Risk of Squamous Cell Carcinoma of the Upper Aerodigestive Tract in Cancer-Free Alcoholic Japanese Men: An Endoscopic Follow-up Study

Epidemiologic studies have consistently shown a strong relation between the development of squamous cell carcinoma (SCC) in the upper aerodigestive tract (UADT) and both alcohol drinking and tobacco smoking (1). By systematically screening a large population of Japanese male alcoholics by upper gastrointestinal endoscopy combined with oropharyngolaryngeal inspection and esophageal iodine staining, we found a high prevalence of SCC in the oropharyngolarynx and distinct iodine-unstained lesions (DIUL) in the esophagus (mainly consisting of dysplasia and SCC; ref. 2).

We subsequently showed that heterozygosity for the inactive form of aldehyde dehydrogenase-2 (ALDH2), encoded by the gene ALDH2*1/*2, and the less-active homozygous form of alcohol dehydrogenase-1B (ADH1B, previously called ADH2), encoded by the gene ADH1B*1/*1, are strong risk factors for SCC at these sites in the alcoholic population (3, 4). Most case-control studies have reported that ALDH2*1/*2 (5, 6) and ADH1B*1/*1 (5-8) are linked to an increased risk of esophageal SCC. Macrocytosis with a mean corpuscular volume (MCV) ≥ 106 is induced by alcoholism, high acetaldehyde exposure, heavy smoking, and poor nutrition. It is a marker for high risk of esophageal SCC in alcoholics, and the combination of ALDH2*1/*2, ADH1B*1/*1, and MCV ≥ 106 increases the risk of esophageal SCC in a multiplicative fashion (9). ALDH2 is the key enzyme responsible for the elimination of genotoxic acetaldehyde, an established animal carcinogen (10) produced during ethanol metabolism (11). After consuming alcohol, persons with inactive ALDH2 exhibit facial flushing as a result of acetaldehydemia, and the results of a simple questionnaire asking about current and past flushing in response to ingestion of a small amount of alcohol have shown that current or former flushing is a highly reliable marker of inactive ALDH2 and predictor of risk of esophageal SCC as a marker of inactive ALDH2 in a general population (12).

We have previously reported that multiple primary intra-esophageal SCC and ALDH2*1/*2 greatly increases the risk of metachronous SCC of the UADT in Japanese alcoholics with esophageal SCC (13, 14), and the presence of multiple non-cancerous DIULs has been reported to be a predictor of metachronous SCC in the UADT in Japanese patients after endoscopic mucosectomy for esophageal SCC (15, 16). Both ALDH2*1/*2 and multiple non-cancerous DIULs, which are intercorrelated (17), increase the risk of multiple cancer in Japanese patients with SCC of the UADT (17, 18). High level of exposure and inefficient degradation of acetaldehyde in the UADT may be associated with multifocal cancerization in the UADT. High-K_m ADH4 (previously called ADH7) is strongly expressed in the UADT and is active in producing acetaldehyde upon exposure to a locally high dose of ethanol, whereas ALDH2 activity in the UADT is extremely weak (19). After ethanol ingestion, acetaldehyde levels in saliva are markedly higher than in blood (20), especially in inactive ALDH2 heterozygotes (21). Normal oral microflora also form acetaldehyde from ethanol and contribute to the acetaldehyde levels in saliva (20).

Detection of SCC of the UADT is not uncommon during follow-up examinations of alcoholic men diagnosed as cancer-free during the initial endoscopic screening, but little is known about whether or to what extent the detection of esophageal dysplasia at the time of the initial screening, ALDH2*1/*2, ADH1B*1/*1, macrocytosis, current/former alcohol flushing, smoking, and nutritional status increase the risk of future development of SCC of the UADT in cancer-free alcoholic men. Hence, there are no established guidelines for follow-up surveillance programs. In the present study, follow-up endoscopic examinations of the UADT in such patients provided us with an opportunity to explore these issues.

The reference population was 3,771 Japanese alcoholic men with no history of cancer of the UADT or stomach who had been diagnosed as cancer-free in the UADT and stomach by an initial screening consisting of upper gastrointestinal endoscopy combined with esophageal iodine staining and oropharyngolaryngeal inspection between 1993 and 2005 at the National Hospital Organization Kurihama Alcoholism Center. A retrospective review of the endoscopy reports revealed that 808 patients had undergone follow-up endoscopic examinations ranging from 1 to 148 months (median, 31 months) after the initial examination, and we evaluated those 808 patients as the subjects of this study. Three major reasons for the follow-up examination were readmission for relapse of alcoholism; detection of non-cancerous lesions, including esophageal dysplasia, during the initial screening; and the health check-ups of outpatients. Mucosal biopsy specimens were taken from DIULs when their greatest diameter was ≥5 mm. Biopsy during the initial screening resulted in a histologic diagnosis of esophageal dysplasia in 112 patients. Table 1 shows the following characteristics of the study subjects at the time of the initial screening: age, ALDH2 genotype, ADH1B genotype, alcohol flushing, MCV, drinking, smoking habit, results of esophageal iodine staining, and body mass index (BMI); they were similar to those of the consecutively diagnosed cancer-free alcoholic men in our previous case-control studies (2, 9). Olympus endoscopes (models Q10, P20, XQ200, XQ230, Q240, and Q240Z, in chronological order of use; Olympus Optical Co. Ltd., Tokyo, Japan) were used to perform the examinations. This study was reviewed and approved by the Ethics Committee of the Center, and informed consent was obtained from the patients who participated.

