VEGF as a therapeutic target The VEGF-A inhibitor bevacizumab is approved in 11 cancer types, but has failed to improve overall survival in 9 of these.


Dec. 2, 2019


ASCO - November 07, 2019.

Does this suggest we should abandon PFS for evaluating new cancer drugs? Specific problems with the use of overall survival arise with patient cross-over (an ethical imperative for clearly efficacious therapies) and in trials of early line therapies in cancers with multiple treatment options such as breast & ovary.

In a recent article in Journal of Clinical Oncology, Tewari et al1 provided high-quality data on the final protocol-specified analysis of overall survival (OS) of 1,873 women treated with chemotherapy and/or bevacizumab for incompletely resected stage III or IV ovarian cancer. This randomized, three-arm, double-blind, placebo-controlled phase III trial (GOG-0218) revealed that after a median follow-up of nearly 9 years, there was no OS benefit from bevacizumab. Women receiving carboplatin and paclitaxel had median survival times equivalent to those receiving concurrent carboplatin, paclitaxel, and bevacizumab and those receiving concurrent carboplatin, paclitaxel, and bevacizumab plus bevacizumab maintenance (hazard ratios [HRs], 1.06 and 0.96; P = .34 and .53, respectively). These OS data are congruent with three earlier phase III trials for this indication: ICON7, AURELIA, and OCEANS (HRs, 0.99, 0.85, and 0.95; P = .85, .17, and .65, respectively). All four trials yielded unequivocal outcomes using the gold standard of OS. The high-probability conclusion is that bevacizumab-based targeting of vascular endothelial growth factor (VEGF) in ovarian cancer fails to increase median lifespan. The additional clinical concern is that bevacizumab contributes to many adverse events, including thromboembolism, GI perforation, hypertension, and proteinuria. These data have important implications for other more prevalent and incurable cancer types,2-4 where the lowered bar of progression-free survival (PFS) has controversially garnered US Food and Drug Administration approval for the use of bevacizumab.

In metastatic breast cancer, for example, all five major randomized bevacizumab-based phase III trials (AVF2119g, E2100, AVADO, RIBBON-1, and RIBBON-2) revealed no significant increase in OS (median HRs between 0.85 and 1.05; P values between .16 and .83). In retrospect, Sledge5 reflected that although “bevacizumab…electrified the field in 2005 with the E2100 trial, which showed a doubling in progression-free survival…why has VEGF-based therapy for breast cancer failed so completely?”5(p133) One reason has remained hidden in plain view. In this pivotal E2100 trial, independent radiologists disagreed 50% of the time as to whether tumor progression had even occurred, starkly exposing the severe limitations of PFS as a proxy for OS, where ambiguity is absent. Sledge concluded that “perhaps it’s time for the field to pack its bags and move on, but perhaps not.”5(p134) The following clinical trials bear on this important question.

In metastatic castration-resistant prostate cancer, the only relevant randomized, double-blind, placebo-controlled phase III trial of bevacizumab is CALGB 90401, where 1,050 men were treated with docetaxel and prednisone with or without bevacizumab. Bevacizumab failed to increase median OS (HR, 0.91; P = .18).

Furthermore, bevacizumab did not provide an OS benefit in metastatic pancreatic cancer (CALGB 80303), metastatic renal cancer (AVOREN and CALGB 90206), metastatic gastric cancer (AVAGAST), glioblastoma (AVAglio and TROG 08254), or non–small-cell lung cancer trials (AVAiL and BeTa) or in five of seven colorectal trials (NO16966, NSABP C-08, ML18147, and MAX). To date, an OS benefit has only occurred in metastatic cervical cancer (GOG 24O6) and unresectable pleural mesothelioma (MAPS7).

The situation at the physician-patient interface is that bevacizumab-based VEGF therapy has failed to increase OS in nine of 11 indications and yet continues to be used in trials and off trial. Since its first approval, sales of bevacizumab have far exceeded US$50 billion in drug costs alone, which represents a financial burden on health care systems worldwide. The latest exemplary ovarian publication1 and those from other cancers highlight the dangers of using PFS versus OS in determining drug efficacy. On the basis of the OS gold standard, the angiogenesis-based VEGF hypothesis is on life support. This is hardly unexpected, because extrapolating the basic molecular and cellular foundations of ligand interactions with normal endothelial cells to the outputs from tumors is hazardous. Perturbation of the multiple isoforms of VEGF will rarely disturb the interactions from heterogeneous tumor outputs and their endothelial targets, especially because tumors themselves adapt by constructing abnormal tumor-derived endothelial cells8 and hence bypass most molecular roadblocks.

Given the preclinical and clinical data described here, do the authors agree that additional large randomized trials in these nine metastatic cancers are not justified, given the lack of OS benefit for bevacizumab combined with its clinical toxicity, and that although surgery, chemotherapy, and radiation treatment all relieve the cancer burden, the exploration of novel approaches in immunotherapy and personalized radiation treatment based on metallomics9 is even more pressing? Should we in the oncology profession now try to disregard the inferior PFS end point and rely solely on OS whenever possible?