Supplemental Figure 1. Distribution of EGFR extracellular domain (ECD) mutations in ctDNA samples from patients with CRC. Supplemental Figure 2. Response to crizotinib in a patient with small bowel adenocarcinoma and GOPC-ROS1 fusion. Supplemental Figure 3. Detection of copy number amplification is associated with increased ctDNA fraction. Supplemental Table 1. List of 62 genes sequenced in this study using the FoundationACT ctDNA assay. Supplemental Table 2. Genomic alterations detected in temporally matched ctDNA and tissue samples.

Background Liquid biopsy offers the ability to non-invasively analyze the genome of a tumor through circulating tumor DNA (ctDNA) to identify targetable and prognostic genomic alterations. Few studies have rigorously analyzed ctDNA results and determined the fidelity with which they recapitulate the genomics of a sequenced tissue sample obtained from the same tumor. The clinical utility study (CUS) for the FoundationACT™ ctDNA assay (Foundation Medicine, Cambridge, MA, USA; NCT02620527) is a multi-center prospective clinical study for multiple solid tumor types to compare genomic profiling of paired tissue and blood samples from the same patient. In this subset of the study, paired specimens from 96 patients with colorectal cancer (CRC) were analyzed with comprehensive genomic profiling (CGP) of the tumor tissue sample (FoundationOne®) and blood sample (FoundationACT™). Methods Both samples underwent CGP using the hybrid capture-based Illumina Hi-Seq technology. Maximum somatic allele frequency (MSAF) was used to estimate the fraction of ctDNA in the sample. The set of genes and targeted regions common to both tumor and liquid were compared for each subject. Results Among these patients, 61% were male; 74% had clinical stage IV disease, 19% had clinical stage III disease, and 7% had clinical stage II disease. Time between the tissue biopsy and liquid biopsy (range, 0-709 days) had a significant impact on the positive percent agreement (PPA) between the two assays. Eighty percent of cases had evidence of ctDNA in the blood (MSAF >0). For all cases with MSAF >0, 171 base substitutions and insertions/deletions (indels) were identified in the tumor, and 79% (PPA) of these identical alterations were also identified in matched ctDNA samples; PPA increased to 87% for cases <270 days between the tissue and liquid biopsy, 95% for <90 days, and 100% PPA for <30 days. All known and likely short variants in KRAS, NRAS, and BRAF were analyzed independently as testing of these genes is recommended by the National Comprehensive Cancer Network (NCCN) for patients with CRC and have therapeutic implications. For NCCN genes, PPA was 80% for all time points for short variants; PPA increased to 90% for cases <270 days between the tissue and liquid biopsy. There was high concordance for KRAS G12X between tissue and liquid: overall percent agreement (97%), PPA (93%), negative percent agreement (NPA) (100%), positive predictive value (PPV) (100%), and negative predictive value (NPV) (96%) for the <270 day cohort. Conclusions In cases where tumor tissue profiling is not possible, these results provide compelling evidence that genomic profiling of ctDNA in late stage CRC shows a high concordance with tumor tissue sequencing results and can be used to identify most clinically relevant alterations capable of guiding therapy for these patients.

