Amanda Katz

A germline DNA polymorphism enhances alternative splicing of the KLF6 tumor suppressor gene and is associated with increased prostate cancer risk

Prostate cancer is a leading and increasingly prevalent cause of cancer death in men. Whereas family history of disease is one of the strongest prostate cancer risk factors and suggests a hereditary component, the predisposing genetic factors remain unknown. We first showed that KLF6 is a tumor suppressor somatically inactivated in prostate cancer and since then, its functional loss has been further established in prostate cancer cell lines and other human cancers. Wild-type KLF6, but not patient-derived mutants, suppresses cell growth through p53-independent transactivation of p21. Here we show that a germline KLF6 single nucleotide polymorphism, confirmed in a tri-institutional study of 3,411 men, is significantly associated with an increased relative risk of prostate cancer in men, regardless of family history of disease. This prostate cancer-associated allele generates a novel functional SRp40 DNA binding site and increases transcription of three alternatively spliced KLF6 isoforms. The KLF6 variant proteins KLF6-SV1 and KLF6-SV2 are mislocalized to the cytoplasm, antagonize wtKLF6 function, leading to decreased p21 expression and increased cell growth, and are up-regulated in tumor versus normal prostatic tissue. Thus, these results are the first to identify a novel mechanism of self-encoded tumor suppressor gene inactivation and link a relatively common single nucleotide polymorphism to both regulation of alternative splicing and an increased risk in a major human cancer.

Frequent inactivation of the tumor suppressor Kruppel-like factor 6 (KLF6) in hepatocellular carcinoma.

Hepatocellular carcinoma (HCC) is a leading cause of cancer death worldwide, reflecting incomplete characterization of underlying mechanisms and lack of early detection. Krüppel‐like factor 6 (KLF6) is a ubiquitously expressed zinc finger transcription factor that is deregulated in multiple cancers through loss of heterozygosity (LOH) and/or inactivating somatic mutation. We analyzed the potential role of the KLF6 tumor suppressor gene in 41 patients who had HCC associated with hepatitis C virus (16 patients), hepatitis B virus (12 patients, one of whom was coinfected with hepatitis C virus), and other etiologies (14 patients) by determining the presence of LOH and mutations. Overall, LOH and/or mutations were present in 20 (49%) of 41 tumors. LOH of the KLF6 gene locus was present in 39% of primary HCCs, and the mutational frequency was 15%. LOH and/or mutations were distributed across all etiologies of HCC evaluated, including patients who did not have cirrhosis. Functionally, wild‐type KLF6 decreased cellular proliferation of HepG2 cells, while patient‐derived mutants did not. In conclusion, we propose that KLF6 is deregulated by loss and/or mutation in HCC, and its inactivation may contribute to pathogenesis in a significant number of these tumors. (HEPATOLOGY 2004;40:1047–1052.)

RET Fusion Lung Carcinoma: Response to Therapy and Clinical Features in a Case Series of 14 Patients

Background RET (rearranged during transfection) fusions have been reported in 1% to 2% of lung adenocarcinoma (LADC) cases. In contrast, KIF5B‐RET and CCDC6‐RET fusion genes have been identified in 70% to 90% and 10% to 25% of tumors, respectively. The natural history and management of RET‐rearranged LADC are still being delineated. Materials and Methods We present a series of 14 patients with RET‐rearranged LADC. The response to therapy was assessed by the clinical response and an avatar model in 2 cases. Patients underwent chemotherapy, targeted therapy, and immunotherapy. Results A total of 14 patients (8 women; 10 never smokers; 4 light smokers; mean age, 57 years) were included. KIF5B‐RET and CCDC6‐RET variants were diagnosed in 10 and 4 cases, respectively. Eight patients had an early disseminated manifestation, seven with KIF5B‐RET rearranged tumor. The features of this subset included bilateral miliary lung metastases, bone metastases, and unusual early visceral abdominal involvement. One such patient demonstrated an early and durable complete response to cabozantinib for 7 months. Another 2 patients treated with cabozantinib experienced a partial response, with rapid significant clinical improvement. Four patients with tumors harboring CCDC6‐RET and KIF5B‐RET fusions showed pronounced and durable responses to platinum‐based chemotherapy that lasted for 8 to 15 months. Two patients’ tumors showed programmed cell death ligand 1‐positive staining but did not respond to pembrolizumab. The median overall survival was 22.8 months. Conclusion RET‐rearranged LADC in our series tended to occur as bilateral disease with early visceral involvement, especially with KIF5B fusion. Treatment with cabozantinib achieved responses, including 1 complete response. However, further studies are required in this group of patients. Micro‐Abstract Data are increasing regarding RET (rearranged during transfection) fusions in lung cancer. We present our experience with the natural history of this disease and its response to targeted therapy and standard chemotherapy in 14 patients. In our series, RET‐rearranged lung adenocarcinoma had an early disseminated presentation, especially with KIF5B fusion. Treatment with cabozantinib achieved responses, including 1 complete response.

