Nomogram with potential clinical use to predict lymph node metastasis in endometrial cancer patients diagnosed incidentally by postoperative pathological assessment

The value of machine learning in preoperative identification of lymph node metastasis status in endometrial cancer: a systematic review and meta-analysis

Background

The early identification of lymph node metastasis status in endometrial cancer (EC) is a serious challenge in clinical practice. Some investigators have introduced machine learning into the early identification of lymph node metastasis in EC patients. However, the predictive value of machine learning is controversial due to the diversity of models and modeling variables. To this end, we carried out this systematic review and meta-analysis to systematically discuss the value of machine learning for the early identification of lymph node metastasis in EC patients.

Methods

A systematic search was conducted in Pubmed, Cochrane, Embase, and Web of Science until March 12, 2023. PROBAST was used to assess the risk of bias in the included studies. In the process of meta-analysis, subgroup analysis was performed according to modeling variables (clinical features, radiomic features, and radiomic features combined with clinical features) and different types of models in various variables.

Results

This systematic review included 50 primary studies with a total of 103,752 EC patients, 12,579 of whom had positive lymph node metastasis. Meta-analysis showed that among the machine learning models constructed by the three categories of modeling variables, the best model was constructed by combining radiomic features with clinical features, with a pooled c-index of 0.907 (95%CI: 0.886-0.928) in the training set and 0.823 (95%CI: 0.757-0.890) in the validation set, and good sensitivity and specificity. The c-index of the machine learning model constructed based on clinical features alone was not inferior to that based on radiomic features only. In addition, logistic regression was found to be the main modeling method and has ideal predictive performance with different categories of modeling variables.

Conclusion

Although the model based on radiomic features combined with clinical features has the best predictive efficiency, there is no recognized specification for the application of radiomics at present. In addition, the logistic regression constructed by clinical features shows good sensitivity and specificity. In this context, large-sample studies covering different races are warranted to develop predictive nomograms based on clinical features, which can be widely applied in clinical practice.

Systematic review registration

https://www.crd.york.ac.uk/PROSPERO, identifier CRD42023420774.

Demographic Factors Predict Risk of Lymph Node Involvement in Patients with Endometrial Adenocarcinoma

The presence of lymph node positivity (LN+) guides adjuvant treatment for endometrial adenocarcinoma (EAC) patients, but recommendations regarding LN evaluation at the time of primary surgery remain variable. Sociodemographic factors in addition to pathologic tumor characteristics may more accurately predict risk of LN+ in EAC patients. Patients diagnosed between 2004 and 2016 with pathologic T1-T2 EAC who had at least one lymph node sampled at the time of surgery in the National Cancer Data Base were included. Pathologic primary tumor predictors of LN+ were identified using logistic regression. To predict overall, pelvic only, and paraaortic and/or pelvic LN+, nomograms were generated. Among the 35,170 EAC patients included, 2864 were node positive. Using multivariable analysis, younger patient age (OR 0.98, 95% CI 0.98-0.99, p < 0.001), black versus white race (OR 1.19, 95% CI 1.01-1.40, p = 0.04), increasing pathologic tumor stage and grade, increase in tumor size, and presence of lymphovascular invasion were predictive of regional LN+. Both black versus white (OR 1.64, 95% CI 1.27-2.09, p < 0.001) and other versus white race (OR 1.54, 95% CI 1.12-2.07, p = 0.006) strongly predicted paraaortic LN+ in the multivariable analysis. Independent subset analyses of black and white women revealed that tumor grade was a stronger predictor of LN+ among black women. In addition to standard pathologic tumor features, patient age and race were associated with a higher risk of regional LN+ generally and paraaortic LN+ specifically. This information may inform adjuvant treatment decisions and guide future studies.

