|
|
| |||||||||||||||||||||||||||||||||||||||
Background Case reports and the results of a recent case-control study have raised questions about the potential neoplastic effects of medications used as treatment for infertility.
Methods We examined the risk of ovarian tumors in a cohort of 3837 women evaluated for infertility between 1974 and 1985 in Seattle. Computer linkage with a population-based tumor registry was used to identify women in whom tumors were diagnosed before January 1, 1992. Data on infertility testing and treatment were abstracted from the medical records of women who had ovarian cancer and those of a randomly selected comparison group. The risk of ovarian tumors associated with exposure to ovulation-inducing medications was assessed through an age-standardized comparison with the rate of ovarian tumors in the general population, and Cox regression analysis was used to compare the risk of cancer among women who received these medications with the risk among infertile women who did not receive them.
Results There were 11 invasive or borderline malignant ovarian tumors, as compared with an expected number of 4.4 (standardized incidence ratio, 2.5; 95 percent confidence interval, 1.3 to 4.5). Nine of the women in whom ovarian tumors developed had taken clomiphene; the adjusted relative risk among these women, as compared with that among infertile women who had not taken this drug, was 2.3 (95 percent confidence interval, 0.5 to 11.4). Five of the nine women had taken the drug during 12 or more monthly cycles. This period of treatment was associated with an increased risk of ovarian tumors among both women with ovarian abnormalities and those without apparent abnormalities (relative risk, 11.1; 95 percent confidence interval, 1.5 to 82.3), whereas treatment with the drug for less than one year was not associated with an increased risk.
Conclusions Prolonged use of clomiphene may increase the risk of a borderline or invasive ovarian tumor.
We performed a case-cohort study to determine whether infertile women have an increased risk of ovarian tumors and, if so, whether that risk is influenced by the apparent cause of the infertility or the treatment received for it.
Methods
Cohort Identification
The cohort was composed of 3837 women evaluated for infertility at participating clinics in Seattle between January 1, 1974, and December 31, 1985. The members of the cohort met the following eligibility criteria: at the time of the evaluation, they resided in the 13-county area of western Washington covered by the Cancer Surveillance System (CSS), a population-based tumor registry operating as part of the Surveillance, Epidemiology and End Results (SEER) program of the National Cancer Institute; they had attempted conception for a period of at least one year; and they had made at least two visits to an infertility clinic participating in the study.
Ascertainment of Cases of Cancer
Identifying information, including name, address, date of birth, and social security number, was collected for each member of the cohort from the records of the infertility clinic where she was evaluated. These data were linked by computer to CSS records to identify women who received a diagnosis of cancer after enrollment in the study and before January 1, 1992. In addition to ascertaining all tumors considered reportable by SEER, since 1974 the CSS has ascertained ovarian tumors of low malignant potential. Also known as borderline ovarian tumors, these neoplasms exhibit some features of cancer and may metastasize but are not invasive7.
Subcohort Selection
A comparison group of 135 women were randomly selected from the cohort in four strata for age at the time of enrollment (<25, 25 to 29, 30 to 34, and 
35 years old) and in three strata for the period of enrollment (1974 through 1977, 1978 through 1981, and 1982 through 1985). For each stratum, the number of women selected was three times as large as the number of women with the most common type of cancer in that stratum.
Ascertainment of Exposure
We reviewed the clinic records of the women with ovarian tumors and the members of the subcohort. The results of this review and the treating physician's interpretation of the results of infertility tests, including the basal body temperature, hysterosalpingography, the postcoital test, analysis of the partner's semen, endometrial biopsy, and laparoscopy, were used to classify the causes of infertility. Information on the menstrual, contraceptive, and reproductive history was also collected, as well as the dosages and types of medications received for the treatment of infertility and the dates of the treatment.
The classification of ovulatory abnormalities was based on the presence of abnormal menstrual cycles (a pattern other than regular cycles of 21 to 35 days) recorded at the time of enrollment in the cohort or on the presence of an abnormality in the basal body temperature (according to the treating physician's interpretation of this test) at a time when no medication was being administered for infertility. Ovulatory infertility was classified as amenorrhea (no menstruation without medication or a period of 6 months or longer between cycles), oligomenorrhea (36 to 180 days between cycles), anovulation (no shift in the basal body temperature, as noted by the physician), or a luteal-phase defect (an elevated temperature for less than 11 days or an abnormal pattern of temperature elevation, as noted by the physician). In addition, we assessed the risk of cancer among the women with polycystic ovaries diagnosed by a physician.
