Benign epithelial ovarian tumors constitute a common gynecologic problem frequently requiring surgical treatment to alleviate symptoms and prevent complications. Although these tumors are diagnosed most often in women in their 30s or 40s, they can affect women of all ages and account for about 55% of all treated epithelial ovarian neoplasms.1 The true incidence is, however, unknown. In the United States in 2002, approximately 44,000 female inpatients had a primary discharge diagnosis of a benign ovarian neoplasm2 and total hospital treatment costs for this group are estimated at approximately $US 264 million. Although this figure includes an unknown proportion of nonepithelial tumors, it may underestimate the incidence of benign epithelial tumors, because many can be asymptomatic and remain undiagnosed and untreated.3
Current evidence suggests that benign serous tumors are unlikely to progress to high-grade serous cancer, although a precursor role in borderline ovarian cancer is possible.4,5 In contrast, benign mucinous tumors seem to have the potential to progress through borderline tumors to invasive cancer.4,5 Better understanding of the causes of benign tumors is needed, but only a few small studies have investigated this6–9 and rarely for serous and mucinous benign tumors separately.9 Given the current lack of information regarding this common condition, we have conducted a comprehensive investigation of the risk factors for benign epithelial ovarian tumors, by histologic subtype.
MATERIALS AND METHODS
The study was based primarily in the Australian state of Queensland where women with benign epithelial ovarian tumors were identified in one of two ways. A number were identified by the Australian Ovarian Cancer Study (see below) and the remainder were ascertained through the single public and two major private pathology laboratories servicing Queensland. Between September 2003 and June 2005 the pathology laboratories conducted a monthly record search to identify all women aged 18 to 79 years newly diagnosed with a benign mucinous or serous ovarian tumor. Women who had not previously been recruited for Australian Ovarian Cancer Study were mailed study information and asked to indicate their willingness to be contacted about the study by returning a “permission to contact” form to the pathology laboratory. The details of those women indicating willingness to be contacted were then passed on to the study investigators who formally invited them to participate in the study. If the “permission to contact” form was not returned after two weeks, a second letter was sent.
The pathology laboratories approached 353 women in total. Of these, 74 (21%) did not respond to the mailed information and four were excluded due to poor health (n=2) or language problems (n=2). Among the remainder, further contact was refused by 40 women, and nine did not complete the questionnaire, giving an overall participation rate of 65%. Subsequent to review of the diagnostic pathology reports, a further six of these women were found to have ineligible pathology.
Additional women with benign ovarian tumors were identified through the Australian Ovarian Cancer Study, a national case–control study of ovarian cancer that recruited women between January 2002 and June 2005. Women with suspected ovarian cancer were invited to participate before surgical treatment and any subsequently found to have benign serous or mucinous ovarian tumors were included in the present study. An additional 55 women were identified in Queensland in this way, and eligible women were also included from the other Australian states: New South Wales (n=25), Victoria (n=21), South Australia (n=4), Western Australia (n=8) and Tasmania (n=25).
Two researchers independently abstracted information on site (ovary, fallopian tube, other), histologic subtype, and tumor behavior (benign, borderline, and invasive) from the pathology report of each case woman, and any discrepancies were resolved by consensus. For a sample of 87 women (including women with benign, borderline, and invasive tumors), the pathology reports and full set of diagnostic slides were formally reviewed by one of a group of gynecologic pathologists. The agreement between the results of the formal review and the abstracted data were 97% for tumor site, 98% for tumor behavior and 99% for histologic subtype. Only benign tumors of serous or mucinous subtype considered to have arisen from the ovary were included in the following analyses.
The comparison group for the present study was selected from the group of women who were recruited as controls for the Australian Ovarian Cancer Study. A computerized script was used to randomly select women from the Australian Electoral Roll (enrolment is compulsory in Australia) after frequency matching on age (in 5-year age bands) and state to the case series of women with ovarian cancer. Selected women were mailed an invitation letter and information brochure explaining the study and then, where possible, followed up by telephone. At least five attempts were made to reach each woman; those women who did not have a listed telephone number were mailed a second invitation letter. Because most women with benign tumors were from Queensland and in general were younger than women with ovarian cancer, we used a subset of the Australian Ovarian Cancer Study controls in the current analyses. We used SAS (SAS Institute Inc., Cary, NC) (proc surveyselect) to randomly select up to five Australian Ovarian Cancer Study controls for each woman with a benign epithelial tumor, after stratifying by state and age group (in 10-year age bands). Women were excluded as controls if they had a history of ovarian cancer or bilateral oophorectomy or (as for cases) if they were unable to provide informed consent or complete the study questionnaires due to language difficulties, illness, or mental incapacity. The overall participation rate for Australian Ovarian Cancer Study was 84% among cases and 47% among controls.
