Search

Immune checkpoint blockade (ICB) has changed the standard of care for many patients with cancer, yet no ICB is approved for ovarian cancer. We hypothesized that maintenance therapy with an IL15 “superagonist” (N-803) and ICB in combination could induce potent immune activation in ovarian cancer. Using flow cytometry, cytometry by time of flight analysis, and cytotoxicity assays, we analyzed patient samples from women with advanced epithelial ovarian cancer treated with N-803 for indications of PD-1/PD-L1 upregulation with this treatment. In addition, ICB and N-803 were evaluated in preclinical studies to determine the functional impact of combination therapy on natural killer (NK) cells in vitro and in vivo. We observed that N-803 stimulated initial NK-cell expansion in patient samples; however, proliferation was not sustained beyond 2 weeks despite continued treatment. This result was reverse translated back to the laboratory to determine the functional relevance of this finding. The addition of ICB with an antibody-dependent cellular cytotoxicity IgG1 antibody against PD-L1 (avelumab) or an IgG4 antibody against PD-1 (pembrolizumab) enhanced N-803 induced NK-cell function in vitro. Using models of human ovarian cancer and NK-cell adoptive transfer in mice, we showed enhanced antitumor control with N-803 and ICB, as well as a combination effect that enhanced NK-cell persistence and expansion in vivo. This work suggests that PD-1/PD-L1 blockade combined with IL15 signaling may overcome resistance to cytokine therapy in ovarian cancer.

Supplemental Table 1 and Supplemental Figures 1-5

Immune checkpoint blockade (ICB) has changed the standard of care for many patients with cancer, yet no ICB is approved for ovarian cancer. We hypothesized that maintenance therapy with an IL15 “superagonist” (N-803) and ICB in combination could induce potent immune activation in ovarian cancer. Using flow cytometry, cytometry by time of flight analysis, and cytotoxicity assays, we analyzed patient samples from women with advanced epithelial ovarian cancer treated with N-803 for indications of PD-1/PD-L1 upregulation with this treatment. In addition, ICB and N-803 were evaluated in preclinical studies to determine the functional impact of combination therapy on natural killer (NK) cells in vitro and in vivo. We observed that N-803 stimulated initial NK-cell expansion in patient samples; however, proliferation was not sustained beyond 2 weeks despite continued treatment. This result was reverse translated back to the laboratory to determine the functional relevance of this finding. The addition of ICB with an antibody-dependent cellular cytotoxicity IgG1 antibody against PD-L1 (avelumab) or an IgG4 antibody against PD-1 (pembrolizumab) enhanced N-803 induced NK-cell function in vitro. Using models of human ovarian cancer and NK-cell adoptive transfer in mice, we showed enhanced antitumor control with N-803 and ICB, as well as a combination effect that enhanced NK-cell persistence and expansion in vivo. This work suggests that PD-1/PD-L1 blockade combined with IL15 signaling may overcome resistance to cytokine therapy in ovarian cancer.

Supplemental Table S2

Animation of PCA for Figure 1

Supplemental Table S1

Supplemental Table S4

Supplemental Table S3

The best known ovarian cancer biomarker, CA125, is neither adequately sensitive nor specific for screening the general population. By using a combination of proteins for screening, it may be possible to increase the sensitivity and specificity over CA125 alone. In this study, we used Proseek Multiplex Oncology II plates to simultaneously measure the expression of 92 cancer-related proteins in serum using proximity extension assays. This technology combines the sensitivity of the PCR with the specificity of antibody-based detection methods, allowing multiplex biomarker detection and high-throughput quantification. We analyzed 1 μL of sera from each of 61 women with ovarian cancer and compared the values obtained with those from 88 age-matched healthy women. Principle component analysis and unsupervised hierarchical clustering separated the ovarian cancer patients from the healthy, with minimal misclassification. Data from the Proseek plates for CA125 levels exhibited a strong correlation with clinical values for CA125. We identified 52 proteins that differed significantly (P < 0.006) between ovarian cancer and healthy samples, several of which are novel serum biomarkers for ovarian cancer. In total, 40 proteins had an estimated area under the ROC curve of 0.70 or greater, suggesting their potential to serve as biomarkers for ovarian cancer. CA125 alone achieved a sensitivity of 93.4% at a specificity of 98%. By adding the Oncology II values for five proteins to CA125 in a multiprotein classifier, we increased the assay sensitivity to 98.4% at a specificity of 98%, thereby improving the sensitivity and specificity of CA125 alone.

There is an urgent need for development of alternative treatment strategies for advanced-stage ovarian cancer. During metastatic spread, single cells or spheroids detach from the primary tumor and are carried by the peritoneal fluid until attaching to the mesothelial cells lining the abdominal cavity. While attachment of ovarian tumor cells and spheroids to the mesothelial layer is known to be mediated by β-integrins and CD44, efforts to block these receptors with antibodies have been insufficient to inhibit disease progression. As graphene oxide nanosheets have previously been shown bind to β-integrins on the surface of mesenchymal stem cells, we hypothesized that they would be able to attach to ovarian cancer spheroids, forming a physical barrier around the spheroids and thus inhibiting their adhesion to the mesothelium. SKOV-3 human ovarian cancer cells were used in this study, as this cell line was derived from ascites fluid. SKOV-3 cells were placed in poly(hydroxyethyl methacrylate) (polyHEMA)-coated plates for 24 hours to form spheroids, followed by 48-hour treatment with the nanosheets. The graphene oxide nanosheets were functionalized with 4-arm polyethylene glycol (PEG) to enhance biocompatibility and provide additional steric hindrance to prevent spheroid adhesion. Scanning electron microscopy confirmed that the nanosheets were able to bind to the tumor spheroids. Association of the nanosheets with the SKOV-3 spheroids reduced their adhesion to extracellular matrix proteins collagen and fibronectin as well as LP-9 human mesothelial cells over a 7-day time period, with the PEG-functionalized nanosheets showing the greatest reduction in adhesion. The nanomaterials did not exhibit cytotoxicity to LP-9 cells at the range of concentrations used to prevent adhesion. Receptor blocking studies confirmed that the nanosheets were predominantly engaging with the tumor spheroids via β-integrins. Finally, for tumor spheroids that did bind to the mesothelial cells, the presence of nanosheets on the surface of the spheroids decreased the rate of mesothelial layer clearance by the tumor cells. Taken together, these results demonstrate that graphene oxide nanomaterials can bind to ovarian tumor spheroids and inhibit their attachment to matrix proteins and mesothelial cells, while also reducing their ability to clear mesothelial cells once bound. Future efforts are focused on testing these materials with human ascites fluid and further modifying the nanosheets to enable ablation of tumor cells bound to the nanosheets to enhance their therapeutic potential. Citation Format: Hak Rae Lee, Melissa A. Geller, Samira M. Azarin. Inhibition of ovarian cancer spheroid adhesion using graphene oxide nanomaterials [abstract]. In: Proceedings of the AACR Special Conference on Advances in Ovarian Cancer Research; 2019 Sep 13-16, 2019; Atlanta, GA. Philadelphia (PA): AACR; Clin Cancer Res 2020;26(13_Suppl):Abstract nr B52.

