Unique Molecular Barcodes Research Articles

Enhanced microbial diversity and chemoradiation response in HPV+ anal cancer.

4 Background: Anal cancer is a rare, but treatable, malignancy caused by the human papilloma virus (HPV). Because anal cancer is uncommon and carries a significant social stigma, this disease has rarely been studied in a prospective fashion, creating a large knowledge gap of which biological factors can improve treatment responses, particularly the microbiome which regulates many facets of oncobiology. Methods: We collected microbiome samples before, during, and after definitive chemoradiation for localized anal cancer as part of an IRB-approved institutional clinical trial (2017-0606). Samples were collected by a non-invasive swab biopsy that enabled collection of tumor cells and microbiome during standard-of-care treatment visits. Bacterial genomic DNA is extracted using MO BIO PowerSoil DNA Isolation Kit (MO BIO Laboratories) and the 16S rDNA V4 region is amplified by PCR and sequenced on the MiSeq platform (Illumina) using the 2x250 bp paired-end protocol, yielding pair-end reads that overlapped almost completely. Sequence read pairs are demultiplexed using unique molecular barcodes, and reads are merged using USEARCH version 7.0.1090 and grouped by organizational taxonomy units (OTU’s) representative of unique bacterial species. Responders are were those patients who had more than 60% tumor shrinkage by MRI at the midpoint of treatment, and all others were considered non-responders. Results: As of June 2019 we enrolled 17 patients and analyzed 14. The majority of patients (10/14, 71.4%) were responders but unfortunately, 4 patients developed in-field recurrences of their anal cancer within a year after their treatment ended and were all T3 or T4 tumors. We found that alpha-diversity of the microbiome did not change during chemoradiation, but non-responders exhibited a lower alpha diversity compared to responders at baseline. Furthermore, specific taxa were correlated with treatment response, which imply a basis for future therapy. Conclusions: The alpha diversity of the microbiome is correlated with treatment response in HPV+ anal cancer. The idenitification of specific bacterial species between responders and non-responder might pave the way for more effective therapeis in the future.

Corrigendum to: Identification of Disease-associated Traits and Clonotypes in the T Cell Receptor Repertoire of Monozygotic Twins Affected by Inflammatory Bowel Diseases

Background and aims Intestinal inflammation in inflammatory bowel diseases [IBD] is thought to be T cell mediated and therefore dependent on the interaction between the T cell receptor [TCR] and human leukocyte antigen [HLA] proteins expressed on antigen presenting cells. The collection of all TCRs in one individual, known as the TCR repertoire, is characterised by enormous diversity and inter-individual variability. It was shown that healthy monozygotic [MZ] twins are more similar in their TCR repertoire than unrelated individuals. Therefore MZ twins, concordant or discordant for IBD, may be useful to identify disease-related and non-genetic factors in the TCR repertoire which could potentially be used as disease biomarkers. Methods Employing unique molecular barcoding that can distinguish between polymerase chain reaction [PCR] artefacts and true sequence variation, we performed deep TCRα and TCRβ repertoire profiling of the peripheral blood of 28 MZ twin pairs from Denmark and Germany, 24 of whom were discordant and four concordant for IBD. Results We observed disease- and smoking-associated traits such as sharing, diversity and abundance of specific clonotypes in the TCR repertoire of IBD patients, and particularly in patients with active disease, compared with their healthy twins. Conclusions Our findings identified TCR repertoire features specific for smokers and IBD patients, particularly when signs of disease activity were present. These findings are a first step towards the application of TCR repertoire analyses as a valuable tool to characterise inflammatory bowel diseases and to identify potential biomarkers and true disease causes.

Alignparse: A Python package for parsing complex features from high-throughput long-read sequencing.

Summary & PurposeAdvances in sequencing technology have made it possible to generate large numbers of long, high-accuracy sequencing reads. For instance, the new PacBio Sequel platform can generate hundreds of thousands of high-quality circular consensus sequences in a single run (Hebert et al., 2018; Rhoads & Au, 2015). Good programs exist for aligning these reads for genome assembly (Chaisson & Tesler, 2012; Li, 2018). However, these long reads can also be used for other purposes, such as sequencing PCR amplicons that contain various features of interest. For instance, PacBio circular consensus sequences have been used to identify the mutations in influenza viruses in single cells (Russell et al, 2019), or to link barcodes to gene mutants in deep mutational scanning (Matreyek et al., 2018). For such applications, the alignment of the sequences to the targets may be fairly trivial, but it is not trivial to then parse specific features of interest (such as mutations, unique molecular identifiers, cell barcodes, and flanking sequences) from these alignments.Here we describe alignparse, a Python package for parsing complex sets of features from long sequences that map to known targets. Specifically, it allows the user to provide complex target sequences in Genbank Flat File format that contain an arbitrary number of user-defined sub-sequence features (Sayers et al., 2019). It then aligns the sequencing reads to these targets and filters alignments based on whether the user-specified features are present with the desired identities (which can be set to different thresholds for different features). Finally, it parses out the sequences, mutations, and/or accuracy (sequence quality) of these features as specified by the user. The flexibility of this package therefore fulfills the need for a tool to extract and analyze complex sets of features in large numbers of long sequencing reads.

A systematic capsid evolution approach performed in vivo for the design of AAV vectors with tailored properties and tropism

Adeno-associated virus (AAV) capsid modification enables the generation of recombinant vectors with tailored properties and tropism. Most approaches to date depend on random screening, enrichment, and serendipity. The approach explored here, called BRAVE (barcoded rational AAV vector evolution), enables efficient selection of engineered capsid structures on a large scale using only a single screening round in vivo. The approach stands in contrast to previous methods that require multiple generations of enrichment. With the BRAVE approach, each virus particle displays a peptide, derived from a protein, of known function on the AAV capsid surface, and a unique molecular barcode in the packaged genome. The sequencing of RNA-expressed barcodes from a single-generation in vivo screen allows the mapping of putative binding sequences from hundreds of proteins simultaneously. Using the BRAVE approach and hidden Markov model-based clustering, we present 25 synthetic capsid variants with refined properties, such as retrograde axonal transport in specific subtypes of neurons, as shown for both rodent and human dopaminergic neurons.

Identification of Disease-associated Traits and Clonotypes in the T Cell Receptor Repertoire of Monozygotic Twins Affected by Inflammatory Bowel Diseases.

