Search

Sensitive and specific detection of low frequency mutations in acute myeloid leukemia (AML) is critical in research of minimal residual disease (MRD). At variant allele frequencies (VAF) below approximately 1%, PCR and sequencing errors result in prohibitive signal-to-noise ratios with next generation sequencing (NGS). Duplex Sequencing (DS) relates the original top and bottom DNA strands to make double stranded consensus sequences to greatly reduce errors. In two experiments we show further improvement of DS of AML-related genes by the use of enzymatic fragmentation (EF) and an updated gene panel that incorporates 2022 European LeukemiaNet (ELN) recommendations. In the first DS study, 29 AML-related genes (59 kb) were targeted in hybrid capture. Mutant cell line DNA was mixed into DNA from a healthy young donor to simulate MRD. Expected VAFs in mixtures were 1.0-0.003%. Samples comprised a mix of 4 insertions and deletions (indel), a mix of 15 single nucleotide variants (SNV), and 4 serial dilutions of a FLT3 ITD plus an NPM1 insertion. Pure diluent DNA was used as a negative control. Each sample was prepared in quadruplicate with mechanical fragmentation (MF) vs EF. DNA input mass was 1,500 ng per replicate, except for 50-250 ng for the FLT3/NPM1 mixes with expected VAF of 1% and 0.1%, respectively. Input masses were set to ensure >95% probability of detection of all mutations in the combined data. All targeted variants down to 0.003% VAF were detected in spike-in mixtures, with expected vs observed VAF highly correlated whether using MF or EF (R2 >0.98 for all mixtures). Duplex molecular depth was 1.2-2.0x higher with EF vs MF across input masses. Panel-wide mean duplex depth per 1,500 ng replicate was 30,376x for MF and 48,036x for EF, for combined mean depths of 121,508x for MF and 192,144x for EF. In the negative control, background mutational calls at spike-in positions were 4/2,993,429 duplex bases using MF and 0/4,780,491 duplex bases with EF, reflecting increased specificity with EF. Next, a revised panel targeting 36 AML-related genes (80 kb) was used for DS with EF. Here differing mutant cell line DNA was mixed into the same diluent from above (the negative control). This mixture harbors 27 variants with expected VAFs of 0.125-0.006% (4 indels, 21 SNVs, a FLT3 ITD and an NPM1 insertion), 17 of which (2 indels, 14 SNVs and the NPM1 insertion) span VAFs of 0.011-0.009% to establish a limit of detection of 0.01%. With 2,000 ng (+/- 10%) input, for 14 samples, panel-wide mean duplex depth was 56,528x. For variants with expected VAF ≥ 0.009%, the assay had >98% sensitivity, >96% specificity, and >98% accuracy at genomic positions of known true positive variants. In summary, DS with EF yields more data per nanogram of DNA than MF, maximizing the use of precious samples. This method exhibits extremely low background mutation signal and a low limit of detection across targets that are valuable in AML MRD research. Citation Format: Elizabeth Schmidt, Devon M. Fitzgerald, Camila Zanette, Gavin D. Meredith, Mark A. McElwain, Raul Burciaga, Kevin C. Vavra, Thomas H. Smith, Jesse J. Salk, Jake Higgins. Accurate detection of low frequency AML-associated mutations in vitro using Duplex Sequencing with enzymatic fragmentation [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2023; Part 2 (Clinical Trials and Late-Breaking Research); 2023 Apr 14-19; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2023;83(8_Suppl):Abstract nr LB254.

Clonal hematopoiesis (CH aka CHIP) is the process whereby otherwise healthy individuals accumulate clones bearing low-frequency somatic mutations in hematologic malignancy (HM) driver genes. Historically thought to be a phenomenon of the elderly that is associated with increased risk of HM and cardiovascular disease, CH has been observed in increasingly younger cohorts as sequencing technology has advanced. CH in hematopoietic stem cell transplant (HSCT) donors has also been associated with adverse recipient outcomes. In this retrospective analysis, we asked whether CH mutations preferentially expanded in recipients vs. donors in paired samples collected a median of 34 years post-HSCT (range 6-45). 16 donor-recipient pairs were analyzed, 11 of which had a donor sample collected pre-HSCT. All individuals were healthy at the time of contemporary sampling. Median donor age was 26 years (range 12-84) at time of HSCT. We took advantage of the ultra-low error rate ( The mean DS error-corrected molecular depth across all samples was 26,731x for AML-associated genes and 32,812x in the neutral regions. CH was detected in 100% of donors and recipients, with a trend toward more CH clones in older vs. younger donors. We defined CH as mutations observed in at least 2 unique DNA molecules, rather than by a threshold variant allele frequency (VAF). Among the 11 donors with matched historical and contemporary DNA samples, CH mutations in the latter had a higher average VAF and comprised more unique mutations than in the former, reflecting clonal expansions over time. The magnitude of CH expansions was greater for the AML-associated genes than for neutral loci, consistent with positive selection. Post-HSCT the overall number of unique mutations was similar between donors and recipients. Most donor CH variants did not preferentially expand in the recipient. A minority of recipients had a modestly increased overall average VAF relative to paired donors in AML genes. While the VAF of some likely pathogenic variants increased in the recipient vs the donor, this was not the case for most CH mutations in most recipients and the reciprocal was sometimes observed. DS represents a powerful tool for the high-resolution study of CH evolution. In a unique historical cohort comprising longitudinal samples from some of the earliest patients and donors to undergo HSCT, we observed limited instances of accelerated CH in recipients relative to donors. While this cohort is biased due to enriched inclusion of long-term survivors, our data exemplifies how healthy hematopoiesis, indistinguishable between donors and recipients, can be maintained almost half a century following transplant. Citation Format: Masumi Ueda, Jake Higgins, Thomas Smith, Elizabeth Schmidt, Fang Y. Lo, Charles C. Valentine, Jesse J. Salk, Rainer Storb, Jerald P. Radich. Duplex Sequencing reveals ubiquitous clonal hematopoiesis and complex donor-recipient clonal dynamics following HSCT [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2022; 2022 Apr 8-13. Philadelphia (PA): AACR; Cancer Res 2022;82(12_Suppl):Abstract nr LB063.