Early detection is the key to increased survival for women with ovarian cancer, yet a screening tool has yet to be developed that is adequately sensitive and specific enough for use in the general population. In contrast, screening for cervical cancer by Pap tests has been routinely performed for over 50 years. In the liquid-based Pap test, cells are collected from the cervix and placed into an alcohol-based fixative and then examined for abnormal cells. Since ovarian cancer cells have been observed in Pap tests, we reasoned that ovarian cancer peptide biomarkers may also be present. Our central hypothesis is that proteins shed by ovarian cancer cells can be detected during routine Pap tests by mass spectrometry (MS)-based proteomics. In particular, when collected at the time of cervical cancer Pap test screening, the alcohol-based Pap test fixative and cervical swabs are ideal for biomarker discovery since they are derived from a site near the ovarian cancer (i.e. proteins may be secreted or shed from the tumor and flow through the fallopian tube into the uterus and out the cervical opening). Recently, the fimbria of the fallopian tube have been suggested to be the true precursor to ovarian cancer, strengthening our hypothesis that ovarian cancer proteins will be found in the lower genital tract, perhaps even at early stages. To demonstrate the feasibility of using Pap tests as a biospecimen for proteomics, we previously examined the proteins present in residual Pap test fixative samples from women with normal cervical cytology by MS and described 152 proteins in the “Normal Pap Proteome.” The objective of this study was to identify and compare the proteins from three different sources from the same ovarian cancer patient: (i) the residual Pap test fixative, (ii) a Merocel swab of the cervix, and (iii) the primary ovarian cancer tumor tissue. Proteins were concentrated from the cell-free supernatant of the Pap test fixative or eluted from the swab, and then trypsin digested using the filter-aided sample preparation method. A total protein extract from the patient's tumor tissue was digested by standard in-solution trypsin digestion. The samples were run on 2D-liquid chromatography MS/MS, followed by bioinformatics integration. We identified over 5000 proteins total in the three samples. More than 2000 proteins were expressed in all three ovarian cancer samples, including several known ovarian cancer biomarkers such as CA125. By Scaffold analysis of the Gene Ontology nomenclature of the proteins, we classified the proteins by both cellular localization and biological processes. Additional matched samples from patients will be used to build a library of proteins and peptides that are specific to ovarian cancer for use in the development of targeted MS assays. We conclude that quantification of proteins from Pap test fixatives and cervical swabs will prove to be a rich source of biomarkers for ovarian cancer detection. Citation Format: Kristin L.M. Boylan, Anna C. Rogers, Melissa A. Geller, Peter A. Argenta, Timothy J. Griffin, and Amy P.N. Skubitz. COMPARISON OF POTENTIAL OVARIAN CANCER BIOMARKERS BY MASS SPECTROMETRY-BASED PROTEOMIC ANALYSIS OF RESIDUAL PAP TEST FLUID, CERVICAL SWABS, AND TUMOR TISSUE FROM AN OVARIAN CANCER PATIENT [abstract]. In: Proceedings of the 12th Biennial Ovarian Cancer Research Symposium; Sep 13-15, 2018; Seattle, WA. Philadelphia (PA): AACR; Clin Cancer Res 2019;25(22 Suppl):Abstract nr DP-002.

