Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 7-Day Trial for You or Your Team.

Learn More →

High-Throughput Detection of Actionable Genomic Alterations in Clinical Tumor Samples by Targeted, Massively Parallel Sequencing

High-Throughput Detection of Actionable Genomic Alterations in Clinical Tumor Samples by... Published OnlineFirst November 7, 2011; DOI: 10.1158/2159-8290.CD-11-0184 Wagle et al. research article Resea Rch aR ticle h igh-t hroughput Detection of a ctionable Genomic a lterations in c linical t umor s amples by t argeted, Massively Parallel s equencing 1–3 3 2 3 3 Nikhil Wagle , Michael F. Berger , Matthew J. Davis , Brendan Blumenstiel , Matthew DeFelice , 1 2 2 2 1–3 Panisa Pochanard , Matthew Ducar  , Paul Van Hummelen , Laura E. MacConaill , William C. Hahn , 1–3 3 1–3 Matthew Meyerson , Stacey B. Gabriel , and Levi A. Garraway OF82 |  CANCER DISCOVERY  JANUARY 2012  www.aacrjournals.org Downloaded from cancerdiscovery.aacrjournals.org on July 7, 2021. © 2012 American Association for Cancer Research. T Published OnlineFirst November 7, 2011; DOI: 10.1158/2159-8290.CD-11-0184 Genomic Profiling of F The BATTLE Trial: Personalizing ormalin-Fix Ther ed, Par apy for Lung Cancer affin-Embedded Tumor Samples by Targeted Sequencing research ar research ar ticle ticle Knowledge of “actionable” somatic genomic alterations present in each tumor a Bst Ract (e.g., point mutations, small insertions/deletions, and copy-number alterations that direct therapeutic options) should facilitate individualized approaches to cancer treatment. However, clinical implementation of systematic genomic profiling has rarely been achieved beyond limited numbers of oncogene point mutations. To address this challenge, we utilized a targeted, massively parallel sequencing approach to detect tumor genomic alterations in formalin-fixed, paraffin-embedded (FFPE) tumor samples. Nearly 400-fold mean sequence coverage was achieved, and single-nucleotide sequence variants, small insertions/deletions, and chromosomal copy- number alterations were detected simultaneously with high accuracy compared with other meth- ods in clinical use. Putatively actionable genomic alterations, including those that predict sensitivity or resistance to established and experimental therapies, were detected in each tumor sample tested. Thus, targeted deep sequencing of clinical tumor material may enable mutation- driven clinical trials and, ultimately, “personalized” cancer treatment. si GNiFic a Nce : Despite the rapid proliferation of targeted therapeutic agents, systematic methods to profile clinically relevant tumor genomic alterations remain underdeveloped. We describe a sequencing- based approach to identifying genomic alterations in FFPE tumor samples. These studies affirm the feasi- bility and clinical utility of targeted sequencing in the oncology arena and provide a foundation for genomics-based stratification of cancer patients. Cancer Discovery; 2(1); 82–93. ©2011 AACR. int Roduction clinical setting. Because many targeted agents in devel- The maturation of cancer genome characterization opment are designed to intercept proteins and/or path- efforts has fueled the notion that many treatment deci- ways commonly perturbed by tumor genetic changes, an sions might ultimately be guided by the genetic makeup urgent need exists to implement robust approaches that of individual tumors (1). Moreover, the rapid prolifera- determine the “actionable” genetic profiles of individual tion of targeted agents in development has called specific tumors. If widely obtained, such information might better attention to the importance of molecular profiling ap- identify those patients most likely to respond to existing proaches that pinpoint in situ those tumors most likely and emerging anticancer regimens. to respond. Knowledge of such alterations in the clinical We and others have developed tumor mutation–profiling and translational arenas—including mutations, somatic platforms that use mass-spectrometric genotyping (2, 3) copy-number alterations, and polymorphisms affecting or allele-specific PCR-based technologies (4). Each of these drug metabolism—should ultimately facilitate individual- approaches interrogates known oncogene or tumor sup- ized approaches to cancer treatment. However, systematic pressor gene mutations present in DNA obtained from genetic profiling of cancers remains underdeveloped in the either frozen or formalin-fixed, paraffin-embedded (FFPE) tumor tissue. However, genotyping-based platforms have certain limitations that may preclude their applicability as 1 2 Authors’ Affiliations: Department of Medical Oncology and Center for definitive cancer diagnostic modalities. These include the Cancer Genome Discovery, Dana-Farber Cancer Institute, Harvard Medical finite number of prespecified point mutations that can School, Boston; Broad Institute of MIT and Harvard, Cambridge, be assayed (designated a priori from a restricted subset of Massachusetts known cancer genes), difficulties in detecting small inser- Note: N. Wagle and M.F. Berger contributed equally to this work; M.F. tions or deletions (“indels”), insensitivity to most tumor Berger is currently in the Department of Pathology, Memorial Sloan- suppressor gene mutations (which may occur anywhere Kettering Cancer Center, New York, New York. within the gene), inability to detect gene amplifications or Note: Supplementary data for this article are available at Cancer deletions, and decreased sensitivity in tumor samples with Discovery Online (http://www.cancerdiscovery.aacrjournals.org). high stromal admixture. At the present time, no systematic Corresponding Author: Levi A. Garraway, Department of Medical mechanism exists whereby clinical tumor specimens might Oncology, Dana-Farber Cancer Institute, 450 Brookline Avenue, D1542, Boston, MA 02215. Phone: 617-632-6689; Fax: 617-582-7880; E-mail: be interrogated in situ for a fully comprehensive panel of [email protected] actionable cancer gene alterations. doi: 10.1158/2159-8290.CD-11-0184 The advent of massively parallel sequencing is trans- ©2011 American Association for Cancer Research. forming the cancer genomics landscape by enabling JANUARY 2012 CANCER DISCOVERY    83 Downloaded from cancerdiscovery.aacrjournals.org on July 7, 2021. © 2012 American Association for Cancer Research. Published OnlineFirst November 7, 2011; DOI: 10.1158/2159-8290.CD-11-0184 Wagle et al. research article comprehensive cancer genome characterization at an selection reaction and sequenced in a single Illumina lane unprecedented scope (1, 5, 6). Concomitantly, hybrid se- with 100-bp paired end reads. lection-based methods that enrich for coding sequences The 11 equimolar DNAs were evenly represented, with prior to sequencing (“exon capture”; refs. 7, 8) are rou- 12 to 17 million purity-filtered reads generated per sample tinely being implemented in discovery-oriented settings (average of 14.6 million purity-filtered reads; Supplementary (5). Here, we describe an adaptation of exon capture and Table S4), whereas the sample present at 50% concentration massively parallel sequencing for robust detection of so- (HT-29, index 2) had 7.8 million purity-filtered reads, as matic genomic alterations in FFPE samples. The approach expected. The percent of bases mapping “on-target” averaged leverages a targeted exon capture technique to enrich 60% (range, 56%–64%) across all samples in the pool, yielding for a cancer-relevant genomic territory consisting of 137 a mean 527× target coverage (range, 441× to 593×) for the genes ( 400,000 coding bases), thereby allowing multiple 11 equimolar samples. More than 95% of target exons barcoded samples to be pooled into a single sequencing exhibited more than 30× coverage after sequencing (suf- reaction while preserving deep (e.g., >300- to 400-fold) ficient to call “high-confidence” variants in a sample with sequencing coverage of targeted regions. This approach 70%–80% tumor purity), whereas only 1% had no coverage simultaneously identifies mutations and chromosomal (Supplementary Fig. S2A and B). In general, poorly captured copy-number alterations in clinical tumor material and exons had greater than 70% GC content, although GC con- may inform a comprehensive means to achieve DNA-based tent did not account for all of the poorly captured targets patient stratification in the clinical and translational on- (Supplementary Fig. S2C). The capture performance for a cology arena. particular target exon was highly reproducible from sample to sample (Supplementary Figure S2D–F). Results Detection of s ingle-Nucleotide Variants, We generated a list of 137 “druggable” or potentially insertions/Deletions, and c opy-Number actionable genes known to undergo somatic genomic a lterations alterations in cancer (Supplementary Table S1). These in- clude targets of existing and novel therapeutics, prognostic In total, 102 single-nucleotide variants and 6 indels markers, and other oncogenes and tumor suppressors that (excluding known germline polymorphisms) were detected are frequently mutated in cancer. In addition, we included in coding sequences across the 10 cell lines, including all 79 pharmacogenomic polymorphisms in 34 genes that 21 single-nucleotide variants and 3 of 4 indels reported for may predict heightened sensitivity/resistance or toxic- these lines in the Catalogue of Somatic Mutations in Cancer ity to conventional cancer therapies (Supplementary Table database (COSMIC) (Supplementary Table S5; ref. 10). The S2). Altogether, these genes comprise 2,372 exons encod- single indel that was not initially identified—a 9-bp deletion ing 433,159 bases. We then designed and synthesized 7,021 in PIK3CA in the NCI-H69 cell line—was readily detected by unique biotinylated RNA baits corresponding to these ge- manual inspection of the raw sequencing data. Therefore, all nomic regions. previously reported point mutations and indels for this small We leveraged a solution-based exon capture/massively par- collection were detectable by this approach. (A complete list- allel sequencing approach in which a pool of long oligonucle- ing of all alterations identified in these cell lines can be found otides complimentary to these exons of interest were used to in the Supplementary Appendix.) reduce the complexity of tumor genomic DNA for clinically- In the absence of paired normal samples, the majority oriented sequencing. Here, a 6-nucleotide DNA barcode was of variants detected are germline alterations. Nonetheless, appended to the ends of DNA fragments during library con- previously unreported variants were still informative in sev- struction, thus allowing multiple samples to be pooled before eral instances. For example, 12 single-nucleotide variants hybrid selection to expand the scope of genomic profiling (9). were detected in the breast cancer cell line MDA-MB-231, The approach is illustrated schematically in Supplementary including all 4 alterations in the COSMIC database (BRAF, Figure S1. TP53, KRAS, and NF2; Fig. 1A and B, Supplementary Table S6; ref. 10). One of the additional alterations was a 1-bp c apture Performance and r eproducibility frameshift insertion involving the NF1 tumor suppressor We first optimized the approach by using genomic predicted to generate a truncated protein product (Fig. DNA from normal samples and tumor cell lines known 1C). This NF1 insertion likely represents a bona-fide cancer- to harbor mutations and/or chromosomal copy-number associated mutation. The MDA-MB-231 cell line has pre- alterations affecting multiple cancer genes represented in viously been shown to lack both an NF1 mRNA isoform our hybrid capture baits. Ten cancer cell lines with well- and the neurofibromin protein (the product of NF1); thus, characterized, mutually exclusive cancer gene mutations these findings may provide a genetic basis for neurofibro- were chosen (Supplementary Table S3) as well as control min loss in this setting (11). diploid genomic DNA. Equimolar amounts of the result- Although detection of point mutations and indels by tar- ing sequencing libraries were pooled together with an ad- geted, massively parallel sequencing has become increasingly ditional library from the HT-29 cell line, which was added common, the simultaneous detection of chromosomal copy- at a 50% molar ratio compared with the other libraries. number alterations by this approach is less well-established, This pool of 12 libraries was subjected to a single hybrid particularly in the clinical arena. To determine copy-number 84 |  CANCER DISCOVERY  JANUARY 2012  www.aacrjournals.org Downloaded from cancerdiscovery.aacrjournals.org on July 7, 2021. © 2012 American Association for Cancer Research. Published OnlineFirst November 7, 2011; DOI: 10.1158/2159-8290.CD-11-0184 Genomic Profiling of FFPE Tumor Samples by Targeted Sequencing research article chr7:140127873-140127899 chr7:140127873-140127899 chr7:7517811-7517837 chr7:7517811-7517837 chr17:26565586-26565612 chr17:26565586-26565612 a B c chr7:140127873-140127899 chr7:7517811-7517837 chr17:26565586-26565612 G464V (C>A) G464V (C>A) R280K (C>T) R280K (C>T) T467Hfs*3 (ins C) T467Hfs*3 (ins C) Reference: 310 Reference: 310 Reference: 1 Reference: 1 Reference: 0 Reference: 0 G464V (C>A) R280K (C>T) T467Hfs*3 (ins C) Variant: 391 Variant: 391 Variant: 768 Variant: 768 Variant: 67 Variant: 67 Reference: 310 Reference: 1 Reference: 0 Variant: 391 Variant: 768 Variant: 67 BRAF BRAF TP53 TP53 NF1 NF1 (1-bp insertion) (1-bp insertion) BRAF TP53 NF1 (1-bp insertion) D Target exon coverage: tumor vs. normal Target exon coverage: tumor vs. normal e Copy-number ratio: sequencing vs. array Copy-number ratio: sequencing vs. array Target exon coverage: tumor vs. normal Copy-number ratio: sequencing vs. array 2400 2400 All target exons All target exons All target genes All target genes CDKN1A CDKN1A CDKN1A CDKN1A All target exons All target genes 2.5 2.5 2000 2000 SRC SRC SRC SRC CDKN1A CDKN1A 2.5 NOTCH4 NOTCH4 NOTCH4 NOTCH4 SRC SRC 1600 1600 PTPRD PTPRD PTPRD PTPRD NOTCH4 NOTCH4 JAK2 JAK2 JAK2 JAK2 1600 PTPRD 2 PTPRD CDKN2A CDKN2A CDKN2A CDKN2A JAK2 JAK2 1200 1200 1.5 1.5 CDKN2A CDKN2A 1200 1.5 800 800 11 400 400 0.5 0.5 400 0.5 00 200 200 400 400 600 600 800 800 1000 1000 1200 1200 00 0.5 0.5 11 1.5 1.5 22 2.5 2.5 Exon coverage in normal cell line (HapMap) Exon coverage in normal cell line (HapMap) Copy-number ratio for MDA-MB-231: SNP array Copy-number ratio for MDA-MB-231: SNP array 0 200 400 600 800 1000 1200 0 0.5 1 1.5 2 2.5 Exon coverage in normal cell line (HapMap) Copy-number ratio for MDA-MB-231: SNP array Figure 1. Genomic alterations in breast cancer cell line MDA-MB-231. a –c , representative genome images from the Integrated Genome Viewer (IGV) for several alterations found in the breast cancer cell line MDA-MB-231. The number of reads for the reference allele and the variant allele are shown for each alteration. a , BRAF oncogene point mutation. B, point mutation in the TP53 tumor suppressor gene. c , a 1-bp insertion in tumor suppressor NF1. D, sequence coverage for each target exon in breast cancer cell line MDA-MB-231 compared with a normal diploid sample. Targets from several genes with copy-number gains and losses are highlighted. e , comparison of gene-level copy-number alterations as detected by exon capture and copy-number data previously obtained with a high-density single-nucleotide polymorphism (SNP) array (Affymetrix SNP 6.0 platform). Several genes with copy-number gains and losses are highlighted. Copy-number data are highly correlated, with a correlation coefficient of 0.94. alterations, the accumulated sequence coverage for each of 1. Amplified exons present in the tumor should have a exon in the tumor sample was compared with the coverage greater number of relative reads and therefore fall above the obtained for the same exon in the diploid normal control diagonal, whereas deleted exons should have fewer reads (after normalization for global differences in “on-target” and fall below the diagonal. sequence coverage). When tumor and normal reads are Guided by this framework, we determined relative copy- displayed as a scatter plot (normal = x-axis and tumor = number ratios for all targeted exons across the cell line col- y-axis), exons with a neutral copy-number across the 2 sam- lection. An example for the MDA-MB-231 breast cancer cell ples should be distributed along a diagonal with a slope line (compared with a normal diploid sample) is shown in JANUARY 2012 CANCER DISCOVERY    85 Downloaded from cancerdiscovery.aacrjournals.org on July 7, 2021. © 2012 American Association for Cancer Research. Exon coverage in tumor cell line (MDA-MB-231) Exon coverage in tumor cell line (MDA-MB-231) Exon coverage