Skip Navigation


Human Molecular Genetics Advance Access originally published online on June 13, 2007
Human Molecular Genetics 2007 16(15):1872-1883; doi:10.1093/hmg/ddm135
This Article
Right arrow Abstract Freely available
Right arrow FREE Full Text (PDF) Freely available
Right arrow All Versions of this Article:
16/15/1872    most recent
ddm135v1
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Add to My Personal Archive
Right arrow Download to citation manager
Right arrowRequest Permissions
Google Scholar
Right arrow Articles by Konings, A.
Right arrow Articles by Van Camp, G.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Konings, A.
Right arrow Articles by Van Camp, G.
Social Bookmarking
Add to CiteULike   Add to Connotea   Add to Del.icio.us  
What's this?

© The Author 2007. Published by Oxford University Press. All rights reserved. For Permissions, please email: journals.permissions@oxfordjournals.org

Association between variations in CAT and noise-induced hearing loss in two independent noise-exposed populations

Annelies Konings1, Lut Van Laer1, Malgorzata Pawelczyk2, Per-Inge Carlsson3,4, Marie-Louise Bondeson5, Elzbieta Rajkowska2, Adam Dudarewicz2, Ann Vandevelde1, Erik Fransen1, Jeroen Huyghe1, Erik Borg4, Mariola Sliwinska-Kowalska2 and Guy Van Camp1,*

1 Department of Medical Genetics, University of Antwerp, B-2610 Antwerp, Belgium, 2 Department of Audiology and Phoniatrics, Nofer Institute of Occupational Medicine, 91-348 Lodz, Poland, 3 Department of Audiology, and 4 Ahlsén Research Institute, Örebro University Hospital, 701 85 Örebro, Sweden and 5 Department of Genetics and Pathology, Uppsala University, 751 85 Uppsala, Sweden

* To whom correspondence should be addressed at: Department of Medical Genetics, University of Antwerp, Campus Drie Eiken, Universiteitsplein 1, B-2610 Antwerp, Belgium. Tel: +32 38202491; Fax: +32 38202566; Email: guy.vancamp{at}ua.ac.be

Received January 31, 2007; Accepted May 16, 2007


   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 RESULTS
 DISCUSSION
 MATERIAL AND METHODS
 REFERENCES
 
Noise-induced hearing loss (NIHL) is an important occupational hazard that results from an interaction between genetic and environmental factors. Although the environmental risk factors have been studied quite extensively, little is known about the genetic factors. On the basis of multiple studies, it was proposed that oxidative stress plays an important role in the development of NIHL. Here, we investigated whether variations (single nucleotide polymorphisms; SNPs) in the catalase gene (CAT), one of the genes involved in oxidative stress, influence noise susceptibility. Audiometric data from 1261 Swedish and 4500 Polish noise-exposed labourers were analysed. DNA samples were collected from the 10% most susceptible and the 10% most resistant individuals. Twelve SNPs were selected and genotyped. Subsequently, the interaction between noise exposure and genotypes and their effect on NIHL were analysed using logistic regression. Significant interactions were observed between noise exposure levels and genotypes of two SNPs for the Swedish population and of five SNPs for the Polish population. Two of these SNPs were significant in both populations. The interaction between predictor haplotypes and tagSNP haplotypes and noise exposure levels and their effect on NIHL were also analysed, resulting in several significant associations. In conclusion, this study identified significant associations between catalase SNPs and haplotypes and susceptibility to development of NIHL. These results indicate that catalase is a NIHL susceptibility gene, but that the effect of CAT polymorphisms can only be detected when noise exposure levels are taken into account.


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
RESULTS
 DISCUSSION
 MATERIAL AND METHODS
 REFERENCES
 
Noise-induced hearing loss (NIHL) is most often caused by continuous and regular exposure to noise, but it can also be the result of single or repeated occasions of acoustic trauma. It is the second most frequent form of sensorineural hearing loss, after age-related hearing impairment, and forms worldwide a high occupational health-risk. In human subjects, a large variation in susceptibility to noise is observed. NIHL is a complex condition, caused by an interaction between genetic and environmental factors. Noise is the best-known and one of the most studied environmental factors causing hearing loss. Noise is harmful from 85 dB and causes both mechanical and metabolic damage (1). Several other environmental factors, such as exposure to chemicals like organic solvents and heavy metals, can augment the effect of noise (2–4). In addition, individual factors such as smoking, high blood pressure and cholesterol levels can influence the susceptibility to noise (5).

As a formal heritability study has not yet been realized for NIHL in humans, the notion that genetic factors play a role in the development of NIHL has been deduced from studies using animals (6). Several studies demonstrated that mouse strains exhibiting age-related hearing impairment are more susceptible to noise than other strains (7–9). In addition, several knockout mice, such as PMCA2 –/– (10), CDH23 +/– (11), SOD1 –/– (12) and GPX1 –/– (13), are more susceptible to noise than their wild-type littermates. Also, spontaneously hypertensive rats have a progressive hearing impairment and are more susceptible to noise than normotensive rats (14).

So far, only a few association studies trying to identify genes involved in human subjects have been performed (15–18). Only two of them have led to a possible identification of NIHL susceptibility genes, i.e. KCNE1 (15) and GSTM1 (17).

Cochlear hair cell damage by reactive oxygen species (ROS) following noise exposure is a potential mechanism for NIHL. Superoxide anion radicals emerge in the stria vascularis after intense noise exposure (19) and hydroxyl radicals significantly increase in the cochlea of animals exposed to noise (12). It is known that antioxidant therapy protects against NIHL, whereas chemicals that produce oxidative stress potentiate NIHL (20–23). In addition, mice lacking either SOD1 or GPX1, two enzymes that are involved in the protection against oxidative stress, are more susceptible to NIHL than their wild-type littermates (12,13). Finally, the expression of glutathione and catalase, two other antioxidant factors, increases after noise exposure as shown in studies using animals (20,24).

Since the cochlea is a metabolically active organ, several ROS are generated under normal metabolic circumstances during the reduction of oxygen into O2–(i.e. O2 into H2O). Antioxidant systems are present to neutralize these ROS. Besides antioxidant enzymes that are active in glutathione metabolism, another set of enzymes is involved in the breakdown of superoxide anions and hydrogen peroxide (H2O2). One of such enzyme is catalase (hydrogen peroxide oxidoreductase; EC 1.11.1.6 [EC] ), a heme-containing homotetrameric protein (25). Catalase decomposes H2O2 in reactions catalyzed by two different modes of enzymatic activity: either catalytic (2H2O2->O2+2H2O) or peroxidatic (H2O2+AH2-> A+2H2O), in which AH2 stands for a proton donor (26,27). Erythrocytes, liver and kidney contain the highest level of catalase while the lowest levels are observed in connective tissue (28–30). In the inner ear, higher levels of catalase are observed in the organ of Corti than in the stria vascularis (24).

