Skip to main content
PMC home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2013 Dec 9.
Published in final edited form as: Endocr Relat Cancer. 2013 Mar 22;20(2):10.1530/ERC-12-0395. doi: 10.1530/ERC-12-0395

Obesity and risk of ovarian cancer subtypes: evidence from the Ovarian Cancer Association Consortium

Catherine M Olsen 1, Christina M Nagle 1, David C Whiteman 1, Roberta Ness 2, Celeste Leigh Pearce 3, Malcolm C Pike 3,4, Mary Anne Rossing 5, Kathryn L Terry 6, Anna H Wu 3; the Australian Cancer Study (Ovarian Cancer)1; Australian Ovarian Cancer Study Group1,7, Harvey A Risch 8, Herbert Yu 9, Jennifer A Doherty 5, Jenny Chang–Claude 10, Rebecca Hein 10, Stefan Nickels 10, Shan Wang–Gohrke 11, Marc T Goodman 12, Michael E Carney 13, Rayna K Matsuno 9, Galina Lurie 9, Kirsten Moysich 14, Susanne K Kjaer 15,16, Allan Jensen 15, Estrid Hogdall 15, Ellen L Goode 17, Brooke L Fridley 17, Robert A Vierkant 17, Melissa C Larson 17, Joellen Schildkraut 18, Cathrine Hoyo 18, Patricia Moorman 18, Rachel P Weber 18, Daniel W Cramer 6, Allison F Vitonis 6, Elisa V Bandera 19, Sara H Olson 4, Lorna Rodriguez–Rodriguez 19, Melony King 20, Louise A Brinton 21, Hannah Yang 21, Montserrat Garcia–Closas 22, Jolanta Lissowska 23, Hoda Anton–Culver 24, Argyrios Ziogas 24, Simon A Gayther 3, Susan J Ramus 3, Usha Menon 25, Aleksandra Gentry–Maharaj 25, Penelope M Webb 1,*, on behalf of the Ovarian Cancer Association Consortium
PMCID: PMC3857135  NIHMSID: NIHMS486473  PMID: 23404857

Abstract

Whilst previous studies have reported that higher body-mass index (BMI) increases a woman’s risk of developing ovarian cancer, associations for the different histological subtypes have not been well defined. As the prevalence of obesity has increased dramatically, and classification of ovarian histology has improved in the last decade, we sought to examine the association in a pooled analysis of recent studies participating in the Ovarian Cancer Association Consortium. We evaluated the association between BMI (recent, maximum, and in young adulthood) and ovarian cancer risk using original data from 15 case-control studies (13,548 cases, 17,913 controls). We combined study-specific adjusted odds ratios (ORs) using a random–effects model. We further examined the associations by histological subtype, menopausal status and post-menopausal hormone use. High BMI (all time-points) was associated with increased risk. This was most pronounced for borderline serous (recent BMI: pooled OR=1.24 per 5kg/m2; 95%CI 1.18–1.30), invasive endometrioid (1.17; 1.11–1.23) and invasive mucinous (1.19; 1.06–1.32) tumours. There was no association with serous invasive cancer overall (0.98; 0.94–1.02), but increased risks for low grade serous invasive tumours (1.13, 1.03–1.25) and in pre-menopausal women (1.11; 1.04–1.18). Among post–menopausal women, the associations did not differ between HRT users and non–users. Whilst obesity appears to increase risk of the less common histological subtypes of ovarian cancer, it does not increase risk of high grade invasive serous cancers, and reducing BMI is therefore unlikely to prevent the majority of ovarian cancer deaths. Other modifiable factors must be identified to control this disease.

Keywords: ovarian cancer, obesity, body mass index

INTRODUCTION

It is widely accepted that being overweight or obese increases a woman’s risk of developing endometrial and post–menopausal breast cancer (Calle and Kaaks 2004). The association with ovarian cancer is less clear, largely because individual studies have had insufficient power to reliably detect moderate effects or to consider the different histological subtypes of ovarian cancer. In 2008, a pooled analysis of cohort studies concluded that BMI was associated with ovarian cancer in pre-menopausal women only, however this analysis only included 2000 cases and thus also had limited power to evaluate the different histological subtypes separately (Schouten, et al. 2008). A recent pooled analysis conducted to overcome these limitations concluded that among women who have not used hormone replacement therapy (HRT), the risk of ovarian cancer increases by 10% for every 5kg/m2 increase in body–mass index (BMI) (Collaborative Group on Epidemiological Studies of Ovarian Cancer, 2012). This association did not vary significantly for the different histological subtypes of ovarian cancer, with the exception of borderline serous cancers where the excess relative risk was substantially greater than for the other tumour types. There was no increase in risk with increasing BMI among women who had used HRT.

However, the mean year of diagnosis of the cases in the studies included in the previous report was 1992 (Collaborative Group on Epidemiological Studies of Ovarian Cancer, 2012) and over the last few decades, most countries have seen dramatic increases in the prevalence of overweight and obesity (Finucane, et al. 2011). Classification of the different histological subtypes of ovarian cancer has also improved in recent years (Gilks and Prat 2009) and it is possible that misclassification in earlier studies might have masked differences between the histological subtypes. In particular, it is now recognized that low and high grade invasive serous cancers are distinct entities and that many cancers previously described as high grade endometrioid tumours should really be classified as high grade serous cancers (Gilks and Prat 2009). We therefore sought to confirm the results of the previous analysis in a second, independent pooled analysis using data from more recent studies that met the inclusion criteria for the Ovarian Cancer Association Consortium (OCAC) collaboration (Ramus, et al. 2008). We examined the associations by histological subtype and tumour grade and by menopausal status and HRT use because, if the effects of obesity on ovarian cancer risk are mediated through oestrogenic pathways, then any association between BMI and risk may be more evident among women who have not used exogenous oestrogens. We also evaluated the relation between body–size at different ages and ovarian cancer risk.

METHODS

OCAC was founded in 2005 to foster collaborative efforts in discovering and validating associations between genetic polymorphisms and ovarian cancer risk. A detailed description has been provided elsewhere (Ramus et al. 2008) but, briefly, studies were eligible for inclusion if they included at least 200 cases of ovarian cancer and 200 controls, with controls from broadly the same population as cases, and provided DNA for genetic analyses. Table 1 summarizes the characteristics of the fifteen case–control studies (fourteen population–based and one clinic–based) that provided data for these analyses (Ziogas, et al. 2000; Royar, et al. 2001; Glud, et al. 2004; Pike, et al. 2004; Terry, et al. 2005; Hoyo, et al. 2005; Risch, et al. 2006; Garcia-Closas, et al. 2007; Rossing, et al. 2007; Kelemen, et al. 2008; Lurie, et al. 2008; Merritt, et al. 2008; Moorman, et al. 2008; Wu, et al. 2009; Balogun, et al. 2011; Bandera, et al. 2011; Ness, et al. 2011). Race/ethnicity was categorized as non–Hispanic White (88%), Hispanic White (3%), Black (4%), Asian (3%), or other (2%). All studies had ethics approval, and all study participants provided informed consent.

Table 1.

Characteristics of the fifteen studies included in the pooled analyses of BMI and ovarian cancer.

