Wang, Liewei

Abstract Genetic risks in breast cancer remain only partly understood. Here, we report the results of a genome-wide association study of germline DNA from 4,658 women, including 252 women experiencing a breast cancer recurrence, who were entered on the MA.27 adjuvant trial comparing the aromatase inhibitors (AI) anastrozole and exemestane. Single-nucleotide polymorphisms (SNP) of top significance were identified in the gene encoding MIR2052HG, a long noncoding RNA of unknown function. Heterozygous or homozygous individuals for variant alleles exhibited a ∼40% or ∼63% decrease, respectively, in the hazard of breast cancer recurrence relative to homozygous wild-type individuals. Functional genomic studies in lymphoblastoid cell lines and ERα-positive breast cancer cell lines showed that expression from MIR2052HG and the ESR1 gene encoding estrogen receptor-α (ERα) was induced by estrogen and AI in a SNP-dependent manner. Variant SNP genotypes exhibited increased ERα binding to estrogen response elements, relative to wild-type genotypes, a pattern that was reversed by AI treatment. Further, variant SNPs were associated with lower expression of MIR2052HG and ERα. RNAi-mediated silencing of MIR2052HG in breast cancer cell lines decreased ERα expression, cell proliferation, and anchorage-independent colony formation. Mechanistic investigations revealed that MIR2052HG sustained ERα levels both by promoting AKT/FOXO3-mediated ESR1 transcription and by limiting ubiquitin-mediated, proteasome-dependent degradation of ERα. Taken together, our results define MIR2052HS as a functionally polymorphic gene that affects risks of breast cancer recurrence in women treated with AI. More broadly, our results offer a pharmacogenomic basis to understand differences in the response of breast cancer patients to AI therapy. Cancer Res; 76(23); 7012–23. ©2016 AACR.

Abstract Background: Rituximab, cyclophosphamide, doxorubicin, vincristine, and prednisone (R-CHOP) therapy is standard of care for patients with newly diagnosed DLBCL. A majority of patients respond to R-CHOP, however ∼9% will have primary refractory disease and outcomes in these patients are extremely poor. While there is some knowledge of the pharmacogenomic (PGx) determinants of individual agents within R-CHOP, little is known about the PGx determinants of combination therapy. Our study aims to elucidate germline PGx determinants of drug bioavailability and undesirable side chain reactions in DLBCL patients treated with R-CHOP having potential to influence refractory disease. Methods: Newly diagnosed, DLBCL patients treated with R-CHOP between 2002 and 2012 with germline DNA and genotyping data were selected from the Molecular Epidemiology Resource within the Iowa/Mayo Lymphoma SPORE. Refractory disease was defined as either i) a best response of stable or progressive disease or ii) progressive disease by the end of the planned number of cycles of R-CHOP. PGx candidates within known pharmacokinetics and pharmacodynamics (PD) pathways for 11 genes (1,040 single nucleotide polymorphisms (SNPs)) were selected for study (CPT 2012;92(4):414-7). Odds ratios (ORs) and 95% confidence intervals (CI) were used to estimate the association of SNPs with refractory disease, controlling for age, gender, and age-adjusted International Prognostic Index. Results: The median age at diagnosis of 424 of DLBCL patients treated with R-CHOP was 62 years (range 18-92); 45.7% were female; and 58.4% had Ann Arbor Stage III-IV. By the end of initial therapy 40 patients had primary refractory disease. Of the 1,040 SNPs evaluated, 120 (12%) were significantly associated with refractory disease at the 5% significance threshold suggesting novel associations with drug metabolism. The top SNP from each PGx candidate gene are shown in Table, with ALDH5A1 and CYP2B6 showing the most significant associations. Conclusions: While needing replication, we identified associations within two PGx candidate genes that are potentially functionally related to cyclophosphamide PD: alleles within CYP2B6 may have relevance in R-CHOP outcomes due to potential competition with other drug metabolizing enzymes shared with prednisone and vincristine biotransformation in the liver, and alleles withinALDH5A1 may impact a deactivating side chain reaction in B lymphocytes. Table: R-CHOP PGx candidates with known side chain reactionsPGx candidaten SNPsTop SNPChrOR* (95% CI)P-valueMAFABCB1365rs3480093571.86 (1.14 – 3.04)1.32E-020.29ADH1A32rs122996841.66 (0.60 - 4.53)3.27E-010.038ALDH3A116rs2228100171.26 (0.71 - 2.22)4.28E-010.24ALDH5A1192rs946103466.30 (2.43 - 16.4)1.55E-040.030CBR110rs2835266212.10 (0.90 - 4.89)8.76E-020.052CBR356rs45527331212.12 (0.91 - 4.98)8.29E-020.050CYP2B6107rs1038376192.38 (1.44 - 3.94)7.74E-040.27CYP3A421rs2898858372.32 (0.80 - 6.78)1.23E-010.028CYP3A531rs778032871.69 (0.68 - 4.16)2.57E-010.054SLC22A16193rs11217403561.80 (1.06 - 3.03)2.78E-020.21SOD116rs1041740211.35 (0.83 - 2.19)2.26E-010.32Chr=chromosome, MAF=minor allele frequency, *Adjusted for age, sex, and age-adjusted International Prognostic Index Citation Format: Matthew K. Breitenstein, Susan L. Slager, Anne J. Novak, Matthew J. Maurer, Carrie A. Thompson, Umar Farooq, Brian K. Link, Thomas M. Habermann, Curtis L. Olswold, Andrew L. Feldman, Grzegorz S. Nowakowski, Stephen M. Ansell, Thomas E. Witzig, Richard M. Weinshilboum, Liewei Wang, James R. Cerhan. Pharmacogenomic of refractory response to R-CHOP therapy in diffuse large B-cell lymphoma (DLBCL) patients. [abstract]. In: Proceedings of the AACR-NCI-EORTC International Conference: Molecular Targets and Cancer Therapeutics; 2015 Nov 5-9; Boston, MA. Philadelphia (PA): AACR; Mol Cancer Ther 2015;14(12 Suppl 2):Abstract nr B143.

Abstract Motivation: Single-cell RNA-sequencing technology allows detection of gene expression at the single-cell level. One typical feature of the data is a bimodality in the cellular distribution even for highly expressed genes, primarily caused by a proportion of non-expressing cells. The standard and the over-dispersed gamma-Poisson models that are commonly used in bulk-cell RNA-sequencing are not able to capture this property. Results: We introduce a beta-Poisson mixture model that can capture the bimodality of the single-cell gene expression distribution. We further integrate the model into the generalized linear model framework in order to perform differential expression analyses. The whole analytical procedure is called BPSC. The results from several real single-cell RNA-seq datasets indicate that ∼90% of the transcripts are well characterized by the beta-Poisson model; the model-fit from BPSC is better than the fit of the standard gamma-Poisson model in > 80% of the transcripts. Moreover, in differential expression analyses of simulated and real datasets, BPSC performs well against edgeR, a conventional method widely used in bulk-cell RNA-sequencing data, and against scde and MAST, two recent methods specifically designed for single-cell RNA-seq data. Availability and Implementation: An R package BPSC for model fitting and differential expression analyses of single-cell RNA-seq data is available under GPL-3 license at https://github.com/nghiavtr/BPSC . Contact: yudi.pawitan@ki.se or mattias.rantalainen@ki.se Supplementary information: Supplementary data are available at Bioinformatics online.

Abstract We previously reported, on the basis of a genome-wide association study for aromatase inhibitor-induced musculoskeletal symptoms, that single-nucleotide polymorphisms (SNPs) near the T-cell leukemia/lymphoma 1A (TCL1A) gene were associated with aromatase inhibitor-induced musculoskeletal pain and with estradiol (E2)-induced TCL1A expression. Furthermore, variation in TCL1A expression influenced the downstream expression of proinflammatory cytokines and cytokine receptors. Specifically, the top hit genome-wide association study SNP, rs11849538, created a functional estrogen response element (ERE) that displayed estrogen receptor (ER) binding and increased E2 induction of TCL1A expression only for the variant SNP genotype. In the present study, we pursued mechanisms underlying the E2-SNP-dependent regulation of TCL1A expression and, in parallel, our subsequent observations that SNPs at a distance from EREs can regulate ERα binding and that ER antagonists can reverse phenotypes associated with those SNPs. Specifically, we performed a series of functional genomic studies using a large panel of lymphoblastoid cell lines with dense genomic data that demonstrated that TCL1A SNPs at a distance from EREs can modulate ERα binding and expression of TCL1A as well as the expression of downstream immune mediators. Furthermore, 4-hydroxytamoxifen or fulvestrant could reverse these SNP-genotype effects. Similar results were found for SNPs in the IL17A cytokine and CCR6 chemokine receptor genes. These observations greatly expand our previous results and support the existence of a novel molecular mechanism that contributes to the complex interplay between estrogens and immune systems. They also raise the possibility of the pharmacological manipulation of the expression of proinflammatory cytokines and chemokines in a SNP genotype-dependent fashion.

