It was almost exactly a year ago that I described the basic results of the global LUX Lung-1 trial that enrolled 585 patients with advanced NSCLC who had gone at least 12 weeks without progression on Tarceva (erlotinib) or Iressa (gefitinib) in a 2:1 fashion to either the oral targeted therapy afatinib (an irreversible inhibitor of the human epidermal receptor (HER) family, of which EGFR and HER2/neu are members, also known as a "pan-HER inhibitor") or placebo. This trial was intended to enroll patients who had developed acquired resistance to an EGFR tyrosine kinase inhibitor (TKI) like Tarceva or Iressa, but the eligibility criteria were frankly a little too loose to really have enrolled a uniform population of patients who we think of as classic "acquired resistance" patients, who should have an EGFR activating mutation and would often have a very good response in terms of tumor shrinkage, or at least very prolonged stable disease. Requiring just stable disease for at least three months of an EGFR TKI allows the study to become a combination of 1) people with true acquired resistance after a very strong clinical benefit, and 2) people with slowly progressing cancer that didn't happen to meet criteria for progression within 12 weeks on an EGFR TKI.

To review the highlights, the median age of patients was 58-59 years old, 60% of the patients were women, nearly two thirds were never-smokers, the median duration of response was about 10 months, and the objective response rate to prior EGFR TKI was about 45%. This is lower than the response rate to EGFR TKIs that is typically 60-75% for trials exclusively of patients with an EGFR mutation, and the median duration of response is a little on the low side, so my estimate is that only about one of every three or four patients squeaked into the trial but wasn't really the target patient for the population that this trial was intending to study. However, this study was designed just before trial results like IPASS made it clear that molecular selection (identifying patients for trials based on the presence of a particular mutation) is clearly superior to clinical selection (identifying a patient based on clinical characteristics like race, sex, and smoking status).

The trial showed that afatinib clearly has some promising activity in this population: the response rate (unconfirmed) was 13% for patients receiving afatinib (7% confirmed: it's not unusual for responses to not be confirmed for practical reasons, since this requires a repeat CT just a few weeks after the first, which isn't always done), compared with 0.5% for patients receiving placebo. The waterfall plots below convincingly illustrate that far more patients experienced minor or major tumor shrinkage if they received afatinib (on the right) compared with placebo (on the left). Waterfall plots show every patient as a vertical bar, which goes up for tumor growth and down for tumor shrinkage, and longer bars mean more growth or shrinkage -- so the most shrinkage is on the far right of the curves, and the worst progression is on the far left of each plot.
(click on image to enlarge)

I can't explain how a few people responded so well while on placebo. I suspect it was an improvement in a pneumonia or area of pneumonitis. Regardless, far more patients had tumor shrinkage with afatinib. In addition, the progression-free survival was statistically and I would say also clinically significantly longer in the recipients of afatinib compared with placebo, at 3.3 vs. 1.1 months (hazard ratio 0.38, log-rank test p-value <0.0001):

However, despite these clear improvements in response rate and progression-free survival, overall survival was no better with afatinib, actually trending toward a non-signficant, slightly lower median survival with the active drug: (10.8 vs. 12.0 months, HR 1.08, log-rank test p-value 0.7428).

How was afatinib tolerated? Generally well, though diarrhea, rash, and mouth sores were pretty common (around 90% of patients developed some degree of rash and diarrhea, and 60% developed mouth sores, though these side effects were severe in only a small minority of them). Meanwhile, patients receiving afatinib had a prolongation of the time before they developed any worsening of cough, shortness of breath, or pain.

The primary endpoint of the trial was overall survival, which afatinib failed to improve, so it's not clear where that leaves us. It's hard to imagine the FDA approving this agent based on the LUX lung trial, since it was technically a negative trial for what it set out to achieve. Still, it's very clear that this agent has meaningful activity, and I'd say that the response rate and progression-free survival results are quite promising.

Why did it fail to show a survival benefit? The short answer is that I don't know -- nobody really does. Could it be that the results would be different if we divided the results of the trial based on whether they had an EGFR mutation or not? Based on the results of the IPASS trial and much of the research on EGFR TKIs in the past 2-3 years, I'd say absolutely yes. I think many oncologists would consider using this drug primarily or only in patients with a known EGFR mutation or a profound response to an EGFR TKI, rather than for someone who had stable disease but no real tumor shrinkage, then progressed shortly after 3 months into treatment.

It's also possible that there was a trend toward longer survival on the placebo arm because patients on another EGFR inhibitor had some rebound very fast progression that caused them to do worse after they stopped it at progression. Investigators from Memorial Sloan-Kettering Cancer Center have reported that a minority (~20-25%) of patient with an EGFR mutation or otherwise with a very good response to an EGFR TKI can deteriorate very quickly after stopping an EGFR TKI, even if they are progressing on it. And it's also possible that patients who were treated with an EGFR TKI again after progressing on placebo could have done well with the "rechallenge" treatment. We know that some patients can respond again to an EGFR TKI after a break from it. We don't know with certainty that the favorable results with afatinib wouldn't have been seen by just giving these same patients Tarceva or Iressa, even if they progressed on it previously.

One final point to make is that, aside from these questions of its value as a single agent, I'd say that the most promising lead in treating patients with acquired resistance to an EGFR TKI is the early results on the combination of afatinib with the monoclonal antibody against EGFR Erbitux (cetuximab). This work was only presented for the first time a few months ago and was based on just a few dozen patients, but the 40% response rate in EGFR TKI-pretreated patients with an EGFR mutation has led to great enthusiasm for a phase III trial. Frankly, the trial I'd most like to see is a randomized study of EGFR mutation patients (no clinical selection) with acquired resistance to afatinib alone or the afatinib/Erbitux combination, though it will likely need another comparator arm of whatever we'd consider standard of care here -- even possibly placebo -- to lead to potential drug approval). But most of all, I'd just like to see more work with this agent, since I think there will likely e a place for it in our treatment of advanced NSCLC. It may well provide some additional value beyond our current EGFR TKIs, or else after them.