A Variable-Clustering-Based Feature Selection to Improve Positive and Negative Discrimination of P53 Protein in Colorectal Cancer Patients.

P53 protein tumor suppressor gene plays a guiding role in the treatment and prognosis of colorectal cancer (CRC). This paper aimed at proposing a feature selection method based on variable clustering to improve positive and negative discrimination of P53 protein in CRC patients. In this approach, we cluster the preprocessed dataset with variables, and then find the features with the largest information value (IV) for each cluster to form a feature subset. We call this method as IV_Cluster. In the actual medical data test, compared with the information value feature selection method, the accuracy of the 10-fold cross-validation logistic regression model increased by 4.4%, 2.0%, and 5.8%, and Kappa values increased by 21.8%, 8.6%, and 22.4%, respectively, under 5, 10, and 15 feature sets.

A novel model for extrapleural cavity metastasis assessment in patients with lung cancer.

Aim: To investigate the clinical value of tumor markers in extrapleural tumor metastasis assessment of newly diagnosed lung cancer patients. Materials & methods: This study retrospectively analyzed 306 patients diagnosed with lung cancer accompanied by tumor metastasis. Patients were grouped into extrapleural tumor metastasis and intrapleural tumor metastasis. Seven serum tumor markers were included for analysis. Results: The area under curves of receiver operating characteristic curve based on binning decision tree algorithm were above 0.8 in both training and validation sets. A scorecard with a score below 3 suggested extrapleural tumor metastasis in newly diagnosed lung cancer patients. Conclusion: The serum tumor marker-derived model is a convenient and fast approach for extrapleural cavity metastasis assessment, which may provide positive implications in newly diagnosed lung cancer patients.

Development of a prediction model with serum tumor markers to assess tumor metastasis in lung cancer.

BACKGROUND: This study aimed to explore the possibility of serum tumor markers (TMs) combinations in assessing tumor metastasis in patients with lung cancer. METHODS: We performed a retrospective analysis of 541 patients diagnosed with lung cancer between January 2016 and December 2017 at the Pneumology Department of Dazhou Central Hospital. Serum carcinoembryonic antigen (CEA), carbohydrate antigen (CA)125, CA153, CA199, CA724, cytokeratin 19 fragment (CYFRA), and neuron-specific enolase (NSE) levels were quantified in each patient at the time of lung cancer diagnosis. Metastasis was confirmed by computed tomography, and/or positron emission tomography, and/or surgery or other necessary methods. Receiver operating characteristic (ROC) curves and calibration curves were used to evaluate the performance of the model. RESULTS: Of the 541 patients eligible for final analysis, 253 were detected with metastasis and 288 were detected without metastasis. Compared with those in nonmetastatic patients, the serum CEA, CA125, CA199, CA153, CYFRA, and NSE levels were notably higher in metastatic patients (P < .05). The ROC curve demonstrated that the CEA-CA125-CA199-CA153-CYFRA-NSE-CA724 combination based on the cut-off value had an optimal area under the curve and specificity in assessing tumor metastasis. The decision tree model is a convenient and valuable tool for guiding the appropriate application of our model to assess metastasis in lung cancer patients. CONCLUSIONS: Our study suggested that the nomogram of the regression model is valuable for assessing tumor metastasis in newly diagnosed lung cancer patients before traditional standard methods are used. These findings could aid in the evaluation of metastasis in the clinic.

The clinical value of carcinoembryonic antigen for tumor metastasis assessment in lung cancer.

BACKGROUND: Carcinoembryonic antigen (CEA) as a diagnostic or prognostic marker has been widely studied in patients with lung cancer. However, the relationship between serum CEA and tumor metastasis in lung cancer remains controversial. This study aimed to investigate the ability of serum CEA to assess tumor metastasis in lung cancer patients. METHODS: We performed a retrospective analysis of 238 patients diagnosed with lung cancer from January to December 2016 at pneumology department of Dazhou Central Hospital (Dazhou, China). Serum CEA levels were quantified in each patient at the time of diagnosis of lung cancer. Metastasis was confirmed by computed tomography (CT), and/or positron emission tomography (PET) and/or surgery or other necessary detecting methods. RESULTS: Of the 213 patients eligible for final analysis, 128 were diagnosed with metastasis and 85 were diagnosed without metastasis. Compared to non-metastatic patients, the serum CEA was markedly higher in patients with metastasis (p < 0.001), and the area under the curve (AUC) was 0.724 (95% CI [0.654-0.793]). Subsequent analyses regarding the number and location of tumor metastases showed that CEA also had clinical value for multiple metastases versus single metastasis (AUC = 0.780, 95% CI [0.699-0.862]) and distant metastasis versus non-distant metastasis (AUC = 0.815, 95% CI [0.733-0.897]). In addition, we found that tumor size, histology diagnosis, age and gender had no impact on the assessment performance of CEA. CONCLUSION: Our study suggested the serum CEA as a valuable marker for tumor metastases assessment in newly diagnosed lung cancer patients, which could have some implications in clinical application.