Abstract Elucidating mechanisms of chemoresistance is critical to improve cancer therapy, especially for the treatment of pancreatic ductal adenocarcinoma (PDAC). Genome-wide association studies have suggested the less studied gene HEAT repeat-containing protein 1 (HEATR1) as a possible determinant of cellular sensitivity to different chemotherapeutic drugs. In this study, we assessed this hypothesized link in PDAC, where HEATR1 expression is downregulated significantly. HEATR1 silencing in PDAC cells increased resistance to gemcitabine and other chemotherapeutics, where this effect was associated with increased AKT kinase phosphorylation at the Thr308 regulatory site. Mechanistically, HEATR1 enhanced cell responsiveness to gemcitabine by acting as a scaffold to facilitate interactions between AKT and the protein phosphatase PP2A, thereby promoting Thr308 dephosphorylation. Consistent with these findings, treatment with the AKT inhibitor triciribine sensitized HEATR1-depleted PDAC cells to gemcitabine, suggesting that this therapeutic combination may overcome gemcitabine resistance in patients with low HEATR1 expression. Clinically, we found that HEATR1 downregulation in PDAC patients was associated with increased AKT phosphorylation, poor response to tumor resection plus gemcitabine standard-of-care treatment, and shorter overall survival. Collectively, our findings establish HEATR1 as a novel regulator of AKT and a candidate predictive and prognostic indicator of drug responsiveness and outcome in PDAC patients. Cancer Res; 76(3); 572–81. ©2015 AACR.

Abstract Breast cancer is the most common invasive cancer in women worldwide and treatment related toxicity has high inter-patient variability. Our goal was to identify additional genetic predictors of toxicity by completing a pathway based analysis taking advantage of the genotyping results obtained through a GWAS of the same patients. For our analysis, we have utilized GWA data from the SUCCESS-A trial in which 3,754 breast cancer patients were treated with adjuvant chemotherapy. We have included SNPs associated with 112 genes involved in drug disposition and 375 genes in selected cancer pathways. We identified 35 SNPs associated with neutropenia (P-value <1×10−4). These patients initially received three cycles consisting of 5-fluorouracil (5-FU), cyclophosphamide, and epirubicin. Currently we are characterizing the function of three SNPs located within an intron of the hyaluronan mediated motility receptor (HMMR) gene that are associated with the neutropenia phenotype (P-value <1×10−4). These SNPs are highly associated with the expression of a gene encoding a B-cell survival factor, TNFSF13B, (P-value <1×10−5). Our in vitro studies using lymphoblastoid cell lines (LCLs) selected based on the genotypes of HMMR gene showed that heterozygous and homozygous variant cells (n = 6) have an increased sensitivity to 5-fluorouracil, but not cyclophosphamide or epirubicin when compared with homozygous wild-type cells (n = 6). LCLs exposed to the previously mentioned chemotherapeutic agents resulted in an upregulation of TNFSF13B gene expression and downregulation of HMMR gene expression. We have utilized a pathway based approach to analyzing GWA data to identify several SNPs in the HMMR gene highly associated with the expression of one of the top signals in the cytopenia GWAS. Our goal is to further characterize the role of HMMR gene variants in chemotherapy response by assessing whether combinations of these agents in vitro (including 5-FU) also have a genotype dependent response to cytotoxicity. Citation Format: Behzad Bidadi, Krishna R. Kalari, Matthias Rubner, Alexander Hein, Matthias W. Beckmann, Peter A. Fasching, Brigitte Rack, Wolfgang Janni, Richard Weinshilboum, Liewei Wang. Utilizing a pathway based analysis of genome wide association data to identify biomarkers of toxicity in breast cancer patients. [abstract]. In: Proceedings of the 106th Annual Meeting of the American Association for Cancer Research; 2015 Apr 18-22; Philadelphia, PA. Philadelphia (PA): AACR; Cancer Res 2015;75(15 Suppl):Abstract nr 5488. doi:10.1158/1538-7445.AM2015-5488

