Create Account | Sign In: Author or Forum

Search Symptoms

Category: Gynecology | Oncology | Pharmacy | Research | Surgery

Back to Articles

Bevacizumab for breast cancer: FDA grants a stay of execution

Author: Dr. A. De la Guerra | Submitted: January 19, 2011. Updated: January 26, 2011.

Dr. A. De la Guerra's avatar

Authors’ disclosures of potential conflicts of interest and author contributions are found at the end of this article.

ABSTRACT: The Food and Drug Administration has after repeated delays proposed removing metastatic breast cancer as an indication from the Avastin label. The proposal is being contested by Roche/Genentech, the drug’s maker. In this article we review the evidence for and against this decision.

The US Food and Drug Administration (FDA) has delayed an expected decision on whether to withdraw approval of Bevacizumab (Avastin®; Genentech) for the treatment of metastatic breast cancer (MBC). The FDA’s resolution was anticipated on September 17, 2010; on that same day Genentech announced the FDA’s decision to extend the review to December 17 in a press release [1]:

“South San Francisco, Calif. -- September 17, 2010 -- Genentech, Inc., a member of the Roche Group (SIX: RO, ROG; OTCQX: RHHBY), today announced that information submitted by the company to the U.S. Food and Drug Administration (FDA) during the review of the supplemental Biologics License Applications (sBLAs) for Avastin® (bevacizumab) for previously untreated (first-line) advanced HER2-negative breast cancer has been deemed a major amendment. In accordance with FDA guidelines, the agency has extended the review period for the sBLAs by an additional 90 days. The company now anticipates FDA action on the sBLAs by December 17, 2010.”

Genentech did not go into details about the new data it had submitted to the FDA; likewise, there was no official press announcement on FDA’s newsroom portal (updated on 10/01/2010). The news had hit the headlines and bevacizumab for breast cancer has since become hotly debated on several media outlets [2-4].


Angiogenesis, the formation of new vasculature from existing blood vessels, is a key process in tumor growth, invasion, and metastasis [5].  Vascular endothelial growth factor A - VEGF-A (known as VEGF) is an important mediator of tumor angiogenesis and is commonly overexpressed in breast tumors [5, 6]. Because cancer cells produce VEGF to stimulate new blood vessel formation, VEGF is an important target of anticancer therapy [7].

Bevacizumab is a recombinant monoclonal antibody (MAb) directed against the VEGF-A ligand by binding circulating VEGF, which inhibits binding to its receptor (VEGFR), thereby preventing activation of signaling cascades to angiogenesis [7]. Bevacizumab was the first VEGF targeting drug approved for cancer therapy. It is currently approved for the following indications in combination with chemotherapy: (a) the treatment of advanced non-small cell lung cancer; (b) advanced renal cell cancer; (c) advanced colorectal cancer; and (d) metastatic breast cancer. Bevacizumab also has approval for use as a single agent in second-line treatment of advanced glioblastoma [8, 9].

Accelerated approval for Bevacizumab in MBC

On February 2008, the FDA granted accelerated approval for Bevacizumab as first-line therapy combined with paclitaxel, for patients with HER2-negative MBC [10]. Approval was based on the results of the Eastern Cooperative Oncology Group trial E2100 [11]; details of the study design are shown in box 1 below.

Box 1. E2100 study design.

Open-label study of paclitaxel +/- bevacizumab.

Population   :

- HER2 negative.

- Recurrent or MBC.

- No prior chemotherapy.

N = 722        :

- Paclitaxel + bevacizumab (368).

- Paclitaxel (354).

Results of the ECOG E2100 trial
The study showed an improvement in progression free survival (PFS) of 5.9 months, but not in overall survival (OS); nevertheless, in the bevacizumab plus paclitaxel arm there was a 20.2% increase in grade 3 to 5 toxicity and 1.7% increase in treatment related deaths. Outcomes and safety data are shown in tables 1 and 2 respectively.

