Safety of testosterone treatment in postmenopausal women

Transcription

1 MODERN TRENDS Edward E. Wallach, M.D. Associate Editor Safety of testosterone treatment in postmenopausal women Glenn D. Braunstein, M.D. Department of Medicine, Cedars-Sinai Medical Center, David Geffen School of Medicine at UCLA, Los Angeles, California Objective: To critically examine the safety of T therapy given to postmenopausal women. Design: MEDLINE literature review, cross-reference of published data, and review of Food and Drug Administration transcripts. Result(s): Although some retrospective and observational studies provide some long-term safety data, most prospective studies have had a duration of 2 years or less. In addition, with the exception of the female-to-male transsexuals, T was administered in conjunction with estrogens or estrogens and progestins, which confound the interpretation of some of the studies. The major adverse reactions are the androgenic side effects of hirsutism and acne. There does not appear to be an increase in cardiovascular risk factors, with the exception of a lowering of high-density lipoprotein with oral T. There are little data on endometrial safety, and most of the experimental data support a neutral or beneficial effect in regards to breast cancer. There does not appear to be an increased risk of hepatotoxicity, neurobehavorial abnormalities, sleep apnea, or fetal virilization (in premenopausal women) with the physiologic treatment doses of T. Conclusion(s): Except for hirsutism and acne, the therapeutic administration of T in physiologic doses is safe for up to several years. However, prospectively collected long-term safety studies are needed to provide a greater degree of assurance. (Fertil Steril Ò 2007;88:1 17. Ó2007 by American Society for Reproductive Medicine.) Key Words: Testosterone, androgens, hirsutism, virilization, breast carcinoma, endometrial carcinoma, lipids For over 60 years, women have been treated with T to improve libido (1 4). Results of the first randomized, doubleblind placebo-controlled trial of T therapy for menopausal women, published in 1950 by Greenblatt and coworkers, demonstrated an improvement in libido (5). Subsequently, numerous trials of T treatment for women with low libido have generally demonstrated efficacy for sexual desire, arousal, pleasure, responsiveness, and satisfaction (6 8). These trials also examined the safety of T administration to women. There is a low incidence of mild androgenic effects (hirsutism and acne) and lowering of HDLs with methyltestosterone (MT) given orally. There have been no reports of serious adverse effects due to T, even with doses that cause moderately supraphysiologic blood T concentrations. However, the duration of most trials is 6 months or less, and concerns about the long-term safety of T treatment in women have been raised by a Food and Drug Administration (FDA) advisory panel (9). The purpose of this article is to Received August 15, 2006; revised and accepted January 11, Dr. Braunstein is a consultant and principal investigator on studies using transdermal testosterone patches for the treatment of hypoactive sexual desire disorder sponsored by Procter and Gamble Pharmaceuticals. Reprint requests: Glenn D. Braunstein, M.D., Room 2119 South Tower, Cedars-Sinai Medical Center, 8700 Beverly Blvd., Los Angeles, CA (FAX: ; braunstein@cshs.org). review the safety concerns and examine the existing data that address these issues. SAFETY CONCERNS The primary safety issues are androgenic side effects; adverse effects on the cardiovascular system; stimulation of the endometrium and breast leading to neoplasia; hepatotoxicity; induction or aggravation of sleep apnea; and behavioral effects (8 11). Some of these concerns are based on problems that have been found in males receiving pharmacological doses of T or anabolic steroids, in women with hyperandrogenic polycystic ovaries, and from the Heart and Estrogen/ Progestin Replacement Study (HERS) and Women s Health Initiative (WHI), in which estrogens and progestins were found to have adverse effects (12, 13). Although these models are useful for providing signals as to the type of safety issues that should be examined, each is fraught with difficulties in trying to extrapolate the findings to women receiving physiologic to moderately supraphysiologic doses of T. Under normal circumstances, circulating T levels in men are times higher than those in women. The administration of supraphysiologic doses of T or anabolic steroids to athletes and body builders has been associated with acne, gynecomastia, testicular atrophy, hepatic inflammation, hepatocellular adenomas and carcinomas, sleep apnea, aggressive /07/$32.00 Fertility and Sterility â Vol. 88, No. 1, July doi: /j.fertnstert Copyright ª2007 American Society for Reproductive Medicine, Published by Elsevier Inc.

