Correspondence
Risk of Ovarian Cancer after Treatment for Infertility
N Engl J Med 1995; 332:1300-1302May 11, 1995
- Article
To the Editor:
The article by Rossing et al. (Sept. 22 issue)1 may have generated undue anxiety in many women undergoing treatment for infertility. Although this study does prompt some interesting hypotheses about the future risks of ovarian cancer in patients with ovulatory dysfunction, it does not convincingly link any risk to the use of clomiphene. There are several confounding variables and biases that weaken the validity of the authors' conclusions.
Patients undergoing treatment for infertility, especially those who take medication to induce ovulation, undergo far more surveillance with ultrasound studies than other patients, solely because of their treatment. Furthermore, because of a suspicion bias, any abnormality in a clomiphene-treated patient will probably be evaluated more closely than would an ovarian mass in a patient in the general population.2 The extremely high ratio of borderline tumors to invasive neoplasms identified in this study (5 of 11 neoplasms were borderline) gives strong evidence of a selection bias. This proportion is significantly higher than that usually found in the general population (approximately 1 in 10).3 Since borderline tumors are often asymptomatic, those detected in this study are probably attributable to the intensive screening these patients underwent.
Even more important than the extremely high ratio of borderline tumors to invasive neoplasms is the inclusion of two patients with granulosa-cell tumors. Granulosa-cell tumors often present with abnormalities of fertility and ovulation, and, as has been reported, they may have been the cause of the clomiphene treatment rather than its result.4 Given the small numbers of patients, excluding these two patients would further diminish the importance of this study.
The use of oral contraceptives is a major confounding variable in analyses of the development of ovarian cancer and is not addressed in this study. Oral contraceptives substantially decrease the risk of ovarian cancer even after only one year of use.5 The women undergoing infertility treatment are likely to have used contraception for less time than the women in the control groups. Other confounders, such as a family history of breast or ovarian cancer, though less important, are also not addressed.
Many of these questions might have been answered had the authors reported the tumor stages at diagnosis and the clinical context. If these characteristics differed substantially from those usually found in patients with ovarian neoplasms, then the conclusions would probably be due to unrecognized confounding variables and potential biases. We hope that this article leads to additional studies in this area. However, we believe that no changes in current practices are indicated on the basis of this report.
5 ReferencesGiuseppe Del Priore, M.D., M.P.H.
Kathleen Robischon, M.D.
William R. Phipps, M.D.
University of Rochester Medical Center, Rochester, NY 146421
Rossing MA ,Daling JR ,Weiss NS ,Moore DE ,Self SG . Ovarian tumors in a cohort of infertile women. N Engl J Med 1994;331:771-776
Full Text | Web of Science | Medline2
Feinstein AR. Clinical epidemiology: the architecture of clinical research. Philadelphia: W.B. Saunders, 1985.
3
Hart WR. Pathology of malignant and borderline (low malignant potential) epithelial tumors of ovary. In: Coppleson M, ed. Gynecologic oncology: fundamental principles and clinical practice. 2nd ed. Vol. 2. Edinburgh, Scotland: Churchill Livingstone, 1992:863-87.
4
Jansen R . Ovarian stimulation and granulosa-cell tumour. Lancet 1993;341:1345-1345
CrossRef | Web of Science | Medline5
Herbst AL ,Berek JS . Impact of contraception on gynecologic cancers. Am J Obstet Gynecol 1993;168:1980-1985
Web of Science | Medline
To the Editor:
We were surprised to see a deviation from the accepted classification system in the article about the relation between an ovulation-inducing agent (clomiphene citrate) and ovarian tumors. In the Results section, the authors refer to two types of ovarian tumors that they identify as invasive or borderline malignant. They proceed to subclassify their five “borderline malignant” tumors into mucinous cystadenomas (two cases), papillary serous cystadenomas (two cases), and papillary mucinous cystadenomas of low malignant potential (one case).
The World Health Organization histologic classification of ovarian tumors considers the terms “borderline” and “low malignant potential” to be equivalent. These neoplasms are considered intermediate in their behavior between benign cystadenomas and frankly malignant invasive carcinomas. Accordingly, only one of the five cases reported by the authors as borderline is, in fact, borderline — the one reported as a “papillary mucinous cystadenoma of low malignant potential.” The other four cases are benign neoplasms, and to include them in the “borderline” category is misleading. We question whether the conclusion that clomiphene citrate has an adverse effect in terms of the development of truly malignant or borderline ovarian tumors would still stand if these four benign ovarian cysts were excluded from the analyses.
Robert Kurman, M.D.
Edward E. Wallach, M.D.
Howard A. Zacur, M.D., Ph.D.
