CAMBRIDGE, Mass. (
(AVEO - Get Report)
has had a difficult time since its lead cancer compound tivozanib was roundly panned at a May 2 FDA advisory panel. By a 13-1 vote, experts on the FDA's panel concluded tivozanib did not demonstrate a favorable risk/benefit for the treatment of advanced kidney cancer. Despite scathing critiques of the design and conduct of Aveo's phase III trial, a deeper look at the data demonstrates tivozanib's efficacy is actually fairly similar to previously approved kidney cancer drugs. While the phase III trial might not be completely predictive of future success, at the very least, the trial is strong enough to justify an AVEO valuation above cash on hand.
There has been a lot written about the
mistakes made by Aveo in designing and conducting the tivozanib study
, and how these errors could have -- and should have -- been avoided. Setting aside that debate, I want to look at the tivozanib data in isolation to see what they say about the actual clinical activity of the kidney cancer drug. In particular, the FDA advisory panel compared tivozanib to seven other drugs approved in kidney cancer:
(PFE - Get Report)
Sutent, Inlyta and Torisel,
(NVS - Get Report)
(GSK - Get Report)
Votrient. Are the tivozanib data dramatically out of sync with the trials results of the approved drugs?
Chemist and pharmaceutical analyst John Tucker (@JohnTuckerPhD on twitter) conducted an analysis correlating progression free survival (PFS) with overall survival (OS) for all seven drugs to determine the tivozanib results from Aveo's phase III trial were anomalous. If the tivozanib trial produced flaw results in terms of either PFS or OS, then it should appear outside of the correlation seen with the approved drugs.
One of the first things to note is the relatively strong correlation between PFS and OS among the seven approved drugs. The blue diamonds represents the approved drugs with the regression line superimposed. The only real outlier is Torisel (the blue diamond sitting by itself below the regression line) with a median PFS of 5.5 months but a median OS of only 10.9 months. Based on the regression line, Torisel's median OS should have been closer to 15 months.
Now, look at how closely tivozanib (the red square) hews to the regression line. The median OS is slightly longer than one would expect given the median PFS but certainly within the margin of error. It's difficult to argue from these data alone that tivozanib is not an active kidney cancer drug -- perhaps the most effective drug compared to its peers, in fact.