Anthropometry and the risk of epithelial ovarian cancer

Authors


Abstract

BACKGROUND

The association between anthropometric factors and ovarian cancer risk was investigated using data from 762 cases and 1348 controls participating in a population-based case-control study in the Delaware Valley from 1994–1998. Because factors such as oral contraceptive (OC), hormone therapy (HT), and parity may affect weight and hormone levels, the associations were examined in women with and without these characteristics.

METHODS

Unconditional logistic regression was used to calculate odds ratios and 95% confidence intervals while controlling for age, race, parity, family history of ovarian cancer, tubal ligation, and OC use.

RESULTS

Compared with controls, cases were taller and heavier in recent years and at age 18. Results did not differ by OC or HT use. However, anthropometric associations differed significantly based on parity, as increasing anthropometric measures were associated with increased ovarian cancer risk among nulliparous women only. Adjusted OR for recent body mass index (BMI) quartile 4 compared with quartile 1 for nulliparous women was 2.53 (95% confidence interval [CI]: 1.39, 4.61) compared with 0.96 (95% CI: 0.70, 1.31) for parous women. Additionally, adult weight gain was significant only for nulliparous women. Adjusted OR for weight change (recent to age 18) quartile 4 compared with quartile 1 for nulliparous women was 3.73 (95% CI: 1.88, 7.42) versus 1.09 (95% CI: 0.78, 1.51) for parous women.

CONCLUSIONS

BMI and weight in women's adult lifetime may be positively associated with ovarian cancer risk. Observations were most apparent for nulliparous women, possibly reflecting an interaction between local inflammation caused by incessant ovulation and increased estrogen exposure on ovarian epithelium. Cancer 2006. © 2006 American Cancer Society.

Although a high body mass index (BMI) has been associated with many types of cancers, its relation with ovarian cancer has been inconsistent.1–4 The association between adiposity and ovarian cancer is plausible, because adiposity has been associated with alterations in hormone levels and ovulatory function,5 infertility,6 polycystic ovarian syndrome (PCOS),7 hyperandrogenism,8 endometriosis,9 and inflammation,10 all of which have been shown to increase ovarian cancer risk. Moreover, height, weight, and BMI have been associated with other hormonally related cancers, such as endometrial and breast cancer, and lifestyle factors such as parity11 may mediate the association between BMI and these cancers.

We investigated whether height, weight, and BMI in a woman's present and in her past (at age 18), as well as changes in weight and BMI between age 18 and present age, were associated with ovarian cancer risk. Because lifestyle factors such as oral contraceptive (OC) use, parity, and hormone therapy (HT) use may affect weight or hormone levels12, 13 and have also been linked to ovarian cancer,11, 14 we further examined interactions between these factors and weight or BMI.

MATERIALS AND METHODS

Study Subjects

This analysis is based on the SHARE Study, a case-control study of contraceptive and reproductive risk factors for epithelial ovarian cancer. Details of this study have been described elsewhere.15 Briefly, cases were women age 20-69 diagnosed with incident epithelial ovarian cancer within the 9 months before interview. Between May 1994 and July 1998, 873 eligible women were identified at 39 hospitals around the Delaware Valley. Fourteen physicians did not consent to patients' participation and 92 women refused to participate. Thus, 767 completed case interviews (88% of potentially eligible, incident cases) were eligible for the analyses presented in this article. For all cases, the diagnosis of epithelial ovarian cancer was confirmed by pathology.

Controls age 65 or younger were ascertained by random digit dialing and frequency-matched to cases by 5-year age groups and 3-digit telephone exchanges. Of the 14,551 telephone numbers screened for this purpose, we identified 1637 households with a potentially eligible control, of whom 1215 (74%) completed interviews. Controls age 65-69 were ascertained through Health Care Financing Administration (HCFA) lists. Of the 263 potentially eligible participants identified, 152 (58%) were interviewed. Therefore, of the 1900 screened and potentially eligible controls, 1367 (72%) were eligible for our analyses.

Institutional Review Board approval was obtained from all hospitals from which subjects were recruited, and study subjects gave informed consent for participation. Trained interviewers conducted a standardized 1.5-hour in-person interview of cases and controls. Interview data detailed information on a subject's medical history, general demographic and anthropometric data, and gynecologic and obstetric history including use of OCs and number of live births. Questions relating to anthropometric data included self-reported information on height, average weight during the last 5 years (when not pregnant), and average weight at age 18.

