Alex Parker’s research while affiliated with Foundation Medicine, Inc. and other places

6539 Background: In many instances, trials may offer the best or only therapeutic option for patients with rare findings. However, conducting clinical trials of novel therapeutics targeting rare molecular variants is challenging. Patient populations are small, distributed, and predominantly in community settings where trial access remains limited by awareness and site availability. These challenges increase costs of drug development and approval, delaying widespread patient access. Methods: Foundation Medicine deployed a trial education and access program, “Precision Enrollment,” with Ignyta (a trial sponsor) and Pharmatech (a site management organization, or SMO, enabling “Just-In-Time” clinical trials) (Wiener, JCO 2007). Infrastructure and algorithms developed at Foundation Medicine (“SmartTrials Engine”) matched sequenced patients (avg n = 800/wk) with activating NTRK, ROS1, or ALK fusions to the phase II study of Entrectinib (NCT02568267). Oncologists at Foundation Medicine, through peer-to-peer outreach, facilitated trial access by providing trial and nearest site information to treating providers of matched patients. Results: 107 treatment-eligible patients with NTRK, ROS1, or ALK fusions were matched by the SmartTrials Engine; 36 (33%) expressed interest in trial participation. One such patient with NSCLC and a CD74-ROS1 fusion was unable to participate at an open trial site due to inability to travel. The patient’s site was part of the “Just-In-Time” network, with IRB and contract pre-approval, and was activated in only 3 days. Total time from patient identification to initiation of therapy was 7 days. Conclusions: We demonstrate a novel methodology for patient matching to trials targeting rare genomic findings, including in community settings. If extended, such innovative partnerships combined with computational matching infrastructure, could improve drug development and therapeutic access.

2514 Background: Genomic findings have diagnostic, prognostic, and predictive utility in clinical oncology. Population studies have been limited by reliance on trials, registries, or institutional chart review, which are costly and represent narrow populations. Integrating electronic health record (EHR) and genomic data collected as part of routine clinical practice may overcome these hurdles. Methods: Patients in the Flatiron Health Database with non-small cell lung cancer (NSCLC) who underwent comprehensive genomic profiling (CGP) by Foundation Medicine were included. EHR processing included structured data harmonization and abstraction of variables from unstructured documents. EHR and CGP data were de-identified and linked in a HIPAA-compliant process. Data included clinical characteristics, alterations across > 300 genes, tumor mutation burden (TMB), therapies and associated real-world responses, progression, and overall survival (OS). Results: The cohort (n = 1619) had expected clinical (mean age 66; 75% with smoking hx; 80% non-squamous) and genomic (18% EGFR; 4% ALK; 1% ROS1) properties of NSCLC. Presence of a driver mutation (EGFR, ALK, ROS1, MET, BRAF, RET, or ERBB2; n = 576) was associated with younger age, female gender, non-smoking, improved OS (35 vs 19 mo, LR p < 0.0001), and prolonged survival when treated with NCCN-recommended therapy (42 vs 28 mo, LR p = 0.001). CGP identified false negative results in up to 30% of single-biomarker tests for EGFR, ALK, and ROS1. CGP accuracy was supported by clinical outcomes. For example, 5 patients with prior negative ALK-fusion testing began ALK-directed therapy after positive CGP results. All 5 exhibited at least a partial response as recorded in the EHR by treating clinicians. Immunotherapy was used in 22% of patients (n = 353). TMB predicted response to nivolumab, including in PD-L1 negative populations. We recapitulated known associations with smoking, histology, and driver mutations. Conclusions: We present and validate a new paradigm for rapidly generating large, research-grade, longitudinal clinico-genomic databases by linking genomic data with EHR clinical annotation. This method offers a powerful tool for understanding cancer genomics and advancing precision medicine.

