Erin Trachet

Antitumor activity and pharmacokinetic properties of PF-00299804, a second-generation irreversible pan-erbB receptor tyrosine kinase inhibitor

Signaling through the erbB receptor family of tyrosine kinases contributes to the proliferation, differentiation, migration, and survival of a variety of cell types. Abnormalities in members of this receptor family have been shown to play a role in oncogenesis, thus making them attractive targets for anticancer treatments. PF-00299804 is a second-generation irreversible pan-erbB receptor tyrosine kinase inhibitor currently in phase I clinical trials. PF-00299804 is believed to irreversibly inhibit erbB tyrosine kinase activity through binding at the ATP site and covalent modification of nucleophilic cysteine residues in the catalytic domains of erbB family members. Oral administration of PF-00299804 causes significant antitumor activity, including marked tumor regressions in a variety of human tumor xenograft models that express and/or overexpress erbB family members or contain the double mutation (L858R/T790M) in erbB1 (EGFR) associated with resistance to gefitinib and erlotinib. Furthermore, PF-00299804 shows exceptional distribution to human tumor xenografts and excellent pharmacokinetic properties across species. [Mol Cancer Ther 2008;7(7):1880–9]

Preclinical Models of Tumor Growth and Response

an important role in cancer drug discovery for more than 60 years. The same models have proven critical as tools for the elucidation of the molecular basis of neoplastic transformation, the processes involved in tumor progression and metastasis, and the determinants of therapeutic success or failure. More recently, transgenic models in particular have been used to “validate” and prioritize new strategies for therapeutic intervention. In vivo cancer models can be considered to fall within two broad classes, transplantable models, and in situ models, each with a number of subtypes (Fig. 1). For pragmatic reasons, transplantable models as a group are the most commonly used for drug evaluation, while in situ models such as cancer-prone transgenic mice provide a rich source of information on cancer etiology. It should be noted that each transplantable model represents the tumor of a single patient, not a tumor type. This discussion is centered on the application of both model types, and the potential impact of imaging technologies for cancer drug discovery. However, with recent advances in preclinical imaging technologies, these models are also proving useful in the development and testing of new imaging techniques and contrast agents. Increasingly, with the expanding role of drugs tied to specific molecular targets, these models are also being used to optimize and validate clinical imaging strategies. Finally, molecular imaging techniques are finding a critical role preclinically in the simultaneous confirmation of mechanism of action and assessment of efficacy. This is particularly true in orthotopic or transgenic model systems. Open image in new window Fig. 1. Schematic representation of the broad categories of preclinical cancer models in use today. In situ tumor models can be subcategorized by the method for induction of the tumor. Transplantable tumor models are commonly subcategorized according to whether the tumor is implanted in the organ in which the cell line originated (orthotopic versus ectopic) and in the species in which it originated (syngeneic versus xenogeneic).

Abstract C232: ALB‐109564, a novel tubulin inhibitor with improved efficacy over vinorelbine, is better tolerated when dosed iv versus ip, leading to improved activity in human tumor xenograft studies

