Novel model for cancer-specific survival of patients with pulmonary sarcomatoid carcinoma: A population-based analysis and external validation.

BACKGROUND/OBJECTIVE: We aimed to develop a comprehensive and effective nomogram for predicting cancer-specific survival (CSS) in patients with pulmonary sarcomatoid carcinoma (PSC). METHODS: Data for patients diagnosed with PSC between 2004 and 2018 from the Surveillance, Epidemiology, and End Results database were retrospectively collected and randomly divided into training and internal validation sets. We then retrospectively recruited patients diagnosed with PSC to construct an external validation cohort from the Southwest Hospital. A prognostic nomogram for CSS was established using independent prognostic factors that were screened from the multivariate Cox regression analysis. The performance of the nomogram was evaluated using area under the receiver operating characteristic (ROC) curves, Harrell's concordance index (C-index), calibration diagrams, and decision curve analysis (DCA). The clinical value of the nomogram and tumor, nodes, and metastases (TNM) staging system was compared using the C-index and net reclassification index (NRI). RESULTS: Overall, 1356 patients with PSC were enrolled, including 876, 377, and 103 in the training, internal validation, and external validation sets, respectively. The C-index and ROC curves, calibration, and DCA demonstrated satisfactory nomogram performance for CSS in patients with PSC. In addition, the C-index and NRI of the nomogram suggested a significantly higher nomogram value than that of the TNM staging system. Subsequently, a web-based predictor was developed to help clinicians obtain this model easily. CONCLUSIONS: The prognostic nomogram developed in this study can conveniently and precisely estimate the prognosis of patients with PSC and individualize treatment, thereby assisting clinicians in their shared decision-making with patients.

Study on the application of percutaneous closed pleural brushing combined with cell block technique in the diagnosis of malignant pleural effusion.

INTRODUCTION: This study was to investigate the diagnostic value of percutaneous closed pleural brushing (CPBR) followed by cell block technique for malignant pleural effusion (MPE) and the predictive efficacy of pleural fluid carcinoembryonic antigen (CEA) for epidermal growth factor receptor (EGFR) mutations in lung adenocarcinoma patients with MPE. METHODS: All patients underwent closed pleural biopsy (CPB) and CPBR followed by cell block examination. MPE-positive diagnostic rates between the two methods were compared. Univariate and multivariate analyses were performed to determine factors influencing the EGFR mutations. Receiver operating characteristic (ROC) curve was used to analyze the predictive efficacy of pleural fluid CEA for EGFR mutations. RESULTS: The cumulative positive diagnostic rates for MPE after single and twice CPBR followed by cell block examination were 80.5% and 89.0%, higher than CPB (45.7%, 54.3%) (P < 0.001). Univariate analysis showed that EGFR mutation was associated with pleural fluid and serum CEA (P < 0.05). Multivariate analysis showed that pleural fluid CEA was an independent risk factor for predicting EGFR mutation (P < 0.001). The area under the curve (AUC) of pleural fluid CEA for EGFR mutation prediction was 0.774, higher than serum CEA (P = 0.043), but no difference with the combined test (P > 0.05). CONCLUSION: Compared with CPB, CPBR followed by the cell block technique can significantly increase the positive diagnostic rate of suspected MPE. CEA testing of pleural fluid after CPBR has a high predictive efficacy for EGFR mutation in lung adenocarcinoma patients with MPE, implying pleural fluid extracted for cell block after CPBR may be an ideal specimen for genetic testing.

Nomogram to predict gas-related complications during transoral endoscopic resection of upper gastrointestinal submucosal lesions.

BACKGROUND: Gas-related complications present a potential risk during transoral endoscopic resection of upper gastrointestinal submucosal lesions. Therefore, the identification of risk factors associated with these complications is essential. AIM: To develop a nomogram to predict risk of gas-related complications following transoral endoscopic resection of the upper gastrointestinal submucosal lesions. METHODS: We collected patient data from the First Affiliated Hospital of the Army Medical University. Patients were randomly allocated to training and validation cohorts. Risk factors for gas-related complications were identified in the training cohort using univariate and multivariate analyses. We then constructed a nomogram and evaluated its predictive performance based on the area under the curve, decision curve analysis, and Hosmer-Lemeshow tests. RESULTS: Gas-related complications developed in 39 of 353 patients who underwent transoral endoscopy at our institution. Diabetes, lesion origin, surgical resection method, and surgical duration were incorporated into the final nomogram. The predictive capability of the nomogram was excellent, with area under the curve values of 0.841 and 0.906 for the training and validation cohorts, respectively. CONCLUSION: The ability of our four-variable nomogram to efficiently predict gas-related complications during transoral endoscopic resection enhanced postoperative assessments and surgical outcomes.

Prognostic nomogram and epidemiological analysis for lung atypical carcinoid: A SEER database and external validation study.

