JOSEPH W. GOLDZIEHER, M,D.t C, BRANDON CHENAULT, M.D. ARMANDO DE LA PENA, M.S. TAZEWELL S. DOZIEIt DUANE C. KRAEMER, D.V.M., PH.D.

Transcription

1 FERTILITY AND STERILITY Copyright,< 1978 The American Fertility Society Vol. 29, No.4, April 1978 Printed in U.SA. COMPARATIVE STUDIES OF THE ETHYNYL ESTROGENS USED IN ORAL CONTRACEPTIVES: EFFECTS WITH AND WITHOUT PROGESTATIONAL AGENTS ON PLASMA ANDROSTENEDIONE, TESTOSTERONE, AND TESTOSTERONE BINDING IN HUMANS, BABOONS, AND BEAGLES* JOSEPH W. GOLDZIEHER, M,D.t C, BRANDON CHENAULT, M.D. ARMANDO DE LA PENA, M.S. TAZEWELL S. DOZIEIt DUANE C. KRAEMER, D.V.M., PH.D.* Southwest Foundation for Research and Education, San Antonio, Texas The effects of ethynylestradiol or mestranol given in cyclic fashion, with and without a progestational compound (norethindrone acetate, dl-norgestrel, or megestrol acetate), on plasma androgens and their binding were examined in adult women, female baboons, and beagles. The two estrogens are equivalent in their effect, and there were essentially no dose-related differences over the range examined. In human subjects, the estrogens increased total testosterone and testosterone binding, and decreased free testosterone. In baboons, estrogen produced a transient decrease in total testosterone and an increase in binding. The levels of progestational agents used did not affect total testosterone in humans, as is commonly observed with commercial agents, but did decrease it in baboons. Percentage binding was decreased in both species by the 19-nor compounds, but not by megestrol. Androstenedione levels were unaffected in human subjects, but effects of both estrogens and progestins were seen in baboons. Because of the very low levels of androgens in female beagles, this species did not lend itself well to a study of this kind. However, an increase in testosterone binding was induced by estrogen even in the absence of testosterone/estrogen-binding globulin. The biologic activity and metabolic disposal of circulating androgens and estrogens are largely governed by their binding to plasma proteins, the most important being the globulin designated as TEBG (testosterone/estrogen-binding globulin) Received October 10, 1977; revised November 21, 1977; accepted November 23,1977. *Clinical studies supported by Contract csdj2821, Office of Population, United States Agency for International Development; animal studies supported by Contract HD , Center for Population Research, National Institute of Child Health and Human Development, United States Public Health Service. treprint requests: Joseph W. Goldzieher, M.D., Southwest Foundation for Research and Education, P.O. Box 28147, 8848 W. Commerce Street, San Antonio, Tex *Present address: Institute of Comparative Medicine, and Department of Veterinary Physiology and Pharmacology, Texas A & M University College of Veterinary Medicine, College Station, Tex or SHBG (sex hormone-binding globulin). The concentration of SHBG is influenced by the prevailing levels of estrogens, androgens, and thyroid hormones and by the administration of synthetic progestational compounds. 1 Thus, the concentration of total plasma testosterone or estrogen may not reflect accurately the concentration of hormone free to interact with the receptors of target tissues. Measurement of the dynamic balance 2 between free and protein-bound hormone as it exists in vivo is not a simple problem. Further obscurity is introduced by the various ways in which steroid-binding or binding capacity is expressed. Finally, laboratory animals differ in regard to SHBG: it is present in man and certain nonhuman primates; absent in rats, guinea pigs, hamsters, cats, and dogs; and qualitatively different (from human SHBG) in rabbits. 3 The situation is less complex for the androgen pre-

2 Vol. 29, No.4 COMPARATIVE STUDIES OF ETHYNYL ESTROGENS 389 hormone, androstenedione, which is only loosely associated with albumin and not with SHBG. It is well known that estrogen or androgen therapy, as well as contraceptive steroids, alter:s the total plasma testosterone level. The effects of various contraceptive formulations differ depending on the respective effects of the components. We have therefore undertaken a study of the effects of the synthetic estrogens ethynylestradiol (EE) and mestranol (ME), with and without concomitant synthetic progestins, to examine the magnitude ofthe changes in total testosterone and androstenedione; the percentage binding of testosterone (by equilibrium dialysis); and-by calculation-the levels of unbound (and presumably biologically active) testosterone in plasma of normal women, baboons, and beagle dogs. MATERIALS AND METHODS The details of the investigational design have been described. 