Graph Perceiver Network for Lung Tumor and Bronchial Premalignant Lesion Stratification from Histopathology.

Bronchial premalignant lesions (PMLs) precede the development of invasive lung squamous cell carcinoma (LUSC), posing a significant challenge in distinguishing those likely to advance to LUSC from those that might regress without intervention. This study followed a novel computational approach, the Graph Perceiver Network, leveraging hematoxylin and eosin-stained whole slide images to stratify endobronchial biopsies of PMLs across a spectrum from normal to tumor lung tissues. The Graph Perceiver Network outperformed existing frameworks in classification accuracy predicting LUSC, lung adenocarcinoma, and nontumor lung tissue on The Cancer Genome Atlas and Clinical Proteomic Tumor Analysis Consortium datasets containing lung resection tissues while efficiently generating pathologist-aligned, class-specific heatmaps. The network was further tested using endobronchial biopsies from two data cohorts, containing normal to carcinoma in situ histology. It demonstrated a unique capability to differentiate carcinoma in situ lung squamous PMLs based on their progression status to invasive carcinoma. The network may have utility in stratifying PMLs for chemoprevention trials or more aggressive follow-up.

Immune evasion before tumour invasion in early lung squamous carcinogenesis.

Early detection and treatment are critical for improving the outcome of patients with cancer1. Understanding the largely uncharted biology of carcinogenesis requires deciphering molecular processes in premalignant lesions, and revealing the determinants of the intralesional immune reaction during cancer development. The adaptive immune response within tumours has previously been shown to be strongest at the earliest stage of carcinoma2,3. Here we show that immune activation and immune escape occur before tumour invasion, and reveal the relevant immune biomarkers of the pre-invasive stages of carcinogenesis in the lung. We used gene-expression profiling and multispectral imaging to analyse a dataset of 9 morphological stages of the development of lung squamous cell carcinoma, which includes 122 well-annotated biopsies from 77 patients. We identified evolutionary trajectories of cancer and immune pathways that comprise (1) a linear increase in proliferation and DNA repair from normal to cancerous tissue; (2) a transitory increase of metabolism and early immune sensing, through the activation of resident immune cells, in low-grade pre-invasive lesions; (3) the activation of immune responses and immune escape through immune checkpoints and suppressive interleukins from high-grade pre-invasive lesions; and, ultimately, (4) the activation of the epithelial-mesenchymal transition in the invasive stage of cancer. We propose that carcinogenesis in the lung involves a dynamic co-evolution of pre-invasive bronchial cells and the immune response. These findings highlight the need to develop immune biomarkers for early detection as well as immunotherapy-based chemopreventive approaches for individuals who are at high risk of developing lung cancer.

Premalignant Progression in the Lung: Knowledge Gaps and Novel Opportunities for Interception of Non-Small Cell Lung Cancer. An Official American Thoracic Society Research Statement.

Rationale: Despite significant advances in precision treatments and immunotherapy, lung cancer is the most common cause of cancer death worldwide. To reduce incidence and improve survival rates, a deeper understanding of lung premalignancy and the multistep process of tumorigenesis is essential, allowing timely and effective intervention before cancer development. Objectives: To summarize existing information, identify knowledge gaps, formulate research questions, prioritize potential research topics, and propose strategies for future investigations into the premalignant progression in the lung. Methods: An international multidisciplinary team of basic, translational, and clinical scientists reviewed available data to develop and refine research questions pertaining to the transformation of premalignant lung lesions to advanced lung cancer. Results: This research statement identifies significant gaps in knowledge and proposes potential research questions aimed at expanding our understanding of the mechanisms underlying the progression of premalignant lung lesions to lung cancer in an effort to explore potential innovative modalities to intercept lung cancer at its nascent stages. Conclusions: The identified gaps in knowledge about the biological mechanisms of premalignant progression in the lung, together with ongoing challenges in screening, detection, and early intervention, highlight the critical need to prioritize research in this domain. Such focused investigations are essential to devise effective preventive strategies that may ultimately decrease lung cancer incidence and improve patient outcomes.

