The pathogenetic mechanisms behind alcohol-associated carcinogenesis in the upper digestive tract remain unclear, as alcohol is not carcinogenic. However, there is increasing evidence that a major part of the tumour-promoting action of alcohol might be mediated via its first, toxic and carcinogenic metabolite acetaldehyde. Acetaldehyde is produced from ethanol in the epithelia by mucosal alcohol dehydrogenases, but much higher levels derive from microbial oxidation of ethanol by the oral microflora. In this study we investigated factors that might alter the composition and quantities of the oral microflora and, consequently, influence microbial acetaldehyde production. Information about dental health, smoking habits, alcohol consumption and other factors was obtained by a questionnaire from 326 volunteers with varying social backgrounds and health status, e.g. oral cavity malignancy. Paraffin-induced saliva was collected and the microbial production of acetaldehyde from ethanol was measured. Smoking and heavy drinking were the strongest factors increasing microbial acetaldehyde production. Whether poor dental status may alter local acetaldehyde production from ethanol remained unanswered. Bacterial analysis revealed that mainly Gram-positive aerobic bacteria and yeasts were associated with higher acetaldehyde production. Increased local microbial salivary acetaldehyde production due to ethanol among smokers and heavy drinkers could be a biological explanation for the observed synergistic carcinogenic action of alcohol and smoking on upper gastrointestinal tract cancer. It offers a new microbiological approach to ethanol-associated carcinogenesis at these anatomic sites.

Introduction

Smoking and alcohol intake are the most important risk factors for oral and pharyngeal cancer (14). Evidence concerning a connection between smoking and cancer has been extensively documented (5). Many of the compounds in tobacco smoke are hazardous to health and some are undoubtedly carcinogenic (5). In contrast, the tumour-promoting effects of alcohol consumption are less well defined. Epidemiological data clearly indicate an independent carcinogenic effect of alcohol intake, but ethanol itself is not carcinogenic (6). Acetaldehyde, the first metabolite of ethanol, is carcinogenic and it has been proposed that the major part of the carcinogenic potency of alcohol is mediated via this compound (7).

In addition to the intracellular formation of acetaldehyde via mucosal alcohol dehydrogenases (ADHs), acetaldehyde is also formed in high concentrations in saliva by the oral microflora (810). Salivary acetaldehyde production shows high interindividual variation (10). Recently, we have been able to demonstrate increased acetaldehyde production in the mouthwashings of patients with upper gastrointestinal tract cancer, but the causal relationships for this finding remain unclear (11).

The aim of the present study was to elucidate the factors that regulate microbial production of acetaldehyde in saliva. Moreover, possible differences in microbial composition and relative concentrations among `high' and `low' acetaldehyde producers were examined.

Materials and methods

Study subjects

A total of 326 volunteers participated in the study. This cohort consisted of 114 healthy volunteers, 24 patients seeking dental examination or treatment (enrolled by the Institute of Dentistry, University of Helsinki), 90 unemployed volunteers from the region of Vantaa, 26 patients with a malignant tumour of the oral cavity (enrolled by the Department of Oral and Maxillofacial Surgery, Helsinki University Central Hospital) and 64 alcoholics (enrolled by a local A-Clinic, Töölö, Helsinki). Eleven of the cancer patients were untreated and 15 were in the follow-up stages.

Questionnaire

A questionnaire was answered by each volunteer. Information concerning age, gender, tobacco use, alcohol use, diet, oral health status, oral hygiene habits and other characteristics were elicited. Tobacco use indicators included the average number of cigarettes, cigars or pipes smoked per day within the past 30 days, duration of smoking in years and the date when possible smoking cessation occurred. Daily tobacco consumption was calculated as cigarettes smoked per day (1 cigar = 3 cigarettes, 1 pipe = 5 cigarettes). Ex-smokers with smoking cessation of >5 years were ranked in further analyses as non-smokers; ex-smokers with a shorter period of cessation were excluded.

Alcohol consumption was estimated as the average number of ingested drinks (~12 g of pure alcohol) for every drinking day during the past 30 days and as the frequency of alcohol intake per week. Based on these data, the average amount of alcohol consumed as g pure ethanol/day was calculated. Volunteers were ranked as non-drinkers (<1 g/day), moderate drinkers (1–30 g/day for females, 1–40 g/day for males) or heavy drinkers (>30 g/day for females, >40 g/day for males). Since the statistical analysis did not reveal significant differences between the teetotallers and moderate drinkers, these groups were combined.

Indicators of oral health included the number of teeth lost (except wisdom teeth), frequency and time since last visit to the dentist, denture wear, frequency of toothbrushing, self-reported periodontitis, self-reported frequency of a dry or burning mouth, frequency of mouthwash use and frequency of eating between meals.

Salivary samples

Stimulated whole saliva was collected between 9 and 12 a.m. after 1 min use of a paraffin chewing gum (Orion Diagnostics, Espoo, Finland). It was immediately frozen at –70°C. Exclusion criteria were as follows: treatment with oral antiseptic or antibiotics in the past month, food or fluid intake, smoking or toothbrushing in the past 90 min, recent alcohol intake or measurable amount of alcohol in the saliva by head space gas chromatography.

