Plasma metabolomic profiling in workers with noise-induced hearing loss: a pilot study

Noise-induced hearing loss (NIHL) remains a leading occupational related disease and is a serious public health problem. Hence, the identification of potential biomarkers for NIHL prevention and diagnosis has become an urgent work. To discover potential metabolic biomarkers of NIHL, plasma metabolomics analysis in 62 NIHL patients and 62 normal hearing controls was performed using ultrahigh-performance liquid chromatography coupled with quadrupole time-of-flight tandem mass spectrometry (UHPLC-Q-TOF MS). Orthogonal partial least square-discriminant analysis (OPLS-DA) model was applied to distinguish metabolite profile alterations in plasma samples between the two groups. The metabolites with a variable importance of projection (VIP) value > 1 and P value < 0.05 were considered to be potential metabolic biomarkers. KEGG database was performed to explore the involved pathways of potential biomarkers. Three autophagy-related genes (PI3K, AKT, and ATG5) were selected for further verification, and mRNA levels were detected using RT-qPCR analysis. Twenty plasma metabolites with VIP > 1 and P < 0.05 were significantly altered between the two groups. Totally, seven metabolic pathways involving the glycerophospholipid metabolism, glycosylphosphatidylinositol (GPI)-anchor biosynthesis, autophagy pathway, choline metabolism, the alpha-linolenic acid metabolism and linoleic acid metabolism, and retrograde endocannabinoid pathway were significantly related to NIHL. Furthermore, verification by RT-qPCR suggested that the mRNA expression levels of PI3K and AKT along with ATG5 were significantly lower in the NIHL patients compared with controls. In summary, the present study provides the first evidence that the identified aberrantly altered metabolites may be the potentially valuable biomarkers of NIHL for occupational noise-exposed workers. Autophagy signal pathway may be involved in the occurrence and development of NIHL. Moreover, this present study may be helpful to further better understand the metabolic changes in NIHL and be helpful for the understanding of pathogenic mechanism.


