Introduction

Esophageal cancer is the fifth and seventh most common cancer in males and females , respectively, worldwide [1], and patient prognosis has been bleak [2]. Due to advances in diagnosis, operative techniques, perioperative management, and chemoradiotherapy, patient prognosis has improved, and more than half of the patients with this cancer appeared to be cured after surgical resection at a high-volume center in Japan [3]. However, early detection as well as prevention will provide the best opportunity to avoid death due to esophageal cancer.

Both alcohol consumption and cigarette smoking are well-established risk factors for esophageal squamous cell carcinoma (ESCC) [4–6], and these two factors have synergistic effects on the development of ESCC [7, 8]. Furthermore, these risk factors are closely associated with multiple occurrences of squamous cell carcinoma (SCC) in the upper aerodigestive tract (UADT) [9–11].

Esophageal cancer susceptibility differs between individuals due to polymorphisms in enzymes that metabolize carcinogens. Therefore, the effects of tobacco and alcohol on the development of cancers are considered to differ between individuals, despite similar exposures to these risk factors [12–14].

We herein discuss the significance of cigarette smoking and alcohol drinking as risk factors for ESCC, paying special attention to the joint effect of these two factors, multiple occurrences of SCC in the UADT, and individual differences in cancer susceptibility.

Risk of ESCC incurred by cigarette smoking and alcohol drinking

Both cigarette smoking and alcohol consumption are well-established risk factors for ESCC [4, 5]. It is estimated that 90% of ESCCs are attributable to tobacco and alcohol in more developed countries, although the relative contribution of these risk factors varies between geographic areas [6]. A large cohort study revealed that odds ratios (ORs) for esophageal cancer mortality were 2.4 for daily alcohol consumption in comparison to that for nondrinkers; and 2.3 for heavy smoking (15 cigarettes per day or more) in comparison with that for nonsmokers, and this study also reported dose–response relationships for alcohol drinking and smoking with ESCC [7]. Another study reported ORs for esophageal cancer mortality in the male population to be 2.71 and 4.36 for former and current smokers, in comparison to people who had never smoked, and 2.43 and 2.40 for ex- and current drinkers, respectively [15]. A cohort study from Shanghai indicated similar results, describing a twofold risk of developing esophageal cancer for both regular drinkers and long-term smokers [16]. A case–control study examining 346 patients with hypopharyngeal or esophageal cancer (cases) and 11,936 male controls reported ORs of hypopharyngeal and esophageal cancers of 2.2 and 3.5 for current smokers, and 4.7 and 4.4 for current drinkers, respectively [17]. We previously conducted a case–control study examining the risk of thoracic ESCC mainly for men living on the island of Kyushu in Japan. As a result, the ORs for smokers and drinkers were 3.9 and 4.5. Furthermore, a dose-relationship between ESCC risk and the cumulative quantity of tobacco or alcohol was described (Fig. 1a) [8]. Strong associations with tobacco and alcohol consumption were also reported regarding female ESCC [18–20]. A case–control study based on 114 women with ESCC and 425 controls reported that ORs for ESCC were 4.5 for heavy smoking and 5.4 for heavy alcohol drinking [20].

Fig. 1
figure 1

Risk for the development of squamous cell carcinoma of the thoracic esophagus in males in relation to cigarette smoking and alcohol drinking, based on a case–control study. a Odds ratios (ORs) of esophageal cancer for smokers and drinkers in comparison to those for nonsmokers and nondrinkers, respectively. b Combined effect of cigarette smoking and alcohol consumption on the occurrence of esophageal cancer. The term “Pack-years” is defined as the number of cigarettes per day divided by 20 with the result multiplied by the number of years of smoking. The drinking index (wk, weeks; yr, year) was calculated according to the alcohol concentration of the beverage. One drink of alcohol corresponds to 180 ml of sake (the most popular local alcoholic beverage), 120 ml of white liquor (shochu), 70 ml of whisky, or 720 ml of beer. Odds ratios were calculated and compared with those for persons who neither drank nor smoked [8] (license number 2374680816767)