Background and AimsIntestinal inflammation in inflammatory bowel diseases [IBD] is thought to be T cell mediated and therefore dependent on the interaction between the T cell receptor [TCR] and human leukocyte antigen [HLA] proteins expressed on antigen presenting cells. The collection of all TCRs in one individual, known as the TCR repertoire, is characterised by enormous diversity and inter-individual variability. It was shown that healthy monozygotic [MZ] twins are more similar in their TCR repertoire than unrelated individuals. Therefore MZ twins, concordant or discordant for IBD, may be useful to identify disease-related and non-genetic factors in the TCR repertoire which could potentially be used as disease biomarkers.MethodsEmploying unique molecular barcoding that can distinguish between polymerase chain reaction [PCR] artefacts and true sequence variation, we performed deep TCRα and TCRβ repertoire profiling of the peripheral blood of 28 MZ twin pairs from Denmark and Germany, 24 of whom were discordant and four concordant for IBD.ResultsWe observed disease- and smoking-associated traits such as sharing, diversity and abundance of specific clonotypes in the TCR repertoire of IBD patients, and particularly in patients with active disease, compared with their healthy twins.ConclusionsOur findings identified TCR repertoire features specific for smokers and IBD patients, particularly when signs of disease activity were present. These findings are a first step towards the application of TCR repertoire analyses as a valuable tool to characterise inflammatory bowel diseases and to identify potential biomarkers and true disease causes.

Abstract 2516: A bioinformatics framework for high-sensitivity detection and monitoring of oncogenic gene fusions in plasma cfDNA

Abstract Gene fusions represent key oncogenic driver alterations in many solid tumors that can be targeted therapeutically. Beyond FDA-approved drugs targeting ALK and ROS1 fusions in lung cancer, investigational agents targeting gene fusions involving RET, NTRK1/2/3, and FGFR2/3 have elicited significant and durable responses in a wide range of cancer types. Cell-free DNA (cfDNA) profiling provides a transformative opportunity to non-invasively and longitudinally monitor responses to these investigational therapies and identify acquired mutations that confer drug resistance. We have developed a pan-cancer, high-sensitivity NGS cancer assay (MSK-ACCESS, presented previously) to detect and monitor somatic mutations in plasma cfDNA. Here, we present an improved bioinformatics pipeline that detects gene fusions and achieves greater sensitivity and accuracy for detecting point mutations and indels. MSK-ACCESS uses ultra-high depth sequencing (~20,000x) coupled with duplex unique molecular indexing (UMI) and dual sample barcodes to achieve high sensitivity through background error suppression. The panel was designed to capture key exons and domains of 129 cancer genes as well as introns harboring recurrent breakpoints in 10 commonly rearranged genes. We have now developed a bioinformatics module for sensitive detection of structural variation from cfDNA, leveraging the high-depth sequencing data and intronic coverage. We have also added a statistical polishing method that assigns confidence to detected somatic mutations by building frequency distributions for those mutations from a large set of normal samples that has been processed with the same laboratory protocol and bioinformatics pipeline as the patient samples. In parallel, we have further improved the wet lab protocols to address challenges of capturing sufficient numbers of cfDNA molecules and removing noise, both required for detection of somatic mutations with high sensitivity and specificity. Using this method, we have successfully detected druggable fusions in plasma cfDNA from patients spanning a diverse set of cancer types. Through longitudinal cfDNA collection efforts embedded within early-phase clinical trials, we demonstrate that the levels of oncogenic driver fusions in cfDNA may scale with the burden of disease. Moreover, the ability to simultaneously call high-confidence mutations at low allele fractions has revealed novel acquired mutations indicative of polyclonal resistance to investigational targeted therapies. Citation Format: Juber Patel, Maysun Hasan, Fanli Meng, Xiaohong Jing, Grittney Tam, Ian Johnson, Youyun Zheng, Chaitanya Bandlamudi, Caitlin Stewart, Helen Won, Oliver Hampton, Alison Schram, Ezra Rosen, Alexander Drilon, Anna Varghese, David Hyman, Dana Tsui, Brian Houck-Loomis, Michael Berger. A bioinformatics framework for high-sensitivity detection and monitoring of oncogenic gene fusions in plasma cfDNA [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 2516.

PF344 HIGH FREQUENCY ASXL1 MUTATIONS AND MULTIPLE SUBCLONAL PIGA MUTATIONS ARE OBSERVED IN CELLULAR AND CELL‐FREE DNA IN PAROXYSMAL NOCTURNAL HAEMOGLOBINURIA

Background:Paroxysmal nocturnal haemoglobinuria (PNH) is a life‐threatening clinical syndrome caused by acquired mutations in PIGA in the haemopoietic compartment. The clinicopathological features of PNH can overlap with aplastic anaemia (AA) and myelodysplastic syndrome (MDS) and these entities may represent different manifestations of a common underlying pathological process. To date, the landscape of acquired mutations in de novo PNH (i.e. not preceded by a formal diagnosis of AA/MDS) remains incompletely characterised.Aims:To characterise the spectrum of acquired mutations in the cellular and cell‐free compartments in a cohort of patients with de novo PNH.Methods:DNA was extracted from peripheral blood leukocytes (n = 19) or bone marrow (n = 2) from 21 patients with a clinicopathological diagnosis of PNH (defined as a detectable PNH clone by flow cytometry, active haemolysis and not meeting formal diagnostic criteria for MDS or AA). Patients with a pre‐existing diagnosis of AA or MDS were excluded. Cell‐free DNA (cfDNA) was also collected from 19 patients. DNA sequencing libraries were prepared from cellular DNA using both the Peter MacCallum Cancer Centre PanHaem hybridisation and Myeloid amplicon panels (sensitivity 5%) which includes sequence variant detection for recurrently mutated genes associated with haematological malignancy and whole genome copy number assessment. A custom panel using Anchored Multiplex PCR target enrichment chemistry and unique molecular barcodes was used for cfDNA sequencing (sensitivity <1%).Results:At time of sampling 18 of 21 patients were two or more years post diagnosis (median time since diagnosis of 8 years) with a median age of 41 years (range 23 ‐ 82). 16 patients were being treated with complement inhibitor therapy. PIGA mutations were detected in 19 patients, with multiple mutations detected in half the cohort (11 patients, range 2 ‐ 9 PIGA mutations per patient). 21 mutations were detected in the cellular DNA, while cfDNA sequencing revealed 30 additional PIGA mutations at low allelic burden (median allele frequency 2.19%, range 0.57% ‐ 4.86%) (20 frameshift, 15 missense, 10 splice site, 6 nonsense). Two patients did not have a detectable PIGA mutation but instead harboured focal copy number losses at the PIGA locus, which also involved ZRSR2 in the minimal deleted region. Remarkably, 7 of 21 patients had truncating ASXL1 mutations detectable in both the cellular and cfDNA (9 frameshift, 3 nonsense); in 4 patients multiple ASXL1 mutations were detected. Clonal pathogenic mutations were also detected in U2AF1 (2 patients) and BCOR (1 patient). Patients with ASXL1 and/or U2AF1 mutations had significantly lower neutrophil counts (mean 1.7 vs 1.08, p = 0.021) compared to those without. Haemoglobin and platelet counts were not statistically different between the two groups.Summary/Conclusion:Multiple subclonal PIGA mutations are detectable in patients with PNH. In addition acquired pathogenic variants in ASXL1, U2AF1 and ZRSR2 were detected in a third of the cohort. Longitudinal studies are required to definitively determine the natural history of patients harbouring these pathogenic genomic variants. Furthermore cfDNA sequencing may represent a more sensitive approach for the detection of clinically relevant mutations in patients with PNH.

