Comprehensive evaluation of published gene expression prognostic signatures for biomarker-based lung cancer clinical studies

Reprogramming exosomes for immunity-remodeled photodynamic therapy against non-small cell lung cancer

Traditional treatments against advanced non-small cell lung cancer (NSCLC) with high morbidity and mortality continue to be dissatisfactory. Given this situation, there is an urgent requirement for alternative modalities that provide lower invasiveness, superior clinical effectiveness, and minimal adverse effects. The combination of photodynamic therapy (PDT) and immunotherapy gradually become a promising approach for high-grade malignant NSCLC. Nevertheless, owing to the absence of precise drug delivery techniques as well as the hypoxic and immunosuppressive characteristics of the tumor microenvironment (TME), the efficacy of this combination therapy approach is less than ideal. In this study, we construct a novel nanoplatform that indocyanine green (ICG), a photosensitizer, loads into hollow manganese dioxide (MnO2) nanospheres (NPs) (ICG@MnO2), and then encapsulated in PD-L1 monoclonal antibodies (anti-PD-L1) reprogrammed exosomes (named ICG@MnO2@Exo-anti-PD-L1), to effectively modulate the TME to oppose NSCLC by the synergy of PDT and immunotherapy modalities. The ICG@MnO2@Exo-anti-PD-L1 NPs are precisely delivered to the tumor sites by targeting specially PD-L1 highly expressed cancer cells to controllably release anti-PD-L1 in the acidic TME, thereby activating T cell response. Subsequently, upon endocytic uptake by cancer cells, MnO2 catalyzes the conversion of H2O2 to O2, thereby alleviating tumor hypoxia. Meanwhile, ICG further utilizes O2 to produce singlet oxygen (1O2) to kill tumor cells under 808 nm near-infrared (NIR) irradiation. Furthermore, a high level of intratumoral H2O2 reduces MnO2 to Mn2+, which remodels the immune microenvironment by polarizing macrophages from M2 to M1, further driving T cells. Taken together, the current study suggests that the ICG@MnO2@Exo-anti-PD-L1 NPs could act as a novel drug delivery platform for achieving multimodal therapy in treating NSCLC.

Identification of a gene expression signature of vascular invasion and recurrence in stage I lung adenocarcinoma via bulk and spatial transcriptomics

Microscopic vascular invasion (VI) is predictive of recurrence and benefit from lobectomy in stage I lung adenocarcinoma (LUAD) but is difficult to assess in resection specimens and cannot be accurately predicted prior to surgery. Thus, new biomarkers are needed to identify this aggressive subset of stage I LUAD tumors. To assess molecular and microenvironment features associated with angioinvasive LUAD we profiled 162 resected stage I tumors with and without VI by RNA-seq and explored spatial patterns of gene expression in a subset of 15 samples by high-resolution spatial transcriptomics (stRNA-seq). Despite the small size of invaded blood vessels, we identified a gene expression signature of VI from the bulk RNA-seq discovery cohort (n=103) and found that it was associated with VI foci, desmoplastic stroma, and high-grade patterns in our stRNA-seq data. We observed a stronger association with high-grade patterns from VI+ compared with VI- tumors. Using the discovery cohort, we developed a transcriptomic predictor of VI, that in an independent validation cohort (n=60) was associated with VI (AUROC=0.86; p=5.42×10-6) and predictive of recurrence-free survival (HR=1.98; p=0.024), even in VI- LUAD (HR=2.76; p=0.003). To determine our VI predictor's robustness to intra-tumor heterogeneity we used RNA-seq data from multi-region sampling of stage I LUAD cases in TRACERx, where the predictor scores showed high correlation (R=0.87, p<2.2×10-16) between two randomly sampled regions of the same tumor. Our study suggests that VI-associated gene expression changes are detectable beyond the site of intravasation and can be used to predict the presence of VI. This may enable the prediction of angioinvasive LUAD from biopsy specimens, allowing for more tailored medical and surgical management of stage I LUAD.

Tobacco-induced hyperglycemia promotes lung cancer progression via cancer cell-macrophage interaction through paracrine IGF2/IR/NPM1-driven PD-L1 expression

Tobacco smoking (TS) is implicated in lung cancer (LC) progression through the development of metabolic syndrome. However, direct evidence linking metabolic syndrome to TS-mediated LC progression remains to be established. Our findings demonstrate that 4-(methylnitrosamino)-1-(3-pyridyl)-1-butanone and benzo[a]pyrene (NNK and BaP; NB), components of tobacco smoke, induce metabolic syndrome characteristics, particularly hyperglycemia, promoting lung cancer progression in male C57BL/6 J mice. NB enhances glucose uptake in tumor-associated macrophages by increasing the expression and surface localization of glucose transporter (GLUT) 1 and 3, thereby leading to transcriptional upregulation of insulin-like growth factor 2 (IGF2), which subsequently activates insulin receptor (IR) in LC cells in a paracrine manner, promoting its nuclear import. Nuclear IR binds to nucleophosmin (NPM1), resulting in IR/NPM1-mediated activation of the CD274 promoter and expression of programmed death ligand-1 (PD-L1). Restricting glycolysis, depleting macrophages, or blocking PD-L1 inhibits NB-mediated LC progression. Analysis of patient tissues and public databases reveals elevated levels of IGF2 and GLUT1 in tumor-associated macrophages, as well as tumoral PD-L1 and phosphorylated insulin-like growth factor 1 receptor/insulin receptor (pIGF-1R/IR) expression, suggesting potential poor prognostic biomarkers for LC patients. Our data indicate that paracrine IGF2/IR/NPM1/PD-L1 signaling, facilitated by NB-induced dysregulation of glucose levels and metabolic reprogramming of macrophages, contributes to TS-mediated LC progression.

Identification of a prognosis-related gene signature and ceRNA regulatory networks in lung adenocarcinoma

The ceRNA network, consisting of both noncoding RNA and protein-coding RNA, governs the occurrence, progression, metastasis, and infiltration of lung adenocarcinoma. Signatures comprising multiple genes can effectively determine survival stratification and prognosis of patients with lung adenocarcinoma. To explore the mechanisms of lung adenocarcinoma progression and identify potential biological targets, we carried out systematic bioinformatics analyses of the genetic profiles of lung adenocarcinoma, such as weighted gene co-expression network analysis (WGCNA), differential expression (DE) assessment, univariate and multivariate Cox proportional hazard regression models, ceRNA modulatory networks generated using the ENCORI and miRcode databases, nomogram models, ROC curve assessment, and Kaplan-Meier survival curve analysis. The ceRNA network encompassed 37 nodes, comprising 12 mRNAs, 22 lncRNAs, and three miRNAs. Simultaneously, we performed integration analysis using the 12 genes from the ceRNA network. Our findings revealed that the signature established by these 12 genes serves as an adverse element in lung adenocarcinoma, contributing to unfavorable patient prognosis. To ensure the credibility of our results, we used in vitro experiments for further verification. In conclusion, our study delved into the potential mechanisms underlying lung adenocarcinoma via the ceRNA regulatory network, specifically focusing on the PIF1 and has-miR-125a-5p axis. Additionally, a signature comprising 12 genes was identified as a biomarker related to the prognosis of lung adenocarcinoma.

A cellular and spatial atlas of TP53 -associated tissue remodeling in lung adenocarcinoma

TP53 is the most frequently mutated gene across many cancers and is associated with shorter survival in lung adenocarcinoma (LUAD). To define how TP53 mutations affect the LUAD tumor microenvironment (TME), we constructed a multi-omic cellular and spatial tumor atlas of 23 treatment-naïve human lung tumors. We found that TP53 -mutant ( TP53 mut ) malignant cells lose alveolar identity and upregulate highly proliferative and entropic gene expression programs consistently across resectable LUAD patient tumors, genetically engineered mouse models, and cell lines harboring a wide spectrum of TP53 mutations. We further identified a multicellular tumor niche composed of SPP1 + macrophages and collagen-expressing fibroblasts that coincides with hypoxic, pro-metastatic expression programs in TP53 mut tumors. Spatially correlated angiostatic and immune checkpoint interactions, including CD274 - PDCD1 and PVR - TIGIT , are also enriched in TP53 mut LUAD tumors, which may influence response to checkpoint blockade therapy. Our methodology can be further applied to investigate mutation-specific TME changes in other cancers.

A surgical Decision-making scoring model for spontaneous ventilation- and mechanical ventilation-video-assisted thoracoscopic surgery in non-small-cell lung cancer patients

BACKGROUNDS

Spontaneous ventilation-video-assisted thoracoscopic surgery (SV-VATS) has been applied to non-small cell lung cancer (NSCLC) patients in many centers. Since it remains a new and challenging surgical technique, only selected patients can be performed SV-VATS. We aim to conduct a retrospective single-center study to develop a clinical decision-making model to make surgery decision between SV-VATS and MV (mechanical ventilation) -VATS in NSCLC patients more objectively and individually.

METHODS

Four thousand three hundred sixty-eight NSCLC patients undergoing SV-VATS or MV-VATS in the department of thoracic surgery between 2011 and 2018 were included. Univariate and multivariate regression analysis were used to identify potential factors influencing the surgical decisions. Factors with statistical significance were selected for constructing the Surgical Decision-making Scoring (SDS) model. The performance of the model was validated by area under the receiver operating characteristic curve (AUC), calibration curves and decision curve analysis (DCA).