Table 1. Outcomes of E2100 trial.




PFS (median, months)



OS (median, months)



ORR (%)



QOL improvement *



(*) Evaluated with FACT-B questionnaire.
PAC, paclitaxel; BVZ, bevacizumab; QOL, quality of life; ORR, overall response rate.

Table 2. Safety results of E2100 trial.




Grade 3



Grade 4



Grade 5






Protocol related deaths (n)



PAC, paclitaxel; BVZ, bevacizumab.

The statistical benefit shown in terms of PFS with a hazard ratio (HR) for progression at 0.60 and P <0.001, was not translated into a significant increase in OS. Despite this, the FDA granted approval in opposition to the recommendations of the Oncologic Drugs Advisory Committee (ODAC) members who voted 5 to 4 against its approval. Box 2 shows the review issues that were exposed by the ODAC during the meeting [12].

Box 2. Review issues with ECOG E2100 trial.

Open-label trial.
No benefit on OS.

Lack of confidence in the PFS result:

  • 10% missing scans.
  • 34% not followed until IRC-PFS event or end of study.
  • 51% discordance between IRF and ECOG.

Safety concerns:

  • Inadequate evaluation of toxicity profile:
  • Grade 1 and 2 data not collected.
  • Laboratory data not collected.
  • 20.2 % increase in grade 3 to 5 toxicity.
  • 1.7 % treatment related death in the BVZ arm.

IRF, independent review facility; IRC, independent review committee.
Under the FDA rules, Genentech was obliged to submit further data to prove efficacy and define the magnitude of the clinical benefit of the drug [13].
Title 21: Food and Drugs. Part 601 - 601.41:

“Approval under this section will be subject to the requirement that the applicant study the biological product further, to verify and describe its clinical benefit, where there is uncertainty as to the relation of the surrogate endpoint to clinical benefit, or of the observed clinical benefit to ultimate outcome.”

Confirmation of approval
The ODAC meeting to review the supplemental Biologics License Applications (sBLAs) for bevacizumab was tentatively scheduled for July 20, 2010 [14]. Two placebo controlled trials were identified by the company to support the conversion from accelerated to regular approval: the AVADO (BO17708) and the RIBBON 1 (AVF3694g) trials [15].

FDA review of Bevacizumab in MBC

On July 20, 2010, the ODAC meeting reviewed the sBLAs for bevacizumab, for two proposed indications:

  • STN* 125085\191: First-line treatment of HER2-negative MBC, in combination with paclitaxel or docetaxel.
  1. STN* 125085\192: First-line treatment of HER2-negative MBC, in combination with taxane-based, anthracycline-based or capecitabine chemotherapy.

* BLA Submission Tracking Number.

The AVADO trial [16] was evaluated for the first proposed indication. The study investigated the combination of bevacizumab and docetaxel; details of study design are shown on box 3.

Box 3. AVADO study design.

Randomized, double-blinded, placebo controlled.

Population   :

- HER2 negative.

- Recurrent or MBC.

- No prior chemotherapy.

N = 736        :

- Docetaxel + bevacizumab 7.5 mg/kg (248).

- Docetaxel + bevacizumab 15 mg/kg (247).
- Docetaxel + placebo (241).

Results of the AVADO trial
The AVADO trial showed a statistically significant increase in PFS of less than a month for both bevacizumab arms; stratified HR of 0.70 (p=0.0035) for the 7.5 mg/kg arm and 0.62 (p=0.0001) for the 15 mg/kg arm. Although significant the differences are marginal, with an increased median PFS of 0.82 and 0.88 months for the 7.5 mg/kg and 15mg/kg arms respectively [17]. This time the benefit in PFS on the AVADO trial was smaller than in the E2100 study; and again, there was no impact on OS. Efficacy data is shown in table 3.

Table 3. Efficacy of AVADO trial.


