Last week we discussed SATURN, the first of 2 recently presented trials testing the role of maintenance Tarceva (erlotinib) in advanced NSCLC patients. Today I will discuss the ATLAS trial, the last of the 4 major maintenance therapy trials (along with immediate versus delayed Taxotere (docetaxel) and maintenance Alimta (pemetrexed)). In the SATURN trial of maintenance Tarceva or placebo after 4 cycles of first-line chemotherapy, Tarceva prolonged progression-free survival (PFS) by about 1 week although the true benefit may have been more robust than that number indicates. The ATLAS trial was of nearly identical design to the SATURN trial. Patients with advanced NSCLC were randomized after completing 4 cycles of platinum doublet chemotherapy with Avastin (bevacizumab) to either maintenance Avastin plus Tarceva or Avastin plus placebo. The goal, or primary endpoint, was to show improved PFS with the combination of Avastin and Tarceva.

ATLAS and SATURN Trial Designs

(click on image to enlarge)

ATLAS also looks pretty similar to another recently presented phase III study, the BeTa (Bevacizumab Tarceva) trial, which randomized advanced NSCLC patients with progressive disease after first line chemotherapy to either second-line Tarceva or Tarceva plus Avastin. The BeTa trial was the first phase III trial to test the hypothesis that adding a vascular endothelial growth factor (VEGF) blocker to an epidermal growth factor receptor (EGFR) blocker had benefit. One key thing to remember about BeTa is that the primary endpoint was improved overall survival (OS), not PFS, which is usually much more difficult to prove (and most would argue more meaningful). BeTa did show a doubling of PFS with the addition of Avastin to Tarceva, but did not show a statistically significant OS benefit to the combination. Now back to ATLAS. The BeTa trial suggested that there is some benefit in PFS to the addition of Avastin to Tarceva, so as the primary endpoint of ATLAS was increased PFS compared to Avastin alone I was not shocked when earlier this year a press release indicated that this trial had met its endpoint. The only mystery was the magnitude of the benefit. So here it is: the median PFS for the combination was 4.76 months versus 3.75 months with Avastin alone, and this difference was statistically significant.

ATLAS Trial Progression-Free Survival

Again, not many people get overwhelmed by a one month improvement in PFS, but the fact that the curves stay separate throughout the duration of the study suggests that there are patients who are getting a significant benefit from Tarceva. At 3 months, 2/3 of patients on the combination arm were alive and progression-free, compared to only 53% of the Avastin alone arm. By 6 months those numbers were 40% versus 28%, still a pretty good gap. So who were these patients that benefited from the addition of Tarceva? Well, the molecular analysis has not been presented yet, but we know that there was benefit seen in men and women, as well as in smokers and non-smokers. However, in SATURN we saw that most, but not all, of the benefit from maintenance Tarceva was seen in the EGFR mutant population. I see no reason to think this will not also be true for ATLAS. So what conclusions can we draw from these trials of maintenance Tarceva, which had such consistent results that I dare say we can refer to them together in any argument for or against this practice? Let’s take a quick look at the chemotherapy maintenance trial results, which would get me shot in a scientific symposium but I’m going to do it here anyway. The early versus delayed Taxotere trial showed an increased PFS with immediate/maintenance Taxotere of about 3 months, with a non-significant improvement in OS that was also about 3 months. The maintenance Alimta trial showed an improvement in PFS of about 2 months, although the OS benefit was a statistically significant 3 months (and 5 months for patients with adenocarcinoma histology). I think many people consider these clinically meaningful improvements and would consider changing their practice based on these results. In contrast (not comparison, mind you, since comparing outcomes across different trials, with different patient populations, is fraught with problems), the maintenance Tarceva trials had 1 month or less improvements in PFS, with no overall survival results presented yet. If the OS is significantly prolonged when these results come out, please come back and taunt me for this post. But my guess is that there will NOT be an OS benefit seen in the overall population, although there probably will be in the EGFR mutant subgroup. I believe that Tarceva is an effective therapy for patients with relapsed NSCLC, which is especially effective in patients with EGFR mutations. However, I don’t see how anyone could look at the results of SATURN or ATLAS and consider changing their practice to incorporate maintenance Tarceva based on these results. There probably is some real prolongation of PFS in this unselected population, but this magnitude of benefit would not, in my opinion, be worth the extra side-effects one gets with Tarceva versus the benefit of a treatment break. And for patients in whom you would consider maintenance therapy, you would have to look at the other maintenance trials and say that the evidence is more compelling with cytotoxic chemotherapy than with Tarceva.