Background Rapid advancements in cancer genomics and in the development of targeted therapies provide expanding opportunities to use genomic profiling to improve patient outcomes. However, most patients do not have access to clinical genomic profiling platforms, and currently available assays capture a small set of known mutations or translocations tailored to specific tumor types. The spectrum of somatic alterations in leukemia, lymphoma, and myeloma includes substitutions, insertions/deletions (indels), copy number alterations (CNAs) and gene fusions; no current assay captures the different types of alterations in a single clinical genomic test. We developed a novel, CLIA-certified next-generation sequencing-based assay designed to provide targeted assessment of the genomic landscape of hematologic malignancies, including identification of all classes of genomic alterations using archived FFPE, blood and bone marrow aspirate samples with high accuracy in a clinically relevant timeframe. Methods DNA and RNA were successfully extracted from 350/362 (96%) specimens from 319 patients, including 57 FFPE samples, 150 blood samples and 142 bone marrow aspirates. The initial sample cohort included 20 ALL, 83 AML, 53 CLL, 57 DLBCL, 48 MDS, 32 MPN and 57 multiple myeloma samples. Adaptor ligated sequencing libraries were captured by solution hybridization using a custom bait-set targeting 374 cancer-related genes and 24 frequently rearranged genes by DNA-seq, and 258 frequently-rearranged genes by RNA-seq. All captured libraries were sequenced to high depth (Illumina HiSeq) in a CLIA-certified laboratory (Foundation Medicine), averaging 590x for DNA and >20M total pairs for RNA, to enable the sensitive and specific detection of substitutions, indels, CNAs and gene fusions. Results Sufficient tumor content (≥20%) was present in 317/350 (91%) of the samples (289/319 patients), and a total of 885 alterations were identified (3.1 alterations per sample), including 555 base substitutions, 213 indels, 36 splice mutations, 51 CNAs and 36 fusions/rearrangements. The most frequent alterations across all hematologic malignancies included mutations in TP53 (9%), ASXL1, KRAS, NRAS, IDH2, TET2, SF3B1, JAK2, MLL2, DNMT3A, RUNX1, and SRSF2 (2-5% each); FLT3 ITDs (2%); MLL PTDs (1%); homozygous loss of CDKN2A/B (3%); and focal amplification of REL (1%). Rearrangements in BCL2/6, MYC, MLL, MLL2, NOTCH2, ABL1 and ETV6 were identified using DNA and RNA targeted sequencing, demonstrating the ability of this platform to reliably identify gene fusions with immediate clinical relevance. Overall high accuracy of the assay for substitutions, indels and CNAs was previously demonstrated by extensive validation studies achieving 95-99% across alteration types with high specificity (PPV>99%) [Frampton et al, Nat Biotech, in press]. Comparison of detected alterations to previous molecular testing for JAK2, NPM1, IDH2, FLT3 and CEBPA in MPN/AML samples demonstrated 97% sensitivity (33/34) in our ability to identify known mutations in these clinical samples. We identified additional clinically relevant mutations that were not detected using standard clinical assays, including alterations in JAK2, FLT3 and IDH2, which can inform therapeutic decisions. The use of our content rich sequencing platform allowed us to identify clinically actionable mutations in hematologic malignancies, including IDH1/2 mutations in a spectrum of myeloid/lymphoid malignancies, recurrent BRAF mutations in refractory CLL and myeloma, and mutations in the JAK-STAT signaling pathway in diffuse-large B cell lymphoma. These results demonstrate that a targeted sequencing platform which includes a large set of known disease alleles/therapeutic targets can identify mutations with therapeutic relevance in disease contexts where gene-specific assays are not currently performed in the clinical setting. Conclusions We have developed a sensitive, high throughput assay to detect somatic alterations in hundreds of genes known to be deregulated in hematologic malignancies, which can be used for clinical sequencing of frozen/paraffin samples. We demonstrate that targeted DNA and RNA sequencing can be used to identify all classes of genomic alterations in genes known to be therapeutic targets in a broad spectrum of hematologic malignancies. Disclosures Lipson: Foundation Medicine, Inc: Employment, Equity Ownership. Nahas:Foundation Medicine, Inc: Employment, Equity Ownership. Otto:Foundation Medicine, Inc: Employment, Equity Ownership. Yelensky:Foundation Medicine, Inc: Employment, Equity Ownership. Wang:Foundation Medicine, Inc: Employment, Equity Ownership. He:Foundation Medicine, Inc: Employment, Equity Ownership. Rampal:Foundation Medicine: Consultancy. Brennan:Foundation Medicine, Inc: Employment, Equity Ownership. Brennan:Foundation Medicine, Inc: Employment, Equity Ownership. Young:Foundation Medicine, Inc: Employment, Equity Ownership. Donahue:Foundation Medicine, Inc: Employment, Equity Ownership. Sanford:Foundation Medicine, Inc: Employment, Equity Ownership. Greenbowe:Foundation Medicine, Inc: Employment, Equity Ownership. Frampton:Foundation Medicine, Inc: Employment, Equity Ownership. Fichtenholtz:Foundation Medicine, Inc: Employment, Equity Ownership. Young:Foundation Medicine, Inc: Employment, Equity Ownership. Erlich:Foundation Medicine, Inc: Employment, Equity Ownership. Parker:Foundation Medicine, Inc: Employment, Equity Ownership. Ross:Foundation Medicine, Inc: Employment, Equity Ownership. Stephens:Foundation Medicine, Inc: Employment, Equity Ownership. Miller:Foundation Medicine, Inc: Employment, Equity Ownership. Levine:Foundation Medicine, Inc: Consultancy.