Abstract Background: The use of anti-estrogen (AE) therapies (selective estrogen receptor modulators [SERMs], aromatase inhibitors [AIs]) in breast cancer (BC) patients was recently associated with risk of developing rheumatoid arthritis (RA) (J Rheum 2015;42:55-9). We attempted to replicate and extend these results using electronic health record (EHR) enhanced cancer registry data that assessed specific AE therapies and also accounted for potential confounding by age, smoking status, and chemotherapy use. Methods: Using cancer registry and EHR data, we identified a cohort of BC patients who were newly diagnosed and treated at Mayo Clinic Rochester from 1998-2011 and had no previous diagnosis of RA. Smoking status at diagnosis, BC treatments and RA at diagnosis and during follow-up were identified using validated EHR-based algorithms. Hazard ratios (HRs) and 95% confidence intervals (CI) from a multivariate Cox model were used to estimate the association of AE use with risk of RA, controlling for age, smoking status at BC diagnosis, and BC chemotherapy use. Results: The analytic cohort of 9,244 newly diagnosed BC patients treated and followed at Mayo Clinic had a median age at diagnosis of 59 years (range 18-97); 32% were smokers; and 29% were treated with chemotherapy. During a median follow-up of 49 months (range 1-191), 19 patients developed RA. Compared to BC patients not receiving any AE therapy, those receiving AI monotherapy, but not SERM monotherapy or both SERMs and AIs, were at significantly increased risk of developing RA (Table). AE Therapy# of RA Cases / # patientsHR* (95% CI)P-valueNo AE therapy6/4,0711.00 (reference)SERMs only5/2,5371.20 (0.36 - 3.95)0.77AIs only5/7126.37 (1.77 - 22.92)0.005SERMs then AIs3/1,8750.90AI then SERMs0/490.91 (0.21 - 3.86)*Adjusted for age, smoking status and use of chemotherapy Conclusions: Our findings indicate that the use of AIs for BC therapy, but not SERMs or SERMs followed by AIs, is associated with increased risk for development of RA, when controlling for smoking status and use of chemotherapy. While needing replication, these findings suggest a specific role for AIs in RA development and suggest pathways that could be targeted to prevent this potential treatment complication. Citation Format: Matthew K. Breitenstein, Ming-Fen Ho, Richard M. Weinshilboum, James R. Cerhan, Jyotishman Pathak, Tim Bongartz, Liewei Wang, James N. Ingle. Association of anti-estrogen therapy in breast cancer patients and subsequent risk of rheumatoid arthritis: An electronic health record study. [abstract]. In: Proceedings of the 106th Annual Meeting of the American Association for Cancer Research; 2015 Apr 18-22; Philadelphia, PA. Philadelphia (PA): AACR; Cancer Res 2015;75(15 Suppl):Abstract nr LB-191. doi:10.1158/1538-7445.AM2015-LB-191

Abstract Introduction: Triple negative breast cancers (TNBC) are characterized as lacking estrogen receptor (ER), progesterone receptor (PR) and human epidermal growth factor receptor 2 (HER2) expression and TNBC patients have higher rates of recurrence and death compared with other breast cancer subtypes. For TNBC patients who fail standard chemotherapy, there are a lack of novel drug therapies, given the absence of well-defined molecular targets. Recently, a microarray meta-analysis identified 7 triple negative subtypes, including the validation of the luminal androgen receptor (LAR) positive subtype [Lehmann, 2011]. However, microarray technology is dependent on probe-target specificity and the 7 subtypes have yet to be validated using RNA sequencing data, and the presences of recurrent genomic alterations in the 7 subtypes are unknown. METHODS: We obtained 1106 breast cancer RNA-Seq bam files from The Cancer Genome Atlas (TCGA) and aligned with Tophat v1.3. The PAM50 intrinsic gene signature was used to extract a cohort of 128 TNBC samples. Consensus clustering of genes, greater than 75th percentile of variance, was performed using Kmeans clustering in Spearman's correlation space. A nearest centroid prediction model was developed from genes differentially expressed among the clusters. Eighty independent TNBC RNA sequencing samples were obtained (British Colombia; BC) [Shah, 2012] which were calibrated to our TNBC conditional quantile normalized cohort and sub-typed by our model. RESULTS: Using RNA-Seq gene expression count