Benchmarking pipelines for subclonal deconvolution of bulk tumour sequencing data

A clinical solution for tracking clonal evolution of acute myeloid leukemia after allogeneic transplantation using bulk next generation sequencing

Clinical next generation sequencing (NGS) typically relies on limited gene panels run on bulk marrow or blood. Current computational tools for inferring clonal relationships is generally limited by the use of a small panel of pathogenic mutations to define clones. We developed an online software (CloneTracker) that uses 'incidentally-sequenced' single nucleotide polymorphisms (SNPs) in the regions of recurrent somatic mutations in addition to conventional mutation data from bulk NGS gene panels to provide detailed visualizations of clonal evolution during cancer treatment, alongside clinical data. Tested on 29 patients who underwent non-myeloablative transplantation for AML, CloneTracker successfully reconstructed the evolutionary dynamics of donor engraftment from bulk NGS and rendered intuitive visualizations of residual patient-derived hematopoiesis and relapsing malignant clones. The software does not require sequencing donor samples, as donor-derived clones are identifiable from post-HCT SNP data. This manuscript aims to introduce CloneTracker to the BMT community and make it available for those who would ascertain its clinical utility, e.g, in BMT trials leveraging molecular minimal residual disease (MRD) monitoring and targeted interventions to pre-empt relapse.

Exome sequencing shows same pattern of clonal tumor mutational burden, intratumor heterogenicity and clonal neoantigen between autologous tumor and Vigil product

Retrospective data support overall survival (OS) advantage to high clonal tumor mutation burden (cTMB), high clonal neoantigen load (cNEO) and low intratumor heterogeneity (ITH) in cancer patients who receive immunotherapy. In order to explore this relationship prospectively with Vigil, a triple function targeted immunotherapy involving ovarian cancer patients in long term follow up of the Phase 2b VITAL trial, we developed an exome sequencing procedure and associated bioinformatics pipeline to determine clonal signal patterns. DNA libraries containing exome sequences tagged with unique molecular identifiers (UMI) were prepared from paired samples and sequenced on Illumina sequencers to high coverage depths of ~ 930X (tumor) and ~ 130X (normal). Raw sequence reads were processed into optimized binary alignment map (BAM) files, using the UMI information. The BAM files were inputted into modules for calling MHC-I alleles, annotating single nucleotide variants (SNVs) and small insertions/deletions (InDels), and for determination of allelic copy number. The outputs were used to predict the sequence of peptide neoantigens and to perform clonality analysis in order to assign each SNV and InDel in a patient tumor sample to a primary clone or subclone. The Clonal Neoantigen pipeline was further assessed using whole exome Illumina sequencing data from three previously published studies. Evaluation of the pipeline using synthetic sequencing data from a sub-clonal deconvolution tool benchmarking study, showed positive predictive value (PPV) and positive percent agreement (PPA) of > 97.5% and > 96.5%, respectively, for SNV and InDel detection with minimum requirements for variant density and allele fraction. Haplotype calls from the Clonal Neoantigen pipeline MHC-I/ MHC-II typing module matched a published benchmark for 91.5% of the calls in a sample of 99 patients. Analysis of exome sequencing data from 14 patients with advanced melanoma revealed a strong correlation between cTMB values determined by the Clonal Neoantigen pipeline as compared to those calculated from the published data (R2 = 0.99). Following validation, the wet lab process and Clonal Neoantigen pipeline was applied to a set of matched normal, tumor, and Vigil product samples from 9 (n = 27 samples) ovarian cancer subjects entered into the VITAL (CL-PTL-119) trial. Results demonstrated marked correlation (R2 = 0.98) of cTMB between tumor used to construct Vigil and Vigil product. Correlation between tumor and Vigil for the cNEO and ITH metrics, showed R2 values of 0.95 and 0.87, respectively. The consistency of the Clonal Neoantigen pipeline results with previously published data as well as the agreement between results for tumor and Vigil for the entire system provide a strong basis of support for utilization of this pipeline for prospective determination of cTMB, cNEO, and ITH values in clinical tumor tissue in order to explore possible correlative relationships with clinical response parameters.

The evolutionary theory of cancer: challenges and potential solutions

The clonal evolution model of cancer was developed in the 1950s-1970s and became central to cancer biology in the twenty-first century, largely through studies of cancer genetics. Although it has proven its worth, its structure has been challenged by observations of phenotypic plasticity, non-genetic forms of inheritance, non-genetic determinants of clone fitness and non-tree-like transmission of genes. There is even confusion about the definition of a clone, which we aim to resolve. The performance and value of the clonal evolution model depends on the empirical extent to which evolutionary processes are involved in cancer, and on its theoretical ability to account for those evolutionary processes. Here, we identify limits in the theoretical performance of the clonal evolution model and provide solutions to overcome those limits. Although we do not claim that clonal evolution can explain everything about cancer, we show how many of the complexities that have been identified in the dynamics of cancer can be integrated into the model to improve our current understanding of cancer.