Aberrant epithelial polarity cues drive the development of precancerous airway lesions.

Molecular events that drive the development of precancerous lesions in the bronchial epithelium, which are precursors of lung squamous cell carcinoma (LUSC), are poorly understood. We demonstrate that disruption of epithelial cellular polarity, via the conditional deletion of the apical determinant Crumbs3 (Crb3), initiates and sustains precancerous airway pathology. The loss of Crb3 in adult luminal airway epithelium promotes the uncontrolled activation of the transcriptional regulators YAP and TAZ, which stimulate intrinsic signals that promote epithelial cell plasticity and paracrine signals that induce basal-like cell growth. We show that aberrant polarity and YAP/TAZ-regulated gene expression associates with human bronchial precancer pathology and disease progression. Analyses of YAP/TAZ-regulated genes further identified the ERBB receptor ligand Neuregulin-1 (NRG1) as a key transcriptional target and therapeutic targeting of ERBB receptors as a means of preventing and treating precancerous cell growth. Our observations offer important molecular insight into the etiology of LUSC and provides directions for potential interception strategies of lung cancer.

Genomic approaches to accelerate cancer interception.

Although major advances have been reported in the last decade in the treatment of late-stage cancer with targeted and immune-based therapies, there is a crucial unmet need to develop new approaches to improve the prevention and early detection of cancer. Advances in genomics and computational biology offer unprecedented opportunities to understand the earliest molecular events associated with carcinogenesis, enabling novel strategies to intercept the development of invasive cancers. This Series paper will highlight emerging big data genomic approaches with the potential to accelerate advances in cancer prevention, screening, and early detection across various tumour types, and the challenges inherent in the development of these tools for clinical use. Through coordinated multicentre consortia, these genomic approaches are likely to transform the landscape of cancer interception in the coming years.

Assessment of microRNA differential expression and detection in multiplexed small RNA sequencing data.

Small RNA sequencing can be used to gain an unprecedented amount of detail into the microRNA transcriptome. The relatively high cost and low throughput of sequencing bases technologies can potentially be offset by the use of multiplexing. However, multiplexing involves a trade-off between increased number of sequenced samples and reduced number of reads per sample (i.e., lower depth of coverage). To assess the effect of different sequencing depths owing to multiplexing on microRNA differential expression and detection, we sequenced the small RNA of lung tissue samples collected in a clinical setting by multiplexing one, three, six, nine, or 12 samples per lane using the Illumina HiSeq 2000. As expected, the numbers of reads obtained per sample decreased as the number of samples in a multiplex increased. Furthermore, after normalization, replicate samples included in distinct multiplexes were highly correlated (R > 0.97). When detecting differential microRNA expression between groups of samples, microRNAs with average expression >1 reads per million (RPM) had reproducible fold change estimates (signal to noise) independent of the degree of multiplexing. The number of microRNAs detected was strongly correlated with the log2 number of reads aligning to microRNA loci (R = 0.96). However, most additional microRNAs detected in samples with greater sequencing depth were in the range of expression which had lower fold change reproducibility. These findings elucidate the trade-off between increasing the number of samples in a multiplex with decreasing sequencing depth and will aid in the design of large-scale clinical studies exploring microRNA expression and its role in disease.

Experimental evolution of an RNA virus in wild birds: evidence for host-dependent impacts on population structure and competitive fitness.