Growth of Triple Negative and Progesterone Positive Breast Cancer Causes Oxidative Stress and Down-Regulates Neuroprotective Transcription Factor NPAS4 and NPAS4-Regulated Genes in Hippocampal Tissues of TumorGraft Mice—an Aging Connection

While the refinement of existing and the development of new chemotherapeutic regimens has significantly improved cancer treatment outcomes and patient survival, chemotherapy still causes many persistent side effects. Central nervous system (CNS) toxicity is of particular concern, as cancer patients experience significant deficits in memory, learning, cognition, and decision-making. These chemotherapy-induced cognitive changes are termed chemo brain, and manifest in more than half of cancer survivors. Moreover, recent studies have emerged suggesting that neurocognitive deficits manifest prior to cancer diagnosis and treatment, and thus may be associated with tumor presence, a phenomenon recently termed “tumor brain.” To dissect the molecular mechanisms of tumor brain, we used TumorGraftTM models, wherein part of a patients tumor is grafted into immune-deficient mice. Here, we analyzed molecular changes in the hippocampal tissues of mice carrying triple negative (TNBC) or progesterone receptor positive (PR+BC) xenografts. TNBC growth led to increased oxidative damage, as detected by elevated levels of 4-hydroxy-2-nonenal, a product of lipid peroxidation. Furthermore, the growth of TNBC and PR+BC tumors altered global gene expression in the murine hippocampus and affected multiple pathways implicated in PI3K-Akt and MAPK signaling, as well as other pathways crucial for the proper functioning of hippocampal neurons. TNBC and PR+BC tumor growth also led to a significant decrease in the levels of neuronal transcription factor NPAS4, a regulator that governs the expression of brain-derived neurotrophic factor (BDNF), and several other key brain neurotrophic factors and pro-survival molecules. The decreased expression of ERK1/2, NPAS4, and BDNF are also seen in neurodegenerative conditions and aging, and may constitute an important tumor brain mechanism.

Growth of malignant extracranial tumors alters microRNAome in the prefrontal cortex of TumorGraft mice

A wide array of central nervous system complications, neurological deficits, and cognitive impairments occur and persist as a result of systemic cancer and cancer treatments. This condition is known as chemo brain and it affects over half of cancer survivors. Recent studies reported that cognitive impairments manifest before chemotherapy and are much broader than chemo brain alone, thereby adding in tumor brain as a component. The molecular mechanisms of chemo brain are under-investigated, and the mechanisms of tumor brain have not been analyzed at all. The frequency and timing, as well as the long-term persistence, of chemo brain and tumor brain suggest they may be epigenetic in nature. MicroRNAs, small, single-stranded non-coding RNAs, constitute an important part of the cellular epigenome and are potent regulators of gene expression. miRNAs are crucial for brain development and function, and are affected by a variety of different stresses, diseases and conditions. However, nothing is known about the effects of extracranial tumor growth or chemotherapy agents on the brain microRNAome. We used the well-established TumorGraft ™ mouse models of triple negative (TNBC) and progesterone receptor positive (PR+BC) breast cancer, and profiled global microRNAome changes in tumor-bearing mice upon chemotherapy, as compared to untreated tumor-bearing mice and intact mice. Our analysis focused on the prefrontal cortex (PFC), based on its roles in memory, learning, and executive functions, and on published data showing the PFC is a target in chemo brain. This is the first study showing that tumor presence alone significantly impacted the small RNAome of PFC tissues. Both tumor growth and chemotherapy treatment affected the small RNAome and altered levels of miRNAs, piRNAs, tRNAs, tRNA fragments and other molecules involved in post-transcriptional regulation of gene expression. Amongst those, miRNA changes were the most pronounced, involving several miRNA families, such as the miR-200 family and miR-183/96/182 cluster; both were deregulated in tumor-bearing and chemotherapy-treated animals. We saw that miRNA deregulation was associated with altered levels of brain-derived neurotrophic factor (BDNF), which plays an important role in cognition and memory and is one of the known miRNA targets. BDNF downregulation has been associated with an array of neurological conditions and could be one of the mechanisms underlying tumor brain and chemo brain. In the future our study could serve as a roadmap for further analysis of cancer and chemotherapy’s neural side effects, and differentially expressed miRNAs should be explored as potential tumor brain and chemo brain biomarkers.

