Eur J ClinPharmacol(1992) 43:543-546
© Springer-Verlag 1992
Menstrual cycle effects on caffeine elimination in the human female J. D. Lane, J. F. Steege, S. L. Rupp, and C. M. Kuhn Departments of Psychiatry,Obstetrics and Gynecology,and Pharmacology, Duke UniversityMedicalCenter,Durham,North Carolina,USA Received:January 13,1992/Accepted in revisedform:May 15, 1992
Summary. Increases in the levels of sex steroids due to pregnancy or oral contraceptive steroid use are known to decrease significantly the rate at which caffeine is eliminated from the body. An investigation has now been made into whether the changes in sex steroid levels that occur during normal menstrual cycling also affect the rate of caffeine elimination, especially whether hormonal shifts in the luteal phase are associated with slower elimination of caffeine. Repeated 24-hour caffeine elimination studies were conducted during the follicular and luteal phases of the menstrual cycle in 10 healthy women. Comparisons of the follicular and luteal phases revealed that systemic clearance of caffeine was slower in the luteal phase, although the tm did not differ. The slowing effect was related to the proximity to onset of menstruation and to levels of progesterone. The evidence suggests that caffeine elimination may be slowed in the late luteal phase, prior to the onset of menstruation. Such a reduction would lead to increased accumulation of caffeine with repeated self-administration during the day, but the effect may be too small to be of clinical significance in the majority of women. Key words: Caffeine, Menstrual cycle; pharmacokinetics, females
Caffeine is without doubt the most frequently ingested natural alkaloid in the world. It is present in a variety of common beverages, including coffee, tea, many soft drinks, in products containing cocoa or chocolate, and in a variety of medications, including headache or pain remedies, and over-the-counter stimulants [1]. It is generally recognized that caffeine can have salutary effects on mental alertness and mood, effects that probably account for its wide use. However, high systemic levels of caffeine can produce symptoms of intoxication, including dysphoric symptoms of physiological excitation, and psychiatric symptoms similar to generalized anxiety or panic disorder [2, 3]. Although high caffeine levels and the resultant intoxication are most likely to occur when high
doses of caffeine are ingested, they could also occur whenever the rate of caffeine elimination is reduced, leading to a greater systemic accumulation of caffeine if intake is maintained at a constant rate. Research has identified two conditions in females that are associated with a clinically significant reduction in the rate of caffeine elimination. The use of oral contraceptive steroids has been shown to lead to a doubling of the elimination half-life of caffeine, normally ranging from 3-5 h but increasing to 6-10 h during use of OCS [4-7]. Caffeine elimination is also significantly slower during the second and third trimesters of pregnancy [8], with the elimination half-life rising to almost three-times the normal value in the weeks prior to parturition. These pharmacokinetic effects have been attributed to the elevated levels of sex steroids that characterise both conditions and that are thought to inhibit the liver functions responsible for the metabolism of caffeine prior to elimination. The present study investigated whether the cyclical increases in sex steroids during the luteat phase of the normal menstrual cycle might also inhibit caffeine metabolism and reduce its clearance from the body. A reduction in the rate of caffeine elimination during the luteal phase prior to the onset of menstruation might lead to the accumulation of higher than normal systemic concentrations of caffeine and so cause caffeine intoxication. The resultant symptoms might contribute to or aggravate other similar symptoms observed in the premenstrual interval. Repeated pharmacokinetic studies were conducted in healthy women during the follicular and luteat phases of the menstrual cycle and standard measures of caffeine elimination pharmacokinetics were compared:
Materials and methods Ten regularly menstruating women of reproductive age (22-42y, mean 36 y) participated after providinginformed consent. Participants were free fromhypertensionand other cardiovasculardiseases and were not taking any hormonal, antihypertensive or cardiac medications.All were normal on physicalexaminationand had nor-
544 Table 1. Subject age and sex steroid levels on the follicular and luteal-phase days of the caffeine pharmacokinetic study Subject
AL AS BC Ctl CU EB IS MS SR TM
Age (y)
36 39 38 36 33 42 39 36 39 22
Follicular phase Cycle Oestrogen day (pg.ml ~) 4 2 5 1 3 5 4 3 5 4
< 25.0 < 25.0 32.9 < 25.0 < 25.0 33.5 < 25.0 < 25.0 < 25.0 36.3
Progesterone (ng.ml *)
Luteat phase Cycle day
Oestrogen (pg.ml -~)
0.25 0.11 0.17 0.50 0.46 0.18 0.15 0.79 0.40 0.54
-11 -10 -7 -2 -12 -7 -7 -5 -6 -5