2 behavior, polycythemia, lipid abnormalities, and accelerated atherosclerosis (14). Some of these effects, such as the liver effects, are related to the type of anabolic steroid used, while others, like polycythemia, are dose related. Women with polycystic ovarian syndrome (PCOS) have endogenous hyperandrogenism and exhibit androgenic effects including hirsutism, acne, and androgenic alopecia, as well as occasional virilization. They also have an increased risk of endometrial carcinoma, metabolic syndrome, diabetes, and cardiovascular disease (15). Although it is tempting to ascribe a cause and effect relationship to the hyperandrogenism and these latter effects, it is important to note that many of these patients are obese and exhibit insulin resistance that may in fact be the proximate pathophysiologic factors for these problems. The hazard ratios of 1.29 for coronary heart disease, 1.41 for stroke, 2.11 for venous thromboembolism, and 1.26 for breast cancer in women receiving estrogen and progestin for an average of 5.2 years in the WHI raised major concerns about sex steroid therapy for postmenopausal women (13).In postmenopausal women with coronary artery disease (CAD) enrolled in HERS, there was an increase in cardiovascular events in the estrogen/progestin group during the first year of treatment and then a reduction in years 3 5 (12). Since most of the studies that evaluated T therapy in postmenopausal women did so in the presence of estrogen (if the patient had undergone a hysterectomy) or estrogen and a progestin (if she had undergone a natural menopause), there has been a concern that adding androgen may compound any adverse cardiovascular effect of the other sex steroid hormones. Additionally, since T is aromatized into E 2 in a variety of peripheral tissues through the action of aromatase, it has been hypothesized that giving T will result in greater tissue exposure to estrogen (10). Thus, it is speculated that estrogen-induced problems may be exacerbated with androgen therapy in which an aromatizable androgen such as T is given. Given this background, it is understandable that an FDA advisory committee might be concerned about potential long-term adverse effects of T therapy for women. However, with the exception of the androgenic side effects, these concerns appear to be unwarranted based on available data, which will be reviewed below. ANDROGENIC EFFECTS Hirsutism Hair can be classified as asexual, which is present in both sexes at birth (e.g., scalp and eyebrow regions), ambisexual, which develops in the axillary and pubic regions in both sexes in response to the rise in adrenal androgens at adrenarche, and sexual, which is on the face and body and responds to the rising androgen concentrations during the pubertal transition in males. The pilosebaceous unit is composed of a hair papilla and associated sebaceous glands. In the presence of androgens, fine vellus hair in androgen-sensitive areas of the body will become dark, thick, terminal hairs, and the sebaceous glands will enhance sebum production (16). Hirsutism, which represents excessive hair growth in the androgen-sensitive areas of women, is an expected consequence of excessive production of endogenous androgens or exposure to supraphysiologic quantities of exogenous androgens (17, 18). Thus, it is not surprising that hirsutism is the major side effect of exogenous androgen administration in women. As reviewed by Gelfand and Wiita, between 1% and 36% of women receiving various doses of MT, with or without estrogen, noted hirsutism (19). However, as summarized in Table 1, there is no clear-cut dose-response effect noted. This may reflect the means by which the hirsutism is reported (selfreport by the patient vs. observation by the investigator) or the duration of androgen administration, as it often takes 4 6 months of exposure to exogenous T before hirsutism becomes apparent (19 30). In a randomized, double-blind study of postmenopausal women receiving conjugated equine estrogens (CEEs; or 1.25 mg/day) or esterified estrogens (EEs) þ MT (low dose, mg EE þ 1.25 mg MT; high dose, 1.25 mg EE þ 2.5 mg MT), facial hair was reported as an adverse event in 3% of women on CEE and in 6% receiving EE þ MT. In a subgroup of 100 women in whom facial hair was absent at the start of the study, 14% TABLE 1 Frequency of hirsutism and acne noted as adverse effects in women receiving oral methyltestosterone with and without EEs. Dose of MT/day, mg Hirsutism N (%) Acne N (%) References 10 mg a 266 (7.5) 266 (11.3) (20 22) 2.5 mg b 196 (11.7) 97 (15.46) (23 28) 1.25 mg b 196 (0.8) 120 (9.2) (25, 26, 29, 30) Note: Percents represent the mean of all data combined; N ¼ number of subjects. a MT administered without estrogens. b MT administered with EEs. Braunstein. Testosterone safety in women. Fertil Steril Braunstein Testosterone safety in women Vol. 88, No. 1, July 2007

3 of the women receiving the low dose of EE þ MT developed facial hirsutism, as did 22% of those receiving the higher dose of EE þ MT at 12 months. (25). However, 20% of the women in both of the estrogen-only control groups also developed facial hair. Similarly, Lobo et al. found no change in the hirsutism score in 107 women receiving mg of EE þ 1.25 mg MT for 4 months versus 109 women receiving mg EE alone in a randomized, double-blind trial (29). Table 2 provides the levels of serum or plasma testosterone, free testosterone, and/or bioavailable testosterone achieved with the various doses of MT or the other androgens used to treat women with low libido (23, 27 53). Intramuscular injections of 150 mg of T with estrogen, which lead to supraphysiologic T levels, have been associated with a 15% 20% incidence of hirsutism (54). In contrast, with a dose of 75 mg of T, less than 5% of women develop hirsutism (55). Subcutaneous implants of E 2 and T also result in high serum T levels and may be associated with hirsutism. Burger noted hirsutism in 6% of women receiving implants of 40 mg E 2 and 100 mg of T over 6 months (52), while Cardozo et al. indicated that 21% of their patients developed hirsutism using implants containing 50 mg of E 2 and 100 mg of T (53). In contrast, Davis et al. treated women with pellets containing 50 mg E 2 and 50 mg T and did not note any hirsutism over 2 years (48). There have been several double-blind, placebo-controlled trials of transdermal T in women who have undergone surgical menopause and who were receiving concomitant estrogen treatment. They have shown no difference in objective changes in facial hair growth in women receiving 300 mg/ day of T via a patch that was changed twice weekly versus those receiving only placebo (35, 36, 38, 39). This dose of T results in median serum free T levels within the range found in reproductive-age normal women (35; Table 2). Combining the data from one phase II and two phase III studies with the T patch, 5.12% (34/664) of women receiving placebo patches and 6.98% (46/659) of women receiving 300 mg/day of T reported hirsutism as an adverse event. However, in one study, the facial depilation rate was significantly higher in women who received the 300-mg/day dose of T for 24 weeks than in the women receiving placebo (35). In a placebo-controlled, crossover study of 12 weeks