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.

The role of epidermal growth factor-containing fibulin-like extracellular matrix protein 1 (EFEMP1) in osteosarcoma remains unknown. Then applying EFEMP1 siRNA, plasmids transfection and adding purified EFEMP1 protein in human osteosarcoma cell lines, and using immunohistochemistry on 113 osteosarcoma tissues, demonstrated that EFEMP1 was a poor prognostic indicator of osteosarcoma; EFEMP1 was specifically upregulated in osteosarcoma and associated with invasion and metastasis in vitro and in vivo. At the same time, we found a direct regulatory effect of EFEMP1 on MMP-2. Moreover, we firstly found the marked induction of EFEMP1 by oncogenic AEG-1. And EFEMP1 expression was inhibited by the selective inhibitor of NF-κB (PDTC) in osteosarcoma cells. Then we thought that NF-κB pathways might be one of the effective ways which EFEMP1 was induced by AEG-1. Thus, we suggested that EFEMP1 played a part as the mediator between AEG-1 and MMP-2. And NF-κB signaling pathway played an important role in this process. In summary, EFEMP1 was associated with invasion, metastasis and poor prognosis of osteosarcoma patients. EFEMP1 might indirectly enhance the expression of MMP-2, providing a potential explanation for the role of AEG-1 in metastasis. NF-κB pathways might be one of the effective ways which EFEMP1 was induced by AEG-1.

Astrocyte elevated gene-1 (AEG-1), known as an oncogene, is overexpressed in various cancers and implicated in tumor progression and metastasis. However, its functional significance and underlying molecular mechanisms in thyroid cancer remain to be elucidated. In the present study, we detected the potential function of AEG-1 in papillary thyroid cancer (PTC). We also investigated the relation between AEG-1 and matrix metalloproteases (MMP)2 and 9 through immunohistochemistry, western blotting, real-time PCR, immunofluorescence staining, zymography and co-immunoprecipitation (Co-IP). We found that overexpression of AEG-1 in PTC was positively correlated with lymph node metastasis and MMP2/9 expression. Knockdown of AEG-1 reduced the capacity of migration and invasion through downregulation of MMP2/9 in thyroid cancer cells. Furthermore, we firstly found that AEG-1 interacted with MMP9 in thyroid cancer cells. AEG-1 was associated with the activation of the nuclear factor κB (NF-κB) signaling pathways in thyroid cancer cells. Overall, our results for the first time showed that AEG-1 interacted with MMP9 in thyroid cancer cells and AEG-1 expression was closely associated with progression and metastasis of papillary thyroid cancer. AEG-1 might be a potential therapeutic target in papillary thyroid cancer.

The role of epidermal growth factor-containing fibulin-like extracellular matrix protein 1 (EFEMP1) inhibiting migration in hepatocellular carcinoma (HCC) remains unknown. Expression of EFEMP1 in HCC cell lines were quantified by western blotting and real-time PCR. The role of EFEMP1 in HCC cell migration was explored in vitro via siRNA and adding purified EFEMP1 protein. The associated molecule expression was detected by western blotting after downregulation of EFEMP1 and also tested by immunohistochemistry. Eight pairs of HCC non-HCC liver samples and 215 HCC samples were subjected to immunohistochemistry. EFEMP1 was highly expressed in 7,721 and HepG2 HCC cell lines while HuH7 HCC cell line expressed the lowest level of EFEMP1 compared with the others. Downregulating EFEMP1 by siRNA markedly increased the migration ability of HCC cells while adding purified EFEMP1 protein inhibited HCC cell migration. Downregulation of EFEMP1 increased the expression of ERK1/2, MMP2 and MMP9. Furthermore, U0126 (a highly selective and potent inhibitor of pERK1/2) could abrogate the migration ability enhanced by siRNA. Accordingly, MMP2 and MMP9 were inversely expressed with EFEMP1 expression by immunohistochemistry. EFEMP1 downregulated in HCC tissues, and lower EFEMP1 expression was significantly associated with HCC patients with ascites (P=0.050), vascular invasion (P=0.044), poorer differentiation (P=0.002) and higher clinical stage (P=0.003).

No abstract