Infertility and Ovarian Cancer Ness et al. Infertility, Fertility Drugs, and Ovarian Cancer: A Pooled Analysis of Case- Control Studies

Transcription

1 American Journal of Epidemiology Copyright 2002 by the Johns Hopkins Bloomberg School of Public Health All rights reserved Vol. 155,. 3 Printed in U.S.A. Infertility and Ovarian Cancer Ness et al. Infertility, Fertility Drugs, and Ovarian Cancer: A Pooled Analysis of Case- Control Studies Roberta B. Ness, 1 Daniel W. Cramer, 2 Marc T. Goodman, 3 Susanne Krûger Kjaer, 4 Kathy Mallin, 5 Berit Jul Mosgaard, 6 David M. Purdie, 7 Harvey A. Risch, 8 Ronald Vergona, 1 and Anna H. Wu 9 Controversy surrounds the relations among infertility, fertility drug use, and the risk of ovarian cancer. The authors pooled interview data on infertility and fertility drug use from eight case-control studies conducted between 1989 and 1999 in the United States, Denmark, Canada, and Australia. Odds ratios and 95% confidence intervals were calculated, adjusting for age, race, family history of ovarian cancer, duration of oral contraception use, tubal ligation, gravidity, education, and site. Included in the analysis were 5,207 cases and 7,705 controls. Among nulligravid women, attempts for more than 5 years to become pregnant compared with attempts for less than 1 year increased the risk of ovarian cancer 2.67-fold (95% confidence interval (CI): 1.91, 3.74). Among nulliparous, subfertile women, neither any fertility drug use (odds ratio (OR) = 1.60, 95% CI: 0.90, 2.87) nor more than 12 months of use (OR = 1.54, 95% CI: 0.45, 5.27) was associated with ovarian cancer. Fertility drug use in nulligravid women was associated with borderline serous tumors (OR = 2.43, 95% CI: 1, 5.88) but not with any invasive histologic subtypes. Endometriosis (OR = 1.73, 95% CI: 1.10, 2.71) and unknown cause of infertility (OR = 1.19, 95% CI: 0, 1.40) increased cancer risk. These data suggest a role for specific biologic causes of infertility, but not for fertility drugs in overall risk for ovarian cancer. Am J Epidemiol 2002;155: fertility agents, female; infertility; ovarian neoplasms Considerable public attention and patient anxiety during the past decade have focused on the possible link between fertility drugs and ovarian cancer risk. Some studies suggested that women who took fertility drugs, yet did not become pregnant, as well as women with longer durations of fertility drug exposure, are at substantially elevated risk of ovarian cancer (1, 2). However, results from other studies have not been consistent, even among potentially high-risk subgroups of women with high lifetime exposures to fertility drugs (3 10). Interpretation of an association between fertility drugs and ovarian cancer is complicated by the fact that infertility Received for publication December 6, 2000, and accepted for publication July 31, Abbreviations: CI, confidence interval; OR, odds ratio. 1 University of Pittsburgh School of Public Health, Pittsburgh, PA. 2 Brigham and Women s Hospital, Harvard Medical School, Boston, MA. 3 Cancer Research Center, University of Hawaii, Honolulu, HI. 4 Danish Cancer Society, Institute of Cancer Epidemiology, Copenhagen, Denmark. 5 University of Illinois at Chicago, Chicago, IL. 6 Department of Obstetrics and Gynecology, Herlev Hospital, University of Copenhagen, Copenhagen, Denmark. 7 Queensland Institute of Medical Research, Royal Brisbane Hospital, Queensland, Australia. 8 Yale University School of Medicine, New Haven, CT. 9 University of Southern California, Los Angeles, CA. Reprint requests to Dr. Roberta B. Ness, University of Pittsburgh, 130 DeSoto Street, 517 Parran Hall, Pittsburgh, PA ( repro@vms.cis.pitt.edu). may elevate ovarian cancer risk (1 3, 6 9, 11 16). Infertility is a heterogeneous end product of a number of biologically distinct conditions. There have been several attempts to segregate infertility by type and to examine each infertility type separately as it relates to ovarian cancer risk. These studies have been inconclusive, with some showing an association to ovarian cancer and ovulatory infertility (2, 16) and some to infertility of unknown cause (3, 7). The increasing use of fertility drugs in many countries suggests an emerging need to separate