Introduction

The United States government has paid more attention to auditory healthcare recently due to the prevalence of adult hearing loss between ages 20 and 69. Approximately 37.5 million of American adults aged 18 and over have suffered hearing loss in the national health interview survey in year 20121. About 28.8 million of them could benefit from wearing hearing devices2,3. However, only approximately 16 percent of adult aged 20–69 have ever used the hearing devices, based on the manufacturers’ voluntary reports of registered devices to the U.S. Food and Drug Administration, 2012. Therefore, strategy to postpone the onset of hearing loss is needed in order to improve quality of life and reduce medical expenses4. There are a myriad of intrinsic and extrinsic factors concerning hearing loss issues, such as age5,6, gene polymorphism5,6, noise and digital music exposure7, poor nutrition and lifestyle6, current ototoxic drugs used5,6, and secondhand smoke8. In addition to these factors, over 50% of patients with hearing impairment have previously unrecognized underlying diseases, such as hypertension9, diabetes mellitus (DM)10, metabolic syndrome11, osteoporosis12,13 and Paget disease14.

Osteoporosis is a metabolic disease that raises alkaline phosphatase (ALP) levels. ALP, a homodimeric enzyme of 86 kilodaltons, plays a vital role in metabolism within the liver and bone. Clinicians use bisphosphonates to reduce the level of ALP, thus preventing a further decrease in bone mineral density. In addition, bisphosphonates also lower the level of ALP in Paget disease15. Recently, researchers have noticed asymptomatic elevated ALP values in normal individuals14. In patients with chronic kidney disease, elevated serum ALP levels affect the inflammatory responses and are directly associated with erythropoietin stimulating agent-resistant anemia16. A chronically elevated ALP level in a normal person may imply presence of a disease, including hearing dysfunction.

Due to the lack of literature concerning the relationship between ALP and frequency distributions of pure-tone air-conduction (PTAC) thresholds to date, our study was designed to provide insight into this issue. Therefore, we investigated the association between ALP and hearing loss using the National Health and Nutrition Examination Survey (NHANES) data from the year 1999 to 2004.

Results

Characteristics of the study population

We included 3877 NHANES participants who represented noninstitutionalized civilian population. The characteristics of the study population are presented in Table 1. Our subjects had a mean age of 39.21 years old (SD = 12.4) and 46.4% of the population was male. Among subjects with higher quartiles of ALP, body mass index (BMI) and uric acid were significantly higher (P < 0.001). DM status was significantly higher in the fourth quartile of ALP (P < 0.001). Figure 1A,B showed aged between 50–80 years old have >25 dB HL (mild hearing loss) in high and low pure-tone thresholds. Figure 1C,D showed non-Hispanic white has >16 dB HL (slight hearing loss) in high pure-tone threshold. Hearing loss continues to be common in older, non-Hispanic whites according to the NHANES data.

Table 1 Characteristics of Study Participants.
Full size table
Figure 1
figure 1

(A,B) showed the high and low pure tone thresholds categorized by age. (C,D) showed the high and low pure tone thresholds categorized by race.

Full size image

Gender difference in the association of ALP and hearing thresholds

Table 2 shows the positive correlation between ALP and pure tone average at high frequencies (high-PTA) in both men and women. After additional adjustment of models 1, 2 and 3, beta coefficients for ALP and high-PTA were 0.005, 0.005, and 0.005 respectively in men and 0.003, 0.004, and 0.004, respectively, in women (P for trend <0.001). In terms of pure tone average at low frequencies (low-PTA), both men and women showed a positive correlation with ALP. The beta coefficients were 0.004, 0.004, and 0.003 (P for trend <0.001) respectively in men and 0.004, 0.004, and 0.004 (P for trend <0.001) respectively in women.

Table 2 Gender difference in the association between alkaline phosphatase with hearing threshold.
Full size table

BMI and the association between ALP and hearing thresholds

Table 2 shows a positive correlation between BMI and high-PTA after additional adjustment of three models. Beta coefficients for ALP and high-PTA were 0.004, 0.004, and 0.004 in BMI<30, and in for BMI>30 (P for trend <0.001). In terms of low-PTA, only BMI<30 was related to ALP, and the beta coefficients were 0.004, 0.004, and 0.004 (P for trend <0.001).

