FedDUS: Lung tumor segmentation on CT images through federated semi-supervised with dynamic update strategy.

BACKGROUND AND OBJECTIVE: Lung tumor annotation is a key upstream task for further diagnosis and prognosis. Although deep learning techniques have promoted automation of lung tumor segmentation, there remain challenges impeding its application in clinical practice, such as a lack of prior annotation for model training and data-sharing among centers. METHODS: In this paper, we use data from six centers to design a novel federated semi-supervised learning (FSSL) framework with dynamic model aggregation and improve segmentation performance for lung tumors. To be specific, we propose a dynamically updated algorithm to deal with model parameter aggregation in FSSL, which takes advantage of both the quality and quantity of client data. Moreover, to increase the accessibility of data in the federated learning (FL) network, we explore the FAIR data principle while the previous federated methods never involve. RESULT: The experimental results show that the segmentation performance of our model in six centers is 0.9348, 0.8436, 0.8328, 0.7776, 0.8870 and 0.8460 respectively, which is superior to traditional deep learning methods and recent federated semi-supervised learning methods. CONCLUSION: The experimental results demonstrate that our method is superior to the existing FSSL methods. In addition, our proposed dynamic update strategy effectively utilizes the quality and quantity information of client data and shows efficiency in lung tumor segmentation. The source code is released on (https://github.com/GDPHMediaLab/FedDUS).

A contrast-enhanced CT radiomics-based model to identify candidates for deintensified chemoradiotherapy in locoregionally advanced nasopharyngeal carcinoma patients.

OBJECTIVES: To develop a contrast-enhanced CT (CECT) radiomics-based model to identify locoregionally advanced nasopharyngeal carcinoma (LA-NPC) patients who would benefit from deintensified chemoradiotherapy. METHODS: LA-NPC patients who received low-dose concurrent cisplatin therapy (cumulative: 150 mg/m2), were randomly divided into training and validation groups. 107 radiomics features based on the primary nasopharyngeal tumor were extracted from each pre-treatment CECT scan. Through Cox regression analysis, a radiomics model and patients' corresponding radiomics scores were created with predictive independent radiomics features. T stage (T) and radiomics score (R) were compared as predictive factors. Combining the N stage (N), a clinical model (T + N), and a substitution model (R + N) were constructed. RESULTS: Training and validation groups consisted of 66 and 33 patients, respectively. Three significant independent radiomics features (flatness, mean, and gray level non-uniformity in gray level dependence matrix (GLDM-GLN)) were found. The radiomics score showed better predictive ability than the T stage (concordance index (C-index): 0.67 vs. 0.61, AUC: 0.75 vs. 0.60). The R + N model had better predictive performance and more effective risk stratification than the T + N model (C-index: 0.77 vs. 0.68, AUC: 0.80 vs. 0.70). The R + N model identified a low-risk group as deintensified chemoradiotherapy candidates in which no patient developed progression within 3 years, with 5-year progression-free survival (PFS) and overall survival (OS) both 90.7% (hazard ratio (HR) = 4.132, p = 0.018). CONCLUSION: Our radiomics-based model combining radiomics score and N stage can identify specific LA-NPC candidates for whom de-escalation therapy can be performed without compromising therapeutic efficacy. CLINICAL RELEVANCE STATEMENT: Our study shows that the radiomics-based model (R + N) can accurately stratify patients into different risk groups, with satisfactory prognosis in the low-risk group when treated with low-dose concurrent chemotherapy, providing new options for individualized de-escalation strategies. KEY POINTS: • A radiomics score, consisting of 3 predictive radiomics features (flatness, mean, and GLDM-GLN) integrated with the N stage, can identify specific LA-NPC populations for deintensified treatment. • In the selection of LA-NPC candidates for de-intensified treatment, radiomics score extracted from primary nasopharyngeal tumors based on CECT can be superior to traditional T stage classification as a predictor.

CPVL suppresses metastasis of nasopharyngeal carcinoma through inhibiting epithelial-mesenchymal transition.

