Search

Validation of predictive response biomarkers

Structures of compounds used in the study

Workflow for filtering cell line genomic data

Additional reproducibility data

Background In epithelial ovarian cancer (EOC), 15-20% of the tumors do not respond to first-line chemotherapy (paclitaxel with platinum-based therapy), and in recurrences this number increases. Our aim is to determine the feasibility of cell proliferation assays of tumor cells isolated from malignant ascites to predict in vitro chemotherapy sensitivity, and to correlate these results with clinical outcome. Materials and methods Ascites was collected from twenty women with advanced EOC. Cell samples were enriched for tumor cells and EOC origin was confirmed by intracellular staining of CK7, surface staining of CA125 and EpCAM, and HE4 gene expression. In vitro sensitivity to chemotherapy was determined in cell proliferation assays using intracellular ATP content as an indirect measure of cell number. In vitro drug response was quantified by calculation of the drug concentration at which cell growth was inhibited with 50%. Clinical outcome was determined using post-treatment CA125 level. Results Cell samples of twenty patients were collected, of which three samples that failed to proliferate were excluded in the analysis (15%). Three other samples were excluded, because clinical outcome could not be determined correctly. In twelve of the fourteen remaining cases (86%) in vitro drug sensitivity and clinical outcome corresponded, while in two samples (14%) there was no correspondence. Conclusions Our study demonstrates the feasibility of drug sensitivity tests using tumor cells isolated from ascites of advanced EOC patients. Larger observational studies are required to confirm the correlation between the in vitro sensitivity and clinical outcome.

Kinase inhibitors form the largest class of precision medicine. From 2013 to 2017, 17 have been approved, with 8 different mechanisms. We present a comprehensive profiling study of all 17 inhibitors on a biochemical assay panel of 280 kinases and proliferation assays of 108 cancer cell lines. Drug responses of the cell lines were related to the presence of frequently recurring point mutations, insertions, deletions, and amplifications in 15 well-known oncogenes and tumor-suppressor genes. In addition, drug responses were correlated with basal gene expression levels with a focus on 383 clinically actionable genes. Cell lines harboring actionable mutations defined in the FDA labels, such as mutant BRAF(V600E) for cobimetinib, or ALK gene translocation for ALK inhibitors, are generally 10 times more sensitive compared with wild-type cell lines. This sensitivity window is more narrow for markers that failed to meet endpoints in clinical trials, for instance CDKN2A loss for CDK4/6 inhibitors (2.7-fold) and KRAS mutation for cobimetinib (2.3-fold). Our data underscore the rationale of a number of recently opened clinical trials, such as ibrutinib in ERBB2- or ERBB4-expressing cancers. We propose and validate new response biomarkers, such as mutation in FBXW7 or SMAD4 for EGFR and HER2 inhibitors, ETV4 and ETV5 expression for MEK inhibitors, and JAK3 expression for ALK inhibitors. Potentially, these new markers could be combined to improve response rates. This comprehensive overview of biochemical and cellular selectivities of approved kinase inhibitor drugs provides a rich resource for drug repurposing, basket trial design, and basic cancer research.

// Marion A.A. Libouban 1, 2 , Jeroen A.D.M. de Roos 1 , Joost C.M. Uitdehaag 1 , Nicole Willemsen-Seegers 1 , Sara Mainardi 2 , Jelle Dylus 1 , Jos de Man 1 , Bastiaan Tops 3 , Jules P.P. Meijerink 4 , Zuzana Storchova 5 , Rogier C. Buijsman 1 , Rene H. Medema 2 , Guido J.R. Zaman 1 1 Netherlands Translational Research Center B.V., Oss, The Netherlands 2 Netherlands Cancer Institute, Amsterdam, The Netherlands 3 Department of Pathology, Radboud University Medical Center, Nijmegen, The Netherlands 4 Princess Maxima Center for Pediatric Oncology, Utrecht, The Netherlands 5 University of Kaiserslautern, Kaiserslautern, Germany Correspondence to: Guido J.R. Zaman, email: guido.zaman@ntrc.nl Keywords: TTK, Mps1, kinase inhibitor, chromosome instability, aneuploidy Received: November 16, 2016 Accepted: March 03, 2017 Published: March 15, 2017 ABSTRACT Inhibition of the spindle assembly checkpoint kinase TTK causes chromosome mis-segregation and tumor cell death. However, high levels of TTK correlate with chromosomal instability (CIN), which can lead to aneuploidy. We show that treatment of tumor cells with the selective small molecule TTK inhibitor NTRC 0066-0 overrides the mitotic checkpoint, irrespective of cell line sensitivity. In stable aneuploid cells NTRC 0066-0 induced acute CIN, whereas in cells with high levels of pre-existing CIN there was only a small additional fraction of cells mis-segregating their chromosomes. In proliferation assays stable aneuploid cells were more sensitive than cell lines with pre-existing CIN. Tetraploids are thought to be an intermediate between diploid and unstable aneuploid cells. TTK inhibitors had the same potency on post-tetraploid and parental diploid cells, which is remarkable because the post-tetraploids are more resistant to mitotic drugs. Finally, we confirm that the reference compound reversine is a TTK inhibitor and like NTRC 0066-0, inhibits the proliferation of patient-derived colorectal cancer organoids. In contrast, treatment with TTK inhibitor did not reduce the viability of non-proliferating T cell acute lymphoblastic leukemia cells samples. Consequently, TTK inhibitor therapy is expected to spare non-dividing cells, and may be used to target stable aneuploid tumors.