The catalase gene (CAT) has been assigned to 11p13 (31). It measures 34 kb and contains 13 exons (32). The coding region consists of 1581 bp (33). Possible associations between CAT polymorphisms and diabetes mellitus, high blood pressure and vitiligo have already been detected (34–38).

The general aim of the current study was to investigate a possible association between CAT variants and NIHL. In an earlier study, we analysed several polymorphisms in a number of oxidative stress genes, including CAT, for association with NIHL, but did not detect significant main effects (18). In this study, when we included the different exposure categories into the analysis and when we performed an in-depth analysis using more markers, we obtained several significant associations for genotypes and haplotypes in two independent noise-exposed populations. These results confirm the hypothesis that oxidative stress plays a role in the development of NIHL, and that variants in CAT influence susceptibility.


   RESULTS
TOP
 ABSTRACT
 INTRODUCTION
 RESULTS
DISCUSSION
 MATERIAL AND METHODS
 REFERENCES
 
SNP selection
The first set of SNPs (SNP 5, 10 and 12; Tables 1 and 2; Fig. 1) has been analysed previously (18). No significant associations between the genotypes of these SNPs and NIHL were obtained in the Swedish population (18). Since there were several strong arguments that oxidative stress is involved in the development of NIHL, further statistical analysis was performed, testing for the interaction between noise exposure levels and genotypes. This resulted in significant P-values for SNP 5 and 12 (Table 2). Subsequently, three additional SNPs (SNP 2, 3 and 4; Tables 1 and 2; Fig. 1), situated in or in the proximity of the promoter region (32), were selected. A third set of SNPs was selected using Hapmap (http://www.hapmap.org) data. On the basis of these data and the forced inclusion of the first six SNPs, the program Tagger (http://www.broad.mit.edu/mpg/tagger/) selected six additional tagSNPs (SNP 1, 6, 7, 8, 9 and 11; Tables 1 and 2; Fig. 1). After aggressive tagging, Tagger also proposed four specific haplotypes of tagSNPs, each of which serves as a proxy for hidden SNPs. These haplotypes are hereafter referred to as predictor haplotypes (Table 3). As all hidden SNP alleles are captured with a mean R2 of 0.975 (with minimum R2 of 0.8), this set of 12 SNPs in combination with the four predictor haplotypes should cover all common genetic variation in CAT.


Figure 1
View larger version (6K):
[in this window]
[in a new window]
[Download PowerPoint slide]
  		Figure 1. Genomic region of catalase. (A) Positions of exons. (B) Position of SNPs. (a) First set of SNPs; (b) second set of SNPs. PrH1: hidden SNP for predictor-haplotype 1, rs10488736; PrH2: hidden SNP for predictor-haplotype 2, rs554576; PrH3: hidden SNP for predictor-haplotype 3, rs566979.

 


View this table:
[in this window]
[in a new window]

  		Table 1. SNP and PCR conditions

 


View this table:
[in this window]
[in a new window]

  		Table 2. Single SNP analysis

 


View this table:
[in this window]
[in a new window]

  		Table 3. Predictor-haplotype and tagSNP-haplotype analysis

 
Single SNP analysis
In total, 12 SNPs were analysed in 205 Swedish and 685 Polish noise-exposed labourers. Table 2 contains the resulting P-values after association analysis for the main effect and for the interaction between genotype and noise exposure level for every single SNP. We detected several significant genotype versus noise exposure level interactions. This means that a significant difference existed in genotype distribution between sensitive and resistant persons for the different noise exposure groups, which suggests a differential effect of the genotype on the noise susceptibility according to the noise exposure level. In case the interaction term was significant, the main genotype effect for each separate noise stratum was calculated. For the SNPs of which the interaction P-value was not significant, the main effect of genotypes on NIHL was calculated on the whole populations. No significant associations were found with these latter calculations, which suggest that none of the SNPs has a significant effect on noise susceptibility across all exposure levels.

In the Swedish population, we have obtained significant interaction P-values for two SNPs, SNP 5 (P = 0.004) and SNP 12 (P = 0.018). For SNP 5, the low-level (<85 dB) exposure group revealed a significant main effect P-value of 0.033. Figure 2A shows that in the low-level exposure group (<85 dB), sensitive workers are more likely to carry the AG genotype, whereas in resistant persons the GG genotype dominates. For the high-level exposure group (>92 dB), an opposite picture was observed: resistant persons are more likely to be AG heterozygotes while the GG genotype appears more often in sensitive subjects. A trend towards significance was calculated (P = 0.057). Although a significant interaction P-value of 0.018 was calculated in the Swedish population for SNP 12, no significant main effect P-values were obtained for the different exposure level strata. Since SNP 5 and 12 are in high linkage disequilibrium (LD) with each other (r2 = 0.981), SNP 12 displays almost an identical pattern as SNP 5 (Fig. 2B versus 2A).


Figure 2
View larger version (19K):
[in this window]
[in a new window]
[Download PowerPoint slide]
  		Figure 2. Bar charts for SNPs 5 and 12 in the Swedish and the Polish population. The two populations were divided into three noise exposure level groups (<85, 85–92, >92 dB). (A) SNP 5; (B) SNP 12. Filled bars represent susceptible subjects, whereas white bars represent resistant subjects. The genotypes are indicated below each bar chart. Numbers of all categories can be found in Table 5. An asterisk indicates that a significant main effect P-value was observed in this noise exposure level group.

 
The statistical analysis of individual SNPs in the Polish population resulted in a significant P-value for the interaction of genotypes with noise exposure for five SNPs: SNP1 (P = 0.022), SNP 2 (P = 0.001), SNP 5 (P = 6.45 x 10–5), SNP 9 (P = 0.024) and SNP 12 (P = 2.16 x 10–4). SNPs 5 and 12 also showed a significant interaction P-value in the Swedish population. However, for the Polish population, a significant main effect P-value for the low level (<85 dB) and the high level (>92 dB) exposure group was detected for both SNP 5 and 12 (for low level, P = 0.031 and 0.022, respectively; for high level: twice P = 0.022; Fig. 2A and B). When calculating the main effect for the different exposure level groups for SNPs 1 and 2, a significant main effect resulted (P = 0.015 and 0.025, respectively) in the low-level exposure group (<85 dB), whereas in the high-level exposure group (>92 dB) a significant main effect was calculated only for SNP 2 (P = 0.018). For SNP 1, the resistant workers are more likely to carry the CC genotype in the low-level exposure group (<85 dB), whereas heterozygous people (CT) are more likely to be susceptible. For SNP 2, heterozygous people (AG) of the low-level exposure group are more likely to be resistant, whereas sensitive samples more often carry the GG genotype (Fig. 3A and B). Again the opposite picture was observed in the high-level exposure group for both SNPs 1 and 2. For SNP 1, CC homozygotes tend to be more susceptible while CT heterozygotes are more often resistant, and for SNP 2, resistant persons are more likely to be GG homozygotes while the AG genotype appears more often in sensitive subjects. Analysis of the different exposure level groups for SNP 9 gave a significant main effect (P = 0.042) for the high-level exposure group (>92 dB). In this exposure level category, the homozygous genotype (AA) prevailed among the sensitive samples, while the heterozygous persons (AG) were more often resistant to noise (Fig. 3C).