DOC + BVZ 7.5 mg/kg

DOC + BVZ 15 mg/kg


PFS (median, months)




OS (median, months)




ORR (%)




DOC, docetaxel; BVZ, bevacizumab; PL, placebo; ORR, overall response rate.

Assessment of the safety data revealed more adverse events and drug discontinuation in the bevacizumab-containing arms. Grade 3 to 5 toxicity was:

  • 78% in the docetaxel + bevacizumab 7.5 mg/kg arm.
  • 75% in the docetaxel + bevacizumab 15 mg/kg arm.
  • 68% in the docetaxel + placebo arm.

The ODAC concluded that the AVADO trial met their primary endpoint, but with significantly more adverse events and no OS improvement, HR favoring placebo arm on survival [18].
Afterwards, the RIBBON 1 trial [19] was evaluated for the second proposed indication: First-line treatment of HER2-negative MBC, in combination with taxane-based, anthracycline-based or capecitabine chemotherapy. The study was designed to investigate the clinical benefit of combining bevacizumab with different first-line chemotherapy regimens for MBC; details on the study design are shown on box 4.

Box 4. RIBBON 1 study design.

Randomized, double-blinded, placebo controlled.

Population :

- HER2 negative.

- Recurrent or MBC.

- No prior chemotherapy.

N = 1237 :

Taxane/Anthracycline cohort (622):
- Placebo (207)
- Bevacizumab (415)

Capecitabine cohort (615).
- Placebo (206)
- Bevacizumab (409)

Results of the RIBBON 1 trial
The efficacy and safety outcomes from the RIBBON 1 study are similar to those seen in the E2100 and AVADO trials: statistically significant improvement in PFS and ORR, but no effect on OS. In addition, there were more adverse events attributed to bevacizumab. Efficacy in each treatment group is shown separately in tables 4a and b.

Table 4a. RIBBON 1 trial outcomes: Taxane/Anthracycline cohort.





PFS (median, months)



<0.0001 (S)

OS (median, months)



0.8298 (NS)

ORR # evaluated (%)

67 (37.9)

177 (51.3)

0.0054 (S)

TAX, taxane; ANT, anthracycline; BVZ, bevacizumab; PL, placebo; ORR, overall response rate; S, significant; NS, non-significant.

Table 4b. RIBBON 1 trial outcomes: Capecitabine cohort.





PFS (median, months)



0.0002 (S)

OS (median, months)



0.2706 (NS)

ORR # evaluated (%)

38 (23.6)

115 (35.4)

0.0097 (S)

CAP, capecitabine; BVZ, bevacizumab; PL, placebo; ORR, overall response rate; S, significant; NS, non-significant.
Safety data collected from each treatment group is presented in table 5. Adverse events occurred more frequently in the bevacizumab-containing arms.

Table 5. Overall safety of RIBBON 1 trial: all cohorts.

Adverse Events










All (%)







Serious (%)







Grade 3-5 (%)







BVZ, bevacizumab; PL, placebo.

The ODAC concluded that the trial reached its primary endpoint: improvement in PFS; however this was not translated into improved OS, in addition to an increased rate of major adverse events attributable to bevacizumab [18].

ODAC's Recommendation

ODAC stated that accelerated approval requires further demonstration that the surrogate endpoint, improvement of PFS (the primary endpoint of all 3 trials), correlates with the true endpoint, which is improvement of OS, an endpoint with clear implications of clinical benefit. Moreover, the condition for definitive approval of bevacizumab was not proof of OS improvement, as it was for all FDA’s previous approvals of first-line therapy for MBC. The FDA asked the company to show evidence that survival was not impaired by the addition of bevacizumab to chemotherapy. Even so, survival analysis in the AVADO and RIBBON 1 yielded hazard ratios for OS favoring the placebo arms, except for the capecitabine cohort, thus data provided was insufficient to establish a favorable risk-benefit analysis for the use of bevacizumab.
When these findings were taken together (table 6), the ODAC determined that the AVADO and RIBBON 1 trials did not provide confirmatory evidence of clinical benefit of bevacizumab in combination with chemotherapy as first-line therapy for MBC [18], and voted 12 to 1 against regular approval of the drug for the indications under review (BLA STN 125085\191 and 192), and recommended to remove it from the Avastin (bevacizumab) label [20].

