We learned several months ago that the ATLAS trial of maintenance Avastin (bevacizumab) with either Tarceva (erlotinib) or placebo did not demonstrate a significant improvement in overall survival (OS) with Tarceva, despite the fact that it was associated with an improvement in progression-free survival (PFS). This is in contrast with the similar trial called SATURN that randomized advanced NSCLC patients after first line chemo to maintenance Tarceva vs. placebo, but without the Avastin during and after first line chemo, as SATURN demonstrated a significant improvement in OS along with a significant improvement in PFS. In Dr. Mark Socinski's great overview of the topic of maintenance therapy, he noted these discrepant results, with no obvious explanation. We received additional information about the OS results at ASCO, where it was reported that at the latest time point of analysis, in July of 2009, the median OS was 15.9 months vs. 13.9 months, favoring the addition of Tarceva. This corresponded with a hazard ratio of 0.90, or 10% improvement, over the entire course of treatment, a result that was not statistically significant. The overall survival curves are shown below: (click on image to enlarge) Admittedly, it doesn't look like a major separation, but there is a modest advantage with the combination. You can decide whether that's a glass half empty or half full.

My take on these results is that the trend was still favorable, even if the trend was non-significant, so they don't really challenge the conclusions of the SATURN trial. This is similar to what we might say about the Fidias trial and maintenance chemotherapy, which demonstrated a greater than 2 month difference in OS that wasn't statistically significant, so the trial was called "negative", but we've seen positive trials that showed less impressive improvements in absolute terms. In addition, the ATLAS trial was stopped early, after it met its primary endpoint of improvement in PFS. One final thing I find intriguing about the ATLAS trial is that certain subgroups of patients receiving the combination of Tarceva and Avastin appeared to have a particular benefit. Namely, Asian/Pacific Islander patients and never-smokers (both groups that we know have a higher incidence of an EGFR mutation) and those specifically tested and found to have an EGFR mutation (less than half of the enrolled patients had tissue available for molecular testing for an EGFR mutation) had a very notable improvement in OS). This is shown as the circles being to the left of the vertical line on the right side of the figure below for those groups. While this is just a retrospective assessment of relatively small subsets, these same groups were the ones who demonstrated a particularly strong OS benefit in the related BeTa trial, which compared Tarceva alone to Tarceva with Avastin as second line treatment for patients who hadn't received prior Avastin. So these findings suggest that the Tarceva/Avastin combination may truly be particularly effective, even compared with Tarceva alone, in patients who have the molecular and clinical features associated with the greatest benefit from an EGFR mutation. Apparently I'm not the only person who has come to this conclusion, because the CALGB, one of the major US-based cooperative oncology research groups, is now planning a trial comparing Tarceva/Avastin to Tarceva alone as first line therapy for patients with an identified EGFR mutation. It will take a few years, but can expect to learn more about this question in the next few years.