Cancer Cell Article Pemetrexed and Gemcitabine as Combination Therapy for the Treatment of Group3 Medulloblastoma Marie Morfouace,1 Anang Shelat,2 Megan Jacus,3 Burgess B. Freeman III,4 David Turner,3 Sarah Robinson,1 Frederique Zindy,1 Yong-Dong Wang,5 David Finkelstein,5 Olivier Ayrault,6 Laure Bihannic,6 Stephanie Puget,7 Xiao-Nan Li,8 James M. Olson,9 Giles W. Robinson,10 R. Kiplin Guy,2 Clinton F. Stewart,3 Amar Gajjar,10 and Martine F. Roussel1,* 1Department of Tumor Cell Biology 2Department of Chemical Biology and Therapeutics 3Department of Pharmaceutical Sciences 4Department of Preclinical Pharmacokinetics 5Department of Computational Biology St. Jude Children’s Research Hospital, Memphis, TN 38105, USA 6Institut Curie/CNRS UMR 3306/INSERM U1005 Building 110, Centre Universitaire, 91405 Orsay, Cedex, France 7AP-HP, Department of Neurosurgery, Necker-Enfants Malades Hospital, Universite´ Rene Descartes, 75015 Paris, France 8Department of Pediatrics, Section of Hematology-Oncology, Baylor College of Medicine, Houston, TX 77030, USA 9Department of Pediatric Hematology-Oncology, Fred Hutchinson Cancer Research Center, Seattle, WA 98109, USA 10Department of NeuroOncology, St. Jude Children’s Research Hospital, Memphis, TN 38105, USA *Correspondence: [email protected] http://dx.doi.org/10.1016/j.ccr.2014.02.009 SUMMARY We devised a high-throughput, cell-based assay to identify compounds to treat Group3 medulloblastoma (G3 MB). Mouse G3 MBs neurospheres were screened against a library of approximately 7,000 compounds including US Food and Drug Administration-approved drugs. We found that pemetrexed and gemcitabine preferentially inhibited G3 MB proliferation in vitro compared to control neurospheres and substantially in- hibited G3 MB proliferation in vivo. When combined, these two drugs significantly increased survival of mice bearing cortical implants of mouse and human G3 MBs that overexpress MYC compared to each agent alone, while having little effect on mouse MBs of the sonic hedgehog subgroup. Our findings strongly suggest that combination therapy with pemetrexed and gemcitabine is a promising treatment for G3 MBs. INTRODUCTION patients with WNT MB tend to be older, female, and to uniformly survive with current therapy (Northcott et al., 2012). In contrast, Medulloblastoma (MB), a tumor of the posterior fossa, is primar- patients with G3 MB tend to be younger, male, have anaplastic ily a pediatric disease, although it occasionally occurs in adults histology, exhibit a higher incidence of metastatic disease, and (Ellison et al., 2011). MB is classified into four major subgroups have a poor prognosis (Dubuc et al., 2013; Kool et al., 2012; Par- based on clinical and molecular profiles (Taylor et al., 2012). sons et al., 2011; Pugh et al., 2012; Rausch et al., 2012; Robin- Two subgroups exhibit constitutive activation of the sonic son et al., 2012). One characteristic feature of G3 MB is their hedgehog (SHH) or wingless (WNT) developmental pathways. high MYC expression in >75% of cases. Indeed, one study The other two are referred to as group 3 (G3) and G4. Impor- shows that MYC expression is elevated in 20 of 26 (77%) G3 tantly, the molecular subgrouping of these tumors also relates MBs (named group C), compared to 1 of 35 (3%) G4 MBs (named to distinct patient demographics, histologic classification, so- group D; Northcott et al., 2011). Whole genome sequencing matic genetic variations, and clinical outcome. For example, studies on 17 G3 MBs reveals only one tumor (5.9%) with true Significance Despite the recent identification of four molecular groups of human MB, patients are currently treated with similar chemo- therapies independent of classification. G3 MB has a high incidence of metastasis and poor prognosis. Thus, more effective therapeutic approaches for these patients are desperately needed. The development of a mouse model of G3 MB enabled production of cultured neurospheres that provided an ideal platform to identify additional chemotherapies. We found two FDA-approved drugs that significantly inhibited mouse and human G3 MB neurosphere cultures, mouse allografts, and xenografts from G3 MB primary patient samples, but not mouse SHH MBs. These findings provide a strong rationale for combination therapy with pemetrexed and gemcitabine to treat patients with G3 MB. 516 Cancer Cell 25, 516–529, April 14, 2014 ª2014 Elsevier Inc. Cancer Cell Pemetrexed and Gemcitabine in G3 Medulloblastoma MYC amplification, whereas two other tumors display an aber- (10 mM) in triplicate. Z-prime and other assay diagnostics were rant copy number gain of the MYC gene. However, 15 of these acceptable (Figure S1C) and the scatterplot of controls and 17 (88%) G3 MBs demonstrate high MYC expression (Robinson compound activities