Abstract
Purpose
Reproductive characteristics, the most established ovarian cancer risk factors, differ markedly between African American and White women. Studies in predominantly White populations suggest that associations between reproductive characteristics and ovarian cancer vary by timing of the events and menopause status. This analysis examined associations between number, duration and timing of reproductive events and epithelial ovarian cancer among African American women.
Methods
Data from a multi-center case-control study of ovarian cancer in African American women (641 cases/752 controls) were used to examine associations with oral contraceptive use and pregnancy characteristics. Odds ratios(OR) and 95% confidence intervals(CI) associated with reproductive characteristics were calculated with logistic regression models.
Results
Oral contraceptive use (OR=0.7, 95%CI 0.5–0.9), parity (OR=0.5, 95% CI 0.3–0.6) and breastfeeding for >12 months (OR=0.3, 95% CI 0.2–0.5) were inversely associated with ovarian cancer. More recent pregnancies and oral contraceptive use had stronger associations with ovarian cancer than pregnancies or oral contraceptive use that occurred earlier in life, especially among pre-menopausal women.
Conclusions
This study provides the first thorough documentation that pregnancy, breastfeeding and oral contraceptive use are inversely associated with ovarian cancer in African American women, similar to what has been observed in White women. The associations with timing of the exposures suggest that these factors have both short and long-term effects.
Keywords: African Americans, breastfeeding, case-control study, oral contraceptives, ovarian cancer, parity, pregnancy
Introduction
Epithelial ovarian cancer has a median age of diagnosis of approximately 63 years.1 Despite being a disease that is more frequently diagnosed among post-menopausal women, the factors that most influence ovarian cancer risk are reproductive characteristics such as pregnancy, oral contraceptive use and breastfeeding that typically occur when a woman is in her twenties or thirties.2 Analyses conducted within predominantly White populations suggest that the associations between reproductive characteristics and ovarian cancer depend upon the timing of the exposure and may differ for ovarian cancer diagnosed before and after menopause.3–12 Most notably, the inverse association with pregnancy seems to be stronger for pre-menopausal women, which may be due to the effect of time since last pregnancy.3,4,6,8,12 More recent pregnancies have been associated with a greater reduction in ovarian cancer risk that appears to be independent of the total number of pregnancies. These findings suggest that reproductive risk factors may operate through multiple biological pathways that may have both short-term and long-term effects on ovarian cancer risk.
African American women differ markedly from White women in their incidence of ovarian cancer (9.8/100,000 and 12.8/100,000 in African Americans and Whites, respectively)1 as well as in many of their reproductive characteristics. On average, African American women experience more total pregnancies,13 an earlier age at first pregnancy,13 less breastfeeding,14 and less oral contraceptive use.15 There are only a few published reports describing the association between reproductive characteristics and ovarian cancer risk in African American women, and all have had very modest sample sizes.16–19 While these studies have reported inverse associations with pregnancy and oral contraceptive use similar to what has been reported in White women, none of them has presented results stratified by menopausal status and all were limited in their ability to examine effects by duration, number or timing of the reproductive events.
The purpose of this report is to describe associations between ovarian cancer and the reproductive characteristics of oral contraceptive use, parity, and breastfeeding stratified by menopausal status, using data from a multi-center, case-control study of ovarian cancer in African American women. We present overall associations with ovarian cancer risk as well as examine the effect of the number of pregnancies, the duration of exposure to oral contraceptives and timing of the exposures.
Methods
The data used in these analyses are from the African American Cancer Epidemiology Study (AACES), a population-based, case-control study of ovarian cancer in African American women in 11 geographic regions: North Carolina, South Carolina, Georgia, Alabama, Tennessee, Louisiana, Texas, New Jersey, Ohio, Chicago and Detroit. Duke University is the lead institution for the study. Institutional review board approval was obtained from the Duke University School of Medicine and all participating institutions. The methods of the study have been previously reported 20 and are described here briefly.
Women with ovarian cancer were identified using rapid case ascertainment systems through state cancer registries, Surveillance, Epidemiology and End Results (SEER) registries or individual hospital registries. Inclusion criteria were self-identified African-American/Black race, aged 20–79 years, diagnosis of invasive, epithelial ovarian cancer, no prior history of ovarian cancer and ability to complete an interview in English. Of 1546 eligible cases identified, physician consent was not obtained for 1% of the women, 17% died before they could be contacted, 16% could not be contacted, 23% refused to participate and 42% were enrolled in the study. Controls were selected using random digit dialing, frequency matched to cases on age and geographic region. Eligibility criteria were similar to cases plus they must not have had bilateral oophorectomy or a prior history of ovarian cancer. Of 1450 eligible controls identified, 0.2% died, 24% could not be contacted for an interview, 24% refused to participate and 52% were enrolled in the study. The current analyses are based on women enrolled from December 2010 through January 2016 and include 641 cases and 752 controls.
Data were collected via an interviewer-administered computer-assisted telephone interview (CATI). Survey information included demographic characteristics; reproductive, gynecologic and medical history; hormone use; family history of cancer; and lifestyle characteristics such as smoking, alcohol consumption and physical activity. For the pregnancy characteristics, women provided detailed information on each pregnancy including outcome, duration, date pregnancy ended and breastfeeding information. A full-term pregnancy for the purposes of these analyses was defined as one lasting more than 6 months. Oral contraceptive information was based on a detailed lifetime contraceptive history of the type and timing of each method used.
Menopausal status was based on self-reported menstrual history. Women were categorized as post-menopausal if their menstrual periods had stopped naturally 12 or more months prior to diagnosis/interview or their periods stopped due to chemotherapy or radiation. Women who had started menopausal hormones before their periods stopped and had been taking them for at least two years or thought that they began menopause at least 4 years prior to diagnosis or interview were categorized as post-menopausal. Women who had a pre-menopausal hysterectomy without bilateral oophorectomy were considered post-menopausal if they were 50 years of age or older at diagnosis/interview or, if they were younger than age 50, at least 4 years had passed since they thought they began menopause.
Demographic and other descriptive characteristics of cases and controls were compared using a chi-squared test. Unconditional logistic regression analyses were used to calculate odds ratios (ORs) and 95% confidence intervals (CIs) for parameters related to oral contraceptive use and pregnancy history. Variables included as potential confounders included study site, age (continuous), family history of breast or ovarian cancer in a first degree relative (yes/no), age at menarche (continuous), tubal ligation (yes/no), and body mass index (BMI in kg/m2, continuous). Analyses that simultaneously examined the timing of exposure and duration of oral contraceptive use or timing and number of pregnancies were restricted to ever users of oral contraceptives and parous women, respectively. To perform tests for trend, categories of the variables were coded as continuous variables. Tests for interaction were conducted by including in the model a product term for menopausal status and the individual reproductive exposure variable. All analyses were conducted using SAS version 9.3 software.
Results
Descriptive characteristics of cases and controls stratified by menopausal status are presented in Table 1. For most characteristics, the direction of the associations were similar for pre- and post-menopausal women, although the magnitude of the differences between cases and controls varied between pre-and post-menopausal cases for several characteristics including family history of cancer, and infertility, which had stronger associations in pre-menopausal women. A notable exception was that post-menopausal cases had higher BMI than controls, which was not the case in pre-menopausal women.
TABLE 1.