4 Briefly, informed normal adult female volunteers were assigned randomly (as far as possible) to treatment with oral doses of 50 or 80 f,-tg of EE or 50, 80, or 100 f,-tg of ME daily in the usual 21-day contraceptive regimen. The women taking the lowest estrogen dose had to be using intrauterine devices (IUDs), which upset random assignment to this group. After six cycles of therapy, the subjects in each of the five groups were separated into three subsets which received, in addition to the same dose of estrogen, a progestational agent consisting either of norethindrone acetate (2.5 mg/day), megestrol acetate (2 mg/day), or dl-norgestrel (0.5 mg/day). This combination was administered for another six cycles. Blood sampling for various endocrine and metabolic studies was always carried out toward the end of the treatment cycle. A group of women using only intrauterine devices was sampled in the same way, as a control population. Inbred female beagles of an average weight of 6.9 kg were assigned randomly to oral treatment regimens consisting of 1 or 4 f,-tg/kg/day of ethynylestradiol or mestranol for 21 days, followed by 7 days without drug, to simulate the human regimen. This program was used for four cycles, after which each of the four groups was subdivided into three groups, and one of the three progestational agents (norethindrone acetate, 1 mg/day; megestrol acetate, 1 mg/day; or dlnorgestrel, 0.5 mg/day) was given together with the estrogen for another four cycles. The treatment was then discontinued, and the animals were studied once again 3 months later. Blood samples were obtained before treatment, during the last week of treatment cycles 1, 2, 4, 5, 6, and 8, and once again 3 months after the end of treatment. A similar design was used for the study of adult female baboons weighing 13.6 kg on the average. A total of 60 beagles and 60 baboons were used. There were two 30-animal replications of each protocol in each species to randomize time- and season-dependent biases. In addition, entry of the 30 animals in each replicate into the experiment was staggered for logistical reasons. This further randomized time-dependent biases. The animals were assigned to the various drug-dose groups by computer-generated random allocation. Pills containing the various dose levels of ethynylestradiol, mestranol, norethindrone acetate, megestrol acetate, or dl-norgestrel were prepared by Wyeth Laboratories, Philadelphia, Pa., with precise quality control to ensure homogeneous bioavailability. All blood samples were obtained in the morning. Baboons were under minimal Sernylan tranquilization; no sedation was required for the beagles. The plasma was deep-frozen immediately. In the human studies, testosterone and androstenedione were separated from plasma by thin-layer chromatography.5 The androstenedione was reduced to testosterone by reaction with 0.02% sodium borohydride for exactly 15 minutes, and re-extracted. Both testosterone fractions were then quantitated by competitive protein binding.5 The studies in baboons and beagles were performed at a later date, and plasma extracts were assayed for testosterone by radioimmunoassay6 using antiserum raised against testosterone-11a-bovine serum albumin. Androstenedione was assayed directly by radioimmunoassay according to the procedure and with the somewhat nonspecific antibody available from Endocrine Sciences, Tarzana, Calif. Protein binding of testosterone and androstenedione was determined at 37 C on a 1:5 plasma dilution by the equilibrium dialysis method of Forest et a1.,7 and "free" testosterone was calculated from the total testosterone concentration and the percentage bound for each sample, without correction for dilution. RESULTS Total Plasma Testosterone Human Subjects. In the subjects using IUDs, the total plasma testosterone concentration

3 390 GOLDZIEHER ET AL. April 1978 '0 "- 0> C 100 == ~feroid ] WOMEN BABOONS O~~=U~~~~~~~~~~~~ o I Cycle FIG. 1. Levels of total plasma testosterone, by treatment cycle, in normal women with an IUD, in women on steroid regimens (all estrogen types and dosage levels combined), and in baboons (estrogen types and dosage levels combined). Ninety-five per cent confidence limits are indicated vertically. Treatment cycles in baboons are plotted to coincide with the treatment schedule in women, although only four cycles of each regimen were used in baboons, compared with six in women. Thus, the fourth (last) cycle of estrogen treatment in baboons is plotted to coincide with the sixth (last) cycle of estrogen treatment in women, and the same for the estrogen-plus-progestin cycles in the two species. The post-treatment values (cycle 11) are also indicated for baboons. averaged 29.7 ng/dl (N = 140) with 95% confidence limits (CL) of 27.7 to The control level for the subjects electing to take the medications averaged 34.8 (CL 32.5 to 37.3) ng/dl (N = 163). Comparison of EE with ME at doses of 50 or 80 