Introduction Intensive multi-agent treatment has boosted survival up to 80% of pediatric T-cell acute lymphoblastic leukemia (T-ALL) patients. Nevertheless, relapsed patients have a poor prognosis due to acquired therapy resistance while most survivors have detrimental chemotherapy-induced side effects. Therefore, novel targeted therapies are urgently needed since further intensification of the current standard treatment regimen is not feasible for refractory/relapsed cases. Protein kinase inhibitors (PKIs) are amongst the most successful cancer treatments. Targetable kinases activated by gene fusions are rare in T-ALL and include subclonal NUP214-ABL1 fusion in 6% of cases or other rare clonal ABL1 fusions. Nevertheless, leukemic blasts rely on enhanced kinase signaling to sustain their dysregulated proliferation. Protein kinases can be hyper-activated even in the absence of defects in their genes. Thus, together with the identification of genomic aberrations, phospho-proteomics can provide information on pathway activation, signaling networks and aberrant kinase activities that offer important opportunities for targeted therapies. Aim Here, we aimed to identify and quantify kinase activation in T-ALL cell lines that may yield differential sensitivity to PKIs in vitro. This approach could pinpoint targetable leukemia vulnerabilities and provide effective (combination) treatment strategies. Methods Protein extracts from 11 T-ALL cell lines were enriched for phospho-peptides by titanium dioxide enrichment and anti-phospho tyrosine immunoprecipitation followed by liquid chromatography - tandem mass spectrometry (MS). Subsequently, the Integrative Inferred Kinase Activity (INKA) pipeline was used to rank activated kinases in our panel (Beekhof et al., 2019). Based on these data, selected kinase inhibitors were tested in vitro as single treatment or in combinations. Eventually, drugs of interest were further tested ex vivo in a cohort of T-ALL patient-derived xenografts (PDXs). Results MS-based phospho-proteome profiling of 11 T-ALL cell lines identified about 3700 tyrosine phospho-sites and more than 13300 serine/threonine phosphorylation sites. We found SRC-family kinases including LCK, SRC, FYN, and YES1 as most activated kinases in many T-ALL cell lines while ABL1, ZAP70, LYN, and FGR were detected only in specific lines. Additionally, other kinases including CDK1/2 and PAK1/2 were found to be activated in all the cell lines while activation of the INSR/IGF-1R axis was detected only in a subset of lines. We then tested cellular response to multiple clinically relevant PKIs based on predicted kinase activities. The in vitro drug screening showed an effective response and G1-arrest following treatment with the CDK1/2 inhibitor milciclib in all the cell lines tested, with IC50 values between 10nM and 1uM. Despite a general SRC-family kinases activation profile, dual SRC/ABL inhibitors like dasatinib reduced cellular viability only in cell lines with ABL1 fusions or LCK translocations (IC50 < 10nM) while other lines lacking ABL or LCK rearrangements were affected at much higher drug concentration (IC50 > 3uM, which is beyond the clinical achievable plasma concentration). Thus, PKIs were tested in combination with other relevant inhibitors based on additional kinase activities detected. Interestingly, the concomitant inhibition of the SRC-family kinases by dasatinib and the INSR/IGF-1R axis by BMS-754807 led to a drastic reduction of cell viability at nanomolar concentrations even in cell lines that did not respond to dasatinib, identifying a novel possible effective combination strategy for T-ALL. Eventually, we tested clinically relevant PKIs in 50 PDXs ex vivo and identified various T-ALL samples with a high sensitivity to dasatinib single treatment (IC50 < 100nM) as previously reported by others (Frismantas et al., 2017). Moreover, 70% of our PDXs efficiently responded to the broad spectrum kinase inhibitor midostaurin (IC50 < 1uM), highlighting the importance of targeting multiple signaling nodes simultaneously to tackle T-ALL vulnerabilities. Conclusions Ranking kinase activities and signaling networks from phospho-proteomic data can guide the use of PKIs as treatment option for T-ALL patients. Moreover, kinase activity profiling can provide insights for efficient treatment combination strategies to develop personalized medicine. Disclosures No relevant conflicts of interest to declare.

Kinases are the major anticancer drug target class of the 21st century with nearly sixty small molecule kinase inhibitors approved for clinical use in the first two decades. While there are more than 500 kinases encoded by the human genome, currently approved inhibitors act primarily through approximately twenty different targets. Key to the success of kinase inhibitor therapy has been the simultaneous development of biomarker assays to enable the selection of patients most likely to respond. Predictive drug response biomarkers can be identified with cancer cell panel profiling, which is the parallel testing of compounds on a large panel of cancer cell lines. By correlating drug sensitivity with genomic information of the cell lines, strategies for patient stratification or drug repurposing have been developed [1-3]. In two earlier studies [2,3], we compared the kinase selectivity and the cellular inhibition profiles of all kinase inhibitors approved by the FDA until May 2018. Here, we will present the cancer cell panel profiling of all twenty small molecule kinase inhibitors approved since then. Several of these inhibitors act through well-known targets, such as ALK (lorlatinib), BRAF (encorafenib), EGFR (dacomitinib) and MEK1 (binimetinib, selumetinib). Others act through kinases for which no small molecule inhibitors have been approved before, such as CSF1R (pexidartinib), FGFR (erdafitinib, pemigatinib), c-MET (capmatinib), RET (selpercatinib, pralsetinib) and TRK (larotrectinib, entrectinib). All compounds were profiled on a panel of 102 cancer cell lines, known as Oncolines, representing a wide range of solid tumors and hematological malignancies, and harboring mutant and wild-type versions of many major cancer driver genes [3]. To determine similarities in the mechanisms underlying the anti-proliferative activity of the inhibitors, their IC50 fingerprint in the cell proliferation assays were compared by hierarchical clustering [4]. Compounds that act through the same primary kinase clustered together in this analysis. Exceptions were investigated further by profiling of additional cell lines, representing cancer gene alterations that were not present in the 102 cell line panel, such as FLT3 mutation and TRK-gene fusions, which occur in relatively small cancer patient populations. To compare the genomic targeting of kinase inhibitors acting on the same biochemical target, the cancer cell lines were classified as either “mutant” or “wild-type” for specific cancer gene alterations and were grouped per cancer gene. The relationship between drug sensitivity and cancer gene mutation status was examined. Detailed genomic biomarker analyses and comparative profiling results of novel BTK, EGFR, FGFR, FLT3 and TRK kinase inhibitors will be presented.[1] Mol Cancer Ther 2017;26:2609-17; [2] PLoS ONE 2014;9:e92146; [3] Mol Cancer Ther 2019;18:470-81; [4] Mol Cancer Ther 2016;15:3097-109 Citation Format: Jeffrey J. Kooijman, Wilhelmina E. van Riel, Martine B. Prinsen, Jelle Dylus, Jeroen A. de Roos, Yvonne Grobben, Nicole Willemsen-Seegers, Michelle Muller, Jos de Man, Yugo Narumi, Yusuke Kawase, Rogier C. Buijsman, Suzanne J. van Gerwen, Guido J. Zaman. Comparative cancer cell panel profiling of kinase inhibitors approved for clinical use from 2018 to 2020 [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2021; 2021 Apr 10-15 and May 17-21. Philadelphia (PA): AACR; Cancer Res 2021;81(13_Suppl):Abstract nr 1480.

