Alcohol consumption, homeostasis model assessment indices and blood pressure in middle-aged healthy men

(2004) 18, 343 350 & 2004 Nature Publishing Group All rights reserved 0950-9240/04 $25.00 www.nature.com/jhh ORIGINAL ARTICLE Alcohol consumption, homeostasis model assessment indices and blood pressure in middle-aged healthy men Y Yamada 1, Y Noborisaka 1, M Ishizaki 1, I Tsuritani 1, R Honda 1 and S Yamada 2 1 Department of Hygiene, Kanazawa Medical University, Uchinada, Ishikawa, Japan; 2 Matsushita Science Center of Industrial Hygiene, Kadoma, Osaka, Japan A total of 1595 middle-aged healthy men consuming alcohol up to 120 ml per day and 538 without alcohol consumption were recruited from an occupational population, and their insulin resistance (IR) and betacell function (BC) were measured using the homeostasis model assessment (HOMA-IR and HOMA-BC), and the associations with alcohol consumption, blood pressure (BP), and serum gamma-glutamyltransferase (GGT) levels were analysed cross-sectionally. Both HOMA-IR and HOMA-BC were decreased with increasing alcohol consumption, but HOMA-BC corresponding to a level of HOMA-IR was 4 10 and 8 20% lower in drinkers consuming less than 60 ml of alcohol per day and those consuming more, respectively, than in nondrinkers, suggesting an altered fasting serum insulin glucose relationship in alcohol consumers. Although BP was higher and HOMA-IR was lower in alcohol consumers than in nonconsumers, BP was higher at higher HOMA- IR irrespective of alcohol consumption. Elevations of serum GGT were positively associated with BP and HOMA-IR in both alcohol consumers and nonconsumers. Multiple regression analyses in the subjects showed that elevated serum GGT was an independent contributor to HOMA-IR elevations, and both serum GGT and HOMA-IR were significantly related to BP elevations after adjusting for alcohol consumption, age, body mass index, cigarette consumption, and physical activity at leisure. Although cross-sectional observations do not provide evidence of causal association, the results suggest that elevated serum GGT in alcohol consumers relates to elevations of IR and that the elevated insulin resistance relates, at least partly, to BP elevations in alcohol consumers. (2004) 18, 343 350. doi:10.1038/sj.jhh.1001678 Published online 22 January 2004 Keywords: alcohol consumption; blood pressure; serum gamma-glutamyltransferase (GGT); homeostasis model assessment (HOMA); insulin resistance Introduction Correspondence: Dr Y Yamada, MD, Department of Hygiene, Kanazawa Medical University, 1-1 Digaku, Uchinada, Ishikawa 920 0293, Japan. E-mail: ulyamada@kanazawa-med.ac.jp Received 17 March 2003; revised 10 November 2003; accepted 17 November 2003; published online 22 January 2004 Although the biological mechanisms involved in elevations of blood pressure (BP) due to alcohol consumption have remained obscure, a significant association between elevations of serum gammaglutamyltransferase (GGT) activity and BP has been observed in alcohol consumers, 1 3 and confirmed by follow-up studies 4 6 and an interventional observation. 7 The findings suggest that alcoholic liver manifestations and the subsequent biological effects may be involved in alcohol-related BP elevations. Elevated serum GGT relates not only to BP elevations but also to other coronary risk factors characteristic of the metabolic syndrome, such as impaired glucose tolerance and dyslipidaemia, even irrespective of alcohol consumption. 8 10 Since insulin resistance (IR) is supposed to be the core pathology in the metabolic syndrome, 11 13 elevated serum GGT in alcohol consumers may also relate to increased IR, and may play a role in BP elevations. Meanwhile, light to moderate alcohol consumption is protective against the development of coronary heart disease (CHD), 14 16 and even against noninsulin-dependent diabetes mellitus (NIDDM) 17,18 although this is still disputed. 19 The protective effects of moderate alcohol consumption on CHD have been attributed to elevated serum HDL-cholesterol levels, suppressed coagulation capacity, and/or the role of antioxidant substances contained in alcoholic beverages. 20,21 More recently, beneficial effects on IR have been proposed as another possible mechanism, 22 implying that improved IR may suppress the metabolic syndrome in moderate alcohol consumers, and thus CHD. This

