Computational immunogenomic approaches to predict response to cancer immunotherapies.

Cancer immunogenomics is an emerging field that bridges genomics and immunology. The establishment of large-scale genomic collaborative efforts along with the development of new single-cell transcriptomic techniques and multi-omics approaches have enabled characterization of the mutational and transcriptional profiles of many cancer types and helped to identify clinically actionable alterations as well as predictive and prognostic biomarkers. Researchers have developed computational approaches and machine learning algorithms to accurately obtain clinically useful information from genomic and transcriptomic sequencing data from bulk tissue or single cells and explore tumours and their microenvironment. The rapid growth in sequencing and computational approaches has resulted in the unmet need to understand their true potential and limitations in enabling improvements in the management of patients with cancer who are receiving immunotherapies. In this Review, we describe the computational approaches currently available to analyse bulk tissue and single-cell sequencing data from cancer, stromal and immune cells, as well as how best to select the most appropriate tool to address various clinical questions and, ultimately, improve patient outcomes.

The oesophageal adenocarcinoma tumour immune microenvironment dictates outcomes with different modalities of neoadjuvant therapy - results from the AGITG DOCTOR trial and the cancer evolution biobank.

A plateau in treatment effect can be seen for the current 'one-size-fits-all' approach to oesophageal adenocarcinoma (OAC) management using neoadjuvant chemoradiotherapy (nCRT) or chemotherapy (nCT). In OAC, the tumour microenvironment (TME) is largely immunosuppressed, however a subgroup of patients with an immune-inflamed TME exist and show improved outcomes. We aimed to understand the overall immune-based mechanisms underlying treatment responses and patient outcomes in OAC, and in relation to neoadjuvant therapy modality. This study included 107 patients; 68 patients were enrolled in the Australian Gastro-Intestinal Trials Group sponsored DOCTOR Trial, and 38 patients were included from the Cancer Evolution Biobank. Matched pre-treatment and post-treatment tumour biopsies were used to perform multi-modality analysis of the OAC TME including NanoString mRNA expression analysis, multiplex and single colour immunohistochemistry (IHC), and peripheral blood mononuclear cell analysis of tumour-antigen specific T cell responses. Patients with the best clinicopathological outcomes and survival had an immune-inflamed TME enriched with anti-tumour immune cells and pathways. Those with the worst survival showed a myeloid T regulatory cell enriched TME, with decreased CD8+ cell infiltration and increased pro-tumour immune cells. Multiplex IHC analysis identified that high intra-tumoural infiltration of CD8+ cells, and low infiltration with CD163+ cells was associated with improved survival. High tumour core CD8+ T cell infiltration, and a low tumour margin infiltration of CD163+ cells was also associated with improved survival. nCRT showed improved survival compared with nCT for patients with low CD8+, or high CD163+ cell infiltration. Poly-functional T cell responses were seen with tumour-antigen specific T cells. Overall, our study supports the development of personalised therapeutic approaches based on the immune microenvironment in OAC. Patients with an immune-inflamed TME show favourable outcomes regardless of treatment modality. However, in those with an immunosuppressed TME with CD163+ cell infiltration, treatment with nCRT can improve outcomes. Our findings support previous studies into the TME of OAC and with more research, immune based biomarker selection of treatment modality may lead in improved outcomes in this deadly disease.

Performance of tumour microenvironment deconvolution methods in breast cancer using single-cell simulated bulk mixtures.

Cells within the tumour microenvironment (TME) can impact tumour development and influence treatment response. Computational approaches have been developed to deconvolve the TME from bulk RNA-seq. Using scRNA-seq profiling from breast tumours we simulate thousands of bulk mixtures, representing tumour purities and cell lineages, to compare the performance of nine TME deconvolution methods (BayesPrism, Scaden, CIBERSORTx, MuSiC, DWLS, hspe, CPM, Bisque, and EPIC). Some methods are more robust in deconvolving mixtures with high tumour purity levels. Most methods tend to mis-predict normal epithelial for cancer epithelial as tumour purity increases, a finding that is validated in two independent datasets. The breast cancer molecular subtype influences this mis-prediction. BayesPrism and DWLS have the lowest combined numbers of false positives and false negatives, and have the best performance when deconvolving granular immune lineages. Our findings highlight the need for more single-cell characterisation of rarer cell types, and suggest that tumour cell compositions should be considered when deconvolving the TME.

Generalising uncertainty improves accuracy and safety of deep learning analytics applied to oncology.

