19 December 2024 in npj Precision Oncology
Immune checkpoint inhibitors (ICI) have become integral to treatment of non-small cell lung cancer (NSCLC). However, reliable biomarkers predictive of immunotherapy efficacy are limited. Here, we introduce HistoTME, a novel weakly supervised deep learning approach to infer the tumor microenvironment (TME) composition directly from histopathology images of NSCLC patients. We show that HistoTME accurately predicts the expression of 30 distinct cell type-specific molecular signatures directly from whole slide images, achieving an average Pearson correlation of 0.5 with the ground truth on independent tumor cohorts. Furthermore, we find that HistoTME-predicted microenvironment signatures and their underlying interactions improve prognostication of lung cancer patients receiving immunotherapy, achieving an AUROC of 0.75 [95% CI: 0.61-0.88] for predicting treatment responses following first-line ICI treatment, utilizing an external clinical cohort of 652 patients. Collectively, HistoTME presents an effective approach for interrogating the TME and predicting ICI response, complementing PD-L1 expression, and bringing us closer to personalized immuno-oncology.
16 October 2024 in Clinical Cancer Research
Purpose: Osteosarcoma is an aggressive bone cancer lacking robust biomarkers for personalized treatment. Despite its scarcity in humans, it is relatively common in adult pet dogs. This study aimed to analyze clinically annotated bulk tumor transcriptomic datasets of canine and patients with human osteosarcoma to identify potentially conserved patterns of disease progression. Experimental Design: Bulk transcriptomic data from 245 pet dogs with treatment-naïve appendicular osteosarcoma were analyzed using deconvolution to characterize the tumor microenvironment (TME). The TME of both primary and metastatic tumors derived from the same dog was compared, and its impact on canine survival was assessed. A machine learning model was developed to classify the TME based on its inferred composition using canine tumor data. This model was applied to eight independent human osteosarcoma datasets to assess its generalizability and prognostic value. Results: This study found three distinct TME subtypes of canine osteosarcoma based on cell type composition of bulk tumor samples: immune enriched, immune enriched dense extracellular matrix-like, and immune desert. These three TME-based subtypes of canine osteosarcomas were conserved in humans and could predict progression-free survival outcomes of human patients, independent of conventional prognostic factors such as percent tumor necrosis post standard of care chemotherapy treatment and disease stage at diagnosis. Conclusions: These findings demonstrate the potential of leveraging data from naturally occurring cancers in canines to model the complexity of the human osteosarcoma TME, offering a promising avenue for the discovery of novel biomarkers and developing more effective precision oncology treatments.
14 October 2024 in Nature Communications
Immune checkpoint blockade (ICB) is a promising cancer therapy; however, resistance frequently develops. To explore ICB resistance mechanisms, we develop Immunotherapy Resistance cell-cell Interaction Scanner (IRIS), a machine learning model aimed at identifying cell-type-specific tumor microenvironment ligand-receptor interactions relevant to ICB resistance. Applying IRIS to deconvolved transcriptomics data of the five largest melanoma ICB cohorts, we identify specific downregulated interactions, termed resistance downregulated interactions (RDI), as tumors develop resistance. These RDIs often involve chemokine signaling and offer a stronger predictive signal for ICB response compared to upregulated interactions or the state-of-the-art published transcriptomics biomarkers. Validation across multiple independent melanoma patient cohorts and modalities confirms that RDI activity is associated with CD8 + T cell infiltration and highly manifested in hot/brisk tumors. This study presents a strongly predictive ICB response biomarker, highlighting the key role of downregulating chemotaxis-associated ligand-receptor interactions in inhibiting lymphocyte infiltration in resistant tumors.
