Abstract
Purpose
In contrast to other hormonally mediated cancers such as those of the breast, prostate, and testes, almost no data are available regarding the relation of prenatal characteristics and exposures to subsequent ovarian carcinogenesis. In a population-based study of 812 women aged 35-74 years with epithelial ovarian cancer diagnosed in Washington State from 2002-2005 and 1,313 controls, we assessed the relation of such factors to disease risk.
Methods
Information was collected through in-person interviews and logistic regression was used to calculate odds ratios (ORs) and 95% confidence intervals (CIs).
Results
Overall, we noted little evidence that prenatal or birth characteristics including birth weight, birth order, maternal age or in utero exposure to cigarette smoking were associated with risk of epithelial ovarian cancer in adulthood. Among women <55 years of age, risk was reduced among those whose weight at birth was < 5.5 pounds (OR and 95% CI= 0.54, 0.31-0.94) relative to those with birth weight 5.5-9 pounds.
Conclusions
In this study, birth weight was associated with risk of epithelial ovarian cancer only among women under 55 years of age. Prenatal influences might be expected to more substantially influence cancer risk at younger ages. Other reports examining associations of ovarian cancer risk with birth weight or other prenatal characteristics are few and have not examined risk separately according to age at diagnosis, suggesting that additional studies may prove useful.
Keywords: ovarian cancer, epidemiology, prenatal
Introduction
Characteristics such as birth weight, birth order and maternal age have been hypothesized to influence risk of hormonally-mediated cancers, perhaps serving as indicators of in utero exposure to hormones such as estrogen and testosterone (1, 2). While the possible contribution of the intrauterine environment to risk of breast, testicular, prostate, and some other cancers has received attention in multiple studies, almost no reports have examined associations with ovarian cancer. Thus, in a population-based case-control study of women aged 35-74 years, we assessed the relation of prenatal and birth characteristics with risk of epithelial ovarian cancer.
Materials and Methods
The study population and methods have been described (3). Female residents of a thirteen-county area of western Washington State, 35-74 years of age, diagnosed with a primary invasive or borderline epithelial ovarian tumor from 2002 through 2005 were identified through a population-based registry that is part of the Surveillance, Epidemiology, and End Results program of the US National Cancer Institute. Of 1,058 eligible case women identified, 812 (76.7%) were interviewed; of the interviewed cases, 595 had invasive disease. Reasons for not obtaining an interview included: physician refusal (n=23); inability to locate the patient (n=10); patient refusal (n=110) and death (n=103). Among interviewed women, the large majority (73%) of those with borderline tumors had localized disease, while 85% of women with invasive tumors had regional or distant disease, according to the SEER staging system; these proportions were quite similar to the distribution among all eligible women (74 and 86%, respectively).
Controls were selected by random digit dialing (4) using stratified sampling in five-year age categories, one-year calendar intervals and two county strata in a 2:1 ratio to women with invasive epithelial ovarian cancer. For 14,561 (82.0%) of the 17,768 telephone numbers belonging to a residence, we determined whether an eligible (i.e., age- and county-eligible and, if so, with at least one ovary and no prior history of ovarian cancer) woman resided there. Of the 1,561 women identified as eligible controls, 1,313 were interviewed (84.1%). To maximize the likely accuracy of available variables as proxy measures of the prenatal environment, we excluded 40 women who had been adopted and an additional 51 women who were not singleton births, resulting in 778 cases and 1,256 controls for this analysis.
The study was approved by the Institutional Review Board of the Fred Hutchinson Cancer Research Center, and all women provided signed informed consent before participating. Information was obtained during an in-person interview. Prenatal characteristics of the study participants were recorded as described below, with the birth of the study participant designated as the index birth. Women were asked their mother's age at index birth and their mother's number of pregnancies resulting in live births prior to the index birth (henceforth referred to as birth order). In addition, women reported whether the index pregnancy had resulted in a twin or multiple birth. They were asked their birth weight in pounds and ounces; these responses were converted to grams and grouped in 500-gram categories for analysis. When the weight was unknown or could only be partially reported, participants were asked if they weighed less than 5.5 pounds, or 9 pounds or more, at birth. Maternal smoking during the index pregnancy was also recorded.
