National Public Health Institute, Department of Mental Health and Alcohol Research, P.O. Box 719, FIN-00101 Helsinki, Finland,
1 Department of Forensic Medicine and Sciences, Mie University School of Medicine, Tsu, Japan and
2 Department of Surgery, Helsinki University Central Hospital, Helsinki, Finland
Received 1 April 1999; in revised form 21 June 1999; accepted 13 July 1999
![]() |
ABSTRACT |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
![]() |
INTRODUCTION |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
It is well known that acute alcohol intake leads to decreased levels of testosterone in normal healthy men (Ylikahri et al., 1974). Investigations on the hormonal effects of alcohol in women are complicated by the menstrual cycle and the use of hormonal preparations. Most studies have focused on a particular time in the menstrual cycle and on women not using hormonal preparations, usually with fewer than ten subjects. Probably as a consequence, these studies (McNamee et al., 1979
; Välimäki et al., 1983
; Becker et al., 1988
) have not reported any significant effect of alcohol, although a tendency to higher testosterone concentrations was displayed in one of them (Välimäki et al., 1983
).
We have shown in an earlier report that alcohol ingestion causes an acute elevation of the total testosterone levels in premenopausal women (Eriksson et al., 1994). This effect was found to be more prominent among women using oral contra-ceptives, but was also seen to a lesser degree during the mid-cycle phase among non-users as well. In line with our findings is a recent observation of testosterone elevations in four women using oral contraceptives 14 h after a large (2.0 g/kg) dose of alcohol (Karila et al., 1996
).
In the present study, we focused on the mid-cycle phase. The aim of the study was to confirm our earlier observation and to find out whether the effect of alcohol on testosterone in premenopausal healthy women is dose-dependent. In addition we wanted to focus on the possible mechanism by measuring androstenedione and dehydroepiandrosterone (DHEA), the principal precursors of testosterone in women. Furthermore, we wanted to find out how the observed acute effect is reflected in free testosterone and dihydrotestosterone levels.
![]() |
SUBJECTS AND METHODS |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Five subjects were excluded on grounds of using a hormonal intrauterine device, using an oral contraceptive containing only the progestin component (minipill), or not reporting the menstrual cycle phase. In the final analysis, users of oral contraceptives (OC+, n = 47) all reported using preparations containing both synthetic ethinyl oestradiol (mean 29 µg, range 2040 µg) and a progestin (75 µg gestoden, n = 16; 150 µg desogestrel, n = 15; 50150 µg levonorgestrel, n = 10; 2 mg cyproterone acetate, n = 6) for several months before entering the study. OC+ subjects age was 26 ± 4 years (mean ± SD, range 1938), body mass index (BMI) was 21.2 ± 2.1 kg/m2 (mean ± SD, range 16.826.4), the median of the reported alcohol consumption was 7 drinks/week (range 030), and that of the menstrual cycle phase was 18 ± 3 days (mean ± SD, range 726). The non-user group (OC, n = 40) did not use any form of hormonal medication. The mean age of the OC subjects was 30 ± 7 years (range 1946), their BMI was 21.5 ± 2.2 kg/m2 (range 17.027.4), the median of their reported alcohol consumption was 6 drinks/week (range 026), and their menstrual cycle phase was 15 ± 4 days (range 821). All subjects had a history of regular menses and none used any medication other than oral contraceptives. None of them had a record of hirsutism or other diseases.
All subjects participated randomly in one placebo and one alcohol drinking event with the intervening time being 28 days. Participation in the sessions was scheduled as close as possible to the mid-cycle phase with reference to the phase reported in the questionnaire. Among subjects not using oral contraceptives mid-cycle ovarian activity was confirmed with oestradiol levels >70 pmol/l. All drinking events took place on Mondays (three times), Tuesdays (nine times), and Wednesdays (twice) starting at 18:00. All the subjects were told not to drink any alcohol on the previous evening. No instructions were given regarding meals and snacks. Alcohol (dose 0.5 g/kg, equal to about 2 to 3 standard drinks) was given diluted in lingonberry juice (10% w/v). The placebo drink contained an equal volume of lingonberry juice only. Blood samples were drawn from the median cubital vein before and at 45 and 90 min after the start of drinking. The participants remained seated throughout the experiment. Drinking time was 30 min. No ethanol was detected before the start of drinking and levels increased to 5.8 ± 2.9 mM (27 ± 13 mg/dl) at 45 min and to 7.1 ± 3.0 mM (33 ± 14 mg/dl) at 90 min from the start of drinking.