As more clinically relevant cancer genes are identified, comprehensive diagnostic approaches are needed to match patients to therapies, raising the challenge of optimization and analytical validation of assays that interrogate millions of bases of cancer genomes altered by multiple mechanisms. Here we describe a test based on massively parallel DNA sequencing to characterize base substitutions, short insertions and deletions (indels), copy number alterations and selected fusions across 287 cancer-related genes from routine formalin-fixed and paraffin-embedded (FFPE) clinical specimens. We implemented a practical validation strategy with reference samples of pooled cell lines that model key determinants of accuracy, including mutant allele frequency, indel length and amplitude of copy change. Test sensitivity achieved was 95-99% across alteration types, with high specificity (positive predictive value >99%). We confirmed accuracy using 249 FFPE cancer specimens characterized by established assays. Application of the test to 2,221 clinical cases revealed clinically actionable alterations in 76% of tumors, three times the number of actionable alterations detected by current diagnostic tests.

Background: As the number of clinically relevant cancer genes grows and the amount of tissue available for analysis decreases, next-generation sequencing (NGS) becomes increasingly attractive as a diagnostic tool, as it can detect all classes of genomic alterations in all cancer genes in a single test. However, for NGS to demonstrate its full utility in the clinic, robust analytical validation approaches and performance benchmarks against gold standard methodologies are required. Methods: We have developed a comprehensive NGS-based diagnostic test to enable accurate detection of clinically-relevant genomic alterations across 4604 exons of 289 cancer genes (∼1.5 Mb) in routine FFPE specimens, including needle biopsies. To validate the test, we created synthetic reference standards spanning key determinants of detection accuracy for somatic alterations, such as a range of stromal admixture, indel length, and amplitude of copy change: For base substitutions, we mixed 2 pools of 10 normal cell-lines, thereby testing 1035 different variants in a broad 10%-100% allele frequency range. For indels, 28 tumor cell lines with 44 alterations (1-40bp) were mixed into 41 variably sized pools, creating a validation set of 163 events. For copy alterations, 7 matched tumor/normal cell-lines bearing a total of 12 amplifications (copy number 8-15) and 7 homozygous deletions were mixed in 5 tumor content ratios 20%-75%. The test was applied blinded to all pools and alteration calls compared to expectation based on constituents in each mix. We then verified that observed performance translated to FFPE by examining test concordance on 185 tumors characterized clinically for mutations in 8 genes including KRAS, BRAF, and EGFR (Sanger, Sequenom, and PCR) and 131 specimens characterized for CNAs in ERBB2, PTEN, AR and CCND1 (IHC and FISH). Results: On synthetic standards, performance reached >99% sensitivity for substitutions (1,035/1,035) and >98% for indels (161/163) at ≥10% allele-frequency with PPV>99%. Sensitivity for CNAs was >95% for focal amplifications (46/48) and >99% for homozygous deletions (28/28) at ≥30% tumor fraction, with PPV>99%. Robust performance translated to FFPE: concordance was 98% for both substitutions and indels (164/168) and CNAs (41/42) relative to prior calls. Conclusions: We present a rigorous validation approach and performance benchmarks for a comprehensive NGS-based test for use in clinical oncology. Given the ability of NGS to detect a much broader range of genomic alterations than currently available technologies, particularly on limited tissue, we suggest that this type of testing can be a direct component of patient care and potentially expand targeted treatment options. Citation Format: Roman Yelensky, Garrett M. Frampton, Alex Fichtenholtz, Sean Downing, Jie He, Frank Juhn, Kristina Brennan, Kai Wang, Geoff Otto, Mirna Jarosz, Alex Parker, Jeffrey S. Ross, John Curran, Maureen T. Cronin, Philip J. Stephens, Doron Lipson. Bringing next generation sequencing (NGS) to the clinic: Analytical validation of a comprehensive NGS-based cancer gene test. [abstract]. In: Proceedings of the 104th Annual Meeting of the American Association for Cancer Research; 2013 Apr 6-10; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2013;73(8 Suppl):Abstract nr 2279. doi:10.1158/1538-7445.AM2013-2279