of treatment of premenopasual women with low libido who received either T cream (10 mg/day) or placebo, there was no significant change in the hirsutism score with either treatment (43). These results indicate that there is little or no increase in objective hair growth in women receiving physiologic to slightly supraphysiologic doses of T compared with placebo. However, as the levels of serum T increase, the frequency of reported hirsutism increases. Acne Acne vulgaris results from androgen-stimulated sebum production, keratinization of the sebaceous duct, and colonization of the duct with the anaerobic diphtheroid Propionibacterium acnes (16). T stimulates sebum production in a dose-response manner (17). Acne has been noted in 1% 45% of women receiving 10 mg of MT orally per day, 8% 30% ingesting 2.5 mg a day, and 5.6% 38.5% taking 1.25 mg daily (Table 1). In the same studies, 0 7% of the women on estrogen alone noted acne (23, 25, 26, 29, 56). Approximately 3% of women receiving a combination of estrogen þ 100 mg of T implants were noted to develop acne (52, 53), while none of 16 patients treated with SC pellets of estrogen and 50 mg of T developed acne (48). In the transdermal T trials, 8.65% (57/659) of the women receiving 300 mg/day noted acne, while 6.66% (44/664) of the women on placebo patches developed acne (36, 38, 39). None of the 31 premenopausal women with low libido treated with 10 mg/day of T cream for 12 weeks developed acne (43). In a study of 53 postmenopausal women who received combined estrogen/progestin hormone therapy þ placebo or 10 mg of a T gel, Nathorst-Boos and coworkers noted no significant difference (17% vs. 20.7%) in acne, facial hair, or hair on the legs after 3 months of treatment in a randomized, double-blind crossover trial (44). Virilization Signs of virilization include deepening of the voice, increase in muscle mass, temporal hair recession, and clitoromegaly. All of these changes occur in women with androgen-producing tumors as well as in women receiving large doses of T as part of the female-to-male transsexual conversion (57 59).In two phase III trials of 300 mg/day of transdermal T or placebo given to surgically menopausal women receiving concomitant estrogen treatment, 4.2% (23/549) of women reported alopecia as an adverse reaction, while 2.9% (16/545) of the women receiving placebo had similar complaints (38, 39). Excessive use of an injectable androgen-estrogen combination that resulted in serum T concentrations that were times the upper limit of normal for premenopausal women and times the upper limit of normal for postmenopausal women were noted to result in hirsutism in eight of nine women, clitoral enlargement in seven of the nine, mood changes in three, and temporal hair balding in one (60, 61). Of interest, up to 2 years were required for the T levels to fall to within the normal range in some of these women (60). Clitoromegaly has also been occasionally noted in women receiving T implants (10). In female-to-male transsexuals who received mg of intramuscular testosterone cypionate every month, the stretched clitoral length increased from an average of about 1.25 cm to approximately 5 cm over time, with a plateau occurring at about one year (58). Table 3 provides the mean serum testosterone levels reached with various doses of intramuscular testosterone injections in female-to-male transsexuals (58, 59). As androgen receptors are present in laryngeal tissue, deepening of the voice would be anticipated with T use; it is certainly seen in androgen-treated female-to-male transsexuals (57). It has also been noted in women treated with Fertility and Sterility â 3

4 4 Braunstein Testosterone safety in women Vol. 88, No. 1, July 2007 TABLE 2 Serum or plasma testosterone, free testosterone, and/or bioavailable testosterone levels in women receiving testosterone preparations for low libido expressed as a percentage of the upper limit of the reference range for the method used. Study Drug Route Dose With estrogen Percent of Upper Limit of Reference Range Total testosterone Free T Bio T Lobo et al. (29) MT PO 1.25 mg/d yes 38% ND 109% 4 mo Dobs et al. (27) MT PO 2.5 mg/d yes ND 96% ND 4 mo Raisz et al. (23) MT PO 2.5 mg/d yes 47% ND ND 9 wks Warnock et al. (28) MT PO 2.5 mg/d yes 37% 63% 112% 2 mo Flöter et al. (31) TU PO 40 mg/d no 107% ND ND 1-4 d Flöter et al. (32) TU PO 40 mg/d yes 163% 231% ND 6 mo Penotti et al. (33) TU PO 40 mg/d yes 96% ND ND 8 mo Miller et al. (34)* T TD-patch 150 mg/d no 171% 87% ND 3 mo Shifren et al. (35) T TD-patch 150 mg/d yes 118% 57% 56% 3 mo Braunstein et al. (36) T TD-patch 150 mg/d yes 89%** 29%** 24%** 6mo Choi et al. (37)*,# T TD-patch 300 mg/d no 76% 115% ND 6 mo Miller et al. (34)* T TD-patch 300 mg/d no 307% 182% ND 3 mo Shifren et al. (35) T TD-patch 300 mg/d yes 189% 87% 90% 3 mo Buster et al. (38) T TD-patch 300 mg/d yes 131%** 42%** 56%** 6mo Simon et al. (39) T TD-patch 300 mg/d yes 140%** 55%** ND 6 mo Braunstein et al. (36) T TD-patch 300 mg/d yes 182%** 51%** 50%** 6mo Davis et al. (40) T TD-patch 300 mg/d yes 126%** 66%** 56%** 6mo Shifren et al. (41) T TD-patch 300 mg/d yes 108%** 38%** 38%** 6mo Braunstein et al. (36) T TD-patch 450 mg/d yes 245%** 81%** 62%** 6mo Davis et al. (42) T TD-gel 2mg yes 83% 73% ND 4 mo Goldstat et al. (43)*** T TD-cream 10 mg no 96% 122% $ ND 3 mo Nathorst-Boos et al. (44) T TD-gel 10 mg yes 233% ND ND 3 mo Braunstein. Testosterone safety in women. Fertil Steril When tested

5 Fertility and Sterility â 5 TABLE 2 Continued. Study Drug Route Dose With estrogen Percent of Upper Limit of Reference Range Total testosterone Free T Bio T When tested Sherwin, Geldfand (45) TE-bah IM 150 mg/3 mo yes 133% ND ND 3 mo Sherwin, Geldfand (45) TE-bah IM 150 mg yes 285% ND ND 8 days Sherwin, Geldfand (45) TE IM 96 mg no 111% ND ND 3 mo Burger et al. (47) T Implant 50 mg yes 117% ND ND 6 wks Davis et al. (48) T Implant 50 mg yes 96% (79%) ND ND 6 mo (2 yrs) Worboys et al. (49) T Implant 50 mg yes 180% 139% $ ND 6 wks Kapetanakis et al. (50) T Implant 75 mg yes 357% ND ND 1 mo Buckler et al. (51)*** T Implant 100 mg/6 mo no 204% ND ND 3.3 years Burger et al. (52) T Implant 100 mg yes 223% 244% ND 1 mo Cardozo et al. (53)*** T Implant 100 mg/4-12 mo yes 167% ND ND 6 mo Cardozo et al. (53) T Implant 100 mg/4-12 mo yes 92 ND ND 6 mo Note: Total testosterone levels in women receiving estrogens tend to be higher than in women not receiving estrogens because of higher levels of sex hormone-binding globulin. Therefore, the free or bioavailable testosterone levels more accurately reflect the testosterone available to the tissues. PO, taken orally; ND, not determined; d, days; mo, months; wks, weeks; Free T, free testosterone level; Bio T, bioavailable testosterone level; T, testosterone; MT, methyltestosterone; TU, testosterone undeconate; TD, transdermal; IM, intramuscularly; TE-bah, testosterone enanthate benzilic acid hydrazone; TE, testosterone enanthate. * HIV infected women. Not all postmenopausal. ** Median and not mean reported. *** Premenopausal women. # Peak total testosterone 134% at 24 hours and peak free testosterone 140% at 24 hours. $ Free androgen index used in place of free testosterone. Braunstein. Testosterone safety in women. Fertil Steril 2007.