Substudy B. Ten non-pregnant healthy Caucasian female student volunteers were recruited. All the subjects, aged 2032 (mean 24) years and weight 5263 (mean 58) kg, reported using oral contraceptives containing both synthetic ethinyl oestradiol and a progestin (brands similar to those used in substudy A) for several months before entering the study. All the subjects had a history of regular menses and none used any medication other than oral contraceptives. None of them had a record of hirsutism or other diseases. The menstrual cycle phase was not documented. All the subjects were light drinkers with no record of heavy drinking or any kind of alcohol problem. Each subject participated randomly in four drinking events, one of which was with a placebo (lingonberry juice) and the other three with different amounts of alcohol (0.34, 0.68 and 1.02 g/kg diluted in lingonberry juice 10% w/v), starting at 18:00. The intervals between each session were longer than 1 week. The conditions were the same as in substudy A, except that venous blood samples were taken before, and at 40, 90 and 150 min from start of, drinking. No ethanol was detected before the start of drinking and levels increased to 5.2 ± 0.6 mM (24 ± 3 mg/dl, at 90 min) with the dose of 0.34 g/kg, to 13.2 ± 1.0 mM (61 ± 5 mg/dl, at 150 min) with the dose of 0.68 g/kg, and to 25.6 ± 0.6 mM (117 ± 3 mg/dl, at 150 min) with the dose of 1.02 g/kg.
Analytical procedures
Blood samples were collected into tubes containing 22.5 mg of sodium fluoride and 22.5 mg of potassium oxalate as anticoagulants for a volume of 10 ml of blood. Plasma samples were prepared within 4 h and stored at 20°C until determinations. Ethanol levels were determined in plasma by headspace gas chromatography (Perkin-Elmer F40). Testosterone levels [within-assay variability 6.6% and between-assay variability 7.0% at the level of 0.96 nmol/l (n = 10), detection limit 0.1 nmol/l], free testosterone levels [within-assay variability 4.3% and between-assay variability 5.5% at the level of 4.6 pmol/l (n = 10), detection limit 0.5 pmol/l] and androstenedione levels [within-assay variability 10.4% and between- assay variability 4.3% at the level of 5.3 nmol/l (n = 10), detection limit 0.14 nmol/l] were determined by standard radioimmunoassay reagent sets (Orion Diagnostica, Finland, for testosterone; Diagnostic Products Corporation, Los Angeles, CA, USA for free testosterone and androstenedione).
Dehydroepiandrosterone levels [DHEA; within-assay variability 5.9% and between-assay variability 8.3% at the level of 7.6 nmol/l (n = 38), detection limit 2 nmol/l] were determined by a method based on extraction into petroleum ether followed by a radioimmunoassay (RIA) using tritiated DHEA as the labelled antigen (DHEA[1,2,6,7-3H(N)]-, NET-814, 250 µCi/250 µl ethanol, NENR Research Products) and Anti-DHEA (cat no. 07-129016 ICN) as the antibody. Sample and standard extracts were dissolved in buffer and incubated at 28°C overnight, and dextran-coated charcoal [0.32% Norit A (active charcoal) and 0.032% Dextran T70 in 0.045 mol/l phosphate-buffered saline (PBS) with 0.1% gelatine, pH 7.0] was used to separate the bound and free steroids.
Dihydrotestosterone levels [within-assay variability 9.1% and between-assay variability 17.8% at the level of 2.5 nmol/l (n = 10), detection limit 0.2 nmol/l] were determined as described by Apter et al. (1976). Briefly, the method is based on extraction into ethylether-ethylacetate (7:3 v/v) twice followed by chromatographic separation on a Lipidex-5000 column (hydroxyalkoxypropyl Sephadex, petroleum-chloroform 98:2 as eluent) and a RIA using tritiated dihydrotestosterone [5-dihydro(1,2,4,5,6,7-3H)testosterone, Amersham TRK 443] as the labelled antigen and antisera raised in rabbits (for immunization procedures see Jänne et al., 1974). Samples taken 45 min and 90 min after the start of drinking were pooled in equal volumes for dihydrotestosterone determinations.