Full size image

Pandeya et al. [21] examined the effect of alcohol intake quantity on the development of esophageal cancer in detail: below 170 g/week alcohol intake, there was no association between alcohol intake and the risk of ESCC. Above this threshold, however, a significant 3% increase in the risk of ESCC was observed for each extra 10 g/week of alcohol intake thereafter. Furthermore, there was a statistically significant multiplicative interaction between alcohol intake and smoking on the risk of ESCC. The authors also analyzed the risk of adenocarcinoma of the esophagus as well as for the esoohagogastric junction; however, there was no association between alcohol intake or smoking-drinking interaction and adenocarcinoma risk.

On the other hand, several researchers have revealed that the dose–response relationship characteristics regarding the risk of esophageal cancer may differ between tobacco and alcohol. The risk for tobacco depends on the duration of the tobacco consumption rather than the number of cigarettes, while the risk for alcohol depends mainly on the mean daily intake and not on the duration of the habit [15, 19, 22].

Joint effect of tobacco and alcohol on the development of esophageal cancer

The joint effect of tobacco smoking and alcohol drinking is considered to be synergistic rather than additive [7, 19]. Our case–control study clearly revealed the synergistic effects: in groups of nonsmokers or nondrinkers, the elevation of the OR of esophageal cancer is small, while marked elevation of the OR was observed in heavily consuming groups. The OR for those who were both heavy smokers and heavy drinkers was 50.1 (95% confidence interval: 14.3–175) (Fig. 1b) [8]. A synergistic effect of smoking and drinking on ESCC development was observed in both males and females [19]. Lee et al. [23] examined the joint effect with generalized additive models in detail. They concluded that a high frequency of heavy alcohol consumption contributed substantially to the development of ESCC. Furthermore, tobacco was found to interact with a light-to-moderate amount of alcohol (0.1–30 g/day) to increase the risk of esophageal cancer in a supramultiplicative way, whereas heavy alcohol consumption involved a simple multiplicative model.

The mechanism of this synergistic effect appears to be multifactorial. One mechanism is a local effect of acetaldehyde. Acetaldehyde metabolized from ethanol is highly carcinogenic by forming DNA adducts [24]. Acetaldehyde concentration in saliva was higher in smokers than in nonsmokers. Furthermore, active smoking plus ethanol challenge resulted in a sevenfold higher level of salivary acetaldehyde than that in nonsmokers [25]. Heavy smoking and drinking causes poor oral hygiene, which changes the bacteria in the oral cavity. As a result, higher acetaldehyde production from ethanol in saliva may cause a UADT cancer [26, 27]. Another explanation is the solvent action of alcohol: tobacco smoke contains many carcinogenic substances including benzo[a]pyrene. A classical animal experiment revealed that alcohol acted as a solvent to increase benzo[a]pyrene transportation to the esophageal mucosa [28]. The activation of cytochrome P450 2E1 appears to be important in the synergistic effect of tobacco and alcohol. Chronic alcohol consumption induces cytochrome P450 2E1, which metabolizes ethanol to acetaldehyde primarily in the liver. The activation of cytochrome not only leads to the generation of reactive oxygen species and reactive nitrogen species but also increases the activation of nitrosamines in tobacco smoke into active carcinogens [24].

Risk of esophageal cancer in relation to tumor location

A large case–control study reported by Takezaki et al. [17] examined the relation between tumor location and risk factors. As a result, cigarette smoking was found to increase the risk of middle esophageal cancer, with ORs of 5.8 for current smokers and 2.6 for chain smokers, while the elevations of ORs for pharyngeal cancers and upper and lower esophageal cancers were minimal (2.2, 2.4, and 2.1, respectively, for current smokers). The same tendencies were also recognized in other factors regarding cigarette smoking, such as the number of cigarettes, years of smoking, and years after quitting smoking. On the other hand, alcohol consumption significantly elevated the risk of pharyngeal cancer as well as that of the esophagus: The ORs of current drinkers for hypopharyngeal cancer and upper, middle, and lower esophageal cancers were 4.7, 6.8, 3.8, and 4.7, respectively. These results are concordant with those in a case–control study from Shanghai, China [29]. These findings suggest the possibility that a higher incidence of mid-thoracic esophageal cancer may be due to a stronger effect of tobacco exposure in the middle third of the esophagus.