Age-adjusted odds ratios (ORs) and 95% confidence intervals (CIs) comparing characteristics of ovarian cancer cases and controls stratified by menopausal status, the African American Cancer Epidemiology Study (AACES), 2010–2016.
| Pre-menopausal | Post-menopausal | All Women | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Cases n=175 n (%) |
Controls n=224 n (%) |
OR | 95% CI | Cases n=466 n (%) |
Controls n=528 n (%) |
OR | 95% CI | Cases n=641 n (%) |
Controls n=752 n (%) |
OR | 95% CI | |
| Age | ||||||||||||
| 20 – 29 | 7 (4.0) | 22 (9.8) | 1 (0.2) | 1 (0.2) | 8(1.3) | 23 (3.1) | ||||||
| 30 – 39 | 27 (15.4) | 55 (24.6) | 1 (0.2) | 2 (0.4) | 28(4.4) | 57 (7.6) | ||||||
| 40 – 49 | 90 (51.4) | 99 (44.2) | 20 (4.3) | 22 (4.2) | 110 (17.2) | 121 (16.1) | ||||||
| 50 – 59 | 48 (27.4) | 46 (20.5) | 165 (35.4) | 235 (44.5) | 213 (33.2) | 281 (37.4) | ||||||
| 60 – 69 | 3 (1.7) | 2 (0.9) | 177 (38.0) | 193 (36.6) | 180 (28.1) | 195 (25.9) | ||||||
| 70 – 79 | 102 (21.9) | 75 (14.2) | 102 (15.9) | 75 (10.0) | ||||||||
| Education | ||||||||||||
| ≤ High school graduate | 69 (39.4) | 69 (30.8) | 1.0 | Reference | 214 (45.9) | 211 (40.0) | 1.0 | Reference | 283 (44.2) | 280 (37.3) | 1.0 | |
| Some post-high school | 56 (32.0) | 84 (37.5) | 0.7 | (0.4–1.1) | 149 (32.0) | 182 (34.5) | 0.9 | (0.7–1.2) | 205 (32.0) | 266 (35.4) | 0.8 | (0.7–1.1) |
| ≥ College graduate | 50 (28.6) | 71 (31.7) | 0.8 | (0.4–1.3) | 103 (22.1) | 134 (25.4) | 0.8 | (0.06–1.1) | 153(23.9) | 205 (27.3) | 0.8 | (0.6–1.1) |
| Missing | 1 | 1 | ||||||||||
|
Family history of breast or ovarian cancer (1st degree) |
||||||||||||
| No | 128 (75.3) | 194 (89.0) | 1.0 | Reference | 333 (73.0) | 400 (78.4) | 1.0 | Reference | 461 (73.6) | 594 (81.6) | 1.0 | Reference |
| Yes | 42 (24.7) | 24 (11.0) | 2.4 | (1.3–4.2) | 123 (27.0) | 110 (21.6) | 1.4 | (1.0–1.9) | 165 (26.4) | 134 (18.4) | 1.6 | (1.2–2.1) |
| Missing | 5 | 6 | 10 | 18 | 15 | 24 | ||||||
| Age at menarche (years) | ||||||||||||
| <12 | 46 (26.3) | 57 (25.5) | 1.0 | Reference | 100 (21.5) | 143 (27.1) | 1.0 | Reference | 146 (22.8) | 200 (26.6) | 1.0 | Reference |
| 12 – 13 | 87 (49.7) | 113 (50.5) | 0.9 | (0.6–1.6) | 239 (51.3) | 249 (47.2) | 1.3 | (1.0–1.8) | 326 (50.9) | 362 (48.1) | 1.2 | (0.9–1.6) |
| >13 | 42 (24.0) | 54 (24.1) | 0.9 | (0.5–1.7) | 127 (27.3) | 136 (25.8) | 1.2 | (0.9–1.8) | 169(26.4) | 190 (25.3) | 1.1 | (0.8–1.5) |
| Age at menopause (yrs) | ||||||||||||
| <45 | 85 (19.0) | 110 (21.1) | 1.0 | Reference | ||||||||
| 45 – 49 | 108 (24.2) | 135 (25.9) | 1.0 | (0.7–1.5) | ||||||||
| 50 – 54 | 182 (40.7) | 203 (39.0) | 1.1 | (0.7–1.5) | ||||||||
| >54 | 72 (16.1) | 73 (14.0) | 1.1 | (0.7–1.7) | ||||||||
| Missing | 19 | 7 | ||||||||||
| Tubal Ligation | ||||||||||||
| No | 122 (69.7) | 148 (66.1) | 1.0 | Reference | 297 (64.0) | 301 (57.1) | 1.0 | Reference | 419 (65.6) | 449 (59.8) | 1.0 | Reference |
| Yes | 53 (30.3) | 76 (33.9) | 0.6 | (0.6–1.0) | 167 (36.0) | 226 (42.9) | 0.7 | (0.6–1.0) | 220 (34.4) | 302 (40.2) | 0.7 | (0.6–0.9) |
| Missing | 2 | 1 | 2 | 1 | ||||||||
|
Doctor diagnosed female infertility |
||||||||||||
| No | 127 (85.2) | 195 (92.4) | 1.0 | Reference | 396 (89.0) | 461 (92.2) | 1.0 | Reference | 523 (88.1) | 656 (92.3) | 1.0 | Reference |
| Yes | 22 (14.8) | 16 (7.6) | 2.5 | (1.2–5.3) | 49 (11.0) | 39 (7.8) | 1.6 | (1.0–2.6) | 71 (12.0) | 55 (7.7) | 1.9 | (1.3–2.8) |
| Missing | 26 | 13 | 21 | 28 | 47 | 41 | ||||||
|
Body mass index (kg/m2) 1 year prior to diagnosis/ interview |
0.25 | 0.11 | 0.11 | |||||||||
| <25 | 27 (15.6) | 46 (20.5) | 1.0 | Reference | 63 (13.6) | 95 (18.0) | 1.0 | Reference | 90 (14.1) | 141 (18.8) | 1.0 | Reference |
| 25 – <30 | 47 (27.2) | 48 (21.4) | 1.4 | (0.7–2.8) | 120 (25.9) | 150 (28.5) | 1.2 | (0.8–1.8) | 167 (26.2) | 198 (26.4) | 1.3 | (0.9–1.8) |
| 30 – <35 | 49 (28.3) | 54 (24.1) | 1.4 | (0.7–2.7) | 130 (28.0) | 135 (25.6) | 1.4 | (0.9–2.1) | 179 (28.1) | 189 (25.2) | 1.4 | (1.0–2.0) |
| ≥35 | 50 (28.9) | 76 (33.9) | 0.9 | (0.5–1.7) | 151 (32.5) | 147 (27.9) | 1.6 | (1.1–2.4) | 201 (31.6) | 223 (29.7) | 1.4 | (1.0–2.0) |
| Missing | 2 | 2 | 1 | 4 | 1 | |||||||
| Histology | ||||||||||||
| Serous | 84 (52.2) | 271 (62.9) | <0.0001† | 355(60.0) | ||||||||
| Endometrioid | 31 (19.3) | 44 (10.2) | 75(12.7) | |||||||||
| Mucinous | 17 (10.6) | 13 (3.0) | 30(5.1) | |||||||||
| Clear cell | 5 (3.1) | 8 (1.9) | 13(2.2) | |||||||||
| Other | 24 (14.9) | 95 (22.0) | 119(20.1) | |||||||||
| Missing | 14 | 35 | 48 | |||||||||
p-value comparing histology in pre-menopausal versus postmenopausal cases.
Table 2 presents associations between patterns of oral contraceptive use and ovarian cancer for all women and stratified by menopausal status. The point estimate for ever use of oral contraceptives was 0.7 (95% CI 0.5 – 0. 9) for all women, and was similar for pre- and post-menopausal women, with odds ratios of 0.6 (95% CI 0.4 – 1.2) and 0.7 (95% CI 0.5 – 0.9), respectively, although the association was statistically significant only among the postmenopausal women. Compared to pre-menopausal women, we observed among postmenopausal women that increasing duration of oral contraceptive use was associated with greater reductions in ovarian cancer risk, test for trend p=0.005. Differences in associations by menopausal status were observed in relation to the timing of oral contraceptive use. Among pre-menopausal women, the data suggested that more recent use was associated with greater reduction in risk with significant trends for age at first use (p=0.01), age at last use (p= 0.01) and years since last use (p=0.0009). In contrast, among post-menopausal women there was not a significant association with age at first use, age at last use and years since last use.
TABLE 2.