in tumor cell line (MDA-MB-231) Copy-number ratio for MDA-MB-231: Sequencing Copy-number ratio for MDA-MB-231: Sequencing Copy-number ratio for MDA-MB-231: Sequencing Published OnlineFirst November 7, 2011; DOI: 10.1158/2159-8290.CD-11-0184 Wagle et al. research article Figure 1D. In total, 8 genes with amplifications (defined as exons targeted had more than 30× coverage after sequencing mean sequence coverage >3-fold greater than the reference and 1% had no coverage. In one sample (FFPE 9; Table 1), 86% normal) and another 8 with deletions (mean sequence cov- of exons showed more than 30× coverage and 2% had zero erage >3-fold lower than the reference normal) were seen coverage—this sample also had the lowest mean coverage of across the cell lines. Comparison of overall copy-number the group (116×). The tumor purity for 8 samples was greater values derived by sequencing to those obtained from high- than 50%, whereas 2 samples had tumor purities of 20% or less density single-nucleotide polymorphism (SNP) array data (FFPE 2 and FFPE 3; Table 1). (Affymetrix SNP 6.0 platform) demonstrated a robust cor- In total, 155 sequence variants and 14 indels were detected relation at the gene level, with correlation coefficients across the samples. In addition, 2 gene amplifications (>3- ranging from 0.89 to 0.98 (Supplementary Table S7). As fold increase in mean sequence read counts compared with an example, the correlation for the MDA-MB-231 cell line a reference normal sample) and 2 gene-level deletions (3-fold (r = 0.94) is shown in Figure 1E. decrease in mean sequence read counts) were seen. (Summary information for all 10 samples is shown in Supplementary Profiling of a rchival t umor s amples by Massively Table S8; a complete listing of all alterations can be found in Parallel s equencing the Supplementary Appendix.) Having established a robust approach for high-throughput Detection of c linically a ctionable Genomic exon capture and massively parallel sequencing of 137 can- a lterations in FFPe t umor s amples cer genes, we next sought to determine whether this approach might prove useful in the clinical setting. As a proof-of-princi- Next, we developed an initial framework to segregate ple, we characterized a pilot collection of 10 FFPE tumor sam- genetic alterations on the basis of their predicted clinical ples from patients with breast or colon cancer. As was the case utility. Toward this end, we designated 3 categories of al- with the aforementioned cell line experiment, each of the 12 terations. One category, termed “actionable in principle,” in- barcoded samples was evenly represented, with a mean cover- cludes variants that predict tumor sensitivity or resistance to age of 391× (Table 1). There was greater variation in the tumor U.S. Food and Drug Administration (FDA)–approved (tier 1) samples compared with the cell lines, with coverage ranging or experimental therapies (tier 2). Another category contains from 116× to 537×. This variance may reflect differences in prognostic or diagnostic variants. The remaining alterations quality of FFPE-derived input DNA. For 11 samples, 94% of are termed “variants of unclear significance,” which may t able 1. s ummary of sequencing results for FFPe samples Percent of Percent of total target bases PF reads Percent Mean with at least Sample Tumor type Tumor purity, % PF reads in pool selected bases target coverage 30× coverage HAPMAP N/A N/A 9,655,996 7 46 394 96 FFPE 1 Colon 60 11,161,868 8 46 457 96 FFPE 2 Colon 10 8,841,660 7 48 353 94 FFPE 3 Colon 20 13,047,230 10 44 498 96 FFPE 4 Colon 60 10,144,562 8 38 300 95 FFPE 5 Breast 80 16,450,558 12 36 472 95 FFPE 6 Breast 70 15,188,624 11 42 532 96 FFPE 7 Colon 50 8,480,282 6 39 250 94 FFPE 8 Breast 80 15,758,604 12 41 537 96 FFPE 9 Colon 60 3,640,236 3 42 116 86 FFPE 10 Colon 50 14,429,284 11 36 410 96 HT-29 (cell line) Colon N/A 7,519,880 6 53 369 96 Abbreviations: N/A, not available; PF, purity filtered. NOTE: Barcoded and pooled genomic DNA from FFPE tumor samples was subjected to exon capture and sequenced in a single 100-bp paired-end Illumina HiSeq2000 lane. PF sequence reads for each sample are shown; the percent of total PF reads shows the relative representation of each sample within the pool. “Percent selected bases” indicates bases that mapped within 250 bp of a target exon, including both on- and near-target sequence. “Mean target coverage” represents the average number of unique reads in which each base was sequenced. 86 |  CANCER DISCOVERY  JANUARY 2012  www.aacrjournals.org Downloaded from cancerdiscovery.aacrjournals.org on July 7, 2021. © 2012 American Association for Cancer Research. Published OnlineFirst November 7, 2011; DOI: 10.1158/2159-8290.CD-11-0184 Genomic Profiling of FFPE Tumor Samples by Targeted Sequencing research article include biologically important mutations without known alterations included a nonsense mutation in MSH2, which therapeutic implications as well as uncharacterized muta- is diagnostic for hereditary nonpolyposis coli and is a prog- tions in genes with presumed clinical relevance. nostic marker in colon cancer, and a nonsense mutation in We detected biologically or clinically meaningful al- SMAD2, which has been suggested to be associated with ad- terations in all 10 FFPE samples, including the 2 samples vanced disease and decreased survival in colon cancer (25). that contained only 10% to 20% tumor cells. These in- Plausibly actionable amplifications of both FGFR1 and clude known somatic mutations in KRAS, BRAF, PIK3CA, CCND1 were observed in a breast tumor sample (Fig. 2A). and CTNNB1; nonsense mutations in the tumor suppres- In preclinical studies, FGFR1 amplification was shown to sors APC, MSH2, SMAD2, TSC1, and TP53; and a 2-bp predict resistance to hormonal therapy in breast cancer (26) deletion in BRCA1. In particular, 12 of the 155 single-nu- and thus may be considered a candidate tier 1 copy-number cleotide variants and 1 of the 14 indels were deemed plau- event for this FDA-approved indication. Clinical trials are sibly actionable (“actionable in principle” or “prognostic/ currently underway to test FGFR inhibitors against tumors diagnostic”; Table 2). KRAS mutations in colon cancer pre- with amplified or overexpressed FGFR1, making FGFR1 am- dict resistance to cetuximab (12, 13) and exemplify tier 1 plification a tier 2 actionable variant as well. Amplification actionable alterations. In addition, mutations in PIK3CA of CCND1 (which encodes the cyclin D1 cell-cycle regulator) have been shown in some studies to promote resistance has also been suggested to predict resistance to hormonal to cetuximab in patients with colon cancer (13–16) and therapy (27, 28). Moreover, this alteration may predict sen- trastuzumab in breast cancer (17, 18), and therefore may sitivity to cyclin-dependent kinase inhibitors (tier 2 action- conceivably represent tier 1 alterations (although this has able event; ref. 29), as well as overall disease prognosis in not been shown definitively). Multiple tier 2 actionable patients with breast cancer (prognostic alteration; refs. 27, alterations (targeted by drugs currently in clinical devel- 28, 30). Lower-level copy-number alterations (between 2- opment) were also seen, including mutations in PIK3CA and 3-fold relative changes) were observed in several known [phosphoinositide 3-kinase (PI3K) pathway inhibitors; or putative cancer genes, including CDK8, GNAS, MYC, and ref. 19], KRAS (MEK inhibitors; ref. 20), TSC1 (TOR inhibi- SRC. Although these events are most likely to ref lect aneu- tors; ref. 21), BRAF (MAPK pathway inhibitors; ref. 22, 23) ploidy, some may represent higher level copy-number altera- and BRCA1 (PARP inhibitors; ref. 24). Other noteworthy tions in samples with low tumor purity. t able 2. a ctionable or prognostic genomic alterations in 10 FFPe tumor samples Sample Tumor Mean target Actionable in principle Prognostic/ type coverage Tier 1 Tier 2 diagnostic b a FFPE 1 Colon 457 KRAS (Q61H) KRAS (Q61h ) a a FFPE 2 Colon 353 KRAS (G13c ) KRAS (G13c ) a a a FFPE 3 Colon 498 KRAS (G13c ) KRAS (G13c ) MSH2 (r 680*) b a PIK3CA (H1047R) PIK3CA (h 1047r ) FFPE 4 Colon 300 BRAF (D594G) TSC1 (E258*) b d b FFPE 5 Breast 472 CCND1 amp CCND1 amp CCND1 amp d a FGFR1 amp FGFR1 amp a a FFPE 6 Breast 532 BRCA1 (2-bp del) BRCA1 (2-bp del) a a FFPE 7 Colon 250 KRAS (G13D) KRAS (G13D) b a PIK3CA (E545K) PIK3CA (e 545K) b a FFPE 8 Breast 537 PIK3CA (H1047R) PIK3CA (h 1047r ) FFPE 9 Colon 116 SMAD2 (S306*) b a FFPE 10 Colon 410 KRAS (Q61H) KRAS (Q61h ) Abbreviations: amp, amplification; del, deletion. The level of evidence for each actionable alteration is denoted by the following footnotes: Clinically validated and approved alterations (for tier 1 or prognostic/diagnostic) or specifically targeted alterations (for tier 2), shown in bold. Limited clinical evidence. Clinical evidence in a different tumor type only. Preclinical evidence only. JANUARY 2012 CANCER DISCOVERY    87 Downloaded from cancerdiscovery.aacrjournals.org on July 7, 2021. © 2012 American Association for Cancer Research. Published OnlineFirst November 7, 2011; DOI: 10.1158/2159-8290.CD-11-0184 Wagle et al. research article a B Target exon coverage: tumor vs. normal Target exon coverage: tumor vs. normal Copy number: QPCR vs. exon capture Copy number: QPCR vs. exon capture 6000 6000 0.40 0.40 All target exons All target exons FGFR1 FGFR1 0.35 0.35 5000 5000 CCND1 CCND1 MDM4 MDM4 0.30 0.30 FGFR1 FGFR1 4000 CEBPA CEBPA 0.25 0.25 NOTCH1 NOTCH1 NKX2-1 NKX2-1 3000 3000 0.20 0.20 CCND1 CCND1 0.15 0.15 2000 2000 0.10 0.10 1000 1000 0.05 0.05 NOTCH1 NOTCH1 0.00 0.00 0 0 0 0 200200 400400 600600 800800 1000 1000 0 0 1.01.0 2.02.0 3.03.0 4.04.0 5.05.0 6.06.0 7.07.0 Exon coverage in normal cell line (HapMap) Exon coverage in normal cell line (HapMap) Relative copy number by exon capture Relative copy number by exon capture Figure 2. Copy-number alterations in an archival breast cancer sample. a , sequence coverage is shown for each target in the tumor sample compared with a normal diploid sample. Exon targets from several genes with copy-number gains and losses are highlighted. B, copy-number correlation between exon capture and QPCR in sample FFPE 5. Quantitative PCR of FGFR1, CCND1, and NOTCH1 with 3 independent sets of primers was performed and average values for each gene were compared to exon capture copy-number. Examination of 79 pharmacogenomic loci facili- c omparison with an e xisting Mutation Profiling tated inspection of plausibly actionable polymorphisms Platform (Supplementary Table S9). The ERCC2-K751QC allele, as- We next wished to compare the sensitivity and specificity sociated with increased risk of FOLFOX-induced grade of targeted hybrid capture/sequencing to an existing mass 3 or 4 hematologic toxicity (31), was present in 2 samples spectrometric genotyping-based platform because this type (i.e., FFPE 2 and FFPE 9). The UGT1A1-G3156A allele was of approach is currently being used in several clinical and found to be heterozygous in 5 samples but homozygous translational oncology settings (2, 33–35). We thus performed in none of them. This allele is associated with irinote- OncoMap, a mass-spectrometric genotyping technology that can-related neutropenia when present as a homozygous interrogates more than 400 known mutations in 33 cancer event (32). genes. Of the 155 single-nucleotide variants seen by hybrid To validate these findings, a representative subset of capture/sequencing of the FFPE samples described previously, alterations (31 nonsynonymous variants and 2 indels; sam- 13 were also interrogated by assays present in OncoMap (Table ples 4–7) were independently queried by mass spectromet- 3). However, when OncoMap was performed on these samples, ric genotyping (2, 3). All 31 single-nucleotide variants and only 10 of these 13 mutations were detected. To determine 2 indels tested were confirmed, demonstrating 100% spec- the basis for this discrepancy, we assayed all 13 mutations by ificity of the targeted exon capture approach in the small an orthogonal genotyping approach that uses distinct reagent subset examined. Copy-number alterations involving 3 chemistry (hME genotyping; see Methods). All 13 mutations genes that were amplified or deleted in sample FFPE 5 were confirmed by this orthogonal genotyping method, sug- (FGFR1, CCND1, NOTCH1) were also tested by quantita- gesting that the 3 mutations not detected by OncoMap were tive PCR with the use of 3 independent primer pairs for false-negative results by mass spectrometric genotyping (Table each gene. As shown in Fig. 2B, the quantitative PCR re- 3; shown in bold). All mutations seen by OncoMap were also sults were highly correlated to the copy-number ratios detected by targeted exon capture. detected by targeted exon capture/sequencing in FFPE 5 2 2 (r = 0.94). The correlation coefficient (r ) for these same d iscussion genes in sample FFPE 9—which has a 2.3-fold amplifica- tion of FGFR1 but no copy-number changes in CCND1 or We have developed a targeted, massively parallel sequenc- NOTCH1—was 0.99 (Supplementary Fig. S3). ing platform to detect actionable genomic alterations in 88 |  CANCER DISCOVERY  JANUARY 2012  www.aacrjournals.org Downloaded from cancerdiscovery.aacrjournals.org on July 7, 2021. © 2012 American Association for Cancer Research. Exon coverage in tumor (FFPE 5) Exon coverage in tumor (FFPE 5) QPCR concentration (ng/mL) QPCR concentration (ng/mL) Published OnlineFirst November 7, 2011; DOI: 10.1158/2159-8290.CD-11-0184 Genomic Profiling of FFPE Tumor Samples by Targeted Sequencing research article t able 3. c omparison of OncoMap and targeted exon capture profiling in FFPe samples Seen via Seen via Validated FFPE Sample Sample type Gene Mutation OncoMap exon capture by hMe FFPE 1 Colon KRAS Q61H Yes Yes Yes FFPE 2 Colon KRAS G13C Yes Yes Yes FFPE 3 Colon KRAS G13D Yes Yes Yes FFPE 3 Colon PIK3CA H1047R Yes Yes Yes FFPE 4 Colon BRAF D594G No Yes Yes FFPE 4 Colon APC Q1367* No Yes Yes FFPE 4 Colon APC Q1378* Yes Yes Yes FFPE 6 Breast TP53 R248Q Yes Yes Yes FFPE 7 Colon KRAS G13D Yes Yes Yes FFPE 7 Colon PIK3CA E545K Yes Yes Yes FFPE 7 Colon CTNNB1 S45F No Yes Yes FFPE 8 Breast PIK3CA H1047R Yes Yes Yes FFPE 10 Colon KRAS Q61H Yes Yes Yes NOTE: Mutations that were not detected by OncoMap are shown in bold. clinical tumor samples. In this initial proof-of-concept effort, were seen by sequencing at multiple loci (including KRAS). we sequenced 137 cancer genes from 10 pooled FFPE tumor The OncoMap approach involves iPLEX genotyping of >500 DNA samples (plus 2 control samples) and achieved 391× mutations followed by hME validation of all candidates (see mean coverage per sample within a single paired-end se- Methods)—the iPLEX method allows increased multiplexing, quencing lane. This depth of coverage afforded robust, simul- but in our hands has proved somewhat less sensitive than taneous detection of base mutations, indels, amplifications, hME genotyping. The fact that all 13 mutations were subse- and deletions. Thus, targeted massively parallel sequencing quently confirmed by hMe chemistry suggests that massively provides a unifying approach for detection of multiple cat- parallel sequencing to several hundred-fold mean coverage egories of actionable genetic alterations. affords enhanced sensitivity compared to mass spectrometric In our pilot study, all of the tumor samples profiled genotyping. Moreover, most alterations found by sequencing contained biologically or clinically meaningful genomic are not assayed by genotyping or allele-specific PCR-based alterations, including several that might predict sensitivity mutation profiling platforms. Thus, the sequencing-based or resistance to targeted agents or provide useful prognos- approach may uncover more actionable options for patients tic information. In particular, 15 alterations (at least one than allele-specific approaches. per sample) were plausibly actionable, and might thus be Hybrid selection approaches have been widely used to predicted to impact clinical decision-making or clinical trial promote gene discovery by reducing genome complexity enrollment if identified as part of an experimental thera- before sequencing (5). In this study, we adapted this tech- peutics or phase I trial program. Several actionable somatic nique to capture a highly restricted genomic territory com- alterations (KRAS, PIK3CA, and MSH2) were detected in posed of 137 known cancer genes and 400,000 coding bases. samples with tumor purity as low as 10% to 20%, highlight- This afforded an expanded depth of coverage (to >400- ing the utility of this approach in “real-world” clinical tu- fold) while also enabling multiple barcoded samples to be mor samples. pooled within a single