Variability in ethanol biodisposition in whites is modulated by polymorphisms in the ADH1B and ADH1C genes


  • Potential conflict of interest: Nothing to report.


Association between genetic variations in alcohol-related enzymes and impaired ethanol biodisposition has not been unambiguously proven, and the effect of many newly described polymorphisms remains to be explored. The aims of this study are to elucidate the influence of genetic factors in alcohol biodisposition and effects. We analyzed alcohol pharmacokinetics and biodisposition after the administration of 0.5 g/kg ethanol; we measured ethanol effects on reaction time and motor time in response to visual and acoustic signals, and we analyzed 13 single nucleotide polymorphism (SNPs) in the genes coding for ADH1B, ADH1C, ALDH2, and CYP2E1 in 250 healthy white individuals. Variability in ethanol pharmacokinetics and biodisposition is related to sex, with women showing a higher area under the curve (AUC) (P = 0.002), maximum concentration (Cmax) (P < 0.001) and metabolic rate (P = 0.001). Four nonsynonymous SNPs are related to decreased alcohol metabolic rates: ADH1B rs6413413 (P = 0.012), ADH1C rs283413 (P < 0.001), rs1693482 (P < 0.001), and rs698 (P < 0.001). Individuals carrying diplotypes combining these mutations display statistically significant decrease in alcohol biodisposition as compared with individuals lacking these mutations. Alcohol effects displayed bimodal distribution independently of sex or pharmacokinetics. Most individuals had significant delays in reaction and motor times at alcohol blood concentrations under 500 mg/L, which are the driving limits for most countries. Conclusion: Besides the identification of new genetic factors related to alcohol biodisposition relevant to whites, this study provides unambiguous identification of diplotypes related to variability in alcohol biodisposition. (HEPATOLOGY 2010;51:491–500.)

Effects of alcohol drinking vary among individuals. Part of the interindividual differences in the response to alcohol may be attributed to variability in pharmacokinetics, because a large interindividual variation in alcohol concentration exists, even when subjects receive weight-adjusted doses.1, 2 Because variations in alcohol pharmacokinetics are inheritable,2, 3 it has been hypothesized that genetic variations in alcohol-metabolizing enzymes may underlie interindividual variability in the response to alcohol. More than 90% of ingested ethanol is metabolized in humans.4 The primary step of alcohol ethanol metabolism in human liver is oxidation to acetaldehyde by two enzyme systems, namely alcohol dehydrogenase (ADH1) and the microsomal ethanol oxidizing system, with a prominent role of cytochrome P450 2E1 (CYP2E1).5 ADH1 is a dimer, and the monomers are encoded by the genes ADH1A, ADH1B, and ADH1C. Common allelic variants caused by single nucleotide polymorphisms (SNPs) have been reported for the ADH1B and ADH1C genes, giving rise to at least 21 different possible forms of ADH1 composed of homodimers or heterodimers.6 Acetaldehyde is subsequently oxidized to acetate predominantly by the aldehyde dehydrogenase (ALDH) enzyme.5 Because acetaldehyde inhibits ethanol metabolism, the functionality of ALDH might modulate the primary metabolism of ethanol.6 Although genetic variations for each of the mentioned enzymes have been reported, and there exists information concerning the catalytic differences of various forms of ADH and ALDH enzymes in vitro,7, 8 data documenting the pharmacokinetic disposition of ethanol in individuals with known genotypes are scarce and controversial. Initial reports pointed to genetic variation in ADH1B as a relevant factor in ethanol concentration through an effect on the peak level rather than on the rate of metabolism.9 Although some reports indicated that the ADH1B 48His polymorphism was related to increased metabolic rates in vitro,7 few studies limited to some specific populations analyzed such association in vivo and reported positive findings,8, 10 whereas other studies reported negative findings.11–13

Recently it has been shown that after ingestion of a moderate amount of ethanol, individuals carrying variant ALDH2 alleles exhibit increased peak ethanol concentration and area under the concentration-time curve (AUC) as compared with noncarriers of variant alleles. This observation was limited to Asian subjects in an extremely small study group.8