All of the alcoholics who participated in this study met the DSM-III-R criteria for alcohol dependence (22). Before the initial endoscopic screening examination, information on the patients' drinking profile and smoking was obtained from the patients and, when available, from their significant others. Daily alcohol consumption during the preceding year was expressed in grams of ethanol per day calculated by using a standard conversion for alcoholic beverages. Beer was assumed to be 5% ethanol (v/v); wine, 12%; sake, 16%; shochu, 25%; and whiskey, 40%.

We interviewed all 693 of the 808 subjects since 1995 when we discovered the ALDH2-associated SCC risk, with regard to flushing by asking two simple questions just before the endoscopic screening: (a) Do you always have flush in the face immediately after drinking a glass of beer (yes or no)? And (b) did you always flush in the face immediately after drinking a glass of beer during the first to second year after you started drinking (yes or no)? “Current flushing” was applied to individuals who answered “yes” to question (a), and “former flushing” to those who answered “no” to question (a) and “yes” to question (b). The remaining subjects were recorded as “never flushing.” Current/former flushing individuals were assumed to have inactive ALDH2. The sensitivity of this classification system was found to be 90% in a Japanese general male population, and its specificity was 88% (12).

In 2005, we reviewed the patients' medical charts and adopted the MCV value at the time of each patient's first visit to the Center for the treatment of alcoholism, which was before the initial endoscopic screening, as the MCV value to use in this study. The medical charts of 765 of the 808 patients were stored. MCV was measured by the electrical impedance method with an autoanalyzer (CELL-DYN 3500, Abbott, North Chicago, IL). We dichotomized the patients into an MCV < 106 fl group and MCV ≥ 106 fl group because macrocytosis with an MCV ≥ 106 fl was associated with increased risk for esophageal SCC in our previous case-control study in Japanese alcoholic men (9).

On their first visit to the Center, body height was measured without shoes, and the patients were weighed wearing light clothing. Height and weight measurements were not done in three disabled patients. The patients were classified into quartiles according to their BMI values.

Since 1996, we have genotyped ALDH2 and ADH1B in the alcoholic patients admitted to the Center on Tuesdays and Thursdays, the days designated for admission for treatment of alcoholism, if informed consent was obtained from the patient. PCR-RFLP methods (3, 23, 24) were used for ALDH2 and ADH1B genotyping of lymphocyte DNA samples. We excluded patients in whom genotyping was done after they were diagnosed with SCC in the UADT, and, ultimately, ALDH2 and ADH1B genotypes were available for 556 of the 808 patients. We used the data for these 556 patients to evaluate the relationship between ALDH2 and ADH1B genotypes and the risk of SCC in the UADT.

The detection rate of SCC was estimated by the Kaplan-Meier method, and the difference between groups was analyzed by the log-rank test. The independent effects of selected variables on the SCC detection rate were assessed by using the multivariate Cox proportional hazards model. All analyses were done with the SAS statistical package (version 9.1; SAS Institute, Cary, NC).

Follow-up endoscopy resulted in the diagnosis of SCC of the UADT in 53 patients [esophagus in 33 and oropharyngolarynx in 30 (hypopharynx in 16, mouth in 6, oropharynx in 4, and larynx in 4)]. Esophageal SCC seemed to have developed at the same site as the initial esophageal dysplasia developed in only nine patients. Esophageal SCC was confined to the epithelium in 23 patients but had invaded the proper mucosal layer in three patients, the muscularis mucosae in two patients, the submucosa in three patients, and the proper muscle layer or deeper in two patients. Oropharyngolaryngeal SCC was limited to the epithelium in seven patients but had invaded to the subepithelium in 13 patients and extended to the proper muscle layer or deeper in 10 patients. Only two of the esophageal SCC patients and three of the oropharyngolaryngeal SCC patients had symptoms attributable to the SCC lesions before the follow-up examination. Twenty-eight of the 33 patients with esophageal SCC were treated by endoscopic mucosectomy alone, and 16 of the 30 patients with oropharyngolaryngeal SCC were treated by endoscopic or endoscope-guided mucosectomy alone.