For this analysis, only women with complete information for height, recent weight, and weight at age 18 were considered. Nine women, all controls, were missing 1 or all 3 of these variables and were thus excluded. Additional women missing information on number of live births (n = 1) and duration of OC use (n = 14) were excluded. Hence, the main analyses reported here include 762 cases and 1348 controls.

Statistical Analyses

Because matching in this study was based on frequencies for only 2 broad criteria (age within 5-year intervals and 3-digit telephone exchange), the “match” was not preserved in the study. We calculated BMI (weight in kg/height in m2) for current age and at age 18. Anthropometric variables were normally distributed. Thus, initially, t-tests were used to assess differences in anthropometric measures between cases and controls. Subjects were then stratified by quartiles of anthropometric measures as determined by the distribution in the control population. Multivariate unconditional logistic regression methods were used to calculate the odds ratio (OR) and 95% confidence intervals (CIs) for the association between ovarian cancer and quartile 4 of each anthropometric measure compared with quartile 1 of the anthropometric measure. Adjustments were made for age, race (white vs. other), ever parous, family history of ovarian cancer, history of tubal ligation, and OC use (ever vs. never use) based on results of univariate analyses and known risk factors for ovarian cancer.

All analyses were repeated after stratification by ever parous, ever use of OCs, and ever use of HT. Parous women were defined as those reporting a live birth. OC and HT users were women who reported using OCs or HT for at least a 1-month period of time. Interactions between anthropometric factors and stratification variables (ever parous, ever use of OCs, and ever use of HT) were assessed with likelihood ratio tests. Based on these results, we performed exploratory analyses by further substratifying lifestyle factors (HT use, OC use) and menopausal status among nulliparous and parous women. Additionally, we compared nulliparous women who had experienced at least 1 pregnancy (gravid) to nulliparous women who had never been pregnant (nulligravid). Finally, we examined differences in rates of tumor types among cases in relation to the highest versus lowest quartiles of anthropometric measures, first among all women, then comparing parous to nulliparous women. All P-values were 2-sided and considered significant at the level of P<0.05. All analyses were conducted using SAS v. 8 (Cary, NC).

RESULTS

Compared with controls, cases were taller (163.69 cm vs. 162.91 cm, P = 0.01) and heavier both in recent years (69.86 kg vs. 68.36 kg, P = 0.04) and at age 18 (55.93 kg vs. 55.00 kg, P = 0.03). However, recent BMI, BMI at age 18, and changes in weight and BMI did not differ between cases and controls.

In multivariable models adjusted for known ovarian cancer risk factors (Table 1), greater recent weight (adjusted OR, 1.29; 95% CI: 0.98, 1.70 for quartile 4 vs. quartile 1) as well as weight at age 18 (adjusted OR, 1.38; 95% CI: 1.06, 1.80 for quartile 4 vs. quartile 1) appeared to be associated with a greater risk of ovarian cancer. Higher BMI was also associated with a modest but nonsignificant increase in ovarian cancer risk, both for recent BMI (adjusted OR, 1.24; 95% CI: 0.95, 1.63 for quartile 4 vs. quartile 1) and for BMI at age 18 (adjusted OR, 1.18; 95% CI: 0.91, 1.53 for quartile 4 vs. quartile 1). Similarly, greater height was modestly, albeit nonsignificantly, related with ovarian cancer (adjusted OR, 1.26; 95% CI: 0.98, 1.61 for quartile 4 vs. quartile 1).

Table 1. Adjusted Odds Ratios and 95% Confidence Intervals For Height, Weight, and Body Mass Index at Various Life Periods for All Women*
 All women (N = 2110)
 CasesControlsORadj95% CI
  • *

    CI: confidence interval; BMI: body mass index.

  • Odds ratio adjusted (ORadj)for age, race (white/other), number of live births, family history of ovarian cancer, tubal ligation, and oral contraceptive use (ever/never).