Overall, the effects of polymorphisms in genes coding for alcohol-related enzymes in alcohol pharmacokinetics are poorly understood, especially in white individuals, and still need conclusive studies.14 Most studies addressing the effect of polymorphisms of ethanol-metabolizing enzymes in ethanol metabolism in vivo have low sample sizes and are limited to a few polymorphisms or to a determined enzyme.7–13 In addition, none have analyzed the interaction of polymorphisms for all major enzymes. For instance, the largest study on ADH1 gene variation and alcohol pharmacokinetics analyzed 103 SNPs in ADH1 genes, but it only studied one nonsynonymous SNP in the ADH1B gene and two nonsynonymous SNPs in the ADH1C gene.12 However, several nonsynonymous SNPs have been mapped to ADH1B and ADH1C genes (see the websites and, respectively), and the effect of most of these SNPs on alcohol pharmacokinetics remain to be studied. In addition, no studies addressed the combined analyses of pharmacokinetic variation and alcohol effects, and therefore, whether alterations in alcohol metabolism correlate with interindividual variations in alcohol effects remains unknown.

Aiming to obtain conclusive evidence on the putative influence of genetic polymorphisms on interindividual variability in the response to ethanol, we analyzed ethanol pharmacokinetics and effects as well as 13 intragenic (12 nonsynonymous) SNPs in the genes coding for the major enzymes related to ethanol metabolism, ADH1B, ADH1C, ALDH, and CYP2E1, in a large enough group of white individuals to detect carriers of SNPs occurring with low frequencies. In addition, ethanol effects were analyzed in all participants to examine the association of both pharmacokinetic variability and polymorphisms in the effects of alcohol.


ADH, alcohol dehydrogenase; ALDH, aldehyde dehydrogenase; AUC, area under the concentration-time curve; Cmax, peak concentration; CYP2E1, cytochrome P450 2E1; SNP, single nucleotide polymorphism; Tmax, time to peak concentration.

Patients and Methods

The study group consisted of 250 white Spanish individuals. Table 1 summarizes the characteristics of participants, who were recruited at the Medical School of the University of Extremadura (Badajoz, Spain) among students and staff. More than 95% of individuals who were invited to participate agreed to participate. Data concerning age, sex, personal or familial antecedents of alcoholism, and smoking and drinking habits were collected for all participants. None of the participants had personal antecedents of alcoholism, and none reported familial antecedents of alcoholism. The inclusion criteria were the following: age over 18, absence of consumption of illicit drugs by self-report, and lack of all the exclusion criteria. Exclusion criteria were pregnancy, diabetes mellitus, history of gastrointestinal, liver, or renal disease. Signed informed consent was obtained for all participants. The protocol of the study was approved by the Ethics Committee of the University Hospital Infanta Cristina (Badajoz, Spain).

Table 1. Descriptive Data of Individuals Included in the Current Study
  1. Smokers include ex-smokers. Individuals drinking over 20 g ethanol per day are classified as regular drinkers.

Age (mean ± SD), years22.5 ± 6.2 (range, 19–57)
Sex (men/ women)66/184
Weight (men)78.4 ± 29.3 (range, 56.0–129.0)
Weight (women)60.5 ± 39.6 (range, 44.5–120.0)
Smokers (No., %) 54 (21.6%)
Body mass index (men)24.6 ± 12.9 (range, 19.4–35.6)
Body mass index (women)22.0 ± 12.4 (range 15.2–34.3)
Nondrinkers (No., %)160 (64.0%)
Moderate drinkers (No., %) 87 (34.8%)
Regular drinkers (No., %)  3 (1.2%) 

