Skip to main content
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2020 Sep 30.
Published in final edited form as: Gynecol Oncol. 2016 Dec 5;144(2):329–335. doi: 10.1016/j.ygyno.2016.11.042

Tumor characteristics and survival outcomes of women with tamoxifen-related uterine carcinosarcoma

Koji Matsuo a,*, Malcolm S Ross b, Stephen H Bush c, Mayu Yunokawa d, Erin A Blake e, Tadao Takano f, Yutaka Ueda g, Tsukasa Baba h, Shinya Satoh i, Masako Shida j, Yuji Ikeda k, Sosuke Adachi l, Takuhei Yokoyama m, Munetaka Takekuma n, Satoshi Takeuchi o, Masato Nishimura p, Keita Iwasaki q, Shiori Yanai r, Merieme M Klobocista s, Marian S Johnson t, Hiroko Machida a, Kosei Hasegawa u, Takahito M Miyake v, Tadayoshi Nagano w, Tanja Pejovic x, Mian MK Shahzad c, Dwight D Im y, Kohei Omatsu z, Frederick R Ueland s, Joseph L Kelley b, Lynda D Roman a
PMCID: PMC7526046  NIHMSID: NIHMS1627567  PMID: 27931750

Abstract

Objective

To examine tumor characteristics and survival outcome of women with uterine carcinosarcoma who had a history of tamoxifen use.

Methods

This is a multicenter retrospective study examining stage I–IV uterine carcinosarcoma cases based on history of tamoxifen use. Patient demographics, tumor characteristics, treatment pattern, and survival outcomes were compared between tamoxifen users and non-users.

Results

Sixty-six cases of tamoxifen-related uterine carcinosarcoma were compared to 1009 cases with no history of tamoxifen use. Tamoxifen users were more likely to be older (mean age, 69 versus 64, P < 0.001) and had a past history of malignancy (100% versus 12.7%, P < 0.001). Tamoxifen-related uterine carcinosarcoma was significantly associated with a higher proportion of stage IA disease (48.4% versus 29.9%) and a lower risk of stage IVB disease (7.8% versus 16.0%) compared to tamoxifen-unrelated carcinosarcoma (P = 0.034). Deep myometrial tumor invasion was less common in uterine carcinosarcoma related to tamoxifen use (28.3% versus 48.8%, P = 0.002). On univariate analysis, tamoxifen use was not associated with progression-free survival (5-year rates 44.5% versus 46.8%, P = 0.48) and disease-specific survival (64.0% versus 59.1%, P =0.39).After adjusting for age, past history of malignancy, stage, residual disease status at surgery, and postoperative treatment patterns, tamoxifen use was not associated with progression-free survival (adjusted-hazard ratio 0.86, 95% confidence interval 0.50 to 1.50, P =0.60)anddisease-specific survival(adjusted-hazard ratio0.68,95% confidence interval 0.36 to 1.29, P = 0.24).

Conclusion

Our study suggests that tamoxifen-related uterine carcinosarcoma may have favorable tumor characteristics but have comparable stage-specific survival outcomes compared to tamoxifen-unrelated uterine carcinosarcoma.

Keywords: Uterine carcinosarcoma, Tamoxifen, Survival outcome

1. Introduction

Uterine carcinosarcoma is a relatively rare malignancy comprising <5% all of uterine tumors [1]. However, it is an aggressive tumor that is responsible for 30% of uterine cancer deaths [2]. Uterine carcinosarcoma is a metaplastic tumor in which the sarcoma component arises from the carcinoma component via transvergent differentiation [3]. Risk factors for uterine carcinosarcoma have been previously described are age, ethnicity, pelvic irradiation, and tamoxifen use [2].

Tamoxifen is a selective estrogen receptor modulator that antagonizes the estrogen receptor in breast tissue; it is often given to women with estrogen receptor positive breast cancer as a maintenance agent following theinitial treatment [4]. However, tamoxifen selectively stimulates the endometrium thus increasing the risk of uterine malignancies [5–9]. The most common uterine malignant diagnosis associated with tamoxifen use is low-grade carcinoma with endometrioid histology [5, 8,9]; and even though carcinosarcoma is not the most common histologic subtype of uterine malignance related to tamoxifen use, there remains a strong association with tamoxifen use and high-grade histologic subtypes [10,11].

Tamoxifen-related uterine carcinosarcoma comprises only 8% of uterine carcinosarcoma, and it has been relatively understudied due to its low incidence [12]. Prior studies have suggested that tamoxifen-related uterine carcinosarcoma may have a poor prognosis; however, these studies were mainly conducted in small series or lacked an adequate control group making the results difficult to interpret [12–15]. To date, there is little data relating to outcomes for patients with tamoxifen-related uterine carcinosarcoma. The objective of the study was to examine the tumor characteristics and survival outcome of women with uterine carcinosarcoma who had a history of tamoxifen use.

2. Patients and methods

This study was a retrospective study utilizing a previously organized large-scale multicenter database for uterine carcinosarcoma obtained via medical record review (8 institutions from the United States and 18 institutions from Japan) [16]. Institutional Review Board approval was obtained at each participating institution. Eligible cases were stage I–IV uterine carcinosarcoma diagnosed and managed between January 1, 1993 and December 31, 2013 with available demographics and treatment information. Cases with no information for medication type at uterine carcinosarcoma diagnosis and incorrect diagnosis were excluded from the study. Eligible cases were then grouped based on history of tamoxifen exposure. The STROBE guidelines for retrospective studies were consulted for the study.

