There were an estimated 25,400 new cancer cases and 14,300 deaths from epithelial ovarian cancer in 2003.1 Oral contraceptives (OCs) have been consistently linked to reduced risk of ovarian cancer.2–6 Cohort analysis of trends in mortality due to ovarian cancer demonstrate that women who were born after 1920, ie, from generations who had used OCs, consistently show decreased rates of ovarian cancer.7 The contraceptive effect of OCs is hypothesized to derive from a suppressed mid-cycle gonadotropin surge and inhibited ovulation.8 According to both the “incessant ovulation” and the “gonadotropin” hypotheses, OC use is anticipated to decrease the risk of epithelial ovarian cancer.8 During the last 3 decades, OC formulations have changed with the introduction of new, lower-dose progestins that possess varying androgenic properties. Early progestins had a 21-carbon skeleton that made them estrogenic in nature.9,10 Subsequently, the 19-carbon nor-testosterones were developed, which possessed both androgenic and progestogenic qualities. More recently, compounds that are less androgenic have gained in popularity.10–12 A number of currently marketed OCs contain androgenic progestins, including Alesse, Triphasil, Levlen, Estrostep, Ortho-Cept, and Lo/Ovral. Over 10 million U.S. women and 100 million women worldwide are currently taking oral contraceptives, and approximately 20–25% of OCs contain androgenic progestins.13
We3 and others14 have explored whether estrogen and progestin dosages might alter the efficacy of OCs in preventing ovarian cancer by comparing risk reduction of OCs that contain various high- or low-dose combinations of estrogens and progestins. However, OCs with high-dose progestins are not necessarily androgenic in nature, and the estrogen component is not known to influence androgenicity. We have found no studies to date that have examined ovarian cancer risk reduction solely in relation to the androgenicity of the synthetic progestins in OCs. A search of MEDLINE from 1966 to November 2004 with the search terms “androgenicity of progestins” and “androgenic progestins” and by combining the search term “ovarian neoplasms” with “progestins” resulted in 111 articles, none of which evaluated the androgenicity of progestins in relation to ovarian cancer risk.
Adverse effects of androgen excess, including acne, weight gain, and unwanted facial hair,10,15,16 are experienced by many women taking androgenic OCs, and according to the literature, these androgenic adverse effects are more important than any other effects in explaining why women discontinue OC usage.17–19 Polycystic ovarian syndrome, a condition characterized by increased circulating serum levels of adrenal androgens, may increase the risk for developing epithelial ovarian cancer.20 Additionally, more than 90% of ovarian cancer tumors express androgen receptors,21–24 and androgens play a role in follicular growth, maturation, and atresia.25–27 Excess androgen exposure may be particularly important as a cancer risk factor for women with endometriosis,26,28,29 a condition that is frequently managed with OCs. In this analysis, we explore whether the protection associated with OCs might be altered by the androgenicity of the progestin component among all women and, specifically, among women with endometriosis.
SUBJECTS AND METHODS
Women in this study were selected from a case-control study of contraceptive and reproductive risk factors for epithelial ovarian cancer (the Steroid Hormones and Reproductions [SHARE] Study). Cases, aged 20–69 years, with epithelial ovarian cancer diagnosed within the 6 months before interview were ascertained from 39 hospitals in the Delaware Valley surrounding Philadelphia between May 1994 and July 1998. Older subjects would have been unlikely to have been exposed to oral contraceptives. A total of 2,418 cases of histologically confirmed borderline or invasive epithelial ovarian cancer were initially identified. After excluding women who were too young or too old (n = 640), resided outside the counties in which referral hospitals were located (n = 342), had a prior diagnosis of ovarian cancer (n = 158), or did not speak English or were mentally incompetent (n = 25), there were 1,253 potentially eligible women. After further excluding those who were diagnosed more than 6 months before interview (n = 296), were critically ill or dead (n = 69), or were untraceable (n = 15), 873 women remained who had incident cancer and were thus eligible for the study. Fourteen physicians did not consent to their patients’ participation, and 92 women refused to participate. Thus, there were 767 completed case interviews (61% of potentially eligible cases and 88% of potentially eligible incident cases).