Introduction
Generally, noise is considered as a set of sounds that could make people irritable or is loud enough to endanger everybody. In environment, there are diverse sources of noise, including transportation noise, occupational noise, and construction noise. Nowadays, noise pollution has been an important and neglected public health issue in the world. Adverse health effects of noise on human beings can be observed on psychologically and physiologically (Muzet 2007). It is estimated that about 20 million adults in Europe suffer from long-term noise annoyance (Wothge et al. 2017). In addition, the World Health Organization reported that mental, behavioral, and neurological diseases affected by noise account for 3% of global deaths and 10% of global burden of disease (Minichilli et al. 2018). Recent evidence showed that long-term exposure to environmental noise could cause some adverse effects, such as sleep disturbance (Halperin 2014;Muzet 2007), annoyance (Licitra et al. 2016;Miedema and Oudshoorn 2001;Minichilli et al. 2018), learning impairment (Erickson and Newman 2017;Minichilli et al. 2018;Zacarias et al. 2013), hypertension (Dratva et al. 2012), and cardiovascular diseases (Babisch et al. 2005;Dimakopoulou et al. 2017).
To our knowledge, in daily life, road traffic noise is the most important noise source influencing quality of human life and health. A growing body of research shows that traffic noise could lead to short-and long-term adverse health effects (Babisch 2006;Browning et al. 1990;Stansfeld et al. 2005). Recent evidence from cross-sectional studies found a significant correlation between road traffic noise and blood pressure in both children and adults Sorensen et al. 2011). A recent meta-analysis reported a 3% higher risk of hypertension per 5-dB(A) increase in road traffic noise (van Kempen and Babisch 2012). Regarding long-term association, Sorensen et al. (2013) found that 14% risk of incident diabetes could be attributed to a 10 dB(A) increase in daily road traffic noise levels. Studies have found that road traffic noise can also contribute to heart failure and atrial fibrillation (Monrad et al. 2016;Seidler et al. 2016). Studies based on large population suggested that road traffic noise significantly increased the risk of heart failure by 2 to 7% per 10 dB (A) rise (Heritier et al. 2017;Seidler et al. 2016). Moreover, study on atrial fibrillation concluded that every 10 dB (A) of road traffic noise increased the risk of atrial fibrillation by 6% (Monrad et al. 2016). Studies have shown that production of road traffic noise is affected by some parameters, such as engines, flow composition, acoustic impedance , tyre model (Licitra et al. 2017), pavement aging , pavement texture (Del , and mixture Praticò and Anfosso-Lédée 2012). In summary, it is extremely important to take appropriate methodologies to mitigate the traffic noise emission and noise exposure and protect human health by controlling theses main parameters.
Moreover, occupational noise is a common harmful factor that seriously affects health of workers in the field of occupational health. Noise-induced hearing loss (NIHL) is one of the worst adverse health effects induced by occupational noise exposure in workplaces (Masterson et al. 2016). The WHO reported that about 10% of the world's population is exposed to high level of noise and is at risk of progressing to NIHL (Basner et al. 2014). A recent review revealed that occupational noise exposure resulted in 7 to 21% of workers' hearing loss, with the lowest incidence in industrialized countries and the highest in developing countries (Lie et al. 2016). Nowadays, NIHL is an urgent health problem and causes a significant impact on the social economy and human health. According to previous study, NIHL is one of the leading occupational related disease in China, accounting for approximately one-sixth of the annual increase of occupational disease (Miao et al. 2019). Although a number of studies were conducted, the exact pathogenic mechanism of NIHL has still not yet been entirely illustrated. Therefore, it is of great significance to search new and potential biomarkers for the understanding of pathogenesis of NIHL.
Metabolites in human body fluids represent the end products of metabolic pathways and reflect final consequences of organisms in response to environmental stimulation and disease stress (Huang et al. 2018). Importantly, endogenous metabolic changes can indicate the direct biological response to stressors, such as environmental exposure, disease, and nutritional imbalances ). Metabolomics has now become a powerful and effective platform to identify extensive biomolecule closely correlated with environmental factors and health effect (Chen et al. 2019;Huang et al. 2018). Thus, identification of metabolic biomarkers could help to better understand the possible molecular mechanism of adverse effects induced by harmful exposure factors and may further contribute to high-risk individuals' identification (Floegel et al. 2013).
It has been reported that a key factor contributing to NIHL is oxidative stress damage to body's sensory hair cells . Oxidative stress is a state of imbalance between oxidation and antioxidant defenses, thus resulting in oxidative damage (Sies 1997). Available studies showed that reactive oxygen species (ROS) may play an essential role in regulating cellular stress and defense pathways. Excessive production of ROS was thought to be a key pathological mechanism involving in the process of inner ear injury, such as exposure to noise and ototoxic drug therapy (Ohlemiller et al. 1999;Yamashita et al. 2004). In addition, related studies found that ROS has the capacity to induce cell defense pathways like autophagy. Autophagy is an important defense process that could pass impaired cell components to lysosomes degradation (Wang and Klionsky 2003;Yang and Klionsky 2009). In our previous studies, we found that inflammation-related gene polymorphisms are associated with the susceptibility to NIHL, revealing that inflammation is an essential stress to the pathogenesis of noise-induced cochlear impairment (Miao et al. 2021a;Miao et al. 2021b). Nevertheless, the related metabolic profiles in occupational noise-exposed workers are still not clear, and whether autophagy is involved in the development of NIHL has yet to be established and needs to be further explored.
In this present study, metabolomics of plasma samples from occupational noise-exposed workers with hearing loss and normal hearing was performed to identify potential metabolic biomarkers and abnormal pathways involving in NIHL using ultrahigh-performance liquid chromatography coupled with quadrupole time-of-flight tandem mass spectrometry (UHPLC-Q-TOF MS) tool. Then, three autophagy-related genes including PI3K, AKT, and ATG5 were selected for further verification, and mRNA expression levels were detected using real-time quantitative PCR (RT-qPCR) analysis.

