Pre-clinical drug screen reveals topotecan, actinomycin D and volasertib as potential new therapeutic candidates for ETMR brain tumor patients

ABSTRACT
Background: Embryonal tumor with multilayered rosettes (ETMR) is a rare and aggressive embryonal brain tumor that solely occurs in infants and young children and has only recently been recognized as a separate brain tumor entity in the WHO classification for CNS tumors. Patients have a very dismal prognosis with a median survival of 12 months upon diagnosis despite aggressive treatment. The aim of this study was to develop novel treatment regimens in a pre-clinical drug screen in order to inform potentially more active clinical trial protocols. Methods: We have carried out an in vitro and in vivo drug screen using the ETMR cell line BT183 and its xenograft model. Furthermore, we have generated the first patient-derived xenograft (PDX) model for ETMR and evaluated our top drug candidates in an in vitro drug screen using this model. Results: BT183 cells are very sensitive to the topoisomerase inhibitors topotecan and doxorubicin, to the epigenetic agents decitabine and panobinostat, to actinomycin D, and to targeted drugs such as the PLK1 inhibitor volasertib, aurora kinase A inhibitor alisertib, and the mTOR inhibitor MLN0128. In xenograft mice, monotherapy with topotecan, volasertib, and actinomycin D, led to a temporarily response in tumor growth and a significant increase in survival. Finally, using multi-agent treatment regimens of topotecan or doxorubicin combined with methotrexate and vincristine the response in tumor growth and survival was further increased compared to mice receiving single treatments. Conclusions: We have identified several promising candidates for combination therapies in future clinical trials for ETMR patients.Embryonal tumors with multilayered rosettes (ETMR) are a tremendous challenge in neurooncological treatment because of their aggressive nature, clinical course, resistance to current chemotherapy regimens, young age, and dismal outcome of patients. There is an urgent need for novel treatment strategies and re-evaluation of current treatment protocols. We performed a detailed in vitro and in vivo drug screen in two ETMR xenograft models to identify potential drug candidates. Most of the conventional chemotherapeutics in our screen had no or little effect on ETMR cells, but they were very sensitive to PLK1 and mTOR inhibition, to actinomycin D, to topoisomerase I inhibition by topotecan and to epigenetic compounds, such as panobinostat and decitabine. In vivo studies revealed topotecan, volasertib, and actinomycin D, as the best candidates to prolong survival of tumor-bearing mice and we thus propose these three drugs as potent candidates for combination chemotherapies in future clinical trials for ETMR.

INTRODUCTION
Embryonal tumor with multilayered rosettes (ETMR) is a rare and aggressive embryonal brain tumor occurring primarily in infants and young children. Formerly described as a variant of central nervous system primitive neuroectodermal tumors (CNS-PNETs) this newly recognized brain tumor entity includes three histological variants: embryonal tumors with abundant neuropil and true rosettes (ETANTR), ependymoblastoma and medulloepithelioma1,2. These tumors are classified by their distinct histology of multilayered rosettes and neoplastic neuropils interspersed with embryonal regions and differentiated areas. Additional diagnostic markers for ETMRs are the amplification of the miRNA cluster C19MC on chromosome 19 and high expression of LIN28A protein3,4.EMTR patients face a very aggressive clinical course with poor outcome and a median survival of 12 months after diagnosis1,8. Current treatment protocols include maximal safe surgery with subsequent chemotherapy, often including high-dose chemotherapy (HDCT) with stem cell rescue, and focal or cranio-spinal radiotherapy, dependent on the age of the patient. Conventional chemotherapy often comprises a systemic multi-agent regimen including cyclophosphamide, methotrexate, vincristine, carboplatin, and etoposide9. Because of the low number of ETMR patients it is difficult to draw any profound conclusions on different treatment regimens and their effects on patient outcome to date. While conventional chemotherapy with HDCT and radiotherapy may give a good initial tumor response, the very aggressive nature of ETMRs is still poorly controlled by the current therapies, leading to a high risk of early relapse with very poor overall survival9-11.

