- Email: [email protected]
Comparative studies of the ethynyl estrogens used in oral contraceptives
Comparative studies of the ethynyl estrogens used in oral contraceptives II.
Antiovulatory
JOSEPH
W.
ARMANDO
GOLDZIEHER, DE
C.
BRANDON
T.
B.
San
Antonio,
potency
M.D.
PERA,
LA
M.S.
CHENAULT,
WOUTERSZ,
M.D. M.D
Texas
The occurrence or inhibition of ovulation was inferred from the plasma progestin level measured in the last week of 430 control cycles and of 4,638 cycles front fertile women receiving various antiovulatory steroids, singly and in combination. The substances tested included: mestranol and ethynylestradiol (EE) of h omogeneous bioavailability, used alone in a range from 50 to IO0 pg per day; these same dose levels combined with various progestins; and finally various proprietary combination and sequential low-dose regimens undergoing clinical trials. Statistical analysis showed ethynylestradiol and mestranol alone to be equipotent over the tested range, although at 50 pg per day superiority of mestranol over EE was suggested. Two preparations of EE at 50 pg per day, one of them a sequential with dimethisterone, showed different potency. At 50 pg per day no estrogen, alone or with a sequential progestin, reached a satisfactory level of effectiveness. However, very small amounts in the range of 20 to 40 pg per day were highly eflective when combined with quantities of various I9-nor progestins which by themselves are well below the antiovulatory level. This indicates that a synergism exists between these two classes of compounds insofar as their antiovulatory eflect is concerned, thus explaining the high contraceptive effectiveness observed with very-low-dose combination regimens.
THE
ANT
IO
V U LATO
RY
&-eCt
Of
Od
Goldzieher, and their associates’-” have established the disproportionately high pituitary-inhibiting potency of the ethynyl estrogens as compared to natural compounds such as estradiol, estriol, conjugated estrogens, and also to nonsteroidal synthetics such as stilbestrol (Table I). Previous studies,4 using urinary pregnanediol excretion as an index of ovulation, demonstrated that a dose-level of about 0.08 mg. per day of an ethynyl estrogen, by itself, was required for acceptable antiovulatory efficacy. The sample size at the 0.05 mg. per day level was too small for a meaningfui evaluation of its efficacy (Table II). Although there are published opinions regarding the required dose, no actual study has been performed, nor has there been any valid clinical comparison of the relative effectiveness of the ethynylestradiol and mestwo ethynyl estrogens,
Con-
traceptives is considered to be their major mechanism of action. This property of both progestational and estrogenic substances has been known for about 50 years; since 1959 the investigations of Kudel, From the Southwest and Education.
Foundation
for
Research
This study was supported by Contract csd/2821, Agency for International Development, and by Wyeth Laboratories, Radnor, Pennsylvania. Received Revised Accepted
for
publication
October October
August
23, 1974.
3, 1974. 8, 1974.
Reprint requests: Joseph W. Coldzieher, M.D., Director, Clinical Sciences and Rebroductiue Biology,.Southwest Foundation fo; Research and Education, P. 0. Box 28147, 8848 W. Commerce St., San Antonio, Texas 78284.
619
620
Gokizieher
et a:.
Table I. Antiovulatory estrogens by urinary
Mestranol
activity of various pregnanediol assay*
0.05
20
0.02
10
0.08
18 4
Estradiol
1.o
Estriol Premarin
2.0 5.0 5.0 1.25 3.75 5.0
Stilbestrol *After Kincl.”
44 20 60
dose
No. of cycles
estradiol days 12t Ethvnvl estradiol 010; x 20 days 59 Mestranol-chlormadinone acetate (0.08 -f 398 2) sequential Mestranol-lynestrenol (0.075 + 2.5) s 170 Mestranol-norethindrone (0.06 + 10) 88 *After Gual and associates.” tFour pregnancies in 68 cycles.
2.3
10 1.7 45
11 18
10 10
39
24
13 86 67
5
10 15 6
18 17 12
Table II. Antiovulatory activity of various estrogens by urinary pregnanediol assay*
Drug and (mg./day)
oral
No. with elevated pregnanedial values
5.9 8.3
oral
Per cent cycles considered ouulatory 2 95% C.L.