Background: Analysis of circulating tumor DNA (ctDNA) provides a noninvasive method to profile tumor-associated genomic aberrations, heterogeneity and evolution. Methods: Peripheral blood from 5 newly diagnosed and 33 relapsed high-risk neuroblastoma patients was serially profiled for ctDNA (n=1-7 samples/patient; total n=122 samples) with the FoundationACT assay that utilizes hybrid capture-based genomic profiling of 62 genes. Samples were sequenced to a median unique coverage depth of at least 3168x and variants were evaluated and compared with temporally-matched tumor sequencing and imaging evaluations. Results: Ninety-three percent (114/122) of peripheral blood samples yielded suitable cell-free DNA for sequencing. ctDNA was detected in 68% (78/114) of samples (maximum somatic allele frequency [MSAF] >0; median MSAF=0.53%; range MSAF 0-79%). At least 1 pathogenic genomic alteration (genomic short-variant or amplification) was found in 54% (61/114) of samples (range 1-6 alterations). Fifty-four percent (13/24) of detected ctDNA genomic-short variants were not present in temporally matched tumor samples at diagnosis or relapse (collected within 3 months of each other; n=23 patients), including pathogenic variants in ALK, TP53, TERT, NF1, FLT3, PTPN11, and PIK3CA. There was 100% concordance between the detection of MYCN (6/6) and ALK (3/3) amplification in paired ctDNA/tissue samples. For example, a newly diagnosed patient with stage 4 neuroblastoma had MYCN amplification noted on both tumor and ctDNA sequencing, however had 3 separate ALKmutations (R1275Q, F1245L, and F1174L) uniquely identified in ctDNA. Twenty-eight patients had multiple ctDNA samples sequenced (range 2-7 samples) and 50% (14/28) had alterations emerge or regress across serial samples. For example, pathogenic variants in ALK, BRCA2, NRAS, PTEN, TP53, CDKN2A, PTPN11, ABL1,CDH1, MET, ERRFl1 and ERBB2 appeared in subsequent ctDNA sequencing that were not present in the initial ctDNA or tumor sequencing. Overall, serial ctDNA profiling identified additional pathogenic variants in driver cancer genes beyond that derived from tumor sequencing in 45% (17/38) of cases, including identifying targetable ALK-RAS-MAPK pathway alterations uniquely in ctDNA in 21% (8/38) of cases. For the 18 patients that had 3 or more ctDNA samples, 33% (6/18) developed ctDNA unique ALK-RAS-MAPK pathway mutations during therapy. Finally, in most cases ctDNA profiling was complementary to standard imaging surveillance, however in 3 cases rising ctDNA allele frequencies occurred prior to clinical or imaging signs of disease relapse; additional correlation of ctDNA data and disease evaluations is ongoing. Conclusions: Sequencing of ctDNA from neuroblastoma patients identified clinically actionable tumor-associated genetic aberrations emerging under the selective pressure of standard and targeted therapies. Citation Format: Kristopher R. Bosse, Samantha Buongervino, Maria Lane, Adam Hyman, Maria Gemino-borromeo, Jennifer Saggio, Alana Fitzsimmons, Brady Forcier, Anne Murphy, John Wick, Matthew Cooke, Jennifer Webster, Russell Madison, Alley Welsh, Vincent A. Miller, Siraj M. Ali, John M. Maris, Yael P. Mosse. Serial profiling of ctDNA identifies clinically actionable genomic evolution in high-risk neuroblastoma [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 3105.

Abstract Purpose: Genomic profiling of tumor biopsies from advanced gastrointestinal and anal cancers is increasingly used to inform treatment. In some cases, tissue biopsy can be prohibitive, and we sought to investigate whether analysis of blood-derived circulating tumor DNA (ctDNA) may provide a minimally invasive alternative. Experimental Design: Hybrid capture–based genomic profiling of 62 genes was performed on blood-based ctDNA from 417 patients with gastrointestinal carcinomas to assess the presence of genomic alterations (GA) and compare with matched tissue samples. Results: Evidence of ctDNA was detected in 344 of 417 samples (82%), and of these, ≥1 reportable GA was detected in 89% (306/344) of samples. Frequently altered genes were TP53 (72%), KRAS (35%), PIK3CA (14%), BRAF (8%), and EGFR (7%). In temporally matched ctDNA and tissue samples available from 25 patients, 86% of alterations detected in tissue were also detected in ctDNA, including 95% of short variants, but only 50% of amplifications. Conversely, 63% of alterations detected in ctDNA were also detected in matched tissue. Examples demonstrating clinical utility are presented. Conclusions: Genomic profiling of ctDNA detected potentially clinically relevant GAs in a significant subset of patients with gastrointestinal carcinomas. In these tumor types, most alterations detected in matched tissue were also detected in ctDNA, and with the exception of amplifications, ctDNA sequencing routinely detected additional alterations not found in matched tissue, consistent with tumor heterogeneity. These results suggest feasibility and utility of ctDNA testing in advanced gastrointestinal cancers as a complementary approach to tissue testing, and further investigation is warranted. Clin Cancer Res; 24(8); 1881–90. ©2018 AACR.

569 Background: The liquid biopsy offers the ability to non-invasively analyze the genome of a tumor through circulating tumor (ct) DNA to identify targetable and prognostic genomic alterations. Few studies have rigorously analyzed ctDNA results and determined the fidelity with which they recapitulate the genomics of the tumor. The clinical utility study (CUS) for FoundationACT (Foundation Medicine, Cambridge, MA; NCT02620527) is a large multi-center prospective clinical study to validate this ctDNA assay for multiple solid tumor types. In this subset of the CUS study, paired specimens from 98 patients with colon cancer were analyzed with comprehensive genomic profiling of the tumor (FoundationOne) and a blood sample (FoundationACT). Methods: Maximum somatic allele fraction (MSAF) was used to estimate the fraction of ctDNA in the sample. The set of genes and targeted regions common to both FoundationOne and FoundationACT were compared for each subject. Results: 60% were male; 73 had stage IV, 17 had stage III, and 8 had stage II disease. 16% of cases had an MSAF value of 0, indicating that no ctDNA were in these samples. For the cases with MSAF > 0, 153 insertion/deletions (indels)/substitutions were identified in the tumor, and 73% of these identical alterations were also identified in the ctDNA samples. As robust analytical performance for FoundationACT has been demonstrated at MSAF > 0.5% for indel/subs, further analysis was conducted using this threshold. 83% of the alterations identified with FoundationOne were identified with FoundationACT. Further, 90% of NRAS/KRAS and 75% BRAF alterations (NCCN guideline genes) were concordant between the two assays. Lastly, there were 11 patients with tumor biopsy and blood draw taken within 30 days of each other, and in these there was 100% concordant. Association of the genomics results with other clinical variables will be presented for the available subset of patients. Conclusions: In cases where tumor profiling is not possible, these results provide compelling evidence that comprehensive molecular profiling of ctDNA in colon cancer can be used to identify most clinically relevant alterations and has fidelity to the genomics of the tumor. Clinical trial information: NCT02620527.