Association of both smoking and drinking with multiple occurrences of SSC in the UADT

Slaughter et al. [30], in 1953, first reported the high incidence of multiple SCCs in the head and neck region and thus proposed the concept of field carcinogenesis. Since then, a large number of studies have demonstrated multiple occurrences of SCC in the UADT, including the esophagus, head and neck region, and bronchi. The incidence of such multiplicities has been reported to be between 9% and 14% [8]. Regarding intraesophageal multiplicity, our pathohistological study of entire resected esophageal specimens revealed the incidence to be 22.3%. Furthermore, the incidence of patients with 3 or more ESCCs was 13.8% [31].

We have analyzed multicentric carcinogenesis in the UADT from pathological, epidemiological, and biomolecular aspects. Kuwano et al. [32] histologically examined whole resected esophageal specimens in detail and reported a higher incidence of intraepithelial carcinoma in the main lesion in earlier rather than advanced cases. They also reported a close relationship between the multiplicity of ESCCs and the coexistence of intraepithelial carcinomas contiguous to the main lesion [33]. We have also previously reported that the coexistence of dysplasia was especially frequent in patients with three or more ESCCs (Fig. 2) [31]. Furthermore, when cytophotometric DNA analysis of nuclear DNA content was performed on different areas of ESCCs, variations in DNA content were shown to be more frequent in intraepithelial components than in invasive components [34]. These findings support the concept of multicentric carcinogenesis in the esophagus. On the other hand, we also reported a close relationship between intraesophageal multiple cancers and concomitant head and neck cancer, which supports the concept of field carcinogenesis within the esophagus that is closely related to multiorgan carcinogenesis in the UADT [35].

Fig. 2
figure 2

Number of dysplastic lesions in the esophagus in patients with solitary or multiple squamous cell carcinoma (SCC) of the esophagus. The single and double asterisks indicate P < 0.05 and P < 0.01, respectively, compared with the solitary SCC group

Full size image

Because cigarette smoking and alcohol drinking are well-established risk factors for UADT cancers, we examined the risk factors of UADT cancers epidemiologically, paying special attention to these two habits as well as familial cancer history. The study herein examined 143 men with SCC cancer of the thoracic esophagus, including 30 with associated cancer of the UADT, and revealed that the ORs of the second cancer in the UADT for smokers and drinkers were 5.3 and 14.7, respectively. The ORs were significantly increased, to 12.7 and 14.7, for heavy smokers and heavy drinkers, respectively. Furthermore, an eightfold risk elevation of second cancers was also present after analysis showed a positive family history of UADT cancers [9]. Similar findings were observed for the risk of secondary UADT cancers associated with carcinoma of the hypopharynx and cervical esophagus [10]. To accurately clarify the relationship between the number of SCCs in the UADT and smoking and drinking habits, we further performed serial histological evaluations using the entire resected esophagus, including the carcinoma. As a result, a close relationship was found between multiplicity and both habits. The incidence of heavy smokers and that of heavy drinkers were found to be extremely high in patients with 3 or more SCCs in the UADT (Fig. 3) [11]. Furthermore, histological examination revealed that not only multiple carcinomas but also the coexistence of dysplasia were closely related to both smoking and drinking habits [36]. These findings support the hypothesis that heavy smoking and drinking are associated with multiple occurrences of carcinoma in the UADT (Fig. 4). Multifactorial analysis examining the risk of a second cancer after esophagectomy for SCC of the thoracic esophagus by Matsubara et al. [37] also supported these findings. The risk of head and neck cancers after esophagectomy was markedly elevated, with a relative risk of 34.9, and independent factors affecting subsequent malignancies included additional cancers detected prior to esophagectomy, being male, alcohol consumption, and smoking. Of note, we also evaluated the risk factors for the multiple occurrence of gastric adenocarcinomas, and a family history of gastric cancer was a risk factor for multiple gastric cancers, although neither smoking nor drinking were found to be risk factors [38].