Odds ratios (ORs) and 95% confidence intervals (CIs) were calculated using unconditional logistic regression. The analyses shown were adjusted for the frequency matching variables of age, year of diagnosis/reference date, and county of residence as well as duration of hormonal contraception and number of full-term births. Other characteristics examined as potentially confounding the associations of interest included: race/ethnicity; cigarette smoking; education; age at menarche; body weight and body mass index (BMI), assessed both at age 30 years and five years before the reference date; tubal ligation; hysterectomy; use of hormone replacement therapy; family history of breast and/or ovarian cancer; and personal history of breast cancer. Additional adjustment for these characteristics did not result in a meaningful change in odds ratio estimates. We also examined potential confounding influences of maternal age at index pregnancy, birth weight, and birth order on each of the other variables. We used polytomous logistic regression for analyses that separated case women according to the degree of invasiveness and/or histologic type of the tumor. Tests for linear trend were done using the maximum likelihood test with the categorical variable of interest entered as a continuous term. STATA was used for analysis (version 9.2; STATA Corporation, College Station, TX).
Results
The majority of cases (89.2%) and controls (90.0%) were non-Hispanic whites. Cases were less likely than controls to have given birth, used hormonal contraception, or to have had a tubal ligation, and family history of ovarian cancer was more commonly reported among cases (data not shown).
Sufficient information to allow assessment of birth weight within 500-gram categories was reported by 579 cases (74.4%) and 930 controls (74.0%). For most of the remaining women, the collected data allowed us to assign birth weight within categories of <5.5, 5.5-9, or >9 pounds (Table 1). Whether or not women with more broadly estimated birth weights were included in the analysis, we observed little evidence that this characteristic was associated with risk in the overall study population. Also, we noted no association with mother's age at index birth, number of births prior to the index birth, or smoking during the index pregnancy.
Table 1.
Risk of epithelial ovarian cancer according to prenatal and birth characteristics among women aged 35-74 years
| Cases
(n= 778) |
Controls
(n=1256) |
|||||
|---|---|---|---|---|---|---|
| n | %1 | n | %1 | OR2 | 95% CI | |
| Birth weight (g) | ||||||
| <2500 | 70 | 12.1 | 123 | 13.2 | 0.87 | 0.61-1.24 |
| 2500-2999 | 124 | 21.4 | 218 | 23.4 | 0.85 | 0.64-1.14 |
| 3000-3499 | 210 | 36.3 | 328 | 35.3 | 1.00 | Ref |
| 3500-3999 | 116 | 20.0 | 171 | 18.4 | 1.02 | 0.75-1.39 |
| ≥4000 | 59 | 10.2 | 90 | 9.7 | 0.97 | 0.66-1.42 |
| missing | 199 | (25.6) | 326 | (26.0) | p trend=0.30 | |
| Birth weight (lbs) | ||||||
| <5.5 | 60 | 7.9 | 110 | 9.0 | 0.89 | 0.63-1.24 |
| 5.5- <9.0 | 640 | 84.5 | 1028 | 84.2 | 1.00 | Ref |
| ≥9.0 | 57 | 7.5 | 83 | 6.8 | 1.06 | 0.74-1.53 |
| missing | 21 | (2.7) | 35 | (2.8) | p trend=0.44 | |
| Mother's age at index birth (yrs) | ||||||
| <20 | 64 | 8.3 | 113 | 9.0 | 0.89 | 0.62-1.28 |
| 20-24 | 222 | 28.7 | 340 | 27.2 | 1.00 | Ref |
| 25-29 | 219 | 28.3 | 378 | 30.2 | 0.88 | 0.69-1.12 |
| 30-34 | 159 | 20.6 | 270 | 21.6 | 0.88 | 0.67-1.15 |
| ≥ 35 | 109 | 14.1 | 149 | 11.9 | 1.07 | 0.78-1.45 |
| missing | 5 | (0.6) | 6 | (0.5) | p trend=0.85 | |
| Births prior to index birth | ||||||
| 0 | 269 | 34.8 | 486 | 38.8 | 1.00 | Ref |
| 1 | 227 | 29.4 | 360 | 28.8 | 1.06 | 0.84-1.34 |
| 2 | 135 | 17.5 | 203 | 16.2 | 1.18 | 0.90-1.55 |
| ≥3 | 142 | 18.4 | 203 | 16.2 | 1.21 | 0.92-1.59 |
| missing | 5 | (0.6) | 4 | (0.3) | p trend=0.12 | |
| Maternal smoking | ||||||
| no | 516 | 72.7 | 846 | 72.9 | 1.00 | Ref |
| yes | 194 | 27.3 | 314 | 27.1 | 0.97 | 0.78-1.21 |
| missing | 68 | (8.7) | 96 | (7.6) | ||
Percentages in parentheses are based on 778 cases and 1256 controls; all others are based on cases and controls with known data for the relevant variable.
