An Anthropomorphic Diagnosis System of Pulmonary Nodules using Weak Annotation-Based Deep Learning.

Purpose: To develop an anthropomorphic diagnosis system of pulmonary nodules (PN) based on Deep learning (DL) that is trained by weak annotation data and has comparable performance to full-annotation based diagnosis systems. Methods: The proposed system uses deep learning (DL) models to classify PNs (benign vs. malignant) with weak annotations, which eliminates the need for time-consuming and labor-intensive manual annotations of PNs. Moreover, the PN classification networks, augmented with handcrafted shape features acquired through the ball-scale transform technique, demonstrate capability to differentiate PNs with diverse labels, including pure ground-glass opacities, part-solid nodules, and solid nodules. Results: The experiments were conducted on two lung CT datasets: (1) public LIDC-IDRI dataset with 1,018 subjects, (2) In-house dataset with 2740 subjects. Through 5-fold cross-validation on two datasets, the system achieved the following results: (1) an Area Under Curve (AUC) of 0.938 for PN localization and an AUC of 0.912 for PN differential diagnosis on the LIDC-IDRI dataset of 814 testing cases, (2) an AUC of 0.943 for PN localization and an AUC of 0.815 for PN differential diagnosis on the in-house dataset of 822 testing cases. These results demonstrate comparable performance to full annotation-based diagnosis systems. Conclusions: Our system can efficiently localize and differentially diagnose PNs even in resource-limited environments with good robustness across different grade and morphology sub-groups in the presence of deviations due to the size, shape, and texture of the nodule, indicating its potential for future clinical translation. Summary: An anthropomorphic diagnosis system of pulmonary nodules (PN) based on deep learning and weak annotation was found to achieve comparable performance to full-annotation dataset-based diagnosis systems, significantly reducing the time and the cost associated with the annotation. Key Points: A fully automatic system for the diagnosis of PN in CT scans using a suitable deep learning model and weak annotations was developed to achieve comparable performance (AUC = 0.938 for PN localization, AUC = 0.912 for PN differential diagnosis) with the full-annotation based deep learning models, reducing around 30%∼80% of annotation time for the experts.The integration of the hand-crafted feature acquired from human experts (natural intelligence) into the deep learning networks and the fusion of the classification results of multi-scale networks can efficiently improve the PN classification performance across different diameters and sub-groups of the nodule.

Establishment and validation of nomograms for predicting survival of lung invasive adenocarcinoma based on the level of pathological differentiation: a SEER cohort-based analysis.

Background: The pathological differentiation of invasive adenocarcinoma (IAC) has been linked closely with epidemiological characteristics and clinical prognosis. However, the current models cannot accurately predict IAC outcomes and the role of pathological differentiation is confused. This study aimed to establish differentiation-specific nomograms to explore the effect of IAC pathological differentiation on overall survival (OS) and cancer-specific survival (CSS). Methods: The data of eligible IAC patients between 1975 and 2019 were collected from the Surveillance, Epidemiology, and End Results (SEER) database, and randomly divided in a ratio of 7:3 into a training cohort and a validation cohort. The associations between pathological differentiation and other clinical characteristics were evaluated using chi-squared test. The OS and CSS analyses were performed using the Kaplan-Meier estimator, and the log-rank test was used for nonparametric group comparisons. Multivariate survival analysis was performed using a Cox proportional hazards regression model. The discrimination, calibration, and clinical performance of nomograms were assessed by area under receiver operating characteristic curve (AUC), calibration plots, and decision curve analysis (DCA). Results: A total of 4,418 IAC patients (1,001 high-differentiation, 1,866 moderate-differentiation, and 1,551 low-differentiation) were identified. Seven risk factors [age, sex, race, tumor-node-metastasis (TNM) stage, tumor size, marital status, and surgery] were screened to construct differentiation-specific nomograms. Subgroup analyses showed that disparate pathological differentiation played distinct roles in prognosis, especially in patients with older age, white race, and higher TNM stage. The AUC of nomograms for OS and CSS in the training cohort were 0.817 and 0.835, while in the validation cohort were 0.784 and 0.813. The calibration curves showed good conformity between the prediction of the nomograms and the actual observations. DCA results indicated that these nomogram models could be used as a supplement to the prediction of the TNM stage. Conclusions: Pathological differentiation should be considered as an independent risk factor for OS and CSS of IAC. Differentiation-specific nomogram models with good discrimination and calibration capacity were developed in the study to predict the OS and CSS in 1-, 3- and 5-year, which could be used predict prognosis and select appropriate treatment options.

The role of lipid metabolic reprogramming in tumor microenvironment.

