Molecular Profiling of the Residual Disease of Triple-Negative Breast Cancers After Neoadjuvant Chemotherapy Identifies Actionable Therapeutic Targets

Published OnlineFirst December 19, 2013; DOI: 10.1158/2159-8290.CD-13-0286

RESEARCH ARTICLE

Molecular Proﬁ ling of the Residual Disease of Triple-Negative Breast Cancers after Neoadjuvant Chemotherapy Identiﬁ es Actionable Therapeutic Targets

Justin M. Balko1 , 5 , Jennifer M. Giltnane2 , 5 , Kai Wang 8 , Luis J. Schwarz1 , 9 , 10 , Christian D. Young1 , Rebecca S. Cook3 , 5, Phillip Owens 3 , Melinda E. Sanders 2 , 5, Maria G. Kuba 2 , Violeta Sánchez 1 , Richard Kurupi 1 , Preston D. Moore 1 , Joseph A. Pinto 9 , Franco D. Doimi 9 , Henry Gómez 10 , Dai Horiuchi 6 , 7, Andrei Goga 6 , 7, Brian D. Lehmann 4 , Joshua A. Bauer 4 , Jennifer A. Pietenpol 4 , 5, Jeffrey S. Ross 8 , Gary A. Palmer 8 , Roman Yelensky 8 , Maureen Cronin8 , Vincent A. Miller 8 , Phillip J. Stephens 8 , and Carlos L. Arteaga 1 , 3 , 5

ABSTRACT Neoadjuvant chemotherapy (NAC) induces a pathologic complete response (pCR) in approximately 30% of patients with triple-negative breast cancers (TNBC). In patients lacking a pCR, NAC selects a subpopulation of chemotherapy-resistant tumor cells. To under- stand the molecular underpinnings driving treatment-resistant TNBCs, we performed comprehensive molecular analyses on the residual disease of 74 clinically defi ned TNBCs after NAC, including next- generation sequencing (NGS) on 20 matched pretreatment biopsies. Combined NGS and digital RNA expression analysis identifi ed diverse molecular lesions and pathway activation in drug-resistant tumor cells. Ninety percent of the tumors contained a genetic alteration potentially treatable with a currently available targeted therapy. Thus, profi ling residual TNBCs after NAC identifi es targetable molecular lesions in the chemotherapy-resistant component of the tumor, which may mirror micrometastases destined to recur clinically. These data can guide biomarker-driven adjuvant studies targeting these micrometastases to improve the outcome of patients with TNBC who do not respond completely to NAC.

SIGNIFICANCE: This study demonstrates the spectrum of genomic alterations present in residual TNBC after NAC. Because TNBCs that do not achieve a CR after NAC are likely to recur as metastatic disease at variable times after surgery, these alterations may guide the selection of targeted therapies immediately after mastectomy before these metastases become evident. Cancer Discov; 4(2); 232–45. ©2013 AACR.

INTRODUCTION have recently shown that these subtypes differ vastly in their post-NAC Ki67 scores, confounding the prognostic utility Neoadjuvant chemotherapy (NAC) is used increasingly in of Ki67 in this setting (10 ), and this has been confi rmed by patients with triple-negative breast cancer (TNBC), a subtype other investigators (9 ). Furthermore, Ki67 scoring is diffi cult lacking expression of estrogen receptor (ER), progesterone to standardize among clinical laboratories and many stud- receptor (PR), or HER2 amplifi cation. The goals of NAC are ies have defi ned different “cutoffs” for patient stratifi cation, to increase the likelihood of breast-conserving surgery and ranging from 14% to 50% (4–6 ). Finally, the Ki67 scoring of to eliminate clinically silent micrometastases. Approximately the post-NAC residual tumor is not actionable as it does not 30% of TNBC patients who receive NAC achieve a pathologic identify a pathogenic driver of the tumor and, as such, a drug complete response (pCR). These patients have a favorable target and rational treatment decision. recurrence-free survival (RFS) and overall survival (OS; refs. Intuition suggests that tumor cells remaining after NAC 1–3 ). The remaining patients with residual viable cancer in contain the cancer cell population intrinsically resistant to the breast or lymph nodes exhibit high rates of metastatic chemotherapy. These tumor cells likely mirror the micrometa- recurrence and an overall poor long-term outcome ( 1–3 ). static component of the disease that is ultimately responsible Immunohistochemistry (IHC) of the proliferation marker for distant metastases, and is unlikely to be highly sensitive Ki67 in post-NAC residual disease has been shown to corre- to further chemotherapy once clinical metastases become evi- late with patient outcome (4–6 ). Previous studies showing the dent. The standard of care for patients with TNBC who have prognostic ability of Ki67 after NAC included all subtypes of residual disease after NAC is observation, as therapies that breast cancer (i.e., HER2-enriched, luminal A, luminal B, and would be effective in reducing recurrences are unknown. Thus, basal-like), which also offer prognostic information ( 7–9 ). We we molecularly profi led the residual disease remaining after NAC in a cohort of 111 TNBCs [including gene expression analysis of 89 tumors and next-generation sequencing (NGS) Authors’ Affi liations: Departments of 1 Medicine, 2 Pathology, Microbi- ology & Immunology, 3 Cancer Biology, and 4 Biochemistry; 5 Breast Can- of 80 tumors, 74 of which were TNBC] to identify lesions that cer Research Program, Vanderbilt-Ingram Cancer Center, Vanderbilt could be therapeutically targeted in adjuvant trials. University, Nashville, Tennessee; Departments of 6Cell & Tissue Biology and 7 Medicine, University of California, San Francisco, San Francisco, California; 8Foundation Medicine, Cambridge, Massachusetts; 9 Oncosalud; RESULTS and 10Instituto Nacional de Enfermedades Neoplásicas (INEN), Lima, Perú Ki67 Does Not Predict Clinical Outcome in TNBCs Note: Supplementary data for this article are available at Cancer Discovery Online (http://cancerdiscovery.aacrjournals.org/). Because TNBC is a heterogeneous subtype of breast can- Corresponding Author: Carlos L. Arteaga, Vanderbilt University Medical cer (11 ), we determined whether Ki67 could predict patient Center, 2200 Pierce Ave, 777 PRB, Nashville, TN 37232-6307. Phone: 615- outcome within this clinical subtype by scoring Ki67 in the 936-3524; Fax: 615-936-1790; E-mail: [email protected] residual disease of a cohort of 111 TNBCs after NAC. Patient doi: 10.1158/2159-8290.CD-13-0286 demographics are listed in Supplementary Table S1. Molecu- ©2013 American Association for Cancer Research. lar subtyping based on gene expression using the PAM50