Mutations of ABL1 are the dominant mechanism of relapse in Philadelphia chromosome-positive acute lymphoblastic leukemia (Ph + ALL). We performed highly accurate Duplex Sequencing of exons 4–10 of ABL1 on bone marrow or peripheral blood samples from 63 adult patients with previously untreated Ph + ALL who received induction with intensive chemotherapy plus a BCR-ABL1 TKI. We identified ABL1 mutations prior to BCR-ABL1 TKI exposure in 78% of patients. However, these mutations were generally present at extremely low levels (median variant allelic frequency 0.008% [range, 0.004%–3.71%] and did not clonally expand and lead to relapse in any patient, even when the pretreatment mutation was known to confer resistance to the TKI received. In relapse samples harboring a TKI-resistant ABL1 mutation, the corresponding mutation could not be detected pretreatment, despite validated sequencing sensitivity of Duplex Sequencing down to 0.005%. In samples under the selective pressure of ongoing TKI therapy, we detected low-level, emerging resistance mutations up to 5 months prior to relapse. These findings suggest that pretreatment ABL1 mutation assessment should not guide upfront TKI selection in Ph + ALL, although serial testing while on TKI therapy may allow for early detection of clinically actionable resistant clones.

Rationale: Accurate and sensitive detection of measurable residual disease (MRD) is critical for clinical management and the development of new therapies for pediatric acute myeloid leukemia (AML). The presence of MRD is a well-established predictor of disease relapse and mortality. Ideally, MRD detection would be both broadly applicable across patients and highly specific, thus enabling clinicians to make timely treatment decisions to improve outcomes. Background: In pediatric AML, unlike adult AML, there are few cross-patient recurrent somatic driver mutations. This makes it impractical to use a fixed gene panel for MRD monitoring. One solution is to perform disease tracking with patient-specific mutations that are present in individual AMLs at time of diagnosis (TOD). However, following modern therapies, should residual disease remain, the telltale mutations will be present in only a very small subpopulation of cells–far below the ~1% limit-of-detection of conventional NGS methods. Duplex sequencing (DS) is an ultra-accurate NGS error-correction method which increases the accuracy of base calls by more than 10,000-fold. Here we demonstrate the use of tumor-informed DS to detect patient-specific MRD in pediatric AML samples with unprecedented sensitivity and specificity. Methods: Marrow aspirates were collected from 13 pediatric AML patients at TOD, during treatment, remission, and at relapse. Among the cohort, 9/13 patients ultimately relapsed. Personalized hybrid capture panels were designed to target hundreds of somatic mutations that were present in each patient’s TOD sample, as identified by whole genome sequencing of tumor-normal pairs. DS was used for longitudinal monitoring of these mutations in blood or marrow at up to 5 time points following initial induction. Results: We detected MRD in one or more samples from all 9 patients who went on to relapse. However, we also observed high-confidence TOD variants in the last collected sample from all 5 patients who did not relapse during at least 24 months of clinical follow-up. As a technical control, we tested a subset of patient panels on a blood sample from a healthy subject and found no disease-specific mutations. Interestingly, we observed a marked degree of heterogeneity of response-to-therapy in the cohort. In some cases, all mutations logarithmically decreased uniformly, whereas in others there was clear dispersion of subclones over time, with some declining 100-times more quickly than others. Conclusions: While DS in conjunction with personalized panels represents a powerful approach for MRD tracking, simply the existence of very low frequency TOD mutations in pediatric AML is not specific enough for predicting relapse without additional interpretation. Lingering mutations in non-relapsers might represent preneoplastic field defects, pre-existing somatic mosaicism, or perhaps even true residual disease under immune suppression. Our study highlights an unappreciated degree of complexity that arises when using ultra-accurate mutation detection for cancer monitoring. Citation Format: Thomas H. Smith, Jake E. Higgins, Fang Y. Lo, Michael J. Hipp, Charles C. Valentine, Amanda R. Leonti, Rhonda E. Ries, Soheil Meshinchi, Benjamin J. Huang, Jesse J. Salk. Longitudinal monitoring of pediatric acute myeloid leukemia using duplex sequencing of patient-specific panels reveals ultra-low frequency MRD that marks persistent carcinogenesis and complex clonal evolution [abstract]. In: Proceedings of the AACR Special Conference on the Evolutionary Dynamics in Carcinogenesis and Response to Therapy; 2022 Mar 14-17. Philadelphia (PA): AACR; Cancer Res 2022;82(10 Suppl):Abstract nr A009.