J.Lg/day showed no statistically significant differences. The data for EE and ME were therefore combined, and the effect of dose was examined. Compared with the control level (34.8 [CL 32.5 to 37.3], N = 163), the testosterone values at 50, 80, and 100 J.Lg/day were not significantly different at any time through cycle 6. It may be concluded that, in this dose range, the two estrogens produce an equivalent effect. The data have been combined in Figure 1, which displays the total testosterone levels over the duration of the experiment. Estrogen treatment produced a total testosterone increase of the order of 4 ng/dl. The effect of the three progestational compounds on the estrogen-treated individuals is summarized in Table 1. None of the progestins had a significant effect, even when tested statistically at each level and by each type of estrogen treatment. Baboons. The control value for total testosterone was 60.6 ± 37.3 ng/dl (N = 60). There was no statistically significant difference between EE and ME nor between the dose levels of 1 and 4 J.Lg/kg/day; the data from all of the animals in the two replicate experiments were combined. The values are shown in Figure 1, with the time scale adjusted to make the treatments comparable to the human schedule. An unexpected, transient fall in the first treatment cycle was observed in all of the dose-level subsets and in both replicates of the entire experiment. After one cycle of exposure to the additional progestational compounds, the level fell from 67.4 ± 37.2 to 47.8 ± 41.0 ng/dl; 15 of the 60 determinations were below the limit of detectability (10 ng/dl). After the fourth cycle of estrogen-progestin treatment the level had fallen to 32.9 ± 23.5 ng/dl, with 11 of 60 determinations below the limit of detectability. At the doses used, the three progestational agents were similar to one another in their depressant effect on total testosterone. After treatment was discontinued, total testosterone returned to control values. Beagles. The total plasma testosterone in female beagles was extremely low to begin with, averaging 20.5 ± 24 ng/dl, with 53% of the measurements below the limit of detectability. On the assumption that the type and dose of estrogen did not make a difference, the data were combined and examined in terms of the percentage of total plasma testosterone determinations which were in the detectable range. This is shown in Figure 2, with statistically significant differences indicated by asterisks. There appears to have been a transient decrease during estrogen treatment; during added progestational treatment the results were totally erratic. The normal level of testosterone in the female beagle, and its variability, do not appear to lend themselves to a study ofthis kind. TABLE 1. Effect of Progestational Compounds on Total Plasma Testosterone Level in Estrogen-Treated Subjects Norethindrone acetate dl-norgestrel Megestrol acetate Testosterone leve}a ngldl 36.5 ( ) 46.7 ( ) 38.4 ( ) avalues in parentheses are confidence limits. Before progestin No. of subjects After 6 cycles of progestin Testosterone level ngldl 39.5 ( ) 35.4 ( ) 30.7 ( ) No. of subjects

4 Vol. 29, No. 4 COMPARATIVE STUDIES OF ETHYNYL ESTROGENS Androstenedione...., IUD CONTROL STEROID ([) ~ g 25 ::0 0 a U C t' 75 Q 50 a I Testosterone ~ 96 ~!: t Testosterone - human...l...;... L... j ~ 93 ct 84 Androstenedione - humon 82 76~~~~~LC~~~~UCl ~Cl~~ o I Cycle a a I CYCLE 8 II FIG. 2. Plasma levels of total testosterone and androstenedione, by treatment cycle, in beagle dogs, Since many of the values were below the limit of detectability (10 ng/dl), the results are shown in terms of the percentage of the determinations which were in the measurable range, An increase or decrease in this percentage presumably reflects an increase or decrease in the mean plasma steroid levels. Statistically significant changes from one cycle to the next are indicated by an asterisk. Percentage Binding of Testosterone Human Subjects. The group using an IUD yielded a mean percentage binding of 93.9 ± 2.4 (N = 140); that for the control cycles of the treated subjects was 94.1 ± 2.7 CN = 162). There was no significant difference between EE and ME, and the data from the two estrogens were combined. At 50 /Lg/day, percentage binding increased significantly by the third cycle (93.6 ± 2.4% [N = 51] versus 97.2 ± 1.2% [N = 49]; P < 0.05). Both the 80- and 100-/Lg/day dose levels produced a significant increase (P < 0.001) after the first cycle. There was no difference between the effects of 80 and 100 /Lg/day. All of the data were combined and are presented in Figure 3, which shows an increase from a control level of 94.1 ± 2.7 CN = 162) to a level of 97.7 ± 1.8 (N = 106) after six cycles. The effect of the progestational compounds was difficult to detect. When analyzed by estrogen and