To develop a nomogram for predicting 3-year overall survival (OS) and outcomes of surgically staged patients with uterine carcinosarcomas (UCS).
This retrospective study analyzed the clinicopathological characteristics, treatment data, and oncological outcomes of 69 patients diagnosed with UCS between January 2002 and September 2018. Significant prognostic factors for OS were identified and integrated to develop a nomogram. Concordance probability (CP) was used as a precision measure. The model was internally validated using bootstrapping samples to correct overfitting.
The median follow-up time was 19.4 months (range, 0.77–106.13 months). The 3-year OS was 41.8% (95% confidence interval [CI], 29.9–58.3%). The International Federation of Gynecology and Obstetrics (FIGO) stage and adjuvant chemotherapy were independent factors for OS. The CP of the nomogram integrating with body mass index (BMI), FIGO stage, and adjuvant chemotherapy was 0.72 (95% CI, 0.70–0.75). In addition, the calibration curves for the probability of 3-year OS demonstrated good agreement between the nomogram-predicted and observed data.
The established nomogram using BMI, FIGO stage, and adjuvant chemotherapy accurately predicted the 3-year OS of patients with UCS. The nomogram was useful for patient counselling and deciding on follow-up strategies.
Uterine carcinosarcomas (UCS), previously known as malignant mixed müllerian tumors (MMMT), are rare, aggressive neoplasms, representing 2–5% of all uterine tumors. UCS are biphasic tumors composed of both malignant epithelial and mesenchymal tissue elements [
The staging system and histological classifications were recently based on the criteria established by the revised International Federation of Gynecology and Obstetrics (FIGO) 2009 staging system and the 2014 World Health Organization (WHO) classification of tumors [
Currently, FIGO staging is used to classify cancer prognosis, which is inaccurate. A nomogram is a graphical illustration of a mathematical model that includes predictors of an endpoint. Nomograms have been widely used for cancer prognosis, including uterine cancers, to anticipate reliable prognostic information for individual patients [
After obtaining approval from the Ethics Committee for Research Involving Human Subjects of the Prince of Songkla University Faculty of Medicine (REC 61-320-12-3), a data search was performed for all patients diagnosed with UCS between January 2002 and September 2018. We included patients at all stages, from those who underwent definitive surgical staging to hysterectomy, bilateral salpingo-oophorectomy, peritoneal washings, pelvic and para-aortic lymphadenectomy in which the removal of 10 or more lymph nodes was considered optimal [
Data extracted from electronic databases for analysis included age at diagnosis, body mass index (BMI), menopausal status, medical disease (diabetes mellitus, hypertension, and/or dyslipidemia), clinical symptoms, FIGO stage, tumor size, MI, LVSI, lower uterine segment involvement, cervical involvement, uterine serosal or ovarian involvement, omental involvement, positive peritoneal cytology, residual tumor, lymph node metastasis, lymph node count, adjuvant radiotherapy, adjuvant chemotherapy, and disease recurrence or progression. Clinical stage and histological classification were based on the criteria established by the revised FIGO 2009 and 2014 WHO classification of tumors.
Survival outcomes were OS and progression-free survival (PFS). OS was defined as the time from surgery to death, regardless of the cause or date of the last follow-up, for patients who were alive and censored. PFS was defined as the time from surgery to disease progression or recurrent disease.
Patient demographics and clinicopathologic characteristics are presented as descriptive data. OS and PFS probabilities were estimated using the Kaplan-Meier method. Univariate and multivariate analyses were performed using the Cox proportion hazards model. Hazard ratios (HR) were estimated with a 95% confidence interval (CI), and probability values of <0.05 were considered statistically significant. To permit nonlinear relationships, continuous variables were modelled using restricted cubic splines. Collinearity was also assessed.
