I recently had the honor of providing the expert commentary at the ASCO 2014 conference on three high impact lung cancer presentations, all on the subject of treatment options for molecularly defined populations (EGFR and ALK). I'll review the findings from these three abstracts and my perspective on each of these, starting with a very provocative presentation by Dr. Kato from Japan that attempted to ask the question of whether we should add Avastin (bevacizumab) to Tarceva (erlotinib) as first line therapy for EGFR mutation-positive advanced NSCLC.

There have been a few large trials that directly tested the combination of Tarceva/Avastin vs. Tarceva alone -- one as second line treatment (called BeTa), and one as maintenance therapy (called ATLAS).  These have failed to demonstrate a significant benefit for the combination, but they were in molecularly unselected patients, meaning that they were a broad population and not just patients with a specific molecular feature like an activating EGFR mutation.  However, both of these trials provided hints that the combination appeared to be especially effective for specific patients, most likely those with an EGFR mutation. For instance, the BeTa trial showed that two particular clinically defined subsets demonstrated an overwhelming benefit for the combination over Tarceva alone -- Asian or Pacific Islander, and never-smokers.  Of course, these are also two subgroups known to be highly enriched for having an activating an EGFR mutation.  The subgroup analysis also showed that the small subgroup with an EGFR mutation also did far better with the combination, corroborating our presumption about why the Asian/Pacific Islander and never-smoker groups did so well.

  Moreover, the same clinically defined subgroups -- Asian or Pacific Islander, as well as never-smokers -- were also the two standout subgroups that appeared to benefit most in the ATLAS trial. The subgroup analysis from that trial did not include an analysis by EGFR mutation, but I suspect it would have shown the same effect as seen in the BeTa trial. 

Dr. Kato and colleagues looked specifically at the potential benefit of adding Avastin at 15 mg/kg IV once every three weeks to daily Tarceva at 150 mg daily as first line systemic therapy for advanced NSCLC patients with a prospectively identified EGFR mutation. They randomized 154 patients, of whom 152 received treatment, and demonstrated a rather remarkably improvement in median progression-free survival (PFS) of more than 6 months, increasing from a median of 9.7 months for Tarceva alone to 16.0 months with the Tarceva/Avastin combination. Though the response rate was not significantly different, about 2/3 of patients showing significant tumor shrinkage (called an objective response) in both groups, the disease control rate, which is the combination of patients who demonstrate tumor shrinkage with those who demonstrate at least stable disease, was different, favoring the combination, at 99% of patients on the combination showing tumor shrinkage or stable disease, vs. 88% for those receiving Tarceva alone.

Also quite interesting was the difference between the two common types of activating mutation.  Specifically, both subtypes of activating EGFR mutation -- exon 19 deletion and L858R on exon 21 -- demonstrated a major improvement in PFS with the combination compared with Tarceva alone. But the patients with an exon 19 deletion had a longer PFS compared with the exon 21 patients, whichever treatment they received.   The improvement in PFS was especially remarkable in exon 19 deletion patients, who actually demonstrated an improvement of nearly 8 months in median PFS, which was a full 18 months with the combination!

While moderately severe or greater side effects were more common with the combination, this was primarily high blood pressure and proteinuria (spilling protein into the urine), rather than more fearsome problems, and there were no treatment-related deaths. And while the discontinuation rate for Avastin was 41% over the course of treatment, higher than the 10-15% seen in earlier trials, this was again often due to more chronic side effects that I suspect were a product of the long time that people were on Avastin, because they were doing so well -- being on a treatment for a long time leads to a higher risk of side effects, especially cumulative ones.

The other issue that could be a real barrier is cost. The patients on the Avastin arm received a median of 16 cycles of treatment, which costs "the system" at least $120,000 on top of the already significant cost of Tarceva from month to month.

Because this wasn't a large, definitive study, it is debatable whether the strength of the evidence is great enough to warrant a change in how we approach EGFR mutation-positive NSCLC. My view was that while the study wasn't so large that it offered incontrovertible proof that Avastin is very helpful, and it doesn't show a survival benefit (the patients are almost all still alive, so we can't know if it improves survival for a while), I consider the magnitude of the improvement in PFS to be very impressive and convincing.  I argued that with targeted therapies given to targeted population, we often see a staggering impact, with response rates and progression-free survival results that are so overwhelmingly better than what we could otherwise expect that you don't NEED a 1000-patient phase III trial and additional years of follow-up to prove a survival benefit.  We know that XALKORI (crizotinib) was amazingly helpful when it gives a 60% response rate and responses that last for >6 months, and the FDA was wise to not deny its benefits to thousands of people while waiting for years of additional studies after the first overwhelmingly convincing early evidence came in. I would consider the benefits here with the Tarceva/Avastin combination to be quite convincing, even if it's only a little more than 150 patients.

Some may want to see a confirmatory study done, perhaps outside of Japan (since Japanese patients may sometimes have genetic differences in how they respond vs. other genetic populations). Some may want to see a survival difference, which may occur, because few patients will cross over to get Avastin later, but these patients will often live for one to several years after initial progression and benefit from later treatments that could diminish the survival effect from a first treatment given.  My concern is that the compulsion for a higher level of evidence isn't really because people remain unconvinced that there is a major benefit from adding Avastin, but that it's a bias introduced by the cost of the treatment.

I noted that I think it's perfectly fair to openly discuss the cost of a drug and ask whether the benefit is enough of a value. However, I don't think it's appropriate to have cost remain just an undercurrent that taints the judgment of people by making them reluctant to accept the credibility of the benefit but not acknowledge that it's the cost that leads to their concern.

Unfortunately, as the cost of new cancer drugs creeps beyond $10,000 month, we're likely to see a higher threshold to have the cancer community become convinced that an observed benefit is  fair value relative to its cost.  In this case, I remain impressed that Avastin provides a meaningful, striking benefit along with Tarceva, but I may find it hard to convince and insurer that the evidence is strong enough.  And if it's not covered by an insurer, adding a $120,000 price tag to treatment is going to mean that people can't benefit from it.

What do you think?