Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Nomogram for Predicting Recurrence-Free Survival in Chinese Women with Endometrial Cancer after Initial Therapy: External Validation

Nomogram for Predicting Recurrence-Free Survival in Chinese Women with Endometrial Cancer after... Hindawi Journal of Oncology Volume 2020, Article ID 2363545, 11 pages https://doi.org/10.1155/2020/2363545 Research Article Nomogram for Predicting Recurrence-Free Survival in Chinese Women with Endometrial Cancer after Initial Therapy: External Validation 1 1 2 3 1 1 Yuan Cheng, Yangyang Dong, Wenjuan Tian, Hua Zhang, Xiaoping Li, Zhiqi Wang, 2 2 1 2 1 Boer Shan, Yulan Ren, Lihui Wei, Huaying Wang , and Jianliu Wang Department of Obstetrics and Gynecology, Peking University People’s Hospital, No. 11 Xizhimen South Street, Xicheng Dist., Beijing 100044, China Department of Gynecology, Fudan University Shanghai Cancer Center, No. 255, Dong’An Road, Shanghai 200032, China Research Center of Clinical Epidemiology, Peking University 0ird Hospital, Xueyuan Rd 38#, Haidian Dist., Beijing 100191, China Correspondence should be addressed to Huaying Wang; wanghuaying270@163.com and Jianliu Wang; wangjianliu@ pkuph.edu.cn Received 11 September 2019; Revised 27 February 2020; Accepted 3 March 2020; Published 29 May 2020 Academic Editor: Francesca De Felice Copyright © 2020 Yuan Cheng et al. /is is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. /is study aimed at developing an available recurrence-free survival (RFS) model of endometrial cancer (EC) for accurate and individualized prognosis assessment. A training cohort of 520 women with EC who underwent initial surgical treatment and an external validation cohort of 445 eligible EC patients from 2006 to 2016 were analyzed retrospectively. Multivariable Cox proportional hazards regression models were used to develop nomograms for predicting recurrence. /e concordance index (C- index) and the area under the receiver operating characteristic curve (AUC) were calculated to determine the discrimination of RFS prognostic scoring systems. Calibration plots were generated to examine the performance characteristics of the predictive nomograms. Regression analysis revealed that an advanced International Federation of Gynecology and Obstetrics (FIGO) stage, histological grade 3, primary tumor diameter≥2 cm, and positive peritoneal cytology were independent prognostic factors for RFS in EC in the training set. /e nomograms estimated RFS according to these four variables, with a C-index of 0.860, which was superior to that of FIGO stage (2009 criteria), at 0.809 (P � 0.034), in the training cohort. Encouragingly, consistent results were observed in the validation set, with a C-index of 0.875 for the nomogram and a C-index of 0.833 for the FIGO staging (P � 0.0137). Furthermore, the calibrations of the nomograms predicting 3- and 5-year RFS strongly corresponded to the actual survival outcome. In conclusion, this study developed an available nomogram with effective external validation and relatively appreciable discrimination and conformity for the accurate assessment of 3- and 5-year RFS in Chinese women with EC. uterine serous carcinoma, clear cell cancer, and carcino- 1. Introduction sarcoma was identified as type II and accounts for about Endometrial cancer (EC) is the most common gynecological 35%. /e 5-year overall survival of type I and type II was malignancy, ranking as the fourth among female tumors in about 85% and 55%, respectively [4]. Although EC is de- developed countries [1]. Epidemiological analysis in China tected early in most cases and patients begin receiving ap- showed that the morbidity and mortality rates of EC have propriate treatment with a good prognosis, the 5-year overall increased over recent years [2, 3]. Endometrioid adeno- survival of patients with stage I–III EC ranges from 57% to carcinoma, also known as type I, is the most frequent his- 91% and with stage IV is 20–26% [5–7]. /e prognosis of tological subtype and accounts for about 65%. endometrial cancer is also affected by many other factors Nonendometrioid endometrial cancer which includes such as age, tumor grade, and positive peritoneal cytology 2 Journal of Oncology [8]. Individual differences in recurrence or death in women recommended for radiotherapy or chemotherapy. If the with EC at 2 to 3 years after primary therapy vary widely patients with more high-risk factors (stage III-IV, type II endometrial cancer), they were undergone combination [9, 10]. Hence, it is urgent to place greater emphasis on precise and individualized prognosis evaluation and mon- regimen (radiotherapy and chemotherapy). itoring strategies for the management of patients with EC. Individualized mathematical nomograms have been 2.2. Treatment and Follow-Up. All women with stage I–IV widely adopted as auxiliary tools to guide clinical decision EC were enrolled if they had undergone initial surgical making in medical fields [11–14]. In 2014, AlHilli et al. treatment, including total hysterectomy with bilateral sal- developed nomograms stratified histologically to predict the pingo-oophorectomy with or without systematic lymph overall survival of EC patients [10]. In 2016, a nomogram node dissection (pelvic± para-aortic lymphadenectomy). predicted a low recurrence rate in women with EC (stages Patients who were at high risk for cancer development and I–III), which could reduce unnecessary treatment by 60% those with an advanced cancer stage underwent postoper- [15]. However, these risk-scoring models have only been ative adjuvant radiotherapy, systemic chemotherapy, or performed in analyses of internal data, and they lack external their combination. Patients were followed up after initial validation in independent samples based on established surgery. And the occurrence of recurrence or death of the mathematical formulas. patients was recorded. Physical examination and diagnostic /is study retrospectively analyzed data from 965 imaging tests were performed according to the findings. women with stage I–IV EC from two large-scale hospitals that have focused on EC treatment in China over the last 10 years. A nomogram with good discrimination and cali- 2.3. Statistical Analysis bration was developed for both internal and external vali- 2.3.1. Definition of RFS. /e clinical outcome was evaluated dation cohorts based on clinicopathological characteristics according to recurrence-free survival (RFS). /e duration of to predict the probability of 3- and 5-year recurrence-free follow-up for RFS was defined as the time from hysterec- survival (RFS) and overall survival in women with EC. tomy-based surgical treatment to the date of first recurrence or last follow-up if there was no recurrence. In 2. Materials and Methods addition, Kaplan–Meier cumulative survival probability was used in this study. Cumulative survival probability was 2.1. Patients. /e retrospective cohort study included 965 calculated by multiplying probabilities for each prior relapse patients who underwent hysterectomy for stage I–IV EC in time [16]. Peking University People’s Hospital (training cohort, n = 520) and Fudan University Shanghai Cancer Center (validation cohort, n = 445) from January 2006 to December 2.3.2. Nomogram of Prediction Model. /e clinical and 2016. /e present study was approved by the Ethics Com- pathological variables were evaluated for an association with mittee of Peking University People's Hospital and Fudan RFS by univariate and multivariate Cox proportional haz- University Shanghai Cancer Center. /e exclusion criteria ards regression analyses. Associations are represented by the were incomplete clinical data or lost to follow-up. /e hazard ratio (HR) and corresponding 95% confidence in- baseline characteristics collected for all patients were as tervals (CIs) assessed from the model. Variables with follows: (1) essential variables: age and menopausal status; P< 0.05 were identified as independent risk factors for RFS (2) clinical and surgical variables: surgical procedure (with and were retained in the final model. Furthermore, the or without lymphadenectomy); (3) pathological variables: selected high-risk variables were included in the Cox pro- FIGO stage, pathological type, differentiation status, tumor portional hazards models of RFS. /e risk coefficient of each size, peritoneal cytology status, lymphovascular space in- factor was calculated and included in the equations of the volvement (LVSI), lymph node metastasis, depth of myo- individual prediction models for each patient and is pre- metrial invasion, and cervical stromal invasion (clinical stage sented as nomograms. and histological grade for all patients were classified in accordance with the 2009 FIGO criteria and pathological 2.3.3. Validation of the Prediction Models. /e discrimi- type followed by the two types of endometrial carcinoma of Bokhman in 1983); and (4) adjuvant therapy information: nation ability of the prediction models was estimated using radiotherapy, chemotherapy, or their combination. Endo- the Harrell C-index. /e C-index was calculated by Cox metrial cancer patients with high-risk factors were per- regression models of 1000 random bootstrap resamples with formed adjuvant treatment after surgery according to the same sample size for assessing model accuracy [17]. /e pathological findings and comprehensive multidisciplinary C-index ranges from 0.5 to 1, with greater than 0.5 defined as discussion based on international guidelines. Generally, having predictive power. Kaplan–Meier curves were plotted high-risk factors usually include age≥ 60, myometrial according to the bisection method for stratified management invasion≥ 50%, grade 3, LVSI positive, and type II endo- by the nomogram scores for the high- and low-risk groups. Calibration plots were examined by graphic charts for metrial cancer. Patients without any high-risk factor were considered as in the low-risk group who just needed follow- monitoring the average and maximal errors between the predicted 3- and 5-year probability of RFS and the actual up. Patients with high-risk factors (age≥ 60, grade 3, myometrial invasion≥ 50%, stage II, and LVSI) were outcome frequencies by the Kaplan–Meier method. /e Journal of Oncology 3 specificity and sensitivity of the models based on the no- Total patients with endometrial cancer mogram compared with the 2009 FIGO stage for predicting in training cohort (n = 692) RFS were evaluated by calculating the area under the receiver operating characteristic (ROC) curve (AUC). Excluded due to 2.3.4. Additional Statistical Analysis. /e follow-up time missing data (n = 