by dose level, dl-norgestrel in combination with 100 /Lg/day of ME reduced the binding from 98.0 ± 0.4% (N = 22) to 95.5 ± 0.7% (N = 8) (P < 0.02), and in combination with EE (80 /Lg/day) FIG. 3. Percentage binding of plasma testosterone and androstenedione in human subjects, by cycle, as determined by equilibrium dialysis. One standard deviation is indicated by each vertical bar. it reduced the level from 97.9 ± 0.5% (N = 26) to 93.9 ± 1.1% CN = 8) (P < 0.001). When all of the data were pooled and compared (cycle 6 against cycle 12), the reduction in binding in norethindrone acetate-treated subjects from 97.8 ± 0.8 (N = 43) to 95.5 ± 1.8% (N = 37) was statistically significant (j> < 0.02), as was the reduction in the entire dl-norgestrel group, from 97.7 ± 0.9 (N = 38) to 94.9 ± 1.4 (N = 40) (P < 0.001). The change observed in megestroltreated subjects, from 97.6 ± 1.0 (N = 34) to 96.8 ± 2.0% (N = 35) was not statistically significant. The f3-error was 0.13, indicating that there is a 13% chance of a "false negative." Baboons. The percentage binding of testosterone in female baboons was higher than that in women. The variance was also considerably less, with a standard deviation of about 0.5%, compared c e 100 (j; 98.2 ([) BABOON ~ 96 '0 Ol c 86 is c iii 84 C t' 82 rf 80 a I II CYCLE BEAGLE FIG, 4. Percentage binding of plasma testosterone, by treatment cycle, in baboons and beagles.

5 392 GOLDZIEHER ET AL. April 1978 CYCLE FIG. 5. Percentage binding of testosterone by beagle plasma, by treatment cycle. Group A (N = 20) received norethindrone acetate starting with cycle 5; group B received dl-norgestrel, and group C received megestrol acetate. The consistency of the changes in the three randomly allocated sets of animals is noteworthy. with 2.5% in women (Fig. 4). The values for estrogen-treatment cycles 2 and 4 (97.7 ± 0.45% and 97.9 ± 0.41%) were higher than the control level; the absolute change was very small and statistical significance could not be demonstrated. The increase, however, was observed in all of the experimental subsets, and it may therefore be a real change. No difference between EE and ME or between 1 ILg/kg and 4 ILg/kg dose levels could be detected. The addition of progestational compounds produced a small decrease in binding, from a starting level of 97.8% ± 1.04% to 96.4% ± 1. 7% after the fourth cycle of combined treatment. The value returned to pretreatment levels by cycle 11. No difference between the three progestational compounds could be detected. Beagles. SHBG is reportedly absent from the plasma of dogs, and this was reflected in the binding level of 83.8% ± 3.1%, which is attributable to low-affinity association with albumin. The variation by cycle is shown in Figure 4. Because of the large variance, these differences were not statistically significant. However, an examination ofthe individual groups (which were assigned at the beginning of the experiment to receive a particular progestational compound from cycle 5), shows a consistency in the.alterations (Fig. 5) which suggests that these changes are real and that the statistical analysis is not appropriate. On the institution of estrogen therapy there was an increase in percentage binding, and there was a consistent and comparable decrease induced by all three progestational compounds. "Free" Plasma Testosterone Human Subjects. The mean control value for free testosterone in the IUD users was 3.4 ± 2.0 ngldl (N = 140) and in the control cycle for the test subjects it was 3.6 ± 2.6 ngldl (N = 162). There was no difference between EE and ME and there were no dose-related differences (except for the delayed effect of the 50-lLg dose on percentage binding). Estrogen treatment (both types and all doses combined) caused a prompt and statistically significant decrease in free testosterone from 3.6 ± 2.6 ngldl (N = 162) to 2.3 ± 1.7 (N = 143). In the third treatment cycle, the level had fallen to 2.0 ± 1.6 (N = 129) (Fig. 6). The variations in the IUD cycles are not statistically significant. The effects of the progestational compounds were slight but definite. When data from all the estrogen regimens used with each progestin were combined, the increase with norethindrone acetate from cycle 6 (1.10 ± 0.99 [N = 43]) to cycle 12 (2.17 ± 1.89 ngldl [N = 37]) was significant (P < 0.01) and the increase withdl-norgestrel from 1.20 ± 0.86 (N = 38) to 1.93 ± 1.34 ngldl (N = 40) was also significant (P < 0.01), whereas the increase with megestrol acetate (from 1.35 ± 0.98 [N = 34] to 1.74 ± 2.10 ngldl [N =35]) was not. The f3-error indicates a 68% chance of this being a false negative. Baboons. Analysis of variance showed no difference between EE and ME and no difference between the two dose levels. The data were combined. On initiation of estrogen treatment, the free plasma testosterone level decreased from a control value of 1.57 ± 0.34 ngldl (N o E U) o Q , IUD ] WOMEN STEROID BABOONS 6 Cycle FIG. 6. Levels of free testosterone, by treatment cycle, in human subjects using an IUD and in human subjects and baboons undergoing the steroid-treatment regimens. Data for all estrogen types and dose levels at each cycle are combined for each species. For correlation of baboon and human treatment regimens, see legend to Figure 1. 12

