Abstract
Medulloblastoma is the most common malignant pediatric brain tumor. Survival rates vary widely between subgroups, with an average overall survival of 70%. Recurrent medulloblastoma is highly aggressive, treatment-resistant, and usually fatal. In addition, current treatments are highly toxic to the developing brain and surviving patients suffer from lifelong side effects. Therefore, novel therapeutic options are urgently needed. To inform risk-based, personalized therapy, we developed a novel platform called DrugSeq, which allows predictions of drug sensitivities in patients across medulloblastoma subgroups. We used a perturbagen-response dataset to calculate transcriptional response signatures for each drug and compared this to patient medulloblastoma tumor gene expression. We then stratified patients by molecular subgroup and used an ANOVA analysis to identify drugs that selectively targeted each subgroup. We found distinct differences in transcriptional profiles and predicted drug sensitivity for each medulloblastoma subgroup. We identified kinase inhibitors, epigenetic inhibitors, and several drugs that have been investigated in drug repositioning studies for cancer. We posit that DrugSeq may identify novel therapies and facilitate patient stratification in clinical trials, leading to more successful targeted medulloblastoma therapies that improve tumor response while minimizing late toxicities. This computational tool can also be used for other cancers to stratify patients based on any clinical or molecular feature.