Good pre-clinical ETMR models are needed to evaluate standard and novel treatment options. The only available pre-clinical model so far is the human ETMR cell-line BT1837. In the study of Spence et al. an in vitro drug library screen was performed, showing that BT183 cells are sensitive to IGF1R, PI3K and mTOR inhibition, as well as cell cycle and DNA damaging compounds7. However, no in vivo follow-up studies have been conducted since then. Here we have carried out a detailed in vitro and in vivo pre-clinical drug screen using BT183 cells to investigate current chemotherapeutic treatment regimens for ETMRs, and to explore other relevant and novel treatment options in order to improve therapy. Furthermore, we have used human ETMR cells derived from a newly established ETMR patient-derived xenograft (PDX) model to validate and confirm our results obtained with the BT183 cell line.For all human patient samples an informed consent was obtained in accordance with research ethics board approval.DNA methylation profiling was performed as described in Sturm et al.16. CpG methylation values for the pediatric brain tumor reference samples were used from the public dataset available under the GEO accession number GSE73801. For unsupervised clustering of pediatric brain tumor reference samples we selected the 5,000 most variably methylated probes across the dataset.Gene expression profiling of the BT183 cell line was performed using the GeneChip® Human Genome U133 Plus 2.0 Array. Data were compared with primary human ETMRs, other pediatric human brain tumor entities, and normal brain (public datasets GSE3526 and GSE73038) by using the R2 platform for analysis and visualization (http://hgserver1.amc.nl/cgi- bin/r2/main.cgi).

BT183 cells were derived from a primary, untreated tumor of a two-year-old ETMR patient and were cultured in low-adhesion cell culture flasks, as described previously7.BT183 cells were plated in 96-well plates (Sarstedt) and inhibitors were added 16 hours later. NCH3602 cells were obtained after dissection and manual dissociation of tumor cells. They were cultured in NeuroCult™ NS-A Proliferation medium and drugs were added 2-4 hours after plating. All inhibitors were commercially purchased (Supplementary Table 1). Treatments were performed in triplicates, each plate contained a non-treatment condition and vehicle condition. 72 hours later cell viability was determined using the CellTiter-Glo® reagent (Promega), luminescence was measured with an automated plate reader (Mithras LB 940, Berthold Technologies). Dose-response curves and IC50 values were calculated with Graphpad Prism software.BT183 cells were plated in collagen-coated wells at a density of 1,250 cells/ml. One hour later inhibitors were added. Colony formation was observed over 8 days. Inhibitors were added again at day 4 to avoid replenishment of inhibitors. On days 3, 5 and 8, the number of colonies was counted per well after staining with 0.1 % crystal violet/H2O. For analysis wells were scanned with a photo scanner (Epson Perfection V850 Pro) and colonies were counted manually using ImageJ software (https://imagej.nih.gov/ij/).For time-course assessment of caspase 3/7 activity, BT183 cells were measured every 12 hours from time point of drug application for 72 hours using the Caspase-Glo® 3/7 assay (Promega) and an automated plate reader.Cell viability assay was performed with the same experimental procedure as for the Caspase 3/7 time course assay using the CellTiter-Glo® assay (Promega).BT183 cells were treated with the respective inhibitors 16 hours after plating. Analysis of apoptosis was done 48 hours later for volasertib, topotecan, doxorubicin and panobinostat, and 72 hours later for alisertib, MLN0128 and decitabine. Apoptosis assay was using the PE Annexin V Apoptosis Detection Kit (BD Biosciences) according to the manufacturer’s protocol.Forty-eight hours after drug treatment, cell cycle analysis was performed using a fluorochrome solution of 0.1% Triton X-100/50 μg/mL propidium iodide/0.1% sodium citrate/H2O.