150) was described using median, ranging from min to max; frequencies and proportions were used for categorical variables. /e clinical features of the cohorts were analyzed using Student’s t-test. P< 0.05 was considered statistically Patients with endometrial cancer identified ( n = 542) significant. Data were collected using Microsoft Excel and converted to .sav files. All analyses were performed using SPSS v20.0 and R v2.15.0 with the Hmisc, rms, and Presence Absence packages. Excluded due to lost to follow-up (n =22) 3. Results 3.1. Training Cohort. We included 520 women with EC in the training cohort (Figure 1). /e percentage of types I EC Patients with endometrial cancer included (n = 520) was 87.5%. /e distribution of women EC was 84.8% with stages I and II and 15.2% with stages III and stage IV. /e Figure 1: Flow diagram of the study participants. number of women with low-grade and high-grade EC was 393 (75.6%) and 127 (24.4%), respectively. /e clinico- pathological characteristics of women are shown in Table 1. the training set. Furthermore, the AUCs for the 3- and 5-year /e median follow-up period for RFS was 53 months (range, RFS nomograms were 0.894 (95% CI: 0.832–0.956) and 1–110); 46 women (8.8%) relapsed, and 474 (91.2%) showed 0.873 (95% CI: 0.812–0.934), respectively, which was su- no recurrence. /e median time from initial therapy to perior to that of the 2009 FIGO classification at 0.849 (95% recurrence was 12 months (range, 1–100). /e mean (SD) 3- CI: 0.777–0.920; P � 0.0268) and 0.816 (95% CI: 0.75–0.89; and 5-year RFS was 92.0%± 1.3% and 90.1%± 1.6%, P � 0.0037), respectively (Figures 3(a) and 3(b)). respectively. 3.4. Validation of the Nomogram. We recruited 445 eligible 3.2. Prediction Nomogram. /e clinicopathological charac- women with EC for the validation cohort. /e clinico- teristics of EC patients from the training sets with or without pathological characteristics of EC patients from the vali- recurrence were analyzed (Supplementary Table 1). /e dation cohort with or without recurrence were analyzed results of the univariate and multivariate analyses revealed (Supplementary Tables 2–4). /e frequency of type I and that four of the screened variables including advanced stage, type II EC was 359 (80.7%) and 86 (19.3%), respectively. /e G3, primary tumor diameter≥ 2 cm, and positive peritoneal distribution of women with EC was as follows: 264 (59.3%) cytology were independent prognostic factors in the training with stage I, 38 (8.5%) with stage II, 111 (25.0%) with stage group (Table 2). /e predictive nomograms were con- III, and 32 (7.2%) with stage IV. /e number of women with structed based on the selected covariates to assess the low-grade and high-grade EC was 297 (66.7%) and 148 probability of 3- and 5-year RFS in the training set (Fig- (33.3%), respectively. /e median follow-up period for RFS ure 2). /e incorporated mathematical formula of the no- was 28 months (range, 1–112); 92 (20.7%) women relapsed, mograms involved FIGO stages II (HR � 2.4; 95% CI: and 353 (79.3%) showed no recurrence. /e median time 0.9–6.7), III (HR � 4.2; 95% CI: 1.9–9.5), and IV (HR � 15.1; from initial therapy to recurrence was 12 months (range, 95% CI: 5.4–42.8), G3 (HR � 4.2; 95% CI: 2.1–8.4), tumor 1–63). /e mean (SD) 3- and 5-year RFS was 78.8% ± 1.2% diameter≥ 2 (HR � 2.9; 95% CI: 1.0–8.4), and positive and 72.8%± 1.6%, respectively. peritoneal cytology (HR � 2.5; 95% CI: 1.2–5.0), with further /e 3- and 5-year RFS rates were also calculated in the score transformation. According to the formulas for the validation cohort based on the nomogram for the training nomograms, the total scores for each patient for 3- and 5- cohort. /e C-index for the RFS nomogram was 0.875 (95% year RFS could be easily and accurately calculated to indi- CI: 0.829–0.921), which was superior to that of the 2009 vidualize the prognosis. FIGO classification at 0.833 (95% CI: 0.785–0.882; P � 0.0137). /e AUCs for 3- and 5-year RFS were 0.875 3.3.ComparisonwithFIGOStage. /e discrimination ability (95% CI: 0.829–0.921) and 0.867 (95% CI: 0.823–0.910), respectively, which were superior to those of the 2009 FIGO of the nomograms was compared with that of FIGO stage. /e C-index of the RFS nomogram was 0.860 (95% CI: classification at 0.833 (95% CI: 0.785–0.882; P � 0.0137) and 0.829 (95% CI: 0.785–0.882; P � 0.0296), respectively 0.797–0.923), which was superior to that of the 2009 FIGO classification, at 0.809 (95% CI: 0.738–0.879; P � 0.034) in (Figures 3(c) and 3(d)). 4 Journal of Oncology Table 1: /e clinicopathological characteristics of the training Calibration plots for the nomograms to predict 3- and 5- cohort. year RFS were calculated in the internal validation. /e predicted 3- and 5-year RFS rates were similar to the actual Training cohort, n � 520 Variables survival rates, with small average error rates of less than 10% (no. of patients) (%) and a lack of bias, as represented by the dotted lines in Age (years) Figures 4(a) and 4(b). Moreover, external validation showed <60 339 (65.2) ≥60 181 (34.8) no dramatic differences between the predicted and actual 3- Menopausal status and 5-year RFS rates (Figures 4(c) and 4(d)). No 177 (34.0) Yes 343 (66.0) Depth of myometrial invasion 3.5. Optimal Nomogram 0reshold and Redistribution. <50% 391 (75.2) /e low- and high-risk groups were defined according to the ≥50% 129 (24.8) optimal threshold of the ROC calculated from the recur- Cervical stromal invasion rence distribution of each probability of the RFS nomograms No 461 (88.7) in the training cohort (P � 0.029). We further analyzed the Yes 59 (11.3) distribution of patients in the low- and high-risk groups Adnexal involvement estimated by the nomogram scores. /e frequency of low- No 492 (94.6) and high-risk EC was 371 (71.3%) and 149 (28.7%), re- Yes 28 (5.4) spectively (Table 3). We found individual differences in FIGO stage recurrence in women with EC after redistribution. In the I 404 (77.7) II 37 (7.1) low-risk group, the characteristic distribution among EC III 62 (11.9) patients with high-risk factors was as follows: advanced stage IV 17 (3.3) (4, 1.1%), grade 3 (22, 5.9%), primary tumor diameter≥ 2 cm Histological grade (192, 51.8%), and positive peritoneal cytology (1, 0.3%). In Grade 1 161 (31.0) the high-risk group, the number of patients with low-risk Grade 2 232 (44.6) factors was as follows: FIGO stage I, G1/G2, primary tumor Grade 3 127 (24.4) diameter< 2 cm, and negative peritoneal cytology was 55 Histological type (36.9%), 44 (29.5%), 10 (6.7%), and 103 (69.1%), Type I 455 (87.5) respectively. Type II 65 (12.5) Tumor diameter <2 cm 189 (36.3) 4. Discussion ≥2 cm 331 (63.7) Peritoneal cytology Predictive nomograms for the assessment of EC prognosis No 473 (91.0) and recurrence have preliminarily been well developed in Yes 47 (9.0) Europe and the United States [10, 15]. Nevertheless, these LVSI prediction models are heterogeneous according to women No 331 (74.4) presenting with EC in different populations [18]. As far as Yes 114 (25.6) we know, a nomogram to predict recurrence in EC pa- Lymph node involvement tients based on the Chinese population has not been No 401 (77.1) established until now. In the current study, we developed a Yes 49 (9.4) nomogram to predict recurrence in women with EC in Unknown 70 (13.5) China, which was well validated in an independent cohort Lymphadenectomy No 70 (13.5) from Fudan University Shanghai Cancer Center. An Yes 450 (86.5) available nomogram for predicting RFS in Chinese Adjuvant therapy women with EC after initial therapy was preliminarily No adjuvant therapy 285 (54.8) developed and externally validated. We have sorted out a Radiotherapy 10 (2.0) table to compare the predicting recurrence-free survival Chemotherapy 165 (31.7) model with previous studies (Table 4). (1) /e predictive Radiotherapy + cChemotherapy 60 (11.5) model was established based on the data from patients Recurrence with endometrial cancer in all stages (I–IV), not just No 474 (91.2) focusing on early stages (I–III) in this study. (2) /e Yes 46 (8.8) independent risk factors of endometrial cancer recurrence Follow-up (months) selected by multivariate analysis in the present study were Median 53 Mean 48.7 some differences from those in the previous studies. Range 1–110 Advanced stage, grade 3, primary tumor diameter≥ 2 cm, Data are expressed as n (%) or the means± SD. FIGO = International and positive peritoneal cytology were independent re- Federation of Gynecology and Obstetrics; LVSI = lymphovascular space current factors in the training group in this study. One or involvement. more of these indicators have been included in previously Journal of Oncology 5 Table 2: Multivariate Cox proportional hazards regression analysis for recurrence-free survival (RFS) in the training cohort. Univariate analysis Multivariate analysis Variables P value P value OR (95% CI) Adjusted OR (95% CI) Age (years) <60 1 (referent) ≥60 2.4 (1.3–4.3) 0.003 Depth of myometrial invasion <50% 1 (referent) ≥50% 2.9 (1.6–5.2) <0.001 Cervical stromal invasion No 1 (referent) Yes 4.1 (2.2–7.6) <0.001 Histological grade Grade 1/2 1 (referent) 1 (referent) Grade 3 9.4 (5.0–18.0) <0.001 4.2 (2.1–8.4) <0.001 Histological type Type I 1 (referent) Type II 8.5 (4.7–15.2) <0.001 Tumor diameter <2 cm 1 (referent) 1 (referent) ≥2 cm 6.6 (2.4–18.4) <0.001 2.9 (1.0–8.4) 0.049 LVSI No 1 (referent) Yes 4.1 (2.2–7.5) <0.001 FIGO stage I 1 (referent) 1 (referent) II 5.4 (2.0–14.3) 0.001 2.4 (0.9–6.7) 0.099 III 8.8 (4.1–19.0) <0.001 4.2 (1.9–9.5) <0.001 IV 84.3 (36.3–195.5) <0.001 15.1 (5.4–42.8) <0.001 Peritoneal cytology No 1 (referent) 1 (referent) Yes 10.1 (5.6–18.3) <0.001 2.5 (1.2–5.0) 0.013 Lymph node involvement No 1 (referent) Yes 8.1 (4.2–15.5) <0.001 Lymphadenectomy No 1 (referent) Yes 0.7 (0.3–1.4) 0.27 Adjuvant therapy No adjuvant therapy 1 (referent) Radiotherapy 9.6 (2.0–45.6) 0.004 Chemotherapy 6.7 (3.0–14.7) <0.001 Radiotherapy + chemotherapy 5.3 (2.0–14.1) <0.001 OR � odds ratio; CI � confidence interval; LVSI � lymphovascular space involvement; FIGO � International Federation of Gynecology and Obstetrics. established models; however, the combination of these of 3-year or 5-year relapse-free survival of patients with four indicators was included in a recurrence model for the endometrial cancer. /is comparison has not been per- first time in this study. (3) Other studies have only pre- formed in previous studies. dicted 3-year relapse-free survival in patients with en- Four independent risk factors, i.e., FIGO stage, histo- dometrial cancer. Our prediction model is focused on logical grade, tumor diameter, and peritoneal cytology both 3-year relapse-free survival and 5-year relapse-free status, were primarily deemed predictive factors of RFS to survival. /e model established in this study proved to be develop a nomogram for patients with stage I–IV EC in the accurate and stable through external verification. /e training set after confounding factors adjustment. 