For eligible cases, the following information was abstracted from the medical records: (i) patient demographics, (ii) histopathology results, (iii) treatment type, and (iv) survival outcomes. (i) For patient demographics, country of institution, patient age, ethnicity, body mass index (BMI, kg/m2), pregnancy history, personal history of malignancy, CA-125 level at diagnosis, tamoxifen use (dose, frequency, and duration), and history of pelvic irradiation, were collected. (ii) Histopathologic findings included carcinoma and sarcoma histology, tumor size, lymphovascular space invasion, depth of myometrial tumor invasion, pelvic/para-aortic lymph node metastasis, dominant component, and cancer stage. (iii) Treatment type was abstracted for use of hysterectomy, status of residual tumor at surgery, and use of neoadjuvant or adjuvant radiotherapy and systemic chemotherapy. (iv) For survival outcomes, progression-free survival and disease-specific survival were recorded.

Cancer stage was re-classified based the 2009 International Federation of Gynecology and Obstetrics (FIGO) system [17]. A sarcoma component comprising >50% of the primary tumor in the hysterectomy specimen was defined as sarcoma dominance [16]. Myometrial tumor invasion >50% was defined as deep invasion. Postoperative radiotherapy type was grouped into whole pelvis radiotherapy and intracavitary brachytherapy. Progression-free survival was defined as the time interval between the date of surgery and the date of the first recurrence/progression of uterine carcinosarcoma or the last follow-up if censored. Disease-specific survival was defined as the time interval between the date of surgical staging and the date of death due to uterine carcinosarcoma. Cases were censored if patient was alive at the last follow-up or died of other causes.

The primary analysis of interest was to examine the tumor characteristics of tamoxifen-related uterine carcinosarcoma. The secondary analysis of interest was to examine survival outcome of surgically-treated women with uterine carcinosarcoma. Women with history of tamoxifen use were statistically compared to women who had no history of tamoxifen use. Sample size estimation was performed by using α = 0.05 and power of 80%. Based on a prior study, median disease-specific survival times for tamoxifen-related uterine carcinosarcoma and non-related counterparts were estimated as 19 and 28 months, respectively, and tamoxifen use rate were approximated as 8% for the study [12]. Although this study did not specify either whether overall survival referred to all-cause mortality or disease-specific survival, we used their results as a surrogate of disease-specific survival in our sample size estimation because the majority of uterine carcinosarcoma patients typically die from their disease [16].Basedonfindings from prior studies [12–15], we hypothesized that women with tamoxifen-related uterine carcinosarcoma have a worse disease-specific survival compared to those with no history of tamoxifen use. In this computation, 62 cases of tamoxifen use and 729 cases of non-use were estimated.

Continuous variables were expressed with mean (± standard deviation) or median (range) based on its normality examined by Kolmogorov-Smirnov test, and statistical differences were examined by Student t-test or Mann-Whitney U test as appropriate. Categorical variables were evaluated with Fisher exact test or chi-square test, and statistical significance was expressed with an odds ratio and 95% confidence interval.

For survival analysis, log-rank test for univariate analysis and a Cox proportional hazard regression model for multivariable analysis were utilized, and statistical significance was expressed with hazard ratio and 95% confidence interval. In the entire cohort, association of tamoxifen-related uterine carcinosarcoma and survival outcome was adjusted for other covariates with P < 0.10 in univariate analysis comparing tamoxifen-related and non-related groups.

The variance inflation factor was determined among covariates for the multivariable analysis, and variance inflation factor ≥2.0 was interpreted as multicollinearity. CA-125 level was not used in the multivariable model due to limited available results. Age and BMI were examined as continuous variables, and other categorical variables were grouped in an a priori manner [16]. Survival curves were constructed with Kaplan-Meier method. A P-value of <0.05 was considered statistically significant (all, 2-tailed). Statistical Package for Social Science software (SPSS, version 12.0, Chicago, IL) was used for all the analyses.

3. Results

There were 1192 cases of uterine carcinosarcoma identified during the study period. Of those, 79 cases lacking medication history at diagnosis and 38 cases without uterine carcinosarcoma diagnosis were removed from the study. Among 1075 cases of uterine carcinosarcoma with available medication records at diagnosis, tamoxifen use was reported in 66 (6.1%, 95% confidence interval 4.7 to 7.6) cases.

Patient demographics comparing tamoxifen users (n = 66) and non-users (n = 1009) are shown in Table 1. Tamoxifen users were more likely to be older (mean age, 69 versus 64, P < 0.001, Student t-test) and have past history of any type of malignancy (100% versus 12.8%, P < 0.001, Fisher exact test). The most common dose and duration of tamoxifen was 20 mg daily (76.9%) and five years (56.3%), respectively (Table 2). Periodical endometrial biopsy during tamoxifen use was only reported in 2.5% of tamoxifen users. Median time interval between tamoxifen completion and uterine carcinosarcoma diagnosis was 27 months. Median time interval between breast cancer diagnosis and uterine carcinosarcoma diagnosis was 79 months.

Table 1.

Patient demographics of tamoxifen-related uterine carcinosarcoma (N = 1075).