Salivary acetaldehyde production capacity

Saliva was thawed and preheated to 37°C before analysis. A sample of 400 μl of saliva was transferred to a gas chromatograph vial. To this was added 50 μl of potassium phosphate buffer (final concentration 100 mM, pH 7.4) containing ethanol (final concentration 22 mM) and the vials were immediately tightly closed. Vials were incubated for 90 min and the reaction was stopped by injecting 50 μl of 6 M perchloric acid through the rubber septum of the vial. All samples were measured as triplicates. For each salivary sample one analysis was carried out by concomitantly adding 50 μl perchloric acid and an ethanol/potassium phosphate buffer mixture, before addition of 400 μl of saliva. These control assays for baseline and artefactual acetaldehyde production were incubated for 90 min and revealed values were subtracted from the acetaldehyde levels after 90 min incubation with ethanol. Head space gas chromatographic conditions (Perkin Elmer, Norwalk, CT) were as follows: column 60/80 Carbopack B/5% Carbowax 20M, 2 m×1/8 inch (Supelco, Bellefonte, PA); oven temperature 85°C; transfer line and detector temperature 200°C; carrier gas flow rate (N2) 20 ml/min.

Bacterial analysis

Among all volunteers, the 10 saliva samples with the lowest and highest acetaldehyde production were identified. The samples were thawed, serially diluted in peptone yeast extract broth and 10 μl of undiluted saliva and the appropriate dilutions were inoculated on several non-selective and selective agar media for the enumeration and isolation of aerobic and anaerobic bacteria. Aerobic blood agar base medium and chocolate agar were used for the determination of total aerobic counts. Vitamin K1- and hemin-supplemented anaerobic Brucella blood agar media were used for the determination of total anaerobic counts. The aerobic plates were incubated at 36°C in an atmosphere containing 5% CO2 for a total of 5–7 days and anaerobic plates in anaerobic jars filled by the evacuation replacement method with mixed gases (85% N2, 10% CO2, 5% H2) were incubated for 7 days for the first inspection and further up to 14 days for the final inspection. Bacterial counts were determined by multiplying the number of colonies by the dilution factor, adjusted for inoculation volume.

Statistical analysis

For statistical evaluation of the data in Table I, the following statistical calculations were done. As a preliminary analysis, and to test for co-linearity of variables, a Spearman correlation matrix was computed for the entire study population. This was followed by multivariate regression analyses. Variables such as age, smoking (the number of cigarettes smoked per day), alcohol (the average amount of alcohol in grams), tooth brushing, tooth loss (number of lost teeth) and eating between meals were correlated as continuous variables, whereas self-reported periodontitis, frequency of dentist visits, mouthwash use, dentures and self-reported dry mouth and burning mouth were calculated as categorical variables. For categorical values, dummy variables were used in the analysis. As co-linearity was obvious for smoking and heavy alcohol intake, the multivariate analysis were re-run for non-smokers (n = 189) and for moderate and non-drinkers (n = 259, <30/40 g alcohol/day, for females/males, respectively) in order to adjust for this confounding factor. A multiple linear regression analysis, a forward stepwise regression analysis (no variable forced into the equation) and a best subsets regression analysis (r2 as best criterion) was run with the best descriptor for all variables, setting acetaldehyde production as the dependent variable.

Different groups of data were analysed by calculating simple ANOVA on ranks, followed by Dunn's test (Figure 1). If two different groups were analysed, a non-parametric unpaired test (Mann–Whitney test) was used (comparison of different bacteria in Figure 2).

Fisher's exact test was used to analyse the distribution of smokers and heavy drinkers in the two groups of `high' and `low' acetaldehyde-producing salivas.

All reported P values derive from two-sided tests. All values are expressed as means ± SEM. For all statistical calculations, statistical software (SigmaStat 2.0; Jandel Scientific, San Rafael, CA) was used.

Results

Acetaldehyde production capacity

Patient characteristics and statistical analyses are summarised in Table I. The different regression analyses clearly show that smoking and heavy alcohol intake are strong predictors of microbial acetaldehyde production. As these factors themselves showed co-linearity, the analysis was repeated for all non-smokers (n = 189) to estimate the influence attributable to alcohol intake alone, and for subjects with only moderate or no alcohol consumption to estimate the influence of smoking alone (n = 259). By this analysis method, both factors could be proved to be independent risk factors for higher acetaldehyde production.

Age (inverse correlation, r = –0.13, P = 0.02) and the reported frequency of dry mouth (positive correlation, r = 0.1, P = 0.06) were other, less compelling factors possibly associated with increased acetaldehyde production. Again, both factors showed co-linearity with smoking (dry mouth, r = 0.17, P = 0.001; age, r = –0.2, P < 0.0001) and alcohol intake (age, r = –0.32, P < 0.0001). After adjustment for confounders, the reported frequency of a dry mouth showed a slight but significant contribution to salivary acetaldehyde production, at least in non-smokers. It can be estimated that a subject with a very frequent dry mouth has ~20% higher salivary acetaldehyde levels than subjects without such symptoms.

None of the other variables contributed significantly to salivary acetaldehyde production. In detail, patients with cancer of the oral cavity did not have salivary acetaldehyde production which differed significantly from the rest of the cohort. This was true both for patients with fresh, untreated tumours as for patients in follow-up after surgical removal of a malignancy.

Patients were ranked according to their alcohol consumption (moderate/non-drinkers versus heavy drinkers) and smoking habits (non-smoker versus smoker). Both smoking and heavy alcohol consumption independently increased salivary acetaldehyde production in comparison with the control groups by 60–75%, and combined misuse further increased it (Figure 1).