Before testing, all subjects were briefed on the experimental design and were instructed not to consume ethanol for 7 days before the study. Parameters known to influence the absorption of ethanol, including the time of day, drinking pattern, dosage form, ethanol concentration in the beverage, and the fasting state,4 were identical for all participants. Subjects arrived at the research unit at 8:00 AM after a 12-hour overnight fast. Starting at approximately 8:30 AM, subjects consumed ethanol (0.5 g/kg) over 2 minutes as apple liquor containing 17% ethanol. After alcohol consumption, subjects were asked to rinse their mouths with water to remove ethanol. For 5 hours after administration, or until the ethanol levels were not detectable (under 0.02 g/L), breath samples were measured every 15 minutes using a breath analyzer Alcotest-7710 (Dräger Safety, Spain) to estimate blood ethanol levels as described elsewhere.15 Before ethanol administration, and then every 30 minutes, participants performed a test to measure their reaction time and motor time by using the reaction test in the Vienna Test System 5.20 (Dr. G. Schuhfried GMBH 2003, Austria). Every reaction test consisted of 16 measurements of reaction and motor time in response to complex visual and acoustic signals. The average of the 16 values was used as the reaction and motor times for every measurement.

Pharmacokinetic analyses were carried out using the WinNonlin 1.1 (Scientific Consulting Inc., Apex, NC). Because at moderate doses ethanol follows a zero-order elimination, the pharmacokinetic model chosen to calculate all parameters was one-compartment with extravascular administration and zero-order elimination.4, 16 Rates of ethanol metabolism were calculated based on the linear rate of decrease in breath ethanol concentrations in the elimination phase.17

Blood samples were collected within the week previous to alcohol challenge, and these were immediately frozen after collection and kept at −80°C until analyzed. Genomic DNA was prepared from peripheral leukocytes and dissolved in sterile 10 mM Tris HCl, pH 8.0, 1 mM ethylenediaminetetra-acetic acid at a final concentration of 400 to 600 μg/mL. The samples were stored at 4°C in sterile plastic vials. Genotyping analyses aimed to detect SNPs in genes coding for ethanol-metabolizing enzymes. The polymorphisms tested, selected on the basis of their allele frequencies in white subjects and expected effect in enzyme activity, are the following: For the ADH1B gene, the SNPs rs1229984 His48Arg, rs1041969 Asn57Lys, rs6413413 Thr60Ser, and rs2066702 Arg370Cys were analyzed. For ADH1C, we analyzed the SNPs rs35385902 Arg48His, rs283413 Gly78X, rs34195308 Pro166Ser, rs1693482 Arg272Gln, rs698 Ile350Phe, and rs35719513 Pro352Thr. For the ALDH2 gene, the SNP rs671 Glu487Lys was analyzed, and for the CYP2E1 gene two variant alleles, namely, CYP2E1*2 rs72559710 Arg76His and CYP2E1*5, a variant allele that contains diverse mutations at the gene promoter (see the website were analyzed. For genotyping details, see supporting material.

Statistical analysis was carried out using the SPSS 15.0 statistical package (SPSS Inc., Chicago, IL). The sample size (250 individuals, 500 genes) was designed to identify variant alleles with low frequencies. For continuous variables, the Kolmogorov-Smirnoff test was used to analyze normality in the distribution. Then, the Student two-sample t test was used for variables that followed a normal distribution, namely, peak concentration (Cmax), AUC, and elimination rate, and the Mann-Whitney test was used for the time to peak concentration (Tmax), because no normal distribution was observed for such parameters. Logistic regression with backward selection was used to analyze the association of variant alleles with pharmacokinetic parameters and alcohol effects. For continuous values, linear regression with backward selection was used. Statistical analyses for multiple comparisons were carried out according to Bonferroni's test. The Hardy-Weinberg equilibrium and the linkage disequilibrium analyses were performed with the genetics package of the statistical software R (version 2.4.0).


Interindividual Variability in Ethanol Metabolism and Effects.

Large interindividual variability in all parameters analyzed is observed. Ethanol pharmacokinetic parameters are summarized in Table 2. Of particular relevance are the 5.6-fold interindividual differences in the AUC and the 4.1-fold interindividual differences in the rate of metabolism. When these findings were stratified according to sex and drinking and smoking habits, a statistically significant higher average of Tmax values were observed in women as compared with men (P = 0.031) and in nonsmokers as compared with smokers (P = 0.002). Cmax and AUC values were higher in women than in men (P = 0.002 and P = 0.001, respectively), and the rate of metabolism was higher in women than in men (P = 0.001). Sex accounts for 7.6% of the variability in AUC and for 4.2% of the variability in the ethanol metabolic rate. The rest of the comparisons were not statistically significant regarding sex or smoking or drinking habits. With regard to body mass index, no statistically significant association with any of the pharmacokinetic parameters was observed. Figure 1 shows the frequency distribution Cmax (Fig. 1A), AUC (Fig. 1B), and the rate of metabolism (Fig. 1C). All of these parameters follow a unimodal distribution in the population analyzed.