Johns Hopkins University, Baltimore, MD 21287-1201To the Editor:
In her editorial (Sept. 22 issue)1 on the association between the use of fertility medications and ovarian cancer, Whittemore states that “concern about the risk of ovarian cancer associated with the use of fertility medications was heightened by . . . findings from a combined analysis” of three case–control studies.2-4 The authors of the three studies used in that analysis5,6 kindly provided me with the previously unpublished data shown in Table 1Table 1
Use of Fertility Drugs in Three Case–Control Studies of Ovarian Cancer..Various decision rules were applied to the pooled analysis (Hartge P: personal communication), and frankly invasive tumors2 and tumors of low malignant potential3 were reported on separately. However, if the data are pooled for simplicity, then among 914 cases there was one exposure to clomiphene. As for the other drugs to which the patients were exposed, estrogens do not increase the risk of ovarian tumors, and when combined with progestogens they decrease it. Thyroid hormone and “speed” (dextroamphetamine and amobarbital) do not merit serious consideration; nor do unknown drugs, since the great majority of the women were treated in the early 1960s, when ovulation-inducing drugs were either unavailable (e.g., clomiphene) or hardly available (e.g., menotropins). Whatever the explanation is for the association described in the combined analysis,5,6 it was not due to clomiphene, and it is exceedingly unlikely that it could have been due to other fertility medications. The data give no grounds for heightened concern.
The validity of the findings of Rossing et al. is also in doubt, because of the acknowledged potential sources of error and bias and for three additional reasons. First, epithelial and granulosa-cell tumors cannot be considered as a single outcome because their embryologic, pathological, and epidemiologic features differ. It is also questionable whether borderline and invasive tumors can be combined. Insufficient numbers of tumors were studied for the three outcomes to be considered separately. Second, 6 of the 11 patients (54 percent) had infertility associated with ovarian abnormalities that may themselves have increased the risk of ovarian cancer; precancerous disease could have “caused” clomiphene therapy, rather than the reverse. Finally, several inferences are based on small numbers and are thus unjustified. For example, the contrast between the receipt of clomiphene for 1 to 11 cycles as compared with 12 or more cycles is based on three and five patients with exposure, respectively, yielding markedly unstable estimates of relative risk of 0.8 and 11.1. This difference was interpreted as substantive when it could easily have been due to chance (P>0.1).
There are legitimate grounds for concern that fertility medications may increase the risk of ovarian cancer, and the matter is being studied. To date, however, the quantitative evidence to support that concern consists of equivocal data from a single study.
6 ReferencesSamuel Shapiro, M.B., F.R.C.P.(E.)
Boston University School of Medicine, Brookline, MA 021461
Whittemore AS . The risk of ovarian cancer after treatment for infertility. N Engl J Med 1994;331:805-806
Full Text | Web of Science | Medline2
Hartge P ,Schiffman MH ,Hoover R ,McGowan L ,Lesher L ,Norris HJ . A case-control study of epithelial ovarian cancer. Am J Obstet Gynecol 1989;161:10-16
Web of Science | Medline3
Cramer DW ,Hutchison GB ,Welch WR ,Scully RE ,Ryan KJ . Determinants of ovarian cancer risk. I. Reproductive experiences and family history. J Natl Cancer Inst 1983;71:711-716
Web of Science | Medline4
Nasca PC ,Greenwald P ,Chorost S ,Richart R ,Caputo T . An epidemiologic case-control study of ovarian cancer and reproductive factors. Am J Epidemiol 1984;119:705-713
Web of Science | Medline5
Whittemore AS, Harris R, Itnyre J, Collaborative Ovarian Cancer Group
Web of Science | Medline6
Harris R, Whittemore AS, Itnyre J, Collaborative Ovarian Cancer Group
Web of Science | Medline
Author/Editor Response
The authors reply:
To the Editor: We agree with Dr. Shapiro that our findings are limited by the small number of tumors that developed in the study cohort, which points to the need for larger studies of this possible association. However, we observed an increased risk of an ovarian tumor associated with the use of clomiphene during 12 or more menstrual cycles, among women either with or without a known ovarian abnormality; this finding suggests that the association is not entirely attributable to the presence of ovarian disease that leads to clomiphene use.
Although Dr. Del Priore et al. note that the ratio of borderline tumors to invasive ovarian cancers is approximately 1:10 among the general population of women, borderline tumors constitute a much higher proportion of the tumors diagnosed in women in the age range of our study population. On the basis of data from the Cancer Surveillance System of western Washington State from 1989 through 1992, 43 percent of the ovarian tumors reportable to the Surveillance, Epidemiology and End Results program of the National Cancer Institute among women 20 to 50 years of age were borderline tumors. Also, although ultrasound screening during infertility treatment could potentially have resulted in the diagnosis of borderline tumors that might otherwise have remained occult, such screening does not appear to account for our study findings: 9 of the 11 women in whom ovarian tumors developed were no longer being treated for infertility at the time of the tumor diagnosis. If we eliminate women with granulosa-cell tumors from our analyses, the elevated risk associated with exposure to 12 or more cycles of clomiphene, though reduced, remains quite high (relative risk, 6.7; 95 percent confidence interval, 0.8 to 58.8). Although uncontrolled confounding by factors such as oral-contraceptive use may have influenced our results, we believe such confounding is not likely to be present to any important degree, because the comparisons were made within the study cohort of women evaluated for infertility.