Characteristic Tamoxifen n = 66 Non-tamoxifen n = 1009 Odds ratio (95%CI) P-value
Age (years) 69 (±9.4) 64 (±11.1) <0.001
<60 11 (16.7%) 335 (33.2%) 1
≥60 55 (83.3%) 674 (66.8%) 2.49 (1.28–4.81)
Race 0.62
Caucasian 29 (43.9%) 363 (36.0%)
African 5 (7.4%) 91 (9.0%)
Hispanic 1 (1.5%) 24 (2.4%)
Asian 30 (45.5%) 494 (49.0%)
Others 1 (1.5%) 37 (3.7%)
Country 0.45
USA 38 (57.6%) 524 (51.9%) 1
Japan 28 (42.4%) 485 (48.1%) 0.80 (0.48–1.32)
BMI (kg/m2) 26.1 (±5.0) 27.2 (±8.6) 0.10
<30 50 (83.3%) 684 (73.3%) 1
≥30 10 (16.7%) 249 (26.7%) 0.55 (0.27–1.10)
Parity 0.52
Nulliparous 54 (84.4%) 794 (80.4%) 1
Multiparous 10 (15.6%) 193 (19.6%) 0.76 (0.38–1.52)
Personal history of cancer <0.001
No 0 879 (87.2%)
Yes 66 (100%) 129 (12.8%)
Types of past cancer <0.001
Breast 63 (95.5%) 59 (46.1%)
Colon 0 18 (14.1%)
Cervical 0 8 (6.3%)
Hematologic 0 8 (6.3%)
Multiple malignancies 3 (4.5%)a 5 (3.9%)b
Gastric 0 5 (3.9%)
Lung 0 3 (2.3%)
Renal 0 3 (2.3%)
Skin 0 3 (2.3%)
Thyroid 0 2 (1.6%)
Liver 0 2 (1.6%)
Anal 0 2 (1.6%)
Brain 0 2 (1.6%)
Ovary 0 2 (1.6%)
Other sites 0 6 (4.7%)
Unknown 0 1 (0.8%)
Prior pelvic radiotherapy 0.33
No 64 (97.0%) 992 (98.3%) 1
Yes 2 (3.0%) 17 (1.7%) 1.82 (0.41–8.07)
Preop CA-125 (IU/L) 23 (6–2121) 25 (1–6000) 0.54
<30 IU/L 24 (57.1%) 359 (54.8%) 1
≥30 IU/L 18 (42.9%) 296 (45.2%) 0.92 (0.50–1.71)

Number (%) per column, mean (±SD), ormedian (range) is shown. Univariate analysis with Student t-test, Mann-Whitney U test, Fisher exact test, or chis-square test for P-values comparing tamoxifen user and non-users. Significant P-values are emboldened. 82 missing for BMI, 24 missing for parity, and 1 missing for personal history of cancer. CA-125 levels were not evaluated in 378 cases. Abbreviations: BMI, body mass index; CA-125, cancer antigen 125; and CI, confidence interval.

a

All cases had history ofbreast cancer.

b

4caseshadhistoryofbreastcancer.

Table 2.

Dose and duration of tamoxifen.

n = 66
Daily dose (mg)
 10 2 (7.7%)
 20 20 (76.9%)
 ≥40 4 (15.4%)
 Unknown 40
Duration (years)
 <2 3 (6.3%)
 2–2.9 6 (12.5%)
 3–3.9 6 (12.5%)
 4–4.9 3 (6.3%)
 5 27 (56.3%)
 >5 3 (6.3%)
 Unknown 18
Periodic endometrial biopsy
 No 39 (97.5%)
 Yes 1 (2.5%)
 Unknown 26
Breast cancer to UCS (months)a 79 (3–372)
 ≤12 6 (12.2%)
 13–36 6 (12.2%)
 37–60 4 (8.2%)
 61–120 21 (42.9%)
 >120 12 (24.5%)
 Unknown 17
Tamoxifen to UCS (months)b 27 (0–132)
 ≤12c 14 (33.3%)
 13–36 10 (23.8%)
 37–60 8 (19.0%)
 61–120 7 (16.7%)
 >120 3 (7.1%)
 Unknown 24

Number (%) or median (range) is shown. Percentages are based on available cases. Abbreviation: UCS, uterine carcinosarcoma.

a

Time interval between breast cancer diagnosis and UCS diagnosis.

b

Time interval between tamoxifen completion and UCS diagnosis.

c

Including 7 cases that UCS was diagnosed during tamoxifen taking.

Among types of past malignancy prior to the uterine carcinosarcoma diagnosis, tamoxifen users were more likely to have had a diagnosis of breast cancer compared to non-users (100% versus 46.1%, P < 0.001, chi-square test). Across the two groups, there were 195 (18.1%, 95% confidence interval 15.8 to 20.4) women who had past history of any malignancy with breast cancer being the most common (n = 129, 64.6%) followed by colon cancer (9.2%). There were eight (0.7%) cases that had multiple malignancies prior to uterine carcinoma diagnosis.

Tumor characteristics were compared between the two groups (Table 3). Across the two groups, the majority of uterine carcinosarcomas were comprised of high-grade carcinoma (71.8%) and homologous sarcoma (58.6%). Histology subtypes, tumor size, lymphovascular space invasion rates, and nodal metastasis rates were similar between tamoxifen-related uterine carcinosarcoma and the non-related counterparts (all, P > 0.05, Fisher exact test). However, uterine carcinosarcoma related to tamoxifen use was associated with a lower risk of deep myometrial tumor invasion (proportion of tumor with >50% invasion, 28.3% versus 48.8%, odds ratio 0.41, 95% confidence interval 0.23 to 0.74, P = 0.002, Fisher exact test). Moreover, women with tamoxifen-related uterine carcinosarcoma were more likely to have stage IA disease (48.4% versus 29.9%) and less likely to have stage IVB disease (7.8% versus 16.0%) compared to non-related counterparts (P = 0.034, chi-square test).

Table 3.

Tumor characteristics of tamoxifen-related uterine carcinosarcoma (N = 1075).

