Association between aldehyde dehydrogenase gene polymorphisms and the phenomenon of field cancerization in patients with head and neck cancer

1Division of Digestive Endoscopy and Gastrointestinal Oncology, National Cancer Center Hospital East, Kashiwanoha 6-5-1, Kashiwa 277-8577, 2Investigative Treatment Division, National Cancer Center Research Institute East, Kashiwa 277-8577, 3Kaken Corp., Horimachi 1044, Mito 310-0903, 4Epidemiology and Biostatistics Division, National Cancer Center Research Institute East, Kashiwa 277-8577, 5Division of Head and Neck Surgery, National Cancer Center Hospital East, Kashiwa 277-8577, Japan


Introduction
Patients with HNSCC, especially in the oral cavity, oropharynx and hypopharynx, often develop second primary ESCC (1)(2)(3)(4)(5).In such patients, widespread epithelial oncogenic alterations, such as several grades of squamous epithelial dysplasia, also frequently occur in the upper aerodigestive tract including the esophagus (6,7).They are clinically detected as multiple Lugolvoiding lesions (multiple LVL) by Lugol chromoendoscopy (8), since Lugol-dye does not stain such lesions (9).These findings have been explained by the concept of 'field cancerization' proposed by Slaughter et al. (10), which is based on the hypothesis that repeated exposure to carcinogens contributes to the development of lesions.Although alcohol and cigarette smoke have been linked to this phenomenon (11,12), patients with HNSCC do not always develop second ESCC even if they consume alcohol and cigarettes.Identification of the etiology underlying the phenomenon of field cancerization might help to explain why some patients with HNSCC have a high likelihood of developing a second ESCC, and this will have a great impact on designing effective prevention, early diagnosis and therapeutic strategies.
There has been increasing evidence that acetaldehyde, the first metabolite of ethanol, rather than alcohol itself, may be responsible for the risk of developing alcohol-related cancers.Studies have suggested direct mutagenic and carcinogenic effects of acetaldehyde in vitro and in experimental animal models (13)(14)(15)(16)(17). Acetaldehyde is formed by the oxidation of ethanol by alcohol dehydrogenase (ADH), and is eliminated by aldehyde dehydrogenase (ALDH) (18).In humans, the ADH3 and ALDH2 genes carry two alleles (ADH3-1 and ADH3-2, ALDH2-1 and ALDH2-2 alleles, respectively) with ethnic differences in the polymorphisms and different kinetic properties (18)(19)(20)(21)(22).The ADH3-1 allele is present in 40-50% of Caucasians, while it is more predominant (~95%) in Japanese (23).In contrast, the ALDH2-2 allele is found in only orientals at a frequency of 50% (22).The ADH3-1 allele, coding for the rapidly acting ADH3, and the ALDH2-2 allele coding for inactive ALDH2, are closely associated with alcoholrelated cancers in the upper aerodigestive tract (24)(25)(26)(27)(28)(29).As blood acetaldehyde concentrations after drinking in individuals carrying the ADH3-1 allele or the ALDH2-2 allele are reported to be significantly higher than in those lacking these alleles, an increase in acetaldehyde concentration appears to play a critical role in alcohol-related carcinogenesis (18,30).However, this raises an important question as to whether patients harboring the ADH3-1 or ALDH2-2 alleles can drink heavily: the accumulation of acetaldehyde plays a protective role against excessive alcohol consumption as it causes unpleasant reactions such as headache, nausea, vomiting, tachycardia, facial flushing, hypotension, and sleepiness (31)(32)(33).To date, there have been no studies with biological measurements of acetaldehyde in relation to the risk of alcohol-related cancer to provide direct evidence that the accumulation of acetaldehyde is carcinogenic to human beings.
In this study, we found that a gene-environmental interaction between the ALDH2-2 allele and alcohol drinking is the most significant contributing factor for the development of multiple LVL, resulting in the development of second ESCC in patients with HNSCC.Ultimately, an increased acetaldehyde level is thus a critical determinant of the phenomenon of 'field cancerization'.