BACKGROUND: Patients with ROC have limited treatment options in later lines of therapy, as many become platinum-resistant or -ineligible. Poly(ADP-ribose) polymerase inhibitor (PARPi) treatment is one option, however these agents are currently not approved for BRCAwt patients in the treatment setting. Niraparib, a PARPi, has demonstrated increased progression-free survival vs placebo as maintenance therapy for platinum-responsive ROC. Niraparib was effective regardless of BRCAmut or homologous recombination deficiency (HRD) status, although an increased treatment effect was observed in the HRDpos population. QUADRA (NCT02354586) evaluated niraparib treatment in heavily pretreated ROC patients, regardless of BRCAmut or HRD status, platinum sensitivity, or prior PARPi use. METHODS: Eligible patients had grade 2-3 serous ROC, ≥3 prior chemotherapy lines, and measurable disease. Patients were evaluated for BRCAmut and HRD status (MyChoice HRD Test). Patients received niraparib 300 mg once daily until progression; dose could be lowered to 200 mg or 100 mg due to treatment-emergent adverse events (TEAEs). The primary endpoint was objective response rate (ORR) per RECIST v1.1 in patients treated in 4th or 5th line who were HRDpos, platinum sensitive, and PARPi naïve. RESULTS: 463 patients were treated. Of these, 161 were platinum refractory, 151 platinum resistant, and 120 platinum sensitive to their last line of platinum therapy. Of note, 83 patients were primary platinum resistant or refractory (42 and 41, respectively). Median number of prior lines of therapy was 4 (range 2-16), and 27% of patients were treated in 6th or later line. 100% of patients received prior platinum, 99.6% received prior taxane, 81% received prior doxorubicin, 63% received prior bevacizumab, and 59% received prior gemcitabine. In the primary efficacy population (N=45), ORR was 29% (95% CI: 16-44, P=0.0003), disease control rate (DCR) was 71% (95% CI: 56-84), and duration of response (DOR) was 9.2 months (95% CI: 5.9-NE); overall survival (OS) was not reached (95% CI: 18.5-NE). The efficacy outcomes in all 456 patients with measurable disease at baseline treated in 4th or later line demonstrated a DCR of 49% with 2 complete responses (CR), 36 partial responses (PR), and 187 patients with stable disease (SD). The responses were durable with a median DOR of 9.4 months, a clinical benefit rate (CR+PR+SD) at 16 weeks of 29%, and median OS of 17.2 months (95% CI: 14.9-19.8) among all patients treated in 4th or later line. 266 (57.5%) patients had grade ≥3 treatment-related TEAEs. Grade ≥3 thrombocytopenia event (thrombocytopenia or decreased platelet count) by dose at onset of the event was 27.5% at 300 mg, 4.7% at 200 mg, and 2.7% at 100 mg. The other most common grade ≥3 TEAEs were anemia (26.3%) and neutropenia (8.2%). Hematologic TEAEs were most frequent in the first month and decreased in frequency and severity after dose reduction during months 2-3. CONCLUSIONS: Niraparib demonstrated durable anticancer activity in this heavily treated (≥3 lines) ROC population, including platinum-resistant or BRCAwt patients. Toxicities were consistent with previous niraparib studies. DISCLOSURE Funded by TESARO, Inc. Citation Format: Kathleen Moore, Angeles Secord, Melissa Geller, David Miller, Noelle Cloven, Gini Fleming, Andrea E. Wahner Hendrickson, Masoud Azodi, Paul Disilvestro, Amit Oza, Mihaela Cristea, Jonathan S. Berek, John Chan, Yong Li, Kaiming Sun, Katarina Luptakova, Ursula A. Matulonis, Bradley J. Monk. QUADRA: A PHASE 2, OPEN-LABEL, SINGLE-ARM STUDY TO EVALUATE SINGLE-AGENT NIRAPARIB TREATMENT IN PATIENTS WITH RELAPSED OVARIAN CANCER (ROC) WHO HAVE RECEIVED ≥3 PRIOR CHEMOTHERAPY REGIMENS [abstract]. In: Proceedings of the 12th Biennial Ovarian Cancer Research Symposium; Sep 13-15, 2018; Seattle, WA. Philadelphia (PA): AACR; Clin Cancer Res 2019;25(22 Suppl):Abstract nr NT-101.

There are no FDA approved screening tools for detecting ovarian cancer in the general population. The two best known ovarian cancer biomarkers, CA125 and HE4, are neither adequately sensitive nor specific when used alone to screen the general population for early stage disease. By using a combination of protein biomarkers for screening, it may be possible to increase the sensitivity and specificity over CA125 alone. In this study, we used Proseek Multiplex Oncology II plates to simultaneously measure the expression of 92 cancer-related proteins in serum using proximity extension assays. This technology combines the sensitivity of the polymerase chain reaction with the specificity of antibody-based detection methods, allowing multiplex biomarker detection and high throughput quantification. We analyzed one microliter of serum from 61 women with advanced stage high grade serous ovarian cancer and compared the values obtained to 88 age-matched healthy women. Principle component analysis and unsupervised hierarchical clustering separated the ovarian cancer patients from the healthy, with minimal misclassification. Data from the Proseek plates for CA125 levels exhibited a strong correlation with previously measured clinical values for CA125 (correlation coefficient of 0.91). CA125 and HE4 were detected at low levels in samples from healthy women, while higher levels were observed in the ovarian cancer cases. We identified 52 proteins that differed significantly (p < 0.006) between ovarian cancer and healthy samples; several of which are novel serum biomarkers for ovarian cancer. In total, 40 proteins had an estimated area under the ROC curve of 0.70 or greater. CA125 alone achieved a sensitivity of 93.4% at a specificity of 98%. However, by adding five proteins to CA125, we increased the assay sensitivity to 98.4%, while holding the specificity fixed at 98%. Our data demonstrate that the Proseek technology can replicate the results established by conventional clinical assays for known biomarkers, identify new candidate biomarkers, and improve the sensitivity and specificity of CA125 alone. Citation Format: Amy P.N. Skubitz, Kristin L.M. Boylan, Kate Geschwind, Qing Cao, Timothy K. Starr, Melissa A. Geller, Joseph Celestino, Robert C. Bast Jr., Karen H. Lu, and Joseph S. Koopmeiners. DEVELOPMENT OF A MULTI-PROTEIN CLASSIFIER FOR OVARIAN CANCER DETECTION BY SIMULTANEOUS MEASUREMENT OF 92 SERUM PROTEINS ON PROSEEK MULTIPLEX ONCOLOGY II PLATES [abstract]. In: Proceedings of the 12th Biennial Ovarian Cancer Research Symposium; Sep 13-15, 2018; Seattle, WA. Philadelphia (PA): AACR; Clin Cancer Res 2019;25(22 Suppl):Abstract nr AP08.