Pathologic primary tumor factors associated with risk of lymph node involvement in patients with non-endometrioid endometrial cancer

PURPOSE/OBJECTIVES

Lymph node (LN) involvement is an important factor in guiding adjuvant treatment for patients with endometrial cancer. Risk factors for LN involvement are fairly well-established for endometrial adenocarcinoma, but it is not as well defined whether these factors similarly predict LN positivity in less common histologies.

MATERIALS/METHODS

Patients diagnosed with pathologic T1-T2 carcinosarcoma, clear cell, uterine papillary serous carcinoma (UPSC), and mixed histologic type endometrial cancer between 2004 and 2016 undergoing primary surgery with at least 1 lymph node sampled in the National Cancer Data Base were identified. Logistic regression was performed to identify primary pathologic tumor predictors of LN positivity. Nomograms were created to predict overall, pelvic only, and paraaortic with or without pelvic LN involvement.

RESULTS

Among 11,390 patients included, 1950 (18%) were node positive. On multivariable analysis, increasing pathologic tumor stage (pT2 versus pT1a, odds ratio [OR] 3.63, 95% confidence interval [CI] 3.15-4.18, p < 0.001), increase in tumor size per centimeter (OR 1.08, 95% CI 1.06-1.10, p < 0.001), and the presence of lymphovascular invasion (LVI) (OR 4.97, 95% CI 4.43-5.57, p < 0.001) were predictive of overall LN positivity. Relative to carcinosarcoma, both clear cell (OR 1.54, 95% CI 1.22-1.95, p < 0.001) and UPSC (OR 1.73, 95% CI 1.48-2.02, p < 0.001) histology were significantly associated with a higher risk of LN positivity while mixed histology was not (OR 1.07, 95% CI 0.92-1.24, p = 0.42).

CONCLUSION

Among patients with non-endometrioid endometrial cancer, predictors of LN positivity are similar to endometrial adenocarcinoma. The nomograms provided could be helpful in making adjuvant treatment decisions for these less common histologies.

Association of Myometrial Invasion With Lymphovascular Space Invasion, Lymph Node Metastasis, Recurrence, and Overall Survival in Endometrial Cancer: A Meta-Analysis of 79 Studies With 68,870 Patients

Background

Myometrial invasion has been demonstrated to correlate to clinicopathological characteristics and prognosis in endometrial cancer. However, not all the studies have the consistent results and no meta-analysis has investigated the association of myometrial invasion with lymphovascular space invasion (LVSI), lymph node metastasis (LNM), recurrence, and overall survival (OS). Therefore, a meta-analysis was performed to evaluate the relationship between myometrial invasion and clinicopathological characteristics or overall survival in endometrial cancer.

Materials and Methods

A search of Pubmed, Embase, and Web of Science was carried out to collect relevant studies from their inception until June 30, 2021. The quality of each included study was evaluated using Newcastle–Ottawa scale (NOS) scale. Review Manager version 5.4 was employed to conduct the meta-analysis.

Results

A total of 79 articles with 68,870 endometrial cancer patients were eligible including 9 articles for LVSI, 29 articles for LNM, 8 for recurrence, and 37 for OS in this meta-analysis. Myometrial invasion was associated with LVSI (RR 3.07; 95% CI 2.17–4.35; p < 0.00001), lymph node metastasis (LNM) (RR 4.45; 95% CI 3.29–6.01; p < 0.00001), and recurrence (RR 2.06; 95% CI 1.58–2.69; p < 0.00001). Deep myometrial invasion was also significantly related with poor OS via meta-synthesis of HRs in both univariate survival (HR 3.36, 95% CI 2.35–4.79, p < 0.00001) and multivariate survival (HR 2.00, 95% CI 1.59–2.53, p < 0.00001). Funnel plot suggested that there was no significant publication bias in this study.

Conclusion

Deep myometrial invasion correlated to positive LVSI, positive LNM, cancer recurrence, and poor OS for endometrial cancer patients, indicating that myometrial invasion was a useful evaluation criterion to associate with clinical outcomes and prognosis of endometrial cancer since depth of myometrial invasion can be assessed before surgery. The large scale and comprehensive meta-analysis suggested that we should pay more attention to myometrial invasion in clinical practice, and its underlying mechanism also deserves further investigation.

