Identification and validation of expressed HLA-binding breast cancer neoepitopes for potential use in individualized cancer therapy

Background

Therapeutic regimens designed to augment the immunological response of a patient with breast cancer (BC) to tumor tissue are critically informed by tumor mutational burden and the antigenicity of expressed neoepitopes. Herein we describe a neoepitope and cognate neoepitope-reactive T-cell identification and validation program that supports the development of next-generation immunotherapies.

Methods

Using GPS Cancer, NantOmics research, and The Cancer Genome Atlas databases, we developed a novel bioinformatic-based approach which assesses mutational load, neoepitope expression, human leukocyte antigen (HLA)-binding prediction, and in vitro confirmation of T-cell recognition to preferentially identify targetable neoepitopes. This program was validated by application to a BC cell line and confirmed using tumor biopsies from two patients with BC enrolled in the Tumor-Infiltrating Lymphocytes and Genomics (TILGen) study.

Results

The antigenicity and HLA-A2 restriction of the BC cell line predicted neoepitopes were determined by reactivity of T cells from HLA-A2-expressing healthy donors. For the TILGen subjects, tumor-infiltrating lymphocytes (TILs) recognized the predicted neoepitopes both as peptides and on retroviral expression in HLA-matched Epstein-Barr virus–lymphoblastoid cell line and BC cell line MCF-7 cells; PCR clonotyping revealed the presence of T cells in the periphery with T-cell receptors for the predicted neoepitopes. These high-avidity immune responses were polyclonal, mutation-specific and restricted to either HLA class I or II. Interestingly, we observed the persistence and expansion of polyclonal T-cell responses following neoadjuvant chemotherapy.

Conclusions

We demonstrate our neoepitope prediction program allows for the successful identification of neoepitopes targeted by TILs in patients with BC, providing a means to identify tumor-specific immunogenic targets for individualized treatment, including vaccines or adoptively transferred cellular therapies.

Reconstructing tumor history in breast cancer: signatures of mutational processes and response to neoadjuvant chemotherapy⋆

BACKGROUND

Different endogenous and exogenous mutational processes act over the evolutionary history of a malignant tumor, driven by abnormal DNA editing, mutagens or age-related DNA alterations, among others, to generate the specific mutational landscape of each individual tumor. The signatures of these mutational processes can be identified in large genomic datasets. We investigated the hypothesis that genomic patterns of mutational signatures are associated with the clinical behavior of breast cancer, in particular chemotherapy response and survival, with a particular focus on therapy-resistant disease.

PATIENTS AND METHODS

Whole exome sequencing was carried out in 405 pretherapeutic samples from the prospective neoadjuvant multicenter GeparSepto study. We analyzed 11 mutational signatures including biological processes such as APOBEC-mutagenesis, homologous recombination deficiency (HRD), mismatch repair deficiency and also age-related or tobacco-induced alterations.

RESULTS

Different subgroups of breast carcinomas were defined mainly by differences in HRD-related and APOBEC-related mutational signatures and significant differences between hormone-receptor (HR)-negative and HR-positive tumors as well as correlations with age, Ki-67 and immunological parameters were observed. We could identify mutational processes that were linked to increased pathological complete response rates to neoadjuvant chemotherapy with high significance. In univariate analyses for HR-positive tumors signatures, S3 (HRD, P < 0.001) and S13 (APOBEC, P = 0.001) as well as exonic mutation rate (P = 0.002) were significantly correlated with increased pathological complete response rates. The signatures S3 (HRD, P = 0.006) and S4 (tobacco, P = 0.011) were prognostic for reduced disease-free survival of patients with chemotherapy-resistant tumors.

CONCLUSION

The results of this investigation suggest that the clinical behavior of a tumor, in particular, response to neoadjuvant chemotherapy and disease-free survival of therapy-resistant tumors, could be predicted by the composition of mutational signatures as an indicator of the individual genomic history of a tumor. After additional validations, mutational signatures might be used to identify tumors with an increased response rate to neoadjuvant chemotherapy and to define therapy-resistant subgroups for future therapeutic interventions.

Transcriptomic silencing as a potential mechanism of treatment resistance

Next-generation sequencing (NGS) has not revealed all the mechanisms underlying resistance to genomically matched drugs. Here, we performed in 1417 tumors whole-exome tumor (somatic)/normal (germline) NGS and whole-transcriptome sequencing, the latter focusing on a clinically oriented 50-gene panel in order to examine transcriptomic silencing of putative driver alterations. In this large-scale study, approximately 13% of the somatic single nucleotide variants (SNVs) were unexpectedly not expressed as RNA; 23% of patients had ≥1 nonexpressed SNV. SNV-bearing genes consistently transcribed were TP53, PIK3CA, and KRAS; those with lower transcription rates were ALK, CSF1R, ERBB4, FLT3, GNAS, HNF1A, KDR, PDGFRA, RET, and SMO. We also determined the frequency of tumor mutations being germline, rather than somatic, in these and an additional 462 tumors with tumor/normal exomes; 33.8% of germline SNVs within the gene panel were rare (not found after filtering through variant information domains) and at risk of being falsely reported as somatic. Both the frequency of silenced variant transcription and the risk of falsely identifying germline mutations as somatic/tumor related are important phenomena. Therefore, transcriptomics is a critical adjunct to genomics when interrogating patient tumors for actionable alterations, because, without expression of the target aberrations, there will likely be therapeutic resistance.

Abstract 2522: RNA rescue somatic mutations and RNA editing in esophageal cancer

The detection of somatic single nucleotide variants is a crucial component to the characterization of the cancer genome. Traditional mutation calling algorithms have focused on comparing the normal and tumor DNA from the same individual. For projects like The Cancer Genome Atlas (TCGA), it became routine to also sequence the tumor RNA. Our computational method, RADIA (RNA and DNA Integrated Analysis), combines the patient-matched normal and tumor DNA with the tumor RNA to detect somatic mutations. The inclusion of the RNA increases the power to detect somatic mutations, especially at low DNA allelic frequencies or when tumor purity is low. By integrating an individual’s DNA and RNA, we are able to detect mutations that would otherwise be missed by traditional algorithms that examine only the DNA. Mutations with high support in the RNA and low support in the DNA are termed RNA Rescue mutations. RNA editing is a post-transcriptional modification of pre-mRNA that has recently been identified as an additional epigenetic mechanism relevant to cancer development and progression. Using patient-matched normal and tumor DNA along with tumor RNA, we are able to identify RNA editing events across the entire transcriptome.

There exists a remarkable geographic variability observed in incidence rates for Esophageal Cancer where more than 80% of cases and deaths occur in developing countries. There is an urgent demand to address the unmet clinical needs for these regions. Mutation of the tumor suppressor gene TP53 is the most frequent genetic alteration in both Esophageal Squamous Cell Cancer (ESCC) and Esophageal Adenocarcinoma (EAC). In a previous study of 59 tumors in Malawi a high proportion of tumors without TP53 mutations was reported. Here, we apply RADIA to a cohort of 61 tumors from Tanzania, and identify RNA Rescue Mutations that were previously missed by DNA mutation callers in significantly mutated genes such as TP53, CDK6, NOTCH1, VEGFA, KMT2D. RNA Editing events in the 3’UTR regions of genes are very common, especially when Alu elements are present, and are known to deregulate microRNA mediated gene regulation. MDM2 is a key oncogene in the p53 pathway with elevated gene expression in many tumor types and is known to be transcriptionally repressed by microRNAs. Here we detect transcriptome-wide RNA Editing events, and identify RNA Editing events in the seed regions of microRNA target sites of genes such as MDM2. We will also show how RNA Editing of certain genes are significantly associated with an RNA-Seq clustering solution. These genes show significantly differential expression between RNA-Seq clusters, indicating that these editing events have a functional impact on the genes they affect. We will further investigate these functional effects and downstream implications on the molecular characterization of the RNA-Seq subtypes.

