No abstract

No abstract

Background Targeted therapy and immunotherapy put forward higher demands for accurate lung cancer classification, as well as benign versus malignant disease discrimination. Digital whole slide images (WSIs) witnessed the transition from traditional histopathology to computational approaches, arousing a hype of deep learning methods for histopathological analysis. We aimed at exploring the potential of deep learning models in the identification of lung cancer subtypes and cancer mimics from WSIs. Methods We initially obtained 741 WSIs from the First Affiliated Hospital of Sun Yat-sen University (SYSUFH) for the deep learning model development, optimization, and verification. Additional 318 WSIs from SYSUFH, 212 from Shenzhen People’s Hospital, and 422 from The Cancer Genome Atlas were further collected for multi-centre verification. EfficientNet-B5- and ResNet-50-based deep learning methods were developed and compared using the metrics of recall, precision, F1-score, and areas under the curve (AUCs). A threshold-based tumour-first aggregation approach was proposed and implemented for the label inferencing of WSIs with complex tissue components. Four pathologists of different levels from SYSUFH reviewed all the testing slides blindly, and the diagnosing results were used for quantitative comparisons with the best performing deep learning model. Results We developed the first deep learning-based six-type classifier for histopathological WSI classification of lung adenocarcinoma, lung squamous cell carcinoma, small cell lung carcinoma, pulmonary tuberculosis, organizing pneumonia, and normal lung. The EfficientNet-B5-based model outperformed ResNet-50 and was selected as the backbone in the classifier. Tested on 1067 slides from four cohorts of different medical centres, AUCs of 0.970, 0.918, 0.963, and 0.978 were achieved, respectively. The classifier achieved high consistence to the ground truth and attending pathologists with high intraclass correlation coefficients over 0.873. Conclusions Multi-cohort testing demonstrated our six-type classifier achieved consistent and comparable performance to experienced pathologists and gained advantages over other existing computational methods. The visualization of prediction heatmap improved the model interpretability intuitively. The classifier with the threshold-based tumour-first label inferencing method exhibited excellent accuracy and feasibility in classifying lung cancers and confused nonneoplastic tissues, indicating that deep learning can resolve complex multi-class tissue classification that conforms to real-world histopathological scenarios.

BackgroundThe strong invasive and metastatic nature of non-small cell lung cancer (NSCLC) leads to poor prognosis. Collagen triple helix repeat containing 1 (CTHRC1) is involved in cell migration, motility and invasion. The object of this study is to investigate the involvement of CTHRC1 in NSCLC invasion and metastasis.MethodsA proteomic analysis was performed to identify the different expression proteins between NSCLC and normal tissues. Cell lines stably express CTHRC1, MMP7, MMP9 were established. Invasion and migration were determined by scratch and transwell assays respectively. Clinical correlations of CTHRC1 in a cohort of 230 NSCLC patients were analysed.ResultsCTHRC1 is overexpressed in NSCLC as measured by proteomic analysis. Additionally, CTHRC1 increases tumour cell migration and invasion in vitro. Furthermore, CTHRC1 expression is significantly correlated with matrix metalloproteinase (MMP)7 and MMP9 expression in sera and tumour tissues from NSCLC. The invasion ability mediated by CTHRC1 were mainly MMP7- and MMP9-dependent. MMP7 or MMP9 depletion significantly eradicated the pro-invasive effects mediated by CTHRC1 on NSCLC cells. Clinically, patients with high CTHRC1 expression had poor survival.ConclusionsCTHRC1 serves as a pro-metastatic gene that contributes to NSCLC invasion and metastasis, which are mediated by upregulated MMP7 and MMP9 expression. Targeting CTHRC1 may be beneficial for inhibiting NSCLC metastasis.Electronic supplementary materialThe online version of this article (10.1186/s12885-018-4317-6) contains supplementary material, which is available to authorized users.

Purpose: Examining the role of developmental signaling pathways in “driver gene–negative” lung adenocarcinoma (patients with lung adenocarcinoma negative for EGFR, KRAS, BRAF, HER2, MET, ALK, RET, and ROS1 were identified as “driver gene–negative”) may shed light on the clinical research and treatment for this lung adenocarcinoma subgroup. We aimed to investigate whether developmental signaling pathways activation can stratify the risk of “driver gene–negative” lung adenocarcinoma. Experimental Design: In the discovery phase, we profiled the mRNA expression of each candidate gene using genome-wide microarrays in 52 paired lung adenocarcinoma and adjacent normal tissues. In the training phase, tissue microarrays and LASSO Cox regression analysis were applied to further screen candidate molecules in 189 patients, and we developed a predictive signature. In the validation phase, one internal cohort and two external cohorts were used to validate our novel prognostic signature. Results: Kyoto Encyclopedia of Genes and Genomes pathway analysis based on whole-genome microarrays indicated that the Wnt/β-catenin pathway was activated in “driver gene–negative” lung adenocarcinoma. Furthermore, the Wnt/β-catenin pathway–based gene expression profiles revealed 39 transcripts differentially expressed. Finally, a Wnt/β-catenin pathway–based CSDW signature comprising 4 genes (CTNNB1 or β-catenin, SOX9, DVL3, and Wnt2b) was developed to classify patients into high-risk and low-risk groups in the training cohort. Patients with high-risk scores in the training cohort had shorter overall survival [HR, 10.42; 6.46–16.79; P < 0.001) than patients with low-risk scores. Conclusions: The CSDW signature is a reliable prognostic tool and may represent genes that are potential drug targets for “driver gene–negative” lung adenocarcinoma.