RESULTS

The Surgical Decision-making Scoring (SDS) model was built guided by the clinical judgment and statistically significant results of univariate and multivariate regression analyses of potential predictors, including smoking status (p = 0.03), BMI (p < 0.001), ACCI (p = 0.04), T stage (p < 0.001), N stage (p < 0.001), ASA grade (p < 0.001) and surgical technique (p < 0.001). The AUC of the training group and the testing group were 0.72 and 0.70, respectively. The calibration curves and the DCA curve revealed that the SDS model has a desired performance in predicting the surgical decision.

CONCLUSIONS

This SDS model is the first clinical decision-making model developed for an individual NSCLC patient to make decision between SV-VATS and MV-VATS.

Evaluation of pre-diagnostic blood protein measurements for predicting survival after lung cancer diagnosis

BACKGROUND

To evaluate whether circulating proteins are associated with survival after lung cancer diagnosis, and whether they can improve prediction of prognosis.

METHODS

We measured up to 1159 proteins in blood samples from 708 participants in 6 cohorts. Samples were collected within 3 years prior to lung cancer diagnosis. We used Cox proportional hazards models to identify proteins associated with overall mortality after lung cancer diagnosis. To evaluate model performance, we used a round-robin approach in which models were fit in 5 cohorts and evaluated in the 6th cohort. Specifically, we fit a model including 5 proteins and clinical parameters and compared its performance with clinical parameters only.

FINDINGS

There were 86 proteins nominally associated with mortality (p < 0.05), but only CDCP1 remained statistically significant after accounting for multiple testing (hazard ratio per standard deviation: 1.19, 95% CI: 1.10-1.30, unadjusted p = 0.00004). The external C-index for the protein-based model was 0.63 (95% CI: 0.61-0.66), compared with 0.62 (95% CI: 0.59-0.64) for the model with clinical parameters only. Inclusion of proteins did not provide a statistically significant improvement in discrimination (C-index difference: 0.015, 95% CI: -0.003 to 0.035).

INTERPRETATION

Blood proteins measured within 3 years prior to lung cancer diagnosis were not strongly associated with lung cancer survival, nor did they importantly improve prediction of prognosis beyond clinical information.

FUNDING

No explicit funding for this study. Authors and data collection supported by the US National Cancer Institute (U19CA203654), INCA (France, 2019-1-TABAC-01), Cancer Research Foundation of Northern Sweden (AMP19-962), and Swedish Department of Health Ministry.

A novel Chr1-miR-200 driven whole transcriptome signature shapes tumor immune microenvironment and predicts relapse in early-stage lung adenocarcinoma

BACKGROUND

In Lung adenocarcinoma (LUAD), targeted therapies and immunotherapies have moved from metastatic to early stage and stratification of the relapse risk becomes mandatory. Here we identified a miR-200 based RNA signature that delineates Epithelial-to-mesenchymal transition (EMT) heterogeneity and predicts survival beyond current classification systems.

METHODS

A miR-200 signature was identified using RNA sequencing. We scored the miR-200 signature by WISP (Weighted In Silico Pathology), used GSEA to identify pathway enrichments and MCP-counter to characterize immune cell infiltrates. We evaluate the clinical value of this signature in our series of LUAD and using TCGA and 7 published datasets.

RESULTS

We identified 3 clusters based on supervised classification: I is miR-200-sign-down and enriched in TP53 mutations IIA and IIB are miR-200-sign-up: IIA is enriched in EGFR (p < 0.001), IIB is enriched in KRAS mutation (p < 0.001). WISP stratified patients into miR-200-sign-down (n = 65) and miR-200-sign-up (n = 42). Several biological processes were enriched in MiR-200-sign-down tumors, focal adhesion, actin cytoskeleton, cytokine/receptor interaction, TP53 signaling and cell cycle pathways. Fibroblast, immune cell infiltration and PDL1 expression were also significantly higher suggesting immune exhaustion. This signature stratified patients into high-vs low-risk groups, miR-200-sign-up had higher DFS, median not reached at 60 vs 41 months and within subpopulations with stage I, IA, IB, or II. Results were validated on TCGA data on 7 public datasets.

CONCLUSION

This EMT and miR-200-related prognostic signature refines prognosis evaluation independently of tumor stage and paves the way towards assessing the predictive value of this LUAD clustering to optimize perioperative treatment.

Systematic analysis of transcriptome signature for improving outcomes in lung adenocarcinoma

PURPOSE

The updated guidelines highlight gene expression-based multigene panel as a critical tool to assess overall survival (OS) and improve treatment for lung adenocarcinoma (LUAD) patients. Nevertheless, genome-wide expression signatures are still limited in real clinical utility because of insufficient data utilization, a lack of critical validation, and inapposite machine learning algorithms.

METHODS

2330 primary LUAD samples were enrolled from 11 independent cohorts. Seventy-six algorithm combinations based on ten machine learning algorithms were applied. A total of 108 published gene expression signatures were collected. Multiple pharmacogenomics databases and resources were utilized to identify precision therapeutic drugs.

RESULTS

We comprehensively developed a robust machine learning-derived genome-wide expression signature (RGS) according to stably OS-associated RNAs (OSRs). RGS was an independent risk element and remained robust and reproducible power by comparing it with general clinical parameters, molecular characteristics, and 108 published signatures. RGS-based stratification possessed different biological behaviors, molecular mechanisms, and immune microenvironment patterns. Integrating multiple databases and previous studies, we identified that alisertib was sensitive to the high-risk group, and RITA was sensitive to the low-risk group.

CONCLUSION

Our study offers an appealing platform to screen dismal prognosis LUAD patients to improve clinical outcomes by optimizing precision therapy.

Construction of an STK11 Mutation and Immune-Related Prognostic Prediction Model in Lung Adenocarcinoma

Background

STK11 mutation in LUAD affects immune cell infiltration in tumor tissue, and is associated with tumor prognosis.

Objective

This study aimed to construct a STK11 mutation and immune-related LUAD prognostic model.

Materials and Methods

The mutation frequency of STK11 in LUAD was queried via cBioPortal in TCGA and PanCancer Atlas databases. The degree of immune infiltration was analyzed by CIBERSORT analysis. DEGs in STK11mut and STK11wt samples were analyzed. Metascape, GO and KEGG methods were adopted for functional and signaling pathway enrichment analysis of DEGs. Genes related to immune were overlapped with DEGs to acquire immune-related DEGs, whose Cox regression and LASSO analyses were employed to construct prognostic model. Univariate and multivariate Cox regression analyses verified the independence of riskscore and clinical features. A nomogram was established to predict the OS of patients. Additionally, TIMER was introduced to analyze relationship between infiltration abundance of 6 immune cells and expression of feature genes in LUAD.

Results

The mutation frequency of STK11 in LUAD was 16%, and the degrees of immune cell infiltration were different between the wild-type and mutant STK11. DEGs of STK11 mutated and unmutated LUAD samples were mainly enriched in immune-related biological functions and signaling pathways. Finally, 6 feature genes were obtained, and a prognostic model was established. Riskscore was an independent immuno-related prognostic factor for LUAD. The nomogram diagram was reliable.

Conclusion

Collectively, genes related to STK11 mutation and immunity were mined from the public database, and a 6-gene prognostic prediction signature was generated.

Establishment of a Lymph Node Metastasis-Associated Prognostic Signature for Lung Adenocarcinoma

Background

Lung adenocarcinoma (LUAD) is the most common histological subtype of non-small cell lung cancer (NSCLC) with a low 5-year survival rate, which may be associated with the presence of metastatic tumors at the time of diagnosis, especially lymph node metastasis (LNM). This study aimed to construct a LNM-related gene signature for predicting the prognosis of patients with LUAD.

Methods

RNA sequencing data and clinical information of LUAD patients were extracted from The Cancer Genome Atlas (TCGA) and Gene Expression Omnibus (GEO) databases. Samples were divided into metastasis (M) and nonmetastasis (NM) groups based on LNM status. Differentially expressed genes (DEGs) between M and NM groups were screened, and then WGCNA was applied to identify key genes. Furthermore, univariate Cox and LASSO regression analyses were conducted to construct a risk score model, and the predictive performance of model was validated by GSE68465, GSE42127, and GSE50081. The protein and mRNA expression level of LNM-associated genes were detected by human protein atlas (HPA) and GSE68465.

Results

A prognostic model based on eight LNM-related genes (ANGPTL4, BARX2, GPR98, KRT6A, PTPRH, RGS20, TCN1, and TNS4) was developed. Patients in the high-risk group had poorer overall survival than those in the low-risk group, and validation analysis showed that this model had potential predictive value for patients with LUAD. HPA analysis supported the upregulation of ANGPTL4, KRT6A, BARX2, RGS20 and the downregulation of GPR98 in LUAD compared with normal tissues.

Conclusion

Our results indicated that the eight LNM-related genes signature had potential value in the prognosis of patients with LUAD, which may have important practical implications.

Identification of lactate metabolism-related subtypes and development of a lactate-related prognostic indicator of lung adenocarcinoma

Background: Increasing evidence supports that lactate plays an important role in tumor proliferation, invasion and within the tumor microenvironment (TME). This is particularly relevant in lung adenocarcinoma (LUAD). Therefore, there is a current need to investigate lactate metabolism in LUAD patients and how lactate metabolism is affected by different therapies.

Methods: Data from LUAD patients were collected from The Cancer Genome Atlas (TCGA) and patients were divided into two subtypes according to 12 lactate metabolism-related genes to explore the effect of lactate metabolism in LUAD. We established a lactate-related prognostic indicator (LRPI) based on different gene expression profiles. Subsequently, we investigated associations between this LRPI and patient survival, molecular characteristics and response to therapy. Some analyses were conducted using the Genomics of Drug Sensitivity in Cancer (GDSC) database.

