As mentioned previously, since 1985, nearly 140 cases of tamoxifen(Drug information on tamoxifen)-associated uterine cancers have been reported. [1-13] The anecdotal nature of many of these small series, while suggesting an association between tamoxifen treatment for breast cancer and the development of endometrial cancer, fails to provide conclusive evidence for its occurrence. Perhaps the strongest data initially implicating tamoxifen use and the subsequent development of endometrial cancer were published in 1989 by Fornander et al.  The authors reviewed the frequency of new primary cancers as recorded in the Swedish Cancer Registry for a group of 1,846 postmenopausal women with early breast cancer who were included in a randomized trial of adjuvant tamoxifen. They noted a 6.4-fold increase in the relative risk of endometrial cancer in 931 tamoxifen-treated patients, compared with 915 patients in the control group. The tamoxifen dosage in this study was 40 mg/day, and the greatest cumulative risk of developing endometrial cancer was after 5 years of tamoxifen use.
NSABP B-14—Fisher et al  recently published the most compelling data to date regarding the association between tamoxifen use and the development of endometrial cancer when they reported the findings of the NSABP B-14 trial. Data regarding the rates of endometrial and other cancers were analyzed on 2,843 patients with node-negative, ER-positive, invasive breast cancer randomly assigned to placebo or tamoxifen (20 mg/day), and on 1,220 tamoxifen-treated patients registered in NSABP B-14 subsequent to randomization. Two of the 1,424 patients assigned to receive placebo developed endometrial cancer; however, both had subsequently received tamoxifen for treatment of breast cancer recurrence. Fifteen patients randomized to tamoxifen treatment developed endometrial cancer. One of these patients never actually accepted tamoxifen therapy. Eight additional cases of uterine cancer occurred in the 1,220 tamoxifen-treated patients, with women age 60 or older accounting for 76% of cases. The mean duration of tamoxifen therapy was 35 months, with 36% of the endometrial cancers developing within 2 years of therapy, and six occurring less than 9 months after treatment was initiated, suggesting that some of the cancers may have been present prior to starting tamoxifen therapy.
The average annual hazard rates for endometrial cancer were 0.2/1,000 in the placebo group and 1.6/1,000 in the randomized tamoxifen-treated group. The relative risk of an endometrial cancer occurring in the randomized, tamoxifen-treated group was 7.5. Similar results were seen in the 1,220 registered patients who received tamoxifen.
Caveats—Although the findings indicate that the average annual and cumulative hazard rates of occurrence of endometrial cancer are greater in tamoxifen-treated patients than in those receiving placebo, the authors point out several caveats.
First, there may have been a bias in detection, as tamoxifen-treated patients are more often symptomatic and more likely, therefore, to seek gynecologic consultation.
Second, the rate of endometrial cancer in the placebo-treated group appeared to be unusually low. Surveillance, Epidemiology, and End Results (SEER) data  would predict that 6.9 cancers would occur in the placebo group. Using SEER data as a comparison, the relative risk of endometrial cancer in the randomized tamoxifen-treated group would be 2.2. In addition, findings from the NSABP B-06 trial noted seven endometrial cancers occurring in 1,159 breast cancer patients with negative lymph nodes who did not receive systemic therapy. This yielded an average annual hazard rate for endometrial cancer of 0.7/1,000. Comparing this with the rate of 1.6/1,000 for endometrial cancer in the randomized tamoxifen-treated group from the B-14 trial, the relative risk for endometrial cancer with tamoxifen use is 2.3, similar to that obtained using SEER data.
Finally, any conclusions regarding the risks of tamoxifen treatment inducing endometrial cancer must weigh the benefits of tamoxifen in reducing breast cancer recurrence and new contralateral breast cancers. In the B-14 trial, the cumulative rate of breast cancer relapse was reduced from 227.8/1,000 in the placebo group to 123.5/1,000 in the randomized tamoxifen-treated group. In addition, the cumulative rate of contralateral breast cancer was reduced from 40.5 to 23.5/1,000, respectively, in the two groups. Taking into account the increased cumulative rate of endometrial cancer, there was a 38% reduction in the 5-year cumulative hazard rate in the tamoxifen-treated group. These results led the authors to conclude that the benefit of tamoxifen therapy for breast cancer outweighs the potential increase in endometrial cancer being reported.
The etiology of tamoxifen-associated endometrial neoplasia has not been established. There are, however, some metabolites of tamoxifen that may act primarily as estrogen agonists, and some investigators favor this mechanism as a possible hypothesis for the development of endometrial neoplasia. Metabolite E is formed by the removal of the aminoethane side chain from tamoxifen. This compound is a weak estrogen agonist that binds the estrogen receptor with low affinity . The presence of a hydroxyl group in this compound destabilizes the ethylene bond, allowing isomerization of the compound to its E isomer, a potent estrogen agonist. The clinical significance of metabolite E is controversial, as it has not been detected in the serum of tamoxifen-treated breast cancer patients, and its role in endometrial neoplasia remains speculative.