455 cases accuracy and the verification of the model were higher of type I and 65 cases of type II were included in this study. than those of other types of models. (4) We also compared However, pathological type was not selected as an inde- our prediction model with the FIGO staging and found pendent risk factor for recurrence in patients with endo- that it was superior to the FIGO staging in the prediction metrial cancer. Accumulating evidence has affirmed that 6 Journal of Oncology 0 102030405060708090 100 risk group and women with low-stage disease presenting Points other high-risk factors allocated to the high-risk group. II VI /erefore, individual differences in prognosis are more FIGO stage I III common in EC patients. G3 Kondalsamy-Chennakesavan et al. developed nomo- Histological grade G1/G2 grams to predict EC recurrence in 2097 patients with stage ≥2 Tumor diameter I–III EC from 1997 to 2009. /e multivariate Cox model <2 indicated that age, FIGO stage, histological grade, LVSI, Yes Peritoneal cytology tumor type, and peritoneal cytology status were independent No prognostic factors of EC relapse [28]. Bendifallah et al. Total points attempted to validate the Kondalsamy-Chennakesavan no- 0 20 40 60 80 100 120 140 160 180 200 220 240 mogram of EC for prognosis evaluation in 271 cases with Linear predictor –1.5 –1 –0.5 0 0.5 1 1.5 234 2.5 3.5 4.5 5.0 stage I–III EC using an independent, multicenter external patient cohort. However, results showed that the nomogram was only partially generalized in another independent 3-year recurrence-free survival 0.95 0.9 0.85 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 population, with a discrimination ability of 0.66 for 3-year 5-year recurrence-free survival recurrence-free survival [18]. /e study reported disparate 0.95 0.9 0.85 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 population characteristics, recommendations for lymph node resection, and adjuvant chemotherapy in multiple Figure 2: Nomograms for predicting 3- and 5-year recurrence-free survival (RFS) in patients with endometrial cancer. In order to regions, and the efficacy of the prediction model in the evaluate the recurrence-free survival rate of each patient, the score external validation was relatively limited. Further optimi- of each variable was calculated by the value of the “Points” axis, and zation and improvement of nomograms and existing risk the sum of the values of all variables was corresponding to the stratification strategies are needed [29]. In the present study, number of the “Total points” axis. /e vertical line of the total score based on the four variables of the model, we verified the was corresponding to 3- and 5-year probability of recurrence-free accuracy of our nomogram in an external cohort, with a survival. C-index of 0.875 for the predicted 3-year RFS of EC. Ad- ditionally, the calibration curve was acceptable despite little LVSI significantly contributes to replapse in stage (I–III) and differences in several clinical characteristics, recurrence high-risk EC patients [9, 14]. However, LVSI was not the outcomes, and follow-up times between the two pop- most appropriate pathological variable for assessing recur- ulations. /e number of patients with advanced endometrial rence risk in our study, which may be due to the limited cancer and the proportion of patients with recurrence in the sample size. validation cohort were higher than those in the training /e influence of the peritoneal cytology status on the cohort. However, the follow-up time of EC patients in the prognosis of women with EC remains and should be validation cohort was shorter. More optimistically, the above further evaluated. Previous multivariate analyses have four covariates used to develop the prediction models in the revealed that positive peritoneal cytology could predict training cohort were also identified as independent prog- relapse and tumor-related death in early-stage EC [19–23]. nostic factors of EC relapse in the validation cohort by However, other studies found that, in low-risk patients multivariate analysis. with EC, positive peritoneal cytology did not affect the 5- Our study has some limitations. Retrospective clinical year disease-free survival rate [24, 25]. Positive peritoneal data with uncertain potential confounding factors could cytology has also previously shown no effects on overall negatively affect the accuracy of the results. Inevitably, many survival or disease-free survival in patients with low- or observations with missing data were deleted, which could intermediate-risk disease [26, 27]. In the present study, we have caused bias. Abu-Rustum et al. included the number of found that positive peritoneal cytology was a critical and lymph nodes removed during comprehensive surgery in a independent prognostic factor of EC recurrence in both predictive model of prognosis for women with EC [14]. cohorts. /e possible reason may be that our study in- AlHilli et al. revealed inadequate/negative lymphadenec- volved EC patients with progressive staging or poor tomy as an independent risk factor in low-risk patients with differentiation. EC [10]. We also assessed the value of surgical treatment and Notably, the results showed that the RFS estimation of adjuvant therapy in the recurrence of patients with EC. the nomograms according to the four variables was superior Univariate analysis showed that they were associated with to that of the 2009 FIGO classification in the training cohort. endometrial cancer recurrence. However, multivariate /e calibrations of the nomograms predicting 3- and 5-year analysis showed that they were not independent risk factors RFS highly corresponded to the actual survival rates, with for endometrial cancer recurrence in our study. /e reason minute average error rates of less than 10% for both. We may be the limited number of cases or other interaction further divided the patients into low- and high-risk groups factors. Additionally, the modeling and validation groups according to the optimal threshold of ROC from the re- differed in clinical characteristics and recurrence rates, currence distribution of each probability of the RFS no- which might also have affected the results. Finally, further mograms. Predictably, we discovered women presenting optimization of this model in a national multicenter study is only one high-risk factor who were represented in the low- needed. Journal of Oncology 7 Compare 2 models Compare 2 models 1.0 1.0 0.8 0.8 0.6 0.6 0.4 0.4 0.2 0.2 0.0 0.0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 1 – specificity 1 – specificity AUC AUC Model 1: 0.894 Model 1: 0.873 Model 2: 0.849 Model 2: 0.816 (a) (b) Compare 2 models Compare 2 models 1.0 1.0 0.8 0.8 0.6 0.6 0.4 0.4 0.2 0.2 0.0 0.0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 1 – specificity 1 – specificity AUC AUC Model 1: 0.875 Model 1: 0.867 Model 2: 0.833 Model 2: 0.829 (c) (d) Figure 3: Area under the receiver operating characteristic curve (AUC) values of the nomogram and the 2009 International Federation of Gynecology and Obstetrics (FIGO) stage system for the training and validation cohorts to assess the 3- and 5-year recurrence-free survival (RFS). (a, b) Comparing the AUCs of nomogram with FIGO stage in the training group (T) for predicting 3- and 5-year recurrence-free survival (RFS) in patients with endometrial cancer. (c, d) Comparing the AUCs of nomogram with FIGO stage in the validation group (V) for predicting 3- and 5-year recurrence-free survival (RFS) in patients with endometrial cancer. Model 1 of black line represents nomogram; model 2 of red line represents FIGO stage. (a) T: 3 RFS-AUC nomogram:FIGO, (b) T: 5 RFS-AUC nomogram:FIGO, (c) V: 3 RFS-AUC nomogram:FIGO, and (d) V: 5 RFS-AUC nomogram:FIGO. Sensitivity Sensitivity Sensitivity Sensitivity 8 Journal of Oncology 1.0 1.0 � � � � 0.9 0.9 0.8 0.8 0.7 0.7 0.6 0.6 � 0.5 0.5 0.5 0.6 0.7 0.8 0.9 1.0 0.5 0.6 0.7 0.8 0.9 1.0 Predicted probability 3-year recurrence-free survival Predicted probability 3-year recurrence-free survival n = 520 d = 46 p = 6, X − resampling optimism added, n = 520 d = 46 p = 6, X − resampling optimism added, 130 subjects per group B = 1000 130 subjects per group B = 999 Gray: ideal Based on observed−predicted Gray: ideal Based on observed−predicted (a) (b) 1.0 1.0 � � 0.8 0.8 0.6 0.6 0.4 0.4 0.2 0.2 0.0 0.0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 Predicted probability 3-year recurrence-free survival Predicted probability 3-year recurrence-free survival n = 445 d = 92 p = 1, X − resampling optimism added, n = 445 d = 92 p = 1, X − resampling optimism added, 110 subjects per group B = 1000 110 subjects per group B = 1000 Gray: ideal Based on observed−predicted Gray: ideal Based on observed−predicted (c) (d) Figure 4: Calibration of the nomograms for 3- and 5-year recurrence-free survival (RFS) in patients with stage I–IV endometrial cancer in the training and validation cohorts. X-axis indicates the predicting probability of nomogram. Y-axis shows the actual 3- or 5-year probability of survival as assessed by Kaplan–Meier curves. Red line represents the predicted probability of nomogram. Gray line represents ideal consistency between the prediction and actual probabilities of 3- or 5-year RFS. Vertical bars represent 95% CI. Blue dots correspond to the accuracy of the prediction. (a, b) Calibrations of the nomogram in the training cohort for predicting 3- and 5-year RFS. (c, d) Calibrations of the nomogram in the validation cohort for predicting 3- and 5-year RFS. We developed a clinically available and relatively precise tumor diameter≥2 cm, and positive peritoneal cytology and model based on nomograms to predict RFS in Chinese helped develop a tool that may be conducive for developing women with EC. /e study specifically incorporated the four individualized therapeutic strategies for Chinese patients independent prognosis covariates of advanced stage, grade 3, with EC. Actual 3-year recurrence-free survival (proportion) Actual 3-year recurrence-free survival (proportion) Actual 3-year recurrence-free survival (proportion) Actual 3-year recurrence-free survival (proportion) Journal of Oncology 9 Table 3: /e distribution of patients in the low- and high-risk cohorts estimated by the nomogram scores