Invasive epithelial ovarian cancer (EOC) is the most lethal gynecologic malignancy. The etiology of EOC remains elusive; however, experimental and epidemiologic data suggest a role for hormone-related exposures in ovarian carcinogenesis and risk factor differences by histologic phenotypes and developmental pathways. Research on prediagnosis androgen concentrations and EOC risk has yielded inconclusive results, and analyses incorporating EOC subtypes are sparse. We conducted a pooled analysis of 7 nested case–control studies in the Ovarian Cancer Cohort Consortium to investigate the association between pre-diagnosis circulating androgens [testosterone, free testosterone, androstenedione, dehydroepiandrosterone sulfate (DHEAS)], sex hormone binding globulin (SHBG), and EOC risk by tumor characteristics (i.e., histology, grade, and stage). The final study population included 1,331 EOC cases and 3,017 matched controls. Multivariable conditional logistic regression was used to assess risk associations in pooled individual data. Testosterone was positively associated with EOC risk (all subtypes combined, ORlog2 = 1.12; 95% confidence interval 1.02–1.24); other endogenous androgens and SHBG were not associated with overall risk. Higher concentrations of testosterone and androstenedione associated with an increased risk in endometrioid and mucinous tumors [e.g., testosterone, endometrioid tumors, ORlog2 = 1.40 (1.03–1.91)], but not serous or clear cell. An inverse association was observed between androstenedione and high grade serous tumors [ORlog2 = 0.76 (0.60–0.96)]. Our analyses provide further evidence for a role of hormone-related pathways in EOC risk, with differences in associations between androgens and histologic subtypes of EOC. Cancer Res; 77(14); 3951–60. ©2017 AACR.

Reproductive history influences risk of ovarian cancer and it has been hypothesized that these associations are mediated by exposure to endogenous hormones, including androgens (1). Data from experimental studies link androgen-related signaling to ovarian cancer through increased cellular proliferation and reduced apoptotic rates (2–4). The relationship between androgens and epithelial ovarian cancer (EOC) risk has been examined in 7 nested case-control studies with the numbers of cases in these studies ranging from 31 to 1,052 (5–10); these studies predominantly investigated EOC as a composite outcome. Emerging data show heterogeneity in risk factors by histologic subtypes (e.g., serous, endometrioid, mucinous, clear cell) and by the hypothesized “dualistic pathway” of ovarian carcinogenesis (defined by differences in the genetic make-up and the morphological architecture of histologic phenotypes; refs. 11–18). The relationship between androgens and EOC risk by disease subtype has been minimally explored. Analyses to date suggest heterogeneity by subtype (9, 10); however, individual studies evaluating EOC by subtype were either limited by small case numbers in subtype analyses (9), or restricted to women pregnant at the time of serum sampling (10).

We pooled and harmonized available data from 6 nested case–control studies within the Ovarian Cancer Cohort Consortium (OC3), plus the Finnish Maternity Cohort (FMC), to investigate the relationship of pre-diagnosis concentrations of androgens [e.g., testosterone, free testosterone, androstenedione, dehydroepiandrosterone-sulfate (DHEAS)] and sex-hormone binding globulin (SHBG) with EOC risk, overall and by subtype. Subtype analyses included analyses by histology, grade and stage, and by the hypothesized dualistic model of EOC development, that is, type I versus type II (19). Our study represents the largest investigation to date including individual-level data from 1,331 EOC cases and 3,017 matched controls, with 61 (clear cell) to 667 (serous) cases represented in the major histologic subtypes.

Study population: ovarian cancer cohort consortium (OC3)

The OC3 has been described previously (12). For this investigation, eligible cohorts were required to have data on a defined set of a priori selected covariates (e.g., menopausal status at blood donation, oral contraceptive use at blood donation, parity) and prediagnosis measurements of testosterone, free testosterone, androstenedione, or DHEAS. In addition to the OC3 cohorts, the FMC, a cohort of women pregnant at blood collection, contributed data to this investigation (for contributing cohorts see Supplementary Table S1). Available biomarker and questionnaire data from each cohort were centrally collated and harmonized at the Data Coordinating Center at the Brigham and Women's Hospital.

Case characteristics

Eligible cases included women diagnosed with invasive EOC [International Classification of Disease Codes (ICD): ICD9 codes 183 and 158; ICD10 code C56] ascertained by self-report with medical record confirmation and/or linkage to cancer registries. Cases were individually matched to two or three controls (free of cancer and alive at the time of diagnosis of the index case) on age, date, menopausal status and day or phase of menstrual cycle at blood collection in premenopausal women, with the exception of the FMC (matched on age and date at blood collection, parity at blood collection and at diagnosis/index date). Histomorphologic data were complete, and the majority of cases had data on stage (82%); grade was available for 36% of the cases. We used histology and grade to classify tumors into type I [(48%, n = 291); low-grade serous and endometrioid, all mucinous and clear cell] and type II [(52%, n = 314); high-grade serous, high-grade endometrioid; ref. 19]. Serous and endometrioid cases missing grade data were excluded from these analyses; mucinous and clear cell tumors were included regardless of grade data availability, as these tumors are classified as type I independent of grade. In a sensitivity analysis, all mucinous and clear cell cases missing grade were excluded from the type I subgroup (after exclusion, case n = 77). The proportion of type I tumors was higher than expected; however, we observed the expected distribution (type I: 28% vs. type II: 72%) after excluding women from the FMC (all missing grade; younger at diagnosis and more frequently diagnosed with mucinous tumors than cases from the other cohorts).

Laboratory methods

In all studies, case–control sets were measured in the same batch and technicians performing the assays were blinded to case–control status and quality control samples. Information on sample type, laboratory assays, and intra- and inter-batch coefficients of variations for each cohort is summarized in Supplementary Table S2. Free testosterone was calculated on the basis of measured concentrations of testosterone and SHBG, with albumin assumed to be a constant 40 g/L, according to the mass law of action (20).

Statistical analyses

Hormone measurements were standardized across studies based on the cohort-specific mean concentrations in controls (see Supplementary Methods; Supplementary Table S3). Conditional logistic regression was used to estimate ORs and 95% confidence intervals (CI). ORs were estimated using log2-transformed biomarker concentrations and study-specific tertiles based on the distribution in controls. A continuous probit score, generating a rank for each participant in each cohort by hormone concentration, was used to test for trend across tertiles. We additionally evaluated associations in quintiles for EOC overall and the serous subtype. Multivariable models included: parity [never, ever, missing (2.8%)] and OC use [never, ever, missing (47%); excluding FMC 2.3% missing]. Additional adjustment for body mass index (BMI; kg/m2) among women with data available (n = 747 cases), did not change the ORs (data not shown).

Statistical analyses were conducted using a two-stage approach: First, ORs were calculated within each cohort and pooled using DerSimonian and Laird random effects meta-analysis models to assess between-study heterogeneity (21). Second, ORs were calculated on the basis of pooled individual participant data (22). ORs estimated from meta-analysis and the data pooling method were similar, and we observed no significant between-study heterogeneity. Therefore, presented results are based on the pooled analysis. The assumption of linearity was tested using restricted cubic splines; no significant deviations from linearity were observed. Statistical heterogeneity of associations across subtypes was assessed via a likelihood ratio test comparing a model allowing the association for the risk factor of interest to vary by subtype versus one assuming the same association across subtype using polytomous conditional logistic regression (23).

We evaluated associations after stratification by menopausal status at blood collection (premenopausal vs. postmenopausal) and age at diagnosis (<55 vs. ≥55 years). Androgen concentrations are relatively stable in pregnancy (24); however, we excluded FMC members in sensitivity analyses given that all women were pregnant at the time they provided a blood sample. Finally, we conducted a sensitivity analysis after exclusion of women diagnosed within two years after blood donation. A more detailed description of statistical procedures is available in the Supplementary Methods.

SAS Statistical Software, version 9.3 (SAS Institute) was used for statistical analyses. P values <0.05 were considered as statistically significant; all statistical tests and P values were two-sided.

In total 1,331 cases and 3,017 matched controls from 7 cohorts were included in this investigation (Table 1). Average age at blood collection ranged from 32 (FMC) to 61 years (CLUE II), and the majority of women were parous (89% cases, 94% controls; Table 1). Average age at diagnosis ranged from 45 (FMC) to 67 (CLUE II; Supplementary Table S4).

