Poster Session – Molecular Targeted Agents II Friday 21 November 2014 159

Conclusion: Taken together, these findings point to therapeutic activationof PP2A as a novel strategy for the treatment of advanced KRAS-mutantNSCLC. Activation of tumor suppressor proteins such as PP2A could offerthe opportunity to identify novel synergistic strategies for the treatment of anumber of cancer types. Nevertheless, translation of a PP2A activationstrategy into clinical medicine has required pharmaceutically tractableagents for development. Our studies represent a first step into that newterritory and highlight the potential for the development of small moleculeactivators of other protein phosphatases and tumor suppressor proteins.

487 POSTEREC-70124, a multi-kinase inhibitor, blocks NF-kB and STAT3

dependent signaling in prostate cancer stem cells

G. Civenni1, P. Costales2, C. Garcıa-Inclan3, D. Albino1, N. Longoni1,L.E. Nunez2, F. Moris2, G.M. Carbone1, C.V. Catapano1. 1Institute ofOncology Research (IOR), Tumor Biology and Experimental Therapeutics,Bellinzona, Switzerland; 2Entrechem, Oviedo, Spain; 3HospitalUniversitario Central de Asturias, Dept. Otolaryngology, Oviedo, Spain

EC-70124 is a novel multi-kinase inhibitor generated by combinatorialengineering of the biosynthetic pathways of glycosylated indolocarbazoles,like staurosporine and rebeccamycin. Biochemical kinase assays showedthat EC-70124 has a unique spectrum of activity against multiple kinasesand with degree of inhibition comparable or better than other clinicallyapproved kinase inhibitors. Furthermore, in vitro testing suggested thatNF-kappa B and STAT3 could be important targets of EC-70124 throughthe inhibition of the upstream activating kinases (e.g., IKK and JAK)of these key oncogenic transcription factors. Both NF-kappa B andSTAT3 signaling pathways are over-active in diverse types of tumors,including prostate cancer, and there is considerable interest in developinginhibitors of these pathways. In this study, we analyzed the activity of EC-70124 in prostate cancer cell lines, including androgen dependent andindependent cell models. EC-70124 showed activity (IC50: 100–400 nM) incell proliferation/viability assays both in androgen receptor (AR) positive(LNCaP, VCaP and 22RV1) and negative (DU145 and PC3) cell lines.Furthermore, EC-70124 was a potent inhibitor of CSC component inthe in vitro prostate-sphere and self-renewal assays (IC50: 100–300 nM).Treatment with EC-70124 reduced STAT3 and IkB phosphorylation, STAT3and NF-kappa B reporter activity, and expression their target genes inprostate cancer cells. Moreover, STAT3 and IkB alpha phosphorylationand expression of specific target genes were greatly reduced in isolatedprostate CSCs in vitro. Importantly, EC-70124 (40mg/kg, PO, daily for2 weeks) inhibited growth of prostate tumor xenografts in immune-deficient mice. Tumor xenografts from EC-70124 derived mice showedmarked reduction of IkB alpha phophorylation and target gene expression.Importantly, this was associated with a marked recuction in the CSCcontent in EC-70124 treated tumor xenografts determined by ex vivosphere formation assay. These results indicate that EC-70124 acts asa dual inhibitor of STAT3 and NF-kappa B signaling pathways and is apotent inhibitor of tumor growth blocking maintenance and self-renewal ofthe prostate CSC component. EC-70124 may provide the basis for newtherapeutic approaches in the treatment of prostate cancer.

488 POSTERDevelopment of a publicly accessible knowledgebase to facilitate

decision support for clinical cancer genomics reporting

K. Shaw1, J. Zeng1, A.M. Bailey1, A. Johnson1, V. Holla1, E.V. Bernstam2,J. Butts3, J. Mendelsohn1, G.B. Mills1, F. Meric-Bernstam1. 1MD AndersonCancer Center, Khalifa Institute for Personalized Cancer Therapy,Houston, USA; 2The University of Texas Health Science Center, Schoolof Biomedical Informatics, Houston, USA; 3MD Anderson Cancer Center,Internet Services, Houston, USA