Controls, aged 65 or younger, were ascertained by random-digit dialing and frequency matched to cases by race, 5-year age groups, and 3-digit telephone exchange. Of the 14,551 telephone numbers screened for this purpose, 6,597 belonged to businesses or were not in service, and 5,640 had no female of eligible age in the household, leaving 2,314 households with potentially eligible participants. Of these, 1,928 households (83%) had a potentially eligible woman who was willing to be further screened. Upon screening, a further 291 either had no resident women eligible on the basis of age (n = 5), resided outside of the target counties (n = 11), had a prior diagnosis of ovarian cancer (n = 9), had a prior bilateral oophorectomy (n = 187), did not speak English or were mentally incompetent (n = 22), had critical illness or death (n = 6), or were untraceable (n = 51). Of the 1,637 screened and potentially eligible controls, 422 declined to be interviewed, and 1,215 (74%) completed interviews. Controls, aged 65–69 years, were ascertained through Health Care Financing Administration (HCFA) lists because we were concerned about reduced random-digit dialing response rates in this group. Four hundred twenty-three women, frequency-matched to cases by county of residence, were initially identified. One hundred sixty were ineligible for the reasons given above. Of the 263 potentially eligible women from HCFA lists, 111 refused to participate and 152 (58%) were interviewed. Therefore, of the total 1,900 screened and potentially eligible controls (1,637 from random-digit dialing and 263 from HCFA lists), 1,367 (72%) interviews were completed.
Cases included 616 women with invasive epithelial ovarian tumors and 151 women with borderline tumors. The diagnosis of ovarian cancer was confirmed by pathology in all cases. Central pathologic review was conducted on a random sample of 120 cases. The reference pathologist agreed with the original pathologic review for invasiveness in 95% of cases and for cell type in 82% of cases. The original pathologic diagnosis was then used for all cases.
Included in our study were women whom we could classify as having either exclusively used androgenic OCs, exclusively used nonandrogenic OCs, used both types of OCs, or reported never having used OCs. Women who could not be classified into 1 of these 4 categories because of missing information on OC formulations (n = 199 cases and 341 controls) were omitted from our analyses. Thus, the final analysis included 568 cases and 1,026 controls (78% of original SHARE cases and controls). Controls included in our analysis were more likely to be at the extremes of the study age ranges (< 40, ≥ 60) (P = .001) and were less likely to have had endometriosis (P = .001) than cases (data not shown). Number of live births, race, education, family history of ovarian cancer, tubal ligation, and hysterectomy did not differ statistically between included and excluded women.
A standardized 1.5-hour, in-person interview of cases and controls provided detailed demographic data as well as information on subjects’ gynecologic and obstetric history, including menstrual history, pregnancy history, tubal ligation, lactation, hysterectomy, family history of breast and ovarian cancers, oral contraceptive use, and hormone replacement therapy use. Women were asked about endometriosis and were asked if this was diagnosed by a physician. A “life” calendar marked with important events that each participant recalled during her lifetime was used to enhance memory of distant events. The reference date was calculated as 6 months before diagnosis (cases) or interview (controls) to ensure that exposures occurred before ovarian cancer diagnoses in cases and within a similar time frame for cases and controls.
All contraceptive use was recorded, including the type of contraception, frequency of use, and duration of use. Additional details obtained for hormonal contraceptives included the brand, reason for use, and reason for stopping use. Oral contraceptive use was categorized as use for contraception, for noncontraceptive uses such as to control abnormal bleeding or menstrual pain, or for both contraception and other uses. Picture books with photographs of oral contraceptives available in the United States (courtesy of Dr. Ruth Peters, University of Southern California, Los Angeles, CA) were used to help women specify the formulations used. For each combined OC preparation, we obtained information on active ingredients and doses by using a variety of existing databases and reference books. For discontinued medications, we made inquiries to pharmaceutical manufacturers. We tested the accuracy of the recall of OC formulation in a subset of 10% of the women participating in the SHARE study by conducting a second interview more than 6 months after initial interview. Agreement on OC use and formulation exceeded 90% between the 2 interviews.