In order to check for possible changes of hormone levels in vitro caused by alcohol as well as for possible interactions of alcohol with hormone assays, ethanol was added to fresh blood from 10 OC and six OC+ 1824 year old healthy female subjects to a final concentration of 10 mM. No significant effects on androstenedione and total testosterone were observed. A negligible reduction of the DHEA levels, confined to higher hormone levels, could be observed (16.8 ± 2.9 and 15.6 ± 2.8 nmol/l, P = 0.017, n = 16), which was perhaps due to the procedure of extraction with petroleum ether. The possible in vitro effect of alcohol was not determined for dihydrotestosterone.
Data analysis
Results are expressed as means ± SEM unless otherwise specified. The magnitude of the effect of alcohol on hormone levels was defined as the change in concentration (%) observed during the placebo session subtracted from the change in concentration observed during the alcohol ingestion session. Statistical significance was tested using two-factor analyses of variance for repeated measures (drug and time as within-group factors) and matched paired t-test. Absolute hormone values were used in the analyses of variance. For correlations, Spearman's rank-order correlation was used. Difference between the two r-values was tested by transforming r to r' (Fisher's transformation) as described (Howell, 1992). Data were analysed using SPSS (version 6.1) and GraphPad Prism (version 2.0) statistical software.
![]() |
RESULTS |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
|
|
The unbound fraction of testosterone expressed in percentages as the ratio of free to total testosterone was seen to decline during alcohol compared with placebo sessions among both OC and OC+ subjects (drug by time interaction; F = 3.2, P = 0.049 for OC and F = 4.3, P = 0.018 for OC+; Fig. 3). The overall ratio of free to total testosterone was clearly lower among OC+ subjects than OC subjects (P < 0.001).
|
|
Among both OC and OC+ subjects, DHEA levels were not statistically significantly different during the alcohol sessions compared with the placebo sessions (Fig. 5, top) and no drug by time interaction was observed. The magnitude of the effect of alcohol among OC subjects was +2% (95% CI = 8, +13) at 45 min and 13% (95% CI = 29, +4) at 90 min, whereas that of alcohol among OC+ subjects was +3% (95% CI = 6, +13) at 45 min and 5% (95% CI = 13, +4) at 90 min. DHEA levels were clearly lower among OC+ subjects than OC subjects (P < 0.001) (Fig. 5
).
|
The change in total testosterone levels (the average of changes seen at 45 and 90 min) during placebo sessions correlated with the change in androstenedione (r = 0.49, P = 0.003 for OC and r = 0.62, P = 0.001 for OC+) and DHEA (r = 0.31, P = 0.07 for OC and r = 0.27, P = 0.09 for OC+) levels. Among OC subjects, these correlations were similar during the alcohol sessions (r = 0.51, P = 0.002 for androstenedione and r = 0.29, P = 0.08 for DHEA), whereas among OC+ subjects the correlations disappeared (r = 0.03, P = 0.85 for androstenedione and r = 0.007, P = 0.96 for DHEA). This reduction in the correlation coefficient among OC+ subjects was statistically significant for androstenedione (z = 3.19, P = 0.005) and a tendency was observed for DHEA (z = 1.22, P = 0.13).
In substudy B, total testosterone levels (Fig. 6) were significantly higher during all the alcohol events than in the placebo session (F = 5.34, P = 0.011, all doses P < 0.001). No significant dose or time effects were observed. The present doseresponse results have previously been presented in preliminary form (Eriksson et al., 1994
).
|
![]() |
DISCUSSION |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Testosterone levels can be affected either by changes in its synthesis in the adrenals or gonads, through changed peripheral conversion from androstenedione and DHEA or through changed catabolism in the liver (Yen and Jaffe, 1991). Since alcohol intake had no effect on DHEA, a major part of which is of adrenal origin (Yen and Jaffe, 1991
), it is unlikely that the testosterone elevation is caused by an acute increase in adrenal androgen synthesis. The finding that the testosterone effect was not alcohol dose-dependent suggests that it is related to the zero-order mechanism of ethanol elimination and mediated by the change in the redox state. It is well known that this effect is rather constant during different dose and time conditions of ethanol oxidation (Forsander, 1970
). The facts that alcohol intake is associated with a decline in androstenedione levels, an elevation in testosterone levels, and an elevation in the testosterone to androstenedione ratio suggest an increase in the conversion of androstenedione to testosterone. The positive correlation between the change in testosterone levels and the change in androstenedione levels after intake of placebo can be explained by the similar circadian rhythm of these hormones (Yen and Jaffe, 1991
). The fact that this positive correlation was significantly reduced during alcohol conditions among women using oral contraceptives, i.e. the group displaying the more pronounced testosterone effect, provides further evidence of an increased androstenedione to testosterone reaction superimposed on the circadian hormonal changes.