Chemo brain or tumor brain - that is the question: The presence of extracranial tumors profoundly affects molecular processes in the prefrontal cortex of TumorGraft mice

Cancer chemotherapy causes numerous persistent central nervous system complications. This condition is known as chemo brain. Cognitive impairments occur even before treatment, and hence are referred to as cancer associated cognitive changes, or tumor brain. There is much yet to be learned about the mechanisms of both chemo brain and tumor brain. The frequency and timing of chemo brain and tumor brain occurrence and persistence strongly suggest they may be epigenetic in nature and associated with altered gene expression. Here we used TumorGraftTM models wherein part of a patients tumor is removed and grafted into immune-deficient mice and conducted global gene expression and DNA methylation analysis. We show that malignant non-central nervous system tumor growth causes profound molecular alterations in the brain. Mice harbouring triple negative or progesterone positive breast cancer TumorGrafts exhibited altered gene expression, decreased levels of DNA methylation, increased levels of DNA hydroxymethylation, and oxidative stress in the prefrontal cortex. Interestingly, chemotherapy did not have any additional synergistic effects on the analyzed processes. The molecular changes observed in this study are known signs of neurodegeneration and brain aging. This study provides an important roadmap for future large-scale analysis of the molecular and cellular mechanisms of tumor brain.

Abstract B40: Mouse clinical trials: integrating PDX models of sarcoma subtypes with genomics to replicate patient responses to cancer therapeutics

Objective: Sarcomas are clinically and genetically heterogeneous tumors that are often difficult to treat. Patient-derived xenograft (PDX or TumorGraft) models have been shown to accurately reflect the characteristics of patient tumors and may be useful tools for developing personalized treatment strategies and deployment in mouse clinical trials assessing novel therapies. We evaluated the accuracy of PDX models in reproducing clinical responses to standard and experimental drugs used for sarcoma treatment. Methods: Fresh tumor tissue (comprising 172 distinct explants) was collected by surgery or biopsy from 150 patients with sarcoma and implanted into immunodeficient mice. Tumors successfully engrafting were screened using next-generation sequencing technology to identify key genomic alterations with therapeutic implications. PDX sensitivity to standard of care and experimental agents was evaluated and tumor growth inhibition/regression values and clinical RECIST outcomes determined. Drug screening results were correlated with individual patient outcomes. Results: Of the 172 implanted tumors, 145 have completed the implantation process, with 86 (59%) successfully establishing a PDX model. Engraftment rate depended on sarcoma subtype and specimen origin (surgical explant versus biopsy). Next generation sequencing of models from major sarcoma subtypes (Ewing sarcoma, leiomyosarcoma, liposarcoma, osteosarcoma, and rhabdomyosarcoma) highlighted alterations in 454 genes, including those informing treatment selection such as PIK3CA, MET, and CDK4. A total of 26 PDX models from 25 patients across the major sarcoma subtypes were screened in 148 drug tests employing 64 FDA-approved drugs/combinations such as ifosfamide, and gemcitabine/docetaxel, and 26 experimental therapies in clinical trial. In 13/13 (100%) cases with available data, a significant correlation between patient clinical response and PDX model outcome was noted (p=0.0004; Fisher9s exact test). Conclusions: Given the close match between patient clinical responses and PDX model outcomes, these results validate the concept of mouse clinical trials for determining the efficacy of novel therapies in sarcoma prior to broad application in expensive human trials. Moreover, the retention of alterations in key genes influencing therapeutic decision-making suggests a use for PDX models in functionally validating genomic hypotheses in a pre-clinical setting. Citation Format: Amanda Katz, Raphael E. Pollock, Leonard H. Wexler, Carlos Rodriguez-Galindo, Jonathan C. Trent, Robert Maki, Jennifer Jaskowiak, Lindsay Ryland, Daniel Ciznadija, Angela Davies, Keren Paz. Mouse clinical trials: integrating PDX models of sarcoma subtypes with genomics to replicate patient responses to cancer therapeutics. [abstract]. In: Proceedings of the AACR Special Conference: Patient-Derived Cancer Models: Present and Future Applications from Basic Science to the Clinic; Feb 11-14, 2016; New Orleans, LA. Philadelphia (PA): AACR; Clin Cancer Res 2016;22(16_Suppl):Abstract nr B40.