Samples were incubated for at least 1 hour at 4°C and then measured with a flow cytometer.Forty-eight hours after drug treatment BT183 cells were lysed and protein concentrations were measured. SDS-PAGE was performed according to standard procedures using 4-12% Bis-Tris gels and afterwards transferred to PVDF western blotting membranes. Membranes were incubated with respective primary antibodies at 4°C overnight (Supplementary Material Table 1). Secondary antibodies (Supplementary Material Table 2) were applied for 1 hour at room temperature and membranes were imaged by chemiluminescence (ECL Detection Reagent, GE Healthcare) and blotted on Fuji medical X-ray films (Fujifilm). Quantification was performed by densitometry analysis using ImageJ software.BT183 cells were treated with the respective inhibitors 16 hours after plating. 72 hours later cell viability was determined using the CellTiter-Glo® assay. Drug synergism was defined according to the Chou-Talalay method using the CompuSyn software (ComboSyn, Inc., Paramus, NJ)17.BT183 cells were transduced with a lentivirus carrying the pGreenFire1 (pGF1) Reporter Vector (System Biosciences) and were sorted by FACS for GFP-positive cells. For cranial transplantation, 6-8 week old female NOD-SCID gamma (NSG) mice were injected with 100,000 pGF1-labelled BT183 cells into the striatum using a stereotactic frame (coordinates: 2.5 mm ML, -1.0 mm AP, -3.0 mm DV to bregma). All animal experiments were performed according to German Laws for Animal Protection and approved by the regional authorities (approval numbers G-64/14, G-259/14).Tumor growth in BT183 transplanted NSG mice was monitored by weekly bioluminescence imaging. Briefly, mice were given 150 mg/kg VivoGlo™ luciferin (Promega), anesthetized with 2.0% isoflurane and imaged using the In Vivo Imaging System (IVIS) Lumina Series III (Caliper Life Sciences). Data was analyzed with the LivingImage software (Caliper Life Sciences).The biopsy of the primary tumor was cultured in Neurobasal A medium (Gibco) supplemented with EGF (20 ng/µl), bFGF (20 ng/µl), B-27, L-Glutamine, Heparin (1 µg/ml), Penicillin andStreptomycin for 24 hours until intracranial transplantation. 1×106 cells were injected into the striatum of 6-8 week old female NSG mice as described above.

Tumor growth was monitored by monthly MRI imaging. Animals were euthanized when they showed severe signs of tumor growth or were symptomatic.Drug treatment in BT183 xenograft mice was started once the total flux, measured by bioluminescence imaging, reached 1,000,000 photons/second in each individual animal. Animals were randomized and treated until they displayed signs of morbidity or toxicity (>20% weight loss) whereupon they were euthanized. For monotherapy treatment animals received topotecan (2.5 mg/kg), doxorubicin (Caelyx®, 7 mg/kg), actinomycin D (0.06 mg/kg), volasertib (25mg/kg), MLN0128 (3 mg/kg), alisertib (30 mg/kg), vincristine (0.5 mg/kg), panobinostat (10 mg/kg), decitabine (2.5 mg/kg). For treatment with combined chemotherapy animals received methotrexate (3 mg/kg), vincristine (0.5 mg/kg), doxorubicin (Caelyx®, 7 mg/kg) or alternatively topotecan (2.5 mg/kg). Further details of the treatments are listed in Supplementary Methods Table 1 and 2).Hematoxylin and Eosin staining and immunohistochemistry were performed on 3-4 µm formalin-fixed paraffin-embedded sections of xenograft tumors using standard protocols. LIN28A immunohistochemistry and FISH analysis for the 19q13.42 locus were performed as previously described3,18. Imaging of sections was done with the Axioskop 40 microscope (Zeiss) using Axivision Rel. 4.8 software (Zeiss). Tumor sections were classified and evaluated by our local neuropathologist.Statistical analyses were performed using GraphPad Prism software. All data are presented as means ± SD unless stated otherwise. Comparisons between different groups were made using Student’s test or ANOVA as appropriate. The statistical significance of Kaplan-Meier survival curves was assessed using the log-rank (Mantel-Cox) test. p values of 0.05 or lower were considered statistically significant for all experiments.Gene expression data reported in this paper has been uploaded to NCBI’s Gene Expression Omnibus (GEO; https://www.ncbi.nlm.nih.gov/geo/) under accession number GEO: GSE94985. CpG methylation values reported in this paper have been uploaded to NCBI’s GEO under accession number GEO: GSE94987.