Table 2. Summary of Ethanol Pharmacokinetic Parameters
inline image
Figure 1.

Frequency histogram for alcohol pharmacokinetics parameters in 250 healthy subjects. (A) Area under the concentration–time curve (AUC); (B) peak concentration (Cmax); (C) rate of alcohol metabolism.

Ethanol effects are summarized in Table 3. In spite of the low Cmax concentrations reached in the current study, significant differences in reaction time and motor time were observed in most subjects when comparing these parameters at Tmax with basal conditions (the average of the results obtained before the administration of ethanol and of the results obtained when ethanol concentrations were less than 0.050 g/L). The average increase in reaction time was 12% (P < 0.001) in overall subjects, 13% in women (P < 0.001) and 11% in men (P = 0.001). The average increase in motor time was approximately 7% (P < 0.001) in overall subjects, and similar increases were observed in women (P = 0.035) and in men (P = 0.387). No sex-related differences were observed in reaction time, but both peak and basal motor times were slower in women than in men (P < 0.001). Figure 2 shows a frequency distribution of the delays in reaction time (Fig. 2A) and motor time (Fig. 2B). Whereas 30 subjects (12.4%) have a moderate improvement in their reaction time at Tmax, most subjects show a delay in their reaction time with a maximum of 386 milliseconds (183% compared with the average basal value). Interestingly, the distribution of the effects of alcohol in reaction time is bimodal. Regarding motor time, a bimodal distribution also can be observed (Fig. 2B). Improvement in the motor times at peak ethanol concentrations was observed in 78 individuals (31.2%), but most subjects have a delay, with a maximum of 170 milliseconds (over 200% compared with basal values). No sex-related differences were observed in the bimodal distribution for reaction or motor times. Ethanol drinking habits did not influence reaction or motor times.

Table 3. Summary of Results Obtained in Reaction and Motor Time Tests
inline image
Figure 2.

Frequency histogram for alcohol effects in 250 healthy subjects. (A) Delay in the reaction time as compared with basal conditions (for each subject, the average of the results obtained before the administration of ethanol and of the results obtained when ethanol concentrations were less than 0.050 g/L); (B) Delay in the motor time as compared with basal conditions.

The extent of delays in the reaction and motor times were not related to the peak ethanol concentration (Pearson's correlation coefficient = 0.373 and 0.170, respectively) or to any other pharmacokinetic parameters analyzed in this study, indicating that factors other than ethanol concentration or pharmacokinetics contribute to the interindividual variability in ethanol effect.

Genotypes of Alcohol Metabolizing Enzymes.

Table 4 summarizes the polymorphism for ethanol-metabolizing enzymes analyzed in the current study. Regarding the ADH1B gene, two of the four SNPs analyzed, namely Asn57Lys and Arg370Cys, were monomorphic in the population study. Because major ADH1B alleles are defined by the presence or absence of the SNPs in positions 48 and 370, and because the SNP in 370 was monomorphic in the population study, the variant alleles ADH1B*1 (Arg48, Arg370) and ADH1B*2 (His48, Arg370), but not ADH1B*3 (Arg48, Cys370), were identified in the population study. In addition, the SNP Thr60Ser proved to be polymorphic in the population study; eight individuals were heterozygous for this SNP, and all of them had the haplotype His48+Ser60. The observed frequency for the Arg48His SNPs is in concordance with that reported for white subjects.18, 19 No previous studies analyzed the SNP Thr60Ser, but the allele frequencies observed in the current study are consistent with those reported in public databases for white individuals (see the website