PURPOSE: Our study aims to delineate the epidemiological distribution of pulmonary carcinoids, including atypical carcinoid (AC) and typical carcinoid (TC), identify independent prognostic factors, develop an integrative nomogram and examine the effects of various surgical modalities on atypical carcinoid-specific survival (ACSS). METHODS: Joinpoint regression model and age-group distribution diagram were applied to determine the epidemiological trend of the pulmonary carcinoids. Univariate and least absolute shrinkage and selection operator (LASSO)-based Cox regression models were used to identify independent factors, and a nomogram and web-based predictor were developed to evaluate prognosis of AC patients individually. We performed Kaplan-Meier survival analyses to compare the scope of various surgical interventions, with and without G-computation adjustment, utilising restricted mean survival time (RMST) to assess survival disparities. RESULTS: A total of 1132 patients were recruited from the Surveillance, Epidemiology, and End Results database (SEER) and a separate medical centre in China. The mean age of AC patients was 63.4 years and a smoking history was identified in 79.8% of AC patients. Joinpoint analysis shows rising annual rates of new AC and carcinoid cases among lung cancers. Both the proportion of pulmonary TC and AC within the total lung cancer population exhibits an L-shaped trend across successive age groups. The nomogram predicted 1, 3 and 5 years of AC with excellent accuracy and discrimination. Kaplan-Meier survival analyses, conducted both pre- and post-adjustment, demonstrated that sublobar resection's survival outcomes were not inferior to those of lobectomy in patients with stage I-II and stage III disease. CONCLUSION: This study is the first to reveal epidemiological trends in pulmonary carcinoids over the past decade and across various age cohorts. For patients with early-stage AC, sublobar resection may be a viable surgical recommendation. The established nomogram and web-based calculator demonstrated decent accuracy and practicality.

Prediction of lung papillary adenocarcinoma-specific survival using ensemble machine learning models.

Accurate prognostic prediction is crucial for treatment decision-making in lung papillary adenocarcinoma (LPADC). The aim of this study was to predict cancer-specific survival in LPADC using ensemble machine learning and classical Cox regression models. Moreover, models were evaluated to provide recommendations based on quantitative data for personalized treatment of LPADC. Data of patients diagnosed with LPADC (2004-2018) were extracted from the Surveillance, Epidemiology, and End Results database. The set of samples was randomly divided into the training and validation sets at a ratio of 7:3. Three ensemble models were selected, namely gradient boosting survival (GBS), random survival forest (RSF), and extra survival trees (EST). In addition, Cox proportional hazards (CoxPH) regression was used to construct the prognostic models. The Harrell's concordance index (C-index), integrated Brier score (IBS), and area under the time-dependent receiver operating characteristic curve (time-dependent AUC) were used to evaluate the performance of the predictive models. A user-friendly web access panel was provided to easily evaluate the model for the prediction of survival and treatment recommendations. A total of 3615 patients were randomly divided into the training and validation cohorts (n = 2530 and 1085, respectively). The extra survival trees, RSF, GBS, and CoxPH models showed good discriminative ability and calibration in both the training and validation cohorts (mean of time-dependent AUC: > 0.84 and > 0.82; C-index: > 0.79 and > 0.77; IBS: < 0.16 and < 0.17, respectively). The RSF and GBS models were more consistent than the CoxPH model in predicting long-term survival. We implemented the developed models as web applications for deployment into clinical practice (accessible through https://shinyshine-820-lpaprediction-model-z3ubbu.streamlit.app/ ). All four prognostic models showed good discriminative ability and calibration. The RSF and GBS models exhibited the highest effectiveness among all models in predicting the long-term cancer-specific survival of patients with LPADC. This approach may facilitate the development of personalized treatment plans and prediction of prognosis for LPADC.

Individual risk and prognostic value prediction by machine learning for distant metastasis in pulmonary sarcomatoid carcinoma: a large cohort study based on the SEER database and the Chinese population.

Background: This study aimed to develop diagnostic and prognostic models for patients with pulmonary sarcomatoid carcinoma (PSC) and distant metastasis (DM). Methods: Patients from the Surveillance, Epidemiology, and End Results (SEER) database were divided into a training set and internal test set at a ratio of 7 to 3, while those from the Chinese hospital were assigned to the external test set, to develop the diagnostic model for DM. Univariate logistic regression was employed in the training set to screen for DM-related risk factors, which were included into six machine learning (ML) models. Furthermore, patients from the SEER database were randomly divided into a training set and validation set at a ratio of 7 to 3 to develop the prognostic model which predicts survival of patients PSC with DM. Univariate and multivariate Cox regression analyses have also been performed in the training set to identify independent factors, and a prognostic nomogram for cancer-specific survival (CSS) for PSC patients with DM. Results: For the diagnostic model for DM, 589 patients with PSC in the training set, 255 patients in the internal and 94 patients in the external test set were eventually enrolled. The extreme gradient boosting (XGB) algorithm performed best on the external test set with an area under the curve (AUC) of 0.821. For the prognostic model, 270 PSC patients with DM in the training and 117 patients in the test set were enrolled. The nomogram displayed precise accuracy with AUC of 0.803 for 3-month CSS and 0.869 for 6-month CSS in the test set. Conclusion: The ML model accurately identified individuals at high risk for DM who needed more careful follow-up, including appropriate preventative therapeutic strategies. The prognostic nomogram accurately predicted CSS in PSC patients with DM.