Table 1.

Case and control characteristics in pooled analysis of prospective data on circulating androgens, SHBG and EOC risk: the Ovarian Cancer Cohort Consortium (OC3)

CohortReferenceNoMean age at blood donation years (SD)Nulliparous%aEver OC use, %aPostmenopausal, %bMean BMI (SD)
Clue IIc  Case 46 60.8 (13.0) 19% 20% 85% 26.3 (5.8) 
  Control 91 61.0 (12.9) 13% 13% 86% 25.4 (4.6) 
EPIC Ose et al. 2015 (9) Case 451 55.9 (8.5) 18% 37% 77% 26.8 (4.9) 
  Control 867 55.9 (8.6) 12% 45% 77% 26.3 (4.7) 
FMC Schock et al. 2014 (10) Case 576 32.5 (4.8) 0% d 0% d 
  Control 1,433 32.5 (4.7) 0% d 0% d 
NHS Tworoger et al. 2008 (7) Case 117 57.7 (6.5) 8% 41% 79% 24.8 (4.8) 
  Control 348 57.7 (6.5) 5% 47% 79% 24.7 (4.1) 
NHS II Tworoger et al. 2008 (7) Case 15 46.1 (4.4) 20% 93% 20% 29.6 (9.8) 
  Control 44 45.8 (4.3) 23% 86% 18% 25.9 (5.8) 
NYUWHS Lukanova et al. 2003 (6) Case 63 52.6 (8.6) 47% 29% 56% 24.5 (3.8) 
  Control 112 52.0 (8.5) 38% 36% 54% 25.9 (4.3) 
WHS Tworoger et al. 2008 (7) Case 63 55.7 (7.2) 25% 65% 75% 24.5 (3.9) 
  Control 122 55.5 (7.0) 15% 71% 70% 25.1 (4.4) 
Total  Case 1,331 45.8 (13.7) 11% 40% 42% 26.2 (5.1) 
  Control 3,017 44.8 (13.7) 6% 47% 39% 25.8 (4.6) 
CohortReferenceNoMean age at blood donation years (SD)Nulliparous%aEver OC use, %aPostmenopausal, %bMean BMI (SD)
Clue IIc  Case 46 60.8 (13.0) 19% 20% 85% 26.3 (5.8) 
  Control 91 61.0 (12.9) 13% 13% 86% 25.4 (4.6) 
EPIC Ose et al. 2015 (9) Case 451 55.9 (8.5) 18% 37% 77% 26.8 (4.9) 
  Control 867 55.9 (8.6) 12% 45% 77% 26.3 (4.7) 
FMC Schock et al. 2014 (10) Case 576 32.5 (4.8) 0% d 0% d 
  Control 1,433 32.5 (4.7) 0% d 0% d 
NHS Tworoger et al. 2008 (7) Case 117 57.7 (6.5) 8% 41% 79% 24.8 (4.8) 
  Control 348 57.7 (6.5) 5% 47% 79% 24.7 (4.1) 
NHS II Tworoger et al. 2008 (7) Case 15 46.1 (4.4) 20% 93% 20% 29.6 (9.8) 
  Control 44 45.8 (4.3) 23% 86% 18% 25.9 (5.8) 
NYUWHS Lukanova et al. 2003 (6) Case 63 52.6 (8.6) 47% 29% 56% 24.5 (3.8) 
  Control 112 52.0 (8.5) 38% 36% 54% 25.9 (4.3) 
WHS Tworoger et al. 2008 (7) Case 63 55.7 (7.2) 25% 65% 75% 24.5 (3.9) 
  Control 122 55.5 (7.0) 15% 71% 70% 25.1 (4.4) 
Total  Case 1,331 45.8 (13.7) 11% 40% 42% 26.2 (5.1) 
  Control 3,017 44.8 (13.7) 6% 47% 39% 25.8 (4.6) 

Abbreviations: BMI, body mass index; CLUE, Washington County, MD Study “Give us a clue to cancer and heart disease”; EPIC, European Prospective Investigation into Cancer and Nutrition; FMC, Finnish Maternity Cohort; NHS, Nurses' Health Study; NYUWHS, New York University Women's Health Study; OC, oral contraceptive; OC3, Ovarian Cancer Cohort Consortium; SHBG, sex hormone binding globulin; WHS, Women's Health Study.

aAmong women with data: parity 2.8% missing; OC use 47% missing (excluding FMC: 2.3% missing).

bAt blood collection.

cData from Clue II have not been published.

dInformation on BMI and OC use was not collected in the FMC.

Androgens and overall EOC risk

A doubling of testosterone (i.e., 1-unit increase in log2-transformed testosterone) was associated with a 12% increase in overall EOC risk (ORlog2, 1.12; 95% confidence interval (CI), 1.02–1.24], and a 25% increase in risk comparing top to bottom tertile (ORT3-T1 = 1.25; 95% CI, 1.06–1.48; Ptrend = 0.03; Table 2). Free testosterone, androstenedione, DHEAS, and SHBG were not associated with overall risk of EOC. Results from analyses evaluating quintiles of androgen and SHBG concentrations were similar to those from models using tertiles (Supplementary Table S5); however, the OR comparing highest vs. lowest quintile of testosterone was not statistically significant (ORQ5-Q1, 1.22; 95% CI, 0.99–1.52).

Table 2.

ORs (95% CI) for EOC overall and by histologic subtypes in tertiles and for doubling of androgen concentrations: The Ovarian Cancer Cohort Consortium (OC3)a