57.1 52.4 43.7 64.6 43.2 69.6 65.0 66.3 52.1 67.8
Progesterone (ng. mt- ~) 16.00 5.45 10.00 6.70 7.81 8.48 11.09 8.87 12.83 2.60"
This luteal phase progesterone value suggests that subject TM may not have ovulated in this cycle. Removal of her data does not change the mean experimental results for caffeine elimination mal values for cholesterol and low- and high-density lipoproteins. Subjects had negative serum [3-HCG assays prior to participation and employed non-hormonal contraceptive methods during the study. Subjects participated in two 24-h studies of caffeine absorption and elimination, one in the early follicular phase (forward cycle Days 1 to 5) and one in the Iuteal phase (reverse cycle Days -2 to -12). The cycle phase chosen for the initial study day was counterbalanced across subjects. Subjects entered the outpatient pavilion of the Clinical Research Unit in the Duke University Medical Center, in the fasting state prior to 08.00 h. Following insertion of an intravenous indwelling catheter, three baseline blood samples were taken at 20-min intervals. Measurements of blood pressure and pulse rate were taken prior to the withdrawal of each sample. Subjects drank an aqueous solution of caffeine (250 rag) and additional samples were collected after 1, 2, 3, 4, 8, and 24 h. Blood pressure and pulse rate were determined prior to each blood sample. Subjects were given a standardised light breakfast after the 1-h blood sample and a standardised lunch at noon. Subjects were also provided with a standard dinner after the 8-h sample and were then permitted to leave. The controlled meals were provided to the subject to prevent the unknowing consumption of additional caffeine during the study periods. No other foods or beverages other than water were consumed during the study period. Oestradiol and progesterone levels were determined by radioimmunoassay using pooled sera from the three baseline samples drawn 20-min apart. The plasma caffeine level in one baseline sample and each post-ingestion sample was determined using the radioimmunoassay procedure developed by Cook [9].
Results Summary data for the studies in the follicular and luteal phases of the menstrual cycle (Table 1) confirm the appropriate timing of study days and hormonal levels in the two menstrual cycle phases. F o r the follicular phase, the cycle day averaged 3.6 days (range 1 to 5 days), oestrogen levels averaged 27.8 pg- m l - 1 (range 25 to 36.3 pg. m l - 1), and progesterone averaged 0 . 3 6 n g . m l -I (range 0.11 to 0.79 ng. ml-1). The luteal phase cycle day averaged -7.2 days from the next onset of menses ( r a n g e - 1 2 t o - 2 days), oestrogen averaged 58.2 pg. m1-1 (range 43.2 to 69.6 pg. ml-1), and progesterone averaged 8.98 ng.m1-1 (range 2.60 to 16.0 ng. ml-2). The caffeine concentration/time curves were analysed by assumption of a first order o n e - c o m p a r t m e n t model after oral administration of a single dose [10]. Least-
squares log-linear regression analysis of the descending portion of the caffeine concentration/time curve gave the elimination rate constant, kd, and the caffeine elimination half-life (tl/2=ln2/kei). T h e area under the concentration/time curve ( A U C ) was calculated by the trapezoidal method, with exponential corrections for the initial and residual (24 h) concentrations if any. Total systemic clearance (CL) was calculated by dividing the dose of caffeine (mg. kg-1) by the A U C . The apparent volume of distribution (V) was calculated as CL/kel. Pharmacokinetic values are presented in Table 2 for the follicular and luteal phases of the menstrual cycle. Comparisons of follicular and luteal values were made within-subject and were evaluated by one-tailed t-tests (dr = 9), consistent with the directional hypothesis that caffeine elimination was slower in the luteal than follicular phase. The directional hypothesis and one-tailed tests provided greater statistical power from the relatively small sample. Evidence of slowed caffeine elimination during the luteal phase was found as a significantly reduced rate of total systemic clearance (t = -2.02, P < 0.035) and a trend to a greater A U C (t = 1.51, P < 0.08). However, neither the elimination tl/2 nor V appeared to be affected by the menstrual cycle phase (t = 0.14 and t = -1.20, respectively), Individual differences in the magnitude of the effect of the phase on kinetic values were apparent, with two subjects demonstrating m a r k e d slowing during the luteal phase and the others less effect. The potential individual differences were further explored by correlational analyses guided by the following hypothesis: If increases in h o r m o n e levels were to contribute to a slowing of caffeine elimination, the higher levels of progesterone and oestrogen should be associated with slower rates of caffeine elimination. The phase-related differences in oestrogen and progestrone levels were not significantly correlated with phaserelated changes in caffeine elimination kinetics. Two trends were observed in the data, namely that the increase in oestrogen f r o m the follicular to the luteal phase was negatively related to the change in systemic clearance (r =-0.52, NS) and positively related to the change in A U C (r = 0.47, NS). Both trends suggested that the magnitude of the increases in oestrogens might be related to