Background: Public efforts characterizing thousands of human tumorsalong with normal germline information have confirmed that any numberand combination of alterations may drive tumorigenesis. With severalhundred genes now on frequently recurring mutation lists and potentiallyactionable gene lists in human cancer, it is nearly impossible for anyindividual to maintain a comprehensive understanding of the alterationsin those genes that may be clinically actionable. With groups reporting thatfewer than 10% of patients with actionable alterations are receiving targetedtreatment, a resource that distills potential therapeutic opportunities andincreases entry on to trials would have wide utility.Materials and Methods: Data for each mutation, relevant therapeuticagents and corresponding clinical trials were annotated. Each variant wasannotated for the level of evidence that associated a specific alteration in apotentially actionable cancer gene with a potential therapeutic opportunitywith appropriate references. Specific mutations, copy number variants

and fusions were linked to targeted agents, clinical trials, and functionaldata. Data were distributed via a publicly accessible website, reports andproactive clinical trial alert notifications.Results: Our group has developed a growing, public resource thatlaunched in April 2014. At the time of abstract submission http://www.personalizedcancertherapy.org/ includes detailed descriptions of 13actionable genes and more than 200 individual somatic and cancer-relevant germline mutations annotated for functional significance. Due todifferent alterations in the same gene having distinct functional effects andtherefore distinct therapeutic liabilities, each variant was annotated for itsfunctional significance. This provides a level of evidence for actionability ata variant level instead of at the gene level. Clinical trial alerts and proactivereports with decision support annotations are designed to improve clinicianrecognition of molecular data in the patient record, increase patient accrualonto targeted trials and the development of novel trials at the institution.Conclusions: Proactive notification of therapeutic opportunities anddetailed functional annotations of tumor molecular data has a goal ofimproving patient assignment to genomically-driven and -informed clinicaltrials. Sharing of the underlying data with the broader community issupported and the underlying infrastructure will be discussed in hopes ofsynergizing with others and developing this as a robust public resourceavailable to patients and physicians around the world.

489 POSTERDevelopment and clinical validation of a quantitative mass

spectrometric assay for PD-L1 protein in FFPE NSCLC samples

E. An1, W. Liao1, S. Thyparambil1, J. Rodriguez2, R. Salgia3, I. Wistuba2,J. Burrows1, T. Hembrough1. 1OncoPlex Diagnostics, Rockville MD,USA; 2MD Anderson Cancer Center, Houston TX, USA; 3Universityof Chicago, Medicine, Chicago IL, USA

Background: Binding of PD-L1 expressed on tumor cells to the PD1 onT lymphocytes transduces immuno-inhibitory signals which cripples theT cell’s ability to combat the tumor. Several anti-PD-L1 and anti-PD1 agentsare in clinical trials and both regimens have reported promising preliminaryresults in NSCLC patients suggesting that tumor expression of PD-L1 isassociated with a response to either anti-PD-L1 and anti-PD1 treatment.IHC is the current method to assess PD-L1 expression in FFPE tissue;however, PD-L1 IHC has yielded mixed results; some studies showed highfalse positive by IHC while another study showed that 13% of the PD-L1negative patients responded to treatment. Moreover, IHC is low throughputand assessing multiple druggable targets by IHC is tissue consuming. Assuch, there is an urgent need to develop quantitative and multiplexed teststo assess biomarker expression. We have developed and clinically validateda quantitative mass spectrometric assay to measure PD-L1 expression inFFPE tissue.Method: We used trypsin digestion mapping of rPD-L1 to identify optimalquantitative peptides. Stable isotope-labeled peptides were synthesized asinternal standards, and standard curves were generated in pyrococcuscomplex matrix to determine LOD, LLOQ, accuracy, precision and linearityof the assay. The PD-L1 assay was pre-clinically validated on 14 cell lineswith known expression levels of PD-L1. The assay was run on archivedFFPE sections from 9 normal lung tissues, 21 early staged and 4 advancedstaged NSCLC patients. The multiplex assays also successfully sub-classify NSCLC samples to adenocarcinoma or squamous cell carcinoma.Results: PD-L1 SRM assay had a LOD of 75 amol and LOQ of 100amol. PD-L1 protein expression was detected in 7 cell lines (7/14) rangingfrom 76–375 amol/mg tumor protein. The regression analysis betweenSRM and mRNA analysis demonstrated excellent correlation (R2 = 0.8894).For the clinical analysis of NSCLC tissue, while no normal lung tissueexpresses detectable level of PD-L1, ~24% of early stage NSCLC (5/21)and 50% of advanced stage NSCLC (2/4) express measurable PD-L1protein. Interestingly, in this initial cohort, all of the PD-L1 positive earlystaged NSCLC were squamous cell carcinoma while in a small set ofadvanced staged NSCLC, PD-L1 expression was seen in both squamouscell carcinoma (1/3) and adenocarcinoma (1/1). Characterization of largercohorts of NSCLC tissue is currently underway.Conclusions: We develop and clinical validate a quantitative proteomicPD-L1 assay to accurately measure PD-L1 expression in FFPE tissue.Additional quantitative assays for both lymphocyte (CD3, CD8, CD68) andimmunotargets (PD1, B7-H3) are under development. This proteomic assayis a critical component of our multiplexed biomarker analysis, and willallow more accurate identification of potential candidates for PD-L1 or PD1targeted therapies.