Subjects
The study subjects were the workers who exposed to occupational noise in the factories. All workers were asked to receive annual occupational health examination, such as general physical examination, peripheral blood samples collection, and pure-tone audiometry (PTA) test. The questionnaire survey was performed to collect important information of all subjects, including general family history of disease, personal disease history, smoking, and alcohol consumption, along with history of drug use. The criteria for inclusion of study subjects were (1) workers exposed to occupational noise higher than 1 year; (2) workers only exposed to occupational noise; and (3) Chinese Han workers. Nevertheless, the exclusion criteria were (1) workers had a family history of deafness and a history of explosive noise exposure; (2) workers carried diseases that could affect normal hearing (otitis media, tinnitus, and skull trauma); (3) workers had treatment with an ototoxic drug (aminoglycosides, quinolones, and aspirin) that damages the normal function of the inner ear; (4) workers had metabolic diseases (e.g., diabetes, hypertension, hyperlipidemia, etc.); and (5) workers who had Meniere's disease, sudden deafness, deafness caused by autoimmunological diseases, contagious diseases, and others.
This study was approved by the Ethics Committee of Zhongda Hospital, Affiliated Hospital of Southeast University. Written informed consent was acquired from all workers before they joined in the study.

Noise exposure evaluation
Exposure level of noise in the workplace was estimated according to equivalent continuous dB(A)-weighted sound pressure levels (L Aeq , 8 h) with class I sound level meter. Cumulative noise exposure (CNE) was conducted to reflect the true level of individual exposure to noise based on L Aeq .

Pure-tone audiometric examination
After stopping noise exposure for over 12 h, each study subject has to accept the PTA examination conducted by professional doctor using Voyager 522 audiometer (Madsen, Taastrup, Denmark) at frequencies of 0.5, 1, 2, 3, 4, and 6 kHz in a noise-proof room. Referring to GB/T7582-2004, the obtained original data were adjusted by sex and age.

Selection of NIHL cases and normal hearing controls
Referring to the Chinese National Occupational Health Standard (GBZ49-2007), NIHL cases of this current study were define as occupational noise-exposed workers with binaural high frequencies (3, 4, and 6 kHz) hearing threshold level greater than 25 dB (A). Nevertheless, those with binaural high frequencies hearing threshold level less than 25 dB (A) were included as the normal hearing controls. Totally, 124 subjects including 62 NIHL cases and 62 normal hearing controls were recruited in this current study. The control group had age, sex, work time with noise, status of smoking and drinking, and exposure level with noise matched with cases.

Chemicals and reagents
High-performance liquid chromatography (HPLC)-grade methanol and acetonitrile were purchased from Merck (Darmstadt, Germany). Both formic acid and ammonium formate were purchased from Thermo Fisher Scientific (Waltham, USA). Besides, deionized water was prepared by Milli-Q water purify system (Millipore, USA).

Plasma sample collection and preparation
Considering that food intake and drinking may result in the alteration of human metabolome, morning peripheral blood (collected after 12-h fasting) is collected for plasma collection. Two milliliters of peripheral venous blood were collected from each study subject and transferred to tubes containing ethylenediaminetetraacetic acid. Plasma samples were isolated and centrifuged at 3500 rpm at room temperature for 15 min. Then, all plasma samples were stored at -80°C until metabolomics analysis.
All collected plasma samples were unfrozen at 4°C and then vortexed for 20 s. A 100 μL aliquot of plasma sample was mixed using three times volume of methanol for protein precipitation. Then, the mixture was further vortexed for 30 s and was centrifuged at 13,000 rpm at 4°C for 20 min. The liquid supernatant was obtained for centrifugation at 13,000 rpm at 4°C for 20 min. Totally, 20 μL aliquot of the supernatant was transferred into a sample vial for metabolomics analysis.