Abnormalities of the fallopian tubes were documented on the basis of evidence of such abnormalities on the hysterosalpingogram or at laparoscopy or laparotomy. Endometriosis was documented on the basis of evidence of this condition at laparoscopy or laparotomy. Male-factor infertility was defined as a minimum of two semen analyses with abnormal results or an abnormal sperm-penetration assay, or a previous vasectomy in the male partner.
Calculation of the Period at Risk
In general, the date of enrollment in the cohort was the date of the second visit to the clinic. However, if a woman had visited a clinic two or more times before one year of attempted conception had elapsed, then the enrollment date was the date of the first visit after one year of attempted conception. If the evaluation of infertility began before 1974, the date of enrollment was the date of the first visit to the clinic in 1974.
To determine whether cohort members were still residing in the CSS area at the end of the follow-up period, a computer-generated list of licensed drivers in Washington State in 1991 was matched to the cohort. For the women not listed as licensed drivers in the state, other procedures were used to determine whether they had moved, including credit-bureau tracings and searches of the National Death Index.
For women residing in the CSS area, the period at risk was calculated as the time from enrollment in the cohort to the end of the study (December 31, 1991). For the women who had moved, the end of the follow-up period was considered to be the first date at which each was known to be residing outside the study area. For the women who died during the study, the end of the follow-up period was considered to be the date of death. Women who were not located were assumed to have remained in the area until the end of the study.
Statistical Analysis
Age-standardized incidence ratios were calculated to compare the rates of cancer in the cohort with those in the general population of the CSS area. The stratified ratio method8 was used to calculate the expected numbers of cancers in subgroups of the cohort, with age-specific cancer rates reported by the CSS from 1974 to 1991. Fisher's exact 95 percent confidence intervals9 were calculated for the age-standardized incidence ratios.
For within-cohort analyses, Cox regression analysis (with age as the time axis) was used to estimate relative risks and 95 percent confidence intervals, with the necessary adjustment of the variance estimates due to cohort sampling10,11,12. Potential confounding factors investigated included parity, gravidity, use of oral contraceptives, age at the birth of the first child, body weight (all assessed at the time of enrollment in the cohort), age at menarche, and type of infertility. Because ovulation-inducing drugs were used primarily during follow-up, such use was treated as a time-dependent variable.
Cohort Follow-up
The women ranged in age from 17 to 44 years at the time of enrollment and contributed 43,438 person-years of observation. Nearly all the women (96.6 percent) were under 50 years of age at the end of follow-up (Table 1). Most (85 percent) were still residing in the CSS area at the end of the study; 10.9 percent were located outside the study area. Only 0.5 percent of the women were known to have died, and 3.6 percent could not be traced. The social security numbers were unknown for one third of the untraced women -- a considerably larger proportion than that of the whole cohort (4.4 percent).
|
Eleven women were identified as having invasive or borderline malignant ovarian tumors. A variety of histologic types were reported, including four invasive epithelial tumors (two mucinous cystadenocarcinomas, one endometrioid carcinoma, and one adenocarcinoma); five borderline epithelial tumors (two mucinous cystadenomas, two papillary serous cystadenomas, and one papillary mucinous cystadenoma of low malignant potential); and two granulosa-cell tumors. Tumors were diagnosed an average of 6.9 years after enrollment in the cohort (range, 1.8 to 16.4). The 11 women ranged in age from 24 to 43 years at the time of the diagnosis (median, 39).
The risk of cancer (at all anatomical sites, in situ tumors excluded) in the cohort was similar to that in the female population of western Washington (age-standardized incidence ratio, 0.9; 95 percent confidence interval, 0.7 to 1.2; 58 observed cases occurring vs. 62.5 expected cases). The relative risk of an ovarian tumor was 2.5 (95 percent confidence interval, 1.3 to 4.5). Although the risk of an invasive epithelial ovarian tumor was somewhat increased (age-standardized incidence ratio, 1.5; 95 percent confidence interval, 0.4 to 3.7), the risk of a borderline tumor was substantially higher than that expected on the basis of rates in the general population of women (age-standardized incidence ratio, 3.3; 95 percent confidence interval, 1.1 to 7.8).