Cancer cell line panels are important tools to characterize the in vitro activity of new investigational drugs. Here, we present the inhibition profiles of 122 anticancer agents in proliferation assays with 44 or 66 genetically characterized cancer cell lines from diverse tumor tissues (Oncolines). The library includes 29 cytotoxics, 68 kinase inhibitors, and 11 epigenetic modulators. For 38 compounds this is the first comparative profiling in a cell line panel. By strictly maintaining optimized assay protocols, biological variation was kept to a minimum. Replicate profiles of 16 agents over three years show a high average Pearson correlation of 0.8 using IC50 values and 0.9 using GI50 values. Good correlations were observed with other panels. Curve fitting appears a large source of variation. Hierarchical clustering revealed 44 basic clusters, of which 26 contain compounds with common mechanisms of action, of which 9 were not reported before, including TTK, BET and two clusters of EZH2 inhibitors. To investigate unexpected clusterings, sets of BTK, Aurora and PI3K inhibitors were profiled in biochemical enzyme activity assays and surface plasmon resonance binding assays. The BTK inhibitor ibrutinib clusters with EGFR inhibitors, because it cross-reacts with EGFR. Aurora kinase inhibitors separate into two clusters, related to Aurora A or pan-Aurora selectivity. Similarly, 12 inhibitors in the PI3K/AKT/mTOR pathway separated into different clusters, reflecting biochemical selectivity (pan-PI3K, PI3Kβγδ-isoform selective or mTOR-selective). Of these, only allosteric mTOR inhibitors preferentially targeted PTEN-mutated cell lines. This shows that cell line profiling is an excellent tool for the unbiased classification of antiproliferative compounds. Mol Cancer Ther; 15(12); 3097–109. ©2016 AACR.

Introduction Poor prednisolone response in the induction phase of treatment is a risk-stratification marker in the current Dutch DCOG ALL-11 treatment protocol. Physiological or mutational activation of the IL7-receptor (IL7R) provokes steroid resistance in T-ALL. Moreover, relapsed T-ALL patients that harbor IL7R or RAS mutations are considered as an ultra-high-risk group with extremely poor outcome. Aim We studied the contribution of two key IL7R downstream pathways (STAT5B and MAPK-ERK respectively) to steroid resistance in order to optimize future treatment regimens. Results To date, IL7-dependent steroid resistance has been proposed to depend on IL7-induced JAK-STAT signaling and subsequent BCL2 upregulation. Here, we demonstrate that the transforming STATBN642H mutation does not impair steroid sensitivity despite a paradoxical upregulation of BCL2 and BCLXL following steroid treatment. We therefore hypothesized that other pathways contribute to IL7-dependent steroid resistance in T-ALL. By studying T-ALL PDX samples, we observed that physiological IL7-signaling can activate the MAPK-ERK pathway in IL7-dependent and IL7-independent steroid resistant PDX cells. Similar to mutational-driven MAPK-ERK signaling (e.g. by activating IL7R, JAK1 or NRAS mutations), IL7-induced MAPK-ERK signaling leads to the inactivation of the pro-apoptotic protein BIM by inhibitory phosphorylation of both BIM-EL and BIM-L isoforms. By mass spectrometry, we observe that phosphorylated BIM specifically decreases its physical interaction with Aurora kinase A and Bcl-2 family members BMF, BCL2, BCLXL and MCL1, which alters the apoptotic threshold and effectuates steroid resistance. The JAK1/2-inhibitor ruxolitinib and MEK-inhibitor selumetinib are considered for future basket-treatment protocols in IL7R/JAK-STAT-activated or MAPK-ERK-activated leukemias respectively. We demonstrate that both ruxolitinib and selumetinib prevent IL7-induced MAPK-ERK signaling in T-ALL PDX cells. Moreover, ruxolitinib and selumetinib abolish BIM phosphorylation in mutant JAK1-overexpressing SUPT-1 cells, which restores the binding of BIM to BCL2, BCLXL and MCL1. Only selumetinib restores the function of BIM in wild type or mutant NRAS overexpressing SUPT-1 cells, indicating that ruxolitinib only blocks MAPK-ERK signaling in the context of 'upstream' (e.g. mutant IL7R/JAK1 or IL7-induced) pathway activation. Surprisingly, none of our 46 PDX samples respond to ruxolitinib treatment in the absence of IL7, while the majority of these samples demonstrate a robust response towards MEK-inhibition. IL7-dependent steroid resistant PDX samples do respond to ruxolitinib treatment in the presence of IL7, while IL7 exposed IL7-independent steroid resistant PDX samples remain unresponsive to ruxolitinib. We therefore conclude that the therapeutic effect of ruxolitinib is dependent on an IL7-protective effect, limiting its therapeutic application to a selective subset of T-ALL patients. When we combine prednisolone treatment with selumetinib or ruxolitinib, we observe that ruxolitinib only synergizes with prednisolone in IL7-dependent steroid resistant PDX samples in the presence of IL7. Importantly, combination treatment with prednisolone and selumetinib acts highly synergistic in IL7-dependent and IL7-independent steroid resistant samples in the presence and absence of IL7. Conclusion and clinical relevance In addition to described JAK-STAT and PI3K-AKT signaling, our study demonstrates that IL7-induced signaling activates the MAPK-ERK signaling pathway, which actively contributes to IL7-dependent steroid resistance. We demonstrate the central importance of MAPK-ERK signaling in T-ALL, whereas all T-ALL PDX samples tested highly benefit from combined prednisolone and selumetinib treatment. Ruxolitinib only acts synergistically with steroids in the context of IL7-dependent steroid resistance. Moreover, our data proposes that this synergistic effect may (in part) depend on the anti-MAPK-ERK effect downstream of JAK1/2-inhibition. Combined, our study strongly supports the enrollment of T-ALL patients in the current phase I/II SeluDex trial (NCT03705507), and contributes to the optimization and stratification of newly designed T-ALL treatment regimens. Disclosures Bourquin: Servier: Other: Travel Support. Vormoor:AstraZeneca: Research Funding.

In epithelial ovarian cancer (EOC), 15-20% of tumors do not respond to first-line chemotherapy, consisting of platinum-based therapy plus paclitaxel, and in recurrences the response rate is even lower. PARP inhibitors synergize with platinum therapy and have been approved for platinum sensitive EOC. Clinical trials with immunotherapies, such as PD-1/PD-L1 blockade, have so far not been successful. There is a great need of novel therapies with improved long term treatment outcome, as well as diagnostic tools and biomarkers to predict chemotherapy response in the clinic. We have used malignant ascites collected from patients with EOC, to identify new biomarkers of chemotherapy response and to support small molecule drug discovery programs on the tryptophan metabolizing enzymes TDO and IDO1 with clinically relevant assays. Modulation of TDO or IDO1 activity with small molecule inhibitors decreases anti-tumor immune response by inducing T cell anergy. Ascites was gathered from eighteen patients with advanced stage EOC. Cells were isolated by centrifugation and tumor cells were separated from immune cells by overnight adherence to tissue culture plates. Both cell fractions were characterized for the expression of immune and cancer cell markers by flow cytometry and/or quantitative PCR. TDO and IDO1 activity was determined using a functional assay and small molecule inhibitors. All samples expressed CA125 or HE4, or both ovarian cancer markers. Flow cytometry analysis of eleven adherent cell samples with anti-EpCAM or CA125 confirmed their EOC origin. Five of six samples showed high cell surface expression of PD-L1. Adherent cell samples were cultured over 2 - 3 passages and further characterized for TDO and IDO1 expression and activity. All cultured cell samples stained negative for immune cell markers (CD45) and only a small fraction (less than 10%) stained positive for a fibroblast marker. Two samples out of eighteen showed high IDO1 gene expression. However, only after stimulation with IFNγ, the tryptophan metabolizing activity in these samples was detectable, and could be inhibited with selective IDO1 inhibitors. In contrast, four samples expressed relatively high TDO and detectable endogenous tryptophan metabolizing activity, which could be inhibited by selective TDO inhibitors. The immune cell composition of the ascites, the expression and activity of TDO, IDO1 and PD-L1 were related to the clinical outcome of the patients from which the ascites was derived. Citation Format: Wilhelmina van Riel, Antoon van Doornmalen, Winfried Mulder, Diep Vu, Jelle Dylus, Judith den Ouden, Anne van Altena, Rogier Buijsman, Guido Zaman. High TDO and IDO1 expression in ovarian cancer-associated cells isolated from malignant ascites [abstract]. In: Proceedings of the Annual Meeting of the American Association for Cancer Research 2020; 2020 Apr 27-28 and Jun 22-24. Philadelphia (PA): AACR; Cancer Res 2020;80(16 Suppl):Abstract nr 5604.