Purpose: ALB‐109564 is a novel semi‐synthetic Vinca alkaloid which is currently in a Phase I clinical trial. The antitumor activity of ALB‐109564 was compared to vinorelbine in human tumor xenograft studies when dosed ip and iv. Methods: The maximum tolerated dose (MTD) of ALB‐109564 and vinorelbine was determined in athymic nude mice using both ip and iv dosing on q4dx4 and q7dx4 schedules. Athymic nude mice (n = 10) were implanted sc with a number of human tumor cell lines. Following establishment of measurable tumors, mice received equivalent therapeutic doses, based on their MTDs, of vinorelbine and ALB‐109564. Vinorelbine and ALB‐109564 were compared by dosing ip (q4dx4) in a panel of five xenografts (H460 NSCLC, Colo205 colon cancer, PC3 prostate cancer, H69 SCLC, and MX‐1 breast cancer). Additionally, a comparison of ALB‐109564 and vinorelbine was run dosing iv on a q7dx4 schedule in two xenograft models (PC3 and H460). Results: It was discovered that ALB‐109564 and vinorelbine were better tolerated when dosed iv rather than ip, allowing higher dose levels of both agents. For example, the MTD for ALB‐109564 using the q4dx4 schedule was 6 mpk ip compared with 14 mpk iv. ALB‐109564 demonstrated antitumor activity superior to vinorelbine when dosed at their respective MTDs ip. Statistically significant tumor growth delays (TGDs) were observed with ALB‐109564 in four of the five xenografts studied while vinorelbine did not significantly delay tumor growth in any of the models when dosing ip. A comparison of the two agents dosed iv in the PC3 and H460 xenograft models showed that both agents induced significant TGDs in both models with ALB‐109564 showing superior efficacy in the H460 study (35.7 days TGD vs. 15.7 days at respective MTDs). Conclusions: Preclinical evaluation of tubulin inhibitor ALB‐109564 (12′‐methylthiovinblastine dihydrochloride) demonstrated in vivo oncolytic activity against several human tumor cell types in xenograft models greater than that seen with vinorelbine. These results demonstrate that minor structural modifications of vinca alkaloids can have a large impact on activity. Based on these results a Phase I clinical study with ALB‐109564 was initiated in August 2008 and is ongoing through dose‐escalation. Citation Information: Mol Cancer Ther 2009;8(12 Suppl):C232.

Abstract 2066: Comparison of ADC MRI, T2-weighted MRI and combined T2-weighted/T1-contrast-enhanced/ADC MRI quantification of cerebral edema in an intracranial glioma model

Proceedings: AACR Annual Meeting 2014; April 5-9, 2014; San Diego, CA Introduction: Brain edema is a prominent feature of brain cancer and contributes to neurologic dysfunction and impaired quality of life. Efficacious treatments with minimal side effects are sought after in this indication. Quantitative, non-invasive methods for detecting and measuring brain tumor associated edema are needed to facilitate therapeutic discovery efforts. In this work, an intracranial U251-luc human glioma model was characterized for edema incidence and progression. Multiple MRI scan protocols were used to detect tumor, edema and fluid and quantify the volume of each. Apparent diffusion coefficient (ADC) based quantification was compared with a combined T2-weighted/T1-contrast-enhanced method. Methods: Female nude mice were implanted intracranially with 1x106 U251-luc (Luc-mCherry) glioma cells. A multi-modal MRI scan approach consisting of T2-weighted (T2w), T1-weighted (T1w) contrast-enhanced and ADC scans were used to detect and delineate tumor, fluid, edema and normal brain tissue. These scans yielded: V(T2): volume defined by hyper intense T2 - assumed to be volume[tumor + edema + fluid] V(CE): contrast enhancing volume - assumed to be volume[tumor] V(ADC_high): volume with ADC > threshold value that delineated fluid - assumed to be volume[fluid] A combinatorial approach to yield the volume of edema, V(edema) was used with: V(edema) = V(T2) - V(CE) - V(ADC_high) This approach was then used to examine the ADC and T2 distributions within each of these volumes to determine the sensitivity of T2 and ADC alone for distinguishing tumor, fluid and edema and quantifying edema volume. Results: The edema incidence was 100% based on T2w and T1w CE scans in orthotopically implanted U251-luc (Luc-mCherry) gliomas. Edema volume could be quantified using the multi-modal MRI scan protocol. Edema progression occurred with increasing tumor volume over time. Regions of edema exhibited greater T2 and ADC values compared to tumor tissue regions (that also had a greater T2 signal than normal tissue). This is consistent with edematous tissue water content, compared with tumor tissue and normal brain. The use of T2 and ADC alone provided good delineation of fluid from other tissues, and reasonable delineation of edema from tumor (based on the initial compartmentalization analysis). However, the combined T2w/ T1w-CE/ADC protocol was considered more precise. Conclusion: These results support the use of intracranial U251-luc (Luc-mCherry) as a reliable model for studying the effects of therapies targeting tumor-related cerebral edema. The application of multi-parametric MRI was an effective method for quantifying cerebral edema longitudinally in vivo. Citation Format: Deanne Lister, Deepa Balagurunathan, Meridith Baugher, Athena Flecha, Erin Trachet, Scott Wise, W.R. Leopold, Patrick McConville. Comparison of ADC MRI, T2-weighted MRI and combined T2-weighted/T1-contrast-enhanced/ADC MRI quantification of cerebral edema in an intracranial glioma model. [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 2066. doi:10.1158/1538-7445.AM2014-2066