Odds ratios (ORs) and 95% confidence intervals (CIs) for associations between patterns of oral contraceptive (OC) use and ovarian cancer, stratified by menopausal status, the African American Cancer Epidemiology Study, 2010-2016
| Pre-menopausal | Postmenopausal | All Women | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Cases n=168 n (%) |
Controls n=218 n (%) |
OR* | 95% CI | Cases n=451 n (%) |
Controls n=506 n (%) |
OR* | 95% CI | Cases n=619 n (%) |
Controls 724 n (%) |
OR* | 95% CI | |
| OC use | ||||||||||||
| Non-user† | 38 (22.6) | 37 (17.0) | 1.0 | Referent | 140 (31.0) | 107 (21.2) | 1.0 | Referent0. | 178 (28.8) | 144 (19.9) | 1.0 | Referent |
| User | 130 (77.4) | 181 (83.0) | 0.6 | (0.4–1.2) | 311 (69.0) | 399 (78.9) | 0.7 | (0.5–0.9) | 441 (71.2) | 580 (80.1) | 0.7 | (0.5–0. 9) |
| p for interaction‡ | 0.95 | |||||||||||
| Years of OC use | ||||||||||||
| <1 | 25 (14.9) | 36 (16.5) | 0.7 | (0.3–1.6) | 71 (15.7) | 73 (14.4) | 0.8 | (0.6–1.3) | 96 (15.5) | 109 (15.1) | 0.8 | (0. 6–1.2) |
| 1 – <5 | 48 (28.6) | 69 (31.6) | 0.6 | (0.3–1.3) | 113 (25.1) | 153 (30.2) | 0.7 | (0.5–0.9) | 161 (26.0) | 222 (30.7) | 0.7 | (0.5–0.9) |
| 5 – <10 | 27 (16.1) | 36 (16.5) | 0.7 | (0.3–1.5) | 70 (15.5) | 87 (17.2) | 0.7 | (0.5–1.1) | 97 (15.7) | 123 (17.0) | 0.7 | (0.5–1.0) |
| ≥10 | 30 (17.9) | 40 (18.4) | 0.6 | (0.3–1.2) | 57 (12.6) | 86 (17.0) | 0.6 | (0.4–0.9) | 87 (14.0) | 126 (17.4) | 0.6 | (0.4–0..9) |
| p-trend | 0.18 | 0.005 | 0.004 | |||||||||
| p for interaction | 0.93 | |||||||||||
| Age at first OC use | ||||||||||||
| <20 | 97 (57.7) | 118 (54.1) | 0.8 | (0.4–1.5) | 119 (26.4) | 210 (41.5) | 0.5 | (0.3–0.7) | 216 (34.9) | 328 (45.3) | 0.6 | (0.5–0.9) |
| 20 – 24 | 24 (14.3) | 46 (21.1) | 0.4 | (0.2–0.9) | 133 (29.5) | 137 (27.1) | 0.8 | (0.5–1.1) | 157 (25.4) | 183 (25.3) | 0.7 | (0.5–1.0) |
| ≥25 | 9 (5.4) | 17 (7.8) | 0.4 | (0.1–1.1) | 59 (13.1) | 52 (10.3) | 1.0 | (0.6–1.5) | 68 (11.0) | 69 (9.5) | 0.8 | (0. 6–1.3) |
| p-trend | 0.01 | 0.84 | 0.23 | |||||||||
| p for interaction | 0.01 | |||||||||||
| Age at last OC use | ||||||||||||
| <20 | 33 (19.6) | 39 (17.9) | 1.0 | (0.5–2.2) | 34 (7.5) | 57 (11.3) | 0.5 | (0.3–0.9) | 67 (10.8) | 96 (13.3) | 0.7 | (0.5–1.1) |
| 20 – 24 | 40 (23.8) | 50 (22.9) | 0.7 | (0.3–1.5) | 99 (22.0) | 122 (24.1) | 0.7 | (0.5–1.1) | 139 (22.5) | 172 (23.8) | 0.7 | (0.5–1.0) |
| 25 – 29 | 20 (11.9) | 35 (16.1) | 0.5 | (0.2–1.2) | 69 (15.3) | 87 (17.2) | 0.7 | (0.5–1.1) | 89 (14.4) | 122 (16.9) | 0.7 | (0.5–1.0) |
| ≥30 | 37 (22.0) | 57 (26.2) | 0.5 | (0.2–1.0) | 109 (24.2) | 133 (26.3) | 0.7 | (0.5–1.0) | 146 (23.6) | 190 (26.2) | 0.7 | (0.5–0.9) |
| p-trend | 0.01 | 0.08 | 0.01 | |||||||||
| p for interaction | 0.50 | |||||||||||
| Years since first use | ||||||||||||
| >20 | 104 (61.9) | 119 (54.5) | 0.7 | (0.4–1.4) | 305 (67.6) | 394 (77.9) | 0.67 | (0.5–0.9) | 409 (66.1) | 513 (70.9) | 0.7 | (0.5–0.9) |
| ≤20 | 26 (15.5) | 62 (28.4) | 0.5 | (0.2–1.1) | 6 (1.3) | 5 (1.0) | 1.44 | (0.4–5.3) | 32 (5.2) | 67 (9.3) | 0.6 | (0.3–1.1) |
| p-trend | 0.08 | 0.03 | 0.007 | |||||||||
| p for interaction | 0.24 | |||||||||||
| Years since last use | ||||||||||||
| >20 | 78 (46.4) | 66 (30.3) | 1.3 | (0.6–2.6) | 278 (61.6) | 356 (70.4) | 0.7 | (0.5–0.9) | 356 (57.5) | 422 (58.3) | 0.7 | (0.5–1.0) |
| 10 – 20 | 29 (17.3) | 56 (25.7) | 0.4 | (0.2–0.9) | 26 (5.8) | 32 (6.3) | 0.9 | (0.5–1.6) | 55 (8.9) | 88 (12.2) | 0.6 | (0.4–0.9) |
| <10 | 23 (13.7) | 59 (27.1) | 0.3 | (0.2–0.7) | 7 (1.6) | 11 (2.2) | 0.7 | (0.3–2.1) | 30 (4.9) | 70 (9.7) | 0.5 | (0.3–0.8) |
| p-trend | 0.0009 | 0.10 | 0.002 | |||||||||
| p for interaction | 0.06 | |||||||||||
|
Duration of use/ years since last use |
||||||||||||
| <5 yrs/any | 73 (43.5) | 105 (48.2) | 0.7 | (0.3–1.2) | 184 (40.8) | 226 (44.7) | 0.7 | (0.5–1.0) | 257 (41.5) | 331 (45.7) | 0.7 | (0.5–1.0) |
| ≥5 yrs/ ≥10 yrs ago | 40 (23.8) | 46 (21.1) | 0.7 | (0.3–1.4) | 123 (27.3) | 165 (32.6) | 0.6 | (0.4–0.9) | 163 (26.3) | 211 (29.1) | 0.7 | (0.5–0.9) |
| ≥ 5yrs/<10 yrs ago | 17 (10.1) | 30 (13.8) | 0.5 | (0.2–1.3) | 4 (0.9) | 8 (1.6) | 0.6 | (0.2–2.0) | 21 (3.4) | 38 (5.3) | 0.6 | (0.3–1.2) |
| p for interaction | 0.81 | |||||||||||
OR adjusted for study site, age, family history of breast or ovarian cancer in first degree relative, age at menarche, tubal ligation, body mass index, and number of full-term pregnancies.
Non-users are the referent group for all comparisons
p for interaction based on product term for menopausal status and each individual reproductive variable
Longer duration of oral contraceptive use tends to correlate with more recent use and later age at last use, but it is inappropriate to include terms for both characteristics of exposure in the same model when unexposed women (i.e., non-users) are the reference group.21 Therefore, we ran logistic regression models restricted to oral contraceptive users that examined associations with characteristics of timing of exposure while controlling for duration of oral contraceptive use (Table 3). Among pre-menopausal oral contraceptive users, significant inverse trends were observed for later age at first use (p=0.02), later age at last use (p=0.02) and years since last use (p=0.0003) when controlling for duration of use. In contrast, among post-menopausal oral contraceptive users, women with earlier use were at lower risk than those with later use of oral contraceptives, controlling for duration of use.
TABLE 3.