Introduction Poly (ADP-ribose) polymerases (PARPs) are important players in DNA damage repair. Inhibition of PARP activity is highly effective against cancers deficient in homologous recombination repair due to, a.o., BRCA1/2 mutations. Clinically, PARP inhibitors (PARPi) are used to treat, a.o., ovarium cancers. Although effective, repeated treatment with PARPi may lead to resistance. To identify new synergistic drug combinations, we developed a two-step approach of screening and confirmation using two of our platform technologies SynergyScreen and SynergyFinder. We first screen for synergy in presence of a fixed concentration of compound (e.g., PARPi), followed by confirmation using dose response curves. The PARPi niraparib was used as a proof-of-concept combinatorial drug at a fixed concentration, and combined with over 150 anti-cancer agents. Material and methods Proliferation of cell lines and patient derived primary ovarium cancer cells was measured using ATPLite 1 Step after 120 hours compound incubation. Growth was calculated relative to vehicle treated cells; relative IC50s were calculated from a four parameter logistics curve. The fixed concentration niraparib used in the SynergyScreen represented 80% viability. Anti-cancer agents were screened in absence and presence of niraparib, followed by analysis of IC50 shifts. Combinations with IC50shifts>2 fold indicated a synergistic hit and were re-examined using equipotent mixtures and calculation of Combination Index (CI) within our SynergyFinder platform.1 CI Results and discussions Results of the SynergyScreen showed various synergistic drug combinations from the high-throughput setup, i.e. niraparib in combination with the topoisomerase I inhibitor irinotecan (CI0.5=0.70) and the DNA alkylating agent temozolomide (CI0.5=0.42). These and other synergies were confirmed in patient samples. Independently from niraparib, the bromodomain inhibitor JQ1 and the histone acetyltransferase inhibitor anacardic acid showed synergy (CI0.5 0.42), which was not demonstrated before. Conclusion Our Synergy platforms can identify novel and reproducible synergistic drug combinations in an unbiased and efficient manner. Proof-of-concept was demonstrated and new synergistic drug combinations were found. Our method is therefore excellently suited for the in vitro discovery and validation of synergistic combinations. Reference Uitdehaag, et al. PLoS ONE 2015;10(5): e0125021.

Introduction: Epithelial ovarian cancer (EOC) is the most lethal gynecologic malignancy. First-line therapy in advanced EOC is surgery in combination with platinum-based chemotherapy and paclitaxel. 15-20% of patients do not respond and in 80% of advanced cases, the disease recurs within three years. PARP inhibitors synergize with platinum therapy and have been approved for platinum sensitive EOC. Clinical trials with immunotherapies, such as PD-1/PD-L1 blockade, have so far not been successful. Currently, the only approved companion diagnostic is BRCA gene mutations for PARP inhibitors. More diagnostic assays to predict the clinical response to chemo- or immunotherapies are needed. We have developed a biomarker discovery platform using ascites of ovarian cancer patients. Experimental procedures Ascites was gathered from patients by punction or during debulking surgery. Cells were collected by centrifugation and characterized by flow cytometry using specific antibodies. Genomic DNA was sequenced using Illumina cancer gene panels. Gene expression was analyzed by quantitative PCR (qPCR). Cellular activity of the tryptophan metabolizing enzymes IDO1 and TDO was measured with NFK Green [1]. Levels of L-tryptophan and its metabolite L-kynurenine in ascites fluid and blood were determined with LC-MS/MS. In vitro data were related to tumor histopathology and clinical response data. Results: Low passage cell samples from twenty patients were profiled for sensitivity to various cytotoxic agents and targeted anti-cancer therapies in cell proliferation assays. In parallel the mutation status of fifty cancer genes including BRCA1 and 2 was assessed by DNA sequencing. The expression of genes implicated in resistance to chemotherapy (CCNE1, ABCB1) or immunotherapy (PD-L1, IDO1, TDO) was determined with qPCR. The immune status of ascites was analyzed by measuring the relative proportion of different immune cell populations, i.e., cytotoxic and regulatory T cells, monocytes, dendritic and natural killer cells. The expression of the immune suppressive markers PD-L1, IDO1 and TDO was related to the immune cell composition of the ascites, kynurenine-tryptophan ratio, and clinical response data. A tumor cell sample derived from a patient with low grade serous ovarian cancer (LGSOC) was heterozygous for an oncogenic NRAS mutation and was much more sensitive to MEK inhibitors than other samples not harboring the mutation. The cells also expressed IDO1 and high levels of PD-L1 at the cell surface. Two other samples derived from high grade serous ovarian cancer (HGSOC) expressed high PD-L1 and one also IDO1. Several HGSOC samples expressed TDO. One HGSOC sample showed high expression of the ABCB1 gene, encoding the multidrug transporter P-glycoprotein. The sample was relatively resistant to paclitaxel, a known substrate of P-glycoprotein Conclusion: Our study shows that in vitro drug sensitivity assays with primary patient samples can be used to confirm or identify predictive drug response biomarkers. In an ongoing study, in which hundred patients with HGSOC will be included, the in vitro drug response of tumor cells from ascites to first-line cytotoxic anti-cancer agents will be determined and compared to the clinical response of patients bearing specific genomic biomarkers. [1] Seegers et al. (2014) J Biomol Screen 19, 1266-1272 Citation Format: Guido J.R. Zaman, Judith E. den Ouden, Jelle Dylus, Antoon M. van Doornmalen, Winfried R. Mulder, Diep Vu-Pham, Suzanne J.C. van Gerwen, Joost C.M. Uitdehaag, Rogier C. Buijsman, Leon F. Massuger, Anne M. van Altena. A precision medicine platform to predict the clinical response to chemo- and immunotherapy for epithelial ovarian cancer [abstract]. In: Proceedings of the AACR-NCI-EORTC International Conference on Molecular Targets and Cancer Therapeutics; 2019 Oct 26-30; Boston, MA. Philadelphia (PA): AACR; Mol Cancer Ther 2019;18(12 Suppl):Abstract nr A044. doi:10.1158/1535-7163.TARG-19-A044