sequencing lane, thereby increasing Comparison with OncoMap, a mass spectrometric ge- throughput and lowering costs. We previously used a simi- notyping platform in current translational use, confirmed lar approach to characterize a frozen tumor sample from robust performance of targeted massively parallel sequenc- a patient with metastatic melanoma who developed resis- ing, even when applied to FFPE tumor specimens. In our tance to the RAF-inhibitor vemurafenib, and identified an previous study, the sensitivity and specificity of OncoMap activating mutation in MEK1 that caused resistance to RAF- in FFPE tissue was 89.3% and 99.4%, respectively, based on and MEK-inhibition (36). Here, we have adapted the ap- a focused comparison with massively parallel sequencing of proach to capture and sequence multiple barcoded samples KRAS (codon 12) in 93 FFPE samples. In the current study, and to identify distinct categories of genomic alterations OncoMap detected 10 of 13 mutations (79% sensitivity) that simultaneously. JANUARY 2012 CANCER DISCOVERY    89 Downloaded from cancerdiscovery.aacrjournals.org on July 7, 2021. © 2012 American Association for Cancer Research. Published OnlineFirst November 7, 2011; DOI: 10.1158/2159-8290.CD-11-0184 Wagle et al. research article An advantage of solution-phase hybrid capture is that Emerging frameworks for clinical interpretation of redesign and synthesis of long oligonucleotides for bait gen- genome sequencing data typically categorize alterations eration is a straightforward process that may be performed based on “actionability” or prognostic utility. Potentially iteratively until an optimal set of baits has been developed. actionable alterations may be further subdivided depending Thus, prioritized genomic regions can be readily amended on the level of evidence about a particular alteration, rang- as new knowledge of cancer gene mutations becomes avail- ing from those with established therapies to others with able. Furthermore, DNA barcoding and pooling decreases sound preclinical evidence. Plausibly actionable alterations the sequencing cost per sample in a manner proportional to may also include those for which the predictive implica- the number of pooled samples present within a sequencing tions within a particular cancer type are not known (e.g., lane. Achieving deep sequencing coverage increases the sen- BRAF mutations in lung cancer), or for which there is no sitivity of mutation detection—particularly in the setting of established clinical proof of concept (e.g., RET mutations high stromal admixture, which can pervade clinical tumor in lung cancer) even though a particular therapy against tissue. As such, this study extends earlier barcoding and hy- the target (sorafenib) may be commercially available. This brid capture/sequencing efforts (36–49) by identifying mul- category may also include mutations in tumor suppressor tiple types of actionable somatic alterations in archival (i.e., genes (e.g., PTEN) hypothesized to predict vulnerability to FFPE) tumor specimens. Because most clinical samples are targeted agents (e.g., PI3K inhibitors). stored as FFPE material, this approach may prove suitable More than 160 variants of unclear significance were for many translational and clinical applications. identified in our sample set. Undoubtedly, many such vari- At the same time, variations in FFPE sample quality may ants represent uncharacterized germline polymorphisms. adversely affect library construction, hybrid selection, or Differentiating somatic from germline alterations is readily sequencing. Potential solutions include the incorporation accomplished by including matched normal samples (36), of additional pre-processing steps to enrich for high-qual- although paired normal material is not always available in ity FFPE DNA, pooling of fewer samples prior to hybrid research settings. Even among alterations that are clearly selection, and/or increasing the overall depth of sequenc- somatic, additional approaches to interpret their potential ing if the starting library complexity is sufficiently high significance and communicate the results to clinicians and (50). The use of orthogonal technologies such as direct patients will be needed. Development of a rigorous formalism genotyping, quantitative PCR, or FISH to validate action- for clinical interpretation of complex genomic data will likely able alterations may prove useful in the short term because become an active research area, with the goal of enabling op- these techniques are used widely in existing clinical labora- timal, genomics-driven decision making for therapy or clini- tories. However, if the superior sensitivity and specificity is cal trial enrollment. confirmed in independent clinical studies, massively paral- Potential applications for targeted hybrid capture/mas- lel sequencing may become increasingly used in diagnostic sively parallel sequencing in translational and clinical or Clinical Laboratory Improvement Amendments (CLIA) oncology research include both retrospective and prospec- laboratory settings. tive profiling of tumor cohorts. Here, the goal may be to Several additional areas for technical and analytical op- identify predictive and prognostic genes or validate pharma- timization remain. Although we generally achieved robust cogenomic polymorphisms. Ultimately, similar approaches sequence coverage of targeted regions, genomic territory with may be used for prospective genomic profiling of cancer very high or very low GC content presents certain challenges. patients to guide clinical decision making. Toward this end, Options to improve coverage of these regions include rede- the potential turnaround time for the current approach is sign or inclusion of additional baits targeting regions that are 2 weeks. Emerging sequencing instruments promise vast difficult to capture. On the analytical side, detection of lon- reductions in turnaround time. Cost, a significant consider- ger indels (such as the 9-bp PIK3CA deletion in the NCI-H69 ation in clinical sequencing, can also be reduced dramatically cell line) remains difficult with current algorithms. Because by sample pooling. Indeed, it is likely that a combination actionable indels occur in multiple genes, including EGFR, of multiplexing together with falling sequencing costs may ERBB2, and KIT, supplemental assays may be needed to ensure ultimately eliminate cost as a limiting barrier to sequencing sensitive indel detection. Moreover, exon-directed capture data generation. approaches do not detect clinically relevant gene rearrange- In conclusion, the results described herein suggest that ments such as those involving ALK, ABL, and PDGFR. One targeted, massively parallel sequencing offers a promis- potential strategy to detect known rearrangements would ing method to detect actionable genetic alterations across a involve design of baits tiled across common translocation large panel of cancer genes in the clinical diagnostic arena. breakpoints. Furthermore, whereas both amplifications and If widely deployed, such implementation may open new op- deletions could be detected in cell line DNA, such events were portunities to link cancer genomics with molecular features, only observed in a single FFPE sample, which had 80% tumor clinical outcomes, and treatment response in a manner that purity. Detection of copy-number aberrations by targeted se- empowers multiple directions in molecular cancer epidemiol- quencing may be more problematic in samples with signifi- ogy. In addition, this approach may ultimately impact clini- cant stromal contamination. Future analytical methods that cal practice by offering a categorical means to identify genetic incorporate allelic information to infer tumor purity may en- changes affecting genes and pathways targeted by existing hance detection of copy gains and losses in samples with vari- and emerging drugs, thereby speeding the advent of personal- able tumor purity. ized cancer medicine. 90 |  CANCER DISCOVERY  JANUARY 2012  www.aacrjournals.org Downloaded from cancerdiscovery.aacrjournals.org on July 7, 2021. © 2012 American Association for Cancer Research. Published OnlineFirst November 7, 2011; DOI: 10.1158/2159-8290.CD-11-0184 Genomic Profiling of FFPE Tumor Samples by Targeted Sequencing research article existing databases including the Catalogue of Somatic Mutations Methods in Cancer (10) and The Cancer Genome Atlas (52). In addition, we High-Throughput, Targeted Deep Sequencing: Overview identified 79 pharmacogenomic polymorphisms described in the lit- Massively parallel sequencing libraries (Illumina) that contain bar- erature, which might predict sensitivity or resistance to conventional coded universal primers (9) were generated with the use of genomic cancer therapies (Supplementary Table S2). DNA from formalin-fixed, paraffin-embedded tumor material. After preamplification and DNA quantification, equimolar pools were Biotinylated RNA Baits generated consisting of 12 barcoded tumor DNAs. These DNA pools The Agilent SureSelect E-array program was used to design 7,021 were subjected to solution-phase hybrid capture with biotinylated unique RNA baits corresponding to the coding sequence of the 137 RNA baits targeting all exons from 137 actionable cancer genes. genes described previously, as well as to the 79 pharmacogenomic Each hybrid capture reaction was sequenced in a single paired-end polymorphisms and to 24 SNPs for fingerprinting. Target loci were lane of an Illumina flow cell. Subsequently, the sequencing data covered with a tiling density of ×2. Baits were replicated 8 times were deconvoluted to match all high-quality barcoded reads with on the 55,000-bait library array. The sequences of all 7,021 baits are the corresponding tumor samples, and genomic alterations (single- listed in the Supplementary Appendix. Biotinylated RNA baits were nucleotide sequence variants, small insertions/deletions, and DNA synthesized by Agilent for the SureSelect Target Enrichment system. copy-number alterations) were identified. The approach is illustrated schematically in Supplementary Figure S1. Pooling and Hybrid Capture DNA libraries were pooled by mixing 300 ng of each library in a Tumor Tissue and Cell Line DNA single 1.5-mL polypropylene sample tube, lyophilizing by the use Discarded and de-identified tumor specimens were obtained of a speedvac evaporator, and resuspending in 4 μL of nuclease-free from the Cooperative Human Tissue Network. An exemption from water. This entire amount (3,600 ng DNA in 4 μL) was used for hy- the Institutional Review Board was obtained for all samples from brid selection. Solution-phase hybrid capture was performed as pre- the Dana-Farber/Partners Cancer Care Office for the Protection of viously described (51) with 3 modifications to the hybrid selection Research Subjects (Protocol 10-380). Genomic DNA was extracted step (Basic Protocol 3). First, instead of 1.5 μL of Blocking Oligo 2.0, from tumor tissue using methods previously described (2). Cell line 0.125 μL of each of 12 additional 200 μM blocking oligonucleotides genomic DNA was purchased directly from the American Type Culture with sequences complementary to the barcodes were added to the hy- Collection (ATCC). Authentication of cell line genomic DNA was per- bridization reaction (see the Supplementary Methods for sequences). formed by ATCC by the use of short tandem repeat profiling, which Second, the biotinylated oligonucleotide baits were diluted 1:8 with uses multiplex PCR to simultaneously amplify the amelogenin gene nuclease-free water from a concentration of 100 ng/μL to 12.5 ng/ and 8 of the most informative polymorphic markers in the human μL immediately before hybridization and 5 μL of this solution was genome. Control genomic DNA was from the HapMap consortium, added to the hybridization reaction. which was purchased from the Coriell Institute for Medical Research. The final volume of the hybridization reaction was 19 μL, consist- ing of the following components: 4 μL of pooled DNA libraries, 2.5 Barcoded Genomic DNA Library Construction μL of 1.0 mg/mL human Cot-1 DNA, 2.5 μL of 10.0 mg/mL salmon Genomic DNA was quantified by the use of Quant-iT PicoGreen® sperm DNA, 1.5 μL of 200 μM blocking oligo 1.0, 1.5 μL of total of dsDNA Assay Kit (Invitrogen, Carlsbad, CA). A total of 1 μg of ge- the twelve 200 μM blocking oligonucleotides, 5.0 μL of 12.5 ng/μL nomic DNA from each sample was sheared by sonication with the biotinylated oligonucleotide baits, 1.0 μL of 20 U/μL Superase-In following conditions: duty cycle 10%, intensity 5, cycles per burst RNAse inhibitor, and 1 μL of nuclease-free water. Third, during PCR 200, and 135 seconds (Covaris S2 instrument). Paired-end adapt- enrichment of the captured DNA (“the catch”), PCR was performed ers for massively parallel sequencing (Illumina) were added as pre- with primers P5 (5′-AAT GAT ACG GCG ACC ACC GA-3′) and P7 viously described (51), with the following modifications to the (5′-CAA GCA GAA GAC GGC ATA CGA-3′), both at 100 μM, in- paired end library preparation step (basic protocol 2). First, the stead of PCR primers PE1.0 and PE2.0. PCR conditions remained as multiplex adapter provided with the Multiplex Paired-End Library described. All custom primers were obtained from Integrated DNA Sample Preparation Kit (Illumina) was used instead of the standard Technologies (IDT). paired-end adapter. Second, PCR enrichment was conducted in 150 μL of total volume with 3 primers from the Multiplexing Sample Sequencing and Analysis Preparation Oligonucleotide Kit (Illumina). Each PCR enrichment We sequenced 100 bases from both ends of library DNA fragments reaction contained 75 μL of Phusion polymerase (Finnzymes), 3 μL by using an Illumina HiSeq 2000 instrument. The sequence reads of Multiplexing PE Primer 1.0 (25 μM), 3 μL of Multiplexing PE were aligned to human reference genome hg18 with the Burrows- Primer 2.0 (0.5 μM), 3 μL of an Index primer (25 μM), 36 μL of Wheeler Alignment tool (53) with use of the following parameters: –q paired-end library, and 30 μL of nuclease-free water. Samples were 5 –l 32 –k 2 –o 1. Artifactual duplicate read pairs were removed with denatured for 5 minutes at 95°C; 18 cycles of 10 seconds at 95°C, Picard tools (picard.sourceforge.net). An average of 450 megabases of 30 seconds at 65°C, and 30 seconds at 72°C; and a final 5 minutes aligned sequence was generated for each library. at 72°C before cooling to 4°C. PCR primers were removed by using Single-nucleotide variants and small insertions/deletions were ×1.8 volume of Agencourt AMPure PCR Purification Kit (Agencourt identified by the use of algorithms from the Genome Analysis Toolkit Bioscience Corporation). developed at the Broad Institute (54). A local multiple sequence align- ment was performed on intervals suspected to harbor indels to de- Selection of Targeted Genes rive the most probable underlying genomic structure of the query We identified 137 genes that are biologically or clinically relevant sample. Single-nucleotide variants were called separately on each sam- in cancer, including targets of new and existing therapies, genes that ple with UnifiedGenotyper and annotated with GenomicAnnotator. predict sensitivity or resistance to therapies, genes that are prog- Variants were discarded if they were present in dbSNP and not in the nostic markers, and oncogenes and tumor suppressors that are COSMIC database (10), they exhibited an unfavorable strand balance known to undergo recurrent somatic genomic alterations in cancer score (> —20), or they were detected in the HapMap normal control. (Supplementary Table S1). These genes were identified by mining Novel recurrent single-nucleotide variants were manually reviewed JANUARY 2012 CANCER DISCOVERY    91 Downloaded from cancerdiscovery.aacrjournals.org on July 7, 2021. © 2012 American Association for Cancer Research. Published OnlineFirst November 7, 2011; DOI: 10.1158/2159-8290.CD-11-0184 Wagle et al. research article to eliminate additional systematic artifacts. Indels were called with U24CA143867 (M. Meyerson), the Snyder Medical Foundation (W.C. IndelGenotyperV2 and were retained if they occurred in protein-cod- Hahn), and the Starr Cancer Consortium (M.F. Berger, L.A. Garraway). ing exons and on both DNA strands, in <2% of reads in the HapMap normal control, and were absent from dbSNP. Received July 25, 2011; revised October 24, 2011; accepted To calculate relative copy-number levels of the 137 target gene November 4, 2011; published OnlineFirst November 7, 2011. loci, we computed the mean sequence coverage for each gene across all protein-coding exons by using the DepthOfCoverage tool in the Genome Analysis Toolkit. All bases in reads with mapping quality r e Fere Nces <5 were ignored, as were any additional bases with base quality <5. Gene-level coverage in each tumor was normalized by the gene-level 1. Macconaill LE, Garraway LA. Clinical implications of the cancer genome. J Clin Oncol 2010;28:5219–28. coverage for an indexed HapMap diploid cell line included in the 2. MacConaill LE, Campbell CD, Kehoe SM, Bass AJ, Hatton C, Niu same pooled hybrid selection experiment (after adjusting for dif- L, et al. Profiling critical cancer gene mutations in clinical tumor ferences in the overall amount of aligned sequence per sample). samples. PLoS One 2009;4:e7887. Sequence-derived estimates of copy number were then compared to 3. Thomas RK, Baker AC, Debiasi RM, Winckler W, Laframboise T, Lin SNP array-derived estimates of copy number for the cancer cell lines. WM, et al. High-throughput oncogene mutation profiling in human cancer. Nat Genet 2007;39:347–51. Mass Spectrometric Genotyping 4. Dias-Santagata D, Akhavanfard S, David SS, Vernovsky K, Kuhlmann Mass spectrometric genotyping was performed with the OncoMap G, Boisvert SL, et al. Rapid targeted mutational analysis of human 3.0 platform as previously described (2, 3) with iPLEX chemistry for tumours: a clinical platform to guide personalized cancer medicine. initial mutation profiling and validation by multibase hME exten- EMBO Mol Med 2010;2:146–58. sion chemistry. Genomic DNA from all tumor samples was quanti- 5. Meyerson M, Gabriel S, Getz G. Advances in understanding cancer fied using Quant-iT PicoGreen® dsDNA Assay Kit (Invitrogen). genomes through second-generation sequencing. Nat Rev Genet To validate alterations detected by massively parallel sequenc- 2010;11:685–96. ing that were not included in the OncoMap assay collection, base 6. Stratton MR, Campbell PJ, Futreal PA. The cancer genome. Nature substitutions and indels were queried using multi-base hME exten- 2009;458:719–24. 