Odds ratios (ORs) and 95% confidence intervals (CIs) for associations between patterns of oral contraceptive (OC) use and ovarian cancer among OC users, stratified by menopausal status, the African American Cancer Epidemiology Study (AACES), 2010–2016.
| Pre-Menopausal | Postmenopausal | All Women | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Cases n=130 n (%) |
Controls n=181 n (%) |
OR | 95% CI | Cases n=311 n (%) |
Controls n=399 n (%) |
OR | 95% CI | Cases n=441 n (%) |
Controls n=580 n (%) |
OR | 95% CI | |
| Age at first OC use | ||||||||||||
| <20 | 97 (74.6) | 118 (65.2) | 1.0 | Reference | 119 (38.3) | 210 (52.6) | 1.0 | Reference | 216 (48.9) | 328 (56.6) | 1.0 | Reference |
| 20 – 24 | 24 (18.5) | 46 (25.4) | 0.5 | (0.2–0.9) | 133 (42.8) | 137 (34.3) | 1.6 | (1.1–2.3) | 157 (35.6) | 183 (31.6) | 1.1 | (0.8–1.5) |
| ≥25 | 9 (6.9) | 17 (9.4) | 0.4 | (0.2–1.2) | 59 (19.0) | 52 (13.0) | 2.0 | (1.2–3.3) | 68 (15.4) | 69 (11.9) | 1.4 | (0.9–2.1) |
| p-trend | 0.02 | 0.004 | 0.17 | |||||||||
| OR for each 5 year duration of OC use |
0.9 | (0. 7–1.2) | 0.9 | (0.8–1.1) | 0.9 | (0.8–1.1) | ||||||
| p for interaction† | 0.003 | |||||||||||
| Age at last OC use | ||||||||||||
| <20 | 33 (25.4) | 39 (21.6) | 1.0 | Reference | 34 (10.9) | 57 (14.3) | 1.0 | Reference | 67 (15.2) | 96 (16.6) | 1.0 | Reference |
| 20 – 24 | 40 (30.8) | 50 (27.6) | 0.6 | (0.3–1.3) | 99 (31.8) | 122 (30.6) | 1.5 | (0.9–2.5) | 139 (31.5) | 172 (29.7) | 1.1 | (0.7–1.6) |
| 25 – 29 | 20 (15.4) | 35 (19.3) | 0.4 | (0.2–1.0) | 69 (22.2) | 87 (21.8) | 1.7 | (0.9–3.1) | 89 (20.2) | 122 (21.0) | 1.1 | (0.7–1.7) |
| ≥30 | 37 (28.5) | 57 (31.5) | 0.3 | (0.1–0.8) | 109 (35.1) | 133 (33.3) | 1.9 | (1.0–3.6) | 146 (33.1) | 190 (32.8) | 1.1 | (0.7–1.8) |
| p-trend | 0.02 | 0.09 | 0.79 | |||||||||
| OR for each 5 year duration of OC use |
1.2 | (0.9–1.8) | 0.8 | (0. 7–1.0) | 0.91 | (0.8–1.1) | ||||||
| p for interaction | 0.38 | |||||||||||
| Years since first use | ||||||||||||
| >20 | 104 (80.0) | 119 (65.8) | 1.0 | Reference | 305 (98.1) | 394 (98.7) | 409 (92.7) | 513 (88.5) | 1.0 | Reference | ||
| ≤20 | 26 (20.0) | 62 (34.3) | 0.7 | (0.3–1.6) | 6 (1.9) | 5 (1.3) | 32 (7.3) | 67 (11.6) | 0.9 | (0.5–1.6) | ||
| OR for each 5 year duration of OC use |
0.9 | (0.7–1.2) | 0.9 | (0.8–1.1) | ||||||||
| p for interaction | 0.08 | |||||||||||
| Years since last use | ||||||||||||
| >20 | 78 (60.0) | 66 (36.5) | 1.0 | Reference | 278 (89.4) | 356 (89.2) | 1.0 | Reference | 356 (80.7) | 422 (72.7) | 1.0 | Reference |
| 10 – 20 | 29 (22.3) | 56 (30.9) | 0.2 | (0.1–0.5) | 26 (8.4) | 32 (8.0) | 1.6 | (0.8–2.9) | 55 (12.5) | 88 (15.2) | 0.9 | (0.6–1.4) |
| <10 | 23 (17.7) | 59 (32.6) | 0.2 | (0.1–0.4) | 7 (2.3) | 11 (2.8) | 1.4 | (0.5–4.1) | 30 (6.8) | 70 (12.1) | 0.7 | (0.4–1.3) |
| p-trend | 0.0003 | 0.20 | 0.23 | |||||||||
| OR for each 5 year duration of OC use |
1.3 | (0.9–1.8) | 0.9 | (0.7–1.0) | 1.0 | (0.8–1.1) | ||||||
| p for interaction | 0.03 | |||||||||||
Each model included the term for age or timing of OC use (age at first use, age at last use, years since first use or years since last use) and duration of OC use, site, age, family history of breast or ovarian cancer, age at menarche, tubal ligation, body mass index, and number of full-term pregnancies.
p for interaction based on product term for menopausal status and each individual reproductive variable
The effects of pregnancy characteristics on ovarian cancer risk are presented in Table 4. The OR for ovarian cancer was significantly reduced among those who reported at least one full-term pregnancy (OR=0.5, 95% CI 0.3–0.6). The inverse association was stronger for pre-menopausal (OR=0.3, 95% CI 0.2 – 0.5) versus post-menopausal women (OR=0.5, 95% CI 0.4–0.8), with significant trends for decreasing risk with increasing number of pregnancies in pre-menopausal women. Analyses of the timing of pregnancy suggested that later pregnancies, defined either by age at last pregnancy or time since last pregnancy, were associated with reduced ovarian cancer risk. Strong inverse associations were observed for later age at last pregnancy and years since last pregnancy among pre-menopausal women. Among post-menopausal women, no significant trends were observed for these measures, although the smallest odds ratios were observed for more recent pregnancies. A composite variable combining the number of pregnancies and time since last pregnancy showed the smallest odds ratio for pre-menopausal women with 3 or more pregnancies and less than 10 years since the last pregnancy (OR=0.1, 95% CI 0.04–0.4). Duration of breastfeeding was inversely associated with risk among pre-menopausal women whereas no significant trend was observed among postmenopausal women.
TABLE 4.