545 Table 2. Caffeine elimination kinetic profiles in the follicular and luteal phases of the menstrual cycle"
Subject AL AS BC CH CU EB IS MS SR TM Mean
Follicular phase AUC t~/2 2304 2.5 3632 4.3 3754 4.9 3463 2.5 4749 2.2 8226 11.5 2182 6.3 5845 5.3 3779 4.6 7289 8.7 4522 5.3
CL 1.63 1.16 0.93 0.82 0.89 0.46 1.74 0.72 0.98 0.62 0.99
V 0.35 0.43 0.40 0.18 0.17 0.46 0.95 0.33 0.39 0.46 0.41
Luteal phase AUC 2392 3431 4289 4484 4572 8556 2937 10207 3906 7008 5178
tl~2 2.4 3.7 4.3 3.9 3.0 11.5 2.5 13.9 2.3 6.9 5.4
CL 1.57 1.23 0.81 0.63 0.92 0.44 f.29 0.41 0.95 0.64 0.89b
V 0.32 0.39 0.31 0.21 0.24 0.44 0.28 0.49 0.19 0.39 0.33
a Drug dose = 250 mg caffeine, administered R O.; AUC = area under the concentration/time curve (gg- ml- 1.min); tli2= elimination half-life (h); CL=systemic clearance rate (mI.kg-a.min ~); V = apparent volume of distribution (1. kg ~)
b Luteal phase systemic clearance is significantly lower than in the follicular phase (P < 0.05, see text)
the magnitude of the reduction in caffeine elimination. No similar trend was observed for changes in progesterone level. In the data from the follicular phase, the oestrogen levels were correlated with the A U C (r = 0.69, P < 0.03) and elimination tl/2 (r = 0.73, P < 0.02), with a similar, but negative, trend for CL (r =-0.55, NS). All correlations suggested that higher levels of oestrogens were associated with lower rates of caffeine elimination within the follicular phase. In contrast, levels of progesterone were not related to the pharmacokinetic variables. Similar effects of oestrogen level appeared within the luteaI phase, although they did not reach statistical significance. There were trends correlating the oestrogen level with the A U C (r = 0.53, NS), tl/2 (r = 0.57, P < 0.09), and with the systemic clearance (r =-0.37, NS), each again suggesting that a higher oestrogen level might be associated with a reduced rate of caffeine elimination. In contrast, there was a trend to a positive relationship between luteal progesterone level and the rate of systemic clearance (r = 0.54, NS), consistent with general observations that oestrogen and progesterone can have opposite effects on many physiological processes. Finally, the effect of luteal phase cycle day was investigated, given that the hypothesis suggested that caffeine elimination should be slower as the woman approached the onset of the menses. This hypothesis was confirmed by the significant correlation observed between luteal phase cycle day, which is counted as a negative number, and systemic clearance rate, CL (r = -0.63, P < 0.05) and supported by similar trends for A U C (r = 0.40, NS) and for tl/2 (r = 0.34, NS). Women who were studied closer to the onset of their next menses had a lower rate of caffeine elimination.
levels that are associated with the phases of the normal menstrual cycle may be associated with variation in the rate of caffeine elimination. Confirming this hypothesis, there was a significant average reduction in total systemic clearance during the luteal phase that was associated with a trend to a larger 24-h AUC. In contrast, no cycle phase effects were observed on the average caffeine elimination tl/2 or the apparent volume of distribution. These observations confirm an earlier report of slowed caffeine elimination during the luteal phase [11] in which a reduction of similar magnitude (roughly 25 %) was found, although significant effects were also observed in a lengthened tip2 as well as reduced systemic clearance and increased AUC. The correlational exploration of the individual variation in cycle effects provided additional evidence that normal variability in sex steroid levels can influence caffeine elimination. Higher oestrogen levels were consistently associated with relatively slower rates of caffeine elimination, while the opposite trend was seen for progesterone levels. The small sample size limited the level of statistical significance reported for many of the correlations, but if the associations are considered in terms of the size of the observed effects (defined as R 2 or the proportion of variability in the pharmacokinetic values that can be accounted for by the menstrual cycle variables), then the relationships are more impressive. In many of the reported relationships, the menstrual cycle variables accounted for 25 % to 50 % of the total variability among subjects observed in the measures of caffeine elimination. The average changes in caffeine elimination associated with menstrual cycle phase were relatively small, compared to changes associated with OCS use or with pregnancy. In three studies of OCS effects [4, 5, 7], OCS use was associated with increases in tl~ of 47 % to 235 % and with reductions in caffeine clearance ranging from 39 % to 66%. The third trimester of pregnancy was associated with a 200 % increase in ti/z and 70 % reduction in clearance [8]. In the present study, the reduction in clearance from the follicular to the luteal phase only averaged 11% and there was no effect on average tl/2. However, it should not be surprising that menstrual cycle effects are smaller
