Last week, updated information on the AVAiL (AVAstin in Lung cancer) trial (see prior post) of cisplatin/gemcitabine with either placebo or a low or higher (full) dose of avastin was presented in a meeting in Stockholm. We had previously heard that this trial was positive for a significant improvement in progression-free survival for both the lower and higher doses of avastin with cis/gemcitabine, but there was no overall survival benefit compared with chemo and placebo.

The presentation last week shed some light on the subject. The highlight of the presentation, or at least the spin, was that all three groups had an unprecedented median overall survival of over 13 months. It was actually 13.4 months for the group receiving chemo with avastin at the standard (high) dose of 15 mg/kg IV every 3 weeks, 13.6 months at the avastin dose of 7.5 mg/kg IV every 3 weeks, and 13.1 months in the placebo arm.

With typical median survival results for advanced NSCLC in the 8-10 month range, these results are really unprecedented for a large phase III trial in advanced NSCLC. How did everyone do so well, even including the patients who received a placebo? And how was a significant improvement in median progression-free survival erased when we look at overall survival.

Although the design of the trial has been rather unclear, one report quoted the presenter of the AVAiL trial, Dr. Christian Managold, as saying that this may be because patients on placebo crossed over to avastin (see report here). If that's the case, we've previously seen that this "crossover" design can contaminate the interpretation of a clinical trial with avastin. The original phase II trial of avastin with carbo/taxol showed that patients on the chemo alone arm who then received avastin sometimes had a prolonged period of non-progression on avastin alone:

(Click to enlarge)

This small amount of data is what I would consider to be the strongest evidence in favor of continuing on maintenance avastin after initial chemo/avastin, as I discuss in a prior post.

In the phase II study (abstract here), just as in AVAiL, the chemo alone arm had an unusually good survival, I believe because they also received benefit from the avastin after the initial therapy.

So now we've seen a median survival that eclipses a year, even just beyond 13 months, for all three groups on the AVAiL trial. I believe this is because patients were excluded if they had squamous NSCLC or brain metastases, both associated with a less favorable survival, and it included a cisplatin-based doublet chemo, which as I've noted previously appears to produce slightly more favorable survival than carboplatin-based chemo combinations (see prior post).

There's still more we need to learn about this trial, but based on what's been presented, including a new mention of the crossover of placebo arm patients to avastin, I think it's hard if not impossible to interpret the impact of avastin on survival in this trial. I had previously considered it a strike against avastin to have failed to show a survival benefit in a second trial, but now I would return to a point of saying that the more interpretable results from the ECOG trial carry more weight, and it's possible to interpret that the remarkably favorable median survival in all three groups on AVAiL actually suggests to me that avastin provided much of that benefit -- even to the placebo arm.

ADDENDUM: Apparently, the report of the survival in the placebo arm being related to crossover and receiving subsequent avastin is untrue. I don't know if the error was in Dr. Manegold's understanding of the trial he himself presented or that he was misunderstood/misquoted by the press that reported on his presentation, but data made recently available illustrate that crossover to avastin or another anti-angiogenic drug occured in only a very small percentage of patients).