It is well established that unopposed estrogen administration is associated with an increased risk of developing endometrial carcinoma. These tend to be predominantly early-stage, low-grade, minimally invasive lesions that have a favorable prognosis . If the effect of tamoxifen on the endometrium is that of a weak estrogen agonist, one could expect associated endometrial cancers to have clinical characteristics comparable to those associated with unopposed estrogen.
The Yale Study—A recent report from the Yale Tumor Registry by Magriples et al  suggested that uterine cancers occurring in breast cancer patients on tamoxifen may behave more aggressively and carry a worse prognosis. The authors identified 53 patients with invasive or in situ breast cancer who subsequently developed uterine cancer. Fifteen of the patients had received adjuvant tamoxifen at a dosage of 40 mg/day for a mean of 4.2 years, while 38 had not received tamoxifen. The mean patient age was 72.3 years for tamoxifen users, which was not statistically different from those not receiving tamoxifen (68.5 years). The interval between the diagnosis of breast and endometrial cancer was significantly lower in the tamoxifen-treated group, compared with those not receiving tamoxifen (5.3 vs 12.3 years). In the tamoxifen-treated group, 67% of the uterine cancers were high-grade lesions (grade 3 adenocarcinoma) or had high-risk histologies (papillary serous, clear cell, mixed mesodermal tumor), compared with 28% of those developing in the 38 breast cancer patients who had not received tamoxifen. In addition, patients in the tamoxifen-treated group were statistically more likely to die of endometrial cancer (33.3% versus 2.6%).
These findings led the authors to conclude that "women receiving tamoxifen as treatment for breast cancer are at risk for high-grade endometrial cancers that have a poor prognosis." In addition, the presence of a high percentage of poor-prognosis histologies, including poorly differentiated adenocarcinoma, papillary serous, and clear cell cancers, along with mixed mesodermal tumors, in the tamoxifen-treated group led the authors to speculate that the mechanism of tamoxifen- induced endometrial neoplasias may be different from that of neoplasias induced by exogenous estrogen, which are associated with more favorable histologies.
Lack of Confirmation—Several recent studies [11-13,25], however, have not been able to confirm the findings of Magriples et al  (Table 2). Barakat et al  reported the Memorial Sloan-Kettering Cancer Center experience in 73 patients with a history of breast cancer who subsequently developed uterine cancer. Twenty-three (32%) had received tamoxifen for at least 1 year, with a median duration of use of 4.5 years, while 50 (68%) did not receive tamoxifen. There was no significant difference in age, mean weight, or median survival following hysterectomy between the two groups of patients. The median interval between the diagnosis of breast and uterine cancer was less in those receiving tamoxifen (4.6 vs 6.7 years), but this was not statistically significant. There was no significant difference in the FIGO stage of the uterine cancers occurring in those patients who had received tamoxifen, compared with nonusers.
In the tamoxifen-treated group, 74% of the corpus cancers were endometrial adenocarcinomas, while 26% were high-risk histologic subtypes, including papillary serous and clear cell carcinomas, as well as uterine sarcomas. This distribution was identical to that seen in the group not receiving tamoxifen. Five women (22%) from the tamoxifen group died of uterine cancer, as did 13 (26%) of those who did not receive tamoxifen. The authors concluded that there was no difference in the stage, grade, or histologic subtype of corpus cancers that develop in breast cancer patients using tamoxifen, compared with nonusers.
Other authors have reported similar results. Fornander et al  recently reported the clinicopathologic findings of endometrial cancers occurring as second primaries in 931 tamoxifen-treated patients with early breast cancer from the Stockholm Adjuvant Tamoxifen Trial .The median duration of tamoxifen use was 24 months, and the drug was given at a dosage of 40 mg/day. On histologic review of these cancers, 82% were FIGO stage I, and all were histologic grade 1 or 2. Three deaths (18%) were attributable to endometrial cancer.
VanLeeuwen et al  recently reported the results of a case-control study from the Netherlands Cancer Registry. There was no difference in the FIGO stage or histologic distribution of endometrial cancers that occurred in 23 breast cancer patients who received tamoxifen, compared with 75 who did not. None of the tamoxifen-treated patients died of endometrial cancer, compared with four of those who did not receive tamoxifen.
Finally, the results of the NSABP B-14 trial  also confirmed that uterine cancers occurring in tamoxifen-treated breast cancer patients are not associated with a higher incidence of adverse histologic features. Eighty-eight percent of the tamoxifen-associated endometrial cancers were FIGO stage I. In addition, 71% were endometrioid adenocarcinomas and 78% were low-grade lesions. Four deaths (16.%) were due to endometrial cancer.
