Petros Giannikopoulos

The Impact of EGFR T790M Mutations and BIM mRNA Expression on Outcome in Patients with EGFR-Mutant NSCLC Treated with Erlotinib or Chemotherapy in the Randomized Phase III EURTAC Trial

Purpose: Concomitant genetic alterations could account for transient clinical responses to tyrosine kinase inhibitors of the EGF receptor (EGFR) in patients harboring activating EGFR mutations. Experimental Design: We have evaluated the impact of pretreatment somatic EGFR T790M mutations, TP53 mutations, and Bcl-2 interacting mediator of cell death (BCL2L11, also known as BIM) mRNA expression in 95 patients with EGFR-mutant non–small-cell lung cancer (NSCLC) included in the EURTAC trial (trial registration: NCT00446225). Results: T790M mutations were detected in 65.26% of patients using our highly sensitive method based on laser microdissection and peptide-nucleic acid-clamping PCR, which can detect the mutation at an allelic dilution of 1 in 5,000. Progression-free survival (PFS) to erlotinib was 9.7 months for those with T790M mutations and 15.8 months for those without, whereas among patients receiving chemotherapy, it was 6 and 5.1 months, respectively (P < 0.0001). PFS to erlotinib was 12.9 months for those with high and 7.2 months for those with low/intermediate BCL2L11 expression levels, whereas among chemotherapy-treated patients, it was 5.8 and 5.5 months, respectively (P = 0.0003). Overall survival was 28.6 months for patients with high BCL2L11 expression and 22.1 months for those with low/intermediate BCL2L11 expression (P = 0.0364). Multivariate analyses showed that erlotinib was a marker of longer PFS (HR = 0.35; P = 0.0003), whereas high BCL2L11 expression was a marker of longer PFS (HR = 0.49; P = 0.0122) and overall survival (HR = 0.53; P = 0.0323). Conclusions: Low-level pretreatment T790M mutations can frequently be detected and can be used for customizing treatment with T790M-specific inhibitors. BCL2L11 mRNA expression is a biomarker of survival in EGFR-mutant NSCLC and can potentially be used for synthetic lethality therapies. Clin Cancer Res; 20(7); 2001–10. ©2014 AACR.

Evolution and clinical impact of co-occurring genetic alterations in advanced-stage EGFR-mutant lung cancers

A widespread approach to modern cancer therapy is to identify a single oncogenic driver gene and target its mutant-protein product (for example, EGFR-inhibitor treatment in EGFR-mutant lung cancers). However, genetically driven resistance to targeted therapy limits patient survival. Through genomic analysis of 1,122 EGFR-mutant lung cancer cell-free DNA samples and whole-exome analysis of seven longitudinally collected tumor samples from a patient with EGFR-mutant lung cancer, we identified critical co-occurring oncogenic events present in most advanced-stage EGFR-mutant lung cancers. We defined new pathways limiting EGFR-inhibitor response, including WNT/β-catenin alterations and cell-cycle-gene (CDK4 and CDK6) mutations. Tumor genomic complexity increases with EGFR-inhibitor treatment, and co-occurring alterations in CTNNB1 and PIK3CA exhibit nonredundant functions that cooperatively promote tumor metastasis or limit EGFR-inhibitor response. This study calls for revisiting the prevailing single-gene driver-oncogene view and links clinical outcomes to co-occurring genetic alterations in patients with advanced-stage EGFR-mutant lung cancer.

Oncogenic activation of the PI3-kinase p110β isoform via the tumor-derived PIK3Cβ(D1067V) kinase domain mutation.