Results: The two LUAD subtypes exhibited different levels of lactate metabolism, in which patients that displayed high lactate metabolism also had a worse prognosis and a poorer immune environment. Indeed, LRPI was shown to accurately predict the prognosis of LUAD patients. Patients with a high LRPI showed a poor prognosis coupled with high sensitivity to chemotherapy using GDSC data. Meanwhile, these patients exhibited a high responsiveness to immunotherapy in TMB (Tumor mutation burden) and TIDE (Tumor Immune Dysfunction and Exclusion) analyses.

Conclusion: We validated the effect of lactate metabolism on the prognosis of LUAD patients and established a promising biomarker. LRPI can predict LUAD patient survival, molecular characteristics and response to therapy, which can aid the individualized treatment of LUAD patients.

Cell Trajectory-Related Genes of Lung Adenocarcinoma Predict Tumor Immune Microenvironment and Prognosis of Patients

Background

Lung adenocarcinoma (LUAD) is the most common subtype of lung cancer which typically exhibits a diverse progression trajectory. Our study sought to explore the cell differentiation trajectory of LUAD and its clinical relevance.

Methods

Utilizing a single-cell RNA-sequencing dataset (GSE117570), we identified LUAD cells of distinct differential status along with differentiation-related genes (DRGs). DRGs were applied to the analysis of bulk-tissue RNA-sequencing dataset (GSE72094) to classify tumors into different subtypes, whose clinical relevance was further analyzed. DRGs were also applied to gene co-expression network analysis (WGCNA) using another bulk-tissue RNA-sequencing dataset (TCGA-LUAD). Genes from modules that demonstrated a significant correlation with clinical traits and were differentially expressed between normal tissue and tumors were identified. Among these, genes with significant prognostic relevance were used for the development of a prognostic nomogram, which was tested on TCGA-LUAD dataset and validated in GSE72094. Finally, CCK-8, EdU, cell apoptosis, cell colony formation, and Transwell assays were used to verify the functions of the identified genes.

Results

Four clusters of cells with distinct differentiation status were characterized, whose DRGs were predominantly correlated with pathways of immune regulation. Based on DRGs, tumors could be clustered into four subtypes associated with distinct immune microenvironment and clinical outcomes. DRGs were categorized into four modules. A total of nine DRGs (SFTPB, WFDC2, HLA-DPA1, TIMP1, MS4A7, HLA-DQA1, VCAN, KRT8, and FABP5) with most significant survival-predicting power were integrated to develop a prognostic model, which outperformed the traditional parameters in predicting clinical outcomes. Finally, we verified that knockdown of WFDC2 inhibited proliferation, migration, and invasion but promoted the apoptosis of A549 cells in vitro.

Conclusion

The cellular composition and cellular differentiation status of tumor mass can predict the clinical outcomes of LUAD patients. It also plays an important role in shaping the tumor immune microenvironment.

Comprehensive evaluation of machine learning models and gene expression signatures for prostate cancer prognosis using large population cohorts

Overtreatment remains a pervasive problem in prostate cancer (PCa) management due to the highly variable and often indolent course of disease. Molecular signatures derived from gene expression profiling have played critical roles in guiding PCa treatment decisions. Many gene expression signatures have been developed to improve the risk stratification of PCa and some of them have already been applied to clinical practice. However, no comprehensive evaluation has been performed to compare the performance of these signatures. In this study, we conducted a systematic and unbiased evaluation of 15 machine learning (ML) algorithms and 30 published PCa gene expression-based prognostic signatures leveraging 10 transcriptomics datasets with 1,558 primary PCa patients from public data repositories. This analysis revealed that survival analysis models outperformed binary classification models for risk assessment, and the performance of the survival analysis methods - Cox model regularized with ridge penalty (Cox-Ridge) and partial least squares regression for Cox model (Cox-PLS) - were generally more robust than the other methods. Based on the Cox-Ridge algorithm, several top prognostic signatures displayed comparable or even better performance than commercial panels. These findings will facilitate the identification of existing prognostic signatures that are promising for further validation in prospective studies and promote the development of robust prognostic models to guide clinical decision-making. Moreover, this study provides a valuable data resource from large primary PCa cohorts, which can be used to develop, validate, and evaluate novel statistical methodologies and molecular signatures to improve PCa management.

Updated Prognostic Factors in Localized NSCLC

Lung cancer is the most common cause of cancer mortality worldwide, and non-small cell lung cancer (NSCLC) represents 80% of lung cancer subtypes. Patients with localized non-small cell lung cancer may be considered for upfront surgical treatment. However, the overall 5-year survival rate is 59%. To improve survival, adjuvant chemotherapy (ACT) was largely explored and showed an overall benefit of survival at 5 years < 7%. The evaluation of recurrence risk and subsequent need for ACT is only based on tumor stage (TNM classification); however, more than 25% of patients with stage IA/B tumors will relapse. Recently, adjuvant targeted therapy has been approved for EGFR-mutated resected NSCLC and trials are evaluating other targeted therapies and immunotherapies in adjuvant settings. Costs, treatment duration, emergence of resistant clones and side effects stress the need for a better selection of patients. The identification and validation of prognostic and theranostic markers to better stratify patients who could benefit from adjuvant therapies are needed. In this review, we report current validated clinical, pathological and molecular prognosis biomarkers that influence outcome in resected NSCLC, and we also describe molecular biomarkers under evaluation that could be available in daily practice to drive ACT in resected NSCLC.

Lung Cancer Computational Biology and Resources

Comprehensive clinical, pathological, and molecular data, when appropriately integrated with advanced computational approaches, are transforming the way we characterize and study lung cancer. Clinically, cancer registry and publicly available historical clinical trial data enable retrospective analyses to examine how socioeconomic factors, patient demographics, and cancer characteristics affect treatment and outcome. Pathologically, digital pathology and artificial intelligence are revolutionizing histopathological image analyses, not only with improved efficiency and accuracy, but also by extracting additional information for prognostication and tumor microenvironment characterization. Genetically and molecularly, individual patient tumors and preclinical models of lung cancer are profiled by various high-throughput platforms to characterize the molecular properties and functional liabilities. The resulting multi-omics data sets and their interrogation facilitate both basic research mechanistic studies and translation of the findings into the clinic. In this review, we provide a list of resources and tools potentially valuable for lung cancer basic and translational research. Importantly, we point out pitfalls and caveats when performing computational analyses of these data sets and provide a vision of future computational biology developments that will aid lung cancer translational research.

Identification of Master Regulators Driving Disease Progression, Relapse, and Drug Resistance in Lung Adenocarcinoma

Backgrounds: Lung cancer is the leading cause of cancer related death worldwide. Current treatment strategies primarily involve surgery, chemotherapy, radiotherapy, targeted therapy, and immunotherapy, determined by TNM stages, histologic types, and genetic profiles. Plenty of studies have been trying to identify robust prognostic gene expression signatures. Even for high performance signatures, they usually have few shared genes. This is not totally unexpected, since a prognostic signature is associated with patient survival and may contain no upstream regulators. Identification of master regulators driving disease progression is a vital step to understand underlying molecular mechanisms and develop new treatments.

Methods: In this study, we have utilized a robust workflow to identify potential master regulators that drive poor prognosis in patients with lung adenocarcinoma. This workflow takes gene expression signatures that are associated with poor survival of early-stage lung adenocarcinoma, EGFR-TKI resistance, and responses to immune checkpoint inhibitors, respectively, and identifies recurrent master regulators from seven public gene expression datasets by a regulatory network-based approach.

Results: We have found that majority of the master regulators driving poor prognosis in early stage LUAD are cell-cycle related according to Gene Ontology annotation. However, they were demonstrated experimentally to promote a spectrum of processes such as tumor cell proliferation, invasion, metastasis, and drug resistance. Master regulators predicted from EGFR-TKI resistance signature and the EMT pathway signature are largely shared, which suggests that EMT pathway functions as a hub and interact with other pathways such as hypoxia, angiogenesis, TNF-α signaling, inflammation, TNF-β signaling, Wnt, and Notch signaling pathways. Master regulators that repress immunotherapy are enriched with MYC targets, E2F targets, oxidative phosphorylation, and mTOR signaling.

Conclusion: Our study uncovered possible mechanisms underlying recurrence, resistance to targeted therapy, and immunotherapy. The predicted master regulators may serve as potential therapeutic targets in patients with lung adenocarcinoma.

Genetic and immunologic features of recurrent stage I lung adenocarcinoma

Although surgery for early-stage lung cancer offers the best chance of cure, recurrence still occurs between 30 and 50% of the time. Why patients frequently recur after complete resection of early-stage lung cancer remains unclear. Using a large cohort of stage I lung adenocarcinoma patients, distinct genetic, genomic, epigenetic, and immunologic profiles of recurrent tumors were analyzed using a novel recurrence classifier. To characterize the tumor immune microenvironment of recurrent stage I tumors, unique tumor-infiltrating immune population markers were identified using single cell RNA-seq on a separate cohort of patients undergoing stage I lung adenocarcinoma resection and applied to a large study cohort using digital cytometry. Recurrent stage I lung adenocarcinomas demonstrated higher mutation and lower methylation burden than non-recurrent tumors, as well as widespread activation of known cancer and cell cycle pathways. Simultaneously, recurrent tumors displayed downregulation of immune response pathways including antigen presentation and Th1/Th2 activation. Recurrent tumors were depleted in adaptive immune populations, and depletion of adaptive immune populations and low cytolytic activity were prognostic of stage I recurrence. Genomic instability and impaired adaptive immune responses are key features of stage I lung adenocarcinoma immunosurveillance escape and recurrence after surgery.