in the training cohort. Variables Low risk, N � 371 (71.3%) High risk, N � 149 (28.7%) P value FIGO stage <0.001 I 349 (94.1) 55 (36.9) II 18 (4.8) 19 (12.8) III 4 (1.1) 58 (38.9) IV 0 (0) 17 (11.4) Histological grade <0.001 Grade 1/2 349 (94.1) 1 44 (29.5) Grade 3 22 (5.9) 105 (70.5) Primary tumor diameter <0.001 <2 cm 179 (48.2) 10 (6.7) ≥2 cm 192 (51.8) 139 (93.3) Peritoneal cytology <0.001 No 370 (99.7) 103 (69.1) Yes 1 (0.3) 1 46 (30.9) Data are expressed as n (%). FIGO � International Federation of Gynecology and Obstetrics. Table 4: Comparison of this study with previous studies of predictive models for recurrence-free survival of endometrial cancer patients. Obermair A Bendifallah S Ouldamer L Wang J Number of cases 2097 396 861 520 Recurrence-free survival 3-year 3-year 3-year 3- and 5-year Histologic type I, II I I, II I, II FIGO stage I–III I–III I–III I–IV Factors Age Yes Yes Yes — Histological grade Yes Yes — Yes FIGO stage Yes — Yes Yes Histologic type Yes — Yes — Tumor diameter ≥ 2 cm — Yes — Yes Myometrial invasion ≥ 50 % — Yes — — LVSI Yes Yes Yes — Peritoneal washing Yes — — Yes Surgical nodal staging — — Yes Comparison with FIGO stage — — — Superior 3-year internal validation Yes (0.86) Yes (0.74) Yes (0.75) Yes (0.89) 5-year internal validation — — — Yes (0.87) 3-year external validation — Yes (0.82) — Yes (0.88) 5-year external validation — — — Yes (0.87) FIGO � International Federation of Gynecology and Obstetrics; LVSI � lymphovascular space involvement. Abbreviations Conflicts of Interest RFS: Recurrence-free survival /e authors declare that they have no conflicts of interest. EC: Endometrial cancer C-index: Concordance index Authors’ Contributions AUC: Area under the receiver operating characteristic Yuan Cheng and Yangyang Dong contributed equally to this curve article. Prof. Jianliu Wang and Prof. Huaying Wang have FIGO: Federation of Gynecology and Obstetrics substantial contributions to the conception and design of EC: Endometrial cancer this project. Yuan Cheng and Yangyang Dong were re- LVSI: Lymphovascular space involvement sponsible for the acquisition of data, or analysis and in- HR: Hazard ratio. terpretation of data, and writing and revising the manuscript. Hua Zhang mainly gave guidance in data sta- Data Availability tistics and R language programming. Wenjuan Tian, Boer /e datasets used and/or analyzed during the current study Shan, and Yulan Ren mainly established the database of are available from the corresponding author on reasonable endometrial cancer from Fudan University Shanghai Cancer request. Center. Prof. Zhiqi Wang, Prof. Xiaoping Li, and Prof. Lihui 10 Journal of Oncology endometrial cancer?” British Journal of Cancer, vol. 112, no. 5, Wei gave good advice for the clinical application of this pp. 793–801, 2015. predicting model. [10] M. M. AlHilli, A. Mariani, J. N. Bakkum-Gamez et al., “Risk- scoring models for individualized prediction of overall sur- Acknowledgments vival in low-grade and high-grade endometrial cancer,” Gy- necologic Oncology, vol. 133, no. 3, pp. 485–493, 2014. /is study was supported by the National Natural Science [11] Y. Dong, Y. Cheng, W Tian et al., “An externally validated Foundation of China (Grant no. 81802607), Special Projects nomogram for predicting lymph node metastasis of presumed for Strengthening Basic Research of Peking University stage I and II endometrial cancer,” Frontiers in Oncology, (Grant no. BMU2018JC005), the Fund for Fostering Young vol. 9, p. 1218, 2019. Scholars of Peking University Health Science Center (Grant [12] Y. Kim, G. A. Margonis, J. D. Prescott et al., “Nomograms to no. BMU2017PY011), the Research and Development Fund predict recurrence-free and overall survival after curative of Peking University People’s Hospital (Grant no. RDY2017- resection of adrenocortical carcinoma,” JAMA Surgery, 12), and National Key Technology Research and Develop- vol. 151, no. 4, pp. 365–373, 2016. [13] R. Rouzier, C. Uzan, A. Rousseau et al., “Multicenter pro- ment Program of the Ministry of Science and Technology of spective evaluation of the reliability of the combined use of China (Grant no. 2015BAI13B06). two models to predict non-sentinel lymph node status in breast cancer patients with metastatic sentinel lymph nodes: Supplementary Materials the MSKCC nomogram and the Tenon score. Results of the NOTEGS study,” British Journal of Cancer, vol. 116, no. 9, Supplementary Table 1: clinicopathological characteristics of pp. 1135–1140, 2017. 520 endometrial cancer patients from the training cohort [14] N. R. Abu-Rustum, Q. Zhou, J. D. Gomez et al., “A nomogram with or without recurrence. Supplementary Table 2: the for predicting overall survival of women with endometrial clinicopathological characteristics of 445 endometrial cancer cancer following primary therapy: toward improving indi- patients of the validation cohort. Supplementary Table 3: vidualized cancer care,” Gynecologic Oncology, vol. 116, no. 3, characteristics of 445 women with endometrial carcinoma in pp. 399–403, 2010. [15] L. Ouldamer, S. Bendifallah, G. Body et al., “Predicting poor the validation cohort with or without recurrence. Supple- prognosis recurrence in women with endometrial cancer: a mentary Table 4: multivariate Cox proportional hazards nomogram developed by the FRANCOGYN study group,” regression analysis for recurrence-free survival (RFS) in the British Journal of Cancer, vol. 115, no. 11, pp. 1296–1303, 2016. validation cohort. (Supplementary Materials) [16] A. Hazra and N. Gogtay, “Biostatistics series module 9: survival analysis,” Indian Journal of Dermatology, vol. 62, References no. 3, pp. 251–257, 2017. [17] F. E. Harrell, K. L. Lee Jr., and D. B. Mark, “Multivariable [1] R. L. Siegel, K. D. Miller, and A. Jemal, “Cancer statistics, prognostic models: issues in developing models, evaluating 2018,” CA: A Cancer Journal for Clinicians, vol. 68, no. 1, assumptions and adequacy, and measuring and reducing pp. 7–30, 2018. errors,” Statistics in Medicine, vol. 15, no. 4, pp. 361–387, 1996. [2] W. Chen, R. Zheng, P. D. Baade et al., “Cancer statistics in [18] S. Bendifallah, G. Canlorbe, E. Raimond et al., “An external China, 2015,” CA: A Cancer Journal for Clinicians, vol. 66, validation study of nomograms designed to predict isolated no. 2, pp. 115–132, 2016. loco-regional and distant endometrial cancer recurrences: [3] W. Chen, K. Sun, R. Zheng et al., “Cancer incidence and how applicable are they?” British Journal of Cancer, vol. 109, mortality in China, 2014,” Chinese Journal of Cancer Research, no. 6, pp. 1498–1503, 2013. vol. 30, no. 1, pp. 1–12, 2018. [19] M. Momomura, H. Matsumoto, S Umezawa et al., “Histologic [4] J. V. Bokhman, “Two pathogenetic types of endometrial grade and peritoneal cytology as prognostic factors in type 1 carcinoma,” Gynecologic Oncology, vol. 15, no. 1, pp. 10–17, endometrial cancer,” International Journal of Clinical On- cology, vol. 22, no. 3, pp. 533–540, 2017. [5] P. Morice, A. Leary, C. Creutzberg, N. Abu-Rustum, and [20] S. Sasada, M. Yunokawa, Y. Takehara et al., “Baseline risk of E. Darai, “Endometrial cancer,” 0e Lancet, vol. 387, recurrence in stage I-II endometrial carcinoma,” Journal of no. 10023, pp. 1094–1108, 2016. Gynecologic Oncology, vol. 29, no. 1, p. e9, 2018. [6] M. Aamir Sheikh, A. D. Althouse, K. E. Freese et al., “USA [21] T. Tuomi, A. Pasanen, A. Leminen, R. Butzow, ¨ and endometrial cancer projections to 2030: should we be con- M. Loukovaara, “Prediction of site-specific tumor relapses in cerned?” Future Oncology, vol. 10, no. 16, pp. 2561–2568, 2014. patients with stage I-II endometrioid endometrial cancer,” [7] S. Bendifallah, L. Ouldamer, V. Lavoue et al., “Patterns of International Journal of Gynecological Cancer, vol. 27, no. 5, recurrence and outcomes in surgically treated women with pp. 923–930, 2017. endometrial cancer according to ESMO-ESGO-ESTRO [22] K. Matsuo, A. Yabuno, M. S. Hom et al., “Significance of Consensus Conference risk groups: results from the FRAN- abnormal peritoneal cytology on survival of women with stage COGYN study Group,” Gynecologic Oncology, vol. 144, no. 1, I-II endometrioid endometrial cancer,” Gynecologic Oncology, pp. 107–112, 2017. vol. 149, no. 2, pp. 301–309, 2018. [8] Y. Dai, Y. Dong, Y Cheng et al., “Prognostic significance of [23] B. L. Seagle, A. L. Alexander, T. Lantsman, and S. Shahabi, lymphovascular space invasion in patients with endometrioid “Prognosis and treatment of positive peritoneal cytology in endometrial cancer: a retrospective study from a single center,” Journal of Gynecologic Oncology, vol. 31, 2019. early endometrial cancer: matched cohort analyses from the National Cancer Database,” American Journal of Obstetrics [9] S. Bendifallah, G. Canlorbe, P. Collinet et al., “Just how ac- curate are the major risk stratification systems for early-stage and Gynecology, vol. 218, no. 3, pp. 329.e1–329.e15, 2018. Journal of Oncology 11 [24] N. Takeshima, H. Nishida, T. Tabata, Y. Hirai, and K. Hasumi, “Positive peritoneal cytology in endometrial cancer: en- hancement of other prognostic indicators,” Gynecologic On- cology, vol. 82, no. 3, pp. 470–473, 2001. [25] S. A. Scott, C. van der Zanden, E. Cai, C. E. McGahan, and J. S. Kwon, “Prognostic significance of peritoneal cytology in low-intermediate risk endometrial cancer,” Gynecologic On- cology, vol. 145, no. 2, pp. 262–268, 2017. [26] B. Lee, D. H. Suh, K. Kim, J. H. No, and Y. B. Kim, “Influence of positive peritoneal cytology on prognostic factors and survival in early-stage endometrial cancer: a systematic review and meta-analysis,” Japanese Journal of Clinical Oncology, vol. 46, no. 8, pp. 711–717, 2016. [27] N. R. Abu-Rustum, A. Iasonos, Q Zhou et al., “Is there a therapeutic impact to regional lymphadenectomy in the surgical treatment of endometrial carcinoma?” American Journal of Obstetrics and Gynecology, vol. 198, no. 4, pp. 457.e1–457.e6, 2008. [28] S. Kondalsamy-Chennakesavan, C. Yu, M. W. Kattan et al., “Nomograms to predict isolated loco-regional or distant re- currence among women with uterine cancer,” Gynecologic Oncology, vol. 125, no. 3, pp. 520–525, 2012. [29] S. Bendifallah, G. Canlorbe, F. Huguet et al., “A risk scoring system to determine recurrence in early-stage type 1 endo- metrial cancer: a French multicentre study,” Annals of Sur- gical Oncology, vol. 21, no. 13, pp. 4239–4245, 2014. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Journal of Oncology Hindawi Publishing Corporation

Nomogram for Predicting Recurrence-Free Survival in Chinese Women with Endometrial Cancer after Initial Therapy: External Validation

Loading next page...