Figure 3
View larger version (12K):
[in this window]
[in a new window]
[Download PowerPoint slide]
  		Figure 3. Bar charts of SNPs 1, 2 and 9 in the Polish population. The population was divided into three noise exposure level groups (<85, 85–92, >92 dB). (A) SNP 1; (B) SNP 2; (C) SNP 9. Rare homozygotes were added to heterozygotes for statistical analysis. Filled bars represent susceptible subjects, whereas white bars represent resistant subjects. The genotypes are indicated below each bar chart. Numbers of all categories can be found in Table 5. An asterisk indicates that a significant main effect P-value was observed in this noise exposure level group.

 
To test whether the differences in the duration of noise exposure between the two populations had a substantial effect on the results of the analysis, we have attempted to make the Polish population more comparable with the Swedish population by including only the samples that were exposed to noise for more than 10 years (n = 586) and repeating the single SNP analysis for the Polish population. In this analysis, the interactions between SNP 5 and 12, noise exposure levels and NIHL remained highly significant (data not shown). The other results were very similar. Only minor borderline switches between significant and non-significant had occurred.

Predictor haplotype and tagSNP haplotype analysis
To statistically analyse the four different predictor haplotypes (each of which serves as a proxy for hidden SNPs), the haplotypes were first inferred by Famhap. The frequency of the specific predictor haplotype (i.e. GTG, GGC, GC and TT for, respectively, predictor haplotypes 2-3-5, 5-7-10, 5-10 and 1-10) was subsequently compared with the frequency of all other haplotypes combined. This is equivalent to an allelic test for the hidden SNP, which has a high correlation with the respective predictor haplotype. In the Swedish population, no significant P-values were obtained for this analysis (Table 3). In the Polish population, however, this analysis resulted in highly significant interaction P-values for predictor haplotypes 1, 2 and 3 (P-values: 0.002, 2.08 x 10–4 and 1.26 x 10–4, respectively; Table 3). For the high-level exposure group (>92 dB), the main effect P-values for these three predictor haplotypes were significant (P-values: 0.022, 0.002 and 0.002, respectively; Table 3). In all the cases, the specific predictor haplotype appears more often among resistant subjects in the high-level exposure group (>92 dB; Fig. 4).


Figure 4
View larger version (14K):
[in this window]
[in a new window]
[Download PowerPoint slide]
  		Figure 4. Bar charts of predictor-haplotypes 1, 2 and 3 in the Polish population. The population was divided into three noise exposure level groups (<85, 85–92, >92 dB). (A) PrH1, predictor-haplotype 1; (B) PrH2, predictor-haplotype 2; (C) PrH3, predictor-haplotype 3. The frequency of the specific haplotype that is representative for the hidden SNP is compared with that of all the other haplotypes for the specific combination of SNPs that composes the predictor-haplotype. Filled bars represent susceptible subjects, whereas white bars represent resistant subjects. Numbers of all categories can be found in Table 5. An asterisk indicates that a significant main effect P-value was observed in this noise exposure level group.

 
In addition to single SNPs and predictor haplotypes, we also analysed the tagSNP haplotypes of CAT (haplotypes composed of all tagSNPs). Only the most frequent haplotypes, with a total frequency of 0.90, were included in the analysis. The other haplotypes were combined into one group. Table 4 gives an overview of the most frequent tagSNP haplotypes and their frequencies in the Swedish and Polish population. Logistic regression resulted in a main P-value of 0.933 for the Swedish population (Table 3). For the Polish population, the interaction between tagSNP haplotype and noise exposure level resulted in a significant P-value (P = 0.005). When calculating the main effect for the separate exposure level groups, a significant effect was detected (P = 0.019) in the high-level exposure group (>92 dB). In the Polish high-level exposure group, more sensitive samples have tagSNP haplotypes 2, 4 and 5, and more resistant samples have tagSNP haplotypes 1, 6 and 7 (Fig. 5).


Figure 5
View larger version (12K):
[in this window]
[in a new window]
[Download PowerPoint slide]
  		Figure 5. Bar charts of the tagSNP-haplotypes for the different noise exposure level groups of the Polish population. (A) low noise level exposure group (<85 dB); (B) middle noise level exposure group (85–92 dB); (C) high noise level exposure group (>92 dB). The specific tagSNP-haplotypes are enlisted in Table 4. Filled bars represent susceptible subjects, whereas white bars represent resistant subjects. Numbers of all categories can be found in Table 5. An asterisk indicates that a significant main effect P-value was observed in this noise exposure level group.

 


View this table:
[in this window]
[in a new window]

  		Table 4. Frequencies of the most frequent tagSNP haplotypes in Sweden and Poland

 


View this table:
[in this window]
[in a new window]

  		Table 5. Description of the Swedish and Polish noise-exposed populations by affection status and noise exposure level

 

   DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 RESULTS
 DISCUSSION
MATERIAL AND METHODS
 REFERENCES
 
In this study, the association between several SNPs in CAT and NIHL, taking into account noise exposure levels, was analysed. A previous association study on the involvement of polymorphisms in oxidative stress genes in NIHL, including CAT, could not detect any significant association (18). However, we have demonstrated here that the effect of CAT polymorphisms on NIHL can only be detected when noise exposure levels are taken into account. Several significant interactions were found, including those for five different tagSNPs, three predictor haplotypes and the tagSNP haplotype. Two of the significant tagSNPs showed interactions in both populations. Overall, when subdividing the population by noise exposure level, and testing for association in the separate noise exposure strata, we observed that the majority of the SNPs showed a significant result in the high-level exposure group (>92 dB), indicating that polymorphisms in CAT have a larger effect when carriers are exposed to higher levels of noise. This is not entirely unexpected because the higher the noise level, the more harmful it might be, and consequently the more effect that will be seen. Whether this is the only NIHL susceptibility gene with this characteristic remains to be elucidated, as up to now only a limited number of studies on the genetic factors involved in NIHL have been performed. Moreover, an analysis taking into account exposure levels in humans has never been done before. Hence, there is a chance that a similar effect might be observed in other genes too. These results illustrate the complexity of NIHL and indicate that accounting for effect modification by exposure levels is an interesting option for future research that aims at the investigation of the genetic factors contributing to NIHL.