Within hosts, RNA viruses form populations that are genetically and phenotypically complex. Heterogeneity in RNA virus genomes arises due to error-prone replication and is reduced by stochastic and selective mechanisms that are incompletely understood. Defining how natural selection shapes RNA virus populations is critical because it can inform treatment paradigms and enhance control efforts. We allowed West Nile virus (WNV) to replicate in wild-caught American crows, house sparrows and American robins to assess how natural selection shapes RNA virus populations in ecologically relevant hosts that differ in susceptibility to virus-induced mortality. After five sequential passages in each bird species, we examined the phenotype and population diversity of WNV through fitness competition assays and next generation sequencing. We demonstrate that fitness gains occur in a species-specific manner, with the greatest replicative fitness gains in robin-passaged WNV and the least in WNV passaged in crows. Sequencing data revealed that intrahost WNV populations were strongly influenced by purifying selection and the overall complexity of the viral populations was similar among passaged hosts. However, the selective pressures that control WNV populations seem to be bird species-dependent. Specifically, crow-passaged WNV populations contained the most unique mutations (~1.7× more than sparrows, ~3.4× more than robins) and defective genomes (~1.4× greater than sparrows, ~2.7× greater than robins), but the lowest average mutation frequency (about equal to sparrows, ~2.6× lower than robins). Therefore, our data suggest that WNV replication in the most disease-susceptible bird species is positively associated with virus mutational tolerance, likely via complementation, and negatively associated with the strength of selection. These differences in genetic composition most likely have distinct phenotypic consequences for the virus populations. Taken together, these results reveal important insights into how different hosts may contribute to the emergence of RNA viruses.

Integrated Genomics Reveals Convergent Transcriptomic Networks Underlying Chronic Obstructive Pulmonary Disease and Idiopathic Pulmonary Fibrosis.

RATIONALE: Despite shared environmental exposures, idiopathic pulmonary fibrosis (IPF) and chronic obstructive pulmonary disease are usually studied in isolation, and the presence of shared molecular mechanisms is unknown. OBJECTIVES: We applied an integrative genomic approach to identify convergent transcriptomic pathways in emphysema and IPF. METHODS: We defined the transcriptional repertoire of chronic obstructive pulmonary disease, IPF, or normal histology lungs using RNA-seq (n = 87). MEASUREMENTS AND MAIN RESULTS: Genes increased in both emphysema and IPF relative to control were enriched for the p53/hypoxia pathway, a finding confirmed in an independent cohort using both gene expression arrays and the nCounter Analysis System (n = 193). Immunohistochemistry confirmed overexpression of HIF1A, MDM2, and NFKBIB members of this pathway in tissues from patients with emphysema or IPF. Using reads aligned across splice junctions, we determined that alternative splicing of p53/hypoxia pathway-associated molecules NUMB and PDGFA occurred more frequently in IPF or emphysema compared with control and validated these findings by quantitative polymerase chain reaction and the nCounter Analysis System on an independent sample set (n = 193). Finally, by integrating parallel microRNA and mRNA-Seq data on the same samples, we identified MIR96 as a key novel regulatory hub in the p53/hypoxia gene-expression network and confirmed that modulation of MIR96 in vitro recapitulates the disease-associated gene-expression network. CONCLUSIONS: Our results suggest convergent transcriptional regulatory hubs in diseases as varied phenotypically as chronic obstructive pulmonary disease and IPF and suggest that these hubs may represent shared key responses of the lung to environmental stresses.

Peritraumatic Distress Mediates the Effect of Severity of Disaster Exposure on Perinatal Depression: The Iowa Flood Study.

Disaster exposure during pregnancy has received limited attention. This study examined the impact of the 2008 Iowa Floods on perinatal maternal depression and well-being, and the role of peritraumatic distress as a possible mechanism explaining this link. Perinatal women (N = 171) completed measures of depressive symptoms and general well-being at 5 timepoints from pregnancy to 30 months postpartum. Objectively assessed prenatal flood exposure was associated with greater depression (r = .15). Further, flood-related peritraumatic distress was uniquely associated with greater depression (r = .23), and was a key mechanism through which flood exposure led to depression. Prenatal flood exposure was also associated with general well-being (r = .18); however, a mechanism other than peritraumatic distress appears to have been responsible for the effect of flood exposure on well-being. We discuss the implications of these findings for informing etiological models and enhancing the efficacy of interventions for maternal psychopathology.

Airway epithelial gene expression in the diagnostic evaluation of smokers with suspect lung cancer.