The androgenicity of the progestin component of each OC formulation was determined by compiling data from various studies that assessed the androgenic potential of each progestin, while factoring in dose of progestin per OC formulation (Browning MC, Anderson J. Effect of oral contraceptives on plasma testosterone concentration [letter]. Br Med J 1977;1:107).1,9–11,16,30–52 Evaluations of androgenicity involve measuring the progestin's affinity for and binding to the androgen receptor, its effect on sex hormone binding globulin levels, its degree of binding to sex hormone binding globulin, and its effect on free testosterone.9,10,38,47,53 The most androgenic progestins used by cases in our study, levonorgestrel and dl-norgestrel, were originally derived from 19-nortestosterone. They have a high affinity for sex hormone binding globulin and decrease free sex hormone binding globulin levels by binding it and displacing testosterone, consequently increasing free testosterone levels.
We classified progestin androgenicity by pharmacokinetic actions of the progestin and dose (Table 1). Each progestin has a different potency, milligram per milligram. A progestin that is considered to be high dose in terms of progestogenicity can be low in terms of androgenicity. A higher-potency progestin may be used in a much smaller dose and thus be equivalent to a larger dose of a less-potent progestin. For example, desogestrel is a very potent and androgenic progestin, but its usual oral contraceptive dose is 0.15 mg. Its progestin potency compared with 1.00 mg of norethindrone would be 0.15 × 9.0 = 1.35 times. For androgenicity, it would be 0.15 × 3.4 = 0.51 or half as androgenic as a pill containing 1 mg of norethindrone. As a cutoff value for progestin androgenicity, we classified as “androgenic” any drug with the equivalent of 1 mg of norethindrone or higher and as “nonandrogenic” anything below the 1-mg equivalent. For example, high doses of norethindrone and norethindrone acetate in OCs gave them the classification of androgenic, such as the 10-mg dose of certain types of Norinyl or Ortho-Novum. Lower doses, such as 0.4 mg norethindrone in Ovcon 35 (28 or 21 day) formulations, with a potency of 0.40, were categorized as nonandrogenic.
Among the 568 cases and 1,026 controls, a total of 125 different formulations of OCs were used. According to dose and potency, we were able to classify all of the progestins used in these 125 formulations in terms of their androgenicity. Table 2 gives an example of the comparative androgenicity of a sample of the OCs that study participants used.
Odds ratios (ORs), with corresponding 95% confidence intervals (CIs), were calculated as the primary measure of effect size. Because the SHARE Study used frequency rather than individual matching and matched on the basis of broad criteria, such as age within 5- to 10-year intervals, we used unconditional logistic regression models to adjust for any additional effects of potential confounding variables that had been determined a priori to affect ovarian cancer risk.2,54–56 Included in the models were age and parity as continuous variables and tubal ligation and family history of ovarian canceras dichotomous (yes/no) variables. Tests for trend (P value) were performed by coding OC duration as a grouped linear variable. Odds ratios for OC exposures were calculated from the estimated β coefficients and their standard errors. Maximum-likelihood ratios were obtained using the STATA BLOGIT function to compare the calculated odds ratios and test for significant differences in their values. All tests of statistical significance were 2-tailed and considered significant at P < .05. All analyses were performed using STATA 8.0 (STATA Corporation, College Station, TX).