Ethynyl 0.02
x 20
3
25
2
3.4
+ 4.6
14
3.5
2 1.8
10
5.9 5 3.5
6
6.8
t 5.3
tranol. Most animal experiments have suggested that the former of these compounds is more potent although the hemicastrated female rat system4 shows mestranol to be more potent, and possibly also qualitatively different. However, results in rodents’, ’ are often not comparable to effects in man and it appears hazardous to extrapolate the animal data uncritically to clinical usage, as is commonly done.? In order to examine this problem further, we undertook a comparative study of the antiovulatory activity of the two ethynyl estrogens, with and without concomitant synthetic progestational steroids. Materials
and
methods
For the comparative dose-response studies of ethynylestradiol and mestranol, bulk steroids were
obtained and tablets containing 0.05, 0.08. CII 0.1 mg. mestranol and 0.05 or 0.08 131&y. OF ethynylestradiol were prepared by Wyeth I,aboratories, using identical formulas for each tablet strength except for adjustment of the amount of diluent due to differences in the amount of active ingredient. The tablet formula and method of manufacture were the same as those used fog a currently marketed contraceptive containing ethynylestradiol. Complete quality control checks were performed, including tablet-to-tablet content uniformity and dissolution, thereby assuring appropriate control of dose content and bioavailability. In a similar manner, tablets containing the same doses of these estrogens, combined with various amounts of norethindrone acetate, cll-norgestrel, or megestrol acetate were prepared. Commercial clinical-trial material of formulations of ethynylestradiol with norgestrel, ethynylestradiol with norethindrone acetate, mestranol with chlormadinone acetate, and a sequential preparation of ethynylestradiol and dimethisterone were made available by Mead Johnson and Co., Parke, Davis and Company, Syntex Laboratories, and Wyeth Laboratories. The clinical samples derive primarily from studies carried out at our facility. In addition, some blood samples from collaborative clinical trials with the ethynylestradiol-norethindrone acetate combinations were provided by other investigators. Since these trials extended over a considerable period of time, no randomized drug assignment was possible. Moreover, in subjects studied at the 0.05 rrlg. per day level of the ethynyl estrogens alone, supplemental contraceptive protection was provided by the use of an IUD. These individuals were a select group willing to use two contraceptive modalities, and may therefore not be homogeneous with the other groups receiving higher doses of mestranol or ethynylestradiol alone. The estrogen-alone regimens (group A to F, Table III) consisted of 21 day cycles of tablets with a 7 day rest period, for a maximum of six cycles. The relatively poor cycle control produced a significant dropout rate. Those who completed the six cycles continued on into combinationtype regimens (series G) , maintaining the same type and dose of estrogen for an additional maximum of six cycles. The commercial preparations (E, H-P) were also in the form of 21 day drug cycles. The women who participated in the ethynyl estrogen dose-response studies wcrc menstrually regular, fertile women of reproductive age who had not used steroidal contraceptives previously. In the
Ethynyl
Table III.
Frequency
of ovulatory
plasma
progestin
levels with
estrogens
various
and
Control
dose
for
(mg./day)
21 day
cycles
I
I-<2
I
contraceptives.
oral contraceptive
No. of cycles with plasma progestin value (ng./ml.) in a given range Drug
in oral
>2
Total
of
No. cycles
II
steroid
621
regimens
Per cent cycles considered ouulatory (plasma P > I ng./ml.), I! 95% C.L.
cycles:
1. Pill 2. IUD
28 18
choosers users
106 251
13 14
147 283
81.0 93.6
?r 6.3 + 2.9
Mestranol
A. B. C.
0.05 0.08 0.10
mg./day mg.jday mg./day
Ethynylestradiol
D. E.
mg./day dimethisterone + 0.25 dimethisterone
Experimental
G. EE
3 4 1
19 6 1
143 176 176
15.4 + 5.9 5.7 + 3.4 1.1 -c 1.5
83
1
27
111
25.2
+
6.7
345
5
36
386
10.6
t
3.1
155
4
3
162
4.3
f
3.1
486
1
19
506
4.n
-t
1.7
266
0
0
266
0.0
+
1.4
53 362 238 525 625 314
1 1 1 4 1 1
5 15 1 6 3 3
59 378 240 535 629 318
10.2 4.2 0.8 1.9 0.6 1.3
+ * + +_ -t 2
7.7 2.0 1.1 1.1 0.6 1.2
99 22
1 0
6 0
105 22
,
(EE):
0.05 With 0.05 With 0.08
F.
121 166 174
combined
(seq.) (seq.) E f P preparations:
Various + norgestrel:
H. EE
0.03/0.3 + norethindrone
I. J. K. L. M. N.
0.02/0.4 O.OZ/l.O O.OZ/l.O* 0.03/0.6 0.03/1.5 0.04/2.0
Mestranol
0. P.
acetate:
+ chlormadinone
acetate:
0.05/0.5 O.lO/l.O
*Recalculation
of
series
J, with
one
subset
including
other study groups previous contraceptive use (with a rest period prior to the initiation of this study) was permitted. Fully informed, written consent was secured in all instances. Plasma samples were obtained during the last 7 days of contraceptive agent intake. The progestin level was measured by a competitive protein-binding method* without chromatographic purification. The procedure yields somewhat higher values than a specific assay for progesterone. Just as reported by the authors of the method,” in a very large experience, we have found that anovulatory cycles or the preovulatory phase of normal cycles yield plasma progestin values of less than 1 ng. per milliliter. We consider a value of 1 to 2 ng. per milliliter “possibly ovulatory” and values of 2 ng. per milliliter or more a positive indication of progesterone production by a corpus luteum and therefore operational evidence for ovulation. Every effort was made to insure reliable medication intake by intensive, monthly monitoring at our
14 ovulations
in
138
cycles
7.3 f 5.0 0.0 + 15.3
omitted.
clinical facility. However, undisclosed errors can and probably do exist both in our studies and in the material obtained from our collaborators. For the purposes of the study, “possibly ovulatory” and definitely ovulatory cycles were lumped together, and the over-all per cent of presumably ovulatory cycles was calculated. Next, for any two groups that were to be compared, a calculation was performed to see what sample size would be required to demonstrate a difference with a 90 per cent certainty of detecting the difference (if it existed) at the P = 0.05 level. If the actual sample size exceeded this figure, a t test showing statistical significance could be accepted with some confidence. If the sample size was less than the minimum calculated requirement, then the significance of t test calculations becomes questionable. On the assumption that the data for any particular regimen represent a random sample of a larger population, a 95 per cent confidence limit was calculated for the per cent of ovulations in that particular regimen
.~., ..-Am.