Metabolic reprogramming is one of the most important hallmarks of malignant tumors. Specifically, lipid metabolic reprogramming has marked impacts on cancer progression and therapeutic response by remodeling the tumor microenvironment (TME). In the past few decades, immunotherapy has revolutionized the treatment landscape for advanced cancers. Lipid metabolic reprogramming plays pivotal role in regulating the immune microenvironment and response to cancer immunotherapy. Here, we systematically reviewed the characteristics, mechanism, and role of lipid metabolic reprogramming in tumor and immune cells in the TME, appraised the effects of various cell death modes (specifically ferroptosis) on lipid metabolism, and summarized the antitumor therapies targeting lipid metabolism. Overall, lipid metabolic reprogramming has profound effects on cancer immunotherapy by regulating the immune microenvironment; therefore, targeting lipid metabolic reprogramming may lead to the development of innovative clinical applications including sensitizing immunotherapy.

The multi-omics analysis identifies a novel cuproptosis-anoikis-related gene signature in prognosis and immune infiltration characterization of lung adenocarcinoma.

Background: Lung adenocarcinoma (LUAD) has emerged as one of the most aggressive lethal cancers. Anoikis serves as programmed apoptosis initiated by the detachment of cells from the extracel-lular matrix. Cuproptosis is distinct from traditional cell death modalities. The above two modes are both closely related to tumor progression, prognosis, and treatment. However, whether they have synergistic effects in LUAD deserves further investigation. Methods: The anoikis-related prognostic genes (ANRGs) co-expressed with cuproptosis-associated genes (CAGs) were screened using correlation analysis, analysis of variance, least absolute shrinkage, and selection operator (LASSO), and COX regression followed by functional analysis, and then LUAD risk score model was constructed. Using consensus clustering, the relationship between different subtypes and clinicopathological features, immune infiltration characteristics, and somatic mutations was analyzed. A nomogram was developed by incorporating clinical information, which provided a prediction of the survival of patients. Finally, a comprehensive analysis of ANRGs was performed and verified by the HPA database. Results: A total of 27 ANRGs associated with cuproptosis were obtained. On this basis, three distinct ANRGs subtypes were identified, and the differences between clinical prognosis and immune infiltration were observed. A risk score model has been constructed by incorporating seven ANRGs signatures (EIF2AK3, IKZF3, ITGAV, OGT, PLK1, TRAF2, XRCC5). A highly reliable nomogram was developed to help formulate treatment strategies based on risk score and the clinicopathological features of LUAD. The seven-gene signature was turned out to be strongly linked to immune cells and validated in single-cell data. Immunohistochemistry proved that all of them are highly expressed in LUAD tissues. Conclusion: This study reveals the potential relationship between cuproptosis-related ANRGs and clinicopathological features, tumor microenvironment (TME), and mutation characteristics, which can be applied for predicting the prognosis of LUAD and help develop individualized treatment strategies.

Identification of adenoid subtype characterized with immune-escaped phenotype in lung squamous carcinoma based on transcriptomics.

Non-small cell lung cancer (NSCLC) is a heterogeneous disease, and its demarcation contributes to various therapeutic outcomes. However, a small subset of tumors shows different molecular features that are in contradiction with pathological classification. Unsupervised clustering was performed to subtype NSCLC using the transcriptome data from the TCGA database. Next, immune microenvironment features of lung adenocarcinoma (LUAD), lung squamous carcinoma (LUSC), and lung adenoid squamous carcinoma (LASC) were characterized. In addition, diagnostic biomarkers to demarcate LASC among LUSC were screened using weighted gene co-expression network analysis (WGCNA) and validated by the in-house cohort. LASC was identified as a novel subtype with adenoid transcriptomic features in LUSC, which exhibited the most immuno-escaped phenotype among all NSCLC subtypes. In addition, FOLR1 was identified as a biomarker for LASC discrimination using the WGCNA analysis, and its diagnostic value was validated by the in-house cohort. Moreover, FOLR1 was related to immuno-escaped tumors in LUSC but not in LUAD. Overall, we proposed a novel typing strategy in NSCLC and identified FOLR1 as a biomarker for LASC discrimination.

Statin shapes inflamed tumor microenvironment and enhances immune checkpoint blockade in non-small cell lung cancer.