Kaplan-Meier estimates of the proportions of patients with SCC of the UADT as a function of time are shown in Fig. 1. Primary SCC of the UADT developed significantly more frequently among ALDH2*1/*2 heterozygotes, patients with current/former flushing, those with an MCV ≥ 106 fl, those who smoked ≥30 cigarettes per day, and those with esophageal dysplasia, and it developed significantly less frequently among those with BMI ≥ 23.2 (highest quartile). There was a marginally significant higher frequency of SCC of the UADT among the ADH1B*1/*1 homozygotes.

The Cox proportional hazards analysis enabled determination of the age-adjusted relative hazards for SCC of the UADT (Table 1): 11.55 [95% confidence interval (95% CI), 5.73-23.3] in ALDH2*1/*2 heterozygotes compared with ALDH2*1/*1 homozygotes, 2.02 (95% CI, 1.02-4.02) in ADH1B*1/*1 homozygotes compared with ADH1B*1/*2 heterozygotes or *2/*2 homozygotes, 2.64 (95% CI, 1.49-4.67) in patients with current/former flushing compared with those who had never experienced flushing, 2.91 (95% CI, 1.63-5.20) in those with an MCV ≥ 106 fl compared with those with an MCV < 106 fl, 2.52 (95% CI, 1.22-5.22) in those who smoked ≥30 cigarettes per day compared with those who smoked 0 to 19 cigarettes per day, 7.26 (95% CI, 3.99-13.23) in those with esophageal dysplasia compared with those without DIUL ≥ 5 mm, and 0.28 (95% CI, 0.09-0.85) in those with BMI ≥ 23.2 compared with those with BMI < 19.0 (lowest quartile).

Similar significant relationships between these factors and risk of SCC were found in both the esophagus and oropharyngolarynx, but the effect of ADH1B and smoking at both sites and of BMI in the oropharyngolarynx did not reach significance (Table 1).

A multivariate analysis that excluded alcohol flushing and MCV, both of which are strongly correlated with ALDH2 genotype, showed that ALDH2/ADH1B genotype, esophageal dysplasia, and BMI independently affected the risk of SCC of the UADT (Table 2). When ALDH2/ADH1B genotype was excluded, the multivariate analysis showed that alcohol flushing, esophageal dysplasia, and BMI were independently related to the risk, and the effect of MCV was marginally significant.

The follow-up endoscopic screening by oropharyngolaryngeal inspection and esophageal iodine staining revealed a high frequency of development of SCC of the UADT in the alcoholic men diagnosed as cancer-free in the initial screening. This is the first follow-up study to show a higher risk of development of SCC of the UADT among cancer-free alcoholics with inactive ALDH2, less-active ADH1B, current/former alcohol flushing, heavier smoking, and MCV ≥ 106 fl. Alcoholics with BMI ≥ 23.2 had a lower risk of SCC. When the risk of esophageal SCC and oropharyngolaryngeal SCC was evaluated separately, essentially the same results were obtained as when SCC was evaluated in the UADT as a whole, thereby supporting the validity of the findings. The results are consistent with the results of previous case-control studies (3-6, 8, 9, 12). The most important advantage of the present follow-up study over the case-controls studies was that the risk factors were measured before the development of SCC, thereby avoiding the debate as to whether the status of risk factors (other than genotype) was altered by the disease. That was especially important with regard to MCV, BMI, esophageal dysplasia, and smoking. Another advantage was that the follow-up study allowed calculation of the frequency of SCC over a long period.

The results also showed that esophageal dysplasia is a good independent marker or predictor of high risk of SCC in the UADT rather than a potential precursor of SCC per se. Esophageal SCC seemed to have developed at the same site as the dysplasia in only 9 of the 22 SCC patients diagnosed with esophageal dysplasia during the initial screening examination. Because esophageal dysplasia was diagnosed in biopsy specimens from DIULs ≥5 mm and is seldom diagnosed without esophageal iodine staining, esophageal iodine staining may be useful for predicting future risk of SCC as well as for improving the rate of detection of early SCC.

The risk associated with current/former flushing was substantially lower than the risk associated with inactive ALDH2 (2.64 versus 11.55), and both risks were similar to the corresponding odds ratios estimated in our case-control study in Japanese alcoholics (3.49 versus 7.45; ref. 9). The weaker relationship is mainly explained by the lower sensitivity of the alcohol flushing for identifying inactive ALDH2 in alcoholics, especially in alcoholics with less-active ADH1B (25). When current/former flushing individuals were assumed to have inactive ALDH2, the sensitivity and specificity for identifying inactive ALDH2 were 77% and 93%, respectively, in the present alcoholics, and 57% and 94%, respectively, in those with less-active ADH1B, which is a risk factor for both alcoholism and esophageal SCC. Alcohol-induced flushing, which may be triggered by a steep initial increase in either blood or cutaneous acetaldehyde after drinking, may not occur in persons with the less-active ADH1B. However, one of the greatest advantages of using the flushing questionnaire and inexpensive MCV testing in lieu of genotyping is that they can be easily used to select high-risk alcoholics for careful follow-up examinations at frequent intervals.

