A Bioinformatics Tool for Identifying Intratumoral Microbes from the ORIEN Dataset.

Evidence supports significant interactions among microbes, immune cells, and tumor cells in at least 10%-20% of human cancers, emphasizing the importance of further investigating these complex relationships. However, the implications and significance of tumor-related microbes remain largely unknown. Studies have demonstrated the critical roles of host microbes in cancer prevention and treatment responses. Understanding interactions between host microbes and cancer can drive cancer diagnosis and microbial therapeutics (bugs as drugs). Computational identification of cancer-specific microbes and their associations is still challenging due to the high dimensionality and high sparsity of intratumoral microbiome data, which requires large datasets containing sufficient event observations to identify relationships, and the interactions within microbial communities, the heterogeneity in microbial composition, and other confounding effects that can lead to spurious associations. To solve these issues, we present a bioinformatics tool, microbial graph attention (MEGA), to identify the microbes most strongly associated with 12 cancer types. We demonstrate its utility on a dataset from a consortium of nine cancer centers in the Oncology Research Information Exchange Network. This package has three unique features: species-sample relations are represented in a heterogeneous graph and learned by a graph attention network; it incorporates metabolic and phylogenetic information to reflect intricate relationships within microbial communities; and it provides multiple functionalities for association interpretations and visualizations. We analyzed 2,704 tumor RNA sequencing samples and MEGA interpreted the tissue-resident microbial signatures of each of 12 cancer types. MEGA can effectively identify cancer-associated microbial signatures and refine their interactions with tumors. SIGNIFICANCE: Studying the tumor microbiome in high-throughput sequencing data is challenging because of the extremely sparse data matrices, heterogeneity, and high likelihood of contamination. We present a new deep learning tool, MEGA, to refine the organisms that interact with tumors.

Transcriptional state dynamics lead to heterogeneity and adaptive tumor evolution in urothelial bladder carcinoma.

Intra-tumor heterogeneity contributes to treatment failure and poor survival in urothelial bladder carcinoma (UBC). Analyzing transcriptome from a UBC cohort, we report that intra-tumor transcriptomic heterogeneity indicates co-existence of tumor cells in epithelial and mesenchymal-like transcriptional states and bi-directional transition between them occurs within and between tumor subclones. We model spontaneous and reversible transition between these partially heritable states in cell lines and characterize their population dynamics. SMAD3, KLF4 and PPARG emerge as key regulatory markers of the transcriptional dynamics. Nutrient limitation, as in the core of large tumors, and radiation treatment perturb the dynamics, initially selecting for a transiently resistant phenotype and then reconstituting heterogeneity and growth potential, driving adaptive evolution. Dominance of transcriptional states with low PPARG expression indicates an aggressive phenotype in UBC patients. We propose that phenotypic plasticity and dynamic, non-genetic intra-tumor heterogeneity modulate both the trajectory of disease progression and adaptive treatment response in UBC.

A bioinformatics tool for identifying intratumoral microbes from the ORIEN dataset.

Evidence supports significant interactions among microbes, immune cells, and tumor cells in at least 10-20% of human cancers, emphasizing the importance of further investigating these complex relationships. However, the implications and significance of tumor-related microbes remain largely unknown. Studies have demonstrated the critical roles of host microbes in cancer prevention and treatment responses. Understanding interactions between host microbes and cancer can drive cancer diagnosis and microbial therapeutics (bugs as drugs). Computational identification of cancer-specific microbes and their associations is still challenging due to the high dimensionality and high sparsity of intratumoral microbiome data, which requires large datasets containing sufficient event observations to identify relationships, and the interactions within microbial communities, the heterogeneity in microbial composition, and other confounding effects that can lead to spurious associations. To solve these issues, we present a bioinformatics tool, MEGA, to identify the microbes most strongly associated with 12 cancer types. We demonstrate its utility on a dataset from a consortium of 9 cancer centers in the Oncology Research Information Exchange Network (ORIEN). This package has 3 unique features: species-sample relations are represented in a heterogeneous graph and learned by a graph attention network; it incorporates metabolic and phylogenetic information to reflect intricate relationships within microbial communities; and it provides multiple functionalities for association interpretations and visualizations. We analyzed 2704 tumor RNA-seq samples and MEGA interpreted the tissue-resident microbial signatures of each of 12 cancer types. MEGA can effectively identify cancer-associated microbial signatures and refine their interactions with tumors.