an association of infertility with ovarian cancer from that of infertility treatment. Key obstacles to studying this question include the rarity of the disease and the relatively infrequent prescription of ovulation-inducing drugs. We report the results of an international pooled analysis of infertility and fertility drug use in eight ovarian cancer case-control investigations. Our analysis included a large sample of subjects in which to 1) determine whether specific infertility diagnoses were associated with ovarian cancer and 2) examine whether, among women with subfertility, fertility drugs elevated ovarian cancer risk. MATERIALS AND METHODS Data collection Table 1 summarizes the characteristics of the eight casecontrol studies included in this analysis. All of the studies were conducted between 1989 and 1999, and each used population-based case and control selection. Four studies were 217

2 218 Ness et al. TABLE 1. Study sites and their characteristics included in the pooled analysis of eight case-control studies, Authors (reference no.) Cramer et al. (18) Location Massachusetts and New Hampshire Period of ascertainment Case selection Treatment centers Control selection Random digit dialing and town lists. 563 Cases % recruited Controls % recruited 72 Age range (years) Matching parameters Individual age, residence Mallin et al.* Cook County, Illinois Hospitals Random digit, HCFA for 64 years Frequency age, race Goodman et al. (17) Hawaii and Los Angeles County, California SEER registries, HCFA files Frequency age, race, and location (Hawaii, Los Angeles) Kjaer et al.* Mosgaard et al. (8) Ness et al. (19) Purdie et al. (21) Risch et al. (20) Denmark, selected counties Denmark Philadelphia, Pennsylvania Australia Ontario, Canada conducted among women living on the mainland United States (17 19) (one of which included women living in Hawaii (17)), two were conducted in Denmark (8) (Kjaer et al., Danish Cancer Society, unpublished data), one in Canada (20), and one in Australia (21). Only one study separately published an analysis of fertility drug use and ovarian cancer (8). Six additional studies were excluded from consideration in this pooled analysis because they did not meet the timing criteria of conduct on or after 1989 or did not collect data on fertility drugs or both (five studies) or because they had not completed data collection (one study). All data in this report were analyzed in Pittsburgh, Pennsylvania, using datasets stripped of individual identifiers. Infertility and fertility drug use Hospitals National Hospitals Treatment centers Cancer registry Standardized in-person interviews were conducted in all but one (8) of the studies. That study mailed self-administered questionnaires. In four of the studies (17, 19) (Kjaer et al., Danish Cancer Society, and Mallin et al., University of Illinois at Chicago, unpublished data) a modification of a single questionnaire was used. This questionnaire based recall for reproductive events on a life calendar approach in which each woman marked a calendar with important events during her life and used these to enhance her memory for date-related information. The studies by Risch et al. (20) and Purdie et al. (21) also used a life calendar as a memory aid. Data were checked for internal consistency, and when necessary, clarifications were requested from the investigators. Random digit dialing, HCFA files * Unpublished data. HCFA, Health Care Financing Administration; SEER, Surveillance, Epidemiology, and End Results. Percent recruited in Hawaii. Percent recruited in Los Angeles, California We requested that each study provide data on the following independent variables in the following format: sought medical attention for infertility (yes/no), infertility tests done (partner/self/both/neither), infertility diagnosis (type), fertility drug use (duration), specific fertility drug used (duration), and histology (cases only). Time spent attempting pregnancy (continuous) was also obtained. In most studies, this had been collapsed into a single variable encompassing cumulative duration of trying over a lifetime. Covariates Each study also provided data on potentially important covariates that we have previously found to be significantly