 
/lp/hindawi-publishing-corporation/nomogram-for-predicting-recurrence-free-survival-in-chinese-women-with-QrMJeGLCYf

References (36)

Publisher
Hindawi Publishing Corporation
Copyright
Copyright © 2020 Yuan Cheng et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
ISSN
1687-8450
eISSN
1687-8469
DOI
10.1155/2020/2363545
Publisher site
See Article on Publisher Site

Abstract

Hindawi Journal of Oncology Volume 2020, Article ID 2363545, 11 pages https://doi.org/10.1155/2020/2363545 Research Article Nomogram for Predicting Recurrence-Free Survival in Chinese Women with Endometrial Cancer after Initial Therapy: External Validation 1 1 2 3 1 1 Yuan Cheng, Yangyang Dong, Wenjuan Tian, Hua Zhang, Xiaoping Li, Zhiqi Wang, 2 2 1 2 1 Boer Shan, Yulan Ren, Lihui Wei, Huaying Wang , and Jianliu Wang Department of Obstetrics and Gynecology, Peking University People’s Hospital, No. 11 Xizhimen South Street, Xicheng Dist., Beijing 100044, China Department of Gynecology, Fudan University Shanghai Cancer Center, No. 255, Dong’An Road, Shanghai 200032, China Research Center of Clinical Epidemiology, Peking University 0ird Hospital, Xueyuan Rd 38#, Haidian Dist., Beijing 100191, China Correspondence should be addressed to Huaying Wang; wanghuaying270@163.com and Jianliu Wang; wangjianliu@ pkuph.edu.cn Received 11 September 2019; Revised 27 February 2020; Accepted 3 March 2020; Published 29 May 2020 Academic Editor: Francesca De Felice Copyright © 2020 Yuan Cheng et al. /is is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. /is study aimed at developing an available recurrence-free survival (RFS) model of endometrial cancer (EC) for accurate and individualized prognosis assessment. A training cohort of 520 women with EC who underwent initial surgical treatment and an external validation cohort of 445 eligible EC patients from 2006 to 2016 were analyzed retrospectively. Multivariable Cox proportional hazards regression models were used to develop nomograms for predicting recurrence. /e concordance index (C- index) and the area under the receiver operating characteristic curve (AUC) were calculated to determine the discrimination of RFS prognostic scoring systems. Calibration plots were generated to examine the performance characteristics of the predictive nomograms. Regression analysis revealed that an advanced International Federation of Gynecology and Obstetrics (FIGO) stage, histological grade 3, primary tumor diameter≥2 cm, and positive peritoneal cytology were independent prognostic factors for RFS in EC in the training set. /e nomograms estimated RFS according to these four variables, with a C-index of 0.860, which was superior to that of FIGO stage (2009 criteria), at 0.809 (P � 0.034), in the training cohort. Encouragingly, consistent results were observed in the validation set, with a C-index of 0.875 for the nomogram and a C-index of 0.833 for the FIGO staging (P � 0.0137). Furthermore, the calibrations of the nomograms predicting 3- and 5-year RFS strongly corresponded to the actual survival outcome. In conclusion, this study developed an available nomogram with effective external validation and relatively appreciable discrimination and conformity for the accurate assessment of 3- and 5-year RFS in Chinese women with EC. uterine serous carcinoma, clear cell cancer, and carcino- 1. Introduction sarcoma was identified as type II and accounts for about Endometrial cancer (EC) is the most common gynecological 35%. /e 5-year overall survival of type I and type II was malignancy, ranking as the fourth among female tumors in about 85% and 55%, respectively [4]. Although EC is de- developed countries [1]. Epidemiological analysis in China tected early in most cases and patients begin receiving ap- showed that the morbidity and mortality rates of EC have propriate treatment with a good prognosis, the 5-year overall increased over recent years [2, 3]. Endometrioid adeno- survival of patients with stage I–III EC ranges from 57% to carcinoma, also known as type I, is the most frequent his- 91% and with stage IV is 20–26% [5–7]. /e prognosis of tological subtype and accounts for about 65%. endometrial cancer is also affected by many other factors Nonendometrioid endometrial cancer which includes such as age, tumor grade, and positive peritoneal cytology 2 Journal of Oncology [8]. Individual differences in recurrence or death in women recommended for radiotherapy or chemotherapy. If the with EC at 2 to 3 years after primary therapy vary widely patients with more high-risk factors (stage III-IV, type II endometrial cancer), they were undergone combination [9, 10]. Hence, it is urgent to place greater emphasis on precise and individualized prognosis evaluation and mon- regimen (radiotherapy and chemotherapy). itoring strategies for the management of patients with EC. Individualized mathematical nomograms have been 2.2. Treatment and Follow-Up. All women with stage I–IV widely adopted as auxiliary tools to guide clinical decision EC were enrolled if they had undergone initial surgical making in medical fields [11–14]. In 2014, AlHilli et al. treatment, including total hysterectomy with bilateral sal- developed nomograms stratified histologically to predict the pingo-oophorectomy with or without systematic lymph overall survival of EC patients [10]. In 2016, a nomogram node dissection (pelvic± para-aortic lymphadenectomy). predicted a low recurrence rate in women with EC (stages Patients who were at high risk for cancer development and I–III), which could reduce unnecessary treatment by 60% those with an advanced cancer stage underwent postoper- [15]. However, these risk-scoring models have only been ative adjuvant radiotherapy, systemic chemotherapy, or performed in analyses of internal data, and they lack external their combination. Patients were followed up after initial validation in independent samples based on established surgery. And the occurrence of recurrence or death of the mathematical formulas. patients was recorded. Physical examination and diagnostic /is study retrospectively analyzed data from 965 imaging tests were performed according to the findings. women with stage I–IV EC from two large-scale hospitals that have focused on EC treatment in China over the last 10 years. A nomogram with good discrimination and cali- 2.3. Statistical Analysis bration was developed for both internal and external vali- 2.3.1. Definition of RFS. /e clinical outcome was evaluated dation cohorts based on clinicopathological characteristics according to recurrence-free survival (RFS). /e duration of to predict the probability of 3- and 5-year recurrence-free follow-up for RFS was defined as the time from hysterec- survival (RFS) and overall survival in women with EC. tomy-based surgical treatment to the date of first recurrence or last follow-up if there was no recurrence. In 2. Materials and Methods addition, Kaplan–Meier cumulative survival probability was used in this study. Cumulative survival probability was 2.1. Patients. /e retrospective cohort study included 965 calculated by multiplying probabilities for each prior relapse patients who underwent hysterectomy for stage I–IV EC in time [16]. Peking University People’s Hospital (training cohort, n = 520) and Fudan University Shanghai Cancer Center (validation cohort, n = 445) from January 2006 to December 2.3.2. Nomogram of Prediction Model. /e clinical and 2016. /e present study was approved by the Ethics Com- pathological variables were evaluated for an association with mittee of Peking University People's Hospital and Fudan RFS by univariate and multivariate Cox proportional haz- University Shanghai Cancer Center. /e exclusion criteria ards regression analyses. Associations are represented by the were incomplete clinical data or lost to follow-up. /e hazard ratio (HR) and corresponding 95% confidence in- baseline characteristics collected for all patients were as tervals (CIs) assessed from the model. Variables with follows: (1) essential variables: age and menopausal status; P< 0.05 were identified as independent risk factors for RFS (2) clinical and surgical variables: surgical procedure (with and were retained in the final model. Furthermore, the or without lymphadenectomy); (3) pathological variables: selected high-risk variables were included in the Cox pro- FIGO stage, pathological type, differentiation status, tumor portional hazards models of RFS. /e risk coefficient of each size, peritoneal cytology status, lymphovascular space in- factor was calculated and included in the equations of the volvement (LVSI), lymph node metastasis, depth of myo- individual prediction models for each patient and is pre- metrial invasion, and cervical stromal invasion (clinical stage sented as nomograms. and histological grade for all patients were classified in accordance with the 2009 FIGO criteria and pathological 2.3.3. Validation of the Prediction Models. /e discrimi- type followed by the two types of endometrial carcinoma of Bokhman in 1983); and (4) adjuvant therapy information: nation ability of the prediction models was estimated using radiotherapy, chemotherapy, or their combination. Endo- the Harrell C-index. /e C-index was calculated by Cox metrial cancer patients with high-risk factors were per- regression models of 1000 random bootstrap resamples with formed adjuvant treatment after surgery according to the same sample size for assessing model accuracy [17]. /e pathological findings and comprehensive multidisciplinary C-index ranges from 0.5 to 1, with greater than 0.5 defined as discussion based on international guidelines. Generally, having predictive power. Kaplan–Meier curves were plotted high-risk factors usually include age≥ 60, myometrial according to the bisection method for stratified management invasion≥ 50%, grade 3, LVSI positive, and type II endo- by the nomogram scores for the high- and low-risk groups. Calibration plots were examined by graphic charts for metrial cancer. Patients without any high-risk factor were considered as in the low-risk group who just needed follow- monitoring the average and maximal errors between the predicted 3- and 5-year probability of RFS and the actual up. Patients with high-risk factors (age≥ 60, grade 3, myometrial invasion≥ 50%, stage II, and LVSI) were outcome frequencies by the Kaplan–Meier method. /e Journal of Oncology 3 specificity and sensitivity of the models based on the no- Total patients with endometrial cancer mogram compared with the 2009 FIGO stage for predicting in training cohort (n = 692) RFS were evaluated by calculating the area under the receiver operating characteristic (ROC) curve (AUC). Excluded due to 2.3.4. Additional Statistical Analysis. /e follow-up time missing data (n = 150) was described using median, ranging from min to max; frequencies and proportions were used for categorical variables. /e clinical features of the cohorts were analyzed using Student’s t-test. P< 0.05 was considered statistically Patients with endometrial cancer identified ( n = 542) significant. Data were collected using Microsoft Excel and converted to .sav files. All analyses were performed using SPSS v20.0 and R v2.15.0 with the Hmisc, rms, and Presence Absence packages. Excluded due to lost to follow-up (n =22) 3. Results 3.1. Training Cohort. We included 520 women with EC in the training cohort (Figure 1). /e percentage of types I EC Patients with endometrial cancer included (n = 520) was 87.5%. /e distribution of women EC was 84.8% with stages I and II and 15.2% with stages III and stage IV. /e Figure 1: Flow diagram of the study participants. number of women with low-grade and high-grade EC was 393 (75.6%) and 127 (24.4%), respectively. /e clinico- pathological characteristics of women are shown in Table 1. the training set. Furthermore, the AUCs for the 3- and 5-year /e median follow-up period for RFS was 53 months (range, RFS nomograms were 0.894 (95% CI: 0.832–0.956) and 1–110); 46 women (8.8%) relapsed, and 474 (91.2%) showed 0.873 (95% CI: 0.812–0.934), respectively, which was su- no recurrence. /e median time from initial therapy to perior to that of the 2009 FIGO classification at 0.849 (95% recurrence was 12 months (range, 1–100). /e mean (SD) 3- CI: 0.777–0.920; P � 0.0268) and 0.816 (95% CI: 0.75–0.89; and 5-year RFS was 92.0%± 1.3% and 90.1%± 1.6%, P � 0.0037), respectively (Figures 3(a) and 3(b)). respectively. 3.4. Validation of the Nomogram. We recruited 445 eligible 3.2. Prediction Nomogram. /e clinicopathological charac- women with EC for the validation cohort. /e clinico- teristics of EC patients from the training sets with or without pathological characteristics of EC patients from the vali- recurrence were analyzed (Supplementary Table 1). /e dation cohort with or without recurrence were analyzed results of the univariate and multivariate analyses revealed (Supplementary Tables 2–4). /e frequency of type I and that four of the screened variables including advanced stage, type II EC was 359 (80.7%) and 86 (19.3%), respectively. /e G3, primary tumor diameter≥ 2 cm, and positive peritoneal distribution of women with EC was as follows: 264 (59.3%) cytology were independent prognostic factors in the training with stage I, 38 (8.5%) with stage II, 111 (25.0%) with stage group (Table 2). /e predictive nomograms were con- III, and 32 (7.2%) with stage IV. /e number of women with structed based on the selected covariates to assess the low-grade and high-grade EC was 297 (66.7%) and 148 probability of 3- and 5-year RFS in the training set (Fig- (33.3%), respectively. /e median follow-up period for RFS ure 2). /e incorporated mathematical formula of the no- was 28 months (range, 1–112); 92 (20.7%) women relapsed, mograms involved FIGO stages II (HR � 2.4; 95% CI: and 353 (79.3%) showed no recurrence. /e median time 0.9–6.7), III (HR � 4.2; 95% CI: 1.9–9.5), and IV (HR � 15.1; from initial therapy to recurrence was 12 months (range, 95% CI: 5.4–42.8), G3 (HR � 4.2; 95% CI: 2.1–8.4), tumor 1–63). /e mean (SD) 3- and 5-year RFS was 78.8% ± 1.2% diameter≥ 2 (HR � 2.9; 95% CI: 1.0–8.4), and positive and 72.8%± 1.6%, respectively. peritoneal cytology (HR � 2.5; 95% CI: 1.2–5.0), with further /e 3- and 5-year RFS rates were also calculated in the score transformation. According to the formulas for the validation cohort based on the nomogram for the training nomograms, the total scores for each patient for 3- and 5- cohort. /e C-index for the RFS nomogram was 0.875 (95% year RFS could be easily and accurately calculated to indi- CI: 0.829–0.921), which was superior to that of the 2009 vidualize the prognosis. FIGO classification at 0.833 (95% CI: 0.785–0.882; P � 0.0137). /e AUCs for 3- and 5-year RFS were 0.875 3.3.ComparisonwithFIGOStage. /e discrimination ability (95% CI: 0.829–0.921) and 0.867 (95% CI: 0.823–0.910), respectively, which were superior to those of the 2009 FIGO of the nomograms was compared with that of FIGO stage. /e C-index of the RFS nomogram was 0.860 (95% CI: classification at 0.833 (95% CI: 0.785–0.882; P � 0.0137) and 0.829 (95% CI: 0.785–0.882; P � 0.0296), respectively 0.797–0.923), which was superior to that of the 2009 FIGO classification, at 0.809 (95% CI: 0.738–0.879; P � 0.034) in (Figures 3(c) and 3(d)). 4 Journal of Oncology Table 1: /e clinicopathological characteristics of the training Calibration plots for the nomograms to predict 3- and 5- cohort. year RFS were calculated in the internal validation. /e predicted 3- and 5-year RFS rates were similar to the actual Training cohort, n � 520 Variables survival rates, with small average error rates of less than 10% (no. of patients) (%) and a lack of bias, as represented by the dotted lines in Age (years) Figures 4(a) and 4(b). Moreover, external validation showed <60 339 (65.2) ≥60 181 (34.8) no dramatic differences between the predicted and actual 3- Menopausal status and 5-year RFS rates (Figures 4(c) and 4(d)). No 177 (34.0) Yes 343 (66.0) Depth of myometrial invasion 3.5. Optimal Nomogram 0reshold and Redistribution. <50% 391 (75.2) /e low- and high-risk groups were defined according to the ≥50% 129 (24.8) optimal threshold of the ROC calculated from the recur- Cervical stromal invasion rence distribution of each probability of the RFS nomograms No 461 (88.7) in the training cohort (P � 0.029). We further analyzed the Yes 59 (11.3) distribution of patients in the low- and high-risk groups Adnexal involvement estimated by the nomogram scores. /e frequency of low- No 492 (94.6) and high-risk EC was 371 (71.3%) and 149 (28.7%), re- Yes 28 (5.4) spectively (Table 3). We found individual differences in FIGO stage recurrence in women with EC after redistribution. In the I 404 (77.7) II 37 (7.1) low-risk group, the characteristic distribution among EC III 62 (11.9) patients with high-risk factors was as follows: advanced stage IV 17 (3.3) (4, 1.1%), grade 3 (22, 5.9%), primary tumor diameter≥ 2 cm Histological grade (192, 51.8%), and positive peritoneal cytology (1, 0.3%). In Grade 1 161 (31.0) the high-risk group, the number of patients with low-risk Grade 2 232 (44.6) factors was as follows: FIGO stage I, G1/G2, primary tumor Grade 3 127 (24.4) diameter< 2 cm, and negative peritoneal cytology was 55 Histological type (36.9%), 44 (29.5%), 10 (6.7%), and 103 (69.1%), Type I 455 (87.5) respectively. Type II 65 (12.5) Tumor diameter <2 cm 189 (36.3) 4. Discussion ≥2 cm 331 (63.7) Peritoneal cytology Predictive nomograms for the assessment of EC prognosis No 473 (91.0) and recurrence have preliminarily been well developed in Yes 47 (9.0) Europe and the United States [10, 15]. Nevertheless, these LVSI prediction models are heterogeneous according to women No 331 (74.4) presenting with EC in different populations [18]. As far as Yes 114 (25.6) we know, a nomogram to predict recurrence in EC pa- Lymph node involvement tients based on the Chinese population has not been No 401 (77.1) established until now. In the current study, we developed a Yes 49 (9.4) nomogram to predict recurrence in women with EC in Unknown 70 (13.5) China, which was well validated in an independent cohort Lymphadenectomy No 70 (13.5) from Fudan University Shanghai Cancer Center. An Yes 450 (86.5) available nomogram for predicting RFS in Chinese Adjuvant therapy women with EC after initial therapy was preliminarily No adjuvant therapy 285 (54.8) developed and externally validated. We have sorted out a Radiotherapy 10 (2.0) table to compare the predicting recurrence-free survival Chemotherapy 165 (31.7) model with previous studies (Table 4). (1) /e predictive Radiotherapy + cChemotherapy 60 (11.5) model was established based on the data from patients Recurrence with endometrial cancer in all stages (I–IV), not just No 474 (91.2) focusing on early stages (I–III) in this study. (2) /e Yes 46 (8.8) independent risk factors of endometrial cancer recurrence Follow-up (months) selected by multivariate analysis in the present study were Median 53 Mean 48.7 some differences from those in the previous studies. Range 1–110 Advanced stage, grade 3, primary tumor diameter≥ 2 cm, Data are expressed as n (%) or the means± SD. FIGO = International and positive peritoneal cytology were independent re- Federation of Gynecology and Obstetrics; LVSI = lymphovascular space current factors in the training group in this study. One or involvement. more of these indicators have been included in previously Journal of Oncology 5 Table 2: Multivariate Cox proportional hazards regression analysis for recurrence-free survival (RFS) in the training cohort. Univariate analysis Multivariate analysis Variables P value P value OR (95% CI) Adjusted OR (95% CI) Age (years) <60 1 (referent) ≥60 2.4 (1.3–4.3) 0.003 Depth of myometrial invasion <50% 1 (referent) ≥50% 2.9 (1.6–5.2) <0.001 Cervical stromal invasion No 1 (referent) Yes 4.1 (2.2–7.6) <0.001 Histological grade Grade 1/2 1 (referent) 1 (referent) Grade 3 9.4 (5.0–18.0) <0.001 4.2 (2.1–8.4) <0.001 Histological type Type I 1 (referent) Type II 8.5 (4.7–15.2) <0.001 Tumor diameter <2 cm 1 (referent) 1 (referent) ≥2 cm 6.6 (2.4–18.4) <0.001 2.9 (1.0–8.4) 0.049 LVSI No 1 (referent) Yes 4.1 (2.2–7.5) <0.001 FIGO stage I 1 (referent) 1 (referent) II 5.4 (2.0–14.3) 0.001 2.4 (0.9–6.7) 0.099 III 8.8 (4.1–19.0) <0.001 4.2 (1.9–9.5) <0.001 IV 84.3 (36.3–195.5) <0.001 15.1 (5.4–42.8) <0.001 Peritoneal cytology No 1 (referent) 1 (referent) Yes 10.1 (5.6–18.3) <0.001 2.5 (1.2–5.0) 0.013 Lymph node involvement No 1 (referent) Yes 8.1 (4.2–15.5) <0.001 Lymphadenectomy No 1 (referent) Yes 0.7 (0.3–1.4) 0.27 Adjuvant therapy No adjuvant therapy 1 (referent) Radiotherapy 9.6 (2.0–45.6) 0.004 Chemotherapy 6.7 (3.0–14.7) <0.001 Radiotherapy + chemotherapy 5.3 (2.0–14.1) <0.001 OR � odds ratio; CI � confidence interval; LVSI � lymphovascular space involvement; FIGO � International Federation of Gynecology and Obstetrics. established models; however, the combination of these of 3-year or 5-year relapse-free survival of patients with four indicators was included in a recurrence model for the endometrial cancer. /is comparison has not been per- first time in this study. (3) Other studies have only pre- formed in previous studies. dicted 3-year relapse-free survival in patients with en- Four independent risk factors, i.e., FIGO stage, histo- dometrial cancer. Our prediction model is focused on logical grade, tumor diameter, and peritoneal cytology both 3-year relapse-free survival and 5-year relapse-free status, were primarily deemed predictive factors of RFS to survival. /e model established in this study proved to be develop a nomogram for patients with stage I–IV EC in the accurate and stable through external verification. /e training set after confounding factors adjustment. 