Fig. 3
figure 3

Number of squamous cell carcinomas in upper aerodigestive tract (UADT) according to cigarette smoking and alcohol drinking. The drinking index was calculated according to the alcohol concentration of the beverage. One drink of alcohol corresponds to 180 ml of sake (the most popular local alcoholic beverage), 120 ml of white liquor (shochu), 70 ml of whisky, or 720 ml of beer. The single asterisks indicate P < 0.01, in comparison to the control group. The double asterisks indicate P < 0.01, compared with solitary SCC. SCC squamous cell carcinoma [11] (license number 2374670449092)

Full size image
Fig. 4
figure 4

Multicentric carcinogenesis in the upper aerodigestive tract (UADT). Findings such as intraesophageal multiplicity of squamous cell carcinoma (SCC), multiple occurrences of dysplastic lesions, and association with head and neck cancer, all of which were frequently observed in esophageal SCC cases, were each closely related. These data support not only the concept of field carcinogenesis in the esophagus but also the concept of multiorgan carcinogenesis in the UADT. Both cigarette smoking and alcohol drinking, as well as a family history of UADT cancer, are considered to be closely related to multicentric carcinogenesis in the UADT. It is important to clarify the molecular mechanism of carcinogenesis based on basic research. There are also several significant clinical aspects of multicentric carcinogenesis, such as preoperative screening for secondary cancers, patient followup for the development of secondary cancer after treatment of the initial UADT cancer, and determination of the resection line or radiation field

Full size image

A p53 gene mutation has been reported in over half of human cancers, and it appears to occur in early-stage cancers, thus suggesting the important role of such mutations in carcinogenesis of the digestive tract [39]. We have reported that the abnormal expression of p53 or mutations of the p53 gene are closely related not only to the development of esophageal cancers but also to the development of multicentric carcinogenesis of the UADT, particularly when combined with smoking and drinking habits [36, 40–44]. An immunohistochemical study revealed a close relationship between the expression of p53 protein in the tumor and smoking and drinking habits in patients with SCC of the esophagus [40]. Furthermore, the use of both tobacco and alcohol was clearly associated with the expression of p53 in multiple ESCCs [41], as well as in dysplastic lesions [36]. On the other hand, our studies examining the p53 gene mutation pattern revealed a predominance of transversion mutations, in particular the G:C to T:A transversion; a relatively low frequency of transitions; and a relatively high percentage of frameshift mutations. These findings suggest a possible importance of the benzo[a]pyrene metabolite in tobacco smoke and oxidative DNA damage in carcinogenesis [42, 43]. Furthermore, we found a variation in the p53 mutation pattern between various lesions in multiple esophageal carcinomas [44]. All these findings suggest that the p53 gene is a target gene of tobacco as well as alcohol in multicentric carcinogenesis of the esophagus.

The fragile histidine triad (FHIT) gene is another candidate target gene of tobacco and/or alcohol. Mori et al. [45] reported a close relationship between FHIT expression in normal esophageal epithelium in a patient with SCC of the esophagus who had heavy use of both tobacco and alcohol. Furthermore, we found that FHIT expression in the esophageal epithelium was related to the development of multiple ESCCs [46]. Soma et al. [47] found that nicotine induced FHIT gene methylation in esophageal squamous cell lines.