and sample. Tests for significance and P = 0.01 levels were carried the formula t,
=
arc
sin
[820.8
PI”
-
(&
arc
at the P = 0.05 out by means of
sin
Pz”
t $)I”
where P = the percentage of ovulations (in decimal form), N = sample size, and 820.8 is assumed to be the parametric variance of the arc sin transformation. Results
Two control series of fertile, reproductive age women using different contraceptive modalities are shown for comparison purposes (Table III). The frequency of ovulatory cycles is within the accepted range for this population; the difference reflects differences in age distribution of the two groups, and possibly inherent population differences as well, for these are known to exist between pill choosers and IUD choosers. The three dose levels of mestranol (0.05, 0.08, and 0.1 mg. per day [series A, B, and C]) reveal a progressive increase in effectiveness of ovulation inhibition ( 15.4, 5.7, and 1.1 per cent ovulatory cycles). The differences between the groups are significant at the P = 0.01 level, but a larger sample size (200 in each group) would be desirable for the comparison of the 0.05 and 0.08 mg. per day levels. The difference between the two dose levels of ethynylestradiol (0.05 and 0.08 mg. per day [series D and F]) is also significant at the P = 0.01 level; the sample size is adequate. The difference between the two 0.05 mg. per day ethynylestradiol formulations (series D and E) is also significant at P = 0.01. It may represent a difference in bioavailability in the two formulations, an effect of the dimethisterone, or some unknown factor. The comparison of mestranol and ethynylestradiol on a dose-for-dose basis is the first of its kind. At the 0.05 mg. per day level, the ovulation frequency for ethynylestradiol is 25.2 per cent, that for mestranol only 15.4 per cent. However, the difference does not attain statistical significance at the P = 0.05 level. A sample size of 350 in each group would be required for an evaluation that meets our specifications. There is much less difference between the two steroids at the 0.08 mg. per day level-5.7 per cent for mestranol, 4.3 per cent for ethynylestradiol.
J. Obstt I
Gvnrrol
It is not significant at the P = 0.05 level. ii harnplf size of 5,400 in each group would be required to test this small difference adequately. Thus, these results do not confirm animal studies which attribute a significantly higher potency (twoto fivefold) to ethynylestradiol. In fact, they suggest that further experiments at 0.05 mg. per day 01 lower dose levels might demonstrate a higher antiovulatory potency for mestranol. The various combined preparations yielded a range of ovulation frequencies which require careful evaluation. In the group of ethynylestradiol-norethindrone acetate formulations there is no statistically significant difference between the ovulation frequencies of 10.2 per cent (the 0.02/0.4 formulation) and the 4.2 per cent of the 0.02/1.0 formulation. The small sample size in the 0.02JO.4 series produces a very large uncertainty (27.7) in the observed value of 10.2 per cent. Series J (0.02,’ 1.O) ? which gave an over-all value of 4.2 per cent, consisted of groups of samples received from different sources over a period of time. One subset of 138 samples included 1-l of the 15 values in the ovulatory range. The odds are more than 100: 1 against this having occurred simply by chance, suggesting that unadmitted irregularities in drug intake were a factor. Excluding this subset (“series K”) yielded a value of 0.8 per cent for the ovulatory frequency associated with this formulation, a result not significantly different from that of the O.OS/O.S. 0.03/1.5, or 0.04/2.0 formulations (series L. M, N) or from mestranol alone at 0.1 mg. per day. Series H (EE + norgestrel) was significantly different from all the norethindrone acetate formulations, but in contrast to the latter, all the samples were obtained from a San Antonio clinical trial. Thus, differences in patient reliability might have affected the comparison. The small sample size in the mestranolchlormadinone acetate formulations prevents an adequate evaluation. The practical difficulties in trying to establish meaningful differences at these high antiovulatory efficiencies are enormous: It would require two groups of at least 900 samples each for a reliable comparison of series L with M, 1,500 each for J vs. N, and 2,000 for J vs. L. For this reason also, it would be impossible to detect any trend in antiovulatory efficiency in the various norethindrone acetate formulations, should one exist. In practical terms, all the values are in the range associated with clinically satisfactory contraceptive effectiveness.
Volumr Numbe,
122 5
Comment These studies provide the first clinical analysis of the antiovulatory dose-response and relative potencies of ethynylestradiol and mestranol under conditions of standardized bioavailability. The numbers involved in the ethynylestradiol study of Jackson, ain, and Payne” are too small to permit any SP conclusions. A distinct improvement in ovulation inhibition was seen as the dose of mestranol was increased from 0.05 to 0.08 to 0.1 mg. per day, and the same was observed for ethynylestradiol from 0.05 to 0.08 mg. per day. In a dose-for-dose comparison of the two estrogens, mestranol appeared to be more effective than ethynylestradiol at the 0.05 mg. per day level, but the sample sizes did not provide a 90 per cent certainty of seeing a real difference, and the observed, large difference was not statistically significant at P = 0.05. At 0.08 mg. per day, the effectiveness of the two steroids appeared to be identical. These results are strikingly different from potency estimates derived from animal studies; furthermore, they contradict the indirect estimates made by Dickey’ and others with respect to their relative effectiveness in man. The numerical difficulties in measuring a difference in potency at 0.08 mg. per day appear to be prohibitive, but studies at 0.05 mg. per day or lower (with adequate supplemental contraceptive protection) might be desirable to explore further the question of the relative antiovulatory potency of these estrogens in the lower dose range. The data suggesting that mestranol is more potent than ethynylestradiol are compatible with the observation that the action of mestranol is prolonged due to the necessity of enzymatic conversion to the biologically active form? ethynylestradiol. Studies of the metabolic clearance rateslo of the two compounds are not helpful, as they do not measure the biologically active form of mestranol. However, comparative studies of plasma levels of ethynylestradiol after oral administration of both parent compounds are now feasible,” and should yield the necessary pharmacodynamic insight. No significant difference is demonstrable between the antiovulatory effectiveness of the estrogens alone at 0.08 or 0.1 mg. per day and that of combination agents, which yielded antiovulatory rates of about 1 per cent. Very large samples are required for analyses of such small differences, even under optimal conditions, but when the possibility of patient error as a cause of elevated progestin levels enters
Ethynyl
estrogens
in oral
contraceptives.
