Introduction

Esophageal cancer (EC) is a gravely lethal malignancy with poor 5-year survival rates. EC can be subdivided into esophageal adenocarcinoma (EAC), more prevalent in the Western world with gastro-esophageal reflux disease (GERD) and obesity as main risk factors. Esophageal squamous cell carcinoma (ESCC) is more prominent in Asia, predisposed by the use of tobacco and alcohol (ethanol) and local dietary habits.1 So tactics in averting contact with environmental and dietary factors such as alcohol intake are crucial in esophageal carcinogenesis.

Ethanol, although classified by WHO-IARC as a Group-1 carcinogen, in its pure form is less toxic than its principal metabolite acetaldehyde (AA), which is a carcinogen and partly responsible for the cancer risk.2 Acetaldehyde is also detectable in tobacco smoke.3 The key enzymes for the oxidation of ethanol and its metabolite acetaldehyde are the phase I biotransformation enzymes cytochrome P450 2E1 (CYP2E1), alcohol dehydrogenase (ADH) and acetaldehyde dehydrogenase (ALDH). The activity of these enzymes influences AA levels.4 The primarily hepatic CYP2E1 isozyme is induced by ethanol and is also expressed in the esophagus.5 Of the ADH enzymes classes 1, 3, 4 and 7, which are expressed in the esophageal epithelium,6, 7 ADH class 1 is the most relevant as subclasses 3 and 4 have a low affinity for ethanol.4 Within the ALDH superfamily, the ALDH2 isozyme is most relevant, as it is involved in the oxidation of AA.4 ALDH2 primarily is a liver enzyme, although it is also expressed in esophageal mucosa where it is involved in the first pass metabolism of alcohol.8

Functional polymorphisms in the genes of alcohol-metabolizing enzymes may cause alterations in the phenotypes. These polymorphisms can therefore act as a proxy for the exposure to environmental toxins such as alcohol and its main metabolite acetaldehyde.

Previously, a large Asian meta-analysis showed single-nucleotide polymorphisms (SNPs) in ADH genes to be associated with an increased risk of cancer of the upper-aerodigestive tract (UADT).9 Furthermore, a Japanese genome-wide association study (GWAS) reported ADH1B and ALDH2 polymorphisms to be associated with an increased ESCC risk.10 Many case–control studies originate from the Eastern continent, so the correlation between polymorphisms in alcohol-metabolizing enzymes and modulation of ESCC risk is well established in Asian countries.11, 12 In sharp contrast, at present only an American and a French case–control study examined the roles of ADH1C and CYP2E1 genotypes in EAC13 and UADT cancer risk,14 respectively. However, the number of patients included in these studies were small; n=13713 and n=15814 for combined EC and UADT cases. So there is a literary gap concerning the role of functional polymorphisms in alcohol-metabolizing enzymes in Western EAC as well as ESCC cases. Consequently, we hypothesized that variant genotypes of CYP2E1, ADH1B, ADH1C, ADH7 and ALDH2 might alter EAC and ESCC risk in a Dutch Caucasian population. To prove our hypothesis we conducted a case–control study, including 351 patients with EAC or ESCC of Dutch Caucasian descent and 430 age-, sex- and race-matched controls.

Materials and methods

Patients and controls

The study was approved by the Medical Ethical Review Committee, region Arnhem-Nijmegen (CMO approval no. 2002/114) and written informed consent was received from all participants. Blood or tissue samples from 351 Caucasian patients with EC were collected in the period October 2002 to June 2012 at four different hospitals, all located within 30 km distance in the South-East area of the Netherlands.15 Patients were included in the order of entry to the hospital. Only patients with a diagnosis of esophageal carcinoma, as confirmed by a pathologist, were included in the study. As a source of DNA, in 92 cases, tissue biopsies of normal esophagus or stomach from patients were collected after surgery, whereas in 259 cases EDTA blood was collected. Blood and tissue samples were frozen at −20 and −80 °C, respectively. DNA isolation was performed by usage of the High Pure PCR Template Preparation Kit (Roche, Mannheim, Germany) according to the instructions of the manufacturer. Post-extraction DNA was stored at 4 °C. Healthy controls (n=430), matched for age, sex and Caucasian race were recruited from the same geographical area of The Netherlands, after advertisement in local papers as described earlier.15