Odds ratios (ORs) and 95% confidence intervals (CIs) for associations between pregnancy characteristics and ovarian cancer, stratified by menopausal status, the African American Cancer Epidemiology Study (AACES), 2010–2015
| Pre-Menopausal | Postmenopausal | All Women | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Cases n=168 n (%) |
Controls n=218 n (%) |
OR* | 95% CI | Cases n=452 n (%) |
Controls n=507 n (%) |
OR | 95% CI | Cases n=620 n (%) |
Controls n=725 n (%) |
OR | 95% CI | |
| Full-term pregnancy | ||||||||||||
| Never† | 60 (35.7) | 34 (15.6) | 1.0 | Reference | 101 (22.4) | 71 (14.0) | 1.0 | Reference | 161 (26.2) | 105 (14.5) | 1.0 | Reference |
| Ever | 108 (64.3) | 184 (84.4) | 0.3 | (0.2–0.5) | 351 (77.7) | 436 (86.0) | 0.5 | (0.4–0.8) | 459 (74.0) | 620 (85.5) | 0.5 | (0.3–0.6) |
| p for interaction‡ | 0.26 | |||||||||||
| Full-term pregnancies | ||||||||||||
| 1 | 31 (18.5) | 43 (19.7) | 0.4 | (0.2–0.9) | 73 (16.2) | 86 (17.0) | 0.6 | (0.4–1.0) | 104 (16.8) | 129 (17.8) | 0.6 | (0.4–0.8) |
| 2 | 38 (22.6) | 65 (29.8) | 0.2 | (0.1–0.5) | 95 (21.0) | 122 (24.1) | 0.6 | (0.4–0.9) | 133 (21.5) | 187 (25.8) | 0.5 | (0.3–0.7) |
| 3 | 24 (14.3) | 42 (19.3) | 0.2 | (0.1–0.5) | 84 (18.6) | 119 (23.5) | 0.5 | (0.3–0. 8) | 108 (17.4) | 161 (22.2) | 0.4 | (0.3–0.6) |
| >3 | 15 (8.9) | 34 (15.6) | 0.2 | (0.1–0.4) | 99 (21.9) | 109 (21.5) | 0.5 | (0.3–0.8) | 114 (18.4) | 143 (19.7) | 0.4 | (0.3–0.6) |
| p-trend | 0.04 | 0.61 | <0.08 | |||||||||
| p for interaction | 0.56 | |||||||||||
| Age at first pregnancy | ||||||||||||
| <20 | 49 (29.3) | 64 (29.4) | 0.4 | (0.2–0.9) | 164 (36.3) | 201 (39.6) | 0.5 | (0.4–0.8) | 213 (34.4) | 265 (36.6) | 0.5 | (0.4–0.7) |
| 20 – 24 | 24 (14.4) | 54 (24.8) | 0.2 | (0.1–0.5) | 128 (28.3) | 135 (26.6) | 0.7 | (0.4–1.0) | 152 (24.6) | 189 (26.1) | 0.5 | (0.4–0.7) |
| 25 – 29 | 18 (10.8) | 31 (14.2) | 0.3 | (0.1–0.6) | 34 (7.5) | 57 (11.2) | 0.5 | (0.3–0.8) | 52 (8.4) | 88 (12.1) | 0.4 | (0.3–0.7) |
| >29 | 16 (9.6) | 35 (16.1) | 0.2 | (0.1–0. 5) | 25 (5.5) | 43 (8.5) | 0.4 | (0.2–0. 8) | 41 (6.6) | 78 (10.8) | 0.4 | (0.2–0.6) |
| Missing | 1 | 1 | ||||||||||
| p-trend | 0.04 | 0.36 | 0.15 | |||||||||
| p for interaction | 0.21 | |||||||||||
| Age at last pregnancy | ||||||||||||
| <25 | 35 (21.0) | 54 (24.8) | 0.4 | (0.2–0.7) | 126 (27.9) | 131 (25.8) | 0.7 | (0.4–1.0) | 161 (26.0) | 185 (25.5) | 0.6 | (0.4–0.8) |
| 25 – 29 | 29 (17.4) | 42 (19.3) | 0.3 | (0.2–0.7) | 99 (21.9) | 132 (26.0) | 0.5 | (0.3–0. 8) | 128 (20.7) | 174 (24.0) | 0.5 | (0.3–0.7) |
| 30 – 34 | 28 (16.8) | 47 (21.6) | 0.3 | (0.1–0.6) | 80 (17.7) | 100 (19.7) | 0.5 | (0.3–0.8) | 108 (17.5) | 147 (20.3) | 0.5 | (0.3–0.7) |
| >34 | 15 (9.0) | 41 (18.8) | 0.1 | (0.1–0.3) | 46 (10.2) | 73 (14.4) | 0.4 | (0.3–0.7) | 61 (9.9) | 114 (15.7) | 0.3 | (0.2–0.5) |
| Missing | 1 | 1 | ||||||||||
| p-trend | 0.01 | 0.10 | 0.01 | |||||||||
| p for interaction | 0.53 | |||||||||||
|
Years since first pregnancy |
||||||||||||
| >20 | 81 (48.5) | 82 (37.6) | 0.5 | (0.3–1.1) | 344 (76.1) | 420 (82.8) | 0.6 | (0.4–0.8) | 425 (68.7) | 502 (69.2) | 0.6 | (0.4–0.8) |
| 10 – 20 | 21 (12.6) | 73 (33.5) | 0.2 | (0.08–0.3) | 6 (1.3) | 16 (3. 2) | 0.3 | (0.1–0.9) | 27 (4.4) | 89 (12.3) | 0.3 | (0.2–0.4) |
| <10 | 5 (3.0) | 29 (13.3) | 0.1 | (0.04–0.4) | 1 | 0 | 6 (1.0) | 29 (4.0) | 0.2 | (0.08–0.5) | ||
| Missing | 1 | 1 | ||||||||||
| p-trend | 0.17 | 0.63 | 0.06 | |||||||||
| p for interaction | 0.20 | |||||||||||
|
Years since last pregnancy |
||||||||||||
| >20 | 55 (32.9) | 37 (17.0) | 0.9 | (0.4–2.1) | 328 (71.2) | 382 (75.4) | 0.6 | (0.4–0.9) | 383 (61.9) | 419 (57.8) | 0.6 | (0.4–0.8) |
| 10 – 20 | 38 (22.8) | 83 (38.1) | 0.2 | (0.1–0.4) | 22 (5.8) | 51 (10.1) | 0.4 | (0.2–0.8) | 60 (9.7) | 134 (18.5) | 0.3 | (0.2–0.5) |
| < 10 | 14 (8.4) | 64 (29.4) | 0.2 | (0.07–0.3) | 1 (0.2) | 3 (0.6) | 15 (2.4) | 67 (9.2) | 0.2 | (0.1–0.4) | ||
| Missing | 1 | 1 | ||||||||||
| p-trend | 0.02 | 0.26 | 0.01 | |||||||||
| p for interaction | 0.20 | |||||||||||
|
# of pregnancies/ yrs since last pregnancy |
||||||||||||
| <3, ≥10 yrs ago | 59 (35.3) | 72 (33.0) | 0.4 | (0.24–0.7) | 167 (36.6) | 208 (41.0) | 0.6 | (0.4–0.9) | 226 (36.5) | 280 (38.6) | 0.5 | (0.4–0.7) |
| <3, <10 yrs ago | 9 (5.4) | 36 (16.5) | 0.2 | (0.08–0.5) | 1 | 0 | 10 (1.6) | 36 (5.0) | 0.3 | (0.1–0. 7) | ||
| ≤3, ≥10 yrs ago | 34 (20.4) | 48 (22.0) | 0.3 | (0.1–0.7) | 183 (40.4) | 225 (44.4) | 0.5 | (0.3–0.8) | 217 (35.1) | 273 (37.7) | 0.5 | (0.3–0.7) |
| ≥3, <10 yrs ago | 5 (3.0) | 28 (12.8) | 0.1 | (0.04–0.4) | 0 | 3 (0.6) | 5 (0.8) | 31 (4.3) | 0.2 | (0.05–0.4) | ||
| Missing | 1 | 1 | ||||||||||
| Breastfeeding** | ||||||||||||
| 0 months | 79 (47.0) | 99 (45.4) | 0.4 | (0.2–0.8) | 242 (53.5) | 286 (56.4) | 0.6 | (0.4–0.8) | 321 (51.8) | 385 (53.1) | 0.5 | (0.4–0.7) |
| <6 months | 17 (10.1) | 39 (17.9) | 0.3 | (0.1–0.6) | 51 (11.3) | 70 (13.8) | 0.5 | (0.3–0.8) | 68 (11.0) | 109 (15.0) | 0.4 | (0.3–0.6) |
| 6 – 12 months | 7 (4.2) | 18 (8.3) | 0.2 | (0.07–0.6) | 28 (6.2) | 34 (6.7) | 0.6 | (0.3–1.1) | 35 (5.7) | 52 (7.2) | 0. 5 | (0.3–0.8) |
| >12 months | 5 (3.0) | 28 (12.8) | 0.05 | (0.02–0.2) | 30 (6.6) | 46 (9.1) | 0.5 | (0.3–0.8) | 35 (5.7) | 74 (10.2) | 0.3 | (0.2–0.5) |
| p-trend | <0.001 | 0.32 | 0.02 | |||||||||
| p for interaction | 0.10 | |||||||||||
OR adjusted for study site, age, family history of breast or ovarian cancer in first degree relative, age at menarche, tubal ligation, body mass index, and duration of oral contraceptive use.
Never pregnant women are referent category for all comparisons
p for interaction based on product term for menopausal status and each individual reproductive variable
Duration of breastfeeding among parous women. Reference group is women with no full-term pregnancy.
Because women with more pregnancies may be more likely to have had more recent pregnancies, we repeated the analyses of the timing of pregnancy restricting the sample to women with at least one full-term pregnancy and included a term for number of pregnancies within each of the models (Table 5). A significant inverse association was observed for each measure of timing of pregnancy (age at first and last pregnancy, years since first and last pregnancy) among pre-menopausal women, suggesting that more recent pregnancies were associated with reduced risk even when taking into account the total number of pregnancies. Among post-menopausal women, we observed no significant trends in the associations between timing of pregnancies and risk of ovarian cancer.
TABLE 5.