Integrated metabolic and genetic analysis reveals distinct features of human differentiated thyroid cancer.

BACKGROUND: Differentiated thyroid cancer (DTC) affects thousands of lives worldwide each year. Typically, DTC is a treatable disease with a good prognosis. Yet, some patients are subjected to partial or total thyroidectomy and radioiodine therapy to prevent local disease recurrence and metastasis. Unfortunately, thyroidectomy and/or radioiodine therapy often worsen(s) quality of life and might be unnecessary in indolent DTC cases. On the other hand, the lack of biomarkers indicating a potential metastatic thyroid cancer imposes an additional challenge to managing and treating patients with this disease. AIM: The presented clinical setting highlights the unmet need for a precise molecular diagnosis of DTC and potential metastatic disease, which should dictate appropriate therapy. MATERIALS AND METHODS: In this article, we present a differential multi-omics model approach, including metabolomics, genomics, and bioinformatic models, to distinguish normal glands from thyroid tumours. Additionally, we are proposing biomarkers that could indicate potential metastatic diseases in papillary thyroid cancer (PTC), a sub-class of DTC. RESULTS: Normal and tumour thyroid tissue from DTC patients had a distinct yet well-defined metabolic profile with high levels of anabolic metabolites and/or other metabolites associated with the energy maintenance of tumour cells. The consistency of the DTC metabolic profile allowed us to build a bioinformatic classification model capable of clearly distinguishing normal from tumor thyroid tissues, which might help diagnose thyroid cancer. Moreover, based on PTC patient samples, our data suggest that elevated nuclear and mitochondrial DNA mutational burden, intra-tumour heterogeneity, shortened telomere length, and altered metabolic profile reflect the potential for metastatic disease. DISCUSSION: Altogether, this work indicates that a differential and integrated multi-omics approach might improve DTC management, perhaps preventing unnecessary thyroid gland removal and/or radioiodine therapy. CONCLUSIONS: Well-designed, prospective translational clinical trials will ultimately show the value of this integrated multi-omics approach and early diagnosis of DTC and potential metastatic PTC.

SOX9 drives KRAS-induced lung adenocarcinoma progression and suppresses anti-tumor immunity.

The SOX9 transcription factor ensures proper tissue development and homeostasis and has been implicated in promoting tumor progression. However, the role of SOX9 as a driver of lung adenocarcinoma (LUAD), or any cancer, remains unclear. Using CRISPR/Cas9 and Cre-LoxP gene knockout approaches in the KrasG12D-driven mouse LUAD model, we found that loss of Sox9 significantly reduces lung tumor development, burden and progression, contributing to significantly longer overall survival. SOX9 consistently drove organoid growth in vitro, but SOX9-promoted tumor growth was significantly attenuated in immunocompromised mice compared to syngeneic mice. We demonstrate that SOX9 suppresses immune cell infiltration and functionally suppresses tumor associated CD8+ T, natural killer and dendritic cells. These data were validated by flow cytometry, gene expression, RT-qPCR, and immunohistochemistry analyses in KrasG12D-driven murine LUAD, then confirmed by interrogating bulk and single-cell gene expression repertoires and immunohistochemistry in human LUAD. Notably, SOX9 significantly elevates collagen-related gene expression and substantially increases collagen fibers. We propose that SOX9 increases tumor stiffness and inhibits tumor-infiltrating dendritic cells, thereby suppressing CD8+ T cell and NK cell infiltration and activity. Thus, SOX9 drives KrasG12D-driven lung tumor progression and inhibits anti-tumor immunity at least partly by modulating the tumor microenvironment.