Patients
This study was approved by the Institutional Review Board of the National Cancer Center, Tokyo.We excluded patients who had cancer at a site other than the oral cavity, oropharynx or hypopharynx or with histological findings other than squamous cell carcinoma.Seventy-eight consecutive Japanese patients with newly diagnosed HNSCC were entered into the study prospectively from September 1999 to July 2001.After obtaining written informed consent, demographic data were collected from the hospital charts and by interviews with patients.
The history of usage of tobacco and alcohol was carefully documented.The same questionnaire, which aimed to obtain detailed drinking and smoking habits, was used for the interview.Taking into account the different ethanol concentrations, alcohol consumption was estimated as the average amount of ingested alcohol for every drinking day in grams of pure alcohol per day.Indicators of tobacco use included the average number of cigarettes smoked per day, and the duration of smoking in years.The smoking index was calculated as the number of cigarettes smoked per day multiplied by the number of total smoking years.Drinkers and smokers were defined as patients who drank and smoked every day, respectively.Smokers who had ceased smoking for more than 5 years before being treated for HNSCC were categorized as ex-smokers, as in a previous report (34).Non-smokers and non-drinkers were defined as patients who never used or who only rarely used tobacco and alcohol, respectively.

Lugol chromoendoscopy
Lugol chromoendoscopy was performed using Lugol dye staining methods as described previously (8,35,36), using an electronic endoscope (Q230 or Q200, Olympus Optical, Tokyo, Japan).Lesions were defined as multiple LVL when numerous well-defined, irregular-shaped Lugol-voiding lesions were observed endoscopically throughout the entire esophageal mucosa after application of the Lugol dye solution (Figure 1).A single endoscopist (MM) evaluated the incidence of multiple LVL.Concomitant second ESCC was confirmed histologically using biopsy specimens.

Molecular analysis
To determine genetic polymorphism of the ADH3 and ALDH2 genes, genomic DNA was extracted from peripheral lymphocytes using a DNA isolation kit (Roche Diagnostics, Switzerland).Genetic polymorphism was analyzed by the polymerase chain reaction-restriction fragment length polymorphism (PCR-RFLP) method as described previously (8).For ADH3 genotyping, genomic DNA was initially digested with NlaIII (TOYOBO, Osaka, Japan) to prevent amplification of closely related ADH1 and ADH2 genes.The PCR products for ADH3 and ALDH2 genotyping were digested with SspI (TOYOBO, Osaka, Japan) and MboII (TOYOBO, Osaka, Japan) respectively, and separated on a 10% polyacrylamide gel.We divided the genotypes of ADH3 and ALDH2 into two groups: those homozygous for the ADH3-1 allele and those with the ADH3-2 allele; and those homozygous for the ALDH2-1 allele and with the ALDH2-2 allele, respectively.

Measurement of acetaldehyde levels
Previous reports have shown that acetaldehyde circulating in the blood can be expired (37)(38)(39).If HNSCC patients expire high levels of acetaldehyde, direct exposure to acetaldehyde could play critical and specific roles in carcinogenesis in the mucosa of the upper aerodigestive tract.To assess the acetaldehyde levels in the breath, we asked participants to drink 200 ml of 6% ethanol in grapefruit juice over 10 min after at least 2 h of fasting.Among 78 patients enrolled in this study, 26 patients could not participate in this 'acetaldehyde breath test', because of severe dysphasia.Thus, the study population for this acetaldehyde breath test consisted of 52 patients.Breath samples were collected before drinking the alcohol, and 0, 10 and 30 min later.The test period and amount of alcohol required for ingestion to adequately measure acetaldehyde were previously assessed (unpublished data).Exhaled breath samples were obtained in collecting bags with a capacity of approximately 250 ml, which are used routinely for urea breath tests for Helicobacter pylori infection.Acetaldehyde levels in breath aliquots (1 ml each) were determined immediately by gas chromatography.The conditions for analysis were as follows: Porapack Q60/80 glass column (GL Science, Tokyo) 3 mm ϫ 2 m; detector, flame ionization detector (FID); injection temperature, 200°C; column temperature, 150°C; detector temperature, 200°C; carrier gas (He) flow rate, 40 ml/min.Reproducibility of the measurement of acetaldehyde in the breath was determined by repetition of this test twice in six healthy volunteers: two subjects homozygous for ALDH2-1 and four subjects heterozygous for ALDH2-1/2-2.The investigator (MN) analyzing the genetic polymorphism was blinded to the results of the acetaldehyde breath test and endoscopic findings.