Patients with a BMI ≥ 23.2 (highest quartile) had a substantially lower risk of SCC. This finding is consistent with the results of prior reports, which showed that a low BMI increased the risk of oral and oropharyngeal (26) and esophageal (27) SCC. Our earlier case-control study also showed an inverse relationship between BMI and risk of esophageal SCC in alcoholics (9). Although a majority of the esophageal SCCs were in the early stage in that case-control study (9), we were unable to rule out the possibility that the negative relationship was to some extent a consequence of cancer-related weight loss. However, such a possibility is very unlikely in the present follow-up study of cancer-free alcoholics, and the patients' low BMI at the baseline examination may have been mainly attributable to their poor dietary habits that had led to vitamin deficiencies and suppressed immune function.

Because the time course of SCC development in UADT is unknown, there are no guidelines for longitudinal follow-up examinations. In the present follow-up study, in 17 of the 23 patients diagnosed with intraepithelial esophageal SCC at the time of the follow-up examination, the intraepithelial esophageal SCC was detected within a relatively short interval of 2 years after the final endoscopic examination. By contrast, the interval was >2 years in 9 of the 10 patients eventually diagnosed with invasive esophageal SCC. Although 20 of the 30 patients with oropharyngolaryngeal SCC were diagnosed with intraepithelial or subepithelial SCC, the oropharyngolaryngeal SCC tended to be diagnosed at a more advanced stage than esophageal SCC, regardless of the interval between the examinations. Gastrointestinal endoscopists may miss very early cancer or cancer located in dead areas of the oropharyngolarynx. These findings warrant intensive follow-up examinations of high-risk alcoholics (e.g., annually) by endoscopy combined with esophageal iodine staining and oropharyngolaryngeal inspection and/or laryngoscopy by otorhinolaryngologists.

Our study has several potential limitations. Because there was no full follow-up surveillance program for all alcoholics diagnosed as cancer-free by the initial screening, we are unable to report the true incidence of SCC in the entire cohort. For the 808 follow-up patients, the data set of genotypes and alcohol flushing is also incomplete. The reasons for the follow-up examinations included readmission because of a relapse of alcoholism and detection of esophageal dysplasia at the time of the initial screening. The background characteristics of the follow-up patients may reflect their higher risk of SCC compared with all alcoholics diagnosed as cancer-free by the initial screening. Thus, we cannot rule out the possibility that the Kaplan-Meier curves in the follow-up patients overestimated the detection rate of SCC. However, we have reported a prevalence of non-cancerous DIULs ≥5 mm in esophageal-cancer-free alcoholics of 23.2% (2), and the finding of a similar value (23.8%) in the present study implies that they represented cancer-free alcoholics as a whole with regard to DIULs. The age distribution, drinking, smoking, ALDH2/ADH1B genotypes, alcohol flushing, MCV, and BMI characteristics of the subjects of this study were also similar to the characteristics of the esophageal cancer-free controls consecutively recruited in our previous case-control study (9). Five of the 53 SCC patients had symptoms attributable to the SCC lesions before the follow-up screening, and there was concern that that may have had unexpected effects. However, repeating our analyses after excluding these five patients confirmed the results obtained in our overall analyses; thus, it is unlikely that the case recruitment biased the results greatly.

Another limitation of our study is that technical improvements in endoscopes and growing understanding of the endoscopic findings of very early SCC in the oropharyngolarynx (28, 29) have been achieved during the study period (1993-2005). The Olympus Q240 and Q240Z panendoscopes provided much clearer images than the older Olympus models. The results of this study should be confirmed by studies based on prospective designs and using advanced technology and diagnostic methods.

The majority of the SCC patients in this study were treated by endoscopic or endoscope-guided mucosectomy. Treatment by mucosal resection procedures is safe and effective for early SCC of both the esophagus (2, 30) and oropharyngolarynx (28, 29) and is especially useful when definitive therapy (radical surgery and/or chemoradiation therapy) would involve a risk of complications. Patients treated by mucosal resection recover within a few days and can enjoy the same quality of life as before. However, when patients fail to receive regular follow-up examinations (as often happens when alcoholics relapse), SCC in the UADT may go undetected until the late stages. Combining the results of esophageal iodine staining with ALDH2/ADH1B genotypes, replies to the flushing questionnaire, MCV, smoking, and BMI may not only provide a means of estimating the risk of future development of SCC in the UADT but also help persuade high-risk alcoholics to undergo careful long-term follow-up examinations and change their lifestyle choices with regard to drinking, smoking, and diet.