A nomogram was used to predict the 3-year OS probability of patients with UCS. The 3-year predicted survival probability was calculated for each patient using a Cox regression model. The nomogram was constructed based on the results of multivariate Cox regression analysis. Discrimination between patients who had an event and those who did not was estimated by the area under the curve of a receiver operating curve, which was the concordance probability (CP) in the nomogram that reflects the accuracy of the prognostic model [
Calibration was determined by dividing the patients based on their nomogram-predicted probabilities and assessed via a calibration plot by plotting the nomogram’s predicted 3-year OS probability against the patient’s observed or actual probability, as calculated using the Kaplan-Meier method. Validation with 200 bootstrap resamples was used to analyze the model activity. Statistical analysis was conducted using the rms package of R software (version 3.5.3; R Foundation for Statistical Computing, Vienna, Austria).
In total, 81 patients with UCS were identified; of these, only 69 patients who underwent definite surgical staging were enrolled. The mean age was 63.4 years (range, 40–90), and the median BMI was 25.5 kg/m2 (range, 21.3–30.2). Patient demographics and clinical characteristics are presented in
Univariate analysis was performed to identify prognostic variables for oncological outcomes, as shown in
The variables were screened based on clinical evidence obtained from univariate and multivariate analyses before initiating the model, as previously suggested in the literature. No more than three prognostic variables should be incorporated into the model, as recommended by Harrell’s guideline [
For each patient, points were assigned to each of the three variables, and the total score was calculated using the nomogram. The total number of points corresponded to the predicted 3-year OS probabilities. It is important to note that all three variables must be available to use the nomogram. As an example of the clinical utility of the nomogram, a patient with a BMI of 30 kg/m2 (21.25 points) who had received adjuvant chemotherapy (77.5 points) and with FIGO stage III (51.25 points) would have an estimated 3-year OS of approximately 42.5%.
Predictive accuracy of OS after cancer treatment is beneficial for patient counselling and decision-making during follow-up. Previous reports of survival in UCS cases were mostly based on disease stage [
The univariate analysis indicated that higher BMI in patients with UCS was associated with better OS than lower BMI. However, in the multivariate analysis, higher BMI was not a significant prognostic factor for OS. Obesity is associated with the excessive production of steroid hormones, which increases the incidence of endometrial cancer and poor prognosis [
Postoperative adjuvant therapy should be considered for all stages of UCS. However, in patients with early stage I–II UCS, the recommendation for adjuvant chemotherapy is based only on limited data obtained from retrospective studies [
The present study also showed worse prognosis among patients with advanced-stage disease. The extent of the tumor beyond the uterus at the time of diagnosis is the most important unfavorable prognostic factor. Multiple studies have reported that disease stage is the most important predictor of survival in UCS [
There is a limited number of nomogram studies to predict the prognosis of patients with UCS. Abu-Rustum et al. [
This study has some limitations. First is its retrospective nature. Data were collected over several years from a single institution. Due to the rarity of the disease, the number of patients included in the study was relatively small compared to that in the study conducted by Abu-Rustum et al. [
In conclusion, three variables, namely BMI, FIGO stage, and adjuvant chemotherapy, were identified for the construction of a nomogram to predict the 3-year OS of patients with UCS. This model could assist clinicians in making more individualized and specific predictions of patient prognosis. In addition, the nomogram could be beneficial for patient counselling and deciding on follow-up strategies.
No conflict of interests relevant to this article were reported.
This study was approved by the Ethics Committee for Research Involving Human Subjects of the Prince of Songkla University Faculty of Medicine (REC 61-320-12-3).
The requirement for informed consent was waived due to the retrospective study and full anonymity of the participants was maintained.
This research was supported by the Faculty of Medicine, Prince of Songkla University and Ms. Jirawan Jayuphan, Internal Affairs Unit, Faculty of Medicine, Prince of Songkla University.
Nomogram for predicting 3-year overall survival. BMI, body mass index; CMT, adjuvant chemotherapy.
Calibration curves of the nomogram. Solid line represents the actual nomogram; gray line represents the ideal agreement between the actual and predicted probabilities of 3-year overall survival; vertical bars represent 95% confidence interval; dots correspond to apparent predictive accuracy; and crosses mark the bootstrap-corrected estimates.