Genotyping

The extracted DNA was used for SNP detection of ADH1B c.143A>G (rs1229984; Arg48His), ADH1B c.1108C>T (rs2066702; Arg370Cys), ADH1C c.815G>A (rs1693482; Arg272Gln), ADH1C c.1048A>G (rs698; Ile350Val), ADH7 c.275G>C (rs1573496; Gly92Ala), CYP2E1 c.-1295G>C (rs3813867), CYP2E1 c.-1055 C>T (rs2031920) and near ALDH (rs4767364) by means of real-time PCR (rtPCR) techniques. A specific set of primers, which flank the region of the SNP, was used to amplify the DNA. In the TaqMan assay, two probes, labeled with a fluorophore at the 5′-end and a quencher at the 3′-end of the probe, were added to the PCR mixture. The PCR primers and TaqMan probes were designed by Beacon Designer (PREMIER Biosoft International, Palo Alto, CA, USA) and synthesized by Isogen Life Science (De Meern, the Netherlands). The sequences of the primers and probes, as well as the annealing temperatures and MgCl2 concentrations, are given in Table 1. The gene expression was measured using the CFX96 Real-Time PCR Detection System (Bio-Rad Laboratories, Hercules, CA, USA).

Table 1 Sequences of primers and probes and annealing temperatures for the investigated SNPs in the ADH, CYP2E1 and (near) ALDH genes
Full size table

ADLH2 mutation analysis

The ALDH2 gene consists of 13 exons. A specific set of primers, flanking the coding region, was designed for each exon, by using Primer3 software (Whitehead Institute for Biomedical Research, Cambridge, UK). A PCR followed by melting curve analysis was performed for each exon using the Bio-Rad CFX96 Real-Time PCR Detection System. The PCR also contained EvaGreen (Biotium, Hayward, CA, USA), a dye staining double-stranded DNA. The annealing temperature varied between the exons. The melt curves were analyzed using Bio-Rad Precision Melt Analysis software (Bio-Rad Laboratories). The PCR products with an abnormal melt curve and some normal PCR products were isolated after electrophoresis with the Qiaex II Gel Extraction Kit (Qiagen, Venlo, the Netherlands) and were sequenced. The outcome of the DNA sequencing was analyzed using ChromasPro, Technelysium.

Statistical analyses

The independent sample t-test was applied for the differences in continuos variables between characteristics of patients and controls. The chi-square test was used for analyzing nominal variables of patient characteristics and to test for differences of frequencies in predicted enzyme activity genotypes between two groups. Except for ADH1C, the most common genotype was set as reference. Odds ratios (OR) with 95% confidence interval (95% CI) were calculated. Stratified analyses were performed according to tumor histology. All P-values were two-sided and a probability level of P<0.05 was considered to be significant. All analyses were performed with the software SPSS for Windows, version 16.0 (SPSS Inc., Chicago, IL, USA).

Haplotypes were generated using the PLEM program.16 The most common haplotype was taken as reference in the comparison between cases and controls. Only participants with complete genotypes were included in the haplotype analyses. Haploview was used to evaluate the degree of linkage between the ADH polymorphisms.17

Results

Table 2 displays the characteristics of patients and controls. A total of 351 patients with EC, 260 with EAC and 85 with ESCC and 430 healthy controls were included. For six patients, the exact tumor type was not mentioned in the pathology reports. Controls were matched against the whole EC group for ethnicity, age and sex. The female sex was more present in patients with ESCC in comparison to patients with EAC (P<0.001), and in comparison to controls (P=0.003). Otherwise, no significant difference was found between the reported demographics of patients with EAC or ESSC, and controls.

Table 2 Characteristics of patients with esophageal cancer and controls
Full size table

Polymorphisms in the following alcohol-metabolizing enzymes were chosen on the basis of functionality and relevance as reported in earlier EC studies. As ADH1B and ADH1C are the most important ADH enzymes involved in the oxidation of alcohol in the esophageal epithelium, the most common functional polymorphisms ADH1B p.Arg48His, ADH1B p.Arg370Cys, ADH1C p.Arg272Gln and ADH1C p.Ile350Val were selected for genotyping. The ADH1C*1 allele was defined as the 272Arg 350Ile variant, whereas the ADH1C*2 allele is the 272Gln 350Val arrangement.18 ADH7 is also expressed in the esophagus and has a vital role in the first pass metabolism of ethanol,7 and the allelic variant encoded by the ADH7 p.Gly92Ala polymorphism was selected for genotyping. The CYP2E1 c.-1295G>C (rs3813867) and c.-1055 C>T (rs2031920) genetic polymorphisms are known to increase the enzyme activity19 and in vitro expression.20 Both polymorphisms are highly linked, leading to either the most common c1, or the variant c2 allele.20 Consequently, genotypes can be classified as c1/c1, c1/c2 and c2/c2.