Discussion
The present study demonstrates slowing of caffeine elimination in the luteal phase of the normal menstrual cycle, and suggests that the normal fluctuations in sex steroid
546 in magnitude, given that the hormonal changes during pregnancy are an order of magnitude greater than those seen during the normal menstrual cycle. Although the average effect of menstrual cycle phase on caffeine elimination was small, the present study does suggest the possibility that cycle-related changes might be clinically important in some women, particularly those who have a higher oestrogen level and a greater increases in oestrogen level in the luteal phase. Such differences are suggested by the correlational explorations. O f the ten w o m e n in the study, the three that had the greatest increases in oestrogen all showed a decrease in caffeine clearance of m o r e than 23 %, and an increase in A U C of m o r e than 25 % in the luteal phase. However, even in that group, the response was less than that associated with OCS use, and it is unlikely that such effects would lead to such accumulation of caffeine that toxic symptoms would be produced. In conclusion, the evidence suggests that caffeine elimination may fluctuate across the menstrual cycle, with slower elimination in the luteal phase prior to the onset of menstruation. The reduction in caffeine clearance is associated with a slightly increased exposure to ingested caffeine ( A U C ) , that might increase the likelihood of systemic accumulation of caffeine if it were repeatedly ingested throughout the day. The clinical implications of these relatively small increases in caffeine exposure are unknown, but it is unlikely that these effects alone would lead;to symptoms of caffeine intoxication in the luteal phase. However, given that caffeine has been shown to intensify the cardiovascular and neuroendocrine effects of psychological stress in other studies, and that excessive levels of caffeine can produce toxic symptoms, these menstrual cycle effects m a y deserve further attention in investigations of the premenstrual syndrome.
Acknowledgements. We thank Dr. C. E. Cook, Research Triangle Institute, for his kind gift of the antiserum for the caffeine radioimmunoassay and Mr. T. Zimmerman for his diligence in conducting the assays for plasma caffeine level. Data were managed and analysed in part using the CLINFO Data Analysis System of the Clinical Research Unit. Research supported by NIH Research Grant HL29968-05 NIDA Research Grant DA-06857-01, and a grant (RR-30)
from the General Clinical Research Centers Program of the Division of Research Resources, National Institutes of Health.
References 1. Barone JJ, Roberts H (1984) Human consumption of caffeine. In: Dews PB (ed) Caffeine. Springer, Berlin Heidelberg New York, pp 5%73 2. Greden JF (1974) Anxiety or caffeinism: a diagnostic dilemma. Am J Psychiatry 131:1089-1092 3. American Psychiatric Association (1987) Diagnostic and statistical manual of mental disorders, 3rd ed, revised. Washington, DC, pp 138-139 4. Patwardhan RV, Desmond PV, Wilkinson GR, Schenker S (1980) Impaired elimination of caffeine by oral contraceptive steroids. J Lab Clin Med 95:604-608 5. Callahan MM, Robertson RS, Branfman AR, McComish MF, Yesair DW (1983) Comparison of caffeine metabolism in three nonsmoking populations after oral administration of radiolabeled caffeine. Drug Metab Dispos 11:211-217 6. Reitveld EC, Broekman MMM, Houben JJG, Eskes TKAB, van Rossum JM (1984) Rapid onset of an increase in caffeine residence time in young women due to oral contraceptive steroids. Eur J Clin Pharmaco126:371-373 7. Abernethy DR, Todd EL (1985) Impairment of caffeine clearance by chronic use of low-dose oestrogen-containing oral contraceptives. Eur J Ctin Pharmaco128:425-428 8. Aldridge A, Bailey J, Neims AH (1981) The disposition of caffeine during and after pregnancy. Semin Perinatol 5:310-314 9. Cook CE, Taltent CR, Amerson E, Myers MW, Kepler JA, Taylor GC, Christensen HD (1976) Caffeine in salvia and plasma by a radioimmune assay procedure. J Pharmacol Exp Ther 199: 679686 10. Neims AH, von Borstel RW (1983) Caffeine: metabolism and biochemical mechanisms of action. In: Wurtman R J, Wurtman JJ (eds) Nutrition and the brain, vol 6. Raven Press, New York, pp 1-29 11. Balogh A, Irmisch E, Klinger G, Splinter FK, Hoffmann A (1987) Untersuchungen zur Elimination von Coffein und Metamizol im Menstruationszyklus der fertilen Frau. ZentralbI Gyn~iko1109:1135-1142 J. D. Lane, PhD Department of Psychiatry, Box 3830 Duke University Medical Center Durham, NC 27710 USA