Study Diagnosis Years Age range Geographic location Number of:
Case Sources (Response Rate) Histology b Invasive cases (%) Borderline cases (%) Control Sources (Response Rate) BMI Measurement
Cases a Controls a Recent Early adulthood Maximum
Clinic–based
Mayo Clinic Ovarian Cancer Case Control Study (MAY) 2000–2008 20–91 Upper Midwest, USA 715 945 Mayo Clinic (84%) Ser
Muc
End
CC
Other		 405 (57%)
19 (3%)
100 (14%)
44 (6%)
47 (7%)		 Ser
Muc
Other		 59 (8%)
27 (4%)
14 (2%)		 Women seeking general examinations (65%) Y
Population–based
Australian Ovarian Cancer Study and Australian Cancer Study (Ovarian Cancer) (AUS) 2002–2006 18–80 Australia 1579 1485 Cancer registries, treatment centres (84%) Ser
Muc
End
CC
Other		 756 (48%)
49 (3%)
150 (9%)
98 (6%)
192 (12%)		 Ser
Muc
Other		 150 (9%)
169 (11%)
15 (1%)		 Electoral roll (47%) Y Y Y
Connecticut Ovary Study (CON) 1998–2003 34–81 Connecticut, USA 483 551 Cancer registries, pathology departments (69%) Ser
Muc
End
CC
Other		 221 (46%)
19 (4%)
74 (15%)
35 (7%)
25 (5%)		 Ser
Muc
Other		 69 (14%)
36 (7%)
4 (1%)		 Random digit dialling (61%) Y Y
Diseases of the Ovary and their Evaluation Study (DOV) 2002–2005 35–74 Washington, USA 1569 1848 Cancer Surveillance System, SEER (77%) Ser
Muc
End
CC
Other		 672 (43%)
33 (2%)
187 (12%)
87 (6%)
176 (11%)		 Ser
Muc
Other		 234 (15%)
156 (10%)
24 (2%)		 Random digit dialling (69%) Y Y Y
German Ovarian Cancer Study (GER) 1993–1998 21–75 Germany 254 519 Hospital admissions (58%) Ser
Muc
End
CC
Other		 106 (42%)
26 (10%)
26 (10%)
6 (2%)
63 (25%)		 Ser
Muc
Other		 15 (6%)
9 (4%)
3 (1%)		 Population registries (51%) Y Y
Hawaii Ovarian Cancer Study (HAW) 1993–2008 18–93 Hawaii, US 883 1089 Cancer registry (78%) Ser
Muc
End
CC
Other		 312 (35%)
70 (8%)
116 (13%)
81 (9%)
122 (14%)		 Ser
Muc
Other		 88 (10%)
87 (10%)
7 (1%)		 Department of Health annual survey (80%) Y Y
Hormones and Ovarian Cancer Prediction (HOP) 2003– 25–80 NY, OH and PA, US 771 1803 Cancer registries, pathology databases, physician offices (69%) Ser
Muc
End
CC
Other		 364 (47%)
36 (5%)
97 (13%)
52 (7%)
125 (16%)		 Ser
Muc
Other		 58 (8%)
29 (4%)
10 (1%)		 Random digit dialling (81%) Y Y Y
The Danish Malignant Ovarian Tumour Study (MAL) 1994–1999 35–79 Denmark 744 1552 Danish Cancer Registry, 16 gynecologic departments 79%) Ser
Muc
End
CC
Other		 337 (45%)
50 (7%)
75 (10%)
43 (6%)
39 (5%)		 Ser
Muc
Other		 103 (14%)
87 (12%)
10 (1%)		 Danish Central Population Register (67%) Y Y
North Carolina Ovarian Cancer Study (NCO) 1999–2007 20–75 North Carolina, US 1087 1083 North Carolina Central Cancer Registry (70%) Ser
Muc
End
CC
Other		 470 (43%)
43 (4%)
138 (13%)
88 (8%)
122 (11%)		 Ser
Muc
Other		 155 (14%)
64 (6%)
5 (0%)		 Random digit dialling (63%) Y Y Y
New England–based Case– Control Study of Ovarian Cancer (NEC) 1992 – 2008 18–78 New England, US 1960 2097 Hospital tumour boards, State cancer registries (72%) Ser
Muc
End
CC
Other		 819 (43%)
89 (5%)
296 (16%)
192 (10%)
70 (4%)		 Ser
Muc
Other		 242 (13%)
145 (8%)
35 (2%)		 Random digit dialling and townbook selection (69%) Y Y
New Jersey Ovarian Cancer Study (NJO) 2004–2008 23–96 New Jersey, US 224 448 NJ State Cancer Registry (47%) Ser
Muc
End
CC
Other		 129 (58%)
11 (5%)
31 (14%)
30 (13%)
23 (10%)		 Random digit dialling, Medicare and Medicaid lists, area sampling (40%) Y Y Y
Polish Ovarian Cancer Study (POL) 2001–2003 24–74 Poland 283 1071 Hospitals in Warsaw and Lodz (71%) Ser
Muc
End
CC
Other		 116 (41%)
19 (7%)
39 (14%)
10 (4%)
78 (28%)		 Ser
Muc
Other		 17 (6%)
3 (1%)
1 (0%)		 Electoral roll (67%) Y Y
UC Irvine Ovarian Cancer Study (UCI) 1994–2004 18–86 Orange and San Diego counties, US 588 565 Orange County Cancer Surveillance Program, Tumour registry (70%) Ser
Muc
End
CC
Other		 211 (36%)
28 (5%)
72 (12%)
37 (6%)
43 (7%)		 Ser
Muc
Other		 122 (21%)
74 (13%)
1 (0%)		 Random digit dialling (80%) Y Y
UK Ovarian Cancer Population Study (UKO) 2006– 50–76 United Kingdom 687 1026 Gynecologic oncology NHS centres (86%) Ser
Muc
End
CC
Other		 348 (51%)
69 (10%)
106 (15%)
65 (9%)
99 (14%)		 Postmenopausal women participating in UKCTOPCS* (97%) Y
Los Angeles County Case–Control Studies of Ovarian Cancer (USC) 1993– 19–86 Los Angeles, US 1721 1831 Cancer Surveillance Program of Los Angeles (73%) Ser
Muc
End
CC
Other		 826 (48%)
112 (7%)
183 (11%)
87 (5%)
110 (6%)		 Ser
Muc
Other		 240 (14%)
158 (9%)
5 (0%)		 Neighbourhood controls (73%) Y Y
Open in a new tab
*

United Kingdom Collaborative Trial of Ovarian Cancer Screening;

a

Numbers of participants with body–size information;

b

S=serous; M=mucinous; E=endometrioid; CC=clear cell; Other includes both ‘other’ histologies and subjects with unknown histology

Analysis Variables

There was some variation in the way weight information was collected by the individual studies (Supplementary Table A). Weight in early adulthood was reported by 14 studies (all except MAY); this was reported as weight at age 18 for nine studies and at age 20 for two studies (AUS, GER), while three studies reported weight ‘in your 20s’ (CON, MAL, USC). Recent weight was reported by 11 studies (AUS, CON, DOV, HOP, MAL, MAY, NCO, NJO, NEC, UCI, USC); for most studies this was reported as weight one year prior to diagnosis/reference date, but five years prior to diagnosis/reference date was used for four studies (CON, DOV, MAL, USC). To minimize overlap between our analyses of recent weight and the previous pooled analysis,3 we excluded two studies (GER, HAW) that were included in the previous analysis, but included two (NEC, USC) that had contributed only part of their data to the previous analysis (total overlap ~1200 cases). Maximum weight was reported by 8 studies (AUS, DOV, GER, HAW, HOP, NCO, NJO, POL). Body mass index (BMI), calculated as weight in kilograms divided by the square of height in metres (kg/m2), was classified using the World Health Organization (WHO) definitions of obesity (<18.5 ‘underweight’; 18.5–24.9 ‘normal weight’; 25–29.9 ‘overweight’; 30–34.9 ‘class I obesity’; 35–39.9 ‘class II obesity’; and ≥40 ‘class III obesity’) (W.H.O., 1995). For subgroup analyses there were small numbers in the upper classes of obesity for BMI in early adulthood so these groups were combined.

Covariate Information

Each case–control study provided information on potential confounding variables including age, cancer grade, race/ethnicity, parity, breastfeeding, oral contraceptive (OC) and hormone replacement therapy (HRT) use, family history of breast or ovarian cancer in a first degree relative, menopausal status, and history of hysterectomy or tubal ligation. All data were cleaned and checked for internal consistency and clarification was provided by the original investigators when needed.