All participants gave consent and were asked to complete a health and lifestyle questionnaire, including questions relating to demographic and physical characteristics, family history, personal medical and surgical history, lifestyle habits (including smoking and alcohol consumption), and reproductive and contraceptive histories. Missing information or inconsistencies were clarified during a subsequent telephone interview. A small number of women who did not return the full questionnaire completed an abbreviated questionnaire covering key exposures. Ethics approval was obtained from the Queensland Institute of Medical Research Human Research Ethics Committee and all relevant hospitals.
Effects were estimated by odds ratios (ORs) with 95% confidence intervals (CIs) calculated using unconditional multiple logistic regression to simultaneously adjust for potential confounding factors. Serous and mucinous tumors were initially examined separately, but because many of the associations were similar for serous and mucinous tumors, we have also reported results for the combined tumor group. To investigate linear trends for ordinal variables, the median value for each category was used in a continuous term, and the associated P value was assessed.10 Stratified analyses were undertaken to investigate possible effect modification, and the statistical significance of potential interactions was assessed by including a multiplicative term in the models. Models were adjusted for age, hormonal contraceptive use, parity, education, hysterectomy status, and smoking status. Other potential confounding factors that changed the point estimates by less than 10% were not included in the final models. All statistical analyses were conducted using SAS 9.1 software (SAS Institute Inc.)
Use of combined oral, progestin-only, and injected (depot and implant formulations) contraceptives was combined for the analyses of the effects of hormonal contraceptives. Hormone replacement therapy included estrogen-only and combined estrogen–progestin regimes. For analyses of parity we included pregnancies of 6 or more months' duration; those lasting less than 6 months were considered to be incomplete pregnancies (including miscarriages, ectopic pregnancies, and induced abortions). Only women who had ever had a live birth were included in analyses of breastfeeding, and only those who had had a pregnancy of duration 6 or more months were included in analyses of the effect of age at first birth. The relation with obesity was assessed using measures of body mass index (BMI) calculated from self-reported height, weight at age 20, and weight 1 year before diagnosis (or first contact for controls) by dividing weight in kilograms by the square of height in meters. We used standard BMI categories for analysis (18.5–24.9 kg/m2, “normal”; 25–29.9 kg/m2, “overweight”; 30 kg/m2 or more, “obese”). Risk associated with amount of smoking was assessed by calculating pack-years of smoking (the number of cigarettes smoked per day, multiplied by the number of years smoked, divided by 20). We estimated amount of perineal talc use by multiplying frequency of use by years of use. We excluded all exposures in the 12 months before diagnosis for cases (or first approach for controls) because we considered them unlikely to be associated with tumor causation.
The final case group included 225 women with benign serous tumors, 127 with benign mucinous tumors, and six women with one serous and one mucinous tumor. The latter group were included in analyses of both subtypes. Data from 754 control women were included. To ensure the 83 interstate women with benign ovarian tumors recruited through Australian Ovarian Cancer Study were not materially different from the population-based Queensland sample, we compared the groups across key variables. The interstate women were on average older than those from Queensland (57 years compared with 52 years, P<.001) reflecting the higher referral rate of older women to gynecologic oncology clinics, but the groups did not differ significantly with respect to parity, use of contraceptives, level of education, hysterectomy, tubal ligation, smoking status, or BMI (data not shown) and were therefore combined for analyses.
Women with benign mucinous tumors were, on average, younger than women with benign serous tumors (50 years compared with 55 years, P=.003). The average age of control women was 55 years. More than 95% of women were white, and ethnicity did not vary between the cases and controls.
We examined the relation between lifestyle and personal characteristics and risk of the different tumor types (Table 1). Level of education, as a marker of socioeconomic status, was not significantly related to risk of either tumor type. Smoking status was, however, strongly and significantly associated with the risk of benign tumors, both mucinous and serous (OR 3.3, 95% CI 2.0–5.3 and OR 2.3, 95% CI 1.5–3.5, respectively, for current smokers compared with never smokers). Significant trends of increasing risk with increasing pack-years of smoking were observed for both tumor types, although the test for linear trend suggested a closer association with mucinous tumors (P for trend=.001) than serous tumors (P for trend=.02).