The Role of Screening
The published data would appear to support an association between tamoxifen and the development of both benign and malignant endometrial neoplasia. The increased risk of endometrial cancer associated with tamoxifen use will lead to increased morbidity in breast cancer patients, but this does not appear to outweigh the significant advantage that tamoxifen confers by controlling breast cancer. This issue is clear for the patient with a diagnosis of breast cancer. Of greater concern is the possible implication of using tamoxifen in healthy women considered to be at high risk for developing breast cancer based on family history. Large-scale trials of tamoxifen as a chemopreventive agent are currently underway in England and the United States. Whether the postulated reduction in breast cancer outweighs the risk of endometrial cancer in this population remains to be determined.
Additional studies are also required to evaluate the issue of dose and duration of tamoxifen use. In a meta-analysis of the effect of estrogen replacement therapy on the risk of breast cancer, Steinberg et al  noted an increased risk of breast cancer based on dosage and duration of tamoxifen use. Whether the dose level of tamoxifen or duration of use is important remains to be determined. In the study by Magriples et al  and in the Swedish trial , the tamoxifen dosage was 40 mg/day, twice the conventional dosage used in this country. Fornander et al , noted that the greatest cumulative frequency of uterine cancer was seen among the patients who were allocated to 5 years of tamoxifen treatment. VanLeeuwen et al  noted that women who used tamoxifen for more than 2 years had a 2.3 (range, 0.9 to 5.9) times greater risk of endometrial cancer than nonusers, and the risk rose to 3.0 (range, 0.6 to 15.8) for those treated for more than 5 years. None of the risk estimates for duration of use reached statistical significance, however.
What, if any, recommendations can be made at this point regarding screening for endometrial cancer in breast cancer patients on tamoxifen? To begin with, the expected annual risk of endometrial cancer is approximately 2/1,000, as defined by the B-14 trial . A screening program may detect premalignant endometrial precursors, such as atypical hyperplasia or benign endometrial conditions including polyps, the incidence of which will be higher than 2/1,000. The best method for screening, however, remains to be determined. Some have proposed annual endometrial sampling, although this is not without its difficulties. Gal et al  were unable to perform office endometrial biopsies with a Novak curette in 44% of 89 postmenopausal patients, due to atrophic changes. Should these patients be subjected to the inherent morbidity of a fractional D&C under general anesthesia?
Suction Biopsy—Perhaps larger screening studies such as the one that is currently being performed at Memorial Sloan-Kettering Cancer Center, which uses an endometrial suction biopsy device (Pipelle, Unimar, Wilmington, Conn), may help answer this question. Certainly, all patients with abnormal bleeding should seek immediate gynecologic evaluation. As reported by Gibson et al , all cases of endometrial carcinoma detected by D&C in tamoxifen-treated breast cancer patients occurred in women who presented with abnormal bleeding.
Sonography—Transvaginal sonography may provide a noninvasive means of screening for endometrial pathology in tamoxifen-treated breast cancer patients. The definition of an abnormal endometrial stripe remains to be determined. As reported by Lahti et al , if a cutoff of 5 mm or more were used to define an abnormal endometrial echo, about 50% of patients undergoing endometrial sampling on that basis would be found to have no abnormal pathology. Kedar et al  reported a predictive value of 100% (16/16) for atypical hyperplasia or polyps with an endometrial stripe of 8 mm or more. These findings suggest that premalignant changes can be detected with transvaginal sonography; however, the use of ultrasound and/or endometrial sampling to screen for endometrial neoplasia needs to be evaluated in large prospective trials before screening recommendations can be made.
Care must be taken not to overinterpret the ultrasonographic findings of the endometrium in tamoxifen-treated patients. Goldstein  recently reported five postmenopausal tamoxifen-treated patients who, on routine surveillance with vaginal probe ultrasonography, were described as having heterogeneous, bizarre-appearing endometria with multiple sonolucent areas suggestive of a polyp. Because of concerns regarding tamoxifen use and endometrial neoplasia, the first patient was referred for a curettage and hysteroscopy. Minimal tissue was obtained, and hysteroscopic evaluation revealed a smooth atrophic endometrium. When the abnormal sonographic appearance persisted, the patient underwent a sonohysterogram, which involves the instillation of 3 to 10 mL of saline at the time of sonography. The fluid enhancement revealed that the changes originally interpreted as endometrial were actually subendometrial in origin. Four additional patients with similar abnormal sonographic findings were also found to have subendometrial abnormalities on sonohysterogram. It is unclear what these abnormal areas represent, as none of the patients have undergone hysterectomy, although it was speculated that they may represent adenomyomatous-like changes. Further studies regarding the sonographic appearance of the endometrium in tamoxifen-treated patients are warranted.
For now, all women with breast cancer, whether or not they are receiving tamoxifen, should be encouraged to undergo annual gynecologic evaluation, which should include endometrial sampling in the presence of abnormal vaginal bleeding.