Odds ratios (ORs) and 95% confidence intervals (CIs) for associations between pregnancy characteristics and ovarian cancer among parous women, stratified by menopausal status, the African American Cancer Epidemiology Study (AACES), 2010–2015
| Pre-Menopausal | Postmenopausal | All | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Cases n=108 n (%) |
Controls n=184 n (%) |
OR* | 95% CI | Cases n=351 n (%) |
Controls n=436 n (%) |
OR | 95% CI | Cases n=459 n (%) |
Controls n=620 n (%) |
OR | 95% CI | |
| Age at first pregnancy | ||||||||||||
| <20 | 49 (45.8) | 64 (34.8) | 1.0 | Reference | 164 (46.7) | 201 (46.1) | 1.0 | Reference | 213 (46.5) | 265 (42.7) | 1.0 | Reference |
| 20 – 24 | 24 (22.4) | 54 (29.4) | 0.4 | (0.2–0. 9) | 128 (36.5) | 135 (31.0) | 1.2 | (0.8–1.7) | 152 (33.2) | 189 (30.5) | 1.0 | (0.7–1.3) |
| 25 – 29 | 18 (16.8) | 31 (16.9) | 0.5 | (0.2–1.2) | 34 (9.7) | 57 (13.1) | 0.8 | (0.5–1.3) | 52 (11.4) | 88 (14.2) | 0.7 | (0.5–1.1) |
| ≥30 | 16 (15.0) | 35 (19.0) | 0.3 | (0.1–0.6) | 25 (7.1) | 43 (9.9) | 0.7 | (0.4–1.2) | 41 (9.0) | 78 (12.6) | 0.6 | (0.4–1.0) |
| Missing | 1 | 1 | ||||||||||
| p-trend | 0.004 | 0.22 | 0.02 | |||||||||
| OR for each additional birth |
0.6 | (0.5–0.9) | 0.9 | (0.8–1.1) | 0.8 | (0.7–1.0) | ||||||
| p for interaction† | 0.18 | |||||||||||
| Age at last pregnancy | ||||||||||||
| <25 | 35 (32.7) | 54 (29.4) | 1.0 | Reference | 126 (35.9) | 131 (30.1) | 1.0 | Reference | 161 (35.2) | 185 (29.8) | 1. 0 | Reference |
| 25 – 29 | 29 (27.1) | 42 (22.8) | 1.0 | (0.5–2.1) | 99 (28.2) | 132 (30.3) | 0.8 | (0.5–1.1) | 128 (28.0) | 174 (28.1) | 0.8 | (0. 6–1.1) |
| 30 – 34 | 28 (26.2) | 47 (25.5) | 1.0 | (0. 5–2.1) | 80 (22.8) | 100 (22.9) | 0.8 | (0.5–1.2) | 108 (23.6) | 147 (23.7) | 0.9 | (0.6–1.2) |
| ≥35 | 15 (14.0) | 41 (22.3) | 0.4 | (0.1–0.9) | 46 (13.1) | 73 (16.7) | 0.7 | (0.4–1.1) | 61 (13.3) | 114 (18.4) | 0.6 | (0.4–0.9) |
| Missing | 1 | 1 | ||||||||||
| p-trend | 0.04 | 0.12 | 0.03 | |||||||||
| OR for each additional birth |
0.8 | (0.6–1.1) | 1.0 | (0.9–1.2) | 0.9 | (0.8–1.1) | ||||||
| p for interaction | 0.49 | |||||||||||
|
Years since first pregnancy |
||||||||||||
| >20 | 81 (75.7) | 82 (46.3) | 1.0 | Reference | 344 (98.0) | 420 (96.3) | 1.0 | Reference | 425 (92.8) | 502 (81.9) | 1.0 | Reference |
| 10 – 20 | 21 (19.6) | 73 (41.2) | 0.3 | (0.2–0.7) | 6 (1.7) | 16 (3.7) | 0.5 | (0.2–1.5) | 27 (5.9) | 89 (14.5) | 0.5 | (0.3–0.8) |
| <10 | 5 (4.7) | 22 (12.4) | 0.3 | (0.08–1.3) | 1 (0.3) | 0 | 6 (1.3) | 22 (3.6) | 0.5 | (0.2–1.4) | ||
| Missing | 1 | 7 | 1 | 7 | ||||||||
| p-trend | 0.009 | 0.008 | ||||||||||
| OR for each additional birth |
0.7 | (0.5–0. 9) | 0.96 | (0.8–1.1) | 0.9 | (0.8–1.0) | ||||||
| p for interaction | 0.11 | |||||||||||
|
Years since last pregnancy |
||||||||||||
| >20 | 54 (51.9) | 36 (21.4) | 1.0 | Reference | 323 (93.4) | 382 (87.6) | 1.0 | 377 (83.8) | 418 (69.2) | 1.0 | Reference | |
| 10 – 20 | 38 (36.5) | 82 (48.8) | 0.3 | (0.1–0.6) | 23 (6.7) | 51 (11.7) | 0.8 | (0.4–1.4) | 61 (13.6) | 133 (22.0) | 0.6 | (0.4–0.9) |
| < 10 | 12 (11.5) | 50 (29.8) | 0.2 | (0.08–0.7) | 0 | 3 (0.8) | 12 (2.7) | 53 (8.8) | 0.3 | (0.2–0.7) | ||
| Missing | 2 | 13 | 2 | 13 | ||||||||
| p-trend | 0.01 | 0.002 | ||||||||||
| OR for each additional birth |
0.8 | (0.6–1.1) | 1.0 | (0.8–1.1) | 0.9 | (0.8–1.1) | ||||||
| p for interaction | 0.06 | |||||||||||
Models included terms for pregnancy-related characteristic (i.e., age at first pregnancy, age at last pregnancy, years since first pregnancy, years since last pregnancy or breastfeeding), parity, age, race, family history of breast or ovarian cancer, age at menarche, tubal ligation, body mass index, oral contraceptive (OC) duration and site. Parity was 1 degree of freedom coded 1-2-3-4, with 4+ included as 4.
p for interaction based on product term for menopausal status and each individual reproductive variable
Discussion
The associations between reproductive characteristics and ovarian cancer that we evaluated in the present study have previously been well-established in studies of predominantly White women. In this largest and most thorough investigation yet of these associations in African American women, our analyses showed that oral contraceptive use and parity are inversely associated with ovarian cancer in African American women. This clearly documents for the first time what would be expected, namely that oral contraceptive use and parity are strongly inversely associated with ovarian cancer risk in both African American and White women.
Previous studies that reported on associations between reproductive characteristics and ovarian cancer in African American women were limited in their ability to examine characteristics of pregnancy or oral contraceptive use because of their small sample sizes (number of cases ranging from 84 to 143). 16–19 While each of the studies reported that parity and oral contraceptive use were inversely associated with ovarian cancer, none of them examined the effect of the timing of the exposures or whether associations differed by age or menopausal status. With the larger sample size in the current study, we were able to demonstrate, as has been shown in several studies of predominantly White populations, that the timing of the exposures appears to influence the associations, with stronger inverse associations with more recent pregnancies or oral contraceptive use as compared to exposures that occurred earlier in life.
Our study enrolled only African American women, so comparisons with findings in White women are necessarily between rather than within studies. There are some differences between our study and many of the studies of predominantly White women, specifically in regard to geographic location. Although our study recruited across a broad geographic region, a large proportion of the study population was from Southern states, reflecting areas with a higher percentage of African Americans in their population. Despite some differences in the geographic location from which study participants were recruited, the women in the present study should be fairly representative of African American women and comparisons with White women would be reasonable.
The overall association with oral contraceptive use was similar for pre and post-menopausal women, but differences in associations by menopausal status were noted when examining timing of exposure. Use later in life, whether measured by age at last use or years since last use, was associated with stronger inverse associations among pre-menopausal women, whereas no such trend was noted among post-menopausal women. Pregnancy related characteristics were more strongly associated with pre-menopausal than postmenopausal disease, and inverse associations appeared stronger for pregnancies at older ages or more recent pregnancies. Because pre-menopausal women have had their pregnancies more recently than post-menopausal women, the stronger association with pregnancy characteristics would be expected.