7. Gnirke A, Melnikov A, Maguire J, Rogov P, LeProust EM, Brockman sion chemistry with plexing of ≤6 assays per pool. Conditions for W, et al. Solution hybrid selection with ultra-long oligonucleotides hME validation were implemented as described previously (2, 3). for massively parallel targeted sequencing. Nat Biotechnol Primers and probes used for hME validation were designed with the 2009;27:182–9. Sequenom MassARRAY Assay Design 3.0 software, applying default 8. Hodges E, Rooks M, Xuan Z, Bhattacharjee A, Benjamin Gordon D, multi-base extension parameters but with the following modifica- Brizuela L, et al. Hybrid selection of discrete genomic intervals on tions: maximum multiplex level input equal to 6; maximum pass it- custom-designed microarrays for massively parallel sequencing. Nat eration base adjusted to 200. Protoc 2009;4:960–74. 9. Craig DW, Pearson JV, Szelinger S, Sekar A, Redman M, Corneveaux Microarray Analysis of Chromosomal Copy Number JJ, et al. Identification of genetic variants using bar-coded Chromosomal copy-number information was obtained from the multiplexed sequencing. Nat Methods 2008;5:887–93. Broad-Novartis Cancer Cell Line Encyclopedia project, which has 10. Forbes SA, Bindal N, Bamford S, Cole C, Kok CY, Beare D, et al. high-density SNP array data from the Affymetrix SNP 6.0 platform COSMIC: mining complete cancer genomes in the Catalogue of for all cancer cell lines profiled in this study (55). Somatic Mutations in Cancer. Nucleic Acids Res 2011;39(Database issue):D945–50. Quantitative PCR Analysis of Chromosomal Copy Number 11. Ogata H, Sato H, Takatsuka J, De Luca LM. Human breast cancer MDA-MB-231 cells fail to express the neurofibromin protein, lack Quantitative PCR was performed with the SYBR Green PCR its type I mRNA isoform and show accumulation of P-MAPK and Master Mix Kit (Applied Biosystems) according to the manufactur- activated Ras. Cancer Lett 2001;172:159–64. er’s instructions. To determine the chromosomal copy number of 12. Allegra CJ, Jessup JM, Somerfield MR, Hamilton SR, Hammond EH, each gene, 3 sets of gene-specific primers were designed to interrogate Hayes DF, et al. American Society of Clinical Oncology provisional the genetic locus. Primers recognizing LINE sequences were used for clinical opinion: testing for KRAS gene mutations in patients reference amplification/normalization as described previously (56). with metastatic colorectal carcinoma to predict response to anti- Primer sequences are provided in the Supplementary Methods. Male epidermal growth factor receptor monoclonal antibody therapy. J genomic DNA (Promega) was included as a standard, and HapMap Clin Oncol 2009;27:2091–6. DNA (Coriell) was used as a normal diploid control. Quantitative 13. Bardelli A, Siena S. Molecular mechanisms of resistance to cetuximab PCRs were performed in triplicate for each sample using an ABI 7300 and panitumumab in colorectal cancer. J Clin Oncol 2010;28:1254–61. instrument, in 25-μL reactions containing 0.5 ng of genomic DNA 14. Sartore-Bianchi A, Di Nicolantonio F, Nichelatti M, Molinari F, De and forward and reverse primers each at a concentration of 600 nM. Dosso S, Saletti P, et al. Multi-determinants analysis of molecular alterations for predicting clinical benefit to EGFR-targeted Disclosure of Potential c onflicts of interest monoclonal antibodies in colorectal cancer. PLoS One 2009;4:e7287. 15. Sartore-Bianchi A, Martini M, Molinari F, Veronese S, Nichelatti M, Consultant/advisory role: Foundation Medicine (N. Wagle, Artale S, et al. PIK3CA mutations in colorectal cancer are associated M.F. Berger, M.J. Davis, M. Meyerson, L.A. Garraway), Novartis with clinical resistance to EGFR-targeted monoclonal antibodies. (W.C. Hahn, M. Meyerson, L.A. Garraway), Daiichi Sankyo (L.A. Cancer Res 2009;69:1851–7. Garraway). Ownership interest: Foundation Medicine (N. Wagle, M. 16. De Roock W, Claes B, Bernasconi D, De Schutter J, Biesmans B, Meyerson, L.A. Garraway). Research support: Novartis (W.C. Hahn, Fountzilas G, et al. Effects of KRAS, BRAF, NRAS, and PIK3CA M. Meyerson, L.A. Garraway). Patents: Laboratory Corporation of mutations on the efficacy of cetuximab plus chemotherapy in America (M. Meyerson). Honoraria: Illumina (M.F. Berger). chemotherapy-refractory metastatic colorectal cancer: a retrospective consortium analysis. Lancet Oncol 2010;11:753–62. a cknowledgments 17. Berns K, Horlings HM, Hennessy BT, Madiredjo M, Hijmans EM, This work was supported by the NIH Director’s New Innovator Beelen K, et al. A functional genetic approach identifies the PI3K Award DP2OD002750 (L.A. Garraway), the National Cancer Institute pathway as a major determinant of trastuzumab resistance in breast R33CA126674 (L.A. Garraway), the National Cancer Institute cancer. Cancer Cell 2007;12:395–402. 92 |  CANCER DISCOVERY  JANUARY 2012  www.aacrjournals.org Downloaded from cancerdiscovery.aacrjournals.org on July 7, 2021. © 2012 American Association for Cancer Research. Published OnlineFirst November 7, 2011; DOI: 10.1158/2159-8290.CD-11-0184 Genomic Profiling of FFPE Tumor Samples by Targeted Sequencing research article 18. Wang L, Zhang Q, Zhang J, Sun S, Guo H, Jia Z, et al. PI3K pathway 37. Bell CJ, Dinwiddie DL, Miller NA, Hateley SL, Ganusova EE, Mudge activation results in low efficacy of both trastuzumab and lapatinib. J, et al. Carrier testing for severe childhood recessive diseases by next- BMC Cancer 2011;11:248. generation sequencing. Sci Transl Med 2011;3:65ra4. 19. Courtney KD, Corcoran RB, Engelman JA. The PI3K pathway as drug 38. Berg JS, Evans JP, Leigh MW, Omran H, Bizon C, Mane K, et al. target in human cancer. J Clin Oncol. 2010;28:1075–83. Next generation massively parallel sequencing of targeted exomes to identify genetic mutations in primary ciliary dyskinesia: implications 20. Duffy A, Kummar S. Targeting mitogen-activated protein kinase kinase (MEK) in solid tumors. Target Oncol 2009;4:267–73. for application to clinical testing. Genet Med 2011;13:218–29. 39. Chmielecki J, Peifer M, Jia P, Socci ND, Hutchinson K, Viale A, et al. 21. Wagner AJ, Malinowska-Kolodziej I, Morgan JA, Qin W, Fletcher CD, Vena N, et al. Clinical activity of mTOR inhibition with sirolimus Targeted next-generation sequencing of DNA regions proximal to a conserved GXGXXG signaling motif enables systematic discovery of in malignant perivascular epithelioid cell tumors: targeting the pathogenic activation of mTORC1 in tumors. J Clin Oncol tyrosine kinase fusions in cancer. Nucleic Acids Res 2010;38:6985–96. 40. Grossmann V, Kohlmann A, Klein HU, Schindela S, Schnittger S, 2010;28:835–40. 22. Paik PK, Arcila ME, Fara M, Sima CS, Miller VA, Kris MG, et al. Dicker F, et al. Targeted next-generation sequencing detects point mutations, insertions, deletions and balanced chromosomal Clinical characteristics of patients with lung adenocarcinomas harboring BRAF mutations. J Clin Oncol 2011;29:2046–51. rearrangements as well as identifies novel leukemia-specific fusion genes in a single procedure. Leukemia 2011;25:671–80. 23. Smalley KS, Xiao M, Villanueva J, Nguyen TK, Flaherty KT, Letrero R, et al. CRAF inhibition induces apoptosis in melanoma cells with 41. Johansson H, Isaksson M, Sorqvist EF, Roos F, Stenberg J, Sjoblom T, et al. Targeted resequencing of candidate genes using selector probes. non-V600E BRAF mutations. Oncogene 2009;28:85–94. 24. Fong PC, Boss DS, Yap TA, Tutt A, Wu P, Mergui-Roelvink M, et al. Nucleic Acids Res 2011;39:e8. 42. Kiialainen A, Karlberg O, Ahlford A, Sigurdsson S, Lindblad-Toh K, Inhibition of poly(ADP-ribose) polymerase in tumors from BRCA mutation carriers. N Engl J Med 2009;361:123–34. Syvanen AC. Performance of microarray and liquid based capture methods for target enrichment for massively parallel sequencing and 25. Xie W, Rimm DL, Lin Y, Shih WJ, Reiss M. Loss of Smad signaling in human colorectal cancer is associated with advanced disease and SNP discovery. PLoS One 2011;6:e16486. 43. Liao GJ, Lun FM, Zheng YW, Chan KC, Leung TY, Lau TK, et al. poor prognosis. Cancer J 2003;9:302–12. 26. Turner N, Pearson A, Sharpe R, Lambros M, Geyer F, Lopez- Targeted massively parallel sequencing of maternal plasma DNA permits efficient and unbiased detection of fetal alleles. Clin Chem Garcia MA, et al. FGFR1 amplification drives endocrine therapy resistance and is a therapeutic target in breast cancer. Cancer Res 2011;57:92–101. 44. Meder B, Haas J, Keller A, Heid C, Just S, Borries A, et al. Targeted 2010;70:2085–94. 27. Bostner J, Ahnstrom Waltersson M, Fornander T, Skoog L, next-generation sequencing for the molecular genetic diagnostics of cardiomyopathies. Circ Cardiovasc Genet 2011;4:110–22. Nordenskjold B, Stal O. Amplification of CCND1 and PAK1 as predictors of recurrence and tamoxifen resistance in postmenopausal 45. Nijman IJ, Mokry M, van Boxtel R, Toonen P, de Bruijn E, Cuppen E. Mutation discovery by targeted genomic enrichment of multiplexed breast cancer. Oncogene 2007;26:6997–7005. 28. Lundgren K, Holm K, Nordenskjold B, Borg A, Landberg G. Gene barcoded samples. Nat Methods 2010;7:913–5. 46. Shearer AE, DeLuca AP, Hildebrand MS, Taylor KR, Gurrola J, 2nd, products of chromosome 11q and their association with CCND1 gene amplification and tamoxifen resistance in premenopausal Scherer S, et al. Comprehensive genetic testing for hereditary hearing loss using massively parallel sequencing. Proc Natl Acad Sci U S A breast cancer. Breast Cancer Res 2008;10:R81. 29. Finn RS, Dering J, Conklin D, Kalous O, Cohen DJ, Desai AJ, et al. 2010;107:21104–9. 47. Summerer D, Schracke N, Wu H, Cheng Y, Bau S, Stahler CF, PD 0332991, a selective cyclin D kinase 4/6 inhibitor, preferentially inhibits proliferation of luminal estrogen receptor-positive human et al. Targeted high throughput sequencing of a cancer-related exome subset by specific sequence capture with a fully automated breast cancer cell lines in vitro. Breast Cancer Res 2009;11:R77. 30. Roy PG, Pratt N, Purdie CA, Baker L, Ashfield A, Quinlan P, et al. High microarray platform. Genomics 2010;95:241–6. 48. Teer JK, Bonnycastle LL, Chines PS, Hansen NF, Aoyama N, Swift CCND1 amplification identifies a group of poor prognosis women with estrogen receptor positive breast cancer. Int J Cancer 2010;127:355–60. AJ, et al. Systematic comparison of three genomic enrichment methods for massively parallel DNA sequencing. Genome Res 31. Boige V, Mendiboure J, Pignon JP, Loriot MA, Castaing M, Barrois M, et al. Pharmacogenetic assessment of toxicity and outcome in patients 2010;20:1420–31. 49. Zang ZJ, Ong CK, Cutcutache I, Yu W, Zhang SL, Huang D, et al. with metastatic colorectal cancer treated with LV5FU2, FOLFOX, and FOLFIRI: FFCD 2000–05. J Clin Oncol 2010;28:2556–64. Genetic and structural variation in the gastric cancer kinome revealed through targeted deep sequencing. Cancer Res 2011;71:29–39. 32. Lara PN, Jr., Natale R, Crowley J, Lenz HJ, Redman MW, Carleton JE, et al. Phase III trial of irinotecan/cisplatin compared with 50. Fisher S, Barry A, Abreu J, Minie B, Nolan J, Delorey TM, et al. A scalable, fully automated process for construction of sequence-ready etoposide/cisplatin in extensive-stage small-cell lung cancer: clinical and pharmacogenomic results from SWOG S0124. J Clin Oncol human exome targeted capture libraries. Genome Biol 2011;12(1):R1. 51. Blumenstiel B, Cibulskis K, Fisher S, DeFelice M, Barry A, Fennell T, 2009;27:2530–5. 33. Arcila M, Lau C, Nafa K, Ladanyi M. Detection of KRAS and BRAF et al. Targeted exon sequencing by in-solution hybrid selection. Curr Protoc Hum Genet 2010;Chapter 18:Unit 18.4. mutations in colorectal carcinoma roles for high-sensitivity locked nucleic acid-PCR sequencing and broad-spectrum mass spectrometry 52. The Cancer Genome Atlas. Available from: cancergenome.nih.gov. 53. Li H, Durbin R. Fast and accurate short read alignment with genotyping. J Mol Diagn 2011;13:64–73. 34. Brevet M, Johnson ML, Azzoli CG, Ladanyi M. Detection of EGFR Burrows-Wheeler transform. Bioinformatics 2009;25:1754–60. 54. McKenna A, Hanna M, Banks E, Sivachenko A, Cibulskis K, mutations in plasma DNA from lung cancer patients by mass spectrometry genotyping is predictive of tumor EGFR status and Kernytsky A, et al. The Genome Analysis Toolkit: a MapReduce framework for analyzing next-generation DNA sequencing data. response to EGFR inhibitors. Lung Cancer 2011;73:96–102. 35. Beadling C, Heinrich MC, Warrick A, Forbes EM, Nelson D, Genome Res 2010;20:1297–303. 55. Broad-Novartis Cancer Cell Line Encylopedia. Available from: www. Justusson E, et al. Multiplex mutation screening by mass spectrometry: evaluation of 820 cases from a personalized cancer broadinstitute.org/ccle. 56. Garraway LA, Widlund HR, Rubin MA, Getz G, Berger AJ, medicine registry. J Mol Diagn. 2011;13:504–13. 36. Wagle N, Emery C, Berger MF, Davis MJ, Sawyer A, Pochanard P, Ramaswamy S, et al. Integrative genomic analyses identify MITF as a lineage survival oncogene amplified in malignant melanoma. Nature et al. Dissecting therapeutic resistance to raf inhibition in melanoma by tumor genomic profiling. J Clin Oncol 2011;29:3085–96. 2005;436:117–22. JANUARY 2012 CANCER DISCOVERY    93 Downloaded from cancerdiscovery.aacrjournals.org on July 7, 2021. © 2012 American Association for Cancer Research. Published OnlineFirst November 7, 2011; DOI: 10.1158/2159-8290.CD-11-0184 High-Throughput Detection of Actionable Genomic Alterations in Clinical Tumor Samples by Targeted, Massively Parallel Sequencing Nikhil Wagle, Michael F. Berger, Matthew J. Davis, et al. Cancer Discov 2012;2:82-93. Published OnlineFirst November 7, 2011. Access the most recent version of this article at: Updated version doi:10.1158/2159-8290.CD-11-0184 Access the most recent supplemental material at: Supplementary http://cancerdiscovery.aacrjournals.org/content/suppl/2011/11/07/2159-8290.CD-11-0184.DC Material This article cites 53 articles, 19 of which you can access for free at: Cited articles http://cancerdiscovery.aacrjournals.org/content/2/1/82.full#ref-list-1 This article has been cited by 90 HighWire-hosted articles. Access the articles at: Citing articles http://cancerdiscovery.aacrjournals.org/content/2/1/82.full#related-urls Sign up to receive free email-alerts related to this article or journal. E-mail alerts To order reprints of this article or to subscribe to the journal, contact the AACR Publications Reprints and Department at [email protected]. Subscriptions To request permission to re-use all or part of this article, use this link Permissions http://cancerdiscovery.aacrjournals.org/content/2/1/82. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site. Downloaded from cancerdiscovery.aacrjournals.org on July 7, 2021. © 2012 American Association for Cancer Research. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Cancer Discovery Unpaywall