RESULTS
The human ETMR cell line BT183 derived from a two-year-old ETMR patient grows stably in cell culture and can be transplanted intracranially into immune-deficient NOD scid gamma (NSG) mice to form tumors within 2-3 months that recapitulate ETMR biology7. In addition, we have generated another ETMR patient-derived xenograft (PDX) model (NCH3602) from a two- year-old patient with a supratentorial tumor mass in the right hemisphere. Tumor tissue was obtained at the time of initial surgery of the patient and tumor cells were transplanted into the striatum of four NSG mice. Three out of four mice developed tumors within 7-11 months after transplantation. To confirm the ETMR origin of our BT183 cell line, BT183 xenografts, and the established NCH3602 PDX model, we performed DNA methylation profiling (including copy- number analysis and genotyping) and compared the data with a panel of other ETMRs and various other pediatric brain tumors. ETMR methylation profiles are highly distinct from other brain tumors and the methylation profiles of the BT183 cells (unlabeled cell line (‘cell line’) and GFP-luciferase labeled cell line (‘labeled cell line’), the BT183 and NCH3602 xenografts and their matching primary tumors cluster well within the group of ETMR tumors (Fig. 1A). The C19MC amplification, present in the primary tumors, is retained in both BT183 and NCH3602 xenografts, as well as gain of chromosome 2 (Fig. 1B-E). ETMR histology of the primary tumors and the xenograft tumors for both models was confirmed by H&E staining, high LIN28A expression, and amplification of the C19MC miRNA cluster was re-validated by FISH (Supplementary Fig. 1A-N). Our data emphasize that the BT183 cell line can be used as an in vitro and in vivo model to study ETMR biology and to perform pre-clinical tests. The newly established NCH3602 ETMR PDX model, despite its long tumor latency, can serve as a useful tool for validation.

For our drug screen in ETMR cells we selected 35 compounds, including drugs currently used in treatment of ETMR and CNS-PNET patients and targeted drugs potentially relevant in ETMR tumors (Supplementary Table 1) based on the following criteria: Firstly, gene expression analyses of ETMRs, other pediatric brain tumor entities and normal brain identifying several pathways and drug targets being de-regulated in ETMRs, including the mTOR pathway (4EBP1), the SHH pathway (PTCH1), the NOTCH pathway (HES5, NOTCH1), MYCN, DNMT3B (DNA methyltransferase 3B), EZH2 (enhancer of zeste homolog 2) and topoisomerase inhibitors (Supplementary Fig. S2)4,5,7. Secondly, pre-clinical and clinical status of potential compounds targeting these pathways and proteins was considered, including efficacy for treatment in humans, preferentially in children with CNS malignancies, and approval in pediatrics or ongoing clinical studies. Thirdly, the efficacy of crossing of the blood-brain barrier (BBB) was taken into account.All compounds were tested at concentrations ranging from 16-50,000 nM using the BT183 cells and compounds were ranked by their respective IC50 (Fig. 2A). Individual IC50 curves for each compound can be seen in Supplementary Fig. S3A-E. Among the top hits in our drug screen with an IC50 <120 nM in BT183 cells were approved compounds in pediatrics, such as actinomycin D, vincristine, doxorubicin and topotecan. Other top hits included novel drugs, such as the PLK1 inhibitor volasertib, the aurora kinase A inhibitor alisertib, and the mTOR inhibitor MLN0128. Furthermore, with an IC50 <300 nM we identified several epigenetic drugs, such as nanaomycin A, decitabine and panobinostat.

