Heterogeneous extrachromosomal amplification of mutant PDGFRA is associated with an aggressive phenotype in glioblastoma
Berezovsky AD, Transou AD, Irtenkauf SM, Hasselbach LA, Koeman J, Kim H, Verhaak RG, Mikkelsen T, Poisson LM, and DeCarvalho AC. Heterogeneous extrachromosomal amplification of mutant PDGFRA is associated with an aggressive phenotype in glioblastoma. Cancer Res 2018; 78(13).
Background & Objective: Receptor tyrosine kinase (RTK) signaling is altered in over 80% of glioblastoma (GBM) cases, frequently through gene amplification. About 14% of GBMs carry amplification in the gene coding for platelet-derived growth factor receptor A (PDGFRA), according to TCGA. PDGFRA plays a key role in brain development and is associated with GBM proneural (PN) subtype. Despite the frequency of oncogenic RTK signaling in GBMs, RTK inhibitors have not yet achieved sufficient efficacy in clinical trials to earn FDA approval. Imatinib, a multi-kinase inhibitor that targets PDGFRA, has not shown efficacy in earlier clinical trials for GBM, in which amplification status was not an inclusion criteria. It has been reported that treatment of GBMs carrying extrachromosomal (ecDNA) amplification with EGFR inhibitors leads to a decrease in ecDNA copy number as a mechanism of resistance. Here, our objective was to assess the heterogeneity of ecDNA PDGFRA amplification in a newly diagnosed PN GBM tumor (HF3253), and to evaluate whether similar modulation of ecDNA copy number could be attained by Imatinib treatment of patient-derived xenografts (PDX). Experimental Approaches & Results: PDGFRA amplification was detected by low pass whole genome sequencing and confirmed to be extrachromosomal by fluorescent in situ hybridization (FISH) in metaphase spreads using PDGFRA and centromere 4 labeled DNA probes. The amplified PDGFRA also harbored a novel missense mutation corresponding to the extracellular domain. PDGFRA FISH signals/number ranged from 3 to 100 in the HF3253 samples, with high signal (> 20) evident in 67%, 39%, and 43% and low amplification signal (6 < x < 9) evident in only 0%, 6%, and 19% of nuclei from biopsy, neurosphere, and xenograft respectively. HF3253 neurospheres were orthotopically implanted into the brains of immunocompromised nude mice. Imatinib mesylate was administered by oral gavage at a 75 mg/kg/day dosage in 5-day cycles with 2-day drug holiday intervals, starting 2 weeks post implant. Control mice received vehicle gavage under the same schedule. Mice were monitored daily and sacrificed on the basis of weight loss or neurological symptoms. HF3253 PDX treated with Imatinib mesylate did not have a significant survival advantage (median survival: 47.2 days) relative to control mice (median survival: 45.8 days; log rank test p value = 0.7825). The untreated and treated PDXs exhibited high levels of phospho-PDGFRA by IHC. Two independent in vitro dose response assays showed no difference in the IC50 for Imatinib between PDGFRA positive and negative GBM neurospheres: (9.447 μ M, 8.384 μ M) and (8.972 μ M, 6.896 μ M). Conclusions: We have developed a patient derived model of glioblastoma that retains ecDNA PDGFRA amplification and high levels of expression and RTK activation. Although PDGFRA is a driver of malignancy, our results show that better inhibitors are needed.