The functional polymorphism ALDH2 p.Glu487Lys leads to the inactive allelic variant ALDH2*2 (487Lys), which is incapable of oxidizing acetaldehyde.4 However, this ALDH2*2 variant, although common in Asia, is rare in Caucasians.21 Therefore, we also performed mutation analyses of the coding sequences of the ALDH2 gene to investigate whether other unknown polymorphism do exist in our Dutch Caucasian population.

Recently, a GWAS reported the 12q24 variant rs4767364, near the ALDH gene, being directly correlated to fluctuations in breath and blood concentrations of alcohol and it significantly increased ESSC risk.22 Consequently, we involved this SNP in our current study.

All polymorphisms were distributed according to the Hardy–Weinberg equilibrium. The genotype frequencies are depicted in Table 3. The distribution of the CYP2E1, ADH1B, ADH1C, ADH7 or near ALDH rs4767364 genotypes were not significantly different between patients with EAC or ESSC and controls. The ADH1B*3 allele and the ADH1B*2*2 genotype did not occur in our study population. Furthermore, homozygosity for ADH7 92Ala was not found in patients with ESCC.

Table 3 Genotype distribution in alcohol-metabolizing enzymes and corresponding odds ratios and confidence intervals for patients with ESCC or EAC, compared with controls
Full size table

ADLH2 mutation analysis revealed no polymorphisms in the exons and splice sites of the ALDH2 gene in our study population. The ADH haplotype with the variant ADH7 92Ala showed a tendency to decrease EAC risk (OR 0.65; 95% CI 0.42–1.02). No haplotype significantly modified EAC or ESCC risk (Table 4).

Table 4 ADH haplotypes with corresponding odds ratios and confidence intervals for patients with ESCC or EAC compared with controls
Full size table

Table 5 shows the results of earlier published case–control studies on alcohol-metabolizing enzymes associated with EC risk. Most studies originated from Asia, and thus ESCC was the predominant histological subtype. Four studies included patients without histological specification, however, all studies were from Asia, making ESCC the most probable subtype. Only Terry et al.13 reported on EAC risk.

Table 5 Results of this study in comparison to earlier case–control studies on polymorphisms in alcohol-metabolizing enzymes and esophageal cancer risk
Full size table

Figure 1 shows the linkage disequilibrium (LD) patterns of ADH1B p.Arg48His, ADH1C p.Arg272Gln, ADH1C p.Ile350Val and ADH7 p.Gly92Ala. A nearly total linkage disequilibrium (LD) exists between both ADH1C polymorphisms and a complete LD was found between both the ADH1C and the ADH7 polymorphisms. The ADH1B p.Arg48His had low LD values with the three other ADH polymorphisms.

Figure 1
figure 1

Linkage disequilibrium (LD) patterns between the ADH1B Arg48His (rs1229984), ADH1C Ile350Val (rs698), ADH1C Arg272Gln (rs1693482) and ADH7 Gly92Ala (rs1573496) loci. LD patterns of the single-nucleotide polymorphisms (SNPs) along the 4q22 region encompassing the ADH gene, generated by Haploview. The LD between any two SNPs is shown in the cross cell. A darker color and higher number within the cell indicates a higher degree of LD. The two dark cells without numbers indicate a complete LD.

Full size image

Discussion

This study aimed to clarify the role of altered activity genotypes of alcohol-metabolizing enzymes in EC susceptibility in Caucasians. We did not detect significant associations between variant genotypes of CYP2E1, ADH1B and ADH1C, ADH7 or near ALDH (rs4767364) and EAC or ESSC risk in a Dutch population. Moreover, mutation analysis revealed no polymorphisms in the ALDH2 gene in our study population. Our results are in contrast to many reports on EC from Asian countries, where polymorphisms in alcohol-metabolizing enzymes have a modulating role in cancer risk (Table 5). Strong associations in European populations between the ADH1B, ADH1C and ADH7 polymorphisms and the risk of UADT squamous cell carcinomas have also been appreciated by a large European GWAS and the subgroup analyses demonstrated the associations were consistent regarding the anatomical site of esophagus.22, 23 However, a major limitation of these studies is the lack of stratified analyses according to histology, as the GWAS solely included SCC type esophageal tumors. At present, one case–control study investigated the association between ADH and EAC.13 However, their Western study population was small (n=114), patients with gastric cardia carcinomas were also included and only associations with ADH1C were examined. So a large case–control study emphasizing EAC risk was still lacking.