Table 6. Efficacy summary of trials.












+ 5.5 mos


+ 1.7 mos


DOC +/- BVZ 7.5


+ 0.8 mos


- 1.1 mos

DOC +/- BVZ 15


+ 0.9 mos


-1.7 mos




+ 1.2 mos





+ 2.9 mos


+ 2.9 mos

BVZ, bevacizumab; PAC, paclitaxel; DOC, docetaxel; TAX, taxane; ANT, anthracycline; CAP, capecitabine; HR, hazard ratios; mos, months; N/A, not available.
(*) No difference calculated since OS was not reached in the placebo arm.


Three drugs were approved for first-line therapy of MBC: trastuzumab, gemcitabine and bevacizumab; the first two, under FDA’s regular approval program, were based on OS benefit (a true endpoint). Bevacizumab is the only agent granted accelerated approval for this indication, based on PFS benefit (a surrogate endpoint). In general, the AVADO and RIBBON 1 trials showed a marginal effects on PFS (less than a month) with no OS improvement and more significant adverse events in patients receiving bevacizumab. As a result the ODAC’s recommendations to confirm the effects of the E2100 trial in terms of OS were not met.
Currently, MBC is considered incurable, thus the main goal of treatment is to extend survival with acceptable levels of toxicity [21, 22]. Based on risk-benefit analysis, ODAC’s recommendation to drop the endorsement for bevacizumab as first-line therapy in MBC seems appropriate. The commentary by Dr. Ron Richardson (ODAC’s temporary voting member) during the meeting stresses the advisory panel position [23]:

“ has been said many times, survival trumps everything, and we haven’t shown a survival benefit here”.

According to the company, FDA rescheduled the final decision on bevacizumab for MBC because information submitted to the agency “has been deemed a major amendment”. This announcement raises some questions:

1. If there are no new phase III trials on this subject, where did the new information come from?
2. If the new data comes from a subgroup analysis within the AVADO and/or RIBBON 1 trials, do these new results show real clinical benefits, like OS improvement?
3. If new data from subgroup analysis shows clinical benefits, were these trials sufficiently powered to measure treatment effect and interactions within subgroups reliably?
4. Finally, if there is new information, when is it going to be made publicly available?

We must wait until December to get the answers. Meanwhile let’s examine some issues regarding cancer drug evaluation that may give us some idea of what to expect.

First, there aren’t any new trials examining bevacizumab for MBC in the first-line setting, the only phase III trials are those discussed here. Therefore, data could only come from the records of the patients included in the E2100, AVADO, and RIBBON 1 studies. One possibility is to undertake a meta-analysis to find out if larger data demonstrates an OS improvement; however a recent meta-analysis of 2,447 patients from those trials did not showed a survival benefit for bevacizumab [24]. Moreover, a systematic review and meta-analysis of five trials (3,163 patients) evaluating bevacizumab for MBC showed the same results [25].

On the other hand the use of subgroup analyses to detect patients for whom the addition of bevacizumab could be beneficial is risky. There are many publications warning on the risk of bogus results derived from chance when exploring differences in treatment outcomes based on patient characteristics, especially when subgroup analyses was not planned before data collection, thus leading to false assumptions of positive clinical effects [26-32].

Currently the most controversial subject is validity of surrogate endpoints as primary endpoints in phase III trials evaluating new cancer drugs, particularly PFS as surrogate for OS.