27 October 2022 in The American Journal of Pathology
Osteosarcomas (OSs) are aggressive bone tumors with many divergent histologic patterns. During pa-thology review, OSs are subtyped based on the predominant histologic pattern; however, tumors often demonstrate multiple patterns. This high tumor heterogeneity coupled with scarcity of samples compared with other tumor types render histology-based prognosis of OSs challenging. To combat lower case numbers in humans, dogs with spontaneous OSs have been suggested as a model species. Herein, a convolutional neural network was adversarially trained to classify distinct histologic patterns of OS in humans using mostly canine OS data during training. Adversarial training improved domain adaption of a histologic subtype classifier from canines to humans, achieving an average multiclass F1 score of 0.77 (95% CI, 0.74-0.79) and 0.80 (95% CI, 0.78-0.81) when compared with the ground truth in canines and humans, respectively. Finally, this trained model, when used to characterize the histologic land-scape of 306 canine OSs, uncovered distinct clusters with markedly different clinical responses to standard-of-care therapy. (Am J Pathol 2023, 193: 60-72; https://doi.org/10.1016/ j.ajpath.2022.09.009)
30 November 2017 in PLOS Computational Biology
Guilt-by-association codifies the empirical observation that a gene’s function is informed by its neighborhood in a biological network. This would imply that when a gene’s network context is altered, for instance in disease condition, so could be the gene’s function. Although context-specific changes in biological networks have been explored, the potential changes they may induce on the functional roles of genes are yet to be characterized. Here we analyze, for the first time, the network-induced potential functional changes in breast cancer. Using transcriptomic samples for 1047 breast tumors and 110 healthy breast tissues from TCGA, we derive sample-specific protein interaction networks and assign sample-specific functions to genes via a diffusion strategy. Testing for significant changes in the inferred functions between normal and cancer samples, we find several functions to have significantly gained or lost genes in cancer, not due to differential expression of genes known to perform the function, but rather due to changes in the network topology. Our predicted functional changes are supported by mutational and copy number profiles in breast cancers. Our diffusion-based functional assignment provides a novel characterization of a tumor that is complementary to the standard approach based on functional annotation alone. Importantly, this characterization is effective in predicting patient survival, as well as in predicting several known histopathological subtypes of breast cancer. Proteins that cooperate to perform a specific cellular function often interact with each other and are clustered together in a protein-protein interaction network (PIN). Thus, a protein’s function can be inferred from the prevailing functions of its network neighbors. However, the network neighborhood is highly dynamic and so, we argue, a protein’s function should be. Based on this premise, we evaluate how a protein’s function might change in a specific dynamic context of breast cancer. This novel approach uniquely reveals several functions that are significantly lost or gained in cancer. Moreover, the PIN-induced sample-specific functional activity provides a novel characterization of cancer sample that can help predict patient survival as well as several known subtypes of breast cancer.
29 September 2023 in bioRxiv
Osteosarcoma is a relatively rare but aggressive cancer of the bones with a shortage of effective biomarkers. Although less common in humans, Osteosarcomas are fairly common in adult pet dogs and have been shown to share many similarities with their human analogs. In this work, we analyze bulk transcriptomic data of 213 primary and 100 metastatic Osteosarcoma samples from 210 pet dogs enrolled in nation-wide clinical trials to uncover three Tumor Microenvironment (TME)-based subtypes: Immune Enriched (IE), Immune Enriched Dense Extra-Cellular Matrix-like (IE-ECM) and Immune Desert (ID) with distinct cell type compositions, oncogenic pathway activity and chromosomal instability. Furthermore, leveraging bulk transcriptomic data of canine primary tumors and their matched metastases from different sites, we characterize how the Osteosarcoma TME evolves from primary to metastatic disease in a standard of care clinical setting and assess its overall impact on clinical outcomes of canines. Most importantly, we find that TME-based subtypes of canine Osteosarcomas are conserved in humans and predictive of progression free survival outcomes of human patients, independently of known prognostic biomarkers such as presence of metastatic disease at diagnosis and percent necrosis following chemotherapy. In summary, these results demonstrate the power of using canines to model the human Osteosarcoma TME and discover novel biomarkers for clinical translation.