The Evolutionary Interplay of Somatic and Germline Mutation Rates

Novel sequencing technologies are making it increasingly possible to measure the mutation rates of somatic cell lineages. Accurate germline mutation rate measurement technologies have also been available for a decade, making it possible to assess how this fundamental evolutionary parameter varies across the tree of life. Here, we review some classical theories about germline and somatic mutation rate evolution that were formulated using principles of population genetics and the biology of aging and cancer. We find that somatic mutation rate measurements, while still limited in phylogenetic diversity, seem consistent with the theory that selection to preserve the soma is proportional to life span. However, germline and somatic theories make conflicting predictions regarding which species should have the most accurate DNA repair. Resolving this conflict will require carefully measuring how mutation rates scale with time and cell division and achieving a better understanding of mutation rate pleiotropy among cell types.

Longitudinal molecular profiling elucidates immunometabolism dynamics in breast cancer

Although metabolic reprogramming within tumor cells and tumor microenvironment (TME) is well described in breast cancer, little is known about how the interplay of immune state and cancer metabolism evolves during treatment. Here, we characterize the immunometabolic profiles of tumor tissue samples longitudinally collected from individuals with breast cancer before, during and after neoadjuvant chemotherapy (NAC) using proteomics, genomics and histopathology. We show that the pre-, on-treatment and dynamic changes of the immune state, tumor metabolic proteins and tumor cell gene expression profiling-based metabolic phenotype are associated with treatment response. Single-cell/nucleus RNA sequencing revealed distinct tumor and immune cell states in metabolism between cold and hot tumors. Potential drivers of NAC based on above analyses were validated in vitro. In summary, the study shows that the interaction of tumor-intrinsic metabolic states and TME is associated with treatment outcome, supporting the concept of targeting tumor metabolism for immunoregulation.

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.

Spatiotemporal genomic profiling of intestinal metaplasia reveals clonal dynamics of gastric cancer progression

Intestinal metaplasia (IM) is a pre-malignant condition of the gastric mucosa associated with increased gastric cancer (GC) risk. Analyzing 1,256 gastric samples (1,152 IMs) across 692 subjects from a prospective 10-year study, we identify 26 IM driver genes in diverse pathways including chromatin regulation (ARID1A) and intestinal homeostasis (SOX9). Single-cell and spatial profiles highlight changes in tissue ecology and IM lineage heterogeneity, including an intestinal stem-cell dominant cellular compartment linked to early malignancy. Expanded transcriptome profiling reveals expression-based molecular subtypes of IM associated with incomplete histology, antral/intestinal cell types, ARID1A mutations, inflammation, and microbial communities normally associated with the healthy oral tract. We demonstrate that combined clinical-genomic models outperform clinical-only models in predicting IMs likely to transform to GC. By highlighting strategies for accurately identifying IM patients at high GC risk and a role for microbial dysbiosis in IM progression, our results raise opportunities for GC precision prevention and interception.

Clonal KEAP1 mutations with loss of heterozygosity share reduced immunotherapy efficacy and low immune cell infiltration in lung adenocarcinoma

BACKGROUND

KEAP1 mutations have been associated with reduced survival in lung adenocarcinoma (LUAD) patients treated with immune checkpoint inhibitors (ICIs), particularly in the presence of STK11/KRAS alterations. We hypothesized that, beyond co-occurring genomic events, clonality prediction may help identify deleterious KEAP1 mutations and their counterparts with retained sensitivity to ICIs.

PATIENTS AND METHODS

Beta-binomial modelling of sequencing read counts was used to infer KEAP1 clonal inactivation by combined somatic mutation and loss of heterozygosity (KEAP1 C-LOH) versus partial inactivation [KEAP1 clonal diploid-subclonal (KEAP1 CD-SC)] in the Memorial Sloan Kettering Cancer Center (MSK) MetTropism cohort (N = 2550). Clonality/LOH prediction was compared to a streamlined clinical classifier that relies on variant allele frequencies (VAFs) and tumor purity (TP) (VAF/TP ratio). The impact of this classification on survival outcomes was tested in two independent cohorts of LUAD patients treated with immunotherapy (MSK/Rome N = 237; DFCI N = 461). Immune-related features were studied by exploiting RNA-sequencing data (TCGA) and multiplexed immunofluorescence (DFCI mIF cohort).