Despite advances in standard of care treatment, a significant number of ovarian cancer patients will exhibit disease recurrence, either due to the presence of chemoresistant cells or the reawakening of dormant tumor cells. Currently there are no methods to predict which patients are likely to recur, and the mechanisms that allow a unique subpopulation of cancer cells to enter dormancy and survive chemotherapy remain unclear. Since chemotherapy is more effective at destroying highly proliferative cells, it is theorized that tumor cells capable of entering a dormant state are also the same population able to evade initial treatment and ultimately engender chemoresistance. A silica gel encapsulation platform that uses physical confinement to inhibit proliferation has previously demonstrated the ability to distinguish between dormancy-capable and dormancy-incapable breast cancer cell lines. In this work, we sought to apply this platform to an ovarian cancer cell line and determine if it could be used to select for the dormancy-capable subset of cells and to detect differences upon enrichment of dormant or chemoresistant cell populations. Upon encapsulation in silica gels, surviving OVCAR-3 cells demonstrated hallmarks of dormancy, including reduction in Ki67 expression, modulation of key cell cycle regulatory genes and dormancy-associated signaling pathways, and ability to resume proliferation when the stress of physical confinement was removed. Moreover, enrichment of the dormant subpopulation using hypoxia treatment or serum starvation prior to silica gel encapsulation resulted in enhanced survival in gels relative to cells grown in standard culture conditions. These results suggest that silica gel encapsulation may be used as a simple method to isolate dormancy-capable cells for further applications such as gene or protein expression analysis and drug screening. Similarly, cells already in a dormant state demonstrated lower susceptibility to cisplatin drug treatment relative to actively proliferating cells, further supporting the idea that dormant cancer cells are less susceptible to chemotherapy and that chemoresistance may stem from a subpopulation of dormant cancer cells. After extraction from silica gels and subsequent return to a proliferative state, cells surviving prolonged encapsulation also demonstrated enhanced survival after cisplatin treatment relative to control populations. This suggests that the dormancy-capable cells selected for within silica gels are indeed more chemoresistant, even when no longer dormant, and that the ability to enter dormancy and chemoresistance in ovarian cancer are likely somehow connected. Lastly, when the drug-resistant phenotype was selected for through cisplatin drug treatment prior to encapsulation, a significant difference in viability was observed upon encapsulation between cisplatin-treated and untreated cells. Overall, these results indicate that the silica gel technology and the behavior of ovarian cancer cells after encapsulation may be an indicator of cells inherently capable of entering dormancy and less susceptible to platinum-based chemotherapies. In future applications, silica gel encapsulation of patient ovarian cancer cells could potentially be used as a predictive clinical tool to identify patients at risk of early recurrence or as a facile in vitro platform to investigate mechanistic links between dormancy, chemoresistance, and recurrence. Citation Format: Tiffany Lam, Hak Rae Lee, Melissa A. Geller, Alptekin Aksan, Samira Azarin. SILICA GEL ENCAPSULATION AS A POTENTIAL TOOL TO IDENTIFY DORMANT AND DRUG-RESISTANT CELL POPULATIONS IN OVARIAN CANCER [abstract]. In: Proceedings of the 12th Biennial Ovarian Cancer Research Symposium; Sep 13-15, 2018; Seattle, WA. Philadelphia (PA): AACR; Clin Cancer Res 2019;25(22 Suppl):Abstract nr DP-008.

Overcoming the immunosuppressive tumor microenvironment and localizing adoptive cell and checkpoint blockade therapies to solid tumors remain major impediments to successful cancer immunotherapy. Humans experience many viral infections. Once controlled, the host retains memory T-cells throughout the entire body to sense reinfection or recrudescence. Mouse models have demonstrated that when that same virus is reencountered, these T-cells sound an alarm that induces a local immunostimulatory environment that activates and recruits many arms of the immune system. We observed that, like healthy tissue, human tumors are commonly surveyed by memory T-cells specific for previously encountered viral infections. Antiviral T-cell immune surveillance of tumors was recapitulated in mouse cancer models. We tested and discovered that local delivery of adjuvant-free peptide, derived from previously encountered mouse viral infections, recapitulated the sensing and alarm T-cell function within the tumor: recruiting and activating both the innate and adaptive immune system. This approach induced intratumoral accumulation of granzyme B+ CD8+ T and NK cells, and activated dendritic cells within the tumor and subsequently within the tumor draining lymph node. In addition to stimulating the tumor microenvironment, preliminary data suggest activated antiviral T-cells directly killed peptide coated tumor cells. Viral peptide administration arrested rapidly growing and poorly immunogenic B16 melanomas in vivo and this treatment synergized with anti-PD-L1 checkpoint blockade to eliminate measurable tumors, and prevented recurrence in 34% of mice. Most cured mice rejected subsequent B16 tumor challenges at distant sites, indicating that effective systemic tumor-specific immunity was established. In support of the hypothesis that this approach could translate to human cancer immunotherapy, we found that viral peptide alarm therapy of freshly isolated human tumors drove similar immune activation to that observed in mice. This study demonstrates that natural and existing antiviral immunity can be repurposed to fight tumors without the need for adjuvant, vaccination, or personalized identification of immunogenic tumor neoantigens. Citation Format: Pamela C. Rosato, Sathi Wijeyesinghe, J. Michael Stolley, Christine Nelson, Rachel L. Davis, Luke S. Manlove, Christopher A. Pennell, Bruce R. Blazar, Clark C. Chen, Melissa A. Geller, Vaiva Vezys, David Masopust. Repurposing antiviral T-cells to fight tumors [abstract]. In: Proceedings of the Fourth CRI-CIMT-EATI-AACR International Cancer Immunotherapy Conference: Translating Science into Survival; Sept 30-Oct 3, 2018; New York, NY. Philadelphia (PA): AACR; Cancer Immunol Res 2019;7(2 Suppl):Abstract nr B182.