Microbial analysis

Microbial analysis of the saliva of `high' and `low' acetaldehyde producers showed a clear trend in aerobic conditions. Total counts of aerobes were significantly increased among `high' producers. Aerobic species that were significantly associated with higher acetaldehyde production were Streptococcus salivarius, hemolytic Streptococcus viridans var., Corynebacterium sp., Stomatococcus sp. and yeasts (Figure 2). Yeasts were not only found at higher concentrations, but also more frequently among the subjects with higher acetaldehyde production (eight of 10 versus two of 10, P = 0.02). Corynebacterium sp. were found in all `high' acetaldehyde producers, but in only six of 10 `low' producers (P = 0.08), whereas the other facultative commensals Streptococcus mutans, Haemophilus sp. and Staphylococcus sp. were equally distributed in both groups. No bacterial species was found to be significantly more frequent in the saliva of `low' acetaldehyde producers (Figure 2). Although the total anaerobic counts were slightly increased in `high' producers, there was no correlation between single bacterial species and acetaldehyde production (data not shown).

The selection criterion for `high' and `low' acetaldehyde producers was exclusively the measured acetaldehyde level and there was no significant difference in the proportion of smokers (P = 0.19) or heavy drinkers (P = 0.62, calculated by Fisher's exact test) in the two groups.

Discussion

Heavy drinking and smoking are the main causes of upper gastrointestinal tract cancer and they have been estimated to account for ~80% of all cases (1). Other possible risk factors for upper gastrointestinal tract cancer are poor nutritional status and intake of micronutrients, hereditary factors, certain papilloma viruses, occupational hazards and poor oral hygiene (1,1216). Dentition, tooth loss, poor dental status and oral hygiene habits have frequently been associated with higher risks, especially for oral cavity cancer (1417). It is generally agreed that the influence of poor oral hygiene as a risk factor is much less compelling than alcohol and smoking, but there is some experimental evidence that its influence might become clinically more important among alcoholics, where poor dental status and hygiene is a common problem (17). The reason for this finding is unclear.

Although alcohol and tobacco smoke are well-known independent and strong risk factors for upper gastrointestinal tract cancer, their combined action on these epithelia is poorly understood. There is epidemiological evidence indicating that alcohol and tobacco act together in a more multiplicative rather than in an additive manner and, accordingly, seem to have synergistic tumour-promoting effects (13,18,19). As alcohol is involved synergistically in the attributable risk of both smoking and poor oral hygiene, it is conceivable to suggest a unifying pathogenetic mechanism of alcohol drinking behind these epidemiological findings. This could be the local production of carcinogenic acetaldehyde from ethanol by oral microbes.

There is increasing evidence for acetaldehyde to be the ultimate carcinogenic substance behind alcohol intake. Acetaldehyde has been shown to be highly toxic, mutagenic and carcinogenic in different cell cultures and animal models (7,2023). In experimental animals, histopathological changes after acetaldehyde treatment have been shown to mimic those known to occur after treatment with alcohol (24). Stronger evidence for acetaldehyde as the major factor behind ethanol-associated carcinogenesis is derived from studies linking the genotypes of ethanol-metabolizing enzymes with tumour risk. Rapid metabolizing ADHs (ADH3), leading to higher and quicker production of cellular acetaldehyde, and lack of the low Km aldehyde dehydrogenases (ALDHs) ALDH2, leading to a longer and delayed exposure to acetaldehyde, have recently been shown to be associated with increased cancer risk in the upper gastrointestinal tract (2529). In a very recent study among Orientals, a possible correlation between ALDH2 genotype mutation and cancer risk in alcoholics has been expanded to all possible alcohol-related cancers. In this study, the frequency of a mutant ALDH2-2 allele was significantly higher in alcoholics with oropharyngeal, laryngeal, oesophageal, stomach, colon and lung cancer, but not liver or other cancers (29). This is very interesting, as these organs are covered by microbes and microbial production of acetaldehyde from ethanol has been described (811,30). Thus, it is possible that the hampered detoxification of acetaldehyde from ethanol in ALDH2-deficient subjects might only become clinically relevant in cases of marked acetaldehyde production by microbes. Hence, there is conclusive experimental support for microbial acetaldehyde production from ethanol as a major factor in alcohol-associated carcinogenesis.

Recently, local acetaldehyde production in the saliva by microbes has been described (10). Microbial salivary acetaldehyde production shows high interindividual variation, but there exists a significant positive correlation between salivary ethanol and acetaldehyde levels. Moreover, in vivo salivary acetaldehyde levels correlate very significantly with the levels that are produced in vitro. This offers the opportunity to use the in vitro salivary test as a tool to investigate possible variables which might influence salivary acetaldehyde production. Salivary acetaldehyde levels after ethanol intake strikingly exceed those known to be derived from endogenous metabolism of ethanol (10). Salivary acetaldehyde may reach, via normal distribution and evaporation, all target tissues of the upper aerodigestive tract, such as the larynx, pharynx, oral cavity, oesophagus and even the stomach. Consequently, we suggest that the major part of the carcinogenic role of alcohol is caused by its first metabolite, acetaldehyde, which is microbially produced.