Adjusted for age, reference year, county, number of full-term births, and duration of hormonal contraception.
Because we hypothesized that prenatal influences might more substantially influence cancer risk at younger ages, we conducted subanalyses to separately assess risk in women <55 years of age (the approximate midpoint of the age distribution of cases) and women 55 years or older. Among women <55 years, risk of epithelial ovarian cancer was reduced by nearly 50% among women who weighed <5.5 pounds at birth (Table 2). Also, in this age group, a trend of increasing risk with increasing birth weight was observed, both in an analysis limited to women whose birth weight could be placed within a 500-gram category (p trend= 0.03) and when the analysis also included women whose birth weight was estimated in one of three broader categories (p trend= 0.01). The relative risks among women with birth weight <5.5 pounds, relative to 5.5-9 pounds, were 0.29 (95% CI 0.09-0.97) and 0.69 (CI 0.37-1.30) among women aged 35-44 years and 45-54 years, respectively (data not shown). Risk did not vary according to mother's age at index birth, birth order, or maternal smoking during pregnancy. Among women ≥ 55 years, we noted no association with any of the prenatal or birth characteristics studied (Table 3).
Table 2.
Risk of epithelial ovarian cancer according to prenatal and birth characteristics, among women < 55 years of age
| Cases
(n=380) |
Controls
(n=505) |
|||||
|---|---|---|---|---|---|---|
| n | %1 | n | %1 | OR2 | 95% CI | |
| Birth weight (g) | ||||||
| <2500 | 26 | 8.9 | 52 | 13.0 | 0.58 | 0.33-1.02 |
| 2500-2999 | 68 | 23.4 | 105 | 26.3 | 0.79 | 0.52-1.20 |
| 3000-3499 | 104 | 35.7 | 136 | 34.1 | 1.00 | Ref |
| 3500-3999 | 60 | 20.6 | 70 | 17.5 | 1.01 | 0.64-1.60 |
| ≥4000 | 33 | 11.3 | 36 | 9.0 | 1.15 | 0.65-2.02 |
| missing | 89 | (23.4) | 106 | (21.0) | p trend=0.03 | |
| Birth weight (lbs) | ||||||
| <5.5 | 22 | 6.0 | 50 | 10.0 | 0.54 | 0.31-0.94 |
| 5.5- <9.0 | 314 | 85.6 | 417 | 83.7 | 1.00 | Ref |
| ≥9.0 | 31 | 8.4 | 31 | 6.2 | 1.39 | 0.80-2.39 |
| missing | 13 | (3.4) | 7 | (1.4) | p trend=0.01 | |
| Mother's age at index birth (yrs) | ||||||
| <20 | 35 | 9.2 | 53 | 10.5 | 0.85 | 0.50-1.42 |
| 20-24 | 109 | 28.7 | 132 | 26.2 | 1.00 | Ref |
| 25-29 | 108 | 28.4 | 150 | 29.8 | 0.82 | 0.57-1.20 |
| 30-34 | 72 | 18.9 | 106 | 21.0 | 0.76 | 0.50-1.14 |
| ≥ 35 | 56 | 14.7 | 63 | 12.5 | 0.99 | 0.62-1.58 |
| missing | 0 | (0.0) | 1 | (0.2) | p trend=0.76 | |
| Births prior to index birth | ||||||
| 0 | 108 | 28.5 | 154 | 30.6 | 1.00 | Ref |
| 1 | 121 | 31.9 | 152 | 30.2 | 1.18 | 0.82-1.69 |
| 2 | 71 | 18.7 | 97 | 19.3 | 1.10 | 0.73-1.66 |
| ≥3 | 79 | 20.8 | 100 | 19.9 | 1.21 | 0.81-1.82 |
| missing | 1 | (0.3) | 2 | (0.4) | p trend=0.42 | |
| Maternal smoking | ||||||
| no | 239 | 67.3 | 313 | 67.2 | 1.00 | Ref |
| yes | 116 | 32.7 | 153 | 32.8 | 1.02 | 0.75-1.39 |
| missing | 25 | (6.6) | 39 | (7.7) | ||
Percentages in parentheses are based on 380 cases and 505 controls; all others are based on cases and controls with known data for the relevant variable.