Technological advances allow genomic and transcriptomic analyses to be conducted at the single-cell level. Analysis of gene expression and DNA sequence at this detailed level could lead to prognostic and predictive biomarkers as well as an enhanced understanding of cancer cell and stromal subpopulations including stem cells and chemotherapy-resistant populations. This knowledge will improve our ability to implement a “precision medicine” approach to the treatment of women with ovarian cancer. We have initiated a project to prospectively study the transcriptome and DNA exome of freshly isolated, high-grade serous ovarian cancer solid tumor samples at the single-cell level. To date we have enrolled 8 patients and have initiated single-cell whole-exome DNA sequencing, using the Fluidigm C1 chip, and single-cell whole-transcriptome RNA sequencing, using the 10X Genomics platform. For comparison of RNA sequencing technologies, we previously performed RNA sequencing of a patient’s sample using the Fluidigm C1 chip. In addition to sequencing the primary tumor samples, we are concurrently attempting to establish patient-derived xenografts (PDXs) and cell lines for each patient. We have successfully generated PDX mice from the first patient and are currently treating a subset of the mice with carboplatin/paclitaxel. Patients will be followed prospectively and information on their clinical response will be incorporated into the analysis. Analysis of the RNA sequencing results from the first patient using the Fluidigm C1 platform identified several distinct populations of cells characterized by gene expression patterns that correlate with specific signaling pathways and disease states. We could clearly differentiate stromal cells from cancer epithelial cells based on gene expression patterns. Furthermore, using known functional markers, we could define subsets within each population, including three subgroups of cancer cells and four subgroups of stromal cells. Based on their gene expression patterns, we could determine the frequency of cancer epithelial cells compared to cancer cells undergoing epithelial-to-mesenchymal transition, as well as activated and nonactivated fibroblasts and myofibroblasts. We were also able to identify rare cells expressing stem cell markers. Using the 10X Genomics platform, we have completed RNA sequencing of five patients (~370 million reads/patient). We quantified gene expression on an average of 8,707 cells/patient and 1,723 genes/cell. Using graph-based clustering combined with the t-Distributed Stochastic Neighbor Embedding technique for high dimensionality reduction, we have identified approximately 9 to 15 subsets of cells within these cancer samples based on an unbiased analysis of their gene expression patterns. Using bioinformatic tools, including CellRanger, Seurat, Ingenuity Pathway Analysis, and Gene Set Enrichment Analysis, we can define several of these subsets based on known functional markers, including subsets of immune cells, stromal cells, and cancer epithelial cells. We can estimate the frequency of each subset and can also subdivide the groups using cell-type specific markers, for example, by distinguishing macrophages from dendritic cells within the immune subset. We will present unpublished data, including RNA sequencing and DNA exome sequencing of single cells on the first 8 enrolled patients. In addition to sequencing the primary tumor samples, we plan on performing single-cell sequencing of matched platinum-resistant tumors generated by treating the PDX mice with carboplatin and paclitaxel. We will also correlate presence and percentage of cell subpopulations with clinical outcomes of the patients. Our long-term goal is to use single-cell data as a prognostic biomarker for chemotherapy resistance as well as a tool for predicting effective therapeutic options, including targeted therapy and immune therapy. Citation Format: Boris J. Winterhoff, Christopher R. Clark, Sidharth Ramesh, Mihir Shetty, Locke Uppendahl, Amit Kumra Mitra, Attila Sebe, Martina Bazzaro, Melissa A. Geller, Juan E. Abrahante, Klein Molly, Raffaele Hellweg, Sally Mullany, Kenneth Beckman, Jerry Daniel, Timothy K. Starr. Single-cell sequencing as a prognostic and predictive tool for ovarian cancer therapy. [abstract]. In: Proceedings of the AACR Conference: Addressing Critical Questions in Ovarian Cancer Research and Treatment; Oct 1-4, 2017; Pittsburgh, PA. Philadelphia (PA): AACR; Clin Cancer Res 2018;24(15_Suppl):Abstract nr B59.

Currently, there are no FDA-approved screening tools for detecting ovarian cancer in the general population. The two most common ovarian cancer biomarkers, CA125 and HE4, when used individually are neither adequately sensitive nor specific enough to be used for screening. We hypothesize that by using a combination of biomarkers for screening, it will be possible to increase the sensitivity and specificity over CA125 alone. Development of a blood-based assay for ovarian cancer detection could significantly improve the survival of patients if it identified cancers earlier. In this study, we used Proseek Multiplex Oncology II plates to simultaneously measure the expression of 92 cancer-related proteins in serum using proximity extension assays. This technology combines the sensitivity of the polymerase chain reaction with the specificity of antibody-based detection methods, allowing multiplex biomarker detection and high-throughput quantification. In a review of the literature, we found that 44 of these 92 proteins had been studied individually and were present at elevated levels in ovarian cancer samples (e.g., serum or tissues). This study is the first to examine the levels of 92 proteins simultaneously using a large cohort of ovarian cancer serum samples. We analyzed one microliter of serum from each of 60 women with late stage (Stages III and IV) high-grade serous ovarian cancer and compared the values obtained to those from 241 age-matched healthy women. The data from these 301 serum samples were normalized across four Proseek Multiplex Oncology II 96-well plates and across the different sources of sera. Principal component analysis and unsupervised hierarchical clustering separated patients into the two major groups of ovarian cancer and healthy, with minimal misclassification. Data from the Proseek plates for CA125 levels exhibited a strong correlation with previously measured clinical values for CA125 (correlation coefficient of 0.90, with a 95% confidence interval of 0.88-0.92). CA125 and HE4 were detected at low levels in healthy samples, while higher levels were found in the ovarian cancer cases. Overall, we identified 45 proteins that showed a significant difference (p < 0.001) between ovarian cancer and healthy samples, many of which could serve as novel serum biomarkers for ovarian cancer. In total, 46 proteins had an estimated area under the ROC curve (AUC) of 0.7 or greater. CA125 alone achieved a sensitivity of 0.92 and a specificity of 0.98. In addition, we generated a multiprotein classifier using 10 proteins (including CA125) that increased the sensitivity to 0.97, while holding the specificity fixed at 0.98. This 5% increase in sensitivity will have a significant effect on the number of women correctly identified when screening a large population. Our data demonstrate that the Proseek technology can replicate the results established by conventional clinical assays for known biomarkers, identify new candidate biomarkers, and improve the sensitivity and specificity of CA125 alone. Additional studies using a larger cohort of early-stage ovarian cancer patients will allow for validation of these biomarkers and lead to the development of a screening tool for detecting early-stage ovarian cancer in the general population. Citation Format: Amy P.N. Skubitz, Kristin L.M. Boylan, Kate Geschwind, Qing Cao, Timothy K. Starr, Melissa A. Geller, Robert C. Bast, Jr., Karen H. Lu, Joseph S. Koopmeiners. Simultaneous measurement of 92 serum protein biomarkers for the development of a multiprotein classifier for ovarian cancer detection. [abstract]. In: Proceedings of the AACR Conference: Addressing Critical Questions in Ovarian Cancer Research and Treatment; Oct 1-4, 2017; Pittsburgh, PA. Philadelphia (PA): AACR; Clin Cancer Res 2018;24(15_Suppl):Abstract nr B20.