344 hypothesis is mainly based on the findings of low insulin levels in the fasting blood of moderate alcohol consumers. 23 26 From these conflicting contentions on the relation between IR and alcohol consumption and the diverse effects on health, including hypertension, we aimed to clarify the associations of IR with BP elevations in light to moderate regular alcohol consumers who are often found in a healthy population. The exact measurement of IR, such as using an insulin clamp method or an intravenous glucose load test, however, is not applicable to population-based studies. In this study, therefore, IR and the pancreas beta-cell function (BC) of secreting insulin were evaluated by the homeostasis model assessment (HOMA) proposed by Matthews et al 27 in 1985. The associations of the HOMA indices with alcohol consumption, serum GGT, and BP were analysed cross-sectionally in middle-aged healthy men recruited from an occupational population. Subjects and methods HOMA indices and BP in alcohol consumers The target subjects were selected from 2656 male workers aged between 35 and 64 years in an electronic-parts factory who participated in an annual health check-up programme, and comprised 98.3% of the male workers of that age range in this factory. Among them, we excluded 276 men who had been confirmed by the health records to have diseases possibly affecting the study results, such as cardiovascular disease, chronic liver disease, renal disease, and dyslipidaemia, diabetes mellitus and hypertension being treated with medicines. Also, 121 men who did not have records of those disease but did show body mass index (BMI) of 30 or higher, or BP of 180/110 mmhg or higher, 28 or fasting serum glucose of 126 mg/dl (7.0 mmol/l) or higher 29 in the health check-ups, were excluded so as to avoid the strong effects of marked obesity, severe hypertension, and latent diabetes mellitus on the study results. Owing to incompleteness of the measurements in the health check-ups, an additional 126 men were excluded as well. Totally, 2133 middleaged men (78.9% of the total male workers of that age range) were defined as healthy subjects. Written informed consent was obtained from all the subjects. The subjects were measured in the morning after fasting 12 h or longer for height and body weight in light clothes with the shoes removed, and the BMI was calculated. Systolic and diastolic BP were measured with a sphygmomanometer in the sitting position after resting on a chair for 5 min or longer by experienced nurses. Serum sample was separated soon after blood was obtained by venous puncture and kept frozen at 201C until the analyses. The activity of serum GGT (U/l) was measured using an automatic analyser (HITACH 7450, Hitachi, Japan) as well as the other serum hepatic enzymes activities and lipid concentrations. Fasting serum glucose concentration (mmol/l) was also measured with a hexokinase/glucose-6-phosphate dehydrogenase method using the automatic analyser. Serum insulin concentration (mu/ml) was measured with an enzyme immunoassay using an automatic analyser, IMx (Dinabot, Japan). The HOMA indices of insulin resistance (HOMA-IR) and beta-cell function (HOMA-BC) were calculated according to the following formulae: HOMA-IR ¼ (insulin glucose)/ 22.5 and HOMA-BC (%) ¼ 20 insulin/(glucose 3.5). All the subjects showed serum glucose concentration above 3.5 mmol/l, thus permitting calculation of HOMA-BC, and serum insulin concentration above the detectable limit of this method of 0.8 mu/ml. The data of serum GGT, insulin, and the HOMA indices were logarithmically transformed for the statistical analyses. The data of alcohol and cigarette consumption and physical activity at leisure were obtained by a questionnaire and confirmed by experienced nurses. The average volume of ethanol consumed per day in the subjects was calculated from the data of usual alcohol consumption during the preceding year, that is, the frequency of alcohol consumption, the kinds of beverages, and the volume consumed on a single occasion, excluding the subjects who consume alcohol less than once a month as well as abstainers and teetotalers who were classified as nondrinkers. In all, 10 subjects who had quit drinking (exdrinkers) were also included in the nondrinkers. The subjects who consume alcohol more than once a month were classified into four categories according to the average volume of alcohol consumed per day: less than 30, 30 59, 60 89, and 90 ml or more. Cigarette consumption was classified into five categories as nonsmokers, ex-smokers, current smokers consuming less than 1 pack a day, and those consuming less than two packs, and those consuming two packs or more. The subjects were also classified into four categories of physical activity at leisure, namely doing any kind of exercise lasting 30 min or longer not more often than once a month, doing the exercise once a week or less, 2 4 times a week, 5 times or more a week. The differences in the means of age and BMI in the subjects divided by the five categories of alcohol consumption were tested by a one-way analysis of variance (ANOVA), and those in the frequencies of current smokers and physically inactive subjects in the five alcohol consumption categories were tested by a w 2 -test. The means of systolic and diastolic BP, serum GGT, glucose and insulin, and the HOMA indices in the five categories of alcohol consumption after adjusting for age, BMI, cigarette consumption, and physical activity were calculated and tested for differences by a generalized linear model (GLM) analysis. The means of HOMA-BC corresponding to HOMA-IR levels, those of systolic BP corresponding to the HOMA indices, and those of systolic BP and the HOMA indices corresponding to serum GGT levels in nondrinkers, relatively light