High-Throughput Detection of Actionable Genomic Alterations in Clinical Tumor Samples by Targeted, Massively Parallel Sequencing

Download PDF
13 pages

Loading next page...
 
/lp/unpaywall/high-throughput-detection-of-actionable-genomic-alterations-in-GdT3NRHhN4

References (59)

Publisher
Unpaywall
ISSN
2159-8274
DOI
10.1158/2159-8290.cd-11-0184
Publisher site
See Article on Publisher Site

Abstract

Published OnlineFirst November 7, 2011; DOI: 10.1158/2159-8290.CD-11-0184 Wagle et al. research article Resea Rch aR ticle h igh-t hroughput Detection of a ctionable Genomic a lterations in c linical t umor s amples by t argeted, Massively Parallel s equencing 1–3 3 2 3 3 Nikhil Wagle , Michael F. Berger , Matthew J. Davis , Brendan Blumenstiel , Matthew DeFelice , 1 2 2 2 1–3 Panisa Pochanard , Matthew Ducar  , Paul Van Hummelen , Laura E. MacConaill , William C. Hahn , 1–3 3 1–3 Matthew Meyerson , Stacey B. Gabriel , and Levi A. Garraway OF82 |  CANCER DISCOVERY  JANUARY 2012  www.aacrjournals.org Downloaded from cancerdiscovery.aacrjournals.org on July 7, 2021. © 2012 American Association for Cancer Research. T Published OnlineFirst November 7, 2011; DOI: 10.1158/2159-8290.CD-11-0184 Genomic Profiling of F The BATTLE Trial: Personalizing ormalin-Fix Ther ed, Par apy for Lung Cancer affin-Embedded Tumor Samples by Targeted Sequencing research ar research ar ticle ticle Knowledge of “actionable” somatic genomic alterations present in each tumor a Bst Ract (e.g., point mutations, small insertions/deletions, and copy-number alterations that direct therapeutic options) should facilitate individualized approaches to cancer treatment. However, clinical implementation of systematic genomic profiling has rarely been achieved beyond limited numbers of oncogene point mutations. To address this challenge, we utilized a targeted, massively parallel sequencing approach to detect tumor genomic alterations in formalin-fixed, paraffin-embedded (FFPE) tumor samples. Nearly 400-fold mean sequence coverage was achieved, and single-nucleotide sequence variants, small insertions/deletions, and chromosomal copy- number alterations were detected simultaneously with high accuracy compared with other meth- ods in clinical use. Putatively actionable genomic alterations, including those that predict sensitivity or resistance to established and experimental therapies, were detected in each tumor sample tested. Thus, targeted deep sequencing of clinical tumor material may enable mutation- driven clinical trials and, ultimately, “personalized” cancer treatment. si GNiFic a Nce : Despite the rapid proliferation of targeted therapeutic agents, systematic methods to profile clinically relevant tumor genomic alterations remain underdeveloped. We describe a sequencing- based approach to identifying genomic alterations in FFPE tumor samples. These studies affirm the feasi- bility and clinical utility of targeted sequencing in the oncology arena and provide a foundation for genomics-based stratification of cancer patients. Cancer Discovery; 2(1); 82–93. ©2011 AACR. int Roduction clinical setting. Because many targeted agents in devel- The maturation of cancer genome characterization opment are designed to intercept proteins and/or path- efforts has fueled the notion that many treatment deci- ways commonly perturbed by tumor genetic changes, an sions might ultimately be guided by the genetic makeup urgent need exists to implement robust approaches that of individual tumors (1). Moreover, the rapid prolifera- determine the “actionable” genetic profiles of individual tion of targeted agents in development has called specific tumors. If widely obtained, such information might better attention to the importance of molecular profiling ap- identify those patients most likely to respond to existing proaches that pinpoint in situ those tumors most likely and emerging anticancer regimens. to respond. Knowledge of such alterations in the clinical We and others have developed tumor mutation–profiling and translational arenas—including mutations, somatic platforms that use mass-spectrometric genotyping (2, 3) copy-number alterations, and polymorphisms affecting or allele-specific PCR-based technologies (4). Each of these drug metabolism—should ultimately facilitate individual- approaches interrogates known oncogene or tumor sup- ized approaches to cancer treatment. However, systematic pressor gene mutations present in DNA obtained from genetic profiling of cancers remains underdeveloped in the either frozen or formalin-fixed, paraffin-embedded (FFPE) tumor tissue. However, genotyping-based platforms have certain limitations that may preclude their applicability as 1 2 Authors’ Affiliations: Department of Medical Oncology and Center for definitive cancer diagnostic modalities. These include the Cancer Genome Discovery, Dana-Farber Cancer Institute, Harvard Medical finite number of prespecified point mutations that can School, Boston; Broad Institute of MIT and Harvard, Cambridge, be assayed (designated a priori from a restricted subset of Massachusetts known cancer genes), difficulties in detecting small inser- Note: N. Wagle and M.F. Berger contributed equally to this work; M.F. tions or deletions (“indels”), insensitivity to most tumor Berger is currently in the Department of Pathology, Memorial Sloan- suppressor gene mutations (which may occur anywhere Kettering Cancer Center, New York, New York. within the gene), inability to detect gene amplifications or Note: Supplementary data for this article are available at Cancer deletions, and decreased sensitivity in tumor samples with Discovery Online (http://www.cancerdiscovery.aacrjournals.org). high stromal admixture. At the present time, no systematic Corresponding Author: Levi A. Garraway, Department of Medical mechanism exists whereby clinical tumor specimens might Oncology, Dana-Farber Cancer Institute, 450 Brookline Avenue, D1542, Boston, MA 02215. Phone: 617-632-6689; Fax: 617-582-7880; E-mail: be interrogated in situ for a fully comprehensive panel of [email protected] actionable cancer gene alterations. doi: 10.1158/2159-8290.CD-11-0184 The advent of massively parallel sequencing is trans- ©2011 American Association for Cancer Research. forming the cancer genomics landscape by enabling JANUARY 2012 CANCER DISCOVERY    83 Downloaded from cancerdiscovery.aacrjournals.org on July 7, 2021. © 2012 American Association for Cancer Research. Published OnlineFirst November 7, 2011; DOI: 10.1158/2159-8290.CD-11-0184 Wagle et al. research article comprehensive cancer genome characterization at an selection reaction and sequenced in a single Illumina lane unprecedented scope (1, 5, 6). Concomitantly, hybrid se- with 100-bp paired end reads. lection-based methods that enrich for coding sequences The 11 equimolar DNAs were evenly represented, with prior to sequencing (“exon capture”; refs. 7, 8) are rou- 12 to 17 million purity-filtered reads generated per sample tinely being implemented in discovery-oriented settings (average of 14.6 million purity-filtered reads; Supplementary (5). Here, we describe an adaptation of exon capture and Table S4), whereas the sample present at 50% concentration massively parallel sequencing for robust detection of so- (HT-29, index 2) had 7.8 million purity-filtered reads, as matic genomic alterations in FFPE samples. The approach expected. The percent of bases mapping “on-target” averaged leverages a targeted exon capture technique to enrich 60% (range, 56%–64%) across all samples in the pool, yielding for a cancer-relevant genomic territory consisting of 137 a mean 527× target coverage (range, 441× to 593×) for the genes ( 400,000 coding bases), thereby allowing multiple 11 equimolar samples. More than 95% of target exons barcoded samples to be pooled into a single sequencing exhibited more than 30× coverage after sequencing (suf- reaction while preserving deep (e.g., >300- to 400-fold) ficient to call “high-confidence” variants in a sample with sequencing coverage of targeted regions. This approach 70%–80% tumor purity), whereas only 1% had no coverage simultaneously identifies mutations and chromosomal (Supplementary Fig. S2A and B). In general, poorly captured copy-number alterations in clinical tumor material and exons had greater than 70% GC content, although GC con- may inform a comprehensive means to achieve DNA-based tent did not account for all of the poorly captured targets patient stratification in the clinical and translational on- (Supplementary Fig. S2C). The capture performance for a cology arena. particular target exon was highly reproducible from sample to sample (Supplementary Figure S2D–F). Results Detection of s ingle-Nucleotide Variants, We generated a list of 137 “druggable” or potentially insertions/Deletions, and c opy-Number actionable genes known to undergo somatic genomic a lterations alterations in cancer (Supplementary Table S1). These in- clude targets of existing and novel therapeutics, prognostic In total, 102 single-nucleotide variants and 6 indels markers, and other oncogenes and tumor suppressors that (excluding known germline polymorphisms) were detected are frequently mutated in cancer. In addition, we included in coding sequences across the 10 cell lines, including all 79 pharmacogenomic polymorphisms in 34 genes that 21 single-nucleotide variants and 3 of 4 indels reported for may predict heightened sensitivity/resistance or toxic- these lines in the Catalogue of Somatic Mutations in Cancer ity to conventional cancer therapies (Supplementary Table database (COSMIC) (Supplementary Table S5; ref. 10). The S2). Altogether, these genes comprise 2,372 exons encod- single indel that was not initially identified—a 9-bp deletion ing 433,159 bases. We then designed and synthesized 7,021 in PIK3CA in the NCI-H69 cell line—was readily detected by unique biotinylated RNA baits corresponding to these ge- manual inspection of the raw sequencing data. Therefore, all nomic regions. previously reported point mutations and indels for this small We leveraged a solution-based exon capture/massively par- collection were detectable by this approach. (A complete list- allel sequencing approach in which a pool of long oligonucle- ing of all alterations identified in these cell lines can be found otides complimentary to these exons of interest were used to in the Supplementary Appendix.) reduce the complexity of tumor genomic DNA for clinically- In the absence of paired normal samples, the majority oriented sequencing. Here, a 6-nucleotide DNA barcode was of variants detected are germline alterations. Nonetheless, appended to the ends of DNA fragments during library con- previously unreported variants were still informative in sev- struction, thus allowing multiple samples to be pooled before eral instances. For example, 12 single-nucleotide variants hybrid selection to expand the scope of genomic profiling (9). were detected in the breast cancer cell line MDA-MB-231, The approach is illustrated schematically in Supplementary including all 4 alterations in the COSMIC database (BRAF, Figure S1. TP53, KRAS, and NF2; Fig. 1A and B, Supplementary Table S6; ref. 10). One of the additional alterations was a 1-bp c apture Performance and r eproducibility frameshift insertion involving the NF1 tumor suppressor We first optimized the approach by using genomic predicted to generate a truncated protein product (Fig. DNA from normal samples and tumor cell lines known 1C). This NF1 insertion likely represents a bona-fide cancer- to harbor mutations and/or chromosomal copy-number associated mutation. The MDA-MB-231 cell line has pre- alterations affecting multiple cancer genes represented in viously been shown to lack both an NF1 mRNA isoform our hybrid capture baits. Ten cancer cell lines with well- and the neurofibromin protein (the product of NF1); thus, characterized, mutually exclusive cancer gene mutations these findings may provide a genetic basis for neurofibro- were chosen (Supplementary Table S3) as well as control min loss in this setting (11). diploid genomic DNA. Equimolar amounts of the result- Although detection of point mutations and indels by tar- ing sequencing libraries were pooled together with an ad- geted, massively parallel sequencing has become increasingly ditional library from the HT-29 cell line, which was added common, the simultaneous detection of chromosomal copy- at a 50% molar ratio compared with the other libraries. number alterations by this approach is less well-established, This pool of 12 libraries was subjected to a single hybrid particularly in the clinical arena. To determine copy-number 84 |  CANCER DISCOVERY  JANUARY 2012  www.aacrjournals.org Downloaded from cancerdiscovery.aacrjournals.org on July 7, 2021. © 2012 American Association for Cancer Research. Published OnlineFirst November 7, 2011; DOI: 10.1158/2159-8290.CD-11-0184 Genomic Profiling of FFPE Tumor Samples by Targeted Sequencing research article chr7:140127873-140127899 chr7:140127873-140127899 chr7:7517811-7517837 chr7:7517811-7517837 chr17:26565586-26565612 chr17:26565586-26565612 a B c chr7:140127873-140127899 chr7:7517811-7517837 chr17:26565586-26565612 G464V (C>A) G464V (C>A) R280K (C>T) R280K (C>T) T467Hfs*3 (ins C) T467Hfs*3 (ins C) Reference: 310 Reference: 310 Reference: 1 Reference: 1 Reference: 0 Reference: 0 G464V (C>A) R280K (C>T) T467Hfs*3 (ins C) Variant: 391 Variant: 391 Variant: 768 Variant: 768 Variant: 67 Variant: 67 Reference: 310 Reference: 1 Reference: 0 Variant: 391 Variant: 768 Variant: 67 BRAF BRAF TP53 TP53 NF1 NF1 (1-bp insertion) (1-bp insertion) BRAF TP53 NF1 (1-bp insertion) D Target exon coverage: tumor vs. normal Target exon coverage: tumor vs. normal e Copy-number ratio: sequencing vs. array Copy-number ratio: sequencing vs. array Target exon coverage: tumor vs. normal Copy-number ratio: sequencing vs. array 2400 2400 All target exons All target exons All target genes All target genes CDKN1A CDKN1A CDKN1A CDKN1A All target exons All target genes 2.5 2.5 2000 2000 SRC SRC SRC SRC CDKN1A CDKN1A 2.5 NOTCH4 NOTCH4 NOTCH4 NOTCH4 SRC SRC 1600 1600 PTPRD PTPRD PTPRD PTPRD NOTCH4 NOTCH4 JAK2 JAK2 JAK2 JAK2 1600 PTPRD 2 PTPRD CDKN2A CDKN2A CDKN2A CDKN2A JAK2 JAK2 1200 1200 1.5 1.5 CDKN2A CDKN2A 1200 1.5 800 800 11 400 400 0.5 0.5 400 0.5 00 200 200 400 400 600 600 800 800 1000 1000 1200 1200 00 0.5 0.5 11 1.5 1.5 22 2.5 2.5 Exon coverage in normal cell line (HapMap) Exon coverage in normal cell line (HapMap) Copy-number ratio for MDA-MB-231: SNP array Copy-number ratio for MDA-MB-231: SNP array 0 200 400 600 800 1000 1200 0 0.5 1 1.5 2 2.5 Exon coverage in normal cell line (HapMap) Copy-number ratio for MDA-MB-231: SNP array Figure 1. Genomic alterations in breast cancer cell line MDA-MB-231. a –c , representative genome images from the Integrated Genome Viewer (IGV) for several alterations found in the breast cancer cell line MDA-MB-231. The number of reads for the reference allele and the variant allele are shown for each alteration. a , BRAF oncogene point mutation. B, point mutation in the TP53 tumor suppressor gene. c , a 1-bp insertion in tumor suppressor NF1. D, sequence coverage for each target exon in breast cancer cell line MDA-MB-231 compared with a normal diploid sample. Targets from several genes with copy-number gains and losses are highlighted. e , comparison of gene-level copy-number alterations as detected by exon capture and copy-number data previously obtained with a high-density single-nucleotide polymorphism (SNP) array (Affymetrix SNP 6.0 platform). Several genes with copy-number gains and losses are highlighted. Copy-number data are highly correlated, with a correlation coefficient of 0.94. alterations, the accumulated sequence coverage for each of 1. Amplified exons present in the tumor should have a exon in the tumor sample was compared with the coverage greater number of relative reads and therefore fall above the obtained for the same exon in the diploid normal control diagonal, whereas deleted exons should have fewer reads (after normalization for global differences in “on-target” and fall below the diagonal. sequence coverage). When tumor and normal reads are Guided by this framework, we determined relative copy- displayed as a scatter plot (normal = x-axis and tumor = number ratios for all targeted exons across the cell line col- y-axis), exons with a neutral copy-number across the 2 sam- lection. An example for the MDA-MB-231 breast cancer cell ples should be distributed along a diagonal with a slope line (compared with a normal diploid sample) is shown in JANUARY 2012 CANCER DISCOVERY    85 Downloaded from cancerdiscovery.aacrjournals.org on July 7, 2021. © 2012 American Association for Cancer Research. Exon coverage in tumor cell line (MDA-MB-231) Exon