The relationship between different quartiles of ALP and hearing thresholds

We performed a quartile-based analysis by dividing ALP into quartiles, as shown in Table 3. The subjects in the first quartile were set as our reference group. The trends showed significance in these three models regardless of high-PTA or low-PTA. Beta coefficients for Q4 versus Q1 at high-PTA were 0.039 in model 1, −0.040 in model 2 and 0.038 in model 3 (P for trend <0.001). The beta coefficients for ALP comparing Q4 to Q1 at low-PTA were 0.050 in model 1, 0.049 in model 2 and 0.045 in model 3 (P for trend <0.001).

Table 3 Regression coefficients of the quartiles of bone alkaline phosphatase with hearing threshold.
Full size table

Discussion

By evaluating representative data for the US population, we discovered the fourth quartile of ALP was highly related to high-PTA when we weighed the differences between the first, second and third quartiles’ ALP levels. This connection was found not only in obese (BMI>30) individuals but also in underweight, normal weight, and overweight (BMI<30) individuals.

Our results showed a worse hearing threshold in low-PTA was not only observed in women with BMI less than 30 but also in men. As far as we know, women who have low body mass indices are at risk for osteoporosis17. ALP has been found at higher levels in osteoporotic postmenopausal women due to high bone turnover rate12. In addition, high bone turnover states are known to raise plasma lead levels. A high turnover rates in women who had a higher bone lead level in Q2 and Q3 of ALP compared to Q1 in peri- and postmenopausal women18. Therefore, the association between high ALP and high BLL in postmenopausal women in their study corresponds to a worse hearing threshold of low-PTA in low BMI women in our study.

In recent years, scientists have recognized osteoporosis in men. Men suffer from osteoporotic fractures and have twice the 1-year fatality rate of women due to decreased lower body strength19. In addition, osteoporotic men trended to have lower free testosterone, higher levels of ALP and sex hormone-binding globulin than osteopenic and normal bone mineral density groups. This finding also supports the correlation between hearing loss and ALP in men, although the association is weak20.

Age-related hearing loss is one of the most common conditions affecting older adults. However, hearing loss is also found in younger people without evidence of hair cell loss or absence of auditory nerve synapses in their temporal bones7,21. The mechanism is related to the intense metabolic activity such as infection and inflammation, which drive the formation of free radicals and inflammatory cytokines, in both healthy and disease22,23,24,25. The elevated serum ALP is associated with C-reactive protein (CRP), which indicates ALP level may be a marker of low-grade chronic inflammation26. Chronic TNF-alpha release that enters the inner ear may cause bone conduction threshold elevation at the high frequencies27. In addition, otitis media-related inflammation also increased air conduction and bone conduction thresholds during childhood and adults stages28. While bacterial infection, eosinophilic inflammation, and chronic otitis media deteriorated bone conduction threshold29. The increases in stiffness and in mass of the ossicular chain affect bone conduction threshold shifts, which are less likely to accompany aging than are changes in air conduction thresholds30. The temporal bone houses the structures of the auditory system31. The malleus, incus, and stapes bones are the ossicles, which function to mechanically convert the vibrations into the inner ear. While ALP activity was detected in the ganglion cells and near the temporal bone31. Therefore, temporal bone and the ossicles, which contain the ALP activity in the ear, may take part in bone conduction threshold shift32.

The mechanism of hearing loss in regard to the increment of ALP remains an uncertainty. To date, ALP with reactive oxygen species (ROS), inflammation, and insulin resistance are possible mechanisms of hearing loss, subsequent to the disturbance of microcirculation in the cochlea. Scientist have established the role of the ROS signaling pathway in osteoblast (OB) differentiation. They demonstrated that the activation of NADPH oxidase was the key reaction to initiate ALP expression. The ROS system therefore produces oxidative stress and induces mitochondrial morphological transition33. In addition, vascular calcification raised ALP concentration asymptomatically. A statistically significant correlation between insulin resistance and serum bone ALP levels in vascular calcification, and ALP can be a predictor of cardiovascular events and mortality in patients with DM34. Moreover, ALP is capable of dephosphorylating nucleotide phosphatase and lipopolysaccharide to reduce inflammation and coagulopathy, as it shares a similar amino acid sequence with von Willebrand factor and collagen35. The positive correlation between inflammation and ALP level mostly presents in healthy elderly people with unrecognized Paget’s disease5, osteoporosis12,13,17, and presbycusis5. The sensorineural hearing loss in Paget disease includes a variety of cochlear lesions36 and inferior cochlear vein occlusion that lead to pathologic changes in the stria vascularis and spiral ligament37. Based on the findings of Nomura et al., ALP shows strong activity in spiral ligament and stria vascularis, and the microcirculatory disturbance in the stria prominence, causes the impaired function in hearing38. As previously mentioned, the increased in ALP was parallel to the hearing loss due to the evidence of ALP activity in the temporal bone and cochlea.