drinkers consuming alcohol less than 60 ml per day, and heavy drinkers consuming more, adjusting for the confounding factors, were also calculated and tested by GLM analyses. Multiple regression analyses were performed to test the independent contribution of serum GGT to the HOMA indices and those of serum GGT and the HOMA indices to elevations of systolic BP in the subjects after adjusting for alcohol consumption together with the other confounding factors. A program package of SPSS version 11.0 for Windows (SPSS Japan, Tokyo) was used for those statistical analyses, and Po0.05 was defined as significance. Results In all, 1595 subjects (74.8%) were defined as alcohol consumers and 538 (25.2%) as nonconsumers. Of the former, 479 (22.5%) consume less than 30 ml of alcohol per day, 581 (27.2%) 30 59 ml, 375 (17.6%) 60 89 ml, and 160 (7.5%) 90 ml or more but up to 120 ml per day. For cigarette consumption, 435 (20.4%) are nonsmokers, 417 (19.5%) are ex-smokers, 822 (38.5%) consume less than one pack per day, 441 (20.7%) less than two packs, and 18 (0.8%) two packs or more. For physical activity at leisure, 1495 (70.1%) exercise not more often than once a month, 365 (17.1%) once a week or less, 199 (9.3%) 2 4 times a week, and 74 (3.4%) 5 times a week or more. The means and standard deviations of age and BMI in the subjects divided by the five categories of alcohol consumption with the frequencies of current smokers and physically inactive subjects who exercise not more often than once a month are shown in Table 1. No significant differences in the means were detected by a one-way ANOVA. Smokers were found more in heavier alcohol consumers, while physically inactive subjects were found more in heavier drinkers and nondrinkers, and those differences were significant by w 2 -tests (Po0.001). Table 2 shows the means and standard errors of HOMA indices and BP in alcohol consumers systolic and diastolic BP, serum GGT, glucose and insulin, and the HOMA indices in the subjects of the five alcohol consumption categories calculated by a GLM analysis adjusting for age, BMI, cigarette consumption, and physical activity. The means of BP and serum GGT increased linearly with increasing alcohol consumption, and those of serum glucose were higher in drinkers consuming 60 ml or more of alcohol per day. In contrast, serum insulin, HOMA-IR, and HOMA-BC decreased with increases in alcohol consumption. All the differences in the parameters in the five categories were significant (Po0.001). Figure 1 illustrates the means of HOMA-BC corresponding to HOMA-IR levels and those of systolic BP corresponding to HOMA-IR and HOMA-BC levels separately in nondrinkers, relatively light drinkers consuming less than 60 ml of alcohol per day, and heavy drinkers consuming more calculated by GLM analyses adjusting for the confounding factors, and tested for the differences in the means of the three alcohol consumption levels. The five (I V) subgroups of HOMA-IR and HOMA-BC were defined according to the quintile of those distributions, that is, low to high 20% each, in the whole subjects. The means of HOMA-BC were elevated with elevations of HOMA-IR in both alcohol consumers and nonconsumers, but the means were 4 10 and 8 20% lower in the light and heavy drinkers, respectively, than in nondrinkers at a level of HOMA-IR, showing different relationships between fasting serum insulin and glucose depending on the volume of alcohol consumed. The means of systolic BP were 3 4 mmhg and 6 8 mmhg higher in the light and heavy drinkers, respectively, than in nondrinkers at any level of HOMA-IR, but were elevated with elevations of HOMA-IR irrespective of alcohol consumption. The differences in systolic BP between the lowest and the highest HOMA-IR groups were roughly 5 mmhg in both alcohol consumers and nonconsumers. Also, at any level of HOMA-BC, systolic BP was signifi- 345 Table 1 Means and standard deviations (s.d.) of age and BMI, and numbers (N) and percentage of current smokers and physically inactive subjects in middle-aged healthy men divided by the average volume of alcohol consumed per day Nondrinker Less than 30 ml 30 59 ml 60 89 ml 90 ml or more n ¼ 538 n ¼ 479 n ¼ 581 n ¼ 375 n ¼ 160 Parameters Mean s.d. Mean s.d. Mean s.d. Mean s.d. Mean s.d. Age (year) a 44.4 6.60 44.4 6.97 44.9 6.27 45.7 6.29 44.7 6.14 BMI a 23.1 2.76 23.0 2.58 22.9 2.29 23.2 2.46 23.1 2.52 N % N % N % N % N % Current smoker b 303 56.3 250 52.2 356 61.3 254 67.7 118 73.8 Physically inactive b,c 416 77.3 297 62.0 390 67.1 273 72.8 119 74.4 a The means of age and BMI were not significantly different (PX0.10) in the five alcohol consumption categories tested by a one-way ANOVA. b The frequencies of current smokers and physically inactive subjects were both significantly different (Po0.001) in the five alcohol consumption categories tested by a w 2 -test. c Subjects who do physical exercise at leisure time not more often than once a month.