Multiple theories for the genesis of ovarian cancer have been advanced, which variously posit that ovarian carcinogenesis is linked to incessant ovulation, inflammation or levels of gonadotropins or progesterone.22–27 With increasing recognition that epithelial ovarian cancer is comprised of distinct subtypes and that the fallopian tube may be the cell of origin for many ovarian cancers, there has been a re-examination and modification of these theories. 28–30 For example, Fathalla originally hypothesized that the rupture and subsequent repair of the ovarian surface epithelium with each ovulation led to the development of inclusion cysts that underwent malignant transformation.22 This mechanism has been discounted by more recent investigations which have provided evidence that ovarian cancers probably are not derived from surface epithelium cells.29 An alternative hypothesis is that epithelial cells from the fimbria of the fallopian tube, which are in close contact with the ovaries, may be dislodged and implant in the ovary when the surface epithelium of the ovary is disrupted during ovulation. 29 The inflammatory response that accompanies ovulation may further contribute to ovarian carcinogenesis.30 Thus, while the carcinogenic mechanism that was originally proposed for incessant ovulation seems less likely, ovulation remains as a key element in the cascade of events leading to ovarian cancer.
Although biological and epidemiologic evidence continues to point to the role of ovulation in ovarian cancer etiology, it has been noted that the reported reduction in risk associated with pregnancies or oral contraceptive use is greater than what can be explained simply by the reduction in number of ovulatory cycles.24 Our findings that the timing of the exposures appears to modify the reduction in risk corroborates that some other aspects of pregnancy or oral contraceptive use beyond their effect on ovulation play into ovarian cancer risk reduction.
One of the alternative theories of ovarian cancer carcinogenesis focuses on the role of progesterone,26 which is present in higher levels during pregnancy and is a component of oral contraceptive pills. Progesterone has been shown in experimental systems to induce apoptosis and have a “clearing” effect on transformed cells.26,27,31 This mechanism suggests that later or more recent exposure to progesterone would have a stronger protective effect for ovarian cancer. Our observations of greater reductions in risk associated with more recent pregnancies or oral contraceptive use, particularly for pre-menopausal women, are consistent with this hypothesis.
Our data suggest that pregnancy and oral contraceptive use have both long-term and shorter term effects on the risk for ovarian cancer. The observation of reduced risk among parous women or oral contraceptive users regardless of age is indicative of a persistent, long-term effect of these exposures. The observations that women with more recent pregnancies or oral contraceptive use had greater reductions in risk than women with earlier exposure, especially among pre-menopausal women, suggest that these exposures also have effects that diminish over time. This effect of timing of the exposures was less prominent among the postmenopausal women, which is not surprising considering that most of these women had not been pregnant or used oral contraceptives in more than 20 years.
We did observe a difference in the distribution of histologic types between pre- and post-menopausal cases, with endometrioid and mucinous sub-types occurring more frequently in the younger women. While the small sample sizes for individual histologic subtypes within our study precluded doing meaningful analyses stratified by histology, it is unlikely that this was an explanation for the observed differences between pre- and post-menopausal women. Studies that have examined risk factors for ovarian cancer by histologic sub-type have tended to find similar associations with oral contraceptives and pregnancy across all histologic sub-types, with the possible exception of a weaker association between OC use and mucinous cancers.32–35
A limitation of our analyses is that it is impossible to disentangle the effects of age and menopausal status. Although our analyses were stratified by menopausal status, it cannot be discerned whether the observed differences reflect the endogenous hormonal environment or simply that post-menopausal women were older. Because pregnancies and oral contraceptive use occur only in pre-menopausal women (with rare exceptions for use of oral contraceptives in post-menopausal women), postmenopausal women would have experienced these exposures in the more distant past than pre-menopausal women. The suggestion of stronger associations among pre-menopausal women could reflect the time since exposure rather than differences in the hormonal environment. When analyses were repeated stratifying by age (<50 vs. >50 years), the results were very similar to those obtained when stratifying by menopausal status (data not shown). A related limitation is that the sample size of pre-menopausal women was relatively modest, therefore few of the tests for interaction by menopausal status were statistically significant. A much larger sample size than the one in the present study would be required to determine whether the associations are more strongly related to age or to menopausal status.
An additional potential limitation of the study is the possible selection bias that could result from non-response among both cases and controls. The proportion of women with ovarian cancer that could not be contacted because they were deceased (~17%) is much higher than what has been reported in studies of predominantly white women, suggesting that our case group was not entirely representative of African American women with ovarian cancer. Despite the relatively low response rates among both cases and controls, we did observe the expected associations with most established ovarian cancer risk factors, suggesting that selection bias was not likely a major problem.
The associations between ovarian cancer and oral contraceptive use and parity have been well-established for some time, but the unique contribution of this study is that it is the first to examine in detail the effects of these reproductive characteristics on ovarian cancer risk in African American women. African American women differ from White women with respect to many of these characteristics, including having more pregnancies, earlier age at first pregnancies, and less use of oral contraceptives. Although the characteristics of these exposures differ by race, we showed associations with oral contraceptives and parity that are generally similar to those that have been reported in White women. A possible exception is that the estimates associated with parity were stronger for the pre-menopausal African American women than what has been reported in most other studies, which may reflect a higher average number of pregnancies. The higher average number of pregnancies also may contribute to the lower overall incidence of epithelial ovarian cancer among African American women. In contrast, breastfeeding and oral contraceptive use, while also showing inverse associations with ovarian cancer, have a lower prevalence among African American women. Further evaluation of these associations in studies directly comparing African American and White women taking into account the magnitude of the associations as well as the prevalence of reproductive and other risk factors may provide further insight into reasons for racial differences in ovarian cancer incidence.
Conclusion
The present study clearly documents for the first time that pregnancy, breastfeeding and oral contraceptive use are associated with a substantially reduced risk for ovarian cancer in African American women. These data now provide the evidence-base to support what previously could only be assumed based on data from other populations, namely that these reproductive characteristics are at least as strongly inversely associated with ovarian cancer risk in African American women as they are in women of European ancestry. The effects appear to be affected by the timing of the exposures, especially among pre-menopausal women, suggesting that these factors have both short and long-term effects, which may be mediated by different mechanisms.
Acknowledgments
The AACES study was funded by National Cancer Institute (NCI) Grant (CA142081). Additional support was provided by the Metropolitan Detroit Cancer Surveillance System (MDCSS) with federal funds from the NCI under Contract No HHSN261201000028C and the Epidemiology Research Core, supported in part by NCI Center Grant (P30CA22453) to the Karmanos Cancer Institute, Wayne State University School of Medicine.