In the present study, we were able to demonstrate that smoking and heavy alcohol consumption significantly increase salivary acetaldehyde production. Smoking showed a positive linear correlation and it can be estimated that a smoker with a daily consumption of ~20 cigarettes has an increased salivary acetaldehyde production of ~50–60%. This implies that smokers, even after moderate alcohol intake, produce much higher levels of carcinogenic acetaldehyde in the oral cavity than non-smokers. The evidence for increased microbial salivary acetaldehyde production in smokers, together with an epidemiological description of the multiplicative carcinogenic action of alcohol and smoking, suggests that salivary acetaldehyde production mediated by microbes could be a biologically plausible pathogenetic mechanism for these findings. Alcohol seems to interact and increase salivary acetaldehyde production only if consumption is heavy (>40 g/day); when an increase is observed it is dose dependent. Smoking and alcohol together increase salivary acetaldehyde production by ~100% as compared with non-smokers and moderate alcohol consumers.

It has been demonstrated that smoking and alcohol intake affect and slightly decrease saliva flow (31). It is well known that in the presence of a low saliva flow bacterial concentrations increase, and this could be the explanation for the observed higher total counts in these subjects. However, there are obviously also qualitative changes in the microbial flora in high acetaldehyde producers, as has already been described for smokers (3240). For instance, an increase in yeast infections, e.g. Candida albicans, has been demonstrated for smokers (3537). This is in accordance with our finding of an increased incidence and also higher loads of yeasts among high acetaldehyde producers. On the other hand, Neisseriae spp. have been reported to occur less frequently in the oral cavity of smokers and this species is not associated with higher acetaldehyde production (38,39). In general, a microbial `switch' with a significant increase in the proportion of Gram-positive versus Gram-negative bacteria has been described in smokers (34,3840). This is in line with our observation that almost all aerobic Gram-positive bacteria were significantly increased in `high' acetaldehyde producers (with the facultative commensals Staphylococcus sp. and S.mutans as the only exceptions), whereas the Gram-negative aerobic bacteria Haemophilus sp. and the already mentioned Neisseria sp. were not associated with higher acetaldehyde production. Thus, there is in general a good link between our microbial observations that some species are associated with higher acetaldehyde production and the well-known effect of smoking on the oral microflora.

Microbial changes in the oral microflora of alcoholics have been less intensively described (17,41). Epidemiological studies have shown that heavy drinking is associated with poor oral hygiene. It has been suggested that this may lead to bacterial overgrowth, but so far no study has convincingly proved this hypothesis and no bacterial species have been associated with high alcohol consumption (17). As high salivary acetaldehyde production was observed only among heavy drinkers, enzyme induction might be an explanation for this finding. Bacteria are known to be easily able to induce the corresponding metabolizing enzymes, however, it remains speculative whether this effect accounts for our observations.

Poor oral health status is a weak risk factor for oral cavity cancer (1416). Bacterial overgrowth and, in the case of concomitant alcohol intake, high bacterial acetaldehyde production have been suggested as pathogenetic mechanisms. Our study could not confirm these suggestions and the only indirect parameter of oral health which was weakly associated with higher acetaldehyde production was the self-reported frequency of dry mouth, which again could lead to higher bacterial loads per millilitre due to decreased flow. However, it can be questioned how reliably a self-reported questionnaire takes into account the factors representing actual oral status. The question whether salivary microbial acetaldehyde production might be the mechanism behind the higher cancer risk in alcoholics with poor oral hygiene remains unanswered and further studies that include actual dental status and a control for confounding factors are needed.

In the microbial fermentation of glucose to alcohol the conversion from acetaldehyde to ethanol is catalysed by ADHs and the reaction is reversible. Thus, in the case of substrate (ethanol) excess and in the presence of oxygen, the reaction runs in the opposite direction, with acetaldehyde as the end product. In this study, the total anaerobic counts included both facultative and microaerophilic bacteria that can grow under anaerobic conditions, but otherwise no anaerobic species were associated with higher acetaldehyde production. On the other hand, our microbial analyses revealed some Gram-positive aerobic bacterial strains and yeasts which where found at higher loads among high acetaldehyde producers. Moreover, yeasts (and possibly Corynebacterium sp.) were also found more frequently in this group. Recent experience with intestinal bacteria indicate that there are some strains which have a much higher acetaldehyde production capacity and ADH activity than others (30). Thus, our findings can be regarded as a first hint of high acetaldehyde-producing microbes. Moreover, our findings support the hypothesis that bacteria and yeasts are involved in ethanol-associated carcinogenesis and may represent the main metabolic source for the production of highly carcinogenic acetaldehyde from ethanol. This may open a new microbiological approach to the pathogenesis of the oral cavity and upper gastrointestinal tract cancer.

In contrast to our previous findings, we could not demonstrate a higher acetaldehyde production in patients with a malignancy of the oral cavity (11). First, saliva, and not mouthwash, was used as the probe in this study and we were able to demonstrate recently that in vitro salivary acetaldehyde is the method of choice to reliably reflect the acetaldehyde levels in vivo. Secondly, in this study a multivariate analysis was performed among 326 volunteers, whereas the number of patients in our previous study was limited (n = 53) and only univariate statistical analysis could be performed. Thirdly, controls and cancer patients were well balanced in our study with respect to alcohol consumption and smoking, whereas these risk factors were twice as high in the cancer cohort of our previous study (11). Thus, it is possible that our previous findings of higher acetaldehyde production among cancer patients might just have reflected the higher intake of tobacco products and alcohol in the cancer cohort rather than increased acetaldehyde production by the tumour or its influence on the microflora itself.