For SNP 12 in the Swedish population, a significant interaction P-value was calculated, but no significant main effect could be detected when the analysis was performed on the separate noise exposure level groups. This may be due to small sample numbers in the different noise exposure level groups. Power calculations for the Swedish population, which consisted of a selection derived from ~1200 samples, yielded a power of 80% for the detection of a causative allele with a relative risk of 2 when we assume an additive model (Genetic Power Calculator; http://www.statgen.iop.kcl.ac.uk/gpc). When this population is divided into three noise exposure groups, it is possible that the size of the separate noise exposure groups is too small to be able to detect an effect.

Although the strongest interaction P-values (SNPs 5 and 12) can be observed in both the Swedish and the Polish population, this is not the case for all significant interaction P-values. Three are only observed in the Polish population (SNPs 1, 2 and 9). However, in order to confirm the association of a gene with a disease, it is not necessary that identical SNPs are associated in all populations under study. Different SNPs associated in different populations, but within the same gene, can be regarded as a replication (39).

Even though SNPs 5 and 12 resulted in significant interaction P-values in both the Swedish and Polish population, the direction of the genotype effect in the different noise exposure level groups differs in both populations (Fig. 2A and B) (40, 41). The fact that a certain SNP allele is protective in one population but disease causing in another population has already been observed. A theoretical model demonstrated that these ‘flip-flop’ associations could occur when the investigated variant is correlated, through interactive effects or through LD, with a causal variant at another locus and that these associations may indeed be confirmations (42). We cannot rule out the possibility of a false-positive observation. Alternatively, the effect modification by other unknown environmental factors that differ between the two populations may explain this observation (43).

The differences observed in the two populations may also be due to differences in sample selection. Different criteria were used for the two populations. For the Swedish population, the hearing threshold levels (HTLs) at 3 kHz were used as a selection criterion, whereas for the Polish population the HTLs at 4 and 6 kHz were considered. Another difference between the two populations is the fact that for the Swedish population the resistant and susceptible samples were matched for age and level of noise exposure, whereas in the Polish population the selection of resistant and susceptible samples was based on the ISO 1999 norms, which take into account age, tenure of exposure and noise exposure level. This procedure is therefore not the same, but is comparable with the matching method that was used in the Swedish population. Another factor that may play a role in the differences that were observed between the two populations is the kind of noise that the samples were exposed to. It is well known that impulse noise is more damaging than steady-state noise (44,45). The two populations consist of labourers from different types of factories and it is likely that some factories have more impulse noise than others. However, the type of noise was only documented for the Polish population.

Quite remarkable is the opposite effect of the same genotype between the low and the high exposure group for SNP 1, 2, 5 and 12. This finding seems unusual, but similar observations have been done previously (46). The interactions between day care exposure in the first 6 months of life (environmental factor) and genotypes of several candidate genes (genetic factor) and their effects on cytokine response profiles and on the development of atopic phenotypes in the first year of life have been investigated in a cohort of children with asthma (46). The effects of a particular genotype on the phenotype were opposite depending on whether the child attended day care or not. The authors therefore suggested that these genotypic effects at these particular loci were modified by the environment in such a manner that the same genotype was associated with either protection or risk for a phenotype, depending on the exposure to day care (43).

Another hypothesis that can explain this remarkable observation is the possibility that genotype CC for SNP 1, AG for SNP 2, GG for SNP 5 and AA for SNP 12 gives people a better hearing under normal conditions, i.e. under low noise exposure, while this same genotype makes people more susceptible to hearing loss under conditions of high noise exposure. In other words, these specific genotypes would give people a more sensitive hearing that is also more vulnerable to loud noise.

Alternatively, this observation could be the result of sound conditioning. Moderate level noise exposure can protect the ear from a following high-level exposure (47). This sound conditioning can be induced by leisure time noise such as music listening or it might be a matter of the temporal schedule of noise exposure at work. Since it has been implicated that conditioning and the acquired resistance to noise is the consequence of an enhanced antioxidant response system, it is tempting to speculate that certain genotypes in CAT might have different effects on noise susceptibility depending on whether the subject has been conditioned to noise or not (24).

SNP 2 is a polymorphism that is located in the promotor region of CAT. Computer analysis has indicated that the two alleles of this variant bind different transcription factors (48). In addition, higher transcriptional activity of the A variant was observed in human hepatocellular liver carcinoma cell line 2 (HepG2) and human erythroleukemia cells (K562 cells), which was confirmed by the catalase levels observed in red blood cells (48). This higher transcriptional activity of the A variant could not be replicated and even an opposite effect was demonstrated in another study on red blood cells (49). According to this study, the A variant of catalase shows a decreased catalase activity in red blood cells. Further studies are required to explain these opposite findings, but these results might again indicate that the same genotype could have an opposite effect depending on the type of exposure to certain environmental factors. This demonstrates the importance of including the analysis of environmental factors into the investigation of complex traits.

Three predictor haplotypes that are proxies for three different hidden SNPs had a significant interaction P-value in the Polish population. The three hidden SNPs lie between SNPs 10 and 11. The latter two SNPs are in high LD with each other and most likely also with the hidden SNPs. It is tempting to speculate that the significant results obtained for the three predictor haplotypes could all refer to the same causative SNP.

In addition to the significant results for the predictor haplotypes, a significant interaction P-value was obtained for the tagSNP haplotype in the Polish population. This was not replicated in the Swedish population. Since eight, different tagSNP haplotypes were needed to cover 90% of all tagSNP haplotypes, this non-significant finding may be due to a large number of degrees of freedom, decreasing the power in an already small population.

On the basis of the current findings, a single putative causative variant cannot be identified. However, there might be many variants in CAT influencing noise susceptibility. For example, in case catalase functioning is crucially dependent on expression level, one would expect that any variant influencing expression would show an association. The SNPs that gave significant results in our association study were spread all over the gene. In addition, none of the directly analysed or hidden SNPs were non-synonymous coding SNPs, i.e. SNPs that change an amino acid at the protein level. Only two SNPs were synonymous (SNPs 4 and 10). Synonymous SNPs close to the border of the exon might be located in exonic splicing enhancers or silencers that could affect splicing (50). The causative variants may not be (non-)synonymous coding SNPs, but may be non-coding SNPs with an effect on the expression level or the activity of catalase. Polymorphisms located in a regulatory region or in an internal promotor may affect expression levels of catalase. Also, intronic variations located in the vicinity of exons can sometimes induce cryptic splicing. This may result in loss or gain of a limited number of amino acids, which might lead to a small effect on the function of the gene.

Recently, mice lacking catalase have been generated (51). In view of the results that we have obtained in this study, it would be most interesting to expose these mice to noise and measure their susceptibility.

