EGFR and HER2 (ERBB2) exon 20 mutations occur in approximately 3.6% of NSCLC, and patients with tumors harboring these mutations have historically experienced poor response rates to clinically available TKIs. Given the poor clinical responses in these patient populations, a deeper understanding of the effect of exon 20 mutations on the drug-binding pocket, sensitivity to available TKIs, and the genomic landscape of exon 20 mutations is greatly needed. We hypothesized that while exon 20 mutations are prevalent in NSCLC, these mutations also occur in other cancer types and alter the drug-binding pocket, resulting in de novo drug resistance across cancers. To test these hypotheses, we performed an analysis of eleven databases (N=212,000) to determine the prevalence of exon 20 mutations across cancer types and utilized in silico, in vitro, and in vivo models to investigate structural alterations induced by exon 20 mutations and identify effective inhibitors. Through this analysis we found that EGFR and HER2 exon 20 mutations occur in 28 different types of cancers, and that exon 20 mutations comprise 0.6% of all cancers, amounting to approximately 16,000 patients per year in the United States. Molecular modeling and molecular dynamics simulations showed that exon 20 insertions in both EGFR and HER2 reduced the overall volume of the drug-binding pocket, which correlated with decreased sensitivity to TKIs. Through in vitro screening using more than 14 EGFR TKIs, we found that poziotinib was the most potent inhibitor tested in EGFR (N=20) and HER2 (N=6) exon 20 insertion models with IC50 values of 1.5nM and 2.5nM, respectively. In our extensive panel of Ba/F3 cells engineered to express various EGFR/HER2 mutations, poziotinib was found to be the most selective TKI for the majority of EGFR and HER2 exon 20 mutants compared to WT EGFR (Mutant/WT IC50 ratio = 0.5). In vivo, poziotinib caused 70% and 85% reduction in tumor burden in PDX models of EGFR exon 20 mutant NSCLC models harboring EGFR S768dupSVD and EGFR H773insNPH mutations after 10 days of treatment. Using genetically engineered mouse models (GEMMs) of EGFR exon 20 mutant NSCLC, poziotinib reduced tumor volume in EGFR (D770insNPG) and HER2 (Y772dupYVMA) mutant tumors by 80% and 60%, respectively, after 4 weeks of treatment. In addition, we observed that low-dose poziotinib caused an upregulation in cell surface expression of HER2 exon 20 mutants and sensitized HER2 exon 20 mutant-expressing cells to T-DM1 treatment. To exploit this, we tested the combination of low-dose poziotinib (2.5mg/kg) and a single dose of T-DM1 (10mg/kg) in an HER2 mutant NSCLC PDX model (HER2 Y772dupYVMA). We observed complete tumor regression in 20/20 mice, compared to 2/9 mice receiving T-DM1 alone or 0/12 mice receiving low-dose poziotinib by day 15 (p<0.0001). Median progression-free survival (mPFS, tumor doubling from best response) was 3 days, 15 days, and 27 days in vehicle control, low-dose poziotinib, and T-DM1 treated groups, whereas the mPFS had not been reached by day 45 in the combination-treated group. To validate these findings in an additional model of HER2 exon 20 mutant NSCLC, we tested low-dose poziotinib, T-DM1, and the combination in a GEMM of NSCLC harboring Y772dupYVMA. Recapitulating results seen in the PDX model, mice receiving either poziotinib or T-DM1 had on average of an 11% increase in tumor growth, whereas mice receiving the combination of low-dose poziotinib and T-DM1 had an average 47% reduction in tumor burden after four weeks. Lastly, to validate the activity of poziotinib, a phase II investigator-initiated trial (NCT03066206) testing poziotinib in patients with EGFR or HER2 exon 20 mutated NSCLC was opened. In the EGFR cohort, there was an objective response rate (ORR) of 43% and mPFS of 5.5 months in 44 evaluable patients. While the HER2 cohort is still ongoing, in the first twelve evaluable patients, there was an ORR of 42% and a mPFS of 5.6 months. Taken together, these data demonstrate that poziotinib is an effective and clinically active inhibitor for both EGFR and HER2 exon 20 mutant NSCLC and that poziotinib in combination with drug-antibody conjugates may have increased efficacy. Further, these studies demonstrate that clinical studies testing poziotinib alone and in combination with antibody-drug conjugates in other EGFR and HER2 exon 20 mutant cancers are warranted.