Height, cm     
 Quartile 1<157.51793791 
 Quartile 2157.5–162.62313761.260.98, 1.62
 Quartile 3162.7–167.6921771.070.78, 1.48
 Quartile 4167.62604161.260.98, 1.61
   trend P .16 
Recent weight. kg     
 Quartile 1<57.61703391 
 Quartile 257.6–65.32073731.130.87, 1.46
 Quartile 365.4–76.52003241.311.00, 1.72
 Quartile 4>76.51853121.290.98, 1.70
   trend P .04 
Weight at age 18     
 Quartile 1<49.51924011 
 Quartile 249.5–54.02103951.070.83, 1.37
 Quartile 354.1–58.51682671.321.01, 1.74
 Quartile 4>58.51892831.381.06, 1.80
   trend P .006 
Recent BMI     
 Quartile 1<21.81733301 
 Quartile 221.8–24.61963411.100.85, 1.44
 Quartile 324.6–28.71923391.140.87, 1.49
 Quartile 4>28.72013381.240.95, 1.63
   trend P .12 
BMI at age 18     
 Quartile 1<18.71893501 
 Quartile 218.7–20.21653480.920.70, 1.20
 Quartile 320.3–21.91973101.220.94, 1.59
 Quartile 4>22.02083381.180.91, 1.53
   trend P .07 
Weight change, recent; age 18     
 Quartile 1<4.51443001 
 Quartile 24.5–11.22574001.351.03, 1.75
 Quartile 311.3–20.21763371.140.86, 1.52
 Quartile 4>20.21823091.381.04, 1.85
   trend P .11  
BMI change, recent; age 18     
 Quartile 1<1.71913351 
 Quartile 21.7–4.11923371.000.77, 1.30
 Quartile 34.2–7.51853390.990.76, 1.29
 Quartile 4>7.51913351.110.85, 1.46
   trend P .49 

Stratification by ever parity revealed that anthropometric measures were associated with a greater risk of ovarian cancer among nulliparous women (Tables 2, 3). In particular, among nulliparous women higher recent BMI was associated with a 2 1/2-fold increase in ovarian cancer risk (adjusted OR, 2.53; 95% CI: 1.39, 4.61 for quartile 4 vs. quartile 1); among parous women, the relation was nonsignificant (OR, 0.96; 95% CI: 0.70, 1.31 for quartile 4 vs. quartile 1). In addition, weight change since age 18 was significantly related with ovarian cancer among nulliparous women (OR, 3.73; 95% CI: 1.88, 7.42 for quartile 4 vs. quartile 1) but not among parous women (OR, 1.09; 95% CI: 0.78, 1.51 for quartile 4 vs. quartile 1). BMI change was also significantly related with ovarian cancer among nulliparous women only (OR, 2.67; 95% CI: 1.40, 5.11 for quartile 4 vs. quartile 1) and not parous women (OR, 0.91, 95% CI: 0.67, 1.23 for quartile 4 vs. quartile 1). Whereas all interactions between anthropometric variables and parity were significant or of borderline significance, significant interactions were not noted between weight and height variables and OC use, HT use, and menopausal status (data not shown).

Table 2. Adjusted Odds Ratios and 95% Confidence Intervals for Ovarian Cancer by Height, Weight, and BMI at Various Life Periods by Ever Parous*
 CasesControlsNulliparous (n = 406)Parous (n = 1704)
ORadj95% CICasesControlsORadj95% CI
  • *

    CI: confidence interval; BMI: body mass index; OC: oral contraceptives.

  • Adjusted for age, race (white/other), family history of ovarian cancer, tubal ligation, and OC use (ever/never).

  • Adjusted for all preceding variables plus number of live births.