Lung cancer is the leading cause of death from cancer in the US and the world. The high mortality rate (80-85% within 5 years) results, in part, from a lack of effective tools to diagnose the disease at an early stage. Given that cigarette smoke creates a field of injury throughout the airway, we sought to determine if gene expression in histologically normal large-airway epithelial cells obtained at bronchoscopy from smokers with suspicion of lung cancer could be used as a lung cancer biomarker. Using a training set (n = 77) and gene-expression profiles from Affymetrix HG-U133A microarrays, we identified an 80-gene biomarker that distinguishes smokers with and without lung cancer. We tested the biomarker on an independent test set (n = 52), with an accuracy of 83% (80% sensitive, 84% specific), and on an additional validation set independently obtained from five medical centers (n = 35). Our biomarker had approximately 90% sensitivity for stage 1 cancer across all subjects. Combining cytopathology of lower airway cells obtained at bronchoscopy with the biomarker yielded 95% sensitivity and a 95% negative predictive value. These findings indicate that gene expression in cytologically normal large-airway epithelial cells can serve as a lung cancer biomarker, potentially owing to a cancer-specific airway-wide response to cigarette smoke.

A computed tomography-based score indicative of lung cancer aggression (SILA) predicts lung adenocarcinomas with low malignant potential or vascular invasion.

BACKGROUND: Histologic grading of lung adenocarcinoma (LUAD) is predictive of outcome but is only possible after surgical resection. A radiomic biomarker predictive of grade has the potential to improve preoperative management of early-stage LUAD. OBJECTIVE: Validate a prognostic radiomic score indicative of lung cancer aggression (SILA) in surgically resected stage I LUAD (n= 161) histologically graded as indolent low malignant potential (LMP), intermediate, or aggressive vascular invasive (VI) subtypes. METHODS: The SILA scores were generated from preoperative CT-scans using the previously validated Computer-Aided Nodule Assessment and Risk Yield (CANARY) software. RESULTS: Cox proportional regression showed significant association between the SILA and 7-year recurrence-free survival (RFS) in a univariate (p< 0.05) and multivariate (p< 0.05) model incorporating age, gender, smoking status, pack years, and extent of resection. The SILA was positively correlated with invasive size (spearman r= 0.54, p= 8.0 × 10 - 14) and negatively correlated with percentage of lepidic histology (spearman r=-0.46, p= 7.1 × 10 - 10). The SILA predicted indolent LMP with an area under the receiver operating characteristic (ROC) curve (AUC) of 0.74 and aggressive VI with an AUC of 0.71, the latter remaining significant when invasive size was included as a covariate in a logistic regression model (p< 0.01). CONCLUSIONS: The SILA scoring of preoperative CT scans was prognostic and predictive of resected pathologic grade.

Graph Attention-Based Fusion of Pathology Images and Gene Expression for Prediction of Cancer Survival.

Multimodal machine learning models are being developed to analyze pathology images and other modalities, such as gene expression, to gain clinical and biological insights. However, most frameworks for multimodal data fusion do not fully account for the interactions between different modalities. Here, we present an attention-based fusion architecture that integrates a graph representation of pathology images with gene expression data and concomitantly learns from the fused information to predict patient-specific survival. In our approach, pathology images are represented as undirected graphs, and their embeddings are combined with embeddings of gene expression signatures using an attention mechanism to stratify tumors by patient survival. We show that our framework improves the survival prediction of human non-small cell lung cancers, outperforming existing state-of-the-art approaches that leverage multimodal data. Our framework can facilitate spatial molecular profiling to identify tumor heterogeneity using pathology images and gene expression data, complementing results obtained from more expensive spatial transcriptomic and proteomic technologies.

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.

A Lung Cancer Mouse Model Database.

