Hepatic Metastasis in Newly Diagnosed Esophageal Cancer: A Population-Based Study

Use machine learning to predict bone metastasis of esophageal cancer: A population-based study

Objective

The objective of this study is to develop a machine learning (ML)-based predictive model for bone metastasis (BM) in esophageal cancer (EC) patients.

Methods

This study utilized data from the Surveillance, Epidemiology, and End Results database spanning 2010 to 2020 to analyze EC patients. A total of 21,032 confirmed cases of EC were included in the study. Through univariate and multivariate logistic regression (LR) analysis, 10 indicators associated with the risk of BM were identified. These factors were incorporated into seven different ML classifiers to establish predictive models. The performance of these models was assessed and compared using various metrics including the area under the receiver operating characteristic curve (AUC), accuracy, sensitivity, specificity, F-score, precision, and decision curve analysis.

Results

Factors such as age, gender, histological type, T stage, N stage, surgical intervention, chemotherapy, and the presence of brain, lung, and liver metastases were identified as independent risk factors for BM in EC patients. Among the seven models developed, the ML model based on LR algorithm demonstrated excellent performance in the internal validation set. The AUC, accuracy, sensitivity, and specificity of this model were 0.831, 0.721, 0.787, and 0.717, respectively.

Conclusion

We have successfully developed an online calculator utilizing a LR model to assist clinicians in accurately assessing the risk of BM in patients with EC. This tool demonstrates high accuracy and specificity, thereby enhancing the development of personalized treatment plans.

Outcomes and Prognostic Factors of Patients with Unresectable or Metastatic Esophageal Squamous Cell Carcinoma Undergoing Immunotherapy- Versus Chemotherapy-Based Regimens: Systematic Review and Pooled Analyses

OBJECTIVE

Immunotherapy-based regimens (IMT) versus cytotoxic chemotherapy (CHT) improved overall survival (OS) of patients with unresectable or metastatic esophageal squamous cell carcinoma (mESCC), but the role of prognostic variables is unclear. The study aims to explore the interaction of prognostic factors with survival after IMT or CHT.

METHODS

A systematic review was performed to select trials comparing IMT and CHT regimens in mESCC patients. A meta-analysis of upfront IMT + CHT vs. CHT trials evaluated the overall effect size and heterogeneity between studies. In view of the expected differences between chemotherapy and immunotherapy on the survival curve, to better explore the effect of any prognostic variables on OS, before and after progression, the treatment arms were evaluated as independent cohorts, and ten baseline variables were extracted and assessed by linear regression.

RESULTS

Fourteen trials were identified. Seven studies compared upfront CHT + IMT vs. CHT documenting longer OS for CHT + IMT (HR 0.69, CI 0.65-0.72), without heterogeneity (Q = 1.43, p value = 0.968) or differences in the most represented subgroups. Twenty-nine study cohorts were selected from the 14 trials. Median OS and PPS, but not PFS, were significantly increased after IMT compared with CHT. The analysis of baseline variables after CHT documented a favorable prognostic effect for advanced age (β = 0.768, p value = 0.016), involvement of 0-1 metastasis sites (β = 0.943, p value = 0.005), and absence of previous radiation therapy (β =  - 0.939, p value = 0.006), while none of them influenced prognosis after IMT.

CONCLUSION

The introduction of upfront IMT prolonged mESCC patients OS, mostly improving the outcomes of young patients, with multiple metastasis sites and without previous radiotherapy.

Prediction of Early Mortality in Esophageal Cancer Patients with Liver Metastasis Using Machine Learning Approaches

Patients with esophageal cancer liver metastasis face a high risk of early mortality, making accurate prediction crucial for guiding clinical decisions. However, effective predictive tools are currently limited. In this study, we used clinicopathological data from 1897 patients diagnosed with esophageal cancer liver metastasis between 2010 and 2020, which were sourced from the SEER database. Prognostic factors were identified using univariate and multivariate logistic regression, and seven machine learning models, including extreme gradient boosting (XGBoost) and support vector machine (SVM), were developed to predict early mortality. The models were evaluated using Receiver Operating Characteristic (ROC) curves, calibration curves, decision curve analysis (DCA), and F1 scores. Results showed that 40% of patients experienced all-cause early mortality and 38% had cancer-specific early mortality. Key predictors of early mortality included age, location, chemotherapy, and lung metastasis. Among the models, XGBoost performed best in predicting all-cause early mortality, while SVM excelled in predicting cancer-specific early mortality. These findings demonstrate that machine learning models, particularly XGBoost and SVM, can serve as valuable tools for predicting early mortality in patients with esophageal cancer liver metastasis, aiding clinical decision making.