coverage in tumor cell line (MDA-MB-231) Exon coverage in tumor cell line (MDA-MB-231) Copy-number ratio for MDA-MB-231: Sequencing Copy-number ratio for MDA-MB-231: Sequencing Copy-number ratio for MDA-MB-231: Sequencing Published OnlineFirst November 7, 2011; DOI: 10.1158/2159-8290.CD-11-0184 Wagle et al. research article Figure 1D. In total, 8 genes with amplifications (defined as exons targeted had more than 30× coverage after sequencing mean sequence coverage >3-fold greater than the reference and 1% had no coverage. In one sample (FFPE 9; Table 1), 86% normal) and another 8 with deletions (mean sequence cov- of exons showed more than 30× coverage and 2% had zero erage >3-fold lower than the reference normal) were seen coverage—this sample also had the lowest mean coverage of across the cell lines. Comparison of overall copy-number the group (116×). The tumor purity for 8 samples was greater values derived by sequencing to those obtained from high- than 50%, whereas 2 samples had tumor purities of 20% or less density single-nucleotide polymorphism (SNP) array data (FFPE 2 and FFPE 3; Table 1). (Affymetrix SNP 6.0 platform) demonstrated a robust cor- In total, 155 sequence variants and 14 indels were detected relation at the gene level, with correlation coefficients across the samples. In addition, 2 gene amplifications (>3- ranging from 0.89 to 0.98 (Supplementary Table S7). As fold increase in mean sequence read counts compared with an example, the correlation for the MDA-MB-231 cell line a reference normal sample) and 2 gene-level deletions (3-fold (r = 0.94) is shown in Figure 1E. decrease in mean sequence read counts) were seen. (Summary information for all 10 samples is shown in Supplementary Profiling of a rchival t umor s amples by Massively Table S8; a complete listing of all alterations can be found in Parallel s equencing the Supplementary Appendix.) Having established a robust approach for high-throughput Detection of c linically a ctionable Genomic exon capture and massively parallel sequencing of 137 can- a lterations in FFPe t umor s amples cer genes, we next sought to determine whether this approach might prove useful in the clinical setting. As a proof-of-princi- Next, we developed an initial framework to segregate ple, we characterized a pilot collection of 10 FFPE tumor sam- genetic alterations on the basis of their predicted clinical ples from patients with breast or colon cancer. As was the case utility. Toward this end, we designated 3 categories of al- with the aforementioned cell line experiment, each of the 12 terations. One category, termed “actionable in principle,” in- barcoded samples was evenly represented, with a mean cover- cludes variants that predict tumor sensitivity or resistance to age of 391× (Table 1). There was greater variation in the tumor U.S. Food and Drug Administration (FDA)–approved (tier 1) samples compared with the cell lines, with coverage ranging or experimental therapies (tier 2). Another category contains from 116× to 537×. This variance may reflect differences in prognostic or diagnostic variants. The remaining alterations quality of FFPE-derived input DNA. For 11 samples, 94% of are termed “variants of unclear significance,” which may t able 1. s ummary of sequencing results for FFPe samples Percent of Percent of total target bases PF reads Percent Mean with at least Sample Tumor type Tumor purity, % PF reads in pool selected bases target coverage 30× coverage HAPMAP N/A N/A 9,655,996 7 46 394 96 FFPE 1 Colon 60 11,161,868 8 46 457 96 FFPE 2 Colon 10 8,841,660 7 48 353 94 FFPE 3 Colon 20 13,047,230 10 44 498 96 FFPE 4 Colon 60 10,144,562 8 38 300 95 FFPE 5 Breast 80 16,450,558 12 36 472 95 FFPE 6 Breast 70 15,188,624 11 42 532 96 FFPE 7 Colon 50 8,480,282 6 39 250 94 FFPE 8 Breast 80 15,758,604 12 41 537 96 FFPE 9 Colon 60 3,640,236 3 42 116 86 FFPE 10 Colon 50 14,429,284 11 36 410 96 HT-29 (cell line) Colon N/A 7,519,880 6 53 369 96 Abbreviations: N/A, not available; PF, purity filtered. NOTE: Barcoded and pooled genomic DNA from FFPE tumor samples was subjected to exon capture and sequenced in a single 100-bp paired-end Illumina HiSeq2000 lane. PF sequence reads for each sample are shown; the percent of total PF reads shows the relative representation of each sample within the pool. “Percent selected bases” indicates bases that mapped within 250 bp of a target exon, including both on- and near-target sequence. “Mean target coverage” represents the average number of unique reads in which each base was sequenced. 86 |  CANCER DISCOVERY  JANUARY 2012  www.aacrjournals.org Downloaded from cancerdiscovery.aacrjournals.org on July 7, 2021. © 2012 American Association for Cancer Research. Published OnlineFirst November 7, 2011; DOI: 10.1158/2159-8290.CD-11-0184 Genomic Profiling of FFPE Tumor Samples by Targeted Sequencing research article include biologically important mutations without known alterations included a nonsense mutation in MSH2, which therapeutic implications as well as uncharacterized muta- is diagnostic for hereditary nonpolyposis coli and is a prog- tions in genes with presumed clinical relevance. nostic marker in colon cancer, and a nonsense mutation in We detected biologically or clinically meaningful al- SMAD2, which has been suggested to be associated with ad- terations in all 10 FFPE samples, including the 2 samples vanced disease and decreased survival in colon cancer (25). that contained only 10% to 20% tumor cells. These in- Plausibly actionable amplifications of both FGFR1 and clude known somatic mutations in KRAS, BRAF, PIK3CA, CCND1 were observed in a breast tumor sample (Fig. 2A). and CTNNB1; nonsense mutations in the tumor suppres- In preclinical studies, FGFR1 amplification was shown to sors APC, MSH2, SMAD2, TSC1, and TP53; and a 2-bp predict resistance to hormonal therapy in breast cancer (26) deletion in BRCA1. In particular, 12 of the 155 single-nu- and thus may be considered a candidate tier 1 copy-number cleotide variants and 1 of the 14 indels were deemed plau- event for this FDA-approved indication. Clinical trials are sibly actionable (“actionable in principle” or “prognostic/ currently underway to test FGFR inhibitors against tumors diagnostic”; Table 2). KRAS mutations in colon cancer pre- with amplified or overexpressed FGFR1, making FGFR1 am- dict resistance to cetuximab (12, 13) and exemplify tier 1 plification a tier 2 actionable variant as well. Amplification actionable alterations. In addition, mutations in PIK3CA of CCND1 (which encodes the cyclin D1 cell-cycle regulator) have been shown in some studies to promote resistance has also been suggested to predict resistance to hormonal to cetuximab in patients with colon cancer (13–16) and therapy (27, 28). Moreover, this alteration may predict sen- trastuzumab in breast cancer (17, 18), and therefore may sitivity to cyclin-dependent kinase inhibitors (tier 2 action- conceivably represent tier 1 alterations (although this has able event; ref. 29), as well as overall disease prognosis in not been shown definitively). Multiple tier 2 actionable patients with breast cancer (prognostic alteration; refs. 27, alterations (targeted by drugs currently in clinical devel- 28, 30). Lower-level copy-number alterations (between 2- opment) were also seen, including mutations in PIK3CA and 3-fold relative changes) were observed in several known [phosphoinositide 3-kinase (PI3K) pathway inhibitors; or putative cancer genes, including CDK8, GNAS, MYC, and ref. 19], KRAS (MEK inhibitors; ref. 20), TSC1 (TOR inhibi- SRC. Although these events are most likely to ref lect aneu- tors; ref. 21), BRAF (MAPK pathway inhibitors; ref. 22, 23) ploidy, some may represent higher level copy-number altera- and BRCA1 (PARP inhibitors; ref. 24). Other noteworthy tions in samples with low tumor purity. t able 2. a ctionable or prognostic genomic alterations in 10 FFPe tumor samples Sample Tumor Mean target Actionable in principle Prognostic/ type coverage Tier 1 Tier 2 diagnostic b a FFPE 1 Colon 457 KRAS (Q61H) KRAS (Q61h ) a a FFPE 2 Colon 353 KRAS (G13c ) KRAS (G13c ) a a a FFPE 3 Colon 498 KRAS (G13c ) KRAS (G13c ) MSH2 (r 680*) b a PIK3CA (H1047R) PIK3CA (h 1047r ) FFPE 4 Colon 300 BRAF (D594G) TSC1 (E258*) b d b FFPE 5 Breast 472 CCND1 amp CCND1 amp CCND1 amp d a FGFR1 amp FGFR1 amp a a FFPE 6 Breast 532 BRCA1 (2-bp del) BRCA1 (2-bp del) a a FFPE 7 Colon 250 KRAS (G13D) KRAS (G13D) b a PIK3CA (E545K) PIK3CA (e 545K) b a FFPE 8 Breast 537 PIK3CA (H1047R) PIK3CA (h 1047r ) FFPE 9 Colon 116 SMAD2 (S306*) b a FFPE 10 Colon 410 KRAS (Q61H) KRAS (Q61h ) Abbreviations: amp, amplification; del, deletion. The level of evidence for each actionable alteration is denoted by the following footnotes: Clinically validated and approved alterations (for tier 1 or prognostic/diagnostic) or specifically targeted alterations (for tier 2), shown in bold. Limited clinical evidence. Clinical evidence in a different tumor type only. Preclinical evidence only. JANUARY 2012 CANCER DISCOVERY    87 Downloaded from cancerdiscovery.aacrjournals.org on July 7, 2021. © 2012 American Association for Cancer Research. Published OnlineFirst November 7, 2011; DOI: 10.1158/2159-8290.CD-11-0184 Wagle et al. research article a B Target exon coverage: tumor vs. normal Target exon coverage: tumor vs. normal Copy number: QPCR vs. exon capture Copy number: QPCR vs. exon capture 6000 6000 0.40 0.40 All target exons All target exons FGFR1 FGFR1 0.35 0.35 5000 5000 CCND1 CCND1 MDM4 MDM4 0.30 0.30 FGFR1 FGFR1 4000 CEBPA CEBPA 0.25 0.25 NOTCH1 NOTCH1 NKX2-1 NKX2-1 3000 3000 0.20 0.20 CCND1 CCND1 0.15 0.15 2000 2000 0.10 0.10 1000 1000 0.05 0.05 NOTCH1 NOTCH1 0.00 0.00 0 0 0 0 200200 400400 600600 800800 1000 1000 0 0 1.01.0 2.02.0 3.03.0 4.04.0 5.05.0 6.06.0 7.07.0 Exon coverage in normal cell line (HapMap) Exon coverage in normal cell line (HapMap) Relative copy number by exon capture Relative copy number by exon capture Figure 2. Copy-number alterations in an archival breast cancer sample. a , sequence coverage is shown for each target in the tumor sample compared with a normal diploid sample. Exon targets from several genes with copy-number gains and losses are highlighted. B, copy-number correlation between exon capture and QPCR in sample FFPE 5. Quantitative PCR of FGFR1, CCND1, and NOTCH1 with 3 independent sets of primers was performed and average values for each gene were compared to exon capture copy-number. Examination of 79 pharmacogenomic loci facili- c omparison with an e xisting Mutation Profiling tated inspection of plausibly actionable polymorphisms Platform (Supplementary Table S9). The ERCC2-K751QC allele, as- We next wished to compare the sensitivity and specificity sociated with increased risk of FOLFOX-induced grade of targeted hybrid capture/sequencing to an existing mass 3 or 4 hematologic toxicity (31), was present in 2 samples spectrometric genotyping-based platform because this type (i.e., FFPE 2 and FFPE 9). The UGT1A1-G3156A allele was of approach is currently being used in several clinical and found to be heterozygous in 5 samples but homozygous translational oncology settings (2, 33–35). We thus performed in none of them. This allele is associated with irinote- OncoMap, a mass-spectrometric genotyping technology that can-related neutropenia when present as a homozygous interrogates more than 400 known mutations in 33 cancer event (32). genes. Of the 155 single-nucleotide variants seen by hybrid To validate these findings, a representative subset of capture/sequencing of the FFPE samples described previously, alterations (31 nonsynonymous variants and 2 indels; sam- 13 were also interrogated by assays present in OncoMap (Table ples 4–7) were independently queried by mass spectromet- 3). However, when OncoMap was performed on these samples, ric genotyping (2, 3). All 31 single-nucleotide variants and only 10 of these 13 mutations were detected. To determine 2 indels tested were confirmed, demonstrating 100% spec- the basis for this discrepancy, we assayed all 13 mutations by ificity of the targeted exon capture approach in the small an orthogonal genotyping approach that uses distinct reagent subset examined. Copy-number alterations involving 3 chemistry (hME genotyping; see Methods). All 13 mutations genes that were amplified or deleted in sample FFPE 5 were confirmed by this orthogonal genotyping method, sug- (FGFR1, CCND1, NOTCH1) were also tested by quantita- gesting that the 3 mutations not detected by OncoMap were tive PCR with the use of 3 independent primer pairs for false-negative results by mass spectrometric genotyping (Table each gene. As shown in Fig. 2B, the quantitative PCR re- 3; shown in bold). All mutations seen by OncoMap were also sults were highly correlated to the copy-number ratios detected by targeted exon capture. detected by targeted exon capture/sequencing in FFPE 5 2 2 (r = 0.94). The correlation coefficient (r ) for these same d iscussion genes in sample FFPE 9—which has a 2.3-fold amplifica- tion of FGFR1 but no copy-number changes in CCND1 or We have developed a targeted, massively parallel sequenc- NOTCH1—was 0.99 (Supplementary Fig. S3). ing platform to detect actionable genomic alterations in 88 |  CANCER DISCOVERY  JANUARY 2012  www.aacrjournals.org Downloaded from cancerdiscovery.aacrjournals.org on July 7, 2021. © 2012 American Association for Cancer Research. Exon coverage in tumor (FFPE 5) Exon coverage in tumor (FFPE 5) QPCR concentration (ng/mL) QPCR concentration (ng/mL) Published OnlineFirst November 7, 2011; DOI: 10.1158/2159-8290.CD-11-0184 Genomic Profiling of FFPE Tumor Samples by Targeted Sequencing research article t able 3. c omparison of OncoMap and targeted exon capture profiling in FFPe samples Seen via Seen via Validated FFPE Sample Sample type Gene Mutation OncoMap exon capture by hMe FFPE 1 Colon KRAS Q61H Yes Yes Yes FFPE 2 Colon KRAS G13C Yes Yes Yes FFPE 3 Colon KRAS G13D Yes Yes Yes FFPE 3 Colon PIK3CA H1047R Yes Yes Yes FFPE 4 Colon BRAF D594G No Yes Yes FFPE 4 Colon APC Q1367* No Yes Yes FFPE 4 Colon APC Q1378* Yes Yes Yes FFPE 6 Breast TP53 R248Q Yes Yes Yes FFPE 7 Colon KRAS G13D Yes Yes Yes FFPE 7 Colon PIK3CA E545K Yes Yes Yes FFPE 7 Colon CTNNB1 S45F No Yes Yes FFPE 8 Breast PIK3CA H1047R Yes Yes Yes FFPE 10 Colon KRAS Q61H Yes Yes Yes NOTE: Mutations that were not detected by OncoMap are shown in bold. clinical tumor samples. In this initial proof-of-concept effort, were seen by sequencing at multiple loci (including KRAS). we sequenced 137 cancer genes from 10 pooled FFPE tumor The OncoMap approach involves iPLEX genotyping of >500 DNA samples (plus 2 control samples) and achieved 391× mutations followed by hME validation of all candidates (see mean coverage per sample within a single paired-end se- Methods)—the iPLEX method allows increased multiplexing, quencing lane. This depth of coverage afforded robust, simul- but in our hands has proved somewhat less sensitive than taneous detection of base mutations, indels, amplifications, hME genotyping. The fact that all 13 mutations were subse- and deletions. Thus, targeted massively parallel sequencing quently confirmed by hMe chemistry suggests that massively provides a unifying approach for detection of multiple cat- parallel sequencing to several hundred-fold mean coverage egories of actionable genetic alterations. affords enhanced sensitivity compared to mass spectrometric In our pilot study, all of the tumor samples profiled genotyping. Moreover, most alterations found by sequencing contained biologically or clinically meaningful genomic are not assayed by genotyping or allele-specific PCR-based alterations, including several that might predict sensitivity mutation profiling platforms. Thus, the sequencing-based or resistance to targeted agents or provide useful prognos- approach may uncover more actionable options for patients tic information. In particular, 15 alterations (at least one than allele-specific approaches. per sample) were plausibly actionable, and might thus be Hybrid selection approaches have been widely used to predicted to impact clinical decision-making or clinical trial promote gene discovery by reducing genome complexity enrollment if identified as part of an experimental thera- before sequencing (5). In this study, we adapted this tech- peutics or phase I trial program. Several actionable somatic nique to capture a highly restricted genomic territory com- alterations (KRAS, PIK3CA, and MSH2) were detected in posed of 137 known cancer genes and 400,000 coding bases. samples with tumor purity as low as 10% to 20%, highlight- This afforded an expanded depth of coverage (to >400- ing the utility of this approach in “real-world” clinical tu- fold) while also enabling multiple barcoded samples to be mor