Development and analytical validation of a 523-gene clinical assay for cell-free DNA.

3039 Background: Liquid biopsies are emerging as a powerful complement to tumor biopsies for the clinical management of cancer patients. A large gene panel with robust analytical performance that accurately assesses variants, tumor mutational burden (TMB), and microsatellite instability in plasma would be of high value for immunotherapy studies, monitoring minimal residual disease and early cancer detection. To this end, we have completed the initial validation for the cell-free DNA (cfDNA) assay, TruSight Oncology 500 (TSO500), which interrogates the full coding region of 523 genes plus selected intronic regions for fusion detection in 23 driver genes. Methods: Cell-free DNA was extracted from plasma collected from Streck or EDTA blood tubes and quantitated to achieve an assay input of ≥10 ng. Libraries were constructed using unique molecular identifiers (UMIs) and duplex barcodes for error correction, then enriched by target capture and sequenced on a NovaSeq 6000. Healthy donor (HD) specificity assessment used matched white blood cell results to filter germline and clonal hematopoiesis variants. Contrived specimens were used to evaluate sensitivity. Single nucleotide variants (SNVs) (n = 36), insertion/deletions (indels) (n = 19), copy number variants (CNVs) (n = 6), and fusions (n = 5) were tested in 2 multi-site replicates. Results: Sensitivity of detection at 0.5% variant allele fraction (VAF) was > 95% and > 97% for SNVs and indels, respectively. All expected CNVs were identified at the targeted threshold of ≥1.3X change and showed strong correlation with matched digital PCR results. All fusions were identified at ≥0.4% VAF. Specificity in HD was > 99.99%. In 22 temporally matched tumor and blood samples from late-stage patients, 58% of all reportable mutations in tumor were identified in cfDNA. Preliminary TMB analysis identified one TMB high case with POLE p.P286R observed in both tissue and cfDNA. Conclusions: In this initial validation study the TSO500 cfDNA assay exhibited high sensitivity and specificity consistent with requirements for clinical applications. Ongoing studies will further evaluate TSO500 as a complement or potential alternative to tissue biopsy for the genomic profiling of cancer patients.

Non-invasive detection of acquired resistance to FGFR inhibition in patients with cholangiocarcinoma harboring FGFR2 alterations.

4096 Background: FGFR2 alterations are present in 14% of cholangiocarcinomas (CCA) and are promising targets of investigational FGFR-directed therapies. Cell-free DNA profiling has emerged as a non-invasive approach to monitor disease and longitudinally characterize tumor evolution. We describe the use of circulating tumor DNA (ctDNA) among patients (pts) with FGFR2-altered CCA receiving FGFR-targeted therapy in the identification of acquired FGFR2 mutations (mut) at resistance. Methods: Serial blood samples were collected from 8 pts with FGFR-altered CCA for ctDNA isolation and next generation sequencing. Plasma ctDNA collected at baseline and resistance to FGFR-targeted therapy were sequenced using a custom ultra-deep coverage cfDNA panel, MSK-ACCESS, incorporating dual index primers and unique molecular barcodes to enable background error suppression and high-sensitivity mut detection. The assay was enhanced to include all protein-coding exons and relevant introns of FGFR2. In 5/8 pts, genomic profiling of an initial tumor biopsy was performed. Results: 8 pts with FGFR2-altered CCA (7 gene fusions, 1 amplification) were treated with FGFR-targeted therapies. 7/8 pts exhibited stable disease or partial response. 19 total acquired mut in FGFR2 were detected at resistance in 5/8 pts (between 1-9 unique mut identified in each sample). All mut were located in the kinase domain. Conclusions: Acquired mut in FGFR2 are seen in pts who have developed resistance to targeted therapy. CtDNA can be used to identify these mut at the time of acquired resistance. The multitude of FGFR2 mut observed within individual pts suggest heterogeneity and evolutionary convergence of resistance mechanisms. Our results illustrate the utility of ctDNA as a less invasive way to monitor for signs of resistance and to identify other potential targetable alterations. [Table: see text]

Duplex Sequencing Accurately Detects Variants below 1/100,000 in Genes Recurrently Mutated in Acute Myeloid Leukemia