Integrated metabolic and genetic analysis reveals distinct features of primary differentiated thyroid cancer and its metastatic potential in humans.

Differentiated thyroid cancer (DTC) affects thousands of lives worldwide every year. Typically, DTC is a treatable disease with a good prognosis. Yet, some patients are subjected to partial or total thyroidectomy and radioiodine therapy to prevent local disease recurrence and metastasis. Unfortunately, thyroidectomy and/or radioiodine therapy often worsen(s) the quality of life and might be unnecessary in indolent DTC cases. This clinical setting highlights the unmet need for a precise molecular diagnosis of DTC, which should dictate appropriate therapy. Here we propose a differential multi-omics model approach to distinguish normal gland from thyroid tumor and to indicate potential metastatic diseases in papillary thyroid cancer (PTC), a sub-class of DTC. Based on PTC patient samples, our data suggest that elevated nuclear and mitochondrial DNA mutational burden, intratumor heterogeneity, shortened telomere length, and altered metabolic profile reflect the potential for metastatic disease. Specifically, normal and tumor thyroid tissues from these patients had a distinct yet well-defined metabolic profile with high levels of anabolic metabolites and/or other metabolites associated with the energy maintenance of tumor cells. Altogether, this work indicates that a differential and integrated multi-omics approach might improve DTC management, perhaps preventing unnecessary thyroid gland removal and/or radioiodine therapy. Well-designed, prospective translational clinical trials will ultimately show the value of this targeted molecular approach. TRANSLATIONAL RELEVANCE: In this article, we propose a new integrated metabolic, genomic, and cytopathologic methods to diagnose Differentiated Thyroid Cancer when the conventional methods failed. Moreover, we suggest metabolic and genomic markers to help predict high-risk Papillary Thyroid Cancer. Both might be important tools to avoid unnecessary surgery and/or radioiodine therapy that can worsen the quality of life of the patients more than living with an indolent Thyroid nodule.

Truncated FGFR2 is a clinically actionable oncogene in multiple cancers.

Somatic hotspot mutations and structural amplifications and fusions that affect fibroblast growth factor receptor 2 (encoded by FGFR2) occur in multiple types of cancer1. However, clinical responses to FGFR inhibitors have remained variable1-9, emphasizing the need to better understand which FGFR2 alterations are oncogenic and therapeutically targetable. Here we apply transposon-based screening10,11 and tumour modelling in mice12,13, and find that the truncation of exon 18 (E18) of Fgfr2 is a potent driver mutation. Human oncogenomic datasets revealed a diverse set of FGFR2 alterations, including rearrangements, E1-E17 partial amplifications, and E18 nonsense and frameshift mutations, each causing the transcription of E18-truncated FGFR2 (FGFR2ΔE18). Functional in vitro and in vivo examination of a compendium of FGFR2ΔE18 and full-length variants pinpointed FGFR2-E18 truncation as single-driver alteration in cancer. By contrast, the oncogenic competence of FGFR2 full-length amplifications depended on a distinct landscape of cooperating driver genes. This suggests that genomic alterations that generate stable FGFR2ΔE18 variants are actionable therapeutic targets, which we confirmed in preclinical mouse and human tumour models, and in a clinical trial. We propose that cancers containing any FGFR2 variant with a truncated E18 should be considered for FGFR-targeted therapies.

Inference on spatial heterogeneity in tumor microenvironment using spatial transcriptomics data.

In the tumor microenvironment (TME), functional interactions among tumor, immune, and stromal cells and the extracellular matrix play key roles in tumor progression, invasion, immune modulation, and response to treatment. Intratumor heterogeneity is ubiquitous not only at the genetic and transcriptomic levels but also in the composition and characteristics of TME. However, quantitative inference on spatial heterogeneity in the TME is still limited. Here, we propose a framework to use network graph-based spatial statistical models on spatially annotated molecular data to gain insights into modularity and spatial heterogeneity in the TME. Applying the framework to spatial transcriptomics data from pancreatic ductal adenocarcinoma samples, we observed significant global and local spatially correlated patterns in the abundance score of tumor cells; in contrast, immune cell types showed dispersed patterns in the TME. Hypoxia, EMT, and inflammation signatures contributed to intra-tumor spatial variations. Spatial patterns in cell type abundance and pathway signatures in the TME potentially impact tumor growth dynamics and cancer hallmarks. Tumor biopsies are integral to the diagnosis and clinical management of cancer patients; our data suggest that owing to intra-tumor non-genetic spatial heterogeneity, individual biopsies may underappreciate the extent of clinically relevant, functional variations across geographic regions within tumors.