samples. pooled within a single sequencing lane, thereby increasing Comparison with OncoMap, a mass spectrometric ge- throughput and lowering costs. We previously used a simi- notyping platform in current translational use, confirmed lar approach to characterize a frozen tumor sample from robust performance of targeted massively parallel sequenc- a patient with metastatic melanoma who developed resis- ing, even when applied to FFPE tumor specimens. In our tance to the RAF-inhibitor vemurafenib, and identified an previous study, the sensitivity and specificity of OncoMap activating mutation in MEK1 that caused resistance to RAF- in FFPE tissue was 89.3% and 99.4%, respectively, based on and MEK-inhibition (36). Here, we have adapted the ap- a focused comparison with massively parallel sequencing of proach to capture and sequence multiple barcoded samples KRAS (codon 12) in 93 FFPE samples. In the current study, and to identify distinct categories of genomic alterations OncoMap detected 10 of 13 mutations (79% sensitivity) that simultaneously. JANUARY 2012 CANCER DISCOVERY    89 Downloaded from cancerdiscovery.aacrjournals.org on July 7, 2021. © 2012 American Association for Cancer Research. Published OnlineFirst November 7, 2011; DOI: 10.1158/2159-8290.CD-11-0184 Wagle et al. research article An advantage of solution-phase hybrid capture is that Emerging frameworks for clinical interpretation of redesign and synthesis of long oligonucleotides for bait gen- genome sequencing data typically categorize alterations eration is a straightforward process that may be performed based on “actionability” or prognostic utility. Potentially iteratively until an optimal set of baits has been developed. actionable alterations may be further subdivided depending Thus, prioritized genomic regions can be readily amended on the level of evidence about a particular alteration, rang- as new knowledge of cancer gene mutations becomes avail- ing from those with established therapies to others with able. Furthermore, DNA barcoding and pooling decreases sound preclinical evidence. Plausibly actionable alterations the sequencing cost per sample in a manner proportional to may also include those for which the predictive implica- the number of pooled samples present within a sequencing tions within a particular cancer type are not known (e.g., lane. Achieving deep sequencing coverage increases the sen- BRAF mutations in lung cancer), or for which there is no sitivity of mutation detection—particularly in the setting of established clinical proof of concept (e.g., RET mutations high stromal admixture, which can pervade clinical tumor in lung cancer) even though a particular therapy against tissue. As such, this study extends earlier barcoding and hy- the target (sorafenib) may be commercially available. This brid capture/sequencing efforts (36–49) by identifying mul- category may also include mutations in tumor suppressor tiple types of actionable somatic alterations in archival (i.e., genes (e.g., PTEN) hypothesized to predict vulnerability to FFPE) tumor specimens. Because most clinical samples are targeted agents (e.g., PI3K inhibitors). stored as FFPE material, this approach may prove suitable More than 160 variants of unclear significance were for many translational and clinical applications. identified in our sample set. Undoubtedly, many such vari- At the same time, variations in FFPE sample quality may ants represent uncharacterized germline polymorphisms. adversely affect library construction, hybrid selection, or Differentiating somatic from germline alterations is readily sequencing. Potential solutions include the incorporation accomplished by including matched normal samples (36), of additional pre-processing steps to enrich for high-qual- although paired normal material is not always available in ity FFPE DNA, pooling of fewer samples prior to hybrid research settings. Even among alterations that are clearly selection, and/or increasing the overall depth of sequenc- somatic, additional approaches to interpret their potential ing if the starting library complexity is sufficiently high significance and communicate the results to clinicians and (50). The use of orthogonal technologies such as direct patients will be needed. Development of a rigorous formalism genotyping, quantitative PCR, or FISH to validate action- for clinical interpretation of complex genomic data will likely able alterations may prove useful in the short term because become an active research area, with the goal of enabling op- these techniques are used widely in existing clinical labora- timal, genomics-driven decision making for therapy or clini- tories. However, if the superior sensitivity and specificity is cal trial enrollment. confirmed in independent clinical studies, massively paral- Potential applications for targeted hybrid capture/mas- lel sequencing may become increasingly used in diagnostic sively parallel sequencing in translational and clinical or Clinical Laboratory Improvement Amendments (CLIA) oncology research include both retrospective and prospec- laboratory settings. tive profiling of tumor cohorts. Here, the goal may be to Several additional areas for technical and analytical op- identify predictive and prognostic genes or validate pharma- timization remain. Although we generally achieved robust cogenomic polymorphisms. Ultimately, similar approaches sequence coverage of targeted regions, genomic territory with may be used for prospective genomic profiling of cancer very high or very low GC content presents certain challenges. patients to guide clinical decision making. Toward this end, Options to improve coverage of these regions include rede- the potential turnaround time for the current approach is sign or inclusion of additional baits targeting regions that are 2 weeks. Emerging sequencing instruments promise vast difficult to capture. On the analytical side, detection of lon- reductions in turnaround time. Cost, a significant consider- ger indels (such as the 9-bp PIK3CA deletion in the NCI-H69 ation in clinical sequencing, can also be reduced dramatically cell line) remains difficult with current algorithms. Because by sample pooling. Indeed, it is likely that a combination actionable indels occur in multiple genes, including EGFR, of multiplexing together with falling sequencing costs may ERBB2, and KIT, supplemental assays may be needed to ensure ultimately eliminate cost as a limiting barrier to sequencing sensitive indel detection. Moreover, exon-directed capture data generation. approaches do not detect clinically relevant gene rearrange- In conclusion, the results described herein suggest that ments such as those involving ALK, ABL, and PDGFR. One targeted, massively parallel sequencing offers a promis- potential strategy to detect known rearrangements would ing method to detect actionable genetic alterations across a involve design of baits tiled across common translocation large panel of cancer genes in the clinical diagnostic arena. breakpoints. Furthermore, whereas both amplifications and If widely deployed, such implementation may open new op- deletions could be detected in cell line DNA, such events were portunities to link cancer genomics with molecular features, only observed in a single FFPE sample, which had 80% tumor clinical outcomes, and treatment response in a manner that purity. Detection of copy-number aberrations by targeted se- empowers multiple directions in molecular cancer epidemiol- quencing may be more problematic in samples with signifi- ogy. In addition, this approach may ultimately impact clini- cant stromal contamination. Future analytical methods that cal practice by offering a categorical means to identify genetic incorporate allelic information to infer tumor purity may en- changes affecting genes and pathways targeted by existing hance detection of copy gains and losses in samples with vari- and emerging drugs, thereby speeding the advent of personal- able tumor purity. ized cancer medicine. 90 |  CANCER DISCOVERY  JANUARY 2012  www.aacrjournals.org Downloaded from cancerdiscovery.aacrjournals.org on July 7, 2021. © 2012 American Association for Cancer Research. Published OnlineFirst November 7, 2011; DOI: 10.1158/2159-8290.CD-11-0184 Genomic Profiling of FFPE Tumor Samples by Targeted Sequencing research article existing databases including the Catalogue of Somatic Mutations Methods in Cancer (10) and The Cancer Genome Atlas (52). In addition, we High-Throughput, Targeted Deep Sequencing: Overview identified 79 pharmacogenomic polymorphisms described in the lit- Massively parallel sequencing libraries (Illumina) that contain bar- erature, which might predict sensitivity or resistance to conventional coded universal primers (9) were generated with the use of genomic cancer therapies (Supplementary Table S2). DNA from formalin-fixed, paraffin-embedded tumor material. After preamplification and DNA quantification, equimolar pools were Biotinylated RNA Baits generated consisting of 12 barcoded tumor DNAs. These DNA pools The Agilent SureSelect E-array program was used to design 7,021 were subjected to solution-phase hybrid capture with biotinylated unique RNA baits corresponding to the coding sequence of the 137 RNA baits targeting all exons from 137 actionable cancer genes. genes described previously, as well as to the 79 pharmacogenomic Each hybrid capture reaction was sequenced in a single paired-end polymorphisms and to 24 SNPs for fingerprinting. Target loci were lane of an Illumina flow cell. Subsequently, the sequencing data covered with a tiling density of ×2. Baits were replicated 8 times were deconvoluted to match all high-quality barcoded reads with on the 55,000-bait library array. The sequences of all 7,021 baits are the corresponding tumor samples, and genomic alterations (single- listed in the Supplementary Appendix. Biotinylated RNA baits were nucleotide sequence variants, small insertions/deletions, and DNA synthesized by Agilent for the SureSelect Target Enrichment system. copy-number alterations) were identified. The approach is illustrated schematically in Supplementary Figure S1. Pooling and Hybrid Capture DNA libraries were pooled by mixing 300 ng of each library in a Tumor Tissue and Cell Line DNA single 1.5-mL polypropylene sample tube, lyophilizing by the use Discarded and de-identified tumor specimens were obtained of a speedvac evaporator, and resuspending in 4 μL of nuclease-free from the Cooperative Human Tissue Network. An exemption from water. This entire amount (3,600 ng DNA in 4 μL) was used for hy- the Institutional Review Board was obtained for all samples from brid selection. Solution-phase hybrid capture was performed as pre- the Dana-Farber/Partners Cancer Care Office for the Protection of viously described (51) with 3 modifications to the hybrid selection Research Subjects (Protocol 10-380). Genomic DNA was extracted step (Basic Protocol 3). First, instead of 1.5 μL of Blocking Oligo 2.0, from tumor tissue using methods previously described (2). Cell line 0.125 μL of each of 12 additional 200 μM blocking oligonucleotides genomic DNA was purchased directly from the American Type Culture with sequences complementary to the barcodes were added to the hy- Collection (ATCC). Authentication of cell line genomic DNA was per- bridization reaction (see the Supplementary Methods for sequences). formed by ATCC by the use of short tandem repeat profiling, which Second, the biotinylated oligonucleotide baits were diluted 1:8 with uses multiplex PCR to simultaneously amplify the amelogenin gene nuclease-free water from a concentration of 100 ng/μL to 12.5 ng/ and 8 of the most informative polymorphic markers in the human μL immediately before hybridization and 5 μL of this solution was genome. Control genomic DNA was from the HapMap consortium, added to the hybridization reaction. which was purchased from the Coriell Institute for Medical Research. The final volume of the hybridization reaction was 19 μL, consist- ing of the following components: 4 μL of pooled DNA libraries, 2.5 Barcoded Genomic DNA Library Construction μL of 1.0 mg/mL human Cot-1 DNA, 2.5 μL of 10.0 mg/mL salmon Genomic DNA was quantified by the use of Quant-iT PicoGreen® sperm DNA, 1.5 μL of 200 μM blocking oligo 1.0, 1.5 μL of total of dsDNA Assay Kit (Invitrogen, Carlsbad, CA). A total of 1 μg of ge- the twelve 200 μM blocking oligonucleotides, 5.0 μL of 12.5 ng/μL nomic DNA from each sample was sheared by sonication with the biotinylated oligonucleotide baits, 1.0 μL of 20 U/μL Superase-In following conditions: duty cycle 10%, intensity 5, cycles per burst RNAse inhibitor, and 1 μL of nuclease-free water. Third, during PCR 200, and 135 seconds (Covaris S2 instrument). Paired-end adapt- enrichment of the captured DNA (“the catch”), PCR was performed ers for massively parallel sequencing (Illumina) were added as pre- with primers P5 (5′-AAT GAT ACG GCG ACC ACC GA-3′) and P7 viously described (51), with the following modifications to the (5′-CAA GCA GAA GAC GGC ATA CGA-3′), both at 100 μM, in- paired end library preparation step (basic protocol 2). First, the stead of PCR primers PE1.0 and PE2.0. PCR conditions remained as multiplex adapter provided with the Multiplex Paired-End Library described. All custom primers were obtained from Integrated DNA Sample Preparation Kit (Illumina) was used instead of the standard Technologies (IDT). paired-end adapter. Second, PCR enrichment was conducted in 150 μL of total volume with 3 primers from the Multiplexing Sample Sequencing and Analysis Preparation Oligonucleotide Kit (Illumina). Each PCR enrichment We sequenced 100 bases from both ends of library DNA fragments reaction contained 75 μL of Phusion polymerase (Finnzymes), 3 μL by using an Illumina HiSeq 2000 instrument. The sequence reads of Multiplexing PE Primer 1.0 (25 μM), 3 μL of Multiplexing PE were aligned to human reference genome hg18 with the Burrows- Primer 2.0 (0.5 μM), 3 μL of an Index primer (25 μM), 36 μL of Wheeler Alignment tool (53) with use of the following parameters: –q paired-end library, and 30 μL of nuclease-free water. Samples were 5 –l 32 –k 2 –o 1. Artifactual duplicate read pairs were removed with denatured for 5 minutes at 95°C; 18 cycles of 10 seconds at 95°C, Picard tools (picard.sourceforge.net). An average of 450 megabases of 30 seconds at 65°C, and 30 seconds at 72°C; and a final 5 minutes aligned sequence was generated for each library. at 72°C before cooling to 4°C. PCR primers were removed by using Single-nucleotide variants and small insertions/deletions were ×1.8 volume of Agencourt AMPure PCR Purification Kit (Agencourt identified by the use of algorithms from the Genome Analysis Toolkit Bioscience Corporation). developed at the Broad Institute (54). A local multiple sequence align- ment was performed on intervals suspected to harbor indels to de- Selection of Targeted Genes rive the most probable underlying genomic structure of the query We identified 137 genes that are biologically or clinically relevant sample. Single-nucleotide variants were called separately on each sam- in cancer, including targets of new and existing therapies, genes that ple with UnifiedGenotyper and annotated with GenomicAnnotator. predict sensitivity or resistance to therapies, genes that are prog- Variants were discarded if they were present in dbSNP and not in the nostic markers, and oncogenes and tumor suppressors that are COSMIC database (10), they exhibited an unfavorable strand balance known to undergo recurrent somatic genomic alterations in cancer score (> —20), or they were detected in the HapMap normal control. (Supplementary Table S1). These genes were identified by mining Novel recurrent single-nucleotide variants were manually reviewed JANUARY 2012 CANCER DISCOVERY    91 Downloaded from cancerdiscovery.aacrjournals.org on July 7, 2021. © 2012 American Association for Cancer Research. Published OnlineFirst November 7, 2011; DOI: 10.1158/2159-8290.CD-11-0184 Wagle et al. research article to eliminate additional systematic artifacts. Indels were called with U24CA143867 (M. Meyerson), the Snyder Medical Foundation (W.C. IndelGenotyperV2 and were retained if they occurred in protein-cod- Hahn), and the Starr Cancer Consortium (M.F. Berger, L.A. Garraway). ing exons and on both DNA strands, in <2% of reads in the HapMap normal control, and were absent from dbSNP. Received July 25, 2011; revised October 24, 2011; accepted To calculate relative copy-number levels of the 137 target gene November 4, 2011; published OnlineFirst November 7, 2011. loci, we computed the mean sequence coverage for each gene across all protein-coding exons by using the DepthOfCoverage tool in the Genome Analysis Toolkit. All bases in reads with mapping quality r e Fere Nces <5 were ignored, as were any additional bases with base quality <5. Gene-level coverage in each tumor was normalized by the gene-level 1. Macconaill LE, Garraway LA. Clinical implications of the cancer genome. J Clin Oncol 2010;28:5219–28. coverage for an indexed HapMap diploid cell line included in the 2. MacConaill LE, Campbell CD, Kehoe SM, Bass AJ, Hatton C, Niu same pooled hybrid selection experiment (after adjusting for dif- L, et al. Profiling critical cancer gene mutations in clinical tumor ferences in the overall amount of aligned sequence per sample). samples. PLoS One 2009;4:e7887. Sequence-derived estimates of copy number were then compared to 3. Thomas RK, Baker AC, Debiasi RM, Winckler W, Laframboise T, Lin SNP array-derived estimates of copy number for the cancer cell lines. WM, et al. High-throughput oncogene mutation profiling in human cancer. Nat Genet 2007;39:347–51. Mass Spectrometric Genotyping 4. Dias-Santagata D, Akhavanfard S, David SS, Vernovsky K, Kuhlmann Mass spectrometric genotyping was performed with the OncoMap G, Boisvert SL, et al. Rapid targeted mutational analysis of human 3.0 platform as previously described (2, 3) with iPLEX chemistry for tumours: a clinical platform to guide personalized cancer medicine. initial mutation profiling and validation by multibase hME exten- EMBO Mol Med 2010;2:146–58. sion chemistry. Genomic DNA from all tumor samples was quanti- 5. Meyerson M, Gabriel S, Getz G. Advances in understanding cancer fied using Quant-iT PicoGreen® dsDNA Assay Kit (Invitrogen). genomes through second-generation sequencing. Nat Rev Genet To validate alterations detected by massively parallel sequenc- 2010;11:685–96. ing that were not included in the OncoMap assay collection, base 6. Stratton MR, Campbell PJ, Futreal PA. The cancer genome. Nature substitutions and indels were queried using multi-base hME exten- 2009;458:719–24. 