FEBRUARY 2014CANCER DISCOVERY | 233

RESEARCH ARTICLE Balko et al. centroids ( 7 ) revealed a predominance of tumors with basal- (trend, P = 0.0697), and JAK2 amplifi cations (trend, P = 0.08) like gene expression (Supplementary Fig. S1A). After adjust- in the residual disease. Amplifi cations in CDK6 and CCND1, ing for seven tumors that exhibited HER2 amplifi cation (see CCND2, and CCND3 were collectively enriched as well (24% below), 70% of TNBCs were basal-like, which is similar to in post-NAC TNBC vs. 10% in TCGA basal-like tumors). This previously published rates in larger datasets ( 12 ). Basal-like difference suggests that these alterations are present at higher status was associated with a trend toward worse RFS and frequency in chemotherapy-treated TNBCs, and may play a OS (log-rank test, P = 0.12 and 0.058, respectively; Supple- role in de novo or acquired therapeutic resistance. However, it mentary Fig. S1B). As we have previously demonstrated (10 ), is important to note that these comparisons of CNAs with the the Ki67 score in the residual disease varied signifi cantly TCGA data are made between platforms [NGS vs. Affymetrix among molecular subtypes within this TNBC cohort, but single-nucleotide polymorphism (SNP) arrays], and thus some was not prognostic (Supplementary Fig. S1C and S1D). Ki67 variation in calling rates and detection of alterations may be staining decreased signifi cantly in response to chemotherapy platform-specifi c. ( P < 0.0001, paired t test; Supplementary Fig. S1E), but this Identifi ed alterations were categorized into several key path- change was not different among the molecular subtypes way or functional groups: cell-cycle alterations (amplifi cations (Supplementary Fig. S1F). Tumor cellularity was signifi cantly in CDK4 , CDK6 , CCND1, CCND2, CCND3 , CCNE1 , or AURKA decreased between the pre- and post-NAC samples (paired and loss of CDKN2BCDKN2A , , or RB1); phosphoinositide t test, P < 0.0001; Supplementary Fig. S1G). Node status at 3-kinase (PI3K)/mTOR alterations (amplifi cations of AKT1, surgery (an established prognostic marker), but not a change AKT2, AKT3 , PIK3CA , RAPTOR, or RICTOR ; loss or mutation in Ki67, was predictive of both RFS and OS, although this of PTEN ; truncations or nonsense mutations in TSC1 ; ampli- effect seemed to be confi ned only to postmenopausal women fi cations or mutations in PIK3CA or PIK3R1 ); growth factor (Supplementary Fig. S2). These data suggest that the underly- receptor (GFR) amplifi cations ( EGFR , MET , KIT , FGFR1, 2 , and ing molecular subtype may confound the prognostic ability 4 , or IGF1R ); RAS/mitogen-activated protein kinase (MAPK) of Ki67 score in the residual disease after NAC. alterations (amplifi cations/gains of KRAS , BRAF , or RAF1 , or truncations of NF1 ); or DNA repair alterations (truncations, Genomic Alterations in Drug-Resistant loss or mutations of BRCA1 or BRCA2, or mutations in ATM; Residual Cancers after NAC Fig. 1B ). Importantly, more than 90% of the patients had alter- To identify targetable molecular lesions present in breast ations in at least one of these clinically targetable pathways. cancers after NAC, we performed targeted NGS of 3,320 exons of 182 oncogenes and tumor suppressors plus 37 introns of 14 Gene Expression Analysis genes frequently rearranged in cancer (Supplementary Table NanoString gene expression analysis was performed in S2), including intervallic targets throughout the genome for 104 of 111 samples; 89 samples (86%) passed quality control copy-number alteration (CNA) analysis. NGS analysis was metrics (Supplementary Table S5). Sixty-fi ve samples were attempted in 85 formalin-fi xed, paraffi n-embedded (FFPE) can- analyzed by both NGS and NanoString. Overall, 450 tran- cers with suffi ciently high tumor cellularity (>20%); 81 (95%) scripts were quantifi ed. These 450 transcripts were selected were successfully analyzed. Paucicellular samples were enriched on the basis of their inclusion in published gene expression by macrodissection such that the sampled region was as close to signatures or based on their association with the post-NAC 20% tumor nuclei as possible, or greater. Mean depth of cover- Ki67 score we reported recently ( 10 ). Specifi cally, we included age across all samples was 609× (range, 131–1,215). Six samples the PAM50 genes ( 7 ), a signature of MAP–ERK kinase (MEK) lacked suffi cient depth of coverage (<200×) to make calls in activation ( 15 ), a signature of TGF-β activation ( 16 ), and CNAs with confi dence. Seven tumors harbored HER2 ampli- genes we have previously shown to correlate with the post- fi cation (confi rmed by FISH) and were excluded from further NAC Ki67 score ( 10 ). These signatures were selected on the analysis. All remaining post-NAC tumors were ER- and PR-neg- basis of our previous studies demonstrating association of ative by IHC. Thus, 74 tumors had evaluable NGS data, 68 of DUSP4 loss with the MEK activation signature, and with the which also had CNA data. No obvious differences between the enrichment of TGF-β–inducible genes after NAC (10 , 17 ). NGS-evaluable population and the entire cohort were observed There was excellent concordance between gene expression in terms of outcome or clinical characteristics. NGS analysis and CNAs or mutations in cases where both were assessed revealed a diversity of lesions, many of which were present in (Supplementary Fig. S3). Gene expression data were used to less than 5% of samples ( Fig. 1A and Supplementary Table S3). predict the molecular subtype using the PAM50 centroids. Alterations in TP53 were identifi ed in 72 of 81 samples (89%), Of the 89 samples, 10 were predicted to be of the luminal which is similar to other studies of basal-like breast cancer or subtype. These samples were confi rmed ER- and PR-negative TNBC, and The Cancer Genome Atlas (TCGA) dataset (∼85%; by IHC. Gene expression analysis confi rmed low ESR1 and refs. 13, 14 ). The next most common alterations included PGR expression for all samples, with the exception of two MCL1 (54%) and MYC (35%) gene amplifi cations. MYC ampli- outliers for ESR1 mRNA expression (basal-like and luminal fi cations were detected primarily in basal-like tumors (42% B, respectively, both ER-negative by IHC; data not shown). basal vs. 10% all others; Fisher exact test, P = 0.018) and with Neither of these samples was included in the NGS analysis. a similar frequency as in the basal-like cohort in the TCGA This phenomenon has been noted and discussed elsewhere (Supplementary Table S4). Compared with basal-like primary (18 ), and one possible explanation to the presence of ER/PR– tumors in the TCGA, we detected a higher frequency of MCL1 negative samples with luminal-like gene expression patterns amplifi cations (54% in post-NAC TNBC vs. 19% in TCGA as the “Luminal Androgen Receptor” subtype that expresses basal-like tumors, P = 0.0006), PTEN deletions or mutations the androgen receptor (AR) hormone receptor ( 11 ).

234 | CANCER DISCOVERYFEBRUARY 2014 www.aacrjournals.org

Sequencing of Triple-Negative Breast Tumors after Chemotherapy RESEARCH ARTICLE

AB 40 TSC1 RB1 AKT3 80 AKT2 30 AURKA AKT1 60 RAPTOR CDNK2A RICTOR

CCNE1 40 20 PTEN CCND3 FGFR4 20 ATM CCND2 MET PIK3R1 NF1 FGFR1 10 BRCA2 KIT CRAF CCND1 IGF1R BRAF 0 PIK3CA BRCA1 CDK6 FGFR2 Number of samples with aberrations Number of samples with aberrations KRAS CDK4 EGFR NF1 RB1 0 MYC JAK2 TP53 AKT3 MCL1 PTEN CDK6 KRAS EGFR IGF1R BRCA1 CCNE1

CCND1 CCND2 CCND3 PI3K/mTOR DNA repair RAS/MAPK Cell cycle GFRs PIK3CA CDKN2A Gene

Basal-like C Patients HER2-enriched I II III Molecular Luminal A subtype Luminal B D

Normal-like 80 80 Molecular subtype AR 60 60 HER2 IHC H-score PR 40 40 ER MCL1 amplified 20 20 MYC amplified Ye s % Cellularity sample in gross 0 0 PI3K/mTOR altered % Cellularity in sampled region IIIIII IIIIII DNA-repair altered No Cluster designation Cluster designation Cell-cycle altered NA GFR amplified 100 100 RAS/MAPK altered 80 MEK activation score 50 60 0 40 –50 20

Row Z-score –100 0 gene score TGF- β responsive 6 Ki67 in posttreatment sample (%) IIIIII IIIIII Cluster designation Cluster designation 4

2 40 2 Genes 0 20 –2 1 –4 0 0

–6 –20 –1 MEK score –40 –2 DUSP4 gene expression –60 –3 IIIIII IIIIII Cluster designation Cluster designation

Figure 1. Targetable alterations and pathways in TNBCs after NAC. A, most common recurrently altered genes detected by NGS, representing amplifi - cations, deletions, rearrangements, and known somatic mutations. B, organization and representation of altered genes ( n = 81 tumors) into fi ve functional and targetable pathways. A total of 118 genomic alterations were identifi ed across 81 tumors (1.5 alterations/tumor). C, integrated molecular analysis of residual tumors, using unsupervised clustering based on gene expression patterns (NanoString). D, scatterplots depicting the differences among the clusters identifi ed in C for cellularity in the entire FFPE block cross-section; cellularity in the sampled (macrodissected) hotspot; Ki67 score; TGF-β response signature; MEK signature; and DUSP4 gene expression. *, P < 0.05; **, P < 0.01; ***, P < 0.001.

FEBRUARY 2014CANCER DISCOVERY | 235

RESEARCH ARTICLE Balko et al.

Visualization of expression patterns identifi ed dis- and PI3K/mTOR pathway genes. Although the number of tinct gene signatures that did not correlate with the breast alterations in GFRs and DNA repair genes were low in this cancer molecular subtype or pathway alterations identifi ed subset, signifi cant gains and enrichments were noted in by NGS, but seemed to correlate with the MEK signature these pathways as well. score ( Fig. 1C ). To explore these patterns, we identifi ed the Some paired samples demonstrated gains or enrichments three most prominent gene expression clusters (clusters I–III; across several lesions, suggesting a difference in regional Fig. 1C ) by hierarchical clustering. These clusters contained sampling or tumor purity between the pre- and posttherapy core gene sets of 84, 6, and 30 genes, respectively, expressed specimens. To accommodate these variations, we normalized within each cluster (Supplementary Table S6). Tumor cel- each sample to its estimated tumor purity (see Methods) to lularity in the gross specimen seemed to be a defi ning fac- calculate a fold change in allele or copy-number frequency tor of these clusters, where cluster II represented the most across each tumor pair. This produced an expected pattern paucicellular tumors with the lowest Ki67 staining (Fig. 1D ). of normal distribution around zero of changes in alterations Importantly, the cellularity was less of a defi ning feature after as a result of NAC, assuming that most alterations should considering the cellularity of the “hotspot” regions that were not be selected for or against by chemotherapy (Fig. 2A and macrodissected for gene expression analysis. Thus, the gene B). When analyzed by this method, several alterations were expression patterns may be infl uenced by tumor sampling but highly enriched relative to other within-sample alterations. could also be refl ective of the underlying microenvironment These included two mutations in ATM : R337H and R2443Q, resulting from a strong antitumor effect from neoadjuvant TP53 T253fs*11, a CDH1 splice site deletion, KDM6A L214fs*, therapy. Clusters I and III seemed more similar in terms of cel- AR A401V, and DPYD S175W. When examining CNAs in lularity and Ki67 staining ( Fig. 1D ). However, cluster I had a tumor pairs, we found that copy numbers of AKT and CCND distinct lack of expression of TGF-β–responsive genes. Cluster family members were increased in three of four tumors each. III had a high MEK signature score. This cluster contained a Although copy number of MYC and MCL1 was enriched in group of tumors with low expression of DUSP4 , a negative- several cases following NAC, this effect was not consistent in regulator of the MAPK pathway, offering at least one possible all tumor pairs. Furthermore, there was no clear concordance mechanism of MEK activation (Fig. 1D ). Importantly, survival of case-specifi c enrichment with the therapeutic agents used of the patients comprising these clusters was not signifi cantly for NAC. However, because the frequency of MCL1 amplifi ca- different (data not shown). tions was higher in this post-NAC cohort relative to primary Bioinformatic exploration of these genesets with the tumors in the TCGA, this discordance suggests that MCL1 Molecular Signatures Database ( http://www.broadinstitute amplifi cation may be associated with de novo resistance to .org/gsea/msigdb) suggested that cluster I was driven prima- chemotherapy, but is not enriched further upon treatment. rily by luminal-like breast expression patterns (Supplemen- tary Table S7), despite a lack of ER, PR, or AR IHC staining Coamplifi cation of MYC and MCL1 in these tumors. Furthermore, several overlapping gene signa- in the Residual Disease of TNBC tures suggested that trimethylated H3K27 (H3K27me3) genes The antiapoptosis MCL1 protein is dynamically regulated were highly expressed in this cluster. H3K27me3 maintains during cell-cycle progression and shows rapid turnover rates epigenetic silencing of developmental genes in stem cells, in cancer cells (21 ). To determine whether MCL1 CNAs leaving them poised for expression upon differentiation (19 ). contribute to higher protein levels in breast cancer, we per- Thus, we speculate that cluster I is composed of more dif- formed IHC for MCL1 on tissue microarrays (TMA) of this ferentiated tumors, with high Ki67 and low TGF-β and MEK cohort. MCL1 amplifi cation was signifi cantly associated with activity. In contrast, cluster III–expressed genes overlapped increased protein expression ( P = 0.01; Fig. 3A and B ). How- with invasive signatures across many types of cancer, includ- ever, MCL1 amplifi cation does not seem to be the sole factor ing signatures of poorly differentiated cancers, suggesting in modulating protein expression in breast cancer, as several that this cluster refl ects tumors maintained in a less differ- samples showed high MCL-1 protein levels by IHC in the entiated state and toward a higher stem cell–like hierarchy. absence of CNAs. We also detected three frameshift or non- The expression patterns in cluster III, refl ective of high MEK sense mutations in FBXW7 , encoding the E3 ubiquitin ligase and TGF-β activation (including tumors with DUSP4 loss), responsible for targeting MCL-1 (and MYC) for proteasome- are consistent with stem-like phenotypes induced by these mediated degradation (22 ). However, presence of these pathways, as we have previously demonstrated (10 , 17 , 20 ). mutations was not associated with higher protein levels of MCL-1 ( Fig. 3A ). Selection of Oncogenic Alterations We detected a high degree of concordance between CNAs by Chemotherapy in both MYC and MCL1. MCL1 expression has been shown To quantify enrichment of alterations during NAC, to facilitate MYC-induced lung cancers and leukemogen- we analyzed 20 matched pretreatment biopsies by NGS. esis (23–25 ), although this interaction has not been shown We detected gain (not detected in pretreatment sample in breast cancer. Indeed, 83% of MYC-amplifi ed tumors but detected in posttreatment sample) and enrichment also showed CNAs at MCL1 ( P = 0.001; Fig. 3C ). Cooc- (detected in pretreatment sample but increased in post- curence of MYC and MCL1 amplifi cation was not associ- treatment sample) in mutational allele frequencies and ated with altered prognosis (RFS or OS) as compared copy-number estimations in 41 patient-specifi c alterations with the patients with amplifi cation of either gene alone (Supplementary Fig. S4). Many of these enrichments and in this dataset (data not shown). This co-occurrence was gains occurred in genes comprising cell-cycle regulators also present in the basal-like breast cancers in the TCGA