Characteristic Tamoxifen n = 66 Non-tamoxifen n = 1009 Odds ratio (95%CI) P-value
Carcinoma component 0.87
 Low-grade 13 (26.5%) 241 (28.3%) 1
 High-grade 36 (73.5%) 610 (71.7%) 1.09 (0.57–2.01)
Sarcoma component 0.46
 Homologous 26 (53.1%) 501 (58.9%) 1
 Heterologous 23 (46.9%) 350 (41.1%) 1.27 (0.71–2.26)
Sarcoma dominance 0.14
 No 25 (49.0%) 525 (59.7%) 1
 Yes 26 (51.0%) 355 (40.3%) 1.54(0.87–2.71)
Tumor size (cm) 5.2 (0.1–20) 6.0 (0.1–21.5) 0.68
 <5 cm 23 (39.0%) 337 (36.1%) 1
 ≥5 cm 36 (61.0%) 596 (63.9%) 0.89 (0.52–1.52)
Myometrial invasion 0.002
 ≤50% 43 (71.7%) 482 (51.2%) 1
 >50% 17 (28.3%) 460 (48.8%) 0.41 (0.23–0.74)
LVSI 0.35
 No 28 (45.9%) 376 (39.5%) 1
 Yes 33 (54.1%) 576 (60.5%) 0.77 (0.46–1.29)
Pelvic node metastasis 0.99
 No 38 (77.6%) 572 (76.4%) 1
 Yes 11 (22.4%) 177 (23.6%) 0.94 (0.47–1.87)
Sampled pelvic nodes 13.5 (1–52) 16 (1–84) 0.08
Para-aortic node metastasis 0.60
 No 25 (89.3%) 426 (83.5%) 1
 Yes 3 (10.7%) 84 (16.5%) 0.58 (0.17–1.97)
Sampled para-aortic nodes 8 (1–38) 8 (1–72) 0.94
Stage 0.034
 IA 31 (48.4%) 297 (29.9%)
 IB 8 (12.5%) 160 (16.1%)
 II 3 (4.7%) 75 (7.5%)
 IIIA 3 (4.7%) 90 (9.1%)
 IIIB 3 (4.7%) 24 (2.4%)
 IIIC1 9 (14.1%) 98 (9.9%)
 IIIC2 2 (3.1%) 76 (7.6%)
 IVA 0 15 (1.5%)
 IVB 5 (7.8%) 159 (16.0%)

Number (%) per column or median (range) is shown. Univariate analysis with Mann-Whitney U test, Fisher exact test, or chis-square test for P-values comparing tamoxifen user and non-users. Significant P-values are emboldened. Among 1047 cases underwent hysterectomy, 116 missing for sarcoma dominance, 55 missing for tumor size, 45 missing for depth of myometrial tumor invasion, and 34 missing for LVSI. Pelvic and para-aortic lymphadenectomy were performed in 798 and 538 cases, respectively. Archived slide review for carcinoma and sarcoma components was available for 906 cases, and six cases were removed due to missing tamoxifen information. Abbreviations: LVSI, lympho-vascular space invasion; and CI, confidence interval.

Treatment patterns for uterine carcinosarcoma were compared between tamoxifen users and non-users (Table 4). Across the two groups (N = 1075), the majority of women with uterine carcinosarcoma underwent hysterectomy (97.4%). Tamoxifen-related uterine carcinosarcoma cases were less likely to have residual disease at hysterectomy (4.6% versus 12.0%, odds ratio 0.36, 95% confidence interval 0.11 to 1.15, P = 0.07, Fisher exact test) although this did not reach statistical significance. Across the two groups, approximately a quarterof cases received whole pelvic radiotherapy after surgery (24.0%), and approximately two thirds of cases received postoperative systemic chemotherapy (66.8%). Tamoxifen users were more likely to receive whole pelvic radiotherapy (35.4% versus 23.3%, odds ratio 1.81, 95% confidence interval 1.06 to 3.07, P = 0.035, Fisher exact test) but less likely to receive systemic chemotherapy (50.0% versus 67.9%, odds ratio 0.46, 95% confidence interval 0.28 to 0.76, P = 0.004, Fisher exact test) after surgery for uterine carcinosarcoma.

Table 4.

Treatmentpatterns of tamoxifen-related uterine carcinosarcoma (N = 1075).

Characteristic Tamoxifen n = 66 Non-tamoxifen n = 1009 Odds ratio (95%CI) P-value
Hysterectomy 0.71
 None 1 (1.5%) 17 (1.7%)
 Aborted 0 10 (1.0%)
 Yes 65 (98.5%) 982 (97.3%)
Neoadjuvant therapy 0.15
 No 62 (95.4%) 943 (96.0%)
 Chemotherapy 1 (1.5%) 31 (3.2%)
 Radiotherapy 2 (3.1%) 8 (0.8%)
Residual disease at surgery 0.07
 No 62 (95.4%) 861 (86.1%) 1
 Yes 3 (4.6%) 117 (12.0%) 0.36 (0.11–1.15)
Postop WPRT 0.035
 No 42 (64.6%) 745 (76.7%) 1
 Yes 23 (35.4%) 226 (23.3%) 1.81 (1.06–3.07)
Postop chemotherapy 0.004
 No 32 (50.0%) 312 (32.1%) 1
 Yes 32 (50.0%) 661 (67.9%) 0.46 (0.28–0.76)
Chemotherapy cycle 6 (1–6) 6 (1–10) 0.80

Number (%) per column is shown. Univariate analysis with Mann-Whitney U test, Fisher exact test, or chis-square test for P-values comparing tamoxifen user and non-users. Significant P-values are emboldened. Neoadjuvant therapy, residual disease, postoperative WPRT, and postoperative chemotherapy are for 1047 cases that underwent hysterectomy: 4 missing for residual disease, 10 missing for postoperative WPRT, and 10 missing for postoperative chemotherapy. Abbreviation: CI, confidence interval; and WPRT, whole pelvic radiotherapy.

Survival analyses were performed among 1005 women who underwent primary surgical treatment after excluding 18 women due to no surgical treatment and 42 women due to neoadjuvant therapy prior to hysterectomy. Across the two groups, there were 453 (45.3%) recurrences and 324 (32.2%) deaths due to uterine carcinosarcoma in the study population. Median follow-up times were 36.3 months for women who were alive at the last visit and 13.1 months for women who died of uterine carcinosarcoma. Median time to the first recurrence/progression was 8.6 months. There were no deaths due to breast cancer reported in this study population. In the tamoxifen group (n = 62), recurrence and death due to uterine carcinosarcoma were recorded in 27 and 18 women, respectively.