Acute myeloid leukemia (AML) is a challenging disease to treat: most patients achieve remission after induction chemotherapy, but the majority eventually relapse. Minimal residual disease (MRD) after initial treatment is the best predictor of relapse and is thus a critical metric around which to develop new treatments. However, conventional MRD diagnostics, including cytology and flow cytometry, are of variable sensitivity and often only perform well in specialty centers, and there is no gold standard. Molecular tests developed to measure trace MRD in other hematological malignancies (i.e. CML) are high-resolution but assay a single, universally-present mutation, while many different genetic drivers exist in AML and these are spread among dozens of genes. Virtually every AML patient harbors a unique combination of mutations, making it difficult to design an effective universal assay. As such, most reported molecular AML MRD assays are either sensitive for mutations that are only found in a narrow subset of patients, or can screen many potential sites of mutation, but with low sensitivity.Here we present a broadly applicable Duplex Sequencing-based AML MRD assay that can readily detect mutant allele frequencies (MAF) below 1/10,000 across a large panel of genes, and below 1/100,000 in a focused panel. Conventional next generation sequencing (NGS) introduces errors during amplification and sequencing, creating a background of artifactual noise that obscures true mutations present below ~1%. Duplex Sequencing improves accuracy >100,000-fold through a molecular tagging approach whereby both strands of each original DNA duplex are ligated with a unique molecular barcode and amplified such that the reads generated from each strand can be related back to their unique original duplex. Reads can also be distinguished from those of their mate strand, thus the two strands of each DNA duplex can be compared and any discrepant nucleotide positions are discounted as errors.Our complete AML panel targets 151 exons or hotspot codons in 29 genes with a 59 kilobase (kb) hybrid-capture footprint. This region comprises loci containing single-nucleotide and short indel mutations found in approximately 90% of adult AMLs. A mean Duplex error-corrected sequence depth of 10,837 and a maximum Duplex depth of 14,967 was obtained across these targets from a single library preparation using 250 ng of sheared leukocyte DNA (Fig. 1). Duplex depth can be readily increased by preparing additional Duplex libraries from the same source DNA to achieve proportionally higher sensitivity for rarer variants. This stands in contrast to conventional NGS where, beyond a modest level, an increase in depth simply increases the number of background errors identified (Fig. 2A).We simulated low-level residual disease by mixing control DNA from a healthy young blood donor with DNA from 9 human cell lines harboring known AML mutations at dilutions from 1:100 to 1:100,000 (Table 2). The genomic loci of these 9 mutations in NRAS, KRAS and TP53 were captured with a small 1 kb probe panel. This mixture was sequenced to a mean Duplex depth of >1,000,000-fold, with the highest and lowest MAFs shown in Fig. 2B. All were close to expected frequencies (r2=0.96) with MAF as low as 6x10-6 (Fig. 3). As proof of specificity, we examined all coding nucleotide positions (excluding the 9 expected variants) and identified only 241 background variant counts out of 414,452,402 total Duplex BP, for an aggregate mutation frequency of 5.8x10-7, consistent with the estimated background of normal human aging.Our Duplex Sequencing-based AML MRD assay is flexible, broadly applicable and extremely sensitive. The assay is easily implemented using standard NGS equipment and automated cloud-based analysis software. The ~90% of AML patients served by this SNV-focused panel can be expanded to nearly 100% with complementary indel detection via targeted NGS RT-PCR. Optionally, when a patient's mutation profile from time-of-diagnosis is known, MRD testing can focus exclusively on those targets using a subset of pre-validated probes to reduce sequencing cost. Improved MRD testing will facilitate accurate prognostication, better selection among treatment options, and could serve as a surrogate endpoint in clinical trials to bring new treatments to patients faster. We are currently evaluating Duplex Sequencing MRD tests in both retrospective and prospective clinical trials. [Display omitted] DisclosuresHiggins:TwinStrand Biosciences: Employment. Williams:TwinStrand Biosciences: Employment. Buckley:CTI Biopharma: Employment; TwinStrand Biosciences: Consultancy. Radich:TwinStrand Biosciences: Research Funding. Salk:TwinStrand Biosciences: Employment, Equity Ownership.

Identifying shared patterns in the T cell receptor repertoire specific to IE-1 CMV

Introduction Cytomegalovirus (CMV) continues to be a major problem for organ transplant recipients. T cells specific to the CMV antigens immediate-early (IE-1) and pp65 CMV govern the adaptive immune response to CMV. Molecular studies indicate that CMV-specific responses usually comprise clonally expanded viral-specific T cell receptors (TCR) that can be conserved among subjects with a similar HLA haplotype. However, our understanding of the CMV-specific T cell receptor (TCR) repertoire is incomplete and the characteristics of TCR repertoire conferring protection to CMV are not known. The aim of this study is to identify specificity groups in the IE-1-specific TCR repertoire of CMV seropositive A02:01 subjects.Methods: PBMC from three CMV seropositive, IE-1-ELISPOT+ A02:01 healthy subjects were stimulated with IE-1 peptides. Antigen-reactive T cells were FACS-sorted based on the expression of CD137 for CD8+ T cells and CD154 for CD4+ T cells, or the loss of CFSE expression in T cells that had been prelabeled with CFSE prior to stimulation. We sequenced the IE-1 specific TCRβ chains following TCRb CDR3 segment amplification using a PCR template switch methodology that minimizes bias through the use of unique molecular barcodes. IE-1 TCRb sequences were analyzed using the recently published Grouping of Lymphocyte Interactions Paratope Hotspots (GLIPH) algorithm. Results We obtained a total of 2003 TCRβ sequences from IE-1-specific CD8+ T cells from three CMV seropositive subjects. The IE-1-specific TCR repertoire analysis revealed clonally expanded populations for two of the three subjects, with ~50% of the repertoire comprised of 1-10 dominant clones. In contrast, the corresponding non-IE-1 specific T cells had highly diverse TCR repertoires and a minimal number of expanded clones (<1%). Analysis of all CD8+ IE-1-specific TCRβ sequences by GLIPH identified multiple possible clusters. The best scoring identity cluster grouped 95 TCR sequences across two subjects within either the CD137+, or the CFSE-populations. To note, cluster 1 motiff was found in a published IE-1 TCR sequence in the literature, binding to an HLA A2:01 restricted epitope (VLEETSVML).Cluster 2 grouped 43 TCR sequences across two subjects within either the CD137+, or the CFSE-, populations. Importantly, 58 TCR sequences were found to cluster across the IE-1-specific CD8+ T cell populations in all three donors. Here we demonstrate that is feasible to characterize the CMV-specific TCR repertoire in an antigen-specific manner. This methodology has the potential to cluster TCR sequences in transplant patients controlling CMV replication.

Abstract 5574: A comprehensive, targeted next-generation sequencing method that rapidly and accurately detects circulating tumor DNA variants at 0.1% frequency in plasma samples