Activation of the phosphoinositide 3-kinase (PI3K) pathway occurs widely in human cancers. Although somatic mutations in the PI3K pathway genes PIK3CA and PTEN are known to drive PI3K pathway activation and cancer growth, the significance of somatic mutations in other PI3K pathway genes is less clear. Here, we establish the signaling and oncogenic properties of a recurrent somatic mutation in the PI3K p110β isoform that resides within its kinase domain (PIK3CβD1067V). We initially observed PIK3CβD1067V by exome sequencing analysis of an EGFR-mutant non-small cell lung cancer (NSCLC) tumor biopsy from a patient with acquired erlotinib resistance. On the basis of this finding, we hypothesized that PIK3CβD1067V might function as a novel tumor-promoting genetic alteration, and potentially an oncogene, in certain cancers. Consistent with this hypothesis, analysis of additional tumor exome data sets revealed the presence of PIK3CβD1067V at low frequency in other patient tumor samples (including renal cell carcinoma, glioblastoma multiforme, head and neck squamous cell carcinoma, melanoma, thyroid carcinoma and endometrial carcinoma). Functional studies revealed that PIK3CβD1067V promoted PI3K pathway signaling, enhanced cell growth in vitro, and was sufficient for tumor formation in vivo. Pharmacologic inhibition of PIK3Cβ with TGX-221 (isoform-selective p110β inhibitor) specifically suppressed growth in patient-derived renal-cell carcinoma cells with endogenous PIK3CβD1067V and in NIH-3T3 and human EGFR-mutant lung adenocarcinoma cells engineered to express this mutant PI3K. In the EGFR-mutant lung adenocarcinoma cells, expression of PIK3CβD1067V also promoted erlotinib resistance. Our data establish a novel oncogenic form of PI3K, revealing the signaling and oncogenic properties of PIK3CβD1067V and its potential therapeutic relevance in cancer. Our findings provide new insight into the genetic mechanisms underlying PI3K pathway activation in human tumors and indicate that PIK3CβD1067V is a rational therapeutic target in certain cancers.

Evolution and clinical impact of genetic epistasis within EGFR-mutant lung cancers

The current understanding of tumorigenesis is largely centered on a monogenic driver oncogene model. This paradigm is incompatible with the prevailing clinical experience in most solid malignancies: monotherapy with a drug directed against an individual oncogenic driver typically results in incomplete clinical responses and eventual tumor progression1-7. By profiling the somatic genetic alterations present in over 2,000 cases of lung cancer, the leading cause of cancer mortality worldwide8,9, we show that combinations of functional genetic alterations, i.e. genetic collectives dominate the landscape of advanced-stage disease. We highlight this polygenic landscape and evolution of advanced-stage non-small cell lung cancer (NSCLC) through the spatial-temporal genomic profiling of 7 distinct tumor biopsy specimens and 6 plasma specimens obtained from an EGFR-mutant NSCLC patient at (1) initial diagnosis of early-stage disease, (2) metastatic progression, (3) sequential treatment and resistance to 2 EGFR inhibitors, (4) death. The comprehensive genomic analysis of this case, coupled with circulating free (cf) tumor DNA profiling of additional advanced-stage EGFR-mutant NSCLC clinical cohorts with associated treatment responses uncovered features of evolutionary selection for multiple concurrent gene alterations: including the presence of EGFR inhibitor-sensitive (EGFRL858R;EGFRexon19del) or inhibitor-resistant (EGFRT790M;EGFRC797S) forms of oncogenic EGFR along with cell cycle gene alterations (e.g. in CDK4/6, CCNE1, RB1) and activating alterations in WNT/β-catenin and PI3K pathway genes, which our data suggest can cooperatively impart non-redundant functions to limit EGFR targeted therapy response and/or promote tumor progression. Moreover, evidence of an unanticipated parallel evolution of both EGFRT790M and two distinct forms of oncogenic PIK3CA was observed. Our study provides a large-scale clinical and genetic dataset of advanced-stage EGFR-mutant NSCLC, a rationale for specific polytherapy strategies such as EGFR and CDK4/6 inhibitor co-treatment to potentially enhance clinical outcomes, and prompts a re-evaluation of the prevailing paradigm of monogenic-based molecular stratification for targeted therapy. Instead, our findings highlight an alternative model of genetic collectives that operate through epistasis to drive lung cancer progression and therapy resistance.