Genomic landscape of lymphatic malformations: a case series and response to the PI3Kα inhibitor alpelisib in an N-of-1 clinical trial.

Background: Lymphatic malformations (LMs) often pose treatment challenges due to a large size or a critical location that could lead to disfigurement, and there are no standardized treatment approaches for either refractory or unresectable cases. Methods: We examined the genomic landscape of a patient cohort of LMs (n = 30 cases) that underwent comprehensive genomic profiling using a large-panel next-generation sequencing assay. Immunohistochemical analyses were completed in parallel. Results: These LMs had low mutational burden with hotspot PIK3CA mutations (n = 20) and NRAS (n = 5) mutations being most frequent, and mutually exclusive. All LM cases with Kaposi sarcoma-like (kaposiform) histology had NRAS mutations. One index patient presented with subacute abdominal pain and was diagnosed with a large retroperitoneal LM harboring a somatic PIK3CA gain-of-function mutation (H1047R). The patient achieved a rapid and durable radiologic complete response, as defined in RECIST1.1, to the PI3Kα inhibitor alpelisib within the context of a personalized N-of-1 clinical trial (NCT03941782). In translational correlative studies, canonical PI3Kα pathway activation was confirmed by immunohistochemistry and human LM-derived lymphatic endothelial cells carrying an allele with an activating mutation at the same locus were sensitive to alpelisib treatment in vitro, which was demonstrated by a concentration-dependent drop in measurable impedance, an assessment of cell status. Conclusions: Our findings establish that LM patients with conventional or kaposiform histology have distinct, yet targetable, driver mutations. Funding: R.P. and W.A. are supported by awards from the Levy-Longenbaugh Fund. S.G. is supported by awards from the Hugs for Brady Foundation. This work has been funded in part by the NCI Cancer Center Support Grants (CCSG; P30) to the University of Arizona Cancer Center (CA023074), the University of New Mexico Comprehensive Cancer Center (CA118100), and the Rutgers Cancer Institute of New Jersey (CA072720). B.K.M. was supported by National Science Foundation via Graduate Research Fellowship DGE-1143953. Clinical trial number: NCT03941782.

Smoking-associated Downregulation of FILIP1L Enhances Lung Adenocarcinoma Progression Through Mucin Production, Inflammation, and Fibrosis.

Lung adenocarcinoma (LUAD) is the major subtype in lung cancer, and cigarette smoking is essentially linked to its pathogenesis. We show that downregulation of Filamin A interacting protein 1-like (FILIP1L) is a driver of LUAD progression. Cigarette smoking causes its downregulation by promoter methylation in LUAD. Loss of FILIP1L increases xenograft growth, and, in lung-specific knockout mice, induces lung adenoma formation and mucin secretion. In syngeneic allograft tumors, reduction of FILIP1L and subsequent increase in its binding partner, prefoldin 1 (PFDN1) increases mucin secretion, proliferation, inflammation, and fibrosis. Importantly, from the RNA-sequencing analysis of these tumors, reduction of FILIP1L is associated with upregulated Wnt/ß-catenin signaling, which has been implicated in proliferation of cancer cells as well as inflammation and fibrosis within the tumor microenvironment. Overall, these findings suggest that down-regulation of FILIP1L is clinically relevant in LUAD, and warrant further efforts to evaluate pharmacologic regimens that either directly or indirectly restore FILIP1L-mediated gene regulation for the treatment of these neoplasms. Significance: This study identifies FILIP1L as a tumor suppressor in LUADs and demonstrates that downregulation of FILIP1L is a clinically relevant event in the pathogenesis and clinical course of these neoplasms.