On univariate analysis, tamoxifen use was not associated with progression-free survival (median time 44.4 versus 41.9 months, 5-year rates 44.5% versus 46.8%, unadjusted-hazard ratio 0.87, 95% confidence interval 0.59 to 1.28, P = 0.48, log-rank test; Fig. 1A) and disease-specific survival (median time not reached, 5-year rates 64.0% versus 59.1%, unadjusted-hazard ratio 0.81, 95% confidence interval 0.50 to 1.31, P = 0.39, log-rank test; Fig. 1B). After controlling for other covariates exhibiting a difference between tamoxifen-related and unrelated groups (Table 5),tamoxifen-related uterinecarcinosarco-ma was not associated with progression-free survival (adjusted-hazard ratio 0.86, 95% confidence interval 0.50 to 1.50, P = 0.60) and disease-specific survival (adjusted-hazard ratio 0.68, 95% confidence interval 0.36 to 1.29, P = 0.24).

Fig. 1.

Fig. 1.

Survival outcomes of tamoxifen-related uterine carcinosarcoma. Adjusted-hazard ratio (HR) and P-value are shown for Cox proportional hazard regression models in multivariable analysis. A) progression-free survival (8 missing) and B) disease-specific survival (1 missing) for 1005 women who underwent primary surgical treatment are shown based on tamoxifen use.

Table 5.

Multivariable analysis for survival outcome of women who underwent primary surgical treatment (N = 1005).

Characteristic No. Progression-free survival
Disease-specific survival
5-y (%) HR (95%CI) P-value 5-y (%) HR (95%CI) P-value
Tamoxifen use
 No 943 46.8% 1 59.1% 1
 Yes 62 44.6% 0.86 (0.50–1.50) 0.60 64.0% 0.68 (0.36–1.29) 0.24
Age (continuous) 1005 1.03 (1.02–1.04) <0.001 1.03 (1.02–1.04) <0.001
Past cancer history
 No 816 46.0% 1 59.1% 1
 Breast cancer 125 48.2% 0.89 (0.60–1.33) 0.57 64.2% 1.06 (0.69–1.64) 0.78
 Others 63 52.2% 1.03 (0.67–1.58) 0.89 55.0% 1.49 (0.95–2.33) 0.08
Stage
 I 489 64.8% 1 74.5% 1
 II 74 47.6% 2.12 (1.43–3.15) <0.001 65.4% 2.11 (1.28–3.46) 0.003
 III 294 29.3% 2.92 (2.28–3.75) <0.001 47.2% 3.31 (2.44–4.49) <0.001
 IV 135 11.1% 4.27 (3.08–5.91) <0.001 24.5% 4.85 (3.29–7.15) <0.001
Deep myometrial invasion
 No 514 57.4% 1 70.5% 1
 Yes 451 33.6% 1.49 (1.22–1.83) <0.001 46.4% 1.56 (1.22–2.00) <0.001
Residual disease
 No 889 50.0% 1 64.5% 1
 Yes 109 13.6% 1.96 (1.47–2.62) <0.001 19.1% 2.56 (1.87–3.50) <0.001
Postop WPRT
 No 745 46.2% 1 57.4% 1
 Yes 247 49.3% 0.77 (0.61–0.97) 0.03 66.5% 0.69 (0.52–0.91) 0.009
Postop chemotherapy
 No 332 42.3% 1 51.6% 1
 Yes 660 48.8% 0.45 (0.35–0.56) <0.001 63.4% 0.36 (0.27–0.47) <0.001

Cox proportional hazard regression models for multivariable analysis. Associations of tamoxifen-related uterine carcinosarcoma and progression-free survival and disease-specific survival among 1005 women who underwent primary surgical treatment were adjusted for covariates demonstrating P < 0.10 in univariate analysis comparing tamoxifen-related and unrelated groups in Tables 1, 3, and 4. Variance of inflation factor < 2.0 between stage and deep myometrial tumor invasion indicates absence of multicollinearity. Significant P-values are emboldened. Abbreviations: 5-y (%), 5-year survival proportion; HR, hazard ratio; CI, confidence interval; and WPRT, whole pelvic radiotherapy.

Survival analysis was performed by stratifying cancer stage. Among 489 women with stage I uterine carcinosarcoma who underwent primary surgical treatment, progression-free survival (5-year rates, 63.9% versus 64.9%, P = 0.99) and disease-specific survival (77.6% versus 74.2%, P = 0.83) were similar between tamoxifen users and non-users. Similarly, among 503 women with stage II-V disease, tamoxifen use was not associated with progression-free survival (2-year rates, 43.0% versus 46.9%, P = 0.89)and disease-specific survival (62.2% versus 59.7%, P = 0.84).

Among 125 women with a personal history of breast cancer who underwent primary surgical treatment, tamoxifen use was not associated with progression-free survival (5-year rates, 44.5% versus 51.4%, P = 0.81)and disease-specific survival (64.0% versus 64.2%, P =0.34).Lastly, Duration of tamoxifen use (≥5 versus <5 years) was not associated with histology type, stage, tumor size, depth of invasion, progression-free interval, and disease-specific survival among tamoxifen users (all, P > 0.05). The time interval between tamoxifen completion and uterine carcinoma diagnosis was not associated with patient demographics, tumor characteristics, and survival outcomes (all, P > 0.05).

4. Discussion

Key findings of this study include more early-stage disease and less distant metastatic disease in tamoxifen-related uterine carcinosarcomas when compared to tamoxifen-unrelated uterine carcinosarcomas. An additional salient finding is that a history of tamoxifen use was not associated with survival outcomes of women with uterine carcinosarcoma.