Chromatographic and mass spectrometric analysis
Metabonomic profiles of plasma samples from occupational noise-exposed workers were performed by ACQUITY ultrahigh-performance liquid chromatography (UHPLC) (Waters, Milford, USA) system equipped with AB SCIEX Triple TOF 5600 System (AB SCIEX, Framingham, USA).
Chromatographic separation was implemented on an ACQUITY UPLC BEH C8 column (2.1 mm × 100 mm × 1.7 μm, Waters, Milford, USA). For positive ion mode, the mobile phase was made up of water with 0.1% formic acid (A) and acetonitrile (B). For negative ion mode, the mobile phase was 5 mM ammonium formate aqueous solution (C) and acetonitrile (D). The procedures of gradient elution were as follows: 5% solution B for 0-0.5 min, 5-20% solution B for 0.5-2 min, 20-25% solution B for 2-4 min, 25-60% solution B for 4-10 min, and 60-100% solution B for 10-15 min, 100% solution B for 15-16 min, and 5% solution B for 16-19 min. The delivery flow rate was set at 0.3 mL/min. Besides, the injection volume was 5 μL. All analyzed samples were kept at 4°C, and the temperature of column was set at 40°C.

Data processing and analysis
Firstly, the original data were processed using Progenesis QI software for baseline removing, peak identification, peak alignment and integration, and retention time adjustment. Then, a data matrix composed of retention time, mass-tocharge ratio, intensity of peak, and information of sample was performed for further statistical analyses. For each sample, the total peak intensity of the sample was used to further adjust the peak intensity corresponding to metabolite.
In this study, SIMCA-P 14.1 (Umetrics, Umea, Sweden) was applied for multivariate statistical analyses. Both principal component analysis (PCA) and orthogonal partial least square-discriminant analysis (OPLS-DA) models were applied for distinguishing controls and NIHL patients after unit variance scaling. Model validation of OPLS-DA was conducted by resampling the model 200 times via random permutation tests in MATLAB software. The metabolites with a variable importance of projection (VIP) value > 1 and P value < 0.05 were considered to be potential metabolic biomarkers. A heat map of metabolites was generated with multi-experiment viewer (MEV) software.
Some further analyses were also conducted in this study. MetaboAnalyst was applied for related metabolic pathway analysis (http://www.metaboanalyst.ca). Pathway plots analysis was performed based on the Kyoto Encyclopedia of Genes and Genomes (KEGG) database. In addition, Gene Set Enrichment Analysis Software (GSEA, Broad Institute, Cambridge, MA, USA) was used for gene enrichment analysis. Furthermore, to characterize the gene functions, human metabolome database (HMDB) was performed.

Statistical analyses
Statistical analysis was carried out to ensure the potential metabolites significantly changed between NIHL cases and controls by using paired non-parametric test on MATLAB software (MathWorks). Otherwise, statistical analysis was conducted with the SPSS 23.0 software (SPSS, Chicago, IL, USA). The statistical significance criterion was set with a two-sided P value < 0.05.

Characteristics of subjects
A total of 62 NIHL cases and 62 healthy controls were recruited in this study. In terms of age, sex, time with noise exposure, habit of smoking and drinking, systolic blood pressure (SBP), and diastolic blood pressure (DBP) levels, along with exposure level with noise, no significant differences were found between two groups (P > 0.05). There was an obvious difference in high-frequency hearing threshold between the two groups. The results showed that NIHL patients had a significantly higher hearing threshold of high frequency in both ears (53.24 ± 13.01) compared with the controls (18.73 ± 3.96; P < 0.001). The information of all study subjects is summarized in Table 1.

Plasma metabonomic profiling of NIHL
The non-targeted metabolomic profile was explored in 124 plasma samples obtained from 62 NIHL patients and 62 normal hearing controls with UHPLC-Q-TOF/MS with ESI positive ion mode and negative ion mode, respectively. Considering the metabolites detected by positive and negative ion modes were complementary, the data obtained from two modes were combined into a matrix for analysis. After removing missing values > 50% ion peaks, 6777 ion peaks in positive ion mode and 5404 ion peaks in negative ion mode were obtained. A total of 4009 metabolites (2277 in positive ion mode and 1732 in negative ion mode) were detected by accurate mass, fragmentation patterns, and retention time. Univariate statistical analyses were performed for all metabolite to determine the changes of plasma metabolome between the two groups. As shown in Fig. 1, obvious separation suggests a significant difference between NIHL patients and controls at the plasma metabolomic levels.