In conclusion, we have demonstrated increased microbial salivary acetaldehyde production in smokers and heavy drinkers. Numerous studies support the hypothesis that acetaldehyde is the substance behind the tumour-promoting effect of alcohol on the mucosa of the oral cavity. Salivary microbial production is supposed to be one of the major sources of acetaldehyde from ethanol. Thus, our finding could be a biologically plausible mechanism to explain the synergistic and multiplicative manner by which the attributable cancer risks of alcohol and smoking act. Several bacteria and yeasts have been found in significantly higher numbers in the saliva among high acetaldehyde producers, which offers a new microbiological approach to the pathogenesis of alcohol-associated carcinogenesis.

Table I.

Characteristics and effects on salivary acetaldehyde production of the different variables of the study population

Variable  Number (n Salivary acetaldehyde (μmol/l)  P value 
Variables were arranged in categories to improve readabily but the reported P values derive from calculations without previous ranking. Reported P values are as follows: first row, Spearman correlation matrix; second row, multiple linear regression analysis; third row, forward stepwise regression analysis. If only not significant (n.s.) was noted, this was true for all three tests. Best subset regression analyses were not mentioned, as no major difference to forward stepwise regression analysis was noted. Numbers in parentheses represent values after adjustment for the co-linear variable (see text). 
Total  326  119.4 ± 5.1   
Gender       
Female  159  110.3 ± 6.5  n.s. 
Male  167  128.2 ± 7.8   
Age (years)       
<41  93 (44)  127.2 ± 9.0 (84.7 ± 8.2)  <0.02 
41–58  162 (84)  119.7 ± 7.6 (89.6 ± 6.6)  n.s. (n.s.) 
>58  71 (47)  108.8 ± 10.5 (102.5 ± 14.0)  n.s. (n.s.) 
Smoking status       
No  115 (107)  94.6 ± 6.9 (88.2 ± 6.3)  <0.0001 
Yes  137 (82)  151.0 ± 8.8 (137.5 ± 12.2)  0.03 (0.008) 
Ex-smoker  74 (66)  99.7 ± 9.1 (98.0 ± 9.8)  0.008 (0.006) 
Alcohol status       
Teetotaller  50 (40)  111.4 ± 12.0 (102.3 ± 12.4)  <0.0001 
Moderate alcohol  209 (133)  104.0 ± 6.1 (88.8 ± 5.9)  0.01 (<0.001) 
Heavy drinker  67 (14)  171.6 ± 11.2 (156.0 ± 28.4)  0.009 (0.001) 
Tooth brushing       
Twice daily  205  117.3 ± 6.2  n.s. 
Once  106  122.1 ± 9.8   
Less frequently  15  131.1 ± 23.2   
Self-reported periodontitis       
Yes  48  114.1 ± 11.0  n.s. 
No  278  120.4 ± 5.7   
Tooth loss       
<3  190  122.6 ± 7.3  n.s. 
3–11  93  112.6 ± 7.5   
>11  43  120.2 ± 13.5   
Frequency of dentist visits       
At least once per year  137  124.0 ± 7.3  n.s. 
Once every 2/3 years  104  110.3 ± 8.3   
Less frequently  65  122.3 ± 13.0   
Last dentist visit       
Past 12 months  202  118.8 ± 6.6  n.s. 
>1 year ago  48  96.6 ± 9.5   
>2 years ago  76  134.2 ± 11.2   
Eating between meals (daily)       
<2  78  125.3 ± 9.8  n.s. 
2–3  228  115.6 ± 5.9   
>3  20  140.0 ± 30.7   
Mouthwash use       
Yes  22  130.4 ± 23.1  n.s. 
No  302  118.5 ± 5.2   
Dentures       
Yes  73  119.1 ± 5.8  n.s. 
No  253  120.7 ± 10.4   
Self-reported dry mouth       
Never  192 (124)  113.5 ± 6.5 (92.6 ± 6.5)  0.06 
Seldom  77 (38)  120.6 ± 8.6 (91.2 ± 9.6)  n.s. (0.03) 
Frequently/always  57 (27)  138.6 ± 15.0 (122.5 ± 19.5)  n.s. (0.02) 
Self-reported burning mouth       
Never  260  119.5 ± 5.9  n.s. 
Seldom  43  128.1 ± 12.8   
Frequently/always  23  102.4 ± 15.7   
Oral cavity cancer       
No  300  120.0 ± 5.4  n.s. 
Yes  26  103.9 ± 14.0   
Untreated case  11  113.1 ± 22.9   
Case in follow-up  15  91.3 ± 14.7   
Variable  Number (n Salivary acetaldehyde (μmol/l)  P value 
Variables were arranged in categories to improve readabily but the reported P values derive from calculations without previous ranking. Reported P values are as follows: first row, Spearman correlation matrix; second row, multiple linear regression analysis; third row, forward stepwise regression analysis. If only not significant (n.s.) was noted, this was true for all three tests. Best subset regression analyses were not mentioned, as no major difference to forward stepwise regression analysis was noted. Numbers in parentheses represent values after adjustment for the co-linear variable (see text). 
Total  326  119.4 ± 5.1   
Gender       
Female  159  110.3 ± 6.5  n.s. 
Male  167  128.2 ± 7.8   
Age (years)       
<41  93 (44)  127.2 ± 9.0 (84.7 ± 8.2)  <0.02 
41–58  162 (84)  119.7 ± 7.6 (89.6 ± 6.6)  n.s. (n.s.) 
>58  71 (47)  108.8 ± 10.5 (102.5 ± 14.0)  n.s. (n.s.) 
Smoking status       
No  115 (107)  94.6 ± 6.9 (88.2 ± 6.3)  <0.0001 
Yes  137 (82)  151.0 ± 8.8 (137.5 ± 12.2)  0.03 (0.008) 
Ex-smoker  74 (66)  99.7 ± 9.1 (98.0 ± 9.8)  0.008 (0.006) 
Alcohol status       
Teetotaller  50 (40)  111.4 ± 12.0 (102.3 ± 12.4)  <0.0001 
Moderate alcohol  209 (133)  104.0 ± 6.1 (88.8 ± 5.9)  0.01 (<0.001) 
Heavy drinker  67 (14)  171.6 ± 11.2 (156.0 ± 28.4)  0.009 (0.001) 
Tooth brushing       
Twice daily  205  117.3 ± 6.2  n.s. 
Once  106  122.1 ± 9.8   
Less frequently  15  131.1 ± 23.2   
Self-reported periodontitis       
Yes  48  114.1 ± 11.0  n.s. 
No  278  120.4 ± 5.7   
Tooth loss       
<3  190  122.6 ± 7.3  n.s. 
3–11  93  112.6 ± 7.5   
>11  43  120.2 ± 13.5   
Frequency of dentist visits       
At least once per year  137  124.0 ± 7.3  n.s. 
Once every 2/3 years  104  110.3 ± 8.3   
Less frequently  65  122.3 ± 13.0   
Last dentist visit       
Past 12 months  202  118.8 ± 6.6  n.s. 
>1 year ago  48  96.6 ± 9.5   
>2 years ago  76  134.2 ± 11.2   
Eating between meals (daily)       
<2  78  125.3 ± 9.8  n.s. 
2–3  228  115.6 ± 5.9   
>3  20  140.0 ± 30.7   
Mouthwash use       
Yes  22  130.4 ± 23.1  n.s. 
No  302  118.5 ± 5.2   
Dentures       
Yes  73  119.1 ± 5.8  n.s. 
No  253  120.7 ± 10.4   
Self-reported dry mouth       
Never  192 (124)  113.5 ± 6.5 (92.6 ± 6.5)  0.06 
Seldom  77 (38)  120.6 ± 8.6 (91.2 ± 9.6)  n.s. (0.03) 
Frequently/always  57 (27)  138.6 ± 15.0 (122.5 ± 19.5)  n.s. (0.02) 
Self-reported burning mouth       
Never  260  119.5 ± 5.9  n.s. 
Seldom  43  128.1 ± 12.8   
Frequently/always  23  102.4 ± 15.7   
Oral cavity cancer       
No  300  120.0 ± 5.4  n.s. 
Yes  26  103.9 ± 14.0   
Untreated case  11  113.1 ± 22.9   
Case in follow-up  15  91.3 ± 14.7   
Fig. 1.