6 TABLE 3 Mean serum testosterone levels before and after intramuscular testosterone therapy in female-to-male transsexuals following one year of therapy. Author Drug Dose Mean Serum testostone (ng/dl) Meyer et al. (58) None 93 ng/ml T cypionate mg/2 weeks 817 T cypionate 200 mg/2 weeks 922 T cypionate 400 mg/2 weeks 1765 Elbers et al. (59) None 46 T esters 250 mg/2 weeks 896 Note: To convert ng/dl to pmol/l divide by T ¼ Testosterone. Braunstein. Testosterone safety in women. Fertil Steril nandrolone decanoate added to menopausal hormonal therapy (62). Voice changes have been reported in 10% 20% of women using 10 mg of MT orally a day for 6 weeks to 3 years (20, 22, 63). In the transdermal T studies, 2.2% (12/ 545) of participants receiving placebo noted voice change, while 2.7% (15/549) of the women receiving T had similar complaints (38, 39). CARDIOVASCULAR EFFECTS Epidemiological Studies Several studies have examined the relationship between endogenous serum T levels and CAD in women (64). Slowinska-Srzednicka and colleagues studied 21 women who had sustained a myocardial infarction and had coronary artery stenosis on angiography and compared them with 14 women with an abnormal exercise stress test but normal coronary arteries and nine healthy women (65). All women were premenopausal. The women with CAD had lower levels of DHEAS and higher fasting insulin levels and insulin response to an oral glucose tolerance test, but there were no significant differences between the groups in regards to the concentration of E 2, androstenedione, or T. Another study of 60 postmenopausal women undergoing angiography used multiple regression analysis to evaluate the relationship between the degree of CAD and several variables including lipids, E 2, total T, free T, DHEAS, sex hormone binding globulin (SHBG), and insulin. Only free T and total cholesterol showed a significant relationship to the degree of CAD (66). Another cross-sectional study examined the relationship between carotid artery intimal-medial thickness and DHEAS, androstenedione, total and free T, and insulin in 101 pre- and postmenopausal women. Of interest, DHEAS, androstenedione, and free T were inversely related to carotid artery thickness, suggesting that within the physiological range, DHEAS and androgens in women are correlated with a lower risk of carotid artery atherosclerosis (67). Low serum levels of SHBG have been associated with the metabolic syndrome and cardiovascular risk factors in women with PCOS. However, this syndrome is also associated with hyperinsulinism, and the insulin excess may inhibit hepatic SHBG production. In the Study of Women Across the Nation (SWAN), low SHBG and/or high free androgen index were related to a variety of cardiovascular risk factors including higher insulin and glucose, higher plasminogen activator inhibitor type 1, tissue plasminogen activator, highly sensitive C-reactive protein, and activated factor VII-c (68, 69). In the Gothenburg, Sweden, longitudinal study of randomly selected women, Lapidus and coworkers evaluated the SHBG levels in a cohort of postmenopausal women not receiving hormone therapy and noted no relationship between the 12- year incidence of myocardial infarction, angina pectoris, or stroke but did find a significant correlation between low SHBG levels and overall mortality (70). A more direct examination of the relationship among SHBG, androgens, and cardiovascular disease was carried out by Hauner et al. They found no differences in the serum concentrations of T and SHBG in 30 women with CAD on coronary angiograms, 21 women undergoing angiography but found to be free of major stenosis, and 25 healthy women without signs of CAD (71).In a similar study, Reinecke and colleagues evaluated 87 consecutive postmenopausal women undergoing coronary angiograms and found that in comparison with the 32 women without CAD, the 55 women with CAD had significantly higher basal insulin levels, higher insulin/glucose ratio, lower insulin sensitivity, and lower SHBG levels but no difference in T or DHEAS concentrations (72). A multivariate analysis by logistic regression indicated that low levels of SHBG were associated with CAD independently of insulin, obesity markers, and dyslipidemia. A relationship between low SHBG concentrations and atherosclerosis was also demonstrated in a casecontrol study carried out in a group of 182 participants in the Atherosclerosis Risk in Communities cohort of postmenopausal women who were at or above the 95th percentile for carotid intimal-medial thickness and who were compared with 182 controls with carotid thickness less than the 75th percentile. Cases had significantly higher concentrations of estrone and androstenedione and lower SHBG levels than controls, but no differences were noted with DHEAS, total 6 Braunstein Testosterone safety in women Vol. 88, No. 1, July 2007

7 T, or total T/SHBG ratio. Compared with participants in the lowest quartile of total T, those in the highest quartile had lower odds of atherosclerosis, as did women in the highest quartile for SHBG. Thus, total T levels in the high normal range were associated with less atherosclerosis than those with lower T levels (73). In a nested case-control study of 200 postmenopausal women who developed cardiovascular disease and who were participating in the Women s Health Study, low SHBG and elevated free androgen index were noted in women who were not using hormone therapy. However, after adjustment for known cardiovascular risk factors, an independent association was no longer present (74). A 19-year prospective study carried out in the Rancho Bernardo, California, community examined the relationship between androstenedione, total and bioavailable T, estrone, and total and bioavailable E 2 concentrations and the risk of death from cardiovascular and ischemic heart disease. None of the age-adjusted hormone concentrations differed significantly in women with or without a history of heart disease at baseline, and none predicted cardiovascular death (75). There are very little data available concerning the longterm cardiovascular safety of androgen administration to women. Postmarketing data on a combination product containing EEs and MT (1.25 or 2.5 mg) did not identify an increased risk of cardiovascular disease in women (19, 56). Van Kestern et al. performed a retrospective analysis of 293 female-to-male transsexuals with a mean age of 34 years (range, years) who were treated with large doses of T for a total of 2,418 patient-years (range, 2 months to 41 years). They did not find any cardiovascular mortality or an increase in myocardial infarction or hypertension over that expected in the population at large (76). Blood Pressure The