© Springer-Verlag 1992
Menstrual cycle effects on caffeine elimination in the human female J. D. Lane, J. F. Steege, S. L. Rupp, and C. M. Kuhn Departments of Psychiatry,Obstetrics and Gynecology,and Pharmacology, Duke UniversityMedicalCenter,Durham,North Carolina,USA Received:January 13,1992/Accepted in revisedform:May 15, 1992
Summary. Increases in the levels of sex steroids due to pregnancy or oral contraceptive steroid use are known to decrease significantly the rate at which caffeine is eliminated from the body. An investigation has now been made into whether the changes in sex steroid levels that occur during normal menstrual cycling also affect the rate of caffeine elimination, especially whether hormonal shifts in the luteal phase are associated with slower elimination of caffeine. Repeated 24-hour caffeine elimination studies were conducted during the follicular and luteal phases of the menstrual cycle in 10 healthy women. Comparisons of the follicular and luteal phases revealed that systemic clearance of caffeine was slower in the luteal phase, although the tm did not differ. The slowing effect was related to the proximity to onset of menstruation and to levels of progesterone. The evidence suggests that caffeine elimination may be slowed in the late luteal phase, prior to the onset of menstruation. Such a reduction would lead to increased accumulation of caffeine with repeated self-administration during the day, but the effect may be too small to be of clinical significance in the majority of women. Key words: Caffeine, Menstrual cycle; pharmacokinetics, females
Caffeine is without doubt the most frequently ingested natural alkaloid in the world. It is present in a variety of common beverages, including coffee, tea, many soft drinks, in products containing cocoa or chocolate, and in a variety of medications, including headache or pain remedies, and over-the-counter stimulants [1]. It is generally recognized that caffeine can have salutary effects on mental alertness and mood, effects that probably account for its wide use. However, high systemic levels of caffeine can produce symptoms of intoxication, including dysphoric symptoms of physiological excitation, and psychiatric symptoms similar to generalized anxiety or panic disorder [2, 3]. Although high caffeine levels and the resultant intoxication are most likely to occur when high
doses of caffeine are ingested, they could also occur whenever the rate of caffeine elimination is reduced, leading to a greater systemic accumulation of caffeine if intake is maintained at a constant rate. Research has identified two conditions in females that are associated with a clinically significant reduction in the rate of caffeine elimination. The use of oral contraceptive steroids has been shown to lead to a doubling of the elimination half-life of caffeine, normally ranging from 3-5 h but increasing to 6-10 h during use of OCS [4-7]. Caffeine elimination is also significantly slower during the second and third trimesters of pregnancy [8], with the elimination half-life rising to almost three-times the normal value in the weeks prior to parturition. These pharmacokinetic effects have been attributed to the elevated levels of sex steroids that characterise both conditions and that are thought to inhibit the liver functions responsible for the metabolism of caffeine prior to elimination. The present study investigated whether the cyclical increases in sex steroids during the luteat phase of the normal menstrual cycle might also inhibit caffeine metabolism and reduce its clearance from the body. A reduction in the rate of caffeine elimination during the luteal phase prior to the onset of menstruation might lead to the accumulation of higher than normal systemic concentrations of caffeine and so cause caffeine intoxication. The resultant symptoms might contribute to or aggravate other similar symptoms observed in the premenstrual interval. Repeated pharmacokinetic studies were conducted in healthy women during the follicular and luteat phases of the menstrual cycle and standard measures of caffeine elimination pharmacokinetics were compared:
Materials and methods Ten regularly menstruating women of reproductive age (22-42y, mean 36 y) participated after providinginformed consent. Participants were free fromhypertensionand other cardiovasculardiseases and were not taking any hormonal, antihypertensive or cardiac medications.All were normal on physicalexaminationand had nor-
544 Table 1. Subject age and sex steroid levels on the follicular and luteal-phase days of the caffeine pharmacokinetic study Subject
AL AS BC Ctl CU EB IS MS SR TM
Age (y)
36 39 38 36 33 42 39 36 39 22
Follicular phase Cycle Oestrogen day (pg.ml ~) 4 2 5 1 3 5 4 3 5 4
< 25.0 < 25.0 32.9 < 25.0 < 25.0 33.5 < 25.0 < 25.0 < 25.0 36.3
Progesterone (ng.ml *)
Luteat phase Cycle day
Oestrogen (pg.ml -~)
0.25 0.11 0.17 0.50 0.46 0.18 0.15 0.79 0.40 0.54
-11 -10 -7 -2 -12 -7 -7 -5 -6 -5
57.1 52.4 43.7 64.6 43.2 69.6 65.0 66.3 52.1 67.8