Citation Format: Amie J. Radenbaugh, J Zachary Sanborn, Yulia Newton, Charlie Vaske, Katherine Van Loon, Eric Collisson. RNA rescue somatic mutations and RNA editing in esophageal cancer [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2019; 2019 Mar 29-Apr 3; Atlanta, GA. Philadelphia (PA): AACR; Cancer Res 2019;79(13 Suppl):Abstract nr 2522.

Evidence for selective silencing of MHC-binding neoepitopes to avoid immune surveillance

2591

Background: Overall response rates to immune checkpoint inhibition (ICI) are < 50% even in TMB-high patients (e.g. Checkmate-227), suggesting other mechanisms of immune escape exist beyond expressing checkpoints. At least 18% of somatic-specific exonic DNA variants are not expressed into mRNA (Rabizadeh, 2018), yet the selection criteria for which variants to silence remains unclear. We sought to determine if immunogenicity of variants factors into their suppression. Methods: Somatic-specific single nucleotide variants (SNVs) were identified from paired tumor/normal whole-exome sequencing (WES), and annotated as expressed if observed in > = 2 RNAseq reads. MHC1 binding affinity for 9-mer neoepitope peptides resulting from said SNVs were predicted using NetMHC within presented HLA-types. Cases with > 200 non-synonymous exonic mutations were designated as TMB-high in accordance with Rizvi et al, 2015. Tumor immune activity was inferred by RNAseq expression of 6 checkpoint/TME markers, as well as by estimating immune infiltration using RNAseq deconvolution of immune genesets (Bindea et al 2013). Significant associations between TMB, neoantigen-load, expressed neoepitope binding affinities, and immune activity were analyzed. Results: Within a clinical database of 1,363 cases with T/N/R sequencing, a total of 147,015 potential neoepitopes were identified. A small but significant enrichment was observed for silencing neoepitopes that are predicted to bind MHC1 (OR = 1.22, p = 2.4e-78 one-sided Fishers test). The silencing rate was similar between the 17% of patients with high TMB vs others, but was increased in 35% of all patients with high inferred immune infiltration (N = 490, OR = 1.30, p = 1.8e-31). A further silencing enrichment was observed in 19% of all patients displaying high immune activity but low PDL1 expression (N = 263, OR = 1.44, p = 4.0e-45). Conclusions: We observe significant preferential silencing of MHC binding neoepitopes. Specifically, when tumor infiltrating immune cells are activated, silencing neoepitopes may be an alternative to checkpoint expression for avoiding an immune cascade. Patients with TILs and silenced neoepitopes may benefit from epigenetic priming therapy prior to ICI therapy.

Abstract P2-09-04: Identification of a neoantigen targeted by tumor-infiltrating lymphocytes in a patient with Her2+ breast cancer

Background: Recent studies have demonstrated that the number of tumor infiltrating lymphocytes (TILs) positively correlates with outcome and response to chemotherapy in patients with HER2+ and Triple-Negative Breast Cancer (TNBC). Furthermore, first studies of immune-checkpoint inhibitors showed promising results in those patients. However, the targets of those TILs remain unknown. Neoantigens, which arise in the process of tumorigenesis, appear as potential targets. They can elicit high avidity, tumor-specific T-cell responses. Thus, it is the aim of our study to ascertainif these TILs are directed against tumor-specific mutations.

Methods: TILs from breast cancer biopsies taken at the time point of diagnosis were expanded by unspecific stimulation. Additionally, we used the Gentle Macs Dissociator in combination with flow cytometry to investigate the number of TILs in the tumor tissue. Furthermore, we performed whole-genome sequencing of tumor tissue and as reference autologous blood cells to determine tumor-specific mutations. Mutations leading to a non-synonymous amino acid change were analyzed for RNA expression of the encoding gene as well as to determine potential neoantigens. Neoantigens were evaluated for their potential binding to the patient's specific HLA molecules. Peptides for potential neoantigens were synthesized, loaded onto autologous antigen presenting cells (APCs) and cocultured with TILs. All IFNγ producing T-cells were clonally expanded and retested for peptide specificity to identify neoantigen specific T-cell clones.

Results: Our flow cytometric analysis of the tumor biopsy for more than 300 patients showed higher frequencies of TILs in TNBC as compared to other types of breast cancer or patients without malignancy. Screening for neoantigen specific T-cells in one patient led to identification of three peptide-specific CD4+ T-cell clones isolated from HER2+ breast cancer tissue taken at the time point of diagnosis. All T-cell clones specifically recognized the same tumor-specific mutation and not the wildtype counterpart. Furthermore, we demonstrated that these T-cell clones also recognized the endogenously expressed mutated antigen. This verified the ability of processing and presentation of the respective protein. Interestingly, we could also isolate a T-cell clone recognizing the same neoantigen in the resected tumor tissue after neoadjuvant therapy. Based on CDR3 sequencing we could prove that the four T-cell clones represented individual clones. This confirms the polyclonal nature of the immune response. Moreover, we showed that the same neoepitope was presented in two different HLA restriction molecules of the patient with three of the clones recognizing it in HLA-DPB1*0401 and one in HLA-DPB1*0201. These results further underline the immunogenicity of this neoantigen.

Conclusion: In conclusion, our data demonstrate tumor-specificity of TILs in a patient with HER2+ breast cancer. Furthermore, we show the feasibility to identify individual cancer specific T-cell targets in breast cancer patients. These results may contribute to the development of targeted patient-specific immunotherapies in the future.

Citation Format: Reimann H, Nguyen A, Hübner H, Erber R, Bausenwein J, Van der Meijden ED, Lux MP, Jud S, Griffioen M, Rauh C, Sanborn JZ, Benz SC, Rabizadeh S, Beckmann MW, Mackensen A, Rübner M, Fasching PA, Kremer AN. Identification of a neoantigen targeted by tumor-infiltrating lymphocytes in a patient with Her2+ breast cancer [abstract]. In: Proceedings of the 2018 San Antonio Breast Cancer Symposium; 2018 Dec 4-8; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2019;79(4 Suppl):Abstract nr P2-09-04.

Genomic and immune infiltration differences between MSI and MSS GI tumors

528

Background: Dysregulation of DNA mismatch repair pathway can lead to microsatellite instability in many GI tumors, and microsatellite instability is an important diagnostic and prognostic marker. Microsatellite instable (MSI) tumors comprise about 15% of colorectal malignancies and can be found in other gastrointestinal (GI) tumor types. We present results of analysis of genomic and immune infiltration differences between MSI and microsatellite stable (MSS) GI tumors spanning multiple cancer types. Methods: A total of 521 GI patients with deep whole exome sequencing (WES) of tumor and blood samples, and whole transcriptomic sequencing (RNA-Seq) (∼200M reads per tumor) were available for this analysis from a commercial database. Variant calling was performed through joint probabilistic analysis of tumor and normal DNA reads, with germline status of variants being determined by heterozygous or homozygous alternate allele fraction in the germline sample. Results: Gene expression and pathway analysis found significantly higher immune signaling in MSI cohort and higher metabolic signaling in MSS cohort. We also found upregulation of structural cellular integrity pathways in MSI tumors. Per-sample deconvolution of immune infiltration using cell type gene markers shows some MSI samples with high CD8 T-cells. Co-expression analysis of checkpoint and TME genes shows higher correlation of FOXP3 and CTLA4 in the MSS cohort compared to the MSI samples, whereas correlation between FOXP3 and PDL1 is decreased. TIM3, LAG3, and OX40 are significantly more expressed in MSI samples than MSS samples. Within the subset of colorectal tumors, additional checkpoints are significantly differentially overexpressed in MSI malignancies. 50 somatic variants are significantly differential in MSI tumors. Conclusions: MSI tumors demonstrably exhibit higher immune signaling, with many immune and checkpoint markers expressed at higher levels in MSI tumors. Some cellular integrity pathways also appear to be up in MSI cohort. A number of potentially important somatic variants are associated with MSI samples.

