Tumor hypoxia has been shown to predict poor patient outcomes in several cancer types, partially because it reduces radiation’s ability to kill cells. We hypothesized that some of the clinical effects of hypoxia could also be due to its impact on the tumor microbiome. Therefore, we examined the RNA sequencing data from the Oncology Research Information Exchange Network database of patients with colorectal cancer treated with radiotherapy. We identified microbial RNAs for each tumor and related them to the hypoxic gene expression scores calculated from host mRNA. Our analysis showed that the hypoxia expression score predicted poor patient outcomes and identified tumors enriched with certain microbes such as Fusobacterium nucleatum. The presence of other microbes, such as Fusobacterium canifelinum, predicted poor patient outcomes, suggesting a potential interaction between hypoxia, the microbiome, and radiation response. To experimentally investigate this concept, we implanted CT26 colorectal cancer cells into immune-competent BALB/c and immune-deficient athymic nude mice. After growth, in which tumors passively acquired microbes from the gastrointestinal tract, we harvested tumors, extracted nucleic acids, and sequenced host and microbial RNAs. We stratified tumors based on their hypoxia score and performed a metatranscriptomic analysis of microbial gene expression. In addition to hypoxia-tropic and -phobic microbial populations, analysis of microbial gene expression at the strain level showed expression differences based on the hypoxia score. Thus, hypoxia gene expression scores seem to associate with different microbial populations and elicit an adaptive transcriptional response in intratumoral microbes, potentially influencing clinical outcomes. Significance: Tumor hypoxia reduces radiotherapy efficacy. In this study, we explored whether some of the clinical effects of hypoxia could be due to interaction with the tumor microbiome. Hypoxic gene expression scores associated with certain microbes and elicited an adaptive transcriptional response in others that could contribute to poor clinical outcomes.

The microbiome affects cancer, from carcinogenesis to response to treatments. New evidence suggests that microbes are also present in many tumors, though the scope of how they affect tumor biology and clinical outcomes is in its early stages. A broad survey of tumor microbiome samples across several independent datasets is needed to identify robust correlations for follow-up testing. We created a tool called {exotic} for “exogenous sequences in tumors and immune cells” to carefully identify the tumor microbiome within RNA sequencing (RNA-seq) datasets. We applied it to samples collected through the Oncology Research Information Exchange Network (ORIEN) and The Cancer Genome Atlas. We showed how the processing removes contaminants and batch effects to yield microbe abundances consistent with non–high-throughput sequencing–based approaches and DNA-amplicon–based measurements of a subset of the same tumors. We sought to establish clinical relevance by correlating the microbe abundances with various clinical and tumor measurements, such as age and tumor hypoxia. This process leveraged the two datasets and raised up only the concordant (significant and in the same direction) associations. We observed associations with survival and clinical variables that are cancer specific and relatively few associations with immune composition. Finally, we explored potential mechanisms by which microbes and tumors may interact using a network-based approach. Alistipes, a common gut commensal, showed the highest network degree centrality and was associated with genes related to metabolism and inflammation. The {exotic} tool can support the discovery of microbes in tumors in a way that leverages the many existing and growing RNA-seq datasets.Significance: The intrinsic tumor microbiome holds great potential for its ability to predict various aspects of cancer biology and as a target for rational manipulation. Here, we describe a tool to quantify microbes from within tumor RNA-seq and apply it to two independent datasets. We show new associations with clinical variables that justify biomarker uses and more experimentation into the mechanisms by which tumor microbiomes affect cancer outcomes.

Emerging evidence supports the important role of the tumor microbiome in oncogenesis, cancer immune phenotype, cancer progression, and treatment outcomes in many malignancies. In this study, we investigated the metastatic melanoma tumor microbiome and its potential roles in association with clinical outcomes, such as survival, in patients with metastatic disease treated with immune checkpoint inhibitors (ICI). Baseline tumor samples were collected from 71 patients with metastatic melanoma before treatment with ICIs. Bulk RNA sequencing (RNA-seq) was conducted on the formalin-fixed, paraffin-embedded and fresh frozen tumor samples. Durable clinical benefit (primary clinical endpoint) following ICIs was defined as overall survival ≥24 months and no change to the primary drug regimen (responders). We processed RNA-seq reads to carefully identify exogenous sequences using the {exotic} tool. The age of the 71 patients with metastatic melanoma ranged from 24 to 83 years, 59% were male, and 55% survived >24 months following the initiation of ICI treatment. Exogenous taxa were identified in the tumor RNA-seq, including bacteria, fungi, and viruses. We found differences in gene expression and microbe abundances in immunotherapy-responsive versus nonresponsive tumors. Responders showed significant enrichment of bacteriophages in the phylum Uroviricota, and nonresponders showed enrichment of several bacteria, including Campylobacter jejuni. These microbes correlated with immune-related gene expression signatures. Finally, we found that models for predicting prolonged survival with immunotherapy using both microbe abundances and gene expression outperformed models using either dataset alone. Our findings warrant further investigation and potentially support therapeutic strategies to modify the tumor microbiome in order to improve treatment outcomes with ICIs. Significance: We analyzed the tumor microbiome and interactions with genes and pathways in metastatic melanoma treated with immunotherapy and identified several microbes associated with immunotherapy response and immune-related gene expression signatures. Machine learning models that combined microbe abundances and gene expression outperformed models using either dataset alone in predicting immunotherapy responses.