Height, cm        
 Quartile 155321 1243471 
 Quartile 263600.590.33, 1.041683161.491.12, 1.98
 Quartile 340191.310.64, 2.70521580.910.62, 1.34
 Quartile 477600.790.45, 1.391833561.391.05, 1.85
  trend P .99   trend P .14  
      interaction P .002  
Recent weight, kg        
 Quartile 147531 1232861 
 Quartile 266481.300.74, 2.271413251.040.77, 1.41
 Quartile 362351.801.00, 3.241382891.140.84, 1.54
 Quartile 460351.780.99, 3.231252771.100.80, 1.51
 Quartile 1 trend P .03   trend P 0.45  
      interaction P .44  
Weight at age 18        
 Quartile 148521 1443491 
 Quartile 262441.390.79, 2.451483510.970.73, 1.29
 Quartile 358193.311.69, 6.461102481.070.79, 1.46
 Quartile 465551.220.70, 2.101242281.391.03, 1.89
  trend P .26   trend P .03  
      interaction P .004  
Recent BMI        
 Quartile 145551 1282751 
 Quartile 266481.600.92, 2.791302930.960.71, 1.30
 Quartile 355371.670.92, 3.021373020.990.73, 1.34
 Quartile 469312.531.39, 4.611323070.960.70, 1.31
  trend P .003   trend P .85  
      interaction P .06  
BMI at age 18        
 Quartile 142471 1473031 
 Quartile 254341.720.93, 3.171113140.760.56, 1.03
 Quartile 366342.191.20, 4.001312761.020.75, 1.37
 Quartile 471551.400.80, 2.451372831.090.81, 1.46
  trend P .26   trend P .29  
      interaction P .03  
Weight change, recent; age 18        
 Quartile 144611 1002391 
 Quartile 286552.141.26, 3.641713451.170.86, 1.59
 Quartile 353351.810.98, 3.341233020.980.71, 1.36
 Quartile 450193.731.88, 7.421322901.090.78, 1.51
  trend P.001   trend P .99  
       interaction P .03 
BMI change recent; age 18        
 Quartile 159621 1322731.0 
 Quartile 262501.300.77, 2.201302870.930.68, 1.25
 Quartile 359371.530.86, 2.751263020.870.64, 1.19
 Quartile 453212.671.40, 5.111383140.910.67, 1.23
  trend P .003   trend P 0.48  
       interaction P .05 
Table 3. Anthropometric Measures among Cases according to Tumor Type, The SHARE Study 1994–1998
CharacteristicSerous n = 356Endometrioid n = 138Mucinous n = 112Clear Cell n = 60Other n = 96P
MeanSDMeanSDMeanSDMeanSDMeanSD
  1. BMI: body mass index.

Height, cm163.86.6163.05.8164.36.5164.26.8162.75.6.115
Recent weight, kg71.317.269.316.169.118.366.913.968.315.9.233
Weight at age 1856.39.655.89.856.410.554.97.754.87.4.590
Recent BMI26.66.326.15.925.46.324.95.125.86.1.196
BMI at age 1821.03.221.03.720.83.520.42.820.72.9.689
Weight change, recent; age 1814.914.713.413.212.813.312.011.413.012.8.356
BMI change recent; age 185.65.55.05.04.74.94.54.24.94.8.307

We further examined gravidity among nulliparous women (Table 4). Relations between weight and BMI in adulthood and ovarian cancer were evident among both women who had and had not experienced a pregnancy. Among nulligravid nulliparous women, higher recent BMI was associated with an almost 3-fold increase in ovarian cancer risk (adjusted OR, 2.80; 95% CI: 1.39, 5.66 for quartile 4 vs. quartile 1); among gravid nulliparous women, the associated increase in ovarian cancer risk was 2-fold (OR, 1.99; 95% CI: 0.58, 6.88 for quartile 4 vs. quartile 1). Additionally, nulligravid nulliparous women had significant associations of ovarian cancer with BMI change (adjusted OR, 2.49; 95% CI: 1.18, 5.29 for quartile 4 vs. quartile 1). Overall, however, there were no significant interactions with gravidity for nulliparous women in relation to any anthropometric measure examined.

Table 4. Anthropometric Measures among Nulliparous Women Stratified by Gravidity
All Nulliparous Women (n = 406)
CharacteristicNulliparous never gravid (n = 299)Nulliparous gravid ≥1 time (n = 107)
Cases n = 176Controls n = 123ORadj*95% CICases n = 59Controls n = 48ORadj95% CI
  • BMI: body mass index; OC: oral contraceptives.

  • *

    Odds ratio adjusted (ORadj) for age, race (white/other), family history of ovarian cancer, tubal ligation, and OC use (ever/never).

  • Adjusted for age, race (white/other), tubal ligation, and OC use (ever/never). Family history of ovarian cancer was a constant for all nulliparous gravid women.