related to ovarian cancer risk (22): age (continuous), gravidity (continuous), first-degree family history of ovarian cancer (yes/no), race (Black, White, Hispanic, Asian, other), educational attainment (less than high school, high school, more than high school), tubal ligation (ever/never), oral contraceptive use duration (continuous), and study code (categorical). Statistical analysis 1,562 1,721 1, Pooled odds ratios, with corresponding 95 percent confidence intervals, were calculated as the primary measure of effect size. Because most studies used frequency rather than individual matching and matched on the basis of broad criteria, such as age within 5- to 10-year intervals, we used Frequency age Individual age, residence Frequency age, residence Frequency age, urban/rural Frequency age Am J Epidemiol Vol. 155,. 3, 2002

3 Infertility and Ovarian Cancer 219 unconditional logistic regression models to adjust for any additional effects of age as well as for gravidity, race, education, history of ovarian cancer in any first-degree relative, tubal ligation, oral contraceptive use duration, and study, as indicated above. We checked the reasonableness of pooling estimates by calculating odds ratios within individual studies and calculating a Woolf test for heterogeneity among studies for all major results (23). In none of the significant associations between infertility type and ovarian cancer and in none of the fertility drug and ovarian cancer analyses did we find statistically significant heterogeneity among studies. We did find heterogeneity when comparing four or more pregnancies with zero pregnancies and in comparing neverpregnant women trying for less than 1 year to get pregnant with those trying for more than 5 years. In both situations, we provide the heterogeneity statistic and range of odds ratios in Results. In evaluating risk associated with fertility drugs, we restricted analyses to women who had a history of 2 or more years of trying to conceive over a lifetime and/or of seeking medical attention for infertility. Both seeking infertility treatment and prolonged episodes of trying to conceive without success (typically for at least a 1-year period) are standard markers of infertility. The term subfertility rather than infertility is used here to describe women who tried to conceive for 2 or more years and/or sought medical attention because our data on trying comprises a cumulative period over a lifetime rather than the duration of a single episode. Confining our analyses solely to care seekers might have focused on an unrepresentative group of women with enhanced access to health care and higher socioeconomic status. Furthermore, preliminary analyses revealed that, although almost all women using fertility drugs tried for 2 or more years to get pregnant, 29.9 percent of the controls and 33.5 percent of the cases used fertility drugs yet did not report seeking medical attention specifically for infertility. One data set (Mallin et al., unpublished data) did not capture length of pregnancy attempts, but instead asked about trying for at least 1 year; that definition was used to identify subfertility among those women. We analyzed use and duration of use of fertility drugs in general and then duration of the specific drugs clomiphene and human menopausal gonadotrophin. Too few women reported exposure to other specific drugs to present those results. Logistic regression was used to explore the interaction of infertility and fertility drug use by modeling each level of interaction between the pairs of variables infertility (yes/no) and fertility drug use (yes/no) using subjects who were fertile and who never used fertility drugs as the reference category. Although other variables were fairly readily combined across studies, the variable for which categorization differed most between studies was type of infertility. One study (8) had no data on infertility types and, therefore, was not included in these analyses. For other datasets, whenever possible, we collapsed medically similar categories, but for some categories we could not include datasets in which information was unavailable. Datasets that were not included in specific analyses were as follows: uterus development problems (18, 21), menstruation problems (18), and Am J Epidemiol Vol. 155,. 