Sensitive and specific detection of measurable residual disease (MRD) after treatment in pediatric acute myeloid leukemia (AML) is prognostic of relapse and is important for clinical decision making. Mutation-based methods are increasingly being used, but are hampered by the limited number of common driver gene mutations to target as clone markers. Additional targets would greatly increase MRD detection power. However, even in cases with many AML-defining mutations, it is the limited accuracy of current molecular methods which establishes the lower bounds of sensitivity. Here we describe an ultrasensitive approach for disease monitoring with personalized hybrid capture panels targeting hundreds of somatic mutations identified by whole genome sequencing (WGS), and using extremely accurate Duplex Sequencing (DS) in longitudinal samples. In a pilot cohort of 13 patients we demonstrate detection sensitivities several orders of magnitude beyond currently available single locus testing or less accurate sequencing. With multi-target panels, overall power for MRD detection is cumulative across sites. For example, if a patient has MRD at a true frequency of 1/30,000, sequencing a single mutant site to 10,000x molecular depth would be unlikely to detect MRD. However, sequencing 10 sites each to 10,000x would effectively total 100,000x informative site depth, increasing power to >95%. However, standard sequencing assays are insufficiently accurate to achieve this theoretical limit of detection (LOD). DS enables accurate detection of individual variants to Marrow aspirates were collected from 13 uniformly treated pediatric AML patients at time of diagnosis (TOD), during treatment (end of induction, EOI), in remission (end of therapy, EOT), and at relapse. 9/13 patients relapsed. DNA from TOD was analyzed by WGS. Germline variants were excluded and somatic single nucleotide variants (SNVs) were targeted by a custom probe panel designed for each patient. An average of 170 SNVs were targeted per patient (range 53-200). More than 90% of the SNVs were noncoding. Longitudinal samples were then analyzed with DS, which compares sequences from both strands of each DNA molecule to eliminate technical noise and reveal biological mutation signal with extreme accuracy and sensitivity. A median of 82% of WGS SNVs were validated by DS in the TOD DNA, and the vast majority of those were also present at relapse. Relapsers had more SNVs at diagnosis than non-relapsers. EOT samples were sequenced to an average Duplex molecular depth of 29,400x, with a maximum of 61,283x. The figure below shows time course plots tracking SNVs at diagnosis, EOT and relapse for 2 patients. Among mutations validated in TOD samples, a median of only 8 (5%) were detected per EOT sample (range 0-66 mutations). MRD was detected in 8/9 relapsers. Targeting 1 or even 10 SNVs would therefore have missed MRD in the majority of these patients. Among relapsers, median EOT SNV VAF was 0.069%. The lowest single VAF detected per EOT sample ranged from 0.036% to 0.002%. The presence of an SNV at diagnosis and relapse implies that it must truly be present at EOT, whether or not it is detected. Therefore, if a small minority of leukemic mutations are detected at EOT, the true overall MRD frequency is much lower than the LOD at any single site. In the only relapser where MRD was not detected, targeted SNVs were present at diagnosis and relapse, so additional sequencing depth at EOT would eventually reveal ultra-low frequency mutations. Among patients that did not relapse by the end of the study, median VAF at EOI (the latest time point DNA available) was 0.0258%. Therefore, non-relapsers have a lower median VAF at EOI than relapsers do even later at EOT, potentially indicating very early on that treatment is more successful. This study shows excellent performance of DS-based assays for detecting MRD with patient-specific panels. We have demonstrated that among large panels of validated somatic SNVs present at time of diagnosis, a median of 5% are identified at EOT in eventual relapsers. DS detected MRD in 8/9 patients, and at a median VAF well below the limit of detection of any other sequencing technology. Comprehensive personalized hybrid selection panels coupled with DS represents a powerful option for MRD monitoring in pediatric AML and potentially other cancers. Figure Disclosures Higgins: TwinStrand Biosciences: Current Employment. Lo:TwinStrand Biosciences: Current Employment. Hipp:TwinStrand Biosciences: Current Employment. Valentine:TwinStrand Biosciences: Current Employment. Williams:TwinStrand Biosciences: Current Employment. Radich:TwinStrand Biosciences: Research Funding. Salk:TwinStrand Biosciences: Current Employment.