majority of epidemiological studies have not found a relationship between endogenous serum androgen or SHBG concentrations and systolic or diastolic blood pressure (67, 70, 75, 77). In the SWAN cohort, there was a correlation between total T and free androgen index and diastolic but not systolic blood pressure (68). None of the treatment trials with a combination of oral estrogens and MT or T undeconate, SC T pellets, or T patches have shown an increase in blood pressure after androgen administration, and one noted a decrease with EEs and MT given for 2 months (24, 25, 35, 36, 38, 39, 49, 78). In a 3-month trial of oral T undecanoate given to postmenopausal women in doses to raise the serum T into the premenopausal female range, there was a significant decrease in diastolic blood pressure (79). In addition, the studies carried out on female-to-male transsexuals who received doses of exogenous T to raise the serum T level into the mid-to-high male range have not shown an increase in blood pressure (58, 59, 76). Lipids The effect of exogenous T on lipids in women depends on the dose, whether estrogens or estrogen þ a progestin are concurrently given, and the route of administration. Female-to-male transsexuals using pharmacologic doses of T, which increases the serum T into the mid-to-high normal range for males, experience increased total cholesterol, low-density lipoprotein (LDL) cholesterol, apolipoprotein- B, and triglycerides and decreased high-density lipoprotein (HDL) cholesterol with pharmacological quantities of IM T, which increase the serum T into the high normal range for males (58, 80). In the trials of postmenopausal women receiving estrogen and oral MT, the total cholesterol decreased by 6% 17%; HDL was reduced as much as 26%; and the changes in LDL were variable, going from a 7% increase to a 19.3% decrease. In most studies, there was a reduction in triglycerides (Table 4; 23, 24, 26 30, 33, 35, 44, 47, 48, 52, 79, 81, 82). Thus, the major negative effect on cardiovascular risk factors with oral T therapy is the reduction in HDL cholesterol, which is due to the first-pass effect on the liver. Whether this is offset by the reduction in LDL cholesterol and triglycerides is unknown. Parental T, given by SC implants, or by IM injection in conjunction with estrogen, results in a lowering of total cholesterol, LDL cholesterol, and triglycerides, with a slight increase in HDL cholesterol, which results in a mildly beneficial, antiatherogenic lipid profile (Table 4; 82). Buckler et al. treated 22 patients with severe premenstrual syndrome with 100 mg SC T implants every 6 months for a mean duration of 3.3 years and compared them with 22 age-matched control patients with severe premenstrual syndrome who had not received T therapy (51). Of interest, all patients continued to have regular menses. The T-treated group had a reduction in apolipoprotein A1 and HDL cholesterol, an elevation in very low density lipoprotein cholesterol, and no difference in total serum cholesterol and triglycerides or LDL cholesterol, apolipoprotein B, lipoprotein(a), lecithin:cholesterol acyltransferase, and cholesteryl ester transfer protein activity compared with the control women (51). The effects of transdermal T on lipid values appear to be neutral compared with estrogen therapy alone or with placebo (Table 4; 35, 36, 38 40). Vascular Reactivity Studies in animals have shown that T may induce acute vasodilatation by a mechanism that is too rapid to be mediated through the nuclear androgen receptor and that, hence, has been thought to work through a nongenomic pathway (83). That a similar mechanism is present in humans is supported by the demonstration of an acute anti-ischemic effect of a 5-minute IV infusion of T and increased coronary artery diameter and blood flow after an infusion of T into the right coronary artery (84, 85). Sarrel and Wiita compared the effects of 1.25 mg EEs alone or 1.25 mg EE þ 2.5 mg MT given daily for 8 weeks on postocclusion fingertip blood flow and vaginal blood flow. They found a slight but nonsignificant increase in blood Fertility and Sterility â 7

8 TABLE 4 Effects of various androgen preparations on total cholesterol (TC), high-density lipoprotein cholesterol (HDL), low-density lipoprotein cholesterol (LDL), and triglycerides (TG) in women. Therapy (dose) (reference) Route n TC HDL LDL TG MT (1.25 mg) þ EE (0.625 mg) Oral (14, 26, 29, 30) MT (2.5 mg) þ EE (1.25 mg) Oral (23, 26 28, 30) TU (40 mg/day) (79) a Oral 21 NC 9 NC þ18 TU (40 mg/day) þ E 2 (2 mg/day) Oral NC (79) a TU (40 mg/day) þ TDE 2 Oral (50 mg/day) þ MPA (33) b T (300 mg/day) þ CEE (9, 35) c TD patch T (10 mg) þ E þ P (44) d TD-gel 53 þ þ3.5 þ4.2 T (50 mg/3 months) þ E 2 SC NC (50 mg) (48) T (100 mg) þ E 2 (50 mg) (81) e SC þ T (150 mg) þ E 2 (82) f IM þ NC Note: Numbers represent the mean percent change from baseline from studies in which the androgen was administered for a minimum of 1 month. a Extrapolated from Figure 2 in reference 79. b Patients received TD E 2 plus medroxyprogesterone acetate (10 mg/day) orally for 12 days every 2 months. c All patients on daily oral CEEs. Not all patients had each lipid subtype measured. d Patients were receiving combined estrogen/progestogen hormone therapy, cyclic or continuous, for at least 2 months. e Values based on lipid results at 2 months after injection, which represents the time of peak serum T. f Injections contained 150 mg T enanthate, 7.5 mg E 2 dienanthate, and 1 mg E 2 benzoate. Data are extrapolated from Figures 1 3 in reference 82. Braunstein. Testosterone safety in women. Fertil Steril flow with the estrogen-t preparation in comparison with the estrogen-only group, which indicates that the androgen treatment did not diminish the vasodilator effects of estrogen treatment in postmenopausal women (86). Another study by Penotti et al. examined the pulsatility index of the internal carotid artery and the middle cerebral artery in 40 postmenopausal women who were receiving 50 mg/day of transdermal E 2 þ medroxyprogesterone acetate 10 mg/day every other month. Half of the women received 40 mg/day of oral T undecanoate, while the other 20 served as controls. After 8 months, there was no significant difference between the groups in the pulsatility index of the internal carotid artery, while there was a significant increase in the index of the middle cerebral artery, which suggests that androgens may have reduced some of the positive effects of estrogen (33). The discrepancies between these two studies may relate to the vessels being studied, the duration of androgen administration, the level of androgen that the vessels were exposed to, and the use of estrogen þ P þ T in the Penotti et al. study as opposed to