HOMA indices and BP in alcohol consumers 346 Table 2 Means and standard errors (s.e.) of blood pressure, serum GGT activity, fasting serum glucose and insulin, and the HOMA indices in middle-aged healthy men divided by the average volume of alcohol consumed per day, adjusting for the effects of age, BMI, cigarette consumption, and physical activity at leisure Nondrinker Less than 30 ml 30 59 ml 60 89 ml 90 ml or more n ¼ 538 n ¼ 479 n ¼ 581 n ¼ 375 n ¼ 160 Parameters Mean s.e. Mean s.e. Mean s.e. Mean s.e. Mean s.e. SBP (mmhg) a 115.8 0.53 117.6 0.57 118.4 0.51 120.9 0.64 123.9 0.98 DBP (mmhg) a 72.9 0.41 74.5 0.44 75.1 0.40 76.5 0.50 78.7 0.76 GGT (U/l) a,b 17.8 1.03 21.6 1.03 25.8 1.02 34.7 1.03 37.4 1.05 Glucose (mmol/l) a 4.86 0.04 4.92 0.04 4.90 0.03 5.06 0.04 5.01 0.05 Insulin (mu/ml) a,b 4.85 1.02 4.53 1.03 4.23 1.02 4.07 1.03 3.96 1.04 HOMA-IR a,b 1.05 1.02 0.99 1.02 0.93 1.02 0.92 1.03 0.88 1.05 HOMA-BC (%) a,b 74.6 1.03 66.4 1.02 62.7 1.02 53.7 1.03 54.8 1.05 a The increases in BP, GGT, and glucose and the decreases in serum insulin and the HOMA indices with increasing alcohol consumption were all statistically significant (Po0.001). b Geometric means and geometric standard errors. Figure 1 Means of HOMA-BC at five levels of HOMA-IR and those of systolic BP at five levels of HOMA-IR and HOMA-BC adjusted for confounders in the subjects of different alcohol consumption levels. The I V levels of HOMA-IR and HOMA-BC correspond to the quintiles of the distributions, low to high in 20% increments, in the subjects. Open circles represent nondrinkers, closed circles drinkers consuming alcohol less than 60 ml per day, and closed triangles drinkers consuming more. cantly higher in alcohol consumers than in nonconsumers, and higher in heavy drinkers than in light drinkers. Although systolic BP seemed markedly higher in alcohol consumers than in nonconsumers at lower levels of HOMA-BC but not so at higher levels, no significant association of HOMA- BC with systolic BP was detected in the whole subjects, and the interactive effect of alcohol consumption and HOMA-BC on systolic BP was not significant (P ¼ 0.08). Figure 2 illustrates the means of systolic BP, HOMA-IR, and HOMA-BC in alcohol consumers and nonconsumers in the five subgroups of serum GGT levels based on the quintile of the distribution in the whole subjects; group I: up to 13 U/l, II: 14 17 U/l, III: 18 25 U/l, IV: 26 39 U/l, V: 40 U/l or above. Serum GGT of 40 U/l just corresponded to the 95% upper limit value obtained in nondrinkers with BMI less than 25. Systolic BP was elevated with increasing serum GGT in both alcohol consumers and nonconsumers, although it was significantly higher in alcohol consumers than in nonconsumers. The differences in BP between alcohol consumers and nonconsumers and between light and heavy alcohol consumers seemed more marked in higher serum GGT levels, but the interactive effect between alcohol consumption and serum GGT on systolic BP was not significant (P ¼ 0.19). The means of HOMA- IR corresponding to any serum GGT level were lower in alcohol consumers than in nonconsumers, but HOMA-IR was elevated with increasing serum GGT in both alcohol consumer and nonconsumers. On the other hand, the means of HOMA-BC were also lower in alcohol consumers than in nonconsumers with no significant association with serum GGT. The results of multiple regression analyses on the determinants of the HOMA indices in the whole subjects are shown in Table 3. Possible independent variables selected here were age, BMI, alcohol and cigarette consumption, physical activity, and serum GGT. For alcohol and cigarette consumption and physical activity, the subjects were scored 0 4 or 0 3 according to these categories. As shown in the