236 | CANCER DISCOVERYFEBRUARY 2014 www.aacrjournals.org

Sequencing of Triple-Negative Breast Tumors after Chemotherapy RESEARCH ARTICLE

A A C T M F 1.5

1.0

0.5

0.0

–0.5

Change in tumor allele frequency –1.0

–1.5 AR Q58L AR A401V RB1 R358* PTEN Y65* TP53 I255T TP53 I195T ATM D130E ATM WT1 V368A ATM R337H ATM FGFR3 A3T APC E663D TP53 E224* TP53 E339* MYC A287T TP53 M237I ATR M2266I ATR KDR R347H INSR L233F DDR2 splice TP53 V157F IRS2 S903G TP53 W146* TP53 K164E TP52 R110P TP53 R110P TP53 C242Y TP53 R342P TP53 H214R PTEN Q245* MLH1 L729V MSH6 F432L TNKS R498T ATM R2443Q ATM TP53 R174W TP53 R282W NTRK S136R IRS2 S1001A EPHA6 T145I DDR2 E844Q CDH2 P469Q CDH1 D514G PKHD1 splice SUFU G315R IRS2 G1142A NKX2-1 P10R DPYD S175W IKBKE P227Q TP53 N131del TNKS2 S840T RUNX1 Q154* RUNX1 EPHA3 H582D PIK3CA_splice EPHA6 G142E CHEK2 E504Q BRCA1 L1230* SMAD3 R142H TP53 G59fs*64 SMAD4 H132Q PIK3CA R916C MSH6 Y166fs*8 DOT1L K1255N DOT1L TP53 p153fs*18 CEBPA P189del CEBPA CEBPA P188del CEBPA NOTCH1 L433R NOTCH1 RICTOR L1427F RICTOR CDH1_splice-del PDGFRB N130K PIK3CA H1047R PIK3CA H1047R PIK3CA H1047R CDH1 P126fs*89 NOTCH1 L1575P NOTCH1 NOTCH1 S1004L NOTCH1 NOTCH1 L1934M NOTCH1 EPHB6 L227fs*97 BRCA1 S1253fs*1 PIK3R1 L573fs*19 BRCA2 V2118fs*10 BRCA1 P1136fs*19 EPHB6 S171_S172del PTEN C71fs*27_splice RPTOR W165_F167del RPTOR NKX2-1 G263_G265ins BRCA2 C2117_N2119del B A C T M F 10

–5 Change in tumor copy number Change in tumor copy

–10 AR MYC MYC MYC MYC MYC MYC MYC JAK2 AKT2 AKT3 AKT2 AKT1 RAF1 MCL1 MCL1 MCL1 BRAF MCL1 MCL1 MCL1 MCL1 PTEN MCL1 MCL1 PTEN MCL1 PTEN MCL1 CDK4 MCL1 PTEN MCL1 MCL1 KRAS MDM2 IGF1R LRP1B ERBB2 MYCL1 CCNE1 CCND2 CCND3 CCND3 CCND1 RPTOR PIK3CA CDKN2A RAPTOR

Figure 2. Quantitative changes in gene alterations in TNBC tumor pairs before and after NAC. A, change in allele frequency of known and likely somatic mutations during NAC, adjusted for tumor purity assessment. Each segment (n = 20) represents a patient. Type of NAC is depicted with a different color at the top: A, adriamycin/doxorubicin; C, cyclophosphamide; T, taxane; M, methotrexate; F, ﬂ uorouracil. B, change in copy number during NAC, adjusted for tumor purity assessment.

( P < 0.01; Fig. 3D ), although the frequency of MCL1 altera- setting. However, knockdown of these oncogenes increased tions was lower. the fractional killing at lower doses of doxorubicin relative To test whether MCL1 overexpression facilitates MYC- to nontargeting siRNA (siCONTROL)–treated cells ( Fig. 3I ). induced transformation in breast cells, we stably expressed Furthermore, lentiviral-mediated overexpression of MCL1 MCL1 or GFP (control) in MCF10A cells and then trans- increased resistance to doxorubicin and docetaxel (Fig. 3J and duced the cells with a doxycycline-inducible MYC vector Supplementary Fig. S5A and S5B). Resistance to doxorubicin (Fig. 3E ). Although MYC induction induced sporadic trans- was mediated in part by decreased baseline and doxorubicin- formation of MCF10A ascertained by anchorage-independent mediated apoptosis ( Fig. 3K ). Thus, MYC and MCL1 enhance growth assays, concurrent overexpression of MCL1 mark- cell fi tness, and MCL1 additionally protects TNBCs from edly increased MCF10A colony formation (Fig. 3F and G). chemotherapy-induced apoptosis. Furthermore, in TNBC cell lines demonstrating gains or amplifi cations in MYC and/or MCL1 including HCC1143 Molecular Alterations in the Residual Disease ( MYC-amplifi ed and MCL1 gains), HCC1395 ( MYC-ampli- after NAC Correlate with Patient Outcome fi ed), and MDA-436 (MYC -amplifi ed and MCL1-amplifi ed; Next, we explored the prognostic impact of genomic alter- ref. 13 ), siRNA knockdown of MYC and MCL1 reduced ations and gene expression signatures identifi ed by NGS cell viability ( Fig. 3H and I ). siRNA-targeting of MYC and and NanoString analysis, respectively (15 , 17 ). Gene-specifi c MCL1 did not alter the relative sensitivity to doxorubicin, a alterations occurring in at least eight (>10%) analyzed tumors commonly used chemotherapeutic agent in the neoadjuvant were tested for prognostic impact (RFS and OS) by the

FEBRUARY 2014CANCER DISCOVERY | 237

RESEARCH ARTICLE Balko et al.

ABMCL1 IHC CDMCL1 status P = 0.01 Altered in 31 (38%) of cases Normal 3 P Amplified MYC 31% < 0.01 P 2 = 0.001 MCL1 19% Fisher exact test 100 Amplification 1 20 CNAs are putative.

0 Normal MCL1 20 50 –1 E MCL1 signal (a.u.) 28

% of cases GFP MCL1 –2 – + – + DOX 4 0 MYC–HA Normal Amplified Normal MCL1–V5 MCL1 status Amplified Amplified MCL1 MYC status Calnexin F G H P < 0.01 Vehicle DOX (100 ng/mL) 200