Aristolochic acids (AA) are a potent class of mutagen that occur in plants of the genera Aristolochia and Asarum. Exposures to AA are caused both by intentional ingestion of traditional herbal medicines containing Aristolochia and through inadvertent consumption from contaminated crops. The carcinogenic potential and epidemiologic connection of AA exposure with upper-tract urothelial cancers (UTUC), liver cancer and kidney failure have been known for years; however, the extent of AA exposure and its influence on cancer risk and mortality have yet to be fully explored in many world populations. Using whole exome sequencing of tumor samples, a recent study determined that 78% of HCCs from Taiwan exhibit distinct mutation signatures of AA exposure (Ng, Poon, Huang, et al., 2017). Exposure to AA is geographically widespread and there is a major opportunity for primary and secondary prevention of AA-associated diseases if it were possible to detect the exposure through non-invasively sampled body fluids such as blood or urine sediment. Although AA adducts can be measured in body fluids, this measurement only reflects recent ingestion, not the life-integrated exposure that better approximates disease risk. A mutation burden-based assay could circumvent this problem; however, a major challenge is that AA-mediated mutations are extremely rare in bulk tissue or fluid samples from non-tumor sources. Duplex Sequencing (DS) is a tag-based NGS error-correction method that enables the detection of ultra-rare mutations that exist at the mutation frequency of normal tissues (~10e-8). DS can be applied to DNA from any tissue of any organism without the need for tumor formation or ex vivo cloning. We developed a DS-based assay that measures mutations across a 50 kb genome-representative panel and applied it to the tissues and liquid biopsies of cancer patients of known AA-exposure status. We sampled the blood, urine, kidney, normal-adjacent urothelium and ureteral tumors from 12 UTUC patients where 6 of the patients were exposed to AA, and 6 were not. Under blinded conditions, we positively identified AA mutation signatures in the normal adjacent urothelium tissues of 5 AA-exposed patients. Remarkably, we also detected AA signatures in all available blood and urine samples from the same patients. The proportion of mutations attributed to AA-exposure ranged from 2.5% in blood to 67.8% in kidney. These mutations, on the order of one-in-a-million, were exposure-related, not tumor derived. We did not detect AA mutation signatures in any samples from the non-AA-exposed patients. To our knowledge, this is the first time an AA mutation signature has been detected in non-invasively sampled body fluids. DS is a promising approach for the detection of mutagenic signatures caused by environmental carcinogens, and we foresee it being a powerful metric that can be used to measure life-integrated carcinogenic processes and cancer risk. Citation Format: Fang Yin Lo, Charles Valentine, Elizabeth Schmidt, Lindsey Williams, Arnoud Boot, Steve Rozen, Jesse Salk. Non-invasive detection of aristolochic acid exposure using ultra-sensitive duplex sequencing [abstract]. In: Proceedings of the Annual Meeting of the American Association for Cancer Research 2020; 2020 Apr 27-28 and Jun 22-24. Philadelphia (PA): AACR; Cancer Res 2020;80(16 Suppl):Abstract nr 3147.

Hematologic malignancies (HM) are driven by clonal evolution—from preneoplastic clones through acquired drug resistance. Clonal hematopoiesis of indeterminate potential (CHIP) is a process whereby healthy individuals accumulate low-frequency clones with driver gene mutations. CHIP is associated with increased risk of HM but is challenging to detect. Measurable residual disease (MRD) after therapy in acute myeloid leukemia (AML) and other HM is a strong predictor of relapse; however, current MRD assays either have limited sensitivity or are narrowly applicable. Tyrosine kinase inhibitors (TKI) target BCR-ABL1 in chronic myeloid leukemia (CML) and Ph+ acute lymphocytic leukemia (ALL), yet TKI-resistance (TKI-R) mutations can drive relapse. Early detection of MRD and TKI-R mutations can inform important clinical interventions. Next-generation sequencing (NGS) can detect HM subclones, but not below a frequency of approximately 1%. Common error correction methods have difficulty detecting mutations below 0.1%. Duplex sequencing (DS) is a sophisticated NGS error correction assay capable of detecting individual mutations down to ultralow frequencies of at least 0.001% with a background error frequency below 1 in 10 million bases. We used DS for three clinically relevant HM applications: (1) CHIP: We performed DS on DNA from 10 hematopoietic cell transplant (HCT) donor/recipient pairs 7-46 years post-HCT. We targeted 29 genes recurrently mutated in AML, many of which have been described in CHIP. DS error-corrected depth up to 45,402x was generated. We identified dozens of clonal exonic variants (CEV) in every donor and recipient, most nonsynonymous, down to variant allele frequency (VAF) of 5.7E-5. The majority of the 154 shared donor/recipient CEVs were present at VAF This abstract is also being presented as Poster A48. Citation Format: Jake Higgins, Lindsey N. Williams, Charles C. Valentine, Gabriel Pratt, Rashmi Kanagal-Shamanna, Nicholas J. Short, Masumi Ueda, Rainer Storb, Jerald P. Radich, Jesse J. Salk. Clonal landscapes of hematologic malignancies redefined by ultrasensitive duplex sequencing [abstract]. In: Proceedings of the AACR Special Conference on Advances in Liquid Biopsies; Jan 13-16, 2020; Miami, FL. Philadelphia (PA): AACR; Clin Cancer Res 2020;26(11_Suppl):Abstract nr PR04.

Cumulative evidence indicates that a significant proportion of cancer evolution may occur before the development of histological abnormalities. While recent improvements in DNA sequencing technology have begun to reveal the presence of these early preneoplastic clones, the concept of 'premalignant field' was already introduced by Slaughter more than half a century ago. Also referred to as 'field effect', 'field defect' or 'field cancerization', these terms describe the phenomenon by which molecular alterations develop in normal-appearing tissue and expand to form premalignant patches with the potential to progress to dysplasia and cancer. Field effects have been well-characterized in ulcerative colitis, an inflammatory bowel disease that increases the risk of colorectal cancer. The study of the molecular alterations that define these fields is informative of mechanisms of tumor initiation and progression and has provided potential targets for early cancer detection. Herein, we summarize the current knowledge about the molecular alterations that comprise the field effect in ulcerative colitis and the clinical utility of these fields for cancer screening and prevention.