showed adequate separation between et al., 2012). Despite these pronounced differences, patients signal and noise for both Myc1 and Trp53-null (Figure S1D, with MB are typically treated with uniform surgery, radiotherapy, left). Receiver operator characteristic analysis indicated that and adjuvant chemotherapy including vincristine, cisplatin, and the assay demonstrated acceptable discriminatory power be- cyclophosphamide (Packer et al., 2013) or lomustine and carbo- tween true-positive and true-negative results, with the area platin (Massimino et al., 2012). These therapies fail to cure one- under the curve (AUC) >0.8 for both lines, and that an assay cut- third of all patients and carry widespread morbidities that impair off of >50% returned 70% of all true-positive results for Myc1 survivor’s quality of life. (Figure S1D, right). A total of 690 of the 7,389 compounds, Molecular subgrouping has the potential to improve risk strat- including all with inhibition >50% in the primary screen with ification and tailor therapy to reduce toxicities to potential survi- Myc1, analogs of these hits, and other compounds of interest, vors. Paramount to this strategy is the development of accurate were tested in dose-response experiments in triplicate using models that recapitulate the subgroups for preclinical therapeu- concentrations ranging from 4 nM to 10 mM. Of the 690 hits, tic testing. We developed a mouse model of G3 MB through we identified 65 compounds with potency <1 mM against Myc the orthotopic transplantation of transgenic cerebellar granule 1(Figure 1; Table S1). neuronal progenitors (GNPs) in the cortices of naive recipient animals. GNPs were purified by percoll density gradient from Screening Results of FDA-Approved Drugs the cerebella of 5- to 7-day-old Trp53À/À;Cdkn2cÀ/À mice and To accelerate the transition of potential therapeutics into the infected with retroviruses encoding MYC (Kawauchi et al., clinic, we prioritized the FDA-approved drugs with oncology in- 2012). MB tumors develop within 30 days of transplantation dications and potencies below or near 1 mM for further study. with as few as 100 tumor cells that recapitulate the high level We conducted dose-response experiments on 35 FDA- of trimethylation of histone H3 at lysine 27 (H3K27me3) seen in approved drugs in triplicate on Myc1, Trp53-null, HepG2, and human G3 and G4 (Robinson et al., 2012). Tumors grow as neu- BJ cell lines (Figure S2A; Table 1). Drugs were grouped into rospheres that, when transplanted into cortices of recipient nine activity classes: folate pathway inhibitors, other inhibitors mice, induce secondary MBs that mimic the primary tumors of DNA/RNA synthesis, purine antimetabolites, microtubule in- (Kawauchi et al., 2012). Because neurospheres can be passaged hibitors, sterol biosynthesis inhibitors, topoisomerase inhibitors, repeatedly while maintaining their functional and molecular epigenetic regulators, proteasome inhibitors, and RNA polymer- properties, they provide a unique platform to conduct screens ase inhibitors. Myc1 was extremely sensitive to folate pathway of compounds to identify those with therapeutic potential against inhibitors with pemetrexed, methotrexate, and raltitrexed being human G3 MB. We here report the outcome of screening a the most selective. The DNA/RNA synthesis inhibitor gemcita- library of compounds that included US Food and Drug Adminis- bine was highly potent in both Myc1 and Trp53-null, but tration (FDA)-approved drugs and candidate compounds in achieved 100% efficacy in only Myc1. The purine antimetabolite development. cladribine showed promising activity against Myc1, but a nar- rower therapeutic window against the control cell lines HepG2 RESULTS and BJ compared to other drugs. The proteasome inhibitor bortezomib and the RNA polymerase inhibitor dactinomycin High-Throughput Screen Using Mouse G3 MB were equipotent and equally efficacious against both Myc1 Neurospheres and Trp53 null. Microtubule inhibitors, including vincristine and Tumor cells purified from several independently derived primary vinblastine, and topoisomerase inhibitors, such as doxorubicin, mouse G3 MBs were grown as neurospheres for four to five etoposide, and topotecan, were active in G3 MB. Because they passages providing lines with comparable cell proliferation char- are already used in the clinic for the treatment of MB, they were acteristics. Two lines, derived from independent tumors and not considered further. Cerivastatin and fluvastatin, two sterol infections, were hereafter referred to as ‘‘Myc1’’ and ‘‘Myc2.’’ biosynthesis inhibitors targeting 3-hydroxy-3-methyl-glutaryl- Neurospheres from the cerebellum of 7-day-old (P7) Trp53À/À; coenzyme A reductase, had high efficacy but low potency rela- Cdkn2cÀ/À mice (hereafter