The effect of genetic polymorphisms of aldehyde dehydrogenase 2 on susceptibility to alcohol in the development of esophageal cancer

Most carcinogens in alcoholic beverages as well as in tobacco smoke are not carcinogenic per se; metabolic activation is required for a carcinogenic action. Drug-metabolic enzymes that activate or detoxify are considered to play an important role in exogenously induced carcinogenesis [24]. In alcohol metabolism, acetaldehyde, which is first metabolized from ethanol and then is oxidized to acetate, is considered to be carcinogenic, and forms acetaldehyde DNA adducts [48] and causes mutations in tumor suppressor genes [49]. These reactions are primarily catalyzed by alcohol dehydrogenase (ADH) and aldehyde dehydrogenase (ALDH), respectively, in the liver. Aldehyde dehydrogenase 2 (ALDH2) is the key enzyme of ethanol metabolism. In 40%–50% of Oriental peoples, ALDH2 has extremely low activity due to an amino acid substitution of lysine for glutamine at position 487 of the protein following a G-A single nucleotide polymorphism (SNP) within the coding region of the ALDH2 gene [24]. Among the Japanese population, the incidence of the inactive ALDH2 subunit originating from the ALDH2*2 allele is 42% [50] and the peak blood acetaldehyde concentration after alcohol ingestion is closely related to this ALDH2 subtype [51]. Furthermore, severe acetaldehydemia in those with the ALDH2*2 allele results in the flushing response that includes facial flushing, tachycardia, headaches, and other unpleasant symptoms after consuming alcohol [52].

Yokoyama and colleagues examined the risk of UADT cancer, paying special attention to the polymorphisms of alcohol-metabolizing enzymes in patients with alcoholism. These studies have revealed that the presence of heterozygous ALDH2*1/2 is a risk factor for esophageal cancer as well as oropharyngeal cancer [12, 53, 54]: the ORs for esophageal cancer and oropharyngeal cancer were 11.14 and 12.50 in the presence of the ALDH2*2 allele [12]. These findings are consistent with those of a study examining a general population: the ORs of esophageal cancer for people with the ALDH2*1/2 allele compared with those with active ALDH2 (ALDH2*1/1) were reported to be 5.82, 10.01, and 8.56 in light, moderate, and heavy drinkers, respectively [13]. The heterozygous ALDH2*1/2 was also considered to be a risk factor for the multiple occurrences of SCC in the UADT among alcoholic patients [55], as well as in the general population [56, 57]. Clinically, endoscopy with Lugol staining allows the visualization of multiple Lugol-voiding lesions, which reflects multiple occurrences of either SCC or dysplasia in the esophagus (Fig. 5). These findings were found to be closely related to the inactive ALDH2 subtype [56, 57].

Fig. 5
figure 5

Endoscopic findings of multiple Lugol-voiding lesions; detection of the lesions was difficult without Lugol staining (a). However, numerous Lugol-voiding lesions, which histologically reflect either dysplasia or squamous cell carcinoma, were recognized in the same esophagus after Lugol staining (b)

Full size image

Blood acetaldehyde and salivary acetaldehyde play an important role in UADT carcinogenesis [26, 27]. Salivary acetaldehyde concentrations were found to be much higher than blood acetaldehyde concentrations. Furthermore, they were higher in ALDH2 heterozygotes than in homozygotes [58]. We performed immunohistochemical staining with an ALDH2 antibody in resected esophageal specimens from patients with esophageal cancer. We found that the expression of ALDH2 in the esophageal epithelium was closely related to the dose of consumed alcoholic beverages and was lower in patients showing the flushing response, which reflects inactive ALDH2 [59]. These findings support the possibility that a direct effect of salivary aldehyde in the esophageal mucosa may cause the development of carcinoma in the UADT, and the findings also explain the difference in UADT susceptibility according to ALDH genotype. To clarify the mechanism of inactive ALDH2 on carcinogenesis, we examined the tissue distribution of ALDH2 using an ALDH2 knockout mouse [60]. Furthermore, tissue toxicity after acetaldehyde inhalation was found to be more severe in ALDH2 knockout mice than in wild-type mice [61]. This animal model is useful to elucidate the mechanism of alcohol-induced carcinogenesis.