Abstract 4945: Evaluating the feasibility and throughput of quintuple modality imaging in a prostate cancer bone metastasis model with PET, SPECT, CT, MRI, and bioluminescence imaging

Proceedings: AACR Annual Meeting 2014; April 5-9, 2014; San Diego, CA Background: Multimodality imaging presents a unique opportunity to image the efficacy of candidate compounds against various models of cancer. The challenge is in integrating the complimentary data that these modalities can provide. The purpose of this study was to evaluate the feasibility of quintuple modality image in order to optimize workflows for maximal efficacy testing of candidate compounds using PET, SPECT, CT, MRI, and Bioluminescence imaging (BLI). Materials and Methods: Male nude mice were injected into the left ventricle with 3x106 PC-3M-luc-C6 cells. BLI was performed to track disease progression and ensure that animals exhibited established metastatic disease prior to initiation of multimodality imaging. Animals were injected with 500µCi Tc99m-MDP and allowed a 1h uptake period prior to imaging. The animals were then placed into an Animal Handling System (AHS, ASI Instruments) that coupled to multiple integrated Positioning Receiver Assemblies (PRAs, ASI Instruments) designed for the PET, SPECT, CT, and MRI scanners, and a 15min, multi-pinhole SPECT scan was acquired. After completion of the SPECT scan, 200µCi of 18F-FDG was administered under anesthesia. MicroCT and MRI anatomical images were acquired during the FDG uptake period. Body temperature was maintained via the integrated warm water circulation in the AHS to ensure adequate FDG uptake and clearance during the procedures. Whole body CT images were acquired at 512 projections, 75kVp, and 220µA. A series of T2-weighted and T1-weighted gadolinium-enhanced images were acquired using a 7T MRI system. At 1h post-FDG administration, a 10min static PET emission scan was acquired and reconstructed using a 3DOSEM/MAP algorithm. Results: Animals were maintained under anesthesia for a total of approximately 2 hours in order to acquire data from all the modalities. Data was able to be co-registered in various forms in order to verify the location and extent of disease, as well as the usefulness of 18F-FDG PET and Tc99m-MDP to serve as biomarker probes for tumor cell metabolism, and osteolytic and osteoblastic activity in bone lesions. Conclusions: Animals tolerated the length of anesthesia well. The feasibility of this imaging was only made possible through the use of the multimodality AHS bed and docking system that permitted reproducible imaging across the different platforms. In order to further optimize workflows, multiple AHS bed systems should be employed so that data can be acquired on the various modalities in parallel, instead of sequentially, thereby greatly increasing animal throughputs. Citation Format: John L. Chunta, Deanne Lister, Chris Bull, Deepa Balagurunathan, Erin Trachet, Chris Chiodo, Scott Wise, Dick Leopold, Patrick McConville. Evaluating the feasibility and throughput of quintuple modality imaging in a prostate cancer bone metastasis model with PET, SPECT, CT, MRI, and bioluminescence imaging. [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 4945. doi:10.1158/1538-7445.AM2014-4945

Abstract 3929: Orthotopic human choroidal melanoma model characterization with bioluminescence and magnetic resonance imaging for therapeutic efficacy evaluation