Invasive EOCSerousEndometrioidMucinousClear cell
SetsOR (95%CI)PtrendSetsOR (95%CI)PtrendSetsOR (95%CI)PtrendSetsOR (95%CI)PtrendSetsOR (95%CI)Ptrend
Testosterone 
 T1 398 ref  222 ref  35 ref  45 ref  15 ref  
 T2 443 1.20 (1.02–1.41)  229 1.16 (0.92–1.46)  60 1.46 (0.88–2.42)  61 1.34 (0.86–2.08)  27 1.65 (0.73–3.73)  
 T3b 460 1.25 (1.06–1.48) 0.03 204 0.97 (0.76–1.24) 0.56 69 1.80 (1.08–3.01) 0.06 84 1.94 (1.25–3.02) 0.05 17 0.82 (0.34–2.00) 0.68 
 Doublingc 1,301 1.12 (1.02–1.24) 0.02 655 0.96 (0.84–1.11) 0.61 164 1.40 (1.03–1.91) 0.03 190 1.29 (1.01–1.66) 0.04 59 1.12 (0.69–1.80) 0.65 
Phetd               0.06 
Free testosterone 
 T1 286 ref  155 ref  25 ref  35 ref  11 ref  
 T2 287 1.05 (0.85–1.28)  151 1.04 (0.79–1.38)  32 0.94 (0.48–1.86)  48 1.46 (0.85–2.52)  10 0.82 (0.27–2.48)  
 T3b 292 1.06 (0.87–1.31) 0.04 129 0.83 (0.62–1.10) 0.63 36 1.03 (0.53–1.99) 0.45 50 1.50 (0.88–2.54) 0.03 20 2.02 (0.67–6.12) 0.26 
 Doublingc 865 1.10 (1.00–1.21) 0.05 435 0.97 (0.84–1.11) 0.61 93 1.11 (0.80–1.53) 0.53 133 1.33 (1.04–1.72) 0.03 41 1.32 (0.82–2.11) 0.26 
Phetd               0.12 
Androstenedione 
 T1 450 ref  235 ref  46 ref  56 ref  21 ref  
 T2 387 0.86 (0.73–1.02)  204 0.88 (0.70–1.12)  45 0.67 (0.40–1.13)  51 0.90 (0.57–1.42)  15 0.93 (0.41–2.10)  
 T3b 470 1.07 (0.90–1.28) 0.13 217 0.99 (0.77–1.28) 0.79 73 0.99 (0.59–1.67) 0.82 84 1.57 (1.01–2.42) 0.03 24 0.81 (0.37–1.77) 0.62 
 Doublingc 1,307 1.08 (0.97–1.19) 0.16 656 0.97 (0.84–1.12) 0.69 164 1.04 (0.76–1.43) 0.82 191 1.33 (1.03–1.72) 0.03 60 1.07 (0.69–1.66) 0.77 
Phetd               0.17 
DHEAS 
 T1 227 ref  128 ref  18 ref  ref  ref  
 T2 245 1.08 (0.86–1.36)  127 1.06 (0.77–1.45)  23 0.81 (0.36–1.82)  16 1.04 (0.35–3.08)  12 3.36 (0.88–12.8)  
 T3b 219 0.95 (0.74–1.23) 0.87 111 0.83 (0.59–1.18) 0.45 28 1.02 (0.42–2.47) 0.85 20 1.53 (0.50–4.71) 0.29 14 3.50 (1.03–11.9) 0.10 
 Doublingc 691 0.99 (0.89–1.10) 0.82 366 0.93 (0.81–1.08) 0.36 69 1.05 (0.72–1.53) 0.78 44 1.34 (0.81–2.23) 0.26 33 1.52 (0.87–2.65) 0.14 
Phetd               0.22 
SHBG 
 T1 311 ref  147 ref  30 ref  56 ref  19 ref  
 T2 250 0.82 (0.67–1.00)  141 0.89 (0.67–1.18)  27 0.98 (0.49–1.93)  28 0.61 (0.36–1.05)  12 0.75 (0.30–1.91)  
 T3b 325 1.09 (0.89–1.33) 0.56 157 1.14 (0.86–1.52) 0.39 37 1.49 (0.78–2.85) 0.44 51 0.96 (0.58–1.57) 0.79 12 0.75 (0.29–1.97) 0.50 
 Doublingc 886 1.02 (0.91–1.14) 0.76 445 1.06 (0.91–1.25) 0.43 94 1.16 (0.80–1.67) 0.43 135 0.93 (0.68–1.28) 0.65 43 0.77 (0.46–1.30) 0.33 
Phetd               0.43 
Invasive EOCSerousEndometrioidMucinousClear cell
SetsOR (95%CI)PtrendSetsOR (95%CI)PtrendSetsOR (95%CI)PtrendSetsOR (95%CI)PtrendSetsOR (95%CI)Ptrend
Testosterone 
 T1 398 ref  222 ref  35 ref  45 ref  15 ref  
 T2 443 1.20 (1.02–1.41)  229 1.16 (0.92–1.46)  60 1.46 (0.88–2.42)  61 1.34 (0.86–2.08)  27 1.65 (0.73–3.73)  
 T3b 460 1.25 (1.06–1.48) 0.03 204 0.97 (0.76–1.24) 0.56 69 1.80 (1.08–3.01) 0.06 84 1.94 (1.25–3.02) 0.05 17 0.82 (0.34–2.00) 0.68 
 Doublingc 1,301 1.12 (1.02–1.24) 0.02 655 0.96 (0.84–1.11) 0.61 164 1.40 (1.03–1.91) 0.03 190 1.29 (1.01–1.66) 0.04 59 1.12 (0.69–1.80) 0.65 
Phetd               0.06 
Free testosterone 
 T1 286 ref  155 ref  25 ref  35 ref  11 ref  
 T2 287 1.05 (0.85–1.28)  151 1.04 (0.79–1.38)  32 0.94 (0.48–1.86)  48 1.46 (0.85–2.52)  10 0.82 (0.27–2.48)  
 T3b 292 1.06 (0.87–1.31) 0.04 129 0.83 (0.62–1.10) 0.63 36 1.03 (0.53–1.99) 0.45 50 1.50 (0.88–2.54) 0.03 20 2.02 (0.67–6.12) 0.26 
 Doublingc 865 1.10 (1.00–1.21) 0.05 435 0.97 (0.84–1.11) 0.61 93 1.11 (0.80–1.53) 0.53 133 1.33 (1.04–1.72) 0.03 41 1.32 (0.82–2.11) 0.26 
Phetd               0.12 
Androstenedione 
 T1 450 ref  235 ref  46 ref  56 ref  21 ref  
 T2 387 0.86 (0.73–1.02)  204 0.88 (0.70–1.12)  45 0.67 (0.40–1.13)  51 0.90 (0.57–1.42)  15 0.93 (0.41–2.10)  
 T3b 470 1.07 (0.90–1.28) 0.13 217 0.99 (0.77–1.28) 0.79 73 0.99 (0.59–1.67) 0.82 84 1.57 (1.01–2.42) 0.03 24 0.81 (0.37–1.77) 0.62 
 Doublingc 1,307 1.08 (0.97–1.19) 0.16 656 0.97 (0.84–1.12) 0.69 164 1.04 (0.76–1.43) 0.82 191 1.33 (1.03–1.72) 0.03 60 1.07 (0.69–1.66) 0.77 
Phetd               0.17 
DHEAS 
 T1 227 ref  128 ref  18 ref  ref  ref  
 T2 245 1.08 (0.86–1.36)  127 1.06 (0.77–1.45)  23 0.81 (0.36–1.82)  16 1.04 (0.35–3.08)  12 3.36 (0.88–12.8)  
 T3b 219 0.95 (0.74–1.23) 0.87 111 0.83 (0.59–1.18) 0.45 28 1.02 (0.42–2.47) 0.85 20 1.53 (0.50–4.71) 0.29 14 3.50 (1.03–11.9) 0.10 
 Doublingc 691 0.99 (0.89–1.10) 0.82 366 0.93 (0.81–1.08) 0.36 69 1.05 (0.72–1.53) 0.78 44 1.34 (0.81–2.23) 0.26 33 1.52 (0.87–2.65) 0.14 
Phetd               0.22 
SHBG 
 T1 311 ref  147 ref  30 ref  56 ref  19 ref  
 T2 250 0.82 (0.67–1.00)  141 0.89 (0.67–1.18)  27 0.98 (0.49–1.93)  28 0.61 (0.36–1.05)  12 0.75 (0.30–1.91)  
 T3b 325 1.09 (0.89–1.33) 0.56 157 1.14 (0.86–1.52) 0.39 37 1.49 (0.78–2.85) 0.44 51 0.96 (0.58–1.57) 0.79 12 0.75 (0.29–1.97) 0.50 
 Doublingc 886 1.02 (0.91–1.14) 0.76 445 1.06 (0.91–1.25) 0.43 94 1.16 (0.80–1.67) 0.43 135 0.93 (0.68–1.28) 0.65 43 0.77 (0.46–1.30) 0.33 
Phetd               0.43 

NOTE: Bold numbers indicate statistically significant association.

Abbreviations: DHEAS, dehydroepiandrosterone sulfate; SHBG, sex hormone binding globulin.

aResults were derived from conditional logistic regression models, additionally adjusted for OC use (never/ever/missing) and parity (never/ever/missing).

bThe P value for trend across tertiles is based on acontinuous probit score (generating a rank for each person in each cohort by hormone level).

cLinear trends for doubling of hormone concentrations were estimated on log2 scale.

dPairwise heterogeneity tests were performed, using the likelihood ratio test comparing models assuming (i) the same association between exposure and outcomes compared with (ii) a model assuming different associations for each subtype.

Histologic subtypes

The association between testosterone and EOC risk differed by histologic subtype (Phet = 0.06). Higher concentrations of circulating testosterone were associated with increased risk of endometrioid and mucinous tumors (e.g., endometrioid tumors: ORlog2, 1.40; 95% CI, 1.03–1.91), but not with serous or clear cell tumors (e.g., serous tumors: ORlog2, 0.96; 95% CI, 0.84–1.11). Free testosterone and androstenedione were associated with increased risk of mucinous tumors (e.g., androstenedione: ORlog2, 1.33; 95% CI, 1.03–1.72; Table 2), but not with any of the other histologic subtypes (e.g., androstenedione and endometrioid tumors: ORlog2, 1.04; 95% CI, 0.76–1.43). DHEAS and SHBG were not associated with any of the examined histologic subtypes.

Tumor grade and developmental pathways

We observed significant heterogeneity in the association between androstenedione and low grade EOC and high-grade serous disease; androstenedione was significantly inversely associated with high grade serous EOC (Phet = 0.02; all low-grade cases: ORlog2, 1.41; 95% CI, 0.86–2.31]; high-grade serous ORlog2, 0.76; 95% CI, 0.60–0.96; Table 3). The association between SHBG and EOC risk differed significantly by grade (Phet = 0.02); however, the individual effect estimates were not statistically significant.

Table 3.