Obesity (BMI 30 or more) at age 20 years and at 1 year before diagnosis was related to an increased risk of both tumor types, but significantly so only for serous tumors (OR 4.4, 95% CI 1.9–10.2 compared with normal BMI at age 20 and OR 1.9, 95% CI 1.3–2.9 compared with normal BMI 1 year ago). Overall, there was a significant trend of increasing risk of serous tumors with increasing BMI at both time points (P=.002) but not for mucinous tumors (P=.1 for BMI 1 year ago, and P=.2 for BMI at age 20). When we stratified by age (50 years or younger compared with more than 50 years), the association between recent obesity and serous tumors was stronger for those 50 years or younger at diagnosis (OR 2.4, 95% CI 1.1–4.9) than among those over 50 (OR 1.6, 95% CI 1.0–2.7), although the difference with age was not statistically significant. Neither age at menarche nor a family history of breast or ovarian cancer nor use of talc in the perineal region were significantly related to risk of either tumor type (Table 1).
Women who reported at least one term pregnancy had a nonsignificant 40% reduction in risk of mucinous tumors and a 38% reduction in risk of serous tumors (Table 2). No clear trend of decreasing risk with increasing numbers of term pregnancies was seen for either subtype, although the odds ratio was less than 1 for all categories of parity. In contrast, increasing numbers of incomplete pregnancies were associated with a significantly increased risk of mucinous tumors (OR 1.8, 95% CI 1.1–3.1 for two or more compared with none, P for trend was .006), but did not seem to be associated with serous tumors. Having a first term pregnancy at less than 20 years of age was associated with a small increase in risk of serous tumors (OR 1.6, 95% CI 1.0–2.4 for age of first birth less than 20 years compared with greater than 20 years), but there was no associated trend for either tumor type. Age at last birth (data not shown) and breastfeeding were not significantly related to risk of serous or mucinous benign tumors.
Ever use of hormonal contraceptives was not associated with risk of either serous (OR 1.3, 95% CI 0.8–2.0) or mucinous benign tumors (OR 1.1, 95% CI 0.6–1.9). Although not statistically significant, there seemed to be somewhat different associations with disease depending upon the duration of use. There was a suggestion that very long-term use (more than 15 years) might decrease the risk of both tumor types compared with no use, (OR 0.7, 95% CI 0.3–1.5 for mucinous tumors and OR 0.6, 95% CI 0.3–1.2 for serous), whereas the odds ratios were generally above unity for shorter durations of use (Table 3). Too few women had used long-acting progestins to independently assess their relation with tumor occurrence, but excluding women who had used these from the analyses did not substantially change the odds ratios associated with hormonal contraceptive use. The use of hormone replacement therapy was inversely associated with risk of serous tumors (OR 0.7, 95% CI 0.5–1.1 for ever compared with never use) but there was no trend of further decrease with longer duration of use (P for trend=.3). Mucinous tumors had no apparent relation with hormone replacement therapy use (OR 0.9, 95% CI 0.5–1.5 for ever compared with never use, P for trend=.2)
We found no evidence that tubal ligation was associated with risk of either mucinous or serous benign tumors, and hysterectomy was unrelated to risk of mucinous tumors (Table 4). However, having a hysterectomy was associated with an almost three-fold increase in risk of benign serous tumors (OR 2.8, 95% CI 1.9–4.0). We also examined this in relation to indication for hysterectomy. If the reason for hysterectomy was a nonhormonal condition such as prolapse or cervical dysplasia, then the risk for serous tumors was not increased (OR 1.1, 95% CI 0.5–2.7). However, if the indication for hysterectomy was a hormonally responsive condition such as endometriosis, menorrhagia, or adenomyosis, then the associated risk for serous tumors increased three-fold (OR 3.0, 95% CI 2.1–4.5).
A self-reported history of endometriosis or fibroids was unrelated to benign tumor risk. However, being diagnosed with polycystic ovary syndrome (PCOS) was associated with nonsignificant increases in risk of both mucinous and serous tumors (OR 1.7, 95% CI 0.6–5.3 for mucinous tumors and OR 2.2, 95% CI 0.9–5.5 for serous tumors). Further adjustment of this association for BMI at age 20 years weakened the association with mucinous tumors (OR 1.1, 95% CI 0.3–3.8), but if anything strengthened the relation with serous tumors (OR 2.3, 95% CI 0.9–5.9). Reported history of abnormal Pap tests, genital warts, genital herpes, pelvic inflammatory disease (Chlamydia), or other sexually transmitted infections were unrelated to either tumor type (data not shown).