A methylomics-correlated nomogram predicts the recurrence free survival risk of kidney renal clear cell carcinoma

BACKGROUND

Various studies have suggested that the DNA methylation signatures were promising to identify novel hallmarks for predicting prognosis of cancer. However, few studies have explored the capacity of DNA methylation for prognostic prediction in patients with kidney renal clear cell carcinoma (KIRC). It's very promising to develop a methylomics-related signature for predicting prognosis of KIRC.

METHODS

The 282 patients with complete DNA methylation data and corresponding clinical information were selected to construct the prognostic model. The 282 patients were grouped into a training set (70%, n = 198 samples) to determine a prognostic predictor by univariate Cox proportional hazard analysis, least absolute shrinkage and selection operator (LASSO) and multivariate Cox regression analysis. The internal validation set (30%, n = 84) and an external validation set (E-MTAB-3274) were used to validate the predictive value of the predictor by receiver operating characteristic (ROC) analysis and Kaplan-Meier survival analysis.

RESULTS

We successfully identified a 9-DNA methylation signature for recurrence free survival (RFS) of KIRC patients. We proved the strong robustness of the 9-DNA methylation signature for predicting RFS through ROC analysis (AUC at 1, 3, 5 years in internal dataset (0.859, 0.840, 0.817, respectively), external validation dataset (0.674, 0.739, 0.793, respectively), entire TCGA dataset (0.834, 0.862, 0.842, respectively)). In addition, a nomogram combining methylation risk score with the conventional clinic-related covariates was constructed to improve the prognostic predicted ability for KIRC patients. The result implied a good performance of the nomogram.

CONCLUSIONS

we successfully identified a DNA methylation-associated nomogram, which was helpful in improving the prognostic predictive ability of KIRC patients.

Competing risk nomogram predicting cancer-specific mortality for endometrial cancer patients treated with hysterectomy

Background

The incidence of endometrial cancer has tended to increase in recent years. However, competing risk nomogram combining comprehensive factors for endometrial cancer patients treated with hysterectomy is still scarce. Therefore, we aimed to build a competing risk nomogram predicting cancer‐specific mortality for endometrial cancer patients treated with hysterectomy.

Methods

Patients diagnosed with endometrial cancer between 2010 and 2012 were abstracted from the Surveillance, Epidemiology, and End Results (SEER) database. Competing risk model was performed to select prognostic variables to build the competing risk nomogram to predict the cumulative 3‐ and 5‐year incidences of endometrial cancer‐specific mortality. Harrell's C‐index, receiver operating characteristic (ROC) curve, and calibration plot were used in the internal validation. And decision curve analysis was applied to evaluate clinical utility.

Results

A total of 10,447 patients were selected for analysis. The competing risk nomogram identified eight prognostic variables, including age at diagnosis, race, marital status at diagnosis, grade, histology, tumor size, FIGO stage, and number of regional nodes positive. The C‐index of the competing risk nomogram was 0.857 (95% confidence interval [CI]: 0.854–0.859), and the calibration plots were adequately fitted. When the threshold probabilities were between 1% and 57% for 3‐year prediction and between 2% and 67% for 5‐year prediction, the competing risk nomogram was of good clinical utility.

Conclusions

A competing risk nomogram for endometrial cancer patients treated with hysterectomy was successfully built and internally validated. It was an accurately predicted and clinical useful tool, which could play an important role in consulting and health care management of endometrial cancer patients.

Computer-Aided Segmentation and Machine Learning of Integrated Clinical and Diffusion-Weighted Imaging Parameters for Predicting Lymph Node Metastasis in Endometrial Cancer

Simple Summary

Computer-aided segmentation and machine learning added values of clinical parameters and diffusion-weighted imaging radiomics for predicting nodal metastasis in endometrial cancer, with a diagnostic performance superior to criteria based on lymph node size or apparent diffusion coefficient.