Table 3 provides demographic information on ovarian cancer cases and controls for which complete OC data were obtained. As expected, cases were less likely to use oral contraceptives (40.0% versus 58.3%), to have had children (67.8% versus 86.2%), and to have had a tubal ligation (15.0% versus 32.4%) or hysterectomy (9.7% versus 13.0%). They were more likely to report a family history of ovarian cancer (3.5% versus 1.6%) and a personal history of endometriosis (7.9% versus 5.2%). Androgenic OCs were disproportionately used by younger women, whereas older women tended to have used nonandrogenic OCs, reflecting the chronology of the marketing of OC brands as new progestins were developed. Endometriosis was more commonly diagnosed among women who had used OCs (9.9% versus 2.1% of controls). However, the proportion of women diagnosed with endometriosis was not markedly different among androgenic versus nonandrogenic OC users (5.5% versus 7.5% of controls).
For androgenic-only OC users and users of both types of OCs, increasing duration of use was associated with a reduction in ovarian cancer risk (Table 4). In the nonandrogenic group, longer duration did not appear to confer a significant dose-response relationship (OR 0.56 for < 5 years; OR 0.73 for ≥ 5 years).
Independent of the OC androgenicity, early OC use (before age 20) was associated with a greater reduction in risk than later age at first use (Table 4). The point estimate associated with androgenic-only OC use was nonsignificantly smaller than for the nonandrogenic-only use (OR 0.42 versus 0.54).57 Oral contraceptive use within the last 10 years appeared to be more strongly and inversely associated with ovarian cancer than use more than 10 years in the past, regardless of the androgenic content of the formulation. However, even past use that was discontinued more than 10 years previously was associated with a significant reduction in risk, and this observed relationship was independent of the androgenic potency of the particular OC formulation.
Compared to never-users, use of androgenic-only OC formulations was inversely associated with ovarian cancer (adjusted OR 0.52, 95% CI 0.35–0.76) (Table 4). A similar inverse association was observed with use of nonandrogenic-only OC formulations (adjusted OR 0.59, 95% CI 0.45–0.78) and with exposure to both androgenic and nonandrogenic OCs (adjusted OR 0.29, 95% CI 0.17–0.48).
Among women with endometriosis (45 cases, 53 controls), androgenic formulations appeared to confer less protection than nonandrogenic formulations (OR 0.46 versus 0.23), although the sample size of these OC users was small (Table 5). Four cases and 10 controls with histories of endometriosis used both androgenic and nonandrogenic OCs and achieved significant risk reduction (OR 0.11, 95% CI 0.02–0.69, adjusted for age, number of live births, family history of ovarian cancer, and tubal ligation). Further adjustment for OC duration of all models in Table 5 did not alter any findings (data not shown).
Finally, the androgenicity of the OC formulations was not associated with invasiveness or with histologic subtype, and all our findings were similar when we limited cases to women with invasive disease (data not shown).
Independent of the androgenicity of its progestin component, OC use was associated with approximately a 40–50% overall decrease in ovarian cancer risk. Increasing duration of use, early age at first use, and recentness of OC use all provided increased protection against ovarian cancer, regardless of the androgenic potential of the progestin in the OC formulation used. In fact, longer duration of use of androgenic-only OCs appeared to confer increased protection against ovarian cancer when compared to nonandrogenic OC use. Among the limited number of women with endometriosis taking either androgenic-only or nonandrogenic-only OCs (27 cases, 34 controls), androgenic formulations appear to confer less protection than nonandrogenic formulations (OR 0.46 versus 0.23).57
Although our results are reassuring regarding the lack of impact from the androgenicity of progestins in OCs overall, they are somewhat surprising in light of several findings linking androgens to increased ovarian cancer risk. In particular, we recently reported that, for women with endometriosis, danazol, a synthetic androgen that binds to androgenic receptors and sex hormone binding globulin resulting in a 3-fold increase in free testosterone levels, was an independent risk factor for ovarian cancer and was associated with a risk 3.2 times greater than among women who had never used the drug.28 In the current study, among women with endometriosis, use of androgenic-only OCs was associated with somewhat diminished, but still apparent, protection. The small sample size limits the interpretation of this finding.