Adjusted for age, reference year, county, number of full-term births, and duration of hormonal contraception.
Table 3.
Risk of epithelial ovarian cancer according to prenatal and birth characteristics, among women ≥ 55 years of age
| Cases
(n=398) |
Controls
(n=751) |
|||||
|---|---|---|---|---|---|---|
| n | %1 | n | %1 | OR2 | 95% CI | |
| Birth weight (g) | ||||||
| <2500 | 44 | 15.3 | 71 | 13.4 | 1.17 | 0.74-1.86 |
| 2500-2999 | 56 | 19.4 | 113 | 21.3 | 0.91 | 0.60-1.37 |
| 3000-3499 | 106 | 36.8 | 192 | 36.2 | 1.00 | Ref |
| 3500-3999 | 56 | 19.4 | 101 | 19.0 | 1.04 | 0.68-1.58 |
| ≥4000 | 26 | 9.0 | 54 | 10.2 | 0.81 | 0.47-1.40 |
| missing | 110 | (27.6) | 220 | (29.3) | p trend=0.51 | |
| Birth weight (lbs) | ||||||
| <5.5 | 38 | 9.7 | 60 | 8.3 | 1.24 | 0.80-1.93 |
| 5.5- <9.0 | 326 | 83.6 | 611 | 84.5 | 1.00 | Ref |
| ≥9.0 | 26 | 6.7 | 52 | 7.2 | 0.85 | 0.52-1.41 |
| missing | 8 | (2.0) | 28 | (3.7) | p trend=0.23 | |
| Mother's age at index birth (yrs) | ||||||
| <20 | 29 | 7.4 | 60 | 8.0 | 0.89 | 0.53-1.50 |
| 20-24 | 113 | 28.8 | 208 | 27.9 | 1.00 | Ref |
| 25-29 | 111 | 28.2 | 228 | 30.6 | 0.93 | 0.67-1.30 |
| 30-34 | 87 | 22.1 | 164 | 22.0 | 1.02 | 0.72-1.46 |
| ≥ 35 | 53 | 13.5 | 86 | 11.5 | 1.16 | 0.76-1.77 |
| missing | 5 | (1.3) | 5 | (0.7) | p trend=0.41 | |
| Births prior to index birth | ||||||
| 0 | 161 | 40.9 | 332 | 44.3 | 1.00 | Ref |
| 1 | 106 | 26.9 | 208 | 27.8 | 1.03 | 0.76-1.41 |
| 2 | 64 | 16.2 | 106 | 14.2 | 1.32 | 0.91-1.92 |
| ≥ 3 | 63 | 16.0 | 103 | 13.8 | 1.25 | 0.86-1.82 |
| missing | 4 | (1.0) | 2 | (0.3) | p trend=0.12 | |
| Maternal smoking | ||||||
| no | 277 | 78.0 | 533 | 76.8 | 1.00 | Ref |
| yes | 78 | 22.0 | 161 | 23.2 | 0.93 | 0.68-1.28 |
| missing | 43 | (10.8) | 57 | (7.6) | ||
Percentages in parentheses are based on 398 cases and 751 controls; all others are based on cases and controls with known data for the relevant variable.
Adjusted for age, reference year, county, number of full-term births, and duration of hormonal contraception.
In most cases, additional adjustment of our analyses for other prenatal characteristics had no appreciable impact on the results. Among women under 55 years, adjustment of birth order by maternal age slightly increased the strength of association (e.g., risk among women who were the third or subsequent births= 1.42, 95% CI 0.88-2.28; p trend=0.17); however, wide confidence bounds suggest this may reflect a chance effect.