Cancer is a disease of somatic evolution, characterized by the natural selection of genomic mutations that facilitate enhanced cell survival and proliferation. Although many thousands of tumor genomes have now been sequenced, our ability to identify early genetic patterns of clonal selection in both humans and model organisms have been hampered by inadequately sensitive methodologies for identifying mutations during the long period between their occurrence and the final outgrowth of a clinically apparent tumor. Sensitive molecular tests are capable of measuring minute levels of cancer cells in tissue samples, however no existing method is satisfactory at scaling to high-throughput detection at the rate of one cancer-associated variant per more than a million normal cells. NGS is used to study the genetic variation in cell populations, although the accuracy of standard NGS methods limit our ability to detect sub-clones below ~1%. Duplex Sequencing (DS) is a sensitive NGS error-correction method which increases the accuracy of base calls by more than 100,000-fold. DS enables precise reconstruction of the original double-stranded source molecule while overcoming both chemical and technical artifacts that arise during library preparation and sequencing. Here we present the use of Duplex Sequencing, in both human and mouse tissues, for measuring sub-clones at allelic fractions less than 1×10-4 for an early glimpse into pre-neoplastic evolution. We illustrate how, merely weeks after mutagen exposure, we observe emerging clones of cells bearing cancer-driving mutations in histologically normal mouse tissue. Furthermore, we illustrate how similar patterns of clonal selection can be seen in multiple otherwise healthy tissues of humans as part of normal aging. We discuss how variations in the pattern and rate of accumulation of rare cancer-associated mutations offers a new preclinical tool for evaluating carcinogenicity of chemicals, as well as a potential clinical tool for assessing life-integrated carcinogenic processes and cancer risk in humans. Citation Format: Charles C. Valentine, Mark Fielden, Robert Young, Jake Higgins, Lindsey Williams, Tan Li, Rohan Kulkarni, Sheroy Minocherhomji, Rosana Risques, Patrick Danaher, Jesse Salk. Duplex Sequencing detects rare subclonal variants that mark early carcinogenesis and preneoplastic clonal evolution [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 4649.

The purpose of this study is to develop a universally applicable Duplex Sequencing (DS) assay for extremely sensitive detection of minimal residual disease (MRD) in acute myeloid leukemia (AML). The presence of MRD plays a critical role in AML recurrence and mortality, and is an important prognostic marker to guide clinical care and clinical trial risk stratification. Unfortunately, current MRD detection methods, including light microscopy, cytogenetics, qPCR, and flow cytometry have shortcomings in terms of sensitivity, general applicability, and/or inter-laboratory reproducibility. Duplex Sequencing eliminates technical artefacts observed with other sequencing technologies and delivers an unprecedented level of accuracy and sensitivity, on widely available NGS platforms. Here, we report validation of our DS MRD mutation detection assay, with a background mutation level more than 100-fold below single-stranded tag-based error correction (SSCS). Our complete DS MRD probe panel targets mutation hotspots or full coding sequences in 29 genes recurrently mutated in AML, representing mutations found in approximately 90% of patients. Supplementing with a targeted RNA-seq panel for recurrent gene fusions would approach 100%. To simulate low frequency mutations in MRD, we generated 2 DNA dilution series: either AML patient DNA or cell line DNA with known AML mutations that were serially diluted into normal human DNA. Replicate libraries were prepared with custom targeted probe panels and sequenced to over 50,000x or over 1,000,000x total Duplex depth (original Duplex DNA molecules sequenced per site), respectively. Expected mutations were detected down to MAFs of 1.9x10-5 and 6.0x10-6. Background mutation levels of 5.7x10-7 and 5.8x10-7, were respectively seen in each series, and likely reflect the true age-associated biological background, rather than technical noise. SSCS artefacts in these samples were approximately 100x higher than the frequency of DS mutations, and obscured expected mutations in the lower dilutions. We are further validating our method in both retrospective and prospective studies. In retrospective samples from a completed clinical trial, we are comparing the sensitivity and accuracy of DS vs. multicolor flow cytometry for MRD with known clinical outcomes. The relative prognostic value of DNA from blood vs. bone marrow will be quantified with DS as well. We will also perform DS on prospective samples to examine the prognostic value of specific genes and MAFs to eliminate uninformative variants, which will also allow accurate quantification of low level mutations in non-leukemic clones that accumulate with aging and can confound detection of MRD. Elimination of sequencing artefacts is critical for the application of NGS assays to MRD detection, especially when expected mutations are unknown, and DS is an important new tool in high sensitivity cancer diagnostics. Citation Format: Jacob E. Higgins, Lindsey N. Williams, Christopher S. Hourigan, Jerald P. Radich, Jesse J. Salk. Duplex sequencing for MRD detection in acute myeloid leukemia [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 422.