Current status, trends, and predictions in the burden of silicosis in 204 countries and territories from 1990 to 2019.

Background: Silicosis, a severe lung disease caused by inhaling silica dust, predominantly affects workers in industries such as mining and construction, leading to a significant global public health challenge. The purpose of this study is to analyze the current disease burden of silicosis and to predict the development trend of silicosis in the future the world by extracting data from the GBD database. Methods: We extracted and analyzed silicosis prevalence, incidence, mortality, and disability-adjusted life years (DALYs) data from the Global Burden of Disease 2019 program for 204 countries and territories from 1990 to 2019. The association between the Sociodemographic Index (SDI) and the burden of age-standardized rates (ASRs) of DALYs has been examined at the regional level. Jointpoint regression analysis has been also performed to evaluate global burden trends of silicosis from 1990 to 2019. Furthermore, Nordpred age-period-cohort analysis has also been projected to predict future the burden of silicosis from 2019 to 2044. Results: In 2019, global ASRs for silicosis prevalence, incidence, mortality, and DALYs were 5.383, 1.650, 0.161, and 7.872%, respectively which are lower than that in 1990. The populations of 45-59 age group were more susceptible to silicosis, while those aged 80 or above suffered from higher mortality and DALY risks. In 2019, the most impacted nations by the burden of silicosis included China, the Democratic People's Republic of Korea, and Chile. From 1990 to 2019, most regions observed a declining burden of silicosis. An "M" shaped association between SDI and ASRs of DALYs for silicosis was observed from 1990 to 2019. The age-period-cohort analysis forecasted a decreasing trend of the burden of silicosis from 2019 to 2044. Conclusion: Despite the overall decline in the global silicosis burden from 1990 to 2019, some regions witnessed a notable burden of this disease, emphasizing the importance of targeted interventions. Our results may provide a reference for the subsequent development of appropriate management strategies.

Development and External Validation of Machine Learning-Based Models for Predicting Lung Metastasis in Kidney Cancer: A Large Population-Based Study.

The accuracy of indices widely used to evaluate lung metastasis (LM) in patients with kidney cancer (KC) is insufficient. Therefore, we aimed at developing a model to estimate the risk of developing LM in KC based on a large population size and machine learning algorithms. Demographic and clinicopathologic variables of patients with KC diagnosed between 2004 and 2017 were retrospectively analyzed. We performed a univariate logistic regression analysis to identify risk factors for LM in patients with KC. Six machine learning (ML) classifiers were established and tuned using the ten-fold cross-validation method. External validation was performed using clinicopathologic information from 492 patients from the Southwest Hospital, Chongqing, China. Algorithm performance was estimated by analyzing the area under the receiver operating characteristic curve (AUC), accuracy, sensitivity, specificity, precision, recall, F1 score, clinical decision analysis (DCA), and clinical utility curve (CUC). A total of 52,714 eligible patients diagnosed with KC were enrolled, of whom 2,618 developed LM. Variables of age, sex, race, T stage, N stage, tumor size, histology, and grade were identified as important for the prediction of LM. The extreme gradient boosting (XGB) algorithm performed better than other models in both the internal validation (AUC: 0.913, sensitivity: 0.873, specificity: 0.809, and F1 score: 0.325) and the external validation (AUC: 0.904, sensitivity: 0.750, specificity: 0.878, and F1 score: 0.364). This study established a predictive model for LM in KC patients based on ML algorithms which showed high accuracy and applicative value. A web-based predictor was built using the XGB model to help clinicians make more rational and personalized decisions.

Using machine learning to predict lymph node metastasis in patients with renal cell carcinoma: A population-based study.

Background: Lymph node (LN) metastasis is strongly associated with distant metastasis of renal cell carcinoma (RCC) and indicates an adverse prognosis. Accurate LN-status prediction is essential for individualized treatment of patients with RCC and to help physicians make appropriate surgical decisions. Thus, a prediction model to assess the hazard index of LN metastasis in patients with RCC is needed. Methods: Partial data were extracted from the Surveillance, Epidemiology, and End Results (SEER) database. Data of 492 individuals with RCC, collected from the Southwest Hospital in Chongqing, China, were used for external validation. Eight indicators of risk of LN metastasis were screened out. Six machine learning (ML) classifiers were established and tuned, focused on predicting LN metastasis in patients with RCC. The models were integrated with big data analytics and ML algorithms. Based on the optimal model, we developed an online risk calculator and plotted overall survival using Kaplan-Meier analysis. Results: The extreme gradient-boosting (XGB) model was superior to the other models in both internal and external trials. The area under the curve, accuracy, sensitivity, and specificity were 0.930, 0.857, 0.856, and 0.873, respectively, in the internal test and 0.958, 0.935, 0.769, and 0.944, respectively, in the external test. These parameters show that XGB has an excellent ability for clinical application. The survival analysis showed that patients with predicted N1 tumors had significantly shorter survival (p < 0.0001). Conclusion: Our study shows that integrating ML algorithms and clinical data can effectively predict LN metastasis in patients with confirmed RCC. Subsequently, a freely available online calculator (https://xinglinyi.shinyapps.io/20221004-app/) was built, based on the XGB model.