Abnormalities of ovulation and male-factor infertility were the most common causes of infertility in the subcohort (Table 2). Although treatment with clomiphene citrate was most common among the women with ovulatory abnormalities (85.4 percent, n = 41), more than half the women without such abnormalities had taken this medication (52.9 percent, n = 46). Approximately one fourth of the subcohort received human chorionic gonadotropin, whereas only 4.4 percent received human menopausal gonadotropin. Among the women using clomiphene, the duration of treatment was unknown for five subcohort members and one woman with an ovarian tumor.
|
12 cycles, 55.6 percent) became pregnant. The women who used clomiphene were more likely than the unexposed women to have ovulatory abnormalities (47.1 percent vs. 14.9 percent). Male-factor infertility and endometriosis were more common among the long-term users than among the short-term users of clomiphene (50.0 percent vs. 29.7 percent and 38.9 percent vs. 15.6 percent, respectively). Among the short-term users, the most common type of infertility was an ovulatory abnormality (diagnosed in 51.6 percent of the short-term users, as compared with 33.3 percent of the long-term users). Both the women with infertility due to ovulatory abnormalities and those with infertility due to other causes were at increased risk for ovarian tumors, as compared with the general population of women in western Washington (Table 3). The women who had used clomiphene or other ovulation-induction agents were at increased risk (age-standardized incidence ratio associated with clomiphene use, 3.1), whereas little increase in risk was observed among women with no history of exposure to these drugs (age-standardized incidence ratio, 1.4).
|
|
|
12 cycles, as compared with the risk associated with clomiphene use for 0 to 11 cycles, was 9.1 (95 percent confidence interval, 1.0 to 86.5); among the women with ovulatory abnormalities, the relative risk associated with long-term use was 7.4 (95 percent confidence interval, 1.0 to 53.1). Among nulligravid women, the risk of ovarian tumors associated with the use of clomiphene for 12 cycles, as compared with the risk associated with clomiphine use for 0 to 11 cycles, was 10.8 (95 percent confidence interval, 1.5 to 77.9), whereas among gravid women, the relative risk was 17.0 (95 percent confidence interval, 1.2 to 242.8). There was no increase in the risk of ovarian tumors associated with the use of human chorionic gonadotropin when case patients were compared with subcohort members (adjusted relative risk, 1.0; 95 percent confidence interval, 0.2 to 4.3). Furthermore, we observed no trend in the risk associated with the duration of treatment (data not shown). The median duration of treatment among both case patients and subcohort members was 3.0 months. Because of the low prevalence of exposure to other drugs that induce ovulation, we did not estimate the relative risks associated with their use.
Discussion
Our study had a number of limitations, among which was the imprecision of our estimate of the number of person-years of risk for the occurrence of ovarian tumors among women in western Washington. Most of the women in the cohort (72 percent) were identified as residents of the study area on the basis of 1991 records of licensed drivers in Washington State. Women identified as living in western Washington on the basis of credit tracing were also considered residents of the study area. In addition, we included the small proportion of unlocated cohort members (3.6 percent) as residents of western Washington. To the extent that we have overestimated the years during which cohort members resided in western Washington, we have overestimated the expected number of ovarian tumors occurring in the cohort. Such an overestimation would result in an underestimation of age-standardized incidence ratios.
An additional potential source of error is the occurrence of oophorectomy among cohort members. If the proportion of women in the cohort who underwent this procedure was higher or lower than that in the general population, the comparisons with population rates are biased. Also, if certain subgroups of infertile women (e.g., those with ovulatory abnormalities or those who had taken clomiphene) underwent oophorectomy more frequently or less frequently than did other infertile women, our within-cohort comparisons may also be affected.
In some instances, particularly in the case of women whose surnames had changed and whose social security numbers were unknown, we may have failed to match cohort members correctly to CSS records. However, given the small proportion of cohort members with unknown social security numbers (and the high proportion of CSS records containing social security numbers), we expect that a minimal proportion of tumors occurring among cohort members residing in the study area were not identified.
Some of the infertile women did not undergo all the diagnostic procedures, and some of these women may have had unrecognized abnormalities. Such a misclassification would be likely to be nondifferential (i.e., occurring to an equal extent in cases and noncases) and would reduce our ability to observe a relation between the risk of tumor and a particular type of abnormality.
Women may receive ovulation-inducing drugs from more than one doctor. In general, information about prior treatment at infertility clinics not participating in the study was available in the medical charts, although a small proportion of the women had taken clomiphene for an unknown period before enrollment in the cohort. Also, some women may have initiated treatment at a new clinic after undergoing treatment at a study clinic. Information not contained in the medical records or concerning events that occurred after the evaluation for infertility at the study clinics was not available for this analysis.
Only a small number of tumors developed during the follow-up period, limiting the precision of our risk estimates. Also, our ability to examine the relation between the risk of cancer and specific types of ovulatory abnormalities and between the use of clomiphene and histologic subtypes of tumors is limited.