Measurement of metabolite levels and treatment-induced changes in hepatic metastases of gastro-esophageal cancer using 7-T phosphorus magnetic resonance spectroscopic imaging

Methods for early treatment response evaluation to systemic therapy of liver metastases are lacking. Tumor tissue often exhibits an increased ratio of phosphomonoesters to phosphodiesters (PME/PDE), which can be noninvasively measured by phosphorus magnetic resonance spectroscopy (31P MRS), and may be a marker for early therapy response assessment in liver metastases. However, with commonly used 31P surface coils for liver 31P MRS, the liver is not fully covered, and metastases may be missed. The objective of this study was to demonstrate the feasibility of 31P MRS imaging (31P MRSI) with full liver coverage to assess 31P metabolite levels and chemotherapy-induced changes in liver metastases of gastro-esophageal cancer, using a 31P whole-body birdcage transmit coil in combination with a 31P body receive array at 7 T. 3D 31P MRSI data were acquired in two patients with hepatic metastases of esophageal cancer, before the start of chemotherapy and after 2 (and 9 in patient 2) weeks of chemotherapy. 3D 31P MRSI acquisitions were performed using an integrated 31P whole-body transmit coil in combination with a 16-channel body receive array at 7 T, with a field of view covering the full abdomen and a nominal voxel size of 20-mm isotropic. From the 31P MRSI data, 12 31P metabolite signals were quantified. Prior to chemotherapy initiation, both PMEs, that is, phosphocholine (PC) and phosphoethanolamine (PE), were significantly higher in all metastases compared with the levels previously determined in the liver of healthy volunteers. After 2 weeks of chemotherapy, PC and PE levels remained high or even increased further, resulting in increased PME/PDE ratios compared with healthy liver tissue, in correspondence with the clinical assessment of progressive disease after 2 months of chemotherapy. The suggested approach may present a viable tool for early therapy (non)response assessment of tumor metabolism in patients with liver metastases.

Development and validation of a prognostic model for esophageal cancer patients with liver metastasis: a cohort study based on surveillance, epidemiology, and end results database

PURPOSE

Our objective is to examine the independent prognostic risk factors for patients with Esophageal Cancer with Liver Metastasis (ECLM) and to develop a predictive model.

METHODS

In this study, clinical data were obtained from the Surveillance, Epidemiology, and End Results (SEER) database. Cox regression analysis was employed to identify independent prognostic factors and construct nomograms based on the results of multivariate regression. The predictive performance of the nomograms was assessed using several methods, including the consistency index (C-index), calibration curve, time-dependent receiver-operating characteristic curve (ROC), and decision curve analysis (DCA). Additionally, Kaplan-Meier survival curves were generated to demonstrate the variation in overall survival between groups.

RESULTS

A total of 1163 ECLM patients were included in the study. Multivariate Cox analysis revealed that age, tumor differentiation grade, bone metastasis, therapy, and income were independently associated with overall survival (OS) in the training set. Subsequently, a prognostic nomogram was constructed based on these independent predictors. The C-index values were 0.739 and 0.715 in the training and validation sets, respectively. The area under the curve (AUC) values at 0.5, 1, and 2 years were all higher than 0.700. Calibration curves indicated that the nomogram accurately predicted OS. Decision curve analysis (DCA) showed moderately positive net benefits. Kaplan-Meier survival curves demonstrated significant differences in survival between high- and low-risk groups, which were divided based on the nomogram risk score.

CONCLUSIONS

The nomogram we developed for ECLM patients has demonstrated good predictive capability, allowing clinicians to accurately evaluate patient prognosis and identify those at high risk, thereby facilitating the development of personalized treatment plans.