II
623
the picture, the numbers required to demonstrate differences in drug effectiveness become astronomical. Calculations we have made for comparisons of pregnancy rates’l apply also in the present context. In any event, it is clear from the data that very small amounts of ethynyl estrogens, in combination with relatively small amounts of progestins, have as much antiovulatory activity as far larger quantities of ethynyl estrogens by themselves. It is well known that “microdoses” of progestins, even if given continuously, have relatively little antiovulatory activity. Therefore a synergism between the two types of steroid must exist at the hypothalamopituitary level. This was observed in castrated women by Wallach, Root, and Garcia.l’ Extensive clinical trials with the low-dose combinations have demonstrated their contraceptive effectiveness; our data provide an insight into the major mechanism of action. Qualitative differences in the effect of estrogens and progestins on the gonadotropic mechanism have been described previously,12-I’ and studies carried out in conjunction with the present experiments will be reported in a separate publicati0n.l” These studies serve to point up some of the problems encountered in apparently simple clinical pharmacologic investigations. The problem of bioavailability of the drugs under test has been alluded to previouslyl” and may well be crucial in studies of orally ingested steroids. Neglect of this factor may cause substantial errors in potency estimates. Another factor which has a major influence on the validity of the results is the degree of patient reliability in taking medication as prescribed. Human unreliability in this regard is well documented, and although clever medication intake monitors have been described,15 it has not been possible so far to convince the sponsors of such research that the perfection and use of a monitoring device would represent a major step forward in clinical pharmacology. Finally, it should be pointed out that the simplistic use of t tests and similar procedures may be quite misleading. It must be demonstrated, before tests of significance are applied, that the sample size is sufficient to insure that a difference will be detected, with a specified degree of assurance, if it exists. Such calculations may also demonstrate that the testability of certain differences may be impractical, because of the time or resources required to accumulate the necessary number of measurements.
We wish also to acknowledge the participation of Dr. Louis E. Moses, who was the clinical supervisor of this project prior to his untimely death, the assistance of Dr. Stephen Preston in the collaborative studies on
REFERENCES
1.
2.
3.
4.
5. 6. 7. 8.
Goldzieher, J. W., Martinez-Manatou, J., Livingston, N. B., Jr., Moses, L. E., and Rice-Wray, E.: West. J. Surg. Obstet. Gynecol. 71: 187, 1963. Martinez-Manautou, J., Gual, C., Goldzieher, J. W., and Rudel, H. W.: Proc. 46th Meet. Endocrinol. Sot. 24: 118, 1964. Rudel, H. W., and Kincl, F. A.: Int. Encyclo. Pharmacol. Ther. Sec. 48, Vol. II, Chap. 34, pp. 385-470, 1972. Edgren, R. A.: In Lednicer, D., editor: Contraception: The Chemical Control of Fertility, New York, 1969, M. Dekker, p. 46. Kincl, F. A.: Int. Encyclo. Pharmacol. Ther. Sec. 48, Vol. II, Chap. 33, pp. 347-384, 1972. Gual, C., Becerra, C., Rice-Wray, E., and Goldzieher, J. W.: AM. J. OBSTET. GYNECOL. 97: 443, 1967. Dickey, R. P.: Seminar, Am. Coll. Obstet. Gynecol. 1972. Johannsson, E. D. B.: Acta Endocrinol. 61: 592, 1969.
norethindrone acetate combinations, the statistica: adticc, of Mr. Tazewell Dozier, and a generous gift of nustrarco! from Syntex Laboratories.
9. 10. 11. 12. 13. 14. 15.
16.
17. 18.
Jackson, J. L., Spain, W. T., and Payne, II.: Fertil. Steril. 19: 649, 1968. Bird, C. E., and Clark, A. F.: AM. J. OBSTET. GYNECOL. 36: 296, 1973. Hines, D. C., and Goldzieher, J. W.: AM. J. OBSTET. GYNECOL. 105: 450, 1969. Wallach, E. E., Root, A. W., and Garcia, C. R.: J. Clin. Endocrinol. 31: 376, 1970. Goldzieher, J. W., Kleber, J. W., Moses, L. E., and Rathmacher, R. P.: Contraception 2: 225, 1970. Stevens, V. C., Goldzieher, J. W., and Vorys, N.: AM. J. OBSTET. GYNECOL. 102: 95, 1968. Goldzieher, J. W., Maqueo, M., Chenault, C. B., and Woutersz, T. B.: AM. J. OBSTET. GYNECOL. 122: 615, 1975. Goldzieher, J. W., de la Pefia, A., Chenault, C. B., and Cervantes, A.: AM. 1. OBSTET. GYNECOL. 122: 625, 1975. Moulding, T.: Am. Rev. Resp. Dis. 85: 755, 1962. de la Peiia, A., and Goldzieher, J. W.: Steroids. 1975. In press.