Lung cancer, the leading cause of cancer mortality, exhibits diverse histological subtypes and genetic complexities. Numerous preclinical mouse models have been developed to study lung cancer, but data from these models are disparate, siloed, and difficult to compare in a centralized fashion. Here we established the Lung Cancer Mouse Model Database (LCMMDB), an extensive repository of 1,354 samples from 77 transcriptomic datasets covering 974 samples from genetically engineered mouse models (GEMMs), 368 samples from carcinogen-induced models, and 12 samples from a spontaneous model. Meticulous curation and collaboration with data depositors have produced a robust and comprehensive database, enhancing the fidelity of the genetic landscape it depicts. The LCMMDB aligns 859 tumors from GEMMs with human lung cancer mutations, enabling comparative analysis and revealing a pressing need to broaden the diversity of genetic aberrations modeled in GEMMs. Accompanying this resource, we developed a web application that offers researchers intuitive tools for in-depth gene expression analysis. With standardized reprocessing of gene expression data, the LCMMDB serves as a powerful platform for cross-study comparison and lays the groundwork for future research, aiming to bridge the gap between mouse models and human lung cancer for improved translational relevance.

Interrogating the precancerous evolution of pathway dysfunction in lung squamous cell carcinoma using XTABLE.

Lung squamous cell carcinoma (LUSC) is a type of lung cancer with a dismal prognosis that lacks adequate therapies and actionable targets. This disease is characterized by a sequence of low- and high-grade preinvasive stages with increasing probability of malignant progression. Increasing our knowledge about the biology of these premalignant lesions (PMLs) is necessary to design new methods of early detection and prevention, and to identify the molecular processes that are key for malignant progression. To facilitate this research, we have designed XTABLE (Exploring Transcriptomes of Bronchial Lesions), an open-source application that integrates the most extensive transcriptomic databases of PMLs published so far. With this tool, users can stratify samples using multiple parameters and interrogate PML biology in multiple manners, such as two- and multiple-group comparisons, interrogation of genes of interests, and transcriptional signatures. Using XTABLE, we have carried out a comparative study of the potential role of chromosomal instability scores as biomarkers of PML progression and mapped the onset of the most relevant LUSC pathways to the sequence of LUSC developmental stages. XTABLE will critically facilitate new research for the identification of early detection biomarkers and acquire a better understanding of the LUSC precancerous stages.

Lung squamous cell carcinoma is the second most common lung cancer. However, very little is known about how normal tissues in the lung develop in to these tumours. Like many cancers, this transformation comprises of an intermediate phase where healthy cells begin to form lesions that may (or may not) progress in to tumours. Understanding the biology of these lesions in lung squamous cell carcinoma may help clinicians detect them before they become cancerous. Knowing which genes are switched on and off during this intermediary phase can provide clues as to how these lesions form. There are already some publicly available transcriptional datasets showing the activity of tens of thousands of genes in pre-cancerous lesions extracted from patients with lung squamous cell carcinoma. But not every laboratory has the bioinformatic tools and skills required to interrogate these extensive databases. To address this, Roberts et al. built an open-source platform called XTABLE (short for Exploring Transcriptomes of Bronchial Lesions) which can analyse transcriptional datasets in multiple ways depending on the needs of the user. For instance, the tool can stratify the data into groups based on different parameters, such as the lesions potential to progress in to cancer, to see how the genes of the groups compare. It can also analyse the activity of individual genes and sets of genes involved in the same biological processes. Using XTABLE, Roberts et al. showed that a biological process linked to lung squamous cell carcinoma is also involved in the formation of pre-cancerous lesions. This suggests that molecules and genes associated with this process could potentially help scientists design prevention strategies. XTABLE will help researchers to better understand the biology of pre-cancerous lesions and how they develop in to tumours. Moreover, it will make it easier for scientists to validate their hypotheses using data collected from patients. The tool could also be useful for scientists interested in other types of lung cancers that share a similar biology.

Convergence of YAP/TAZ, TEAD and TP63 activity is associated with bronchial premalignant severity and progression.