To explore novel treatment options for ETMR patients, we have focused on the top hits of our screen and elaborated their function in vitro and in vivo in our BT183 xenograft mice. We excluded bortezomib since it does not cross the blood-brain barrier, and nanaomycin A which is not available for pre-clinical or clinical studies. Hence we ended up with eight compounds: actinomycin D, volasertib, alisertib, MLN0128, decitabine and panobinostat and the topoisomerase inhibitors doxorubicin and topotecan (Fig. 2B). To investigate the mechanism of action of the top compounds we performed cell viability and caspase 3/7 activity assays over a time course of 72 hours after treatment of the cells. For all compounds we found a reduced cell viability accompanied with an increased caspase 3/7 activity during the entire time course (Fig. 2C,E). Apoptotic activity was confirmed by Annexin and 7- AAD flow cytometry (Supplementary Fig. S4A-F) and by western blot analysis of BAX and cleaved PARP (Supplementary Fig. S4G-L). For alisertib and volasertib we found an accumulation of cells in G2/M phase 72 hours after drug treatment (Supplementary Fig. S4M,N). To investigate the effect of selected drugs on colony formation we performed a soft-agar colony formation assay and observed a strong reduction in colony formation for all drugs (Fig. 2D). Finally, we verified the mechanism-of-action of the drugs by immunoblotting of their specific targets. Both topoisomerase inhibitors topotecan and doxorubicin led to an increase in phosphorylation of H2A.X, a marker for DNA double-strand breaks (Supplementary Fig. S5A- D). As expected, volasertib reduced expression of PLK1-3 (Supplementary Fig. S5E-H). Alisertib reduced phosphorylation of aurora kinase A and MYCN expression (Supplementary Fig. S5I-L). MLN0128 reduced 4E-BP1 and phospho-4E-BP1, a downstream target of the mTOR pathway (Supplementary Fig. S5M-O). Altogether, our drug screen showed that only a few of the conventional chemotherapeutics used in pediatric neurooncology kill ETMR cells at low concentrations, like actinomycin D, vincristine and doxorubicin, but we identified several potential new compounds for therapy, like topoisomerase I, Aurora kinase A and PLK1 inhibitors.

The potential inhibitory effects of the eight selected compounds was further tested in vivo in our BT183 xenograft model. NSG mice were transplanted intracranially with 1x106 GFP-Luciferase labelled BT183 cells and tumor growth was monitored weekly via bioluminescence imaging. As an interventional approach treatments were started when mice had tumors with a luciferase signal of at least 1x106 pixel/second. Treatment with topotecan, volasertib, and actinomycin D, resulted in a significant increase in median survival of 55.5, 40 and 39 days, respectively, compared to 24 days of median survival in the vehicle group (Fig. 3A,B). Other compounds that showed a slight but not significant increase in survival included MLN0128, alisertib, doxorubicin, vincristine and panobinostat (Fig. 3A,B). For the decitabine treatment we encountered major toxicities in the animals even at low doses resulting in a shorter survival as compared to the vehicle-treated animals (Fig. 3A). For panobinostat and alisertib we encountered signs of toxicity only at a later stage of treatment suggesting that long-term treatment is not tolerable in these mice. The mode of application of MLN0128 (per os) was also not well- tolerated and two animals had to be sacrificed early in treatment. Growth kinetic analyses of tumors showed that treatment with topotecan, volasertib and actinomycin D initially resulted in a strong reduction of tumor growth, but tumors eventually grew again (Fig. 3C-E). In summary, our in vivo data suggest that topotecan, volasertib and actinomycin D monotherapy are sufficient to induce an initial partial response on tumor growth and prolong survival of orthotopic ETMR xenograft mice.