Introduction: Combination drug treatment in cancer therapy aims to improve response rate and decrease the development of drug resistance. The discovery of novel drug combinations is constrained by the cost and effort of carrying out large unbiased screens and is hampered by poor translation towards the clinic. We investigate if computational models combined with screening can more efficiently reveal synergistic combinations and improve translation towards the clinic. The models are based on three different biological views of synergy. First, compounds that exploit similar vulnerabilities in cancer cells frequently show synergy in a particular genetic context (targeted synergy, model 1) [1,2]. Secondly, drugs with non-overlapping resistance mechanisms often yield clinically relevant drug combinations (model 2) [3]. Finally, drugs that mimic synthetic lethal interactions will often act synergistically (model 3) [4]. Experimental procedures Oncolines™ is a panel of 102 genetically characterized cell lines from diverse tumor origins, on which proliferation assays are run in parallel in nine-point dose-response curves. In this cell panel we profiled 162 different cancer therapeutics, including many standard of care, chemotherapeutic agents, epigenetic modulators, and approved and pre-clinical drugs such as CDK4/6, ALK and PI3K-inhibitors [5]. Cancer hotspot mutations and gene expression data were downloaded from the DepMap and CCLE databases. ANOVA and Pearson correlations were used to analyze the single agent response and genetic data in the statistical software R, to determine gene mutation (model 1) and resistance markers (model 2). For model 3, clinically relevant synthetic lethal pairs were combined with gene expression data [4]. The methods were used to predict results from published synergy screens, e.g. DREAM [6] and results from an independent experiment (NTRC SynergyScreen™) in which fixed concentrations of the poly-ADP ribose polymerase (PARP) inhibitor niraparib and the BET bromodomain inhibitor JQ1, were combined with nine-point dose-response curves of 150 anticancer agents. Synergies were quantified using curve-shift and CI-index. Results: Using a cancer cell line IC50 profile of a compound (such as Oncolines™), all three computational models of synergy prediction can be applied. Method 1 is successful in predicting combinations that augment a targeted effect, such as combining BRAF and MEK inhibitors in BRAF-mutant cancer [1]. Method 2 does not distinguish between additive or synergistic combinations but can identify clinically relevant combinations [3]. For method 3, a tool was developed that computationally assesses if drug pairs can pharmacologically mimic clinically validated synthetic lethal interactions [4]. This can predict DREAM data with a high AUC of 0.67 in a ROC curve. The tool successfully predicts synergy between niraparib and the SUMO inhibitor 2-D08, and niraparib and BCL2 inhibitors, which we observed in our screen. Conclusions: Computational tools for synergy have predictive value and can be useful to prioritize libraries for empirical combination screening. References: [1] Uitdehaag et al. (2015). PLoS ONE 10(5): e0125021. [2] Seashore-Ludlow et al. (2015) Cancer Discovery, 5: 1210-1223. [3] Palmer et al. (2017) Cell 171: 1678-1691. [4] Lee et al. (2018). Nature Communications 9: 2546. [5] Uitdehaag et al. (2016). Mol. Cancer Ther. 15: 3097-3109. [6] Menden et al. (2019) Nature Communications 10: 2674. Citation Format: Joost C.M. Uitdehaag, Martine B.W. Prinsen, Derek W. van Tilborg, Jeffrey J. Kooijman, Jelle Dylus, Jeroen A.D.M. de Roos, Suzanne J.C. van Gerwen, Jos de Man, Rogier C. Buijsman, Guido J.R. Zaman. Computational models of synergy contribute to efficient combination screening [abstract]. In: Proceedings of the AACR-NCI-EORTC International Conference on Molecular Targets and Cancer Therapeutics; 2019 Oct 26-30; Boston, MA. Philadelphia (PA): AACR; Mol Cancer Ther 2019;18(12 Suppl):Abstract nr A141. doi:10.1158/1535-7163.TARG-19-A141

In cancer therapy, combination drug treatment aims to improve response rate and decrease the development of drug resistance. The discovery of new effective drug combinations is constrained by the cost and effort of carrying out large unbiased screens and by poor translation of results towards the clinic. Here we describe how focusing on the biological mechanisms underlying the activity of drug candidates may aid a priori selection of promising synergy candidates and help in translate synergistic combinations towards a clinical situation. We have previously shown [1] that curve shift analysis as developed by Straetemans et al. [2] is a better method than combination-matrix screening. Also a dose based score such as the isobologram or the CI-index more robustly assesses synergy than an effect-based score such as the Bliss-additivity [1]. On this basis, we developed a two-step synergy screening approach, called SynergyScreen™. By distinguishing separate synergy screening and synergy confirmation stages, this setup capitalizes on insights from high throughput screening to discover robust and reproducible pharmacologically synergistic pairs. To further improve the efficiency of synergy screening, we focused on pre-selecting compounds in our screening library according to their biological mechanism. We profiled a library of more than 160 anti-cancer agents in a cell panel of 102 cell lines from diverse tumor origins [3]. Agents were clustered according to response and so-called exemplars were collected into a focused library that represents the spectrum of biological mechanisms of current cancer therapy. This synergy screening library includes many standard of care chemotherapeutic agents, approved and pre-clinical kinase inhibitors, epigenetic modulators and compounds acting by other mechanisms. Finally, we harnassed recent insights into the biology of synergy to understand and predict synergistic pairs. A tool was developed that uses the response of a compound in a 102 cell line panel to pick potential synergistic partners from the database of preprofiled compounds. It does this by computationally assessing if pairs can pharmacologically mimick clinically validated synthetic lethal interactions [4]. We optimized prediction accuracy using the results of internal and external synergy screens. The tool was applied to specifically enrich test libraries and to predict synergies at clinically relevant doses, including the results of a SynergyScreen™ with the poly-ADP ribose polymerase (PARP) inhibitor niraparib and the BET bromodomain inhibitor JQ1. References [1] Uitdehaag et al. (2015). PLoS ONE 10(5): e0125021. [2] Straetemans et al. (2005). Biometrical J. 47, 299-308. [3] Uitdehaag et al. (2016). Mol. Cancer Ther. 15, 3097-3109. [4] Lee et al. (2018). Nature Communications 9, 2546. Citation Format: Joost C. Uitdehaag, Derek W. van Tilborg, Martine B.W. Prinsen, Jeffrey J. Kooijman, Jelle Dylus, Jeroen A.D.M. de Roos, Suzanne J.C. van Gerwen, Jos de Man, Rogier C. Buijsman, Guido J.R. Zaman. Combining cell panel profiling and synthetic lethality data to efficiently screen for synergistic combinations [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2019; 2019 Mar 29-Apr 3; Atlanta, GA. Philadelphia (PA): AACR; Cancer Res 2019;79(13 Suppl):Abstract nr 2158.