Proceedings: AACR 103rd Annual Meeting 2012‐‐ Mar 31‐Apr 4, 2012; Chicago, IL Sequence analysis and quantitative allele specific PCR (QPCR) methods permit genetic profiling of cancer for targeted therapeutic selection; such personalized treatments have been associated with improved outcomes in cancer. Circulating tumor cells (CTC) offer a minimally-invasive opportunity for serial patient sampling, and potentially a means of tracking the molecular evolution that underlies the behavior and response phenotype of the disease including potential therapeutic response markers. Acquiring these types of patient profiles requires a platform and workflow providing reliable detection and recovery of small numbers of mutation-bearing CTC from a blood sample. Availability of such an enabling platform is a necessary prerequisite to the clinical correlation studies needed to demonstrate the utility of mutation-bearing CTC to patient care. We have successfully purified CTCs and converted them into DNA template of sufficient purity and quality to support multiple non-overlapping advanced molecular characterizations. Beginning with whole human blood spiked with defined numbers of cultured cancer cells as surrogates for CTCs, cells were successfully fluid-phase labeled using an anti-EpCAM antibody ferrofluid. Using a proprietary microfluidic sheath flow technology, EpCAM positive, cytokeratin staining cells were selected from 2 to 4 ml of labeled blood. This method produced sufficient DNA template for multiple analyses per patient sample. QPCR analysis of these templates demonstrated reproducible detection of fewer than 1% target cells in a background of non-target cells, allowing detection of the KRAS G12S mutation from as few as 5 recaptured cancer cells. The same DNA templates were then used for hybrid capture and next-generation sequencing of a panel of more than 200 cancer-related genes. This sequencing platform was able to detect multiple somatic mutations in genomic DNA templates produced from samples containing as few as 10 cancer cells per milliliter of blood. Together these data provide initial proof-of-concept for a system capable of detecting and characterizing mutations across any specified set of genes within purified CTC populations. Citation Format: {Authors}. {Abstract title} [abstract]. In: Proceedings of the 103rd Annual Meeting of the American Association for Cancer Research; 2012 Mar 31-Apr 4; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2012;72(8 Suppl):Abstract nr 4554. doi:1538-7445.AM2012-4554

Proceedings: AACR 103rd Annual Meeting 2012‐‐ Mar 31‐Apr 4, 2012; Chicago, IL Molecular diagnostics are growing in importance to clinical oncology as an increasing number of therapies targeting specific molecular alterations become available. This trend has led to a proliferation of focused biomarker tests delivered using a variety of technologies, which are generally restricted in their breadth of mutational status assessment and constrained by scarce tissue material. We deployed massively parallel sequencing (or “NGS”) technology to develop a pan-cancer, comprehensive genomic profiling assay interrogating 189 relevant cancer genes, using minimal DNA from fixed surgical or biopsy specimens. With this assay we identify all classes of DNA alterations (base substitutions, insertions and deletions, copy number alterations and rearrangements at specific loci) in a single test with high sensitivity and specificity. To assess the potential for comprehensive genomic profiling to inform clinical decisions, we sequenced 58 archival FFPE tumor specimens, including 16 breast, 16 colon, 14 lung, and 12 renal cell cancers to an average of depth of 850x and considered the resultant mutation profiles in the context of current treatment guidelines and clinical trial information. In total, 121 mutations and 30 copy number alterations were identified, of which less than one third could be identified by traditional “hotspot” analysis. The observed mutation frequencies were consistent with published studies for these tissue types: 56% of breast cancers carried TP53 mutations, 38% had PIK3CA mutations, and 13% harbored HER2 amplifications. In colon cancers, 75% and 63% carried APC and TP53 mutations, while 38%, 31%, and 25% harbored KRAS, BRAF, and PIK3CA mutations respectively. Finally, 57% and 42% of NSCLCs carried mutations in TP53 and KRAS, while 17% of RCCs had mutations in VHL. Nearly 70% of the 58 cases carried one or more mutations that could plausibly confer sensitivity or resistance to approved or experimental targeted therapies. Of these potentially clinically actionable mutations, 57% occurred at known sites of oncogene activation, 27% involved gene amplification or deletion, 14% resulted in tumor suppressor loss, and one resulted in the EML4-ALK fusion gene. Our results demonstrate technical feasibility of comprehensive cancer gene characterization through massively parallel sequencing in clinical FFPE specimens. The prevalence and diversity of potentially actionable mutations we observed highlights the benefits of adopting this approach. As massively-parallel sequencing is the only practical means to detect all classes of somatic mutation in a small, clinically relevant sample, we suggest that this type of testing will be an essential ingredient in bringing targeted therapies to cancer patient care. Citation Format: {Authors}. {Abstract title} [abstract]. In: Proceedings of the 103rd Annual Meeting of the American Association for Cancer Research; 2012 Mar 31-Apr 4; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2012;72(8 Suppl):Abstract nr 967. doi:1538-7445.AM2012-967