Gene Expression Profiling as a Potential Tool for Precision Oncology in Non-Small Cell Lung Cancer

Recent technological advances and the application of high-throughput mutation and transcriptome analyses have improved our understanding of cancer diseases, including non-small cell lung cancer. For instance, genomic profiling has allowed the identification of mutational events which can be treated with specific agents. However, detection of DNA alterations does not fully recapitulate the complexity of the disease and it does not allow selection of patients that benefit from chemo- or immunotherapy. In this context, transcriptional profiling has emerged as a promising tool for patient stratification and treatment guidance. For instance, transcriptional profiling has proven to be especially useful in the context of acquired resistance to targeted therapies and patients lacking targetable genomic alterations. Moreover, the comprehensive characterization of the expression level of the different pathways and genes involved in tumor progression is likely to better predict clinical benefit from different treatments than single biomarkers such as PD-L1 or tumor mutational burden in the case of immunotherapy. However, intrinsic technical and analytical limitations have hindered the use of these expression signatures in the clinical setting. In this review, we will focus on the data reported on molecular classification of non-small cell lung cancer and discuss the potential of transcriptional profiling as a predictor of survival and as a patient stratification tool to further personalize treatments.

Meta-Analysis of Microdissected Breast Tumors Reveals Genes Regulated in the Stroma but Hidden in Bulk Analysis

Transcriptome data provide a valuable resource for the study of cancer molecular mechanisms, but technical biases, sample heterogeneity, and small sample sizes result in poorly reproducible lists of regulated genes. Additionally, the presence of multiple cellular components contributing to cancer development complicates the interpretation of bulk transcriptomic profiles. To address these issues, we collected 48 microarray datasets derived from laser capture microdissected stroma or epithelium in breast tumors and performed a meta-analysis identifying robust lists of differentially expressed genes. This was used to create a database with carefully harmonized metadata that we make freely available to the research community. As predicted, combining the results of multiple datasets improved statistical power. Moreover, the separate analysis of stroma and epithelium allowed the identification of genes with different contributions in each compartment, which would not be detected by bulk analysis due to their distinct regulation in the two compartments. Our method can be profitably used to help in the discovery of biomarkers and the identification of functionally relevant genes in both the stroma and the epithelium. This database was made to be readily accessible through a user-friendly web interface.

A Novel Immune-Related Seventeen-Gene Signature for Predicting Early Stage Lung Squamous Cell Carcinoma Prognosis

With the increasingly early stage lung squamous cell carcinoma (LUSC) being discovered, there is an urgent need for a comprehensive analysis of the prognostic characteristics of early stage LUSC. Here, we developed an immune-related gene signature for outcome prediction of early stage LUSC based on three independent cohorts. Differentially expressed genes (DEGs) were identified using CIBERSORT and ESTMATE algorithm. Then, a 17-immune-related gene (RPRM, APOH, SSX1, MSGN1, HPR, ISM2, FGA, LBP, HAS1, CSF2, RETN, CCL2, CCL21, MMP19, PTGIS, F13A1, C1QTNF1) signature was identified using univariate Cox regression, LASSO regression and stepwise multivariable Cox analysis based on the verified DEGs from 401 cases in The Cancer Genome Atlas (TCGA) database. Subsequently, a cohort of GSE74777 containing 107 cases downloaded from Gene Expression Omnibus (GEO) database and an independent data set consisting of 36 frozen tissues collected from National Cancer Center were used to validate the predictive value of the signature. Seventeen immune-related genes were identified from TCGA cohort, which were further used to establish a classification system to construct cases into high- and low-risk groups in terms of overall survival. This classifier was still an independent prognostic factor in multivariate analysis. In addition, another two independent cohorts and different clinical subgroups validated the significant predictive value of the signature. Further mechanism research found early stage LUSC patients with high risk had special immune cell infiltration characteristics and gene mutation profiles. In conclusion, we characterized the tumor microenvironment and established a highly predictive model for evaluating the prognosis of early stage LUSC, which may provide a lead for effective immunotherapeutic options tailored for each subtype.

A novel 12-gene signature as independent prognostic model in stage IA and IB lung squamous cell carcinoma patients

BACKGROUND

There is currently no formal consensus on the administration of adjuvant chemotherapy to stage I lung squamous cell carcinoma (LUSC) patients despite the poor prognosis. The side effects of adjuvant chemotherapy need to be balanced against the risk of tumour recurrence. Prognostic markers are thus needed to identify those at higher risks and recommend individualised treatment regimens.

METHODS

Clinical and sequencing data of stage I patients were retrieved from the Lung Squamous Cell Carcinoma project of the Cancer Genome Atlas (TCGA) and three tissue microarray datasets. In a novel K-resample gene selection algorithm, gene-wise Cox proportional hazard regressions were repeated for 50 iterations with random resamples from the TCGA training dataset. The top 200 genes with the best predictive power for survival were chosen to undergo an L1-penalised Cox regression for further gene selection.

RESULTS

A total of 602 samples of LUSC were included, of which 42.2% came from female patients, 45.3% were stage IA cancer. From an initial pool of 11,212 genes in the TCGA training dataset, a final set of 12 genes were selected to construct the multivariate Cox prognostic model. Among the 12 selected genes, 5 genes, STAU1, ADGRF1, ATF7IP2, MALL and KRT23, were adverse prognostic factors for patients, while seven genes, NDUFB1, CNPY2, ZNF394, PIN4, FZD8, NBPF26 and EPYC, were positive prognostic factors. An equation for risk score was thus constructed from the final multivariate Cox model. The model performance was tested in the sequestered TCGA testing dataset and validated in external tissue microarray datasets (GSE4573, GSE31210 and GSE50081), demonstrating its efficacy in stratifying patients into high- and low-risk groups with significant survival difference both in the whole set (including stage IA and IB) and in the stage IA only subgroup of each set. The prognostic power remains significant after adjusting for standard clinical factors. When benchmarked against other prominent gene-signature based prognostic models, the model outperformed the rest in the TCGA testing dataset and in predicting long-term risk at eight years in all three validation datasets.

CONCLUSION

The 12-gene prognostic model may serve as a useful complementary clinical risk-stratification tool for stage I and especially stage IA lung squamous cell carcinoma patients to guide clinical decision making.

Pan-cancer application of a lung-adenocarcinoma-derived gene-expression-based prognostic predictor

Gene-expression profiling can be used to classify human tumors into molecular subtypes or risk groups, representing potential future clinical tools for treatment prediction and prognostication. However, it is less well-known how prognostic gene signatures derived in one malignancy perform in a pan-cancer context. In this study, a gene-rule-based single sample predictor (SSP) called classifier for lung adenocarcinoma molecular subtypes (CLAMS) associated with proliferation was tested in almost 15 000 samples from 32 cancer types to classify samples into better or worse prognosis. Of the 14 malignancies that presented both CLAMS classes in sufficient numbers, survival outcomes were significantly different for breast, brain, kidney and liver cancer. Patients with samples classified as better prognosis by CLAMS were generally of lower tumor grade and disease stage, and had improved prognosis according to other type-specific classifications (e.g. PAM50 for breast cancer). In all, 99.1% of non-lung cancer cases classified as better outcome by CLAMS were comprised within the range of proliferation scores of lung adenocarcinoma cases with a predicted better prognosis by CLAMS. This finding demonstrates the potential of tuning SSPs to identify specific levels of for instance tumor proliferation or other transcriptional programs through predictor training. Together, pan-cancer studies such as this may take us one step closer to understanding how gene-expression-based SSPs act, which gene-expression programs might be important in different malignancies, and how to derive tools useful for prognostication that are efficient across organs.

A Novel Immune-Related Prognostic Model for Response to Immunotherapy and Survival in Patients With Lung Adenocarcinoma

Lung adenocarcinoma is one of the most malignant diseases worldwide. The immune checkpoint inhibitors targeting programmed cell death protein 1 (PD-1) and programmed cell death-ligand 1 (PD-L1) have changed the paradigm of lung cancer treatment; however, there are still patients who are resistant. Further exploration of the immune infiltration status of lung adenocarcinoma (LUAD) is necessary for better clinical management. In our study, the CIBERSORT method was used to calculate the infiltration status of 22 immune cells in LUAD patients from The Cancer Genome Atlas (TCGA). We clustered LUAD based on immune infiltration status by consensus clustering. The differentially expressed genes (DEGs) between cold and hot tumor group were identified. Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysis were performed. Last, we constructed a Cox regression model. We found that the infiltration of M0 macrophage cells and follicular helper T cells predicted an unfavorable overall survival of patients. Consensus clustering of 22 immune cells identified 5 clusters with different patterns of immune cells infiltration, stromal cells infiltration, and tumor purity. Based on the immune scores, we classified these five clusters into hot and cold tumors, which are different in transcription profiles. Hot tumors are enriched in cytokine–cytokine receptor interaction, while cold tumors are enriched in metabolic pathways. Based on the hub genes and prognostic-related genes, we developed a Cox regression model to predict the overall survival of patients with LUAD and validated in other three datasets. In conclusion, we developed an immune-related signature that can predict the prognosis of patients, which might facilitate the clinical application of immunotherapy in LUAD.