HOMA indices and BP in alcohol consumers 347 Figure 2 Means of systolic BP, HOMA-IR, and HOMA-BC at five levels of serum GGT adjusted for confounders in the subjects of different alcohol consumption levels. The I V levels of serum GGT correspond to the quintiles of the distribution in the subjects; I: up to 13 U/l, II: 14 17 U/l, III: 18 25 U/l, IV: 26 39 U/l, V: 40 U/l or above. Open circles represent nondrinkers, closed circles drinkers consuming alcohol less than 60 ml per day, and closed triangles drinkers consuming more. Table 3 Results of multiple regression analyses on the determinants of HOMA-IR and HOMA-BC in middle-aged healthy men HOMA-IR b HOMA-BC b Determinants b c t-value P b c t-value P Age (year) 0.056 2.90 0.004 0.132 6.60 o0.001 BMI 0.417 20.90 o0.001 0.319 15.40 o0.001 Alcohol consumption a 0.126 6.00 o0.001 0.183 8.33 o0.001 Cigarette consumption a 0.064 3.23 0.001 0.002 0.10 0.919 Physical activity a 0.090 4.66 o0.001 0.061 3.03 0.002 Serum GGT (U/l) b 0.093 4.28 o0.001 0.009 0.39 0.700 R 2 ¼ 0.22 (Po0.001) R 2 ¼ 0.16 (Po0.001) a The subjects were scored 0 4 for alcohol and cigarette consumption and 0 3 for physical activity. b Logarithmically transformed. c Standardized partial regression coefficient. table, BMI was a major determinant of elevations of both HOMA-IR and HOMA-BC. Increases in age, alcohol consumption, and physical activity were related to decreases in both HOMA-IR and HOMA- BC, and cigarette consumption was related to decreases in HOMA-IR. Elevation of serum GGT was shown to be a significant independent determinant of increases in HOMA-IR, but not of HOMA-BC. The determinants of systolic BP elevations in the subjects were also evaluated by multiple regression analyses (Table 4) in three models; model A included age, BMI, alcohol and cigarette consumption, physical activity, and serum glucose and insulin as the independent variables, model B included serum GGT as an independent variable together with those variables, and model C included HOMA-IR instead of serum glucose and insulin. In model A, a great contribution to systolic BP elevations was made by serum glucose and alcohol consumption, with BMI, age, and serum insulin also related to systolic BP elevations. In model B, serum GGT showed a great contribution following serum glucose, and the contribution of alcohol consumption to systolic BP became smaller. In model C, HOMA-IR showed a significant independent contribution to BP elevations, but the contribution was not as great as those of serum glucose and serum GGT. Discussion Changes in insulin resistance may be involved in the development of both favourable and unfavourable cardiovascular effects of alcohol consumption. However, the knowledge available from population-based studies on this issue has been very limited, mainly because of technical difficulties in assessing IR. The HOMA indices are calculated from glucose and insulin concentrations in the fasting blood, and are thus applicable to population studies. The HOMA indices were originally obtained by repeated measurements of glucose and insulin in the blood in a limited number of subjects, 27 while the present HOMA values were obtained by a single measurement in a large number of subjects. Therefore, although a certain validity of the HOMA indices measured in populations as approximates of real IR and BC can be expected from the findings

HOMA indices and BP in alcohol consumers 348 Table 4 Results of multiple regression analyses on the determinants of systolic BP in middle-aged healthy men: model A including age, BMI, alcohol consumption, and serum glucose and insulin as the independent variables; model B including serum GGT together with these variables; and model C including HOMA-IR instead of serum glucose and insulin Model A Model B Model C Determinants b c t-value P b c t-value P b c t-value P Age (year) 0.094 4.50 o0.001 0.094 4.55 o0.001 0.115 5.60 o0.001 BMI 0.135 5.92 o0.001 0.106 4.58 o0.001 0.096 4.11 o0.001 Alcohol consumption a 0.161 7.62 o0.001 0.109 4.81 o0.001 0.128 5.63 o0.001 Cigarette consumption a 0.016 0.77 0.443 0.028 1.35 0.176 0.035 1.66 0.098 Physical activity a 0.030 1.47 0.142 0.034 1.66 0.097 0.035 1.68 0.093 Serum glucose (mmol/l) 0.171 7.94 o0.001 0.159 7.42 o0.001 Serum insulin (mu/ml) b 0.066 2.78 0.006 0.058 2.47 0.014 Serum GGT (U/l) b 0.138 5.95 o0.001 0.149 6.42 o0.001 HOMA-IR b 0.125 5.39 o0.001 R 2 ¼ 0.12 (Po0.001) R 2 ¼ 0.13 (Po0.001) R 2 ¼ 0.10 (Po0.001) a The subjects were scored 0 4 for alcohol and cigarette consumption and 0 3 for physical activity. b Logarithmically transformed. c Standardized partial regression