Imclone put out a press release yesterday that the previously described, US-based BMS-099 trial of carboplatin-taxane (either taxol (paclitaxel) or taxotere (docetaxel), investigator's discretion) with or without the EGFR monoclonal antibody erbitux (ceteuximab) has failed to demonstrate a statistically significant improvement in overall survival. Just over a year ago, we first learned that this trial did not meet its primary endpoint (see prior post): I thought this bode very poorly for the future of erbitux in lung cancer. Shortly thereafter, though, Imclone put out a press release that the European FLEX trial of cisplatin/navelbine (vinorelbine) with or without erbitux (see prior post), although we had to wait nearly nine months to get more information. The results of the FLEX trial were ultimately presented at the Plenary Session of ASCO 2008 (summary post here), showing a statistically significant improvement in survival from the addition of erbitux to this chemo regimen, but with an improvement in median survival of only just over a month, many people are left wondering whether this difference is truly clinically significant, especially factoring in the added side effects, weekly ongoing treatment, and expense.

In the wake of a trial that was so marginal for demonstrating a meaningful benefit in the very important clinical variable of overall survival, the results from other erbitux trials are helpful. I previously described a smaller randomized trial (131 patients) who platinum (cisplatin or carboplatin, investigator's discretion)/gemcitabine, with or without erbitux, which demonstrated what I would consider to be convincing improvements in response rates, progression-free survival, and overall survival with the addition of erbitux (prior post here). The survival results of the BMS-099 trial, with 676 patients, would also be very helpful to know, but they have been unavailable until now.

The press release essentially noted that the survival mirrors the results of the FLEX trial very well -- for better or for worse. Compared with the FLEX trial that showed a 1.2 month improvement in survival, the BMS-099 trial showed a 1.3 month longer median survival (9.7 vs. 8.4 months) that was not statistically significant. Although a large trial, the BMS-099 trial wasn't nearly as large as the FLEX trial, with 1125 patients, and the former wasn't powered to demonstrate an improvement in overall survival, but rather focused on the endpoint of progression-free survival. Perhaps also significant was the fact that the BMS-099 trial didn't exclude patients based on EGFR protein expression by immunohistochemistry (IHC). While the FLEX trial eliminated 15% of patients because they showed no EGFR protein expression on their tumor cells, the BMS-099 trial, along with other erbitux trials in lung cancer, had no requirement for molecular testing to enroll.

At the end of the day, I would consider the results of the BMS-099 trial exactly what you'd expect to see: a slight improvement in survival that, whether reaching a point of statistical significance or not, is quite modest in the clinical benefit it offers. On the other hand, it is consistent, as shown in the following figure, where the yellow boxes all fall to the left of the vertical line, showing the same magnitude of rather mild survival benefit in all of the studies (hazard ratio (HR) is a measure of that benefit over the whole time of the trial, and a number lower than one corresponds to the erbitux arm doing better -- a HR of 0.80 means that it gave a 20% improvement in survival):

Overall, I doubt that these results will change many people's opinions either way. If you felt that the survival benefit from FLEX was enough, the BMS-099 trial shows the same degree of benefit, now not quite statistically significant (p = 0.17, HR 0.89 vs. HR 0.87 on FLEX). If you weren't impressed with FLEX, you might say that BMS-099 was still a large randomized trial that failed to show a statistically significant overall survival benefit. One thing these results do show me is that it doesn't seem to matter what chemo you give with erbitux, since a very similar benefit is seen with cisplatin/navelbine, carboplatin/taxol or taxotere, or cisplatin or carboplatin/gemcitabine.

But for now, we've got a classic "is the glass half empty or half full?" situation. I'm inclined to discuss and offer erbitux to most patients with first-line advanced NSCLC, without a very forceful recommendation for it, and I would be more inclined to pursue it in patients who at least have EGFR protein expression by IHC. I would have been more enthusiastic about recommending it had the BMS-099 emerged as positive for a survival benefit, but it fell right in line with the rest of the ambiguous, borderline findings with erbitux in NSCLC thus far.