The effect of reproductive risk factors on serous and mucinous tumors in our study were similar and largely in keeping with previous studies.7–9,11 Term pregnancy was associated with a decreased risk of both tumor types, but increasing parity did not seem to confer additional protection, and use of hormonal contraceptives and breastfeeding were not associated with risk. This is in stark contrast to invasive epithelial ovarian cancer, which is strongly inversely associated with these exposures.12–14
The association we observed with smoking was more marked for mucinous than serous benign tumors, similar to the associations reported for ovarian cancer.15,16 In addition, we found body mass index, both recent and at age 20, to be more strongly related to serous than mucinous tumors. Hysterectomy and PCOS were also associated with the risk of serous but not mucinous tumors.
Population attributable risk percents estimated using these results suggest that approximately 20% of all benign mucinous tumors could be attributed to smoking and 17% of benign serous tumors could be attributed to obesity. If the incidence of benign epithelial tumors is 44,000 per year in the United States, and assuming that mucinous tumors comprise one third and serous tumors two thirds of those, then our results suggest that approximately 8,000 benign epithelial tumors could be prevented each year if smoking and obesity could be eliminated in the population.
Considering potential sources of error in our results, the response rates among both cases and controls were less than optimal at 65% and 47%, respectively, raising the possibility that women who took part differed systematically from those who did not. To assess this we compared our control group's responses to data from the 2001 Australian National Health Survey (NHS), with a response rate of approximately 90%.17 Distributions of parity and body mass index among our control women were almost identical to those surveyed in the NHS. Prevalence of use of the oral contraceptive in women aged younger than 50 years was approximately 5% higher among our controls than women in the NHS, suggesting that any inverse association with long-term use is even weaker than observed. The age-standardized rate of hysterectomy in our controls was slightly lower than in the NHS (19% compared with 24%), and if true, then the magnitude of the observed risk of serous tumors associated with hysterectomy may have been overestimated, but sensitivity analyses suggested that this could not fully explain the elevated risk. This underestimation could also have masked a stronger inverse association with mucinous tumors. Finally, the prevalence of current smoking was lower among our controls compared with the NHS women (11% compared with 17%). Sensitivity analyses suggest that this difference could account for much of the apparent increase in risk of serous tumors associated with current smoking, but is insufficient to explain the stronger association observed for mucinous tumors (although the true magnitude of the association may be somewhat lower than that observed) (personal communication, N. Pandeya, research scholar, Queensland Institute of Medical Research, August 2006).
An appropriate comparison population for our case group is not available, but given that surgery is curative for this condition, it seems likely that response patterns in this group are not dissimilar to those of unaffected women. Accordingly, higher levels of smoking would be expected in the nonresponders,18,19 whereas distributions of BMI are likely to be similar among responders and nonresponders.18,20 Thus, the suboptimal response is unlikely to distort our main findings, and, in practice, such nondifferential error would mean that the true associations might be stronger than those observed. Although the possibility of confounding cannot entirely be dismissed, it is not likely to be a major source of error here, because results adjusted for key factors differed little from results adjusted for age only.
Finally detection bias would not explain the association between benign tumors and PCOS and hysterectomy because associations were limited to serous tumors and because increased medical surveillance of women with other gynecologic conditions such as abnormal Pap tests and genital herpes yielded no similar significant associations with benign ovarian tumors.
While the results of our study do not suggest a strong link between pregnancy, hormonal contraceptives, or breastfeeding and benign serous tumors, they do suggest a possible causal role for sex steroid hormones. Hysterectomy performed for hormonally responsive conditions was related to increased risk of serous tumors, whereas PCOS and obesity, which have well-documented effects on sex steroids and insulin,21 were also associated with serous tumor occurrence. The observed associations might therefore implicate androgens or insulin in the pathogenesis of these tumors. Because adjustment for BMI did not substantially change the relation between serous tumors and PCOS, our results suggest independent effects of PCOS and obesity.
In conclusion, we have found that parity is associated with a modestly decreased risk of both serous and mucinous benign ovarian tumors, but increasing parity does not further decrease risk. Obesity, PCOS, and hysterectomy were related to increased risk of serous tumors, whereas smoking was strongly associated with the occurrence of mucinous tumors. These results require replication but suggest that along with many other conditions, a reduction in smoking and obesity could help prevent women from developing benign serous and mucinous ovarian tumors and thus could ultimately lessen the effect of these relatively common conditions on modern gynecologic practice.