455 cases accuracy and the verification of the model were higher of type I and 65 cases of type II were included in this study. than those of other types of models. (4) We also compared However, pathological type was not selected as an inde- our prediction model with the FIGO staging and found pendent risk factor for recurrence in patients with endo- that it was superior to the FIGO staging in the prediction metrial cancer. Accumulating evidence has affirmed that 6 Journal of Oncology 0 102030405060708090 100 risk group and women with low-stage disease presenting Points other high-risk factors allocated to the high-risk group. II VI /erefore, individual differences in prognosis are more FIGO stage I III common in EC patients. G3 Kondalsamy-Chennakesavan et al. developed nomo- Histological grade G1/G2 grams to predict EC recurrence in 2097 patients with stage ≥2 Tumor diameter I–III EC from 1997 to 2009. /e multivariate Cox model <2 indicated that age, FIGO stage, histological grade, LVSI, Yes Peritoneal cytology tumor type, and peritoneal cytology status were independent No prognostic factors of EC relapse [28]. Bendifallah et al. Total points attempted to validate the Kondalsamy-Chennakesavan no- 0 20 40 60 80 100 120 140 160 180 200 220 240 mogram of EC for prognosis evaluation in 271 cases with Linear predictor –1.5 –1 –0.5 0 0.5 1 1.5 234 2.5 3.5 4.5 5.0 stage I–III EC using an independent, multicenter external patient cohort. However, results showed that the nomogram was only partially generalized in another independent 3-year recurrence-free survival 0.95 0.9 0.85 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 population, with a discrimination ability of 0.66 for 3-year 5-year recurrence-free survival recurrence-free survival [18]. /e study reported disparate 0.95 0.9 0.85 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 population characteristics, recommendations for lymph node resection, and adjuvant chemotherapy in multiple Figure 2: Nomograms for predicting 3- and 5-year recurrence-free survival (RFS) in patients with endometrial cancer. In order to regions, and the efficacy of the prediction model in the evaluate the recurrence-free survival rate of each patient, the score external validation was relatively limited. Further optimi- of each variable was calculated by the value of the “Points” axis, and zation and improvement of nomograms and existing risk the sum of the values of all variables was corresponding to the stratification strategies are needed [29]. In the present study, number of the “Total points” axis. /e vertical line of the total score based on the four variables of the model, we verified the was corresponding to 3- and 5-year probability of recurrence-free accuracy of our nomogram in an external cohort, with a survival. C-index of 0.875 for the predicted 3-year RFS of EC. Ad- ditionally, the calibration curve was acceptable despite little LVSI significantly contributes to replapse in stage (I–III) and differences in several clinical characteristics, recurrence high-risk EC patients [9, 14]. However, LVSI was not the outcomes, and follow-up times between the two pop- most appropriate pathological variable for assessing recur- ulations. /e number of patients with advanced endometrial rence risk in our study, which may be due to the limited cancer and the proportion of patients with recurrence in the sample size. validation cohort were higher than those in the training /e influence of the peritoneal cytology status on the cohort. However, the follow-up time of EC patients in the prognosis of women with EC remains and should be validation cohort was shorter. More optimistically, the above further evaluated. Previous multivariate analyses have four covariates used to develop the prediction models in the revealed that positive peritoneal cytology could predict training cohort were also identified as independent prog- relapse and tumor-related death in early-stage EC [19–23]. nostic factors of EC relapse in the validation cohort by However, other studies found that, in low-risk patients multivariate analysis. with EC, positive peritoneal cytology did not affect the 5- Our study has some limitations. Retrospective clinical year disease-free survival rate [24, 25]. Positive peritoneal data with uncertain potential confounding factors could cytology has also previously shown no effects on overall negatively affect the accuracy of the results. Inevitably, many survival or disease-free survival in patients with low- or observations with missing data were deleted, which could intermediate-risk disease [26, 27]. In the present study, we have caused bias. Abu-Rustum et al. included the number of found that positive peritoneal cytology was a critical and lymph nodes removed during comprehensive surgery in a independent prognostic factor of EC recurrence in both predictive model of prognosis for women with EC [14]. cohorts. /e possible reason may be that our study in- AlHilli et al. revealed inadequate/negative lymphadenec- volved EC patients with progressive staging or poor tomy as an independent risk factor in low-risk patients with differentiation. EC [10]. We also assessed the value of surgical treatment and Notably, the results showed that the RFS estimation of adjuvant therapy in the recurrence of patients with EC. the nomograms according to the four variables was superior Univariate analysis showed that they were associated with to that of the 2009 FIGO classification in the training cohort. endometrial cancer recurrence. However, multivariate /e calibrations of the nomograms predicting 3- and 5-year analysis showed that they were not independent risk factors RFS highly corresponded to the actual survival rates, with for endometrial cancer recurrence in our study. /e reason minute average error rates of less than 10% for both. We may be the limited number of cases or other interaction further divided the patients into low- and high-risk groups factors. Additionally, the modeling and validation groups according to the optimal threshold of ROC from the re- differed in clinical characteristics and recurrence rates, currence distribution of each probability of the RFS no- which might also have affected the results. Finally, further mograms. Predictably, we discovered women presenting optimization of this model in a national multicenter study is only one high-risk factor who were represented in the low- needed. Journal of Oncology 7 Compare 2 models Compare 2 models 1.0 1.0 0.8 0.8 0.6 0.6 0.4 0.4 0.2 0.2 0.0 0.0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 1 – specificity 1 – specificity AUC AUC Model 1: 0.894 Model 1: 0.873 Model 2: 0.849 Model 2: 0.816 (a) (b) Compare 2 models Compare 2 models 1.0 1.0 0.8 0.8 0.6 0.6 0.4 0.4 0.2 0.2 0.0 0.0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 1 – specificity 1 – specificity AUC AUC Model 1: 0.875 Model 1: 0.867 Model 2: 0.833 Model 2: 0.829 (c) (d) Figure 3: Area under the receiver operating characteristic curve (AUC) values of the nomogram and the 2009 International Federation of Gynecology and Obstetrics (FIGO) stage system for the training and validation cohorts to assess the 3- and 5-year recurrence-free survival (RFS). (a, b) Comparing the AUCs of nomogram with FIGO stage in the training group (T) for predicting 3- and 5-year recurrence-free survival (RFS) in patients with endometrial cancer. (c, d) Comparing the AUCs of nomogram with FIGO stage in the validation group (V) for predicting 3- and 5-year recurrence-free survival (RFS) in patients with endometrial cancer. Model 1 of black line represents nomogram; model 2 of red line represents FIGO stage. (a) T: 3 RFS-AUC nomogram:FIGO, (b) T: 5 RFS-AUC nomogram:FIGO, (c) V: 3 RFS-AUC nomogram:FIGO, and (d) V: 5 RFS-AUC nomogram:FIGO. Sensitivity Sensitivity Sensitivity Sensitivity 8 Journal of Oncology 1.0 1.0 � � � � 0.9 0.9 0.8 0.8 0.7 0.7 0.6 0.6 � 0.5 0.5 0.5 0.6 0.7 0.8 0.9 1.0 0.5 0.6 0.7 0.8 0.9 1.0 Predicted probability 3-year recurrence-free survival Predicted probability 3-year recurrence-free survival n = 520 d = 46 p = 6, X − resampling optimism added, n = 520 d = 46 p = 6, X − resampling optimism added, 130 subjects per group B = 1000 130 subjects per group B = 999 Gray: ideal Based on observed−predicted Gray: ideal Based on observed−predicted (a) (b) 1.0 1.0 � � 0.8 0.8 0.6 0.6 0.4 0.4 0.2 0.2 0.0 0.0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 Predicted probability 3-year recurrence-free survival Predicted probability 3-year recurrence-free survival n = 445 d = 92 p = 1, X − resampling optimism added, n = 445 d = 92 p = 1, X − resampling optimism added, 110 subjects per group B = 1000 110 subjects per group B = 1000 Gray: ideal Based on observed−predicted Gray: ideal Based on observed−predicted (c) (d) Figure 4: Calibration of the nomograms for 3- and 5-year recurrence-free survival (RFS) in patients with stage I–IV endometrial cancer in the training and validation cohorts. X-axis indicates the predicting probability of nomogram. Y-axis shows the actual 3- or 5-year probability of survival as assessed by Kaplan–Meier curves. Red line represents the predicted probability of nomogram. Gray line represents ideal consistency between the prediction and actual probabilities of 3- or 5-year RFS. Vertical bars represent 95% CI. Blue dots correspond to the accuracy of the prediction. (a, b) Calibrations of the nomogram in the training cohort for predicting 3- and 5-year RFS. (c, d) Calibrations of the nomogram in the validation cohort for predicting 3- and 5-year RFS. We developed a clinically available and relatively precise tumor diameter≥2 cm, and positive peritoneal cytology and model based on nomograms to predict RFS in Chinese helped develop a tool that may be conducive for developing women with EC. /e study specifically incorporated the four individualized therapeutic strategies for Chinese patients independent prognosis covariates of advanced stage, grade 3, with EC. Actual 3-year recurrence-free survival (proportion) Actual 3-year recurrence-free survival (proportion) Actual 3-year recurrence-free survival (proportion) Actual 3-year recurrence-free survival (proportion) Journal