6 Vol. 29, No.4 COMPARATIVE STUDIES OF ETHYNYL ESTROGENS ) to a value of 1.11 ± 0.07 ng/dl (]V = 60) at the end of the first treatment cycle (P < 0.05). The level then rose during the subsequent estrogen cycle and returned to the control level by the fourth cycle of estrogen exposure. It was unaffected by the addition of any of the progestational compounds (Fig. 6). Beagles. The average value for detectable plasma free testosterone was 3.3 ng/dl, but there were 31 animals in which the total, and therefore the free testosterone level, was below the limit of detectability. There was a significant increase in the number of undetectable values after the first cycle of estrogen treatment, but no change thereafter. When progestational compounds were added to the drug regimen, the changes were erratic. Plasma Androstenedione Human Subjects. In the subjects using IUDs, the plasma androstenedione level averaged 90.9 ± 78.7 ng/dl (]V = 140). The control level for the subjects electing to take medications was ± 94.3 ng/dl (N = 162). Statistical comparison of EE and ME at 50 J,Lg/day and 80 J,Lg/day showed no significant differences, and the data for the two types of estrogens were combined. It may be seen (Fig. 7) that neither the estrogens nor the progestational compounds produced any change. Baboons. The plasma androstenedione level averaged 146 ± 62.1 ng/dl (N = 60) in the control cycle. Comparisons of EE and ME and the two dose levels showed no differences, and the data were combined. In the first cycle of 250,..., IUD CONTROL] WOMEN < STEROID... BABOONS ~ I ~ 200 : ~. i;:!: l -OQl~ 150 l,:.~. ~ ~ ~ /(11) ~ -: r: \~ ~ '1:,// 1 _:l T -~ j: > v, E 4) (5)'",: ~ 1/ 2 I 00 >"'~~~---f'-""":4-_-~-o.: i -g.. t... L <:::( ~ 50 (f) o ii o ~~.)..."..."...; 8) Cycle FIG. 7. Levels of androstenedione, by treatment cycle, in human subjects using an IUD and in human subjects and baboons undergoing the steroid-treatment regimens. Data for all estrogen types and dose levels at each cycle are combined for each species. For correlation of baboon and human treatment regimens, see legend to Figure estrogen treatment, the level fell significantly (P < 0.01); it rose again in the second cycle and remained at about this level for the remainder of the estrogen exposure. As in the case of testosterone, this transient decrease was observed in the dose- and estrogen-type subsets and in both replicates. It is not likely to have been of random or methodologic origin. In the first cycle of progestational treatment there was no change, but the values fell thereafter from 129 ± 63 to 105 ± 42 ng/dl (P < 0.02) and remained low. There was no difference among the three progestins with regard to their effect on lowering the level. There was a rebound to 185 ± 56 ng/dl 3 months after the end of treatment (Fig. 7). Beagles. The control value averaged 25.9 ± 21 ng/dl, but 18 of the 57 measurements were in the undetectable range. The fall in the number of detectable values in shown in Figure 2. Interpretation of values so close to the limits of the methodology (10 ng/dl) is risky. DISCUSSION Published studies comparing the effects of EE and ME in humans have almost invariably relied on commercial preparations of combined oral contraceptive tablets, taking the stated estrogen content at face value. Since there is no evidence that these estrogens are present in bioequivalent form, nor any readily available data as to the variance of the estrogen content, such comparisons are open to serious question. In previous studies 5 with tablets of controlled quality, we have shown that EE and ME were biologically equivalent in women over the dosage range of 50 to 100 J,Lg/day with respect to their effect on endometrial growth, inhibition of ovulation, suppression of plasma folliclestimulating hormone and luteinizing hormone, and cortisol binding by corticosteroid-binding globulin (CBG). We have now demonstrated equivalence in producing a slight elevation (about 4 ng/dl) of total plasma testosterone in women. Others 8 have already shown that this estrogenic response is rapid, and our findings in human subjects confirm that a stable level is reached within the first 21-day cycle of treatment. In baboons, there was a significant fall in total testosterone followed by a return to normal levels. Both the experimental design and the statistical analysis of the data suggest that this transient fall was not a random event or a

