IntroductionA decentralized and multi-platform-compatible molecular diagnostic tool for kidney transplant biopsies could improve the dissemination and exploitation of this technology, increasing its clinical impact. As a first step towards this molecular diagnostic tool, we developed and validated a classifier using the genes of the Banff-Human Organ Transplant (B-HOT) panel extracted from a historical Molecular Microscope® Diagnostic system microarray dataset. Furthermore, we evaluated the discriminative power of the B-HOT panel in a clinical scenario.Materials and MethodsGene expression data from 1,181 kidney transplant biopsies were used as training data for three random forest models to predict kidney transplant biopsy Banff categories, including non-rejection (NR), antibody-mediated rejection (ABMR), and T-cell-mediated rejection (TCMR). Performance was evaluated using nested cross-validation. The three models used different sets of input features: the first model (B-HOT Model) was trained on only the genes included in the B-HOT panel, the second model (Feature Selection Model) was based on sequential forward feature selection from all available genes, and the third model (B-HOT+ Model) was based on the combination of the two models, i.e. B-HOT panel genes plus highly predictive genes from the sequential forward feature selection. After performance assessment on cross-validation, the best-performing model was validated on an external independent dataset based on a different microarray version.ResultsThe best performances were achieved by the B-HOT+ Model, a multilabel random forest model trained on B-HOT panel genes with the addition of the 6 most predictive genes of the Feature Selection Model (ST7, KLRC4-KLRK1, TRBC1, TRBV6-5, TRBV19, and ZFX), with a mean accuracy of 92.1% during cross-validation. On the validation set, the same model achieved Area Under the ROC Curve (AUC) of 0.965 and 0.982 for NR and ABMR respectively.DiscussionThis kidney transplant biopsy classifier is one step closer to the development of a decentralized kidney transplant biopsy classifier that is effective on data derived from different gene expression platforms. The B-HOT panel proved to be a reliable highly-predictive panel for kidney transplant rejection classification. Furthermore, we propose to include the aforementioned 6 genes in the B-HOT panel for further optimization of this commercially available panel.

BACKGROUND Major obstacles that have impeded the development of effective new therapies for GBM include inter- and intratumoral heterogeneity, the blood-brain-barrier and use of sub-optimal cell line-based preclinical models. Taking these hurdles into account, we have set up a patient-derived GBM drug-screening platform. We optimized protocols to improve cell culture success rate and retrospectively assessed the predictive power of our assay for patient response to TMZ. A large panel of GBM cells was screened for sensitivity to available oncological agents. Drugs of interest were selected based on favorable physicochemical properties for BBB crossing and potent activity in (a subset of) GBM cultures. Finally, we determined the success rate of performing a small-scale screen with 20 selected agents within 4 weeks of receiving tumor tissue. RESULTS By combining both CUSA and tissue piece-derived dissociation protocols, culture success increased to 95%, ensuring representation of the near-complete spectrum of GBM subtypes. Single-cell sequencing studies confirmed heterogeneity in our low-passage cell cultures. In vitro screening of TMZ on a large cohort (n = 55) identified 3 response categories (responders/intermediates/non-responders) for which Cox regression analysis revealed significantly different overall survival curves of corresponding patients. Screening of 107 FDA-approved anticancer agents on 45 GBM cultures underscored the tremendous intertumoral heterogeneity in drug sensitivities. We identified 20 potent agents each effective at clinically-achievable concentrations in (a subset of) GBM cultures and having favorable BBB penetration properties (CNS-MPO score). Screening of these agents on a per patient basis within 4 weeks of receiving tissue was successful in 18 out of 24 (75%) tested tumors. In the remaining cases the tumor cells grew very slowly and longer culture times were required. CONCLUSION Our drug screening platform offers a tool to predict TMZ response and assess sensitivity to candidate treatments, either for GBM subsets or on a per patient basis.