PURPOSE: Distant metastasis is the main obstacle to treating nasopharyngeal carcinoma (NPC). Tumor distance metastasis is a complex process involving the jointly participation of multiple oncogenes, tumor suppressor genes, and metastasis-associated genes. Enough accurate prognostic genes for evaluating metastasis risk are lacking. We aimed to identify more precise biomarkers for NPC metastasis. METHODS: We performed weighted gene co-expression network analysis, differentially expressed gene analysis, univariate and multivariate stepwise Cox regression, and Kaplan-Meier (K-M) survival analyses, on data obtained from RNA sequencing of 10 NPC samples and the public database, to identify key genes correlated with NPC metastasis. Wound healing assays, transwell assays, and immunohistochemistry were conducted to validate our bioinformatic conclusions. Western blotting was performed to evaluate and quantify the effect of identified EMT genes on epithelial-mesenchymal transition (EMT) of NPC. RESULTS: Combined our own RNA sequencing data and public data, we determined carboxypeptidase vitellogenic-like protein (CPVL) as a tumor suppressor for NPC. Pathway enrichment analyses indicated that genes associated with CPVL are involved in EMT. NPC with low CPVL expression had high tumor purity and low levels of immune cells. Experimental results showed that CPVL protein predominantly expressed in cytoplasmic and membranous and it exhibited higher expression levels in NPC tissues without distant metastasis than those with distant metastasis. CPVL inhibits the migration and invasive capability of NPC cells. Overexpression of CPVL upregulates E-cadherin and ZO-1, whereas it downregulates vimentin, suggesting that CPVL suppresses tumor metastasis by inhibiting EMT. CONCLUSION: CPVL inhibits migration and invasion of NPC cells and is associated with tumor metastasis suppression through upregulating epithelial marker and inhibiting mesenchymal marker expression and could be a prognostic biomarker for metastasis risk evaluation in NPC.

An overall survival predictive nomogram to identify high-risk patients among locoregionally advanced nasopharyngeal carcinoma: Developed based on the SEER database and validated institutionally.

Objective: Locoregionally advanced nasopharyngeal carcinoma (LA-NPC) patients, even at the same stage, have different prognoses. We aim to construct a prognostic nomogram for predicting the overall survival (OS) to identify the high-risk LA-NPC patients. Materials and methods: Histologically diagnosed WHO type II and type III LA-NPC patients in the Surveillance, Epidemiology, and End Results (SEER) database were enrolled as the training cohort (n= 421), and LA-NPC patients from Shantou University Medical College Cancer Hospital (SUMCCH) served as the external validation cohort (n= 763). Variables were determined in the training cohort through Cox regression to form a prognostic OS nomogram, which was verified in the validation cohort, and compared with traditional clinical staging using the concordance index (C-index), Kaplan-Meier curves, calibration curves and decision curve analysis (DCA). Patients with scores higher than the specific cut-off value determined by the nomogram were defined as high-risk patients. Subgroup analyses and high-risk group determinants were explored. Results: Our nomogram had a higher C-index than the traditional clinical staging method (0.67 vs. 0.60, p<0.001). Good agreement between the nomogram-predicted and actual survival were shown in the calibration curves and DCA, indicating a clinical benefit of the nomogram. High-risk patients identified by our nomogram had worse prognosis than the other groups, with a 5-year overall survival (OS) of 60.4%. Elderly patients at advanced stage and without chemotherapy had a tendency for high risk than the other patients. Conclusions: Our OS predictive nomogram for LA-NPC patients is reliable to identify high-risk patients.

A composite score based on immune-related gene prognostic index and m6A risk score of head and neck squamous cell carcinoma.

Background: Immunotherapy has been demonstrated favorable in head and neck squamous cell carcinoma (HNSCC). Studies indicated that immune-related gene prognostic index (IRGPI) was a robust signature, and N6-methyladenosine (m6A) methylation had a significant impact on the tumor immune microenvironment (TIME) and immunotherapy of head and neck squamous cell carcinoma. Thus, combining indicated that immune-related gene prognostic index with m6A status should offer a better predictive power for immune responses. Methods: Head and neck squamous cell carcinoma samples from the cancer genome atlas (TCGA, n = 498) and gene expression omnibus database (GSE65858, n = 270) were used in this study. Cox regression analysis was used to construct the indicated that immune-related gene prognostic index through immune-related hub genes which were identified by weighted gene co-expression network analysis (WGCNA). The m6A risk score was constructed by least absolute shrinkage and selection operator (LASSO) regression analysis. Principal component analysis was used to construct a composite score, and systematically correlate subgroups according to tumor immune microenvironment cell-infiltrating characteristics. Results: A composite score was determined based on indicated that immune-related gene prognostic index and m6A risk score. Head and neck squamous cell carcinoma patients in the cancer genome atlas were divided into four subgroups: A (IRGPI-High&m6A-risk-High, n = 127), B (IRGPI-High&m6A-risk-Low, n = 99), C (IRGPI-Low&m6A-risk-High, n = 99), and D (IRGPI-Low&m6A-risk-Low, n = 128), and overall survival (OS) was significantly different between subgroups (p < 0.001). The characteristics of tumor immune microenvironment cell infiltration in the four subgroups were significantly different in subgroups (p < 0.05). The receiver operating characteristic (ROC) curves show the predictive value of composite score for overall survival was superior to other scores. Conclusion: The composite score is a promising prognostic signature which might distinguish immune and molecular characteristics, predict prognosis, and guide more effective immunotherapeutic strategies for head and neck squamous cell carcinoma.