Ethanol oxidation has earlier been shown to be coupled to steroid reduction in the liver (Andersson et al., 1986). More specifically, ethanol oxidation was shown to cause an increased rate of the reduction catalysed by the liver NAD-dependent 17ß-hydroxysteroid dehydrogenase type 2 enzyme (Andersson and Moghrabi, 1997
) with a secondary change in the equilibrium between conjugated 17-hydroxy- and 17-ketosteroids. These findings on conjugated steroids in men are similar to our results on unconjugated testosterone/ androstenedione and oestradiol/oestrone in premenopausal women (Sarkola et al., 1999
). The finding that the effect in the present study was pronounced among women using oral contraceptives may be explained by the fact that the 17ß-hydroxysteroid type 2 enzyme is induced by the synthetic progestins found in the contraceptive preparation (Tseng and Gurpide, 1979
). Thus, the present results suggest that the testosterone effect is the result of an inhibited catabolism in the liver, i.e. a decreased overall oxidation of testosterone due to the increased reduction of androstenedione to testosterone, mediated by the alcohol-induced elevation in the [NADH]:[NAD+] ratio (Fig. 7
).
|
No acute effect of alcohol on dihydrotestosterone levels was observed during the present dose and time conditions. In view of the testosterone effect, one could expect an elevation in dihydrotestosterone as well. The lack of an elevation in dihydrotestosterone is, however, not surprising, in view of the fact that, in women, the main source of dihydrotestosterone in plasma is androstenedione, with <20% being derived from testosterone (Ito and Horton, 1971). The present dihydrotestosterone finding may thus be the net result of a decline and an elevation caused by androstenedione and testosterone respectively.
An important question is how long the elevating effect of alcohol on testosterone levels could be expected to last. In view of the present results, it seems reasonable to postulate that the acute effect could last at least throughout the period of ethanol elimination. This is supported by the finding (Karila et al., 1996) that the testosterone level was found to be elevated 14 h after intake of alcohol (2.0 g/kg) but not at 38 h (T. Karila, personal communication), compared to levels preceding the day of alcohol intake.
The present finding with consistently lower basal androgen levels in women using oral contraceptives has been reported earlier (Jung-Hoffman et al., 1988). The reduced basal testosterone and androstenedione level is probably caused by the inhibition of the ovarian function (Jung-Hoffman et al., 1988
). The mechanism of the reduced basal DHEA level, which mostly originates from the adrenals (Yen and Jaffe, 1991
), is to our knowledge unknown. The fact that the use of oral contraceptives is associated with an increase in SHBG (Jung-Hoffman et al., 1988
) may well explain the finding of a reduced free to total testosterone ratio in this study group.
In conclusion, the present androgen effects of alcohol are suggested to be the result of a change in the steroid metabolism in the liver. Whether repeated doses of alcohol would have a cumulative effect on androgen levels or whether heavy alcohol consumption is merely associated with repeated testosterone elevations in women cannot be determined from the present results. Both possibilities may explain earlier positive associations between drinking and androgen levels (Dorgan et al., 1994; Cigolini et al., 1996
) as well as findings of hyperandrogenicity (Pettersson et al., 1990
; Välimäki et al., 1990
) and loss of female sexual characteristics (Van Thiel and Lester, 1979
) among women drinking heavy amounts of alcohol.
![]() |
ACKNOWLEDGEMENTS |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
![]() |
FOOTNOTES |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
* Author to whom correspondence should be addressed.
