Background
Risk-reducing salpingo-oophorectomy is an effective ovarian cancer risk reduction strategy. However, bilateral oophorectomy has also been associated with increased long-term nonneoplastic sequelae, effects suggested to be mediated through reductions in systemic sex steroid hormone levels. Currently, it is unclear whether the postmenopausal ovary contributes to the systemic hormonal milieu or whether postmenopausal ovarian volume or other factors, such as body mass index and age, affect systemic hormone levels.
Objective
We examined the impact of oophorectomy on sex steroid hormone levels in postmenopausal women. Furthermore, we explored how well ovarian volume measured by transvaginal ultrasound correlated with direct ovarian measures obtained during surgical pathology evaluation and investigated the association between hormone levels and ovarian volumes.
Study Design
Postmenopausal women who underwent risk-reducing salpingo-oophorectomy (180 cases) or ovarian cancer screening (38 controls) enrolled in an international, prospective study of risk-reducing salpingo-oophorectomy and risk of ovarian cancer algorithm–based screening among women at increased risk of ovarian cancer (Gynecologic Oncology Group-0199) were included in this analysis. Controls were frequency matched to the cases on age at menopause, age at study entry, and time interval between blood draws. Ovarian volume was calculated using measurements obtained from transvaginal ultrasound in both cases and controls and measurements recorded in surgical pathology reports from cases. Serum hormone levels of testosterone, androstenedione, androstenediol, dihydrotestosterone, androsterone, dehydroepiandrosterone, estrone, estradiol, and sex hormone–binding globulin were measured at baseline and follow-up. Spearman correlation coefficients were used to compare ovarian volumes as measured on transvaginal ultrasound and pathology examinations. Correlations between ovarian volumes by transvaginal ultrasound and measured hormone levels were examined using linear regression models. All models were adjusted for age. Paired t tests were performed to evaluate individual differences in hormone levels before and after risk-reducing salpingo-oophorectomy.
Results
Ovarian volumes measured by transvaginal ultrasound were only moderately correlated with those reported on pathology reports (Spearman rho [ρ]=0.42). The median time interval between risk-reducing salpingo-oophorectomy and follow-up for the cases was 13.3 months (range, 6.0–19.3), and the median time interval between baseline and follow-up for the controls was 12.7 months (range, 8.7–13.4). Sex steroid levels decreased with age but were not correlated with transvaginal ultrasound ovarian volume, body mass index, or time since menopause. Estradiol levels were significantly lower after risk-reducing salpingo-oophorectomy (percentage change, −61.9 post-risk-reducing salpingo-oophorectomy vs +15.2 in controls; P =.02), but no significant differences were seen for the other hormones.
Conclusion
Ovarian volumes measured by transvaginal ultrasound were moderately correlated with volumes directly measured on pathology specimens and were not correlated with sex steroid hormone levels in postmenopausal women. Estradiol was the only hormone that declined significantly after risk-reducing salpingo-oophorectomy. Thus, it remains unclear whether the limited post–risk-reducing salpingo-oophorectomy changes in sex steroid hormones among postmenopausal women impact long-term adverse outcomes.
Introduction
Specific hereditary cancer predisposition syndromes are associated with markedly increased risks of developing ovarian cancer. Estimated lifetime ovarian cancer risk associated with pathogenic variants in BRCA1 and BRCA2 ( BRCA1/2 ) are as high as 40% to 50% and 12% to 25%, respectively. , Other hereditary ovarian cancer predisposition genes include the mismatch repair genes associated with Lynch syndrome , and several other DNA repair genes. For females with a BRCA1/2 pathogenic variant, risk-reducing salpingo-oophorectomy (RRSO) reduces ovarian and fallopian tube cancer risk (by 80% to 90%) and cancer-specific and overall mortality. RRSO is considered the most effective option for ovarian cancer risk management in this setting. Bilateral oophorectomy for benign indications may induce a risk of nonneoplastic conditions that are likely mediated through reductions in systemic hormone levels. In premenopausal women, most androgens (in the inactive precursors androstenedione, dehydroepiandrosterone [DHEA], and DHEA sulfate [DHEA-S]) are produced largely in the adrenals and the remaining in the ovary. At menopause, adrenal production of DHEA and ovarian production of DHEA-S, androstenedione, testosterone, and estrogens decline dramatically. However, the contribution of the ovary to maintaining the postmenopausal hormonal milieu is uncertain. A meta-analysis of >6000 postmenopausal women from 13 prospective studies showed that sex hormone levels were lower in older women than younger women, with the largest differences in DHEA-S and androgens (especially testosterone). Moreover, postmenopausal hormones have been shown to be related to body mass index (BMI), oophorectomy, hysterectomy, cigarette smoking, and alcohol consumption. Hormone levels were lower in women with bilateral oophorectomy than naturally postmenopausal women, suggesting a role for the postmenopausal ovary in androgen production. Furthermore, free testosterone and free estradiol levels among postmenopausal women undergoing oophorectomy were lower postoperatively, , further suggesting that the postmenopausal ovary contributed to hormone production. However, an investigation of sex hormone sources in postmenopausal women suggested that androgens were produced mainly by the adrenal glands rather than the ovaries. This issue is particularly important for women contemplating RRSO, although previous reports have been inconsistent regarding the relationship between benign oophorectomy and adverse health outcomes, including increased mortality.