Relationship between diet and the development of esophageal cancer in relation to smoking and drinking

A high intake of fruits and vegetables has been considered to be effective in cancer prevention [16, 17, 62, 63]. A prospective cohort study of middle-aged Japanese men revealed that an increase in consumption of total fruit and vegetables by 100 g/day was associated with an 11% decrease in the development of ESCC and that cruciferous vegetables especially decreased the risk of ESCC [64]. The authors concluded that the beneficial effects of fruits and vegetables were not able to completely offset the harmful effect of tobacco and alcohol. On the other hand, a multinational case–control study performed by the International Agency for Research on Cancer (IARC) suggested that fruit and vegetable intake may modulate the effect of tobacco and alcohol on oropharyngeal carcinogenesis [65].

Dietary folate intake is reported to decrease the risk of cancers in various organs, including ESCC. A case–control study from Italy revealed that SCC of the esophagus was inversely associated with a high intake of folate, at an OR of 0.68, and that the decrease of the cancer risk caused by dietary folate intake was significant in drinkers with heavy alcohol consumption[66]. Furthermore, a Japanese epidemiological study revealed that serum folate was markedly decreased and mean corpuscular volume (MCV) was strongly elevated by alcohol consumption, especially in the ALDH2-inactive subtype population [67]. These findings suggest not only that acetaldehyde plays a significant role in decreased serum folate after alcohol consumption, but also that an elevated MCV, which reflects insufficient folate, may be a predictor of the risk of alcohol-related cancers such as SCC of the esophagus.

Prevention and early detection of ESCC

The prognosis of patients with esophageal cancer is still not ideal, although treatment and perioperative management have evolved in recent years with dramatic advances in diagnostic techniques, operative methods, and combined chemoradiotherapy [3, 68]. In order to improve patient prognosis, early detection and prevention are critical. Cigarette smoking and alcohol drinking are risk factors for the development of esophageal cancer as well as being risk factors for the multicentric occurrence of UADT cancer. The incidence of smokers has gradually decreased in Japan; however, 45.8% of male adults and 13.8% of female adults were smokers in 2005. On the other hand, the incidence of alcohol drinkers has gradually increased. Approximately seventy million Japanese consume alcoholic beverages, and 2.3 million were found to be heavy drinkers who consumed 150 ml of alcohol or more per day [69]. Esophageal cancer development depends on the cumulative quantity of tobacco and alcohol, and the effects of these two factors are synergistic. The duration of the smoking habit is reported to be especially significant [15, 19, 22]. The cessation of smoking habits for current smokers, as well as education on the adverse effects of tobacco for children, will be an effective strategy for smoking prevention. The quantity of daily alcohol intake as well as the intake of strong alcoholic beverages is a factor in alcohol-induced carcinogenesis [15, 19, 22]. Furthermore, people with the inactive ALDH2 subtype are susceptive to UADT cancers. A questionnaire for the facial flushing response is a simple and reliable indicator of this genotype [70]. In order to prevent the development of ESCC, it is important not only to avoid heavy drinking but also to avoid consuming alcohol together with tobacco, especially for those who have experienced the flushing response after drinking alcohol.

Endoscopic therapy, including endoscopic mucosal resection and endoscopic submucosal dissection, is less invasive than open surgery and is a curative treatment for early esophageal cancer, and early detection of these lesions enables early therapy [71]. Endoscopic examinations with Lugol iodine staining are extremely useful for the early detection of esophageal cancer, as well as for the detection of secondary cancers associated with UADT cancers [56, 57, 72]. High-resolution narrow band imaging endoscopy, which improves visualization of the microvasculature and mucosal patterns in the alimentary tract, has recently been applied for the early detection of esophageal cancer as well as pharyngeal cancer [73, 74]. In order to improve the prognosis of patients with ESCC, either endoscopy with Lugol staining or narrow band imaging endoscopy is strongly recommended for high-risk populations, such as smokers, heavy drinkers, people with experience of flushing after drinking alcohol, and patients with UADT cancer.