The clinical relevance of gene expression based prognostic signatures in colorectal cancer

Colorectal cancer (CRC) is one of the most prevalent cancers, with more than one million new cases every year. In the last few decades, several advancements in therapeutic and preventative levels have reduced the mortality rate, but new biomarkers are required for improved prognosis. The alterations at the genetic and epigenetic level have been recognized as major players in tumorigenesis. The products of gene expression in the form of mRNA, microRNA, and long-noncoding RNA, have started to emerge as important regulatory molecules, playing an important role in cancer. Gene-expression based prognostic risk scores, which quantify and compare their expression, have emerged as promising biomarkers with enormous clinical value. These composite multi-gene models in which more than one gene is used to predict prognosis have been shown to be significantly effective in identifying patients with multiple clinico-pathological risks like overall mortality, response to chemotherapy, risk of metastasis, etc. The advent of microarray and advanced sequencing technologies have led to the generation of large datasets like TCGA (The Cancer Genome Atlas) and GEO (Gene Expression Omnibus), which have fueled the search for new biomarkers. Continuous evaluation of these candidate biomarkers in clinical settings is promising to improve the management of CRC. These composite gene signatures provide potential in identifying high-risk patients, which might help clinicians to better manage these patients and design appropriate personalized therapeutic interventions. In this review, we emphasize on composite prognostic scores from diverse resources with clinical utility in CRC.

A multi-scale integrated analysis identifies KRT8 as a pan-cancer early biomarker

An early biomarker would transform our ability to screen and treat patients with cancer. The large amount of multi-scale molecular data in public repositories from various cancers provide unprecedented opportunities to find such a biomarker. However, despite identification of numerous molecular biomarkers using these public data, fewer than 1% have proven robust enough to translate into clinical practice. One of the most important factors affecting the successful translation to clinical practice is lack of real-world patient population heterogeneity in the discovery process. Almost all biomarker studies analyze only a single cohort of patients with the same cancer using a single modality. Recent studies in other diseases have demonstrated the advantage of leveraging biological and technical heterogeneity across multiple independent cohorts to identify robust disease biomarkers. Here we analyzed 17149 samples from patients with one of 23 cancers that were profiled using either DNA methylation, bulk and single-cell gene expression, or protein expression in tumor and serum. First, we analyzed DNA methylation profiles of 9855 samples across 23 cancers from The Cancer Genome Atlas (TCGA). We then examined the gene expression profile of the most significantly hypomethylated gene, KRT8, in 6781 samples from 57 independent microarray datasets from NCBI GEO. KRT8 was significantly over-expressed across cancers except colon cancer (summary effect size=1.05; p < 0.0001). Further, single-cell RNAseq analysis of 7447 single cells from lung tumors showed that genes that significantly correlated with KRT8 (p < 0.05) were involved in p53-related pathways. Immunohistochemistry in tumor biopsies from 294 patients with lung cancer showed that high protein expression of KRT8 is a prognostic marker of poor survival (HR = 1.73, p = 0.01). Finally, detectable KRT8 in serum as measured by ELISA distinguished patients with pancreatic cancer from healthy controls with an AUROC=0.94. In summary, our analysis demonstrates that KRT8 is (1) differentially expressed in several cancers across all molecular modalities and (2) may be useful as a biomarker to identify patients that should be further tested for cancer.

Sex-Biased Molecular Signature for Overall Survival of Liver Cancer Patients

Sex/gender disparity has been shown in the incidence and prognosis of many types of diseases, probably due to differences in genes, physiological conditions such as hormones, and lifestyle between the sexes. The mortality and survival rates of many cancers, especially liver cancer, differ between men and women. Due to the pronounced sex/gender disparity, considering sex/gender may be necessary for the diagnosis and treatment of liver cancer. By analyzing research articles through a PubMed literature search, the present review identified 12 genes which showed practical relevance to cancer and sex disparities. Among the 12 sex-specific genes, 7 genes (BAP1, CTNNB1, FOXA1, GSTO1, GSTP1, IL6, and SRPK1) showed sex-biased function in liver cancer. Here we summarized previous findings of cancer molecular signature including our own analysis, and showed that sex-biased molecular signature CTNNB1^High, IL6^High, RHOA^High and GLIPR1^Low may serve as a female-specific index for prediction and evaluation of OS in liver cancer patients. This review suggests a potential implication of sex-biased molecular signature in liver cancer, providing a useful information on diagnosis and prediction of disease progression based on gender.

Transcriptome Analysis Identifies Novel Prognostic Genes in Osteosarcoma

Osteosarcoma (OS), a malignant primary bone tumor often seen in young adults, is highly aggressive. The improvements in high-throughput technologies have accelerated the identification of various prognostic biomarkers for cancer survival prediction. However, only few studies focus on the prediction of prognosis in OS patients using gene expression data due to small sample size and the lack of public datasets. In the present study, the RNA-seq data of 82 OS samples, along with their clinical information, were collected from the TARGET database. To identify the prognostic genes for the OS survival prediction, we selected the top 50 genes of contribution as the initial candidate genes of the prognostic risk model, which were ranked by random forest model, and found that the prognostic model with five predictors including CD180, MYC, PROSER2, DNAI1, and FATE1 was the optimal multivariable Cox regression model. Moreover, based on a multivariable Cox regression model, we also developed a scoring method and stratified the OS patients into groups of different risks. The stratification for OS patients in the validation set further demonstrated that our model has a robust performance. In addition, we also investigated the biological function of differentially expressed genes between two risk groups and found that those genes were mainly involved with biological pathways and processes regarding immunity. In summary, the identification of novel prognostic biomarkers in OS would greatly assist the prediction of OS survival and development of molecularly targeted therapies, which in turn benefit patients' survival.

Development and validation of a robust immune-related prognostic signature in early-stage lung adenocarcinoma

Background

The incidence of stage I and stage II lung adenocarcinoma (LUAD) is likely to increase with the introduction of annual screening programs for high-risk individuals. We aimed to identify a reliable prognostic signature with immune-related genes that can predict prognosis and help making individualized management for patients with early-stage LUAD.

Methods

The public LUAD cohorts were obtained from the large-scale databases including 4 microarray data sets from the Gene Expression Omnibus (GEO) and 1 RNA-seq data set from The Cancer Genome Atlas (TCGA) LUAD cohort. Only early-stage patients with clinical information were included. Cox proportional hazards regression model was performed to identify the candidate prognostic genes in GSE30219, GSE31210 and GSE50081 (training set). The prognostic signature was developed using the overlapped prognostic genes based on a risk score method. Kaplan–Meier curve with log-rank test and time-dependent receiver operating characteristic (ROC) curve were used to evaluate the prognostic value and performance of this signature, respectively. Furthermore, the robustness of this prognostic signature was further validated in TCGA-LUAD and GSE72094 cohorts.

Results

A prognostic immune signature consisting of 21 immune-related genes was constructed using the training set. The prognostic signature significantly stratified patients into high- and low-risk groups in terms of overall survival (OS) in training data set, including GSE30219 (HR = 4.31, 95% CI 2.29–8.11; P = 6.16E−06), GSE31210 (HR = 11.91, 95% CI 4.15–34.19; P = 4.10E−06), GSE50081 (HR = 3.63, 95% CI 1.90–6.95; P = 9.95E−05), the combined data set (HR = 3.15, 95% CI 1.98–5.02; P = 1.26E−06) and the validation data set, including TCGA-LUAD (HR = 2.16, 95% CI 1.49–3.13; P = 4.54E−05) and GSE72094 (HR = 2.95, 95% CI 1.86–4.70; P = 4.79E−06). Multivariate cox regression analysis demonstrated that the 21-gene signature could serve as an independent prognostic factor for OS after adjusting for other clinical factors. ROC curves revealed that the immune signature achieved good performance in predicting OS for early-stage LUAD. Several biological processes, including regulation of immune effector process, were enriched in the immune signature. Moreover, the combination of the signature with tumor stage showed more precise classification for prognosis prediction and treatment design.

Conclusions

Our study proposed a robust immune-related prognostic signature for estimating overall survival in early-stage LUAD, which may be contributed to make more accurate survival risk stratification and individualized clinical management for patients with early-stage LUAD.

Transcriptomics and solid tumors: The next frontier in precision cancer medicine

Transcriptomics, which encompasses assessments of alternative splicing and alternative polyadenylation, identification of fusion transcripts, explorations of noncoding RNAs, transcript annotation, and discovery of novel transcripts, is a valuable tool for understanding cancer mechanisms and identifying biomarkers. Recent advances in high-throughput technologies have enabled large-scale gene expression profiling. Importantly, RNA expression profiling of tumor tissue has been successfully used to determine clinically actionable molecular alterations. The WINTHER precision medicine clinical trial was the first prospective trial in diverse solid malignancies that assessed both genomics and transcriptomics to match treatments to specific molecular alterations. The use of transcriptome analysis in WINTHER and other trials increased the number of targetable -omic changes compared to genomic profiling alone. Other applications of transcriptomics involve the evaluation of tumor and circulating noncoding RNAs as predictive and prognostic biomarkers, the improvement of risk stratification by the use of prognostic and predictive multigene assays, the identification of fusion transcripts that drive tumors, and an improved understanding of the impact of DNA changes as some genomic alterations are silenced at the RNA level. Finally, RNA sequencing and gene expression analysis have been incorporated into clinical trials to identify markers predicting response to immunotherapy. Many issues regarding the complexity of the analysis, its reproducibility and variability, and the interpretation of the results still need to be addressed. The integration of transcriptomics with genomics, proteomics, epigenetics, and tumor immune profiling will improve biomarker discovery and our understanding of disease mechanisms and, thereby, accelerate the implementation of precision oncology.