RESULTS

Clonality/LOH inference in the MSK MetTropism cohort overlapped with a clinical classification model defined by the VAF/TP ratio. In the ICI-treated MSK/Rome discovery cohort, predicted KEAP1 C-LOH mutations were associated with shorter progression-free survival (PFS) and overall survival (OS) compared to KEAP1 wild-type cases (PFS log-rank P = 0.001; OS log-rank P < 0.001). Similar results were obtained in the DFCI validation cohort (PFS log-rank P = 0.006; OS log-rank P = 0.014). In both cohorts, we did not observe any significant difference in survival outcomes when comparing KEAP1 CD-SC and wild-type tumors. Immune deconvolution and multiplexed immunofluorescence revealed that KEAP1 C-LOH and KEAP1 CD-SC differed for immune-related features.

CONCLUSIONS

KEAP1 C-LOH mutations are associated with an immune-excluded phenotype and worse clinical outcomes among advanced LUAD patients treated with ICIs. By contrast, survival outcomes of patients whose tumors harbored KEAP1 CD-SC mutations were similar to those with KEAP1 wild-type LUADs.

Distinct Mechanisms of Mismatch-Repair Deficiency Delineate Two Modes of Response to Anti-PD-1 Immunotherapy in Endometrial Carcinoma

Mismatch repair-deficient (MMRd) cancers have varied responses to immune-checkpoint blockade (ICB). We conducted a phase II clinical trial of the PD-1 inhibitor pembrolizumab in 24 patients with MMRd endometrial cancer (NCT02899793). Patients with mutational MMRd tumors (6 patients) had higher response rates and longer survival than those with epigenetic MMRd tumors (18 patients). Mutation burden was higher in tumors with mutational MMRd compared with epigenetic MMRd; however, within each category of MMRd, mutation burden was not correlated with ICB response. Pretreatment JAK1 mutations were not associated with primary resistance to pembrolizumab. Longitudinal single-cell RNA-seq of circulating immune cells revealed contrasting modes of antitumor immunity for mutational versus epigenetic MMRd cancers. Whereas effector CD8+ T cells correlated with regression of mutational MMRd tumors, activated CD16+ NK cells were associated with ICB-responsive epigenetic MMRd tumors. These data highlight the interplay between tumor-intrinsic and tumor-extrinsic factors that influence ICB response.

SIGNIFICANCE

The molecular mechanism of MMRd is associated with response to anti-PD-1 immunotherapy in endometrial carcinoma. Tumors with epigenetic MMRd or mutational MMRd are correlated with NK cell or CD8+ T cell-driven immunity, respectively. Classifying tumors by the mechanism of MMRd may inform clinical decision-making regarding cancer immunotherapy. This article is highlighted in the In This Issue feature, p. 247.

Predicting colorectal cancer tumor mutational burden from histopathological images and clinical information using multi-modal deep learning

MOTIVATION

Tumor mutational burden (TMB) is an indicator of the efficacy and prognosis of immune checkpoint therapy in colorectal cancer (CRC). In general, patients with higher TMB values are more likely to benefit from immunotherapy. Though whole-exome sequencing is considered the gold standard for determining TMB, it is difficult to be applied in clinical practice due to its high cost. There are also a few DNA panel-based methods to estimate TMB; however, their detection cost is also high, and the associated wet-lab experiments usually take days, which emphasize the need for faster and cheaper alternatives.

RESULTS

In this study, we propose a multi-modal deep learning model based on a residual network (ResNet) and multi-modal compact bilinear pooling to predict TMB status (i.e. TMB high (TMB_H) or TMB low(TMB_L)) directly from histopathological images and clinical data. We applied the model to CRC data from The Cancer Genome Atlas and compared it with four other popular methods, namely, ResNet18, ResNet50, VGG19 and AlexNet. We tested different TMB thresholds, namely, percentiles of 10%, 14.3%, 15%, 16.3%, 20%, 30% and 50%, to differentiate TMB_H and TMB_L.For the percentile of 14.3% (i.e. TMB value 20) and ResNet18, our model achieved an area under the receiver operating characteristic curve of 0.817 after 5-fold cross-validation, which was better than that of other compared models. In addition, we also found that TMB values were significantly associated with the tumor stage and N and M stages. Our study shows that deep learning models can predict TMB status from histopathological images and clinical information only, which is worth clinical application.

Predicting cancer evolution for patient benefit: Renal cell carcinoma paradigm

Evolutionary features of cancer have important clinical implications, but their evaluation in the clinic is currently limited. Here, we review current approaches to reconstruct tumour subclonal structure and discuss tumour sampling method and experimental design influence. We describe clear-cell renal cell carcinoma (ccRCC) as an exemplar for understanding and predicting cancer evolutionary dynamics. Finally, we discuss how understanding cancer evolution can benefit patients.