3, 2002 cervical mucous/inflamed cervix (21). We also defined a group of women as having an unknown cause of infertility by taking all women so designated in the data and adding any women who had infertility tests, but did not report a specific diagnosis. To account for multiple infertility diagnoses, we limited the comparison group to women with no infertility diagnosis and present analyses adjusted for standard confounders plus all types of infertility beyond the one of interest. Finally, for infertility types that appeared to elevate ovarian cancer risk and for fertility drug use, we separately analyzed histologic subtypes, including borderline serous (n 489), invasive serous (n 1,782); mucinous (n 773), endometrioid and clear cell (n 985), and undifferentiated and all other types (n 420). RESULTS Among the 7,705 controls, 10.2 percent were never pregnant, and among the 5,207 cases, 19.6 percent were never pregnant. In adjusted analyses, women who had never been pregnant were 2.42 times (95 percent confidence interval (CI): 1.86, 3.14) as likely to have ovarian cancer as were those with four or more pregnancies (table 2). Estimates of this association were heterogeneous between studies (p < 0.005), with odds ratios ranging from 1.69 (95 percent CI: 1.25, 2.29) to 3.75 (95 percent CI: 2.80, 5.01). Seeking medical attention for fertility problems was a modest risk factor for ovarian cancer, with the same odds ratio, 1.2, among ever pregnant and never pregnant women. Prolonged episodes of trying also elevated ovarian cancer risk, particularly among never pregnant women, wherein trying for more than 5 years versus less than 1 year without achieving pregnancy elevated the risk of ovarian cancer 2.7-fold (95 percent CI: 1.91, 3.74). Estimates for more than 5 years versus less than 1 year of trying among nulligravid women were also heterogeneous (p < 0.005), with odds ratios ranging from 0.85 (95 percent CI: 0.28, 2.59) to (95 percent CI: 8.91, 28.60). In those trying for more than 5 years who became pregnant, the risks were times higher than among women trying for less than 1 year and were not significant. Two types of infertility (table 3), endometriosis (odds ratio (OR) 1.73, 95 percent CI: 1.10, 2.71) and unknown cause of infertility (OR 1.19, 95 percent CI: 0, 1.43), were independently associated with elevations in risk after adjustment for standard confounding factors. Further adjustment for each of the other types of infertility did not greatly alter the strength of these associations (table 3); however, with reduced power, only unknown cause of infertility continued to be significant. Fertility drug use overall did not raise the risk of ovarian cancer in adjusted analyses (table 4). Among the 2,397 women defined as subfertile (those who sought medical attention for infertility or tried for at least 2 years to get pregnant), 14.6 percent took fertility drugs. The odds ratio of ovarian cancer for subfertile women ever taking these medications versus not taking them was 0.97 (95 percent CI: 0.76, 1.25). Among never pregnant women, the risk for ever

4 220 Ness et al. TABLE 2. Fertility measures among cases and controls, crude and adjusted odds ratios for ovarian cancer for eight case-control studies, Variable. of cases. of controls Crude OR* 95% CI Adjusted OR 95% CI. of pregnancies , , ,020 2,224 1,830 2, , , , , , , , , 3.14 Sought medical attention Never pregnant Ever pregnant Time spent trying to get pregnant (years) Never pregnant <1 1 2 >2 5 >5 One pregnancy (years) <1 1 2 >2 5 >5 Two pregnancies (years) <1 1 2 >2 5 >5 Three or more pregnancies (years) <1 1 2 >2 5 > , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , 1.68 * Odds ratios (OR) of represent the referent categories. They are adjusted for age, gravidity, race, education, history of ovarian cancer, tubal ligation, oral contraceptive use duration and research site. CI, confidence interval. Excludes Cramer et al. (18). Excludes Cramer et al. (18) and Mallin et al. (unpublished data). taking fertility drugs was elevated, but not significantly so (OR 1.60, 95 percent CI: 0.90, 2.87), and among ever pregnant women, the risk was less than (OR 0.82, 95 percent