Introduction: Genomic profiling informs selection of matched targeted therapies as part of routine clinical care in NSCLC. Tissue biopsy is the criterion standard; however, genomic profiling of blood-derived circulating tumor DNA (ctDNA) has emerged as a minimally invasive alternative. Methods: Hybrid capture-based genomic profiling of 62 genes was performed on blood-based ctDNA from 1552 patients with NSCLC. Results: Evidence of ctDNA was detected in 80% of samples, and in 86% of these cases, at least one reportable genomic alteration (GA) was detected. Frequently altered genes were tumor protein p53 gene (TP53) (59%), EGFR (25%), and KRAS (17%). Comparative analysis with a tissue genomic database (N = 21,500) showed similar frequencies of GAs per gene, although KRAS mutation and EGFR T790M were more frequent in tissue and ctDNA, respectively (both p < 0.0001), likely reflecting the use of liquid versus tissue biopsy after relapse during targeted therapy. In temporally matched ctDNA and tissue samples from 33 patients with evidence of ctDNA in their blood, 64% of GAs detected in tissue were also detected in ctDNA, including 78% of short variants (58 of 74) and 100% of rearrangements (four of four), but only 16% of amplifications (four of 25). Conclusions: Genomic profiling of ctDNA detected clinically relevant GAs in a significant subset of NSCLC cases. Most alterations detected in matched tissue were also detected in ctDNA. These results suggest the utility of ctDNA testing in advanced NSCLC as a complementary approach to tissue testing. Blood-based ctDNA testing may be particularly useful at the time of progression during targeted therapy. (C) 2018 International Association for the Study of Lung Cancer. Published by Elsevier Inc.

This retrospective study was undertaken to determine if the plasma circulating tumor DNA (ctDNA) level and tumor biological features in patients with advanced solid tumors affected the detection of genomic alterations (GAs) by a plasma ctDNA assay.

453 Background: Emerging data suggest a role for targeted therapy in patients (pts) with mUC whose tumors contain actionable genomic alterations (GA). Liquid biopsy of ctDNA from blood provides an attractive alternative for CGP in pts in whom tissue-based testing is not feasible. Methods: Hybrid capture-based genomic profiling was performed using FoundationACT assay in a CLIA-certified, CAP-accredited, NY State-approved laboratory. Cell free DNA was extracted from plasma using 20 ml whole blood and underwent CGP of 62 genes. Sequencing was performed to a median unique coverage of 6756× using the Illumina HiSeq 2500/4000 platform. Barcode-based error correction enabled analysis of GA at low allele frequency (AF), including base substitutions and short in/dels (AF≥0.1% for both), rearrangements, and copy number amplification. For several pts, non-temporally matched tissue-based CGP data was available. Comparisons to other CGP datasets of UC ctDNA and tissue were performed. Results: 66 pts with median age 68 (range 29-86) underwent ctDNA assessment as clinical test. There was evidence for ctDNA in 56 pts (85%), and ≥1 GA was noted in 48 pts (73%). The estimated median ctDNA fraction in plasma was 1.9%. In cases with detectable ctDNA, the most frequently altered genes were TP53 (68%), TERT-promoter (38%), PIK3CA (13%), FGFR3 (13%), KRAS (11%), NF1 (6%), ERBB2 (6%). Observed GA frequencies were comparable to prior studies of ctDNA and tissue. Both tissue and ctDNA-based CGP data with clinical correlation was available for several pts. A pt with FGFR3 GA in baseline tumor tissue had disappearance of FGFR3 GA and detection of new TP53 GA in ctDNA after targeted treatment with FGFR3 inhibitor. In a pt with ERBB2 and TP53 GAs in baseline tumor tissue, ctDNA at time of resistance to cisplatin-based therapy showed persistence of ERBB2 and TP53 GAs and new NF1 GA. Conclusions: Most pts with mUC had detectable ctDNA, and frequencies were comparable to prior data sets Detection of potentially targetable GA support further clinical utility assessment. Concordance between tissue and ctDNA in temporally matched samples with clinical annotation warrants further investigation.