In this study, we tested 12 SNPs and 4 predictor haplotypes for association with NIHL and tested for noise exposure level versus genotype interaction. One could argue that a multiple testing correction should be done to keep the type I error under control. However, it is quite controversial how this should be done. The Bonferroni correction uses the number of independent hypothesis to correct the P-value. Owing to high LD between the SNPs, however, the 16 tests that we have performed are not independent, and the Bonferroni correction would be too strict. The false discovery rate (FDR) correction is a good alternative (52). When an FDR correction is applied to our data, using n = 16 (single SNPs and predictor haplotypes) and only considering interaction P-values, all the P-values remain significant for the Swedish and Polish populations. As genetic epidemiologists have not yet come to a consensus how to correct for multiple testing for the investigation of a single gene and as replication in different populations is more important than having very low P-values (38,53), we believe that the findings in the current study are indicative of a true association between the catalase gene and NIHL.

In conclusion, several associations were detected between variations in catalase and NIHL susceptibility in two independent, noise-exposed populations. In particular, significant interactions between genotypic variations and noise exposure levels and their effect on NIHL were obtained, often resulting in significant main effects in the high-level noise exposure group. Further association analysis and functional studies will contribute to the identification of the underlying mechanism of NIHL.


   MATERIAL AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 RESULTS
 DISCUSSION
 MATERIAL AND METHODS
REFERENCES
 
Samples
Swedish population
A detailed description of the Swedish sample population can be found elsewhere (16,18). In brief, 1261 male noise-exposed workers from two paper pulp mills and one steel factory in the mid-western part of Sweden were divided into nine categories (three age-ranges, < 35, 35–50 and > 50 years, and three occupational noise exposure categories, ≤ 85 dBA, 86–91 dBA and ≥ 92 dBA, all leq, 8 h, 5 days a week). From each category, the 10% most-resistant and the 10% most-sensitive persons were selected using the HTL of the left ear at 3 kHz as a measure of noise susceptibility. A noise level of 3 kHz was preferred for the selection of susceptible individuals over 4 or 6 kHz for several reasons. Firstly, an increase in damage leads to a widening of the initial 4–6 kHz notch to lower frequencies (ISO 1999—International Organization for Standardization, 1990). Furthermore, the HTL at 3 kHz continues to increase over a longer period of time than does the HTL at 4 and 6 kHz (54) and the majority of the Swedish subjects (79%) had been exposed to noise for ≥ 20–30 years. In addition, the ISO 1999 norm shows that individuals who have been exposed to noise (≥ 90 dBA) for ≥ 20 years have a higher HTL at 3 kHz than at 4 and 6 kHz in the 0.1 fractile. Blood samples were taken from a total of 218 subjects: 104 noise susceptible and 114 noise resistant subjects. After genotyping, data polishing was initiated that consisted of the removal of samples with more than 10% missing genotypes and the removal of genetic outliers detected with the programs CHECKHET (http://www.smd.qmul.ac.uk/statgen/dcurtis/software.html) and GRR (55). CHECKHET detects subjects with a different genetic background compared with the majority of the tested population, while GRR detects individuals with similar allele patterns which could be indicative of related individuals. A total of 205 samples, consisting of 98 noise susceptible and 107 noise resistant subjects (Table 5), were used for further analysis.

Polish population
Information concerning the audiometric status, noise exposure and exposure to chemicals was gathered from ~4500 Polish workers from different industries, including a coal mine, an electric power station, a dockyard, a glass bottle factory and a lacquer and paint factory. An inclusion criterion for this study was an exposure to noise of at least 1 year. Subjects with a history of middle ear disease, conductive hearing loss or skull trauma and subjects with a family history of hearing loss were excluded. Unlike for the selection in the Swedish population, HTLs at 4 and 6 kHz, the two frequencies that are most easily affected by NIHL, were evaluated. For each subject, a Z-score that corrects for age, gender, tenure of exposure and noise exposure level was generated based on the ISO 1999 norms (ISO 1999—International Organization for Standardization, 1990). This Z-score will represent the amount of hearing loss of each labourer taking the different corrections into account. Details concerning this Z-score generation have been published elsewhere (56). Subsequently, the 10% most-susceptible and the 10% most-resistant subjects were selected at the two extremes of the phenotypic spectrum, resulting in 354 samples from the highly susceptible and 348 from the highly resistant category. After data polishing, 347 noise susceptible and 338 noise resistant subjects were used for further analysis (Table 5).

Although, for the Polish population, the interaction noise exposure level versus genotype was tested using noise exposure level as a continuous variable, we subsequently estimated the stratum-specific main effect of the genotype by subsetting the samples of the Polish population into the same noise exposure level categories as were used in the Swedish population.

SNP selection
Two sets of three SNPs were selected based on the data in dbSNP (http://www.ncbi.nlm.nih.gov/) or Genbank (http://www.ncbi.nlm.nih.gov/Genbank/). To fully cover the common variations in CAT, these six initial SNPs were supplied with six tagSNPs, selected from the International HapMap Database—phase II. These tagSNPs were selected using the program Tagger hereby forcing in the first two sets of three SNPs. The aggressive tagging algorithm ensures the selection of the most informative set of SNPs, minimizing the genotyping effort. In addition, several specific combinations (haplotypes) of either two or three tagSNPs were obtained, each of which serves as a proxy for hidden SNPs. In this study, the latter haplotypes will be referred to as predictor haplotypes. The total set of 12 SNPs and four predictor haplotypes should cover all common variations that are present in CAT gene.

Genotyping methods
Genomic DNA was extracted from all blood samples using standard procedures. PCR primers were designed with Primer3 software (http://www.biotools.umassmed.edu/bioapps/primer3_www.cgi; Table 1). SNP primers consisted of the nucleotide sequences immediately adjacent to the SNP, either on the forward or on the reverse strand, with a minimal melting temperature of 60°C (Table 1). PCR was performed with 50 ng template DNA in 20 mM Tris–HCl, pH 8.4, 50 mM KCl, 1.5 mM MgCl2, 200 µM dNTPs (Amresco Inc., Solon, OH, USA), 0.5 or 1 µM of each PCR primer (Table 1), and 0.0250 or 0.0125 U/µl Taq DNA polymerase (Table 1; Invitrogen Life Technologies, San Diego, CA, USA). PCR was initiated by 5 min denaturation at 94°C, followed by 35 cycles of 1 min denaturation at 94°C, 1 min annealing at 53, 55 or 59°C (Table 1) and 1 min extension at 72°C, and ended with an extension step of 10 min at 72°C. The concentration of the PCR product was subsequently estimated by agarose gel electrophoresis using Massruler (Invitrogen Life Technologies) as a reference.