estrogen þ T in the Sarrel and Wiita study. The ability of the blood vessels to dilate in response to the reactive hyperemia after release of occlusion of the brachial artery is referred to as flow-mediated vasodilatation (FMD) and reflects shear stress-induced endogenous release of nitric oxide from the endothelium (endothelium-dependent vasodilatation). A reduction in FMD is associated with atherosclerosis (87). Female-to-male transsexuals receiving large doses of T did not exhibit a significant difference in FMD compared with age-matched normal females (59). However, endothelium-independent vasodilatation, as assessed by the change in brachial artery diameters after the administration of nitroglycerine, was significantly lower in the transsexuals than in normal women, which suggests some impairment in vascular reactivity (59). Of interest, the female-to-male transsexual patients had significantly larger baseline vascular diameter than did the female controls, which may offset any of the negative effects of the impaired endothelium-independent vasodilatation (59). In 33 postmenopausal women on estrogen therapy for more than 6 months, Worboys et al. found that 6 weeks after the SC implantation of 50 mg of T, there was a significant mean increase (42%) in brachial artery FMD in comparison with the pre-t baseline and in comparison with the response of the controls (postmenopausal women not using estrogen, 8 Braunstein Testosterone safety in women Vol. 88, No. 1, July 2007

9 progestin, or androgen therapy) (49). There was also a significant increase in nitroglycerine-induced vasodilatation after T administration, while no change was noted in the brachial artery diameter in the T-treated group (49). It is unknown whether these positive effects of T are due to a direct androgen effect on the vasculature or are mediated by conversion of T to estrogen by the aromatase present in vascular smooth muscle cells (88). Viscosity, Coagulation Factors, and Hemoglobin Increased plasma viscosity is a risk factor for cardiovascular disease and is primarily determined by levels of triglyceride and fibrinogen (89 92). Basaria and colleagues found that in 20 women receiving 1.25 mg of EEs þ 2.5 mg daily oral MT, there was an increase in serum fibrinogen, a decrease in triglycerides, and a significant decrease in plasma viscosity (93). Obviously, more studies are necessary to examine the effect of T on the factors that affect viscosity, but at present, there are no data to suggest an adverse effect of T on this risk factor. An increase in coagulation factors or a decrease in fibrinolysis are associated with cardiovascular disease (51). In the Buckler et al. study, 22 women with severe premenstrual syndrome who were treated every 6 months with a 100-mg SC T implant for over 2 years. They were compared with 22 women matched for age, and there was no difference in prothrombin time or levels of fibrinogen, antithrombin-iii, protein C, total protein S, free protein S, tissue plasminogen activator, plasminogen activator inhibitor, beta thromboglobulin, or prothrombin fragments 1 and 2, and all levels remained within the normal range. The mean serum T level in the T-treated group was ng/dl (4.5 nmol/l), while that in the control group was 54.8 ng/dl (1.9 nmol/l) (51). Four months of high-dose T treatment of 17 female-to-male transsexuals resulted in a reduction in normalized activated protein C sensitivity ratios, increases in protein S total and free antigen, and no change in protein C antigen or prothrombin levels (94). Overall, this treatment, which increased the serum T from 60.1 ng/dl (2.1 nmol/l) to 952 ng/dl (33 nmol/l), resulted in a mild antithrombotic effect. An elevated hemoglobin concentration also is a risk factor for cardiovascular disease. Since androgen therapy stimulates erythropoiesis and has been used therapeutically to treat the anemia of chronic diseases, it is reasonable to anticipate the possibility that some women might develop polycythemia with androgen treatment (95). Although slight increases in hemoglobin have been found in some patients (34), neither polycythemia nor clinically significant increases in hemoglobin have been found in the majority of studies that have examined the effect of physiological to slightly superphysiological doses of androgen given to women (24 26, 35, 36, 38 42, 78). Even with the large pharmacological doses of T used to treat female-to-male transsexuals, the hemoglobin levels only rise to the levels seen in biological males (96). Together, these results indicate that T treatment does not lead to adverse cardiovascular consequences through adverse changes in serum viscosity, coagulation or fibrinolytic factors, or the induction of polycythemia. Insulin Resistance Insulin resistance is a major etiologic component of the metabolic syndrome, which is associated with cardiovascular disease. In a case-control study of carotid atherosclerosis carried out in a subset of women participating in the Atherosclerosis Risk in Communities Study, Golden and coworkers found that the free androgen index, but not total T, was positively associated with the metabolic syndrome, particularly the hyperinsulinemia and hyperglycemia components of the syndrome (97). Several studies have examined the effects of androgen administration on metabolic factors in women. In otherwise normal surgically menopausal women, fasting glucose and insulin levels demonstrated no significant changes for up to 6 months in the phase II and phase III trials on transdermal T (35, 36, 38 40). In a placebo-controlled randomized trial of HIV-infected women with recent weight loss, 300 mg/day of transdermal T did not affect insulin sensitivity as assessed by glucose and insulin levels during a 3-hour IV glucose tolerance test (98). The administration of 30 mg of nandrolone decanoate every 2 weeks for 9 months in 10 healthy obese women on a low-fat diet led to an increase in lean body mass, a decrease in fat mass, and increased visceral fat but no change in fasting blood sugar or insulin sensitivity assessed by the Minmod method (99). Zang and colleagues performed an open-label study on 63 naturally postmenopausal women, who were randomized to receive T undecanoate (40 mg orally every second day), E 2 valerate (2 mg orally/day), or the combination for 3 months. There were no significant changes in fasting glucose or insulin in any of the groups. Basal insulin sensitivity was increased in the estrogen-only group and unchanged in the others. During the hyperinsulinemic euglycemic clamp study, glucose uptake was reduced by about 15% 20% in the T-treated groups but not in the estrogen-only group. There were no differences in the treatment effect among the three groups, which indicates that any decrease in insulin sensitivity was mild (79). Diamond et al. studied normal, regularly menstruating, nonobese women before and after days of treatment with 5 mg 3 times a day of MT. They assessed insulin sensitivity with the hyperinsulinemic hyperglycemic and euglycemic clamp techniques and found no change in fasting glucose, insulin, c-peptide, glucagons, or glucose turnover. During the high-dose insulin infusion, but not the low-dose insulin infusion, whole-body glucose uptake was significantly decreased, and the rate of glucose uptake per unit of insulin also decreased, which indicates that there was a reduction of insulin sensitivity at supraphysiologic insulin levels (100). Similar results were noted in 13 female-to-male transsexuals who were studied before and after 4 months of IM injections every 2 weeks of 250 mg of Testers (101). Endogenous glucose production was not altered, but there was a significant reduction Fertility and Sterility â 9