Abstract Introduction: Mutation analysis of circulating tumor DNA (ctDNA) in blood-based liquid biopsies provides a minimally invasive approach to detect and monitor disease. Existing next-generation sequencing (NGS) liquid biopsy techniques have laborious and/or inefficient workflows, heuristic error-correction algorithms, and variable performance with clinical tumor-plasma samples. We describe a method that combines a kittable and efficient wet-bench workflow with accurate dry-bench analytics to reduce costs and turnaround time, and is relevant to clinical research and patient testing. Methods: We developed a comprehensive, targeted NGS technology to enable the ultra-sensitive detection of variants from liquid biopsy samples. Input DNA molecules from Seraseq™ ctDNA v2 reference materials (SeraCare) and &amp;gt;50 matched FFPE and plasma samples were uniquely tagged with a random molecular barcode (MBC), amplified in an efficient PCR protocol, and sequenced using a targeted panel covering &amp;gt;500 somatic mutation hotspots. Sequence-ready libraries were prepared from input DNA within 9 hours. Data were analyzed with a robust bioinformatics pipeline tailored to the library chemistry to correct for multiple background errors. Variants were identified with a site-specific model, which effectively eliminated recurring non-biological aberrations that remained despite MBC-facilitated error-suppression. We verified variant calls in plasma by Droplet Digital™ PCR (Bio-Rad). To evaluate the limit of detection, healthy control and mutation-positive plasma admixtures were prepared and sequenced. Results: Input template molecules were efficiently recovered. The median number of unique MBCs across all amplicons was &amp;gt;90% of expectation based on input DNA quantities for both the tumor-associated plasma and Seraseq™ ctDNA v2 material. SNVs and indels were recognized with &amp;gt;90% sensitivity and PPV at 0.5% allele frequency (AF) in the Seraseq™ ctDNA mutation mixes. In both neat plasma samples and plasma admixtures, we correctly identified tumor mutations down to ~0.1% AF while maintaining a low false-positive rate (sensitivity and PPV remained high for AF ≥0.1%). Reference ctDNA material and tumor-associated plasma had analogous template amplification and variant frequency rates. Conclusions: A fast, efficient, and sensitive NGS panel approach was developed and evaluated, and it demonstrated the reliable and specific detection of rare variants in liquid biopsy specimens. The method is able to distinguish a low frequency ctDNA signal from the overwhelming background noise in plasma cell-free DNA using a streamlined workflow and purpose-built bioinformatics pipeline. This technology may provide an easy-to-use, high-performance, and adaptable NGS diagnostic framework for disease detection and therapeutic intervention monitoring. Citation Format: Jessica L. Larson, Liangjing Chen, Lando Ringel, Blake Printy, Farol L. Tomson, Yves Konigshofer, Sarah Statt, Joseph K. Kaplan, Shobha Gokul, Jeffrey Shelton, Gary J. Latham, Brian C. Haynes. A comprehensive, targeted next-generation sequencing method that rapidly and accurately detects circulating tumor DNA variants at 0.1% frequency in plasma samples [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 5574.

In vivo insertion pool sequencing identifies virulence factors in a complex fungal-host interaction.

Large-scale insertional mutagenesis screens can be powerful genome-wide tools if they are streamlined with efficient downstream analysis, which is a serious bottleneck in complex biological systems. A major impediment to the success of next-generation sequencing (NGS)-based screens for virulence factors is that the genetic material of pathogens is often underrepresented within the eukaryotic host, making detection extremely challenging. We therefore established insertion Pool-Sequencing (iPool-Seq) on maize infected with the biotrophic fungus U. maydis. iPool-Seq features tagmentation, unique molecular barcodes, and affinity purification of pathogen insertion mutant DNA from in vivo-infected tissues. In a proof of concept using iPool-Seq, we identified 28 virulence factors, including 23 that were previously uncharacterized, from an initial pool of 195 candidate effector mutants. Because of its sensitivity and quantitative nature, iPool-Seq can be applied to any insertional mutagenesis library and is especially suitable for genetically complex setups like pooled infections of eukaryotic hosts.

Abstract 5695: Increased sensitivity for detection of oncogenic mutations present in circulating cell-free tumor DNA

Abstract Molecular pathways involved in tumor survival and progression are often activated by genetic alterations, and anticancer agents specifically targeting many oncogenic pathways have entered clinical trials. Precision cancer medicine takes advantage of genomic technologies (e.g., massively parallel sequencing, MPS) to perform robust tumor genomic profiling in the clinic, enabling identification of “druggable” mutations. To date, the vast majority of genomic profiling has been performed on DNA extracted from solid tumor biopsies. However, liquid biopsies offer advantages over solid tumor biopsies; chief among them are their non-invasive nature and quick turn-around time when assessed using MPS. Furthermore, liquid biopsies allow for sequential analysis throughout a course of therapy permitting real-time monitoring of treatment effects on tumor expression and subsequent surveillance for development of new oncogenic alterations. Circulating cell-free tumor DNA (cfDNA) derived from plasma represents a convenient assay material that has been effectively analyzed for oncogenic mutations in specific forms of cancer such as non-small cell lung cancer. However, introduction cfDNA for clinical genotyping is limited by sensitivity requirements. cfDNA is typically present at 20-100 ng/mL of plasma, of which the cancer fraction can be below 1%. Error rates associated with PCR amplification and base calling (Illumina HiSeq platform) are approximately 2%, preventing confidence in rare variants with allele fractions (AF) at or below that percentage. We conducted an evaluation of commercially available kits to determine 1) kit sensitivity and if necessary 2) modify the protocol to achieve sensitivity of &amp;lt;= 1% AF. To evaluate kit sensitivity gDNA from a translocation positive cell line was mixed at 5% and 20% with normal cell line gDNA. Libraries were prepared for each percent mix using 10 ng of input. Raw mean target coverage (MTC) was similar for evaluated kits (MTC, 250-300X); Percent Duplication, ~34%. Comparison of 20% and 100% translocation samples showed similar levels of unique single nucleotide variant (SNV) detection. However, 5% mixtures showed a ~10 fold reduction in SNV detection compared to both 20% and 100% translocation samples. Furthermore, translocation rearrangement was only detectable in one third of the 5% translocation samples. Increased sensitivity requires unique molecular barcodes coupled with refinements in bioinformatics analysis for clinical implementation of these assays. Citation Format: Aaron R. Thorner, Michael K. Slevin, Andrea Clapp, Haley A. Coleman, Suzanne R. McShane, Samantha D. Drinan, Edwin Thai, Bruce M. Wollison, Robert T. Burns, Alex Frieden, Matthew D. Ducar, Anwesha Nag, Laura E. MacConaill, Matthew L. Meyerson. Increased sensitivity for detection of oncogenic mutations present in circulating cell-free tumor DNA [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 5695. doi:10.1158/1538-7445.AM2017-5695

CRISPR\u2013Cas9-targeted fragmentation and selective sequencing enable massively parallel microsatellite analysis

Microsatellites are multi-allelic and composed of short tandem repeats (STRs) with individual motifs composed of mononucleotides, dinucleotides or higher including hexamers. Next-generation sequencing approaches and other STR assays rely on a limited number of PCR amplicons, typically in the tens. Here, we demonstrate STR-Seq, a next-generation sequencing technology that analyses over 2,000 STRs in parallel, and provides the accurate genotyping of microsatellites. STR-Seq employs in vitro CRISPR–Cas9-targeted fragmentation to produce specific DNA molecules covering the complete microsatellite sequence. Amplification-free library preparation provides single molecule sequences without unique molecular barcodes. STR-selective primers enable massively parallel, targeted sequencing of large STR sets. Overall, STR-Seq has higher throughput, improved accuracy and provides a greater number of informative haplotypes compared with other microsatellite analysis approaches. With these new features, STR-Seq can identify a 0.1% minor genome fraction in a DNA mixture composed of different, unrelated samples.