7 394 GOLDZIEHER ET AL. April 1978 methodologic artifact. In female beagles the base line of plasma testosterone was extremely low, ranging about 20 ng/dl; the data suggest that estrogen may have produced a transient decrease in total testosterone; thus the human response differs from that of both animal species investigated. The addition of the progestational agent to the estrogen regimen produced no significant changes in the total testosterone level of the human subjects at the dose levels used. In baboons, on the other hand, all three progestins produced a significant and equal fall of almost 50%. Such changes could not have been detected in the beagles, if they occurred, for technical reasons. Since other experiments in humans have shown depressant effects of progestins, these data suggest that the baboon may be a qualitatively similar, but more sensitive, indicator of this effect than is the human. The percentage binding of testosterone in the human subjects averaged 94% ± 2.5%, or, conversely, the percentage unbound was 6.0% ± 0.16%. Other published values for free testosterone, some on small numbers of subjects and based on a variety of techniques, range to values as low as 0.90%. Methodologic problems will have to be resolved before consistent figures are established. In our human studies, the percentage binding rose on the institution of estrogen therapy, as has been reported previously. The level stabilized at about 97.5% binding (2.5% unbound), regardless of dosage or number of cycles. The addition of nor ethisterone acetate produced a decrease of about 2% in binding, and dl-norgestrel a decrease of about 3%. The decrease with megestrol acetate was not statistically significant. In baboons, the percentage binding was higher (about 97%). The two estrogens at 1 and 4 jlg/kg/day were equipotent, and produced a prompt, small (0.5%) increase in percentage binding. The progestational compounds produced changes which resembled the observations in human subjects: all three had an equal effect in reducing the percentage binding by about 1.4%. In the beagles, the percentage binding, presumably related entirely to albumin, ranged around 84%. To our surprise, this binding also showed an increase on estrogen treatment and a decrease from this level on the addition of progestational compounds, although no changes were observed in total protein (albumin was not measured specifically). The mechanism of these changes is not apparent, and requires further investigation. The level of free (unbound) testosterone was calculated from the total testosterone value and percentage binding for each sample. In human subjects it ranged around a value of 3.5 ng/dl. All of the estrogen regimens had a similar effect, depressing the level to about 2.3 ng/dl. The effect stabilized promptly. In the group of subjects treated with norethindrone acetate, the estrogen-depressed free testosterone rose by about 1 ng/dl and in the subjects treated with dl-norgestrel it rose about 0.7 ng/dl. The increase of 0.4 ng/dl with megestrol acetate was not statistically significant. In baboons the level of 1.57 ng/dl of free testosterone fell to 1.11 ng/dl in the first cycle of estrogen therapy and then gradually returned to control levels during continued estrogen treatment. There was no difference in the effect of the various estrogen regimens. In contrast to total testosterone measurements, no effect of the added progestational steroids could be detected. In the female beagles, an attempt was made to estimate the effect on the basis of the number of animals in which the level fell below the limit of detectability. There was an increase in the number of such extremely low values on estrogen treatment. No inferences could be drawn from the values obtained during progestational therapy. Plasma androstenedione levels in women were unchanged by treatment with estrogen or estrogen-progestin combinations, in agreement with previous observations. However, in hirsute women with elevated testosterone and androstenedione levels, combinations of ethynylestrogen and norethindrone have been shown to decrease the plasma level of both androgens. 8 In baboons, there was a decrease in total androstenedione in the first cycle of estrogen treatment, followed by a return to control values. The progestational compounds produced an equivalent and significant fall, which, however, did not take place until the second treatment cycle. In the beagles, estrogen treatment increased the number of undetectable values. The (nonspecific) binding of androstenedione in the human subjects averaged 78% ± 3% and was unaffected by hormone treatments. This value is very close to the 77% binding reported by Forest et au Since this low-affinity binding does not appear to have any biologic significance, it was not examined further.