7. Gnirke A, Melnikov A, Maguire J, Rogov P, LeProust EM, Brockman sion chemistry with plexing of ≤6 assays per pool. Conditions for W, et al. Solution hybrid selection with ultra-long oligonucleotides hME validation were implemented as described previously (2, 3). for massively parallel targeted sequencing. Nat Biotechnol Primers and probes used for hME validation were designed with the 2009;27:182–9. Sequenom MassARRAY Assay Design 3.0 software, applying default 8. Hodges E, Rooks M, Xuan Z, Bhattacharjee A, Benjamin Gordon D, multi-base extension parameters but with the following modifica- Brizuela L, et al. Hybrid selection of discrete genomic intervals on tions: maximum multiplex level input equal to 6; maximum pass it- custom-designed microarrays for massively parallel sequencing. Nat eration base adjusted to 200. Protoc 2009;4:960–74. 9. Craig DW, Pearson JV, Szelinger S, Sekar A, Redman M, Corneveaux Microarray Analysis of Chromosomal Copy Number JJ, et al. Identification of genetic variants using bar-coded Chromosomal copy-number information was obtained from the multiplexed sequencing. Nat Methods 2008;5:887–93. Broad-Novartis Cancer Cell Line Encyclopedia project, which has 10. Forbes SA, Bindal N, Bamford S, Cole C, Kok CY, Beare D, et al. high-density SNP array data from the Affymetrix SNP 6.0 platform COSMIC: mining complete cancer genomes in the Catalogue of for all cancer cell lines profiled in this study (55). Somatic Mutations in Cancer. Nucleic Acids Res 2011;39(Database issue):D945–50. Quantitative PCR Analysis of Chromosomal Copy Number 11. Ogata H, Sato H, Takatsuka J, De Luca LM. Human breast cancer MDA-MB-231 cells fail to express the neurofibromin protein, lack Quantitative PCR was performed with the SYBR Green PCR its type I mRNA isoform and show accumulation of P-MAPK and Master Mix Kit (Applied Biosystems) according to the manufactur- activated Ras. Cancer Lett 2001;172:159–64. er’s instructions. To determine the chromosomal copy number of 12. Allegra CJ, Jessup JM, Somerfield MR, Hamilton SR, Hammond EH, each gene, 3 sets of gene-specific primers were designed to interrogate Hayes DF, et al. American Society of Clinical Oncology provisional the genetic locus. Primers recognizing LINE sequences were used for clinical opinion: testing for KRAS gene mutations in patients reference amplification/normalization as described previously (56). with metastatic colorectal carcinoma to predict response to anti- Primer sequences are provided in the Supplementary Methods. Male epidermal growth factor receptor monoclonal antibody therapy. J genomic DNA (Promega) was included as a standard, and HapMap Clin Oncol 2009;27:2091–6. DNA (Coriell) was used as a normal diploid control. Quantitative 13. Bardelli A, Siena S. Molecular mechanisms of resistance to cetuximab PCRs were performed in triplicate for each sample using an ABI 7300 and panitumumab in colorectal cancer. J Clin Oncol 2010;28:1254–61. instrument, in 25-μL reactions containing 0.5 ng of genomic DNA 14. Sartore-Bianchi A, Di Nicolantonio F, Nichelatti M, Molinari F, De and forward and reverse primers each at a concentration of 600 nM. Dosso S, Saletti P, et al. Multi-determinants analysis of molecular alterations for predicting clinical benefit to EGFR-targeted Disclosure of Potential c onflicts of interest monoclonal antibodies in colorectal cancer. PLoS One 2009;4:e7287. 15. Sartore-Bianchi A, Martini M, Molinari F, Veronese S, Nichelatti M, Consultant/advisory role: Foundation Medicine (N. Wagle, Artale S, et al. PIK3CA mutations in colorectal cancer are associated M.F. Berger, M.J. Davis, M. Meyerson, L.A. Garraway), Novartis with clinical resistance to EGFR-targeted monoclonal antibodies. (W.C. Hahn, M. Meyerson, L.A. Garraway), Daiichi Sankyo (L.A. Cancer Res 2009;69:1851–7. Garraway). Ownership interest: Foundation Medicine (N. Wagle, M. 16. De Roock W, Claes B, Bernasconi D, De Schutter J, Biesmans B, Meyerson, L.A. Garraway). Research support: Novartis (W.C. Hahn, Fountzilas G, et al. Effects of KRAS, BRAF, NRAS, and PIK3CA M. Meyerson, L.A. Garraway). Patents: Laboratory Corporation of mutations on the efficacy of cetuximab plus chemotherapy in America (M. Meyerson). Honoraria: Illumina (M.F. Berger). chemotherapy-refractory metastatic colorectal cancer: a retrospective consortium analysis. Lancet Oncol 2010;11:753–62. a cknowledgments 17. Berns K, Horlings HM, Hennessy BT, Madiredjo M, Hijmans EM, This work was supported by the NIH Director’s New Innovator Beelen K, et al. A functional genetic approach identifies the PI3K Award DP2OD002750 (L.A. Garraway), the National Cancer Institute pathway as a major determinant of trastuzumab resistance in breast R33CA126674 (L.A. Garraway), the National Cancer Institute cancer. Cancer Cell 2007;12:395–402. 92 |  CANCER DISCOVERY  JANUARY 2012  www.aacrjournals.org Downloaded from cancerdiscovery.aacrjournals.org on July 7, 2021. © 2012 American Association for Cancer Research. Published OnlineFirst November 7, 2011; DOI: 10.1158/2159-8290.CD-11-0184 Genomic Profiling of FFPE Tumor Samples by Targeted Sequencing research article 18. Wang L, Zhang Q, Zhang J, Sun S, Guo H, Jia Z, et al. PI3K pathway 37. Bell CJ, Dinwiddie DL, Miller NA, Hateley SL, Ganusova EE, Mudge activation results in low efficacy of both trastuzumab and lapatinib. J, et al. Carrier testing for severe childhood recessive diseases by next- BMC Cancer 2011;11:248. generation sequencing. Sci Transl Med 2011;3:65ra4. 19. Courtney KD, Corcoran RB, Engelman JA. The PI3K pathway as drug 38. Berg JS, Evans JP, Leigh MW, Omran H, Bizon C, Mane K, et al. target in human cancer. J Clin Oncol. 2010;28:1075–83. Next generation massively parallel sequencing of targeted exomes to identify genetic mutations in primary ciliary dyskinesia: implications 20. Duffy A, Kummar S. Targeting mitogen-activated protein kinase kinase (MEK) in solid tumors. Target Oncol 2009;4:267–73. for application to clinical testing. Genet Med 2011;13:218–29. 39. Chmielecki J, Peifer M, Jia P, Socci ND, Hutchinson K, Viale A, et al. 21. Wagner AJ, Malinowska-Kolodziej I, Morgan JA, Qin W, Fletcher CD, Vena N, et al. Clinical activity of mTOR inhibition with sirolimus Targeted next-generation sequencing of DNA regions proximal to a conserved GXGXXG signaling motif enables systematic discovery of in malignant perivascular epithelioid cell tumors: targeting the pathogenic activation of mTORC1 in tumors. J Clin Oncol tyrosine kinase fusions in cancer. Nucleic Acids Res 2010;38:6985–96. 40. Grossmann V, Kohlmann A, Klein HU, Schindela S, Schnittger S, 2010;28:835–40. 22. Paik PK, Arcila ME, Fara M, Sima CS, Miller VA, Kris MG, et al. Dicker F, et al. Targeted next-generation sequencing detects point mutations, insertions, deletions and balanced chromosomal Clinical characteristics of patients with lung adenocarcinomas harboring BRAF mutations. J Clin Oncol 2011;29:2046–51. rearrangements as well as identifies novel leukemia-specific fusion genes in a single procedure. Leukemia 2011;25:671–80. 23. Smalley KS, Xiao M, Villanueva J, Nguyen TK, Flaherty KT, Letrero R, et al. CRAF inhibition induces apoptosis in melanoma cells with 41. Johansson H, Isaksson M, Sorqvist EF, Roos F, Stenberg J, Sjoblom T, et al. Targeted resequencing of candidate genes using selector probes. non-V600E BRAF mutations. Oncogene 2009;28:85–94. 24. Fong PC, Boss DS, Yap TA, Tutt A, Wu P, Mergui-Roelvink M, et al. Nucleic Acids Res 2011;39:e8. 42. Kiialainen A, Karlberg O, Ahlford A, Sigurdsson S, Lindblad-Toh K, Inhibition of poly(ADP-ribose) polymerase in tumors from BRCA mutation carriers. N Engl J Med 2009;361:123–34. Syvanen AC. Performance of microarray and liquid based capture methods for target enrichment for massively parallel sequencing and 25. Xie W, Rimm DL, Lin Y, Shih WJ, Reiss M. Loss of Smad signaling in human colorectal cancer is associated with advanced disease and SNP discovery. PLoS One 2011;6:e16486. 43. Liao GJ, Lun FM, Zheng YW, Chan KC, Leung TY, Lau TK, et al. poor prognosis. Cancer J 2003;9:302–12. 26. Turner N, Pearson A, Sharpe R, Lambros M, Geyer F, Lopez- Targeted massively parallel sequencing of maternal plasma DNA permits efficient and unbiased detection of fetal alleles. Clin Chem Garcia MA, et al. FGFR1 amplification drives endocrine therapy resistance and is a therapeutic target in breast cancer. Cancer Res 2011;57:92–101. 44. Meder B, Haas J, Keller A, Heid C, Just S, Borries A, et al. Targeted 2010;70:2085–94. 27. Bostner J, Ahnstrom Waltersson M, Fornander T, Skoog L, next-generation sequencing for the molecular genetic diagnostics of cardiomyopathies. Circ Cardiovasc Genet 2011;4:110–22. Nordenskjold B, Stal O. Amplification of CCND1 and PAK1 as predictors of recurrence and tamoxifen resistance in postmenopausal 45. Nijman IJ, Mokry M, van Boxtel R, Toonen P, de Bruijn E, Cuppen E. Mutation discovery by targeted genomic enrichment of multiplexed breast cancer. Oncogene 2007;26:6997–7005. 28. Lundgren K, Holm K, Nordenskjold B, Borg A, Landberg G. Gene barcoded samples. Nat Methods 2010;7:913–5. 46. Shearer AE, DeLuca AP, Hildebrand MS, Taylor KR, Gurrola J, 2nd, products of chromosome 11q and their association with CCND1 gene amplification and tamoxifen resistance in premenopausal Scherer S, et al. Comprehensive genetic testing for hereditary hearing loss using massively parallel sequencing. Proc Natl Acad Sci U S A breast cancer. Breast Cancer Res 2008;10:R81. 29. Finn RS, Dering J, Conklin D, Kalous O, Cohen DJ, Desai AJ, et al. 2010;107:21104–9. 47. Summerer D, Schracke N, Wu H, Cheng Y, Bau S, Stahler CF, PD 0332991, a selective cyclin D kinase 4/6 inhibitor, preferentially inhibits proliferation of luminal estrogen receptor-positive human et al. Targeted high throughput sequencing of a cancer-related exome subset by specific sequence capture with a fully automated breast cancer cell lines in vitro. Breast Cancer Res 2009;11:R77. 30. Roy PG, Pratt N, Purdie CA, Baker L, Ashfield A, Quinlan P, et al. High microarray platform. Genomics 2010;95:241–6. 48. Teer JK, Bonnycastle LL, Chines PS, Hansen NF, Aoyama N, Swift CCND1 amplification identifies a group of poor prognosis women with estrogen receptor positive breast cancer. Int J Cancer 2010;127:355–60. AJ, et al. Systematic comparison of three genomic enrichment methods for massively parallel DNA sequencing. Genome Res 31. Boige V, Mendiboure J, Pignon JP, Loriot MA, Castaing M, Barrois M, et al. Pharmacogenetic assessment of toxicity and outcome in patients 2010;20:1420–31. 49. Zang ZJ, Ong CK, Cutcutache I, Yu W, Zhang SL, Huang D, et al. with metastatic colorectal cancer treated with LV5FU2, FOLFOX, and FOLFIRI: FFCD 2000–05. J Clin Oncol 2010;28:2556–64. Genetic and structural variation in the gastric cancer kinome revealed through targeted deep sequencing. Cancer Res 2011;71:29–39. 32. Lara PN, Jr., Natale R, Crowley J, Lenz HJ, Redman MW, Carleton JE, et al. Phase III trial of irinotecan/cisplatin compared with 50. Fisher S, Barry A, Abreu J, Minie B, Nolan J, Delorey TM, et al. A scalable, fully automated process for construction of sequence-ready etoposide/cisplatin in extensive-stage small-cell lung cancer: clinical and pharmacogenomic results from SWOG S0124. J Clin Oncol human exome targeted capture libraries. Genome Biol 2011;12(1):R1. 51. Blumenstiel B, Cibulskis K, Fisher S, DeFelice M, Barry A, Fennell T, 2009;27:2530–5. 33. Arcila M, Lau C, Nafa K, Ladanyi M. Detection of KRAS and BRAF et al. Targeted exon sequencing by in-solution hybrid selection. Curr Protoc Hum Genet 2010;Chapter 18:Unit 18.4. mutations in colorectal carcinoma roles for high-sensitivity locked nucleic acid-PCR sequencing and broad-spectrum mass spectrometry 52. The Cancer Genome Atlas. Available from: cancergenome.nih.gov. 53. Li H, Durbin R. Fast and accurate short read alignment with genotyping. J Mol Diagn 2011;13:64–73. 34. Brevet M, Johnson ML, Azzoli CG, Ladanyi M. Detection of EGFR Burrows-Wheeler transform. Bioinformatics 2009;25:1754–60. 54. McKenna A, Hanna M, Banks E, Sivachenko A, Cibulskis K, mutations in plasma DNA from lung cancer patients by mass spectrometry genotyping is predictive of tumor EGFR status and Kernytsky A, et al. The Genome Analysis Toolkit: a MapReduce framework for analyzing next-generation DNA sequencing data. response to EGFR inhibitors. Lung Cancer 2011;73:96–102. 35. Beadling C, Heinrich MC, Warrick A, Forbes EM, Nelson D, Genome Res 2010;20:1297–303. 55. Broad-Novartis Cancer Cell Line Encylopedia. Available from: www. Justusson E, et al. Multiplex mutation screening by mass spectrometry: evaluation of 820 cases from a personalized cancer broadinstitute.org/ccle. 56. Garraway LA, Widlund HR, Rubin MA, Getz G, Berger AJ, medicine registry. J Mol Diagn. 2011;13:504–13. 36. Wagle N, Emery C, Berger MF, Davis MJ, Sawyer A, Pochanard P, Ramaswamy S, et al. Integrative genomic analyses identify MITF as a lineage survival oncogene amplified in malignant melanoma. Nature et al. Dissecting therapeutic resistance to raf inhibition in melanoma by tumor genomic profiling. J Clin Oncol 2011;29:3085–96. 2005;436:117–22. JANUARY 2012 CANCER DISCOVERY    93 Downloaded from cancerdiscovery.aacrjournals.org on July 7, 2021. © 2012 American Association for Cancer Research. Published OnlineFirst November 7, 2011; DOI: 10.1158/2159-8290.CD-11-0184 High-Throughput Detection of Actionable Genomic Alterations in Clinical Tumor Samples by Targeted, Massively Parallel Sequencing Nikhil Wagle, Michael F. Berger, Matthew J. Davis, et al. Cancer Discov 2012;2:82-93. Published OnlineFirst November 7, 2011. Access the most recent version of this article at: Updated version doi:10.1158/2159-8290.CD-11-0184 Access the most recent supplemental material at: Supplementary http://cancerdiscovery.aacrjournals.org/content/suppl/2011/11/07/2159-8290.CD-11-0184.DC Material This article cites 53 articles, 19 of which you can access for free at: Cited articles http://cancerdiscovery.aacrjournals.org/content/2/1/82.full#ref-list-1 This article has been cited by 90 HighWire-hosted articles. Access the articles at: Citing articles http://cancerdiscovery.aacrjournals.org/content/2/1/82.full#related-urls Sign up to receive free email-alerts related to this article or journal. E-mail alerts To order reprints of this article or to subscribe to the journal, contact the AACR Publications Reprints and Department at [email protected]. Subscriptions To request permission to re-use all or part of this article, use this link Permissions http://cancerdiscovery.aacrjournals.org/content/2/1/82. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site. Downloaded from cancerdiscovery.aacrjournals.org on July 7, 2021. © 2012 American Association for Cancer Research.

Journal

Cancer Discovery – Unpaywall

Published: Jan 1, 2012

There are no references for this article.