Background: Kinase domain (KD) mutations in ABL1 are the dominant mechanism of relapse in patients (pts) with Philadelphia chromosome-positive acute lymphoblastic leukemia (Ph+ ALL). Previous studies in Ph+ ALL have suggested that some pts harbor subclonal T315I mutations prior to tyrosine kinase inhibitor (TKI) treatment. However, nearly all reports have relied upon inherently error prone RT-PCR to generate template cDNA prior to mutation analysis. We hypothesized that conventional assays might over-estimate the incidence of pre-existing resistance mutations, and that improved sequencing accuracy might yield important information for risk stratification and TKI selection. Methods: Duplex sequencing (DS) is a molecular tagging method that improves the accuracy of conventional next-generation sequencing by more than 10,000-fold, by comparing the nucleotide sequences of each strand of double-stranded molecules. ABL1 DS was performed on genomic DNA from 64 pts with newly diagnosed Ph+ ALL treated with hyper-CVAD plus a TKI. DS of exons 4-10 of ABL1 was performed to an average molecular depth of >10,000x. Among pts who relapsed, using RNA extracted from relapse samples, the KD (codons 221 through 500) of the BCR-ABL1 fusion transcript was sequenced by the Sanger method using a nested PCR approach, with a detection limit of 10-20%. Results: The median age of the cohort was 54 years (range, 20-80 years). The TKI used was imatinib in 5 pts, dasatinib in 38 pts, and ponatinib in 21 pts. All pts achieved complete remission, and 12 pts (19%) underwent allogeneic stem cell transplantation in first remission. A total of 115 pretreatment ABL1 KD mutations were detected among 47 pts (73%). The median number of pretreatment ABL1 KD mutations was 2 (range, 0-6 mutations). The median variant allelic frequency (VAF) of the detected somatic mutations was 0.008% (range, 0.004%-0.649%). Five mutations (4%) and 40 mutations (35%) were present at a VAF ≥0.1% and ≥0.01%, respectively. Eleven mutations known to confer resistance to at least one TKI were detected in 7 pts (11%), and included: F317L in 4 pts, E225K in 2 pts, and E225V, L384M, M244V, Q252H and T315I in 1 pt each. Five mutations were detected in ≥1 pretreatment sample (F317L in 4 pts, M244V in 3 pts, and E255K, E459K and V355V in 2 pts each). Of these 7 pts with pretreatment resistance mutations, 5 pts received a TKI known not to be sensitive to the mutation(s); 2 pts who received ponatinib had mutations at least intermediately sensitive to ponatinib (1 pt with F317L and 1 pt with both E255K and M244V mutations). With a median duration of follow-up of 54 months (range, 1-124 months), 18 pts have relapsed. None of the 7 pts with known pretreatment resistance mutations relapsed. There was no difference in the number of pretreatment mutations between pts who relapsed and those who did not (median mutations: 1 [range, 0-4] and 2 [range, 0-6], respectively; P=0.26). Of the 18 pts who relapsed, 14 underwent Sanger sequencing for ABL1 KD mutations at the time of relapse. Clonal resistance mutations were detected at relapse in 9 pts (64% of sequenced samples): T315I in 6 pts, and F317I, V229L and V338G in 1 pt each (all in pts treated with dasatinib, except V338G in pt on imatinib). Relapse mutations were not observed in pretreatment samples in any of the pts. Conclusions: DS identified very low level pretreatment ABL1 KD mutations in a majority of pts with newly diagnosed Ph+ ALL but these appear inconsequential. Ninety percent of mutations identified have not been described as resistance mutations, suggesting that they may be either synonymous or functionally neutral amino acid changes resulting from normal aging. There was no apparent association of these mutations and risk of relapse, even in the minority of cases in which known resistance mutations were detected at baseline. These data suggest that pretreatment testing for ABL1 KD mutations in Ph+ ALL is unlikely to affect treatment decisions. Disclosures Short: Takeda Oncology: Consultancy. Sasaki:Otsuka Pharmaceutical: Honoraria. Ravandi:Jazz: Honoraria; Orsenix: Honoraria; Orsenix: Honoraria; Seattle Genetics: Research Funding; Sunesis: Honoraria; Xencor: Research Funding; Bristol-Myers Squibb: Research Funding; Amgen: Honoraria, Research Funding, Speakers Bureau; Abbvie: Research Funding; Astellas Pharmaceuticals: Consultancy, Honoraria; Seattle Genetics: Research Funding; Bristol-Myers Squibb: Research Funding; Astellas Pharmaceuticals: Consultancy, Honoraria; Amgen: Honoraria, Research Funding, Speakers Bureau; Macrogenix: Honoraria, Research Funding; Jazz: Honoraria; Abbvie: Research Funding; Xencor: Research Funding; Macrogenix: Honoraria, Research Funding; Sunesis: Honoraria. Cortes:Pfizer: Consultancy, Research Funding; Astellas Pharma: Consultancy, Research Funding; Novartis: Consultancy, Research Funding; Daiichi Sankyo: Consultancy, Research Funding; Arog: Research Funding. Konopleva:Stemline Therapeutics: Research Funding. Radich:TwinStrand Biosciences: Research Funding. Jabbour:Takeda: Consultancy, Research Funding; Bristol-Myers Squibb: Consultancy, Research Funding; Novartis: Research Funding; Pfizer: Consultancy, Research Funding; Abbvie: Research Funding.

Clinical oncology is being revolutionized by the increasing use of molecularly targeted therapies. This paradigm holds great promise for improving cancer treatment; however, allocating specific therapies to the patients who are most likely to derive a durable benefit continues to represent a considerable challenge. It is becoming increasingly clear that cancers are characterized by extensive intratumour genetic heterogeneity, and that patients being considered for treatment with a targeted agent might, therefore, already possess resistance to the drug in a minority of cells. Indeed, multiple examples of pre-existing subclonal resistance mutations to various molecularly targeted agents have been described, which we review herein. Early detection of pre-existing or emerging drug resistance could enable more personalized use of targeted cancer therapy, as patients could be stratified to receive the therapies that are most likely to be effective. We consider how monitoring of drug resistance could be incorporated into clinical practice to optimize the use of targeted therapies in individual patients.