coefficient. in the previous studies, 30,31 the considerable limitation of this methodology should be kept in mind when interpreting the present study results. The associations of alcohol consumption with the HOMA indices have been examined by Kiechl et al 25 in 820 middle-aged men and women who consumed up to 100 g or more of ethanol per day, and where both HOMA-IR and HOMA-BC, as well as serum insulin, decreased dose-dependently with increasing alcohol consumption, while BP and glucose were elevated with increasing alcohol consumption. Their findings are in good accordance with the present results in middle-aged healthy men who consume up to 120 ml of alcohol per day. Kiechl et al, however, did not analyse the association between the HOMA indices and BP elevations in alcohol consumers. The present study added new findings, namely that the relationship between HOMA-IR and HOMA-BC differed between alcohol consumers and nonconsumers, and that although HOMA-IR was lower in alcohol consumers, the significant positive association with BP remained similar to that in nonconsumers, except that BP at any given level of HOMA-IR was much higher in alcohol consumers. The associations of HOMA-IR with BP in alcohol consumers and nonconsumers beg the question as to whether the much higher BP at a given level of HOMA-IR in alcohol consumers is a reflection of the pressor effects of alcohol other than those related to IR, or whether the HOMA-IR values in alcohol consumers in fact underestimate the real IR. We could not exclude the latter possibility because alcohol consumption not only lowered serum insulin levels resulting in lower HOMA-IR but also caused lower HOMA-BC, that is, 4 10 and 8 20% lower HOMA-BC in light and heavy drinkers, respectively, than in nondrinkers at a given level of HOMA-IR, mainly due to higher serum glucose in heavier drinkers. The biological implications of the low HOMA-IR along with high BP and low HOMA-BC in alcohol consumers remain obscure at present. However, apart from the differences in the absolute values of HOMA-IR and HOMA-BC between alcohol consumers and nonconsumers and between light and heavy alcohol consumers, systolic BP was always higher in those showing higher HOMA-IR. The findings suggest that BP elevations may depend greatly on high IR even in alcohol consumers who show a low HOMA-IR, if HOMA-IR reflects IR in the subjects having each level of alcohol consumption, that is, nondrinkers, light, or heavy drinkers, and that increased IR may be a major causative factor of BP elevations in alcohol consumers, if the values of HOMA-IR in alcohol consumers underestimate their real IR. The present study also showed that elevations of serum GGT were related to elevations of HOMA-IR both in alcohol consumers and nonconsumers and in both light and heavy drinkers, although HOMA- IR at a serum GGT level was higher in nondrinkers than in drinkers and higher in light drinkers than in heavy drinkers. If elevated HOMA-IR in the subjects of each alcohol consumption level reflects relatively higher IR, the findings are consistent with the previous findings of a significant association of elevated serum GGT with coronary risk factors characteristic of the metabolic syndrome including hypertension, irrespective of alcohol consumption. 8 10 The results of multiple regression analyses suggested that elevated serum GGT was a significant independent determinant of the elevation of HOMA- IR, although the contribution of serum GGT was smaller than that of increased BMI to HOMA-IR, and that elevated HOMA-IR made a significant contribution to systolic BP elevation, although it was not as great as those of serum glucose and serum GGT. The statistically small contribution of serum GGT to