Statistical Analysis

We used Stukel’s two–stage method of analysis to obtain study–specific odds ratios (ORs) and pooled odds ratios (pORs) and 95% confidence intervals (CIs) (Stukel, et al. 2001). In the first stage, each study was analyzed separately, controlling for study–specific confounders. The pooled exposure effect was estimated in a second–stage using a meta–analytic approach. A weighted average of the log relative risk (RR) was estimated, taking into account the random effects using the method of DerSimonian and Laird (DerSimonian and Laird 1986). Statistical heterogeneity among studies was evaluated using the Cochran Q test and I2 statistics (Higgins and Thompson 2002). All models were stratified by age in 5–year groups and adjusted for parity (0, 1, 2, 3, 4+ full–term births), oral contraceptive use (0, ≤60, >60 months), and family history of breast or ovarian cancer in a first degree relative. We also adjusted study–specific results for race/ethnicity (non–Hispanic white, Hispanic white, black, Asian, other) where more than 10% of the study population was not classified as non–Hispanic white and inclusion of a term for race/ethnicity altered the odds ratio by 10% or more. Other potential confounders considered but not included in final models since they did not make any material change to the BMI associations were: breastfeeding, history of hysterectomy, tubal ligation, menopausal status and HRT. Adjusting for history of endometriosis made no material change to the pooled estimates for the endometrioid or clear cell subtypes and thus it was not included in final models. Data on smoking status were not available for all studies, however including smoking status in models where it was available did not result in significant changes to the pooled estimates and thus it was not included in final models. Covariate data were mostly complete and uniformly coded for all studies with a few exceptions. The parity variable included all full–term births (live and still births) for all studies except MAY which recorded only live births. Secondly, tubal ligation and breastfeeding data were unavailable for the MAY study. These missing covariates were therefore not included in the first stage models for this study.

We initially computed odds ratios for each of the primary exposure variables for invasive and borderline cancers separately and then further classified tumours by their histological subtype (serous, mucinous, endometrioid, clear cell). In the subtype–specific models, adjacent levels of confounders were collapsed where necessary to avoid zero cells in the two–stage models. Where heterogeneity was evident, we examined the data for potential sources of this heterogeneity including type of control group (population versus hospital–based) and style of questionnaire (self–completed versus in–person interview). The relative risk of ovarian cancer per 5 kg/m2 increase in body mass index was estimated by fitting a log–linear trend across categories of body mass index (18.5–<20,20–,22.5–,25–,27.5–,30–,32.5–,35–,37.5–,40+ kg/m2) using the overall median value within each category, except for the top category where we used the site–specific median as this varied between sites. Since we were interested in the effects of being overweight and speculated that the relation between BMI and cancer risk might not be linear at very low BMI levels, these analyses excluded women in the ‘underweight’ range (BMI<18.5 kg/m2).

We also conducted subgroup analyses to assess the interaction between recent BMI, menopausal status and use of any hormone replacement therapy (pre/peri–menopausal, postmenopausal and never used HRT, postmenopausal and had used HRT). There was some heterogeneity in how menopausal status was defined across studies, so we also conducted analyses stratified by age at diagnosis (<50, ≥50 years). To avoid problems with zero cells in some studies in these and other sub–group analyses, we pooled all data and computed ORs using logistic regression stratified by study site and age in 5–year groups in order to maximize the statistical power. The statistical significance of any observed stratum–specific differences was then assessed by including a cross–product term (using the continuous BMI variables defined above) in regression models.

Analyses were conducted using SAS (SAS Institute, Cary, North Carolina, USA) and Stata 10 (College Station, TX, USA).

RESULTS

Eleven studies contributed to analyses of recent BMI, eight studies for maximum BMI and 14 studies for BMI in early adulthood (Table 1). Using the two–stage method of analysis, we observed significantly increased risks of both invasive and borderline ovarian cancers associated with higher BMI at all three time–points. The association was modest for invasive tumours with an increase in risk of 4% per 5 kg/m2 for recent BMI and 8% for BMI in early adulthood, but was stronger for borderline tumours with increases of 15–18% per 5 kg/m2 for the different time–points (Table 2).

Table 2.

Adjusteda pooled odds ratios (95% confidence intervals) for ovarian cancer in relation to BMI, by tumour behaviourb

BMI (kg/m2) Invasive
Borderline
Studies I2 (%) Cases Controls pOR (95%CI) Studies I2 (%) Cases Controls pOR (95%CI)
Recent BMI
<18.5 11 26.6 183 282 1.08 (0.84–1.39) 10 0·0 57 281 1.13 (0.82–1.55)
18.5–24.9 (Ref) 11 4020 6796 1.0 10 1080 6599 1.0
25–29.9 11 31.2 2500 4077 1.00 (0.92–1.09) 10 13.8 662 3930 1.23 (1.09–1.39)
30–34.9 11 0.0 1166 1808 1.06 (0.97–1.16) 10 1.1 379 1741 1.61 (1.40–1.85)
35–39.9 11 1.7 511 692 1.21 (1.07–1.38) 10 0.0 150 672 1.68 (1.37–2.06)
≥ 40 11 0.0 383 503 1.22 (1.05–1.41) 9 0.0 137 486 1.96 (1.57–2.46)
 Per 5 kg/m2 c 11 47.7 1.04 (1.00–1.08)* 9 0.0 1.18 (1.14–1.23)
Maximum BMI
<18.5 6 0.0 24 33 1.22 (0.69–2.14) 3 0.0 5 19 1.00 (0.33–3.03)
18.5–24.9 (Ref) 8 1393 2683 1.0 7 296 2548 1.0
25–29.9 8 6.2 1427 2566 1.02 (0.92–1.13) 7 19.7 275 2409 1.13 (0.91–1.41)
30–34.9 8 0.0 823 1335 1.17 (1.04–1.31) 7 17.0 199 1236 1.58 (1.24–2.03)
35–39.9 8 60.2 388 592 1.29 (0.99–1.68)* 6 0.0 105 544 1.70 (1.30–2.22)
≥ 40 8 15.9 310 490 1.16 (0.96–1.41) 5 30.5 108 455 1.90 (1.35–2.68)
 Per 5 kg/m2 c 8 45.3 1.06 (1.01–1.11) 7 35.8 1.17 (1.08–1.26)
BMI early adulthood
<18.5 14 0.0 1646 2931 0.94 (0.88–1.01) 12 13.0 416 2718 0.97 (0.85–1.11)
18.5–24.9 (Ref) 14 7278 12364 1.0 12 1819 11245 1.0
25–29.9 14 0.0 788 1151 1.12 (1.01–1.24) 11 0.0 248 983 1.27 (1.09–1.49)
30–34.9 14 0.0 176 233 1.21 (0.98–1.49) 10 0.0 66 210 1.32 (0.98–1.78)
≥35 12 0.0 76 108 1.08 (0.78–1.49) 10 0.0 43 100 1.86 (1.25–2.78)
 Per 5 kg/m2 c 14 1.08 (1.03–1.14) 12 1.15 (1.08–1.24)
Open in a new tab
a

Stratified by age in 5–year groups and adjusted for parity (0,1,2,3,4+ full–term births), hormonal contraceptive use (0, ≤ 60, >60 months), family history of breast or ovarian cancer in a first degree relative and, where appropriate, race/ethnicity.

b

Numbers may not sum to total because of missing data

c

Excludes women in the underweight range (BMI < 18.5 kg/m2)

*

Significant heterogeneity noted (p–value for heterogeneity <0·05)