Abstract IA06: Genomic approaches to drug discovery

Abstract The principles of genomics (systematic, high throughput approaches to biology) can increasingly be brought to bear on drug discovery. Our group is developing two new approaches: Connectivity Map and PRISM. The concept of the Connectivity Map (CMap) is that disease states, cellular states and the action of pharmacologic or genetic perturbation can be captured by gene expression profiling. In order to make large-scale perturbational profiling feasible, we developed a high-throughput, ultra-low cost profiling method (named L1000) whereby the abundance of 1000 transcripts are quantified by hybridization to Luminex microspheres. As part of the NIH LINCS program, we have now generated over 2 million such L1000 profiles representing knock-down and over-expression of over 3000 genes and perturbation with over 4000 small molecules in more than 15 cell types. Examples of discoveries emanating from this next generation Connectivity Map will be presented. The PRISM method takes inspiration from recent large-scale drug sensitivity profiling studies demonstrating the value of determining the pattern of compound killing across a large panel of genetically characterized cancer cell lines. In order to increase the scale at which such studies can be performed, we inserted unique molecular barcodes (24-mer oligonucleotides) into the genomes of cancer cell lines, such that the lines could then be pooled and the pooled lines exposed to a compound of interest. Drug sensitivity is then quantified by changes in abundance of the molecular barcodes. We have demonstrated feasibility of pooling more than 500 cancer cell lines in a single pool, and have demonstrated feasibility of high throughput screening using the PRISM method. These experiments suggest that PRISM can be used as a starting point for oncology drug discovery and for the characterization of lead compounds for the genotype-specific patterns of cancer cell killing. Examples of recent PRISM discoveries will be presented. Citation Format: Todd R. Golub. Genomic approaches to drug discovery. [abstract]. In: Proceedings of the AACR Precision Medicine Series: Targeting the Vulnerabilities of Cancer; May 16-19, 2016; Miami, FL. Philadelphia (PA): AACR; Clin Cancer Res 2017;23(1_Suppl):Abstract nr IA06.

Rare Pre-Existing MDS Subclones Contribute to Secondary AML Progression

Background: MDS is composed of a dominant founding clone and typically one or more subclones that descend from the founding clone. Using whole genome sequencing (WGS) of secondary AML patients who were initially diagnosed with MDS, we previously showed that AML progression always coincided with expansion of at least one subclone. However, this emergent subclone was not always detectable in the antecedent MDS samples using high-coverage sequencing. It is currently unclear whether these emerging secondary AML-specific clones represent new mutations acquired in cells during treatment or selection of a rare subclone that was below the level of detection in the diagnostic MDS sample. Identification of such pre-existing subclones at MDS diagnosis could provide important prognostic information. Using an ultra-sensitive, error-corrected sequencing approach coupled with high coverage depths, we sought to determine whether secondary AML-specific mutations could be detected in antecedent MDS samples.Methods: We studied 4 MDS patients who eventually progressed to secondary AML (mean 601 days to transformation; range 131-955) that had DNA from initial MDS banking, secondary AML, and skin (as a normal control). DNA from the secondary AML sample was whole genome (3 cases) or exome sequenced (1 case). Custom enrichment panels were designed to target somatic mutations identified in the secondary AML samples and used to sequence the prior MDS samples. All 4 MDS cases had at least one subclone present in the secondary AML sample that was not detected by standard panel-based re-sequencing of the initial MDS sample (539x average coverage). We designed custom molecular barcode-based reagents to perform high-sensitivity, error-corrected sequencing targeting mutations that were detected only at progression to secondary AML. These mutation sites were then sequenced to ultra-deep coverage in prior MDS samples. To determine the sensitivity and specificity of the assay, we first sequenced known dilutions of cell line DNA and showed a sensitivity of 75% to detect variant allele fractions (VAFs) of 0.06% (one mutant cell in 800 cells) while maintaining a specificity of >99.9%. Only reads with unique molecular barcodes that were seen at least three times and had >90% identical reads within a read family were considered. Samples were sequenced to an average total coverage depth of 187,000x (range 107k-474k fold) corresponding to 10,398 unique molecular barcodes of which 3,549 passed filtering (at least two unique read families were required to call a variant), resulting in a mean predicted maximum VAF sensitivity of 0.05% (one mutant cell in 1,000 cells).Results: Using error-corrected barcode sequencing all 4 MDS cases had a subset of mutations detected that were originally only detected in the secondary AML; the mean VAF of detected mutations was 0.25% (range 0.02% to 3.9%) in the antecedent MDS samples, corresponding to 1 mutant cell in 200 cells. However, not all of the ’secondary AML-specific’ mutations were detected in MDS samples. On average, only 17% (range 10-39%) of the ’secondary AML-specific’ subclone mutations were detected in each MDS sample. This suggests that while some mutations pre-exist in an ancestral subclone in the MDS sample, others are acquired during the intervening months to years between MDS and progression to secondary AML. As an example, in two patients who were given decitabine prior to secondary AML progression, mutations detected in the secondary AML subclone that were not detected in the antecedent MDS sample using error-corrected barcode sequencing were more frequently C->G transversions compared to commonly acquired C->T transitions seen in MDS samples (p=.005; Mantel-Haenszel test). Increased C->G transversions are associated with DNA damage induced by decitabine, suggesting that in these two patients an ancestral subclone pre-existed in MDS that acquired additional decitabine-induced mutations during treatment (i.e., C->G transversions).Conclusion: Collectively, the data indicate that very rare subclones pre-exist at MDS diagnosis. These pre-existing subclones can subsequently give rise to the dominant clone at progression to secondary AML. Additional mutations -- some induced by chemotherapy treatment --- are also acquired in ancestral subclones that pre-exist in MDS samples. The combination of subclonal mutations may ultimately contribute to disease progression. DisclosuresJacoby:Sunesis: Research Funding; Quintiles: Consultancy; Celgene: Speakers Bureau.