Why was this study conducted?
It is unclear how closely ovarian volume measured by transvaginal ultrasound (TVUS) reflects the actual volume or whether postmenopausal ovarian volume correlates with sex hormone levels.
Key findings
In postmenopausal women, ovarian volumes by TVUS were moderately correlated with volumes by pathology reports. Risk-reducing salpingo-oophorectomy (RRSO) was associated with decreased estradiol, but not other sex steroid hormone levels. Sex steroid hormone levels were not correlated with ovarian volume, body mass index, or time since menopause.
What does this add to what is known?
Ovarian volume by TVUS might not reflect the actual volume. Ovarian volume was not correlated with sex steroid hormone levels in postmenopausal women. RRSO seemed to substantially affect estradiol levels, but not the other sex steroid hormones.
Data regarding the relationships between postmenopausal ovarian volume and steroid hormone levels are limited. Previous studies suggested that postmenopausal women with large noncystic ovaries were at increased risk of breast and endometrial cancers. , Among women with endometrial cancer, larger ovarian volumes were associated with higher circulating steroid hormone levels. ,
Postmenopausal ovarian volume, as measured by transvaginal ultrasound (TVUS), in healthy women was shown to be positively associated with bone mineral density and negatively associated with sex hormone–binding globulin (SHBG). However, how well the measurements on TVUS correlate with volume measured in pathology specimens is not known.
Similar to women undergoing oophorectomy for benign indications, postmenopausal women at increased ovarian cancer risk who choose RRSO may experience adverse effects secondary to altered postoperative hormone levels. Understanding the impact of oophorectomy on sex steroid hormone levels and the determinants of these changes might be important in identifying women at risk of adverse consequences who could be targeted for mitigating interventions.
The prospective study of risk-reducing salpingo-oophorectomy and longitudinal cancer antigen 125 (CA-125) screening among women at increased risk of ovarian cancer (Gynecologic Oncologic Group Protocol 0199 [GOG-0199]) was implemented in 2003. All participants had TVUS, and serum was collected at enrollment and periodically during follow-up. The participants were elected to undergo either RRSO or ovarian cancer screening (OCS). Here, we evaluated the correlation between the TVUS volume of noncystic benign ovaries and measurements of circulating androgens and estrogens in postmenopausal women and the effect of oophorectomy on circulating hormone levels. Moreover, we sought other possible independent determinants of hormone levels, including age and BMI. In addition, this unique patient population allowed comparing volumes measured on preoperative TVUS with the gross descriptions in corresponding surgical pathology reports.
Materials and Methods
Study population
GOG-0199 was a multi-institution, prospective cohort study. Detailed eligibility criteria have been published. Eligible women had increased breast and/or ovarian cancer risk based on personal and/or family cancer history, age ≥30 years, no previous history of ovarian, fallopian tube, or peritoneal cancer, and at least 1 intact ovary. At enrollment, the participants chose either RRSO or OCS. The GOG-0199 screening algorithm included CA-125 measurements and the risk of ovarian cancer algorithm (ROCA) score calculations every 3 months, in addition to an annual TVUS. The participants in the RRSO cohort underwent surgery within 90 days of enrollment and had CA-125 measurements and ROCA score calculations every 6 months for the 5 years of follow-up. The participants in the OCS cohort had the option to crossover to the RRSO cohort after enrollment, either electively or as prompted by screening results or clinical findings.