CI: 0.62, 9). Further adjustment for each of the types of infertility resulted in odds ratios of 1.75 (95 percent CI: 0.73, 4.21) for nulligravid women and 0.69 (95 percent CI: 0.48, 0) for gravid women. Longer duration (>12 months) of fertility drug exposure did not significantly elevate ovarian cancer risk, even among nulligravid women (OR 1.54, 95 percent CI: 0.45, 5.27), nor did longer duration of specific drugs (i.e., clomiphene or human menopausal gonadotrophin) significantly elevate risk. Separate results for histologic subtypes of ovarian cancers did not reveal any additional associations, with the following exceptions. Among nulligravid women, fertility drug use was significantly associated with borderline serous tumors (OR 2.43, 95 percent CI: 1, 5.88), but not with invasive serous tumors (OR 1.11, 95 percent CI: 0.51, 2.42) or with borderline or invasive mucinous tumors, endometrioid or clear cell tumors, or undifferentiated or other tumors. Among gravid women, fertility drugs were not significantly associated with any of these histologic types. Serous borderline tumors were associated with ovarian cysts (OR 3.75, 95 percent CI: 2.15, 6.51), uterine development prob- Am J Epidemiol Vol. 155,. 3, 2002

5 Infertility and Ovarian Cancer 221 TABLE 3. Infertility types and ovarian cancer risk for eight case-control studies, Variable. of cases. of controls Crude OR* 95% CI* Adjusted OR 95% CI Adjusted OR 95% CI Low sperm count , , , 1.75 Ovulation problems , , , 1.26 Ovarian cysts , , , 4.46 Blocked tubes Uterine development problems Cervical mucous and/or inflamed cervix# Menstrual problems** Endometriosis Unknown cause of infertility 84 2, , , , , , , , , , , , , , , , , , 1.43 * OR, odds ratio; CI, confidence interval. Odds ratios were adjusted for age, gravidity, race, education, history of ovarian cancer, tubal ligation, duration of oral contraceptive use, and research site. Odds ratios were adjusted for age, gravidity, race, education, history of ovarian cancer, tubal ligation, duration of oral contraceptive use, and research site plus each other infertility type. Excludes Mosgaard et al. (8). Excludes Mosgaard et al. (8), Cramer et al. (18), and Purdie et al (21). # Excludes Mosgaard et al. (8) and Purdie et al. (21). ** Excludes Mosgaard et al. (8) and Cramer et al. (18) , , , , , , 1.46 lems (OR 2.73, 95 percent CI: 7, 7.00), menstrual problems (OR 2.34, 95 percent CI: 1.33, 4.14), endometriosis (OR 2.63, 95 percent CI: 1.10, 6.26), and unknown cause of infertility (OR 1.86, 95 percent CI: 1.26, 2.74). Beyond this effect, which may well represent a surveillance bias, we found endometrioid/clear cell tumors associated with endometriosis (OR 3.41, 95 percent CI: 1.94, 5.99) and unknown cause of infertility (OR 1.94, 95 percent CI: 1.49, 2.54). Furthermore, unknown cause of infertility related to serous invasive ovarian cancer (OR 1.86, 95 percent CI: 1.26, 2.74). We further assessed the independent and joint relations among infertility, fertility drug use, and ovarian cancer in a logistic regression model that adjusted for standard confounders. Infertility was significantly related to ovarian cancer (OR 1.26, 95 percent CI: 1.14, 1.39). Fertility drug use was not related to ovarian cancer (OR 1.13, 95 percent CI: 0.56, 2.29) nor was the multiplicative interaction between infertility and fertility drugs (OR 0.79, 95 percent CI: 0.37, 1.66) related to ovarian cancer. DISCUSSION In our pooled analysis, we found no association between fertility drug use and the overall risk of ovarian cancer. Neither longer duration of fertility drug use nor unsuccessful fertility drug use was independently associated with significant elevations in adjusted cancer risk. We did find that, Am J Epidemiol Vol. 155,. 3, 2002