Liquid biopsy has been established as an important diagnostic step in identifying EGFR T790M resistance to 1st and 2nd generation EGFR TKIs. This has led to significant interest in exploring the value of liquid biopsy in patients with other oncogene-addicted cancers and acquired resistance to targeted therapy. This study is being conducted at 6 Canadian centres (NCT03576937) using a validated cell-free DNA next-generation sequencing assay that identifies variants in 74 cancer-associated genes, including fusions and copy number gain (Guardant 360TM). In a discovery cohort (N=60), patients that failed targeted therapy for oncogene-addicted lung cancer receive a liquid biopsy (LB), paired with standard tumour tissue (TT) biopsy with molecular profiling if possible. Patients are required to have measurable disease (RECIST version 1.1), and, if failing a 1st or 2nd generation EGFR TKI, must demonstrate the absence of EGFR T790M in tissue (and plasma if standard of care). Exploratory analysis of incremental targetable alterations identified by LB is presented. Genomic alterations are categorized as actionable, (targeted therapy available for indication), potentially actionable (clinical trials available and/or known resistance mechanism) or non-actionable. 56 of 60 patients have been enrolled. Median age of the cohort is 58 years (range 23-86), 36 (64%) are female, 48 (86%) are never smokers and all have adenocarcinoma. Seven (13%) patients have actionable targets and 9 (16%) potentially actionable targets identified through LB. Response data will be presented at the meeting. Liquid biopsy provides a minimally invasive initial approach to the molecular characterization of resistance in patients with oncogene addicted lung cancer failing targeted therapy, yielding actionable or potentially actionable results in 20% of patients beyond EGFR T790M.Table: 1195POncogenic driverNActionable Target in LBPotentially ActionableNo. pts with Non-ActionableALK fusion*92 ALK G1202R1 ALK C1156Y**1 EGFR amp6EGFR T790M-221 CCDC6-RET fusion1 MET amp**2 T790M1 KRAS G12D1 CDK4 amp1 FGFR1 amp**17EGFR T790M+*201 MET amp1 BRAF V600E1 FGFR3-TACC fusion1 BRAF L597R3 EGFR C797S**14ERBB2ins2--2MET ex14skip11 KRAS G12DEGFR ex20ins1--1NRG fusion11 CCND2 amp1 EGFR amp*** ALK fusion+ pts received prior alectinib (median 3 prior TKI); EGFR T790M+ pts received prior osimertinib (median 2 prior TKI).**concurrent aberrations (G1202R+C1156Y; MET amp+ T790M; FGFR1 amp + CDK4 amp; C797S + BRAF; EGFR amp+CCND2 amp) Open table in a new tab

Abstract Identification of ERBB2 (HER2) overexpression in metastatic breast cancer informs utilization of HER2 targeted therapy. The NCCN recommends HER2 expression re-evaluation at the first disease recurrence in patients with negative or equivocal tissue status given results discrepancies due to inadequate tissue biopsy, tumoral heterogeneity, biopsy technique or fixation as well as discordance in ERBB2 (HER2) expression between primary and metastatic lesions. We examined the incidence of ERBB2 (HER2) negative to positive “flips” (e.g. to ERBB2-amplified in plasma) in a cohort of patients who underwent a blood-based cell-free DNA (cfDNA) assay at a CLIA-certified/CAP-accredited/NYSDOH-approved molecular diagnostic laboratory. Laboratory database was queried for samples from patients with a breast cancer diagnosis. The query was filtered to ensure patients with multiple cfDNA timepoints were counted only once. Patients without a pathology report submitted at any cfDNA collection timepoint or the pathology report did not include ERBB2 (HER2) status, results were inconclusive or quantity not sufficient were excluded. Between March 2014 and April 2017, 1,853 unique patients were identified with reported ERBB2 (HER2) status. For patients with more than one cfDNA timepoint collected (N=349; 18.8%), the earliest pathology report was referenced. 1,386 patient tumor samples were negative for HER2 overexpression (74.8%), 325 (17.5%) were positive, and 142 (7.7%) were equivocal. Twenty-nine of the 1,386 patients with reported tumor negative HER2 status had amplification on subsequent cfDNA analysis (2.1%). All 29 patients were female. Most patients (N=21) had a single cfDNA timepoint collected. Median age at cfDNA blood draw was 58 years (range 28–68). Median length of time between reported tissue negative status and cfDNA blood draw was 405 days (range 21–4,060). Median plasma ERBB2 copy number was 2.44 (greater than 50th-centile per laboratory data) (range 2.15–16.5). Clinical follow-up was obtained for 19 patients (65%). Nine patients were lost to follow-up or succumbed to disease prior to initiation of a new therapeutic regimen. One patient was known HER2 positive prior to receipt of the cfDNA results. In the remaining nine patients, six initiated targeted HER2 therapy following receipt of the cfDNA results, with five of six (83%) demonstrating a clinical response. In one patient with known ER/PR positive, HER2 negative disease, progressing through multiple lines of therapy, addition of trastuzumab and pertuzumab to her paclitaxel regimen following identification of the cfDNA ERBB2 amplification resulted in a significant reduction in CEA levels (238 to 37.9 ng/mL) by week five. In a second patient, following identification of the cfDNA ERBB2 amplification, she was treated with trastuzumab and pertuzumab along with docetaxel and had a dramatic response. She continues on trastuzumab and pertuzumab alone. Although a modest sample size, this is the second cfDNA series demonstrating that ERBB2 (HER2) status may flip from negative to positive upon recurrence or metastasis, and that targeting plasma-detected ERBB2 amplification with anti-HER2 has clinical benefit. cfDNA is a viable alternative to tissue rebiopsy in this patient population. Citation Format: Raymond VM, Diaz J, Banks KC, Ahn E, Brufsky A, Ellis M, Lippman M, Lee C, Pluard T, Schreeder M, Schwab R, Lanman RB. Cell free DNA analysis identifies actionable ERBB2 amplifications in patients with HER2 negative breast cancer [abstract]. In: Proceedings of the 2017 San Antonio Breast Cancer Symposium; 2017 Dec 5-9; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2018;78(4 Suppl):Abstract nr P2-02-12.