Since we did not observe a complete response in our xenograft model we wanted to test for synergistic drug combinations in BT183 cells. We performed an in vitro drug combination screen and tested several drug combinations among the top compounds plus a few of the conventional chemotherapeutics. Overall, 25 combinations were tested, but for 23 of these we did not find a synergistic effect (Supplementary Table 2). For combinations among our best-performing in vivo compounds, topotecan and volasertib, actinomycin D and volasertib, we could not see a synergistic effect in vitro (Fig. 4A,B). But the combination of actinomycin D and topotecan was synergistic (Fig. 4C). A significant reduction of the IC50 for actinomycin D was observed in presence of topotecan (Fig. 4D) and the IC50 for topotecan was significantly reduced in presence of actinomycin D (Fig. 4E). The combination of topotecan and decitabine was also synergistic and the IC50 for decitabine was significantly reduced in the presence of topotecan and vice versa (Fig. 4F-H). Unfortunately we were unable to test the synergistic effect of topotecan and decitabine in our xenograft mice, because of the very strong toxicity of decitabine in our NSG mice even at low concentrations. For the combination of topotecan with actinomycin D dosing tests in our xenograft are needed to avoid severe toxicity under long-term treatment with actinomycin D.Since our drug screen showed good responses in vitro and in vivo for several topoisomerase inhibitors we wanted to further investigate the potential therapeutic effect of two of these, topotecan and doxorubicin, in a multi-agent treatment setting. Topotecan is currently being applied in pediatrics for soft-tissue sarcoma, but is rarely used for brain tumors so far, while doxorubicin is thought to play an important role in the treatment of rhabdoid tumors, such as AT/RT12,13.

The efficacy of these two drugs was tested in a multimodal treatment regimen in combination with the conventional chemotherapeutics vincristine and methotrexate in a two- weeks-schedule (Fig. 5A). Vincristine is the only conventional chemotherapeutic that showed a very good IC50 in our in vitro screen and methotrexate is currently applied as a backbone in a clinical trial for recurrent or progressive malignant brain tumors, including AT/RT and ETMR (NCT02684071). This treatment schedule was repeated until animals showed signs of toxicity or were moribund. In the topotecan combined chemotherapy (TCC) group animals had a significant longer survival compared to animals in the doxorubicin combined chemotherapy (DCC) group (Fig. 5B,C). Moreover, in both treatment regimens we saw an improvement in survival as compared to the monotherapy treatments (Fig. 3A-B, 5B-C). DCC treated mice had a median survival of 34 days, whereas the doxorubicin monotherapy showed a median survival of only 27.5 days. For TCC mice the median survival was 68 days compared to 55.5 days in topotecan monotherapy treatment. Furthermore the tumor growth in TCC mice was slower than in DCC mice detected by bioluminescence imaging (Fig. 5D-F).To validate our drug screening results we used our ETMR PDX model NCH3602. We obtained two PDX tumors from transplantation of the patient-derived tumor cells and performed an in vitro drug screen with the dissociated tumor cells after sacrificing the animals. We tested efficacy of our top compounds and standard chemotherapeutic drugs and could show a very good drug response with vincristine, panobinostat, doxorubicin and topotecan, all with an IC50 well below 500 nM (Fig. 6 and Supplementary Fig. S6A,B). MLN0128 and volasertib had also a good response with an IC50 of 1,319 nM and 3,100 nM, respectively. The standard chemotherapeutics, such as carboplatin, etoposide and methotrexate showed no response in NCH3602 cells, confirming our results in the screen with the BT183 cells. Surprisingly, alisertib and decitabine, both effective at low concentrations in BT183 cells, showed only a low or no effect at all in NCH3602 cells. The changes of IC50 in this screen compared to our initial screen is probably due to a heterogeneous cell population in the wells, including the presence of murine cells and a sub- optimal condition of the cells, due to the tumor cell isolation. Still, we could confirm most of our screening results in this new ETMR cell source supporting our findings that topoisomerase I and II, PLK1 and mTOR inhibitors and epigenetic compounds are promising drug candidates in ETMR.