Immune checkpoint blockade (ICB) therapy has achieved breakthroughs in the treatment of advanced non-small cell lung cancer (NSCLC). Nevertheless, the low response due to immuno-cold (i.e., tumors with limited tumor-infiltrating lymphocytes) tumor microenvironment (TME) largely limits the application of ICB therapy. Based on the glycolytic/cholesterol synthesis axis, a stratification framework for EGFR-WT NSCLC was developed to summarize the metabolic features of immuno-cold and immuno-hot tumors. The cholesterol subgroup displays the worst prognosis in immuno-cold NSCLC, with significant enrichment of the cholesterol gene signature, indicating that targeting cholesterol synthesis is essential for the therapy for immuno-cold NSCLC. Statin, the inhibitor for cholesterol synthesis, can suppress the aggressiveness of NSCLC in vitro and in vivo and can also drastically reverse the phenotype of immuno-cold to an inflamed phenotype in vivo. This change led to a higher response to ICB therapy. Moreover, both our in-house data and meta-analysis further support that statin can significantly enhance ICB efficacy. In terms of preliminary mechanisms, statin could transcriptionally inhibit PD-L1 expression and induce ferroptosis in NSCLC cells. Overall, we reveal the significance of cholesterol synthesis in NSCLC and demonstrate the improved therapeutic efficacy of ICB in combination with statin. These findings could provide a clinical insight to treat NSCLC patients with immuno-cold tumors.

HLA class II molecule HLA-DRA identifies immuno-hot tumors and predicts the therapeutic response to anti-PD-1 immunotherapy in NSCLC.

BACKGROUND: Immune checkpoint blockade (ICB) only works well for a certain subset of patients with non-small cell lung cancer (NSCLC). Therefore, biomarkers for patient stratification are desired, which can suggest the most beneficial treatment. METHODS: In this study, three datasets (GSE126044, GSE135222, and GSE136961) of immunotherapy from the Gene Expression Omnibus (GEO) database were analyzed, and seven intersected candidates were extracted as potential biomarkers for ICB followed by validation with The Cancer Genome Atlas (TCGA) dataset and the in-house cohort data. RESULTS: Among these candidates, we found that human leukocyte antigen-DR alpha (HLA-DRA) was downregulated in NSCLC tissues and both tumor and immune cells expressed HLA-DRA. In addition, HLA-DRA was associated with an inflamed tumor microenvironment (TME) and could predict the response to ICB in NSCLC. Moreover, we validated the predictive value of HLA-DRA in immunotherapy using an in-house cohort. Furthermore, HLA-DRA was related to the features of inflamed TME in not only NSCLC but also in most cancer types. CONCLUSION: Overall, HLA-DRA could be a promising biomarker for guiding ICB in NSCLC.

Statins Lower Lipid Synthesis But Promote Secretion of Cholesterol-Enriched Extracellular Vesicles and Particles.

Lipid droplets are lipid-rich cytosolic organelles that play roles in cell signaling, membrane trafficking, and many other cellular activities. Recent studies revealed that lipid droplets in cancer cells have various biological functions, such as energy production, membrane synthesis, and chemoresistance, thereby fostering cancer progression. Accordingly, the administration of antilipemic agents could improve anti-cancer treatment efficacy given hydrophobic chemotherapeutic drugs could be encapsulated into lipid droplets and then expelled to extracellular space. In this study, we investigated whether statins could promote treatment efficacy of lipid droplet-rich ovarian SKOV-3 cells and the potential influences on generation and composition of cell-derived extracellular vesicles and particles (EVP). Our studies indicate that statins can significantly lower lipid biosynthesis. Moreover, statins can inhibit proliferation, migration, and invasion of SKOV-3 cells and enhance chemosensitivity in vitro and in vivo. Furthermore, statins can lower EVP secretion but enforce the release of cholesterol-enriched EVPs, which can further lower lipid contents in parental cells. It is the first time that the influence of statins on EVP generation and EVP-lipid composition is observed. Overall, we demonstrated that statins could inhibit lipid production, expel cholesterol to extracellular space via EVPs, and improve chemosensitivity.

Recurrent lung adenocarcinoma benefits from microwave ablation following multidisciplinary treatments: A case with long-term survival.

Lung cancer has become the leading cause of cancer death all over the world. Nowadays, there is a consensus that the treatment of non-small cell lung cancer (NSCLC) prefers a combination of multidisciplinary comprehensive treatment and individualized treatment, which can significantly improve the prognosis of patients. Here, we report a female patient with recurrence-prone NSCLC. She had a decade-long disease course, during which the lesion recurred twice and finally cured with Multi-Disciplinary Treatment (MDT). An elderly female patient was admitted to the hospital after diagnosis of lung cancer, and treated with surgery and postoperative adjuvant chemotherapy. Five years later, suspicious lesions were found by computed tomography (CT) reexamination, and then confirmed tumor recurrence by puncture biopsy. Based on the genetic test results, gefitinib was used for subsequent targeted therapy, and the lesion gradually shrunk to disappear. However, the lesion appeared again two years later, after consultation the microwave ablation was adopted and the curative effect was excellent. At last, regular reexamination showed no abnormality, the patient has survived so far. The case proves the great benefit of multidisciplinary comprehensive treatment, especially microwave ablation for patient with recurrence-prone NSCLC. And the effect of systemic anti-tumor immune response induced by microwave ablation on lung cancer also needs to be further explored.