Statistical analysis
All statistical analyses were performed using the Stat View software package for Macintosh (Version 5; Abacus Concepts, Berkeley, CA).Quantitative data were compared using Student's t-test.For categorical variables, we used Fisher's exact test.A P-value of Ͻ0.05 was considered statistically significant.Logistic regression was used for the statistical comparisons of discrete variables between the patients with and without multiple LVL.Multiple logistic regression analysis was used to examine the multivariate-adjusted odds ratio for the development of multiple LVL in patients with HNSCC.

Study population
Table I shows the patients' characteristics.The patients included 66 males and 12 females, with mean Ϯ SD ages of 61 Ϯ 8.4 years.The most common primary sites of tumors were the oral cavity and hypopharynx, and 68 of the 78 patients were drinkers (78%).All of the drinkers had a longterm history of alcohol drinking for more than 20 years.Fiftythree (68%) of the patients currently smoked cigarettes and all but one had a smoking-index history of Ͼ400.Seventeen (22%) of the patients were ex-smokers and eight (10%) were non-smokers.patients, respectively (Table I).Patients with multiple LVL had a higher prevalence of second ESCC compared with patients without multiple LVL (72% vs. 0%, P Ͻ 0.0001).

Patient characteristics correlated with multiple LVL
To identify any patient characteristics correlated with the development of multiple LVL, we first compared gender, age, site of the primary tumor, alcohol consumption and smoking habits (Table I).Age and gender were identical between the patients with and without multiple LVL.For primary tumors, the hypo-and oropharynx were the most prevalent sites in patients with multiple LVL.All of the patients with multiple LVL were smokers or ex-smokers, and a smoking index Ͼ1000 was closely associated with multiple LVL.As for drinking, all of the patients with multiple LVL were drinkers.Although weak but significant associations were observed in the amount of daily alcohol intake and duration of drinking habits between the patients with and without multiple LVL, these differences seemed to depend largely on the inclusion of non-drinkers.To examine the actual effects of drinking, we compared these factors among only drinkers, and no differences were seen.
To investigate the gene-environmental interaction with alcohol drinking for the risk of the development of multiple 1761 LVL, we compared the prevalence of the ADH3 and ALDH2 gene polymorphisms.The ALDH2-2 allele was significantly more prevalent in the patients with multiple LVL compared with those without multiple LVL (79% vs. 20%: P Ͻ 0.0001).In contrast, no differences were observed for ADH3 gene polymorphism.
Clinical and genetic variables correlated with the development of multiple LVL were determined using a univariate logistic regression analysis (Table II).ALDH2 genotype (ALDH2-2 allele present) had the strongest statistical association with multiple LVL (odds ratio 15.0, 95% CI 4.80-46.55:P Ͻ 0.001).The primary site of the tumor and the smoking index were also significantly associated with multiple LVL (odds ratio 9.04 and 3.58; 95 % CI 2.92-27.94and 1.36-9.39:P Ͻ 0.001 and Ͻ0.008, respectively).Although the ADH3-2 allele not the ADH3-1 allele showed a weak association with multiple LVL, the difference was not statistically significant (odds ratio 2.80; 95% CI 0.80-9.84:P ϭ 0.11).Patient age, gender and daily alcohol consumption were not significantly associated with the development of multiple LVL.
ALDH2 genotype, smoking index and site of primary tumor were then examined in a multivariate logistic regression model to define independent variables associated with the   development of multiple LVL in HNSCC patients (Table III).
In the multivariate analysis, ALDH2 genotype (ALDH2-2 allele present) was associated with the greatest overall risk for the development of multiple LVL (odds ratio 17.6; 95% CI 4.72-65.29:P Ͻ 0.0001).Smoking index (ജ1000) and site of the primary tumor were also associated with the risk of the development of multiple LVL, with odds ratios of 6.4 (95% CI 1.65-25.15:P Ͻ 0.005) and 3.8 (95% CI, 0.99-14.57:P Ͻ 0.05), respectively.