3533 Background: We hypothesized that NGS could identify genomic alterations that guide selection of targeted therapies and discover novel drug targets for primary and metastatic CRC. Methods: DNA was extracted and sequenced from 4 X 10 μ FFPE sections from 40 (32 primary and 8 metastatic) CRCs obtained from 2004-10 (52% male; 48% female; mean age 60 years; 10% Stages I/II; 40% Stage III; 40% Stage IV; 10% Stage unknown) using a targeted NGS assay in a CLIA laboratory (Foundation Medicine). 2574 exons of 145 cancer-related genes plus 37 introns from 14 genes often rearranged in cancer were fully sequenced for point mutations, insertions/deletions, copy number alterations (CNAs) and select gene fusions. Results: NGS revealed 125 genomic alterations in 39/40 tumors (mean 3.1, range1-7) in 21 genes, including 80 base substitutions (64%), 39 insertions/deletions (31%), 4 CNAs (3%) and 2 gene fusions (1.6%). TP53 and APC were altered in 80% (32/40) and 67.5% (27/40) of CRCs respectively, with both mutated more frequently than reported in COSMIC. Alterations associated with potential sensitivity to targeted therapies were identified in 21 (52.5%) of CRCs including: 10 KRAS (TK/MEK inhibitors), 6 BRAF (BRAF inhibitors), 5 FBXW7 (mTOR inhibitors); 2 PIK3CA (PI3K/mTOR inhibitors); 2 BRCA2 (PARP inhibitors); 2 GNAS (MEK/ERK inhibitors) and 1 CDK8 (CDK inhibitors). In 1 CRC, a 5.2 Mb tandem duplication generated a novel C2orf44-ALK gene fusion starting at the canonical exon 20 recombination site previously reported for the majority of ALK gene fusions. cDNA sequencing identified an 90-fold increase in 3’ ALK expression, suggesting the C2orf44-ALK fusion results in ALK kinase domain overexpression and contains the same intracellular domain as other ALK fusions including the ALK inhibitor sensitive EML4-ALK. Conclusions: NGS of broad, cancer-related gene content from FFPE CRC samples uncovered an unexpectedly high frequency of genomic alterations, many of which may be clinically actionable by informing treatment decisions, including a novel ALK gene fusion. This previously unrecognized subset of CRC patients may be candidates for clinical trials of crizotinib or other ALK inhibitors.

Applying a next-generation sequencing assay targeting 145 cancer-relevant genes in 40 colorectal cancer and 24 non-small cell lung cancer formalin-fixed paraffin-embedded tissue specimens identified at least one clinically relevant genomic alteration in 59% of the samples and revealed two gene fusions, C2orf44-ALK in a colorectal cancer sample and KIF5B-RET in a lung adenocarcinoma. Further screening of 561 lung adenocarcinomas identified 11 additional tumors with KIF5B-RET gene fusions (2.0%; 95% CI 0.8-3.1%). Cells expressing oncogenic KIF5B-RET are sensitive to multi-kinase inhibitors that inhibit RET.