data, we identified 5 clusters, all of which were stable, including the LAR cluster. Signaling pathway impact analysis (SPIA) implicated cytokine-cytokine receptor interaction, leukocyte transendothelial migration, and regulation of actin cytoskeleton pathways commonly altered in the non-LAR TNBC subtypes. In contrast, cell cycle, ECM-receptor interaction, endocrine regulated calcium reabsorption, and insulin signaling pathways were altered in the LAR versus non-LAR subtypes. Neuroactive ligand-receptor interactions were observed to be altered commonly between all sub-types. We then applied our model to the Shah, et al cohort. In this cohort, the LAR subtype was consistent with Shah's classification of ‘other’ TNBC and contained no basal samples by PAM50 intrinsic modeling. Analysis of sub-type specific mutation data from the BC cohort demonstrates an increased mutational load in ECM-related proteins, particularly the myosins, along with increased TP53 clonality in the non-LAR subtypes. CONCLUSIONS: Using TCGA RNASeq data, we have confirmed the presence of 5 major TNBC subtypes, including the LAR; which was negligible in basal composition by PAM50 intrinsic modeling. SPIA pathway analysis indicates a core set of pathways demonstrating altered expression across the TNBC sub-types and the identification of molecular targets within each subtype is ongoing. Citation Format: Kevin J. Thompson, Xiaojia Tang, Zhifu Sun, Jason P. Sinnwell, Hugues Sicotte, Douglas W. Mahoney, Steven Hart, Peter T. Vedell, Poulami Barman, Jeanette E. Eckel Passow, Eric D. Wieben, James N. Ingle, Judy C. Boughey, Liewei Wang, Richard Weinshilboum, Krishna R. Kalari, Matthew P. Goetz. Molecular classification of triple negative breast cancer via RNA-sequencing data. [abstract]. In: Proceedings of the 105th Annual Meeting of the American Association for Cancer Research; 2014 Apr 5-9; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2014;74(19 Suppl):Abstract nr 5592. doi:10.1158/1538-7445.AM2014-5592

Abstract INTRODUCTION Next generation sequencing (NGS) of patients has significantly changed our ways to study cancer genomics as it provides precise estimates of gene expression, fusion transcripts, expressed single nucleotide variants (eSNVs), splice variants and copy number variants. The Breast Cancer Genome Guided Therapy (BEAUTY) is an ongoing clinical study in which RNA sequencing (RNAseq) and whole exome sequencing (WES) are performed prior to, during and after neoadjuvant chemotherapy. Here we report the use of these sequencing technologies to investigate gene expression levels and mutational profiles in a triple negative breast cancer (TNBC) patient enrolled in BEAUTY whose disease did not respond to neoadjuvant paclitaxel and anthracycline/cycphosphamide. METHODS Computational approaches were used to integrate WES and RNAseq data obtained before therapy (V1T), after 12 weekly paclitaxel treatments (V2T) and after anthracycline-based regimen at surgical resection (V3T) to study a single patient with persistent TNBC disease after neoadjuvant chemotherapy. RESULTS Using RNA-Seq data, we identified an inter-chromosomal fusion transcript between chromosome 20 and 22 (GNAS-TTC38) that was highly expressed at V1T, V2T and V3T. We also identified intra-chromosomal fusion transcripts that were expressed at two time points, such as fusion transcript (KANSL1-ARL17A) on chromosome 17 for V1T and V2T and fusion transcript (RBM12B-LINC00535) on chromosome 8 for V2T and V3T time points. Several gene expression changes were also observed. Gene expression analysis of V1T, V2T and V3T tumors was performed. Differential temporal gene expression profiles of 9884 genes that were significant and varying at different time points were obtained for pathway analysis. Pathway analysis of 9884 genes identified up regulation and down regulation of several transcription factors with a fold change of 2x or more. When compared to blood, DNA tumor and RNA-Seq data, we identified 81 common somatic eSNVs that were expressed in both V1T and V2T time points and we are in the process of investigating V3T data. We found alterations of key transporter domains (CD225, Coatamer_beta_C, DUF2435, Dynamitin, EI24, GLTP, LMF1, Porin_3, V-ATPase_C) in our V1T and