8 Vol. 29, No.4 COMPARATIVE STUDIES OF ETHYNYL ESTROGENS 395 Comparison of the effects observed by us in human subjects with reports in the literature is made difficult by the variety of ways in which binding is reported and by the fact that commercial contraceptive preparations were almost invariably used. Briggs 9 reported a doubling of binding capacity with 30 j.tg/day of ethynylestradiol as compared with controls; the effect of 50 j.tg/day was the same, but 75 j.tg/day produced a trebling of binding capacity; absolute values were not given. The combination contraceptives are reported to produce increases, no change, or decreases in binding capacity, depending on the amount of the estrogen and the amount and type of progestational agent. The unbound testosterone concentration may show no change or a decrease, especially if the level was initially elevated, as in women with hirsutism or polycystic ovarian disease.8, 10, 11 Injected microcrystalline medroxyprogesterone acetate may produce no change or a decrease in binding. N orgestrel, chlormadinone acetate, and lynestrenol also tend to decrease binding capacity, but the dosages and treatment regimens are so various that meaningful comparisons are not possible. In general, the baboon data parallel these observations except for the puzzling transient decrease in total and free testosterone in the first estrogen cyle, at a time when the percentage binding is increasing slightly but significantly. In the baboon the decrease in free testosterone also was not maintained throughout estrogen therapy, as was observed in human subjects. The implications of these changes with respect to the disposition of testosterone are not simple. In general, total and free testosterone levels are parallel because SHBG concentration does not vary systematically during the human menstrual cycle. There is a relationship between the concentration of unbound testosterone and the metabolic clearance rate, and this relationship changes if the binding capacity is altered. 12 Mercier-Bodard et al. I3 believe that SHBG also has a direct effect on the uptake and metabolism of testosterone (by the human prostate), independent of its effect on the plasma binding of testosterone. Steroids also affect the metabolism of testosterone in other ways: estrogens inhibit the 50' reduction of testosterone to dihydrotestosterone and also the conversion of androstenedione to testosterone Progestins, on the other hand, increase 5a-reductase activity.16 In addition, Rosenfield 17 has shown that the metabolic clearance of testosterone in women is a function of adiposity, and others have shown that the aromatization of androstenedione is also affected by the size of the body fat deposits. In the dog, the lung is a very important organ in the inter conversion of androstenedione and testosterone, and the factors which affect steroid metabolism by this organ are as yet unexplored. 18 Although both SHBG and CBG are synthesized in the liver and increased by estrogen treatment, the effect of progestational compounds on the two proteins is not parallel. Van Kammen et al. I9 and Givens et al,1 ' II have reported that norgestrel decreased SHBG but not CBG binding capacity. APPENDIX Statistical Analysis of the Data. The data on human total plasma testosterone conformed best to a lognormal distribution; the results reported herein have been retransformed from the log values, and the variances expressed as 95% confidence limits in the usual fashion. Human and baboon androstenedione and baboon testosterone values showed a distribution which did not depart strongly from normality. The values for free testosterone showed a very narrow range, and the Kolmogorov-Smirnov test could not discriminate between normal and lognormal distributions; the data were therefore treated as normally distributed. Percentage binding data in all instances were studied after arc sin transformation. The plasma testosterone and androstenedione values for beagles were so low, and yielded so many values below the limit of detectability, that there was no valid method for determining the distribution of the data; they were arbitrarily treated under the assumption of normal distribution. The next task was to compare by t-test each experimental point (species, cycle, dose) for the two estrogens, to determine whether the type of estrogen, at equal dosage, produced any difference. Then, a similar set of analyses was performed to compare different doses of the same estrogen, by species and by cycle, for each of these variables. In cases where no differences were observed between types and/or doses of estrogen, it was considered legitimate to combine treatment groups, thereby increasing sample size. Next, a three-way analysis of variance examined the effect of estrogen dose, cycle of use, effect of the added progestational compound, interaction between estrogen dose and cycle number, between the estrogen dose and type of progestational agent, and interaction of estrogen, cycle number, and progestational agent. Finally, an analysis for f3-error was carried out to estimate the likelihood of missing real differences (i.e., to estimate the probability of "false negatives").