Abstract

Precise risk stratification in lymphadenectomy is important for patients with endometrial cancer (EC), to balance the therapeutic benefit against the operation-related morbidity and mortality. We aimed to investigate added values of computer-aided segmentation and machine learning based on clinical parameters and diffusion-weighted imaging radiomics for predicting lymph node (LN) metastasis in EC. This prospective observational study included 236 women with EC (mean age ± standard deviation, 51.2 ± 11.6 years) who underwent magnetic resonance (MR) imaging before surgery during July 2010–July 2018, randomly split into training (n = 165) and test sets (n = 71). A decision-tree model was constructed based on mean apparent diffusion coefficient (ADC) value of the tumor (cutoff, 1.1 × 10⁻³ mm²/s), skewness of the relative ADC value (cutoff, 1.2), short-axis diameter of LN (cutoff, 1.7 mm) and skewness ADC value of the LN (cutoff, 7.2 × 10⁻²), as well as tumor grade (1 vs. 2 and 3), and clinical tumor size (cutoff, 20 mm). The sensitivity and specificity of the model were 94% and 80% for the training set and 86%, 78% for the independent testing set, respectively. The areas under the receiver operating characteristics curve (AUCs) of the decision-tree was 0.85—significantly higher than the mean ADC model (AUC = 0.54) and LN short-axis diameter criteria (AUC = 0.62) (both p < 0.0001). We concluded that a combination of clinical and MR radiomics generates a prediction model for LN metastasis in EC, with diagnostic performance surpassing the conventional ADC and size criteria.

Is It Possible to Develop a Prediction Model for Lymphovascular Space Invasion in Endometrioid Endometrial Cancer?

OBJECTIVE

The purpose of this retrospective study was to define a risk index that would serve as a surrogate marker of lymphovascular space invasion (LVSI) in women with endometrioid endometrial cancer (EC).

MATERIALS AND METHODS

Final pathology reports of 498 women who underwent surgery with curative intent for endometrioid EC between January 2008 and June 2018 were retrospectively reviewed. Logistic regression was used to investigate clinicopathologic factors associated with positive LVSI status. Independent risk factors for LVSI were used to build a risk model and "risk of LVSI index" was defined as "tumor grade×primary tumor diameter×percentage of myometrium involved." The scores used in the "risk of LVSI index" were weighted according to the odds ratios assigned for each variable. The risk of LVSI index was noted for each patient. The diagnostic performance of the model was expressed as sensitivity, specificity, positive predictive value, negative predictive value, positive likelihood ratio, and negative likelihood ratio.

RESULTS

According to the "risk of LVSI index," presence of LVSI was correctly estimated in 89 of 104 LVSI-positive women at a cutoff of 161.0 (sensitivity 85.5%, specificity 79.4%, negative predictive value 95.4%, positive predictive value 52.3%, positive likelihood ratio 4.15, negative likelihood ratio 0.18). The area under curve of the receiver-operating characteristics was 0.90 (95% confidence interval, 0.87-0.93) at this cutoff.

CONCLUSIONS

It seems possible to predict the presence of LVSI in women with endometrioid EC when the "risk of LVSI index" is calculated. However, external validation of this model is warranted.

Prognostic significance of lymphovascular space invasion in patients with endometrioid endometrial cancer: a retrospective study from a single center

OBJECTIVE

This study aims to analyze factors associated with lymphovascular space invasion (LVSI) and evaluate the prognostic significance of LVSI in Chinese endometrioid endometrial cancer (EEC) patients.

METHODS

Five-hundred eighty-four EEC patients undergoing surgery in our center from 2006 to 2016 were selected for analysis. Univariate analysis and multivariate logistic regression were used to examine relevant factors of LVSI. To evaluate the prognostic role of LVSI, survival analyses were conducted. In survival analyses, both multivariate Cox regression and propensity score matching were used to control the confounders.