Computational Staining of Pathology Images to Study the Tumor Microenvironment in Lung Cancer

The spatial organization of different types of cells in tumor tissues reveals important information about the tumor microenvironment (TME). To facilitate the study of cellular spatial organization and interactions, we developed Histology-based Digital-Staining, a deep learning-based computation model, to segment the nuclei of tumor, stroma, lymphocyte, macrophage, karyorrhexis, and red blood cells from standard hematoxylin and eosin-stained pathology images in lung adenocarcinoma. Using this tool, we identified and classified cell nuclei and extracted 48 cell spatial organization-related features that characterize the TME. Using these features, we developed a prognostic model from the National Lung Screening Trial dataset, and independently validated the model in The Cancer Genome Atlas lung adenocarcinoma dataset, in which the predicted high-risk group showed significantly worse survival than the low-risk group (P = 0.001), with a HR of 2.23 (1.37-3.65) after adjusting for clinical variables. Furthermore, the image-derived TME features significantly correlated with the gene expression of biological pathways. For example, transcriptional activation of both the T-cell receptor and programmed cell death protein 1 pathways positively correlated with the density of detected lymphocytes in tumor tissues, while expression of the extracellular matrix organization pathway positively correlated with the density of stromal cells. In summary, we demonstrate that the spatial organization of different cell types is predictive of patient survival and associated with the gene expression of biological pathways. SIGNIFICANCE: These findings present a deep learning-based analysis tool to study the TME in pathology images and demonstrate that the cell spatial organization is predictive of patient survival and is associated with gene expression.See related commentary by Rodriguez-Antolin, p. 1912.

PenDA, a rank-based method for personalized differential analysis: Application to lung cancer

The hopes of precision medicine rely on our capacity to measure various high-throughput genomic information of a patient and to integrate them for personalized diagnosis and adapted treatment. Reaching these ambitious objectives will require the development of efficient tools for the detection of molecular defects at the individual level. Here, we propose a novel method, PenDA, to perform Personalized Differential Analysis at the scale of a single sample. PenDA is based on the local ordering of gene expressions within individual cases and infers the deregulation status of genes in a sample of interest compared to a reference dataset. Based on realistic simulations of RNA-seq data of tumors, we showed that PenDA outcompetes existing approaches with very high specificity and sensitivity and is robust to normalization effects. Applying the method to lung cancer cohorts, we observed that deregulated genes in tumors exhibit a cancer-type-specific commitment towards up- or down-regulation. Based on the individual information of deregulation given by PenDA, we were able to define two new molecular histologies for lung adenocarcinoma cancers strongly correlated to survival. In particular, we identified 37 biomarkers whose up-regulation lead to bad prognosis and that we validated on two independent cohorts. PenDA provides a robust, generic tool to extract personalized deregulation patterns that can then be used for the discovery of therapeutic targets and for personalized diagnosis. An open-access, user-friendly R package is available at https://github.com/bcm-uga/penda.

The hopes of precision medicine rely on our capacity to measure individual molecular information for personalized diagnosis and treatment. These challenging perspectives will be only possible with the development of efficient methodological tools to identify patient-specific molecular defects from the many precise molecular information that one can access at the single-individual, single tissue or even single-cell levels. Such methods will provide a better understanding of disease-specific biological mechanisms and will promote the development of personalized therapeutic strategies. Here we describe a novel method, named PenDA, to perform differential analysis of gene expression at the individual level. Based on a realistic benchmark of simulated tumors, we demonstrated that PenDA reaches very high efficiency in detecting sample-specific deregulated genes. We then applied the method to two large cohorts associated with lung cancer. A detailed statistical analysis of the results allowed to isolate genes with specific deregulation patterns, like genes that are up-regulated in all tumors or genes that are expressed but never deregulated in any tumors. Given their specificities, these genes are likely to be of interest in therapeutic research. In particular, we were able to identified 37 new biomarkers associated to bad prognosis that we validated on two independent cohorts.

Immunoscore Predicts Survival in Early-Stage Lung Adenocarcinoma Patients

Background: The lung cancer staging system is insufficient for a comprehensive evaluation of patient prognosis. We constructed a novel immunoscore model to predict patients with high risk and poor survival.

Method: Immunoscore was developed based on z-score transformed enrichment score of 11 immune-related gene sets of 109 immune risk genes. The immunoscore model was trained in lung adenocarcinoma cohort from The Cancer Genome Atlas (TCGA-LUAD) (n = 400), and validated in other two independent cohorts from Gene Expression Omnibus (GEO), GSE31210 (n = 219) and GSE68465 (n = 356). Meta-set (n = 975) was formed by combining all training and testing sets.

Result: High immunoscore conferred worse prognosis in all sets. It was an independent prognostic factors in multivariate Cox analysis in training, testing and meta-set [hazard ratio (HR) = 2.96 (2.24–3.9), P < 0.001 in training set; HR = 1.99 (1.21–3.26), P = 0.006 in testing set 1; HR = 1.48 (1.69–2.39), P = 0.005 in testing set 2; HR = 2.01 (1.69–2.39), P < 0.001 in meta-set]. Immunoscore-clinical prognostic signature (ICPS) was developed by integrating immunoscore and clinical characteristic, and had higher C-index than immunoscore or stage alone in all sets [0.72 (ICPS) vs. 0.7 (immunoscore) or 0.59 (stage) in training set; 0.75 vs. 0.72 or 0.7 in testing set 1; 0.65 vs. 0.61 or 0.62 in testing set 2; 0.7 vs. 0.66 or 0.64 in meta-set]. Genome analysis revealed that immunoscore was positively correlated with tumor mutation burden (R = 0.22, P < 0.001). Besides, high immunoscore was correlated with high proportion of carcinoma-associated fibroblasts (R = 0.32, P < 0.001) in tumor microenvironment but fewer CD8+ cells infiltration (R = −0.28, P < 0.001).

Conclusion: The immunoscore and ICPS are potential biomarkers for evaluating patient survival. Further investigations are required to validate and improve their prediction accuracy.

Reference genome and annotation updates lead to contradictory prognostic predictions in gene expression signatures: a case study of resected stage I lung adenocarcinoma

RNA-sequencing enables accurate and low-cost transcriptome-wide detection. However, expression estimates vary as reference genomes and gene annotations are updated, confounding existing expression-based prognostic signatures. Herein, prognostic 9-gene pair signature (GPS) was applied to 197 patients with stage I lung adenocarcinoma derived from previous and latest data from The Cancer Genome Atlas (TCGA) processed with different reference genomes and annotations. For 9-GPS, 6.6% of patients exhibited discordant risk classifications between the two TCGA versions. Similar results were observed for other prognostic signatures, including IRGPI, 15-gene and ORACLE. We found that conflicting annotations for gene length and overlap were the major cause of their discordant risk classification. Therefore, we constructed a prognostic 40-GPS based on stable genes across GENCODE v20-v30 and validated it using public data of 471 stage I samples (log-rank P < 0.0010). Risk classification was still stable in RNA-sequencing data processed with the newest GENCODE v32 versus GENCODE v20-v30. Specifically, 40-GPS could predict survival for 30 stage I samples with formalin-fixed paraffin-embedded tissues (log-rank P = 0.0177). In conclusion, this method overcomes the vulnerability of existing prognostic signatures due to reference genome and annotation updates. 40-GPS may offer individualized clinical applications due to its prognostic accuracy and classification stability.

An Immune-Related Signature Predicts Survival in Patients With Lung Adenocarcinoma

We investigated the local immune status and its prognostic value in lung adenocarcinoma. In total, 513 lung adenocarcinoma samples from TCGA and ImmPort databases were collected and analyzed. The R package coxph was employed to mine immune-related genes that were significant prognostic indicators using both univariate and multivariate analyses. The R software package glmnet was then used for Lasso Cox regression analysis, and a prognosis prediction model was constructed for lung adenocarcinoma; clusterProfiler was selected for functional gene annotations and KEGG enrichment analysis. Finally, correlations between the RiskScore and clinical features or signaling pathways were established. Sixty-four immune-related genes remarkably correlated with patient prognosis and were further applied. Samples were hierarchically clustered into two subgroups. Accordingly, the LASSO regression algorithm was employed to screen the 14 most representative immune-related genes (PSMD11, PPIA, MIF, BMP5, DKK1, PDGFB, ANGPTL4, IL1R2, THRB, LTBR, TNFRSF1, TNFRSF17, IL20RB, and MC1R) with respect to patient prognosis. Then, the prognosis prediction model for lung adenocarcinoma patients (namely, the RiskScore equation) was constructed, and the training set samples were incorporated to evaluate the efficiency of this model to predict and classify patient prognosis. Subsequently, based on functional annotations and KEGG pathway analysis, the 14 immune-related genes were mainly enriched in pathways closely associated with lung adenocarcinoma and its immune microenvironment, such as cytokine–cytokine receptor interaction and human T-cell leukemia virus 1 infection. Furthermore, correlations between the RiskScore and clinical features of the training set samples and signaling pathways (such as p53, cell cycle, and DNA repair) were also demonstrated. Finally, the test set sample data were employed for independent testing and verifying the model. We established a prognostic prediction RiskScore model based on the expression profiles of 14 immune-related genes, which shows high prediction accuracy and stability in identifying immune features. This could provide clinical guidance for the diagnosis and prognosis of different immunophenotypes, and suggest multiple targets for precise advanced lung adenocarcinoma therapy based on subtype-specific immune molecules.