Progesterone (ng. mt- ~) 16.00 5.45 10.00 6.70 7.81 8.48 11.09 8.87 12.83 2.60"
This luteal phase progesterone value suggests that subject TM may not have ovulated in this cycle. Removal of her data does not change the mean experimental results for caffeine elimination mal values for cholesterol and low- and high-density lipoproteins. Subjects had negative serum [3-HCG assays prior to participation and employed non-hormonal contraceptive methods during the study. Subjects participated in two 24-h studies of caffeine absorption and elimination, one in the early follicular phase (forward cycle Days 1 to 5) and one in the Iuteal phase (reverse cycle Days -2 to -12). The cycle phase chosen for the initial study day was counterbalanced across subjects. Subjects entered the outpatient pavilion of the Clinical Research Unit in the Duke University Medical Center, in the fasting state prior to 08.00 h. Following insertion of an intravenous indwelling catheter, three baseline blood samples were taken at 20-min intervals. Measurements of blood pressure and pulse rate were taken prior to the withdrawal of each sample. Subjects drank an aqueous solution of caffeine (250 rag) and additional samples were collected after 1, 2, 3, 4, 8, and 24 h. Blood pressure and pulse rate were determined prior to each blood sample. Subjects were given a standardised light breakfast after the 1-h blood sample and a standardised lunch at noon. Subjects were also provided with a standard dinner after the 8-h sample and were then permitted to leave. The controlled meals were provided to the subject to prevent the unknowing consumption of additional caffeine during the study periods. No other foods or beverages other than water were consumed during the study period. Oestradiol and progesterone levels were determined by radioimmunoassay using pooled sera from the three baseline samples drawn 20-min apart. The plasma caffeine level in one baseline sample and each post-ingestion sample was determined using the radioimmunoassay procedure developed by Cook [9].
Results Summary data for the studies in the follicular and luteal phases of the menstrual cycle (Table 1) confirm the appropriate timing of study days and hormonal levels in the two menstrual cycle phases. F o r the follicular phase, the cycle day averaged 3.6 days (range 1 to 5 days), oestrogen levels averaged 27.8 pg- m l - 1 (range 25 to 36.3 pg. m l - 1), and progesterone averaged 0 . 3 6 n g . m l -I (range 0.11 to 0.79 ng. ml-1). The luteal phase cycle day averaged -7.2 days from the next onset of menses ( r a n g e - 1 2 t o - 2 days), oestrogen averaged 58.2 pg. m1-1 (range 43.2 to 69.6 pg. ml-1), and progesterone averaged 8.98 ng.m1-1 (range 2.60 to 16.0 ng. ml-2). The caffeine concentration/time curves were analysed by assumption of a first order o n e - c o m p a r t m e n t model after oral administration of a single dose [10]. Least-
squares log-linear regression analysis of the descending portion of the caffeine concentration/time curve gave the elimination rate constant, kd, and the caffeine elimination half-life (tl/2=ln2/kei). T h e area under the concentration/time curve ( A U C ) was calculated by the trapezoidal method, with exponential corrections for the initial and residual (24 h) concentrations if any. Total systemic clearance (CL) was calculated by dividing the dose of caffeine (mg. kg-1) by the A U C . The apparent volume of distribution (V) was calculated as CL/kel. Pharmacokinetic values are presented in Table 2 for the follicular and luteal phases of the menstrual cycle. Comparisons of follicular and luteal values were made within-subject and were evaluated by one-tailed t-tests (dr = 9), consistent with the directional hypothesis that caffeine elimination was slower in the luteal than follicular phase. The directional hypothesis and one-tailed tests provided greater statistical power from the relatively small sample. Evidence of slowed caffeine elimination during the luteal phase was found as a significantly reduced rate of total systemic clearance (t = -2.02, P < 0.035) and a trend to a greater A U C (t = 1.51, P < 0.08). However, neither the elimination tl/2 nor V appeared to be affected by the menstrual cycle phase (t = 0.14 and t = -1.20, respectively), Individual differences in the magnitude of the effect of the phase on kinetic values were apparent, with two subjects demonstrating m a r k e d slowing during the luteal phase and the others less effect. The potential individual differences were further explored by correlational analyses guided by the following hypothesis: If increases in h o r m o n e levels were to contribute to a slowing of caffeine elimination, the higher levels of progesterone and oestrogen should be associated with slower rates of caffeine elimination. The phase-related differences in oestrogen and progestrone levels were not significantly correlated with phaserelated changes in caffeine elimination kinetics. Two trends were observed in the data, namely that the increase in oestrogen f r o m the follicular to the luteal phase was negatively related to the change in systemic clearance (r =-0.52, NS) and positively related to the change in A U C (r = 0.47, NS). Both trends suggested that the magnitude of the increases in oestrogens might be related to