150 HCC1143 HCC1395 MDA436 cells)

4 –– ++ ––++ – – ++ siMYC 100 +––– – ++ ––+++ siMCL1

pLX302–GFP pINDUCER-MYC 50 (per 1 × 10

Number of colonies MCL1 0 cMYC GFP MCL1

GFP + DOX Calnexin MCL1 + DOX pLX302–MCL1 pINDUCER-MYC

Figure 3. Coamplifi cation and interaction of MYC and MCL1 in TNBCs. A, MCL1 IHC score as quantifi ed on TMAs of TNBCs after NAC. Signal intensity (a.u., arbitrary units) was normalized for tumor area and number of nuclei. FBXW7 -mutant tumors are shown as green triangles. B, example images of high and low MCL1–expressing tumors by IHC. C, coamplifi cation of MCL1 and MYC in residual breast tumors assessed in this study. Absolute numbers of tumors are shown in bars (P = 0.001, Fisher exact test). D, coamplifi cation of MCL1 and MYC in primary basal-like breast tumors in the TCGA (13 , 37 ). E, Western blot analysis of MCF10A cells expressing pLX302–GFP (control) or pLX302–MCL1 (V5-tagged) and pINDUCER22–MYC (HA tagged) ± doxycycline treatment. F, soft agar colony formation assay of MCF10A cells in E ± doxycycline (DOX) where indicated. G, quantifi cation of colonies from F. Each bar represents the mean colony number of triplicate wells ± SD. H, siRNA knockdown of MYC and MCL1 in HCC1143 ( MYC-amplifi ed and MCL1 -gain), HCC1395 ( MYC-amplifi ed), and MDA-436 (MYC- and MCL1-amplifi ed) cells (13 ). (continued on following page)

likelihood ratio test (Table 1). Alterations in pathways, gene cooperate with the MYC oncogene ( 26, 27 ). Thus, we tested expression signature scores, and clinical variables were also the possibility that these perturbations may interact with tested. Of note, JAK2 amplifi cation was strongly associated one another. When the interaction term was tested by Cox with poor RFS (P = 0.006; HR, 3.36), whereas BRCA1 trunca- proportional hazards analysis, a signifi cant interaction was tions/mutations and JAK2 amplifi cation predicted poor OS noted for RFS but not for OS (P = 0.03 and 0.83, respec- ( P = 0.041; HR, 2.5 and P = 0.002; HR, 4.16, respectively). tively). Kaplan–Meier analysis confi rmed this association PTEN alterations were a favorable prognostic factor for OS (Fig. 4A–C ). ( P = 0.03; HR, 0.14). MYC amplifi cations also demonstrated The effect of the interaction between MYC and MEK acti- a trend toward worse OS (P = 0.084; HR, 1.78). When altera- vation on patient outcome suggested a mechanistic inter- tions were categorized into functional pathways, DNA repair action linking these pathways to tumor progression. This alterations were weakly associated with poor OS (P = 0.09; cooperation has been demonstrated in experimental models, HR, 1.89). Interestingly, a high MEK activation score ( 15 ) where MEK stabilizes MYC expression ( 27–34 ). For example, predicted poor RFS and OS (P = 0.059; HR, 1.758 and P = c-MYC overexpression in transgenic mice results in spon- 0.013; HR, 2.264, respectively). These data offer insights into taneous breast tumors that activate MEK through the gen- molecular alterations that may predict the natural history of eration of KRAS mutations ( 30 ). To test this interaction on a TNBC after NAC. molecular level, we used MCF10A cells stably transduced with MYC (5XMYC; ref. 35 ). Stable expression of MYC induced Prognostic Interaction of MEK the formation of anchorage-independent MCF10A colonies. Activation and MYC Amplifi cation However, treatment with a single dose of a MEK1/2 inhibitor MYC amplifi cations in the residual disease trended toward (GSK1120212/trametinib or AZD6244/selumetinib) com- an association with poor OS, whereas high MEK transcrip- pletely abolished the ability of MYC to induce MCF10A tional output was a risk factor for reduced RFS and OS. colonies ( Fig. 4D and Supplementary Fig. S6A and S6B). Importantly, the RAS–MAPK pathway has been shown to This effect was MEK-specifi c, as treatment with the

238 | CANCER DISCOVERYFEBRUARY 2014 www.aacrjournals.org

Sequencing of Triple-Negative Breast Tumors after Chemotherapy RESEARCH ARTICLE

IJHCC1143 HCC1395 MDA436 Anova P < 0.0001 Anova P < 0.0001 Anova P < 0.0001 SUM159PT MDA468 1.2 1.2 0.3 0.15 0.25 GFP GFP 1.0 MCL1 1.0 MCL1 0.20 0.8 0.8 0.2 0.10 0.15 0.6 0.6 0.10 0.1 0.05 0.4 0.4 Relative viability Relative Relative viability Relative 0.05 Relative viability (a.u.) Relative Relative viability (a.u.) Relative 0.2 0.2 Relative viability (a.u.) Relative 0.0 0.00 0.00 0.0 0.0 012345 012345 iMYC MCL1 siMYC s iMCL1 MYC Log [doxorubicinl] (nmol/L) Log [doxorubicinl] (nmol/L) siMCL1 siMCL1 s si iCONTROL s siCONTROL siCONTROL MYC + siMCL1 si siMYC + si siMYC + siMCL1 K SUM159PT MDA468 HCC1143 HCC1395 MDA436 2.5 4 1.5 1.5 1.5 siCONTROL siMCL1 GFP 2.0 siMYC MCL1 siMCL1 + siMYC 3 1.0 1.0 1.0 1.5 2 0.5 0.5 0.5 1.0 1 0.5 0.0 0.0 0.0 Relative caspase 3/7 activity Relative Relative caspase 3/7 activity Relative Fraction control viability Fraction Fraction control viability Fraction Fraction control viability Fraction 0.0 0

/L) –0.5 –0.5 –0.5 ol/L) ol/L) DMSO DMSOmol/L) DMSO DMSOmol/L) 012345 012345 012345 μ μ Dox [log10 nmol/L] Dox [log10 nmol/L] Dox [log10 nmol/L] OX (2 OX (2 DOX (2 μmol/L)D DOX (2 μmol/L)D DOX (10 μm DOX (10 μmol/L) DOX (10 μm DOX (10 μmol

Figure 3. (Continued) I, baseline viability (top) and response to dose titration of 72 hours of doxorubicin (bottom) of cell lines after siRNA knockdown. Viability curves are shown as relative to siCONTROL, DMSO-treated controls. *, p < 0.05; **, p < 0.01; ***, p < 0.001. J, viability curves of cells transduced with MCL-1 or GFP control and treated for 48 hours with a dose titration of doxorubicin. Viability was measured with CellTiter-Glo (Promega). K, caspase cleavage in cells from J after 5 hours doxorubicin at the indicated doses. Caspase-3/7 cleavage was measured with Caspase-Glo (Promega).

pan-PI3K inhibitor BKM120 had no effect on MCF10A Of note, we detected two tumors (2.5%) with presumed loss- anchorage-independent growth. Treatment of 5XMYC cells of-function alterations in TSC1 : a truncation of an intergenic with a MEK inhibitor resulted in the formation of polarized region between intron 8 and the 3′-untranslated region (3′- normal acini, as observed by immunofl uorescence for basal UTR) contained in the 23rd intron (Supplementary Fig. S7), (CK5, vimentin), luminal (CK8, e-cadherin), and a tight- which was detected in matched pre- and post-NAC samples in junction marker (ZO-1; Fig. 4E and Supplementary Fig. S6C 1 patient, and a nonsense mutation (TSC1 _Q516*, 20% allele and S6D). These data suggest that MYC cooperates with frequency) in another patient (Supplementary Table S2). Loss RAS/MAPK to drive anchorage-independent growth in breast of TSC1 function has been proposed as a basis for signifi cant cancer. clinical responses in metastatic bladder cancer to the TORC1 inhibitor everolimus ( 36 ), and may also represent a thera- Molecular Profi ling for Rational Selection of peutically actionable target in patients with breast cancer. Adjuvant Targeted Therapies Importantly, loss-of-function TSC1 deletions, truncations, or Despite the high likelihood of recurrence, the current nonsense mutations were not identifi ed in the TCGA breast standard of care for patients with TNBC who do not achieve cancer study (13 , 37 ). a pCR after NAC is a watchful waiting. Patients who recur with metastatic cancer are less likely to exhibit prolonged responses to conventional anticancer therapy, as the microme- DISCUSSION tastases that generated these clinical recurrences have already Herein, we have described the genomic landscape of drug- been exposed to chemotherapy in the neoadjuvant setting. resistant tumor cells remaining in the breasts of patients This unmet need suggests that the identifi cation of action- with TNBC after anticancer chemotherapy. We also per- able molecular targets in the residual disease could, in turn, formed serial analysis to detect changes in CNAs and muta- be explored in adjuvant trials after NAC and mastectomy. tions before and after NAC. These data provide insights into Because the spectrum of alterations present in such tumors is genomic alterations that may predict de novo or acquired highly diverse, we integrated existing preclinical and clinical resistance to standard anticancer therapies in TNBC and data into an “actionability” table of rational therapies for the could inform on the effective use of rational molecularly tar- identifi ed alterations ( Table 2 and Supplementary Table S6). geted agents in adjuvant trials. In an effort not to confound

FEBRUARY 2014CANCER DISCOVERY | 239

RESEARCH ARTICLE Balko et al.

Table 1. Prognostic ability of clinical factors and molecular alterations

RFS OS

n Events HRP n Events HRP Clinical (all patients)a Ki67 score > 15 105 60 0.915 0.772 108 52 1.179 0.623 Node positive 104 60 1.664 0.068 107 52 1.534 0.147 Age > 50 y 105 60 0.932 0.790 108 52 0.872 0.630 Postmenopausal 105 60 0.681 0.138 108 52 0.711 0.221 Neoadjuvant taxane 105 60 1.255 0.390 108 52 1.157 0.625 Genes (TNBC only) MCL1 = amp 64 40 0.620 0.130 67 36 0.620 0.140 MYC = amp 64 40 1.370 0.400 67 36 1.780 0.084 PIK3CA = amp/mut 64 40 0.530 0.380 67 36 0.300 0.210 BRCA1 = trunc/mut 64 40 1.700 0.260 67 36 2.500 0.041 RB1 = del/trunc 64 40 0.910 0.850 67 36 0.760 0.660 PTEN = del/trunc/mut 64 40 0.570 0.280 67 36 0.140 0.030 JAK2 = amp 64 40 3.360 0.006 67 36 4.160 0.002 Pathways (TNBC only) Cell-cycle altered 64 40 0.580 0.100 67 36 0.560 0.110 DNA repair altered 64 40 1.570 0.253 67 36 1.890 0.090 PI3K/mTOR altered 64 40 0.813 0.540 67 36 0.660 0.280 RAS/MAPK altered 64 40 1.000 0.980 67 36 0.860 0.755 GFR ampliﬁ ed 64 40 1.030 0.920 67 36 1.220 0.600 Gene signatures (TNBC only) MEK signature = int/high 79 42 2.120 0.035 81 37 3.170 0.004 TGF-β signature = int/ 79 42 1.250 0.500 81 37 1.570 0.207 high