Antiovulatory
JOSEPH
W.
ARMANDO
GOLDZIEHER, DE
C.
BRANDON
T.
B.
San
Antonio,
potency
M.D.
PERA,
LA
M.S.
CHENAULT,
WOUTERSZ,
M.D. M.D
Texas
The occurrence or inhibition of ovulation was inferred from the plasma progestin level measured in the last week of 430 control cycles and of 4,638 cycles front fertile women receiving various antiovulatory steroids, singly and in combination. The substances tested included: mestranol and ethynylestradiol (EE) of h omogeneous bioavailability, used alone in a range from 50 to IO0 pg per day; these same dose levels combined with various progestins; and finally various proprietary combination and sequential low-dose regimens undergoing clinical trials. Statistical analysis showed ethynylestradiol and mestranol alone to be equipotent over the tested range, although at 50 pg per day superiority of mestranol over EE was suggested. Two preparations of EE at 50 pg per day, one of them a sequential with dimethisterone, showed different potency. At 50 pg per day no estrogen, alone or with a sequential progestin, reached a satisfactory level of effectiveness. However, very small amounts in the range of 20 to 40 pg per day were highly eflective when combined with quantities of various I9-nor progestins which by themselves are well below the antiovulatory level. This indicates that a synergism exists between these two classes of compounds insofar as their antiovulatory eflect is concerned, thus explaining the high contraceptive effectiveness observed with very-low-dose combination regimens.
THE
ANT
IO
V U LATO
RY
&-eCt
Of
Od
Goldzieher, and their associates’-” have established the disproportionately high pituitary-inhibiting potency of the ethynyl estrogens as compared to natural compounds such as estradiol, estriol, conjugated estrogens, and also to nonsteroidal synthetics such as stilbestrol (Table I). Previous studies,4 using urinary pregnanediol excretion as an index of ovulation, demonstrated that a dose-level of about 0.08 mg. per day of an ethynyl estrogen, by itself, was required for acceptable antiovulatory efficacy. The sample size at the 0.05 mg. per day level was too small for a meaningfui evaluation of its efficacy (Table II). Although there are published opinions regarding the required dose, no actual study has been performed, nor has there been any valid clinical comparison of the relative effectiveness of the ethynylestradiol and mestwo ethynyl estrogens,
Con-
traceptives is considered to be their major mechanism of action. This property of both progestational and estrogenic substances has been known for about 50 years; since 1959 the investigations of Kudel, From the Southwest and Education.
Foundation
for
Research
This study was supported by Contract csd/2821, Agency for International Development, and by Wyeth Laboratories, Radnor, Pennsylvania. Received Revised Accepted
for
publication
October October
August
23, 1974.
3, 1974. 8, 1974.
Reprint requests: Joseph W. Coldzieher, M.D., Director, Clinical Sciences and Rebroductiue Biology,.Southwest Foundation fo; Research and Education, P. 0. Box 28147, 8848 W. Commerce St., San Antonio, Texas 78284.
619
620
Gokizieher
et a:.
Table I. Antiovulatory estrogens by urinary
Mestranol
activity of various pregnanediol assay*
0.05
20
0.02
10
0.08
18 4
Estradiol
1.o
Estriol Premarin
2.0 5.0 5.0 1.25 3.75 5.0
Stilbestrol *After Kincl.”
44 20 60
dose
No. of cycles
estradiol days 12t Ethvnvl estradiol 010; x 20 days 59 Mestranol-chlormadinone acetate (0.08 -f 398 2) sequential Mestranol-lynestrenol (0.075 + 2.5) s 170 Mestranol-norethindrone (0.06 + 10) 88 *After Gual and associates.” tFour pregnancies in 68 cycles.
2.3
10 1.7 45
11 18
10 10
39
24
13 86 67
5
10 15 6
18 17 12
Table II. Antiovulatory activity of various estrogens by urinary pregnanediol assay*
Drug and (mg./day)
oral
No. with elevated pregnanedial values
5.9 8.3
oral
Per cent cycles considered ouulatory 2 95% C.L.