The collective effort of cancer research has resulted in a large diversity of small molecule therapeutics that are tested clinically as precision medicine in specific patient populations. The applications of these therapeutics could be further extended and refined by the identification of new genomic biomarkers that are predictive for response. One of the best ways to do this, is by in vitro profiling in cancer cell line panels. Oncolines is a panel of 102 genetically characterized cell lines from diverse tumour origins, on which proliferation assays are run in parallel. Earlier analysis has shown that the Oncolines workflow generates highly reproducible data, as required for biomarker discovery [1]. In this cell panel, we profiled 162 different cancer therapeutics, including many standard of care chemotherapeutic agents, approved and pre-clinical kinase inhibitors, epigenetic modulators and compounds acting by other mechanisms. Assays were based on ATP-lite read-out, with a nine-point duplicate dilution series of the compounds. Drug response was quantified by manual curve fitting. Response was associated with the genomic status of the cell lines as retrieved from the COSMIC and Cancer Cell Line Encyclopedia (CCLE) databases. Mutations in patient hotspot locations and copy number changes in well-characterized cancer genes were used as input for ANOVA tests. Basal gene expression levels of 383 clinically actionable genes and sets of perturbation-response genes were applied to pathway analysis. Because various alternative metrics were proposed recently to quantify cell line response, we first investigated which metric resulted in the most sensitive identification of biomarkers. Responses were quantified as IC50, GI50, growth rate inhibition (GR50), DSS score, AUC and maximum effect. Known responder populations were analysed such as BRAF(V600E) mutant cell lines for BRAF inhibitors and EGFR amplified lines for EGFR inhibitors, to identify the optimal response metric for biomarker identification. The full spectrum of cell line responses was analyzed with a variety of clustering and principal component methods to define groups of therapeutic agents based on common biological mechanisms. We further analyzed novel clusters of the Poly (ADP-ribose) polymerase (PARP) inhibitors olaparib, niraparib, rucaparib, and talazoparib, the proteasome inhibitors MG-132, bortezomib and carfilzomib, the ubiquitin activating enzyme inhibitor MLN-7243 (TAK-243) and the BET-family inhibitors JQ1 and I-BET-762 and identified several response biomarkers based on mutation, gene expression and pathway analysis. References [1] Uitdehaag et al. (2016). Cell Panel Profiling Reveals Conserved Therapeutic Clusters and Differentiates the Mechanism of Action of Different PI3K/mTOR, Aurora Kinase and EZH2 Inhibitors.Mol. Cancer Ther. 15, 3097-3109. Citation Format: Joost C. Uitdehaag, Jeffrey J. Kooijman, Jeroen A.D.M. de Roos, Martine B.W. Prinsen, Jelle Dylus, Nicole Willemsen-Seegers, Jos de Man, Suzanne J.C. van Gerwen, Rogier C. Buijsman, Guido J.R. Zaman. Cell panel profiling of 162 small molecule therapeutics identifies response biomarkers for PARP, BET-family and proteasome inhibitors [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2019; 2019 Mar 29-Apr 3; Atlanta, GA. Philadelphia (PA): AACR; Cancer Res 2019;79(13 Suppl):Abstract nr 279A.

Epithelial ovarian cancer (EOC) is the most lethal gynecologic malignancy. Standard treatment of advanced EOC is surgery in combination with chemotherapy, consisting of a platinum-based drug (carboplatin or cisplatin) and paclitaxel. Although most women show a good response to first line treatment, tumors do not respond in 15-20% of patients, whereas in 80% of cases of advanced EOC, the disease recurs within three years. Patients who have responded to first line treatment are referred to as ‘platinum sensitive’, and are treated with a PARP inhibitor in second line. Several other new targeted inhibitors are investigated in clinical trials, including PI3-kinase and bromodomain inhibitors. There is great need of new, more effective therapies with improved long-term treatment outcome, and diagnostic assays and biomarkers to predict chemotherapy response in the clinic. We have determined whether tumor cells isolated from ascites of patients with EOC can be used to develop in vitro cell proliferation assays. We also characterized the immune status of ascites by cytokine analysis and used flow cytometry to analyze the immune cell types in ascites. In a pilot study, ascites was gathered from 18 patients with advance stage EOC. Cells were isolated by centrifugation and tumor cells were separated from immune cells by adhering them to tissue culture plates. Sensitivity of tumor cells to platinum compounds, paclitaxel, PARP inhibitors, and other second line and investigational drugs, as well as drug combinations, was determined in cell proliferation assays, using intracellular ATP content as an indirect read-out of cell number [1]. The mutant status of a number of known oncogenes and tumor suppressor genes in the patient-derived cell samples was determined by DNA sequence analysis. Mutation status was related to histopathological data and in vitro drug response. We demonstrate that ascites from EOC patients contains many tumor cells and that these tumor cells can be used to perform drug sensitivity assays in vitro. In an ongoing study, in which ascites from hundred EOC patients will be included, the in vitro sensitivity to standard-of-care chemotherapy will be related to the immune status of the ascites and to the clinical outcome of the patient, which is defined by platinum sensitivity. An update of this study will be presented at the conference. [1] Uitdehaag et al. (2014) PLoS ONE 9(3): e92146 Citation Format: Guido Zaman, Judith E. den Ouden, Jelle Dylus, Antoon M. van Doornmalen, Rogier C. Buijsman, Astrid Eijkelenboom, Leon F. Massuger, Anne M. van Altena. Chemotherapy sensitivity of tumor cells from ascites of ovarian cancer patients: Relationship with immune status and clinical response [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2019; 2019 Mar 29-Apr 3; Atlanta, GA. Philadelphia (PA): AACR; Cancer Res 2019;79(13 Suppl):Abstract nr 2221.