Purpose/Objective(s): Microbiome has been shown to affect tumorigenesis by promoting inflammation. However, the association between the upper aerodigestive microbiome and head and neck squamous cell carcinoma (HNSCC) is not well established. Hypoxia is a modifiable factor associated with poor radiation response. Our study analyzed the HNSCC tumor samples from The Cancer Genome Atlas (TCGA) to investigate the relationship between different HNSCC tumor subsites, hypoxia, and local tumor microbiome composition. Results: A total of 357 patients were included [Oral cavity (OC) = 226, Oropharynx (OPx) = 53, and Larynx/Hypopharynx (LHPx) = 78], of which 12.8%, 71.7%, and 10.3%, respectively, were HPV positive. The mean (SD) hypoxia scores were 30.18 (11.10), 24.31 (14.13), and 29.53 (12.61) in OC, OPx, and LHPx tumors, respectively, with higher values indicating greater hypoxia. The hypoxia score was significantly higher for OC tumors compared to OPx (p = 0.044) and LHPx (p = 0.002). There was no significant correlation between hypoxia and HPV status. Pseudomonas sp. in OC, Actinomyces sp. and Sulfurimonas sp. in OPx, and Filifactor, Pseudomonas and Actinomyces sp. in LHPx had the strongest association with the hypoxia score. Materials/Methods: Tumor RNAseq samples from TCGA were processed, and the R package “tmesig” was used to calculate gene expression signature, including the Buffa hypoxia (BH) score, a validated hypoxia signature using 52 hypoxia-regulated genes. Microbe relative abundances were modeled with primary tumor location and a high vs. low tertile BH score applying a gamma-distributed generalized linear regression using the “stats” package in R, with adjusted p-value < 0.05 considered significant. Conclusions: In our study, oral cavity tumors were found to be more hypoxic compared to other head and neck subsites, which could potentially contribute to their radiation resistance. For each subsite, distinct microbial populations were over-represented in hypoxic tumors in a subsite-specific manner. Further studies focusing on an association between microbiome, hypoxia, and patient outcomes are warranted.

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.

The incidence of colorectal cancer in young adults (age of diagnosis <50 years) has been rapidly increasing. Although ∼20% of early-onset colorectal cancer (EOCRC) cases are due to germline mutations, the etiology of the majority of EOCRC cases remains poorly understood. Nongenetic factors such as environmental exposure and lifestyle changes are likely to have a direct link to the increased incidence of sporadic EOCRC. We hypothesize that such factors may be observable as alterations in the epigenome, microbiome, and immunome. We characterized the DNA methylation (DNAm) signature and measured DNAm age in EOCRC by using The Cancer Genome Atlas (TCGA). Furthermore, we carefully identified intratumoral microbes from TCGA and the Oncology Research Information Exchange Network datasets and then related the microbes to deconvolved immune cell abundances in EOCRC. We observed that the DNAm age in the EOCRC cohort was 12 years older when compared with the average-onset colorectal cancer (AOCRC) cohort, using three different epigenetic clocks. Differentially methylated sites associated with gene expression include cAMP-responsive element binding protein signaling in neurons, G protein–coupled receptor signaling, phagosome formation, and S100 family signaling. These differences were validated in the gene expression data from TCGA and the Oncology Research Information Exchange Network. When comparing the intratumoral microbes between EOCRC and AOCRC, no consistent differences were observed. Interestingly, the most abundant microbes interacted with the immune systems differently between the EOCRC and AOCRC tumors, characterized by more and larger positive correlations in EOCRC. These data suggest that epigenetic modulation and accelerated aging may play a key role in the development of EOCRC. Significance: We investigated whether environmentally driven factors contribute to EOCRC. We observed accelerated epigenetic aging in EOCRC and epigenetic changes associated with chronic inflammation. Tumor immune cell abundances correlated more strongly with microbes in EOCRC than in AOCRC. These data suggest a dysregulation of the immune response in EOCRC, driving chronic inflammation and tissue aging.