One of the high profile presentations in the lung cancer track at ASCO 2014 was from Dr. James Yang of a pooled analysis of the LUX-Lung 3 and LUX-Lung 6 trials, each comparing Gilotrif (afatinib) to standard chemotherapy as first line treatment of EGFR mutation-positive advanced NSCLC, which for the first time demonstrated an actual survival benefit not seen in similarly designed trials with Iressa (gefitinib) or Tarceva (erlotinib). We should expect to see a huge marketing campaign built on this result, implying that Gilotrif is now the leading choice of oral EGFR inhibitors that would be considered for EGFR mutation-positive patients. But is this a fair claim?
The LUX-Lung 3 trial compared Gilotrif to cisplatin/Alimta (pemetrexed) in a global trial, while LUX-Lung 6 compared Gilotrif to cisplatin/gemcitabine in an Asian trial, each with over 300 EGFR mutation-positive patients. Both showed a highly significant improvement in response rate and progression-free survival that largely reproduced the same results seen in several preceding trials with Iressa (gefitinib) or Tarceva (erlotinib) vs. other chemo regimens over the past few years. However, all of the preceding trials have failed to demonstrate a difference in overall survival (OS) from the EGFR tyrosine kinase inhibitor (TKI). This has been presumed to have been because the vast majority of patients receiving initial chemotherapy have crossed over to an EGFR TKI at progression, with the presumption that these patients then demonstrate the same kind of dramatic and long-lasting response that we hope to see in EGFR mutation-positive patients starting on an EGFR TKI as initial treatment. In other words, we might presume that it doesn't matter if you receive an EGFR TKI as first or subsequent treatment if you have an EGFR mutation, as long as you get it, so this "crossover" effect should negate any survival benefit from an EGFR TKI given as first line therapy.
The LUX-Lung trials did not demonstrate a significant improvement as individual studies, and in fact, the LUX-Lung 6 trial didn't even show an improvement in median OS for the Gilotrif arm over chemotherapy. Overall, when we look at the absolute results of the two trials, at least in terms of median OS, we see that they did very comparably to preceding trials with other EGFR TKIs -- in fact, the best median OS was seen in a Japanese trial with Iressa.
Now, these trials individually didn't show a survival benefit, but there was an OS benefit when the results for the two trials were pooled together and the minority of patients (about 10-12%) with rare mutations (not the common "activating" mutations on exon 19 or exon 21) were removed, there was a modest improvement in overall survival:
Noting that the issue of crossover to EGFR TKI from chemo was a leading explanation for the lack of a survival benefit in earlier trials, I immediately wondered whether the significant improvement in OS with the pooled LUX-Lung 3 and LUX-Lung 6 was a product of the unusually low crossover in these trials -- just 62% in the pooled trials (and an astonishingly and tragically low 56% on LUX-Lung 6) -- compared with the crossover of over 90% in the Japanese trials with Iressa. But in fact, the crossover rates in Europe were only in the 75% range -- better than the LUX-Lung trials, but still disappointing when we know that an EGFR TKI is likely to be overwhelmingly beneficial for patients with an EGFR mutation. It's important to underscore that the crossover limitation was NOT because of medical complications intervening and patients missing the opportunity to receive the agent -- it's a practical issue limited by the health care system in question. In Japan, not only did nearly all of the patients on the Iressa trials cross over from chemo to EGFR TKI, but 100% of the Japanese patients on the LUX-Lung trials crossed over after chemotherapy to an EGFR TKI -- while only 52% of patients in countries where EGFR TKIs were poorly covered crossed over. But the patients from Japan, who all crossed over, also showed a survival benefit of Gilotrif.
What was especially striking in this presentation was that the survival benefit was entirely limited to the patients with an exon 19 mutation, and that the patients with an exon 21 mutation trended toward a worse survival with Gilotrif vs. chemotherapy:
This point of the striking difference in results depending on whether patients have one type of activating mutation or another has been reported many years ago but largely forgotten (and will be the subject of another post soon). It suggests that we should really stop lumping together these two groups as "activating EGFR mutations", but rather that we should see them as separate groups, even if both demonstrate a striking benefit of higher response rate and longer progression-free survival with EGFR TKI therapy vs. chemotherapy.
What is also especially interesting and somewhat puzzling is that the patients with an L858R substitution, an exon 21 mutation, showed a clear benefit with afatinib over chemo early on, but the exon 21 patients assigned to chemo had a clear trend toward a more favorable survival. This, along with the finding of the Japanese patients having a superior survival if they started with Gilotrif vs. crossing over from chemo to an EGFR TKI later, suggests to me that there may be more importance to sequence of therapy than most of us had presumed. I hope we learn more about this over the coming years.