DISCUSSION
Despite the aggressive chemotherapeutic intervention in ETMR patients, we are still facing a very poor clinical outcome and there is a need to reconsider and re-evaluate current treatment protocols. A previous study from Spence et al. already identified several potential drug candidates, including mTOR/PI3K/IGF1R, topoisomerase and HDAC inhibitors, but these were only tested in vitro7. To follow up on this study and to validate potential ETMR-specific drug targets we have performed a detailed pre-clinical in vitro and in vivo drug screen with the BT183 cell line and in the newly patient-derived xenograft cells NCH3602. In line with the study of Spence et al. we also identified topoisomerase inhibitors, HDAC inhibitors and compounds targeting the mTOR pathway to be effective in vitro at low concentrations, but we also identified several other candidate drugs like actinomycin D, alisertib for Aurora kinase inhibition and volasertib for PLK1 inhibition. Based on past clinical experience, it was not surprising that most of the conventional chemotherapeutics in our screen had no or little effect on the BT183 cells. Notably, actinomycin D, doxorubicin and topotecan, all effective at low nanomolar concentrations in vitro, significantly increase survival of mice in single or multi-agent treatments, but are all not applied in ETMR patients to date. One of the new drugs identified in the in vitro screen was the PLK1 inhibitor volasertib, which also significantly increased survival of mice and slowed down tumor growth in comparison to vehicle treated animals. However, eventually all tumors progressed under the various treatments suggesting that drugs identified in our study should be combined with other compounds or treatment modalities. In vitro we found a synergistic effect for the combination of topotecan and decitabine, and of topotecan and actinomycin D. Both are promising drug combinations for future pre-clinical and clinical studies. We were unable to study the effect of topotecan and decitabine in vivo due to the severe toxicity of decitabine in the NSG mice. It is possible that the observed toxicity is mouse strain related, because decitabine is used already in pediatric therapies, for example in high-risk relapsed or refractory acute myeloid leukemia (AML)19.

Treating ETMR xenograft animals with a multimodal chemotherapy including methotrexate, vincristine and topotecan or doxorubicin resulted in an increased survival for both treatment regimens compared to the vehicle animals. Most importantly, topotecan multimodal chemotherapy performed better than the doxorubicin multimodal chemotherapy, indicating that topotecan, as a topoisomerase I inhibitor, might be the better choice than doxorubicin in a multimodal ETMR treatment, even as an upfront medication alone. Also SN-38, the active metabolite of irinotecan, a topoisomerase I inhibitor as well, performed very well in our in vitro drug screen and should be considered for future pre-clinical testing. The overall good performance of topoisomerase inhibition in ETMR cells might indicate a loss or impairment of DNA strand break repair function in ETMR, which will be of high interest for further work studying ETMR biology.Our study also highlighted the treatment potential of some of the epigenetic drugs such as panobinostat and decitabine, and targeted drugs like MLN0128 for mTOR inhibition and alisertib for aurora A kinase inhibition. Due to toxicity issues and difficulties in drug application in NSG mice for decitabine, panobinostat and MLN0128, we were unable to see a possible benefit in tumor-bearing mice. The in vivo efficacy of these drugs needs to be followed up in a more detailed toxicity screen, probably even in other mouse strains that are less sensitive to the compounds. Two other drug candidates shown to be very effective in vitro, bortezomib and nanaomycin A, might be of high interest for future treatment strategies in ETMR patients. Unfortunately bortezomib does not pass the blood-brain barrier20,21. Nanaomycin A, a specific DNMT3B inhibitor was not tested by us in vivo yet, because of unknown dosing schedules and no experience in pre-clinical and clinical studies. DNMT3B expression is exclusively high in ETMR and has been implicated already in the downstream signaling pathway of the TTHY1- C19MC fusion and amplification of the miRNA cluster, which makes it a good potential candidate for ETMR treatment5.

In conclusion, we have demonstrated that ETMRs are very sensitive to PLK1 inhibition by volasertib, topoisomerase I and II inhibition by topotecan and doxorubicin and the transcription inhibitor actinomycin D. All these drugs are already approved or in clinical trials for malignancies in pediatric oncology and priority should thus be given to include them in multi- modal therapies in future clinical trials for ETMR patients. Interestingly, a Phase II clinical trial in pediatric neurooncology has already been started testing the use of methotrexate, cyclophosphamide and topotecan in recurrent or progressive malignant brain tumors, including AT/RT and ETMR (NCT02684071). Still, more pre-clinical studies are needed and other drug combinations and treatment modalities have to be evaluated, but including topotecan, volasertib and actinomycin D in current therapies is an important first step in order to improve the clinical course of ETMR MLN0128 patients.