BACKGROUND: Bronchial premalignant lesions (PMLs) are composed primarily of cells resembling basal epithelial cells of the airways, which through poorly understood mechanisms have the potential to progress to lung squamous cell carcinoma (LUSC). Despite ongoing efforts that have mapped gene expression and cell diversity across bronchial PML pathologies, signaling and transcriptional events driving malignancy are poorly understood. Evidence has suggested key roles for the Hippo pathway effectors YAP and TAZ and associated TEAD and TP63 transcription factor families in bronchial basal cell biology and LUSC. In this study we examine the functional association of YAP/TAZ, TEADs and TP63 in bronchial epithelial cells and PMLs. METHODS: We performed RNA-seq in primary human bronchial epithelial cells following small interfering RNA (siRNA)-mediated depletion of YAP/TAZ, TEADs or TP63, and combined these data with ChIP-seq analysis of these factors. Directly activated or repressed genes were identified and overlapping genes were profiled across gene expression data obtained from progressive or regressive human PMLs and across lung single cell RNA-seq data sets. RESULTS: Analysis of genes regulated by YAP/TAZ, TEADs, and TP63 in human bronchial epithelial cells revealed a converged transcriptional network that is strongly associated with the pathological progression of bronchial PMLs. Our observations suggest that YAP/TAZ-TEAD-TP63 associate to cooperatively promote basal epithelial cell proliferation and repress signals associated with interferon responses and immune cell communication. Directly repressed targets we identified include the MHC Class II transactivator CIITA, which is repressed in progressive PMLs and associates with adaptive immune responses in the lung. Our findings provide molecular insight into the control of gene expression events driving PML progression, including those contributing to immune evasion, offering potential new avenues for lung cancer interception. CONCLUSIONS: Our study identifies important gene regulatory functions for YAP/TAZ-TEAD-TP63 in the early stages of lung cancer development, which notably includes immune-suppressive roles, and suggest that an assessment of the activity of this transcriptional complex may offer a means to identify immune evasive bronchial PMLs and serve as a potential therapeutic target.

Integrative genetic and genomic networks identify microRNA associated with COPD and ILD.

Chronic obstructive pulmonary disease (COPD) and interstitial lung disease (ILD) are clinically and molecularly heterogeneous diseases. We utilized clustering and integrative network analyses to elucidate roles for microRNAs (miRNAs) and miRNA isoforms (isomiRs) in COPD and ILD pathogenesis. Short RNA sequencing was performed on 351 lung tissue samples of COPD (n = 145), ILD (n = 144) and controls (n = 64). Five distinct subclusters of samples were identified including 1 COPD-predominant cluster and 2 ILD-predominant clusters which associated with different clinical measurements of disease severity. Utilizing 262 samples with gene expression and SNP microarrays, we built disease-specific genetic and expression networks to predict key miRNA regulators of gene expression. Members of miR-449/34 family, known to promote airway differentiation by repressing the Notch pathway, were among the top connected miRNAs in both COPD and ILD networks. Genes associated with miR-449/34 members in the disease networks were enriched among genes that increase in expression with airway differentiation at an air-liquid interface. A highly expressed isomiR containing a novel seed sequence was identified at the miR-34c-5p locus. 47% of the anticorrelated predicted targets for this isomiR were distinct from the canonical seed sequence for miR-34c-5p. Overexpression of the canonical miR-34c-5p and the miR-34c-5p isomiR with an alternative seed sequence down-regulated NOTCH1 and NOTCH4. However, only overexpression of the isomiR down-regulated genes involved in Ras signaling such as CRKL and GRB2. Overall, these findings elucidate molecular heterogeneity inherent across COPD and ILD patients and further suggest roles for miR-34c in regulating disease-associated gene-expression.

Single-Cell Characterization of Pulmonary Nodules Implicates Suppression of Immunosurveillance across Early Stages of Lung Adenocarcinoma.