Our study had a few limitations. First, the cross-sectional study design with data captured from a single point in time for ALP and hearing loss was a limitation of our study design compared to hearing threshold shift in longitudinal study designs. Second, the self-reported questionnaire of inherited genetic defects and past medical history may have introduced recall bias. Third, the NHANES data only provided the current use of ototoxic medication, instead of hormonal drugs, antidepressants, and anti-inflammatory medicine.

Conclusion

We concluded that a positive relationship between bone ALP and hearing loss does exist for the general population. For an individual with a BMI<30 and hearing that has remained within the normal range, bone ALP levels seem to be a tool for detecting hearing loss. Because the causal mechanism associated ALP and hearing loss remains unclear, further research should be carried out in order to achieve a better understanding of the underlying pathophysiology of hearing loss.

Methods

Ethics statement

The NHANES data were adequately secured and approved by the National Center for Health Statistics Institutional Review Board. Informed consents for all eligible subjects were obtained before the start of a series of surveys with physical and laboratory examination. All methods were performed in accordance with the relevant guidelines and regulations of NHANES.

Study population

We included 3877 NHANES participants from ages 20 to 69 who represented the noninstitutionalized civilian population in the United States. This population were enrolled from NHANES from the period between 1999 and 2004 and had undergone a series of household interview surveys, physical examinations, and laboratory tests at a mobile examination center (MEC).

Audiometric measurements

Half of our target populations had undergone audiometric testing randomly after a series of surveys with physical and laboratory tests. Subjects who could not tolerate auditory headphones in MEC sound-isolated rooms were excluded. The test was performed by technicians who had received training from the National Institute for Occupational Safety and Health. They used an AD226 audiometer, TDH-39P headphones and EARTone 3 A earphones for both ears and recorded threshold values from −10 dB to 120 dB. According to the American Speech Language Hearing Association (ASHA), the normal range for hearing thresholds is −10 to 15 dB and the abnormal hearing thresholds (worse ear) is ≥26 dB. The frequency distributions of PTAC thresholds is 0.25–8 Hz39. We averaged the thresholds at 3,000, 4,000, 6,000, and 8,000 Hz for the high-PTA while the thresholds at 500, 1,000, and 2,000 Hz were included in the low-PTA, according to NHANES data40. In the regression model, we chose the pure-tone thresholds of the worse ear for analysis.

Covariates

We collected age, sex and race (non-Hispanic black, non-Hispanic white or other) as our demographic data. BMI was calculated as weight divided by the square of the height (kg/m2). BMI>30 was referred to as obese. The self-reported questionnaire contained questions about present medical illnesses that had ever been diagnosed by a physician, such as DM, heart disease and stroke, and the current use of ototoxic drugs.

An otoscopic screening examination was used to determine the abnormality of the outer canal of the ear. In addition, tympanometry was abnormal if middle-ear peak pressure showed a value lower than −150 daPa or the compliance was lower than 0.3 ml. The Hitachi analyzer was used to determine biochemical data, such as uric acid and LDL-C. All measurements were standardized according to the guidelines that were recommended by the Centers for Disease Control and Prevention (CDC).

Statistical analysis

The study used sample weights to account for the complex sampling design and to allow approximations of the United States population, following guidelines of National Center for Health Statistics. The NHANES 1999 to 2004 sample weights adjusted for the differential probabilities of selection and nonresponse in the survey sample. The weight represented the number of individuals in the target population each sample participant was estimated to represent41,42. We used the Statistical Package for the Social Sciences version 18.0 to analyze our data. We set two-sided alpha values, with values less than 0.05 as our significant p-values. The dependent variables were the normalized hearing thresholds (high-PTA and low-PTA), which were transformed using the logarithm function. Moreover, we further divided the serum ALP into quartiles as follows: Q1 <10.6 ug/L, 10.6 ug/L ≤ Q2 <13.8 ug/L, 13.8 ug/L ≤Q3 <18.2 ug/L and Q4 ≥ 18.2 ug/L. In the multivariate regression models, we adjusted for age, gender and race for model 1; age, gender, race, BMI, LDL-C and uric acid for model 2; and age, gender, race, BMI, LDL-C, uric acid, stroke, tympanometry status, ear condition, DM status, histories of heart disease, and ototoxic status for model 3.