Abstract B39: Identification and targeting mediators of chemoresistance using the patient-derived xenograft model of ovarian cancer

Abstract

Abstract Objectives: The objectives of this study were to use the heterogeneity of the patient-derived xenograft (PDX) model to characterize the surviving population for de novo mediators of chemotherapy resistance, and subsequently target pathways contributing to that resistance. Methods: Tumors removed during primary tumor-reductive surgery were implanted directly into SCID mice, either subcutaneously and intraperitoneally. Mice with PDX tumors were treated with combination carboplatin/paclitaxel for 4 weeks. In addition, chemoresistant models were developed by treatment until complete response was achieved, and recurrent tumors retreated until resistant. Patient tumors, developing PDX tumors, and post-chemo surviving residual tumors were subjected to qPCR arrays, RNA-Seq, Methyl-Seq, and proteomic analysis; and IHC for Ki-67 and cancer stem cell markers CD133, ALDH1, and CD44. The Pol-I inhibitor CX-5461 was used to treat five PDX models. Results: At the mRNA level, the original patient tumors were similar to PDX tumors in mice as assessed by an 84-oncogene expression panel. Subcutaneous and intraperitoneal tumors were also remarkably similar. Patient responses correlated well with responses in mice. Proteomic analysis identified 14,023 unique proteins, of which approximately 30% were murine, and 70% were human in each sample. A total of 1,202 of human proteins were found to be differently expressed (p-value < 0.01) between patient and PDX tumors. Of these, 996 (82.9%) were reduced in the PDX model, and predominantly participants in the immune system. PDX tumors collected after chemotherapy showed that the surviving population was significantly more dormant than the original tumor (Ki-67 67% versus 31%, P<0.01), and enriched in all three cancer stem cell populations (2.1-4.7 fold, p<0.05). Principle component analysis of both RNA-Seq and proteomic data show that sensitive and resistant tumors were much more similar to one another than to tumors from different patients, suggesting a limited number of genes contributing to chemoresistance. IPA analysis of RNA-Seq data from treated and untreated PDX tumor pairs demonstrated that the most commonly enriched pathways are those of Protein Kinase A, GNRH, Sphingosine-1-phosphate, and alpha-adrenergic signaling. Additionally, numerous regulators of ribosomal synthesis were significantly altered by treatment. In vivo, 5 PDX models were treated with CX-5461; two had progression, one had stable disease, one had a 60% reduction in tumor growth, and one had a complete response to single agent therapy, still without recurrence after 3 months. Conclusion: The ovarian-PDX model exhibits consistent molecular and biologic similarity to patient tumors. Post-chemo tumors reveal multiple pathways to be consistently altered. Taking this information back to the PDX model demonstrated that inhibition of ribosomal RNA synthesis was highly effective, offering a novel opportunity to preferentially target the chemoresistant population in ovarian cancer. Citation Format: Zachary Dobbin, Robert Cornelison, Ashwini K. Katre, David A. Schneider, Karen Watters, Fabian Coscia, Ernst Lengyel, Aaron Buechlein, Fang Fang, Kenneth Nephew, Charles N. Landen, Jr.. Identification and targeting mediators of chemoresistance using the patient-derived xenograft model of ovarian cancer. [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 B39.