Abstract 2173: Robust estimation of mutation burden

Developing a more robust approach to measure mutational burden is of central importance to improving the characterization of the molecular profile of tumor and may improve our ability to predict tumor progression or response to therapy in patients. Mutational burden is typically calculated as a direct enumeration of called somatic mutations per megabase covered. However, there is a growing appreciation that tumor purity, variable sequencing coverage, and copy number alterations can substantially impact the accurate identification any specific somatic mutation. Furthermore, population genetic theory and empirical data indicate that in many cases the vast majority of somatic mutations appear in only a small subpopulation of tumor cells, a context in which there is a high likelihood that an individual subclonal mutation may not be identified by conventional analysis. This tendency to miss low frequency mutations is highly variable and dependent, in part, upon sample purity and results in a strong source of bias not addressed in existing methods to measure variant allele frequencies. We present a novel computational method that incorporates these sources of bias in a coherent probabilistic framework that enables maximum-likelihood inference of relevant population parameters such as mutation burden. We apply our method to simulated data as well as patient tumor samples diluted with varying known proportions of normal DNA. We show that our approach allows us to generate estimates of mutation burden that are robust to the substantial variations in purity and sequencing coverage that are frequently encountered in patient tumor analysis. Hence, our novel method may improve the accurate detection and quantification of variant alleles in patient tumors to better understand their genetic landscape and guide clinical management. Citation Format: Oscar Westesson, Rasmus Nielsen, John St John, Aleah Caulin, Nicholas Hahner, Stewart Stewart, Catherine Foo, Kimberly Lung, Jeff Catalano, Mandy Lee, Petros Giannikopoulos, Will Polkinghorn, Jonathan Wiessman, Aviv Regev, Trever Bivona. Robust estimation of mutation burden. [abstract]. In: Proceedings of the 106th Annual Meeting of the American Association for Cancer Research; 2015 Apr 18-22; Philadelphia, PA. Philadelphia (PA): AACR; Cancer Res 2015;75(15 Suppl):Abstract nr 2173. doi:10.1158/1538-7445.AM2015-2173

Abstract 4707: Comprehensive integrated genomic analysis

Proceedings: AACR Annual Meeting 2014; April 5-9, 2014; San Diego, CA Despite recent advances in the understanding of the biology and genetics of lung cancer, and despite the introduction of multiplex somatic mutation testing in the clinic, the long-term survival for all lung cancer patients, particularly for those with advanced disease, remains low. Lung cancer is the leading cause of cancer death globally, resulting in 1.4 million deaths annually, including 165,000 patients in the United States per year. In order to address the critical need for comprehensive profiling of these patients, we developed a novel, CLIA-certified, whole exome and low-coverage whole genome sequencing assay that applies a disease-focused, integrated approach to identify therapeutically actionable drivers of disease. A panel (12) of surgically resected NSCLC specimens along with corresponding adjacent normal tissue underwent DNA extraction in a clinical (CLIA) environment. Tumor and normal genomic DNA was prepared for whole exome sequencing using the using the Agilent SureSelectXT Human All Exon V5 kit according to the manufacturers instructions, and libraries were sequenced on the Illumina HiSeq2500 at an average depth of 500X. Genomic DNA was then prepared for whole genome sequencing using Illuminas Nextera system and run on the Illumina HiSeq2500 platform at an average depth of 1-2X. Somatic variants were detected using Strelka and somatic copy number alterations (SCNAs) were identified using a novel algorithm comparing normalized read counts within genomic segments as well as genes in the tumor to a panel of normal tissues. In parallel, the same tumor/normal specimens were analyzed by two separate CLIA laboratories via 1) a clinically validated Ion Torrent AmpliSeq Cancer Panel assay, and 2) a clinically validated cancer-focused, high-resolution comparative genome hybridization (CGH) array. In addition, a well-characterized panel of 10 germline samples obtained from the 1000 Genomes Project were pooled to simulate a broad spectrum of somatic single nucleotide variant and indel allele frequencies. Sequencing, data analysis and clinical reporting were completed for all 12 cases with an average turnaround time of less than 3 weeks. Single nucleotide variants and indels were identified with an accuracy of greater than 99%, with a limit of detection of 5-10% mutant allele frequency. Somatic copy number alterations were observed with an overall accuracy of greater than 95%. Actionable variants were identified by cross-referencing individual results with our internally developed, lung-cancer focused therapeutic association database. Citation Format: Catherine K. Foo, John St. John, Nicholas Hahner, Oscar Westesson, Mitchell E. Skinner, Urvish Parikh, Kimberly Lung, Aleah F. Cauhlin, Jeffrey P. Catalano, Anne S. Wellde, Jonathan K. Barry, George W. Wellde, Patrick Ma, Rafael Rosell, Andres Felipe Cardona Zorilla, William R. Polkinghorn, Trever G. Bivona, Jonathan S. Weissman, Petros Giannikopoulos. Comprehensive integrated genomic analysis. [abstract]. In: Proceedings of the 105th Annual Meeting of the American Association for Cancer Research; 2014 Apr 5-9; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2014;74(19 Suppl):Abstract nr 4707. doi:10.1158/1538-7445.AM2014-4707