Finally, we thank Dr. Kurman and colleagues for the opportunity to clarify our description of the histologic features of borderline tumors. The problematic phrase should have read: “Two mucinous cystadenomas, two papillary serous cystadenomas, and one papillary mucinous cystadenoma, each of low malignant potential.”
Mary Anne Rossing, D.V.M., Ph.D.
Janet R. Daling, Ph.D.
Noel S. Weiss, M.D., Dr.P.H.
Fred Hutchinson Cancer Research Center, Seattle, WA 98104Author/Editor Response
In attempting to piece together more precise information on fertility drugs from three studies included in the analysis by our Collaborative Ovarian Cancer Group,1 Shapiro has incorrectly inferred the specific numbers of subjects using fertility-drug therapy, because he did not follow our rules for the exclusion of subjects. Four of the case patients included in his table (three studied by Hartge et al. and one studied by Cramer et al.) had borderline tumors and were excluded from our analysis of invasive ovarian cancers. Nor did we count as exposed the one case patient (and one control) in the study by Cramer et al. who used diethylstilbestrol only during a pregnancy, or the one case patient studied by Hartge et al. who did not see a physician for her infertility. Thus, these two studies contributed 8 exposed case patients with invasive ovarian cancer, whereas Nasca et al. contributed 12 case patients (not 6). In addition, many of the controls who had undergone hysterectomy in each of the three studies were excluded because of the uncertain status of their ovaries. (All the exclusion rules applicable to all subjects, regardless of disease or exposure status, were determined by the investigators before the start of any analyses.) As a result, the total number of controls used in our study was 1101, not 1364 as indicated by Shapiro, and the number of exposed controls was 5 in the studies by Hartge and Cramer and 6 in that by Nasca. Thus, the majority of the exposed case patients and controls were contributed by the study by Nasca et al., in which information on the specific fertility drugs used was not requested. In all, 16 of the 20 exposed case patients either were not asked or could not recall the medications they used, and 12 of these women were not asked when they used them.
Overall, fertility-drug treatment was reported by 20 of 622 case patients with invasive ovarian cancer (3.2 percent), as compared with 11 of 1101 controls (1.0 percent) — a significant difference that continues to be the basis of our concern about such exposure in relation to the risk of ovarian cancer. At this point, we do not think that ad hoc reanalyses of the data (especially those done without knowing the details of the study) are of particular use in addressing what, in Shapiro's own words, are “legitimate grounds for concern that fertility medications may increase the risk of ovarian cancer.”
2 ReferencesDaniel W. Cramer, M.D.
Brigham and Women's Hospital, Boston, MA 02115Patricia Hartge, Sc.D.
National Cancer Institute, Rockville, MD 28092Philip C. Nasca, Ph.D.
University of Massachusetts, Amherst, MA 01003Alice S. Whittemore, Ph.D.
Stanford University School of Medicine, Stanford, CA 943051
Whittemore AS, Harris R, Itnyre J, Collaborative Ovarian Cancer Group
Web of Science | Medline2
Rossing MA ,Daling JR ,Weiss NS ,Moore DE ,Self SG . Ovarian tumors in a cohort of infertile women. N Engl J Med 1994;331:771-776
Full Text | Web of Science | Medline
- Citing Articles (7)
Citing Articles
1
Nikos F. Vlahos, Konstantinos P. Economopoulos, George Creatsas. (2010) Fertility drugs and ovarian cancer risk: a critical review of the literature. Annals of the New York Academy of Sciences 1205:1, 214-219
CrossRef2
Louise A. Brinton, Kamran S. Moghissi, Bert Scoccia, Carolyn L. Westhoff, Emmet J. Lamb. (2005) Ovulation induction and cancer risk. Fertility and Sterility 83:2, 261-274
CrossRef3
Raffi Chalian, Fred Licciardi, Andrei Rebarber, Giuseppe Del Priore. (2004) Successful Infertility Treatment in a Cancer Patient with a Significant Personal and Family History of Cancer. Journal of Women's Health 13:2, 235-237
CrossRef4
GIUSEPPE PRIORE, J. RICHARD SMITH, DEBORAH C.M. BOYLE, DAVID J. CORLESS, FARIS B.P. ZACHARIA, DAVID A. NOAKES, THOMAS DIFLO, JAMES A. GRIFO, JOHN J. ZHANG. (2001) Uterine Transplantation, Abdominal Trachelectomy, and Other Reproductive Options for Cancer Patients. Annals of the New York Academy of Sciences 943:1, 287-295
CrossRef5
W PHIPPS. (2001) Polycystic Ovary Syndrome And Ovulation Induction. Obstetrics and Gynecology Clinics of North America 28:1, 165-182
CrossRef6
Adeola Olaitan, Amanda Mocroft, Ian Jacobs. (2000) Patterns in the incidence of age-related ovarian cancer in South East England 1967-1996. BJOG: An International Journal of Obstetrics and Gynaecology 107:9, 1094-1096
CrossRef7
Kirsten Duckitt, Allan A. Templeton. (1998) Cancer in women with infertility. Current Opinion in Obstetrics and Gynaecology 10:3, 199-203
CrossRef