Prior studies have concluded that tamoxifen-related uterinecarcino-sarcoma has a poor prognosis [12–15]. However, these studies were conducted without adequate control groups which might have impacted the conclusions. Two relatively large studies examined survival outcomes based on history of tamoxifen use but these studies were conducted in a heterogeneous population mixing all histology subtypes other than carcinosarcoma [18,19]. There was one study conducted with a proper study design which compared tamoxifen users and non-users in uterine carcinosarcoma; however, this study included only a small number of women with tamoxifen-related uterine carcinosarcoma, therefore making the results difficult to generalize [12]. By providing a proper and large-size control group, this study found that tamoxifen-related uterine carcinosarcoma has comparable survival outcomes compared to tamoxifen-unrelated uterine carcinosarcoma.

Our study showed that the majority of tamoxifen-related uterine carcinosarcoma was of early stage. This finding is similar to other studies demonstrating that stage I disease was the most common stage in tamoxifen-related endometrial cancer [5,8,9]. However, the most common stage of uterine carcinosarcoma in general is stage I [16]. Similarly, the majority of endometrial cancer is usually stage I disease [20].Conse-quently, it has not been known whether tamoxifen-related uterine carcinosarcoma is more likely to be stage I as compared to tamoxifen-unrelated uterine carcinosarcoma. Therefore, our study is noteworthy in that tamoxifen-related uterine carcinosarcoma has favorable tumor factors with early-stage disease and less myometrial tumor invasion compared to tamoxifen-unrelated counterparts.

Although conducted via a different study design, our results partly support a prior study that examined the survival outcome of women with endometrial cancer who had a history of tamoxifen use for breast cancer [21]. Their study found no difference in endometrial cancer-specific survival outcome between tamoxifen users and non-users. Our results were similar in that progression-free and disease-specific survival for uterine carcinosarcoma were similar between tamoxifen users and non-users. However, their study had only 13% of uterine carcinosarcoma in the study population [21]. Our study design was more specific and allowed a direct comparison of tamoxifen-related uterine carcinosarcoma to tamoxifen-unrelated counterparts, providing relevant information for this topic.

Our results showed that women with tamoxifen-related uterine carcinosarcoma were more likely to receive adjuvant pelvic irradiation and less likely to receive adjuvant chemotherapy compared to non-related counterparts. This difference may reflect the difference in pattern of adjuvant therapy choicefor early-stage uterine carcinosarcoma shown in a recent nation-wide study [22]. That is, radiotherapy is more frequently used for women with early-stage uterine carcinosarcoma and chemotherapy is more commonly prescribed to women with advanced-stage disease. Because tamoxifen-related uterine carcinosarcoma is predominantly early-stage disease, it is speculated that these patients received more radiotherapy than chemotherapy in our study.

Various causalities can be considered for favorable tumor factors in uterine carcinosarcoma related to tamoxifen use. First, tamoxifen indirectly activates the PIK3CA/AKT/mTOR pathway via estrogen receptor to induce insulin growth factor 1 signaling that is recognized as the oncogenic risk factor for endometrial cancer [19,23,24]. Because gynecologic malignancy with the PIK3CA/AKT/mTOR pathway alteration is known to be associated with better prognosis [25,26], this may be a possible biological plausibility to support a high proportion of early-stage disease in tamoxifen-related uterine carcinosarcoma. Nevertheless, this study found that survival outcomes of women with uterine carcinosarcoma for both early-stage and advanced-stage disease were similar between tamoxifen users and non-users. This finding is likely due to the nature of poor prognosis in uterine carcinosarcoma even in early-stage disease [2,16].

However, this non-significance in theresults of our studyfor survival outcome of women with tamoxifen-related uterine carcinosarcoma compared to the unrelated counterparts may be possibly due to the lack in study power. While our study cases exceeded the estimated number, our findings were not consistent with prior studies that were used for sample size estimation. Indeed, if an assumption was made that there is a true difference in progression-free survival between the two groups, the power of our study was <30% for progression-free survival by computing the hazard ratio of 0.87 for tamoxifen users compared non-users (α = 0.05). In the same line, based on our results and statistics, estimated sample size to detect 80% power of the statistical difference of this hazard ratio is 554 cases for tamoxifen users and 8477 cases for non-users. Contrary, if the power analysis of this study was performed based upon an assumption that there is no true difference in progression-free survival between the two groups (non-inferiority), our sample size is adequate with a power >80% for progression-free survival (utilized a-level of 0.025 and non-inferiority margin of 2.3% based on the 5-years rate difference between the two groups). Collectively, further studies with larger sample will be warranted to validate the results of our study.

A strength of the study is that this is one of the largest studies examining tamoxifen-related uterine carcinosarcoma, and our study fills an information gap in the literature on this relatively under-studied area. We have also noticed a number of limitations in this study. Besides a possibility for under-powering as mentioned above, there is a weakness related to the nature of retrospective observational study that may miss potential confounding factors. For instance, the details regarding tamoxifen dose and duration could not be abstracted from records in manycases. Lack of central pathology review is another major limitation especially in multicenter international study like ours where there is a possibility of difference in the practice of pathology for rare tumor types. However, in each participating center, all slides were pulled and reviewed by a gynecologic pathologist with expertise in gynecologic malignancy. Additionally, we did not address the question of outcome in relation to hereditary cancer syndrome status due to the heterogeneity of testing and change in screening practices over time; however, the effect of BRCA status on outcome of patients with tamoxifen related carcinosarcoma is an interesting topic for future studies.

In summary, this study provides insight into the possible biology of tamoxifen-related uterine carcinosarcoma. For instance, our study showed that tamoxifen-related uterine carcinosarcoma was less likely to invade deeply into the myometrium and to have distant metastasis. This suggests further investigation into biomarkers for stromal invasion and lymphatic/hematogenous metastasis to better elucidate the biologic effects of tamoxifen on carcinogenic process of the endometrium.

HIGHLIGHTS.