9 396 GOLDZIEHER ET AL. April 1978 REFERENCES 1. Anderson DC: Sex-hormone-binding globulin. Clin Endocrinol (Oxf) 3:69, Heyns W, De Moor P: Kinetics of dissociation of 17f3- hydroxysteroids from the steroid binding f3-globulin of human plasma. J Clin Endocrinol Metab 32:147, Corvol P, Bardin CW: Species distribution of testosterone-binding globulin. Bioi Reprod 8:277, Goldzieher JW, de la Peiia A, Chenault CB, Cervantes A: Comparative studies of the ethynyl estrogens used in oral contraceptives. Effect on plasma gonadotropins. Am J Obstet GynecoI122:625, Rosenfield RL, Eberlein WR, Bongiovanni AM: Measurement of plasma testosterone by means of competitive protein binding analysis. J Clin Endocrinol Metab 29:854, Hillier SG, Brownsey BG, Cameron EHD: Some observations on the determination of testosterone in human plasma by radioimmunoassay using antisera against testosterone-3-bsa and testosterone-11a-bsa. Steroids 21:735, Forest MG, Rivarola MA, Migeon CJ: Percentage binding of testosterone, androstenedione and dehydroisoandrosterone in human plasma. Steroids 12:323, Andersen RN, Givens JR, Wiser WL, Umstot ES: Response of the binding capacity of plasma testosteroneestradiol-binding globulin to norethindrone, 2 mg., and mestranol, 0.1 mg., in polycystic ovarian disease. Am J Obstet.Gynecol 125:166, Briggs MH: Effects of oral progestogens on estrogeninduced changes in plasma proteins. J Reprod Med 15(3):100, Givens JR, Andersen RN, Ragland JB, Umstot ES: Effects of norgestrel and metyrapone on pituitaryadrenal-ovarian function. Obstet Gynecol 48:392, Givens JR, Andersen RN, Wiser WL, Umstot ES, Fish SA: The effectiveness of two oral contraceptives in suppressing plasma androstenedione, testosterone, LH, and FSH, and in stimulating plasma testosteronebinding capacity in hirsute women. Am J Obstet Gynecol 124:333, Vermeulen A, Verdonck L: Influence of TeBG binding of testosterone on its metabolic clearance rate. Ann Endocrinol (Paris) 31:831, Mercier-Bodard C, Marchut M, Perrot M, Picard MT, Baulieu EE, Robel P: Influence of purified plasma proteins on testosterone uptake and metabolism by normal and hyperplastic human prostate in "constantflow organ culture." J Clin Endocrinol Metab 43:374, Mauvais-Jarvis P: Etudes in vivo sur Ie metabolisme des androgenes apres administration de steroides inhibiteurs de l'ovulation. Acta Endocrinol (Kbh) 53:37, Bird CE, Green RN, Clark AF: Effect of the administration of estrogen on the disappearance of 3H-testosterone in the plasma of human subjects. J Clin Endocrinol Metab 29:123, Southren AL, Gordon GG: Rhythms and testosterone metabolism. J Steroid Biochem 6:809, Rosenfield RL: Studies of the relation of plasma androgen levels to androgen action in women. J Steroid Biochem 6:695, Chapdelaine A: Sites of in vivo extraction and interconversion of testosterone and androstenedione in dogs. J Clin Invest 48:2063, van Kammen E, Thijssen JHH, Rademaker B, Schwarz F: The influence of hormonal contraceptives on sex hormone binding globulin (SHBG) capacity. Contraception 11:53, 1975