6 222 Ness et al. TABLE 4. Fertility drug use and ovarian cancer risk* for eight case-control studies, Variable. of cases. of controls Crude OR 95% CI Adjusted OR 95% CI Fertility drugs (all) , , , 1.25 Fertility drugs Never pregnant , , 2.87 Ever pregnant , , 9 Duration of fertility drug use (months) Never pregnant Never >12 Ever pregnant Never >12 Chlomiphene duration (months) Never >12 HMG duration (months) Never > , , among women who never achieved a pregnancy, fertility drugs were related to an elevated risk of borderline serous tumors but not to invasive tumors. Although the association between fertility drugs and borderline serous tumors may be real and important, it might also be a significant result arising from multiple comparisons. Furthermore, the association may reflect surveillance or diagnostic bias with enhanced surveillance occurring among young women who have invasive, fertility-related procedures; borderline tumors generally arise among younger women who have an excellent life expectancy after tumor excision (24, 25). This assertion is strengthened by the many nonspecific associations between serous borderline tumors and various infertility types. Overall, our findings conflict with some previous reports (1, 2, 9) but not with others (3, 6, 7, 10), suggesting , , , , , , , , , , , , , , , , , , , , , , , , * Includes only women who tried to get pregnant for 2 or more years and/or sought medical attention for infertility. Odds ratios (OR) were adjusted for age, gravidity, race, education, history of ovarian cancer, tubal ligation, duration of oral contraceptive use, and research site. CI, confidence interval; HMG, human menopausal gonadotropin. Excludes Purdie et al. (21). Excludes Mosgaard et al. (8) and Purdie et al. (21). that it may be premature to ascribe a causal link between fertility drug use and the subsequent development of ovarian cancer (26 29). We did find strong relations among nulligravidity, prolonged attempts to become pregnant (although these results were heterogeneous among studies), and ovarian cancer risk, suggesting that subfertile women are at increased risk of ovarian cancer even after adjustment for contraception and pregnancies. Both lack of gravidity and attempts at pregnancy were more strongly related to risk than seeking medical attention for infertility among nulligravid women. These results may be explained by the observation that most women who sought medical attention subsequently became pregnant. At the same time, many women who were nulligravid never sought medical attention (table 2). Am J Epidemiol Vol. 155,. 3, 2002

7 Infertility and Ovarian Cancer 223 Endometriosis and unknown cause of infertility have been linked to ovarian cancer in a variety of clinical and epidemiologic studies, including our own (3, 4, 7, 22, 30 33). Indeed, endometriosis has specifically been linked to endometrioid and clear cell tumors, perhaps as part of a natural progression. That we did not have data on time between infertility workup and ovarian cancer diagnosis introduces the possibility that endometriosis may represent a precursor of ovarian cancer. It is also possible that inflammation plays a role in ovarian carcinogenesis. This may explain the link between ovarian cancer and endometriosis, with its marked local inflammation (34, 35). It may further explain the link to infertility of unknown origin, perhaps through abnormal prostaglandin responses, which mediate both ovulation and inflammation (36, 37). We did not find that either ovulatory infertility or menstrual infertility (perhaps a surrogate marker for anovulation) elevated invasive ovarian cancer risk. Although such relations have been reported in the past (2, 16), current hypotheses regarding the etiology of ovarian cancer do not predict a causal association. Both the ovulation hypothesis and the gonadotrophin hypothesis postulate that fewer ovulations resulting from ovulatory infertility and/or lowerpeak estrogen and progesterone levels associated with anovulation would reduce, not elevate, ovarian cancer risk. Strengths of this pooled analysis include its size, generalizability, population-based ascertainment strategy, and the use of structured, detailed data collection methods, including data on infertility and fertility drug use. Weaknesses include the relatively small proportion of women using fertility drugs among the entire population (2.7 percent), even among subfertile women (14.6 percent), and the even smaller number of women on whom we had specific information regarding specific types of fertility drugs used. Another study weakness is reliance on self-reports for fertility evaluations and fertility drug use. We know of no data regarding distant recall