Background: The advantages of plasma-based tumor mutational burden (TMB) include non-invasive, real-time assessment of mutational load, without the limitations of insufficient tissue. However, at low levels of tumor DNA shedding, TMB may be underestimated if a fraction of the genomic alterations is below assay limit of detection. Currently available blood TMB panels report a 1% tumor content limit of detection, which would result in about half of all clinical plasma samples (based on > 30,000 patients) with unevaluable TMB. Hence, clinically effective blood-based diagnostics must be highly sensitive and account for tumor shedding. Here, we present a comprehensive cfDNA-based TMB using a 500-gene (GuardantOMNI) and a 73-gene (Guardant360) panel. Methods: We developed a statistical model to calculate TMB on plasma samples with low cell-free circulating tumor DNA (ctDNA) content. Theoretical panel performance was assessed in silico by subsetting mutations from whole exome sequencing (WES) to the Guardant panel space (2Mb for GuardantOMNI and 200Kb for Guardant360) from 9,104 TCGA samples and 30 lung cancer samples with published immunotherapy outcomes. Sensitivity was evaluated using 50 serially diluted cfDNA specimens. Analytical validation was performed against tissue-based WES TMB using matched plasma and tissue samples across multiple tumor types. Results: High correlation was observed between TMB called on the Guardant panel and WES mutations from the TCGA dataset (r = 0.99 with GuardantOMNI; r = 0.92 with Guardant360). Subsetting WES from clinical outcome cohorts to each panel recapitulated the association with PFS on immunotherapy (HR = 0.41 with GuardantOMNI; HR = 0.27 with Guardant360). The sensitivity of detection was assessed down to 0.3% tumor content and 5ng cfDNA input. Lastly, we show high quantitative concordance between matched plasma and tissue WES samples for both GuardantOMNI and Guardant360. Conclusions: We describe a plasma-based TMB score that correlates with tissue-derived TMB at tumor fractions down to 0.3%, enabling TMB calculation on > 70% of all clinical samples. Accurate reporting of TMB from a plasma sample has the potential to accelerate clinical trial enrollment and improve outcomes. Legal entity responsible for the study: Guardant Health, Inc. Funding: Guardant Health, Inc. Disclosure: K. Quinn, E. Helman, T. Nance, C. Artieri, J. Yen, J. Zhao, S. Fairclough, M. Sikora, D. Chudova, R.B. Lanman, A. Talasaz: Employee and ownership (stock): Guardant Health, Inc.

Background: RET rearrangements (RETr) are uncommon yet emerging oncogenic targets found in RAS/BRAF wild-type colorectal cancers (CRCs), particularly right-sided, MSI-high tumors. We describe the molecular landscape of metastatic CRCs harboring a somatic RETr detected by next generation sequencing (NGS) of cell-free DNA (cfDNA). cfDNA data complement previous descriptions of RETr in tissue, which are limited by small series of patients (pts) and cfDNA also provides a summary of genomic alterations (alts) across multiple metastatic sites in pts typically exposed to one or more lines of therapy. Methods: Between 2/2015-3/2018, somatic RETr were identified among 4,234 consecutive metastatic CRC pts with tumor DNA detected on a 68-73 gene cfDNA assay (Guardant360®, Redwood City, CA). This validated NGS assay evaluates single nucleotide variants, and select indels, fusions, and copy number gains with high sensitivity and analytic specificity. Relevant clinicopathologic correlates were obtained from clinicians. Results: Seventeen RETr were detected in 16 pts (0.4%). Functionally significant alts in other cancer genes were found in 88% of samples with a RETr (median 9.5 additional alts, max 23). There was a high co-occurrence of canonical KRAS alts (14 alts in 7/16 pts), 5 of whom also had alts in NRAS (n = 4pts), the EGFR extracellular domain (ECD, n = 4), and/or BRAF V600E (n = 3). For 4 pts with available comprehensive tissue NGS results, the RETr was detected in cfDNA only. Two also had KRAS, NRAS, and EGFR ECD alts, one had an EGFR activating alt, and the fourth had a FGFR3 fusion, all detected only in cfDNA. Prior to cfDNA collection, all 4 had progressed on anti-EGFR therapy after a median of 12mo (range 