Proceedings: AACR Annual Meeting 2014; April 5-9, 2014; San Diego, CA BACKGROUND: Intraocular melanomas in the West represents 70% of all primary eye cancers with a mortality rate greater than 50%. Current treatment options include enucleation, plaque radiotherapy, and proton beam radiotherapy. While enucleation is a highly invasive surgical procedure that removes the eye, the side effects of the two different radiotherapies include cataracts, retinopathy, cystoid macular edema, and secondary retinal detachment. Due to the relatively protected nature of the eye within the orbit, most topical treatment applications are precluded and systemic treatments increases the chance and incidence of off-target side effects. Up and coming targeted therapies that may not be constrained by such limitations require a non-invasive, imaging-relevant model in order to evaluate their efficacy. Therefore, the aim of this study was to characterize the imaging of orthotopic luc-enabled OCM-1 human choroidal melanoma tumors using bioluminescence imaging (BLI) and magnetic resonance imaging (MRI) for future use in anti-cancer drug efficacy testing. MATERIALS & METHODS: Human choroidal melanoma cells were modified to express luciferase. 8-12 week old female NIH-Foxn1rnu rats were implanted in the suprachoroidal space with OCM-1-luc tumor spheroids on Day 0. BLI and MRI imaging was performed every 5-7 days starting at Day 7 after implantation of the OCM-1-luc spheroids until animals reached a moribund state to monitor disease progression. For BLI, animals were injected with 150mg/kg luciferin, anesthetized with isoflurane in air and imaged at 10 minutes following luciferin administration. Anatomical MRI was performed using a gradient-echo pulse sequence on a 7T MRI system. After euthanasia, eyes were removed and preserved in formalin for histological analysis. RESULTS & CONCLUSIONS: Animals tolerated the ocular tumor implant well. As a result of these efforts, we have successfully characterized the growth of orthotopic luc-enabled OCM-1 choroidal melanomas using both bioluminescence and magnetic resonance imaging. This provides a simple approach with which to monitor tumor growth and treatment response in a highly efficient way which should allow a high daily animal throughput. These promising results will serve as a solid foundation with which dose routes and levels of anti-cancer drugs can be optimized and their efficacy evaluated with noninvasive imaging. Citation Format: John L. Chunta, Meridith Baugher, Deanne Lister, Erin Trachet, Kevin P. Guley, Chris Bull, Scott Wise, Wilbur R. Leopold, Patrick McConville. Orthotopic human choroidal melanoma model characterization with bioluminescence and magnetic resonance imaging for therapeutic efficacy evaluation. [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 3929. doi:10.1158/1538-7445.AM2014-3929

Abstract B152: MR-based assessment and quantification of cerebral edema in an orthotopic mouse glioma model.

Introduction: Brain edema is a prominent feature of brain cancer and contributes to neurologic dysfunction and impaired quality of life. While corticosteroids are the standard of care, improved treatments that are more efficacious and reduce side effects are sought after. In discovery of novel therapies for edema, animal brain tumor models that show relevant, reproducible and high incidence of tumor and cerebral edema are needed. Additionally, quantitative, non-invasive methods for detecting and measuring brain tumor associated edema are needed to facilitate therapeutic discovery efforts. In this work, an intracranial U251-luc human glioma model was characterized for edema incidence and progression. MRI-based relaxivity and contrast-enhancement protocols for edema detection and delineation from tumor tissue, were tested and characterized in the model. Methods: Female nude mice were implanted intracranially with 1×10∧6 U251-luc (Luc-mCherry) glioma cells. Tumor volume assessment was performed by manual segmentation of T1-weighted, gadolinium-enhanced anatomical brain images based on the assumption of tumor enhancement due to leaky vessels or compromised blood-brain barrier. Edema assessment was performed by manual segmentation of T2-weighted anatomical brain images based on the assumption that enhancing regions include tumor tissue in addition to regions of brain edema. The difference in T2-weighted and T1-weighted volumes was assumed to be edema. T2 maps were also generated over the whole brain to measure T2 values over the respective regions of interest, and confirm the differentiation between tumor tissue and edema. Results: Edema volume was successfully distinguished from contrast enhancing tumor tissue and quantified in orthotopically implanted U251-luc (Luc-mCherry) glioma. Incidence was 100%. Edema progression occurred with increasing tumor volume over time. Regions of edema exhibited greater T2 values compared to tumor tissue regions (that also had greater T2 than normal tissue). This is consistent with edematous tissue water content, compared with tumor tissue and normal brain. These results support the use of intracranial U251-luc (Luc-mCherry) as a reliable model for studying the effects of therapies targeting tumor-related cerebral edema, and the application of MR imaging for quantifying those effects real-time in vivo. Citation Information: Mol Cancer Ther 2013;12(11 Suppl):B152. Citation Format: Deanne R. Lister, Deepa Balagurunathan, Meridith Baugher, Erin Trachet, Patrick McConville, Scott Wise, W.R. Leopold. MR-based assessment and quantification of cerebral edema in an orthotopic mouse glioma model. [abstract]. In: Proceedings of the AACR-NCI-EORTC International Conference: Molecular Targets and Cancer Therapeutics; 2013 Oct 19-23; Boston, MA. Philadelphia (PA): AACR; Mol Cancer Ther 2013;12(11 Suppl):Abstract nr B152.