Abstract A8: The ImmunoGraft: A humanized mouse model for translational assessment of immunotherapy in solid tumors

Background: Therapeutics reactivating the immune system have demonstrated promise, with durable objective responses in patients with a variety of solid tumors. Despite these successes, current animal models do not reliably identify immunotherapeutic targets with the greatest clinical potential, due in part to differences between human and murine immune systems. Hence, development of robust preclinical tools to test such drugs against human tumors in the context of an allogeneic immune system remains an imperative. We have previously demonstrated the generation of its ImmunoGraftTM platform, whereby two technologies, the patient-derived xenograft (PDX) and humanized mice (immunodeficient mice reconstituted with a human immune system), are combined in a single platform. We now report on the utility of the ImmunoGraftTM for assessing the effect of immune-modulating agents in solid tumors. Materials and Methods: Immune-compromised NOG (PrkdcscidIl2rgtm1Sug) mice were reconstituted with human CD34+ cells and monitored for the expansion of human immune cells (humanized). Humanized mice were engrafted with solid tumors that had been subjected to histocompatibility typing and characterized for a number of molecular markers, including PD-L1 expression. Tumor growth in the ImmunoGraftsTM was compared against non-humanized counterparts, as well as the level of immune reconstitution. Finally, ImmunoGraftsTM were treated with drugs blocking the immune checkpoints CTLA4 and PD1 and human immune activation and tumor growth inhibition evaluated. Results: Mature human CD45+ cells comprised close to 50% of the leukocytes detected in the circulation and lymphoid organs of humanized mice. Solid tumors, including NSCLC, melanoma, and head and neck cancer, were successfully engrafted in the humanized mice. Moderate to high expression of PD-L1 was found in approximately 80% of these tumors. ImmunoGraftsTM treated with anti-CTLA4 or anti-PD1 antibodies exhibited systemic immune responses characterized by robust proliferation of splenic and circulating huCD3+ T cells, as well as activated huCD4+ Th1 cells. There was also an increase in tumor-infiltrating huCD8+ cytotoxic T lymphocytes and huCD68+ macrophages, along with elevated secretion of human-specific cytokines. Tumor growth inhibition, and in some instances tumor regression, was demonstrated in treated ImmunoGraftsTM. The magnitude of growth inhibition correlated with the level of immune activation. Conclusion: The ImmunoGraftTM is an innovative pre-clinical model enabling immunotherapeutic agents to be evaluated for efficacy in solid tumors. This platform is more reflective of the human tumor microenvironment (both immune and non-immune cell-based) and may be one of the most translationally-relevant models to date for screening therapies targeting the immune system. To gauge the clinical potential of the ImmunoGraftTM, a retrospective analysis is currently ongoing using PDX models developed from patients treated with immuno-oncology drugs. The ImmunoGraftTMhas the potential to revolutionize translational drug discovery and development for immunotherapeutic agents in oncology. Citation Format: Gilson Baia, David Vasquez, David Cerna, Daniel Ciznadija, David Sidransky, Amanda Katz, Keren Paz. The ImmunoGraft: A humanized mouse model for translational assessment of immunotherapy in solid tumors. [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 A8.

Abstract A14: Molecular fidelity of patient derived xenograft (PDX) models to original human tumor and to the cancer genome atlas (TCGA)