10 in glucose uptake during a hyperinsulinemic euglycemic clamp, which indicates some induction of insulin resistance. However, the interpretation is confounded by weight gain in the patients over the 4 months of treatment (101). Elbers et al. also studied 17 female-to-male transsexuals who were given 250 mg of T esters every 2 weeks for 1 year. Using the hyperinsulinemic euglycemic clamp technique, they found no change in fasting blood sugar or insulin levels and no change in insulin sensitivity in the physiological range. They did note a statistically significant reduction in the whole-body glucose uptake during the infusion of supraphysiological levels of insulin, consistent with the findings of Diamond and coworkers (59). Thus, T administration in otherwise healthy women that results in physiologic to slightly superphysiologic serum free T levels does not result in clinically significant insulin resistance or the metabolic syndrome. EFFECTS ON THE ENDOMETRIUM Estrogen stimulation of the endometrium, unopposed by the protective effect of luteal phase P levels, increases the risk of endometrial cancer (102). Although androgens do not directly stimulate endometrial cells ( ), endometrial cancer cells exhibit aromatase activity (104, 107, 108). Thus, there is a concern that androgens may be converted into estrogens by endometrial cancer cells and thereby stimulate endometrial cancer cellular proliferation (109, 110). In a case-controlled study of women with endometrial cancer, Potischman and colleagues found significantly higher levels of androstenedione and lower levels of SHBG in pre- and postmenopausal women with endometrial carcinoma (109). Similarily, Lukanova et al. carried out a case-control study on 124 postmenopausal women with endometrial carcinoma and found significantly increased odds ratios for endometrial cancer for quartiles with the highest hormone levels relative to the lowest for E 2, estrone, androstenedione, T, and DHEAS and an inverse relationship with SHBG (111). However, adjustment of androstenedione and T for E 2 or estrone levels resulted in a loss of statistically significant relationships. In patients with endometrial carcinoma, there is no relationship between the proliferation activity and the serum levels of androstenedione or T (106). A comparison of endometrial biopsies of 26 women randomly assigned to receive 6 months of EE alone or with 1.25 mg/day of MT, without progestins, showed estrogenstimulated proliferative growth in both groups, with no difference in distribution of biopsy scores (30). In a 4-month randomized, double-blind trial of 150 women with natural menopause, Lobo and colleagues found endometrial hyperplasia in one of 74 women taking EE and 1.25 mg/day of MT, compared with one of 76 women who were given EE alone (29). No change in endometrial thickness was noted at any time period in 53 women receiving combined estrogen/progestin treatment after natural menopause who were enrolled in a double-blind, randomized, crossover study of 10 mg/day of transdermal T or placebo for 3 months each (44). Similarly, 8 months of 40 mg/day of oral T undeconate along with 50 mg of transdermal E 2 and 10 mg/day oral medroxyprogesterone acetate given for 12 days every other month resulted in a nonsignificant 16.5% decrease in endometrial thickness, while the control group had a 2.7% decrease in thickness (33). It is difficult to draw firm conclusions about the endometrial safety of T from these studies as androgens were studied against a background of estrogens or estrogens þ progestin. Futterweit and Deligdisch studied 19 female-to-male transsexuals who received large doses of T enanthate (usually 400 mg IM every 3 4 weeks). They found that 12 of the 19 (63.2%) had a proliferative endometrium, three of which were associated with mild glandular cystic hyperplasia, and seven (36.8%) of which had an inactive endometrium (112). Thus, pharmacological doses of androgens appear to be associated with endometrial stimulation. BREAST CANCER The increased risk of breast cancer found in postmenopausal women using estrogens and progestins as postmenopausal hormone therapy in the WHI and the Million Woman Study (as well as in other trials) (13, ) has raised concerns that androgens may either directly or indirectly stimulate the development or growth of breast cancer, despite the fact that androgens had been used to treat breast cancer in the past. Normal mammary epithelial tissue and breast cancers contain androgen receptors ( ) that are coexpressed with the estrogen and P receptors. The periglandular breast fat contains the aromatase enzyme complex, which theoretically could transform circulating T into E 2, potentially stimulating the breast glandular tissue in a paracrine fashion. However, studies have shown that aromatase activity in breast cancer tissue is inversely related to the estrogen status of the women those with low serum estrogen concentrations have higher aromatase activity, while women receiving estrogens have lower aromatase activity, which suggests that androgen administration given on a background of concomitant estrogen treatment is unlikely to result in much local biotransformation to estrogens ( ). Additionally, in the transdermal T patch and gel studies, the mean serum estradiol level following testosterone administration averaged 104% of the baseline, with a range of 79% 125% in the various studies (34 36, 38 44). Following the intramuscular administration of testosterone esters in female-to-male transsexuals, the average serum estradiol level was 99.2% of the baseline, with a range of 96% 113% (56, 57). Thus, there appears to be little aromatization of testosterone to estradiol, at least as assessed by serum estradiol levels. Also, since T binds to SHBG more avidly than does E 2, it is conceivable that raising the circulating concentrations of androgens would displace E 2 from SHBG. However, as noted above in the transdermal T studies, there was no elevation in either the total or free E 2 levels after T administration compared with the baseline values (36, 38 42). 10 Braunstein Testosterone safety in women Vol. 88, No. 1, July 2007