Height, cm        
 Quartile 143271 1251 
 Quartile 249420.730.38, 1.3814180.230.05, 0.98
 Quartile 327141.250.55, 2.831351.170.22, 6.22
 Quartile 457400.930.49, 1.7620200.400.10, 1.66
  trend P .83   trend P .75  
      interaction P .48  
Recent BMI        
 Quartile 133421 12131 
 Quartile 250341.790.94, 3.4016141.280.40, 4.15
 Quartile 343252.101.05, 4.2012120.740.21, 2.62
 Quartile 450222.801.39, 5.661991.990.58, 6.88
  trend P .004   trend P .43  
      interaction P .83  
BMI at age 18        
 Quartile 129331 13141 
 Quartile 244261.850.91, 3.751081.610.43, 5.97
 Quartile 352232.441.20, 4.9714112.100.60, 7.31
 Quartile 450401.360.70, 2.6421151.570.50, 4.90
  trend P .40   trend P .43  
      interaction P .83  
BMI change, recent; age 18       
 Quartile 144441 15181 
 Quartile 246371.300.70, 2.4116131.630.55, 4.81
 Quartile 347251.800.91, 3.5512120.770.22, 2.72
 Quartile 438162.491.18, 5.291553.911.01, 15.20
  trend P .01   trend P .14  
      interaction P .78  

Finally, we examined height, weight, BMI, weight change, and BMI change by histologic type (serous, endometrioid, mucinous, clear cell, and other) among cases in relation to the highest versus the lowest quartiles of anthropometric measures (data not shown). There were no differences based on histologic type. Stratifying by parity, the relation of anthropometric measures to ovarian cancer was most evident among nulliparous women, where we observed many of the same relations across tumor types (Table 5). For example, for nulliparous women, weight change was associated with serous, mucinous, and other histologic subtypes, and BMI change was associated with both serous and mucinous tumor types. In contrast, parous women only had moderately significant associations of serous tumors with height and significant associations of serous tumors with weight at age 18.

Table 5. Anthropometric Measures Among Cases According to Tumor Type, Nulliparous Versus Parous, The Share Study 1994-1998
 Serous (n = 356)
 Nulliparous  Parous  
 CasesControlsORadj*95% CICasesControlsORadj95% CI
  • OR: odds ratio; 95% CI: 95% confidence interval; BMI: body mass index; OC: oral contraceptives.

  • *

    Adjusted for age, race (white/other), number of live births, family history of ovarian cancer, tubal ligation, and OC use.

  • Adjusted for age, race (white/other), family history of ovarian cancer, tubal ligation, and OC use.

  • P = .003.

  • §

    P <.001.

  • P = .002.

  • P = .03.

  • #

    P = .01.

  • **

    P = .04.