545 Table 2. Caffeine elimination kinetic profiles in the follicular and luteal phases of the menstrual cycle"
Subject AL AS BC CH CU EB IS MS SR TM Mean
Follicular phase AUC t~/2 2304 2.5 3632 4.3 3754 4.9 3463 2.5 4749 2.2 8226 11.5 2182 6.3 5845 5.3 3779 4.6 7289 8.7 4522 5.3
CL 1.63 1.16 0.93 0.82 0.89 0.46 1.74 0.72 0.98 0.62 0.99
V 0.35 0.43 0.40 0.18 0.17 0.46 0.95 0.33 0.39 0.46 0.41
Luteal phase AUC 2392 3431 4289 4484 4572 8556 2937 10207 3906 7008 5178
tl~2 2.4 3.7 4.3 3.9 3.0 11.5 2.5 13.9 2.3 6.9 5.4
CL 1.57 1.23 0.81 0.63 0.92 0.44 f.29 0.41 0.95 0.64 0.89b
V 0.32 0.39 0.31 0.21 0.24 0.44 0.28 0.49 0.19 0.39 0.33
a Drug dose = 250 mg caffeine, administered R O.; AUC = area under the concentration/time curve (gg- ml- 1.min); tli2= elimination half-life (h); CL=systemic clearance rate (mI.kg-a.min ~); V = apparent volume of distribution (1. kg ~)
b Luteal phase systemic clearance is significantly lower than in the follicular phase (P < 0.05, see text)
the magnitude of the reduction in caffeine elimination. No similar trend was observed for changes in progesterone level. In the data from the follicular phase, the oestrogen levels were correlated with the A U C (r = 0.69, P < 0.03) and elimination tl/2 (r = 0.73, P < 0.02), with a similar, but negative, trend for CL (r =-0.55, NS). All correlations suggested that higher levels of oestrogens were associated with lower rates of caffeine elimination within the follicular phase. In contrast, levels of progesterone were not related to the pharmacokinetic variables. Similar effects of oestrogen level appeared within the luteaI phase, although they did not reach statistical significance. There were trends correlating the oestrogen level with the A U C (r = 0.53, NS), tl/2 (r = 0.57, P < 0.09), and with the systemic clearance (r =-0.37, NS), each again suggesting that a higher oestrogen level might be associated with a reduced rate of caffeine elimination. In contrast, there was a trend to a positive relationship between luteal progesterone level and the rate of systemic clearance (r = 0.54, NS), consistent with general observations that oestrogen and progesterone can have opposite effects on many physiological processes. Finally, the effect of luteal phase cycle day was investigated, given that the hypothesis suggested that caffeine elimination should be slower as the woman approached the onset of the menses. This hypothesis was confirmed by the significant correlation observed between luteal phase cycle day, which is counted as a negative number, and systemic clearance rate, CL (r = -0.63, P < 0.05) and supported by similar trends for A U C (r = 0.40, NS) and for tl/2 (r = 0.34, NS). Women who were studied closer to the onset of their next menses had a lower rate of caffeine elimination.
levels that are associated with the phases of the normal menstrual cycle may be associated with variation in the rate of caffeine elimination. Confirming this hypothesis, there was a significant average reduction in total systemic clearance during the luteal phase that was associated with a trend to a larger 24-h AUC. In contrast, no cycle phase effects were observed on the average caffeine elimination tl/2 or the apparent volume of distribution. These observations confirm an earlier report of slowed caffeine elimination during the luteal phase [11] in which a reduction of similar magnitude (roughly 25 %) was found, although significant effects were also observed in a lengthened tip2 as well as reduced systemic clearance and increased AUC. The correlational exploration of the individual variation in cycle effects provided additional evidence that normal variability in sex steroid levels can influence caffeine elimination. Higher oestrogen levels were consistently associated with relatively slower rates of caffeine elimination, while the opposite trend was seen for progesterone levels. The small sample size limited the level of statistical significance reported for many of the correlations, but if the associations are considered in terms of the size of the observed effects (defined as R 2 or the proportion of variability in the pharmacokinetic values that can be accounted for by the menstrual cycle variables), then the relationships are more impressive. In many of the reported relationships, the menstrual cycle variables accounted for 25 % to 50 % of the total variability among subjects observed in the measures of caffeine elimination. The average changes in caffeine elimination associated with menstrual cycle phase were relatively small, compared to changes associated with OCS use or with pregnancy. In three studies of OCS effects [4, 5, 7], OCS use was associated with increases in tl~ of 47 % to 235 % and with reductions in caffeine clearance ranging from 39 % to 66%. The third trimester of pregnancy was associated with a 200 % increase in ti/z and 70 % reduction in clearance [8]. In the present study, the reduction in clearance from the follicular to the luteal phase only averaged 11% and there was no effect on average tl/2. However, it should not be surprising that menstrual cycle effects are smaller