Natural Killer (NK) cells are components of innate immune response that are involved in tumor surveillance and clearance. Due to their ability to lyse tumor targets without prior priming they have gained some notoriety in the tumor immunotherapy setting. Although NK cells sense malignant transformation of cells through a number of stochastic receptors they posses no clonotypic receptors to drive specific recognition of tumor antigens, somewhat limiting their therapeutic applications. In order to address the antigen specificity issue we generated molecules, termed Tri-Specific Killer Engagers (TriKEs), that contain three arms: one arm contains a single chain variable fragment (scFv) that binds a potent NK cell activating receptor (CD16); one arm contains an scFv that binds a tumor antigen; and one arm contains the cytokine IL-15 to drive NK cell priming, expansion and survival. Initial pre-clinical studies with these first generation molecules showed that we could use them to target hematological malignancies (via a CD33 binding arm) both in vitro and in vivo or solid malignancies (via an EpCAM binding arm) in vitro. However, to improve on this platform we altered the molecule by swapping out the anti-CD16 scFv arm with a humanized camelid single domain antibody against CD16 (termed cam16 henceforth). This second generation cam16 TriKE platform potently enhanced NK cell activation against tumor targets and induced far greater NK cell expansion when compared to the first generation TriKE. Using our bacterial expression system we generated and isolated cam16 TriKEs against a number of different tumor antigens (CD33, EpCAM, Mesothelin and ROR1) to display the ability of the second generation TriKE platform to target a variety of cancers. To test these molecules we used flow cytometry based NK cell activation assays, where NK cells co-cultured with tumor targets +/- cam16 TriKEs are evaluated for NK cell degranulation (by expression of surface CD107a) and NK cell inflammatory cytokine production (by expression of intracellular IFNγ). We also used a real-time imaging platform, the IncuCyte Zoom, to evaluate the ability of NK cells to kill tumor targets in the presence or absence of cam16 TriKEs over a 2-3 day period. Our data indicates that cam16 TriKEs drive NK cell specific degranulation, inflammation, and killing of AML (CD33), ovarian cancer (EpCAM and Mesothelin), lung cancer (Mesothelin and EpCAM) and sarcoma (ROR1) tumor targets in vitro. Furthermore cam16 TriKE in vivo efficacy could be seen in humanized mouse models using both luciferase-labeled cell lines and patient derived xenografts. Taken together, these findings clearly highlight the translational potential of cam16 TriKEs in the tumor immunotherapy setting. Citation Format: Martin Felices, Behiye Kodal, Alexander J. Lenvik, Brianna Ettestad, Laura Bendzick, Caitlin Ryan, Dawn K. Schirm, Ron McElmurry, Craig E. Eckfeldt, Melissa A. Geller, Todd R. Lenvik, Daniel A. Vallera, Jeffrey S. Miller. Driving natural killer cell antigen-specific targeting of cancer via next generation trivalent molecules [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2018; 2018 Apr 14-18; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2018;78(13 Suppl):Abstract nr 4939.

We performed RNA sequencing of single cells derived from a high grade serous ovarian cancer (HGSOC) specimen to determine the extent of heterogeneity and to determine if it was feasible to identify cancer stem cells or gene expression signatures of chemo resistance. To perform RNA sequencing we enzymatically digested a fresh specimen from an HGSOC derived from the ovary. Immune cells were depleted by flow cytometry and single cell sequencing was performed using the Fluidigm C1 chip in tandem with Illumina HiSeq 2500 sequencing. Multiple bioinformatics tools were used to identify subgroups and activated pathways. Immunohistochemistry was performed on an adjacent tumor section to analyze markers of epithelium, stroma and stem cells. We found that gene expression patterns in single cells could be used to separate cells into stroma-like and epithelial-like groups. Gene set enrichment analysis identified proliferative genesets (oxidative phosphorylation and MYC targets) associated with the epithelial-like cells while epithelial-to-mesenchymal-transition (EMT) genes associated with the stroma-like cells. Neither group was significantly associated with genesets derived from chemo-resistant cells. Using known marker analysis, we could identify a small percentage of cells that expressed ovarian cancer stem cell markers and we could group cells into functional categories. Using four molecular subtypes established from large-scale bulk sequencing studies we show that single cells from a single patient are heterogeneous and each molecular subtype is represented. In conclusion, we show the feasibility of performing single cell sequencing on an epithelial ovarian cancer and reveal a heterogeneous population of cells. Expanding these findings to a larger cohort of patients could allow for identification of targetable sub-populations of cells that were previously undetectable in studies that use bulk samples to interrogate the transcriptome and genome of ovarian cancer patients. Citation Format: Timothy K. Starr, Boris Winterhoff, Makayla Maile, Kenneth Beckman, Jerry Daniel, Melissa Geller, Martina Bazzaro, Molly Klein, Raffaele Hellweg, Juan Abrahante, Amit K. Mitra, Atilla Sebe, Sally A. Mullany. Single cell sequencing of high grade serous ovarian cancer [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2017; 2017 Apr 1-5; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2017;77(13 Suppl):Abstract nr 5402. doi:10.1158/1538-7445.AM2017-5402