Comprehensive Analysis of the Immunogenomics of Triple-Negative Breast Cancer Brain Metastases From LCCC1419

Background

Triple negative breast cancer (TNBC) is an aggressive variant of breast cancer that lacks the expression of estrogen and progesterone receptors (ER and PR) and HER2. Nearly 50% of patients with advanced TNBC will develop brain metastases (BrM), commonly with progressive extracranial disease. Immunotherapy has shown promise in the treatment of advanced TNBC; however, the immune contexture of BrM remains largely unknown. We conducted a comprehensive analysis of TNBC BrM and matched primary tumors to characterize the genomic and immune landscape of TNBC BrM to inform the development of immunotherapy strategies in this aggressive disease.

Methods

Whole-exome sequencing (WES) and RNA sequencing were conducted on formalin-fixed, paraffin-embedded samples of BrM and primary tumors of patients with clinical TNBC (n = 25, n = 9 matched pairs) from the LCCC1419 biobank at UNC—Chapel Hill. Matched blood was analyzed by DNA sequencing as a comparison for tumor WES for the identification of somatic variants. A comprehensive genomics assessment, including mutational and copy number alteration analyses, neoantigen prediction, and transcriptomic analysis of the tumor immune microenvironment were performed.

Results

Primary and BrM tissues were confirmed as TNBC (23/25 primaries, 16/17 BrM) by immunohistochemistry and of the basal intrinsic subtype (13/15 primaries and 16/19 BrM) by PAM50. Compared to primary tumors, BrM demonstrated a higher tumor mutational burden. TP53 was the most frequently mutated gene and was altered in 50% of the samples. Neoantigen prediction showed elevated cancer testis antigen- and endogenous retrovirus-derived MHC class I-binding peptides in both primary tumors and BrM and predicted that single-nucleotide variant (SNV)-derived peptides were significantly higher in BrM. BrM demonstrated a reduced immune gene signature expression, although a signature associated with fibroblast-associated wound healing was elevated in BrM. Metrics of T and B cell receptor diversity were also reduced in BrM.

Conclusions

BrM harbored higher mutational burden and SNV-derived neoantigen expression along with reduced immune gene signature expression relative to primary TNBC. Immune signatures correlated with improved survival, including T cell signatures. Further research will expand these findings to other breast cancer subtypes in the same biobank. Exploration of immunomodulatory approaches including vaccine applications and immune checkpoint inhibition to enhance anti-tumor immunity in TNBC BrM is warranted.

Phasing analysis of lung cancer genomes using a long read sequencer

Chromosomal backgrounds of cancerous mutations still remain elusive. Here, we conduct the phasing analysis of non-small cell lung cancer specimens of 20 Japanese patients. By the combinatory use of short and long read sequencing data, we obtain long phased blocks of 834 kb in N50 length with >99% concordance rate. By analyzing the obtained phasing information, we reveal that several cancer genomes harbor regions in which mutations are unevenly distributed to either of two haplotypes. Large-scale chromosomal rearrangement events, which resemble chromothripsis events but have smaller scales, occur on only one chromosome, and these events account for the observed biased distributions. Interestingly, the events are characteristic of EGFR mutation-positive lung adenocarcinomas. Further integration of long read epigenomic and transcriptomic data reveal that haploid chromosomes are not always at equivalent transcriptomic/epigenomic conditions. Distinct chromosomal backgrounds are responsible for later cancerous aberrations in a haplotype-specific manner.

Clonality and timing of relapsing colorectal cancer metastasis revealed through whole-genome single-cell sequencing

Recurrence of tumor cells following local and systemic therapy is a significant hurdle in cancer. Most patients with metastatic colorectal cancer (mCRC) will relapse, despite resection of the metastatic lesions. A better understanding of the evolutionary history of recurrent lesions is required to identify the spatial and temporal patterns of metastatic progression and expose the genetic and evolutionary determinants of therapeutic resistance. With this goal in mind, here we leveraged a unique single-cell whole-genome sequencing dataset from recurrent hepatic lesions of an mCRC patient. Our phylogenetic analysis confirms that the treatment induced a severe demographic bottleneck in the liver metastasis but also that a previously diverged lineage survived this surgery, possibly after migration to a different site in the liver. This lineage evolved very slowly for two years under adjuvant drug therapy and diversified again in a very short period. We identified several non-silent mutations specific to this lineage and inferred a substantial contribution of chemotherapy to the overall, genome-wide mutational burden. All in all, our study suggests that mCRC subclones can migrate locally and evade resection, keep evolving despite rounds of chemotherapy, and re-expand explosively.