The analyses reported here used a subset of GOG-0199 participants. Individuals were included as cases if they (1) were postmenopausal at RRSO, (2) had TVUS at baseline, (3) had baseline serum available, (4) had RRSO during study participation (either at baseline or during crossover), (5) were not using exogenous hormones at baseline or after RRSO, (6) had at least 1 blood draw within 12 months after RRSO, and (7) did not have a malignancy (ovarian, fallopian tube, or peritoneal) diagnosed at the time of RRSO. A smaller group of controls was included to evaluate the possibility that the declines in hormone levels after RRSO were a function of aging and time since menopause rather than a direct effect of oophorectomy. The eligibility criteria for the control group were the same as the cases except for the RRSO requirement. The control group was frequency matched to the cases on age at menopause (<44, 45–49, 50–54, 55–59, 60–64, and >64), age at baseline (<44, 45–49, 50–54, 55–59, 60–64, and>64), and time interval between blood draws (<12 months and ≥12 months).
All participants signed a written informed consent for GOG-0199 (NCI Protocol 02-C-0268; ClinicalTrials.gov Identifier: NCT-00043472). The GOG-0199 protocol was approved by institutional review boards at the National Cancer Institute, GOG, and 151 participating GOG institutions (the United States and Australia).
Hormone measurements
Serum hormone levels were measured at baseline for both the RRSO and control groups. Furthermore, the measurements were obtained approximately 12 months after surgery for the RRSO group and at a matched postenrollment interval blood draw for the controls. A panel of 8 serum hormones (testosterone, androstenedione, androstenediol, dihydrotestosterone, androsterone, DHEA, estrone, and estradiol) were assayed using gas chromatography-mass spectrometry (GC-MS). SHBG was quantified by radioimmunoassay using the immunoradiometric assay count SHBG kit (Siemens Healthcare Diagnostics Inc, Mississauga, Ontario, Canada). All hormone measurements were done in the laboratory of Dr Chantal Guillemette, at Laval University, Quebec City, Canada ( Supplemental Data ).
Samples with levels measured below the lower limits of quantification (LLOQ) were assigned a value equal to half the LLOQ or the actual measured value, whichever was higher.
Ovarian volume measurements
Volume by transvaginal ultrasound
For both the RRSO and control groups, ovarian volume was estimated using the prolate ellipsoid formula (length×height×width×0.523) for each ovary, using dimensions recorded on baseline TVUS. If both ovaries were present, volumes were added to determine total ovarian volume. For participants in the RRSO group with missing measurements on TVUS on one or both ovaries for whom measurements from pathology reports were available, the volume of the ovary with missing values was imputed using measurements from pathology reports and the pathology-to-TVUS volume ratio from all participants with complete data. For participants with only 1 ovary, the volume measured by either TVUS or imputed was used as the total volume. For participants in the control group with missing TVUS data, measures from TVUS at another time point were used. Ovaries with hypoechoic, hemorrhagic, complex, or septate cysts reported on TVUS (class 2 or 3 cysts) were excluded from this analysis (n=14).
Volume by pathology
Ovarian volume was estimated using the prolate ellipsoid formula (length×height×width×0.523), using the ovary dimensions recorded in the gross surgical pathology report.
Statistical analysis
Spearman correlation coefficients were used to compare ovarian volumes as measured by TVUS and pathology examination. The Fisher exact test was used to assess differences between RRSO and control patients according to baseline characteristics, including sociodemographic and reproductive factors, personal and family history, history of chemopreventive agents or hormonal therapies, and use of alcohol and cigarette. The Kruskal-Wallis test was used to assess differences for continuous variables, including age at enrollment, BMI, age at menarche, age at menopause, and years since menopause.
Linear regression models were used to assess the relationship between ovarian volume by TVUS (in quartiles and as a continuous variable) and hormones. Analysis of variance models were used to compare hormone concentrations in women before and after RRSO and over time in women who did not undergo surgery. The models were adjusted for age at first blood draw (continuous), BMI (continuous), previous hormone use (ever or never), time since menopause (in years), hysterectomy (yes or no), and personal and family history of cancer. Finally, analyses were stratified by age group, time since menopause, and BMI (normal, overweight, or obese).
Paired t tests were performed to evaluate changes in hormone levels before and after RRSO and for controls to evaluate hormone changes from baseline to follow-up. The absolute change in hormone levels for each individual was calculated; differences between RRSO participants and controls were explored to assess whether hormone changes after RRSO were, in part, because of age-related hormonal declines rather than oophorectomy.