Abstract B15: Increased clinical relevance of orthotopic glioma models through bioluminescence and MR imaging.

Glioblastoma multiforme (GBM) is the most common and most aggressive form of malignant primary brain tumors, affecting nearly 35,000 people in the United States. Most preclinical studies in glioma utilize survival as the primary endpoint to study, which provides limited information about disease progression, tumor burden response to treatment (a primary clinical end point). We have characterized two human glioma cell lines, Gli36 and LN827 that were modified to express luciferase in order to enable in vivo monitoring of disease progression and response to treatment using bioluminescence imaging (BLI). Anatomical magnetic resonance imaging (MRI) was also performed to directly correlate bioluminescence signal with tumor volume. Both models exhibited >90% tumor take-rate and responded to treatment with temozolomide, a clinical standard of care. Analysis of lifespan, tumor volume doubling times, and tumor growth delay all indicated that BLI is a reliable indicator of disease progression and response to treatment. BLI-based endpoints also showed good correlation with MR-based endpoints. These results support the use of in vivo imaging in these modified cell lines for longitudinal monitoring of tumor progression and response to therapy. Imaging was not only a reliable method for quantifying tumor burden, but enabled clinically relevant end points that could not be accessed through non-imaging means or lifespan determination alone. Further utility can be driven in these models through the use of other functional imaging end points. Citation Information: Mol Cancer Ther 2013;12(11 Suppl):B15. Citation Format: Deanne Lister, Mary Anne Meade, Tracey Woolliscroft, Deepa Balagurunathan, Erin Trachet, Wilbur Leopold, Patrick McConville. Increased clinical relevance of orthotopic glioma models through bioluminescence and MR imaging. [abstract]. In: Proceedings of the AACR-NCI-EORTC International Conference: Molecular Targets and Cancer Therapeutics; 2013 Oct 19-23; Boston, MA. Philadelphia (PA): AACR; Mol Cancer Ther 2013;12(11 Suppl):Abstract nr B15.

Abstract B145: Use of μCT imaging in the PyMT breast cancer model to monitor lung metastasis development and determine therapeutic benefit in real time.

Breast cancer is the most frequently diagnosed form of cancer and the second leading cause of death in women. Death and most of the complications associated with breast cancer are due to the metastasis of the primary tumor to the lung, as well as other tissues. To visualize the metastatic process in real time, the genetically engineered mouse mammary tumor virus-polyoma middle T (MMTV-PyMT) mouse model was employed along with in vivo µCT (computed tomography) imaging to evaluate and characterize the size and incidence of lung nodules. µCT images are able to detect lung nodules as small as approximately 0.5mm3. The MMTV-PyMT model is widely used to evaluate metastatic breast cancer; the model produces lung metastases (80-90% incidence rate) in a more clinically relevant manner in a syngeneic mouse with intact immune system. In this study female MMTV-PyMT mice were allowed to develop primary mammary fat pad tumors (>3grams) which were then excised and processed into single cell suspension. Syngeneic female mice (FVB/J) were implanted with the cell suspension into mammary fat pad #4 and the resulting tumors were resected once the primary tumor burden surpassed 750mg (late stage). Standard twice weekly intravenous therapy with eribulin at 1mg/kg was initiated following resection and continued the duration of the experiment. µCT images were used to assess the presence and size of lung metastases in vivo starting at 3 weeks post-primary tumor resection. Mice were monitored daily via cage side observations and imaged a second time once labored breathing was apparent. Treatment with eribulin (1mg/kg, IV) not only decreased the incidence of lung metastases by 75% compared to the vehicle control group but also inhibited the ability of the primary tumor to regrow. Coupling the MMTV-PyMT model with µCT imaging allowed for the longitudinal evaluation of therapeutic efficacy on progressive lung metastases. Visualizing lung metastases in real time with in vivo imaging is advantageous since traditional pharmacology endpoints of the MMTV-PyMT model can be extremely long (20-40 weeks). Citation Information: Mol Cancer Ther 2013;12(11 Suppl):B145. Citation Format: Mary Anne Meade, Deepa Balagurunathan, Chris Bull, Deanne Lister, Erin Trachet, Bryan Smith, Daniel Flynn, Scott C. Wise. Use of μCT imaging in the PyMT breast cancer model to monitor lung metastasis development and determine therapeutic benefit in real time. [abstract]. In: Proceedings of the AACR-NCI-EORTC International Conference: Molecular Targets and Cancer Therapeutics; 2013 Oct 19-23; Boston, MA. Philadelphia (PA): AACR; Mol Cancer Ther 2013;12(11 Suppl):Abstract nr B145.