RESULTS

The incidence of LVSI was 12.16% (71/584). Diabetes history (p=0.021), lymph node metastasis (p=0.005), deep myometrial invasion (p<0.001) and negative PR expression (p=0.007) were independently associated with LVSI. Both Kaplan-Meier method and univariate Cox regressions showed LVSI negative and positive cases had similar tumor-specific survival (TSS) and disease-free survival (DFS). After adjusting for the influence of adjuvant therapy and other clinicopathological factors with multivariate Cox regressions, LVSI still could not bring additional survival risk to the patients (p=0.280 and p=0.650 for TSS and DFS, respectively). This result was verified by Kaplan-Meier survival analyses after propensity score matching (p=0.234 and p=0.765 for TSS and DFS, respectively).

CONCLUSION

LVSI does not significantly compromise the survival outcome of Chinese EEC patients.

An Externally Validated Nomogram for Predicting Lymph Node Metastasis of Presumed Stage I and II Endometrial Cancer

Background: Optimal management for endometrial cancer in patients with clinically negative lymph nodes is still under debate. Several prediction models for lymphatic dissemination of early-stage endometrial cancer have been developed. However, external validation is rare, and decision curve analysis has hardly been applied for these models.

Objective: To develop and validate a nomogram to predict lymph node metastasis of presumed stage I and II endometrial cancer.

Study Design: The prediction nomogram was developed by using multivariable logistic regression with data for 700 EC patients who underwent initial surgery from 2006 to 2017 at Peking University People's Hospital (training dataset), Beijing. External validation was performed in 727 eligible patients from Fudan University Shanghai Cancer Center (validation dataset), Shanghai.

Results: For the 700 women in the training dataset, the lymph node metastasis rate was 8.0% (56/700). Lymphovascular space invasion, histological grade, cervical stromal invasion, and myometrial invasion were independent prognostic factors in the training dataset. We generated a nomogram based on these pathological factors. To determine the clinical usefulness of our nomogram, we compared it with the Mayo criteria. For our nomogram, the area under the receiver operating characteristic curve (AUC) was 0.85 as compared with 0.63 for the Mayo criteria. In the validation dataset, the AUC was 0.78 as compared with 0.57 for the Mayo criteria. The nomogram was well-calibrated in both the training and validation datasets. At a 10% probability threshold, our nomogram decreased almost 29 unnecessary lymphadenectomies per 100 patients than the Mayo criteria without missing more metastatic disease.

Conclusion: We developed a nomogram to predict lymph node metastasis in patients with early-stage endometrial cancer in China. This prediction model may help clinicians in decision-making for patients with early-stage endometrial cancer, especially for the patient with incomplete surgery, reducing overtreatment, and medical costs.

Diagnostic Accuracy of Clinical Biomarkers for Preoperative Prediction of Lymph Node Metastasis in Endometrial Carcinoma: A Systematic Review and Meta-Analysis

BACKGROUND

In endometrial carcinoma (EC), preoperative classification is based on histopathological criteria, with only moderate diagnostic performance for the risk of lymph node metastasis (LNM). So far, existing molecular classification systems have not been evaluated for prediction of LNM. Optimized use of clinical biomarkers as recommended by international guidelines might be a first step to improve tailored treatment, awaiting future molecular biomarkers.

AIM

To determine the diagnostic accuracy of preoperative clinical biomarkers for the prediction of LNM in endometrial cancer.