SNP genotyping was performed either with the AcycloPrime-Fluorescence Polarization (FP) SNP detection system (PerkinElmer Life Sciences, Boston, MA, USA) or with the ABI PRISM® SNaPshot TM Multiplex Kit (Applied Biosystems, Foster City, CA, USA; Table 1). FP was performed according to the manufacturer's instructions. Briefly, PCR products were cleared of excess primers and dNTPs by enzymatic digestion with PCR clean-up reagent (PerkinElmer Life Sciences) during 1 h at 37°C, followed by 15 min heat- inactivation at 80°C. Next, the appropriate Acycloprime-FP mix and SNP primers were added. The primer extension reaction consisted of 2 min initial denaturation at 95°C, followed by cycles of 15 s denaturation at 95°C and 30 s annealing extension at 55°C. The fluorescence polarization was read in a Victor 1420 Multilabel Counter (PerkinElmer Life Sciences) after 15, 25 and 35 cycles of primer extension. SNPscorer software (PerkinElmer Life Sciences) was used to analyse the results.

To perform SNaPshot reactions with the ABI PRISM® SNaPshot TM Multiplex Kit (Applied Biosystems), PCR products were first cleared of excess primers and dNTPs by enzymatic digestion with 0.33 U/ml Exonuclease I (England Biolabs® Inc., Ipswich, MA, USA) and 0.16 U/µl calf intestine alkaline phosphatase (CIAP; Amersham Biosciences, Piscataway, NJ, USA) in 50 mM Tris–HCl (pH 9.0) and 1 mM MgCl2 during 1 h at 37°C, followed by 15 min heat inactivation at 72°C. Subsequently, SNaPshot Ready Reaction Mix (Applied Biosystems), diluted 1:9 with Big DyeTM dilutionbuffer (200 mM Tris–HCl, pH 9.0, 5 mM MgCl2; Applied Biosystems), together with 0.1 µM SNP primer, were added to the reaction. Primer extension was achieved through 25 cycles of 10 s denaturation at 96°C, 5 s hybridization at 50°C and 30 s extension at 60°C. After a final clean-up step using 0.5 U CIAP (1 h at 37°C, 15 min at 72°C), the products were size separated using an ABI PRISM® 3130xl Genetic analyser (Applied Biosystems) and the results were analysed using ABI PRISM® GeneMapperTM Software, Version 3.0 (Applied Biosystems).

Statistical analysis
Hardy–Weinberg equilibrium was checked for all individual SNPs using a {chi}2test for goodness-of-fit (HWE version 1.2) (57). To test for associations between single SNPs and noise susceptibility, stepwise backward logistic regression was performed, accounting for exposure and age, allowing for interactions and quadratic effects (age2, age x noise exposure level, noise exposure level2 and noise exposure level x genotype). Since age2, age x noise exposure level and noise exposure level2 were always significant, interaction between the level of noise exposure and genotypes was tested by a likelihood ratio test, comparing the model that includes the interaction term noise exposure level x genotypes with the model only including main effects for genotype, exposure levels and age and the interaction terms age2, age x noise exposure level and noise exposure level2. In case the interaction term exposure level x genotypes was significant, the main effect of the genotype was calculated for each separate noise stratum. In case the interaction term exposure level x genotypes was not significant, the main effect of the genotype was calculated for the whole population. The significance of the main effects of the SNP genotype was tested using a likelihood ratio test comparing a model including genotype, age, noise exposure levels and the interaction terms age2, age x noise exposure level and noise exposure level2 with a model only including age, exposure levels, age2, age x noise exposure level and noise exposure level2.

In the case of SNPs with a low minor allele frequency, where the logistic regression model did not converge, the rare homozygous group was added to the heterozygous group for this SNP.

In addition to single SNPs, two different haplotype associations were tested: (i) the predictor haplotypes and (ii) the tagSNP haplotypes. The predictor haplotypes test a hidden SNP, where one specific haplotype corresponds to one allele of the hidden SNPs. The tagSNP haplotypes consist of the haplotypes composed of all the tagSNPs. Famhap (58) was used to infer haplotypes and their likelihood for each individual. During haplotype association testing, inferred haplotypes were always weighted by their likelihood as estimated by Famhap. The specific haplotypes of the predictor haplotypes were tested against all other possible haplotypes of the selected SNPs. For the analysis of tagSNP haplotypes, the 10% least-frequent haplotypes were combined into one group to reduce the degrees of freedom and maximize the power of the test. The interaction between noise exposure level and predictor and tagSNP haplotypes, respectively, and the main effect of the predictor and tagSNP haplotypes, respectively, on NIHL were statistically analysed using logistic regression as described for the single SNP analysis. All statistical analyses were performed using SPSS 12.0 (SPSS 12.0 for Windows, SPSS Inc., Chicago, IL, USA).


   ACKNOWLEDGEMENTS
 
This research was supported by a grant of the British Royal Institute for Deaf and Hard of Hearing People (RNID) to G.V.C. and L.V.L., a TOP grant of the University of Antwerp to G.V.C., a Bilateral Scientific Cooperation grant between Belgium and Poland, the State Scientific Committee for Research (Grant No. PB 0911/P05/2004/26), the 6th European Framework Project under the Marie Curie Host Fellowship for the Transfer of Knowledge ‘NoiseHear’ (Contract No. MTKD-CT-2004-003137) and the Swedish Council for Working Life and Social Research.

Conflict of Interest statement. None declared.


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 RESULTS
 DISCUSSION
 MATERIAL AND METHODS
 REFERENCES
 

  1. Lim D.J. Effects of noise and ototoxic drugs at the cellular level in the cochlea: a review. Am. J. Otolaryngol. (1986) 7:73–99.[ISI][Medline]

  2. Fechter L.D. Promotion of noise-induced hearing loss by chemical contaminants. J. Toxicol. Environ. Health A. (2004) 67:727–740.[CrossRef][ISI][Medline]

  3. Campo P., Lataye R. Noise and solvent, alcohol and solvent: two dangerous interactions on auditory function. Noise Health (2000) 3:49–57.[Medline]

  4. Sliwinska-Kowalska M., Zamyslowska-Szmytke E., Szymczak W., Kotylo P., Fiszer M., Wesolowski W., Pawlaczyk-Luszczynska M., Bak M., Gajda-Szadkowska A. Effects of coexposure to noise and mixture of organic solvents on hearing in dockyard workers. J. Occup. Environ. Med. (2004) 46:30–38.[CrossRef][ISI][Medline]

  5. Toppila E., Pyykko I.I., Starck J., Kaksonen R., Ishizaki H. Individual risk factors in the development of noise-induced hearing loss. Noise Health (2000) 2:59–70.[Medline]

  6. Borg E., Canlon B., Engstrom B. Noise-induced hearing loss. Literature review and experiments in rabbits. Morphological and electrophysiological features, exposure parameters and temporal factors, variability and interactions. Scand. Audiol. Suppl. (1995) 40:1–147.[Medline]

  7. Erway L.C., Shiau Y.W., Davis R.R., Krieg E.F. Genetics of age-related hearing loss in mice. III. Susceptibility of inbred and F1 hybrid strains to noise-induced hearing loss. Hear. Res. (1996) 93:181–187.[CrossRef][ISI][Medline]