The key question is whether Gilotrif is clearly the best EGFR inhibitor for EGFR mutation-positive patients, or at least those with an exon 19 deletion. The difference in survival for those patients is quite impressive, but I'd contend that it's not a very justified statement. The survival benefit wasn't seen in either trial independently, which is the way that endpoints on trials are intended to be measured. The survival benefit wasn't seen in the entire study population, but rather only after removing patients who didn't do as well, whether you focus on the removal of just the "rare EGFR mutation" patients who were included but did better with chemo, or the exon 21 patients who are part of the approval for Gilotrif. Having a large study population gave the opportunity for "multiple shots on goal" -- the chance to look in multiple ways for a result that shows what you want to show. The survival benefit was shown to be significant in a pooled analysis of two trials, which are each about twice as large as any of the other preceding trials of this design. But because statistical significance is a function of both the magnitude of the difference and the size of the population, you can make a less impressive difference statistically significant if you enlarge the size of the population (in fact, there are many trials in oncology, especially breast cancer, that demonstrate statistically significant differences that represent remarkably minimal differences in absolute terms). Pooling the two largest studies made it possible to engineer a statistically significant difference.
At the same time, it's worth noting that three of the four trials with Iressa or Tarceva that I included in the first figure above stopped early, with just 165 to 230 patients, because an interim analysis demonstrated remarkable differences in response rate and progression-free survival that favored the EGFR TKI arm and made it arguably unethical to continue the trials by randomizing patients away from what appeared to clearly be a better treatment. We should consider this a laudable intervention -- the people running the study compromised their ability to demonstrate a bigger effect by ending a trial with lower numbers of patients, then completely compromised their ability to demonstrate a survival benefit by crossing over as many chemo patients as possible to EGFR TKI therapy. In contrast, the LUX-Lung 3 and 6 trials were initiated at a time when EGFR TKIs were very widely felt to be the best therapy for EGFR TKI-positive patients -- in fact, the US-based National Comprehensive Cancer Network (NCCN) guidelines listed Tarceva as the only recommended option for initial treatment of EGFR mutation-positive NSCLC, not including chemo as an option. But the Gilotrif trials randomized over 750 patients around the world to chemo vs. EGFR TKI in this setting, and more than a third of the chemo patients never received an EGFR TKI. It may have been legal, and it may have been acceptable by the FDA as a route to approval, but it makes me uncomfortable to think that its approval was only made possible by systematically capitalizing on many people with an EGFR mutation being unable to receive the EGFR TKI that just about anyone could have expected to be the superior choice in these patients. At the very least, I feel we shouldn't punish Iressa and Tarceva for terminating the EGFR TKI vs. chemo trials early based on ethical concerns that weren't shared in the conduct of the LUX-Lung trials, even though more than a third of the chemo recipients on the LUX-Lung 3 and 6 trials failed to receive an EGFR TKI ever.
I'll close by saying that the improvement in survival with Gilotrif for exon 19 patients is, in my mind, impressive and even rather surprising to me, especially for the patients who actually crossed over to another EGFR TKI, but the comparison of these trials was to chemo, while the truly relevant comparison for Gilotrif is to other EGFR TKIs. Right now, the marketing push of this survival benefit is based only on indirect evidence of a benefit not seen with other EGFR TKIs compared with chemo, but we don't know whether Iressa and/or Tarceva would show the same survival benefit vs. chemo if they had done a trial with over 750 patients rather than stop their trials due to ethical concerns. Meanwhile, there is another trial, called LUX-Lung 7, that randomized over 300 EGFR mutation-positive Asian patients to either Iressa or Gilotrif. THIS is the trial that will really tell us whether Gilotrif is superior to other TKIs, and it completed enrollment in July, 2013 and is awaiting analysis. It will be critical to look not only at differences in efficacy but also at tolerability, since Gilotrif is widely felt to have considerably greater side effects than Tarceva or certainly Iressa (which is the best tolerated of the 3).
Until then, Gilotrif has gained some points in my mind as an effective option for EGFR mutation-positive patients with an exon 19 deletion, but it hasn't demonstrated clear superiority over its true competition, which is another EGFR TKI, not chemotherapy, and I am also more impressed by the results of the combination of Tarceva with Avastin in this population that was the subject of another recent post.