Discussion
The present study demonstrates slowing of caffeine elimination in the luteal phase of the normal menstrual cycle, and suggests that the normal fluctuations in sex steroid
546 in magnitude, given that the hormonal changes during pregnancy are an order of magnitude greater than those seen during the normal menstrual cycle. Although the average effect of menstrual cycle phase on caffeine elimination was small, the present study does suggest the possibility that cycle-related changes might be clinically important in some women, particularly those who have a higher oestrogen level and a greater increases in oestrogen level in the luteal phase. Such differences are suggested by the correlational explorations. O f the ten w o m e n in the study, the three that had the greatest increases in oestrogen all showed a decrease in caffeine clearance of m o r e than 23 %, and an increase in A U C of m o r e than 25 % in the luteal phase. However, even in that group, the response was less than that associated with OCS use, and it is unlikely that such effects would lead to such accumulation of caffeine that toxic symptoms would be produced. In conclusion, the evidence suggests that caffeine elimination may fluctuate across the menstrual cycle, with slower elimination in the luteal phase prior to the onset of menstruation. The reduction in caffeine clearance is associated with a slightly increased exposure to ingested caffeine ( A U C ) , that might increase the likelihood of systemic accumulation of caffeine if it were repeatedly ingested throughout the day. The clinical implications of these relatively small increases in caffeine exposure are unknown, but it is unlikely that these effects alone would lead;to symptoms of caffeine intoxication in the luteal phase. However, given that caffeine has been shown to intensify the cardiovascular and neuroendocrine effects of psychological stress in other studies, and that excessive levels of caffeine can produce toxic symptoms, these menstrual cycle effects m a y deserve further attention in investigations of the premenstrual syndrome.
Acknowledgements. We thank Dr. C. E. Cook, Research Triangle Institute, for his kind gift of the antiserum for the caffeine radioimmunoassay and Mr. T. Zimmerman for his diligence in conducting the assays for plasma caffeine level. Data were managed and analysed in part using the CLINFO Data Analysis System of the Clinical Research Unit. Research supported by NIH Research Grant HL29968-05 NIDA Research Grant DA-06857-01, and a grant (RR-30)
from the General Clinical Research Centers Program of the Division of Research Resources, National Institutes of Health.
References 1. Barone JJ, Roberts H (1984) Human consumption of caffeine. In: Dews PB (ed) Caffeine. Springer, Berlin Heidelberg New York, pp 5%73 2. Greden JF (1974) Anxiety or caffeinism: a diagnostic dilemma. Am J Psychiatry 131:1089-1092 3. American Psychiatric Association (1987) Diagnostic and statistical manual of mental disorders, 3rd ed, revised. Washington, DC, pp 138-139 4. Patwardhan RV, Desmond PV, Wilkinson GR, Schenker S (1980) Impaired elimination of caffeine by oral contraceptive steroids. J Lab Clin Med 95:604-608 5. Callahan MM, Robertson RS, Branfman AR, McComish MF, Yesair DW (1983) Comparison of caffeine metabolism in three nonsmoking populations after oral administration of radiolabeled caffeine. Drug Metab Dispos 11:211-217 6. Reitveld EC, Broekman MMM, Houben JJG, Eskes TKAB, van Rossum JM (1984) Rapid onset of an increase in caffeine residence time in young women due to oral contraceptive steroids. Eur J Clin Pharmaco126:371-373 7. Abernethy DR, Todd EL (1985) Impairment of caffeine clearance by chronic use of low-dose oestrogen-containing oral contraceptives. Eur J Ctin Pharmaco128:425-428 8. Aldridge A, Bailey J, Neims AH (1981) The disposition of caffeine during and after pregnancy. Semin Perinatol 5:310-314 9. Cook CE, Taltent CR, Amerson E, Myers MW, Kepler JA, Taylor GC, Christensen HD (1976) Caffeine in salvia and plasma by a radioimmune assay procedure. J Pharmacol Exp Ther 199: 679686 10. Neims AH, von Borstel RW (1983) Caffeine: metabolism and biochemical mechanisms of action. In: Wurtman R J, Wurtman JJ (eds) Nutrition and the brain, vol 6. Raven Press, New York, pp 1-29 11. Balogh A, Irmisch E, Klinger G, Splinter FK, Hoffmann A (1987) Untersuchungen zur Elimination von Coffein und Metamizol im Menstruationszyklus der fertilen Frau. ZentralbI Gyn~iko1109:1135-1142 J. D. Lane, PhD Department of Psychiatry, Box 3830 Duke University Medical Center Durham, NC 27710 USA