  8. Davis R.R., Newlander J.K., Ling X., Cortopassi G.A., Krieg E.F., Erway L.C. Genetic basis for susceptibility to noise-induced hearing loss in mice. Hear. Res. (2001) 155:82–90.[CrossRef][ISI][Medline]

  9. Harding G.W., Bohne B.A., Vos J.D. The effect of an age-related hearing loss gene (Ahl) on noise-induced hearing loss and cochlear damage from low-frequency noise. Hear. Res. (2005) 204:90–100.[CrossRef][ISI][Medline]

  10. Kozel P.J., Davis R.R., Krieg E.F., Shull G.E., Erway L.C. Deficiency in plasma membrane calcium ATPase isoform 2 increases susceptibility to noise-induced hearing loss in mice. Hear. Res. (2002) 164:231–239.[CrossRef][ISI][Medline]

  11. Holme R.H., Steel K.P. Progressive hearing loss and increased susceptibility to noise-induced hearing loss in mice carrying a Cdh23 but not a Myo7a mutation. J. Assoc. Res. Otolaryngol. (2004) 5:66–79.[CrossRef][ISI][Medline]

  12. Ohlemiller K.K., McFadden S.L., Ding D.L., Flood D.G., Reaume A.G., Hoffman E.K., Scott R.W., Wright J.S., Putcha G.V., Salvi R.J. Targeted deletion of the cytosolic Cu/Zn-superoxide dismutase gene (Sod1) increases susceptibility to noise-induced hearing loss. Audiol. Neurootol. (1999) 4:237–246.[CrossRef][Medline]

  13. Ohlemiller K.K., McFadden S.L., Ding D.L., Lear P.M., Ho Y.S. Targeted mutation of the gene for cellular glutathione peroxidase (Gpx1) increases noise-induced hearing loss in mice. J. Assoc. Res. Otolaryngol. (2000) 1:243–254.[CrossRef][Medline]

  14. Borg E. Noise-induced hearing loss in normotensive and spontaneously hypertensive rats. Hear. Res. (1982) 8:117–130.[CrossRef][ISI][Medline]

  15. Van Laer L., Carlsson P.I., Ottschytsch N., Bondeson M.L., Konings A., Vandevelde A., Dieltjens N., Fransen E., Snyders D., Borg E., et al. The contribution of genes involved in potassium-recycling in the inner ear to noise-induced hearing loss. Hum. Mutat. (2006) 27:786–795.[CrossRef][ISI][Medline]

  16. Carlsson P.I., Borg E., Grip L., Dahl N., Bondeson M.L. Variability in noise susceptibility in a Swedish population: the role of 35delG mutation in the Connexin 26 (GJB2) gene. Audiol. Med. (2004) 2:123–130.[CrossRef]

  17. Rabinowitz P.M., Pierce Wise J., sr, Hur Mobo B., Antonucci P.G., Powell C., Slade M. Antioxidant status and hearing function in noise-exposed workers. Hear. Res. (2002) 173:164–171.[CrossRef][ISI][Medline]

  18. Carlsson P.I., Van Laer L., Borg E., Bondeson M.L., Thys M., Fransen E., Van Camp G. The influence of genetic variation in oxidative stress genes on human noise susceptibility. Hear. Res. (2005) 202:87–96.[CrossRef][ISI][Medline]

  19. Yamane H., Nakai Y., Takayama M., Iguchi H., Nakagawa T., Kojima A. Appearance of free radicals in the guinea pig inner ear after noise-induced acoustic trauma. Eur. Arch. Otorhinolaryngol. (1995) 252:504–508.[CrossRef][Medline]

  20. Yamasoba T., Harris C., Shoji F., Lee R.J., Nuttall A.L., Miller J.M. Influence of intense sound exposure on glutathione synthesis in the cochlea. Brain. Res. (1998) 804:72–78.[CrossRef][ISI][Medline]

  21. Henderson D., McFadden S.L., Liu C.C., Hight N., Zheng X.Y. The role of antioxidants in protection from impulse noise. Ann. NY Acad. Sci. (1999) 884:368–380.[Abstract/Free Full Text]

  22. Lautermann J., Crann S.A., McLaren J., Schacht J. Glutathione-dependent antioxidant systems in the mammalian inner ear: effects of aging, ototoxic drugs and noise. Hear. Res. (1997) 114:75–82.[CrossRef][ISI][Medline]

  23. Pouyatos B., Gearhart C.A., Fechter L.D. Acrylonitrile potentiates hearing loss and cochlear damage induced by moderate noise exposure in rats. Toxicol. Appl. Pharmacol. (2005) 204:46–56.[CrossRef][ISI][Medline]

  24. Jacono A.A., Hu B., Kopke R.D., Henderson D., Van De Water T.R., Steinman H.M. Changes in cochlear antioxidant enzyme activity after sound conditioning and noise exposure in the chinchilla. Hear. Res. (1998) 117:31–38.[CrossRef][ISI][Medline]

  25. Chance B., Sies H., Boveris A. Hydroperoxide metabolism in mammalian organs. Physiol. Rev. (1979) 59:527–605.[Free Full Text]

  26. Keilin D., Hartree E.F. Coupled oxidation of alcohol. Proc. R. Soc. Lond. B. Biol. Sci. (1936) 119:141–159.

  27. Keilin D., Hartree E.F. Properties of catalase. Catalysis of coupled oxidation of alcohols. Biochem. J. (1945) 39:293–301.[ISI][Medline]

  28. Aebi H.E., Wyss S.R. Acatalasemia. In: The Metabolic Basis of Inherited Disease—Stanbury J.B., Wyngaarden J.B., Fredrickson D.S., eds. (1978) New York: McGraw-Hill. 1792–1807.