INTRODUCTION Little progress has been made in the development of effective new therapies for glioblastoma (GBM) the past decades. Fresh patient-derived GBM cell culture models have become the gold standard for GBM drug discovery and development. One of the major obstacles in identifying novel candidate drugs against GBM remains the blood-brain barrier (BBB). Therefore, it is crucial to select drugs with favourable physicochemical properties to cross BBB and reach the tumour tissue in therapeutically effective concentrations. In current drug repurposing approach, we evaluated available anti-cancer agents in our patient-derived drug screening platform against GBM. METHODS The FDA-approved Oncology Drug Set II library was tested on 45 primary GBM cell cultures. We developed a drug shortlisting pipeline combining efficacy data with pharmacodynamic and pharmacokinetic characteristics of each compound. The therapeutic efficacy of the selected agent was assessed in an orthotopic mouse PDX model, while penetration into the CNS by LC/MS/MS. RESULTS Omacetaxine mepesuccinate (OMA) was ranked as one of the most promising candidates applying our drug selection approach. In vitro, OMA revealed anti-tumour activity at IC50 values well-below reported Cmax plasma values in approximately 80% of GBM cultures. NanoString nCounter analysis, revealed DNA damage repair as the main pathway involved in OMA’s anti-tumour effect. Activation of caspase 3/7 activity and decrease of glioma cell invasiveness were also linked to its anti-tumour effect. In vivo, 1mg/kg dose of OMA was found to reach the brain tumour tissue in concentrations similar to the reported IC50 values in vitro. No adverse reactions were noted and a survival benefit was observed in a proportion of the treated mice. CONCLUSIONS At 1 mg/kg, OMA reaches the tumour brain tissue in therapeutically effective concentrations in mice while a moderate therapeutic benefit was observed. Additional in vivo experiments are ongoing investigating higher dosages of OMA and longer exposure.

Background: The brain tumor glioblastoma (GBM) is one of the most aggressive forms of cancer. The dismal prognosis of these patients, with a median survival of less than 15 months despite maximal therapy makes the need for new therapeutic approaches urgent. Clinical trials employing oncolytic viruses (OVs) have shown encouraging results, however, it appears that for each OV only a small group of patients responds to treatment. As inter- and intra-tumoral heterogeneity is a hallmark of GBM, we hypothesized that fresh patient-derived GBM cell cultures will reflect this inter-tumoral variability in response and allow identification of potential biomarkers of susceptibility to specific OVs. Furthermore, we established a co-culture system of primary GBM cultures with autologous peripheral blood mononuclear cells (PBMCs) to capture the degree of OV-induced oncolysis in conjunction with subsequent immune activation. Using these model systems, we attempt to develop tools which may guide future personalized trials of OV treatment for GBM. Methods: We tested the oncolytic potency of four OVs derived from different viral families (DNX2401, rQnestin34.5 V1, wild type Reovirus, lentogenic NDV-f0-GFP) on a panel of 19 molecularly characterized GBM cultures and calculated the half maximal effective concentration (EC50) for each virus on each cell culture. Quantitative PCR was performed to assess cytokine expression in tumor cells after infection with the 4 different OVs. OV-induced changes in the gene and protein expression of immune associated genes were assessed in co-cultures of GBM cells with PBMCs using Nanostring nCounter System and Elisa. Results: Screening of the 4 OVs on the panel of patient-derived GBM cell cultures revealed great inter-tumoral variability in oncolysis and cytokine response to the 4 different OVs with some degree of OV specific cytokine response profiles. Correlation analysis of transcriptome data with susceptibility to the four OVs shows that genes involved in distinct pathways are related to specific OV-sensitivity. In particular, cell cycle and immune related biological processes discriminate responders and non-responders. The co-culture of OV-infected glioma cells with PBMCs suggests that infection with different OVs leads to expression of distinct sets of genes and proteins in PBMCs; indicating that each OV mounts a specific immune response. Conclusion: Heterogeneity in OV sensitivity is demonstrated in primary GBM cultures, in terms of oncolysis, cytokine induction and in virus-specific changes in gene and protein expression in OV-infected tumor cells/PBMCs co-cultures. These results support the hypothesis that improving the response rates in oncolytic virotherapy for GBM may require a personalized approach. Citation Format: Eftychia Stavrakaki, Anne Kleijn, Wouter B. van den Bossche, Rutger K. Balvers, Lisette B. Vogelezang, Jie Ju, Andrew Stubbs, Yunlei Li, Dana Mustafa, Federica Fabro, Bernadette van den Hoogen, Rob Hoeben, William F. Goins, Hiroshi Nakashima, E. Antonio Chiocca, Clemens M. Dirven, Martine L. Lamfers. Towards personalized oncolytic virotherapy: Differential response of four oncolytic viruses in primary glioblastoma cultures [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2022; 2022 Apr 8-13. Philadelphia (PA): AACR; Cancer Res 2022;82(12_Suppl):Abstract nr 3560.