Salivary acetaldehyde production in correlation with smoking and drinking. *P < 0.05 versus non-smokers, moderate alcohol consumption.

Fig. 1.

Salivary acetaldehyde production in correlation with smoking and drinking. *P < 0.05 versus non-smokers, moderate alcohol consumption.

Fig. 2.

Colony forming units (CFU) per ml of saliva of aerobic microbes in the saliva of `high' and `low' acetaldehyde producers. *P < 0.05, **P < 0.005, ***P < 0.0005. (1), Facultative commensal, which was significantly (P = 0.02 by Fisher's exact test) more frequently present in `high' acetaldehyde producers. (2), Facultative commensal, but no significant quantitative differences in positive carriers between the two groups. (3), Corynebacterium sp. were detectable in all `high' but only 60% of `low' producers (P = 0.08).

Fig. 2.

Colony forming units (CFU) per ml of saliva of aerobic microbes in the saliva of `high' and `low' acetaldehyde producers. *P < 0.05, **P < 0.005, ***P < 0.0005. (1), Facultative commensal, which was significantly (P = 0.02 by Fisher's exact test) more frequently present in `high' acetaldehyde producers. (2), Facultative commensal, but no significant quantitative differences in positive carriers between the two groups. (3), Corynebacterium sp. were detectable in all `high' but only 60% of `low' producers (P = 0.08).

4
To whom correspondence should be addressed Email: mikko.salaspuro@helsinki.fi

We thank Timo Pessi for his valuable advice concerning the statistical analysis of the data. We also thank all volunteers participating in this study. Special thanks are due to all workers of the Töölö A Clinic. The work was supported in part by grants from the Dr Mildred Stiftung, Deutsche Krebshilfe e.V., Germany (NH), the Finnish Foundation for Studies on Alcohol, the Yrjö Jahnsson Foundation and the Sigrid Juselius Foundation.