8-16mo) of treatment. Four additional RETr/KRAS+ pts had either a KRAS alt that was not detected in tissue (n = 2) or co-occurred with an ALK fusion and/or BRAF V600E in pts who had prior anti-EGFR therapy (n = 2). Conclusions: RETr commonly co-occur with RAS/RAF alts in cfDNA, a novel observation. The alt pattern and clinicopatholgoic history suggest RETr contribute to acquired resistance to anti-EGFR therapy in metastatic CRC. Our data also raise the question of whether driver RETr may be associated with primary resistance to anti-EGFR therapy. Legal entity responsible for the study: Guardant Health. Funding: Has not received any funding. Disclosure: T.A. Rich, V.M. Raymond, Y. Shiotsu, R.B. Lanman: Employee, Stock owner: Guardant Health. Y. Kagawa, Y. Nakamura: Research support: Guardant Health. W. Okamoto: Research support: Guardant Health; Grant funding: MSD. T. Yoshino: Research support: Guardant Health; Grants: MSD KK, Sanofi KK, Sumitomo Dainippon Pharma Co Ltd, Chugai Pharmaceutical Co Ltd, GlaxoSmithKline KK, Nippon Boehringer Ingelheim Co Ltd; Consulting fees: Sanofi KK, Chugai Pharmaceuticals Co Ltd, Eli Lilly Japan, Merck Serono Co Ltd. All other authors have declared no conflicts of interest.

Background: Circulating cell-free DNA (cfDNA) next-generation sequencing (NGS) may supplement when tissue biopsy is insufficient or infeasible. Activating MET alterations (alts) can be oncogenic drivers. MET amp and missense alts mutations may also be acquired as a mechanism of resistance to tyrosine kinase inhibitorsTKI resistance alts. However, their prevalences in cfDNA among East Asian patients (pts) are not well established. Methods: MET amps, exon 14 skipping, and nonsynonymous missense mutations alts predicted to be oncogenic were identified from 338 consecutive samples from 328 East Asian pts with advanced solid tumors in first and higher subsequent lines of therapy. We utilized a comprehensive cfDNA NGS test (68-73-genes, Guardant360) between May 2015-December 2017. This 68-73 gene assay evaluates single nucleotides variants, and selected indels, fusions, and copy number gains. Results: MET alts were found in 32 pts with 11 different cancer types: NSCLC (18/180), colorectal (6/34), prostate (2/5), and 1 each of 6 other cancer types. 28 pts had amps, 3 had exon 14 skipping, and one had both D1246N D1228N and amp. Among the 19 NSCLC samples with MET alts, 13 co-occurred with a classic NSCLC driver: EML4-ALK fusion-1; EGFR L858R-3; EGFR exon 19 deletion-9 (of which 3 also had EGFR T790M). 3/7 MET alts detected in samples from colorectal cancer patients were found with co-existing KRAS or BRAF V600E missense alts. Among the 30 samples with MET amp, 23 had co-amp of at least one other gene also located on chromosome 7 (EGFR, BRAF, and/or CDK6), 11 of which had amp of all 4 of these genes. Conclusions: Somatic MET alts are found in cfDNA of nearly 10% of generally pre-treated East Asian patients with advanced solid tumorss. Acquired MET alts frequently occur with other oncogenic drivers requiring rational combination therapy approaches. Also, co-amplification of other chromosome 7 genes (suggesting aneuploidy rather than focal MET amp) will likely attenuate targeted therapy responses. CfDNA NGS enables non-invasive identification of all three classes of MET alterations, which may have higher prevalence and clinical relevance than previously appreciated in multiple advanced solid tumor cancers in East Asians. Legal entity responsible for the study: Tokyo Medical and Dental University. Funding: Has not received any funding. Disclosure: S. Ikeda: Corporate-sponsored research: ACT genomics, MSD, Chugai; Advisor board: Genodiv pharma T.A. Rich, Y. Shiotsu, A. Franovic, V.M. Raymond, R.B. Lanman: Employment: Guardant Health, Inc. R. Kurzrock: Research funding: Incyte, Genentech, Merck Serono, Pfizer, Sequenom, Foundation Medicine, Guardant Health Consultation: Sequenom, Loxo, Actuate Therapeutics; Ownership interest: CureMatch, Inc. T.S.K. Mok: Honoraria and consulting: AZ, BI, Roche/Genentech, Pfizer, Lilly, MerckSerono, MSD, Novartis, SFJ,ACEA, Vertex, BMS, Oncogenex, Celgene, Ignyta, Cirina, Fishawack Facilitate, Takeda, Janssen, ChiMed. All other authors have declared no conflicts of interest.