Abstract 5693: Tumor whole-transcriptome sequencing and multiplex immunohistochemistry of immune cell populations in 158 Asian colorectal cancers

Background: Colorectal cancer (CRC) is a cancer largely refractory to immune checkpoint inhibition. There is substantial interpatient molecular heterogeneity in colorectal cancer reported from studies on tumor RNA and studies on immunohistochemical analyses of fixed tumor tissue. Several major transcriptomic analyses from microarrays and RNA-seq data have identified major transcriptomic subtypes and major activated pathways and deconvoluted cell-type enrichments. Immunohistochemical (IHC) analyses of formalin-fixed, paraffin-embedded (FFPE) tissue microarrays have identified several different histomorphologic/spatial patterns of immune infiltrates in the tumor microenvironment. Here, we perform large-scale -omic analyses on tumor RNA and multiplex IHC simultaneously on 158 Asian colorectal cancers.

Methods: We performed whole-genome sequencing (WGS) (60x tumor, 30x normal) and deep whole-transcriptomic sequencing (RNA-seq) (∼200x106 reads per tumor) on 158 colorectal cancers. To evaluate the spatial patterns of the tumor microenvironment, we constructed a tissue microarray comprising the tumor core, tumor edge and normal adjacent tissue of these 158 CRCs. We performed H&E analyses of the TMA and multiplex immunohistochemistry to simultaneously evaluate 7 markers, i.e., cytokeratin (CK), CD3, CD8, FOXP3, CD68, PD-L1, DAPI, using the an Opal Multiplex fIHC kit. Images were acquired using a Vectra 3 pathology imaging system microscope (PerkinElmer, Waltham, MA, USA).

Results: 32 are microsatellite instability high (MSI-H) tumors and 126 are microsatellite stable. (MSS). The major transcriptomic subtypes (Consensus molecular subtypes (CMS 1-4) and CRC assigner (Goblet-like, Enterocyte, Stem-like, Inflammatory, Transit-amplifying subtypes) were identified with good concordance between both classification systems. CMS1 and Inflammatory subtypes were enriched amongst MSI-H tumors. Major oncogenic pathway activations (RAS, Wnt), cell-cycle and inflammatory signatures (interferon-rich) were also identified across the populations. We deconvoluted cell-type enrichment scores from the transcriptomic data to identify different cell-type enrichment patterns across the cohort. On the TMAs, we identified cell type populations and immune infiltrate patterns in the tumor core and invasive edge across the cohort. Correlations across these analyses will be presented at the meeting.

Conclusions: There is substantial interindividual variability in the transcriptomic landscape and spatial patterns of immune cell infiltrates in CRCs.

Citation Format: Clarinda Wei Chua, Engin Cukuroglu, Elisa Fontana, Si Lin Koo, Joe Poh Yeong, Andy Nguyen, J. Zachary Sanborn, Steve Benz, Emile John Tan, Ronnie Mathew, Ee-Lin Toh, Sarah Boon Ng, Tony Kiat Lim, Anders Jacobsen Skanderup, Shahrooz Rabizadeh, Anguraj Sadanandam, Jonathan Göke, Iain Bee Tan. Tumor whole-transcriptome sequencing and multiplex immunohistochemistry of immune cell populations in 158 Asian colorectal cancers [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2018; 2018 Apr 14-18; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2018;78(13 Suppl):Abstract nr 5693.

Abstract 2253: Identifying pathogenic germline variants in 1,172 cancer patients utilizing a novel variant phasing tool and strict public database curation

The identification of pathogenic or likely pathogenic (P/LP) germline variants in cancer patients is vital in assessing potential genetic causes of cancer risk, as well as incidental rare disease risk. However, sequencing limitations, lack of pedigree data, and unreliable phenotypic data can hinder the discovery of these putative causal germline variants. In order to improve the identification of such variants, we have developed a germline analysis pipeline that phases variants and identifies rare P/LP variants utilizing data from the GnomAD and ClinVar databases.

In order to ensure our pipeline is annotating variants with accurate phenotypic data, we rate the quality of submitters to ClinVar based on their submission history. This history includes review status, agreement between population frequencies and stated clinical significance, and the quality of supporting evidence provided. The variant-phenotype relationship data extracted from ClinVar is valuable in directing us towards cancer-contributing variants, as well as incidental disease risk. However, there are many discrepancies between various institutions' assertion criteria submitted to ClinVar. Many variant entries in ClinVar are submitted as P/LP, yet a large number of these submissions lack evidence or criteria supporting the variant's clinical significance. Our pipeline identifies low-quality submissions, allowing for the inclusion of only high-quality P/LP variant annotations.

In addition to recognizing known P/LP variants from the ClinVar archive, our pipeline provides additional identification of potentially pathogenic novel germline variants via haplotype phasing. Haplotype phasing of germline variants is vital when determining the impact of multiple heterozygous variants within the same gene, but it is difficult to perform such phasing outside of family studies. Our pipeline utilizes normal DNA, tumor DNA, and tumor RNA to predict the phase of variants without the need for full pedigree information.

The pipeline was tested on 1,172 patient samples, with the goal of phasing clinically-associated cancer genes and discovering rare P/LP germline variants. We were able to phase variants in over 50% of the patients, allowing us to identify the nature of compound heterozygosity as it relates to disease risk in these patients. Additionally, our pipeline identified at least 1 high-quality ClinVar P/LP variant per patient in over 25% of the patients. In the vast majority of patients, we were able to identify rare homozygous germline variants. Gene panels specific to cancer type can be used to further investigate which rare variants most likely factor in the patient's disease. Based on these results, as well as our analysis of ClinVar, we assert the need for deep introspection of ClinVar submissions and highlight the utility of RNA data in variant phasing.

Citation Format: Amanda Polley, Charles Vaske, Steve Benz, Patrick Soon-Shiong, Shahrooz Rabizadeh, J Zachary Sanborn. Identifying pathogenic germline variants in 1,172 cancer patients utilizing a novel variant phasing tool and strict public database curation [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2018; 2018 Apr 14-18; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2018;78(13 Suppl):Abstract nr 2253.

Abstract 5725: Systematic identification of tumour-specific neoantigens(by whole-genome sequencing) and correlation between tumour neoantigen burden, PD-L1 expression and immune infiltrates in 158Asian colorectal cancers

Background: Somatic mutations are attractive therapeutic targets for “individualized neoantigen vaccines” because of lack of host central tolerance and reduced risk of autoimmunity. Here, we perform large-scale-omic analyses to assess the neoantigen landscape of colorectal cancer (CRC), a cancer largely refractory to immune-checkpoint inhibition.

