auditory brainstem response, which may provide additional information on neural conduction pathways. Sixth, although tinnitus is associated with depressive symptoms as mentioned in the Introduction section, we did not include these psychological factors as covariates in the logistic regression analysis. Finally, we did not assess tinnitus psychoacoustic parameters such as tinnitus matching, residual inhibition, or minimum masking levels, which may have provided further characterization of tinnitus perceptual features.

In conclusion, our study provides evidence that auditory function and genetic makeup jointly influence the risk of tinnitus in noise-exposed workers. This study also underlines tinnitus as an occupational health outcome that deserves our attention. From a preventive standpoint, continued efforts to reduce workplace noise exposure remain paramount as does audiometric monitoring. In addition, awareness of genetic susceptibility factors could lead to personalized protective measures.

CRediT authorship contribution statement

Perng-Jy Tsai: Writing – review & editing, Methodology, Investigation, Formal analysis. Jiunn-Liang Wu: Writing – review & editing, Supervision, Resources, Methodology. Cheng-Yu Lin: Writing – review & editing, Supervision, Resources, Methodology, Investigation, Funding acquisition, Conceptualization. Yue Leon Guo: Writing – review & editing, Supervision, Resources, Methodology, Investigation, Funding acquisition, Conceptualization.

Statement on the Use of AI Tools

During the preparation of this work the authors used ChatGPT in order to do English editing. After using this service, the authors reviewed and edited the content as needed and take full responsibility for the content of the publication.

Conflicts of interest

The authors declare that they have no conflicts of interest.