Results
A total of 225 participants undergoing RRSO were eligible for inclusion in this study, and 50 controls were selected from the eligible nonsurgical participants. Of these, 45 participants in the RRSO group and 12 participants in the control group were excluded because of oral contraceptive use at enrollment; anastrozole, tamoxifen, or raloxifene use within 30 days of enrollment; or inadequate ovarian volume information ( Figure 1 ). Mean ages at baseline hormone measurement were similar between RRSO and controls (54.2 vs 56.4 years; P =.12). Participants in the RRSO group were younger at menopause (46.3 vs 49.1 years; P =.006) and more likely to report previous breast cancer (60% vs 42.1%; P =.049). BMI, age at menarche, parity, and use of tamoxifen and raloxifene were similar between cases and controls ( Table 1 ). The median time interval between RRSO and follow-up for the cases was 13. 3 months (range, 6.0–19.3 months), and the median time interval between baseline and follow-up for the controls was 12.7 months (range, 8.7–13.4 months).
| Patient characteristic | RRSO (n=180) | Controls (n=38) | P value |
|---|---|---|---|
| Age at enrollment (y) | 54.2±6.2 | 56.3±7.0 | .12 |
| Race | |||
| White | 173 (96.1) | 37 (97.4) | 1.0 |
| Black | 6 (3.3) | 1 (2.6) | |
| Other or not specified | 1 (0.6) | — | |
| BMI (kg/m 2 ) | 28.0±6.7 | 26.7±6.1 | .23 |
| Age at menarche (y) | 12.5±1.5 | 12.4±1.6 | .87 |
| Parity | |||
| 0 | 30 (18.2) | 6 (16.7) | .70 |
| 1 | 32 (19.4) | 9 (25.0) | |
| 2 | 58 (35.2) | 9 (25.0) | |
| 3 | 30 (18.2) | 9 (25.0) | |
| ≥4 | 15 (9.1) | 3 (8.3) | |
| History of contraceptive use | |||
| Yes | 142 (78.9) | 27 (71.1) | .29 |
| No | 38 (21.1) | 11 (28.9) | |
| Menopausal status | |||
| Age at menopause (y) | 46.3±6.9 | 49.1±6.5 | .006 |
| Time since menopausal in years | 8.0±6.1 | 7.2±6.8 | .22 |
| Postmenopausal hormone use | |||
| No | 80 (44.4) | 17 (44.7) | 1.000 |
| Yes | 100 (55.6) | 21 (55.3) | |
| History of tamoxifen use | .655 | ||
| Yes | 37 (20.6) | 6 (15.8) | |
| No | 143 (79.4) | 32 (84.2) | |
| History of raloxifene use | |||
| Yes | 9 (5.0) | 0 (0) | .365 |
| No | 171 (95.0) | 38 (100.0) | |
| Tubal ligation before enrollment | .851 | ||
| No | 116 (65.5) | 26 (68.4) | |
| Yes | 61 (34.5) | 12 (31.6) | |
| Hysterectomy before enrollment | .636 | ||
| No | 146 (82.0) | 33 (86.8) | |
| Yes | 32 (18.0) | 5 (13.2) | |
| Hysterectomy at time of RRSO | 67 (37.2) | — | |
| History of breast cancer | .049 | ||
| No | 72 (40.0) | 22 (57.9) | |
| Yes | 108 (60.0) | 16 (42.1) |
Correlation of ovarian volume measured by transvaginal ultrasound and in pathology reports
A total of 12 participants had unilateral oophorectomy before enrollment; 1 participant in the RRSO group had unilateral oophorectomy and part of the contralateral ovary resected previously. Ovarian volume from TVUS was available for 141 of 180 participants (78%) in the RRSO group and 35 of 38 participants (92%) in the nonsurgical group. RRSO pathology reports provided ovarian volume for 164 of 180 women (91%). The proportions of participants with hypoechoic, hemorrhagic, complex, or septated cysts (class 2 or 3) reported on TVUS, measured mean ovarian volumes, and mean ovarian volumes based on measured plus imputed values were similar between the RRSO and control groups ( Table 2 ). The Spearman correlation between TVUS-measured and pathology-measured volumes was 0.42 ( Figure 2 ).