Methods: We performed whole genome sequencing (WGS) (60x tumor, 30x normal) and deep whole transcriptomic sequencing (RNA-Seq) (∼200x106 reads per tumor) on 158 colorectal cancers of which 32 are microsatellite instability high (MSI-H) tumours and 126 are microsatellite stable (MSS). Whole exome sequencing (200x tumor, 100x normal) was also performed on 120 tumours. HLA typing, somatic mutations, gene expression and neoepitope predictions were computationally evaluated. Inferred HLA-A alleles were orthogonally validated with Pacbio long-read sequencing. Tissue microarrays (TMAs) with tumour core, tumour edge and normal adjacent tissue of these 158 CRCs were constructed. Histopathological analyses using multiplex immunohistochemistry (mIHC) to simultaneously evaluate 7 markers, i.e. cytokeratin (CK), CD3, CD8, FOXP3, CD68, PD-L1, DAPI, have been performed.

Results: The most common HLAs were, by allele count: A*11:01: 56; A*33:03: 38; B*58:01: 33; B*46:01: 29; B*40:01: 26; C*01:02: 41; C*07:02: 33. Inferred HLA-A alleles from WGS data was largely concordant (&gt;90%) with Pacbio long-read sequencing. There were a median of 2,850 (1229-6909) [MSI] & 213 (27-13,835) [MSS] coding variants, from which 10,487 (4,307-27,365) [MSI] & 726.5 (50-59,096) [MSS] possible neoepitopes were derived, after accounting for epitope processing, the normal proteome and general population variome based on dbSNP, Of these, 5,707 (2,608-15,218) [MSI] & 320 (14-25,243) [MSS] neoepitopes are expressed (based on RNA-Seq). Epitope prediction algorithms revealed a median of 423 (17-1,056) [MSI] & 26 (0-1,102) [MSS] bound & expressed neoepitopes. 5 MSS tumors did not have any predicted bound nor expressed neoepitopes, 112 of 126 (89%) of MSS tumors had at least 5 predicted bound, expressed neoepitopes. Histopathological correlations between extent of immune infiltrates in fixed tissues, tumour PD-L1 expression and neoantigen burden is ongoing.

Conclusions: There is substantial variability in the neoantigen landscape amongst MSI & MSS CRCs. MSI contains multiple-fold higher neo-antigens. Amongst MSS tumours, 89% of patients have at least 5 predicted bound and expressed neo-epitopes that could be targeted in neoantigen-based vaccines for personalized immunotherapy.

Citation Format: Si-Lin Koo, Joe Poh Sheng Yeong, Andy Nguyen, Clarinda Wei Ling Chua, J Zachary Sanborn, Steve Benz, Wah Siew Tan, Choong Leong Tang, Su Yan, Min Hoe Chew, Brian Goh, Chung Yip Chan, Xiao Qing Koh, Alexander Lezhava, Tony Kiat Hon Lim, Shahrooz Rabizadeh, Anders Skanderup, Iain Beehuat Tan. Systematic identification of tumour-specific neoantigens(by whole-genome sequencing) and correlation between tumour neoantigen burden, PD-L1 expression and immune infiltrates in 158Asian colorectal cancers [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2018; 2018 Apr 14-18; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2018;78(13 Suppl):Abstract nr 5725.

Abstract 640: Subsets of HLA alleles are capable of binding neoantigens derived from mutations within cancer driving genes such as KRAS and EGFR

Background: Immunoncology has shown great promise as a low toxicity tool to combat several cancers. Use of checkpoint inhibitors against PD1 or CTLA4 unlocks the immune system’s ability to recognize tumor antigens and, more specifically, neoantigens caused by random mutations within cancers. The vast majority of neoantigens consist of private mutations unique to a patient’s tumor genome, but several cancers harbor recurrent mutations. Mutations in the KRAS gene, such as p.G12V, occur in roughly 25% of colorectal cancers. Mutations in EGFR occur in 10% and 35% of patients with non-small cell lung cancer in the US and East Asia, respectively. Even more prevalent are mutations within the TP53 tumor suppressor gene, with roughly 23000 unique protein variants reported to date. If these mutations in cancer driving genes are so prevalent in cancers, why are neoantigens against these targets not more readily available?

Results: We collected recurrent mutations across a variety of cancer driving genes such as KRAS, EGFR, TP53 and MYC and performed binding analysis using netMHC 3.4 to see which HLA alleles are capable of binding specific cancer mutations such as KRAS p.G12V. Using this method, we report all possible HLA alleles capable of binding these recurrent mutations within cancer genes. We further performed 3-dimensional modeling to determine whether complexes created by the HLA alleles and cancer neoepitopes are stable.

Conclusions: Several HLA alleles are capable of binding recurrent cancer mutations. These include both MHC Class 1 and Class 2 alleles. The variation in alleles capable of binding commonly mutated genes such as EGFR may explain the difference in prevalence of these mutations between geographic populations. Determining whether a certain HLA allele confers resistance to common cancer gene mutations may lead to identification of immune cells within these populations that can recognize neoantigens from commonly mutated cancer genes.

Citation Format: Andrew Nguyen, J Zachary Sanborn, Charles J. Vaske, Shahrooz Rabizadeh, Kayvan Niazi, Patrick Soon-Shiong, Steve Benz. Subsets of HLA alleles are capable of binding neoantigens derived from mutations within cancer driving genes such as KRAS and EGFR [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2017; 2017 Apr 1-5; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2017;77(13 Suppl):Abstract nr 640. doi:10.1158/1538-7445.AM2017-640

Abstract P2-04-26: Identifying patient-specific neoepitopes for cell-based and vaccine immunotherapy across breast cancer classifications reveals rarely shared recurrent neoepitopes

Introduction: Targeted therapies for breast cancers such as trastuzumab and everolimus have durable clinical benefits for patients that express the relevant biomarkers (HER2 and mTOR respectively). Triple negative breast cancer patients lack these biomarkers and are left with few options. Recent advances in immunotherapy agents against PD-1/CTLA4 for patients with melanoma have yielded amazing clinical benefits for a subset of patients and may have similar results in breast cancer patients, but again the vast majority of patients still undergo disease progression. We analyzed whole genome sequencing (WGS) and RNA sequencing data from The Cancer Genome Atlas (TCGA) to identify neoepitopes among breast cancer patients that could be used to develop next-generation, patient-specific cancer immunotherapies. Neoepitopes are tumor specific markers that arise from mutations acquired from cancer and may represent a path to targeted therapies even in triple negative breast cancers.

Results: We analyzed 99 breast cancer patients from TCGA, containing a mixture of PR+/HER2+/ER+ and TNBC classifications. These breast cancer patient samples were selected by the availability of whole genome sequencing (WGS) data, RNA-sequencing data as well as clinical outcome data. We identified an average of 680 potential neoepitopes per patient based solely on WGS data. To further refine and select high quality neoepitopes we restricted these neoepitopes based on gene expression yielding an average of 304 expressed neoepitopes per patient. We predicted each patient's HLA typing using only omics data, which we then used to predict HLA-expressed neoepitope binding analysis resulting in an average of 11 high-quality tumor specific neoepitopes per patient. We identified few recurrent neoepitopes that were bound and expressed, indicating the need for a personalized medicine approach.

Conclusions: Within the TCGA dataset, the majority of neoepitopes among patients with breast cancer were unique to each patient. Rarely within subsets of breast cancers such as HER2+, we identify neoepitopes that are shared between patients. For breast cancer patients who do not respond to targeted therapies, high-throughput identification of neoepitopes could serve as the basis for the development of next-generation, patient-specific immunotherapies.

Citation Format: Nguyen A, Sanborn JZ, Vaske CJ, Rabizadeh S, Niazi K, Soon-Shiong P, Benz SC. Identifying patient-specific neoepitopes for cell-based and vaccine immunotherapy across breast cancer classifications reveals rarely shared recurrent neoepitopes [abstract]. In: Proceedings of the 2016 San Antonio Breast Cancer Symposium; 2016 Dec 6-10; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2017;77(4 Suppl):Abstract nr P2-04-26.