  • ~6% of women with uterine carcinosarcoma (UCS) have history of tamoxifen use.

  • Tamoxifen-related UCS is more likely to be early-stage disease.

  • Tamoxifen-related UCS and unrelated UCS have similar survival outcomes.

Acknowledgements

Authors thank to the following researchers for administrative support and provision of study patients: Iwona Podzielinski, Marina M. Frimer, Tadaaki Nishikawa, Kazuko Fujiwara, Yukiko Hazama, Daisuke Kadogami, Melissa N. Moffitt, Stephen W. Li, Kiyoshi Yoshino, Ken Yamaguchi, Tetsuro Oishi, Kei Kawana, Kazuaki Suda, Akihiko Wakatsuki, Toru Sugiyama, Koichiro Shimoya, Masaaki Andoh, Takayuki Enomoto, Keiichi Fujiwara, Mikio Mikami, Muneaki Shimada, Yasuhiko Shiki Ikuo Konishi, and Tadashi Kimura.

Financial support: Ensign Endowment for Gynecologic Cancer Research (K.M.)

Footnotes

Disclosure statement

The authors did not report any potential conflicts of interest.

References

  • [1].Boll D, Verhoeven RH, van der Aa MA, Pauwels P, Karim-Kos HE, Coebergh JW, van Doorn HC, Incidence and survival trends of uncommon corpus uteri malignancies in the Netherlands, 1989–2008, Int. J. Gynecol. Cancer 22 (2012) 599–606. [DOI] [PubMed] [Google Scholar]
  • [2].Cantrell LA, Blank SV, Duska LR, Uterine carcinosarcoma: a review of the literature, Gynecol. Oncol. 137 (2015) 581–588. [DOI] [PubMed] [Google Scholar]
  • [3].Shah SH, Jagannathan JP, Krajewski K, O'Regan KN, George S, Ramaiya NH, Uterine sarcomas: then and now, AJR Am. J. Roentgenol. 199 (2012) 213–223. [DOI] [PubMed] [Google Scholar]
  • [4].Riggs BL, Hartmann LC, Selective estrogen-receptor modulators — mechanisms of action and application to clinical practice, N. Engl. J. Med. 348 (2003) 618–629. [DOI] [PubMed] [Google Scholar]
  • [5].Fisher B, Costantino JP, Wickerham DL, Cecchini RS, Cronin WM, Robidoux A, Bevers TB, Kavanah MT, Atkins JN, Margolese RG, Runowicz CD, James JM, Ford LG, Wolmark N, Tamoxifen for the prevention of breast cancer: current status of the National Surgical Adjuvant Breast and Bowel Project P-1 study, J. Natl. Cancer Inst. 97 (2005) 1652–1662. [DOI] [PubMed] [Google Scholar]
  • [6].Davies C, Pan H, Godwin J, Gray R, Arriagada R, Raina V, Abraham M, Medeiros Alencar VH, Badran A, Bonfill X, Bradbury J, Clarke M, Collins R, Davis SR, Delmestri A, Forbes JF, Haddad P, Hou MF, Inbar M, Khaled H, Kielanowska J, Kwan WH, Mathew BS, Mittra I, Muller B, Nicolucci A, Peralta O, Pernas F, Petruzelka L, Pienkowski T, Radhika R, Rajan B, Rubach MT, Tort S, Urrutia G, Valentini M, Wang Y, Peto R, Long-term effects of continuing adjuvant tamoxifen to 10 years versus stopping at 5 years after diagnosis of oestrogen receptor-positive breast cancer: ATLAS, a randomised trial, Lancet 381 (2012) 805–816. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [7].Wickerham DL, Fisher B, Wolmark N, Bryant J, Costantino J, Bernstein L, Runowicz CD, Association of tamoxifen and uterine sarcoma, J. Clin. Oncol. 20 (2002) 2758–2760. [DOI] [PubMed] [Google Scholar]
  • [8].Fisher B, Costantino JP, Wickerham DL, Redmond CK, Kavanah M, Cronin WM, Vogel V, Robidoux A, Dimitrov N, Atkins J, Daly M, Wieand S, Tan-Chiu E, Ford L, Wolmark N, Tamoxifen for prevention of breast cancer: report of the National Surgical Adjuvant Breast and Bowel Project P-1 Study, J. Natl. Cancer Inst. 90 (1998) 1371–1388. [DOI] [PubMed] [Google Scholar]
  • [9].Fisher B, Costantino JP, Redmond CK, Fisher ER, Wickerham DL, Cronin WM, Endometrial cancer in tamoxifen-treated breast cancer patients: findings from the National Surgical Adjuvant Breast and Bowel Project (NSABP) B-14, J. Natl. Cancer Inst. 86 (1994) 527–537. [DOI] [PubMed] [Google Scholar]
  • [10].Bergman L, Beelen ML, Gallee MP, Hollema H, Benraadt J, van Leeuwen FE, Risk and prognosis of endometrial cancer after tamoxifen for breast cancer. Comprehensive Cancer Centres' ALERT Group. Assessment of liver and endometrial cancer risk following tamoxifen, Lancet 356 (2000) 881–887. [DOI] [PubMed] [Google Scholar]
  • [11].Swerdlow AJ, Jones ME, Tamoxifen treatment for breast cancer and risk of endometrial cancer: a case-control study, J. Natl. Cancer Inst. 97 (2005) 375–384. [DOI] [PubMed] [Google Scholar]
  • [12].Uehara T, Onda T, Togami S, Amano T, Tanikawa M, Sawada M, Ikeda S, Kato T, Kasamatsu T, Prognostic impact of the history of breast cancer and of hormone therapy in uterine carcinosarcoma, Int. J. Gynecol. Cancer 22 (2012) 280–285. [DOI] [PubMed] [Google Scholar]