Abstract B142: Evaluating the feasibility and throughput of quintuple modality imaging in disseminated models of cancer with PET, CT, MRI, bioluminescence, and fluorescence molecular tomography.

Background: Multimodality imaging presents a unique opportunity to image the efficacy of candidate compounds against various models of cancer. The challenge is in integrating the complimentary data that these modalities can provide. The purpose of this study was to evaluate the feasibility of quintuple modality imaging in order to optimize workflows for maximal efficacy testing of candidate compounds using PET, CT, MRI, Bioluminescence imaging (BLI), and Fluorescence Molecular Tomography (FMT). Materials and Methods: Female SCID-Beige mice were implanted IV with 5×10∧6 MM1s-luc cells and female NIH III mice were implanted with 5×10∧6 5TGM-1-luc cells. The presence and extent of disease was confirmed with BLI concurrently or before multimodality imaging was initiated 28-35 days post-implant. The day prior to imaging, mice were injected IV with 2nmol of Integrisense 750 (PerkinElmer) to allow for 24 hours of uptake prior to FMT imaging. The next day, animals were anesthetized with 2.0% isoflurane, were injected with 200µCi 18F-FDG, and were immediately positioned in an FMT cassette for imaging. Whole body FMT data was acquired with excitation/emission at 755nm/775nm. MicroCT data was then acquired by placing the FMT cassette into a custom Animal Handling System (AHS) dock that coupled the FMT cassette to an integrated bed holder (ASI Instruments) for the remaining modalities. Whole body CT images were acquired at 512 projections, 75kVp, and 220µA. A ten minute static PET data set was then acquired with a Seimen9s Inveon system after a 1 hour tracer uptake period, and reconstructed with a 3D-OSEM algorithm. MicroPET imaging was followed by a series of T1-weighted, gadolinium-enhanced, and T2-weighted anatomical MR images using an Agilent 7 Tesla MRI. Results: Animals were maintained under anesthesia for a total of approximately 2 hours in order to acquire data from all the modalities. Data was able to be co-registered in various forms in order to verify the location and extent of disease, as well as the usefulness of 18F-FDG PET and Integrisense to serve as biomarker probes. Conclusions: Animals tolerated the length of anesthesia well. Initial difficulties were found in performing the BLI on the same day as the remaining modalities, as anesthesia administration to animals inside the FMT cassette while acquiring BLI perturbed animal positioning. When BLI was acquired the day prior to multimodality imaging, the reliability of positioning was increased, permitting more facile fusion of the various modalities. The feasibility of this imaging was only made possible through the use of the AHS bed and dock that permitted reproducible imaging across the different platforms. In order to further optimize workflows, multiple AHS bed systems should be employed so that data can be acquired on the various modalities in parallel, instead of sequentially, thereby greatly increasing animal throughputs. Citation Information: Mol Cancer Ther 2013;12(11 Suppl):B142. Citation Format: John L. Chunta, Deanne Lister, Chris Bull, Deepa Balagurunathan, Erin Trachet, Chris Chiodo, Scott Wise, Wilbur R. Leopold, Patrick McConville. Evaluating the feasibility and throughput of quintuple modality imaging in disseminated models of cancer with PET, CT, MRI, bioluminescence, and fluorescence molecular tomography. [abstract]. In: Proceedings of the AACR-NCI-EORTC International Conference: Molecular Targets and Cancer Therapeutics; 2013 Oct 19-23; Boston, MA. Philadelphia (PA): AACR; Mol Cancer Ther 2013;12(11 Suppl):Abstract nr B142.