ORs (95% CI) for EOC for doubling of androgen concentrations and stratified by grade at diagnosis overall and for serous tumors: The Ovarian Cancer Cohort Consortium (OC3)a

SetsOR (95%CI)
Testosterone 
 Low grade 55 1.28 (0.80–2.07) 
 High grade   
  All 407 0.94 (0.79–1.12) 
  Serous 260 0.84 (0.67–1.04) 
 Phetb 0.25 
 Phetc 0.12 
Free testosterone 
 Low grade 38 1.34 (0.79–2.27) 
 High grade   
  All 277 0.95 (0.80–1.13) 
  Serous 180 0.92 (0.74–1.13) 
 Phetb 0.24 
 Phetc 0.19 
Androstenedione 
 Low grade 55 1.41 (0.86–2.31) 
 High grade   
  All 406 0.84 (0.69–1.01) 
  Serous 259 0.76 (0.60–0.96) 
 Phetb 0.05 
 Phetc 0.02 
DHEAS 
 Low grade 49 1.32 (0.89–1.97) 
 High grade   
  All 374 0.93 (0.81–1.07) 
  Serous 234 0.91 (0.76–1.08) 
 Phetb 0.07 
 Phetc 0.06 
SHBG 
 Low grade 38 0.59 (0.33–1.03) 
 High grade   
  All 286 1.12 (0.93–1.36) 
  Serous 185 1.17 (0.92–1.49) 
 Phetb 0.02 
 Phetc 0.02 
SetsOR (95%CI)
Testosterone 
 Low grade 55 1.28 (0.80–2.07) 
 High grade   
  All 407 0.94 (0.79–1.12) 
  Serous 260 0.84 (0.67–1.04) 
 Phetb 0.25 
 Phetc 0.12 
Free testosterone 
 Low grade 38 1.34 (0.79–2.27) 
 High grade   
  All 277 0.95 (0.80–1.13) 
  Serous 180 0.92 (0.74–1.13) 
 Phetb 0.24 
 Phetc 0.19 
Androstenedione 
 Low grade 55 1.41 (0.86–2.31) 
 High grade   
  All 406 0.84 (0.69–1.01) 
  Serous 259 0.76 (0.60–0.96) 
 Phetb 0.05 
 Phetc 0.02 
DHEAS 
 Low grade 49 1.32 (0.89–1.97) 
 High grade   
  All 374 0.93 (0.81–1.07) 
  Serous 234 0.91 (0.76–1.08) 
 Phetb 0.07 
 Phetc 0.06 
SHBG 
 Low grade 38 0.59 (0.33–1.03) 
 High grade   
  All 286 1.12 (0.93–1.36) 
  Serous 185 1.17 (0.92–1.49) 
 Phetb 0.02 
 Phetc 0.02 

NOTE: Bold numbers indicate statistically significant association.

Abbreviations: DHEAS, dehydroepiandrosterone sulfate; SHBG, sex hormone binding globulin.

aResults were derived from conditional logistic regression models, additionally adjusted for OC use (never/ever/missing) and parity (never/ever/missing). Pairwise heterogeneity tests were performed, using the likelihood ratio test comparing models assuming (i) the same association between exposure and outcomes compared with (ii) a model assuming different associations for each subtype.

bComparing all high grade subtypes to low grade.

cComparing high grade serous to all low grade.

The association between androgens and EOC risk differed by developmental pathway (type I vs. type II tumors, Phet, testosterone: 0.02; free testosterone: 0.01; androstenedione: <0.01; DHEAS: <0.01; Fig. 1). Overall, higher concentrations of androgens were associated with increased risk of type I tumors, and reduced risk of type II tumors (e.g., androstenedione: type I: ORlog2, 1.29; 95% CI, 1.05–1.60; cases n = 287; type II: ORlog2, 0.74; 95% CI, 0.59–0.92; cases n = 307; Phet < 0.01). Significant heterogeneity for androstenedione (P < 0.01) and DHEAS (P = 0.03) remained after exclusion of mucinous and clear cell cases missing data on grade from the type I subgroup (before exclusion, n = 291 case–control sets; after exclusion, n = 77 case–control sets). However, although of the same general magnitude, the effect estimates were no longer statistically significant (Supplementary Fig. S1).

Figure 1.

ORs (95% CI) for EOC for doubling of androgen concentrations and EOC risk by the type I and type II classification: the Ovarian Cancer Cohort Consortium (OC3). Results were derived from conditional logistic regression models, additionally adjusted for OC use (never/ever/missing) and parity (never/ever/missing). Pairwise heterogeneity tests were performed, using the likelihood ratio test comparing models assuming (i) the same association between exposure and outcomes compared with (ii) a model assuming different associations for each subtype. DHEAS, dehydroepiandrosterone sulfate; SHBG, sex hormone binding globulin.

Figure 1.

ORs (95% CI) for EOC for doubling of androgen concentrations and EOC risk by the type I and type II classification: the Ovarian Cancer Cohort Consortium (OC3). Results were derived from conditional logistic regression models, additionally adjusted for OC use (never/ever/missing) and parity (never/ever/missing). Pairwise heterogeneity tests were performed, using the likelihood ratio test comparing models assuming (i) the same association between exposure and outcomes compared with (ii) a model assuming different associations for each subtype. DHEAS, dehydroepiandrosterone sulfate; SHBG, sex hormone binding globulin.

Close modal

Sensitivity and subgroup analyses

We observed some evidence of heterogeneity for the androgens and SHBG and overall EOC by menopausal status at blood collection (androstenedione, Phet = 0.05; SHBG, Phet = 0.02) and age at diagnosis (<55 years vs. ≥55 years: androstenedione, Phet = 0.02; DHEAS, Phet = 0.05; SHBG, Phet = 0.05). Both androstenedione and SHBG were positively associated with risk only among women premenopausal at blood collection (androstenedione: premenopausal women, ORlog2, 1.18; 95% CI, 1.03–1.35; postmenopausal women ORlog2, 0.95; 95% CI, 0.82–1.12; SHBG: premenopausal women, ORlog2, 1.18; 95% CI, 1.00–1.39; postmenopausal women ORlog2, 0.89; 95% CI, 0.76–1.04). No further significant heterogeneity was observed by menopausal status at blood collection. Androstenedione was associated with increased risk of EOC among women diagnosed before age 55 years, but not among women diagnosed at age 55 or older (<55 at diagnosis, ORlog2, 1.21; 95% CI, 1.05–1.40; ≥55 years at diagnosis, ORlog2, 0.95; 95% CI, 0.82–1.10). Although the association between DHEAS and SHBG and EOC differed by age at diagnosis, the ORs were not statistically significant in either age at diagnosis subgroup (e.g., SHBG, <55 at diagnosis, ORlog2, 1.16; 95% CI, 0.98–1.3; ≥55 years at diagnosis, ORlog2, 0.92; 95% CI, 0.79–1.07).

We observed no heterogeneity in analyses by stage at diagnosis. We observed an attenuation of the association between testosterone and EOC after excluding the FMC (n = 576 cases, 43% of sample; after exclusion: ORlog2, 1.06; 95% CI, 0.93–1.21). Overall, ORs were similar for the histologic subtypes after this exclusion, however, no longer statistically significant (e.g., testosterone and endometrioid tumors, before exclusion: n = 164, ORlog2, 1.40; 95% CI, 1.03–1.91; after exclusion: n = 73, ORlog2, 1.39; 95% CI, 0.81–2.36). The most substantial attenuation was for the association between androstenedione and mucinous tumors (before exclusion: n = 191 cases, ORlog2, 1.33; 95% CI, 1.03–1.72; after exclusion: n = 49 cases, ORlog2, 1.19; 95% CI, 0.74–1.92). Excluding women diagnosed within 2 years after blood donation did not meaningfully impact the results (data not shown).

We investigated prediagnosis circulating concentrations of androgens and risk of EOC overall (n = 1,331 cases) and by subtype (case range, n = 61 clear cell to 667 serous), in a collaborative reanalysis of 7 nested case–control studies. The association between testosterone and risk of EOC differed by histologic subtype: endogenous androgens were predominantly associated with increased risk of endometrioid and mucinous tumors, whereas no significant associations were observed for serous or clear cell tumors, although some androgens were inversely associated with high-grade serous and endometrioid (type II) disease.