The most androgenic progestins are also the most progestogenic.42 Possibly, the tumorigenicity induced by an androgenic progestin could be mollified by the apoptotic effect from its progestogenic nature. The change in the progestin component is important because a growing body of evidence suggests that it is the progestin portion of OCs that may provide some of its protective benefit against ovarian cancer. Progestin-only oral contraceptives, which do not totally suppress ovulation, are as protective against ovarian cancer as estrogen-progestin formulas.58 In a study of cynomolgus macaques (Macaca fascicularis), animals randomized to receive estrogen plus progestin or progestin-only pills had a 4- to 6-fold increase in the proportion of apoptotic cells,5 with the maximum 6-fold effect seen in the progestin-only group. These data suggest that progestins induce apoptosis and also support the observation that the protection against ovarian cancer afforded by OCs extends beyond that of ovulation suppression.26 As further evidence, a recent case-control study showed that high-dose progestin OC formulations may be more protective than low-dose formulations,14 although, in a separate study, we failed to find such a difference.3
It is also possible that, although progestins appear to have variable androgenicity, the net effects on androgenic hormones in vivo may be similar. A study comparing levonorgestrel to gestodene, a newer and less androgenic progestin, found that sex hormone binding globulin was elevated 2-fold in the levonorgestrel group and 3-fold in the gestodene group. After administering each preparation, serum levels of luteinizing hormone, follicle-stimulating hormone, estradiol, and progesterone were depressed, with greater reductions seen for gestodene. However, equal decreases were found in testosterone, androstenedione, and dehydroepiandrosterone sulfate with both preparations.11 Thus, although each progestin possessed inherently different androgenic potencies, in the end, both of them reduced circulating levels of androgens similarly.
The strengths of our study include the population-based ascertainment of cases and controls; the large number of incident ovarian cancer cases; and the use of life-events calendars, comprehensive picture books, and structured interviews to enhance the recollection of medical information and contraceptive preparations used. All of the methodological features limited the potential for selection bias and information bias. Moreover, because both participants and our interviewers were unaware of the research question addressed here, it is unlikely that recall bias or interviewer bias could account for our results.
Our study was limited by the small number of women who had taken androgenic progestin oral contraceptives. Only 224 women had ever used androgenic OCs, 108 exclusively and 92 in a combination of androgenic and nonandrogenic life OC use. Additionally, there were many women who did not know the exact formulation of oral contraceptive that they had been taking. Nonetheless, we were able to classify 78% of OC users according to the androgenic properties of the progestin formulation. The rest could not be classified because of insufficient information on the exact OC formulation. Other studies that have attempted to assess the relationship between OC formulations and ovarian cancer have also suffered from this limitation.14,59 Although validation studies have found that recall of use and timing of OCs is quite accurate, recall for specific formulations is less so.60–62
A second potential limitation was that we did not adjust for polycystic ovary syndrome (PCOS) or the use of the medication Estratest in our analyses, which might have added a bias in androgenic exposures. However, a total of only 5 study participants had used Estratest in their lifetime, and 13 participants reported a history of PCOS, which should not have altered our results.
Another limitation of our study comes from the nature of the discovery and marketing of new progestins over the course of the past 40 years. The study was conducted in the late 1990s, and although the cohort of women who participated did contain present OC users, many of the women were past their childbearing years. The newer progestins, such as desogestrel and norgestimate, have not been available in OCs for nearly as long as older progestins, such as norethindrone or norgestrel. Of all of our pill users, only 15 had taken OCs containing low-dose, monophasic desogestrel, and 9 had taken low-dose, tri-phasic norgestimate. On the other hand, hundreds of women had taken OCs containing norgestrel or norethindrone. However, we believe that our calculation of OC androgenicity was highly accurate and would apply to the newer progestins with equivalent validity. Norgestimate has an androgenic level of 1.9 when prescribed in a 1-mg dose. For the 9 women who had used norgestimate in our study in a tri-phasic form (0.18, 0.215, 0.25 mg), the highest dose (0.25 mg) had an androgenic potency of 0.47. Thus, it was categorized as nonandrogenic.