Ethynyl 0.02
x 20
3
25
2
3.4
+ 4.6
14
3.5
2 1.8
10
5.9 5 3.5
6
6.8
t 5.3
tranol. Most animal experiments have suggested that the former of these compounds is more potent although the hemicastrated female rat system4 shows mestranol to be more potent, and possibly also qualitatively different. However, results in rodents’, ’ are often not comparable to effects in man and it appears hazardous to extrapolate the animal data uncritically to clinical usage, as is commonly done.? In order to examine this problem further, we undertook a comparative study of the antiovulatory activity of the two ethynyl estrogens, with and without concomitant synthetic progestational steroids. Materials
and
methods
For the comparative dose-response studies of ethynylestradiol and mestranol, bulk steroids were
obtained and tablets containing 0.05, 0.08. CII 0.1 mg. mestranol and 0.05 or 0.08 131&y. OF ethynylestradiol were prepared by Wyeth I,aboratories, using identical formulas for each tablet strength except for adjustment of the amount of diluent due to differences in the amount of active ingredient. The tablet formula and method of manufacture were the same as those used fog a currently marketed contraceptive containing ethynylestradiol. Complete quality control checks were performed, including tablet-to-tablet content uniformity and dissolution, thereby assuring appropriate control of dose content and bioavailability. In a similar manner, tablets containing the same doses of these estrogens, combined with various amounts of norethindrone acetate, cll-norgestrel, or megestrol acetate were prepared. Commercial clinical-trial material of formulations of ethynylestradiol with norgestrel, ethynylestradiol with norethindrone acetate, mestranol with chlormadinone acetate, and a sequential preparation of ethynylestradiol and dimethisterone were made available by Mead Johnson and Co., Parke, Davis and Company, Syntex Laboratories, and Wyeth Laboratories. The clinical samples derive primarily from studies carried out at our facility. In addition, some blood samples from collaborative clinical trials with the ethynylestradiol-norethindrone acetate combinations were provided by other investigators. Since these trials extended over a considerable period of time, no randomized drug assignment was possible. Moreover, in subjects studied at the 0.05 rrlg. per day level of the ethynyl estrogens alone, supplemental contraceptive protection was provided by the use of an IUD. These individuals were a select group willing to use two contraceptive modalities, and may therefore not be homogeneous with the other groups receiving higher doses of mestranol or ethynylestradiol alone. The estrogen-alone regimens (group A to F, Table III) consisted of 21 day cycles of tablets with a 7 day rest period, for a maximum of six cycles. The relatively poor cycle control produced a significant dropout rate. Those who completed the six cycles continued on into combinationtype regimens (series G) , maintaining the same type and dose of estrogen for an additional maximum of six cycles. The commercial preparations (E, H-P) were also in the form of 21 day drug cycles. The women who participated in the ethynyl estrogen dose-response studies wcrc menstrually regular, fertile women of reproductive age who had not used steroidal contraceptives previously. In the
Ethynyl
Table III.
Frequency
of ovulatory
plasma
progestin
levels with
estrogens
various
and
Control
dose
for
(mg./day)
21 day
cycles
I
I-<2
I
contraceptives.
oral contraceptive
No. of cycles with plasma progestin value (ng./ml.) in a given range Drug
in oral
>2
Total
of
No. cycles
II
steroid
621
regimens
Per cent cycles considered ouulatory (plasma P > I ng./ml.), I! 95% C.L.
cycles:
1. Pill 2. IUD
28 18
choosers users
106 251
13 14
147 283
81.0 93.6
?r 6.3 + 2.9
Mestranol
A. B. C.
0.05 0.08 0.10
mg./day mg.jday mg./day
Ethynylestradiol
D. E.
mg./day dimethisterone + 0.25 dimethisterone
Experimental
G. EE
3 4 1
19 6 1
143 176 176
15.4 + 5.9 5.7 + 3.4 1.1 -c 1.5
83
1
27
111
25.2
+
6.7
345
5
36
386
10.6
t
3.1
155
4
3
162
4.3
f
3.1
486
1
19
506
4.n
-t
1.7
266
0
0
266
0.0
+
1.4
53 362 238 525 625 314
1 1 1 4 1 1
5 15 1 6 3 3
59 378 240 535 629 318
10.2 4.2 0.8 1.9 0.6 1.3
+ * + +_ -t 2
7.7 2.0 1.1 1.1 0.6 1.2
99 22
1 0
6 0
105 22
,
(EE):
0.05 With 0.05 With 0.08
F.
121 166 174
combined
(seq.) (seq.) E f P preparations:
Various + norgestrel:
H. EE
0.03/0.3 + norethindrone
I. J. K. L. M. N.
0.02/0.4 O.OZ/l.O O.OZ/l.O* 0.03/0.6 0.03/1.5 0.04/2.0
Mestranol
0. P.
acetate:
+ chlormadinone
acetate:
0.05/0.5 O.lO/l.O
*Recalculation
of
series
J, with
one
subset
including
other study groups previous contraceptive use (with a rest period prior to the initiation of this study) was permitted. Fully informed, written consent was secured in all instances. Plasma samples were obtained during the last 7 days of contraceptive agent intake. The progestin level was measured by a competitive protein-binding method* without chromatographic purification. The procedure yields somewhat higher values than a specific assay for progesterone. Just as reported by the authors of the method,” in a very large experience, we have found that anovulatory cycles or the preovulatory phase of normal cycles yield plasma progestin values of less than 1 ng. per milliliter. We consider a value of 1 to 2 ng. per milliliter “possibly ovulatory” and values of 2 ng. per milliliter or more a positive indication of progesterone production by a corpus luteum and therefore operational evidence for ovulation. Every effort was made to insure reliable medication intake by intensive, monthly monitoring at our
14 ovulations
in
138
cycles
7.3 f 5.0 0.0 + 15.3
omitted.
clinical facility. However, undisclosed errors can and probably do exist both in our studies and in the material obtained from our collaborators. For the purposes of the study, “possibly ovulatory” and definitely ovulatory cycles were lumped together, and the over-all per cent of presumably ovulatory cycles was calculated. Next, for any two groups that were to be compared, a calculation was performed to see what sample size would be required to demonstrate a difference with a 90 per cent certainty of detecting the difference (if it existed) at the P = 0.05 level. If the actual sample size exceeded this figure, a t test showing statistical significance could be accepted with some confidence. If the sample size was less than the minimum calculated requirement, then the significance of t test calculations becomes questionable. On the assumption that the data for any particular regimen represent a random sample of a larger population, a 95 per cent confidence limit was calculated for the per cent of ovulations in that particular regimen
.~., ..-Am.
