Roche's Avastin misses key data point in brain cancer study
ZURICH | Sat Nov 17, 2012 5:37pm EST
Nov 17 (Reuters) - Swiss drugmaker Roche's cancer drug Avastin failed to make a statistically significant difference to survival rates of patients with a common form of brain cancer in a late-stage study, the company said on Saturday.
The Phase III AVAglio study presented at the Society for Neuro-Oncology congress in Washington on Saturday showed Avastin, combined with radiation and chemotherapy, reduced the risk of cancer worsening or death, the Basel-based drugmaker said in a news release.
But the drug did not reach statistical significance in overall survival, a key data point. Further data are expected next year, Roche said.
"The interim results for overall survival (OS), the other co-primary endpoint, did not reach statistical significance," the company said in the statement.
Earlier data from the study published in August showed Avastin significantly extended progression-free survival of people with an aggressive form of brain cancer.
Avastin is Roche's third-biggest seller and is approved to treat several types of cancer, including breast, kidney, colorectal and ovarian cancers. (Reporting by Katharina Bart and Caroline Copley; editing by Jason Webb)
You have a curious way of arguing. You (and apparently marketmaker, though I have a harder time discerning what point in particular he is trying to make) have taken the position that the "remarkable variation" in the efficacy data will preclude approval, and that unless another large efficacy trial is commenced, the program is doomer. You state this position as a certainty, but you do not offer any evidence whatsoever to support your position. In response, I have pointed out that I am aware of zero instances in which the FDA has denied a drug on efficacy grounds when the 2 phase 3 efficacy studies have come in at p values of .025 or lower on the pivotal efficacy endpoint (or .05, for that matter) -- regardless of how much variability there is below that number. I have also pointed out that the variance between .025 and .001 is well within the variance that the placebo effect has yielded for a particular compound. You now ask me to do more follow-up research for you. You, on the other hand, continue to point to nothing but your own word to support your somewhat melodramatic assertion that the program is doomed. You know say that Nektar will have to show these trials were not tainted by fraud. Well gee, no kidding. Don't you think it is a threshold requirement of any sponser to demonstrate integrity of the data? Of course it is, and of course this will apply to Nektar as well. Assuming these data are not marred by fraud, we have our difference clearly defined -- I think these data will support approval on efficacy grounds notwithstading variability, and you are confident that they will not. I hope we have a chance to see it play out.