Multiple Hereditary Infundibulocystic Basal Cell Carcinoma Syndrome Associated With a Germline SUFU Mutation

IMPORTANCE

Multiple hereditary infundibulocystic basal cell carcinoma syndrome (MHIBCC) is a rare genodermatosis in which numerous indolent, well-differentiated basal cell carcinomas develop primarily on the face and genitals, without other features characteristic of basal cell nevus syndrome. The cause is unknown. The purpose of the study was to identify a genetic basis for the syndrome and a mechanism by which the associated tumors develop.

OBSERVATIONS

Whole-exome sequencing of 5 tumors and a normal buccal mucosal sample from a patient with MHIBCC was performed. A conserved splice-site mutation in 1 copy of the suppressor of fused gene (SUFU) was identified in all tumor and normal tissue samples. Additional distinct deletions of the trans SUFU allele were identified in all tumor samples, none of which were present in the normal sample.

CONCLUSIONS AND RELEVANCE

A germline SUFU mutation was present in a patient with MHIBCC, and additional acquired SUFU mutations underlie the development of infundibulocystic basal cell carcinomas. The downstream location of the SUFU gene within the sonic hedgehog pathway may explain why its loss is associated with relatively well-differentiated tumors and suggests that MHIBCC will not respond to therapeutic strategies, such as smoothened inhibitors, that target upstream components of this pathway.

Abstract 4512: High-throughput identification of neoepitopes for the development of patient-specific cancer immunotherapies

Introduction: Immunotherapies such as checkpoint inhibitors, CAR T cells, NK cells, and therapeutic vaccines are revolutionizing cancer medicine. Although these agents have resulted in remarkable responses in some patients, many fail to respond, suggesting that a patient-specific approach is needed for immunotherapies to realize their full potential. Here we report the analysis of whole genome sequencing (WGS) and RNA sequencing (RNAseq) data from The Cancer Genome Atlas (TCGA) to identify neoepitopes that may serve as viable immunotherapeutic targets and that could be used to develop next-generation, patient-specific cancer immunotherapies.

Methods: HLA-A compatible neoepitopes were identified by creating all permutations of 9-mer amino acid strings derived from identified single nucleotide variants (SNVs) occurring in expressed genes. All neoepitopes were filtered against a database of 9-mers created using every known human protein along with common variations obtained through dbSNP to reduce off-target effects. HLA typing was determined from WGS data using the HLA forest algorithm. NetMHC 3.4 was used to obtain the predicted binding affinities of neoepitopes to HLA-A alleles

Results: We analyzed 750 patients across 23 cancer types using WGS data and RNAseq data, when available. Mutational and neoepitope loads varied across cancer types, with skin cutaneous melanoma and thyroid carcinoma having the highest and lowest mutational and neoepitope loads, respectively. Neoepitope counts identified by WGS correlated with neoepitope expression identified by RNAseq across a wide variety of cancers (Pearson's r = 0.99 for all cancers combined). The distribution of HLA-A and HLA-DRB1 alleles determined from TCGA were generally comparable to that in the US population; however, the frequency of HLA-DRB1*15:01 (associated with several diseases) was ∼2-fold higher than the US population. Among patients who expressed HLA-A*02:01 (n = 143), the numbers of predicted neoepitopes identified by WGS, neoepitopes expressed per RNAseq, and neoepitopes having binding affinity to HLA-A*02:01 were 23272, 9619, and 138, respectively. Almost all neoepitopes were unique to each patient, with a maximum of only 2-shared neoepitopes among any cancer type. Among 26 triple negative breast cancer (TNBC) samples, the numbers of predicted neoepitopes, expressed neoepitopes, and neoepitopes with affinity to HLA-A*02:01 were 17925, 8184, and 228, respectively. There were no shared neoepitopes among TNBC samples. Across all cancers, ∼6% of neoepitopes occurred in cancer driver genes.

Conclusions: Neoepitopes are rarely shared among cancers, and almost all are unique to each patient. For patients with limited treatment options and poor outcomes, such as TNBC, high-throughput identification of neoepitopes could serve as the basis for the development of next-generation, patient-specific immunotherapies.

Citation Format: Andrew Nguyen, J Zachary Sanborn, Charles J. Vaske, Shahrooz Rabizadeh, Kayvan Niazi, Patrick Soon-Shiong, Steven C. Benz. High-throughput identification of neoepitopes for the development of patient-specific cancer immunotherapies. [abstract]. In: Proceedings of the 107th Annual Meeting of the American Association for Cancer Research; 2016 Apr 16-20; New Orleans, LA. Philadelphia (PA): AACR; Cancer Res 2016;76(14 Suppl):Abstract nr 4512.

Abstract P6-04-14: Integrating whole genome sequencing data with RNAseq, pathway analysis, and quantitative proteomics to determine prognosis after standard adjuvant treatment with trastuzumab and chemotherapy in primary breast cancer patients

Background: Despite improvements in the treatment of HER2+ breast cancer (BC), almost all patients (pts) progress in the metastatic setting. Three examples of resistance mechanisms are: PI3K mutations, lack of ADCC, or low expression of HER2. We recently showed that among 237 pts who had HER2 amplifications, 49% had normal or low levels of HER2 RNA. In addition, quantification of HER2 protein by selected reaction monitoring mass spectrometry (SRM-MS) accurately predicted HER2 expression status compared with IHC (3+)/ISH (≥2.0). Here we report a comprehensive panomic approach that integrates whole genome sequencing (WGS), RNASeq, quantitative proteomics, and pathway analysis to determine associations between tumor molecular profiles and prognosis among HER2+ pts.

Methods: Matched tumor-normal samples (FFPE tumors and blood) were obtained from 58 pts with HER2+ BC who had received standard adjuvant chemotherapy and trastuzumab. Pts were divided into 2 groups: those who had no recurrence after 5 years and those who had developed metastases. The HER2 status of each pt was previously determined using IHC/FISH. Samples underwent WGS and RNASeq according to NantOmics CLIA-approved assay specifications. WGS data were processed using Contraster; RNASeq data confirmed the presence of gene mutations and was used to identify mutational and transcript abundance. PARADIGM was used to reveal associations between gene mutations and pathway levels. SRM-MS was used for proteomics analysis of a panel of 53 proteins. Tumor areas from FFPE tissue sections were analyzed after laser microdissection. Absolute protein quantitation was accomplished through simultaneous detection of endogenous target and synthetic labeled heavy peptide identical to analytical targets. Genetic alterations in germline and tumor DNA were compared in pts with vs without recurrence.

Results: There was no statistically significant difference in the mean concentration of HER2 in the tumors of pts with vs without recurrence: 2.34 fmol/µL vs 2.56 fmol/µL. Other analyzed proteins did not appear to be associated with recurrence; however, expected correlations between pt and tumor characteristics and protein expression were found. With regard to clinically relevant mutations, we found one germline BRCA2 mutation in a pt with no family history of this mutation. The most commonly found somatic mutations were in TP53 (11 pts), AMBRA1 (11 pts), MORC4 (10 pts), SETD2 (8 pts), CDC27 (6 pts), BCLAF1 (5 pts), ZNF479 (4 pts) , PIK3CA (3 pts), PIK3R1 (3 pts), RUNX1 (3 pts), and GATA3 (3 pts).

Conclusion: Whereas HER2 expression status was predictive of OS and PFS in pts treated with trastuzumab (Nuciforo et al. Mol Onc. 2015), in this small exploratory study of HER2+ BC pts, HER2 expression status was not predictive of recurrence. To better understand the molecular mechanisms driving recurrence beyond HER2 status alone, genomic sequencing may define a signature of recurrence after anti-HER2 therapy.