PURPOSE. Early detection is the key to increased survival of ovarian cancer patients, however, a screening tool that is adequately sensitive and specific for use in the general population has yet to be developed. CA125 and HE4 are two biomarkers that are currently approved by the FDA for monitoring recurrence of ovarian cancer. We hypothesize that by simultaneously screening dozens of cancer–related biomarkers for their levels of expression in serum samples in a standardized multiplex assay, we will be able to develop an algorithm to screen for early stages of ovarian cancer. EXPERIMENTAL PROCEDURES. In this project, we used a multiplex approach to identify candidate biomarkers for early stages of ovarian cancer. We used a Proseek® Multiplex Oncology I plate (Olink Bioscience) to analyze the expression of 92 biomarkers (including CA125 and HE4) in one microliter of serum collected from 22 women in each of four groups: healthy, benign ovarian tumors, early stage I/II serous ovarian cancer, and late stage III/IV serous ovarian cancer. Biomarker levels were simultaneously quantified based on the Proximity Extension Assay. This innovative technology combines the sensitivity of the polymerase chain reaction with the specificity of antibody–based detection methods, allowing multiplex biomarker detection and high throughput quantification. SUMMARY OF THE DATA. We found that CA125 and HE4 were present at very low levels in healthy controls and the benign cases, while higher levels were found in the early stage cases, and the highest levels of CA125 and HE4 were found in the late stage cases. Overall, statistically significant trends were observed for 37 of the 92 markers (i.e. p < 0.001), many of which have yet to be documented as candidate serum biomarkers for ovarian cancer. The CA125 data from the Proseek® plate exhibited a strong correlation with the CA125 values obtained by ELISA which were abstracted from the patients' medical records, with a Pearson's correlation coefficient of 0.84 (95% CI: 0.76, 0.89). In spite of the small sample size (22 cases each for early stage ovarian cancer vs. healthy), the area under the ROC curve (AUC) for CA125 was determined to be 0.98 (95% CI: 0.95, 1.0) and the AUC for HE4 was determined to be 0.84 (95% CI: 0.71, 0.95). In total, 22 biomarkers had an estimated AUC greater than 0.7, when comparing early stage ovarian cancer vs. healthy, and were significantly associated with cancer status, thus motivating further investigation. CONCLUSIONS. These data provide proof–of–principle that the Proseek® technology can replicate the results of conventional ELISAs for these known oncology biomarkers. In future studies, we plan to use a larger sample size in order to refine our model and validate our results. Our ultimate goal is to develop an algorithm of biomarkers that can be used to screen women for early stages of ovarian cancer, when the likelihood of long term survival is greatest. Citation Format: Kristin L.M. Boylan, Kate Geschwind, Joseph S. Koopmeiners, Melissa A. Geller, and Amy P.N. Skubitz. A MULTIPLEX PLATFORM FOR THE IDENTIFICATION OF SERUM BIOMARKERS FOR THE EARLY DETECTION OF OVARIAN CANCER [abstract]. In: Proceedings of the 11th Biennial Ovarian Cancer Research Symposium; Sep 12-13, 2016; Seattle, WA. Philadelphia (PA): AACR; Clin Cancer Res 2017;23(11 Suppl):Abstract nr DPOC-011.

Early detection is the key to increased survival for women with ovarian cancer, yet a screening tool has yet to be developed that is adequately sensitive and specific enough for use in the general population. A need also exists for a reliable diagnostic method for women who present with an abdominal mass. In contrast, screening for cervical cancer by Pap tests has been routinely performed for over 50 years. In the liquid-based Pap test, cells are collected from the cervical os and placed into an alcohol-based fixative and then examined for abnormal cells. Since ovarian cancer cells have been observed in Pap tests, we reasoned that ovarian cancer peptide biomarkers may also be present. Our central hypothesis is that proteins shed by ovarian cancer cells can be detected in the Pap test fixative by Mass Spectrometry (MS)-based proteomics. The Pap test fixative is ideal for biomarker discovery since it is derived from a site near the ovarian cancer (i.e. proteins may be secreted or shed from the tumor and flow through the fallopian tube into the uterus and out the cervical opening). Recently, the fimbria of the fallopian tube have been suggested to be the true precursor to ovarian cancer, strengthening our hypothesis that ovarian cancer proteins will be found in the lower genital tract, perhaps even at early stages. To demonstrate the feasibility of this approach, we previously examined the proteins present in residual Pap test fixative samples from women with normal cervical cytology by MS, and described 152 proteins in the “Normal Pap Proteome.” (Boylan et al., Clinical Proteomics, 2014). The objective of this study is to discover ovarian cancer biomarkers in Pap test fixative samples by MS-based proteomic techniques. We used isobaric tags to quantify the proteins in liquid Pap test samples from women with ovarian cancer compared to women with benign gynecological disease and healthy women. Proteins were concentrated from “Mock” Pap tests, and samples were trypsin digested utilizing the STrap technique to increase the likelihood that proteins would be recovered. We then performed Tandem Mass Tag™ (TMT) 10-plex labeling of Pap test fixative samples from post-menopausal women: 3 with ovarian cancer, 3 with benign gynecological conditions, 3 healthy, and a Reference Control consisting of pooled samples. The 10 labeled samples were combined and run on 2D LC-MS/MS, followed by bioinformatics integration to discover proteins in Pap test fixatives that are unique and/or differentially expressed in the samples. We identified >3800 proteins total, and 22 proteins that were expressed at least 2-fold higher in all three ovarian cancer samples compared to the controls. We conclude that quantification of proteins from Pap test samples will prove a rich source of biomarkers for ovarian cancer detection. Citation Format: Amy P.N. Skubitz, Somi Afiuni, Kristin L.M. Boylan, Melissa Geller, Peter Argenta, Samantha Hoffman, Timothy Griffin. Tandem Mass Tag 10-plex isobaric labeling of Pap test proteins: A novel method for the identification of ovarian cancer protein biomarkers by mass spectrometry. [abstract]. In: Proceedings of the AACR Special Conference on Advances in Ovarian Cancer Research: Exploiting Vulnerabilities; Oct 17-20, 2015; Orlando, FL. Philadelphia (PA): AACR; Clin Cancer Res 2016;22(2 Suppl):Abstract nr B34.