Ovarian cancer is comprised of four predominant histologic subtypes: serous, mucinous, endometrioid, and clear cell. These histologic subtypes differ substantially by molecular alterations at diagnosis and presumed tissue of origin. The majority of serous tumors are high-grade neoplasms; this subtype represents the majority of invasive EOCs. Separate etiologic pathways are hypothesized for low- and high-grade serous EOC. It is hypothesized that a proportion of low-grade serous carcinomas develop from distal epithelium of the fallopian tube that implants on the ovarian surface epithelium (∼80%), whereas high-grade serous tumors may arise from serous tubal intraepithelial carcinomas (STIC) within the fimbriated end of the fallopian tube (25, 26). Mucinous carcinomas are hypothesized to develop from the gastrointestinal mucosa or from transitional-type epithelium located at the tubal–peritoneal junction; borderline mucinous ovarian tumors are established precursors for this subtype (19). Both endometrioid and clear cell tumors have been proposed to arise from endometrial tissue, and have been associated with endometriosis and retrograde menstruation (19, 27).

Beyond histologic subgroups, two hypothesized developmental pathways of tumorigenesis (type I and type II) have been defined using tumor molecular genetic characteristics (19, 25); in the absence of data on the tumor molecular profile, EOC is classified as type I or type II based on data on histology and grade. Type I tumors include low-grade serous, low-grade endometrioid, mucinous and malignant Brenner tumors (commonly present with KRAS, BRAF, PTEN, PIK3CA, CTNNB1, and ERBB2 mutations)—subtypes that have been hypothesized to develop in a step-wise manner from borderline tumors or endometriosis within or on the surface of the ovary, and are typically diagnosed at earlier disease stage (27). Type II tumors include high-grade serous, high-grade endometrioid, malignant mixed and undifferentiated tumors (typically present with TP53 mutations, but none of the mutations observed in type I disease; ref. 19). These latter tumors comprise the majority of EOCs, are aggressive, and typically present at an advanced stage.

Prior epidemiologic data suggest risk factor differences by EOC subtype defined by histology (e.g., refs. 12, 15–18) and the developmental pathway (11, 14). Consistent differences by histologic subtype of invasive EOC are observed for hormone-related risk factors, including duration of OC use (lower risk of all histologic subtypes but mucinous; refs. 12, 15), older age at menopause (higher risk of all but mucinous; ref. 12), smoking (higher risk of mucinous, lower risk of clear cell; refs. 12, 17), parity (more strongly protective in non-serous subtypes; ref. 12), postmenopausal hormone therapy (HT) use (higher risk of serous and endometrioid subtypes only; refs. 12, 18), and adiposity (among non-HT users; higher risk of serous and endometrioid subtypes only; ref. 16). Data by the type I/II classification are sparse, but consistently show stronger associations between parity and type I, relative to type II, disease (11, 14). Three prospective studies evaluated circulating estrogens (10, 28) and/or androgens (9, 10) and invasive EOC risk by subtype. Higher concentrations of both estrogens and androgens were associated with increased risk of non-serous EOC subtypes (9, 10, 28), whereas higher concentrations of androstenedione had opposing effects on risk of type I (higher risk) and type II (lower risk) EOC (9).

In women, androgens are produced in the ovary, adrenal glands, and via peripheral conversion of androgen precursors (e.g., DHEA); in turn, androgens are the substrate for estrogen production by aromatase. DHEAS is a pre-androgen synthesized in the adrenal gland, and subsequently metabolized toward androstenedione and testosterone, or estradiol (29). Androstenedione, an intermediate between DHEA and DHEAS and testosterone, is produced in both the ovary (premenopausal women: 40%; postmenopausal women: 20%–30%) and the adrenal gland. In premenopausal women, approximately 25% of circulating testosterone originates in the ovary, 25% in the adrenal glands, and 50% is metabolized from precursors such as androstenedione in peripheral tissues (e.g., liver, adipose tissue; refs. 29, 30); the proportion of testosterone of ovarian origin is higher in postmenopausal women (∼50%; ref. 29). These androgens are correlated with each other (e.g., r = 0.54 between DHEAS and androstenedione to r = 0.69 between DHEAS and testosterone; adjusted for menopausal status; ref. 6) and weakly correlated with estradiol [e.g., estradiol and testosterone: premenopausal women: r = 0.08; ref. 31; postmenopausal women, r = 0.23–0.38; refs. 32, 33; and body mass index (r = 0.07–0.13); refs. 31–33].

Androgens may (i) directly influence ovarian carcinogenesis through androgen receptor (AR) signaling, or (ii) impact risk through their role as estrogen precursors; associations with estrogens may be most evident in the context of progesterone insufficiency as observed in polycystic ovarian syndrome (PCOS). ARs and estrogen (ER) receptors are expressed in the normal ovary, including ovarian surface epithelial cells and cortical inclusion cysts, and the fallopian tube (34–36). In vivo data show that ovarian cancer preferentially develops in a hormonal milieu enriched with androgens (e.g., testosterone induces epithelial neoplasms in guinea pigs; ref. 37) or estrogens (e.g., estrogen-induced tumor growth in high-grade serous ovarian cancers; refs. 38, 39). The hyperandrogenic PCOS is characterized by functional ovarian hyperandrogenism, with an excess of testosterone produced in the ovarian thecal cells (40); up to 45% of cases additionally present with adrenal hyperandrogenism (41). Estimates of PCOS prevalence range from 5% to 15% (30); the syndrome has highest prevalence among reproductive-age women. PCOS-related androgen excess is observed in both pre- and postmenopausal women (42). Progesterone deficiency is a hallmark of PCOS, resulting in a higher ratio of estrogens to progesterone. PCOS (43, 44) and relatively high levels of estrogens unopposed by progesterone are associated with increased endometrial cancer risk (i.e., estrogen-alone HT; ref. 45; relatively high endogenous estrogens in postmenopausal women; refs. 33, 46). These associations with endometrial cancer may be most relevant to the endometrioid or clear cell EOC, given endometrial tissue is a proposed tissue of origin for these subtypes. PCOS itself has not consistently been associated with ovarian cancer (43, 44, 47), although data by subtype are limited. Data to date suggest both estrogen-alone and estrogen plus progesterone HT are associated with increased risk of endometrioid EOC (18).

In the current study, we evaluated three members of the androgen synthesis pathway, DHEAS, androstenedione, testosterone, and EOC risk by histology (i.e., accounting for hypothesized differences in cell of origin) and developmental pathway (i.e., “less” relative to “more” aggressive disease). We observed a significant positive association between testosterone and risk of endometrioid ovarian cancer. There is limited in vitro evidence to support a role of androgens in the etiology of endometrioid EOC (34, 48). However, given the possible common tissue of origin, it is plausible that androgens impact risk similarly in both endometrial cancer and endometrioid EOC. With respect to endometrial cancer, recent in vivo data have demonstrated that androgens induce epithelial proliferation in the mouse uterus (49), and epidemiologic data provide some support for an association between androgens and endometrial cancer risk (50). Together, these data on endometrial cancer provide indirect evidence supporting an association between androgens and endometrioid EOC. Androgens are an intermediate on the estrogen-synthesis pathway, and estrogen exposure unopposed by progesterone may be the underlying biological mechanism linking androgens to endometrioid EOC, particularly if in the context progesterone deficiency, as in PCOS and in postmenopausal women. Prior research has linked higher early pregnancy estradiol concentrations to a 2.5-fold increase in risk of endometrioid EOC (10), and postmenopausal HT use (12, 18) and adiposity (16) are associated with increased risk of this subtype. We adjusted for BMI in a sensitivity analysis, given (i) the association between PCOS and obesity and (ii) adipose tissue is a key site of metabolism of androgens to estrogens in postmenopausal women. Adjustment for BMI did not impact the results. Data on history of PCOS were not available.

Higher concentrations of all investigated androgens, except DHEAS, were significantly associated with increased risk of mucinous tumors. Emerging data suggest the ovarian stroma proximal to mucinous EOC has higher concentrations of sex-steroid producing enzymes than distant stroma, providing support for a role for sex steroids in the development of mucinous disease (35). Androgens (directly, or after conversion to estrogens) may contribute to growth promotion in the early stages of mucinous disease; however, to our knowledge, the androgen responsiveness of mucinous tumors is not well characterized, and data on ER expression are limited (51, 52). The precise biological mechanisms underlying the observed associations between androgens and mucinous tumors remain an open question.