of Oncology 9 Table 3: /e distribution of patients in the low- and high-risk cohorts estimated by the nomogram scores in the training cohort. Variables Low risk, N � 371 (71.3%) High risk, N � 149 (28.7%) P value FIGO stage <0.001 I 349 (94.1) 55 (36.9) II 18 (4.8) 19 (12.8) III 4 (1.1) 58 (38.9) IV 0 (0) 17 (11.4) Histological grade <0.001 Grade 1/2 349 (94.1) 1 44 (29.5) Grade 3 22 (5.9) 105 (70.5) Primary tumor diameter <0.001 <2 cm 179 (48.2) 10 (6.7) ≥2 cm 192 (51.8) 139 (93.3) Peritoneal cytology <0.001 No 370 (99.7) 103 (69.1) Yes 1 (0.3) 1 46 (30.9) Data are expressed as n (%). FIGO � International Federation of Gynecology and Obstetrics. Table 4: Comparison of this study with previous studies of predictive models for recurrence-free survival of endometrial cancer patients. Obermair A Bendifallah S Ouldamer L Wang J Number of cases 2097 396 861 520 Recurrence-free survival 3-year 3-year 3-year 3- and 5-year Histologic type I, II I I, II I, II FIGO stage I–III I–III I–III I–IV Factors Age Yes Yes Yes — Histological grade Yes Yes — Yes FIGO stage Yes — Yes Yes Histologic type Yes — Yes — Tumor diameter ≥ 2 cm — Yes — Yes Myometrial invasion ≥ 50 % — Yes — — LVSI Yes Yes Yes — Peritoneal washing Yes — — Yes Surgical nodal staging — — Yes Comparison with FIGO stage — — — Superior 3-year internal validation Yes (0.86) Yes (0.74) Yes (0.75) Yes (0.89) 5-year internal validation — — — Yes (0.87) 3-year external validation — Yes (0.82) — Yes (0.88) 5-year external validation — — — Yes (0.87) FIGO � International Federation of Gynecology and Obstetrics; LVSI � lymphovascular space involvement. Abbreviations Conflicts of Interest RFS: Recurrence-free survival /e authors declare that they have no conflicts of interest. EC: Endometrial cancer C-index: Concordance index Authors’ Contributions AUC: Area under the receiver operating characteristic Yuan Cheng and Yangyang Dong contributed equally to this curve article. Prof. Jianliu Wang and Prof. Huaying Wang have FIGO: Federation of Gynecology and Obstetrics substantial contributions to the conception and design of EC: Endometrial cancer this project. Yuan Cheng and Yangyang Dong were re- LVSI: Lymphovascular space involvement sponsible for the acquisition of data, or analysis and in- HR: Hazard ratio. terpretation of data, and writing and revising the manuscript. Hua Zhang mainly gave guidance in data sta- Data Availability tistics and R language programming. Wenjuan Tian, Boer /e datasets used and/or analyzed during the current study Shan, and Yulan Ren mainly established the database of are available from the corresponding author on reasonable endometrial cancer from Fudan University Shanghai Cancer request. Center. Prof. Zhiqi Wang, Prof. Xiaoping Li, and Prof. Lihui 10 Journal of Oncology endometrial cancer?” British Journal of Cancer, vol. 112, no. 5, Wei gave good advice for the clinical application of this pp. 793–801, 2015. predicting model. [10] M. M. AlHilli, A. Mariani, J. N. Bakkum-Gamez et al., “Risk- scoring models for individualized prediction of overall sur- Acknowledgments vival in low-grade and high-grade endometrial cancer,” Gy- necologic Oncology, vol. 133, no. 3, pp. 485–493, 2014. /is study was supported by the National Natural Science [11] Y. Dong, Y. Cheng, W Tian et al., “An externally validated Foundation of China (Grant no. 81802607), Special Projects nomogram for predicting lymph node metastasis of presumed for Strengthening Basic Research of Peking University stage I and II endometrial cancer,” Frontiers in Oncology, (Grant no. BMU2018JC005), the Fund for Fostering Young vol. 9, p. 1218, 2019. Scholars of Peking University Health Science Center (Grant [12] Y. Kim, G. A. Margonis, J. D. Prescott et al., “Nomograms to no. BMU2017PY011), the Research and Development Fund predict recurrence-free and overall survival after curative of Peking University People’s Hospital (Grant no. RDY2017- resection of adrenocortical carcinoma,” JAMA Surgery, 12), and National Key Technology Research and Develop- vol. 151, no. 4, pp. 365–373, 2016. [13] R. Rouzier, C. Uzan, A. Rousseau et al., “Multicenter pro- ment Program of the Ministry of Science and Technology of spective evaluation of the reliability of the combined use of China (Grant no. 2015BAI13B06). two models to predict non-sentinel lymph node status in breast cancer patients with metastatic sentinel lymph nodes: Supplementary Materials the MSKCC nomogram and the Tenon score. Results of the NOTEGS study,” British Journal of Cancer, vol. 116, no. 9, Supplementary Table 1: clinicopathological characteristics of pp. 1135–1140, 2017. 520 endometrial cancer patients from the training cohort [14] N. R. Abu-Rustum, Q. Zhou, J. D. Gomez et al., “A nomogram with or without recurrence. Supplementary Table 2: the for predicting overall survival of women with endometrial clinicopathological characteristics of 445 endometrial cancer cancer following primary therapy: toward improving indi- patients of the validation cohort. Supplementary Table 3: vidualized cancer care,” Gynecologic Oncology, vol. 116, no. 3, characteristics of 445 women with endometrial carcinoma in pp. 399–403, 2010. [15] L. Ouldamer, S. Bendifallah, G. Body et al., “Predicting poor the validation cohort with or without recurrence. Supple- prognosis recurrence in women with endometrial cancer: a mentary Table 4: multivariate Cox proportional hazards nomogram developed by the FRANCOGYN study group,” regression analysis for recurrence-free survival (RFS) in the British Journal of Cancer, vol. 115, no. 11, pp. 1296–1303, 2016. validation cohort. (Supplementary Materials) [16] A. Hazra and N. Gogtay, “Biostatistics series module 9: survival analysis,” Indian Journal of Dermatology, vol. 62, References no. 3, pp. 251–257, 2017. [17] F. E. Harrell, K. L. Lee Jr., and D. B. Mark, “Multivariable [1] R. L. Siegel, K. D. Miller, and A. Jemal, “Cancer statistics, prognostic models: issues in developing models, evaluating 2018,” CA: A Cancer Journal for Clinicians, vol. 68, no. 1, assumptions and adequacy, and measuring and reducing pp. 7–30, 2018. errors,” Statistics in Medicine, vol. 15, no. 4, pp. 361–387, 1996. [2] W. Chen, R. Zheng, P. D. Baade et al., “Cancer statistics in [18] S. Bendifallah, G. Canlorbe, E. Raimond et al., “An external China, 2015,” CA: A Cancer Journal for Clinicians, vol. 66, validation study of nomograms designed to predict isolated no. 2, pp. 115–132, 2016. loco-regional and distant endometrial cancer recurrences: [3] W. Chen, K. Sun, R. Zheng et al., “Cancer incidence and how applicable are they?” British Journal of Cancer, vol. 109, mortality in China, 2014,” Chinese Journal of Cancer Research, no. 6, pp. 1498–1503, 2013. vol. 30, no. 1, pp. 1–12, 2018. [19] M. Momomura, H. Matsumoto, S Umezawa et al., “Histologic [4] J. V. Bokhman, “Two pathogenetic types of endometrial grade and peritoneal cytology as prognostic factors in type 1 carcinoma,” Gynecologic Oncology, vol. 15, no. 1, pp. 10–17, endometrial cancer,” International Journal of Clinical On- cology, vol. 22, no. 3, pp. 533–540, 2017. [5] P. Morice, A. Leary, C. Creutzberg, N. Abu-Rustum, and [20] S. Sasada, M. Yunokawa, Y. Takehara et al., “Baseline risk of E. Darai, “Endometrial cancer,” 0e Lancet, vol. 387, recurrence in stage I-II endometrial carcinoma,” Journal of no. 10023, pp. 1094–1108, 2016. Gynecologic Oncology, vol. 29, no. 1, p. e9, 2018. [6] M. Aamir Sheikh, A. D. Althouse, K. E. Freese et al., “USA [21] T. Tuomi, A. Pasanen, A. Leminen, R. Butzow, ¨ and endometrial cancer projections to 2030: should we be con- M. Loukovaara, “Prediction of site-specific tumor relapses in cerned?” Future Oncology, vol. 10, no. 16, pp. 2561–2568, 2014. patients with stage I-II endometrioid endometrial cancer,” [7] S. Bendifallah, L. Ouldamer, V. Lavoue et al., “Patterns of International Journal of Gynecological Cancer, vol. 27, no. 5, recurrence and outcomes in surgically treated women with pp. 923–930, 2017. endometrial cancer according to ESMO-ESGO-ESTRO [22] K. Matsuo, A. Yabuno, M. S. Hom et al., “Significance of Consensus Conference risk groups: results from the FRAN- abnormal peritoneal cytology on survival of women with stage COGYN study Group,” Gynecologic Oncology, vol. 144, no. 1, I-II endometrioid endometrial cancer,” Gynecologic Oncology, pp. 107–112, 2017. vol. 149, no. 2, pp. 301–309, 2018. [8] Y. Dai, Y. Dong, Y Cheng et al., “Prognostic significance of [23] B. L. Seagle, A. L. Alexander, T. Lantsman, and S. Shahabi, lymphovascular space invasion in patients with endometrioid “Prognosis and treatment of positive peritoneal cytology in endometrial cancer: a retrospective study from a single center,” Journal of Gynecologic Oncology, vol. 31, 2019. early endometrial cancer: matched cohort analyses from the National Cancer Database,” American Journal of Obstetrics [9] S. Bendifallah, G. Canlorbe, P. Collinet et al., “Just how ac- curate are the major risk stratification systems for early-stage and Gynecology, vol. 218, no. 3, pp. 329.e1–329.e15, 2018. Journal of Oncology 11 [24] N. Takeshima, H. Nishida, T. Tabata, Y. Hirai, and K. Hasumi, “Positive peritoneal cytology in endometrial cancer: en- hancement of other prognostic indicators,” Gynecologic On- cology, vol. 82, no. 3, pp. 470–473, 2001. [25] S. A. Scott, C. van der Zanden, E. Cai, C. E. McGahan, and J. S. Kwon, “Prognostic significance of peritoneal cytology in low-intermediate risk endometrial cancer,” Gynecologic On- cology, vol. 145, no. 2, pp. 262–268, 2017. [26] B. Lee, D. H. Suh, K. Kim, J. H. No, and Y. B. Kim, “Influence of positive peritoneal cytology on prognostic factors and survival in early-stage endometrial cancer: a systematic review and meta-analysis,” Japanese Journal of Clinical Oncology, vol. 46, no. 8, pp. 711–717, 2016. [27] N. R. Abu-Rustum, A. Iasonos, Q Zhou et al., “Is there a therapeutic impact to regional lymphadenectomy in the surgical treatment of endometrial carcinoma?” American Journal of Obstetrics and Gynecology, vol. 198, no. 4, pp. 457.e1–457.e6, 2008. [28] S. Kondalsamy-Chennakesavan, C. Yu, M. W. Kattan et al., “Nomograms to predict isolated loco-regional or distant re- currence among women with uterine cancer,” Gynecologic Oncology, vol. 125, no. 3, pp. 520–525, 2012. [29] S. Bendifallah, G. Canlorbe, F. Huguet et al., “A risk scoring system to determine recurrence in early-stage type 1 endo- metrial cancer: a French multicentre study,” Annals of Sur- gical Oncology, vol. 21, no. 13, pp. 4239–4245, 2014.

Journal

Journal of OncologyHindawi Publishing Corporation

Published: May 29, 2020

There are no references for this article.