The reduction in risk among younger women with birth weight <5.5 pounds was evident for both borderline and invasive ovarian tumors (ORs and 95% CIs for weight <5.5 pounds relative to 5.5-<9 pounds among women <55 were 0.46, 0.20-1.06, and 0.59, 0.31-1.11, respectively). While our ability to examine risks among histologic subtypes was limited by small numbers, we observed no clear differences in the association with birth weight or other prenatal characteristics according to tumor histology (serous, mucinous, endometrioid, clear cell, or other epithelial), whether in analyses that included all women, or separately examined women less than 55 years and 55 years of age or older, or further stratified tumors according to the degree of invasiveness (data not shown).
Discussion
Differences in characteristics of cases and controls who did or did not choose to participate might influence the results of this study, as might errors in recall of prenatal and birth characteristics. Little information is available regarding the accuracy of self-reported birth weight or other prenatal exposures. In one study (5), recall of birth weight by study subjects was fairly similar to that reported by their mothers (Spearman correlation coefficient =0.77). Some evidence suggests that the accuracy of recalled birth weight may be greater in younger women (6). While we have no external means of validating recalled birth weight in the current study, a greater proportion of younger than older women (among both cases and controls) were able to provide sufficient information in answer to an open-ended question to allow an examination of birth weight within 500-gram categories, while larger proportions of older cases and controls could only be grouped according to three pre-supplied broad categories of birth weight.
The mechanisms of prenatal influences on cancer risk are not well understood, but may involve in utero hormonal influences or an altered number of cells at risk for later carcinogenesis (2). Some studies of breast and testicular cancer suggest an increased risk among firstborn individuals, possibly reflecting higher bioavailable estrogen due to less sex-hormone binding globulin in nulliparous women (1, 2, 7). Higher birth weight may reflect maternal obesity, which may be associated with a relatively estrogenic in utero environment; or birth weight may serve as a proxy of other aspects of fetal growth, gestation length, or other pregnancy hormone levels such as insulin-like growth factors (1). Reduced fetal growth has been associated with a number of ovarian alterations later in life, including a reduced ovarian fraction of primordial follicles and anovulation in late adolescence (8, 9). In one study (10), mean ovarian volume was reduced by 38% among adolescent girls who were born small for gestational age relative to similarly-aged girls who were not. Also, in a related study (11), girls born small for gestational age had lower average birth weights than girls who were not (mean weights of 2416 and 3301 grams in the two groups, respectively), and exhibited a serum hormonal profile indicative of ovarian hyporesponsiveness to follicle stimulating hormone. It is possible that some of these alterations may have relevance to ovarian cancer risk: e.g., reduced ovarian volume may indicate a lesser number of epithelial cells at risk, or ovarian hormone production may be reduced in ovaries with a smaller stromal compartment.
Almost no epidemiologic evidence exists to support or refute the hypothesis that prenatal factors may be involved in ovarian carcinogenesis. Similar to the current study, no association with maternal smoking was noted in a case-control study conducted in Hawaii (12). In a retrospective cohort analysis of 5,585 women that employed medical-records-based assessments of birth weight and weight at one year of age (13), no association of birth weight with risk of fatal ovarian cancer was observed, while women in the highest category of weight at one year of age were at increased risk (>24 lbs; standardized mortality ratio= 3.6, CI: 1.7, 6.5). In a cohort of 5,346 Swedish women (of whom 89 developed ovarian cancer), birth characteristics standardized for gestational age were examined, with no associations with ovarian cancer risk seen for birth weight, birth length, or head circumference; no data regarding differences in ovarian cancer risk by age at diagnosis were presented (14).
While we observed no clear association of birth weight with ovarian cancer risk overall, among women under 55 years of age, risk was reduced among women with low birth weight. Prenatal influences on ovarian cancer risk might operate more strongly among women who develop this disease at a younger age, as such a pattern has been observed in studies of birth weight and risk of breast cancer (1). Our results may also, to some extent, reflect improved accuracy of recalled birth weight among younger women. Other studies examining this potential association are few and have not examined risk separately according to age at cancer diagnosis. The limited extent of published data, coupled with the findings of our study, suggest that additional studies of prenatal factors and ovarian cancer risk may prove useful.