References

1
La Vecchia,C., Tavani,A., Franceschi,S., Levi,F., Corrao,G. and Negri,E. (
1997
) Epidemiology and prevention of oral cancer.
Eur. J. Cancer Oral Oncol.
 ,
33
,
302
–312.
2
Blot,W.J., Mc Laughlin,J.K., Winn,D.M., Austin,D.F., Greenberg,F.S., Preston-Martin,S., Bernstein,L., Schoenberg,J.B. Stemhagen,A. and Fraumeni,J.F.Jr (
1988
) Smoking and drinking in relation to oral and pharyngeal cancer.
Cancer. Res.
 ,
48
,
3282
–3287.
3
Brugere,J., Guenel,P., Leclerc,A. and Rodriquez,J. (
1986
) Differential effects of tobacco and alcohol in cancer of the larynx, pharynx and mouth.
Cancer
 ,
57
,
391
–395.
4
Franceschi,S., Talamini,R., Barra,S., Baron,A.E., Negri,E., Bidoli,E., Serraino,D. and La Vecchia,G. (
1990
) Smoking and drinking in relation to cancers of the oral cavity, pharynx, larynx and esophagus in Northern Italy.
Cancer. Res.
 ,
50
,
6502
–6507.
5
IARC (1986) IARC Monographs on the Evaluation of Carcinogenic Risks to Humans, Vol. 38, Tobacco Smoking. IARC, Lyon.
6
IARC (1988) IARC Monographs on the Evaluation of the Carcinogenic Risk of Chemicals to Humans, Vol. 44, Alcohol Drinking. IARC, Lyon.
7
IARC (1985) IARC Monographs on the Evaluation of the Carcinogenic Risk of Chemicals to Humans, Vol. 36, Allyl Components, Aldehydes, Epoxides and Peroxides. IARC, Lyon, pp. 101–132.
8
Pikkarainen,P.H., Baraona,E., Jauhonen,P., Seitz,K.H. and Lieber,C.S. (
1981
) Contribution of oropharynx flora and of lung microsomes to acetaldehyde in expired air after alcohol ingestion.
J. Lab. Clin. Med.
 ,
97
,
631
–636.
9
Miyakawa,H., Baraona,E., Chang,J.C., Lesser,M.D. and Lieber,C.S. (
1986
) Oxidation of ethanol to acetaldehyde by bronchopulmonary washings: role of bacteria.
Alcohol. Clin. Exp. Res.
 ,
10
,
517
–520.
10
Homann,N., Jousimies-Somer,H., Jokelainen,K., Heine,R. and Salaspuro,M. (
1997
) High acetaldehyde levels in saliva after ethanol consumption: methodological aspects and pathogenetic implications.
Carcinogenesis
 ,
18
,
1739
–1743.
11
Jokelainen,K., Heikkonen,E., Roine,R., Lehtonen,H. and Salaspuro,M. (
1996
) Increased acetaldehyde production by mouthwashings from patients with oral cavity, laryngeal or pharyngeal cancer.
Alcohol. Clin. Exp. Res.
 ,
20
,
1206
–1210.
12
Marshall,J.R. and Boyle,P. (
1996
) Nutrition and oral cancer.
Cancer Causes Control
 ,
7
,
101
–111.
13
Elmore,J.G. and Horwitz,R.I. (
1995
) Oral cancer and mouthwash use: evaluation of the epidemiologic evidence.
Otolarnygol. Head Neck Surg.
 ,
113
,
253
–261.
14
Bundgaard,T., Wildt,J., Frydenberg,M., Elbrond,O. and Nielsen,J.E. (
1995
) Case-control study of squamous cell cancer of the oral cavity in Denmark.
Cancer Causes Control
 ,
6
,
57
–67.
15
Marshall,J.R., Graham,S., Haughey,B.P., Shedd,D., O'Shea,R., Brasure,J., Wilkinson,G.S. and West,D. (
1992
) Smoking, alcohol, dentition and diet in the epidemiology of oral cancer.
Eur. J. Cancer Oral Oncol.
 ,
28B
,
9
–15.
16
Graham,S., Dayal,H., Rohrer,T., Swanson,M., Sultz,H., Shedd,D. and Fischman,S. (
1977
) Dentition, diet tobacco and alcohol in the epidemiology of oral cancer.
J. Natl Cancer Inst.
 ,
59
,
1611
–1618.
17
Maier,H., Zöller,J., Herrmann,A., Kreiss,M. and Heller,W.D. (
1993
) Dental status and oral hygiene in patients with head and neck cancer.
Otolaryngol. Head Neck Surg.
 ,
108
,
655
–661.
18
Mashberg,A., Bofetta,P., Winkelman,R. and Garfinkel,L. (
1993
) Tobacco smoking, alcohol drinking and cancer of the oral cavity and oropharynx among U.S. veterans.
Cancer
 ,
72
,
1369
–1375.
19
Bofetta,P., Mashberg,A., Winkelmann,R. and Garfinkel,L. (
1992
) Carcinogenic effect of tobacco smoking and alcohol drinking on anatomic sites of the oral cavity and oropharynx.
Int. J. Cancer
 ,
52
,
530
–533.
20
Helander,A. and Lindahl-Kiessling,K. (
1991
) Increased frequency of acetaldehyde-induced sister-chromatid exchanges in human lymphocytes treated with an aldehyde dehydrogenase inhibitor.
Mutat. Res.
 ,
264
,
103
–107.
21
Dellarco,V.L. (
1988
) A mutagenicity assessment of acetaldehyde.
Mutat. Res.
 ,
195
,
1
–20.
22
Obe,G., Jonas,R. and Schmidt,S. (
1986
