Perhaps the most unexpected clinical trial result in lung cancer over the past 5 years was the finding in the large Southwest Oncology Group (SWOG) 0023 trial that randomized several hundred patients to maintenance therapy with either the oral EGFR inhibitor Iressa (gefitinib) or a placebo after chemo/radiation concurrently and then consolidation taxotere (docetaxel). While just about everyone in the lung cancer community expected to see either a significant benefit or, at worst, no real effect from maintenance Iressa, the actual trial was stopped early and demonstrated a statistically and I would say clinically significant decrease in overall survival with maintenance Iressa. The median overall survival (OS) in the final publication was a full 12 months lower in patients who received Iressa compared with those who received the placebo . To me, not only did this study demonstrate that giving consolidation EGFR inhibitor therapy was probably a bad idea, at least outside of a clinical trial, it also suggested that we don't necessarily know as much as we presume we do about how trials will turn out, so it makes sense to do the studies rather than just start a new strategy without the evidence to back it up.

However, it's fair to point out that this trial was with Iressa and not Tarceva, the agent that has actually demonstrated a significant overall survival benefit compared with placebo in advanced NSCLC, while Iressa failed to do that in an international trial. And the SWOG 0023 trial was in a very broad, unselected patient population rather than a subset we might predict would get the most benefit, such as those with an EGFR mutation. While we might presume that patients with an EGFR mutation will do especially well with maintenance Iressa or Tarceva, I'm not falling for that trick and presuming anything right now, since the SWOG trial showed me that we could do real harm by presuming the outcome of an approach. At the World Conference on Lung Cancer in San Francisco a couple of months ago, Dr. James Rigas from Dartmouth-Hitchcock Medical Center presented the results of a trial that tested maintenance Tarceva vs. placebo after chemo and radiation for patients with unresectable stage III NSCLC. This trial enrolled "all comers", a broad population that was not selected by smoking status or EGFR mutation or any other factor that might predict a higher chance of benefit with Tarceva. All patients received chemotherapy (in this case low dose weekly cisplatin and taxotere (docetaxel)) with concurrent full dose chest radiation over about 6 weeks, and then half of the patients were randomized to receive daily Tarceva at 150 mg daily, and the other half received a placebo. The trial had actually been developed several years ago, long before the (very) negative trial results with SWOG were presented, and it continued to try to enroll, but it had a hard time enrolling well after the value of maintenance therapy with an EGFR inhibitor emerged as so questionable or even detrimental. The trial closed to enrollment at about 60% of its target enrollment, due to these various challenges, but it still accrued 253 patients, which was enough to draw some conclusions. The trial didn't really show any surprising issues with side effects, and unfortunately it also didn't show any survival benefit for the maintenance Tarceva. Both arms did pretty well, with the median OS 26.9 months on the placebo arm, and 23.6 months on the Tarceva arm: However, the numbers and curve above reflect what is called an "intent-to-treat" analysis, which is based on the assignment of patients right from the beginning of the trial, and it includes patients who were assigned to Tarceva but never actually got it. The analysis looks a little different and overall more favorable for Tarceva (non-significantly) when the analysis is limited to the patients who actually received the maintenance therapy they were assigned to be administered. Here are the curves for the time before patients showed cancer progression: However, overall deaths were more comparable, because it seemed that while fewer patients on Tarceva died from progressing cancer, more on Tarceva died due to a "non-cancer related cause", not elaborated further. Here are the OS curves for the patients who actually received maintenance therapy after chemo/radiation: So where does this leave us? For lack of a more straightforward conclusion, I’d have to say confused. The trial by Rigas and colleagues with Tarceva was only about one third the size of the SWOG trial, and the limited size of this work really limits what we can conclude from it. Importantly, this study did not show a suggestion of a decrease in survival with Tarceva the way Iressa did in the SWOG trial, but that's not really high praise. It’s hard to come up with a plausible explanation for why Iressa led to a worse survival, so perhaps it was just a fluke, even if a very unlikely one by chance alone. The trial with Tarceva suggested an improvement in time to progression, particularly in the subset of patients who reached the point of actually taking the maintenance oral therapy they were intended to receive. Survival with erlotinib really didn’t show a clear signal of being superior to placebo, with a suggestion that any improvement in controlling the cancer may have been offset by other complications, though we still need to learn more about this. Overall, my sense is that the world is changing toward molecularly-based management of lung cancer, so it becomes hard to interpret the effectiveness of a targeted therapy, particularly an EGFR inhibitor like Tarceva, without considering how targeted the treated population is. Within these trials, there are likely some patients who received a striking benefit, while others received little benefit and even possible harm. We need to conduct future trials with a better consideration of the relevant subsets so that we can channel the right drugs to the particular patients who will do best with them. In the meantime, with one trial showing a net detrimental effect and the other only looking favorable enough to show no significant survival effect, I am not inclined to presume that an EGFR inhibitor as a maintenance therapy after chemo/radiation for locally advanced NSCLC is going to help them. But I’m hopeful that in the near future we’ll be able to understand these varied effects better as we delve into the molecular differences of the cancers we’re actually treating in these patients.