NOTE: P ≤ 0.05 (signifi cant) or P ≤ 0.1 (statistical trend) defi ned in bold italics. a Clinical data analysis of clinically defi ned TNBC includes 7 samples later identifi ed as HER2 amplifi ed by NGS.

our results with tumors with variable residual cancer burden, These included PTEN alterations (PI3K and AKT inhibitors), we focused on a cohort of cancers with signifi cant macro- and amplifi cations of JAK2 (ruxolinib or tofacitinib), CDK6 , scopic residual disease after NAC. Indeed, this represents a CCND1, CCND2 , CCND3 (CDK4/6 inhibitors), and IGF1R cohort with a particularly poor prognosis (median survival (dalotuzumab). Importantly, several patients’ tumors showed ∼18 months). an enrichment of AKT family CNAs and CCND family CNAs Several molecular insights were gained through this analysis. after NAC, suggesting an association of these alterations with We showed that the Ki67 score after NAC does not provide resistance to chemotherapy. TSC1 truncations and mutations prognostic information in patients with TNBC. Furthermore, were also identifi ed. These alterations have been associated we confi rmed our previous report demonstrating that Ki67 with high sensitivity to the TORC1 inhibitor everolimus in the residual disease is intimately related to the underlying in other tumor types ( 36 ), suggesting they generate tumor molecular subtype ( 10 ). We also found frequent coamplifi cation dependence on the mTOR pathway. of MCL1 and MYC that conferred an advantage in anchorage- Overall, this analysis provides new information on the independent growth. Importantly, these coamplifi cations were molecular alterations present in chemotherapy-resistant more frequent in this study as compared with those previously tumor cells within TNBCs. As supported by the poor outcome reported in primary basal-like breast tumors. Amplifi cation of of patients with TNBC that recurs with metastatic disease the MYC oncogene coinciding with gene expression signatures after an incomplete response to NAC, we surmise that these of MEK activity identifi ed a group of patients with very poor persistent tumor cells are resistant to conventional cytotoxic prognosis. Furthermore, MEK inhibitors potently inhibited chemotherapies without the addition of novel agents target- three-dimensional growth of MYC-overexpressing cells, suggesting these oncogenic pathways. Furthermore, these data sug- ing a role for MEK inhibitors in MYC -amplifi ed breast cancers. gest that molecular analysis of TNBCs not achieving a pCR We also detected a higher frequency of several potentially to NAC should be performed routinely to stratify patients targetable alterations in this cohort of posttreatment TNBCs according to this information to rational adjuvant trials with compared with basal-like primary breast cancers in the TCGA. molecularly targeted agents.

240 | CANCER DISCOVERYFEBRUARY 2014 www.aacrjournals.org

Sequencing of Triple-Negative Breast Tumors after Chemotherapy RESEARCH ARTICLE

A B C 100 100 100 N N N = 86 = 62 = 62 P P = n.s. P = 0.03 80 = 0.01 80 80

60 60 60

40 40 40

20 20 20 Percentage recurrence-free Percentage Percentage recurrence-free Percentage Percentage recurrence-free Percentage 0 0 0 0246810 024 6810 0246810 Time (y) Time Time (y) Low MEK score MYC not amplified MYC not amplified AND low MEK score High MEK score Amplified MYC amplified OR high MEK score MYC amplified AND high MEK score

DE150 Vector 5× MYC E-cadherin/vimentin/DAPI

100 cells) 4

50 (per 1 × 10 Number of colonies

0 5 × MYC Vector

DMSO DMSO 1% DMSO 4 h 24 h 48 h GSK112, 100 nmol/L

AZD6244, 100BKM120, nmol/L 100 nmol/LAZD6244, 100BKM120, nmol/L 100 nmol/L GSK1120212, 10 nmol/L GSK1120212, 10 nmol/L

Figure 4. Interaction of MYC amplifi cation with MEK pathway activity correlates with poor prognosis in TNBCs. A, Kaplan–Meier analysis of RFS in patients with a high MEK transcriptional signature (ref. 15 ; highest 66%) versus all others (lowest 33%). B, Kaplan–Meier analysis of RFS in MYC -amplifi ed tumors versus those with normal MYC copy number. C, combined Kaplan–Meier analysis of patients with a high MEK transcriptional signature and MYC amplifi cation versus those with either or neither alteration. D, quantifi cation of 3-week soft-agar colony formation assays using MCF10A cells stably transduced with MYC (5× MYC) versus vector control, plated in the presence or absence of a single dose of AZD6244/selumetinib, GSK1120212/trametinib, or the pan-PI3K inhibitor BKM120 at the indicated concentrations. Bars represent the mean colony number ± SD of three replicates. E, immunofl uorescence of E-cadherin, vimentin, and DAPI in cells from D grown on chamber slides and treated with 100 nmol/L GSK1120212/trametinib. Scale bars represent 50 μm.

METHODS (m7240; DAKO) was used at a 1:75 dilution overnight. Visualiza- tion was performed using the 4 Plus Detection System (Biocare) and Patients and Tumor Specimens 3,3′-diaminobenzidine (DAB; DAKO) as the chromogen. The section Surgically resected tumor samples (N = 111) were from patients with was scanned at ×100 magnifi cation and the area containing the high- TNBC diagnosed and treated with NAC at the Instituto Nacional de est number of positive cells was selected. Positive and negative tumor Enfermedades Neoplásicas (Lima, Perú). Clinical and pathologic data cells were manually counted at ×400; the percentage of positive cells were retrieved from medical records under an institutionally approved was calculated with at least 700 viable cells. protocol (INEN 10-018). Tumors were determined to be triple-negative Antigen retrieval for ER and PR was performed using citrate if they were negative for ER, PR, and HER2 overexpression measured by buffer (pH 6) in a decloaking chamber (Biocare Medical). The ER IHC. A subset of cases was subjected to HER2 FISH to resolve discrep- (6F11; Vector Laboratories) and PR (PgR636; DAKO) antibodies ant fi ndings between the HER2 IHC results and the PAM50 subtype were used at 1:200 and 1:50 dilutions, respectively, for a 1-hour assignment. The results were further verifi ed by comparison with the incubation. Visualization for both antibodies was performed using NGS results. The diagnostic biopsy (pre-NAC) was obtained for NGS the Envision Detection System (DAKO) and DAB (DAKO) as the analysis in a subset (n = 20) of these patients. chromogen. The percentage of invasive tumor cells with nuclear staining and the average intensity of all positively staining tumor Immunohistochemistry cells in the section were manually counted as per the CAP/ASCO Antigen retrieval for Ki67 was performed using HpH Buffer (pH (College of American Pathologists/American Society of Clinical 9.0) in a decloaking chamber (Biocare Medical). The Ki67 antibody Oncology) guidelines ( 38 ).

FEBRUARY 2014CANCER DISCOVERY | 241

RESEARCH ARTICLE Balko et al.

Table 2. Actionability of lesions identiﬁ ed in at least three post-NAC specimens

Gene symbol # Altered Category Potential therapy TP53 73 D Prognostic (poor, potentially sensitive to WEE1 inhibitors, e.g., MK1775) MCL1 40 C Resistance to anti-tubulins, e.g., paclitaxel, MCL1 inhibitor in development MYC 24 C Aurora kinase inhibitors, e.g., MLN8237, AMG 900; possible sensitivity to CDK inhibitors PIK3CA 13 B PI3K/mTOR inhibitors, e.g., everolimus, temsirolimus, and others PTEN 12 B PI3K/mTOR inhibitors, e.g., GSK2636771, everolimus, temsirolimus, and others BRCA1 9 B PARP inhibitors, e.g., olaparib, CEP-9722, rucaparib, and others RB1 9 D Prognostic JAK2 8 D JAK2 inhibitors, e.g., ruxolitinib, and others ERBB2 7 A Herceptin, lapatinib, and others CDKN2A/B 7 E CDK4/6 inhibitors, e.g., PD0332991, LEE011, P276-00 NF1 5 C MAPK/PI3K/mTOR inhibitors, e.g., MSC1936369B, everolimus, temsirolimus, and others AKT3 5 C AKT inhibitors, e.g., MK2206, PI3K/mTOR inhibitors, e.g., everolimus, temsirolimus KRAS 5 A Resistance to cetuximab, MEK inhibitors, e.g., MEK162 CCND1 5 C CDK4/6 inhibitors, e.g., PD0332991, LEE011, P276-00 CCND3 4 C CDK inhibitors, kinetin riboside CCNE1 4 C CDK2/4/6 inhibitors, e.g., ABT-888, PD0332991, LEE011, P276-00 CCND2 4 C CDK inhibitors, kinetin riboside CDK6 4 C CDK4/6 inhibitors, e.g., PD0332991, LEE011, P276-00 IGF1R 4 C IGF-IR inhibitors, e.g., AMG-479, BMS-754808, MK-0646, IMC A12, and others LRP1B 3 E Biologically relevant, presently no known targeted therapies PIK3R1 3 C PI3K pathway inhibitors ATM 3 C PARP inhibitors, e.g., olaparib, CEP-9722, rucaparib BRCA2 3 B PARP inhibitors, e.g., olaparib, CEP-9722, rucaparib, and others EGFR 3 A Cetuximab, panitumumab, and others FBXW7 3 C Resistance to anti-tubulins, potential sensitivity to PI3K/mTOR inhibitors CDK4 3 C CDK4/6 inhibitors, e.g., PD0332991, LEE011, P276-00 RPTOR 3 E Biologically relevant, possible sensitivity to mTORC1 and mTORC2 inhibitors Category A: approved/standard alterations that predict sensitivity or resistance to approved/standard therapies Category B: alterations that are inclusion or exclusion criteria for speciﬁ c experimental therapies Category C: alterations with limited evidence that predict sensitivity or resistance to standard or experimental therapies Category D: alterations with prognostic or diagnostic utility Category E: alterations with clear biologic signiﬁ cance in cancer (i.e., driver mutations) without clear clinical implications to date

Abbreviation: IGF-IR, insulin-like growth factor-I receptor.