Background: Patient-derived xenograft (PDX) models, also known as Champions TumorGraft® models, maintain the complex intra-tumoral biology of the primary tumor. Over 250 of the Champions models, ranging over a wide variety of solid tumors and passaging generations, have been analyzed using whole exome sequencing (WES) and RNA sequencing (RNAseq). SNPs, InDels and copy number alterations (CNAs) data have been generated for each model, following the Genome Analysis Toolkit (GATK). While several publications compare small numbers of PDX models and human tumors on the molecular level, this is the first known comprehensive analysis whereby the molecular fidelity of the PDX platform is corroborated across several cancer types and throughout different mouse generations. Method and Results: First, we compared PDXs to their human original counterparts using a preliminary group of four PDX models with available matching human patient WES data. Patient tumor source included dedifferentiated liposarcoma, synovial sarcoma, renal cell carcinoma and squamous cell carcinoma of the lung. PDX passages ranged from 2 to 4. We compared called mutations and a high percentage of identified human tumor mutations were present in the PDX models (42-82%), with the lowest scoring model also showing signs of normal contamination in the human tumor sample. For CNAs in oncogenic sites, we saw an average of 65% of human tumor alterations recurring in the PDX models. This was observed, despite inherent difficulties due to exome- based CNA analysis methods. Encouraged by the individual patient results, we subjected our largest (per cancer type) PDX cohorts to a molecular comparison with the equivalent TCGA cohorts. More than 200 of the sequenced models, grouped into colorectal adenocarcinoma (COADREAD), lung adenocarcinoma (LUAD), breast carcinoma (BRCA), head and neck squamous cell carcinoma (HNSC) and ovarian serous carcinoma (OV) cohorts were compared. We applied mutation category (MC) and significantly mutated genes (SMG) analysis, as well as comparison of mutation population frequencies for TCGA SMG. Results showed high correlation between the TCGA and the Champions PDX cohorts, although the level of matching varied between cancer types. For instance, COADREAD was highly correlative, while other cancer types, such as BRCA, showed bias toward CpG site mutations. In SMG analysis and population frequency analysis, major SMGs recur across the cohorts, while, as expected, weaker signals from the TCGA were often missed in the smaller cohorts. Conclusions: Detailed comparison of several PDX models to the human tumor counterpart demonstrated high fidelity, not only at the gene level but also the mutation and CNA level. Cohort comparisons were correlative as well, but a certain bias was discerned in both MC and SMG analyses. There could be several causes for this, including statistical artifacts due to small cohort sizes, clinical and demographic differences between the Champions and TCGA patient profiles, or biological factors such as clonal selection and engraftment pressure. Further analysis is ongoing to better understand the model at a molecular level and maximize its utility as a robust translational research tool. Citation Format: Ido Sloma, Ido Ben-zvi, Tin Khor, Daniel Ciznadija, Amanda Katz, David Vasquez, Jennifer Jaskowiak, Lindsay Ryland, Angela Davies, David Sidransky, Keren Paz. Accurate molecular fidelity of patient-derived xenograft (PDX) models to original human tumors and to The Cancer Genome Atlas (TCGA). [abstract]. In: Proceedings of the AACR Special Conference: Patient-Derived Cancer Models: Present and Future Applications from Basic Science to the Clinic; Feb 11-14, 2016; New Orleans, LA. Philadelphia (PA): AACR; Clin Cancer Res 2016;22(16_Suppl):Abstract nr A21.

Abstract 3219: A patient-centric repository of PDX models for translational oncology research

Proceedings: AACR 106th Annual Meeting 2015; April 18-22, 2015; Philadelphia, PA Patient-derived xenograft (PDX) models maintain the complex intra-tumoral biology and heterogeneity of an intact malignancy, as well as the interplay with stromal components and other cells fluxing into the tumor environment. This intrinsic cross-talk between different elements of the tumor makes PDX models a superior tool for translational drug discovery research and personalized oncology studies. Champions PDX models were originally developed for personalizing cancer treatments through the different Champions clinical programs. These models accurately reflect the population of patients enrolling in clinical trials. We describe herein our extensive TumorBank of PDX models, a valuable resource for translational oncology research to predetermine target populations for intervention with novel therapeutics in specific cancer subtypes. Tumor tissue from over 950 patients with a variety of primary and metastatic solid malignancies, across all ages and ethnicities and encompassing both treatment-naive and heavily-pretreated individuals, has been implanted into immunodeficient mice with successful engraftment observed in ∼72% of cases. Comprehensive and translational-relevant clinical annotations have been maintained for these PDX models, including patient demographics, disease stage, anatomic location, tumor grade and histology, and treatment history. Importantly, whole exome and RNA sequencing, tissue histopathology, and protein immunohistochemistry have all been applied to 297 of these models. Finally, 70 of the models were screened against the corresponding patients treatment used in the clinic, demonstrating a sensitivity of 98%, specificity of 76%, positive predictive value of 89% and negative predictive value of 96%. This wealth of information can be accessed through the Champions TumorGraft Database. The combination of extensive molecular and clinical annotation, together with opportunities for unlimited prospective preclinical testing, makes Champions TumorBank a pioneering resource for pharmaceutical companies seeking to identify target populations for therapeutic intervention. Citation Format: Tin O. Khor, Ido Ben Zvi, Amanda Katz, David Vasquez-Dunddel, Ido Sloma, Daniel Ciznadija, David Sidransky, Keren Paz. A patient-centric repository of PDX models for translational oncology research. [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 3219. doi:10.1158/1538-7445.AM2015-3219