In line with two prior prospective studies (9, 10), both included in this analysis, we observed no association with pre-diagnosis androgen concentrations and increased risk of serous carcinomas. Recent data on estrogens and ovarian cancer are in line with our results on androgens, with no association observed between estrogens and risk of invasive serous tumors in the FMC (first-trimester estrogens; ref. 10) or among postmenopausal women in the Women's Health Initiative (28). We observed no associations with clear cell disease. However, sample size for this subtype was limited.

We observed significant heterogeneity in the strength of associations between androgens and risk of type I vs. type II tumors; higher androgen concentrations were associated with higher risk of type I, but lower risk of type II (predominantly high grade serous), tumors. These results are in agreement with the single prior study on endogenous androgens and EOC risk using the dualistic model classification (9); these data from the European Prospective Investigation into Cancer and Nutrition (EPIC) were included in the current analysis. There is indirect evidence for differences in hormone dependency in type I and type II tumors, based on the variation of ER expression between low-grade (ER expression: 58%) and high-grade serous carcinomas (ER expression: 27%; ref. 53). However, the mechanisms linking androgen concentrations to lower risk of type II tumors in our study are unclear. Although chance and residual confounding may explain the results, future work should explicitly examine the impact of androgens on type II tumors.

Given the large sample size, our study was powered to investigate risk associations for less common tumors (e.g., mucinous tumors) and by developmental pathway (type I/type II). A general weakness of pooled analyses is the difference in data availability of covariates and differences in laboratory methods. In this investigation, data from each cohort were centrally compiled and harmonized and we addressed differences in absolute biomarker concentrations (i) using study-specific tertiles and (ii) standardizing hormone measurements using study-specific mean concentrations. Results were robust regardless of whether we calculated ORs from the pooling of individual data or from meta-analysis. For some of the investigated hormones, the number of sets that could be used was reduced for subgroup analyses, which resulted in reduced power. In our primary analyses using the developmental pathway classification, we included all mucinous and clear cell tumors in the “type I” classification, as their classification is independent of grade. If there were systematic differences in the observed associations with type I disease in cases with and without grade data, this may result in a biased interpretation of the differences between type I and type II EOC. However, the associations observed in our primary analysis and in a sensitivity analysis restricted to women with complete data on grade were of similar magnitude. Many statistical tests are reported; therefore, some significant observations may be due to chance. However, all statistical analyses were hypothesis driven. In line with the majority of other epidemiological studies, a single measurement of biomarkers was used to assess risk associations. This single measurement may not reflect long-term average concentrations and the storage time and conditions may impact the true value of the biochemical indicators. However, the stability of androgen measurements over time has been shown previously for a period over at least 2–3 years: (i) premenopausal women [ICC ranged from 0.58 (androstenedione) up to 0.81 (DHEAS; ref. 54) and (ii) postmenopausal women [ICC ranged from 0.66 (androstenedione) up to 0.92 (SHBG; ref. 55).

The testosterone synthesis pathway (e.g., DHEAS, androstenedione, testosterone) may play an important role in the onset and progression of a subset of epithelial invasive ovarian carcinomas. Androgens may either have a direct impact on ovarian carcinogenesis, or act through increased synthesis of other steroid hormones (e.g., estrogens); this is an area for future epidemiologic research. Although androgens were associated with increased risk of non-serous tumors, we observed an inverse association between androstenedione and high grade serous tumors. In addition to providing novel findings on hormone-related pathways in ovarian carcinogenesis, this study supports emerging data on the heterogeneity of epithelial invasive ovarian cancer and underscores the importance of examining etiologic differences for subtypes.

No potential conflicts of interest were disclosed.

The authors assume full responsibility for analyses and interpretation of these data.

Conception and design: M. Lehtinen, L. Dossus, N.C. Onland-Moret, E. Weiderpass, N. Wentzensen, S.S. Tworoger, R. Kaaks, R.T. Fortner

Development of methodology: J. Ose, M. Lehtinen, L. Dossus, E. Weiderpass

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): E.M. Poole, A.A. Arslan, A. Zeleniuch-Jacquotte, K. Visvanathan, K. Helzlsouer, J.E. Buring, I.-M. Lee, L. Dossus, G. Masala, N.C. Onland-Moret, E. Weiderpass, A. Idahl, R.C. Travis, S.S. Tworoger, R. Kaaks

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): J. Ose, E.M. Poole, H. Schock, M. Lehtinen, A. Zeleniuch-Jacquotte, K. Visvanathan, L. Dossus, N.C. Onland-Moret, E. Weiderpass, B. Trabert, N. Wentzensen, S.S. Tworoger, R.T. Fortner

Writing, review, and/or revision of the manuscript: J. Ose, E.M. Poole, H. Schock, M. Lehtinen, A.A. Arslan, A. Zeleniuch-Jacquotte, K. Visvanathan, K. Helzlsouer, J.E. Buring, I.-M. Lee, A. Tjønneland, L. Dossus, A. Trichopoulou, N.C. Onland-Moret, E. Weiderpass, E.J. Duell, A. Idahl, R.C. Travis, S. Rinaldi, M.A. Merritt, B. Trabert, N. Wentzensen, S.S. Tworoger, R. Kaaks, R.T. Fortner

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): E.M. Poole, A.A. Arslan, L. Dossus, E. Weiderpass, A. Idahl, S.S. Tworoger, R. Kaaks

Study supervision: L. Dossus, I.-M. Lee, E. Weiderpass, R.T. Fortner

We thank the participants and staff of the Nurses' Health Study and Nurses' Health Study II for their valuable contributions as well as the following state cancer registries for their help: Alabama, Arizona, Arkansas, California, Colorado, Connecticut, Delaware, Florida, Georgia, Idaho, Illinois, Indiana, Iowa, Kentucky, Louisiana, Maine, Maryland, Massachusetts, Michigan, Nebraska, New Hampshire, New Jersey, New York, North Carolina, North Dakota, Ohio, Oklahoma, Oregon, Pennsylvania, Rhode Island, South Carolina, Tennessee, Texas, Virginia, Washington, and Wyoming.

We thank the participants and staff of CLUE II for their contributions as well as the Maryland Cancer Registry, Center for Cancer Surveillance and Control, Department of Health and Mental Hygiene, (http://phpa.dhmh.maryland.gov/cancer). We acknowledge the State of Maryland, the Maryland Cigarette Restitution Fund, and the National Program of Cancer Registries of the Centers for Disease Control and Prevention for the funds that support the collection and availability of the cancer registry data.