Height, cm        
 First quartile2532  61347  
 Last quartile34600.69(0.34, 1.38)933561.49(1.03, 2.17)
Recent weight, kg        
 First quartile1653  53286  
 Last quartile27352.17(0.97, 4.85)712771.43(0.94, 2.19)
Weight, age 18        
 First quartile2252  62349  
 Last quartile26551.07(0.52, 2.17)662281.68(1.12, 2.50)#
Recent BMI        
 First quartile1555  55275  
 Last quartile29313.44(1.52, 7.80)733071.16(0.76, 1.77)
BMI, age 18        
 First quartile1747  65303  
 Last quartile30551.46(0.70, 3.04)672831.15(0.78, 1.71)
Weight change recent: age 18        
 First quartile1661  47239  
 Last quartile25195.62(2.26, 13.98)§732901.33(0.87, 2.06)
BMI change recent: age 18        
 First quartile2062  64273  
 Last quartile25213.82(1.61, 9.05)753141.01(0.68, 1.50)
 Endometrioid (n = 138)
 Nulliparous  Parous  
 ControlsCasesORadj*95% CICasesControlsORadj95% CI
Height, cm        
 First quartile3211  25347  
 Last quartile60150.69(0.27, 1.73)313561.16(0.66, 2.04)
Recent weight, kg        
 First quartile5310  23286  
 Last quartile35111.13(0.40, 3.22)212771.1(0.58, 2.09)
Weight, age 18        
 First quartile527  35349  
 Last quartile55152.02(0.74, 5.53)222280.99(0.56, 1.74)
Recent BMI        
 First quartile558  20275
 Last quartile31152.71(0.98, 7.56)233071.18(0.60, 2.29)
BMI, age 18        
 First quartile477  28303  
 Last quartile55182.28(0.85, 6.15)242831(0.56, 1.79)
Weight change recent: age 18        
 First quartile6110  21239  
 Last quartile1971.69(0.49, 5.79)242901.03(0.54, 1.96)
BMI change recent: age 18        
 First quartile6213  28273  
 Last quartile2181.42(0.47, 4.35)263140.89(0.50, 1.61)
 Mucinous (n = 112)
 Nulliparous  Parous  
 CasesControlsORadj*95% CICasesControlsORadj95% CI
Height, cm        
 First quartile832  13347  
 Last quartile11600.82(0.28, 2.38)313562.12(1.07, 4.17)
Recent weight, kg        
 First quartile853  18286  
 Last quartile10352.13(0.72, 6.27)132770.95(0.44, 2.04)
Weight, age 18        
 First quartile652  18349  
 Last quartile9551.07(0.34, 3.40)162281.47(0.72, 3.00)
Recent BMI        
 First quartile955  24275  
 Last quartile11312.3(0.82, 6.43)153070.73(0.35, 1.51)
BMI, age 18        
 First quartile547  25303  
 Last quartile9551.39(0.42, 4.63)212831.05(0.56, 1.95)
Weight change recent: age 18        
 First quartile861  15239  
 Last quartile8193.74(1.13, 12.42)132900.90(0.40, 2.01)
BMI change recent: age 18        
 First quartile1062  19273  
 Last quartile10213.13(1.04, 9.41)**153140.84(0.40, 1.74)
 Clear Cell (n = 60)
 Nulliparous  Parous  
 CasesControlsORadj*95% CICasesControlsORadj95% CI
Height, cm        
 First quartile732  4347  
 Last quartile12601.12(0.38, 3.31)83561.78(0.53, 6.04)
Recent weight, kg        
 First quartile953  6286  
 Last quartile6351.01(0.31, 3.29)62771.05(0.33, 3.39)
Weight at age 18        
 First quartile952  9349  
 Last quartile10550.98(0.36, 2.67)52280.91(0.30, 2.78)
Recent BMI        
 First quartile955  8275  
 Last quartile7311.19(0.38, 3.69)53070.45(0.14, 1.47)
BMI at age 18        
 First quartile1047  10303  
 Last quartile9550.74(0.27, 2.00)82830.96(0.37, 2.49)
Weight change recent: age 18        
 First quartile761  5239  
 Last quartile5192.68(0.69, 10.42)72901.15(0.35, 3.80)
BMI change recent: age 18        
 First quartile1162  5273  
 Last quartile5211.37(0.39, 4.79)73141.22(0.37, 4.03)
 Other (n = 96)
 Nulliparous  Parous  
 CasesControlsORadj*95% CICasesControlsORadj95% CI
Height, cm        
 First quartile432  21347  
 Last quartile5600.78(0.18, 3.41)203560.88(0.46, 1.70)
Recent weight, kg        
 First quartile453  23286  
 Last quartile6352.02(0.49, 8.28)142770.58(0.28, 1.19)
Weight, age 18        
 First quartile452  20349  
 Last quartile5551.22(0.29, 5.05)152281.29(0.63, 2.64)
Recent BMI        
 First quartile455  21275  
 Last quartile7312.80(0.71, 11.02)163070.49(0.23, 1.02)
BMI at age 18        
 First quartile347  19303  
 Last quartile5551.36(0.30, 6.19)172831.05(0.52, 2.11)
Weight change recent: age 18        
 First quartile361  12239  
 Last quartile5195.50(1.08, 27.97)152900.85(0.37, 1.94)
BMI change recent: age 18        
 First quartile562  16273  
 Last quartile5212.68(0.64, 11.23)153140.68(0.32, 1.47)

DISCUSSION

Increased weight both early in adulthood and later in life appears to be associated with ovarian cancer. Our observations were most apparent for nulliparous women.