11 Studies on the effects of androgens on breast cancer cell lines have shown variable responses that are dependent on the cell line used, the type and concentration of androgen tested, and the presence or absence of estrogen in the media (128, 129). As summarized by Somboonporn and Davis, most studies indicate that androgens inhibit estrogen-mediated proliferation and stimulate apoptosis of the breast cancer cells (128). Similarly, in animal models of breast cancer, androgens inhibit breast cancer growth (128, 129). Dimitrakakis and coworkers studied the effects of hormone replacement on breast epithelial proliferation in oophorectomized rhesus monkeys (130). With replacement of E 2 or E 2 þ P in quantities that raised the serum levels to those seen in cycling rhesus monkeys, the breast proliferation index rose about threefold, whereas the addition of physiological concentrations of T to E 2 reduced the stimulation by almost 50%. These investigators also showed that the administration of the pure antiandrogen flutamide to cycling rhesus monkeys for 3 months resulted in a doubling of the mammary epithelium proliferation rate. Testosterone administered to ovariectomized rhesus monkeys decreased estrogen-stimulated estrogen receptoralpha expression (120) Thus, in the primate model, androgens antagonize the effects of estrogen or estrogen þ P on the breast tissue. Multiple prospective case-control studies in women have attempted to examine the relationship between the endogenous levels of T and breast cancer. Interpretation of the results is confounded by several variables, including low numbers of cases in some studies, the use in some studies of T assays that were not optimized for the low levels of T seen in women, especially the older cohort, and, in several of the reports, inadequate adjustment for important variables, especially the independent effect of estrogen. Indeed, many studies have shown that T and E 2 track together when E 2 levels are high, so are T levels, which should not be too surprising since T serves as a prohormone for E 2 ( ). Most of the studies have shown no significant increase in odds ratios or relative risk ratios between quartiles of serum T levels and breast cancer, especially when adjusted for other variables such as age, body mass index, age of menarche, family history of breast cancer, and E 2 levels (131, 132, ), while a few have shown increased risk with increasing T levels (133, ), even when corrected for E 2 levels (133, ). Prospective cohort and case-control studies of women with polycystic ovarian disease and endogenous hyperandrogenism have not shown an increased risk of breast cancer, which provides additional evidence that elevated serum T concentrations are not a risk factor for breast cancer ( ). None of the prospective controlled clinical trials of T treatment in postmenopausal women have been adequately powered to determine whether exogenous T administration alters the risk of breast cancer. A prospective cohort study in the Nurses Health Study noted a 1.77 relative risk for breast cancer in women using estrogen þ T in comparison with never users, which was significantly greater than the risk of estrogen-only therapy (relative risk of 1.15) but not different than estrogen þ progestin users (relative risk of 1.58) (150). However, the increased risk with estrogen þ T was only significant during the first 5 years of therapy and was not significant for those using the therapy for 5 or more years. Only 2.4% of the estrogen þ T users had never used other postmenopausal hormone therapy, and the data were not analyzed to ferret out the effect of the prior hormone use; therefore, the potential for recall bias in this study was substantial. Testosterone users were also noted to be younger, leaner, and more likely to have benign breast disease and consumed more alcohol than never users, additional factors that confound the interpretation (150). Three retrospective series have attempted to evaluate the risk but suffer from low numbers of subjects or lack of appropriate controls ( ). One showed no increased risk with MT þ CEEs (151), while another indicated that there was an increased risk with injectable T and E 2 but not with injections of T, E 2, and progestin (152). The third suggested a decreased risk in women receiving estrogen þ T implants when the cases of breast cancer/100,000 women-years were compared with the reported incidence from the WHI and the Million Women study estrogen þ progestin users or even those women in that study who never used postmenopausal hormone treatment (153). Two additional studies concerning the effects of T on the breast are worth noting. Burgess and Shousha examined mastectomy specimens from 29 female-to-male transsexuals who had received long-term androgen therapy before surgery and compared the histological findings with normal female breast tissue obtained at the time of reduction mammoplasty. There were no differences noted in the prevalence of normal acini and ducts, fibrosis, cysts and apocrine metaplasia, estrogen and P receptors, or components of cyst fluid (154). The only difference was a significant increase in microcalcifications in the breasts of transsexual patients. Hofling and coworkers performed fine-needle aspirations of breast tissue before and after 6 months of continuous combined estrogen-progestin therapy with either 300 mg/day of T delivered transdermally via a patch or a placebo patch. They examined the proliferation index of the epithelial and stromal components and noted that the estrogen þ progestin therapy led to a significant increase in proliferation, which was markedly attenuated in women receiving estrogen þ progestin þ T (155). Although the recent data from the Nurses Health Study raise a concern about T and the risk of breast cancer, most of the available data do not support the concept that T treatment in women will increase the risk of breast cancer; rather, the data suggest that either there is no effect or T may actually reduce the risk of breast cancer by antagonizing the effects of estrogen on the mammary tissue. LIVER TOXICITY Elevations of transaminases, cholestatic jaundice, peliosis hepatis, and liver adenomas have been reported in femaleto-male transsexuals receiving high doses of MT orally (74, Fertility and Sterility â 11