Table 4. Genotypes Identified in the Population Study
inline image

Regarding the ADH1C gene, three of the six SNPs analyzed, namely Arg48His, Pro166Ser, and Pro352Thr, were monomorphic in the population study (Table 4). Although no published studies have analyzed the occurrence of these SNPs in white subjects, these findings are in agreement with the absence or extremely rare occurrence of these SNPs in public databases for white subjects ( Conversely, three polymorphisms were identified, including Gly78X, which is studied here for the first time, Arg272Gln, and Ile350Phe (designated in the literature as ADH1C*2), which displayed allele frequencies similar to those described in independent studies.20 The ALDH2 SNP Glu487Lys, designated as ALDH2*2, was monomorphic in the population study (Table 4). This SNP is common in Asian subjects21 but is extremely rare in white individuals ( Finally, two CYP2E1 variant alleles designated as CYP2E1*2 and CYP2E1*5 were analyzed. The variant allele CYP2E1*2, which consists of a nucleotide change G1132A, causes the amino acid substitution Arg76His, and is associated with impaired activity,22 was monomorphic in the study population (Table 4), whereas the CYP2E1*5 allele, which contains silent mutations and is associated with increased activity, was polymorphic, with allele frequencies similar to those described for white subjects.23 No sex-related differences were observed in the frequencies for any of the polymorphisms identified in the study. With the exception of the ADH1C SNPs 272 and 350, which were strongly linked (P < 0.001), no linkage was observed between the polymorphisms studied, and no significant departures from Hardy Weinberg's equilibrium were observed for any of the SNPs studied.

Table 5 summarizes the distribution of pharmacokinetic parameters stratified by polymorphisms of ethanol-metabolizing enzymes. No differences in any of the parameters were identified regarding the ADH1B Arg48His polymorphism. Subjects carrying ADH1B Ser60 showed lower metabolic rates as compared with nonmutated control subjects (P = 0.012, t test). However, the difference in metabolic rates between genotypes does not appear to correlate with other parameters such as Tmax, Cmax, and AUC (Table 5). Thus, the physiological importance of ADH1B Ser60 is uncertain. The most striking differences in ethanol pharmacokinetics are related to the ADH1C polymorphisms (Table 5). Carriers of the ADH1C 78X, ADH1C 272Gln, or ADH1C 350Phe alleles display lower rates of ethanol metabolism as compared with noncarriers of the respective polymorphisms (P < 0.001 for all comparisons that remain significant after Bonferroni's correction), thus indicating that the ADH1C polymorphisms identified in this study are relevant to the rate of ethanol metabolism. In contrast, no differences in any of the pharmacokinetic parameters analyzed were observed with regard to CYP2E1 polymorphisms (Table 5). Multiple comparison analyses for all pharmacokinetic parameters were carried out, including the relevant polymorphisms, that is, ADH1B Thr60, ADH1C 78X, ADH1C 272Gln, and ADH1C 350Phe. Backward logistic regression indicates that none of these polymorphisms was related to Tmax or Cmax. The polymorphism ADH1C 350Phe was related to increased AUC (P = 0.01), and the polymorphisms ADH1C 78X and ADH1C 350Phe were related to decreased ethanol metabolic rate (P < 0.001 for both SNPs, with or without correction for sex).

Table 5. Pharmacokinetic Parameters According to ADH1B Genotypes
inline image

Smoking or drinking habits did not influence the association of the ADH1B Thr60, ADH1C 78X, ADH1C 272Gln, or ADH1C 350Phe polymorphisms with the rate of ethanol metabolism. The effects of ethanol on reaction time and motor time did not show correlation with any of the polymorphisms analyzed (see Supporting materials), thus supporting the hypothesis that the interindividual differences in ethanol effects observed in this study are not related to pharmacokinetic parameters.

In Table 5, each of the SNPs has been treated in isolation, but it should be kept in mind that each individual has a diplotype that would be a cassette of combinations of all of the SNPs studied. If we consider the four SNPs that have shown association with ethanol metabolism in the current study, 16 different diplotypes were observed in the study group. Figure 3 shows the ethanol metabolism rate in individuals according the commonest diplotypes. All common variant diplotypes are associated with decreased alcohol metabolic rate. Among common diplotypes, the lowest metabolic rates were observed in individuals that carried simultaneously mutations at three different positions (78, 272, and 350) in the ADH1C gene (diplotype 3 in Fig. 3). Among rare diplotypes, three individuals carried simultaneously the three mutations mentioned in the ADB1C gene plus a mutation at position 60 in the ADH1B gene. These subjects showed an extremely low alcohol metabolic rate with a mean ± standard deviation equal to 98.34 ± 5.43 mg/L/hour (P < 0.001 as compared with noncarriers of mutations).