Acknowledgments
Funding: provided by the National Institutes of Health (RO1 CA87538)
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.Grotmol T, Weiderpass E, Tretli S. Conditions in utero and cancer risk. European Journal of Epidemiol. 2006;21:561–570. doi: 10.1007/s10654-006-9036-7. [DOI] [PubMed] [Google Scholar]
- 2.Ekbom A. Growing evidence that several human cancers may originate in utero. Seminars in Cancer Biology. 1998;8:237–244. doi: 10.1006/scbi.1998.0073. [DOI] [PubMed] [Google Scholar]
- 3.Rossing MA, Cushing-Haugen KL, Wicklund KG, Doherty JA, Weiss NS. Menopausal hormone therapy and risk of epithelial ovarian cancer. Cancer Epidemiol Biomarkers Prev. 2007 doi: 10.1158/1055-9965.EPI-07-0550. in press. [DOI] [PubMed] [Google Scholar]
- 4.Waksberg J. Random digit dialing sampling for case-control studies. In: Gail MH, Benichou J, editors. Encyclopedia of Epidemiologic Methods. New York: John Wiley & Sons; 2000. pp. 749–753. [Google Scholar]
- 5.Michels K, Trichopoulos D, Robins J, et al. Birthweight as a risk factor for breast cancer. Lancet. 1996;348:1542–46. doi: 10.1016/S0140-6736(96)03102-9. [DOI] [PubMed] [Google Scholar]
- 6.Allen DS, Ellison GTH, dos Santos Silva I, De Stavola BL, Fentiman IS. Determinants of the availability and accuracy of self-reported birth weight in middle-aged and elderly women. American Journal of Epidemiology. 2002;155:379–84. doi: 10.1093/aje/155.4.379. [DOI] [PubMed] [Google Scholar]
- 7.Garner MJ, Turner MC, Ghadirian P, Krewski D. Epidemiology of testicular cancer: an overview. Int J Cancer. 2005;116:331–339. doi: 10.1002/ijc.21032. [DOI] [PubMed] [Google Scholar]
- 8.DeBruin J, Dorland M, Bruinse H, Spliet W, Nikkels P, Te Velde E. Fetal growth retardation as a cause of impaired ovarian development. Early Human Dev. 1998;51:39–46. doi: 10.1016/s0378-3782(97)00073-x. [DOI] [PubMed] [Google Scholar]
- 9.Ibanez L, de Zegher F, Potau N. Anovulation after precocious pubarche: early markers and time course in adolescence. J Clin Endocrinol Metab. 1999;84:2691–2695. doi: 10.1210/jcem.84.8.5883. [DOI] [PubMed] [Google Scholar]
- 10.Ibanez L, Potau N, Enriquez G, DeZegher F. Reduced uterine and ovarian size in adolescent girls born small for gestational age. Pediatr Res. 2000;47:575–577. doi: 10.1203/00006450-200005000-00003. [DOI] [PubMed] [Google Scholar]
- 11.Ibanez L, Potau N, De Zegher F. Ovarian hyporesponsiveness to follicle stimulating hormone in adolescent girls born small for gestational age. J Clin Endocrinol Metab. 2000;85:2624–2626. doi: 10.1210/jcem.85.7.6765. [DOI] [PubMed] [Google Scholar]
- 12.Goodman MT, Tung KH. Active and passive tobacco smoking and the risk of borderline and invasive ovarian cancer (United States) Cancer Causes Control. 2003;14:569–77. doi: 10.1023/a:1024828309874. [DOI] [PubMed] [Google Scholar]
- 13.Barker D, Winder P, Osmond C, Phillips D, Sultan H. Weight gain in infancy and cancer of the ovary. Lancet. 1995;345:1087–88. doi: 10.1016/s0140-6736(95)90821-8. [DOI] [PubMed] [Google Scholar]
- 14.McCormack VA, dos SS I, Koupil I, Leon DA, Lithell HO. Birth characteristics and adult cancer incidence: Swedish cohort of over 11,000 men and women. Int J Cancer. 2005;115:611–17. doi: 10.1002/ijc.20915. [DOI] [PubMed] [Google Scholar]