Abstract Inactivating germline mutations in the NF1 gene (encoding neurofibromin) cause neurofibromatosis type 1. In addition to peripheral nervous system tumors, NF1 patients are at higher risk for other cancers, including breast cancer. Tumor exome-sequencing studies demonstrate that approximately 20% of all human cancers have somatic NF1 mutations. NF1 has been best known for its ability to inactivate Ras as a GAP (GTPase Activating Protein). However, this function is served by a small GAP domain in a very large protein. Recurrent missense mutations inactivating the GAP activity are infrequent. In contrast, it is common to detect frameshift (FS) and nonsense (NS) NF1 mutations, which can create an NF1-null state deleting not only GAP, but also, potentially, undefined NF1 functions whose loss could also drive tumorigenesis. As we reported at SABCS previously, in 600+ patients treated by tamoxifen adjuvant monotherapy, we found that FS/NS NF1 mutations independently correlate with relapse risk (HR=2.6, p=0.03). To explore this finding, we silenced NF1 in preclinical models of ER+ breast cancer, which markedly enhanced ER transcriptional activities, causing estradiol (E2) hypersensitivity and converted tamoxifen into an agonist (in vitro and in vivo). Most important, these activities depend on ER, but not on NF1's GAP activity. These findings readily explain the poor patient outcomes associated with NS/FS NF1 mutations, and reveal a previously unrecognized function for NF1 in ER regulation. In the presence of an agonist, liganded ER repels co-repressors and recruits co-activators, while the reverse is true with an antagonist such as tamoxifen. Many co-regulators contain leucine/isoleucine rich motifs, which bind directly to the ligand-binding domain (LBD) in ER. NF1 has several of these motifs that are much more highly conserved in species with a functional ER pathway, and some of these are mutated in cancers (e.g., in our patient cohort). Furthermore, we found that NF1 canbind directly to ER, and that this binding is mediated between the ER LBD and the NF1 leucine-rich regions. Like a classic co-repressor, wildtype NF1 (but not mutants lacking GAP activity or the Leu-rich motif) binds to ER, and is recruited by ER to the ERE in the presence of tamoxifen, but not E2. Further preclinical treatment studies indicate that while NF1-deficient ER+ breast cancer should not be treated by tamoxifen or AIs, fulvestrant remains effective. Furthermore, when fulvestrant is combined with dabrafinib and trametinib to inhibit Ras effectors Raf and MEK, apoptosis is induced in vitro, and tumor regression is observed in vivo. In conclusion, we have demonstrated that NF1 is a dual negative regulator at the intersection of two potent oncogenic signaling pathways, Ras and ER, and that NF1-deficient ER+ breast cancer patients may be more effectively treated by co-targeting the Ras and ER signaling. These patients, up to 10% of those with advanced ER+ breast cancer, can be readily identified for treatment by ctDNA analysis. A clinical trial is under development. Citation Format: Chang EC, Zheng Z, Philip L, Burcu C, Lei J, Singh P, Anurag M, Chan D, Li JD, Du XP, Shafaee MN, Banks K, Sacker S, Song W, Nguyen T, Cao J, Chen X, Haricharan S, Kavuri M, Kim B-J, Zhang B, Gutmann DH, Lanman RB, Foulds C, Ellis M. Direct regulation of estrogen receptor-α (ER) transcriptional activity by NF1 [abstract]. In: Proceedings of the 2017 San Antonio Breast Cancer Symposium; 2017 Dec 5-9; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2018;78(4 Suppl):Abstract nr GS2-02.

PARP1/2 inhibitors are effective against BRCA2-deficient tumors. The PARP inhibitor (PARPi) olaparib received FDA breakthrough designation for treatment of metastatic castration-resistant prostate cancers (CRPC) carrying mutations in BRCA1/2 or ATM genes. Emergent resistance to PARPi has been associated with tumor-specific BRCA2 mutations that revert the normal open reading frame rescuing homologous recombination. We describe a case of metastatic CRPC with germline BRCA2 mutation with acquired resistance to olaparib related to biallelic BRCA2 reversion mutations of both the germline and somatic loss of function alleles detected by circulating tumor DNA testing. We also summarize a retrospective analysis of 1,534 prostate cancer cases with ctDNA analysis showing a 1.6% incidence of germline BRCA2 mutations. Within the germline BRCA2-positive cases exposed to platinum chemotherapy or PARP inhibition, the prevalence of reversion mutations was 40%. This report documents the frequency of reversion mutations in a large cohort of prostate cancer patients carrying of BRCA mutations. It also shows the potential utility of ctDNA analyses for early detection of reversion mutation driving tumor resistance.