![]() |
REFERENCES |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Andersson, S., Cronholm, T. and Sjövall, J. (1986) Redox effects of ethanol on steroid metabolism. Alcoholism: Clinical and Experimental Research 10 (Suppl.), 55S63S.[Medline]
Apter, D., Jänne, O., Karvonen, P. and Vihko, R. (1976) Simultaneous determination of five sex hormones in human serum by radioimmunoassay after chromatography on Lipidex-5000. Clinical Chemistry 22, 3238.[Abstract]
Becker, U., Gluud, Ch., Bennet, P., Micic, S., Svenstrup, B., Winkler, Kj., Christensen, N. J. and Hardt, F. (1988) Effect of alcohol and glucose infusion on pituitarygonadal hormones in normal females. Drug and Alcohol Dependence 22, 141149.[ISI][Medline]
Cigolini, M., Targher, G., Bergamo Andreis, I. A., Tonoli, M., Filippi, F., Muggeo, M. and De Sandre, G. (1996) Moderate alcohol consumption and its relation to visceral fat and plasma androgens in healthy women. International Journal of Obesity 20, 206212.[ISI][Medline]
Dorgan, J. F., Reichman, M. E., Judd, J. T., Brown, Ch., Longcope, Ch., Schatzkin, A., Campbell, W., Franz, Ch., Kahle, L. and Taylor, P. R. (1994) The relation of reported alcohol ingestion to plasma levels of estrogens and androgens in premenopausal women (Maryland, United States). Cancer Causes Control 5, 5360.[ISI][Medline]
Eriksson, C. J. P., Fukunaga, T. and Lindman, R. (1994) Sex hormone response to alcohol. Nature 369, 711.[ISI][Medline]
Forsander, O. A. (1970) Influence of ethanol on the redox state of the liver. Quarterly Journal of Studies on Alcohol 31, 550570.[ISI][Medline]
Howell, D. (1992) Statistical Methods for Psychology, 3rd edn. Duxbury Press, Belmont, California.
Ito, T. and Horton, R. (1971) The source of plasma dihydrotestosterone in man. Journal of Clinical Investigations 50, 16211627.[ISI][Medline]
Jung-Hoffman, C., Heidt, F. and Kuhl, H. (1988) Effect of two oral contraceptives containing 30 µg ethinylestradiol and 75 µg gestodene or 150 µg desogestrel upon various hormonal parameters. Contraception 38, 593603.[ISI][Medline]
Jänne, O., Apter, D. and Vihko, R. (1974) Assay of testosterone, progesterone and 17ß-hydroxyprogesterone in human plasma by radioimmunoassay after separation on hydroxyalkoxypropyl Sephadex. Journal of Steroid Biochemistry 5, 155162.[ISI][Medline]
Karila, T., Kosunen, V., Leinonen, A., Tähtelä, R. and Seppälä, T. (1996) High doses of alcohol increase urinary testosterone-to-epitestosterone ratio in females. Journal of Chromatography B 687, 109116.[ISI][Medline]
McNamee, B., Grant, J., Ratcliffe, J., Ratcliffe, W. and Oliver, J. (1979) Lack of effect of alcohol on pituitarygonadal hormones in women. British Journal of Addiction 74, 316317.[ISI]
Pettersson, P., Ellsinger, B-M. and Sjöberg, C. (1990) Fat distribution and steroid hormones in women with alcohol abuse. Journal of Internal Medicine 228, 311316.[ISI][Medline]
Sarkola, T., Mäkisalo, H., Fukunaga, T. and Eriksson, C. J. P. (1999) Acute effect of alcohol on estradiol, estrone, progesterone, prolactin, cortisol, and luteinizing hormone in premenopausal women. Alcoholism: Clinical and Experimental Research 23, 976982.[ISI][Medline]
Tseng, L. and Gurpide, E. (1979) Stimulation of various 17ß- and 20-hydroxysteroid dehydrogenase activities by progestins in human endometrium. Endocrinology 104, 17451748.[ISI][Medline]
Välimäki, M., Härkönen, M. and Ylikahri, R. (1983) Acute effects of alcohol on female sex hormones. Alcoholism: Clinical and Experimental Research 7, 289293.[ISI][Medline]
Välimäki, M., Pelkonen, R., Härkönen, M., Tuomala, P., Koistinen, P., Roine, R. and Ylikahri, R. (1990) Pituitarygonadal hormones and adrenal androgens in non-cirrhotic female alcoholics after cessation of alcohol intake. European Journal of Clinical Investigations 20, 177181.[ISI][Medline]
Van Thiel, D. and Lester, R. (1979) The effect of chronic alcohol abuse on sexual function. Clinics in Endocrinology and Metabolism 8, 499510.[ISI][Medline]
Yen, S. and Jaffe, R. eds (1991) Reproductive Endocrinology, 3rd edn. W. B. Saunders, Philadelphia.
Ylikahri, R., Huttunen, M., Härkönen, M., Seuderling, U., Onikki, S., Karonen, S-L. and Adlercreutz, H. (1974) Low plasma testosterone values in men during hangover. Journal of Steroid Biochemistry 5, 655658.[ISI]