Demographics and clinicopathologic characteristics of patients with uterine carcinosarcomas
Characteristic | Value (n=69) |
---|---|
Age (yr) | |
<70 | 51 (73.91) |
≥70 | 18 (26.09) |
BMI (kg/m2) | |
≥25 | 37 (53.62) |
<25 | 32 (46.38) |
Menopause | |
Yes | 61 (88.41) |
No | 8 (11.59) |
Medical disease (DM, HT, and/or DLP) | |
Yes | 37 (53.62) |
No | 32 (46.38) |
Symptom | |
AUB | 64 (92.75) |
Pelvic mass | 2 (2.89) |
Pelvic pain | 3 (4.35) |
Other symptom | 3 (4.35) |
Values are presented as number (%).
BMI, body mass index; DM, diabetes mellitus; HT, hypertension; DLP, dyslipidemia; AUB, abnormal uterine bleeding.
Clinicopathologic characteristics and treatment of patients with uterine carcinosarcomas
Value (n=69) | |
---|---|
FIGO stage | |
I | 26 (37.68) |
II | 11 (15.94) |
III | 22 (31.88) |
IV | 10 (14.5) |
Tumor size (cm) | |
<10 | 52 (75.36) |
≥10 | 17 (24.64) |
MI | |
<50 | 29 (42.03) |
≥50 | 40 (57.97) |
LVSI | |
Yes | 43 (62.32) |
No | 26 (37.68) |
Lower uterine segment involvement | |
Yes | 14 (20.29) |
No | 55 (79.71) |
Cervical involvement | |
Yes | 18 (26.08) |
No | 51 (73.92) |
Uterine serosal or ovarian involvement | |
Yes | 14 (20.29) |
No | 55 (79.71) |
Omental involvement | |
Yes | 5 (7.25) |
No | 38 (55.07) |
Unknown | 26 (37.68) |
Peritoneal cytology | |
Positive | 5 (7.24) |
Negative | 46 (66.67) |
Unknown | 18 (26.09) |
Residual tumor | |
Yes | 20 (28.99) |
No | 49 (71.01) |
Lymph node count | |
≥10 | 43 (62.32) |
<10 or no node dissection | 26 (37.38) |
LN metastasis | |
Yes | 19 (27.54) |
No | 36 (52.17) |
Unknown | 14 (20.29) |
Adjuvant radiotherapy | |
Yes | 19 (27.54) |
No | 50 (72.46) |
Adjuvant chemotherapy | |
Yes | 63 (91.3) |
No | 6 (8.7) |
Disease recurrence or progression | |
Yes | 35 (50.72) |
No | 34 (49.28) |
Values are presented as number (%).
FIGO, International Federation of Gynecology and Obstetrics; MI, myometrial invasion; LVSI, lymphovascular space involvement; LN, lymph node.
Univariate analysis for the correlation of clinicopathologic and treatment parameters with PFS and OS
Factor | PFS | OS | ||
---|---|---|---|---|
|
| |||
HR (95% CI) | HR (95% CI) | |||
Age (yr) | ||||
| ||||
<70 | 1 | 1 | ||
| ||||
≥70 | 1.31 (0.62–2.72) | 0.477 | 1.47 (0.70–3.09 | 0.308 |
| ||||
BMI (kg/m2) | ||||
| ||||
<25 | 1 | 1 | ||
| ||||
≥25 | 0.57 (0.29–1.11) | 0.101 | 0.51 (0.26–0.99) | 0.049 |
| ||||
Medical disease | ||||
| ||||
No | 1 | 1 | ||
| ||||
Yes | 0.68 (0.35–1.32) | 0.255 | 0.72 (0.37–1.42) | 0.345 |
| ||||
FIGO stage | ||||
| ||||
I/II | 1 | 1 | ||
| ||||
III/IV | 2.70 (1.36–5.35) | 0.004 |
3.06 (1.50–6.22) | 0.002 |
| ||||
Tumor size (cm) | ||||
| ||||
<10 | 1 | 1 | ||
| ||||
≥10 | 0.54 (0.22–1.29) | 0.165 | 0.56 (0.23–1.35) | 0.197 |
