AT-17 RE-ANALYSIS OF PFS/RESPONSE USING ORIGINAL MACDONALD CRITERIA AND RESPONSE EVALUATION CRITERIA IN SOLID TUMORS IN THE PHASE III AVAGLIO STUDY OF BEVACIZUMAB PLUS RADIOTHERAPY AND TEMOZOLOMIDE IN NEWLY DIAGNOSED GLIOBLASTOMA

BACKGROUND: AVAglio: bevacizumab (BEV) plus radiotherapy (RT) plus temozolomide (TMZ) extended investigator-assessed PFS versus placebo + RT/TMZ (HR = 0.64, p < 0.0001, median 10.6 vs 6.2 months). AVAglio response criteria were similar to Response Assessment in Neuro-Oncology (i.e. included T2/fluid-attenuated inversion recovery [FLAIR] imaging for determination of response/progression). The independent review facility algorithm used in AVAglio (adapted Macdonald criteria [Macdonald-A]), original Macdonald (Macdonald1990) criteria, and RECIST, were strictly applied, retrospectively, to the investigator-assessed AVAglio data. METHODS: Patients (n = 921) received: BEV + RT/TMZ or placebo + RT/TMZ (6wks); 28-day break; BEV + TMZ or placebo + TMZ (x6); BEV or placebo until PD/unacceptable toxicity. Case report form data were retrospectively reassessed by: Macdonald-A (PD: greater than or equal to 25% change in sum of product of diameters of measurable enhancing lesion, unequivocal progression of non-measurable/non-enhancing lesions, new lesion[s], neurologic worsening; Macdonald1990 (PD: as for Macdonald-A, but without non-measurable/non-enhancing lesion assessment); RECIST v1.1 (PD: greater than or equal to 20% and >5mm increase in sum of longest diameters of target lesions, unequivocal progression of non-target lesions, non-measurable/non-enhancing lesions, new lesion[s]). RESULTS: BEV + RT/TMZ versus placebo + RT/TMZ, respectively. Median PFS: Macdonald-A, 10.0 vs 6.0 months (HR = 0.67); Macdonald1990, 10.1 vs 6.0 months (HR = 0.67); RECIST, 11.0 vs 6.7 months (HR = 0.64). In patients with Macdonald-A PD, 316/338 (93.5%) and 336/357 (94.1%) had Macdonald1990 PD, 22/338 (6.5%) and 21/357 (5.9%) did not have Macdonald1990 PD, and no patient had Macdonald1990 PD determined at an earlier time. In patients with Macdonald-A PD, 229/338 (67.8%) and 240/357 (67.2%) had RECIST PD, 101/338 (29.9%) and 93/357 (26.1%) did not have RECIST PD, and 8/338 (2.4%) and 24/357 (6.7%) had RECIST PD determined at an earlier time than Macdonald-A. CONCLUSION: This analysis is limited in its conclusions as it was retrospective and there were no scans collected after PD was determined by the investigator. These results confirm the gain in sensitivity of the progression assessment by the introduction of bi-dimensional measurements and neurologic evaluation.