PATH-15. PROTEOMIC SIGNATURES PREDICT GRADE IN PEDIATRIC AND YOUNG ADULT INFILTRATIVE ASTROCYTOMAS

BACKGROUND: Infiltrative astrocytomas in children and young adults pose a treatment challenge due to the difficulty of achieving gross total resection and tumor resistance to irradiation and chemotherapy. Histopathologic grade is an essential part of determining prognosis and treatment, but it is subjective and provides limited understanding of the molecular mechanisms underlying tumor development and progression. METHODS: We performed liquid chromatography/mass spectrometry (LC/MS-MS) on 28 FFPE samples of primary infiltrative astrocytomas (10 grade II, 8 grade III and 10 grade IV – WHO classification) from Nationwide Children’s Hospital (NCH). Initial unsupervised clustering was performed. Lasso regression yielded a protein signature separating low- and high-grade tumors which was validated using a similar cohort of pediatric and young adult infiltrative astrocytomas from the Proteomic Data Commons (PDC) (n=28) of the National Cancer Institute. RESULTS: Unsupervised clustering of NCH samples essentially recapitulated grade and lasso regression yielded a 10-protein signature that distinguished grade II from grade III/IV tumors. This 10-protein signature when applied to the PDC validation dataset, accurately predicted grade for 89.3% of the tumors (p=0.00014). CONCLUSIONS: We identified a quantitative protein signature that can reliably distinguish between low- and high-grade infiltrative astrocytomas from FFPE tissue. Further validation will enable the development an objective prognostic proteomic clinical test that complements and may outperform current histopathological strategies. Additionally, proteomic profiling of tumors will clarify the molecular mechanisms contributing to treatment resistance and tumor progression and help identify novel treatment targets. Independent functional validation and characterization of proteins is ongoing.