In summary, our findings indicate that, for women in general, the androgenicity of the progestin component does not alter the OC's protective effect. Given the many currently available OCs that contain androgenic progestins, our findings are reassuring. However, the possibility that androgenic OCs may be less effective than nonandrogenic OCs for women with endometriosis warrants further investigation.
2. The reduction in risk of ovarian cancer associated with oral-contraceptive use. The Cancer and Steroid Hormone Study of the Centers for Disease Control and the National Institute of Child Health and Human Development. N Engl J Med 1987;316:650–5.
3. Ness RB, Grisso JA, Klapper J, Schlesselman JJ, Silberzweig S, Vergona R, et al. Risk of ovarian cancer in relation to estrogen and progestin dose and use characteristics of oral contraceptives. SHARE Study Group. Steroid Hormones and Reproductions. Am J Epidemiol 2000;152:233–41.
4. Risch HA, Marrett LD, Howe GR. Parity, contraception, infertility, and the risk of epithelial ovarian cancer. Am J Epidemiol 1994;140:585–97.
5. Rodriguez GC, Walmer DK, Cline M, Krigman H, Lessey BA, Whitaker RS, et al. Effect of progestin on the ovarian epithelium of macaques: cancer prevention through apoptosis? J Soc Gynecol Investig 1998;5:271–6.
6. Vessey MP, Painter R. Endometrial and ovarian cancer and oral contraceptives: findings in a large cohort study. Br J Cancer 1995;71:1340–2.
7. Franceschi S, Parazzini F, Negri E, Booth M, La Vecchia C, Beral V, et al. Pooled analysis of 3 European case-control studies of epithelial ovarian cancer: III Oral contraceptive use. Int J Cancer 1991;49:61–5.
8. Riman T, Persson I, Nilsson S. Hormonal aspects of epithelial ovarian cancer: review of epidemiological evidence. Clin Endocrinol (Oxf) 1998;49:695–707.
9. Ellis J. Low-dose oral contraceptives: progestin potency, androgenicity, and atherogenic potential. Clin Ther 1986;8:607–18.
10. Darney PD. The androgenicity of progestins. Am J Med 1995;98:104S–110S.
11. Refn H, Kjaer A, Lebech AM, Borggaard B, Schierup L. Clinical and hormonal effects of two contraceptives: correlation to serum concentrations of levonorgestrel and gestodene. Contraception 1990;41:259–69.
12. Renaud R, Lellouche D. Combined estrogen-progestogen contraception in 1987 and its complications [in French]. Rev Prat 1987;37:2267–76.
13. Petitti DB. Clinical practice. Combination estrogen-progestin oral contraceptives [published erratum appears in N Engl J Med 2004;350:92]. N Engl J Med 2003;349:1443–50.
14. Schildkraut JM, Calingaert B, Marchbanks PA, Moorman PG, Rodriguez GC. Impact of progestin and estrogen potency in oral contraceptives on ovarian cancer risk. J Natl Cancer Inst 2002;94:32–8.
15. Chu MC, Lobo RA. Formulations and use of androgens in women. Mayo Clin Proc 2004;79(suppl):S3–7.
16. Jones EE. Androgenic effects of oral contraceptives: implications for patient compliance. Am J Med 1995;98:116S–119S.
17. Balassone ML. Risk of contraceptive discontinuation among adolescents. J Adolesc Health Care 1989;10:527–33.
18. de Cetina TC, Reyes LP, Gamboa LV, Dunson TR, Rowan AJ, Waszak CS, et al. A comparative clinical trial of Norinyl 1 + 35 versus Norinyl 1 + 50 in Merida, Yucatan, Mexico. Adv Contracept 1990;6:125–39.
19. Pratt WF, Bachrach CA. What do women use when they stop using the pill? Fam Plann Perspect 1987;19:257–66.
20. Schildkraut JM, Schwingl PJ, Bastos E, Evanoff A, Hughes C. Epithelial ovarian cancer risk among women with polycystic ovary syndrome. Obstet Gynecol 1996;88:554–9.