High-throughput DNA sequencing has fundamentally changed practice standards in clinical oncology. We can now systematically examine the dysfunctional genes and pathways of individual tumors and can often pair these with targeted therapies. Moreover, NGS has improved our ability to resolve molecular heterogeneity, such as results from evolving subclones within a tumor or admixtures of cancer-derived and normal DNA in liquid biopsies. However, this remains imperfect. With standard NGS, sequencing errors create a background that obscures variants present at less than 1-2%. Methods of tag-based error correction improve accuracy, but most rely on hemimolecular tagging and are thus still limited by certain types of errors caused by DNA damage. Our group pioneered Duplex Sequencing, an error correction technology that independently genotypes and compares both single-stranded halves of individual DNA duplexes. This approach improves sequencing accuracy by >100,000-fold and allows detection of a single mutation among ten million unmutated nucleotides. We carried out Duplex Sequencing on uterine lavages from women with and without high-grade serous ovarian cancer to determine the sensitivity for identifying tumor-derived TP53 mutations as a screening test. Under blinded conditions we detected tumor-derived mutations in 80% of cancer cases. However, in addition, we found that nearly all women, including those without cancer, carried very low frequency TP53 mutations that did not originate from a tumor. These could not be explained by technical errors, as they exhibited a strong signature of positive selection, indicated by higher-than-random non-synonymous fraction, substantial clustering in cancer-associated hotspots and a spectrum, exonic distribution and predicted pathogenicity closely mirroring TP53 mutations in cancer databases. Furthermore, abundance of these “biological background” mutations increased with age. To more broadly investigate the extent of clonal evolution in normal aging, we used Duplex Sequencing to examine tissue specimens from two individuals without cancer, differing in age by 101 years. We assessed relative mutation load in 22 cancer-associated genes and one non-cancer associated gene in 19 different tissues. Initial results indicate that the extent of age-associated mutation accumulation varies by organ and the pattern of apparent clonal selection is influenced by both tissue type and genetic context. Our findings illustrate the near-universal presence of cancer-like mutations arising from age-associated processes in multiple non-neoplastic tissues. Somatic clonal evolution appears to be a ubiquitous part of human life. Beyond many biological implications, this highlights an important specificity challenge that must be accounted for when using ultra-accurate sequencing technologies for cancer screening and other sensitive diagnostics. Citation Format: Jesse J. Salk, Elisabeth Maritschnegg, Paul Speiser, Robert Zeillinger, Rosana Risques, Lawrence A. Loeb. Duplex Sequencing detects cancer-associated mutations arising during normal aging: Clonal evolution over a century of human lifetime [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 3041. doi:10.1158/1538-7445.AM2017-3041

The extent of heterogeneity in DNA sequence within human cancers has been difficult to fully assess, due to the absence of methods with sufficient sensitivity to detect minority variants. For example, next-generation sequencing could in principle detect sub-clonal mutations at any genomic position; however due to amplification errors and inaccuracies in the sequencing platform itself, such approaches are limited to detection of mutations present in >1% of cells. We have overcome this limitation by developing an approach, Duplex Sequencing, which improves the accuracy of next-generation sequencing by >100,000 fold. Duplex Sequencing is based upon separately tagging and sequencing the two strands of single molecules of duplex DNA. True mutations are present at the same position in both DNA strands and are complementary, whereas artifacts arising from amplification or sequencing errors are seen in only one strand. The calculated error rate of our approach is less than one artifactual mutation per billion nucleotides sequenced (ref. 1). We have also developed an enrichment approach based on sequential rounds of hybridization to biotinylated probes which enables efficient sequencing of targeted regions of the genome (ref. 2). Duplex Sequencing thereby enables exceptionally sensitive detection of sequence heterogeneity within any set of genes. We have now applied Duplex Sequencing to study of sequence heterogeneity in both normal and malignant samples. In acute myeloid leukemia (AML), we find multiple sub-clonal mutations which fall below the detection limit of conventional approaches, some of which encode mutations that can drive chemotherapy resistance and cancer progression. Our results indicate that prior studies have under-estimated the burden of sub-clonal mutations in AML by more than 1,000-fold. We have additionally studied chronic myeloid leukemia (CML); treatment of CML with inhibitors directed against the Abl kinase represents the prototypical targeted cancer therapy. With Duplex Sequencing, we find that Abl mutations are extremely uncommon at the time of CML diagnosis. In contrast, refractory patients who fail therapy frequently possess multiple sub-clonal mutations within the Abl kinase, most of which are below the resolution of conventional approaches but are readily detected with our method. In ongoing work, we have found multiple sub-clonal mutations within human prostate cancer and colon cancer, suggestive of widespread intratumor heterogeneity that may limit the efficacy of single-agent therapy in these diseases. References: 1. Schmitt MW, Kennedy SR, Salk JJ, Fox EJ, Hiatt JB, Loeb LA (2012). Detection of ultra-rare mutations by next-generation sequencing. Proceedings of the National Academy of Sciences USA. 109(36):14508-13. 2. Schmitt MW, Fox EJ, Prindle MJ, Reid-Bayliss KS, True LD, Radich JP, Loeb LA (2015). Sequencing small genomic targets with high efficiency and extreme accuracy. Nature Methods. 12(5):423-5. Citation Format: Michael W. Schmitt, Jesse J. Salk, Edward J. Fox, Justin R. Pritchard, J Graeme Hodgson, Victor M. Rivera, Pamela S. Becker, Jerald P. Radich, Lawrence A. Loeb. Direct assessment of sequence heterogeneity in human cancers by Duplex Sequencing. [abstract]. In: Proceedings of the 107th Annual Meeting of the American Association for Cancer Research; 2016 Apr 16-20; New Orleans, LA. Philadelphia (PA): AACR; Cancer Res 2016;76(14 Suppl):Abstract nr 2395.