Predictive models for the risk and prognosis of bone metastasis in patients with newly-diagnosed esophageal cancer: A retrospective cohort study

Background

Esophageal cancer (EC) is a common malignant tumor worldwide, and patients with both EC and bone metastasis (BM) have a poor prognosis. We aimed to determine the risk and prognostic factors for BM in patients with newly diagnosed EC and to conduct two nomograms to predict the probability of BM and overall survival after BM.

Methods

Data from patients with EC from 2010 to 2015 were reviewed in the Surveillance, Epidemiology, and End Results (SEER) database. We divided participants into training and validation cohorts using univariate and multivariate logistic regression analyses and Cox regression models to explore the risk and prognostic factors of BM, respectively. Moreover, two nomograms were developed for predicting the risk and prognosis of BM in patients with EC. Then we used receiver operating characteristic curves, decision curve analysis, and calibration curves to evaluate the nomogram models. The overall survival of patients with EC and BM was analyzed using the Kaplan-Meier method.

Results

A total of 10,730 patients with EC were involved, 735 of whom had BM at the time of diagnosis. Histologic type, sex, age, N stage, primary site, liver, lung, and brain metastases, and tumor differentiation grade were identified as independent BM risk factors. Histological type, chemotherapy, brain, liver, and lung metastases were identified as prognostic risk factors for patients with EC and BM. We developed diagnostic and prognostic nomograms according to the results. Receiver operating characteristic curves, calibration, and Kaplan-Meier curves, and decision curve analysis all indicated that both nomograms had great clinical predictive ability and good clinical application potential.

Conclusions

Two novel nomograms were constructed to predict the risk and prognosis of BM in patients with EC. These prediction models can effectively assist clinicians in clinical decision-making based on their good accuracy and reliability.

Survival Prediction Model for Patients with Esophageal Squamous Cell Carcinoma Based on the Parameter-Optimized Deep Belief Network Using the Improved Archimedes Optimization Algorithm

Esophageal squamous cell carcinoma (ESCC) is one of the highest incidence and mortality cancers in the world. An effective survival prediction model can improve the quality of patients' survival. Therefore, a parameter-optimized deep belief network based on the improved Archimedes optimization algorithm is proposed in this paper for the survival prediction of patients with ESCC. Firstly, a combination of features significantly associated with the survival of patients is found by the minimum redundancy and maximum relevancy (MRMR) algorithm. Secondly, a DBN network is introduced to make predictions for survival of patients. Aiming at the problem that the deep belief network model is affected by parameters in the construction process, this paper uses the Archimedes optimization algorithm to optimize the learning rate α and batch size β of DBN. In order to overcome the problem that AOA is prone to fall into local optimum and low search accuracy, an improved Archimedes optimization algorithm (IAOA) is proposed. On this basis, a survival prediction model for patients with ESCC is constructed. Finally, accuracy comparison tests are carried out on IAOA-DBN, AOA-DBN, SSA-DBN, PSO-DBN, BES-DBN, IAOA-SVM, and IAOA-BPNN models. The results show that the IAOA-DBN model can effectively predict the five-year survival rate of patients and provide a reference for the clinical judgment of patients with ESCC.

Construction and Validation of a m7G-Related Gene-Based Prognostic Model for Gastric Cancer

Background

Gastric cancer (GC) is one of the most common malignant tumors of the digestive system. Chinese cases of GC account for about 40% of the global rate, with approximately 1.66 million people succumbing to the disease each year. Despite the progress made in the treatment of GC, most patients are diagnosed at an advanced stage due to the lack of obvious clinical symptoms in the early stages of GC, and their prognosis is still very poor. The m7G modification is one of the most common forms of base modification in post-transcriptional regulation, and it is widely distributed in the 5′ cap region of tRNA, rRNA, and eukaryotic mRNA.

Methods

RNA sequencing data of GC were downloaded from The Cancer Genome Atlas. The differentially expressed m7G-related genes in normal and tumour tissues were determined, and the expression and prognostic value of m7G-related genes were systematically analysed. We then built models using the selected m7G-related genes with the help of machine learning methods.The model was then validated for prognostic value by combining the receiver operating characteristic curve (ROC) and forest plots. The model was then validated on an external dataset. Finally, quantitative real-time PCR (qPCR) was performed to detect gene expression levels in clinical gastric cancer and paraneoplastic tissue.