Assessing acetaldehyde levels in the breath
These results strongly suggest that the interaction between drinking and the ALDH2-2 allele is responsible for the development of multiple LVL in patients with HNSCC.To address these issues, we measured acetaldehyde concentrations in the breath of 52 patients (Figure 2).Thirty-three patients were homozygous for ALDH2-1 and the remaining 19 patients were heterozygotes.No acetaldehyde was detected in breath samples taken before drinking in any patients.Immediately after finishing drinking, the acetaldehyde levels in the breath were elevated to significantly higher levels in those patients who had the ALDH2-2 allele than in those homozygous for ALDH2-1 (P Ͻ 0.001).At 10 and 30 min after drinking, this high level of acetaldehyde persisted in all patients with the ALDH2-2 allele, even though it had already decreased to undetectable levels in those homozygous for the ALDH2-1 allele (P Ͻ 0.001).Thus, patients with the ALDH2-2 allele expired significantly higher concentrations of acetaldehyde after drinking.Furthermore, ADH3 genotype modified the concentration of acetaldehyde in the breath.When the ADH3-2 allele was combined with the ALDH2-2 allele, the breath acetaldehyde concentration immediately after finishing drinking was significantly higher than in those without this allele among the ALDH2-2 allele carriers (Figure 3).This high acetaldehyde concentration that was modified by the ADH3-2 allele persisted for at least 30 min after finishing drinking (Figure 3).It was also noted that acetaldehyde levels in the breath in patients with the ALDH2-2 allele showed significant individual differences.These differences had no direct relationship with body surface area, age, sex, or the drinking or smoking habits of each individual (data not shown).