Results of the pooled analyses stratified by histological subtype are presented in Tables 3 and 4 for invasive and borderline tumours respectively. Overall, risk of invasive serous cancer was not associated with any measure of BMI (Table 3). However, stratification by tumour grade (data available for 91% of cases) revealed positive associations between all measures of BMI and risk of low grade (G1) invasive serous tumours (OR=1.13, 1.18 and 1.24 per 5kg/m2 for recent, maximum and young adult BMI respectively, all p<0.01) but not high grade (G2–G4) tumours (OR=0.96, 0.96 and 0.98, respectively). Higher BMI (all BMI variables) was significantly associated with an increased risk of invasive endometrioid ovarian cancer. This association was restricted to low and intermediate grade (G1–G2) tumours (OR per 5kg/m2 1.25, 1.22 and 1.20 for recent, maximum and young adulthood BMI respectively, all p≤0.001) and was not seen for high grade (G3–G4) endometrioid cancers (OR=0.97, 1.02 and 0.90, respectively) (data on grade available for 93% of cases). The associations between BMI and invasive mucinous and clear cell cancers were less clear, with increased risks of both tumour types associated with high recent BMI and, for mucinous cancers, BMI in young adulthood, but not maximum BMI. The results for recent BMI were essentially unaltered when we restricted the analysis to include only studies that assessed weight around 5 years prior to diagnosis to reduce potential bias due to recent weight loss in cases. Considering all non–serous invasive cancers together, the association with recent BMI remained significant after adjusting for maximum BMI or BMI in young adulthood, however after adjusting for recent BMI there was no association with either maximum BMI (OR=1.02, 95%CI 0.95–1.11 per 5kg/m2) or BMI in young adulthood (OR=0.96, 95%CI 0.86–1.08 per 5kg/m2).

Table 3.

Adjusteda pooled odds ratios (95% confidence intervals) for invasive ovarian cancer in relation to BMI, by histological subtypeb

Studies (n) Controls (n) Serous
Mucinous
Endometrioid
Clear Cell
Cases (n) pOR (95% CI) Cases (n) pOR (95% CI) Cases (n) pOR (95% CI) Cases (n) pOR (95% CI)
Recent BMI 11
<18.5 282 91 0.93 (0.72–1.20) 19 2.48 (1.03–4.51) 33 1.47 (0.98–2.21) 18 2.69 (1.34–5.41)
18.5–24.9 (Ref) 6796 2475 1.0 207 1.0 592 1.0 353 1.0
25–29.9 4077 1477 0.93 (0.86–1.02) 134 1.19 (0.95–1.50) 380 1.12 (0.96–1.30) 227 1.05 (0.79–1.40)*
30–34.9 1808 665 0.94 (0.84–1.05) 68 1.48 (0.92–2.37)* 205 1.37 (1.14–1.64) 98 1.14 (0.79–1.63)*
35–39.9 692 275 1.06 (0.90–1.23) 29 2.03 (1.10–3.77) 97 1.74 (1.36–2.23) 48 1.59 (1.14–2.24)
≥ 40 503 170 0.89 (0.74–1.08) 29 2.70 (1.76–4.16) 82 1.86 (1.42–2.24) 37 1.58 (1.04–2.40)
Per 5 kg/m2 c 0.98 (0.94–1.02) 1.19 (1.06–1.32) 1.17 (1.11–1.23) 1.06 (0.96–1.17)*
Maximum BMI 8
18.5–24.9 (Ref) 2683 793 1.0 86 1.0 177 1.0 120 1.0
25–29.9 2566 787 0.93 (0.73–1.17)* 84 1.22 (0.88–1.69) 194 1.20 (0.96–1.45) 112 0.95 (0.72–1.26)
30–34.9 1335 445 1.03 (0.89–1.18) 34 1.08 (0.70–1.67) 129 1.63 (1.26–2.10) 69 1.22 (0.88–1.70)
35–39.9 592 199 1.18 (0.81–1.72)* 20 1.30 (0.74–2.27) 67 1.78 (1.29–2.46) 33 1.30 (0.84–2.00)
≥ 40 490 147 0.98 (0.68–1.41)* 17 1.37 (0.76–2.46) 60 1.82 (1.29–2.56) 27 1.12 (0.70–1.82)
 Per 5 kg/m2 c 1.00 (0.93–1.07) 1.05 (0.94–1.17) 1.18 (1.09–1.28) 1.04 (0.95–1.13)
Early adult 14
<18.5 2931 918 0.94 (0.86–1.03) 102 0.94 (0.74–1.19) 243 0.93 (0.80–1.09) 164 1.08 (0.83–1.39)
18.5–24.9 (Ref) 12364 4161 1.0 465 1.0 1121 1.0 648 1.0
25–29.9 1151 401 1.04 (0.92–1.18) 54 1.20 (0.88–1.64) 150 1.33 (1.10–1.62) 64 1.05 (0.75–1.45)
30–34.9 231 73 1.03 (0.78–1.37) 19 1.90 (1.12–3.21) 39 1.51 (1.03–2.21) 14 1.10 (0.61–1.99)
≥ 35 110 36 1.15 (0.75–1.76) 7 2.18 (0.96–4.95) 18 1.85 (1.05–3.24) 6 2.73 (1.08–6.88)
 Per 5 kg/m2 c 1.02 (0.95–1.10) 1.22 (1.07–1.40) 1.14 (1.04–1.25) 1.02 (0.89–1.16)
Open in a new tab
a

Stratified by age in 5–year groups and adjusted for parity (0,1,2,3,4+ full–term births), hormonal contraceptive use (0, ≤ 60, >60 months), family history of breast or ovarian cancer in a first degree relative and, where appropriate, race/ethnicity; pooled across study sites using random effects models.

b

Numbers may not sum to total because of missing data

c

Excludes women in the underweight range (BMI < 18.5 kg/m2)

*

Significant heterogeneity noted (p–value for heterogeneity <0·05)

Table 4.

Adjusteda pooled odds ratios (95% confidence intervals) for borderline ovarian cancer in relation to BMI, by histological subtype.

Studies (n) Controls (n) Serous b
Mucinous b
Cases (n) pOR (95% CI) Cases (n) pOR (95% CI)
Recent BMI 10
<18.5 281 23 1.12 (0.70–1.79) 33 1.61 (1.08–2.39)
18.5–24.9 (Ref) 6599 568 1.0 454 1.0
25–29.9 3930 403 1.40 (1.22–1.62) 234 1.08 (0.91–1.28)
30–34.9 1741 236 1.86 (1.55–2.24) 122 1.32 (1.05–1.67)
35–39.9 672 101 2.11 (1.66–2.70) 41 1.29 (0.91–1.84)
≥ 40 486 85 2.23 (1.69–2.94) 41 1.68 (1.16–2.43)
 Per 5 kg/m2 c 1.24 (1.18–1.30) 1.09 (1.02–1.16)
Maximum BMI 7
18.5–24.9 (Ref) 2548 135 1.0 138 1.0
25–29.9 2409 153 1.39 (1.00–1.93) 113 0.99 (0.75–1.30)
30–34.9 1236 115 2.00 (1.51–2.65) 78 1.39 (0.99–1.96)
35–39.9 544 66 2.40 (1.71–3.38) 35 1.26 (0.68–2.32)
≥ 40 455 71 2.73 (1.92–3.88) 30 1.29 (0.79–2.11)
 Per 5 kg/m2 c 1.25 (1.17–1.34) 1.09 (0.98–1.21)
BMI early adulthood 12
<18.5 2718 222 0.90 (0.77–1.06) 171 1.07 (0.88–1.31)
18.5–24.9 (Ref) 11245 1034 1.0 699 1.0
25–29.9 983 152 1.40 (1.12–1.74) 86 1.22 (0.95–1.55)
30–34.9 210 40 1.48 (1.03–2.14) 26 1.57 (1.00–2.47)
≥ 35 100 29 2.34 (1.47–3.74) 12 2.00 (1.00–4.01)
 Per 5 kg/m2 c 1.22 (1.12–1.33) 1.11 (0.99–1.24)
Open in a new tab
a

Stratified by age in 5–year groups and adjusted for parity (0,1,2,3,4+ full–term births), hormonal contraceptive use (0, ≤ 60, >60 months), family history of breast or ovarian cancer in a first degree relative and, where appropriate, race/ethnicity; pooled across study sites using random effects models.

b

Numbers may not sum to total because of missing data

c

Excludes women in the underweight range (BMI < 18.5 kg/m2)

Increasing BMI (all BMI variables) was associated with increased risks of both borderline serous and mucinous ovarian cancers, with significant trends with increasing BMI that were stronger for borderline serous cancers (20–25% increase per 5 kg/m2) than borderline mucinous cancers (9–11% per 5 kg/m2; Table 4).