| Variable | RRSO (n=180) | Controls (n=38) | P value |
|---|---|---|---|
| Number of ovaries | .75 | ||
| 1 a | 12 (6.7) | 1 (2.6) | |
| 2 | 168 (93.3) | 37 (97.4) | |
| Ovary cysts | .715 | ||
| Class 0 or 1 | 169 (93.9) | 35 (92.1) | |
| Class 2 or 3 b | 11 (6.1) | 3 (7.9) | |
| Measurements by TVUS | 141 | 35 | .642 |
| Mean ovarian volume in cubic centimeter | 5.9±11.6 | 4.0±3.7 | |
| Imputed measurements by TVUS c | 180 | 38 | .708 |
| Mean ovarian volume in cubic centimeter | 5.7±10.3 | 4.1±3.5 | |
| Measurements by pathology records | 164 | — | — |
| Mean ovarian volume in cubic centimeter | 5.9±8.6 | — |
a One participant had unilateral oophorectomy and part of the second ovary resected before study enrollment
b Excluded from ovarian volume estimates by TVUS and pathology records
Correlations between hormone levels and changes related to surgery and other factors
Most hormone levels were above the LLOQ, except estrone and estradiol, for which baseline values were undetected in 7.2% and 12.5% of participants in the RRSO group ( Supplemental Table 1 ). Correlations between the hormones are shown in Table 3 . Except for androstenediol and DHEA (ρ=0.84) and estrone and estradiol (ρ=0.83), the correlations were modest.
| Variable | Testosterone | Androstenedione | Androstenediol | Dihydrotestosterone | Androsterone | Dehydroepiandrosterone | Estrone | Estradiol | SHBG |
|---|---|---|---|---|---|---|---|---|---|
| Testosterone | 1.00 | 0.51 a | 0.52 a | 0.56 a | 0.46 a | 0.44 a | 0.36 a | 0.26 a | 0.28 a |
| Androstenedione | 1.00 | 0.53 a | 0.5 a | 0.53 a | 0.64 a | 0.48 a | 0.36 a | 0.16 b | |
| Androstenediol | 1.00 | 0.68 a | 0.73 a | 0.84 a | 0.33 a | 0.29 a | 0.2 a | ||
| Dihydrotestosterone | 1.00 | 0.7 a | 0.62 a | 0.28 a | 0.22 a | 0.52 a | |||
| Androsterone | 1.00 | 0.71 a | 0.3 a | 0.27 a | 0.09 a | ||||
| Dehydroepiandrosterone | 1.00 | 0.24 a | 0.17 a | 0.13 c | |||||
| Estrone | 1.00 | 0.83 a | −0.01 | ||||||
| Estradiol | 1.00 | −0.13 | |||||||
| SHBG | 1.00 |
Table 4 shows the percentage changes in pre- and postsurgery hormones for the RRSO group and in baseline to follow-up for nonsurgical controls. Overall, the patterns of decline in steroid hormones before and after surgery (RRSO) and from baseline to follow-up (controls) were similar.
| Variables Hormone (concentration) | RRSO (n=180) | Controls (n=38) | Effect of RRSO P value b | ||||
|---|---|---|---|---|---|---|---|
| Baseline | Post-RRSO | % change a | Baseline | Follow-up | % change a | ||
| Testosterone (ng/mL) | 0.23 | 0.17 | −26.1 | 0.19 | 0.17 | −10.5 | .92 |
| Androstenedione (ng/mL) | 0.57 | 0.49 | −14.0 | 0.51 | 0.48 | −5.9 | .98 |
| Androstenediol (pg/mL) | 291.40 | 278.90 | −4.3 | 273.80 | 240.80 | −12.1 | .72 |
| Dihydrotestosterone (pg/mL) | 39.00 | 37.30 | −4.4 | 40.90 | 37.20 | −9.0 | .84 |
| Androsterone (pg/mL) | 128.30 | 128.40 | 0.1 | 127.80 | 121.80 | −4.7 | .73 |
| Dehydroepiandrosterone (ng/mL) | 2.23 | 2.28 | 2.2 | 2.21 | 1.86 | −15.8 | .28 |
| Estrone (pg/mL) | 30.00 | 28.20 | −6.0 | 23.10 | 24.90 | 7.8 | .31 |
| Estradiol (pg/mL) | 14.70 | 5.60 | −61.9 | 9.20 | 10.60 | 15.2 | .02 |
| SHBG (nmol/L) | 81.00 | 82.50 | 1.9 | 93.70 | 89.40 | −4.6 | .29 |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