  • [13].McCluggage WG, Abdulkader M, Price JH, Kelehan P, Hamilton S, Beattie J, Al-Nafussi A, Uterine carcinosarcomas in patients receiving tamoxifen. A report of 19 cases, Int. J. Gynecol. Cancer 10 (2000) 280–284. [DOI] [PubMed] [Google Scholar]
  • [14].Kloos I, Delaloge S, Pautier P, Di Palma M, Goupil A, Duvillard P, Cailleux PE, Lhomme C, Tamoxifen-related uterine carcinosarcomas occur under/after prolonged treatment: report of five cases and review of the literature, Int. J. Gynecol. Cancer 12 (2002) 496–500. [DOI] [PubMed] [Google Scholar]
  • [15].Evans MJ, Langlois NE, Kitchener HC, Miller ID, Is there an association between long-term tamoxifen treatment and the development of carcinosarcoma (malignant mixed Mullerian tumor) of the uterus? Int. J. Gynecol. Cancer 5 (1995) 310–313. [DOI] [PubMed] [Google Scholar]
  • [16].Matsuo K, Takazawa Y, Ross MS, Elishaev E, Podzielinski I, Yunokawa M, Sheridan TB, Bush SH, Klobocista MM, Blake EA, Takano T, Matsuzaki S, Baba T, Satoh S, Shida M, Nishikawa T, Ikeda Y, Adachi S, Yokoyama T, Takekuma M, Fujiwara K, Hazama Y, Kadogami D, Moffitt MN, Takeuchi S, Nishimura M, Iwasaki K, Ushioda N, Johnson MS, Yoshida M, Hakam A, Li SW, Richmond AM, Machida H, Mhawech-Fauceglia P, Ueda Y, Yoshino K, Yamaguchi K, Oishi T, Kajiwara H, Hasegawa K, Yasuda M, Kawana K, Suda K, Miyake TM, Moriya T, Yuba Y, Morgan T, Fukagawa T, Wakatsuki A, Sugiyama T, Pejovic T, Nagano T, Shimoya K, Andoh M, Shiki Y, Enomoto T, Sasaki T, Mikami M, Shimada M, Konishi I, Kimura T, Post MD, Shahzad MM, Im DD, Yoshida H, Omatsu K, Ueland FR, Kelley JL, Karabakhtsian RG, Roman LD, Significance of histologic pattern of carcinoma and sarcoma components on survival outcomes of uterine carcinosarcoma, Ann. Oncol. 27 (2016) 1257–1266. [DOI] [PubMed] [Google Scholar]
  • [17].Pecorelli S, Revised FIGO staging for carcinoma of the vulva, cervix, and endometrium, Int. J. Gynaecol. Obstet. 105 (2009) 103–104. [DOI] [PubMed] [Google Scholar]
  • [18].Ngo C, Brugier C, Plancher C, de la Rochefordiere A, Alran S, Feron JG, Malhaire C, Scholl S, Sastre X, Rouzier R, Fourchotte V, Clinico-pathology and prognosis of endometrial cancer in patients previously treated for breast cancer, with or without tamoxifen: a comparative study in 363 patients, Eur. J. Surg. Oncol. 40 (2014) 1237–1244. [DOI] [PubMed] [Google Scholar]
  • [19].Tergas AI, Buell-Gutbrod R, Gwin K, Kocherginsky M, Temkin SM, Fefferman A, Lengyel E, Yamada SD, Clinico-pathologic comparison of type II endometrial cancers based on tamoxifen exposure, Gynecol. Oncol. 127 (2012) 316–320. [DOI] [PubMed] [Google Scholar]
  • [20].Wright JD, Barrena Medel NI, Sehouli J, Fujiwara K, Herzog TJ, Contemporary management of endometrial cancer, Lancet 379 (2012) 1352–1360. [DOI] [PubMed] [Google Scholar]
  • [21].Slomovitz BM, Sun CC, Ramirez PT, Bodurka DC, Diaz P, Lu KH, Does tamoxifen use affect prognosis in breast cancer patients who develop endometrial cancer? Obstet. Gynecol. 104 (2004) 255–260. [DOI] [PubMed] [Google Scholar]
  • [22].Rauh-Hain JA, Starbuck KD, Meyer LA, Clemmer J, Schorge JO, Lu KH, Del Carmen MG, Patterns of care, predictors and outcomes of chemotherapy for uterine carcinosarcoma: a National Cancer Database analysis, Gynecol. Oncol. 139 (2015) 84–89. [DOI] [PubMed] [Google Scholar]
  • [23].Klotz DM, Hewitt SC, Korach KS, Diaugustine RP, Activation of a uterine insulinlike growth factor I signaling pathway by clinical and environmental estrogens: requirement of estrogen receptor-alpha, Endocrinology 141 (2000) 3430–3439. [DOI] [PubMed] [Google Scholar]
  • [24].Druckmann R, Rohr UD, IGF-1 in gynaecology and obstetrics: update 2002, Maturitas 41 (2002) S65–S83. [DOI] [PubMed] [Google Scholar]
  • [25].Abe A, Minaguchi T, Ochi H, Onuki M, Okada S, Matsumoto K, Satoh T, Oki A, Yoshikawa H, PIK3CA overexpression is a possible prognostic factor for favorable survival in ovarian clear cell carcinoma, Hum. Pathol. 44 (2012) 199–207. [DOI] [PubMed] [Google Scholar]
  • [26].Akiyama-Abe A, Minaguchi T, Nakamura Y, Michikami H, Shikama A, Nakao S, Sakurai M, Ochi H, Onuki M, Matsumoto K, Satoh T, Oki A, Yoshikawa H, Loss of PTEN expression is an independent predictor of favourable survival in endometrial carcinomas, Br. J. Cancer 109 (2013) 1703–1710. [DOI] [PMC free article] [PubMed] [Google Scholar]

RESOURCES