Citation Format: Benz SC, Rabizadeh S, Cecchi F, Beckman MW, Brucker SY, Hartmann A, Golovato J, Hembrough T, Janni W, Rack B, Sanborn JZ, Schneeweiss A, Vaske CJ, Soon-Shiong P, Fasching PA. Integrating whole genome sequencing data with RNAseq, pathway analysis, and quantitative proteomics to determine prognosis after standard adjuvant treatment with trastuzumab and chemotherapy in primary breast cancer patients. [abstract]. In: Proceedings of the Thirty-Eighth Annual CTRC-AACR San Antonio Breast Cancer Symposium: 2015 Dec 8-12; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2016;76(4 Suppl):Abstract nr P6-04-14.

Exome sequencing of desmoplastic melanoma identifies recurrent NFKBIE promoter mutations and diverse activating mutations in the MAPK pathway

Desmoplastic melanoma is an uncommon variant of melanoma with sarcomatous histology, distinct clinical behavior and unknown pathogenesis. We performed low-coverage genome and high-coverage exome sequencing of 20 desmoplastic melanomas, followed by targeted sequencing of 293 genes in a validation cohort of 42 cases. A high mutation burden (median of 62 mutations/Mb) ranked desmoplastic melanoma among the most highly mutated cancers. Mutation patterns strongly implicate ultraviolet radiation as the dominant mutagen, indicating a superficially located cell of origin. Newly identified alterations included recurrent promoter mutations of NFKBIE, encoding NF-κB inhibitor ɛ (IκBɛ), in 14.5% of samples. Common oncogenic mutations in melanomas, in particular in BRAF (encoding p.Val600Glu) and NRAS (encoding p.Gln61Lys or p.Gln61Arg), were absent. Instead, other genetic alterations known to activate the MAPK and PI3K signaling cascades were identified in 73% of samples, affecting NF1, CBL, ERBB2, MAP2K1, MAP3K1, BRAF, EGFR, PTPN11, MET, RAC1, SOS2, NRAS and PIK3CA, some of which are candidates for targeted therapies.

Glioblastoma adaptation traced through decline of an IDH1 clonal driver and macro-evolution of a double-minute chromosome

BACKGROUND

Glioblastoma (GBM) is the most common malignant brain cancer occurring in adults, and is associated with dismal outcome and few therapeutic options. GBM has been shown to predominantly disrupt three core pathways through somatic aberrations, rendering it ideal for precision medicine approaches.

METHODS

We describe a 35-year-old female patient with recurrent GBM following surgical removal of the primary tumour, adjuvant treatment with temozolomide and a 3-year disease-free period. Rapid whole-genome sequencing (WGS) of three separate tumour regions at recurrence was carried out and interpreted relative to WGS of two regions of the primary tumour.

RESULTS

We found extensive mutational and copy-number heterogeneity within the primary tumour. We identified a TP53 mutation and two focal amplifications involving PDGFRA, KIT and CDK4, on chromosomes 4 and 12. A clonal IDH1 R132H mutation in the primary, a known GBM driver event, was detectable at only very low frequency in the recurrent tumour. After sub-clonal diversification, evidence was found for a whole-genome doubling event and a translocation between the amplified regions of PDGFRA, KIT and CDK4, encoded within a double-minute chromosome also incorporating miR26a-2. The WGS analysis uncovered progressive evolution of the double-minute chromosome converging on the KIT/PDGFRA/PI3K/mTOR axis, superseding the IDH1 mutation in dominance in a mutually exclusive manner at recurrence, consequently the patient was treated with imatinib. Despite rapid sequencing and cancer genome-guided therapy against amplified oncogenes, the disease progressed, and the patient died shortly after.

CONCLUSION

This case sheds light on the dynamic evolution of a GBM tumour, defining the origins of the lethal sub-clone, the macro-evolutionary genomic events dominating the disease at recurrence and the loss of a clonal driver. Even in the era of rapid WGS analysis, cases such as this illustrate the significant hurdles for precision medicine success.

Transcription restores DNA repair to heterochromatin, determining regional mutation rates in cancer genomes

Somatic mutations in cancer are more frequent in heterochromatic and late-replicating regions of the genome. We report that regional disparities in mutation density are virtually abolished within transcriptionally silent genomic regions of cutaneous squamous cell carcinomas (cSCCs) arising in an XPC(-/-) background. XPC(-/-) cells lack global genome nucleotide excision repair (GG-NER), thus establishing differential access of DNA repair machinery within chromatin-rich regions of the genome as the primary cause for the regional disparity. Strikingly, we find that increasing levels of transcription reduce mutation prevalence on both strands of gene bodies embedded within H3K9me3-dense regions, and only to those levels observed in H3K9me3-sparse regions, also in an XPC-dependent manner. Therefore, transcription appears to reduce mutation prevalence specifically by relieving the constraints imposed by chromatin structure on DNA repair. We model this relationship among transcription, chromatin state, and DNA repair, revealing a new, personalized determinant of cancer risk.

The somatic genomic landscape of glioblastoma

We describe the landscape of somatic genomic alterations based on multidimensional and comprehensive characterization of more than 500 glioblastoma tumors (GBMs). We identify several novel mutated genes as well as complex rearrangements of signature receptors, including EGFR and PDGFRA. TERT promoter mutations are shown to correlate with elevated mRNA expression, supporting a role in telomerase reactivation. Correlative analyses confirm that the survival advantage of the proneural subtype is conferred by the G-CIMP phenotype, and MGMT DNA methylation may be a predictive biomarker for treatment response only in classical subtype GBM. Integrative analysis of genomic and proteomic profiles challenges the notion of therapeutic inhibition of a pathway as an alternative to inhibition of the target itself. These data will facilitate the discovery of therapeutic and diagnostic target candidates, the validation of research and clinical observations and the generation of unanticipated hypotheses that can advance our molecular understanding of this lethal cancer.

Double minute chromosomes in glioblastoma multiforme are revealed by precise reconstruction of oncogenic amplicons

DNA sequencing offers a powerful tool in oncology based on the precise definition of structural rearrangements and copy number in tumor genomes. Here, we describe the development of methods to compute copy number and detect structural variants to locally reconstruct highly rearranged regions of the tumor genome with high precision from standard, short-read, paired-end sequencing datasets. We find that circular assemblies are the most parsimonious explanation for a set of highly amplified tumor regions in a subset of glioblastoma multiforme samples sequenced by The Cancer Genome Atlas (TCGA) consortium, revealing evidence for double minute chromosomes in these tumors. Further, we find that some samples harbor multiple circular amplicons and, in some cases, further rearrangements occurred after the initial amplicon-generating event. Fluorescence in situ hybridization analysis offered an initial confirmation of the presence of double minute chromosomes. Gene content in these assemblies helps identify likely driver oncogenes for these amplicons. RNA-seq data available for one double minute chromosome offered additional support for our local tumor genome assemblies, and identified the birth of a novel exon made possible through rearranged sequences present in the double minute chromosomes. Our method was also useful for analysis of a larger set of glioblastoma multiforme tumors for which exome sequencing data are available, finding evidence for oncogenic double minute chromosomes in more than 20% of clinical specimens examined, a frequency consistent with previous estimates.