Figure 3.

Ethanol metabolic rates according the commonest diplotypes for the SNPs related to altered ethanol metabolism. Results correspond to the average ± standard deviation of the rates of alcohol metabolism. Nonmutated: individuals with the following diplotype: ADH1B 60 Thr/Thr + ADH1C 78 Gly/Gly+ ADH1C 272 Arg/Arg + ADH1C 350 Ile/Ile (n = 80 subjects in the study group). Diplotype 1: ADH1B 60 Thr/Thr + ADH1C 78 Gly/Gly + ADH1C 272 Arg/Gln + ADH1C 350 Ile/Phe (n = 77 subjects in the study group). Diplotype 2: ADH1B 60 Thr/Thr + ADH1C 78 Gly/Gly + ADH1C 272 Gln/Gln + ADH1C 350 Phe/Phe (n = 24 subjects in the study group). Diplotype 3: ADH1B 60 Thr/Thr + ADH1C 78 Gly/X + ADH1C 272 Arg/Gln + ADH1C 350 Ile/Phe (n = 15 subjects in the study group). Diplotype 4: ADH1B 60 Thr/Thr + ADH1C 78 Gly/Gly + ADH1C 272 Arg/Arg + ADH1C 350 Ile/Phe (n = 12 subjects in the study group). Other diplotypes: n = 42 individuals. The P values correspond to the two-tailed Student t test for every subgroup as compared with noncarriers of mutations. Crude P values are shown. To correct for the multiple comparison, the significance limit was considered as P < 0.01, and therefore all comparisons remain significant.


Compared with the wide knowledge of interindividual variability in drug metabolism and response, the understanding of the extent and the basis of variability in alcohol metabolism and effects is surprisingly low for the worldwide consumption of alcohol and its health and legal implications. Although it is widely admitted that response to alcohol varies among individuals, even when they receive identical weight-adjusted doses,2, 24 little is known about whether differences in response are attributable to variability in bioavailability, for instance, whether they are caused by variability in absorption or in the first-pass metabolism,25 or whether it is because of variability in the metabolic rates, or even whether variability in the response depends on nonpharmacokinetic parameters.26–34 The understanding of the genetic basis for variability in alcohol metabolism remains elusive, particularly in white subjects. This study addresses this topic, and it has the strength of typing multiple polymorphisms in a large sample size (250 subjects). A major weakness of studies on gene variations with regard to alcohol metabolism is the low minor allele frequencies of many of the SNPs tested. However, with the sample size used in this study, this common weakness is diminished.

Regarding pharmacokinetic parameters, we observed that women had higher Cmax and AUC values than did men. This is not a novel finding, and it is in agreement with findings indicating that the efficiency of first-pass metabolism in women is lower than in men.25 The current study analyzed several putatively relevant SNPs in the genes coding for enzymes related to alcohol metabolism. Many of these were analyzed here for the first time. Overall findings show that four nonsynonymous SNPs are related to the rate of metabolism of alcohol. These are ADH1B Thr60Ser, ADH1C Gly78X, ADH1C Arg272Gln, and ADH1C Ile350Phe, which explain 2.5%, 9.0%, 8.4%, and 12.3% of the variability in the metabolic rate, respectively. Our findings do not support the theoretical models predicting alterations in the AUC, with lowest values in carriers of the ADH1B 48His allele, but are consistent with the model predicting lower AUC in carriers of nonmutated ADH1C enzymes compared with carriers of mutated enzymes.6 The current study confirms the lack of effect of the ADH1B Arg48His in the metabolic rate of alcohol in white subjects, in agreement with independent studies,11–13 and, with the single exception of the polymorphism ADH1B Thr60Ser, it rules out a major effect in vivo for the rest of the ADH1B polymorphisms analyzed. Although consistent evidence has indicated that the ADH1B*3 allele (48 Arg+370 Cys) is associated with variability in alcohol metabolism,35, 36 this allele is specific to African subjects12 and is not relevant to white subjects (Table 4).