The OC3 is supported by Department of Defense Ovarian Cancer Research Program (grant no. W81XWH-12-1-0561). R.T. Fortner was supported by a Marie Curie International Incoming Fellowship of the European Commission's Seventh Framework Programme (MC-IIF-623984). The coordination of EPIC is financially supported by the European Commission (DG-SANCO) and the International Agency for Research on Cancer. The national cohorts are supported by Danish Cancer Society (Denmark); Ligue Contre le Cancer, Institut Gustave Roussy, Mutuelle Générale de l'Education Nationale, Institut National de la Santé et de la Recherche Médicale (INSERM; France); German Cancer Aid, German Cancer Research Center (DKFZ), Federal Ministry of Education and Research (BMBF), Deutsche Krebshilfe, Deutsches Krebsforschungszentrum and Federal Ministry of Education and Research (Germany); the Hellenic Health Foundation (Greece); Associazione Italiana per la Ricerca sul Cancro-AIRC-Italy and National Research Council (Italy); Dutch Ministry of Public Health, Welfare and Sports (VWS), Netherlands Cancer Registry (NKR), LK Research Funds, Dutch Prevention Funds, Dutch ZON (Zorg Onderzoek Nederland), World Cancer Research Fund (WCRF), Statistics Netherlands (The Netherlands); ERC-2009-AdG 232997 and Nordforsk, Nordic Centre of Excellence programme on Food, Nutrition and Health (Norway); Health Research Fund (FIS), PI13/00061 to Granada; PI13/01162 to EPIC-Murcia), Regional Governments of Andalucía, Asturias, Basque Country, Murcia and Navarra, ISCIII RETIC (RD06/0020; Spain); Swedish Cancer Society, Swedish Research Council and County Councils of Skåne and Västerbotten (Sweden); Cancer Research UK (14136 to EPIC-Norfolk; C570/A16491 and C8221/A19170 to EPIC-Oxford), Medical Research Council (1000143 to EPIC-Norfolk, MR/M012190/1 to EPIC-Oxford; United Kingdom). For information on how to submit an application for gaining access to EPIC data and/or biospecimens, please follow the instructions at http://epic.iarc.fr/access/index.php. This work was also supported by the following NIH/NCI grants: R01 CA120061 (Finnish Maternity Cohort); UM1 CA186107, P01 CA87969, UM1 CA176726, and R01 CA67262 (Nurses' Health Study, Nurses' Health Study II); HL043851, HL080467, and HL099355 (Women's Health Study). CLUEII was supported by National Cancer Institute grant numbers: CA-97857, CA-86308 [grant sponsor: M. Jean Goutal (donation)]: grant sponsors: The Woodrow Wilson Foundation/Johnson and Johnson, The Lloyds TSB Charitable Foundation for the Channel Islands.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1.
Lukanova
A
,
Kaaks
R
. 
Endogenous hormones and ovarian cancer: epidemiology and current hypotheses
.
Cancer Epidemiol Biomarkers Prev
2005
;
14
:
98
107
.
2.
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
.
3.
Modugno
F
,
Laskey
RA
,
Smith
AL
,
Andersen
CL
,
Haluska
P
,
Oesterreich
S
. 
Hormone response in ovarian cancer: time to reconsider as a clinical target?
Endocr Relat Cancer
2012
;
19
:
R255
79
.
4.
Edmondson
RJ
,
Monaghan
JM
,
Davies
BR
. 
The human ovarian surface epithelium is an androgen responsive tissue
.
Br J Cancer
2002
;
86
:
879
85
.
5.
Helzlsouer
KJ
,
Alberg
AJ
,
Gordon
GB
,
Longcope
C
,
Bush
TL
,
Hoffman
SC
, et al
Serum gonadotropins and steroid hormones and the development of ovarian cancer
.
JAMA
1995
;
274
:
1926
30
.
6.
Lukanova
A
,
Lundin
E
,
Akhmedkhanov
A
,
Micheli
A
,
Rinaldi
S
,
Zeleniuch-Jacquotte
A
, et al
Circulating levels of sex steroid hormones and risk of ovarian cancer
.
Int J Cancer
2003
;
104
:
636
42
.
7.
Tworoger
SS
,
Lee
IM
,
Buring
JE
,
Hankinson
SE
. 
Plasma androgen concentrations and risk of incident ovarian cancer
.
Am J Epidemiol
2008
;
167
:
211
8
.
8.
Rinaldi
S
,
Dossus
L
,
Lukanova
A
,
Peeters
PH
,
Allen
NE
,
Key
T
, et al
Endogenous androgens and risk of epithelial ovarian cancer: results from the European Prospective Investigation into Cancer and Nutrition (EPIC)
.
Cancer Epidemiol Biomarkers Prev
2007
;
16
:
23
9
.
9.
Ose
J
,
Fortner
RT
,
Rinaldi
S
,
Schock
H
,
Overvad
K
,
Tjonneland
A
, et al
Endogenous androgens and risk of epithelial invasive ovarian cancer by tumor characteristics in the European Prospective Investigation into Cancer and Nutrition
.
Int J Cancer
2015
;
136
:
399
410
.
10.
Schock
H
,
Surcel
HM
,
Zeleniuch-Jacquotte
A
,
Grankvist
K
,
Lakso
HA
,
Fortner
RT
, et al
Early pregnancy sex steroids and maternal risk of epithelial ovarian cancer
.
Endocr Relat Cancer
2014
;
21
:
831
44
.
11.
Fortner
RT
,
Ose
J
,
Merritt
MA
,
Schock
H
,
Tjonneland
A
,
Hansen
L
, et al
Reproductive and hormone-related risk factors for epithelial ovarian cancer by histologic pathways, invasiveness and histologic subtypes: Results from the EPIC cohort
.
Int J Cancer
2015
;
137
:
1196
208
.
12.
Wentzensen
N
,
Poole
EM
,
Trabert
B
,
White
E
,
Arslan
AA
,
Patel
AV
, et al
Ovarian cancer risk factors by histologic subtype: an analysis from the ovarian cancer cohort consortium
.
J Clin Oncol
2016
;
34
:
2888
98
.
13.
Prat
J
. 
Ovarian carcinomas: five distinct diseases with different origins, genetic alterations, and clinicopathological features
.
Virchows Arch
2012
;
460
:
237
49
.
14.
Merritt
MA
,
De Pari
M
,
Vitonis
AF
,
Titus
LJ
,
Cramer
DW
,
Terry
KL
. 
Reproductive characteristics in relation to ovarian cancer risk by histologic pathways
.
Hum Reprod
2013
;
28
:
1406
17
.
15.
Collaborative Group on Epidemiological Studies of Ovarian Cancer
,
Beral
V
,
Doll
R
,
Hermon
C
,
Peto
R
,
Reeves
G
. 
Ovarian cancer and oral contraceptives: collaborative reanalysis of data from 45 epidemiological studies including 23,257 women with ovarian cancer and 87,303 controls
.
Lancet
2008
;
371
:
303
14
.
16.
Collaborative Group on Epidemiological Studies of Ovarian Cancer
. 
Ovarian cancer and body size: individual participant meta-analysis including 25,157 women with ovarian cancer from 47 epidemiological studies
.
PLoS Med
2012
;
9
:
e1001200
.
17.
Collaborative Group on Epidemiological Studies of Ovarian Cancer
,
Beral
V
,
Gaitskell
K
,
Hermon
C
,
Moser
K
,
Reeves
G
, et al
Ovarian cancer and smoking: individual participant meta-analysis including 28,114 women with ovarian cancer from 51 epidemiological studies
.
Lancet Oncol
2012
;
13
:
946
56
.
18.
Collaborative Group On Epidemiological Studies of Ovarian Cancer
. 
Menopausal hormone use and ovarian cancer risk: individual participant meta-analysis of 52 epidemiological studies
.
Lancet
2015
;
385
:
1835
42
.
19.
Kurman
RJ
,
Shih Ie
M
. 
Molecular pathogenesis and extraovarian origin of epithelial ovarian cancer–shifting the paradigm
.
Hum Pathol
2011
;
42
:
918
31
.
20.
Sodergard
R
,
Backstrom
T
,
Shanbhag
V
,
Carstensen
H
. 
Calculation of free and bound fractions of testosterone and estradiol-17 beta to human plasma proteins at body temperature
.
J Steroid Biochem
1982
;
16
:
801
10
.
21.
DerSimonian
R
,
Laird
N
. 
Meta-analysis in clinical trials
.
Control Clin Trials
1986
;
7
:
177
88
.
22.
Smith-Warner
SA
,
Spiegelman
D
,
Ritz