Although there was some heterogeneity among studies for some of the pooled estimates, heterogeneity for the estimates per 5kg/m2 only reached statistical significance for recent BMI and risk of clear cell tumours and the combined group of all invasive tumours; sensitivity analyses by study design features suggested that no single factor could explain this observed heterogeneity.

When we combined all tumour types and stratified by ever use of HRT, we observed a significant association between BMI and cancer risk among women who had not used HRT (OR per 5 kg/m2 = 1.10; 95%CI 1.07–1.14) but no association among women who had used HRT (1.02; 0.97–1.07). However, we saw markedly different patterns of association when we considered pre– and post–menopausal women and the different histological subtypes of cancer separately (Table 5). When we stratified by menopausal status and use of HRT, we saw significant interaction for recent BMI and risk of invasive serous cancers (p≤0.001). A significant trend of increasing risk with increasing BMI was observed in premenopausal women, with no association among postmenopausal women who had never used HRT, and a significant inverse association among those who had used HRT. Further stratification of the pre-menopausal group suggested the positive association was stronger for G1 (OR 1.34, 95%CI 1.14–1.59) but still statistically significant for G2–4 tumors (OR 1.07, 95% CI 1.00–1.15; p<0.05). A similar pattern was seen in analyses of maximum BMI and BMI in young adulthood (data not shown), suggesting the lack of a positive association among post–menopausal women was not simply an artefact due to recent weight loss among women with serous cancer. For all other invasive subtypes combined, the association was somewhat stronger among pre–menopausal women than post–menopausal women but did not differ by HRT use among post–menopausal women. The association with borderline tumours did not vary by menopausal status or HRT use. When we stratified by age at diagnosis (<50, ≥50 years) instead of menopausal status the results did not differ materially (data not shown).

Table 5.

Adjusteda odds ratios (95% confidence intervals) for ovarian cancer in relation to recent BMI, by menopausal status and use of hormone replacement therapy.

Controls (n) Invasive Serousb
All other Invasive Cancersb
All Borderline b
Cases (n) OR (95% CI) Cases (n) OR (95% CI) Cases (n) OR (95% CI)
Premenopausal women
18.5–24.9 (Ref) 2049 484 1.0 514 1.0 529 1.0
25–29.9 919 272 1.23 (1.03–1.47) 275 1.26 (1.06–1.51) 254 1.22 (1.02–1.46)
30–34.9 417 121 1.21 (0.96–1.54) 139 1.40 (1.11–1.76) 147 1.63 (1.30–2.05)
35–39.9 152 55 1.50 (1.07–2.10) 76 1.78 (1.30–2.45) 65 2.00 (1.44–2.78)
≥ 40 136 47 1.43 (0.99–2.06) 72 1.81 (1.30–2.52) 52 1.76 (1.22–2.53)
 Per 5 kg/m2 c 1.11 (1.04–1.18) 1.17 (1.11–1.24) 1.19 (1.12–1.27)
Postmenopausal women, no HRT
18.5–24.9 (Ref) 1343 652 1.0 347 1.0 157 1.0
25–29.9 1054 425 0.87 (0.74–1.01) 312 1.20 (1.00–1.43) 124 1.17 (0.90–1.51)
30–34.9 522 216 0.93 (0.77–1.12) 153 1.24 (0.99–1.55) 82 1.60 (1.19–2.16)
35–39.9 226 87 0.89 (0.67–1.16) 67 1.24 (0.91–1.69) 30 1.36 (0.88–2.09)
≥ 40 157 61 0.87 (0.63–1.21) 65 1.64 (1.18–2.29) 33 2.12 (1.37–3.29)
 Per 5 kg/m2 c 0.97 (0.92–1.03) 1.10 (1.03–1.17) 1.17 (1.08–1.27)
Postmenopausal women who used HRT
18.5–24.9 (Ref) 1650 778 1.0 313 1.0 138 1.0
25–29.9 1123 440 0.86 (0.75–1.00) 221 1.08 (0.89–1.31) 112 1.35 (1.03–1.76)
30–34.9 480 183 0.86 (0.71–1.05) 101 1.19 (0.92–1.54) 60 1.64 (1.18–2.28)
35–39.9 167 75 1.08 (0.78–1.45) 31 1.15 (0.76–1.74) 20 1.67 (1.00–2.78)
≥ 40 111 23 0.49 (0.30–0.77) 31 1.64 (1.06–2.52) 13 1.48 (0.80–2.76)
 Per 5 kg/m2 c 0.92 (0.87–0.98) 1.09 (1.01–1.18) 1.16 (1.05–1.28)
Open in a new tab
a

Stratified by study site (AUS, DOV, HOP, MAY, NEC, NJO, UCI, USC) and age in 5–year groups, and adjusted for parity (0,1,2,3,4+ full–term births), hormonal contraceptive use (0, ≤ 60, >60 months), family history of breast or ovarian cancer in a first degree relative.

b

Numbers may not sum to total because of missing data

c

Excludes women in the underweight range (BMI < 18.5 kg/m2)

DISCUSSION

The results of our pooled analysis confirm that being overweight or obese is associated with an overall increased risk of both invasive and borderline ovarian cancer, however for invasive cancers this association appears to be restricted to the non–serous and low–grade serous subtypes. Furthermore, most of our risk estimates were very consistent with those from a previous pooled analysis (Collaborative Group on Epidemiological Studies of Ovarian Cancer, 2012) with a strong increase in risk of borderline serous cancer (pooled OR/RR=1.24 per 5kg/m2 in our analysis vs. 1.29 in the previous report) and intermediate risks for clear cell (1.06 vs. 1.05) and invasive (1.19 vs. 1.15) and borderline (1.09 vs. 1.06) mucinous cancers. Like the previous report, we saw no increase in risk of invasive serous cancer overall (0.98 vs. 1.00), however we did see an increased risk of low–grade invasive serous cancers (OR=1.13) which are now thought to arise via a different aetiological pathway from their high–grade counterparts. The only subtype for which our results differed appreciably was invasive endometrioid cancers where we saw a 17% increase in risk per 5 kg/m2 overall, and a 25% increase after excluding high–grade endometrioid cancers which are likely to be misclassified serous tumours (Gilks and Prat 2009), compared to only an 8% increase in the previous study (Collaborative Group on Epidemiological Studies of Ovarian Cancer, 2012).

Since endometrioid ovarian tumours are histologically similar to endometrial cancer (Russell 1994), which is strongly associated with obesity (Crosbie, et al. 2010), it seems plausible that obesity might also be a relatively strong risk factor for this subtype of ovarian cancer. The roughly 70–80% risk increases we observed even among the groups of women with highest BMI were, however, considerably lower than the nine–fold risk previously reported for endometrial cancer (Crosbie et al. 2010). Historically, the histopathologic classification of ovarian cancer cell types has only been modestly reproducible (Hernandez, et al. 1984; Cramer, et al. 1987; Sakamoto, et al. 1994), and particularly problematic was the specific diagnosis of serous versus endometrioid carcinomas (Stalsberg, et al. 1988). A recent development is the recognition that many carcinomas formally considered high grade endometrioid are better classified as high grade serous (Gilks and Prat 2009; Kobel, et al. 2010; Madore, et al. 2010). When we excluded high–grade endometrioid tumours from our analysis the associations with BMI were considerably strengthened while, as for invasive serous cancers, we saw no association with high grade endometrioid tumours. It is thus possible that misclassification of serous and endometrioid tumours may explain, in part, why a significant association between obesity and endometrioid ovarian cancers has not previously been consistently reported and why it was not observed in the previous large pooled analysis which included mostly older studies and did not consider tumour grade (Collaborative Group on Epidemiological Studies of Ovarian Cancer, 2012). Time trends in the use of various regimens of HRT, as well as the increasing prevalence of obesity over calendar time, may also play a role.