Antigen retrieval for HER2 was performed using HpH Buffer (pH 9.0) titated using the Ariol SL-50 automated microscope system (Leica in a decloaking chamber (Biocare Medical). The HER2 antibody (#2242; Microsystems) at ×20. Selected areas at original resolution are dis- Cell Signaling Technology) was used at a 1:200 dilution overnight. Visu- played. Immunoreactivity intensity scores were determined in areas alization was performed using the Envision Detection System (DAKO) of residual tumor cells selected by expert breast pathologists (J.M. and DAB (DAKO) as the chromogen. The percentage of invasive tumor Giltnane and M.G. Kuba) and averaged for redundant tissue cores. cells with membranous staining at the highest intensity level was manually assessed and recorded as per the CAP/ASCO guidelines ( 39 ). HER2 FISH Antigen retrieval for MCL1 was performed in citrate buffer (pH FISH for detection of ampliﬁ cation of HER2 was performed using 6.0) under pressure for 15 minutes; endogenous peroxidase activity the PathVysion HER-2 DNA Probe Kit (PathVysion Kit; Abbott Molec- was blocked by incubating with 3% H2 O 2 for 10 minutes. The sections ular) using the Vysis LSI HER-2/neu 17q11.2-12 SpectrumOrange were incubated with MCL1 antibody (Santa Cruz Biotechnology; and Vysis CEP 17 17p11.1-q11.1 SpectrumGreen Alpha Satellite DNA sc-819) at 1:800 dilution overnight at 4°C and developed by using probes. Images were visualized on a Fluorescence Olympus BX60 DAB substrate (Vector Laboratories). Automated slide scanning and Microscope and analyzed using the Genus for Genetic Image Analysis scoring were performed at the Vanderbilt Epithelial Biology Center software, version 3.6. The ratio of HER2 to CEP 17 signals was recorded Imaging Resource (Nashville, TN). Images were captured and quan- and reported as an average ratio as per the CAP/ASCO guidelines ( 39 ).

242 | CANCER DISCOVERYFEBRUARY 2014 www.aacrjournals.org

Sequencing of Triple-Negative Breast Tumors after Chemotherapy RESEARCH ARTICLE

Immunofl uorescence Genomic Alteration Detection Immunofl uorescence staining was performed as described previously Base substitution detection was performed using a Bayesian meth- ( 40 ). All primary and secondary antibodies were diluted in 12% Frac- odology, which allows detection of somatic mutations at a low tion V BSA (RPI-Cat#A30075). The following antibodies and dilutions mutation allele frequency and increased sensitivity for mutations at were used: ZO-1 (Life Technologies; Cat#617300) 1:200, E-cadherin hotspot sites ( 44 ) through the incorporation of tissue-specifi c prior (BD-Cat#610182) 1:200, vimentin (Covance; Cat#PCK-594P) 1:500, expectations: CK5 (Covance; Cat#PRB-160P-100) 1:500, and CK8 (RDI Fitzgerald; = > Cat#20R-CP004) 1:500. Secondary goat antibodies were highly cross- P(Mutation present | Read data “R”) P(Frequency of muation “F” 0 | R) ∝ − = = adsorbed and used at 1:200 (Molecular Probes/Life Technologies). 1 P(R | F 0) P(F 0) Chamber slides were briefl y rinsed in PBS and fi xed in 4% parafor- where P ( R | F) is evaluated with a multinomial distribution of the maldehyde with 0.1% Triton X-100 for 20 minutes. Cells were rinsed observed allele counts using empirically observed error rates and P(F = 0 ) three times with PBS and primary antibodies were applied overnight. is the prior expectation of mutation in the tumor type. To detect indels, Secondary antibodies were incubated for 20 minutes and rinsed four de novo local assembly in each targeted exon was performed using the times in PBS then mounted in SlowFade + DAPI (4′,6-diamidino-2- de-Bruijn approach (45 ). Candidate calls were fi ltered using a series of phenylindole; Molecular Probes/Life Technologies). quality metrics, including strand bias, read location bias, and a custom Tissue Microarrays database of sequencing artifacts derived from normal controls. Germ- line alterations were identifi ed and fi ltered using the Single Nucleotide Triplicate 0.6-mm cores were punched from post-NAC FFPE Polymorphism Database ( dbSNP; version 135; http://www.ncbi.nlm tumor blocks from the patient’s surgical resection specimen using .nih.gov/projects/SNP/), 1000 Genomes (http://www.1000genomes the Beecher Manual Tissue Arrayer MTA-1 (Beecher Scientifi c). Three .org/), and subsequently annotated for known and likely somatic arrays containing approximately 100 cores per array were generated. mutations using the COSMIC database (version 62; http://cancer .sanger.ac.uk/cancergenome/projects/cosmic/). Detection of CNAs Gene Expression Analyses was performed by obtaining a log-ratio profi le of the sample by RNA was isolated from FFPE tumor blocks by macrodissecting normalizing the sequence coverage obtained at all exons against a tumor-rich regions from three to six 10-μm sections. RNA was purifi ed process-matched normal control. The profi le was segmented and using the RNeasy FFPE Kits (Qiagen). Gene expression analysis was interpreted using allele frequencies of approximately 1,800 additional performed by NanoString as previously described ( 10 ). Raw transcript genome-wide SNPs to estimate tumor purity and copy number based counts were subtracted from background (negative input control). on established methods (46–48 ) by fi tting parameters of the equation Normalization of raw transcript counts was performed by dividing the geometric mean of seven housekeeper-control genes: NPAS1 , NAGA, ⎛ + ⎞ ∼ pCCseg ()p) 2 lrrseg N ⎜ log 2 ⎟ POLR1B , CD40 , WAS , B2M, and TUBB . Housekeeper-normalized tran- ⎝ pp×+lid (1 )× 2⎠ script counts were log2 transformed and data were row z -score stand- ardized before further analysis. PAM50 analysis was performed using the PAM50 quantitative real-time PCR (qRT-PCR) centroids ( 7 ) on the where lr seg , C seg , and p are the log-ratios and copy numbers at each segment and sample purity, respectively. Focal amplifi cations are called normalized log2 gene expression data before z -standardization. at segments with ≥6 copies and homozygous deletions at 0 copies, in Sequencing and Primary Sequence Data Analysis samples with tumor cell purity >20%. To normalize for tumor content between pre- and post-NAC– Eighty-fi ve post-NAC tumors and 26 pre-NAC biopsies were submit- matched samples, allele frequencies or copy-number estimations ted for NGS at Foundation Medicine Inc. A 4-μm of hematoxylin and were divided by fractional tumor purity to calculate normalized allele eosin–stained slide was reviewed by an expert pathologist to ensure (i) frequency or copy number for the individual sample. The pre-NAC a sample volume of ≥1 mm3 , (ii) nucleated cellularity ≥80% or ≥30,000 sample-normalized frequency/copy number was subtracted from the cells, and (iii) that ≥20% of the nucleated cells in the sample are derived post-NAC sample-normalized frequency or copy number to calculate from the tumor. DNA was extracted from 40 μm of unstained FFPE the absolute change in allele frequency or copy number. sections, typically 4 μm × 10 μm sections by digestion in a Proteinase K buffer for 12 to 24 hours and then purifi ed with the Promega Max- well 16 Tissue LEV DNA Kit; 50 to 200 ng of double-stranded DNA TSC1 Deletion Verifi cation (dsDNA) in 50 to 100 μL water in microTUBEs were fragmented to PCR primers amplifying across the predicted breakpoint in intron approximately 200 bp by sonication (3 minutes, 10% duty, intensity = 5, 23 (F: ACCCAATCTCACCAAGGTCC; R: CACCTTTTCTCGCTGAAA 200 cycles/burst; Covaris E210) before purifi cation with a 1.8× volume GCA, product: 100 bp) and spanning the truncation from intron of AMPure XP Beads (Agencourt). Solid Phase Reversible Immobilisa- 23-intron 8 (F: AACAACTCTCCAAGAGTATCCGCAG R: TGGTG tion (SPRI) purifi cation and subsequent library construction with the CGAAAAGAGCTGTTGCTT, product: 170 bp) were used for PCR NEBNext Kits (E6040S, NEB), containing mixes for end repair, dA detection of the truncated allele. addition, and ligation, were performed in 96-well plates (Eppendorf) on a Bravo Benchbot (Agilent) using the “with-bead” protocol 43 to maxi- Cell Culture mize reproducibility and library yield. Indexed (6-bp barcodes) sequenc- MCF10A cells were cultured in DMEM/F12 nutrient mix with 5% ing libraries were PCR-amplifi ed with HiFi (Kapa) for 10 cycles, 1.8× horse serum (GIBCO), 20 ng/mL EGF, 0.5 mg/mL hydrocortisone, SPRI-purifi ed and quantifi ed by quantitative PCR (qPCR; Kapa SYBR 100 ng/mL cholera toxin, and 10 μg/mL insulin (all from Sigma). Fast) and sized on a LabChip GX (Caliper); size selection was not per- HCC1143 and HCC1395 were cultured in RPMI + 10% FBS (GIBCO); formed. Paired-end sequencing (49 × 49 cycles) was performed using the MDA-436 and MDA-468 were cultured in Dulbecco’s Modifi ed Eagle HiSeq2000 (Illumina). Sequence data from genomic DNA were mapped Medium (DMEM) + 10% FBS. SUM159PT cells were cultured in to the reference human genome (hg19) using the Burrows-Wheeler DMEM + 5% FBS and 0.5 μg/mL hydrocortisone. All cells were ° Aligner (BWA) ( 41 ). PCR duplicate read removal and sequence metric cultured at 37 C in at 5% CO2 . MCF10A cells were purchased from collection was performed using Picard (http://picard.sourceforge.net ) the American Type Culture Collection (ATCC). All TNBC cell lines and SAMtools (42 ). Local alignment optimization was performed using (HCC1143, HCC1395, MDA-436, MDA-468, and SUM159PT) were the Genome Analysis toolkit (GATK; ref. 43 ). obtained from the sources described in ref. 11 and were confi rmed