J
,
Albanes
D
,
Beeson
WL
,
Bernstein
L
, et al
Methods for pooling results of epidemiologic studies: the Pooling Project of Prospective Studies of Diet and Cancer
.
Am J Epidemiol
2006
;
163
:
1053
64
.
23.
Wang
M
,
Spiegelman
D
,
Kuchiba
A
,
Lochhead
P
,
Kim
S
,
Chan
AT
, et al
Statistical methods for studying disease subtype heterogeneity
.
Stat Med
2016
;
35
:
782
800
.
24.
Schock
H
,
Zeleniuch-Jacquotte
A
,
Lundin
E
,
Grankvist
K
,
Lakso
HA
,
Idahl
A
, et al
Hormone concentrations throughout uncomplicated pregnancies: a longitudinal study
.
BMC Pregnancy Childbirth
2016
;
16
:
146
.
25.
Nik
NN
,
Vang
R
,
Shih Ie
M
,
Kurman
RJ
. 
Origin and pathogenesis of pelvic (ovarian, tubal, and primary peritoneal) serous carcinoma
.
Annu Rev Pathol
2014
;
9
:
27
45
.
26.
Zeppernick
F
,
Meinhold-Heerlein
I
,
Shih Ie
M
. 
Precursors of ovarian cancer in the fallopian tube: serous tubal intraepithelial carcinoma–an update
.
J Obstet Gynaecol Res
2015
;
41
:
6
11
.
27.
Seidman
JD
,
Cho
KR
,
Ronnett
BM
,
Kurman
RJ
. 
Surface epithelial tumors of the ovary
.
In:
Kurman
RJ
,
Hedrick Ellenson
L
,
Ronnett
BM
,
editors
.
Blaustein's Pathology of the Female Genital Tract
, 6th ed:
Springer
; 
2011
.
28.
Trabert
B
,
Brinton
LA
,
Anderson
GL
,
Pfeiffer
RM
,
Falk
RT
,
Strickler
HD
, et al
Circulating estrogens and postmenopausal ovarian cancer risk in the Women's Health Initiative Observational Study
.
Cancer Epidemiol Biomarkers Prev
2016
;
25
:
648
56
.
29.
Yen
SS
,
Jaffe
RB
,
Barbieri
RL
.
Reproductive Endocrinology
. 4th ed.
Philadelphia, PA
:
W.B. Saunders Company
; 
1999
.
30.
Rosenfield
RL
,
Ehrmann
DA
. 
The pathogenesis of polycystic ovary syndrome (PCOS): the hypothesis of PCOS as functional ovarian hyperandrogenism revisited
.
Endocr Rev
2016
;
37
:
467
520
.
31.
Kaaks
R
,
Berrino
F
,
Key
T
,
Rinaldi
S
,
Dossus
L
,
Biessy
C
, et al
Serum sex steroids in premenopausal women and breast cancer risk within the European Prospective Investigation into Cancer and Nutrition (EPIC)
.
J Natl Cancer Inst
2005
;
97
:
755
65
.
32.
James
RE
,
Lukanova
A
,
Dossus
L
,
Becker
S
,
Rinaldi
S
,
Tjønneland
A
, et al
Postmenopausal serum sex steroids and risk of hormone receptor-positive and -negative breast cancer: a nested case-control study
.
Cancer Prev Res
2011
;
4
:
1626
35
.
33.
Lukanova
A
,
Lundin
E
,
Micheli
A
,
Arslan
A
,
Ferrari
P
,
Rinaldi
S
, et al
Circulating levels of sex steroid hormones and risk of endometrial cancer in postmenopausal women
.
Int J Cancer
2004
;
108
:
425
32
.
34.
Gibson
DA
,
Simitsidellis
I
,
Collins
F
,
Saunders
PT
. 
Evidence of androgen action in endometrial and ovarian cancers
.
Endocr Relat Cancer
2014
;
21
:
T203
18
.
35.
Blanco
LZ
 Jr
,
Kuhn
E
,
Morrison
JC
,
Bahadirli-Talbott
A
,
Smith-Sehdev
A
,
Kurman
RJ
. 
Steroid hormone synthesis by the ovarian stroma surrounding epithelial ovarian tumors: a potential mechanism in ovarian tumorigenesis
.
Mod Pathol
2017
;
30
:
563
76
.
36.
Kyriakidis
I
,
Papaioannidou
P
. 
Estrogen receptor beta and ovarian cancer: a key to pathogenesis and response to therapy
.
Archiv Gynecol Obstet
2016
;
293
:
1161
8
.
37.
Silva
EG
,
Tornos
C
,
Fritsche
HA
 Jr
,
el-Naggar
A
,
Gray
K
,
Ordonez
NG
, et al
The induction of benign epithelial neoplasms of the ovaries of guinea pigs by testosterone stimulation: a potential animal model
.
Mod Pathol
1997
;
10
:
879
83
.
38.
Silva
EG
,
Tornos
C
,
Deavers
M
,
Kaisman
K
,
Gray
K
,
Gershenson
D
. 
Induction of epithelial neoplasms in the ovaries of guinea pigs by estrogenic stimulation
.
Gynecol Oncol
1998
;
71
:
240
6
.
39.
Ciucci
A
,
Zannoni
GF
,
Buttarelli
M
,
Lisi
L
,
Travaglia
D
,
Martinelli
E
, et al
Multiple direct and indirect mechanisms drive estrogen-induced tumor growth in high grade serous ovarian cancers
.
Oncotarget
2016
;
7
:
8155
71
.
40.
Ehrmann
DA
. 
Polycystic ovary syndrome
.
N Engl J Med
2005
;
352
:
1223
36
.
41.
Luque-Ramirez
M
,
Escobar-Morreale
HF
. 
Adrenal hyperandrogenism and polycystic ovary syndrome
.
Curr Pharm Des
2016
;
22
:
5588
602
.
42.
Markopoulos
MC
,
Kassi
E
,
Alexandraki
KI
,
Mastorakos
G
,
Kaltsas
G
. 
Hyperandrogenism after menopause
.
Eur J Endocrinol
2015
;
172
:
R79
91
.
43.
Gottschau
M
,
Kjaer
SK
,
Jensen
A
,
Munk
C
,
Mellemkjaer
L
. 
Risk of cancer among women with polycystic ovary syndrome: a Danish cohort study
.
Gynecol Oncol
2015
;
136
:
99
103
.
44.
Barry
JA
,
Azizia
MM
,
Hardiman
PJ
. 
Risk of endometrial, ovarian and breast cancer in women with polycystic ovary syndrome: a systematic review and meta-analysis
.
Hum Reprod Update
2014
;
20
:
748
58
.
45.
Beral
V
,
Bull
D
,
Reeves
G
,
Million Women Study
C
. 
Endometrial cancer and hormone-replacement therapy in the Million Women Study
.
Lancet
2005
;
365
:
1543
51
.
46.
Allen
NE
,
Key
TJ
,
Dossus
L
,
Rinaldi
S
,
Cust
A
,
Lukanova
A
, et al
Endogenous sex hormones and endometrial cancer risk in women in the European Prospective Investigation into Cancer and Nutrition (EPIC)
.
Endocr Relat Cancer
2008
;
15
:
485
97
.
47.
Harris
HR
,
Titus
LJ
,
Cramer
DW
,
Terry
KL
. 
Long and irregular menstrual cycles, polycystic ovary syndrome, and ovarian cancer risk in a population-based case-control study
.
Int J Cancer
2017
;
140
:
285
91
.
48.
Maliqueo
MA
,
Quezada
S
,
Clementi
M
,
Bacallao
K
,
Anido
M
,
Johnson
C
, et al
Potential action of androstenedione on the proliferation and apoptosis of stromal endometrial cells
.
Reprod Biol Endocrinol
2004
;
2
:
81
.
49.
Simitsidellis
I
,
Gibson
DA
,
Cousins
FL
,
Esnal-Zufiaurre
A
,
Saunders
PT
. 
A role for androgens in epithelial proliferation and formation of glands in the mouse uterus
.
Endocrinology
2016
;
157
:
2116
28
.
50.
Clendenen
TV
,
Hertzmark
K
,
Koenig
KL
,
Lundin
E
,
Rinaldi
S
,
Johnson
T
, et al
Premenopausal circulating androgens and risk of endometrial cancer: results of a prospective study
.
Horm Cancer
2016
;
7
:
178
87
.
51.
Tkalia
IG
,
Vorobyova
LI
,
Svintsitsky
VS
,
Nespryadko
SV
,
Goncharuk
IV
,
Lukyanova
NY
, et al
Clinical significance of hormonal receptor status of malignant ovarian tumors
.
Exp Oncol
2014
;
36
:
125
33
.
52.
Sieh
W
,
Kobel
M
,
Longacre
TA
,
Bowtell
DD
,
deFazio
A
,
Goodman
MT
, et al
Hormone-receptor expression and ovarian cancer survival: an Ovarian Tumor Tissue Analysis consortium study
.
Lancet Oncol
2013
;
14
:
853
62
.
53.
Escobar
J
,
Klimowicz
AC
,
Dean
M
,
Chu
P
,
Nation
JG
,
Nelson
GS
, et al
Quantification of ER/PR expression in ovarian low-grade serous carcinoma
.
Gynecol Oncol
2013
;
128
:
371
6
.
54.
Missmer
SA
,
Spiegelman
D
,
Bertone-Johnson
ER
,
Barbieri
RL
,
Pollak
MN
,
Hankinson
SE
. 
Reproducibility of plasma steroid hormones, prolactin, and insulin-like growth factor levels among premenopausal women over a 2- to 3-year period
.
Cancer Epidemiol Biomarkers Prev
2006
;
15
:
972
8
.
55.
Hankinson
SE
,
Manson
JE
,
Spiegelman
D
,
Willett
WC
,
Longcope
C
,
Speizer
FE
. 
Reproducibility of plasma hormone levels in postmenopausal women over a 2–3-year period
.
Cancer Epidemiol Biomarkers Prev
1995
;
4
:
649
54
.