  29. Deisseroth A., Dounce A.L. Catalase: physical and chemical properties, mechanism of catalysis, and physiological role. Physiol. Rev. (1970) 50:319–375.[Free Full Text]

  30. Schonbaum G.R., Chance B. Catalase. In: The Enzymes—Boyer P.D., ed. (1976) 13. New York: Academic Press. 363–408.

  31. Junien C., Turleau C., de Grouchy J., Said R., Rethore M.L., Tenconi R., Dufier J.L. Regional assignment of catalase (CAT) gene to band 11p13. Association with the Aniridia–Wilms’ tumor-gonadoblastoma (WAGR) complex. Ann. Genet. (1980) 23:165–168.[ISI][Medline]

  32. Quan F., Korneluk R.G., Tropak M.B., Gravel R.A. Isolation and characterization of the human catalase gene. Nucleic Acids Res. (1986) 14:5321–5335.[Abstract/Free Full Text]

  33. Bell G.I., Najarian R.C., Mullenbach G.T., Hallewell R.A. cDNA sequence coding for human kidney catalase. Nucleic Acids Res. (1986) 14:5561–5562.[ISI][Medline]

  34. Pask R., Cooper J.D., Walker N.M., Nutland S., Hutchings J., Dunger D.B., Nejentsev S., Todd J.A. No evidence for a major effect of two common polymorphisms of the catalase gene in type 1 diabetes susceptibility. Diabetes Metab. Res. Rev. (2006) 22:356–360.[CrossRef][ISI][Medline]

  35. Chistiakov D.A., Savost'ianov K.V., Turakulov R.I., Shcherbacneva L.N., Mamaeva G.G., Balabolkin M.I., Nosikov V.V. Nucleotide substitution C1167T in the catalase gene and position of nearby polymorphic markers DS11S907 and D11S2008 are connected with development of diabetes mellitus type 2. Mol Biol (Mosk) (2000) 34:863–867.[Medline]

  36. Jiang Z., Akey J.M., Shi J., Xiong M., Wang Y., Shen Y., Xu X., Chen H., Wu H., Xiao J., et al. A polymorphism in the promoter region of catalase is associated with blood pressure levels. Hum. Genet. (2001) 109:95–98.[CrossRef][ISI][Medline]

  37. Casp C.B., She J.X., McCormack W.T. Genetic association of the catalase gene (CAT) with vitiligo susceptibility. Pigment Cell Res. (2002) 15:62–66.[CrossRef][ISI][Medline]

  38. Goth L., Rass F., Pay A. Catalase enzyme mutations and their association with diseases. Mol. Diagn. (2004) 8:141–149.[CrossRef][Medline]

  39. Neale B.M., Sham P.C. The future of association studies: gene-based analysis and replication. Am. J. Hum. Genet. (2004) 75:353–362.[CrossRef][ISI][Medline]

  40. Li Y.J., Oliveira S.A., Xu O., Martin E.R., Stenger J.E., Scherzer C.R., Hauser M.A., Scott W.K., Small G.W., Nance M.A., et al. Glutathione S-transferase omega-1 modifies age-at-onset of Alzheimer disease and Parkinson disease. Hum Mol Genet. (2003) 12:3259–3267.[Abstract/Free Full Text]

  41. Kolsch H., Linnebank M., Lutjohann D., Jessen F., Wullner U., Harbrecht U., Thelen K.M., Kreis M., Hentschel F., Schulz, et al. Polymorphisms in glutathione S-transferase omega-1 and AD, vascular dementia, and stroke. Neurology (2004) 63:2255–2260.[Abstract/Free Full Text]

  42. Lin P.I., Vance J.M., Pericak-Vance M.A., Martin E.R. No gene is an island: the flip-flop phenomenon. Am. J. Hum. Genet. (2007) 80:531–538.[CrossRef][ISI][Medline]

  43. Ober C. Perspectives on the past decade of asthma genetics. J. Allergy Clin. Immunol. (2005) 116:274–278.[CrossRef][ISI][Medline]

  44. Starck J., Pekkarinen J., Pyykko I. Impulse noise and hand-arm vibration in relation to sensory neural hearing loss. Scand. J. Work Environ. Health. (1988) 14:265–271.[ISI][Medline]

  45. Suvorov G., Denisov E., Antipin V., Kharitonov V., Starck J., Pyykko I., Toppila E. Effects of peak levels and number of impulses to hearing among forge hammering workers. Appl. Occup. Environ. Hyg. (2001) 16:816–822.[CrossRef][Medline]

  46. Hoffjan S., Nicolae D., Ostrovnaya I., Roberg K., Evans M., Mirel D.B., Steiner L., Walker K., Shult P., Gangnon R.E., et al. Gene–environment interaction effects on the development of immune responses in the 1st year of life. Am. J. Hum. Genet. (2005) 76:696–704.[CrossRef][ISI][Medline]

  47. Campo P., Subramaniam M., Henderson D. The effect of ‘conditioning’ exposures on hearing loss from traumatic exposure. Hear. Res. (1991) 55:195–200.[CrossRef][ISI][Medline]

  48. Forsberg L., Lyrenas L., de Faire U., Morgenstern R. A common functional C-T substitution polymorphism in the promoter region of the human catalase gene influences transcription factor binding, reporter gene transcription and is correlated to blood catalase levels. Free Radic. Biol. Med. (2001) 30:500–505.[CrossRef][ISI][Medline]

  49. Nadif R., Mintz M., Jedlicka A., Bertrand J.P., Kleeberger S.R., Kauffmann F. Association of CAT polymorphisms with catalase activity and exposure to environmental oxidative stimuli. Free Radic. Res. (2005) 39:1345–1350.[CrossRef][ISI][Medline]

  50. Caputi M., Kendzior R.J. Jr, Beemon K.L. A nonsense mutation in the fibrillin-1 gene of a Marfan syndrome patient induces NMD and disrupts an exonic splicing enhancer. Genes. Dev. (2002) 16:1754–1759.[Abstract/Free Full Text]

  51. Ho Y.S., Xiong Y., Ma W., Spector A., Ho D.S. Mice lacking catalase develop normally but show differential sensitivity to oxidant tissue injury. J. Biol. Chem. (2004) 279:32804–32812.[Abstract/Free Full Text]

  52. Sabatti C., Service S., Freimer N. False discovery rate in linkage and association genome screens for complex disorders. Genetics (2003) 164:829–833.[Abstract/Free Full Text]

  53. Todd J.A. Statistical false positive or true disease pathway? Nat. Genet. (2006) 38:731–733.[CrossRef][ISI][Medline]

  54. Taylor W., Pearson J., Mair A., Burns W. Study of Noise and Hearing in Jute Weaving. J. Acoust. Soc. Am. (1965) 38:113–120.[CrossRef][ISI][Medline]

  55. Abecasis G.R., Cherny S.S., Cookson W.O., Cardon L.R. GRR: graphical representation of relationship errors. Bioinformatics (2001) 17:742–743.[Abstract/Free Full Text]

  56. Sliwinska-Kowalska M., Dudarewicz A., Kotylo P., Zamyslowska-Szmytke E., Pawlaczyk-Luszczynska M., Gajda-Szadkowska A. Individual susceptibility to noise-induced hearing loss: choosing an optimal method of retrospective classification of workers into noise-susceptible and noise-resistant groups. Int. J. Occup. Med. Environ. Health (2006) 19:235–245.[Medline]

  57. Ott J. Analysis of Human Genetic Linkage. (1999) 3rd. Baltimore: Johns Hopkins University Press.

  58. Knapp M., Becker T. Family-based association analysis with tightly linked markers. Hum. Hered. (2003) 56:2–9.[CrossRef][ISI][Medline]


Add to CiteULike CiteULike   Add to Connotea Connotea   Add to Del.icio.us Del.icio.us    What's this?