V2T SNV data. Similar to gene expression analysis, we are in the process of obtaining the list of mutations at various time points to identify driver and passenger mutation candidate genes for this specific TNBC patient. CONCLUSIONS Our initial time-series analysis of eSNV, fusion transcripts and gene expression data demonstrate that intensive analysis for individual patients is feasible. Further investigation of drug transporters and transcription regulators may help develop personalized treatment strategies for patients with disease resistant to current regimens. Citation Format: Krishna R. Kalari, Xiaojia Tang, Kevin J. Thompson, Douglas W. Mahoney, Poulami Barman, Jason P. Sinnwell, Hugues Sicotte, Peter Vedell, Steven N. Hart, Travis J. Dockter, Katie N. Jones, Amy L. Conners, Ann M. Moyer, Daniel W. Visscher, Jia Yu, Bowen Gao, Sarah A. McLaughlin, John A. Copland, Alvaro Moreno-Aspitia, Donald W. Northfelt, Richard J. Gray, Vera J. Suman, Jeanette E. Eckel Passow, Jean-Pierre A. Kocher, Eric D. Wieben, Gianrico Farrugia, Cloann G. Schultz, James N. Ingle, Richard Weinshilboum, Matthew P. Goetz, Liewei Wang, Judy C. Boughey. Analysis of sequencing data to identify potential drug targets for an individual newly diagnosed with basal breast cancer who failed to respond to current standard neoadjuvant chemotherapy. [abstract]. In: Proceedings of the 105th Annual Meeting of the American Association for Cancer Research; 2014 Apr 5-9; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2014;74(19 Suppl):Abstract nr 4185. doi:10.1158/1538-7445.AM2014-4185

Abstract ERRFI1, a negative regulator of the EGFR family, is known to inhibit EGFR activation along with downstream of EGFR signaling. It is recognized as a tumor suppressor. However, ERRFI1 has multiple functions, and here, we present a novel function of ERRFI1. We found that it can directly interact with AKT and positively regulate AKT activity through regulation of AKT473 site by inhibiting PHLPP, the phosphatase access to AKT. The effect of this regulation on cell proliferation and on response to chemotherapy is cell context dependent, e.g. EGFR levels. We hypothesize that in cancer cells with high EGFR expression, down regulation of ERRFI1 enhances EGFR signaling, while on the other hand, increases AKT dephosphorylation. The strength of these two opposing signals will determine the eventual cellular phenotypes. In cancer cells with low EGFR expression, the direct effect of ERRFI1 on AKT is dominant, resulting in activation of AKT. To test these hypotheses, we knockdown ERRFI1 in multiple human cancer cell lines with different EGFR levels. Knockdown of ERRFI1 in high EGFR expressing cells resulted in an increase of cell proliferation and resistance to gemcitabine treatment, and this effect was due to the dominant negative effect of ERRFI1on EGFR, which was confirmed by increased EGFR signaling. Knockdown of ERRFI1 in cell lines with moderate level of EGFR also increased cell proliferation and resistance to gemcitabine, although biochemically, down regulation of ERRFI1 reduced AKT pS473 levels, but increased ERK/MAPK signaling. The addition of EGFR inhibitor, gefitinib, overcame gemcitabine resistance due to downregualtion of ERRFI1. In contrary, in cell lines with low level of EGFR, ERRFI1 positively regulated AKT signaling via directly binding to AKT and blocking PHLPP access to AKT. Therefore, down-regulation of ERRFI1 inhibited cell proliferation and sensitized cells to gemcitabine. In summary, we identified a novel function of ERRFI1 as a positive AKT regulator and its effect on cell proliferation and response to therapy is cell context dependent, e.g. EGFR levels. These results provide us insights into the dual roles of ERRFI1 in regulation of EGFR and AKT pathway and would help us to better individualize cancer therapy. Citation Format: Junmei Hou, Kaustubh N. Bhinge, Liewei Wang. The effect of ERRFI1 as a novel AKT regulator on cell proliferation and response to therapy is cell context dependent. [abstract]. In: Proceedings of the 105th Annual Meeting of the American Association for Cancer Research; 2014 Apr 5-9; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2014;74(19 Suppl):Abstract nr 2782. doi:10.1158/1538-7445.AM2014-2782