We found the ADH1B*3 allele to be absent in our Caucasian population, which is in accordance to the literature, as it is reported to be an African-specific allele.24 The distribution of the ADH1B*1*2 genotype is in agreement with the frequency reported in Europeans.25 Although we found no significant association between the ADH1B genotypes and EC etiology in our European population, the ADH1B*2 allele was found to be protective for ESCC in most Asian studies, except in those of Yang et al.26 and Ding et al.27 This association is most likely due to alcohol avoidance,28, 29, 30 as an increased flushing response has been associated with the ADH1B*2 allele.31 This flushing response is due to the combination of ADH1B*2 and ALDH2*2 alleles. The latter allele is predominant in Asians32 and correlates with increased acetaldehyde blood levels after alcohol consumption.33 The ALDH2*2 allele is not present in our Caucasian study population, we analyzed 781 Dutch Caucasians, both patients and controls, and this may also explain the absence of an ADH1B*2 association with EC in our study. In Asian populations, the ADH1B gene is known to act synergistically with ALDH2.34 Furthermore, nearly all studies described a significantly increased ESCC risk of heterozygous ALDH2 genotypes. The absence of a role for ALDH2 in EC etiology in our Dutch population may also be explained by the extremely high prevalence of the ADH1B*1 allele, associated with a low conversion rate of ethanol to acetaldehyde.25

Furthermore, the combined variant genotypes of the ADH1B and ADH1C, when set off against the wild-type genotypes, had no synergistic effects in increasing EAC risk. Terry et al.13 reported that the ADH1C*1*2 and ADH1C*2*2 genotypes had a tendency to decrease ESCC and EAC risk. However, the ADH1B gene was not analyzed, as ADH1B variants are in linkage disequilibrium with ADH1C polymorphisms.11 Yokoyama et al.25 reported the ADH1C*2 low activity allele to increase ESCC risk and explained this association by the high degree of LD with the ADH1B*1 allele, which was confirmed by Hashibe et al.35 and Li et al.36 Our findings suggest that ADH1C and ADH1B are not in LD, and no significant association between EAC or ESCC risk and ADH1C polymorphisms exists. We also found no association with EAC or ESSC susceptibility and CYP2E1 genotypes in Caucasians. A Western study and a large Asian meta-analysis failed to confirm such an association in Brazilian and Chinese ESCC patients as well.37, 38

All in all it is likely that variant genotypes of the alcohol-metabolizing enzymes ADH1B, ADH1C, ALDH2 and CYP2E1 have a role in ESCC in Asians but not in EAC or ESSC in Caucasians. Explanations may be a different ancestry gene pool, interracial differences in enzyme activity or the difference in dietary habits.

ADH7 92Ala was found to be a protective allele in upper-aerodigestive cancer risk23 and also for ESCC risk.22 We now report the same effect for EAC risk, although our association did not reach statistical significance, probably due to insufficient patient numbers (OR 0.73; 95% CI 0.47–1.11). Interestingly, the haplotype with the variant ADH7 92Ala also showed the same tendency to decrease EAC risk (OR 0.65; 95% CI 0.42–1.02).

Although altered activity genotypes can act as a proxy for the relative risk of ingested carcinogens, that is, alcohol, a limitation of our study is the absence of data on alcohol usage. This is mainly due to ethical reasons, restricting the collection of data on tobacco and alcohol use in patients newly diagnosed with a lethal disease. However, in an earlier study by us on Dutch Caucasian patients with head and neck cancer, having largely similar risk factors, it was established that at least 1–4 alcoholic beverages per day were consumed in 88% of the patients with head and neck cancer and 83% of the controls.39 This indicates that, despite of the missing data, by far most of our EC patients may have consumed alcohol and only a few may have not.

Another limitation is the small group of ESCC patients and the subsequent power of our results on ESCC. This is caused by stratifying according to histology, but was necessary, as most case–control studies concerning polymorphisms in alcohol-metabolizing enzymes originate from Asia and results on adenocarcinoma risk are scarce.

Although acetaldehyde induces development of adenocarcinomas in animal models,10 in humans the role of ethanol in the genesis of EAC is still under debate, as in contrast to earlier reports,40 recent studies suggest that alcohol consumption does not have a role in EAC etiology.41, 42 This may largely explain why no effects of functional polymorphisms in alcohol-metabolizing enzymes were found in modulating EAC risk.

In conclusion, this study suggests that polymorphisms in alcohol-metabolizing enzymes, having a role in ESCC etiology, are not associated with EC susceptibility in an European population.