METHODS

A systematic review was performed according to the Meta-analysis of Observational Studies in Epidemiology (MOOSE) guidelines. Studies identified in MEDLINE and EMBASE were selected by two independent reviewers. Included biomarkers were based on recommended guidelines (cancer antigen 125 [Ca-125], lymphadenopathy on magnetic resonance imaging, computed tomography, and ¹⁸F-fluorodeoxyglucose positron emission tomography/computed tomography [¹⁸FDG PET-CT]) or obtained by physical examination (body mass index, cervical cytology, blood cell counts). Pooled sensitivity, specificity, area under the curve (AUC), and likelihood ratios were calculated with bivariate random-effects meta-analysis. Likelihood ratios were classified into small (0.5-1.0 or 1-2.0), moderate (0.2-0.5 or 2.0-5.0) or large (0.1-0.2 or ≥ 5.0) impact.

RESULTS

Eighty-three studies, comprising 18,205 patients, were included. Elevated Ca-125 and thrombocytosis were associated with a moderate increase in risk of LNM; lymphadenopathy on imaging with a large increase. Normal Ca-125, cytology, and no lymphadenopathy on ¹⁸FDG PET-CT were associated with a moderate decrease. AUCs were above 0.75 for these biomarkers. Other biomarkers had an AUC <0.75 and incurred only small impact.

CONCLUSION

Ca-125, thrombocytosis, and imaging had a large and moderate impact on risk of LNM and could improve preoperative risk stratification.

IMPLICATIONS FOR PRACTICE

Routine lymphadenectomy in clinical early-stage endometrial carcinoma does not improve outcome and is associated with 15%-20% surgery-related morbidity, underlining the need for improved preoperative risk stratification. New molecular classification systems are emerging but have not yet been evaluated for the prediction of lymph node metastasis. This article provides a robust overview of diagnostic performance of all clinical biomarkers recommended by international guidelines. Based on these, at least measurement of cancer antigen 125 serum level, assessment of thrombocytosis, and imaging focused on lymphadenopathy should complement current preoperative risk stratification in order to better stratify these patients by risk.

L1 Cell Adhesion Molecule (L1CAM) expression in endometrioid endometrial carcinomas: A possible pre-operative surrogate of lymph vascular space invasion

Background

Risk stratification of endometrial carcinomas is primarily based on surgical staging that requires extensive retroperitoneal lymph node dissection. One of the most powerful predictor of lymph node involvement is the lymph vascular space invasion (LVSI). The objective of this study was to determine the potential of L1 Cell Adhesion Molecule (L1CAM) to predict LVSI and its association with other risk factors in endometrioid endometrial carcinomas.

Materials and methods

We studied 47 consecutive patients aged 37–88 (61.34±10.52). Twenty-three patients (48.9%) were submitted to complete surgical staging. Nine patients (19.1%) underwent surgical staging without para-aortic dissection. Seven (14.9%) were submitted to hysterectomy with no lymph node dissection. Eight patients (17.0%) only had the biopsy material for analysis. The 32 patients submitted to lymphadenectomy were staged according to the FIGO system and classified among the risk categories of the ESMO-ESGO-ESTRO guidelines. The following histological characteristics were analyzed: tumor size (mm), depth of myometrial infiltration, presence of microcystic, elongated, and fragmented (MELF) pattern of myoinvasion, and lymph vascular space invasion (LVSI). Immunohistochemical analyses of mismatch repair (MMR) proteins MLH1, MSH2, MSH6, and PMS2, p53, and L1CAM were performed in formalin-fixed paraffin embedded whole tumor tissue sections.

Results

LVSI was identified in 26/41 (63,4%) of the cases. L1CAM was positive in 8/47 (17%) cases, all of them positive for LVSI and within the high-risk category of ESMO-ESGO-ESTRO. L1CAM-positive cases were associated with high histological grade and p53 aberrant immunohistochemical profile. Besides, it showed a trend to larger tumors, greater depth of myometrial infiltration, and with a higher frequency of the MELF pattern of myoinvasion. LVSI was also associated with FIGO stage, tumor size, depth of myometrial infiltration, and tumor grade.

Conclusions

L1CAM is highly associated with LVSI and could be used as a pre-operative predictor of lymph node involvement in endometrioid endometrial carcinomas.