Uncertainty estimation is crucial for understanding the reliability of deep learning (DL) predictions, and critical for deploying DL in the clinic. Differences between training and production datasets can lead to incorrect predictions with underestimated uncertainty. To investigate this pitfall, we benchmarked one pointwise and three approximate Bayesian DL models for predicting cancer of unknown primary, using three RNA-seq datasets with 10,968 samples across 57 cancer types. Our results highlight that simple and scalable Bayesian DL significantly improves the generalisation of uncertainty estimation. Moreover, we designed a prototypical metric-the area between development and production curve (ADP), which evaluates the accuracy loss when deploying models from development to production. Using ADP, we demonstrate that Bayesian DL improves accuracy under data distributional shifts when utilising 'uncertainty thresholding'. In summary, Bayesian DL is a promising approach for generalising uncertainty, improving performance, transparency, and safety of DL models for deployment in the real world.

Multi-omic features of oesophageal adenocarcinoma in patients treated with preoperative neoadjuvant therapy.

Oesophageal adenocarcinoma is a poor prognosis cancer and the molecular features underpinning response to treatment remain unclear. We investigate whole genome, transcriptomic and methylation data from 115 oesophageal adenocarcinoma patients mostly from the DOCTOR phase II clinical trial (Australian New Zealand Clinical Trials Registry-ACTRN12609000665235), with exploratory analysis pre-specified in the study protocol of the trial. We report genomic features associated with poorer overall survival, such as the APOBEC mutational and RS3-like rearrangement signatures. We also show that positron emission tomography non-responders have more sub-clonal genomic copy number alterations. Transcriptomic analysis categorises patients into four immune clusters correlated with survival. The immune suppressed cluster is associated with worse survival, enriched with myeloid-derived cells, and an epithelial-mesenchymal transition signature. The immune hot cluster is associated with better survival, enriched with lymphocytes, myeloid-derived cells, and an immune signature including CCL5, CD8A, and NKG7. The immune clusters highlight patients who may respond to immunotherapy and thus may guide future clinical trials.

Efficient detection and monitoring of pediatric brain malignancies with liquid biopsy based on patient-specific somatic mutation screening.

BACKGROUND: Brain cancer is the leading cause of cancer-related death in children. Early detection and serial monitoring are essential for better therapeutic outcomes. Liquid biopsy has recently emerged as a promising approach for detecting these tumors by screening body fluids for the presence of circulating tumor DNA (ctDNA). Here we tested the limits of liquid biopsy using patient-specific somatic mutations to detect and monitor primary and metastatic pediatric brain cancer. METHODS: Somatic mutations were identified in 3 ependymoma, 1 embryonal tumor with multilayered rosettes, 1 central nervous system neuroblastoma, and 7 medulloblastoma patients. The mutations were used as liquid biomarkers for serial assessment of cerebrospinal fluid (CSF) samples using a droplet digital PCR (ddPCR) system. The findings were correlated to the imaging data and clinical assessment to evaluate the utility of the approach for clinical translation. RESULTS: We developed personalized somatic mutation ddPCR assays which we show are highly specific, sensitive, and efficient in detection and monitoring of ctDNA, with a positive correlation between presence of ctDNA, disease course, and clinical outcomes in the majority of patients. CONCLUSIONS: We demonstrate the feasibility and clinical utility of personalized mutation-based liquid biopsy for the surveillance of brain cancer in children. However, even with this specific and sensitive approach, we identified some potential false negative analyses. Overall, our results indicate that changes in ctDNA profiles over time demonstrate the great potential of our specific approach for predicting tumor progression, burden, and response to treatment.

Comprehensive genomic and tumour immune profiling reveals potential therapeutic targets in malignant pleural mesothelioma.

BACKGROUND: Malignant pleural mesothelioma (MPM) has a poor overall survival with few treatment options. Whole genome sequencing (WGS) combined with the immune features of MPM offers the prospect of identifying changes that could inform future clinical trials. METHODS: We analysed somatic mutations from 229 MPM samples, including previously published data and 58 samples that had undergone WGS within this study. This was combined with RNA-seq analysis to characterize the tumour immune environment. RESULTS: The comprehensive genome analysis identified 12 driver genes, including new candidate genes. Whole genome doubling was a frequent event that correlated with shorter survival. Mutational signature analysis revealed SBS5/40 were dominant in 93% of samples, and defects in homologous recombination repair were infrequent in our cohort. The tumour immune environment contained high M2 macrophage infiltrate linked with MMP2, MMP14, TGFB1 and CCL2 expression, representing an immune suppressive environment. The expression of TGFB1 was associated with overall survival. A small subset of samples (less than 10%) had a higher proportion of CD8 T cells and a high cytolytic score, suggesting a 'hot' immune environment independent of the somatic mutations. CONCLUSIONS: We propose accounting for genomic and immune microenvironment status may influence therapeutic planning in the future.