Overall survival (OS), defined as time from randomization to time of death from any cause, is considered the gold standard endpoint for phase III trials evaluating new oncology drug, and the best proof of clinical benefit [33-35]. In contrast progression free survival (PFS) is defined as the time from randomization until tumour progression or death from any cause, whichever occurs earlier. Approval of oncology drugs need significant evidence of efficacy and safety; OS measures both with almost 100% accuracy [36]. Also, is the most reliable endpoint in advanced breast cancer [37].

Studies with surrogate endpoints are tempting because they can be smaller, faster and less costly than those with true cancer endpoints [38, 39]. In 1992 the FDA established the accelerated approval program for serious diseases when no therapy exists or is ineffective [13]. Accelerated approval for a new drug can be granted based on a surrogate endpoint that is likely to predict clinical benefit, but for regular approval the clinical benefit has to be demonstrated by post marketing studies [33, 40], in other words, the surrogate endpoint has to be validated as such.

Many authors, health professionals and statisticians, have indicated a number of problems related to the use of PFS as surrogate endpoint [41-45], and multiple methods have been considered to validate it as a surrogate for OS [46-48]. Most important encountered problems on this topic are shown in box 5.

Box 5. Shortcomings of PFS as endpoint in phase 3 trials evaluating new cancer drugs.

  • No standard definition for progression end points.
  • Interval censorship because assessment of PFS is made at fixed time points.
  • Potential spurious conclusions in presence of harmful side effects drug-related.
  • Estimated effects less reliable when the endpoint of interest occurs at a high event rate.
  • Use of PFS in advanced disease is not beneficial, due to poor prognosis.
  • Trials with PFS as primary endpoint lack statistical power to assess OS.
  • Estimated effects on PFS do not predict if the estimated effects on OS will be clinically meaningful (correlation alone is not enough to validate a surrogate).

Data extracted from references 40-53.

The three trials evaluating bevacizumab for first-line therapy of MBC had PFS as primary endpoint; all recent studies evaluating PFS as a surrogate endpoint for OS in first-line therapy for patients with MBC could not confirmed the validity of surrogacy [49-53]. In contrast to OS assessment, PFS is subject to significant bias and following therapy may change the course of disease.


There is no evidence that bevacizumab improves quality of life or OS in HER2-negative MBC. The ODAC decision was correct [54]. The results of a survey comparing the clinical trial endpoints physicians and patients considered important in MBC treatment decision making was presented at the 7th European Breast Cancer Conference (EBCC-7) last March. Contrary to what physicians consider a worthy improvement in PFS and OS (6 and 2 months), the patients’ requisite to consider a treatment valuable was an improvement of more than 12 months in OS [55].

Companies demonstrating small benefits for their drugs, founded on unconfirmed surrogate endpoints, could gain FDA authorization; but what is required are trials focused on patient benefit not on marketing approval. Also, it is fundamental to find reliable biomarkers to identify subgroups of patients that may benefit from targeted therapies in order to increase the marginal benefits seen so far. This is especially true when considering the stratospheric prices of these new drugs. Until we are able to achieve more reliable targeting, OS must remain the primary endpoint in trials evaluating new cancer drugs.

Why has the FDA delayed their decision? Unfortunately we have no answer yet. But even though the indication of bevacizumab for MBC is finally dropped by the administration, the drug would still be available for their other indications, and eventually for off-label use in patients with MBC when potential benefit is considered by the attending oncologist.

Half of the modern drugs could well be thrown out of the window,
except that the birds might eat them.

Dr. Martin H. Fischer (1879-1962).

Conflict of interest statement

No conflict of interest

Dr. A. De la Guerra. Bevacizumab for breast cancer: FDA grants a stay of execution. Doctors Lounge Website. Available at: Accessed February 28 2017.


Previous: Pregnancy Weight Gain Guidelines in the US - a Historical Overview Next: Ebola Outbreak Highlights Struggle for Science in Africa and Inequalities in Global Health Research

Reader comments on this article are listed below. Review our comments policy.

Submit your opinion:

You must be logged in to post a comment. Please log in or sign up for a free account.

Auto-login on future visits

Forgot your password?