A greater understanding of molecular, cellular, and immunological changes during the early stages of lung adenocarcinoma development could improve diagnostic and therapeutic approaches in patients with pulmonary nodules at risk for lung cancer. To elucidate the immunopathogenesis of early lung tumorigenesis, we evaluated surgically resected pulmonary nodules representing the spectrum of early lung adenocarcinoma as well as associated normal lung tissues using single-cell RNA sequencing and validated the results by flow cytometry and multiplex immunofluorescence (MIF). Single-cell transcriptomics revealed a significant decrease in gene expression associated with cytolytic activities of tumor-infiltrating natural killer and natural killer T cells. This was accompanied by a reduction in effector T cells and an increase of CD4+ regulatory T cells (Treg) in subsolid nodules. An independent set of resected pulmonary nodules consisting of both adenocarcinomas and associated premalignant lesions corroborated the early increment of Tregs in premalignant lesions compared with the associated normal lung tissues by MIF. Gene expression analysis indicated that cancer-associated alveolar type 2 cells and fibroblasts may contribute to the deregulation of the extracellular matrix, potentially affecting immune infiltration in subsolid nodules through ligand-receptor interactions. These findings suggest that there is a suppression of immune surveillance across the spectrum of early-stage lung adenocarcinoma. SIGNIFICANCE: Analysis of a spectrum of subsolid pulmonary nodules by single-cell RNA sequencing provides insights into the immune regulation and cell-cell interactions in the tumor microenvironment during early lung tumor development.

A Graph-Transformer for Whole Slide Image Classification.

Deep learning is a powerful tool for whole slide image (WSI) analysis. Typically, when performing supervised deep learning, a WSI is divided into small patches, trained and the outcomes are aggregated to estimate disease grade. However, patch-based methods introduce label noise during training by assuming that each patch is independent with the same label as the WSI and neglect overall WSI-level information that is significant in disease grading. Here we present a Graph-Transformer (GT) that fuses a graph-based representation of an WSI and a vision transformer for processing pathology images, called GTP, to predict disease grade. We selected 4,818 WSIs from the Clinical Proteomic Tumor Analysis Consortium (CPTAC), the National Lung Screening Trial (NLST), and The Cancer Genome Atlas (TCGA), and used GTP to distinguish adenocarcinoma (LUAD) and squamous cell carcinoma (LSCC) from adjacent non-cancerous tissue (normal). First, using NLST data, we developed a contrastive learning framework to generate a feature extractor. This allowed us to compute feature vectors of individual WSI patches, which were used to represent the nodes of the graph followed by construction of the GTP framework. Our model trained on the CPTAC data achieved consistently high performance on three-label classification (normal versus LUAD versus LSCC: mean accuracy = 91.2 ± 2.5%) based on five-fold cross-validation, and mean accuracy = 82.3 ± 1.0% on external test data (TCGA). We also introduced a graph-based saliency mapping technique, called GraphCAM, that can identify regions that are highly associated with the class label. Our findings demonstrate GTP as an interpretable and effective deep learning framework for WSI-level classification.

Inactivation of LATS1/2 drives luminal-basal plasticity to initiate basal-like mammary carcinomas.

Basal-like breast cancers, an aggressive breast cancer subtype that has poor treatment options, are thought to arise from luminal mammary epithelial cells that undergo basal plasticity through poorly understood mechanisms. Using genetic mouse models and ex vivo primary organoid cultures, we show that conditional co-deletion of the LATS1 and LATS2 kinases, key effectors of Hippo pathway signaling, in mature mammary luminal epithelial cells promotes the development of Krt14 and Sox9-expressing basal-like carcinomas that metastasize over time. Genetic co-deletion experiments revealed that phenotypes resulting from the loss of LATS1/2 activity are dependent on the transcriptional regulators YAP/TAZ. Gene expression analyses of LATS1/2-deleted mammary epithelial cells notably revealed a transcriptional program that associates with human basal-like breast cancers. Our study demonstrates in vivo roles for the LATS1/2 kinases in mammary epithelial homeostasis and luminal-basal fate control and implicates signaling networks induced upon the loss of LATS1/2 activity in the development of basal-like breast cancer.