Chimeric Antigen Receptors (CARs) are used to direct and activate cytotoxic lymphocytes against virally infected or malignant cells. They have been heavily investigated in cytotoxic T cells, and, to a much lesser extent, in natural killer (NK) cells with promising results, especially for anti-CD19 CARs to mediate killing of B cell leukemias. To date, CARs that have been utilized in NK cells have primarily employed the same intracellular signaling components as T cell CARs. However, this may not be the optimal approach for CAR-expressing NK cells due to differences in key signaling pathways. To improve the activity of NK cell-CARs targeted against solid tumors, the intracellular signaling domains of an anti-mesothelin CAR were modified to express NK cell specific signaling molecules and evaluated for activity in NK-92 cells against ovarian cancer cell lines. CARs were assembled using gBlocks encoding various transmembrane, co-activating, and signaling domains. NK cell specific CARs were then transfected into NK-92 cells, a human NK cell tumor line, using Sleeping Beauty. Positive cells were selected for using a drug resistance marker and CAR surface expression was evaluated using flow cytometry. CAR expressing NK-92 cells were then evaluated for degranulation, as indicated by CD107a production, and interferon (INF)-γ release when challenged with mesothelin negative, MA148, and mesothelin positive, A1847, ovarian cell lines. In addition, beads conjugated to mesothelin were used to measure NK cell activation via the CAR alone. In total 10 novel CARs were assembled and transfected into NK-92 cells, 6 of which were properly expressed on the cell surface. All 6 properly expressed CARs enhanced both CD107a and INF-γ production. When CAR activity alone was evaluated using mesothelin conjugated protein A beads, non-transfected NK-92 cells or cells that did not properly express the CAR all had Citation Format: David L. Hermanson, Branden Moriarity, Trevor Argall, Melissa A. Geller, David Largaespada, Dan S. Kaufman. Design and evaluation of novel natural killer cell chimeric antigen receptors. [abstract]. In: Proceedings of the CRI-CIMT-EATI-AACR Inaugural International Cancer Immunotherapy Conference: Translating Science into Survival; September 16-19, 2015; New York, NY. Philadelphia (PA): AACR; Cancer Immunol Res 2016;4(1 Suppl):Abstract nr A175.

Biomarkers that can be used to screen women in the general population for ovarian cancer have yet to be identified. In contrast, screening for cervical cancer by Pap tests has been routinely performed for over 50 years. The liquid-based Pap test consists of collecting cervical cells from the external cervical os and placing them into an alcohol-based fixative. The cells are stained and examined by a pathologist to identify the presence of premalignant and malignant cells. It has been reported that ovarian cancer cells can be identified in the liquid-based Pap test fluid. We hypothesize that proteins shed by ovarian cancer cells can be detected in the Pap test fluid by mass spectrometry (MS)-based proteomic techniques. This biospecimen source is ideal for the discovery phase of biomarker development, since it is readily available at the time of a Pap test, derived from a site near the ovarian cancer, and may not contain the high abundance proteins found in blood that could mask a potential biomarker. In this study, we developed a sample preparation and MS methodology to document the protein profile in residual Pap test fluid. The residual fluid from discarded Pap tests from women over the age of 55 with normal cytology was used to optimize the methodology for sample preparation prior to analysis by MS. The cells were removed by centrifugation and the protein concentration of the cell-free supernatant was determined. The protein composition of individual samples was visualized by separating the proteins on SDS-PAGE gels followed by silver staining. The presence of serum proteins in the residual Pap test fluid was examined by western immunoblot using an antibody against human serum albumin. In order to determine the number of peptides (and proteins) present in the Pap test fluid, samples were precipitated with acetone. The proteins were trypsinized and the resulting peptides were fractionated followed by analysis with tandem MS. The average volume of the residual Pap tests was 1.5 ml and the average protein concentration was 160 ug/ml. By western blot, we showed that the amount of albumin in each sample was significantly reduced compared to normal serum. The number of peptides and proteins present in residual Pap fluid was determined by searching the MS data against the Human Unipro database. Over 300 proteins were identified in samples pooled from 40 patients with normal cytology. When the proteins were classified according to cellular localization and biological function, we found over half of the proteins identified were extracellular or plasma membrane proteins. The largest category of biological function identified was for proteins involved in immunity and defense; including several mucin proteins, such as MUC16 (CA125) which are characteristic of the healthy cervico-vaginal proteome. Approximately 10% of the proteins identified were components of the blood. Using a modification of this method, we have also analyzed the proteins from individual residual Pap test samples by tandem MS, and were able to identify over 1,000 proteins. From these preliminary studies, we conclude that residual Pap test fluid contains a sufficient amount of protein for analysis by MS. Studies to perform MS-based proteomic analysis of proteins from residual Pap test fluid of ovarian cancer patients are in progress. Citation Format: Kristin L.M. Boylan, Melissa A. Geller, Timothy J. Griffin, Stefan E. Pambuccian, Somi Afiuni, Amy P.N. Skubitz. Proteomic analysis of residual Pap test fluid: Novel source for detection of ovarian cancer biomarkers. [abstract]. In: Proceedings of the AACR Special Conference on Advances in Ovarian Cancer Research: From Concept to Clinic; Sep 18-21, 2013; Miami, FL. Philadelphia (PA): AACR; Clin Cancer Res 2013;19(19 Suppl):Abstract nr A21.