21. Kuhnel R, de Graaff J, Rao BR, Stolk JG. Androgen receptor predominance in human ovarian carcinoma. J Steroid Biochem 1987;26:393–7.
22. Chadha S, Rao BR, Slotman BJ, van Vroonhoven CC, van der Kwast TH. An immunohistochemical evaluation of androgen and progesterone receptors in ovarian tumors. Hum Pathol 1993;24:90–5.
23. Cardillo MR, Petrangeli E, Aliotta N, Salvatori L, Ravenna L, Chang C, et al. Androgen receptors in ovarian tumors: correlation with oestrogen and progesterone receptors in an immunohistochemical and semiquantitative image analysis study. J Exp Clin Cancer Res 1998;17:231–7.
24. Ilekis JV, Connor JP, Prins GS, Ferrer K, Niederberger C, Scoccia B. Expression of epidermal growth factor and androgen receptors in ovarian cancer. Gynecol Oncol 1997;66:250–4.
25. Horie K, Takakura K, Imai K, Liao S, Mori T. Immunohistochemical localization of androgen receptor in the human endometrium, decidua, placenta and pathological conditions of the endometrium. Hum Reprod 1992;7:1461–6.
26. Risch HA. Hormonal etiology of epithelial ovarian cancer, with a hypothesis concerning the role of androgens and progesterone. J Natl Cancer Inst 1998;90:1774–86.
27. Evangelou A, Jindal SK, Brown TJ, Letarte M. Down-regulation of transforming growth factor beta receptors by androgen in ovarian cancer cells. Cancer Res 2000;60:929–35.
28. Cottreau CM, Ness RB, Modugno F, Allen GO, Goodman MT. Endometriosis and its treatment with danazol or lupron in relation to ovarian cancer. Clin Cancer Res 2003;9:5142–4.
29. Modugno F, Ness RB, Allen GO, Shildkraut JM, Davis FG, Goodman MG. Oral contraceptive use, reproductive history, and the risk of epithelial ovarian cancer in women with and without endometriosis. Am J Obstet Gynecol 2004;191:733–40.
30. Bancroft J, Sherwin BB, Alexander GM, Davidson DW, Walker A. Oral contraceptives, androgens, and the sexuality of young women II. The role of androgens. Arch Sex Behav 1991;20:121–35.
31. Cheung MC, Walden CE, Knopp RH. Comparison of the effects of triphasic oral contraceptives with desogestrel or levonorgestrel on apolipoprotein A-I-containing high-density lipoprotein particles. Metabolism 1999;48:658–64.
32. Coenen CM, Thomas CM, Borm GF, Rolland R. Comparative evaluation of the androgenicity of four low-dose, fixed-combination oral contraceptives. Int J Fertil Menopausal Stud 1995;40:92–7.
33. Collins D. Selectivity information on desogestrel. Am J Obstet Gynecol 1993;168:1010–6.
34. De Jager E. A new progestogen for oral contraception. Contracept Deliv Syst 1982;3:11–5.
35. Fotherby K, Caldwell AD. New progestogens in oral contraception. Contraception 1994;49:1–32.
36. Gaspard UJ, Romus MA, Gillain D, Duvivier J, Demey-Ponsart E, Franchimont P. Plasma hormone levels in women receiving new oral contraceptives containing ethinyl estradiol plus levonorgestrel or desogestrel. Contraception 1983;27:577–90.
37. Gerstman B. Trends in the content and use of oral contraceptives in the United States, 1964–88. Am J Pub Health 1991;81:90–6.
38. Goldzieher J. Hormonal contraception: pills, injections and implants. London, Ontario, Canada: EMIS-Canada; 1998.
39. Kafrissen ME. A norgestimate-containing oral contraceptive: review of clinical studies. Am J Obstet Gynecol 1992;167:1196–202.
40. Kaplan B. Desogestrel, norgestimate, and gestodene: the newer progestins. Ann Pharmacother 1995;29:736–42.
41. Linn ES. Clinical significance of the androgenicity of progestins in hormonal therapy in women. Clin Ther 1990;12:447–55.