Background: To identify predictive markers for responders in lapatinib-treated patients and to demonstrate molecular changes during lapatinib treatment via cell-free genomics. Patients and methods: We prospectively evaluated the efficacy of combining lapatinib with capecitabine and oxaliplatin as first line neoadjuvant therapy in patients with previously untreated, HER2-overexpressing advanced gastric cancer. A parallel biomarker study was conducted by simultaneously performing immunohistochemistry and next-generation sequencing (NGS) with tumor and blood samples. Results: Complete response was confirmed in 7/32 patients (21.8%), 2 of whom received radical surgery with pathologic-confirmed complete response. Fifteen partial responses (46.8%) were observed, resulting in a 68.6% overall response rate. NGS of the 16 tumor specimens demonstrated that the most common co-occurring copy number alteration was CCNE1 amplification, which was present in 40% of HER2+ tumors. The relationship between CCNE1 amplification and lack of response to HER2-targeted therapy trended toward statistical significance (66.7% of non-responders versus 22.2% of responders harbored CCNE1 amplification; P = 0.08). Patients with high level ERBB2 amplification by NGS were more likely to respond to therapy, compared with patients with low level ERBB2 amplification (P = 0.02). Analysis of cfDNA showed that detectable ERBB2 copy number amplification in plasma was predictive to the response (100%, response rate) and changes in plasma-detected genomic alterations were associated with lapatinib sensitivity and/or resistance. The follow-up cfDNA genomics at disease progression demonstrated that there are emergences of other genomic aberrations such as MYC, EGFR, FGFR2 and MET amplifications. Conclusions: The present study showed that HER2+GC patients respond differently according to concomitant genomic aberrations beyond ERBB2, high ERBB2 amplification by NGS or cfDNA can be a positive predictor for patient selection, and tumor genomic alterations change significantly during targeted agent therapy.

Background: Plasma cell-free DNA (cfDNA) assays are increasingly used in clinic. Despite their rapid adoption, best practices for use and interpretation need better definition. We assessed key clinically-relevant questions using a prospectively collected cohort of EGFR-mutant pts. Methods: Starting in 2015, serial cfDNA testing via Guardant360 NGS was obtained from MGH pts with advanced EGFR-mutant NSCLC across multiple lines of therapy in an IRB-approved project. Medical records were analyzed retrospectively, tissue was genotyped by institutional NGS (SNaPshot), scans were assessed by RECIST. Correlations were tested by Wilcoxon Rank-Sum. Results: 372 plasma samples were collected from 89 pts, covering 150 therapy regimens including 26 drugs (targeted, immune, cytotoxic). To assess genotype correlation, we examined matched cfDNA and tissue biopsies at clinical progression for 60 regimens (51 pts). cfDNA-tissue concordance was 73% for founder EGFR mutations (n = 60), 72% for T790M (n = 60) and 89% for MET amp (n = 53). Excluding 15 samples without detectable founder mutation (presumed “non-shedders”), concordance for T790M=89% and MET=90%. To assess if relative change in cfDNA allelic fraction (AF) correlates with radiographic response, we examined 21 regimens (19 pts) with cfDNA samples at baseline and ≤ 30 days, and 3 aspects of the Guardant360 report. 12/21 regimens yielded PR by scans. Decrease in cfDNA AF in the 1st month of therapy correlated with ultimate PR whether assessing the change in AF of founder EGFR (p = 0.03), largest AF regardless of gene (p = 0.02) or sum of all detected AFs (p = 0.02). Conclusions: CommercialcfDNA assays are readily available, facilitate serial AF monitoring and provide clinically-relevant data at acquired resistance. Among EGFR pts, we found real world cfDNA-tissue correlation of founder mutations and T790M was high and resistance mutation (T790M, MET) results in cfDNA may be most reliable when founder EGFR mutations are detected. Importantly, MET amp had high cfDNA-tissue concordance, which was unexpected. Early AF decrease (within 1st month of therapy) significantly correlates with radiographic response regardless of which aspect of the Guardant360 report is considered. Further investigation is needed to inform optimal use and interpretation of cfDNA assays. Legal entity responsible for the study: MGH Cancer Center. Funding: Guardant Healthcare, Lungevity. Disclosure: N. Marcoux: Honoraria: Bristol-Myers Squibb. L.V. Sequist: Consulting fees: AstraZeneca, Pfizer, Genentech; Institutional research funding: AstraZeneca, Novartis, Merck, Boehringer Ingelheim, Genentech, Merrimack, Incyte. A. Hata: Research/grant support: Novartis, Amgen, Relay Therapeutics. I. Dagogo-Jack: Honoraria: Foundation Medicine; Consulting: Boehringer Ingelheim. R. Nagy, R.B. Lanman: Stock ownership and employee: Guardant Health, Inc. Z. Piotrowska: Consulting/honoraria: AstraZeneca, Ariad/Takeda, GuardantHealth, Novartis, AbbVie; Research support (to institution): Novartis. All other authors have declared no conflicts of interest.