A clonal expression biomarker associates with lung cancer mortality

An aim of molecular biomarkers is to stratify patients with cancer into disease subtypes predictive of outcome, improving diagnostic precision beyond clinical descriptors such as tumor stage1. Transcriptomic intratumor heterogeneity (RNA-ITH) has been shown to confound existing expression-based biomarkers across multiple cancer types2-6. Here, we analyze multi-region whole-exome and RNA sequencing data for 156 tumor regions from 48 patients enrolled in the TRACERx study to explore and control for RNA-ITH in non-small cell lung cancer. We find that chromosomal instability is a major driver of RNA-ITH, and existing prognostic gene expression signatures are vulnerable to tumor sampling bias. To address this, we identify genes expressed homogeneously within individual tumors that encode expression modules of cancer cell proliferation and are often driven by DNA copy-number gains selected early in tumor evolution. Clonal transcriptomic biomarkers overcome tumor sampling bias, associate with survival independent of clinicopathological risk factors, and may provide a general strategy to refine biomarker design across cancer types.

Molecular profiling of transcription factors pinpoints MYC-estrogen related receptor α-regulatory factor X5 panel for characterizing the immune microenvironment and predicting the efficacy of immune checkpoint inhibitors in renal cell carcinoma

Transcription factors (TFs) play key roles in biological processes, and previous studies revealed that they can control oncogenic processes. However, the functional impact of TFs on the prognosis of patients with cancer has not been extensively elucidated. In the context of The Cancer Genome Atlas, few studies have focused on the roles of TFs in tumorigenesis. In the present study, a TF-based robust MYC-estrogen related receptor α-regulatory factor X5 (MYC-ESRRA-RFX5) signature was developed for predicting the survival of patients with renal cell carcinoma. Functional enrichment analysis of this signature revealed that it was associated with the immune system of these patients. Further analysis demonstrated that this panel could characterize the immune microenvironment and potentially predicts the effectiveness of immune checkpoint inhibitors. Therefore, the present study recommends future exploration on TF-based biomarkers for their potential as prognostic predictors. Overall, the highlights of this study are: i) This novel study pinpoints a TF panel for the robust prediction of renal cell carcinoma prognosis, and ii) the MYC-ESRRA-RFX5 panel is proposed as a signature for characterizing the immune microenvironment, and to potentially predict the effectiveness of immune checkpoint inhibitors.

Risk stratification for lung adenocarcinoma on EGFR and TP53 mutation status, chemotherapy, and PD-L1 immunotherapy

The overall survival rates for lung cancer remain unsatisfactorily low, even for patients with biomarkers for which target therapies or immunotherapies are recommended. Better identification of at‐risk patients is needed to achieve more effective personalized treatment. Here, we derived a risk‐stratifying gene signature consisting of five genes that had the greatest differential expression by stage from lung adenocarcinoma (LUAD) transcriptomes. The new gene signature enabled survival prognosis for multiple LUAD datasets from different platforms of transcriptomics and risk stratification for patients with and without a mutation in TP53 or EGFR, with high and low levels of PD‐L1, and with and without adjuvant chemotherapy treatment. Using these evaluations, it was also shown to be more robust compared to several other gene signatures. Functional analysis of the five genes and their protein‐protein interaction partners indicated that they are functionally enriched in cell cycle, endocytosis, and EGFR regulation, which are biological processes associated with lung cancer and drug resistance. Extensive discussions on related experimental studies suggest that the five genes are novel and sensible targets for developing new drugs and/or tackling drug resistance problems for LUAD.

Somatic Alteration Burden Involving Non-Cancer Genes Predicts Prognosis in Early-Stage Non-Small Cell Lung Cancer

The burden of somatic mutations and neoantigens has been associated with improved survival in cancer treated with immunotherapies, especially non-small cell lung cancer (NSCLC). However, there is uncertainty about their effect on outcome in early-stage untreated cases. We posited that the burden of mutations in a specific set of genes may also contribute to the prognosis of early NSCLC patients. From a small cohort of 36 NSCLC cases, we were able to identify somatic mutations and copy number alterations in 865 genes that contributed to patient overall survival. Simply, the number of altered genes (NAG) among these 865 genes was associated with longer disease-free survival (hazard ratio (HR) = 0.153, p = 1.48 × 10^-4). The gene expression signature distinguishing patients with high/low NAG was also prognostic in three independent datasets. Patients with a high NAG could be further stratified based on the presence of immunogenic mutations, revealing a further subgroup of stage I NSCLC with even better prognosis (85% with >5 years survival), and associated with cytotoxic T-cell expression. Importantly, 95% of the highly-altered genes lacked direct relation to cancer, but were implicated in pathways regulating cell proliferation, motility and immune response.

Loss of FGL1 induces epithelial‑mesenchymal transition and angiogenesis in LKB1 mutant lung adenocarcinoma

Liver kinase b1 (LKB1) is a tumor suppressor, and the inactivated mutation frequency of LKB1 in lung adenocarcinoma is ~20%. The present study aimed to explore potential novel biomarkers in LKB1 mutant lung adenocarcinoma. Gene expression data from lung adenocarcinoma patients were downloaded from The Cancer Genome Atlas and the Gene Expression Omnibus databases. R software was used to analyze the gene expression profiles. Reverse transcription‑quantitative PCR (RT‑qPCR), western blot and immunohistochemistry (IHC) analyses were used to examine gene expression and function. Gene function was further explored via gene set enrichment analysis. A colony formation assay was used to evaluate cell proliferation. A wound‑healing assay and immunofluorescence analysis were used to evaluate cell migration and epithelial‑mesenchymal transition (EMT), respectively. Wound healing assay, immunofluorescence, western blot, RT‑qPCR and IHC results for EMT‑associated markers demonstrated that a loss of fibrinogen‑like 1 (FGL1) induced EMT in LKB1 mutant lung adenocarcinoma. RT‑qPCR and IHC analyses of angiogenesis‑related markers revealed that loss of FGL1 promoted angiogenesis in LKB1 mutant lung adenocarcinoma. Overall, the present results demonstrated that loss of FGL1 induced EMT and angiogenesis in LKB1 mutant lung adenocarcinoma. FGL1 may be a novel biomarker to indicate EMT and angiogenesis in patients with LKB1 mutant lung adenocarcinoma.

NETest Liquid Biopsy Is Diagnostic of Lung Neuroendocrine Tumors and Identifies Progressive Disease

BACKGROUND

There are no effective biomarkers for the management of bronchopulmonary carcinoids (BPC). We examined the utility of a neuroendocrine multigene transcript "liquid biopsy" (NETest) in BPC for diagnosis and monitoring of the disease status.

AIM

To independently validate the utility of the NETest in diagnosis and management of BPC in a multicenter, multinational, blinded study.

MATERIAL AND METHODS

The study cohorts assessed were BPC (n = 99), healthy controls (n = 102), other lung neoplasia (n = 101) including adenocarcinomas (ACC) (n = 41), squamous cell carcinomas (SCC) (n = 37), small-cell lung cancer (SCLC) (n = 16), large-cell neuroendocrine carcinoma (LCNEC) (n = 7), and idiopathic pulmonary fibrosis (IPF) (n = 50). BPC were histologically classified as typical (TC) (n = 62) and atypical carcinoids (AC) (n = 37). BPC disease status determination was based on imaging and RECIST 1.1. NETest diagnostic metrics and disease status accuracy were evaluated. The upper limit of normal (NETest) was 20. Twenty matched tissue-blood pairs were also evaluated. Data are means ± SD.

RESULTS

NETest levels were significantly increased in BPC (45 ± 25) versus controls (9 ± 8; p < 0.0001). The area under the ROC curve was 0.96 ± 0.01. Accuracy, sensitivity, and specificity were: 92, 84, and 100%. NETest was also elevated in SCLC (42 ± 32) and LCNEC (28 ± 7). NETest accurately distinguished progressive (61 ± 26) from stable disease (35.5 ± 18; p < 0.0001). In BPC, NETest levels were elevated in metastatic disease irrespective of histology (AC: p < 0.02; TC: p = 0.0006). In nonendocrine lung cancers, ACC (18 ± 21) and SCC (12 ± 11) and benign disease (IPF) (18 ± 25) levels were significantly lower compared to BPC level (p < 0.001). Significant correlations were evident between paired tumor and blood samples for BPC (R: 0.83, p < 0.0001) and SCLC (R: 0.68) but not for SCC and ACC (R: 0.25-0.31).

CONCLUSIONS

Elevated -NETest levels are indicative of lung neuroendocrine neoplasia. NETest levels correlate with tumor tissue and imaging and accurately define clinical progression.

Tumor-based gene expression biomarkers to predict survival following curative intent resection for stage I lung adenocarcinoma

Background

Prognostic biomarkers are needed in clinical setting to predict outcome after resection for early-stage lung adenocarcinoma. The goal of this study is to validate tumor-based single-gene expression biomarkers with demonstrated prognostic value in order to move them along the clinical translation pipeline.

Methods

Prognostic genes were selected from the literature and the best candidates measured by quantitative real-time polymerase chain reaction (qPCR) in tumors of 233 patients with stage I adenocarcinoma. Significant prognostic genes were then validated in an independent set of 210 patients matching the first set in terms of histology, stage, and clinical data.