A growing body of data supports the hypothesis that excessive stimulation of ovarian epithelial cells by androgens or unopposed estrogens may increase the risk of ovarian cancer.11 Obesity has been associated with an increased proportion of free and albumin-bound estrogen due to decreased sex hormone binding globulin and increased transformation of androstenedione in adipose tissue, both of which result in having more biologically active estrogen in circulation.16, 17 Thus, greater weight and higher BMI would be associated with greater levels of unopposed estrogens.

Another explanation for the interaction between weight/weight gain and nulliparity in our data may relate to local inflammation. We have posited that incessant ovulation causes a state of local chronic inflammation that induces rapid cell division in the regional ovarian epithelium.10 Increased rates of cell turnover are associated with a greater likelihood of replication errors and ineffective DNA repair at key regulatory sites, such as within tumor suppressor genes, which may increase the probability of converting DNA lesions to mutations.18 This hypothesis is supported by the finding that local inflammants, such as talc and asbestos, have been positively associated with ovarian cancer and that hysterectomy and tubal ligation, which disrupt the path of inflammants to the ovary, negatively influence ovarian cancer risk.19, 20 Other supportive evidence is that a mutation in the tumor suppressor gene p53, thought to result from spontaneous DNA synthesis errors during cell division, has been correlated with an increased number of lifetime ovulations.21–23 We have also posited that, in endometriosis, excess estrogen interacts with inflammation to escalate malignant transformation.24 Similarly, nulliparity, a marker of anovulation and thus chronic inflammation, may interact with weight/weight gain, a marker of estrogenicity, to influence ovarian cancer risk.

History of gravidity in nulliparous women obviates experiencing at least 1 incomplete pregnancy. Whereas a number of studies have shown a protective effect on ovarian cancer risk with number of incomplete pregnancies,25–27 we and others have shown that incomplete pregnancy does not appear to significantly reduce the risk of developing ovarian cancer.28–30 In our present study, we found that gravidity in nulliparous women did not appear to influence their ovarian cancer risk in relation to anthropometric measures.

Past research has suggested that there is etiological heterogeneity in the subtypes of ovarian cancer.31–33 In particular, mucinous tumors appear to be associated with exogenous exposures, such as cigarette smoke.34 Reproductive factors such as parity and oral contraceptive use, which appear to be protective for other histologic subtypes, have been suggested to be less protective in the risk of developing mucinous tumors.31 In our study, we found that nulliparity combined with high anthropometric measures was a risk for both mucinous and nonmucinous tumor types.

There are potential limitations of our study. First, there may have been possible recall bias of self-reported data for height and weight. Women might tend to underreport weight,37, 38 although such underreporting would bias our findings towards the null. Cases may have lost weight due to disease processes, but again this would bias results towards the null. Moreover, women were asked to report their average weight over the past 5 years before ovarian cancer diagnosis (cases) or interview (controls). Second, the small sample size of gravid yet nulliparous women limited our interpretation of these results.

Third, potential confounders, such as physical activity and diet, were not considered. Whereas a previous case-control study of this same group of women demonstrated that ovarian cancer occurrence was reduced by 27% among women in the highest versus the lowest category of lifetime activity,39 other studies have not replicated these results.40 Additionally, the largest prospective study to date, the Iowa Women's Health Study, followed 31,396 postmenopausal women for 7 years and demonstrated that the most active women had twice the ovarian cancer risk of the least active women in their study.41 Previous research has also shown that BMI and body composition are reflective of physical activity level.42–44 Thus, BMI itself could be perceived as a surrogate measure of greater physical activity as well as lower caloric intake. However, if BMI was still significantly associated with ovarian cancer risk after adjustment for physical activity and caloric intake, it would suggest that other factors possibly related with BMI (such as polycystic ovarian syndrome, or hormonal or genetic factors) were driving the association.

The strengths of our study include our large sample size, population-based ascertainment, adjustment for known confounding factors, and the examination of multiple variables related to weight and weight change, thereby affording an opportunity for internal reproducibility of findings.

In conclusion, increased recent weight and BMI and increased weight gain over the course of a woman's adult years are associated with ovarian cancer risk, particularly for nulliparous women. These findings may reflect the adverse effects of unopposed estrogen stimulation and continuous local inflammation on the ovarian epithelium.

Ancillary