Whole-Exome Sequencing and Targeted Deep Sequencing of cfDNA Enables a Comprehensive Mutational Profiling of Multiple Myeloma

Background. Cell-free DNA (cfDNA) sequencing enables serial temporal sampling, which offers the possibility of following the dynamics of biomarkers and clonal evolution in Multiple Myeloma (MM) over time. The use of cfDNA in clinical practice as a molecular biomarker and for monitoring response/resistance is dependent on a comprehensive profile of matched cfDNA and tumor DNA (tDNA) samples. Here we performed Ultra-Low Pass Whole Genome Sequencing (ULP-WGS) followed by whole-exome sequencing (WES) and targeted deep sequencing of matched cfDNA/tDNA samples from MM patients.Methods. We performed next generation sequencing of matched cfDNA/tDNA samples for 63 patients with newly diagnosed or relapsed MM, SMM, or MGUS. Libraries were constructed using the Kappa Hyper kit and sequenced by ultra-low-pass whole-genome sequencing (ULP-WGS, 0.1x coverage) to quantify tumor fraction within cfDNA. WES was performed on 30 matched samples cfDNA/tDNA/germline DNA from 10 patients with more than 5% of tumor fraction. Libraries were hybridized to the Nextera Rapid Capture Exome kit (Illumina) and then sequenced on HiSeq 4000 (Illumina). Targeted deep sequencing was performed on 32 matched cfDNA/tDNA samples from 16 patients using the HaloPlex HS technology (Agilent), allowing for molecular barcoding. Libraries were constructed according to the manufacturer's instructions and sequenced on HiSeq 2500 (Illumina). Sequencing data were analyzed using the Firehose pipelines, including MuTect, ABSOLUTE, ReCapSeg, GISTIC and MutSig.Results. We first used a cost-effective approach to establish the tumor content of cfDNA in a large-scale manner by ULP-WGS. Among 63 tested samples (53 MM, 6 SMM and 4 MGUS patient samples), the tumor fraction within cfDNA ranged from 0 to 81% with a mean of 10%. About 43% of these samples had tumor fraction greater than 5% within cfDNA. To assess whether cfDNA can capture the genetic diversity of MM and inform clinical management, we performed WES of matched cfDNA/tDNA/germline DNA samples for 10 patients (mean target coverage 194x). Copy number alterations (CNAs) assessed by WES (ReCapSeg) were consistent between cfDNA and tumor DNA. Similarly, focal CNAs assessed by GISTIC were consistent between tDNA and cfDNA. We then examined the overlap of somatic single nucleotide variants (SSNVs) between WES of cfDNA and matched tDNA. We found, on average, 100% of the clonal and 96% of the subclonal (range 54-100%) SSNVs that were detected in the tumor were confirmed to be present in cfDNA. Similarly, for mutations detected in the cfDNA, we found, on average, 100% of the clonal and 99% of the subclonal (range 98-100%) SSNVs were confirmed in the tumor. To assess whether targeted deep sequencing of cfDNA could be a good proxy for tumor biopsy we used a targeted deep sequencing approach of known MM driver genes. Libraries were prepared using unique molecular barcodes to avoid duplication rates, for 32 matched cfDNA/tDNA samples from 16 patients with MM. The mean target coverage was 596x. We found similar frequencies of altered MM driver genes in both cfDNA and tDNA, including KRAS, NRAS, and TP53, indicating that cfDNA can be used for precision medicine.Conclusions. Our study demonstrates that both WES and targeted deep sequencing of cfDNA are consistently representative of tumor DNA alterations in terms of CNAs, focal CNAs and SSNVs. This approach could therefore be used to longitudinally follow clonal evolution across the course of the disease and precision medicine in patients with MM. DisclosuresPalumbo:Takeda: Employment, Honoraria; Janssen Cilag: Honoraria. Kumar:Noxxon Pharma: Consultancy, Research Funding; Celgene: Consultancy, Research Funding; Millennium: Consultancy, Research Funding; Skyline: Honoraria, Membership on an entity's Board of Directors or advisory committees; Janssen: Consultancy, Research Funding; Kesios: Consultancy; Glycomimetics: Consultancy; BMS: Consultancy; Array BioPharma: Consultancy, Research Funding; Sanofi: Consultancy, Research Funding; AbbVie: Research Funding; Onyx: Consultancy, Research Funding. Roccaro:Takeda Pharmaceutical Company Limited: Honoraria. Facon:Amgen: Consultancy, Speakers Bureau; Novartis: Consultancy; Janssen: Consultancy, Speakers Bureau; Bristol: Consultancy; Millenium/Takeda: Consultancy; Celgene: Consultancy, Speakers Bureau; Karyopharm: Consultancy. Ghobrial:Celgene: Honoraria, Research Funding; BMS: Honoraria, Research Funding; Noxxon: Honoraria; Novartis: Honoraria; Takeda: Honoraria; Amgen: Honoraria.

Routine Clinical Testing for Actionable Mutations in CLL Using Endcll Assay V1: The Initial MD Anderson Experience

Background:Recent large scale genomic profiling and clinical correlation studies of chronic lymphocytic leukemia (CLL) identified clinically significant gene mutations with important clinical prognostic and potential therapeutic implications. Integration of individual clonal and subclonal gene mutation information with existing prognostic markers such as somatic hypermutation and chromosomal aberrations provide comprehensive risk stratification for CLL patients. In addition, mutations in BTK and PLCG2 are associated with resistance to targeted therapies using BTK inhibitors (ibrutinib). We implemented routine clinical testing for CLL patients. We describe our initial experience with EndCLL Assay V1, a high yield targeted NGS assay for routine clinical testing of clinically significant mutations in CLL and B-cell neoplasms.Materials and Methods:DNA from total of 376 blood or bone marrow samples containing >10% CLL cells were tested during the panel validation and subsequent routine clinical testing. Targeted sequencing was performed on 29 genes reported to be mutated in CLL and other B-cell neoplasms. These include ATM, BIRC3, BTK, CALR, CARD11, CD79A, CD79B, CHD2, CSMD3, CXCR4, DDX3X, EZH2, FAT1, FBXW7, KLHL6, LRP1B, MAPK1, MUC2, MYD88, NOTCH1, PLCG2, PLEKHG5, POT1, SF3B1, SPEN, TGM7, TP53, XPO1 and ZMYM3. Panel design included a combination of hotspots, limited exons and all exon coverage as required to detect known clinically significant mutations in the genes targeted in the panel. A total of 500 ng input DNA was used to prepare sequencing libraries using Agilent Haloplex HS chemistry, which incorporates unique molecular barcodes in each DNA molecule being sequenced. Sequencing was performed on MiSeq sequencers (Illumina) followed by data analysis on Agilent SureCall v3.0. The average per base coverage depth ranged from 1500x to 4000x based on the number of samples per run. The analytical sensitivity of the assay was determined to be at 2% to 5% for reliable and reproducible detection of mutations depending on the sequencing coverage and percentage of mutant reads.Results:For cases with available information, the average age was 63.3 years (range: 22-87, median: 64). The male:female ratio was 2:1 and the ratio of cases with unmutated:mutated IGHV was 1.2:1. Positive FISH results for -11q, +12, -13q and -17p were detected in 18%, 23%, 59% and 15% cases respectively. Out of 376 samples tested, 249 (66.2%) showed at least one mutation. Average number of genes mutated per sample was 1.1 (range: 0-5, median: 1) and the average number of total mutations per sample was 1.4 (range 0-7, median: 1). Top 10 mutated genes included TP53, SF3B1, NOTCH1, ATM, BIRC3, XPO1, MYD88, SPEN, BTK and KLHL6 (Figure 1). Patterns of co-mutations and mutual exclusivity were observed (Figure 1). Top 10 mutations (hotspots) are listed in the table. Mutations in 3’UTR of NOTCH1 were detected in 5/376 (%) cases. Resistance mutations in BTK codon 481 were detected in 10 patients on ibrutinib treatment, including 5 cases with >1 mutations and 2 cases with subclonal mutations indicating early treatment resistant clone. Clinical findings in all 10 cases were consistent with proven or emerging treatment resistance. No mutations were detected in PLCG2 (codons 646-685). One patient initially on ibrutinib and on idelalisib treatment showed acquisition of 17p deletion and a subclonal TP53 mutation at the time of ibrutinib resistance and progression to an aggressive disease course. Use of molecular barcodes allowed efficient detection of low level subclonal mutations. Correlations with a comprehensive set of clinicopathologic parameters as well as treatment outcomes are in progress.Conclusion:The targeted EndCLL Mutation Assay V1 allowed a high yield detection of clinically significant mutations in 66% of the patient samples using a clinically sustainable panel design accommodating up to 24 multiplexed samples in a single sequencing run. Use of molecular barcodes allows removal of PCR duplicates and enables accurate calling of subclonal mutations for prognostication. This was evident in simultaneous detection of subclonal BTK mutations allows early detection of emerging ibrutinib resistant clone while obtaining additional prognostic information. Overall, the EndCLL Assay V1 is a valuable addition for CLL patient care. [Display omitted] [Display omitted] DisclosuresJain:Incyte: Research Funding; BMS: Research Funding; ADC Therapeutics: Consultancy, Honoraria, Research Funding; Celgene: Research Funding; Novimmune: Consultancy, Honoraria; Pfizer: Consultancy, Honoraria, Research Funding; Seattle Genetics: Research Funding; Abbvie: Research Funding; Infinity: Research Funding; Novartis: Consultancy, Honoraria; Genentech: Research Funding; Pharmacyclics: Consultancy, Honoraria, Research Funding; Servier: Consultancy, Honoraria. Wierda:Gilead: Research Funding; Genentech: Research Funding; Acerta: Research Funding; Novartis: Research Funding; Abbvie: Research Funding.