In contrast to ADH1B, genetic variation in ADH1C seems to be relevant to alcohol metabolism in whites (Table 5), even after the use of correction for the huge multiple comparison problem presented by the set of data presented in this study. There is little information on the effect in vivo of the gene variants associated with decreased ethanol metabolic rate. For example, the ADH1B Thr60Ser seems to be rather conservative, but in contrast the ADH1C Gly78X likely results in largely dysfunctional enzyme. These gene variants have been described very recently, and no functional in vitro studies have been performed so far. The identification of polymorphisms related to decreased alcohol metabolism in whites carried out in this study is likely to have relevant implications, and not only because of the clear relationship between the polymorphisms and the ethanol elimination rate shown in Table 5: Although no major association of these SNPs was observed with the Cmax and the association with the AUC value is weak, the functional significance of the observed decrease in the metabolic rate associated with variant alleles is of crucial importance, because in this study the alcohol challenge was based on a single dose, whereas alcohol consumption follows a pattern of multiple dosing, and therefore alcohol accumulation and eventually higher concentrations in blood are expected to occur in carriers of the variant allozymes as compared with noncarriers.

No effect of CYP2E1 polymorphisms on alcohol pharmacokinetics or effects was observed in this study. We analyzed the most common CYP2E1 variant allele in the 5′ flanking region, CYP2E1*5, which showed a minor allele frequency equal to 3.4%. Another CYP2E1 polymorphism located in the 5′ flanking region, designated as CYP2E1*D, claimed to be related to enzyme inducibility and to nicotine and alcohol dependence,37–39 was not analyzed because, although CYP2E1*D is frequent in Asian and African populations, it shows a low minor allele frequency in subjects of European descent, and because none of the participants in this study had alcohol dependence.

With regard to alcohol effects, it is of interest to note the lack of association with pharmacokinetic parameters, and especially with Cmax. This study provides novel and relevant information on the effects of low doses of alcohol. Law enforcement agencies use fixed limits of alcohol concentrations for driving that, according to our findings summarized in Fig. 2, may not correspond to the same effect for all individuals. In the European Union, driving limits for most countries are 500 mg ethanol per liter of blood, although in the United Kingdom and Ireland, as well as in most states in the United States the limit reaches 800 mg/L. In the current study, we have shown that most participants had significant delays in their reaction times and motor times at peak ethanol concentrations under 500 mg/L (Fig. 2). Only 21 participants (8.4%) reached alcohol concentrations over 500 mg/L, and only two reached concentrations over 600 mg/L. However, 61 participants (24.4%) experienced at the ethanol peak time a delay in reaction time of over 20% of their basal values, and 130 (48%) experienced at the ethanol peak time a delay in motor time of over 20% of their basal values. This suggests that, at least for some individuals, a 500-mg/L ethanol concentration limit might be too high.

The reason for the variability in alcohol effects, including both the magnitude and the direction of change (improvement or deterioration of performance after alcohol intake) (Fig. 2), remains unknown. The bimodal distributions of the reaction and motor time changes after alcohol challenge are unexpected because the participants got used to the tasks before measurements, and because basal values were taken before and after alcohol use. People that seem to improve at ethanol peak concentrations had not particularly good or bad performance before or after alcohol use, as compared with the rest of the participants. Our findings indicate that for all participants variability in alcohol effects is not related to alterations in alcohol pharmacokinetics. Because alcohol effects are mediated by several receptors, including the type A γ-aminobutyric acid receptor,26 glutamate receptors such as N-methyl-D-aspartate27, 28 or kainate,29 glycine receptors,30, 31 P2X4 receptor,32 or type 3 serotonin receptor,33, 34 it is conceivable that part of the interindividual variability in the alcohol effects observed here may be related to alterations in these receptors, and further studies will be conducted to investigate the basis for such variability.


The authors thank Prof. James McCue for assistance in language editing.