Identification of changed endogenous metabolites
Paired non-parametric test was conducted to identify significant differential metabolites for plasma between two groups. As shown in Fig. 2a, the present study identifies 207 differential metabolites (P < 0.05), including 136 up-regulated metabolites and 71 down-regulated metabolites. To further explore which significant differential metabolites contributing to the differences described above through setting a comparatively high stringency (P < 0.05 and VIP > 1), in this way, it showed that 59 metabolites were significantly alterations in NIHL patients compared with the controls. Through examining the original data and feature of ion peaks, 20 identified metabolites with VIP > 1 and P < 0.05 were considered to be potential metabolic biomarkers, including 12 up-regulated and 8 downregulated metabolites (Table 2). Besides, a visual heat map was generated based on these 20 plasma differential metabolites, showing a considerable difference between NIHL cases and controls (Fig. 2b). Among the identified metabolites, some up-regulated metabolites were organic acids, including homodeoxycholic acid and quinolacetic acid, along with 3,4-dihydroxymandelic acid. However, most of downregulated metabolites were lipids, such as PE(15:0/ 20:2(11Z,14Z)), PC(15:0/18:1(11Z)) and PI(O-20:0/18:0).
Pathway enrichment analysis showed that seven metabolic pathways, including glycerophospholipid metabolism, glycosylphosphatidylinositol-anchor biosynthesis, autophagy, choline metabolism, alpha-linolenic acid metabolism, and linoleic acid metabolism, and retrograde endocannabinoid pathway were significantly related to NIHL, suggesting that these pathways may be involved in the development of NIHL (Table 3).

Autophagy related-gene expression in NIHL patients and controls
Among the identified differentially regulated metabolites, PE(15:0/20:2(11Z,14Z)), also known as phosphatidylethanolamine in HMDB, was found to be reduced in plasma samples in NIHL patients, which was involved in autophagy metabolism pathway. In addition, the another metabolite, namely,  FC, fold change PC(15:0/18:1(11Z)) (called phosphatidylcholine in HMDB), was also decreased in plasma samples of NIHL patients and involved in autophagy metabolism pathway. To explore the biological effects of noise exposure on autophagy pathway in NIHL patients, three autophagy-related genes including PI3K, AKT, and ATG5 were selected, and mRNA levels were determined in peripheral white blood cells (WBCs) of NIHL patients and controls. Verification by RT-qPCR suggested that individuals with NIHL had significantly lower expression levels of PI3K and AKT, along with ATG5 than those normal hearing subjects (Fig. 3).