The UCSC Interaction Browser: multidimensional data views in pathway context

High-throughput data sets such as genome-wide protein–protein interactions, protein–DNA interactions and gene expression data have been published for several model systems, especially for human cancer samples. The University of California, Santa Cruz (UCSC) Interaction Browser (http://sysbio.soe.ucsc.edu/nets) is an online tool for biologists to view high-throughput data sets simultaneously for the analysis of functional relationships between biological entities. Users can access several public interaction networks and functional genomics data sets through the portal as well as upload their own networks and data sets for analysis. Users can navigate through correlative relationships for focused sets of genes belonging to biological pathways using a standard web browser. Using a new visual modality called the CircleMap, multiple ‘omics’ data sets can be viewed simultaneously within the context of curated, predicted, directed and undirected regulatory interactions. The Interaction Browser provides an integrative viewing of biological networks based on the consensus of many observations about genes and their products, which may provide new insights about normal and disease processes not obvious from any isolated data set.

Abstract P2-06-05: Single-cell RNA sequencing of paclitaxol-treated breast cancer cell lines to find individual cell response

Cancer treatments act on a population of cells, each of which may experience different individual responses to treatment. Such differential response will result in resistance to treatment even if a majority of cancerous cells are eliminated. To examine differential cell response, we simultaneously profiled the gene expression and mutation spectrum of individual cells from the MDAMB231 cell line using next generation sequencing of isolated RNA. A total of 23 transcriptomes were characterized from paclitaxel-treated and paclitaxel-surviving cells. We found significant different changes in mutation rates between paclitaxel treated cells, with a dose-dependent increase in single nucleotide changes in RNA in paclitaxel-treated cells. Cells undergoing exposure to paclitaxel also showed higher pathway activity in SRC, as well as an integrin switch from ITGB1 to ITGB3. In contrast, cells that survived a high dose of paclitaxel showed an insignificant number of single nucleotide changes, suggesting that these cells either evaded initial paclitaxel exposure or were better able to repair the effects of paclitaxel exposure. Despite the RNA sequence similarity between surviving and untreated cells, there were changes in gene expression and pathway activities including higher PI3K activity. Paclitaxel-surviving cells also showed activation of pathways associated with higher proliferation.

Citation Information: Cancer Res 2012;72(24 Suppl):Abstract nr P2-06-05.

P3-06-07: Integrated Genomic and Pathway Analysis Reveals Key Pathways across Breast Subtypes

Cancer is a disease of genomic perturbations that lead to dysregulation of multiple pathways within the cellular system. While common pathways are believed to be shared within specific cancer types, the mechanisms behind why particular patients respond differently to treatment is not well understood. Genomics studies such as The Cancer Genome Atlas (TCGA) and Stand Up To Cancer (SU2C) attempt to address this issue by collecting large-scale whole-genome measurements of mRNA expression, DNA copy number, and epigenetic features. Common analysis of these measurements integrates data across multiple samples to distinguish signal from noise. However, serious challenges remain in identifying genomic features and pathways significant for prognosis and clinical treatment classifications.

We have created the Five3 Analysis Pipeline to streamline discovery of individual samples’ mutations, small indels, copy number alterations, genome rearrangements, expression changes, and resulting pathway activities. This pipeline is capable of processing and integrating data from both next generation sequencing and microarray platforms in the analysis of single or multiple tumor samples. Our sequence analysis corrects for both tumor sample impurity and germline variation to accurately identify somatic mutations present in the tumor. Our pathway analysis incorporates gene copy number, mutations, expression, and promoter methylation on a superimposed pathway constructed from several curated pathway databases in a sample-specific manner.

By applying this pipeline to the TCGA breast cancer datasets, we recapitulate established breast subtypes at a pathway-dependent level. For example, basal tumors appear enriched for proliferation pathways compared to luminal samples within this cohort. Expanding the pathway analysis to include TCGA lung cancer samples, we find similar subnetworks activated between basal and squamous lung and between luminal and lung adenocarcinomas. This hints at similar genomic mechanisms for these subtypes independent of tissue of origin. Finally, by analyzing genomic alterations across all breast cancers we see mutational clusters in PIK3CA that correspond with publicly-available hotspots [1]. As suggested by previous reports [2], we find that samples with mutations clustered in exon 10 exhibit differential pathway activities relative to those samples with mutations clustered in exon 21, independent of subtype and TP53 mutation status. These results show the power of this integrated genomic platform in elucidating pathway signatures and the need to consider cross cancer analyses to identify shared tumorigenic mechanisms that may suggest common therapeutic targets.

[1] Forbes, S.A et al. COSMIC: mining complete cancer genomes in the Catalogue of Somatic Mutations in Cancer. Nucl. Acids Res. (2011) 39: D945-D950

[2] Vasudevan KM et al. AKT-independent signaling downstream of oncogenic PIK3CA mutations in human cancer. Cancer Cell 2009 Jul.;16(1):21–32.

Citation Information: Cancer Res 2011;71(24 Suppl):Abstract nr P3-06-07.

Abstract 4985: The UCSC Cancer Genomics Browser

The UCSC Cancer Genomics Browser (https://genome-cancer.ucsc.edu) comprises a suite of web-based tools to integrate, visualize and analyze cancer genomics and clinical data. The browser displays whole-genome views of genome-wide experimental measurements for multiple samples alongside their associated clinical information. Multiple datasets can be viewed simultaneously as coordinated “heatmap tracks” to compare across studies or different data modalities. Users can order, filter, aggregate, classify, and display data interactively based on any given feature set including clinical features, annotated biological pathways, and user-contributed collections of genes. Integrated standard statistical tools provide dynamic quantitative analysis within all available datasets. The browser hosts a growing body of publicly available cancer genomics data from a variety of cancer types, including data generated from the Cancer Genome Atlas project. Multiple consortiums use the browser on confidential prepublication data enabled by private installations. Many new features have been added, including the hgMicroscope tumor image viewer, hgSignature for real-time genomic signature evaluation on any browser track, and PARADIGM pathway tracks to display integrative pathway activities. The browser is integrated with the UCSC Genome Browser; thus inheriting and integrating the Genome Browser's rich set of human biology and genetics data that enhances the interpretability of the cancer genomics data.

Citation Format: {Authors}. {Abstract title} [abstract]. In: Proceedings of the 102nd Annual Meeting of the American Association for Cancer Research; 2011 Apr 2-6; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2011;71(8 Suppl):Abstract nr 4985. doi:10.1158/1538-7445.AM2011-4985

The UCSC Cancer Genomics Browser: update 2011

The UCSC Cancer Genomics Browser (https://genome-cancer.ucsc.edu) comprises a suite of web-based tools to integrate, visualize and analyze cancer genomics and clinical data. The browser displays whole-genome views of genome-wide experimental measurements for multiple samples alongside their associated clinical information. Multiple data sets can be viewed simultaneously as coordinated 'heatmap tracks' to compare across studies or different data modalities. Users can order, filter, aggregate, classify and display data interactively based on any given feature set including clinical features, annotated biological pathways and user-contributed collections of genes. Integrated standard statistical tools provide dynamic quantitative analysis within all available data sets. The browser hosts a growing body of publicly available cancer genomics data from a variety of cancer types, including data generated from the Cancer Genome Atlas project. Multiple consortiums use the browser on confidential prepublication data enabled by private installations. Many new features have been added, including the hgMicroscope tumor image viewer, hgSignature for real-time genomic signature evaluation on any browser track, and 'PARADIGM' pathway tracks to display integrative pathway activities. The browser is integrated with the UCSC Genome Browser; thus inheriting and integrating the Genome Browser's rich set of human biology and genetics data that enhances the interpretability of the cancer genomics data.

Inference of patient-specific pathway activities from multi-dimensional cancer genomics data using PARADIGM

Motivation: High-throughput data is providing a comprehensive view of the molecular changes in cancer tissues. New technologies allow for the simultaneous genome-wide assay of the state of genome copy number variation, gene expression, DNA methylation and epigenetics of tumor samples and cancer cell lines.

