Objective:To evaluate preoperative serum ferritin (SF) in predicting the prognosis of patients with esophageal squamous cell carcinoma (ESCC).Methods:A retrospective analysis was conducted on 280 ESCC patients undergoing radical resection of esophageal cancer at Zhejiang Cancer Hospital from Jan 2008 to Dec 2013. Univariate and multivariate analysis were used to investigate the relationship between preoperative SF level and patients′ clinicopathologic characteristics. Kaplain-Meier method was used to analyze the relationship between preoperative SF level and the prognosis.Results:There were 183 cases (65.4%) with low SF level and 97cases (34.6%) with high SF level. The 1-, 3-and 5-year survival rates in low SF patients were 78.7%, 50.3%, 43.2% and that in high SF were 69.1%, 35.1%, 32.0%, respectively (χ 2=4.697, P=0.031). Univariate analysis demonstrated that intravascular cancer embolus, nerve infiltration and the level of preoperative SF were related to ESCC patients prognosis (all P<0.05). The multivariate analysis showed that carcinoma cell embolus ( OR=1.662, 95% CI: 1.239-2.229, P=0.001), nerve infiltration ( OR=1.823, 95% CI: 1.361-2.443, P<0.001) and the level of preoperative SF ( OR=1.504, 95% CI: 1.113-2.032, P=0.008) were independent risk factors for ESCC patients prognosis. Conclusion:Preoperative SF level closely associates with the prognosis of ESCC patients.

Objective To investigate the predictive value of combined PD-L1 expression,radiomics,and deep learning features for the efficacy of immunotherapy in patient with non-small cell lung cancer(NSCLC).Methods A total of 83 NSCLC patients who underwent immunotherapy were analyzed retrospectively.The volume of interest(VOI)was segmented on CT images,and features were extracted through the Pyradiomics and ResNet18 networks.The Radiomics score(Radscore)and deep learning score(Deepscore)were constructed based on the features after dimensionality reduction.Univariate and multivariate analyses were performed on the clinical parameters,Radscore,and Deepscore,and the independent predictive risk factors were selected to establish the clinical model,radiomics model,and combined prediction model,respectively.The receiver operating characteristic(ROC)curve was drawn,and the area under the curve(AUC)of the three models was calculated.Decision curve analysis(DCA)was used to compare the clinical practicability of the three models.Results Three radiomics features and three deep learning features were selected to calculate Radscore and Deepscore,respectively.PD-L1 expression,Radscore,and Deepscore were independent predictors of the efficacy of immunotherapy for NSCLC.The AUC of the combined prediction model in the training set and validation set were 0.885 and 0.877,respectively,which were higher than that of the clinical model(0.654 and 0.640),and the difference in AUC was statistically significant(P=0.006,0.029,respectively).The DCA showed that the combined prediction model achieved better clinical practicability at the threshold of 0-0.25 and 0.3-1.Conclusion The combined prediction model can better predict the efficacy of immunotherapy in NSCLC patients.

The aim of this study was to develop a diagnostic strategy for esophageal squamous cell carcinoma(ESCC) that combines plasma metabolomics with machine learning algorithms.Plasma-based untargeted metabolomics analysis was performed with samples derived from 88 ESCC patients and 52 healthy controls.The dataset was split into a training set and a test set.After identification of differential me-tabolites in training set,single-metabolite-based receiver operating characteristic (ROC) curves and multiple-metabolite-based machine learning models were used to distinguish between ESCC patients and healthy controls.Kaplan-Meier survival analysis and Cox proportional hazards regression analysis were performed to investigate the prognostic significance of the plasma metabolites.Finally,twelve differential plasma metabolites (six up-regulated and six down-regulated) were annotated.The pre-dictive performance of the six most prevalent diagnostic metabolites through the diagnostic models in the test set were as follows:arachidonic acid (accuracy:0.887),sebacic acid (accuracy:0.867),indoxyl sulfate (accuracy:0.850),phosphatidylcholine (PC) (14:0/0:0) (accuracy:0.825),deoxycholic acid(accuracy:0.773),and trimethylamine N-oxide (accuracy:0.653).The prediction accuracies of the ma-chine learning models in the test set were partial least-square (accuracy:0.947),random forest (accu-racy:0.947),gradient boosting machine (accuracy:0.960),and support vector machine (accuracy:0.980).Additionally,survival analysis demonstrated that acetoacetic acid was an unfavorable prognostic factor(hazard ratio (HR):1.752),while PC (14:0/0:0) (HR:0.577) was a favorable prognostic factor for ESCC.This study devised an innovative strategy for ESCC diagnosis by combining plasma metabolomics with machine learning algorithms and revealed its potential to become a novel screening test for ESCC.