Durable Response to PD1 Inhibitor Pembrolizumab in a Metastatic, Metaplastic Breast Cancer.

Metaplastic breast cancer (MBC) is a rare and aggressive subtype of breast cancer. Tumor characteristics typically feature estrogen receptor, progesterone receptor, and HER2-negative, triple-negative breast cancer (TNBC), with a poorer prognosis relative to pure invasive ductal or lobular disease. Resistance to chemotherapy often leads to local recurrence and distant metastasis. Genomic profiling has identified multiple molecular abnormalities that may translate to targetable therapies in MBC. These tumors are known to display higher PD-L1 expressivity than other subtypes of breast cancer, and disease control with pembrolizumab and chemotherapy has been documented. We identify a patient with metastatic, metaplastic TNBC, with mesenchymal components and osseous differentiation, who completed 2 years of pembrolizumab treatment and has remained without evidence of disease after 32 months of observation, while maintaining good quality of life. Future efforts should focus on immunotherapy response with respect to the various subtypes of MBC, and treatment should continue to be incorporated in clinical trials to maximize disease response.

Clinical Features and Multiplatform Molecular Analysis Assist in Understanding Patient Response to Anti-PD-1/PD-L1 in Renal Cell Carcinoma.

Predicting response to ICI therapy among patients with renal cell carcinoma (RCC) has been uniquely challenging. We analyzed patient characteristics and clinical correlates from a retrospective single-site cohort of advanced RCC patients receiving anti-PD-1/PD-L1 monotherapy (N = 97), as well as molecular parameters in a subset of patients, including multiplexed immunofluorescence (mIF), whole exome sequencing (WES), T cell receptor (TCR) sequencing, and RNA sequencing (RNA-seq). Clinical factors such as the development of immune-related adverse events (odds ratio (OR) = 2.50, 95% confidence interval (CI) = 1.05-5.91) and immunological prognostic parameters, including a higher percentage of circulating lymphocytes (23.4% vs. 17.4%, p = 0.0015) and a lower percentage of circulating neutrophils (61.8% vs. 68.5%, p = 0.0045), correlated with response. Previously identified gene expression signatures representing pathways of angiogenesis, myeloid inflammation, T effector presence, and clear cell signatures also correlated with response. High PD-L1 expression (>10% cells) as well as low TCR diversity (≤644 clonotypes) were associated with improved progression-free survival (PFS). We corroborate previously published findings and provide preliminary evidence of T cell clonality impacting the outcome of RCC patients. To further biomarker development in RCC, future studies will benefit from integrated analysis of multiple molecular platforms and prospective validation.

Targetable alterations in invasive pleomorphic lobular carcinoma of the breast.

BACKGROUND: Invasive pleomorphic lobular carcinoma (PLC) of the breast is a subtype of invasive lobular cancer which compromises approximately 1% of all epithelial breast malignancies and is characterized by higher nuclear pleomorphism and poorer prognosis than classic invasive lobular cancer (ILC). Since PLC is more aggressive than classical ILC, we examined the underlying molecular alterations in this subtype of breast cancer to understand the possible benefit from targeted therapies. METHODS: In this study, we investigate the clinical characteristics and molecular alterations in 16 PLC from our institution. Additionally, we examined the clinical and genomic features in 31 PLC from the Cancer Genome Atlas (TCGA). RESULTS: Overall, our analysis of PLC found that 28% had activating ERBB2 mutations, 21% had ERBB2 amplification, and 49% activating PIK3CA mutations. Among cases from our institution, we found 19% with activating ERBB2 mutations, 25% had ERBB2 amplification, and 38% with activating PIK3CA mutations. In data from TCGA, 32% had activating ERBB2 mutations, 19% had ERBB2 amplification, and 55% had activating PIK3CA mutations. While classic ILC in TCGA had similar percentages of PIK3CA alterations compared to PLC, activating ERBB2 alterations were exceedingly rare, with no activating ERBB2 mutations and only one case with ERBB2 amplification. Interestingly, in further examining TCGA data which included FGFR1 and PTEN, 94% of PLC had alterations in ERBB2, FGFR1, or the PI3K pathway. CONCLUSIONS: Our results show a high frequency of ERBB2 and PIK3CA alterations in PLC and suggest all PLC should be tested for potential therapeutic targeting.