) Metabolism of ethanol in vitro produces a compound which induces sister-chromatid exchanges in human peripheral lymphocytes in vitro: acetaldehyde not ethanol is mutagenic.
Mutat. Res.
 ,
174
,
47
–51.
23
Bird,R.P., Draper,H.H. and Basrur,P.K. (
1982
) Effects of malonaldehyde and acetaldehyde on cultured mammalian cells. Production of micronuclei and chromosomal aberrations.
Mutat. Res.
 ,
101
,
237
–246.
24
Homann,N., Kärkkäinen,P., Koivisto,T., Nosova,T., Jokelainen,K. and Salaspuro,M. (
1997
) Effect of acetaldehyde on cell regeneration and differentiation of the upper gastrointestinal tract mucosa.
J. Natl Cancer Inst.
 ,
89
,
1692
–1697.
25
Harty,I.C., Caporaso,N.E., Hayes,R.B., Winn,D.M., Bravo-Otero,E., Blot,W.J., Kleinman,D.W., Brown,L.M., Armenian,H.K., Fraumeni,J.F.Jr and Shields,P.G. (
1997
) Alcohol dehydrogenase 3 genotype and risk of oral cavity and pharyngeal cancers.
J. Natl Cancer Inst.
 ,
89
,
1698
–1705.
26
Yokoyama,A., Muramutsu,T., Ohmori,T., Makuuchi,H., Higuchi,S., Matsushita,S., Yoshino,K., Maruyama,K., Nakano,M. and Ishii,H. (
1996
) Multiple primary esophageal and concurrent upper aerodigestive tract cancer and the aldehyde dehydrogenase type-2 genotype of Japanese alcoholics.
Cancer
 ,
77
,
1986
–1990.
27
Yokoyama,A., Ohmori,T., Muramatsu,T., Higuchi,S., Yokoyama,T., Matsushita,S., Matsumoto,M., Maruyama,K., Hayshida,M. and Ishii,H. (
1996
) Cancer screening of upper aerodigestive tract in Japanese alcoholics with reference to drinking and smoking habits and aldehyde dehydrogenase-2.
Int. J. Cancer
 ,
68
,
313
–316.
28
Yokoyama,A., Muramatsu,T., Ohmori,T., Higuchi,S., Hayashida,M. and Ishii,H. (
1996
) Esophageal cancer and aldehyde dehydrogenase-2 genotypes in Japanese males.
Cancer Epidemiol. Biomarkers Prev.
 ,
5
,
99
–102.
29
Yokoyama,A., Muramatsu,T., Ohmori,T., Yokoyama,T., Okuyama,K., Takahashi,H., Hasegawa,Y., Higuchi,S., Maruyama,K., Shirakura,K. and Ishii,H. (
1998
) Alcohol-related cancers and aldehyde dehydrogenase-2 in Japanese alcoholics.
Carcinogenesis
 ,
19
,
1383
–1387.
30
Jokelainen,K., Siitonen,A., Jousimies-Somer,H., Nosova,T., Heine,R. and Salaspuro,M. (
1996
) In vitro alcohol dehydrogenase-mediated acetaldehyde production by aerobic bacteria representing the normal colonic flora in man.
Alcohol. Clin. Exp. Res.
 ,
20
,
967
–972.
31
Maier,H., Born,I.A., Veith,S., Adler,D. and Seitz,H.K. (
1986
) The effect of chronic ethanol consumption on salivary gland morphology and function in the rat.
Alcohol. Clin. Exp. Res.
 ,
10
,
425
–429.
32
Rivera-Hidalgo,F. (
1986
) Smoking and periodontal diseases.
J. Periodontol.
 ,
57
,
617
–624.
33
Stoltenberg,J.L., Osborn,J.B., Pihlstrom,B.L., Herzberg,M.C., Aeppli,D.M., Wolff,L.F. and Fischer,G.E. (
1993
) Association between cigarette smoking, bacterial pathogens and peridontal status.
J. Periodontol.
 ,
64
,
1225
–1230.
34
McGregor,I.D.M. (
1988
) Smoking, saliva and salivation.
J. Dent.
 ,
18
,
14
–17.
35
Arendorf,T.M., Walker,D.M., Kingdom,R.J., Roll,J.R.S. and Newcombe,R.G. (
1983
) Tobacco smoking and denture wearing in oral candidal leukoplakia.
Br. Dent. J.
 ,
155
,
340
–343.
36
Arendorf,T.M. and Walker,D.M. (
1980
) The prevalence and intra oral distribution of Candida albicans in man.
Arch. Oral Biol.
 ,
25
,
1
–10.
37
Holmstrup,P. and Besserman,M. (
1983
) Clinical, therapeutic and pathogenetic aspects of chronic oral multifocal candiasis.
Oral Surg. Oral Med. Oral Pathol.
 ,
56
,
388
–395.
38
Colman,G., Beighton,D., Chalk,A.J. and Wake,S. (
1976
) Cigarette smoking and the microbial flora of the mouth.
Aust. Dent. J.
 ,
21
,
111
–118.
39
Bardell,D. and Smith,E.J. (
1981
) An in vitro study of mixed populations of normal oropharyngeal bacteria after exposure to cigarette smoke in vitro.
Microbios
 ,
32
,
7
–13.
40
Bastiaan,R.J. and Waite,I.M. (
1978
) Effect of tobacco smoke on plaque development and gingivitis.
J. Periodontol.
 ,
49
,
480
–482.
41
Harris,C.K., Warnakulasuriya,K.A., Johnson,N.W., Gelbier,S. and Peters,T.J. (
1996
) Oral health in alcohol misusers.
Community Dent. Health
 ,
13
,
199
–203.