Discussion
The mechanism of frequent occurrence of second ESCC in patients with HNSCC has been an important unresolved question, because the development of secondary tumors adversely affects the survival of those patients with HNSCC (40)(41)(42).Although there have been numerous clinical reports regarding this phenomenon (1-5), the precise etiology remains unclear.The present data demonstrate a clear association between multiple LVL and second ESCC in patients with HNSCC.This finding suggests that episodes of multiple LVL are precursors of second ESCC in such patients.Therefore, the investigation on how multiple LVL develop could provide essential information about the etiology of the phenomenon of field cancerization.
We found a strong association between the development of multiple LVL and the ALDH2-2 allele in patients with HNSCC.In addition, estimation of the gene-environmental interaction with alcohol drinking demonstrated that only the drinkers developed multiple LVL.A previous case-control study also showed a strong gene and environmental interaction between the ALDH2-2 allele and heavy drinking for the risk of developing esophageal cancer (43).These results suggest that increased acetaldehyde produced in an interaction between the ALDH2-2 allele and drinking is responsible for the development of multiple LVL.Indeed, significantly higher acetaldehyde levels in the breath were detected in all patients with the ALDH2-2 allele compared with those without this allele.The results mean that there is no compensatory mechanism for eliminating acetaldehyde even after long-term alcohol consumption in those patients with the ALDH2-2 allele.Although this finding supports the hypothesis that acetaldehyde is associated with alcohol-related carcinogenesis, it is unknown why the patients harboring the ALDH2-2 allele are heavy drinkers despite the unpleasant reaction to acetaldehyde.To clarify this issue, further investigation is needed.
As inhalation or drinking of acetaldehyde produces nasal and laryngeal carcinomas in experimental animal models (13,16,17), our results imply that direct exposure of the mucosa via expiratory breath to high levels of acetaldehyde may be a critical event in the development of multiple LVL.
These observations can also explain the mechanism of the phenomenon of field cancerization, providing strong evidence for the carcinogenicity of acetaldehyde.
We also observed a weak association between ADH3 gene polymorphism and the development of multiple LVL.Although Coutelle et al. (24) and Harty et al. (25) found a significant association between the ADH3-1 allele and cancers of the upper aerodigestive tract, this association has been controversial because other investigators did not find any such association (44,45).In the present study, the ADH3-2 allele not the ADH3-1 allele was weakly associated with the development of multiple LVL.In addition, harboring this allele actually increased the breath acetaldehyde concentration compared with those lacking this allele even in the ALDH2-2 allele carries.Although the reason was unknown, one possibility was that the combined effects on the alcohol metabolisms by the slow metabolizing ADH3-2 allele and the inactive ALDH2-2 allele might influence the breath acetaldehyde concentration.It would be interesting to investigate the combined influence of ADH3 and ALDH2 genotypes (or other genes) on the phenomenon of field cancerization.However, because of the small sample size of the study and the low frequency (about 5%) of the ADH3-2 allele in Japanese people (22), our results should be interpreted with care until confirmed by larger studies.
Even in those patients homozygous for the ALDH2-1 allele, six (13%) had multiple LVL in our study.As all of these were heavy drinkers and smokers, alcohol and smoking were at least associated with the development of multiple LVL.Although we cannot adequately explain why they developed multiple LVL, it will be interesting to investigate this issue, because the phenomenon of field cancerization in the patients with HNSCC is also seen in Caucasians who do not have any polymorphism in the ALDH2 gene.
We found that a cigarette smoking index of Ͼ1000 is also an independent risk factor for developing multiple LVL.Because the carcinogenic effect of smoking has been established conclusively (46), the result indicates that smoking modifies the risk of the development of multiple LVL.In addition, as cigarette smoke contains significant amounts of acetaldehyde together with other various carcinogens (47,48), the combination of heavy smoking with drinking may partly increase regional acetaldehyde levels in the respiratory tract.
It should be noted that this study focused on those patients with HNSCC, because the phenomenon of field cancerization frequently occurs among them, but not in the general population.To date, it is unknown why this phenomenon occurs.In this limited study, we found that those with a tumor in the oro-or hypopharynx showed a significant higher odds ratio of developing multiple LVL compared with those with a tumor at other sites.If susceptibility to HNSCC is indeed associated with the phenomenon of field cancerization, further investigation is needed to understand the mechanism more clearly.
In summary, the present study reveals strong evidence for genetic and environmental interactions between the ALDH2-2 allele, the ADH3-2 allele and alcohol drinking, for the risk of developing widespread oncogenic epithelial alterations, resulting in the development of secondary ESCC in patients with HNSCC.Ultimately, increased local acetaldehyde exposure is a critical determinant of the phenomenon of field cancerization.This result may provide a new intervention strategy for the prevention of multiple cancers in the upper aerodigestive tract in high-risk populations.

Fig. 1 .
Fig. 1.Multiple LVL.Numerous well-defined, irregular-shaped Lugolvoiding lesions were endoscopically observed throughout the entire esophageal mucosa after application of the Lugol dye solution.

Fig. 2 .
Fig. 2. Acetaldehyde breath test.Acetaldehyde levels in expired breath are measured by gas chromatography.Each point represents one individual who was sampled several times: filled circles, subjects with the ALDH2-2 allele; open circles subjects lacking the ALDH2-2 allele.

Table I .
Comparison of the characteristics of patients with HNSCC acording to presence or absence of multiple LVL a Value are means Ϯ SD.bThe P value was calculated by Student's t-test.cTheP value was calculated by Fisher's exact test.d When ex-smoker was analysed with smoker, the P value was 0.03.

Table II .
Clinical and genetic variables associated with multiple LVL in patients with HNSCC (univariate logistic regression) a CI denotes confidence interval.Table III.Predictors of multiple LVL in patients with HNSCC by multivariate logistic regressionOdds ratio (95% CI) a P value Downloaded from https://academic.oup.com/carcin/article-abstract/23/10/1759/2896664 by guest on 22 July 2018