and sample. Tests for significance and P = 0.01 levels were carried the formula t,
=
arc
sin
[820.8
PI”
-
(&
arc
at the P = 0.05 out by means of
sin
Pz”
t $)I”
where P = the percentage of ovulations (in decimal form), N = sample size, and 820.8 is assumed to be the parametric variance of the arc sin transformation. Results
Two control series of fertile, reproductive age women using different contraceptive modalities are shown for comparison purposes (Table III). The frequency of ovulatory cycles is within the accepted range for this population; the difference reflects differences in age distribution of the two groups, and possibly inherent population differences as well, for these are known to exist between pill choosers and IUD choosers. The three dose levels of mestranol (0.05, 0.08, and 0.1 mg. per day [series A, B, and C]) reveal a progressive increase in effectiveness of ovulation inhibition ( 15.4, 5.7, and 1.1 per cent ovulatory cycles). The differences between the groups are significant at the P = 0.01 level, but a larger sample size (200 in each group) would be desirable for the comparison of the 0.05 and 0.08 mg. per day levels. The difference between the two dose levels of ethynylestradiol (0.05 and 0.08 mg. per day [series D and F]) is also significant at the P = 0.01 level; the sample size is adequate. The difference between the two 0.05 mg. per day ethynylestradiol formulations (series D and E) is also significant at P = 0.01. It may represent a difference in bioavailability in the two formulations, an effect of the dimethisterone, or some unknown factor. The comparison of mestranol and ethynylestradiol on a dose-for-dose basis is the first of its kind. At the 0.05 mg. per day level, the ovulation frequency for ethynylestradiol is 25.2 per cent, that for mestranol only 15.4 per cent. However, the difference does not attain statistical significance at the P = 0.05 level. A sample size of 350 in each group would be required for an evaluation that meets our specifications. There is much less difference between the two steroids at the 0.08 mg. per day level-5.7 per cent for mestranol, 4.3 per cent for ethynylestradiol.
J. Obstt I
Gvnrrol
It is not significant at the P = 0.05 level. ii harnplf size of 5,400 in each group would be required to test this small difference adequately. Thus, these results do not confirm animal studies which attribute a significantly higher potency (twoto fivefold) to ethynylestradiol. In fact, they suggest that further experiments at 0.05 mg. per day 01 lower dose levels might demonstrate a higher antiovulatory potency for mestranol. The various combined preparations yielded a range of ovulation frequencies which require careful evaluation. In the group of ethynylestradiol-norethindrone acetate formulations there is no statistically significant difference between the ovulation frequencies of 10.2 per cent (the 0.02/0.4 formulation) and the 4.2 per cent of the 0.02/1.0 formulation. The small sample size in the 0.02JO.4 series produces a very large uncertainty (27.7) in the observed value of 10.2 per cent. Series J (0.02,’ 1.O) ? which gave an over-all value of 4.2 per cent, consisted of groups of samples received from different sources over a period of time. One subset of 138 samples included 1-l of the 15 values in the ovulatory range. The odds are more than 100: 1 against this having occurred simply by chance, suggesting that unadmitted irregularities in drug intake were a factor. Excluding this subset (“series K”) yielded a value of 0.8 per cent for the ovulatory frequency associated with this formulation, a result not significantly different from that of the O.OS/O.S. 0.03/1.5, or 0.04/2.0 formulations (series L. M, N) or from mestranol alone at 0.1 mg. per day. Series H (EE + norgestrel) was significantly different from all the norethindrone acetate formulations, but in contrast to the latter, all the samples were obtained from a San Antonio clinical trial. Thus, differences in patient reliability might have affected the comparison. The small sample size in the mestranolchlormadinone acetate formulations prevents an adequate evaluation. The practical difficulties in trying to establish meaningful differences at these high antiovulatory efficiencies are enormous: It would require two groups of at least 900 samples each for a reliable comparison of series L with M, 1,500 each for J vs. N, and 2,000 for J vs. L. For this reason also, it would be impossible to detect any trend in antiovulatory efficiency in the various norethindrone acetate formulations, should one exist. In practical terms, all the values are in the range associated with clinically satisfactory contraceptive effectiveness.
Volumr Numbe,
122 5
Comment These studies provide the first clinical analysis of the antiovulatory dose-response and relative potencies of ethynylestradiol and mestranol under conditions of standardized bioavailability. The numbers involved in the ethynylestradiol study of Jackson, ain, and Payne” are too small to permit any SP conclusions. A distinct improvement in ovulation inhibition was seen as the dose of mestranol was increased from 0.05 to 0.08 to 0.1 mg. per day, and the same was observed for ethynylestradiol from 0.05 to 0.08 mg. per day. In a dose-for-dose comparison of the two estrogens, mestranol appeared to be more effective than ethynylestradiol at the 0.05 mg. per day level, but the sample sizes did not provide a 90 per cent certainty of seeing a real difference, and the observed, large difference was not statistically significant at P = 0.05. At 0.08 mg. per day, the effectiveness of the two steroids appeared to be identical. These results are strikingly different from potency estimates derived from animal studies; furthermore, they contradict the indirect estimates made by Dickey’ and others with respect to their relative effectiveness in man. The numerical difficulties in measuring a difference in potency at 0.08 mg. per day appear to be prohibitive, but studies at 0.05 mg. per day or lower (with adequate supplemental contraceptive protection) might be desirable to explore further the question of the relative antiovulatory potency of these estrogens in the lower dose range. The data suggesting that mestranol is more potent than ethynylestradiol are compatible with the observation that the action of mestranol is prolonged due to the necessity of enzymatic conversion to the biologically active form? ethynylestradiol. Studies of the metabolic clearance rateslo of the two compounds are not helpful, as they do not measure the biologically active form of mestranol. However, comparative studies of plasma levels of ethynylestradiol after oral administration of both parent compounds are now feasible,” and should yield the necessary pharmacodynamic insight. No significant difference is demonstrable between the antiovulatory effectiveness of the estrogens alone at 0.08 or 0.1 mg. per day and that of combination agents, which yielded antiovulatory rates of about 1 per cent. Very large samples are required for analyses of such small differences, even under optimal conditions, but when the possibility of patient error as a cause of elevated progestin levels enters
Ethynyl
estrogens
in oral
contraceptives.
