I covered the highlights of the FLEX trial, reported at the Plenary Session of ASCO 2008, a full year ago, but in that time, we never showed the survival curves or covered all of the details. It's time to rectify that, now that it's actually been published, and we're left to reflect on whether certain subgroups benefit more or less, and how the subject of the FLEX trial, the EGFR monoclonal antibody Erbitux (cetuximab) should be used in treating advanced NSCLC.

The European sudy compared standard chemotherapy alone to the same chemotherapy with weekly Erbitux in previously untreated advanced NSCLC. It included over 1100 patients, who were screened as needing to have the EGFR protein on their tumor by a test called immunohistochemistry (IHC), because this is the target that the EGFR monoclonal antibody is supposed to bind to. The definition of a positive test for EGFR by IHC was extremely liberal, since only a single cell with the protein was required for patients to be enrolled. Even with such liberal criteria, 15% of the screened patients had no EGFR protein on a single cell and were excluded from the trial.

As with most recent trials with targeted agents, it was degined to give up to six cycles of chemo with or without the novel agent,in patients who didn't show progresison after six cycles, patients assigned to the experimental arm (with the new agent) would continue on weekly Erbitux as a maintenance therapy. The specific chemo used was cisplatin/navelbine (vinorelbine), a combination rarely used for metastatic NSCLC in the US but a common standard in Europe.

Putting this all together, the trial design is shown here.

FLEX Trial Design

(click on figure to enlarge)

The study showed that the likelihood of significant tumor shrinkage was higher in the recipients of Ebritux with chemo (36% vs. 29%, p = 0.010), and that despite the fact that the median progression-free survival was the same in both arms, there was a significant survival benefit with Erbitux added to chemo.

FLEX Overall Survival Curves

The curves do separate, and the one-year survival rate is higher in the recipients of Erbitux (47% vs. 42%), but it's a puzzling pattern. Notice in the curves above that the separation occurs after more than 6 months, when the majority of patients had already come off of chemo, and the majority were off of Erbitux as well. Why and how did survival increase on the Erbitux arm after the time when the majority were off of first line treatment? We can't really explain that. And how did survival improve when there wasn't an improvement in progression-free survival. Another point that is hard to refute.

If you combine that with the fact that the improvement in median overall survival was only 1.2 months (from 10.1 to 11.3 months), you end up with a trial that left as many questions as answers. Most oncologists have struggled with the real clinical significance of this trial, regardless of its statistical significance.

As we sift through the data, we hope to find groups of patients who did particularly well or poorly, so that we might select patients who are especially likely to benefit, or those particularly unlikely to benefit. Those groups may be defined by race, NSCLC subtype, other clinical factors, or possibly molecular variables. We'll cover the differences among these various groups in a couple of upcoming posts.