The relatively long interval between enrollment in the cohort and the diagnosis of a tumor (mean, 6.9 years) suggests that the observed increase in risk was not due to the diagnosis of preexisting tumors that resulted in infertility. Also, all but two women in whom ovarian tumors developed were no longer being followed at an infertility clinic at the time of the diagnosis, suggesting that most diagnoses were not the result of increased surveillance during the evaluation for infertility.
The relative risk for the development of borderline tumors was somewhat higher than that for the development of invasive cancer, as compared with the rates in the general population. In addition to a particularly strong etiologic relation, the relatively high risk of borderline ovarian tumors may reflect an increased likelihood of a diagnosis at an early stage in the progression of the tumor (if borderline ovarian tumors progress to invasive tumors) or the diagnosis of some borderline tumors that would otherwise have remained occult in women who had previously received care for infertility.
Interpretation of age-standardized incidence ratios is limited by the possibility that infertile women may differ from the general population with regard to characteristics (such as the pregnancy history) that influence the risk of ovarian tumors. However, the calculated age-standardized incidence ratios were generally consistent with the results of within-cohort analyses, which showed a doubling of the risk associated with ovulatory abnormalities and with the use of clomiphene. The increased risk associated with clomiphene was most evident among women who had used the drug during 12 or more menstrual cycles.
Neither of two previous cohort studies13,14 that assessed the risk of ovarian cancer among infertile women reported a clearly increased risk relative to that expected on the basis of rates in the general population. Although not specifically stated, it is likely that only invasive cancers were considered in these studies. Since in one of the studies13 women were evaluated for infertility between 1935 and 1964, their exposure to clomiphene (which was first approved for use in the United States in the mid-1960s15) or other ovulation-inducing agents was likely to have been minimal.
A series of articles on the results of a combined analysis of 12 U.S. case-control studies reported increased risks of invasive epithelial ovarian cancer,4 borderline epithelial ovarian tumors,5 and nonepithelial ovarian cancer6 associated with the use of fertility drugs. Interpretation of these results is limited by the absence of information about the types of drugs used or the duration of their use. Among gravid women, there was little increased risk of invasive epithelial cancer associated with the use of fertility drugs (relative risk, 1.4; 95 percent confidence interval, 0.5 to 3.6), whereas among nulligravid women, the risk was substantially increased (relative risk, 27.0; 95 percent confidence interval, 2.3 to 315.6)4. In contrast, our results suggest that the risk was elevated among both gravid and nulligravid women. A possible explanation for the differences between these two sets of results is that, in the case-control study, the women who did not become pregnant may have used ovulation-inducing drugs longer than those who conceived.
In the present study, clomiphene was commonly used by infertile women, whether or not they had ovulatory abnormalities. The risk of ovarian tumors associated with long-term use of clomiphene was increased among both the women with ovulatory abnormalities and those with no known ovulatory abnormalities. These results suggest that the increased risk associated with the use of clomiphene is not merely a reflection of the presence of ovarian abnormalities that may be indications for treatment with this drug. Although our findings raise the possibility that prolonged use of clomiphene increases the risk of ovarian tumors, additional, larger studies are needed to test this hypothesis.
Supported in part by a grant (R35 CA-39779) from and a contract (NO1-CN-05230) with the National Cancer Institute.
The contents of this article are solely the responsibility of the authors and do not necessarily represent the official views of the National Cancer Institute.
Source Information
From the Departments of Epidemiology (M.A.R., J.R.D., N.S.W.) and Biostatistics (S.G.S.), School of Public Health and Community Medicine, and the Department of Obstetrics and Gynecology, School of Medicine (D.E.M.), University of Washington; and the Fred Hutchinson Cancer Research Center (M.A.R., J.R.D., N.S.W., S.G.S.) -- all in Seattle.
Address reprint requests to Dr. Rossing at Fred Hutchinson Cancer Research Center, 1124 Columbia St., Seattle, WA 98104.
References
| |||||||||||||||||||||||||||||||||||||||
Related Letters:
Risk of Ovarian Cancer after Treatment for Infertility
Del Priore G., Robischon K., Phipps W. R., Kurman R., Wallach E. E., Zacur H. A., Shapiro S., Rossing M. A., Daling J. R., Weiss N. S., Cramer D. W., Hartge P., Nasca P. C., Whittemore A. S.
Extract |
Full Text
N Engl J Med 1995;
332:1300-1302, May 11, 1995.
Correspondence
This article has been cited by other articles:
HOME | SUBSCRIBE | SEARCH | CURRENT ISSUE | PAST ISSUES | COLLECTIONS | PRIVACY | HELP | beta.nejm.org Comments and questions? Please contact us. The New England Journal of Medicine is owned, published, and copyrighted © 2008 Massachusetts Medical Society. All rights reserved. |
Previous