HOMA-IR and that of HOMA-IR to systolic BP elevations may be due to the lower HOMA-IR in alcohol consumers and may not reflect the real strength of the biological association of insulin resistance with higher serum GGT and higher BP in alcohol consumers. According to Arima et al, 32 fasting serum insulin was significantly associated with the development of hypertension only in nondrinkers but not in drinkers recruited from Japanese community adults of both sexes. Such interactive effects on BP elevation were found neither between alcohol consumption and HOMA- IR nor between alcohol consumption and serum GGT in the present middle-aged male workers as shown in Figures 1 and 2. The findings of Arima et al do not conflict with the present results because the implications of fasting serum insulin as an indicator of IR are no longer identical between nondrinkers and drinkers, and do not imply any reduction in the strength of the association between IR and BP elevation in alcohol consumers. This type of cross-sectional observation could not provide any evidence of causal association. In addition, familial factors and lifestyle factors other than smoking and physical activity were not inquired into in the present subjects. Further, the HOMA indices are approximates of IR and BC, and the biological linkage underlying the association between the HOMA indices and biological parameters, including BP and serum GGT, remains obscure. Further investigations are thus required to overcome those limitations, but it can be said from the present results that elevated IR in alcohol consumers who show elevated serum GGT may be related, at least to some extent, to their BP elevations. Both higher serum insulin and higher glucose are supposed to be causative agents in the association between IR and high BP. 33 35 In the present subjects, fasting serum glucose showed a greater contribution to systolic BP elevations than HOMA-IR in the multiple regression analysis. In fact, serum glucose was significantly elevated in drinkers consuming 60 ml or more of alcohol per day who showed 5 8 mmhg higher systolic BP, suggesting a role of high serum glucose in BP elevations in heavy drinkers. The findings of higher systolic BP in the heavy drinkers showing lower HOMA-BC may thus be a reflection of the association between higher serum glucose and higher BP in them. However, BP was significantly elevated even in light drinkers consuming less alcohol and not showing elevated serum glucose. Elevated fasting serum glucose is, therefore, not considered to be the sole determinant of alcoholrelated BP elevation. Further investigations should be conducted on the role of serum insulin and glucose in postprandial or post glucose-loaded state with BP elevations in alcohol consumers. The role of alcoholic liver manifestations in triggering alterations in insulin glucose relationships should also be further investigated. HOMA indices and BP in alcohol consumers Acknowledgements This work was supported by a Grant-in-Aid for Scientific Research (C: 13670391; 2001 2002) from the Japan Ministry of Education, Culture, Science, and Technology. References 1 Peterson B et al. Comparison of gamma-glutamyltransferase and other health screening tests in average middle-aged males, heavy drinkers and alcohol nonusers. Scand J Clin Lab Invest 1983; 43: 141 149. 2 Yanai T. Effects of alcohol consumption on blood pressure levels in the Hisayama residents: Part 2. Relationship between time-related changes in blood pressure and those in the level of g-glutamyl transpeptidase. Fukuoka J Med 1986; 77: 624 630. 3 Yamada Y et al. Relationship between serum gammaglutamyl transpeptidase activity, blood pressure and alcohol consumption. J Hum Hypertens 1989; 3: 409 417. 4 Yamada Y et al. Alcohol, high blood pressure, and serum g-glutamyl transpeptidase level. Hypertension 1991; 18: 819 826. 5 Miura K et al. Serum gamma-glutamyl transferase level in predicting hypertension among male drinkers. J Hum Hypertens 1994; 8: 445 449. 6 Lee DH et al. Gamma-glutamyltransferase, alcohol, and blood pressure: a four year follow-up study. Ann Epidemiol 2002; 12: 90 96. 7 Yamada Y et al. Serum g-glutamyl transferase levels and blood pressure falls after alcohol moderation. Clin Exp Hypertens 1997; 19: 249 268. 8 Noborisaka Y, Ikai E, Yamada Y. Serum g-glutamyl transferase levels and coronary risk factors in middleaged male and female nondrinkers. Obesity Res 1995; 4: 405 408. 9 Rantala AO et al. Gamma-glutamyl transpeptidase and the metabolic syndrome. J Intern Med 2000; 248: 230 238. 