Clono-Specific Evaluation of Minimal Residual Disease in Acute Myeloid Leukemia

Background:The genetic landscape of adult acute myeloid leukemias (AML) has been recently unravelled. This makes achievable the determination of a comprehensive profile of driver lesions for virtually all patients at diagnosis. Recent studies using multi-target minimal residual disease (MRD) strategies with around 1% sensitivity indicate that the clearance of all molecular events after chemotherapy is associated with better survival. To improve the clono-specificity and the sensitivity of this approach, after a precise determination of AML clonal composition, we combined cytogenetic, FISH, and high sensitivity deep sequencing technologies to monitor the MRD in a series of 69 patients.Methods:Forty-five consecutive patients reflecting the genetic diversity of AML were prospectively included and 24 patients were retrospectively studied. All patients received an anthracycline + cytarabine based regimen. The clonal architecture was established at diagnosis based on NGS-targeted resequencing (122 gene panel) and cytogenetic data. Lesions were next investigated in complete remission (CR). Based on the initial clonal composition, targeted resequencing panels were designed to improve the sensitivity by the use of unique molecular barcodes (Haloplex High Sensitivity, or HS-NGS assay). Cytogenetic events were evaluated by FISH.Results:In the 69 patients, a median of 4 genetic or chromosomal events were identified per patient (range 0-10). One patient had no evaluable target allowing MRD evaluation in 68/69 patients.To determine the threshold of detection of the HS NGS assay, we analyzed the frequency of mutant reads in multiple samples expected to be wild type for 31 given SNVs and 2 indels. A consensus threshold of detection was set at a variant allele frequency (VAF) of 0.2% for all lesions.In CR samples, early initiating events frequently persisted after treatment, especially mutations in DNMT3A, TET2, ASXL1, EZH2, IDH1, TP53, SRSF2, and U2AF1. Mutations in FLT3, NRAS, KIT, NPM1, CEBPA, WT1, IDH2 and BCOR were the most frequently cleared events.Seven patients did not reach CR after one course, and two had no available material after one course. In the 59 remaining patients, we tested whether the global response level of all targets was associated with prognosis. We used the median VAF of the first events of all clonal architectures to separate good responder from poor responders (i.e. VAF = 1.66%). At 2 years, there was a trend to lower leukemia free survival (LFS) probability in poor responders (31.7+/-9.9% vs 51.7+/-9.8%, p=0.08) with no translation in overall survival (OS).We next investigated if the persistence of two or more detectable markers was associated with prognosis. The 58 patients with more than one evaluable event were consequently separated in two groups: patients with 0 or 1 marker above the detection threshold after treatment (n=31), and patients with 2 or more detectable lesions (n=27). At 2 years, DFS was 64.9+/-9.3 % in patients with 0 or 1 detectable marker vs 19.8+/-8.7% in patients with 2 or more detectable markers (p=0.001). OS probability was higher in patients with 0 or 1 detectable marker 84+/-6.6% vs 57.1+/-10.5% (p=0.023). When focusing on the 40 patients with intermediate cytogenetics, persistence of 2 or more markers was associated with lower LFS (57+/-11.8% vs 19.4+/-10.5 p=0.0048) and with a trend to lower OS (85+/-8% vs 61+/-11.9% p=0.07). Similar results were observed when restricting the analyses to the 42 prospectively included patients (At 2 years: LFS 73+/-10% vs 24+/-10%, p=0.0026 and OS 90.2+/-6.6% vs 62.8+/-11.5%, p=0.036).In 50 patients with 3 or more identified events, the persistence of 3 or more markers after one course was associated with a very high risk of relapse (DFS 23.5+/-10.3 % vs 75.8 +/-7.5% at one year, p<0.0001; median DFS at 7 months in the non-responder group and not reached after two years in the responder group), and lower OS probability (84.8+/-6.2 vs 45.2+/-13.5% at 2 years, p=0.026)ConclusionOur study shows the high prognostic value of a personalized multi-target clono-specific MRD evaluation that can be used in nearly all AML patients. Detection of two or more events in more than 0.4% cells after one course is associated with lower survival, in particular in intermediate cytogenetic patients. Larger studies are needed to confirm the results and to evaluate if this strategy could be useful to guide treatment decisions. [Display omitted] DisclosuresNo relevant conflicts of interest to declare.