Weakly Supervised Deep Nuclei Segmentation Using Partial Points Annotation in Histopathology Images.

Nuclei segmentation is a fundamental task in histopathology image analysis. Typically, such segmentation tasks require significant effort to manually generate accurate pixel-wise annotations for fully supervised training. To alleviate such tedious and manual effort, in this paper we propose a novel weakly supervised segmentation framework based on partial points annotation, i.e., only a small portion of nuclei locations in each image are labeled. The framework consists of two learning stages. In the first stage, we design a semi-supervised strategy to learn a detection model from partially labeled nuclei locations. Specifically, an extended Gaussian mask is designed to train an initial model with partially labeled data. Then, self-training with background propagation is proposed to make use of the unlabeled regions to boost nuclei detection and suppress false positives. In the second stage, a segmentation model is trained from the detected nuclei locations in a weakly-supervised fashion. Two types of coarse labels with complementary information are derived from the detected points and are then utilized to train a deep neural network. The fully-connected conditional random field loss is utilized in training to further refine the model without introducing extra computational complexity during inference. The proposed method is extensively evaluated on two nuclei segmentation datasets. The experimental results demonstrate that our method can achieve competitive performance compared to the fully supervised counterpart and the state-of-the-art methods while requiring significantly less annotation effort.

All-FIT: allele-frequency-based imputation of tumor purity from high-depth sequencing data.

SUMMARY: Clinical sequencing aims to identify somatic mutations in cancer cells for accurate diagnosis and treatment. However, most widely used clinical assays lack patient-matched control DNA and additional analysis is needed to distinguish somatic and unfiltered germline variants. Such computational analyses require accurate assessment of tumor cell content in individual specimens. Histological estimates often do not corroborate with results from computational methods that are primarily designed for normal-tumor matched data and can be confounded by genomic heterogeneity and presence of sub-clonal mutations. Allele-frequency-based imputation of tumor (All-FIT) is an iterative weighted least square method to estimate specimen tumor purity based on the allele frequencies of variants detected in high-depth, targeted, clinical sequencing data. Using simulated and clinical data, we demonstrate All-FIT's accuracy and improved performance against leading computational approaches, highlighting the importance of interpreting purity estimates based on expected biology of tumors. AVAILABILITY AND IMPLEMENTATION: Freely available at http://software.khiabanian-lab.org. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Non-Genetic Intra-Tumor Heterogeneity Is a Major Predictor of Phenotypic Heterogeneity and Ongoing Evolutionary Dynamics in Lung Tumors.

Impacts of genetic and non-genetic intra-tumor heterogeneity (ITH) on tumor phenotypes and evolvability remain debated. We analyze ITH in lung squamous cell carcinoma at the levels of genome, transcriptome, and tumor-immune interactions and histopathological characteristics by multi-region bulk and single-cell sequencing. Genomic heterogeneity alone is a weak indicator of intra-tumor non-genetic heterogeneity at immune and transcriptomic levels that impact multiple cancer-related pathways, including those related to proliferation and inflammation, which in turn contribute to intra-tumor regional differences in histopathology and subtype classification. Tumor subclones have substantial differences in proliferation score, suggestive of non-neutral clonal dynamics. Proliferation and other cancer-related pathways also show intra-tumor regional differences, sometimes even within the same subclones. Neo-epitope burden negatively correlates with immune infiltration, indicating immune-mediated purifying selection on somatic mutations. Taken together, our observations suggest that non-genetic heterogeneity is a major determinant of heterogeneity in histopathological characteristics and impacts evolutionary dynamics in lung cancer.