10 Hashimoto Y, Futamura A, Nakarai H, Nakahara K. Relationship between response of g-glutamyl transpeptidase to alcohol drinking and risk factors for coronary heart disease. Atherosclerosis 2001; 158: 465 470. 11 Reaven GM. Role of insulin-resistance in human disease. Diabetes 1988; 37: 1595 1607. 12 Kaplan NM. The deadly quartet: upper-body obesity, glucose intolerance, hypertriglyceridemia, and hypertension. Arch Intern Med 1989; 149: 1514 1520. 13 DeFronzo RA, Ferrannini E. Insulin resistance. A multifaced syndrome responsible for NIDDM, obesity, hypertension, dyslipidemia, and atherosclerotic cardiovascular disease. Diabetes Care 1991; 14: 173 194. 14 Dyer AR et al. Alcohol consumption and 17-year mortality in the Chicago Western Electronic Company Study. Prev Med 1980; 9: 78 90. 15 Marmot MG, Rose G, Shipley MJ, Thomas BJ. Alcohol and mortality: a U-shaped curve. Lancet 1981; I: 580 583. 16 Colditz GA et al. Moderate alcohol and decreased cardiovascular mortality in an elderly cohort. Am Heart J 1985; 109: 886 889. 349

350 HOMA indices and BP in alcohol consumers 17 Rimm EB et al. Prospective study of cigarette smoking, alcohol use, and the risk of diabetes in men. Br Med J 1995; 310: 555 559. 18 Perry IJ et al. Prospective study of risk factors for development of non-insulin dependent diabetes in middle aged British men. Br Med J 1995; 310: 560 564. 19 Holbrook TL, Barrett-Connor E, Wingard DL. A prospective population-based study of alcohol use, and non-insulin-dependent diabetes mellitus. Am J Epidemiol 1990; 132: 902 909. 20 Rimm EB, Klasty A, Grobbee D, Stampfer MJ. Review of moderate alcohol consumption and reduced risk of coronary heart disease: is the effect due to beer, wine, or spirits? Br Med J 1996; 312: 731 736. 21 Puddey IB, Croft KD. Alcohol, stroke and coronary heart disease. Are there anti-oxidants and pro-oxidants in alcoholic beverages that might influence the development of atherosclerotic cardiovascular disease? Neuroepidemiology 1999; 18: 292 302. 22 McCarty MF. Does regular ethanol consumption promote insulin sensitivity and leanness by stimulating AMP-activated protein kinase? Med Hypotheses 2001; 57: 405 407. 23 Mayer EJ et al. Alcohol consumption and insulin concentrations. Role of insulin in association of alcohol intake with high-density lipoprotein cholesterol and triglycerides. Circulation 1993; 88: 2190 2197. 24 Razay G, Heaton KW. Alcohol consumption and cardiovascular risk factors in middle aged men. Cardiovasc Risk Factor Int J 1995; 5: 200 205. 25 Kiechl S et al. Insulin sensitivity and regular alcohol consumption: large, prospective, cross sectional population study (Bruneck study). Br Med J 1996; 313: 1040 1044. 26 Lazarus R, Sparrow D, Weiss ST. Alcohol intake and insulin levels. The Normative Aging Study. Am J Epidemiol 1997; 145: 909 916. 27 Matthews DR et al. Homeostasis model assessment: insulin resistance and b-cell function from fasting plasma glucose and insulin concentration in man. Diabetologia 1985; 28: 412 419. 28 Guidelines Subcommittee. 1999 World Health Organization International Society of Hypertension Guidelines for the management of hypertension. J Hypertens 1999; 17: 151 183. 29 The Expert Committee on the Diagnosis and Classification of Diabetes Mellitus. Report of the expert committee on the diagnosis and classification of diabetes mellitus. Diabetes Care 1997; 20: 1183 1197. 30 Haffner SM, Miettinen H, Stern MP. The homeostasis model in the San Antonio Heart Study. Diabetes Care 1997; 20: 1087 1092. 31 Ostgren CJ et al. Association between smoking and b- cell function in a non-hypertensive and non-diabetic population. Skaraborg Hypertension and Diabetes Project. Diabetes Med 2000; 17: 445 450. 32 Arima H et al. Alcohol reduces insulin-hypertension relationship in a general population. The Hisayama study. J Clin Epidemiol 2002; 55: 863 869. 33 Ferrannini E et al. Insulin resistance in essential hypertension. N Engl J Med 1987; 317: 350 357. 34 Capaldo B et al. Skeletal muscle is a primary site of insulin resistance in essential hypertension. Metabolism 1991; 40: 1320 1322. 35 Cubeddu LX, Hoffmann IS. Insulin resistance and upper-normal glucose levels in hypertension: a review. J Hum Hypertens 2002; 16 (Suppl 1): S52 S55.