Analyses of current data sets find that genetic alterations between patients can differ but often involve common pathways. It is therefore critical to identify relevant pathways involved in cancer progression and detect how they are altered in different patients.

Results: We present a novel method for inferring patient-specific genetic activities incorporating curated pathway interactions among genes. A gene is modeled by a factor graph as a set of interconnected variables encoding the expression and known activity of a gene and its products, allowing the incorporation of many types of omic data as evidence. The method predicts the degree to which a pathway's activities (e.g. internal gene states, interactions or high-level ‘outputs’) are altered in the patient using probabilistic inference.

Compared with a competing pathway activity inference approach called SPIA, our method identifies altered activities in cancer-related pathways with fewer false-positives in both a glioblastoma multiform (GBM) and a breast cancer dataset. PARADIGM identified consistent pathway-level activities for subsets of the GBM patients that are overlooked when genes are considered in isolation. Further, grouping GBM patients based on their significant pathway perturbations divides them into clinically-relevant subgroups having significantly different survival outcomes. These findings suggest that therapeutics might be chosen that target genes at critical points in the commonly perturbed pathway(s) of a group of patients.

Availability:Source code available at http://sbenz.github.com/Paradigm

Contact:jstuart@soe.ucsc.edu

Supplementary information:Supplementary data are available at Bioinformatics online.

Abstract 2001: Non-negative matrix factorization (NMF) as a clinical classifier: An example with chemotherapy response in ovarian cancer

Large-scale whole-genome assays such as aCGH and expression profiling are providing an overabundance of data on disease states, yet much of the information is not relevant to driving the disease. There is a need for generating concise explanatory modules with powerful predictive results.

Here we describe a semi-supervised machine-learning method for classifying -omics data with associated clinical features. Using NMF a matrix of -omics data is convoluted step-wise into two matrices; a table of k “metagenes” containing coefficients for membership of each gene, and a table of prediction strengths for each patient into each k-class. Where the class of a patient is known we initially set the prediction strength for that class high, otherwise the prediction strength for each class is set equally. As the NMF algorithm progresses, the class prediction can be swapped if this is supported by the -omics data. Initially the metagene coefficients are unknown and seeded randomly. As NMF does not converge onto a unique solution, multiple metagenes were learned starting at different random coefficients. These multiple models were cross-validated as predictors against held-out sets of patients using a correlation statistic different from the one used to train the model. Where a held-out patient correlates with none of the model's k-classes they are considered members of a novel class, and can be abstained from classification a priori.

We applied our method on the TCGA ovarian serous carcinoma. In this study aCGH and expression profiles were was taken before platinum treatment and the time to relapse recorded. These -omics measurements were then integrated into inferred pathway activities using a computational model of the central dogma (DIGMA). We predicted platinum sensitivity (no relapse or relapse after 12 mo) vs. platinum resistance (relapse within 3mo) on these inferred pathway activities. Consistently, the model with the highest predictive accuracy in a cross-validation setting correctly placed patients into the correct class &gt;80% of the time. Using abstaining it is possible to achieve &gt;90% correct prediction upon a subset of the patients. These results provide useful prognostic indicators as well as indicators for which events drive disease.

Citation Format: {Authors}. {Abstract title} [abstract]. In: Proceedings of the 101st Annual Meeting of the American Association for Cancer Research; 2010 Apr 17-21; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2010;70(8 Suppl):Abstract nr 2001.

Abstract 2015: Patient-specific pathway analysis using DIGMA identifies key activities in multiple cancers

Cancer is a disease of genomic perturbations that lead to dysregulation of multiple pathways within the cellular system. While common pathways are believed to be shared within specific cancer types, the mechanisms of why particular patients respond differently to treatment is not fully understood. Current -omics studies such as The Cancer Genome Atlas (TCGA) and Stand Up To Cancer (SU2C) have attempted to address this issue by using large-scale whole-genome measurements of mRNA expression, DNA copy number, and epigenetic features. Typical analysis of these measurements relies on integrating data from multiple samples to distinguish signal from noise. However, few analytical methods allow for sample-specific differences to identify features and pathways that are significant for prognosis and clinical treatment classifications.

We developed a pathway inference method called DIGMA (Directed and Integrated Graphical Modeling of Activities) to identify patient- and sample-specific pathway activities. DIGMA is capable of inferring individual gene and protein level measurements within pathways as well as overall pathway alterations specific to an individual tumor. DIGMA models each gene's “Central Dogma” within an overall pathway structure using probabilistic graphical models (PGMs). Individual -omic measurements are attached to their represented protein within the network and proteins are connected using interactions defined in curated pathway models. This representation is flexible enough to support any number of -omics measurements, which can be connected to the protein and pathway in biologically meaningful ways.

Applying our method to TCGA ovarian serous carcinoma and glioblastoma multiforme samples identified pathways both significantly activated across a large majority of the cohort as well as pathways useful for classification of samples that responded well to treatment. We used random simulations to measure the significance of the pathway activities and establish a false discovery rate. These results suggest the ability to identify critical intervention points for therapeutics at a sample-specific level.

Citation Format: {Authors}. {Abstract title} [abstract]. In: Proceedings of the 101st Annual Meeting of the American Association for Cancer Research; 2010 Apr 17-21; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2010;70(8 Suppl):Abstract nr 2015.

UCSC cancer genomics browser

Abstract #2022

As experimental techniques for a comprehensive survey of the cancer landscape mature, there is a great demand in the cancer research field to develop advanced analysis and visualization tools for the characterization and integrative analysis of the large, complex genomic datasets arising from different technology platforms.&#x2028; The UCSC Cancer Genomics Browser is a suite of web-based tools designed to integrate, visualize and analyze genomic and clinical data. The secured-access browser, available at https://cancer.cse.ucsc.edu/, consists of three major components: hgHeatmap, hgFeatureSorter, and hgPathSorter. The main panel, hgHeatmap, displays a whole-genome-oriented view of genome-wide experimental measurements for individual and sets of samples/patients alongside their clinical information. hgFeatureSorter and hgPathSorter together enable investigators to order, filter, aggregate and display data interactively based on any given feature set ranging from clinical features to annotated biological pathways to user-edited collections of genes. Standard and advanced statistical tools are available to provide quantitative analysis of whole genomic data or any of its subsets. The UCSC Cancer Genomics Browser is an extension of the UCSC Genome Browser; thus it inherits and integrates the Genome Browser's existing rich set of human biology and genetics data to enhance the interpretability of cancer genomics data.&#x2028; We demonstrate the UCSC Cancer Genomics Browser by integrating several independent studies on breast cancer including the I-SPY chemotherapy clinical trial and other studies focused on chemotherapeutic response or long-term survival. The types of data that are visualized and analyzed by the browser include microarray measurements of gene expression, copy number variation and phosphoprotein expression, MRI imaging measurements, and clinical parameters.&#x2028; Collectively, these tools facilitate a synergistic interaction among clinicians, experimental biologists, and bioinformaticians. They enable cancer researchers to better explore the breadth and depth of the cancer genomics data resources, and to further characterize molecular pathways that influence cellular dynamics and stability in cancer. Ultimately, insights gained by applying these tools may advance our knowledge of human cancer biology and stimulate the discovery of new prognostic and diagnostic markers, as well as the development of therapeutic and prevention strategies.&#x2028; Funding sources: CALGB CA31964 and CA33601, ACRIN U01 CA079778 and CA080098, NCI SPORE CA58207, California Institute for Quantitative Biosciences, NHGRI.

Citation Information: Cancer Res 2009;69(2 Suppl):Abstract nr 2022.