What happened
Shares of AVEO Pharmaceuticals (AVEO +0.00%) closed up 11.7% Monday. That continued a three-day recovery after a drop in the stock price last week from what seemed like bad news about AVEO's phase 3 Tivo-3 study that was disclosed in an 8-K the biotech filed with the Securities and Exchange Commission.
So what
In the SEC document filed last Tuesday, the company noted that some patients were "removed or 'censored' from the [progression-free survival] event count." This worried investors, because the Food and Drug Administration doesn't typically like companies to remove patients from their analyses.
Seeing the need to explain itself a little better, AVEO issued a press release on Thursday giving more details on why the patients were censored; it turns out, they should never have been counted in the first place.
Image source: Getty Images.
Progression-free survival (PFS) measures the time it takes for a patient's tumor to start growing again or when the patient dies, whichever comes first. When patients leave a study, they typically get counted as a failure at that point, but AVEO says that regulatory guidance and the trial's protocol call for patients who leave the trial without document of progression to only be counted as PFS events if they die within eight weeks of their last study assessment and have not started another therapy.
There were 10 patients who were counted as PFS events who shouldn't have been. Since the company is still blinded to whether the patients were treated with its tivozanib or Bayer's Nexavar, AVEO can fix the error while keeping the statistical analysis plan intact.
Now what
The analysis will occur after 255 PFS events. As of Wednesday, the corrected PFS event count was 243, so investors won't have to wait much longer.
While the share price has recovered somewhat, it's still not as high as it was before the company made the disclosure last week, which seems reasonable since the error -- even if it was caught in time -- makes it appear that AVEO is being sloppier than it should be.