With the publishing of the first complete whole genome of a human cancer and its paired normal, we have passed a key milestone in the cancer genome sequencing strategy. The generation of such data will, thanks to technical advances, soon become commonplace. As a significant number of proof-of-concept studies have been published, it is important to analyze now the likely implications of these data and how this information might frame cancer research in the near future. The diversity of genes mutated within individual tumor types, the most striking feature of all studies reported to date, challenges gene-centric models of tumorigenesis. Although cancer genome sequencing will revolutionize certain aspects of personalized care, the value of these studies in facilitating the development of new therapies, their primary goal, seems less promising. Most significantly, however, the cancer genome sequencing strategy, as currently applied, fails to characterize the most relevant genomic features of cancer—the mutational heterogeneity within individual tumors. [Cancer Res 2009;69(12):4948–4950]

Next-Generation DNA Sequencing now allows us to sequence billions of nucleotides in a single experiment and to detect germ-line mutations and driver mutations in human cancer. However, a major limitation is the error rate of approximately 1% to 0.1%. These scattered errors limit its utility in sequencing genetically heterogeneous mixtures such as human cancers or viruses. We have developed a method termed "Duplex Sequencing” that allows for the detection of rare mutations in DNA with unprecedented accuracy. This approach greatly lowers errors by independently sequencing each of the two strands of single DNA molecules. As the two strands are complementary, true mutations are scored only if they are present at the same position and are complementary. Moreover, small deletions and insertions can be easily identified. In contrast, sequencing errors, which occur from PCR - amplification of DNA containing damaged nucleotides, are present in only one strand and thus are not scored as mutations. To demonstrate the power of duplex sequencing we measured the background error frequency in human mitochondrial DNA obtained from cortical regions of the brain of aged normal humans; an overall mutation frequency was 3 X 10-5. These mutations are not random, but rather, show an inordinately large frequency of mutations in the D-loop of the mitochondrial genome. The spectrum of mutations is strikingly different from that reported in the literature using next-generation sequencing. Prior reports have indicated a high frequency of G->T; however, we find that these are likely technical artifacts arising from oxidative DNA damage. In contrast, by Duplex Sequencing, we find a predominance of G->A and T->C single base substitutions. The G->A and T->C mutations could result from cytidine deamination or errors during replication by the mitochondrial DNA polymerase gamma. In order to establish the background mutation frequency in nuclear DNA by Duplex Sequencing we first captured a small portion of the genome, the exons of the replicative DNA polymerases. Sequencing captured DNA from normal human tissues yielded a background mutation frequency of 10-6 to 1O-7 indicating that the error rate of duplex sequencing is less than one to ten mutations in ten million nucleotides. Current effort is to sequence nuclear DNA from different human cancers and the corresponding normal tissues. This result should indicate the distribution of subclonal mutations in human tumors. Subclonal mutations could delineate tumor cell lineage, account for the geographic heterogeneity of mutations in human tumors and serve as a biomarker for the presence of drug resistant cells prior to or during chemotherapy. Citation Format: Lawrence A. Loeb, Edward J. Fox, Scott R. Kennedy, Jessica Kuong, Michael W. Schmitt, Jesse J. Salk. Duplex DNA sequencing detects subclonal and ultra-rare mutations in human tumors. [abstract]. In: Proceedings of the 104th Annual Meeting of the American Association for Cancer Research; 2013 Apr 6-10; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2013;73(8 Suppl):Abstract nr 1789A. doi:10.1158/1538-7445.AM2013-1789A

Chronic inflammation predisposes to a variety of human cancers. Affected tissues slowly accumulate mutations, some of which affect growth regulation and drive successive waves of clonal evolution whereas a far greater number are functionally neutral and serve only to passively mark expanding clones. Ulcerative colitis (UC) is an inflammatory bowel disease in which up to 10% of patients eventually develop colon cancer. We have recently reported on mapping of mutations in non-coding polyguanine tracts in patients with UC to delineate the extent of clonal expansions associated with carcinogenesis. We genotyped colon biopsies for length altering mutations at 28 different polyguanine markers. In eight UC patients without neoplasia, blinded genotyping detected only two mutations in a single individual from among 37 total biopsies. In contrast, in eleven UC patients with neoplasia elsewhere in the colon, blinded genotyping identified 63 mutations in 51 nondysplastic biopsies, and every patient possessed at least one mutant clone. As a biomarker of cancer progression the presence of one or more clones was 100% sensitive and 88% specific. A subset of clones were large and extended over many square centimeters of colon. Of these, some occurred as isolated populations in non-dysplastic tissue, considerably distant from neoplastic lesions. Other large clones included regions of cancer, suggesting that the tumor arose within a pre-existing clonal field. This preliminary work indicated that neutral mutations in polyguanine tracts serve as a powerful tool for identifying clonally expanded populations, and suggested that these clones may be clinically useful for distinguishing a subset of UC patients who are at greatest risk for developing cancer. Here we provide updated results of a new and larger case-control study and discuss ongoing efforts to streamline clone detection using other types of neutral genetic markers and next-generation sequencing. Citation Format: {Authors}. {Abstract title} [abstract]. In: Proceedings of the 101st Annual Meeting of the American Association for Cancer Research; 2010 Apr 17-21; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2010;70(8 Suppl):Abstract nr 4707.