INNV-41. PROSPECTIVE SELECTION OF RECURRENT GLIOBLASTOMA PATIENTS FOR TAILORED THERAPIES. RESULTS OF AN INSTITUTIONAL EXPERIENCE

Abstract

Abstract INTRODUCTION Failure of clinical trials with targeted therapies in glioblastoma (GBM) is probably due to the enrollment of molecularly unselected patients. In preliminary studies, we prospectively selected recurrent GBM patients on the basis of molecular pattern and administered targeted therapy accordingly. This tailored approach gave encouraging results in term of low recurrence rate (RR) and high 6-month progression free survival (PFS-6). Here, we present the long-term results of our work. METHODS On recurrent tumor samples of 34 adult patients, we assessed the expression of VEGF and PTEN through immunohistochemistry and of EGFRvIII through RT-PCR. Patients with VEGF overexpression were treated with bevacizumab (10 mg/Kg i.v. every 2 weeks in 6-week cycles). Patients with EGFRvIII expression and normal PTEN expression added erlotinib (150 mg/day orally). Patients with loss of PTEN expression, irrespective of EGFRvIII status, added sirolimus (1–10 mg/day orally). RESULTS Sixteen patients received bevacizumab alone (bev), 14 bevacizumab plus erlotinib (bev+erl) and 4 bevacizumab plus sirolimus (bev+sir). RR was 50% in the whole cohort and 37.5%, 57.1% and 75% in bev, bev+erl and bev+sir groups, respectively. PFS-6 was 55.9% in the whole cohort and 50%, 64.3% and 50% in bev, bev+erl and bev+sir groups, respectively. Our data compare favorably with those from the large EORTC 26101 trial, which showed a RR of 41.5% and a PFS-6 lower than 30%. When considering sustained response (defined as PFS ≥ 12 months), we overall observed a 20.6% rate, with the highest value in bev+erl subgroup (28.6%). Interestingly, in EORTC 26101 trial, less than 10% of patients achieved PFS ≥ 12 months. CONCLUSIONS Our results confirm that the tailored approach in recurrent GBM provides an advantage in terms of RR, PFS and, above all, of long-term responses, compared with trials without molecular selection.