| ||||
MI | ||||
| ||||
<50 | 1 | 1 | ||
| ||||
≥50 | 0.96 (0.49–1.88) | 0.898 | 0.97 (0.49–1.92) | 0.925 |
| ||||
LVSI | ||||
| ||||
No | 1 | 1 | ||
| ||||
Yes | 1.11 (0.56–2.20) | 0.774 | 1.02 (0.51–2.03) | 0.961 |
| ||||
Lower uterine segment involvement | ||||
| ||||
No | 1 | 1 | ||
| ||||
Yes | 0.85 (0.33–2.21) | 0.744 | 0.94 (0.36–2.45) | 0.902 |
| ||||
Cervical involvement | ||||
| ||||
No | 1 | 1 | ||
| ||||
Yes | 0.83 (0.37–1.83) | 0.639 | 1.02 (0.46–2.29) | 0.955 |
| ||||
Uterine serosal or ovarian involvement | ||||
| ||||
No | 1 | 1 | ||
| ||||
Yes | 2.82 (1.31–6.08) | 0.008 |
2.87 (0.32–6.22) | 0.008 |
| ||||
Omental involvement | ||||
| ||||
No | 1 | 1 | ||
| ||||
Yes | 2.07 (0.70–6.17) | 0.189 | 1.70 (0.57–5.05) | 0.338 |
| ||||
Peritoneal cytology positive | ||||
| ||||
No | 1 | 1 | ||
| ||||
Yes | 5.00 (1.81–13.85) | 0.002 |
6.32 (2.22–17.96) | <0.001 |
| ||||
Residual tumor | ||||
| ||||
No | 1 | 1 | ||
| ||||
Yes | 1.89 (0.95–3.77) | 0.069 | 1.93 (0.96–3.87) | 0.064 |
| ||||
Lymph node metastasis | ||||
| ||||
No | 1 | 1 | ||
| ||||
Yes | 3.41 (1.56–7.44) | 0.002 |
4.06(1.80–9.15) | <0.001 |
| ||||
Lymph node count (nodes) | ||||
| ||||
<10 | 1 | 1 | ||
| ||||
≥10 | 0.58 (0.30–1.13) | 0.107 | 0.38 (0.168–0.88) | 0.024 |
| ||||
Adjuvant radiotherapy | ||||
| ||||
No | 1 | 1 | ||
| ||||
Yes | 0.67 (0.30–1.47) | 0.317 | 0.72 (0.33–1.59) | 0.418 |
| ||||
Adjuvant chemotherapy | ||||
| ||||
No | 1 | 1 | ||
| ||||
Yes | 0.36 (0.14–0.93) | 0.035 |
0.29 (0.11–0.76) | 0.011 |
Values are presented as number (%).
PFS, progression-free survival; HR, hazard ratio; CI, confidence interval; OS, overall survival; BMI, body mass index; FIGO, International Federation of Gynecology and Obstetrics; MI, myometrial invasion; LVSI, lymphovascular space involvement.
Multivariate analysis for the correlation of clinicopathologic and treatment parameters with PFS
Parameter | HR (95% CI) | |
---|---|---|
FIGO stage | ||
I/II | 1 | |
III/IV | 2.80 (1.41–5.55) | 0.003 |
Peritoneal cytology positive | ||
No | 1 | |
Yes | 4.70 (1.42–15.59) | 0.011 |
PFS, progression-free survival; HR, hazard ratio; CI, confidence interval; FIGO, International Federation of Gynecology and Obstetrics.
Multivariate analysis for the correlation of clinicopathologic and treatment parameters with OS
Parameter | HR (95% CI) | |
---|---|---|
BMI (kg/m2) | ||
<25 | 1 | |
≥25 | 0.63 (0.32–1.3) | 0.198 |
FIGO stage | ||
I/II | 1 | |
III | 2.33 (1.03–5.27) | 0.042 |
IV | 5.91 (2.35–14.86) | <0.001 |
Adjuvant chemotherapy | ||
No | 0.1 | |
Yes | 0.24 (0.09–0.65) | 0.005 |
OS, overall survival; HR, hazard ratio; CI, confidence interval; BMI, body mass index; FIGO, International Federation of Gynecology and Obstetrics.