Multiomic profiling of checkpoint inhibitor-treated melanoma: Identifying predictors of response and resistance, and markers of biological discordance.

We concurrently examine the whole genome, transcriptome, methylome, and immune cell infiltrates in baseline tumors from 77 patients with advanced cutaneous melanoma treated with anti-PD-1 with or without anti-CTLA-4. We show that high tumor mutation burden (TMB), neoantigen load, expression of IFNγ-related genes, programmed death ligand expression, low PSMB8 methylation (therefore high expression), and T cells in the tumor microenvironment are associated with response to immunotherapy. No specific mutation correlates with therapy response. A multivariable model combining the TMB and IFNγ-related gene expression robustly predicts response (89% sensitivity, 53% specificity, area under the curve [AUC], 0.84); tumors with high TMB and a high IFNγ signature show the best response to immunotherapy. This model validates in an independent cohort (80% sensitivity, 59% specificity, AUC, 0.79). Except for a JAK3 loss-of-function mutation, for patients who did not respond as predicted there is no obvious biological mechanism that clearly explained their outlier status, consistent with intratumor and intertumor heterogeneity in response to immunotherapy.

Whole-genome sequencing of acral melanoma reveals genomic complexity and diversity.

To increase understanding of the genomic landscape of acral melanoma, a rare form of melanoma occurring on palms, soles or nail beds, whole genome sequencing of 87 tumors with matching transcriptome sequencing for 63 tumors was performed. Here we report that mutational signature analysis reveals a subset of tumors, mostly subungual, with an ultraviolet radiation signature. Significantly mutated genes are BRAF, NRAS, NF1, NOTCH2, PTEN and TYRP1. Mutations and amplification of KIT are also common. Structural rearrangement and copy number signatures show that whole genome duplication, aneuploidy and complex rearrangements are common. Complex rearrangements occur recurrently and are associated with amplification of TERT, CDK4, MDM2, CCND1, PAK1 and GAB2, indicating potential therapeutic options.

CD155 loss enhances tumor suppression via combined host and tumor-intrinsic mechanisms.

Critical immune-suppressive pathways beyond programmed death 1 (PD-1) and programmed death ligand 1 (PD-L1) require greater attention. Nectins and nectin-like molecules might be promising targets for immunotherapy, since they play critical roles in cell proliferation and migration and exert immunomodulatory functions in pathophysiological conditions. Here, we show CD155 expression in both malignant cells and tumor-infiltrating myeloid cells in humans and mice. Cd155-/- mice displayed reduced tumor growth and metastasis via DNAM-1 upregulation and enhanced effector function of CD8+ T and NK cells, respectively. CD155-deleted tumor cells also displayed slower tumor growth and reduced metastases, demonstrating the importance of a tumor-intrinsic role of CD155. CD155 absence on host and tumor cells exerted an even greater inhibition of tumor growth and metastasis. Blockade of PD-1 or both PD-1 and CTLA4 was more effective in settings in which CD155 was limiting, suggesting the clinical potential of cotargeting PD-L1 and CD155 function.

Germline and somatic variant identification using BGISEQ-500 and HiSeq X Ten whole genome sequencing.

Technological innovation and increased affordability have contributed to the widespread adoption of genome sequencing technologies in biomedical research. In particular large cancer research consortia have embraced next generation sequencing, and have used the technology to define the somatic mutation landscape of multiple cancer types. These studies have primarily utilised the Illumina HiSeq platforms. In this study we performed whole genome sequencing of three malignant pleural mesothelioma and matched normal samples using a new platform, the BGISEQ-500, and compared the results obtained with Illumina HiSeq X Ten. Germline and somatic, single nucleotide variants and small insertions or deletions were independently identified from data aligned human genome reference. The BGISEQ-500 and HiSeq X Ten platforms showed high concordance for germline calls with genotypes from SNP arrays (>99%). The germline and somatic single nucleotide variants identified in both sequencing platforms were highly concordant (86% and 72% respectively). These results indicate the potential applicability of the BGISEQ-500 platform for the identification of somatic and germline single nucleotide variants by whole genome sequencing. The BGISEQ-500 datasets described here represent the first publicly-available cancer genome sequencing performed using this platform.