As in the previous pooled analysis, we observed an association between increasing BMI and risk of borderline ovarian tumours, with the strength of the association somewhat stronger for serous than mucinous tumours. High BMI has been associated with benign ovarian tumours (Jordan, et al. 2007), and there is evidence from epidemiological, histopathological and molecular studies that these borderline tumours may develop from benign tumours in a neoplastic progression (Jordan, et al. 2006). Our finding that low grade but not high grade invasive serous tumours were also associated with BMI supports this theory of progression for low grade serous cancers.

We can only speculate as to why we observed heterogeneity in the association between BMI and risk of invasive serous tumours between pre– and post–menopausal women, however this could not be explained by a higher proportion of G1 tumors in the pre-menopausal group. The endocrine consequences of obesity may have differential effects on the pathogenesis of serous ovarian cancer in pre– and postmenopausal women. Whilst postmenopausal obesity is associated with higher levels of endogenous oestrogen due to the synthesis of oestrogen in body fat (Key, et al. 2001), in premenopausal women, obesity lowers sex–hormone binding globulin (Key et al. 2001; Tworoger, et al. 2006) but does not significantly influence the levels of oestrogens and androgens as the ovaries produce more steroids than the peripheral fat tissue. Other hormonal factors that may mediate the relationship between obesity and risk of ovarian cancer include progesterone (Risch 1998) and insulin (Calle and Kaaks 2004). Compared to women of ‘normal’ weight, premenopausal obese women have reduced serum progesterone levels due to an increase in anovulatory cycles (Key et al. 2001), and there is a significant body of evidence suggesting that progesterone plays a protective role in ovarian carcinogenesis (Risch 1998). Obesity is associated with increased insulin levels, which lead to increases in the insulin–like growth factor–1 (IGF–I) (Calle and Kaaks 2004). There is no clear relation between adiposity and IGF-1 however high levels of IGF-1 have been associated with ovarian cancer in women younger than 55 years of age (Lukanova, et al. 2002).

Our observation that the positive association with BMI was stronger among pre-menopausal women is consistent with the earlier analysis of cohort studies (Schouten et al. 2008). However, in contrast to the recent pooled analysis (Collaborative Group on Epidemiological Studies of Ovarian Cancer, 2012), we found no suggestion of effect modification by use of HRT in postmenopausal women. Although the overall association did appear to be restricted to women who had never used HRT, this was driven by the stronger associations seen among pre–menopausal women who rarely use HRT. Similarly, the apparent lack of association among HRT users was driven by the strong inverse association with invasive serous cancers, the most common histological subtype, in this group. For the cancers that showed an overall association with BMI, non–serous invasive cancers and borderline cancers, the risk estimates among post–menopausal women did not differ by use of HRT. Whilst data on recent or current use of menopausal hormonal therapy was not available for the current analyses, the possibility that recent use may modify the relationship between body mass index and ovarian cancer risk deserves further exploration.

Strengths of our study include the large number of cases and controls made possible by pooling data from 15 individual case–control studies. Individual level data were combined into a single dataset following a rigorous data cleaning and harmonization protocol, giving enhanced ability to control for confounding in individual studies (Stukel et al. 2001). Pooling these data increased our statistical power to examine BMI in relation to the different histological subtypes of ovarian cancer, and allowed sub–group analyses to examine the effects by tumour grade, age, menopausal status, and for postmenopausal women, by use of HRT. Additionally, all studies contributing to the pooled analyses were conducted in the past two decades and, aside from early cases from the NEC and USC studies, a total of approximately 1200 cases (10%), there was no overlap with the previous pooled analysis (Collaborative Group on Epidemiological Studies of Ovarian Cancer, 2012). Histological misclassification is likely to be considerably less of a concern for these recent studies than in studies conducted in the more distant past, although some degree of misclassification remains likely.

However, as with any pooled–analysis, some limitations must be acknowledged. First the majority of the studies included in the pooled analyses relied upon retrospective self–reports of weight and height. Research has shown that women with higher BMI are more likely to underestimate weight, whereas underweight women are more likely to overestimate body weight (Kuskowska-Wolk, et al. 1989; Troy, et al. 1995; Lawlor, et al. 2002; Taylor, et al. 2006); this may have attenuated the true associations. We cannot exclude the possibility of selection bias due to self–selection of more health conscious women, who are less likely to be overweight or obese, into control groups; this would have lead to overstated risk estimates. Such misclassification, however, is likely to be non–differential with respect to the different histological subtypes. Finally, weight loss several years before the time of cancer diagnosis would, if present, bias risk estimates towards the null although the similar patterns of risk seen for all three measures of BMI, and for analyses of recent BMI restricted to studies that asked women to report their usual weight approximately five years prior to diagnosis, suggest this has not occurred to any great extent.

In summary, obesity appears to moderately increase the risk of developing the less common histological subtypes of ovarian cancer, particularly borderline and low grade invasive serous cancers and endometrioid cancers. With the possible exception of pre–menopausal women, it does not, however, appear to increase risk of the more common high grade invasive serous cancers that account for the majority of ovarian cancer deaths.

Supplementary Material

supplement

Acknowledgments

Funding

This work was supported by donations by the family and friends of Kathryn Sladek Smith to the Ovarian Cancer Research Fund. The studies that contributed to this analysis were funded by the National Institutes of Health (grant numbers R01 CA13689101A1, CA14089, CA17054, CA61132, CA63464, N01 PC67010 and R03 CA113148 [for USC], R01 CA112523 and R01 CA87538 [for DOV], R01 CA58598, N01 CN67001 and N01 PC35137 [for HAW], 5R01 CA074850 and 5R01 CA080742 [for CON], R01 CA76016 [for NCO], R01–CA–122443 [for MAY], CA054419–10, P40 CA1005009 [for NEC], K07 CA095666, R01 CA83918 and K22 CA138563 CA58860 [for NJO], CA58860, CA92044 [for UCI] and RO1 CA61107 [for MAL]); the California Cancer Research Program (0001389V20170 and 2110200 [for USC]); the California Department of Health Services (sub–contract 050E8709 [for USC]); the UCIBCS component of this research [UCI] was supported by the National Institutes of Health [grant numbers CA58860, CA92044] and the Lon V Smith Foundation [LVS39420]; the German Federal Ministry of Education and Research of Germany, Programme of Clinical Biomedical Research (01GB9401), German Cancer Research Center and University of Ulm [for GER]; the Eve Appeal, Oak Foundation, the University College Hospital National Institute for Health Research Biomedical Research Centre and the Royal Marsden Hospital Biomedical Research Centre [for UKO]; the National Health and Medical Research Council of Australia (199600), U.S. Army Medical Research and Materiel Command (DAMD 170110729 and W81XWH0610220), Cancer Council Tasmania and Cancer Foundation of Western Australia [for AUS]; the US Department of Defence (W81XWH–10–1–0280 [for NEC]); the Cancer Institute of New Jersey [for NJO]; Mermaid 1 and the Danish Cancer Society [for MAL]; Intramural Research Funds of the National Cancer Institute, Department of Health and Human Services, USA [for POL].

