P11.36.A Could a transcriptome profile predict local control for glioblastoma?

Abstract

Abstract Background Glioblastoma (GB) is the most frequent and aggressive primary brain tumor in adults. The standard of care is surgery followed by radiotherapy in combination with temozolomide. GB molecular subtypes (Classical [CL], Mesenchimal [MES] and Proneural [PN]) have been correlated with prognosis. In the preclinical setting, CL and MES glioma stem-like initiating cells have also been shown to be more radioresistant than PN subtype. If confirmed in patients, the molecular subtype could predict the specific benefit of adjuvant treatment. The aim of this work is therefore to investigate whether GB molecular profiles are associated with treatment resistance in vivo. Material and Methods We analyzed the spatiotemporal pattern of recurrence (progression-free survival [PFS] and local control) in 37 GB patients treated with adjuvant radiotherapy (30 x 2 Gy, following EORTC contouring guidelines) plus temozolomide in the context of a multi-centric study. Relapse was defined following the RANO criteria (gadolinium enhanced area on MRI T1 sequence) and classified into 3 groups: in-field relapse (Dmean > 57 Gy), out-of-field relapse (Dmean < 48 Gy) and not-defined relapse (no follow-up MR data or marginal relapse). We used three different classification algorithms (SVM, K-Nearest Neighbor and single sample gene set enrichment analysis) for the transcriptomic classification of tumor samples, and we only assigned a molecular profile when all three classifications agreed, otherwise we called it Mixed profile. Patients with a not-defined spatial pattern of recurrence were included in the PFS analysis, but excluded from the local control analysis. Results The transcriptomic classification resulted in 14 CL, 1 MES, 9 PN and 13 Mixed. Median PFS after radiotherapy was 6 months, with a mean follow-up of 11 months. No differences in PFS were found between CL, PN and Mixed groups (the single MES case was excluded). Local control was analyzed in 24 patients (10 CL, 4 PN, 10 Mixed). Non- statistically significant differences were found between PN/Mixed and CL subtypes, with a median in-field relapse-free survival of 6.9 months (PN), 6.3 months (Mixed) and 4.7 months (CL). Conclusion Although there is no difference in PFS among GB molecular subtypes, these results suggest that the Classical subtypes seem to relapse faster inside the radiotherapy field. This would suggest a higher radioresistance compared to Proneural and Mixed subtypes, in agreement with preclinical data. Due to the low number of patients these differences are statistically non-significant, but they provide the rationale for further investigation in a larger patient cohort. Grant: Marató -TV3