II
623
the picture, the numbers required to demonstrate differences in drug effectiveness become astronomical. Calculations we have made for comparisons of pregnancy rates’l apply also in the present context. In any event, it is clear from the data that very small amounts of ethynyl estrogens, in combination with relatively small amounts of progestins, have as much antiovulatory activity as far larger quantities of ethynyl estrogens by themselves. It is well known that “microdoses” of progestins, even if given continuously, have relatively little antiovulatory activity. Therefore a synergism between the two types of steroid must exist at the hypothalamopituitary level. This was observed in castrated women by Wallach, Root, and Garcia.l’ Extensive clinical trials with the low-dose combinations have demonstrated their contraceptive effectiveness; our data provide an insight into the major mechanism of action. Qualitative differences in the effect of estrogens and progestins on the gonadotropic mechanism have been described previously,12-I’ and studies carried out in conjunction with the present experiments will be reported in a separate publicati0n.l” These studies serve to point up some of the problems encountered in apparently simple clinical pharmacologic investigations. The problem of bioavailability of the drugs under test has been alluded to previouslyl” and may well be crucial in studies of orally ingested steroids. Neglect of this factor may cause substantial errors in potency estimates. Another factor which has a major influence on the validity of the results is the degree of patient reliability in taking medication as prescribed. Human unreliability in this regard is well documented, and although clever medication intake monitors have been described,15 it has not been possible so far to convince the sponsors of such research that the perfection and use of a monitoring device would represent a major step forward in clinical pharmacology. Finally, it should be pointed out that the simplistic use of t tests and similar procedures may be quite misleading. It must be demonstrated, before tests of significance are applied, that the sample size is sufficient to insure that a difference will be detected, with a specified degree of assurance, if it exists. Such calculations may also demonstrate that the testability of certain differences may be impractical, because of the time or resources required to accumulate the necessary number of measurements.
We wish also to acknowledge the participation of Dr. Louis E. Moses, who was the clinical supervisor of this project prior to his untimely death, the assistance of Dr. Stephen Preston in the collaborative studies on
REFERENCES
1.
2.
3.
4.
5. 6. 7. 8.
Goldzieher, J. W., Martinez-Manatou, J., Livingston, N. B., Jr., Moses, L. E., and Rice-Wray, E.: West. J. Surg. Obstet. Gynecol. 71: 187, 1963. Martinez-Manautou, J., Gual, C., Goldzieher, J. W., and Rudel, H. W.: Proc. 46th Meet. Endocrinol. Sot. 24: 118, 1964. Rudel, H. W., and Kincl, F. A.: Int. Encyclo. Pharmacol. Ther. Sec. 48, Vol. II, Chap. 34, pp. 385-470, 1972. Edgren, R. A.: In Lednicer, D., editor: Contraception: The Chemical Control of Fertility, New York, 1969, M. Dekker, p. 46. Kincl, F. A.: Int. Encyclo. Pharmacol. Ther. Sec. 48, Vol. II, Chap. 33, pp. 347-384, 1972. Gual, C., Becerra, C., Rice-Wray, E., and Goldzieher, J. W.: AM. J. OBSTET. GYNECOL. 97: 443, 1967. Dickey, R. P.: Seminar, Am. Coll. Obstet. Gynecol. 1972. Johannsson, E. D. B.: Acta Endocrinol. 61: 592, 1969.
norethindrone acetate combinations, the statistica: adticc, of Mr. Tazewell Dozier, and a generous gift of nustrarco! from Syntex Laboratories.
9. 10. 11. 12. 13. 14. 15.
16.
17. 18.
Jackson, J. L., Spain, W. T., and Payne, II.: Fertil. Steril. 19: 649, 1968. Bird, C. E., and Clark, A. F.: AM. J. OBSTET. GYNECOL. 36: 296, 1973. Hines, D. C., and Goldzieher, J. W.: AM. J. OBSTET. GYNECOL. 105: 450, 1969. Wallach, E. E., Root, A. W., and Garcia, C. R.: J. Clin. Endocrinol. 31: 376, 1970. Goldzieher, J. W., Kleber, J. W., Moses, L. E., and Rathmacher, R. P.: Contraception 2: 225, 1970. Stevens, V. C., Goldzieher, J. W., and Vorys, N.: AM. J. OBSTET. GYNECOL. 102: 95, 1968. Goldzieher, J. W., Maqueo, M., Chenault, C. B., and Woutersz, T. B.: AM. J. OBSTET. GYNECOL. 122: 615, 1975. Goldzieher, J. W., de la Pefia, A., Chenault, C. B., and Cervantes, A.: AM. 1. OBSTET. GYNECOL. 122: 625, 1975. Moulding, T.: Am. Rev. Resp. Dis. 85: 755, 1962. de la Peiia, A., and Goldzieher, J. W.: Steroids. 1975. In press.