Discussion
NIHL is one of the serious harmful health effects induced by high-intensity noise exposure and has become a leading occupation-related disease in the world (Miao et al. 2019).
In the present study, plasma metabolomic characteristics of NIHL patients and normal hearing controls were compared using non-targeted metabolomics approach, which provides a theoretical basis for further exploring the pathogenesis of adverse effects caused by noise. Twenty significantly changed metabolites were identified, revealing disturbances of a variety of biological pathways in the development of NIHL. In addition, three autophagy-related gene expression levels were determined using RT-qPCR, and the results indicated that PI3K, AKT, and ATG5 were significantly down-regulated in NIHL patients compared with controls, suggesting that autophagy pathway plays an essential role in development of NIHL. Taken together, this study firstly provides a new perspective to understand the mechanism and identifies potential biomarkers correlated with NIHL and verified the critical role of autophagy pathway in NIHL.
Metabolomics is a powerful platform for exploring disease phenotype, which provides a wealth of information for the discovery of biomarker, pathogenesis, and personalized treatment (Ye et al. 2014). So far, some metabolomics studies have been performed in noise exposure field (Floegel et al. 2013;Huang et al. 2018;Wang et al. 2017;Zhang et al. 2016). Pudrith and Dudley (2019) found that five metabolites are closely related to glutathione-dependent mercapturic acids in urine, but significant associations were just only found in nonnoise exposure subjects. Fujita et al. (Fujita et al. 2015) revealed that ten metabolites exhibited statistically significant changes in inner ear fluid of guinea pig exposed to loud noise, including amino acid catabolites and lipid compounds. Also, a study by He et al. (2017) demonstrated that multiple metabolic pathways are involved in acoustic trauma, such as arginine, proline, and purine metabolic pathways. However, little studies were conducted to investigate the metabolic signatures induced by occupational noise in humans. In conclusion, plasma samples of 62 NIHL patients and 62 normal hearing controls were analyzed to investigate the metabolomic profiles on  FDR, false discovery rate NIHL. A total of 20 differential metabolites previously unknown were identified to be correlated with NIHL. Bilirubin is one of important products of heme catabolism and is an effective antioxidant that removes hydrogen peroxide (Minetti et al. 1998). Bilirubin and glutathione were reported to have complementary cell-protective and antioxidant effects (Sedlak et al. 2009). Findings from previous studies indicated that serum bilirubin was closely related to cardiovascular disease-related factors, including body mass index, metabolic syndrome, and diabetes (Cheriyath et al. 2010;Horsfall et al. 2012). In addition, studies found that the plasma and serum bilirubin level significantly increased in benzeneexposed workers compared with the controls (Neghab et al. 2015). In our current study, low plasma bilirubin in NIHL patients may act in conjunction with glutathione to protect cells from oxidation. However, the concrete role of bilirubin in the development of NIHL needs to be further explored.
Among the identified differential metabolites, some were phospholipids (PLs), which are major components of cellular membranes and vital bioactive molecules (Cvetkovic et al. 2017). In this study, plasma levels of some crucial PLs (PE(15:0/20:2(11Z,14Z)), PC(15:0/18:1(11Z)), and PI(O-20:0/18:0)) were significantly decreased in NIHL patients compared with controls. It is well known that lipid metabolism is frequently occurred in a variety of diseases, but the most recent evidence found that lipid-related genes may be involved in inflammatory and metabolic diseases (Hirsch et al. 2010). Previous studies have found that the alterations of PLs levels in tissue or plasma may be associated with the risk and progression of all kinds of diseases (Sun et al. 2018). Cvetkovic et al. (2017) reported that the abnormal alterations of PLs profile were associated with non-Hodgkin's lymphoma. In addition, lipidomic studies found that the abnormal altered levels of PLs composition could cause changes in membrane integrity, permeability, cell damage, and cell intimal transport (Leamy et al. 2014). PE, PC, and PI, also known as phosphatidylethanolamine, phosphatidylcholine, and phosphatidylinositol, were important antioxidants and involved in cell morphology, metabolism regulation, signal transduction, and a variety of physiological functions of cells (Hidalgo et al. 2005). In this study, the decreased levels of three metabolites in plasma of NIHL patients may be due to the overproduction of ROS during noise exposure, thereby being consumed to maintain balance between ROS and antioxidant defenses system. Meanwhile, this finding may indicate that oxidative stress is a key factor and important mechanism contributing to NIHL. The results obtained from this study showed that these abnormal metabolites were involved in autophagy pathway, indicating that autophagy may be closely related to the development of NIHL. Recent findings demonstrated that PE is an important substrate for the GPI-anchor biosynthesis that is essential for immune response and plays a vital role in the initiation of autophagy by attaching to the autophagy protein to initiate autophagosome formation (Fracchiolla et al. 2017). The findings from previous studies indicated that ROS have the capacity to induce cell defense pathways like autophagy, which could deliver negative and harmful cellular components to lysosomes for degradation (Wang and Klionsky 2003;Yang and Klionsky 2009). Results of pathway enrichment analysis further indicated that autophagy significantly changed and might be involved in the biological progress of NIHL.
PI3K/AKT is an upstream major modulator of autophagy pathway and participates in extensive cellular process, including cell growth, proliferation, survival, and metabolism (Heras-Sandoval et al. 2014). ATG5, as an essential autophagy related protein, is involved in autophagy at the molecular level (Zheng et al. 2019). Considering the importance of the three genes in autophagy pathway, they were included for further verification. The results displayed that the mRNA levels of PI3K, AKT, and ATG5 were significantly lower in NIHL patients compared with controls. Substantial evidence has suggested that the reduction of PI3K/AKT signaling pathways is correlated with hair cell death and hearing loss after various of injuries and stimuli . The decreased expression levels of PI3K and AKT suggested that the PI3K/AKT signaling was inhibited. Available evidence showed that activation of PI3K/AKT could promote cellular survival, growth, and differentiation, but inhibiting the apoptotic signals (Heras-Sandoval et al. 2014). The findings of this study indicated that the significantly decreased levels of PI3K and AKT might cause the inhibition of PI3K/AKT signaling activity and activation of apoptotic signals, thus causing the death of hair cells in the inner ear and the occurrence of NIHL. Furthermore, data from animal studies showed that deletion of ATG5 could cause hair cell degeneration and serious congenital hearing loss (Fujimoto et al. 2017). A review showed that autophagy is a key factor for the auditory cell fate, but autophagy deficiency may be one of leading causes of hearing impairment (Hayashi et al. 2020). Autophagy is considered to be a common cause of many neurodegenerative diseases (Ghavami et al. 2014;Son et al. 2012). NIHL is also a progressive sensorineural hearing loss; thus, autophagy may be associated with NIHL. The results from RT-qPCR suggested that the expression level of ATG5 was significantly decreased in NIHL patients than controls, which were in accord with previously reported findings, suggesting that autophagy could be considered an essential signaling pathway involving in NIHL development. However, further functional studies on the autophagy mechanisms underlying NIHL are necessary.