FEBRUARY 2014CANCER DISCOVERY | 243

RESEARCH ARTICLE Balko et al. by DNA fi ngerprinting (Cell Line Genetics) on March 24, 2011, to Analysis and interpretation of data (e.g., statistical analysis, consist of a single cell line and to match the DNA fi ngerprint on fi le biostatistics, computational analysis): J.M. Balko, J.M. Giltnane, at ATCC. Frozen stocks archived shortly after fi ngerprinting from the K. Wang, M.E. Sanders, P.D. Moore, J.A. Pinto, H. Gómez, J.A. Bauer, confi rmed cell lines were used in these studies. J.A. Pietenpol, J.S. Ross, R. Yelensky, V.A. Miller, C.L. Arteaga Writing, review, and/or revision of the manuscript: J.M. Balko, Chemicals J.M. Giltnane, K. Wang, R.S. Cook, M.E. Sanders, J.A. Pinto, GSK1120212 (trametinib), AZD6244 (selumetinib), and BKM-120 H. Gómez, J.A. Bauer, J.A. Pietenpol, J.S. Ross, G.A. Palmer, V.A. were purchased from Selleckchem, dissolved in dimethyl sulfoxide Miller, P.J. Stephens, C.L. Arteaga (DMSO) and used at dilutions resulting in a fi nal concentration of Administrative, technical, or material support (i.e., reporting or <0.1% DMSO in all studies in vitro. Doxorubicin was purchased from organizing data, constructing databases): J.M. Balko, L.J. Schwarz, Sigma and was solubilized in DMSO at a concentration of 100 mmol/L. R.S. Cook, P. Owens, V. Sánchez, P.D. Moore, D. Horiuchi, A. Goga, C.L. Arteaga siRNA Knockdown Study supervision: J.M. Balko, M. Cronin, C.L. Arteaga siRNA knockdown was performed as previously described ( 20 ). Cells were transfected with 20 nmol/L siCONTROL (nontargeting Grant Support siRNA), 10 nmol/L siMYC + 10 nmol/L siCONTROL, 10 nmol/L This work was supported by the NIH Breast Cancer Special- si MCL1 + 10 nmol/L siCONTROL, or 10 nmol/L si MYC + 10 nmol/L ized Program of Research Excellence (SPORE) grant P50 CA98131; si MCL1 . Constructs for siMYC and si MCL1 were purchased from Vanderbilt-Ingram Cancer Center Support Grant P30 CA68485 and Ambion (s9129 and s8583, respectively). Core Services Grants CA14837, CA105436, and CA68485; The Lee Jeans Translational Breast Cancer Research Program (to C.L. Arteaga); MCL-1 Overexpression Susan G. Komen for the Cure Foundation grants SAC100013 (to Cells were transduced with lentiviral particles derived from 293FT C.L. Arteaga) and SAC110030 (to J.A. Pietenpol); Komen Career cells transfected with pLX302-MCL1 or GFP. GFP and MCL1 vectors Catalyst grant KG100677 (to R.S. Cook) and Komen Post-Doctoral were purchased from Thermo Scientifi c (Open Biosystems). award PDF12229712 (to J.M. Balko); Department of Defense Post- Doctoral Fellowship award W81XWH-12-1-0026 (to C.D. Young) Soft Agar Colony Formation Assays and Idea Award BC120793 (to R.S. Cook); and NIH R01CA143126 (to R.S. Cook). These assays were carried out in 6- or 12-well dishes using 5 × 104 × 4 or 1 10 cells, respectively. A single-cell suspension in 0.4% agarose Received June 13, 2013; revised December 7, 2013; accepted × in 1 media was layered on the top of a bottom layer of 0.8% agarose December 11, 2013; published OnlineFirst December 19, 2013. in 1× media in the presence of inhibitors or DMSO (control). Fresh 1× media (no drug) was applied to cells every 3 to 4 days to protect against dehydration. Colonies measuring >80 μm were counted after 2 to 3 weeks on a Gelcount Scanner (Oxford Optronix). REFERENCES Immunoblotting 1. Guarneri V , Broglio K , Kau SW , Cristofanilli M , Buzdar AU , Valero V , et al. Prognostic value of pathologic complete response after primary Immunoblotting was carried out as described previously ( 43 ). chemotherapy in relation to hormone receptor status and other fac- Antibodies used for immunoblotting were: p-ERK1/2 (p-T202/Y204; tors. J Clin Oncol 2006 ; 24 : 1037 – 44 . #9101), ERK1/2 (#9102), p-AKT (p-S473; #9271), AKT (# 9272), HA- 2. Kuerer HM , Newman LA , Smith TL , Ames FC , Hunt KK , Dhingra K , tag (#2367), calnexin (#2433; all from Cell Signaling Technology), et al. Clinical course of breast cancer patients with complete patho- and V5-tag (Invitrogen R960-25). logic primary tumor and axillary lymph node response to doxorubicin-based neoadjuvant chemotherapy. J Clin Oncol 1999 ; 17 : 460 – 9 . Statistical Analysis 3. Liedtke C , Mazouni C , Hess KR , Andre F , Tordai A , Mejia JA , et al. Response to neoadjuvant therapy and long-term survival in patients Statistics were performed where indicated using R (49 ) or Graph- with triple-negative breast cancer. J Clin Oncol 2008 ; 26 : 1275 – 81 . Pad Prism (GraphPad Software). P < 0.05 was considered statistically < 4. Guarneri V , Piacentini F , Ficarra G , Frassoldati A , D’Amico R , signifi cant and P 0.1 was considered a statistical trend. Giovannelli S , et al. A prognostic model based on nodal status and Ki-67 predicts the risk of recurrence and death in breast cancer Disclosure of Potential Confl icts of Interest patients with residual disease after preoperative chemotherapy . Ann J.S. Ross is employed as Medical Director at Foundation Medicine, Oncol 2009 ; 20 : 1193 – 8 . Inc., has received a commercial research grant from Foundation 5. Jones RL , Salter J , A’Hern R , Nerurkar A , Parton M , Reis-Filho J S, Medicine, Inc., and has ownership interest (including patents) in the et al. The prognostic signifi cance of Ki67 before and after neo- same. G.A. Palmer, R. Yelensky, and M. Cronin have ownership inter- adjuvant chemotherapy in breast cancer . Breast Cancer Res Treat est (including patents) in Foundation Medicine, Inc. No potential 2009 ; 116 : 53 – 68 . confl icts of interest were disclosed by the other authors. 6. Masuda H , Masuda N , Kodama Y , Ogawa M , Karita M , Yamamura J , et al. Predictive factors for the effectiveness of neoadjuvant chem- Authors’ Contributions otherapy and prognosis in triple-negative breast cancer patients. Cancer Chemother Pharmacol 2011 ; 67 : 911 – 7 . Conception and design: J.M. Balko, M.E. Sanders, J.S. Ross, 7. Parker JS , Mullins M , Cheang MC , Leung S , Voduc D , Vickery T, et al. M. Cronin, P.J. Stephens, C.L. Arteaga Supervised risk predictor of breast cancer based on intrinsic subtypes. Development of methodology: J.M. Balko, J.M. Giltnane, C.D. Young, J Clin Oncol 2009 ; 27 : 1160 – 7 . R.S. Cook, M.E. Sanders, V. Sánchez, D. Horiuchi, J.S. Ross, R. Yelensky, 8. Perou CM , Sorlie T , Eisen MB , van de Rijn M , Jeffrey SS , Rees M. Cronin CA , et al. Molecular portraits of human breast tumours. Nature Acquisition of data (provided animals, acquired and managed 2000 ; 406 : 747 – 52 . patients, provided facilities, etc.): J.M. Balko, C.D. Young, P. Owens, 9. von Minckwitz G , Schmitt WD , Loibl S , Muller BM , Blohmer JU , Sinn M.G. Kuba, R. Kurupi, P.D. Moore, J.A. Pinto, F.D. Doimi, H. Gómez, BV , et al. Ki67 measured after neoadjuvant chemotherapy for primary B.D. Lehmann, J.A. Bauer, J.A. Pietenpol, J.S. Ross, V.A. Miller, C.L. Arteaga breast cancer. Clin Cancer Res 2013 ; 19 : 4521 – 31 .

