Preoperative risk stratification in endometrial cancer (ENDORISK) by a Bayesian network model: A development and validation study

Abstract

Background: Bayesian networks (BNs) are machine-learning-based computational models that visualize causal relationships and provide insight into the processes underlying disease progression, closely resembling clinical decision-making. Preoperative identification of patients at risk for lymph node metastasis (LNM) is challenging in endometrial cancer, and although several biomarkers are related to LNM, none of them are incorporated in clinical practice. The aim of this study was to develop and externally validate a preoperative BN to predict LNM and outcome in endometrial cancer patients.

Methods and findings: Within the European Network for Individualized Treatment of Endometrial Cancer (ENITEC), we performed a retrospective multicenter cohort study including 763 patients, median age 65 years (interquartile range [IQR] 58-71), surgically treated for endometrial cancer between February 1995 and August 2013 at one of the 10 participating European hospitals. A BN was developed using score-based machine learning in addition to expert knowledge. Our main outcome measures were LNM and 5-year disease-specific survival (DSS). Preoperative clinical, histopathological, and molecular biomarkers were included in the network. External validation was performed using 2 prospective study cohorts: the Molecular Markers in Treatment in Endometrial Cancer (MoMaTEC) study cohort, including 446 Norwegian patients, median age 64 years (IQR 59-74), treated between May 2001 and 2010; and the PIpelle Prospective ENDOmetrial carcinoma (PIPENDO) study cohort, including 384 Dutch patients, median age 66 years (IQR 60-73), treated between September 2011 and December 2013. A BN called ENDORISK (preoperative risk stratification in endometrial cancer) was developed including the following predictors: preoperative tumor grade; immunohistochemical expression of estrogen receptor (ER), progesterone receptor (PR), p53, and L1 cell adhesion molecule (L1CAM); cancer antigen 125 serum level; thrombocyte count; imaging results on lymphadenopathy; and cervical cytology. In the MoMaTEC cohort, the area under the curve (AUC) was 0.82 (95% confidence interval [CI] 0.76-0.88) for LNM and 0.82 (95% CI 0.77-0.87) for 5-year DSS. In the PIPENDO cohort, the AUC for 5-year DSS was 0.84 (95% CI 0.78-0.90). The network was well-calibrated. In the MoMaTEC cohort, 249 patients (55.8%) were classified with <5% risk of LNM, with a false-negative rate of 1.6%. A limitation of the study is the use of imputation to correct for missing predictor variables in the development cohort and the retrospective study design.

Conclusions: In this study, we illustrated how BNs can be used for individualizing clinical decision-making in oncology by incorporating easily accessible and multimodal biomarkers. The network shows the complex interactions underlying the carcinogenetic process of endometrial cancer by its graphical representation. A prospective feasibility study will be needed prior to implementation in the clinic.

Conflict of interest statement

The authors declare no potential conflicts of interest.

Figures

Fig 1. Summarizing overview of the BN development and validation.

BN, Bayesian network; MoMaTEC, Markers for the Treatment of Endometrial Cancer; PIPENDO, PIpelle Prospective ENDOmetrial carcinoma.

Fig 2. Final BN for the prediction of LNM and 5-year DSS.

(A.) Probability estimates are shown when no markers were recorded. (B.) Example of probability estimates in a case with preoperative tumor grade (grade 2), cervical cytology (atypical endometrial cells present), L1CAM expression (positive), and Ca-125 serum levels (>35 IU/ml). Probability distributions are shown in the nodes, and dependencies are indicated by the arrows connecting the nodes. If variables are not connected directly or indirectly, they are assumed to be (conditionally) independent. Often, the direction of the arrows can be given causal meaning. Red bars indicate that the specific variable is instantiated, i.e., a specific value or evidence is provided. Blue bars in the bar plots indicate the resulting probabilities of the probability distributions. Because of imputation, probability distributions vary slightly from Table 2. BN, Bayesian network; Ca-125, cancer antigen 125; DSS, disease-specific survival; LNM, lymph node metastasis; LVSI, lymphovascular space invasion; L1CAM, L1 cell adhesion molecule.

Fig 3. Concept web-based interface of the BN.

The baseline probability estimates for LNM and 5-year DSS (visualized in panel A) are interactively updated when variables are provided to the model (visualized in panel B). BN, Bayesian network; Ca-125; cancer antigen 125; DSS, disease-specific survival; ER, estrogen receptor; LNM, lymph node metastasis; L1CAM, L1 cell adhesion molecule; PR, progesterone receptor.

Fig 4. ROC curves.

(A) Prediction of LNM in the MoMaTEC cohort, (B) prediction of 5-year DSS in the MoMaTEC cohort, and (C) prediction of 5-year DSS in the PIPENDO cohort. Calibration plots for (D) prediction of LNM in the MoMaTEC cohort, (E) prediction of 5-year DSS in the MoMaTEC cohort, and (F) prediction of 5-year DSS in the PIPENDO cohort. (G) Concordance statistics of the BN. The solid blue lines represent the ROC curves obtained by ENDORISK. The blue dotted lines represent the ROC curves including obtained by using only preoperative tumor grade as predictor (as a reference). The vertical bars in panel D represent 95% CIs. AUC, area under the curve; BN, Bayesian network; CI, confidence interval; DSS, disease-specific survival; ENDORISK, preoperative risk stratification in endometrial cancer; LNM, lymph node metastasis; MoMaTEC, Markers for the Treatment of Endometrial Cancer; PIPENDO, PIpelle Prospective ENDOmetrial carcinoma; ROC, receiver operating characteristic.