A group of investigators at Dana Farber Cancer Institute in Boston, MA recently published a very newsworthy article in the Journal of the American Medical Association (JAMA) that argues that patients with advanced non-small cell lung cancer (NSCLC) who are over 65 don't appear to benefit from the addition of Avastin (bevacizumab) to standard chemotherapy with carboplatin/Taxol (paclitaxel).  Several years ago, Avastin was demonstrated in the ECOG 4599 trial to lead to a survival benefit when it was added to carbo/Taxol in Avastin-eligible patients, who are a pretty limited subgroup with a good performance status, no brain metastases (a requirement since relaxed with more experience), no significant hemoptysis, and non-squamous NSCLC histology.  However, there have always been elements of the story that have cast some doubt as to how much benefit it really offers, particularly in older patients.  A subset analysis of the ECOG 4599 trial showed that patients over 70 experience disproportionately greater side effects and complications from Avastin and no survival benefit.   Meanwhile, another large randomized phase III trial called AVAiL (AVAstin in Lung cancer) that was conducted in Europe showed a statistically significant but overall very unimpressive improvement in response rate and progression-free survival when Avastin was added to a different standard chemotherapy backbone of cisplatin and gemcitabine, and this study demonstrated no benefit at all in survival.  

Since Avastin was approved by the US FDA in October of 2006, it has been considered a standard of care but not clearly the standard of care, and only about 20-25% of patients in real world clinics actually get it.  The reason it ends up being given to only a minority of patients is a somewhat open question, but in truth, I think that when you disqualify patients with many of the clearer contraindications and then also factor in some relative contraindications such as a cancer next to major blood vessels or a poorly differentiated cancer that is suspected may be of squamous histology, the actual proportion of patients who are really strong candidates for it is probably below 40%.  And then, there is also the question of how much stock clinicians place in the ECOG trial vs. the AVAiL trial that failed to show a benefit...and the fact that the median age of newly diagnosed lung cancer in the US is now around 71 meaning that a very significant fraction of patients with advanced NSCLC are in an age range where the value of Avastin is quite questionable.

In this context, the new publication is very interesting and leads us to further question whether Avastin adds anything but cost and potentially significant side effects in older patients.  In contrast to the ECOG 4599 trial, which looked at a very limited subset of patients in a very controlled situation, this work looks at the Surveillance, Epidemiology, and End Results (SEER) database, a huge collection of data from a wide range of patients from 17 cancer registries in many states, which captures 28% of cancer cases in the US.  This is a reflection of  "real world" cancer care, which is less controlled and may be a strength and a weakness.  Though there are a lot of uncontrolled variables here, the SEER database provides a broad picture of vast numbers of patients and is therefore, in my mind, another valuable tool with its own very real shortcomings, just as heavily controlled clinical trials with highly selected patients have their own shortcomings, such as that there latter may not apply to many of the actual people in a broader cancer population.

This particular study looked retrospectively at Medicare claims within SEER data from patients 65 or older with non-squamous NSCLC histology and some other specific eligibility issues who were diagnosed between 2002 and 2007 and compared results for patients before the availability of Avastin to the results for patients who received treatment after the availability of Avastin in late 2006, and looked at results depending on whether those patients received Avastin or not.   They ended up comparing results from 318 patients treated with carbo/Taxol/Avastin to results from 1184 patients who received carbo/Taxol alone, without Avastin; both of these patient groups were from 2006-2007, and they also had a group of 2666 patients who had received carbo/Taxol from 2002-2006.   

What they found was that the patients who received Avastin were more likely to have fewer other medical problems and more likely to have well or moderately differentiated cancers than the patients in the same time range who received chemotherapy alone.   There were no differences in survival among the three groups, as shown in the graph below:

(click on image to enlarge)

The median survivals were 9.7 months in the chemo/Avastin group (one year overall survival (OS) 39.6%), 8.9 months in the chemo alone group for 2006-2007 (one year OS 40.1%), and 8.0 months for chemo alone in the earlier time range (one year OS 35.6%).  They did multiple different analyses of the numbers that all showed no significant differences favoring addition of Avastin to standard chemotherapy in this population of patients over 65 with advanced non-squamous NSCLC.

It's fair to say that in this analysis there are potential unmeasured variables that are distributed differently between those patients who were prescribed Avastin and those who were not.  Issues like a sense of vaguely better performance status, risk for bleeding, or even just being more proactive in seeking more aggressive vs. less aggressive care are not captured in this analysis.  However, our clear sense is that patients who are better candidates for Avastin have a better prognosis than those who are not, so I suspect that the Avastin group would have benefited, not been penalized, by these unmeasured variables.

What does this mean?  It really remains to be seen, but there has been some speculation that payers may not cover Avastin in older patients.  I believe that oncologists had already been prescribing Avastin sparingly in patients over 70, so with this additional evidence against a benefit, I suspect we'll see only less enthusiasm for including Avastin in the treatment regimen for patients in this age range.  In patients in whom Avastin coverage is denied, it would be hard to challenge that, since I would say that the evidence is now stronger against a benefit, rather than just equivocal.  Previously, the data from the ECOG 4599 subset analysis might have led us to favor just being selective in which older patients we treat with Avastin.

For patients 65-70, we have data from the ECOG trial that showed a demonstrable benefit in patients younger than 70, along with the "real world" data from SEER that suggested against it.  However, the data from the JAMA paper include only 318 recipients of Avastin, and with the uncontrolled nature of those data, I personally consider those results to be of lower quality than the results from the ECOG 4599 trial.   My inclination will be to continue to recommend Avastin for patients under 70 who are otherwise eligible, but we'll need to see whether there is increasing resistance to covering it based on these results.