We are grateful to the family and friends of Kathryn Sladek Smith for their generous support of the International Ovarian Cancer Association Consortium (OCAC) through their donations to the Ovarian Cancer Research Fund. We thank all the individuals who took part in these studies and the project staff of all the participating studies. We thank Ursula Eilber and Tanja Koehler for technical assistance for the German Ovarian Cancer study (GER). The Australian group gratefully acknowledges the contribution of all the clinical and scientific collaborators (see http://www.aocstudy.org). The Connecticut Ovary Study group gratefully acknowledges the cooperation of 30 Connecticut hospitals, including Stamford Hospital, in allowing patient access. The NJ Ovarian Cancer Study group gratefully acknowledges the contribution of collaborators and staff at the New Jersey State Cancer Registry, the Cancer Institute of New Jersey, and Memorial Sloan–Kettering Cancer Center. The POCS thanks Drs. Mark Sherman and Nicolas Wentzensen from the National Cancer Institute, USA, Drs. Neonila Szeszenia–Dabrowska and Beata Peplonska of the Nofer Institute of Occupational Medicine (Lodz, Poland), Witold Zatonski of the Department of Cancer Epidemiology and Prevention, The M. Sklodowska–Curie Cancer Center and Institute of Oncology (Warsaw, Poland), and Pei Chao and Michael Stagner from Information Management Services (Sliver Spring MD, USA), for their valuable contributions to the study. Some of the data used in the CON Study were obtained from the Connecticut Tumor Registry, Connecticut Department of Public Health.

Footnotes

Declaration of Interest

The authors declare that there is no conflict of interest that could be perceived as prejudicing the impartiality of the research reported.