42. Lobo RA. The androgenicity of progestational agents. Int J Fertil 1988;33:6–12.
43. Phillips A, Demarest K, Hahn DW, Wong F, McGuire JL. Progestational and androgenic receptor binding affinities and in vivo activities of norgestimate and other progestins. Contraception 1990;41:399–410.
44. Raudrant D, Rabe T. Progestogens with antiandrogenic properties. Drugs 2003;63:463–92.
45. Runnebaum B. The androgenicity of oral contraceptives: the young patient's concerns. Int J Fertil 1992;37:211–7.
46. Thorneycroft IH. Update on androgenicity. Am J Obstet Gynecol 1999;180:288–94.
47. van der Vange N, Blankenstein MA, Kloosterboer HJ, Haspels AA, Thijssen JH. Effects of seven low-dose combined oral contraceptives on sex hormone binding globulin, corticosteroid binding globulin, total and free testosterone. Contraception 1990;41:345–52.
48. Upton GV, Corbin A. The relevance of the pharmacologic properties of a progestational agent to its clinical effects as a combination oral contraceptive. Yale J Biol Med 1989;62:445–57.
49. Speroff L, Glass RH, Case NG. Clinical gynecologic endocrinology and infertility. Philadelphia (PA): Lippincott Williams & Wilkins; 1999.
50. Hatcher R, Stewart F, Trussell J, Kowal D. Contraceptive technology, 1990–1992. 15th ed. New York (NY): Irvington Publishers; 1990.
51. Dickey RP, Stone SC. Progestational potency of oral contraceptives. Obstet Gynecol 1976;47:106–12.
52. Piper JM, Kennedy DL. Oral contraceptives in the United States: trends in content and potency. Int J Epidemiol 1987;16:215–21.
53. Pasinetti E, Falsetti L. Triphasic pills: variability of endocrine parameters and of sex steroid-binding globulins. Acta Eur Fertil 1993;24:67–70.
54. Riman T, Dickman PW, Nilsson S, Correia N, Nordlinder H, Magnusson CM, et al. Risk factors for invasive epithelial ovarian cancer: results from a Swedish case-control study. Am J Epidemiol 2002;156:363–73.
55. Ness RB, Grisso JA, Vergona R, Klapper J, Morgan M, Wheeler JE. Oral contraceptives, other methods of contraception, and risk reduction for ovarian cancer. Epidemiology 2001;12:307–12.
56. Modugno F, Ovarian Cancer and High-Risk Women Symposium Presenters. Ovarian cancer and high-risk women-implications for prevention, screening, and early detection [published erratum appears in Gynecol Oncol 2004;92:731]. Gynecol Oncol 2003;91:15–31.
57. Altman DG, Bland JM. Interaction revisited: the difference between two estimates. BMJ 2003;326:219.
58. Zeimet AG, Muller-Holzner E, Marth C, Daxenbichler G. Immunocytochemical versus biochemical receptor determination in normal and tumorous tissues of the female reproductive tract and the breast. J Steroid Biochem Mol Biol 1994;49:365–72.
59. Rosenblatt KA, Thomas DB, Noonan EA. High-dose and low-dose combined oral contraceptives: protection against epithelial ovarian cancer and the length of the protective effect. The WHO Collaborative Study of Neoplasia and Steroid Contraceptives. Eur J Cancer 1992;28A:1872–6.
60. West SL, Savitz DA, Koch G, Strom BL, Guess HA, Hartzema A. Recall accuracy for prescription medications: self-report compared with database information. Am J Epidemiol 1995;142:1103–12.
61. Coulter A, Vessey M, McPherson K, Crossley B. The ability of women to recall their oral contraceptive histories. Contraception 1986;33:127–37.
62. Harlow SD, Linet MS. Agreement between questionnaire data and medical records: the evidence for accuracy of recall. Am J Epidemiol 1989;129:233–48.