of infertility type or fertility drugs per se, but there are data on distant recall of contraceptive hormones that suggest women were unlikely to err in recall of ever versus never use of oral contraceptives (38 42). Typically, women with medical record documentation of use reported utilizing these medications 80 percent of the time or more. However, duration of oral contraceptive use and type of oral contraceptives used were less likely to be recalled correctly. On a related topic, it is likely that diagnosis and classification of fertility type may have varied by study site. We did observe variability in the proportion of women with specific fertility diagnoses by study site, yet we did not observe significant heterogeneity in the association between type of infertility and ovarian cancer by site. Nevertheless, these case-control data suggest that infertility per se, but not the use of fertility medications, elevates the overall risk of ovarian cancer. ACKNOWLEDGMENTS The studies involved in this pooled analysis received grant support, and the authors wish to acknowledge contributors as follows: Goodman et al.: grants R01-CA and N01-CN from the National Cancer Institute, Castle Am J Epidemiol Vol. 155,. 3, 2002 Memorial Hospital, Kaiser Foundation Hospital, Kapiolani Medical Center, Kuakini Medical Center, Queens Medical Center, Straub Clinic and Hospital, St. Francis Hospital, Tripler Army Hospital, and Wahiawa General Hospital; Ness et al.: grant R01-CA63748 from the National Cancer Institute; Cramer et al.: grant RO1-CA from the National Cancer Institute; Risch et al.: grant from the National Health Research and Development Program of Health Canada; Purdie et al.: Australian National Health and Medical Research Council and the Queensland Cancer Fund; Mallin et al.: grant R01-CA61093 from the National Cancer Institute; Kjaer et al.: grant R01- CA from the National Cancer Institute; Mosgaard et al.: grants from Løvens Kemiske Fabriks Forskningsfond, Stud. Med. Karsten Hansens Grant, vocare, Fabrikant Einar Willumsens Grant, Danish Hospital Foundation for Medical Research, Region of Copenhagen, The Faroe Island and Greenland (J77/94), Minister Erna Hamiltons Grant for Science and Art, Agnes og Poul Friis-Fund, The Danish Medical Association Research Fund (J086.51), and Max and Anna Friedmanns Grant, Copenhagen, Denmark. The authors also acknowledge the contributions of the following persons: Ness et al.: Jeanne Anne Grisso and Jennifer Klapper; Purdie et al.: Adele Green, Christopher Bain, Victor Siskind, Bruce Ward, Neville Hacker, Michael Quinn, Peter Russell, Gordon Wright, and Beatrice Susil; Mallin et al.: coinvestigators Karin Rosenblatt, Faith Davis, Viswanath, and Candice Zahora; Kjaer et al.: coinvestigator Eva Glud, Jan Blaakaer, and Claus Hoegdall. REFERENCES 1. Whittemore AS, Harris R, Itnyre J. Characteristics relating to ovarian cancer risk: collaborative analysis of 12 US casecontrol studies. II. Invasive epithelial ovarian cancer in white women. Am J Epidemiol 1992;136: Rossing MA, Daling JR, Weiss NS, et al. Ovarian tumors in a cohort of infertile women. N Engl J Med 1994;331: Ron E, Lunenfeld B, Menczer J, et al. Cancer incidence in a cohort of infertile women. Am J Epidemiol 1987;125: Shu WO, Brinton LA, Gao YT, et al. -based casecontrol study in Shanghai. Cancer Res 1989;49: Franceschi S, La Vecchia C, Negri E, et al. Fertility drugs and risk of epithelial ovarian cancer in Italy. Hum Reprod 1994;9: Modan B, Ron E, Lerner-Geva L, et al. Cancer incidence in a cohort of infertile women. Am J Epidemiol 1998;147: Venn A, Watson L, Lumley J, et al. Breast and ovarian cancer incidence after infertility and in vitro fertilization. Lancet 1995; 346: Mosgaard BJ, Lidegaard O, Kjaer SK, et al. Infertility, fertility drugs, and invasive ovarian cancer: a case-control study. Fertil Steril 1997;67: Shushan A, Paltiel O, Iscovich J, et al. Human menopausal gonadotropin and the risk of epithelial ovarian cancer. Fertil Steril 1996;65: Potashnik G, Lerner-Geva L, Genkin L, et al. Fertility drugs and the risk of breast and ovarian cancers: results of a longterm follow-up study. Fertil Steril 1999;71: Cramer DW, Hutchison GB, Welch WR, et al. Determinants of ovarian cancer risk. I. Reproductive experiments and family history. J Natl Cancer Inst 1983;71:

8 224 Ness et al. 12. Hartge P, Schiffman MH, Hoover R, et al. A case-control study of epithelial ovarian cancer. Am J Obstet Gynecol 1989;161: Booth M, Beral V, Smith P. Risk factors for ovarian cancer: a case-control study. Br J Cancer 1989;60: Nasca PC, Greenwald P, Chorost S, et al. An epidemiologic case-control study of ovarian cancer and reproductive factors. Am J Epidemiol 1984;119: Rodriguez C, Tatham LM, Calle EE, et al. Infertility and risk of fatal ovarian cancer in a prospective cohort of US women. Cancer Causes Control 1998;9: Brinton LA, Melton LJ 3rd, Malkasian GD Jr, et al. Cancer risk after evaluation for infertility. Am J Epidemiol 1989;129: Goodman MT, McDuffie K, Kolonel LN, et al. Case-control study of ovarian cancer and polymorphisms in genes involved in catecholestrogen formation and metabolism. Cancer Epidemiol Biomarkers Prev 2001;10: Cramer DW, Harlow BL, Titus Ernstoff L, et al. Over-thecounter analgesics and risk of ovarian cancer. Lancet 1998;351: Ness RB, Grisso JA, Klapper J, et al. Risk of ovarian cancer in relation to estrogen and progestin dose and use characteristics of oral contraceptives. Am J Epidemiol 2000;152: Risch HA, Marrett LD, Howe GR. Parity, contraception, infertility, and the risk of epithelial ovarian cancer. Am J Epidemiol 1994;140: Purdie D, Green A, Bain C, et al. Reproductive and other factors and risk of epithelial ovarian cancer: an Australian casecontrol study. Int J Cancer 1995;62: Ness RB, Grisso JA, Cottreau C, et al. Factors related to inflammation of the ovarian epithelium and risk of ovarian cancer. Epidemiology 2000;11: Woolf B. On estimating the relation between blood group and disease. Ann Hum Genet 1955:19: Kehoe S, Powell J. Long-term follow-up of women with borderline ovarian tumors. Int J Gynecol Obstet 1996;53: Barnhill DR, Kurman RJ, Brady MF, et al. Preliminary analysis of the behavior of stage I ovarian serous tumors of low malignant potential: a Gynecologic Oncology Group study. J Clin Oncol 1995;13: Glud E, Kjaer SK, Troisi R, et al. Fertility drugs and ovarian cancer. Epidemiol Rev 1998;20: Artini PG, Fasciani A, Cela V, et al. Fertility drugs and ovarian cancer. Gynecol Endocrinol 1997;11: Bristow RE, Karlan BY. Ovulation induction, infertility, and ovarian cancer risk. Fertil Steril 1996:66: Kaufman SC, Spirtas R, Alexander NJ. Do fertility drugs cause ovarian tumors? J Womens Health 1995;4: Heaps JM, Nieberg RK, Berek JS. Malignant neoplasms arising in endometriosis. Obstet Gynecol 1990;75: Sainz de la Cuesta R, Eichhom JH, Rice LW, et al. Histologic transformation of benign endometriosis to early epithelial ovarian cancer. Gynecol Oncol 1996;60: Moll CM, Chumas JC, Chalas E, et al. Ovarian carcinoma arising in atypical endometriosis. Obstet Gynecol 1990;75: Brinton LA, Gridley G, Persson I, et al. Cancer risk after a hospital discharge diagnosis of endometriosis. Am J Obstet Gynecol 1997;176: Ness RB, Cottreau C. Possible role of ovarian epithelial inflammation in ovarian cancer. J Natl Cancer Inst 1999;91: Risch HA, Howe GR. Pelvic inflammatory disease and the risk of epithelial ovarian cancer. Cancer Epidemiol Biomarkers Prev 1995;4: Fosslien E. Biochemistry of cyclooxygenase (COX)-2 inhibitors and molecular pathology of COX-2 in neoplasia. Crit Rev Clin Lab Sci 2000;37: Hillier SG, Tetsuka M. An anti-inflammatory role for glucocorticoids in the ovaries. J Reprod Immunol 1998;39: West SL, Savitz DA, Koch G, et al. Recall accuracy for prescription medications: self-reported compared with database information. Am J Epidemiol 1995;142: Coulter A, Vessey M, McPherson K. The ability of women to recall their oral contraceptive histories. Contraception 1986; 33: Harlow SD, Linet MS. Agreement between questionnaire data and medical records: the evidence for accuracy of recall. Am J Epidemiol 1989;129: Stolley PD, Tonascia JA, Sartwell PE, et al. Agreement rates between oral contraceptive users and prescribers in relation to drug use histories. Am J Epidemiol 1978;107: Wingo PA, Lee NC. Use of a life-calendar to enhance the quality of exposure and risk factors histories. (Abstract). Am J Epidemiol 1988;128:921. Am J Epidemiol Vol. 155,. 3, 2002