Background: Microsatellite instability (MSI) is a guideline-recommended biomarker used in assessment of prognosis and treatment choices, including checkpoint inhibitors recently approved for cancers with MSI-high (MSI-H) status. Plasma-based next generation DNA sequencing (NGS) tests are increasingly used for comprehensive genomic profiling of cancer; however, sensitive methods to detect MSI status from cell-free DNA (cfDNA) are not available for clinical patient care. Additionally, the impact of variable tumor shedding on MSI detection has not been evaluated. Methods: We developed an accurate method to assess MSI status using targeted sequencing of cfDNA using the Guardant360® clinical platform across a many cancer types, which allows broad coverage of simple repeats. For each microsatellite locus, the number of differently-sized repeats in experimental samples is quantified using a probabilistic log likelihood-based score designed to accurately discriminate biological signal derived from cfDNA fragments of somatic origin from noise arising from technical artifacts. Loci are considered unstable if the likelihood score is greater than a threshold computed from a cohort of normal samples. MSI status of a sample is determined by the presence of a minimum 5 unstable microsatellite loci among the 91 scored. Results: We simulated MSI high (MSI-H) samples across a range of tumor fractions by combining data from 82 healthy donor samples with in silico spike-ins of differentially sized repeats. Simulated data demonstrates a sensitivity of 94% at 0.2% (limit of detection) tumor content for an expected specificity of 99.9% estimated from healthy donor samples. When applied to a prospective test set of 134 advanced cancer samples, this method demonstrated 98.5% (125/127) specificity and 86% sensitivity (6/7) relative to standard tissue PCR-based MSI assessment across a ctDNA range of 0.1%-15%. Conclusions: Targeted sequencing of cfDNA data can enable highly accurate detection of MSI in cancer samples, even for samples with low tumor shedding. This novel approach enables non-invasive assessment of MSI status concurrent with comprehensive genomic profiling and allows potential access to immunotherapies for patients whose tumor types are not routinely tested for MSI. Legal entity responsible for the study: Guardant Health, Inc. Funding: Guardant Health, Inc. Disclosure: A. Artyomenko, M. Sikora, M. Lefterova, V.M. Raymond, D. Gavino, C. Barbacioru, C. Artieri, E. Helman, D. Chudova, R.B. Lanman, J. Odegaard, A. Talasaz: Employee: Guardant Health, Inc. All other authors have declared no conflicts of interest.

Background: Genomic profiling of ctDNA has proven to be an effective alternative to repeat invasive biopsy in patients (pts) with advanced cancers. There is also the advantage of a more comprehensive approach over tissue-based assays with the ability to provide a summary of tumour heterogeneity. Methods: We performed a retrospective review of Hong Kong pts with advanced/metastatic NSCLC whose physician requested ctDNA-based genomic profiling utilizing the Guardant360 platform between Jan 2016 - Jun 2017. Guardant360 includes all four major types of genomic alterations (point mutations, and selected indels, fusions, and amplifications) and completely sequences exons in 73 target genes. Results: ctDNA testing was performed in 76 pts over this 18-month period (Median Age: 59.5 years (range 42-87), M:F 41:35). Histologies, as reported by the ordering physician, include squamous (SqCC) (n = 7), adenocarcinoma (Adeno) (n = 10), and NSCLC-not otherwise specified (NSCLC-NOS) (n = 58). In SqCC, all 7 pts had multiple detectable variants identified (range: 2-20 variants, median = 6), including FGFR1 amplification (n = 3), ERBB2 (HER2) amplification (n = 2). PIK3CA amplification occurred in combination with either FGFR1 or ERBB2 (HER2) amplification (n = 1 each), or alone (n = 1). In the Adeno and NSCLC-NOS groups combined, 91% of pts (61/68) had variants identified (range: 1-12 variants, median = 3), of which 42% (26/62) had at least one of the seven NCCN recommended lung adenocarcinoma genomic targets (EGFR (21%), EML4-ALK (8.1%), ERBB2 exon 20 insertion (6%), MET Amp (3.2%), ROS1 (2%), BRAF V600E (2%)). Concurrent detection of driver and resistance mutations were identified in 6/13 patients with EGFR driver mutations (T790M (n = 1), T790M/C797S (n = 1), MET amp (n = 2), T790M/MET amp (n = 1) and ERBB2 amp (n = 1)) and in 3/5 patients with EML4-ALK fusion (MET exon 14 skipping (n = 1) and the ALK L1196M gatekeeper mutation (n = 2)). Conclusions: Genomic profiling utilizing ctDNA analysis detected alterations in majority of advanced NSCLC pts, with targetable aberrations and resistance mechanisms identified. This approach has prompted changes in matched therapy in selected pts, and has demonstrated its feasibility in Asia. Legal entity responsible for the study: The Chinese University of Hong Kong Funding: None Disclosure: H. Loong: Advisory Board: Celgene, Novartis, Roche Travel Support: BMS, MSD, Novartis, Roche, TaiHo Speakers' Bureau: Abbvie, Novartis Research Funding: MSD, Mundipharma, V. Raymond: Employee of Guardant Health. T. Yung: Employee of Sanomics Limited. R.B. Lanman: Employee and shareholder of Guardant Health. S. Skrzypczak: Employee of Guardant Health, T.S.K. Mok: Shareholder of Sanomics Limited. All other authors have declared no conflicts of interest.