Strengths and limitations
To our best knowledge, this is the first study that analyzed plasma samples from occupational noise-exposed workers with hearing loss and normal hearing to search potential metabolic biomarkers and abnormal pathways involving in NIHL with UHPLC-Q-TOF MS. The application of metabonomic approach of this present study may be conducive to in-depth understanding of the metabolic changes of NIHL from the perspective of biochemistry and is helpful to further understand its pathogenesis. However, there are some limitations in the present study. First, the subjects recruited were Chinese workers, which inevitably cause a certain degree of selection bias; the findings may not on behalf of other races. Thus, further studies to discover potential metabolic biomarkers related to NIHL in other ethnic populations are required. Second, this was a pilot study on the metabolomics of NIHL, so the significantly altered metabolites have not been quantitatively detected using target analysis. Lastly, it is necessary to conduct functional studies on other metabolites and pathways identified in this study to clarify the potential molecular mechanisms that may be involved in NIHL.

Conclusions
NIHL is generally considered to be a complex hearing disorder, caused by noise mechanical damage and stress responses at the molecular level. Studies on NIHL have been concentrated on identifying possible susceptibility genes and SNPs. Although a large number of studies have been conducted, the pathogenic mechanism has not been illustrated. It is not sufficient to only study SNPs in a single ethnic population or gene. The body fluids, such as plasma, could reflect the changes in various of cells, tissues, and organs; thus, exploring changes in body fluids could act as a research strategy for elucidating the mechanism of NIHL from the perspective of metabolites.
In summary, to discover potential metabolic biomarkers of NIHL, plasma metabolomics analysis in 62 NIHL patients and 62 controls was performed using UHPLC-Q-TOF MS. This study reveals that a total of twenty metabolites related to lipids molecules, fatty acids, and organic compounds were significantly altered in NIHL patients compared to controls, suggesting that the differential metabolites might act as potential biomarkers of NIHL for Chinese noise-exposed workers. The plasma metabonomic profile significantly altered in NIHL patients compared with normal hearing controls, confirming that plasma metabolic profile has a huge advantage in distinguishing patients with NIHL from controls. Furthermore, the metabolic signature alterations of metabolites indicated that the glycerophospholipid metabolism, glycosylphosphatidylinositol (GPI)-anchor biosynthesis, autophagy, choline metabolism, alpha-linolenic acid metabolism and linoleic acid metabolism, and retrograde endocannabinoid pathway were significantly related to NIHL. Meanwhile, the findings show that autophagy might play an essential role in the occurrence and development of NIHL. In conclusion, this study provides evidence for the first time that metabolomics could characterize the metabolites of NIHL. Moreover, this study not only provides a new perspective for the progress of NIHL, but also offers novel clues on the mechanisms underlying NIHL.