References

  1. Balogun N, Gentry-Maharaj A, Wozniak EL, Lim A, Ryan A, Ramus SJ, Ford J, Burnell M, Widschwendter M, Gessler SF, et al. Recruitment of newly diagnosed ovarian cancer patients proved challenging in a multicentre biobanking study. J Clin Epidemiol. 2011;64:525–530. doi: 10.1016/j.jclinepi.2010.07.008. [DOI] [PubMed] [Google Scholar]
  2. Bandera EV, King M, Chandran U, Paddock LE, Rodriguez-Rodriguez L, Olson SH. Phytoestrogen consumption from foods and supplements and epithelial ovarian cancer risk: a population-based case control study. BMC Womens Health. 2011;11:40. doi: 10.1186/1472-6874-11-40. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Calle EE, Kaaks R. Overweight, obesity and cancer: epidemiological evidence and proposed mechanisms. Nat Rev Cancer. 2004;4:579–591. doi: 10.1038/nrc1408. [DOI] [PubMed] [Google Scholar]
  4. Collaborative Group on Epidemiological Studies of Ovarian Cancer. Ovarian cancer and body size: individual participant meta-analysis including 25,157 women with ovarian cancer from 47 epidemiological studies. PLoS Medicine. 2012;9:e1001200. doi: 10.1371/journal.pmed.1001200. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Cramer SF, Roth LM, Ulbright TM, Mazur MT, Nunez CA, Gersell DJ, Mills SE, Kraus FT. Evaluation of the reproducibility of the World Health Organization classification of common ovarian cancers. With emphasis on methodology. Arch Pathol Lab Med. 1987;111:819–829. [PubMed] [Google Scholar]
  6. Crosbie EJ, Zwahlen M, Kitchener HC, Egger M, Renehan AG. Body mass index, hormone replacement therapy, and endometrial cancer risk: a meta-analysis. Cancer Epidemiol Biomarkers Prev. 2010;19:3119–3130. doi: 10.1158/1055-9965.EPI-10-0832. [DOI] [PubMed] [Google Scholar]
  7. DerSimonian R, Laird N. Meta-analysis in clinical trials. Control Clin Trials. 1986;7:177–188. doi: 10.1016/0197-2456(86)90046-2. [DOI] [PubMed] [Google Scholar]
  8. Finucane MM, Stevens GA, Cowan MJ, Danaei G, Lin JK, Paciorek CJ, Singh GM, Gutierrez HR, Lu Y, Bahalim AN, et al. National, regional, and global trends in body-mass index since 1980: systematic analysis of health examination surveys and epidemiological studies with 960 country-years and 9.1 million participants. Lancet. 2011;377:557–567. doi: 10.1016/S0140-6736(10)62037-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Garcia-Closas M, Brinton LA, Lissowska J, Richesson D, Sherman ME, Szeszenia-Dabrowska N, Peplonska B, Welch R, Yeager M, Zatonski W, et al. Ovarian cancer risk and common variation in the sex hormone-binding globulin gene: a population-based case-control study. BMC Cancer. 2007;7:60. doi: 10.1186/1471-2407-7-60. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Gilks CB, Prat J. Ovarian carcinoma pathology and genetics: recent advances. Hum Pathol. 2009;40:1213–1223. doi: 10.1016/j.humpath.2009.04.017. [DOI] [PubMed] [Google Scholar]
  11. Glud E, Kjaer SK, Thomsen BL, Hogdall C, Christensen L, Hogdall E, Bock JE, Blaakaer J. Hormone therapy and the impact of estrogen intake on the risk of ovarian cancer. Arch Intern Med. 2004;164:2253–2259. doi: 10.1001/archinte.164.20.2253. [DOI] [PubMed] [Google Scholar]
  12. Hernandez E, Bhagavan BS, Parmley TH, Rosenshein NB. Interobserver variability in the interpretation of epithelial ovarian cancer. Gynecol Oncol. 1984;17:117–123. doi: 10.1016/0090-8258(84)90065-9. [DOI] [PubMed] [Google Scholar]
  13. Higgins JP, Thompson SG. Quantifying heterogeneity in a meta-analysis. Stat Med. 2002;21:1539–1558. doi: 10.1002/sim.1186. [DOI] [PubMed] [Google Scholar]
  14. Hoyo C, Berchuck A, Halabi S, Bentley RC, Moorman P, Calingaert B, Schildkraut JM. Anthropometric measurements and epithelial ovarian cancer risk in African-American and White women. Cancer Causes Control. 2005;16:955–963. doi: 10.1007/s10552-005-3205-y. [DOI] [PubMed] [Google Scholar]
  15. Jordan S, Green A, Webb P. Benign Epithelial Ovarian Tumours-cancer Precursors or Markers for Ovarian Cancer Risk? Cancer Causes Control. 2006;17:623–632. doi: 10.1007/s10552-005-0370-y. [DOI] [PubMed] [Google Scholar]
  16. Jordan SJ, Green AC, Whiteman DC, Webb PM. Risk factors for benign serous and mucinous epithelial ovarian tumors. Obstet Gynecol. 2007;109:647–654. doi: 10.1097/01.AOG.0000254159.75977.fa. [DOI] [PubMed] [Google Scholar]
  17. Kelemen LE, Sellers TA, Schildkraut JM, Cunningham JM, Vierkant RA, Pankratz VS, Fredericksen ZS, Gadre MK, Rider DN, Liebow M, et al. Genetic variation in the one-carbon transfer pathway and ovarian cancer risk. Cancer Res. 2008;68:2498–2506. doi: 10.1158/0008-5472.CAN-07-5165. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Key TJ, Allen NE, Verkasalo PK, Banks E. Energy balance and cancer: the role of sex hormones. Proc Nutr Soc. 2001;60:81–89. doi: 10.1079/pns200068. [DOI] [PubMed] [Google Scholar]
  19. Kobel M, Kalloger SE, Baker PM, Ewanowich CA, Arseneau J, Zherebitskiy V, Abdulkarim S, Leung S, Duggan MA, Fontaine D, et al. Diagnosis of ovarian carcinoma cell type is highly reproducible: a transcanadian study. Am J Surg Pathol. 2010;34:984–993. doi: 10.1097/PAS.0b013e3181e1a3bb. [DOI] [PubMed] [Google Scholar]
  20. Kuskowska-Wolk A, Karlsson P, Stolt M, Rossner S. The predictive validity of body mass index based on self-reported weight and height. Int J Obes. 1989;13:441–453. [PubMed] [Google Scholar]
  21. Lawlor DA, Bedford C, Taylor M, Ebrahim S. Agreement between measured and self-reported weight in older women. Results from the British Women’s Heart and Health Study. Age Ageing. 2002;31:169–174. doi: 10.1093/ageing/31.3.169. [DOI] [PubMed] [Google Scholar]
  22. Lukanova A, Lundin E, Toniolo P, Micheli A, Akhmedkhanov A, Rinaldi S, Muti P, Lenner P, Biessy C, Krogh V, et al. Circulating levels of insulin-like growth factor-I and risk of ovarian cancer. Int J Cancer. 2002;101:549–554. doi: 10.1002/ijc.10613. [DOI] [PubMed] [Google Scholar]
  23. Lurie G, Wilkens LR, Thompson PJ, McDuffie KE, Carney ME, Terada KY, Goodman MT. Combined oral contraceptive use and epithelial ovarian cancer risk: time-related effects. Epidemiology. 2008;19:237–243. doi: 10.1097/EDE.0b013e31816334c5. [DOI] [PubMed] [Google Scholar]
  24. Madore J, Ren F, Filali-Mouhim A, Sanchez L, Kobel M, Tonin PN, Huntsman D, Provencher DM, Mes-Masson AM. Characterization of the molecular differences between ovarian endometrioid carcinoma and ovarian serous carcinoma. J Pathol. 2010;220:392–400. doi: 10.1002/path.2659. [DOI] [PubMed] [Google Scholar]
  25. Merritt MA, Green AC, Nagle CM, Webb PM. Talcum powder, chronic pelvic inflammation and NSAIDs in relation to risk of epithelial ovarian cancer. Int J Cancer. 2008;122:170–176. doi: 10.1002/ijc.23017. [DOI] [PubMed] [Google Scholar]
  26. Moorman PG, Calingaert B, Palmieri RT, Iversen ES, Bentley RC, Halabi S, Berchuck A, Schildkraut JM. Hormonal risk factors for ovarian cancer in premenopausal and postmenopausal women. Am J Epidemiol. 2008;167:1059–1069. doi: 10.1093/aje/kwn006. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Ness RB, Dodge RC, Edwards RP, Baker JA, Moysich KB. Contraception methods, beyond oral contraceptives and tubal ligation, and risk of ovarian cancer. Ann Epidemiol. 2011;21:188–196. doi: 10.1016/j.annepidem.2010.10.002. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Pike MC, Pearce CL, Peters R, Cozen W, Wan P, Wu AH. Hormonal factors and the risk of invasive ovarian cancer: a population-based case-control study. Fertil Steril. 2004;82:186–195. doi: 10.1016/j.fertnstert.2004.03.013. [DOI] [PubMed] [Google Scholar]
  29. Ramus SJ, Vierkant RA, Johnatty SE, Pike MC, Van Den Berg DJ, Wu AH, Pearce CL, Menon U, Gentry-Maharaj A, Gayther SA, et al. Consortium analysis of 7 candidate SNPs for ovarian cancer. Int J Cancer. 2008;123:380–388. doi: 10.1002/ijc.23448. [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Risch HA. Hormonal etiology of epithelial ovarian cancer, with a hypothesis concerning the role of androgens and progesterone. J Natl Cancer Inst. 1998;90:1774–1786. doi: 10.1093/jnci/90.23.1774. [DOI] [PubMed] [Google Scholar]
  31. Risch HA, Bale AE, Beck PA, Zheng W. PGR +331 A/G and increased risk of epithelial ovarian cancer. Cancer Epidemiol Biomarkers Prev. 2006;15:1738–1741. doi: 10.1158/1055-9965.EPI-06-0272. [DOI] [PubMed] [Google Scholar]
  32. 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;16:2548–2556. doi: 10.1158/1055-9965.EPI-07-0550. [DOI] [PubMed] [Google Scholar]
  33. Royar J, Becher H, Chang-Claude J. Low-dose oral contraceptives: protective effect on ovarian cancer risk. Int J Cancer. 2001;95:370–374. doi: 10.1002/1097-0215(20011120)95:6<370::aid-ijc1065>3.0.co;2-t. [DOI] [PubMed] [Google Scholar]
  34. Russell P. Surface Epithelial-Stromal Tumors of the Ovary. In: Kurman RJ, editor. Blaustein’s Pathology of the Female Genital Tract. 4. New York: Springer-Verlag; 1994. pp. 705–782. [Google Scholar]
  35. Sakamoto A, Sasaki H, Furusato M, Suzuki M, Hirai Y, Tsugane S, Fukushima M, Terashima Y. Observer disagreement in histological classification of ovarian tumors in Japan. Gynecol Oncol. 1994;54:54–58. doi: 10.1006/gyno.1994.1165. [DOI] [PubMed] [Google Scholar]
  36. Schouten LJ, Rivera C, Hunter DJ, Spiegelman D, Adami HO, Arslan A, Beeson WL, van den Brandt PA, Buring JE, Folsom AR, et al. Height, body mass index, and ovarian cancer: a pooled analysis of 12 cohort studies. Cancer Epidemiol Biomarkers Prev. 2008;17:902–912. doi: 10.1158/1055-9965.EPI-07-2524. [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Stalsberg H, Abeler V, Blom GP, Bostad L, Skarland E, Westgaard G. Observer variation in histologic classification of malignant and borderline ovarian tumors. Hum Pathol. 1988;19:1030–1035. doi: 10.1016/s0046-8177(88)80082-0. [DOI] [PubMed] [Google Scholar]
  38. Stukel TA, Demidenko E, Dykes J, Karagas MR. Two-stage methods for the analysis of pooled data. Stat Med. 2001;20:2115–2130. doi: 10.1002/sim.852. [DOI] [PubMed] [Google Scholar]
  39. Taylor AW, Dal Grande E, Gill TK, Chittleborough CR, Wilson DH, Adams RJ, Grant JF, Phillips P, Appleton S, Ruffin RE. How valid are self-reported height and weight? A comparison between CATI self-report and clinic measurements using a large cohort study. Aust N Z J Public Health. 2006;30:238–246. doi: 10.1111/j.1467-842x.2006.tb00864.x. [DOI] [PubMed] [Google Scholar]
  40. Terry KL, De Vivo I, Titus-Ernstoff L, Shih MC, Cramer DW. Androgen receptor cytosine, adenine, guanine repeats, and haplotypes in relation to ovarian cancer risk. Cancer Res. 2005;65:5974–5981. doi: 10.1158/0008-5472.CAN-04-3885. [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Troy LM, Hunter DJ, Manson JE, Colditz GA, Stampfer MJ, Willett WC. The validity of recalled weight among younger women. Int J Obes Relat Metab Disord. 1995;19:570–572. [PubMed] [Google Scholar]
  42. Tworoger SS, Eliassen AH, Missmer SA, Baer H, Rich-Edwards J, Michels KB, Barbieri RL, Dowsett M, Hankinson SE. Birthweight and body size throughout life in relation to sex hormones and prolactin concentrations in premenopausal women. Cancer Epidemiol Biomarkers Prev. 2006;15:2494–2501. doi: 10.1158/1055-9965.EPI-06-0671. [DOI] [PubMed] [Google Scholar]
  43. World Health Organisation (WHO) Physical status: the use and interpretation of anthropometry. Report of a WHO Expert Committee. 1995 [PubMed] [Google Scholar]
  44. Wu AH, Pearce CL, Tseng CC, Templeman C, Pike MC. Markers of inflammation and risk of ovarian cancer in Los Angeles County. Int J Cancer. 2009;124:1409–1415. doi: 10.1002/ijc.24091. [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. Ziogas A, Gildea M, Cohen P, Bringman D, Taylor TH, Seminara D, Barker D, Casey G, Haile R, Liao SY, et al. Cancer risk estimates for family members of a population-based family registry for breast and ovarian cancer. Cancer Epidemiol Biomarkers Prev. 2000;9:103–111. [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

supplement
NIHMS486473-supplement.doc (45KB, doc)

RESOURCES