Results

The model is able to determine the prognosis of GC samples quantitatively and accurately. The ROC analysis of model has an AUC of 0.761 and 0.714 for the 3-year overall survival (OS) in the training and validation sets, respectively. We determined a correlation between risk scores and immune cell infiltration and concluded that immune cell infiltration affects the prognosis of GC patients. NUDT10, METTL1, NUDT4, GEMIN5, EIF4E1B, and DCPS were identified as prognostic hub genes and potential therapeutic agents were identified based on these genes.

Conclusion

The m7G-related gene-based prognostic model showed good prognostic discrimination. Understanding how m7G modification affect the infiltration of the tumor microenvironment (TME) cells will enable us to better understand the TME’s anti-tumor immune response, and hopefully guide more effective immunotherapy methods.

Prognostic Nomogram for Predicting Long-Term Overall Survival of Esophageal Cancer Patients Receiving Neoadjuvant Chemoradiotherapy Plus Surgery: A Population-Based Study

Objective

Neoadjuvant chemoradiotherapy (nCRT) plays an important role in patients with locally advanced esophageal cancer (EC). We aim to determine the prognostic risk factors and establish a reliable nomogram to predict overall survival (OS) based on SEER population.

Methods

Patients with EC coded by 04–15 in the SEER database were included. The data were divided into training group and verification group (7:3). The Cox proportional-risk model was evaluated by using the working characteristic curve (receiver operating characteristic curve, ROC) and the area under the curve (AUC), and a nomogram was constructed. The calibration curve was used to measure the consistency between the predicted and the actual results. Decision curve analysis (DCA) was used to evaluate its clinical value. The best cut-off value of nomogram score in OS was determined by using X-tile software, and the patients were divided into low-risk, medium-risk, and high-risk groups.

Results

A total of 2,209 EC patients who underwent nCRT were included in further analysis, including 1,549 in the training cohort and 660 in the validation group. By Cox analysis, sex, marital status, T stage, N stage, M stage, and pathological grade were identified as risk factors. A nomogram survival prediction model was established to predict the 36-, 60-, and 84-month survival. The ROC curve and AUC showed that the model had good discrimination ability. The correction curve was in good agreement with the prediction results. DCA further proved the effective clinical value of the nomogram model. The results of X-tile analysis showed that the long-term prognosis of patients in the low-risk subgroup was better in the training cohort and the validation cohort (p < 0.001).

Conclusion

This study established an easy-to-use nomogram risk prediction model consisting of independent prognostic factors in EC patients receiving nCRT, helping to stratify risk, identify high-risk patients, and provide personalized treatment options.

Models for Predicting Early Death in Patients With Stage IV Esophageal Cancer: A Surveillance, Epidemiology, and End Results-Based Cohort Study

Background

Despite enormous progress in the stage IV esophageal cancer (EC) treatment, some patients experience early death after diagnosis. This study aimed to identify the early death risk factors and construct models for predicting early death in stage IV EC patients.

Methods

Stage IV EC patients diagnosed between 2010 and 2015 in the Surveillance, Epidemiology, and End Results (SEER) database were selected. Early death was defined as death within 3 months of diagnosis, with or without therapy. Early death risk factors were identified using logistic regression analyses and further used to construct predictive models. The concordance index (C-index), calibration curves, and decision curve analyses (DCA) were used to assess model performance.

Results

Out of 4411 patients enrolled, 1779 died within 3 months. Histologic grade, therapy, the status of the bone, liver, brain and lung metastasis, marriage, and insurance were independent factors for early death in stage IV EC patients. Histologic grade and the status of the bone and liver metastases were independent factors for early death in both chemoradiotherapy and untreated groups. Based on these variables, predictive models were constructed. The C-index was .613 (95% confidence interval (CI), [.573–.653]) and .635 (95% CI, [.596–.674]) in the chemoradiotherapy and untreated groups, respectively, while calibration curves and DCA showed moderate performance.

Conclusions

More than 40% of stage IV EC patients suffered from an early death. The models could help clinicians discriminate between low and high risks of early death and strategize individually-tailed therapeutic interventions in stage IV EC patients.