List of Abbreviations
- AACES
African American Cancer Epidemiology Study
- CATI
computer –assisted telephone interview
- CI
confidence interval
- OR
odds ratio
- SEER
Surveillance Epidemiology and End Results
Footnotes
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
References
- 1.Howlader NNA, Krapcho M, Garshell J, Miller D, Altekruse SF, Kosary CL, Yu M, Ruhl J, Tatalovich Z, Mariotto A, Lewis DR, Chen HS, Feuer EJ, Cronin KA, editors. SEER Cancer Statistics Review, 1975–2012. Bethesda, MD: Institute NC, ed.; 2015. [Google Scholar]
- 2.Permuth-Wey JST. Epidemiology of ovarian cancer. In: MV, editor. Methods of Molecular Biology, Cancer Epidemiology. Vol. 472. Totowa, NJ: Humana Press; 2009. pp. 413–437. [DOI] [PubMed] [Google Scholar]
- 3.Moorman PG, Calingaert B, Palmieri RT, et al. Hormonal risk factors for ovarian cancer in premenopausal and postmenopausal women. Am J Epidemiol. 2008;167(9):1059–1069. doi: 10.1093/aje/kwn006. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Tung KH, Wilkens LR, Wu AH, et al. Effect of anovulation factors on pre- and postmenopausal ovarian cancer risk: revisiting the incessant ovulation hypothesis. Am J Epidemiol. 2005;161(4):321–329. doi: 10.1093/aje/kwi046. [DOI] [PubMed] [Google Scholar]
- 5.Whiteman DC, Siskind V, Purdie DM, Green AC. Timing of pregnancy and the risk of epithelial ovarian cancer. Cancer Epidemiol Biomarkers Prev. 2003;12(1):42–46. [PubMed] [Google Scholar]
- 6.Titus-Ernstoff L, Perez K, Cramer DW, Harlow BL, Baron JA, Greenberg ER. Menstrual and reproductive factors in relation to ovarian cancer risk. Br J Cancer. 2001;84(5):714–721. doi: 10.1054/bjoc.2000.1596. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Kuper H, Cramer DW, Titus-Ernstoff L. Risk of ovarian cancer in the United States in relation to anthropometric measures: does the association depend on menopausal status? Cancer Causes Control. 2002;13(5):455–463. doi: 10.1023/a:1015751105039. [DOI] [PubMed] [Google Scholar]
- 8.Cooper GS, Schildkraut JM, Whittemore AS, Marchbanks PA. Pregnancy recency and risk of ovarian cancer. Cancer Causes Control. 1999;10(5):397–402. doi: 10.1023/a:1008960520316. [DOI] [PubMed] [Google Scholar]
- 9.Beehler GP, Sekhon M, Baker JA, et al. Risk of ovarian cancer associated with BMI varies by menopausal status. J Nutr. 2006;136(11):2881–2886. doi: 10.1093/jn/136.11.2881. [DOI] [PubMed] [Google Scholar]
- 10.Albrektsen G, Heuch I, Kvale G. Reproductive factors and incidence of epithelial ovarian cancer: a Norwegian prospective study. Cancer Causes Control. 1996;7(4):421–427. doi: 10.1007/BF00052668. [DOI] [PubMed] [Google Scholar]
- 11.Fairfield KM, Willett WC, Rosner BA, Manson JE, Speizer FE, Hankinson SE. Obesity, weight gain, and ovarian cancer. Obstet Gynecol. 2002;100(2):288–296. doi: 10.1016/s0029-7844(02)02053-7. [DOI] [PubMed] [Google Scholar]
- 12.Gwinn ML, Lee NC, Rhodes PH, Layde PM, Rubin GL. Pregnancy, breast feeding, and oral contraceptives and the risk of epithelial ovarian cancer. J Clin Epidemiol. 1990;43(6):559–568. doi: 10.1016/0895-4356(90)90160-q. [DOI] [PubMed] [Google Scholar]
- 13.Martin JA, Hamilton BE, Osterman MJ, Curtin SC, Matthews TJ. Births: final data for 2012. Natl Vital Stat Rep. 2013;62(9):1–68. [PubMed] [Google Scholar]
- 14.Centers for Disease C, Prevention. Progress in increasing breastfeeding and reducing racial/ethnic differences - United States, 2000–2008 births. MMWR. Morbidity and mortality weekly report. 2013;62(5):77–80. [PMC free article] [PubMed] [Google Scholar]
- 15.Daniels K, Mosher WD. Contraceptive methods women have ever used: United States, 1982–2010. Natl Health Stat Report. 2013;(62):1–15. [PubMed] [Google Scholar]
- 16.Moorman PG, Palmieri RT, Akushevich L, Berchuck A, Schildkraut JM. Ovarian Cancer Risk Factors in African-American and White Women. Am J Epidemiol. 2009 doi: 10.1093/aje/kwp176. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Ness RB, Grisso JA, Klapper J, Vergona R. Racial differences in ovarian cancer risk. J Natl Med Assoc. 2000;92(4):176–182. [PMC free article] [PubMed] [Google Scholar]
- 18.John EM, Whittemore AS, Harris R, Itnyre J. Characteristics relating to ovarian cancer risk: collaborative analysis of seven U.S. case-control studies. Epithelial ovarian cancer in black women. Collaborative Ovarian Cancer Group. J Natl Cancer Inst. 1993;85(2):142–147. doi: 10.1093/jnci/85.2.142. [DOI] [PubMed] [Google Scholar]
- 19.Wu AH, Pearce CL, Tseng CC, Pike MC. African Americans and Hispanics Remain at Lower Risk of Ovarian Cancer Than Non-Hispanic Whites after Considering Nongenetic Risk Factors and Oophorectomy Rates. Cancer Epidemiol Biomarkers Prev. 2015;24(7):1094–1100. doi: 10.1158/1055-9965.EPI-15-0023. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Schildkraut JM, Alberg AJ, Bandera EV, et al. A multi-center population-based case-control study of ovarian cancer in African-American women: the African American Cancer Epidemiology Study (AACES) BMC cancer. 2014;14:688. doi: 10.1186/1471-2407-14-688. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.McKnight B, Cook LS, Weiss NS. Logistic regression analysis for more than one characteristic of exposure. Am J Epidemiol. 1999;149(11):984–992. doi: 10.1093/oxfordjournals.aje.a009759. [DOI] [PubMed] [Google Scholar]
- 22.Fathalla MF. Incessant ovulation--a factor in ovarian neoplasia? Lancet. 1971;2(7716):163. doi: 10.1016/s0140-6736(71)92335-x. [DOI] [PubMed] [Google Scholar]
- 23.Ness RB, Cottreau C. Possible role of ovarian epithelial inflammation in ovarian cancer. J Natl Cancer Inst. 1999;91(17):1459–1467. doi: 10.1093/jnci/91.17.1459. [DOI] [PubMed] [Google Scholar]
- 24.Fleming JS, Beaugie CR, Haviv I, Chenevix-Trench G, Tan OL. Incessant ovulation, inflammation and epithelial ovarian carcinogenesis: revisiting old hypotheses. Mol Cell Endocrinol. 2006;247(1–2):4–21. doi: 10.1016/j.mce.2005.09.014. [DOI] [PubMed] [Google Scholar]
- 25.Siskind V, Green A, Bain C, Purdie D. Beyond ovulation: oral contraceptives and epithelial ovarian cancer. Epidemiology. 2000;11(2):106–110. doi: 10.1097/00001648-200003000-00005. [DOI] [PubMed] [Google Scholar]
- 26.Risch HA. Hormonal etiology of epithelial ovarian cancer, with a hypothesis concerning the role of androgens and progesterone. J Natl Cancer Inst. 1998;90(23):1774–1786. doi: 10.1093/jnci/90.23.1774. [DOI] [PubMed] [Google Scholar]
- 27.Lukanova A, Kaaks R. Endogenous hormones and ovarian cancer: epidemiology and current hypotheses. Cancer Epidemiol Biomarkers Prev. 2005;14(1):98–107. [PubMed] [Google Scholar]
- 28.Kurman RJ, Shih Ie M. The origin and pathogenesis of epithelial ovarian cancer: a proposed unifying theory. Am J Surg Pathol. 2010;34(3):433–443. doi: 10.1097/PAS.0b013e3181cf3d79. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Li J, Fadare O, Xiang L, Kong B, Zheng W. Ovarian serous carcinoma: recent concepts on its origin and carcinogenesis. J Hematol Oncol. 2012;5:8. doi: 10.1186/1756-8722-5-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Fathalla MF. Incessant ovulation and ovarian cancer - a hypothesis re-visited. Facts Views Vis Obgyn. 2013;5(4):292–297. [PMC free article] [PubMed] [Google Scholar]
- 31.Nguyen H, Syed V. Progesterone inhibits growth and induces apoptosis in cancer cells through modulation of reactive oxygen species. Gynecological endocrinology : the official journal of the International Society of Gynecological Endocrinology. 2011;27(10):830–836. doi: 10.3109/09513590.2010.538100. [DOI] [PubMed] [Google Scholar]
- 32.Collaborative Group on Epidemiological Studies of Ovarian C. 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(9609):303–314. doi: 10.1016/S0140-6736(08)60167-1. [DOI] [PubMed] [Google Scholar]
- 33.Yang HP, Trabert B, Murphy MA, et al. Ovarian cancer risk factors by histologic subtypes in the NIH-AARP Diet and Health Study. Int J Cancer. 2012;131(4):938–948. doi: 10.1002/ijc.26469. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Tung KH, Goodman MT, Wu AH, et al. Reproductive factors and epithelial ovarian cancer risk by histologic type: a multiethnic case-control study. Am J Epidemiol. 2003;158(7):629–638. doi: 10.1093/aje/kwg177. [DOI] [PubMed] [Google Scholar]
- 35.Modugno F, Ness RB, Wheeler JE. Reproductive risk factors for epithelial ovarian cancer according to histologic type and invasiveness. Ann Epidemiol. 2001;11(8):568–574. doi: 10.1016/s1047-2797(01)00213-7. [DOI] [PubMed] [Google Scholar]