244 | CANCER DISCOVERYFEBRUARY 2014 www.aacrjournals.org

Sequencing of Triple-Negative Breast Tumors after Chemotherapy RESEARCH ARTICLE

10. Balko JM , Cook RS , Vaught DB , Kuba MG , Miller TW , Bhola NE , has similarities to claudin-low human breast cancer . Oncogene et al. Profi ling of residual breast cancers after neoadjuvant chemo- 2013 ; 32 : 1296 – 304 . therapy identifi es DUSP4 defi ciency as a mechanism of drug resist- 31. Liu M , Casimiro MC , Wang C , Shirley LA , Jiao X , Katiyar S , et al. ance. Nat Med 2012 ; 18 : 1052 – 9 . p21CIP1 attenuates Ras- and c-Myc-dependent breast tumor epithe- 11. Lehmann BD , Bauer JA , Chen X , Sanders ME , Chakravarthy AB , lial mesenchymal transition and cancer stem cell-like gene expression Shyr Y , et al. Identifi cation of human triple-negative breast cancer in vivo. Proc Natl Acad Sci U S A 2009 ; 106 : 19035 – 9 . subtypes and preclinical models for selection of targeted therapies . J 32. Marampon F , Ciccarelli C , Zani BM . Down-regulation of c-Myc Clin Invest 2011 ; 121 : 2750 – 67 . following MEK/ERK inhibition halts the expression of malignant 12. Rody A , Karn T , Liedtke C , Pusztai L , Ruckhaeberle E , Hanker L , et al. phenotype in rhabdomyosarcoma and in non muscle-derived human A clinically relevant gene signature in triple negative and basal-like tumors. Mol Cancer 2006 ; 5 : 31 . breast cancer. Breast Cancer Res 2011 ; 13 : R97 . 33. Podsypanina K , Politi K , Beverly LJ , Varmus HE . Oncogene coopera- 13. Cerami E , Gao J , Dogrusoz U , Gross BE , Sumer SO , Aksoy BA, e t a l . tion in tumor maintenance and tumor recurrence in mouse mam- The cBio cancer genomics portal: an open platform for exploring mary tumors induced by Myc and mutant Kras. Proc Natl Acad Sci multidimensional cancer genomics data . Cancer Discov 2012 ; 2 : 401 – 4 . U S A 2008 ; 105 : 5242 – 7 . 14. Shah SP , Roth A , Goya R , Oloumi A , Ha G , Zhao Y , et al. The clonal 34. Serra JM , Gutierrez A , Alemany R , Navarro M , Ros T , Saus and mutational evolution spectrum of primary triple-negative breast C, et al. Inhibition of c-Myc down-regulation by sustained extracel- cancers. Nature 2012 ; 486 : 395 – 9 . lular signal-regulated kinase activation prevents the antimetabolite 15. Pratilas CA , Taylor BS , Ye Q , Viale A , Sander C , Solit DB , et al. methotrexate- and gemcitabine-induced differentiation in non– (V600E)BRAF is associated with disabled feedback inhibition of RAF- small-cell lung cancer cells . Mol Pharmacol 2008 ; 73 : 1679 – 87 . MEK signaling and elevated transcriptional output of the pathway. 35. Horiuchi D , Kusdra L , Huskey NE , Chandriani S , Lenburg ME , Proc Natl Acad Sci U S A 2009 ; 106 : 4519 – 24 . Gonzalez-Angulo AM , et al. MYC pathway activation in triple-nega- 16. Padua D , Zhang XH , Wang Q , Nadal C , Gerald WL , Gomis RR , et al. tive breast cancer is synthetic lethal with CDK inhibition . J Exp Med TGFbeta primes breast tumors for lung metastasis seeding through 2012 ; 209 : 679 – 96 . angiopoietin-like 4. Cell 2008 ; 133 : 66 – 77 . 36. Iyer G , Hanrahan AJ , Milowsky MI , Al-Ahmadie H , Scott SN , Jan a- 17. Bhola NE , Balko JM , Dugger TC , Kuba MG , Sanchez V , Sanders M , kiraman M , et al. Genome sequencing identifi es a basis for everolimus et al. TGF-beta inhibition enhances chemotherapy action against sensitivity. Science 2012 ; 338 : 221 . triple-negative breast cancer. J Clin Invest 2013 ; 123 : 1348 – 58 . 37. The Cancer Genome Atlas Network . Comprehensive molecular por- 18. Prat A , Adamo B , Cheang MC , Anders CK , Carey LA , Perou CM. traits of human breast tumours. Nature 2012 ; 490 : 61 – 70 . Molecular characterization of basal-like and non-basal-like triple- 38. Hammond ME , Hayes DF , Dowsett M , Allred DC , Hagerty KL , Badve negative breast cancer. Oncologist 2013 ; 18 : 123 – 33 . S , et al. American Society of Clinical Oncology/College of American 19. Chapman-Rothe N , Curry E , Zeller C , Liber D , Stronach E , Gabra Pathologists guideline recommendations for immunohistochemical H , et al. Chromatin H3K27me3/H3K4me3 histone marks defi ne testing of estrogen and progesterone receptors in breast cancer. Arch gene sets in high-grade serous ovarian cancer that distinguish malig- Pathol Lab Med 2010 ; 134 : 907 – 22 . nant, tumour-sustaining and chemo-resistant ovarian tumour cells . 39. Wolff AC , Hammond ME , Schwartz JN , Hagerty KL , Allred DC , Cote Oncogene 2013 ; 32 : 4586 – 92 . RJ , et al. American Society of Clinical Oncology/College of Ameri- 20. Balko JM , Schwarz LJ , Bhola NE , Kurupi R , Owens P , Miller TW , et al. can Pathologists guideline recommendations for human epidermal Activation of MAPK pathways due to DUSP4 loss promotes cancer growth factor receptor 2 testing in breast cancer. Arch Pathol Lab stem cell-like phenotypes in basal-like breast cancer. Cancer Res Med 2007 ; 131 : 18 – 43 . 2013 ; 73 : 6346 – 58 . 40. Owens P , Pickup MW , Novitskiy SV , Chytil A , Gorska AE , Aakre M E, 21. Ertel F , Nguyen M , Roulston A , Shore GC . Programming cancer cells et al. Disruption of bone morphogenetic protein receptor 2 (BMPR2) for high expression levels of Mcl1. EMBO Rep 2013 ; 14 : 328 – 36 . in mammary tumors promotes metastases through cell autono- 22. Inuzuka H , Shaik S , Onoyama I , Gao D , Tseng A , Maser RS, et al. mous and paracrine mediators. Proc Natl Acad Sci U S A 2012 ; 109 : SCF(FBW7) regulates cellular apoptosis by targeting MCL1 for ubiq- 2814 – 9 . uitylation and destruction. Nature 2011 ; 471 : 104 – 9 . 41. Li H , Durbin R . Fast and accurate short read alignment with Burrows– 23. Allen TD , Zhu CQ , Jones KD , Yanagawa N , Tsao MS , Bishop JM . Wheeler transform. Bioinformatics 2009 ; 25 : 1754 – 60 . Interaction between MYC and MCL1 in the genesis and outcome of 42. Li H , Handsaker B , Wysoker A , Fennell T , Ruan J , Homer N , et al. non–small-cell lung cancer. Cancer Res 2011 ; 71 : 2212 – 21 . The Sequence Alignment/Map format and SAMtools. Bioinformatics 24. Beverly LJ , Varmus HE . MYC-induced myeloid leukemogenesis is 2009 ; 25 : 2078 – 9 . accelerated by all six members of the antiapoptotic BCL family. 43. DePristo MA , Banks E , Poplin R , Garimella KV , Maguire JR , Hartl C , Oncogene 2009 ; 28 : 1274 – 9 . et al. A framework for variation discovery and genotyping using next- 25. Xiang Z , Luo H , Payton JE , Cain J , Ley TJ , Opferman JT , et al. Mcl1 generation DNA sequencing data. Nat Genet 2011 ; 43 : 491 – 8 . haploinsuffi ciency protects mice from Myc-induced acute myeloid 44. Forbes SA , Bindal N , Bamford S , Cole C , Kok CY , Beare D, et al. leukemia. J Clin Invest 2010 ; 120 : 2109 – 18 . COSMIC: mining complete cancer genomes in the Catalogue of 26. Duncan JS , Whittle MC , Nakamura K , Abell AN , Midland AA , Somatic Mutations in Cancer. Nucleic Acids Res 2011 ; 39 : D945 – 50 . Zawistowski JS , et al. Dynamic reprogramming of the kinome in 45. Compeau PE , Pevzner PA , Tesler G . How to apply de Bruijn graphs to response to targeted MEK inhibition in triple-negative breast cancer. genome assembly. Nat Biotechnol 2011 ; 29 : 987 – 91 . Cell 2012 ; 149 : 307 – 21 . 46. Carter SL , Cibulskis K , Helman E , McKenna A , Shen H , Zack T, et al. 27. Sears R , Nuckolls F , Haura E , Taya Y , Tamai K , Nevins JR . Multiple Absolute quantifi cation of somatic DNA alterations in human can- Ras-dependent phosphorylation pathways regulate Myc protein sta- cer. Nat Biotechnol 2012 ; 30 : 413 – 21 . bility. Genes Dev 2000 ; 14 : 2501 – 14 . 47. Van Loo P , Nordgard SH , Lingjaerde OC , Russnes HG , Rye IH , Sun W , 28. Elenbaas B , Spirio L , Koerner F , Fleming MD , Zimonjic DB , Donaher et al. Allele-specifi c copy number analysis of tumors. Proc Natl Acad JL , et al. Human breast cancer cells generated by oncogenic transforma- Sci U S A 2010 ; 107 : 16910 – 5 . tion of primary mammary epithelial cells . Genes Dev 2001 ; 15 : 50 – 65 . 48. Yau C , Mouradov D , Jorissen RN , Colella S , Mirza G , Steers G , et al. A 29. Han G , Wang Y , Bi W . C-Myc overexpression promotes osteosar- statistical approach for detecting genomic aberrations in heterogene- coma cell invasion via activation of MEK-ERK pathway. Oncol Res ous tumor samples from single nucleotide polymorphism genotyping 2012 ; 20 : 149 – 56 . data. Genome Biol 2010 ; 11 : R92 . 30. Hollern DP , Yuwanita I , Andrechek ER . A mouse model with T58A 49. Team RC . R: a language and environment for statistical computing . mutations in Myc reduces the dependence on KRas mutations and Vienna, Austria : R Foundation for Statistical Computing ; 2012 .

FEBRUARY 2014CANCER DISCOVERY | 245

Molecular Profiling of the Residual Disease of Triple-Negative Breast Cancers after Neoadjuvant Chemotherapy Identifies Actionable Therapeutic Targets

Justin M. Balko, Jennifer M. Giltnane, Kai Wang, et al.

Cancer Discovery 2014;4:232-245. Published OnlineFirst December 19, 2013.

Updated version Access the most recent version of this article at: doi:10.1158/2159-8290.CD-13-0286

Supplementary Access the most recent supplemental material at: Material http://cancerdiscovery.aacrjournals.org/content/suppl/2013/12/18/2159-8290.CD-13-0286.DC1 http://cancerdiscovery.aacrjournals.org/content/suppl/2021/03/02/2159-8290.CD-13-0286.DC2

Cited articles This article cites 48 articles, 20 of which you can access for free at: http://cancerdiscovery.aacrjournals.org/content/4/2/232.full#ref-list-1

Citing articles This article has been cited by 49 HighWire-hosted articles. Access the articles at: http://cancerdiscovery.aacrjournals.org/content/4/2/232.full#related-urls

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Department at Subscriptions [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://cancerdiscovery.aacrjournals.org/content/4/2/232. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.