BackgroundWhile deregulation of the cyclin D1-CDK4/6-retinoblastoma pathway is common in hormone receptor positive (HR+) breast cancer, Rb is usually intact in HR+ breast cancer, and targeted CDK 4/6 inhibitors that act upstream of Rb, are routinely being utilized in clinical practice. However, factors that can lead to clinical resistance to CDK 4/6 inhibitors are not known.Patients and methodsWe identified patients who had pre- and post-genotyping in tissue and peripheral blood samples after receiving CDK 4/6 inhibitors. Genotyping was carried out in tumor tissue or blood collected before start of CDK 4/6 inhibitor and after disease progression on CDK 4/6 inhibitor, covering more than 90% of the coding region in RB1.ResultsWe identified detectable acquired RB1 mutations in circulating tumor DNA (ctDNA) after exposure to CDK4/6 inhibitor (palbociclib, palbociclib, ribociclib) for 5, 8, and 13 months, respectively, in three patients. The RB1 mutations included substitution in donor splicing site of exon 8 of the RB1 gene in patient #1; substitution in donor splicing site of exon 22 of RB1 gene, exon 19 deletion, exon 3 insertion in patient #2; and RB1 exon 16 H483Y mutation in patient #3. None of these RB1 mutations were present in the pre-CDK 4/6 specimen highlighting these molecular alterations, which lead to functional loss of Rb1, likely emerged under selective pressure from the CDK4/6 inhibitor potentially confering therapeutic resistance.ConclusionThis is the first clinical report to describe the emergence of somatic RB1 mutations after exposure to palbociclib or ribociclib, in patients with metastatic breast cancer. Further research is needed to validate these findings, identify how these mutations temporally emerge under selective pressure of CDK 4/6 inhibitor, and develop rational therapeutic strategies.

Abstract Background: Next-generation sequencing (NGS) of cell-free circulating tumor DNA (ctDNA) enables a non-invasive option for comprehensive genomic analysis of lung cancer patients. Currently there is insufficient data in regard to the impact of ctDNA analysis on clinical decision making. In this study, we evaluated the clinical utility of ctDNA sequencing on treatment strategy and progression-free survival. Methods: In this retrospective study, data was collected from files of 92 NSCLC patients monitored between the years 2014-2016 at the Thoracic Center Unit at Davidoff Cancer Center, Rabin Medical Center, Israel. Plasma samples from stage IIIb/IV non-small cell lung cancer (NSCLC) patients were analyzed by a commercial test (Guardant 360), using hybrid capture, single molecule barcoding and massively parallel paired-end synthesis to sequence a targeted gene panel. This test allows the detection of somatic alterations such as point mutations, indels, fusions and copy number amplifications. Results: 92 consecutive NSCLC patients were included in this study. Median age at diagnosis was 63 years, male:female ratio was 1:1.6. 40% (37/92) were never-smokers, 84% (77/92) had adenocarcinoma. 38% (35/92) performed ctDNA analysis before 1st line therapy and 62% (57/92) on progression. ctDNA analysis yielded lung cancer related actionable mutations in total 39% (36/92) of the patients; 31% (11/35) among upfront testing and 44% (25/57) among patients at progression on matched therapy. Treatment decision was taken toward targeted therapy subsequent to NGS analysis in 23% (8/35) and 26% (15/57) respectively (total 25%; 23/92). 53 individual actionable genomic alterations were found. The most common genes were sensitizing EGFR mutations (47.2%; 25/53), MET amplifications and/or exon 14 skipping mutations (17%; 9/53) and resistance EGFR mutations (13.2%; 7/53). Response assessment (RECIST) for 18 patients with evaluable response to targeted therapy showed complete response in 6% (1/18), partial response in 39% (7/18), stable disease in 22% (4/18) and progressive disease in 33% (6/18). Total objective response rate (ORR) was 45% and disease control rate was 67%. Conclusions: Comprehensive ctDNA testing revealed possible treatment options for two-thirds of patients analyzed. ctDNA analysis impacted clinical decision making in a quarter of the patients. Although this topic needs to be further assessed in large randomized controlled trials, these positive results emphasize the utility of liquid biopsy analysis to guide clinicians to select the right therapy for the right patient. Citation Format: Smadar Geva, Anna Belilovski Rozenblum, Tal Twito, Addie Dvir, Lior Soussan-Gutman, Maya Ilouze, Laila C. Roisman, Elizabeth Dudnik, Alona Zer, Richard B. Lanman, Nir Peled. The clinical impact of multiplex ctDNA gene analysis in lung cancer [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 5394. doi:10.1158/1538-7445.AM2017-5394