1. Russell P. Surface epithelial-stromal tumors of the ovary. In: Kurman JR, ed. Blaustein's pathology of the female genital tract. New York (NY): Springer-Verlag, 1994:705–82.
2. Kozak LJ, Owings MF, Hall MJ. National Hospital Discharge Survey: 2002 annual summary with detailed diagnosis and procedure data. Vital Health Stat 13 2005:1–199.
3. Crayford TJ, Campbell S, Bourne TH, Rawson HJ, Collins WP. Benign ovarian cysts and ovarian cancer: a cohort study with implications for screening. Lancet 2000;355:1060–3.
4. Bell DA. Origins and molecular pathology of ovarian cancer. Mod Pathol 2005;18 suppl:S19–32.
5. Jordan S, Green A, Webb P. Benign epithelial ovarian tumours—cancer precursors or markers for ovarian cancer risk? Cancer Causes Control 2006;17:623–32.
6. Vessey M, Metcalfe A, Wells C, McPherson K, Westhoff C, Yeates D. Ovarian neoplasms, functional ovarian cysts, and oral contraceptives. Br Med J (Clin Res Ed) 1987;294:1518–20.
7. Booth M, Beral V, Maconochie N, Carpenter L, Scott C. A case–control study of benign ovarian tumours. J Epidemiol Community Health 1992;46:528–31.
8. Parazzini F, Moroni S, Negri E, La Vecchia C, Mezzopane R, Crosignani PG. Risk factors for seromucinous benign ovarian cysts in northern Italy. J Epidemiol Community Health 1997;51:449–52.
9. Westhoff C, Britton JA, Gammon MD, Wright T, Kelsey JL. Oral contraceptive and benign ovarian tumors. Am J Epidemiol 2000;152:242–6.
10. Breslow NE, Day NE. Statistical methods in cancer research. Volume I—The analysis of case–control studies. IARC Sci Publ 1980:5–338.
11. Britton JA, Westhoff C, Howe G, Gammon MD. Diet and benign ovarian tumors (United States). Cancer Causes Control 2000;11:389–401.
12. Whittemore AS, Harris R, Itnyre J. Characteristics relating to ovarian cancer risk: collaborative analysis of 12 US case-control studies. II. Invasive epithelial ovarian cancers in white women. Collaborative Ovarian Cancer Group. Am J Epidemiol 1992;136:1184–203.
13. Risch HA, Marrett LD, Jain M, Howe GR. Differences in risk factors for epithelial ovarian cancer by histologic type. Results of a case–control study. Am J Epidemiol 1996;144:363–72.
14. Tung KH, Goodman MT, Wu AH, McDuffie K, Wilkens LR, Kolonel LN, et al. Reproductive factors and epithelial ovarian cancer risk by histologic type: a multiethnic case-control study. Am J Epidemiol 2003;158:629–38.
15. Green A, Purdie D, Bain C, Siskind V, Webb PM. Cigarette smoking and risk of epithelial ovarian cancer (Australia). Can Causes Control 2001;12:713–9.
16. Kurian AW, Balise RR, McGuire V, Whittemore AS. Histologic types of epithelial ovarian cancer: have they different risk factors? Gynecol Oncol 2005;96:520–30.
18. Tolonen H, Dobson A, Kulathinal S, WHO MONICA Project. Effect on trend estimates of the difference between survey respondents and non-respondents: results from 27 populations in the WHO MONICA Project. Eur J Epidemiol 2005;20:887–98.
19. Jackson R, Chambless LE, Yang K, Byrne T, Watson R, Folsom A, et al. Differences between respondents and nonrespondents in a multicenter community-based study vary by gender ethnicity. The Atherosclerosis Risk in Communities (ARIC) Study Investigators. J Clin Epidemiol 1996;49:1441–6.
20. Heath AC, Howells W, Kirk KM, Madden PA, Bucholz KK, Nelson EC, et al. Predictors of non-response to a questionnaire survey of a volunteer twin panel: findings from the Australian 1989 twin cohort. Twin Res 2001;4:73–80.
21. Kaaks R, Lukanova A, Kurzer MS. Obesity, endogenous hormones, and endometrial cancer risk: a synthetic review. Cancer Epidemiol Biomarkers Prev 2002;11:1531–43.