Profiling of new drug candidates on cancer cell line panels is an important tool to identify candidate drug response biomarkers for clinical studies. We have set up a platform, called Oncolines™, that comprises 102 genetically well-characterized cell lines from diverse tumor tissues. The cell lines are screened in parallel in a high-throughput proliferation assay based on ATP-lite read-out with 9 point dose-response curves in duplicate and manual inspection of curve fitting. We have recently shown that this workflow leads to highly reproducible IC50s which are necessary for genomic biomarker discovery [1]. Here we apply this platform to investigate BTK and CDK4/6 inhibitors. In previous work [1,2] we correlated the IC50 profiles of small molecule inhibitors through Anova to curated databases of somatic mutations and copy numbers. To obtain a more comprehensive view of oncogenic signaling inside a cell, we calculated in this study correlations between the log IC50s and basal gene expression levels for more than 18,000 genes. To reduce the number of false positives, we first considered only genes with a known role in cancer biology or drug resistance. Secondly, correlations were corrected for the average correlations seen in an Oncolines™ database of more than 150 anti-cancer agents [1]. Thirdly, we looked at co-correlation of genes that interact on a protein level (StringDB) or at a pathway level (Gene Set Analysis). Our method, called GeneNominator™, was validated by profiling the microtubule binder vincristine, the EGFR inhibitor gefitinib, and the MDM2 antagonist nutlin, and confirmed known pharmacogenomic relations.In a first study we tested three CDK4/6 inhibitors, ribociclib, palbociclib and abemaciclib. Although their response patterns cluster together, they show significant differences. For instance, palbociclib is more active on cell lines that express the transcription factor GATA3, which is linked to cyclin signalling in neuroblastoma.In a second study we tested the BTK inhibitors ibrutinib and acalabrutinib. Both show very different inhibition profiles across the 102 cell lines. Ibrutinib shows considerable activity in cell lines that overexpress ERBB4, FGFR2 and ERBB2, reflecting its biochemical inhibition of these kinases. Acalabrutinib, which is more selective, is most active in cell lines that highly express IRF4, a known genetic driver in multiple myeloma. and the B-cell receptor subunit CD79B, of which mutations are often found in diffuse large B-cell lymphoma. When mutations and copy number changes are studied, acalabrutinib is particularly active in cell lines with CREBBP or EZH2 mutations, which occur in diffuse large B-cell lymphomas. Our data demonstrate that our in vitro cell panel screening method can uncover new mechanistic information on clinically used anti-cancer agents. [1] Uitdehaag et al. (2016) Mol. Cancer Ther. 15, 3097-3109.[2] Uitdehaag et al. (2014) PLOS ONE 9: e92146 Citation Format: Joost C. Uitdehaag, Jeffrey J. Kooijman, Jeroen A.D.M. de Roos, Martine B.W. Prinsen, Jelle Dylus, Judith de Vetter, Nicole Willemsen-Seegers, Jos de Man, Suzanne J.C. van Gerwen, Rogier C. Buijsman, Guido J.R. Zaman. Cell line panel profiling reveals novel drug response biomarkers for BTK and CDK4/6 inhibitors [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 4907.

Introduction In epithelial ovarian cancer (EOC) patients, 15%–20% of the tumours do not respond to first-line chemotherapy treatment, i.e., paclitaxel combined with a platinum-based therapy. The aim of this pilot study was to validate a new approach to measure the sensitivity of ovarian cancer cells to chemotherapeutics. Furthermore, we aimed to relate the in vitro response of tumour cells from ascites to clinical response of ovarian cancer patients. Material and methods For this pilot study, we selected 18 patients with advanced stage EOC and collected ascites. Proliferation assays on tumour cells isolated from the ascites were performed using intracellular ATP content as an indirect measure of cell number. Clinical outcome was determined using status of remission after primary treatment, disease free survival (DFS), and the platinum-sensitivity. Complete remission was defined as a serum-CA125 Results and discussions The outcome of the proliferation assays, expressed by GI50 and using dose-response curves, corresponded to the clinical outcome in the patient in most cases. Of the 14 ascites samples eventually tested, seven of the in vitro responses were poor as well as the DFS of the patient. Four samples showed good in vitro response and the patient showed a good DFS as well. Only two of the samples tested did not show similarity in in vitro response and in clinical outcome of the patient. We also determined the in vitro response to second line therapies and new targeted therapies, including PARP- and bromodomain inhibitors. Conclusion This pilot study demonstrates the feasibility of using ascites tumour cells to perform drug sensitivity tests. These are promising results because it seems possible to predict accurately the response on first-line chemotherapy in advanced stages of EOC. However, larger observational studies are required to show a statistical significant relation, to determine the best clinical response measurements and to test second line therapies.

Introduction Profiling of drugs on cancer cell line panels can uncover new insights into the mechanisms of drug response. We have established a panel of 102 genetically well-characterised cell lines from distinct tumour origins, called Oncolines. Earlier work on kinase inhibitor profiling showed that our workflow generates highly reproducible data. 1 Here we present the profiling of seventeen kinase inhibitors which have been approved since November 2014 in the Oncolines panel. 2 Material and methods The profiled agents include the ALK inhibitors ceritinib, brigatinib, and alectinib, the CDK4/6 inhibitors palbociclib, abemaciclib, and ribociclib, the BTK inhibitors ibrutinib, and acalabrutinib, and nine other novel marketed inhibitors that were not profiled earlier. 2 The cell lines were screened in parallel in a high-throughput proliferation assay based on ATP-lite read-out, at 9 duplicate concentrations of each inhibitor. Drug response was expressed as IC 50 and GR 50 values. 3 Drug response metrics were associated with genomic alterations in the cell lines. These were retrieved from the Catalogue of Somatic Mutations in Cancer database and filtered for relevance in primary patient tumours. In addition, we calculated correlations between the log IC 50 s and basal gene expression levels of 18 900 genes, retrieved from the Cancer Cell Line Encyclopaedia. Correlations were filtered by the Oncolines database of 150 previously profiled anti-cancer agents. 1 Results and discussions Of the ALK inhibitors, alectinib was more selective than ceritinib and brigatinib. Cell lines harbouring the NPM-ALK translocation were potently inhibited by all three inhibitors. There was a strong correlation between drug sensitivity and high ALK expression levels. Furthermore, elevated expression of JAK3 and other JAK-STAT signalling genes correlated with drug sensitivity, indicating a potential role of JAK-STAT signalling in response to ALK inhibitors. Elevated expression of the transcription factor MYCN was indicative of the sensitivity of NPM-ALK negative cell lines for ceritinib and brigatinib. Conclusion By profiling all clinically approved kinase inhibitors, we were able to compare their selectivity and identify novel response biomarkers. These data will be important in guiding novel applications of these inhibitors as well as future drug development in the kinase field. References Uitdehaag, et al. Mol. Cancer Therap 2016;15:3091–3109. Uitdehaag, et al. PLoS One 2014;9:e92146. Hafner, et al. Nat. Methods 2016;13:521–527.