Results

Eleven genes with demonstrated prognostic value were selected from the literature. Complementary analyses in public databases and our own microarray dataset led to the investigation of six genes associated with good (BTG2, SELENBP1 and NFIB) or poor outcome (RRM1, EZH2 and FOXM1). In the first set of patients, EZH2 and RRM1 were significantly associated with better survival on top of age, sex and pathological stage (EZH2 p = 3.2e-02, RRM1 p = 5.9e-04). The prognostic values of EZH2 and RRM1 were not replicated in the second set of patients. A trend was observed for both genes in the joint analyses (n = 443) with higher expression associated with worse outcome.

Conclusion

Adenocarcinoma-specific mRNA expression levels of EZH2 and RRM1 are associated with poor post-surgical survival in the first set of patients, but not replicated in a clinically and pathologically matched independent validation set. This study highlights challenges associated with clinical translation of prognostic biomarkers.

TUSC3 accelerates cancer growth and induces epithelial-mesenchymal transition by upregulating claudin-1 in non-small-cell lung cancer cells

Lung cancer is the most frequent cause of cancer-related deaths worldwide, but its molecular pathogenesis is poorly understood. The tumor suppressor candidate 3 (TUSC3) gene is located on chromosome 8p22 and is universally acknowledged as a cancer suppressor. However, our research has demonstrated that TUSC3 expression is significantly upregulated in non-small-cell lung cancer compared to benign controls. In this study, we analyzed the consequences of TUSC3 knockdown or overexpression on the biological functions of non-small-cell lung cancer cell lines. To identify the molecules and signaling pathways with which TUSC3 might interact, we completed immunoblotting, quantitative polymerase chain reaction, microarray, co-immunoprecipitation, and immunofluorescence assays. We demonstrated that TUSC3 knockdown leads to decreased proliferation, migration, and invasion, and reduced xenograft tumor growth of non-small-cell lung cancer cell lines, whereas opposite results were observed with overexpression of TUSC3. In addition, TUSC3 knockdown suppressed epithelial-mesenchymal transition by downregulating the expression of claudin-1, which plays an indispensable role in EMT progress. On the contrary, overexpression of TUSC3 significantly enhanced EMT progress by upregulating claudin-1 expression. Overall, our observations suggest that TUSC3 accelerates cancer growth and induces the epithelial-mesenchymal transition in non-small-cell lung cancer cells; we also identified claudin-1 as a target of TUSC3.

The expression level of CSDAP1 in lung cancer and its clinical significance

Expression level of messenger RNA (mRNA) of cold shock domain protein A intronless pseudogene (CSDAP1) in lung cancer tissues was studied. Fresh pathological specimens collected from 317 patients with primary lung cancer through surgical resection from January 2007 to January 2012 were selected. Reverse transcription-polymerase chain reaction (RT-PCR) and western blot analysis were used to detect the transcription and translation of CSDAP1 in lung cancer tissues and cancer-adjacent normal tissues, and the results were analyzed in combination with clinicopathological features and prognosis of lung cancer. Among 317 lung cancer specimens, 105 cases (33.1%) had high expression of CSDAP1. Among 138 cases of pulmonary adenocarcinoma, 59 cases had high expression of CSDAP1, and the high expression rate was 42.8%. Among 170 cases of lung squamous cell carcinoma, 46 cases had high expression of CSDAP1, and the high expression rate was 27.5% (P<0.05). Three cases of large cell carcinoma and 9 cases of small cell carcinoma had extremely low expression or had no expression of CSDAP1. Among the 127 lung cancer patients with regional lymph node metastasis, 53 cases (41.7%) had high expression of CSDAP1, while among the 190 lung cancer patients without regional lymph node metastasis, 52 cases (27.4%) had high expression of CSDAP1 (P<0.05). The results also revealed that the expression of CSDAP1 was also related to tumor-node-metastasis (TNM) staging of lung cancer. One-year, three-year and five-year survival rates of lung cancer patients who had no expression of CSDAP1 were relatively high (P<0.05). The results suggested that CSDAP1 may play an important role in the occurrence, development and judgement of prognosis of lung cancer.

A novel protein-based prognostic signature improves risk stratification to guide clinical management in early-stage lung adenocarcinoma patients

Each of the pathological stages (I-IIIa) of surgically resected non-small-cell lung cancer has hidden biological heterogeneity, manifested as heterogeneous outcomes within each stage. Thus, the finding of robust and precise molecular classifiers with which to assess individual patient risk is an unmet medical need. Here, we identified and validated the clinical utility of a new prognostic signature based on three proteins (BRCA1, QKI, and SLC2A1) to stratify early-stage lung adenocarcinoma patients according to their risk of recurrence or death. Patients were staged according to the new International Association for the Study of Lung Cancer (IASLC) staging criteria (8th edition, 2018). A test cohort (n = 239) was used to assess the value of this new prognostic index (PI) based on the three proteins. The prognostic signature was developed by Cox regression with the use of stringent statistical criteria (TRIPOD: Transparent reporting of a multivariable prediction model for individual prognosis or diagnosis). The model resulted in a highly significant predictor of 5-year outcome for disease-free survival (p < 0.001) and overall survival (p < 0.001). The prognostic ability of the model was externally validated in an independent multi-institutional cohort of patients (n = 114, p = 0.021). We also demonstrated that this molecular classifier adds relevant information to the gold standard TNM-based pathological staging, with a highly significant improvement of the likelihood ratio. We subsequently developed a combined PI including both the molecular and the pathological data that improved the risk stratification in both cohorts (p ≤ 0.001). Moreover, the signature may help to select stage I-IIA patients who might benefit from adjuvant chemotherapy. In summary, this protein-based signature accurately identifies those patients with a high risk of recurrence and death, and adds further prognostic information to the TNM-based clinical staging, even when the new IASLC 8th edition staging criteria are applied. More importantly, it may be a valuable tool for selecting patients for adjuvant therapy. Copyright © 2018 Pathological Society of Great Britain and Ireland. Published by John Wiley & Sons, Ltd.

Histology-dependent prognostic role of pERK and p53 protein levels in early-stage non-small cell lung cancer

Lung tumors represent a major health problem. In early stage NSCLC tumors, surgical resection is the preferred treatment, but 30-55% of patients will relapse within 5 years after surgery. Thus, the identification of prognostic biomarkers in early stage NSCLC patients, especially those which are therapeutically addressable, is crucial to enhance survival of these patients. We determined the immunohistochemistry expression of key proteins involved in tumorigenesis and oncogenic signaling, p53, EGFR, pAKT and pERK, and correlated their expression level to clinicopathological characteristics and patient outcome. We found EGFR expression is higher in the squamous cell carcinomas than in adenocarcinomas (p=0.043), and that nuclear p53 staining correlated with lower differentiated squamous tumors (p=0.034). Regarding the prognostic potential of the expression of these proteins, high pERK levels proved to be an independent prognostic factor for overall (p<0.001) and progression-free survival (p<0.001) in adenocarcinoma patients, but not in those from the squamous histology, and high p53 nuclear levels were identified as independent prognostic factor for progression-free survival (p=0.031) only in squamous cell carcinoma patients. We propose a role as early prognostic biomarkers for pERK protein levels in adenocarcinoma, and for nuclear p53 levels in squamous cell lung carcinoma. The determination of these potential biomarkers in the adequate histologic context may predict the outcome of early stage NSCLC patients, and may offer a therapeutic opportunity to enhance survival of these patients.

Prognostic Model for Resected Squamous Cell Lung Cancer: External Multicenter Validation and Propensity Score Analysis exploring the Impact of Adjuvant and Neoadjuvant Treatment

INTRODUCTION

We developed one of the first clinicopathological prognostic nomograms for resected squamous cell lung cancer (SQLC). Herein, we validate the model in a larger multicenter cohort and we explore the impact of adjuvant and neoadjuvant treatment (ANT).

METHODS

Patients with resected SQLC from January 2002 to December 2012 in six institutions were eligible. Each patient was assigned a prognostic score based on the clinicopathological factors included in the model (age, T descriptor according to seventh edition of the TNM classification, lymph node status, and grading). Kaplan-Meier analysis for disease-free survival, cancer-specific survival (CSS), and overall survival was performed according to a three-class risk model. Harrell's C-statistics were adopted for model validation. The effect of ANT was adjusted with propensity score.

RESULTS

Data on 1375 patients were gathered (median age, 68 years; male sex, 86.8%; T descriptor 1 or 2 versus 3 or 4, 71.7% versus 24.9%; nodes negative versus positive, 53.4% versus 46.6%; and grading of 1 or 2 versus 3, 35.0% versus 41.1%). Data for survival analysis were available for 1097 patients. With a median follow-up of 55 months, patients at low risk had a significantly longer disease-free survival than did patients at intermediate risk (hazard ratio [HR] = 1.67, 95% confidence interval [CI]: 1.40-2.01) and patients at high risk (HR = 2.46, 95% CI: 1.90-3.19); they also had a significantly longer CSS (HR = 2.46, 95% CI: 1.80-3.36 versus HR = 4.30, 95% CI: 2.92-6.33) and overall survival (HR = 1.79, 95% CI: 1.48-2.17 versus HR = 2.33, 95% CI: 1.76-3.07). A trend in favor of ANT was observed for intermediate-risk/high-risk patients, particularly for CSS (p = 0.06 [5-year CSS 72.7% versus 60.8%]).

CONCLUSIONS

A model based on a combination of easily available clinicopathological factors effectively stratifies patients with resected SQLC into three risk classes.