Screening of new drugs on large cell panels is an important tool to study the biologic mechanism of drug response. From a combination of parallel in vitro tests and bio-informatics analysis, important conclusions can be drawn, for instance about drug selectivity in a cellular context, about resistance mechanisms, and about the patient population in which a drug is effective. This makes cell panel screening indispensable in modern drug discovery. We have set up a platform called Oncolines™ that comprises 102 cell lines from diverse tumor tissues. All cell lines are screened in parallel in high-throughput proliferation assays based on ATP-lite™. Compounds are tested with 9 point dose-response curves in duplicate. Curves are visually inspected. In the past, we have shown that this workflow leads to highly reproducible IC50s, which are necessary for genomic biomarker discovery [1]. For instance, the IC50s have been coupled to curated databases of somatic mutations and copy numbers (1). However, these changes only reflect a small percentage of oncogenic transformations. A more comprehensive view of oncogenic signaling inside a cell can be obtained from mRNA expression levels (2). Here, we describe a workflow to investigate drug response in the 102 cell line Oncolines™ panel based on basal gene expression levels. Correlations between gene expression levels and the log IC50s were calculated for more than 18,000 genes. To reduce the number of false-positive correlations, we considered only genes with a known biologic role in cancer or that were clinically actionable. Secondly, we filtered out genes that correlated indiscriminately with drug responses, by correcting for the average correlations seen in our profiling database of more than 150 inhibitors. This filtered gene list was structured graphically by combining it with information on protein-protein interactions (using the StringDB) or information on which genes are part of the same pathways (a method called Gene Set Analysis). Our method reveals that gene overexpression correlates with drug responses for a number of targets. For instance, EGFR, IGFR, or ALK kinase inhibitors have more potent effects in cell lines that overexpress EGFR, IGFR, or ALK, respectively, and the MDM2 antagonist nutlin is more active in MDM2-overexpressing cell lines. This is irrespective of the genomic alterations driving overexpression of these genes. The distinct response profiles of the spectrum-selective BTK/EGFR/ERBB2 inhibitor ibrutinib and the more selective BTK inhibitor alcalabrutinib could be distinguished as well. Finally, the analysis shows mechanisms of drug resistance. For instance, the response of vincristine strongly correlates to cellular ABCB1 expression, the P-glycoprotein/MDR1 drug transporter, but also to ABCC3, another ABC transporter. Combining basal gene expression levels with cell panel data therefore allows discovery of novel mechanisms of drug response and resistance. References: 1. Uitdehaag et al. Cell panel profiling reveals conserved therapeutic clusters and differentiates the mechanism of action of different PI3K/mTOR, Aurora kinase and EZH2 inhibitors. Mol Cancer Therap 2016;15:3097-3109. 2. Rees et al. Correlating chemical sensitivity and basal gene expression reveals mechanism of action. Nature Chem Biol 2016;12:109-16. Citation Format: Joost C.M. Uitdehaag, Jeffrey Kooijman, Jeroen A.D.M. de Roos, Martine B.W. Prinsen, Jelle Dylus, Suzanne J.C. van Gerwen, Rogier C. Buijsman, Guido J.R. Zaman. Combining cell panel screening with analysis of gene expression levels reveals features of drug response and resistance [abstract]. In: Proceedings of the AACR-NCI-EORTC International Conference: Molecular Targets and Cancer Therapeutics; 2017 Oct 26-30; Philadelphia, PA. Philadelphia (PA): AACR; Mol Cancer Ther 2018;17(1 Suppl):Abstract nr B155.

[purpose] Profiling of drug candidates in cell line panels is an important tool to compare the selectivity and targeting of new anti-cancer agents. In addition, comparative profiling may be used to repurpose established therapeutics by identifying new mechanisms of action or cross-selectivities. [experimental procedures] A collection of more than 120 anti-cancer agents, targeting all important oncogenic signaling pathways, including classic cytotoxic agents as well as many targeted kinase inhibitors and epigenetic modulators, was profiled on a panel of 44 or 66 parallel cell line proliferation assays (Oncolines™) [1,2]. The Oncolines™ profiles of the compounds were compared by Pearson correlations of their inhibitor responses. Inhibitor sets were clustered using hierarchical trees. Response profiles were correlated to the genetic background of the cell lines by Analysis of Variance (Anova). [results] Reproducibility of the NTRC Oncolines™ cell panel was validated by monitoring cell growth rate and the variation in IC50s of replicate profiles over a period of three years. The Pearson correlation between replicates ranged between 0.60 and 0.99 for 16 different inhibitors, depending on dose-response curve shape. Correlation analyses of the > 120 profiled anti-cancer agents revealed separate clusters of, a.o., taxanes, platins, topo-isomerase inhibitors, and EGFR, ABL, MEK and BRAF inhibitors. This demonstrates that the Oncolines™ profiles are an unbiased representation of the compound's mechanisms. The profile of the BTK inhibitor ibrutinib correlated with EGFR inhibitors. In biochemical experiments we showed that this due to its cross-reactivity with EGFR. The six Aurora kinase inhibitors profiled fall into two separate clusters, which are related to their biochemical selectivity. Thus, Aurora A-selective inhibitors are relatively more active in cell lines with mutations in cell cycle checkpoint-related genes such as TP53 and RB1; whereas pan-Aurora inhibitors are more active in cell lines with mutations in growth factor signaling pathways, such as NRAS. Profiling of eleven PI3 kinase and mTOR inhibitors revealed four distinct clusters. PI3Kalpha and PI3Kdelta isoform selective inhibitors each target genetically distinct subgroups of cell lines. Rapamycin-analogs, such as everolimus, specifically target PTEN-mutant cell lines. Finally, profiling of EZH2 inhibitors indicates that there are essentially two classes and that these have a cellular profile that is distinct from HDAC, DOT1L or BET inhibitors. [conclusions] Comparative cancer cell line profiling is a powerful tool to rapidly explore the pharmacogenomics of drug action in cancer cells and to identify new or previously unnoted activities of compounds. [references]: [1] Uitdehaag et al. (2014) PLOS ONE 9(3) e92146; [2] Uitdehaag et al. (2015) PLOS ONE 10(6) e0132230. Citation Format: Joost C.M. Uitdehaag, Jeroen A.D.M. de Roos, Martine B.W. Prinsen, Judith R.F. de Vetter, Jelle Dylus, Antoon M. van Doornmalen, Suzanne J.C. van Gerwen, Jos de Man, Rogier C. Buijsman, Guido J.R. Zaman. Comparative cancer cell line profiling differentiates the mechanism of action of different PI3K/mTOR, Aurora kinase and EZH2 inhibitors. [abstract]. In: Proceedings of the 107th Annual Meeting of the American Association for Cancer Research; 2016 Apr 16-20; New Orleans, LA. Philadelphia (PA): AACR; Cancer Res 2016;76(14 Suppl):Abstract nr 4635.