A lot of new data have emerged over the last 2-3 years that have addressed the concept of "maintenance therapy" for patients with advanced NSCLC (see Dr. Socinski's excellent podcast for a general review). The basic idea is that a patient is treated with first line chemo-based treatment, with or without the angiogenesis in inhibitor Avastin (bevacizumab), and the fortunate majority have some degree of tumor shrinkage or at least stable disease. The degree of benefit from each subsequent cycle of chemotherapy tends to decrease with further treatments, so you reach a point of diminishing returns for an often-challenging two or three-drug combination. While the well-established standard up until a few years ago had been to stop first line therapy and observe a patient for progression before initiating second line treatment, several trials in the last few years have demonstrated that lighter ongoing maintenance therapy, more to suppress the cancer at its current state of prior response and non-progression than to markedly shrink the cancer further, provides a very significant improvement in progression-free survival, and a less clear but at least strongly suggested overall survival benefit (the trials showing a significant improvement in overall survival have some substantial flaws). There are two basic concepts of maintenance therapy: one is switch maintenance therapy, in which someone stops all of the first line drugs and switches to a new therapy to maintain the response. It has the theoretical advantage of treating the cancer from a new angle, with an agent that the remaining cancer has not had the opportunity to develop resistance to. The approvals of both Alimta (pemetrexed) and Tarceva (erlotinib) as maintenance therapies are based on this concept. In contrast, continuation maintenance therapy is when one or more of the agents from the first line setting is continued after a fixed number of cycles of first line therapy. We haven't had much data on this concept with modern chemo approaches, but this year at ASCO, a couple of interesting trials were presented that provided some additional insight on this approach (though oncologists often use this approach with Avastin and Alimta, we haven't seen data that demonstrate continuation maintenance with these agents is beneficial).

First, by way of background, we can look at a trial published by Brodowicz and colleagues several years ago, in which 352 previously untreated patients with advanced NSCLC received up to four cycles of cisplatin/gemcitabine, and then patients who hadn't progressed were randomized 2:1 to receive ongoing gemcitabine until progression or stop treatment and be observed, not starting any further treatment until progression. (click on image to enlarge) The study looked at progression-free survival (PFS) as the primary endpoint and noted a near doubling, from 2.0 to 3.6 months, with continuation maintenance therapy with gemcitabine - a result that was highly statistically significant. There wasn't a significant difference in overall survival, but the absolute difference of two months was certainly encouraging and supportive of the concept that maintenance therapy could be of value. A key limitation of this trial was that the number of patients randomized was pretty small - just over 200 - because we know that at least 40% of patients who start first line therapy will drop out due to progression or side effects before getting out to four cycles. Trials that start enrolling after first line (like the trial of Alimta vs. placebo as a maintenance therapy) are somewhat easier to conduct because all of the patients you enroll will have already jumped through the big hurdle of getting through four cycles of first line therapy without progressing or declining clinically, and nearly all can be randomized and analyzed for the benefit of switch maintenance. But the trials that test continuation maintenance need to enroll nearly twice as many patients as they need to be randomized to continuation maintenance therapy, because they must give a uniform first line therapy for four cycles before they can even ask the question of the value of continuation maintenance. Consequently, many trials, like the one described above, tend to be under-powered to demonstrate statistical significance, even if the difference they are showing is clinically significant (and I consider a survival difference of two months for patients with advanced NSCLC to be clinically significant). This year, ASCO featured a couple of trials that looked specifically at continuation maintenance therapy, and one of these even addressed the question of comparing continuation vs. switch maintenance therapy. The first, presented by Dr. Chandra Belani, attempted to enroll 600 patients to receive up to four cycles of carboplatin/gemcitabine, then randomize non-progressing patients 1:1 to either ongoing gemcitabine or observation (really very similar to the Brodowicz trial described above, except a larger study, and done with carboplatin instead of cisplatin for first line treatment). It closed earlier than planned, with 519 patients enrolled, because of slower accrual than expected. What was especially notable about this population was that nearly 2/3 had a marginal performance status of two, which is very different from the population that other maintenance therapy trials have enrolled (all or nearly all with a better performance status of 0 or 1). In this population, fewer than 50% of the patients initially enrolled (255 of the 519) proceeded through four cycles and to the point of being randomized to ongoing gemcitabine or observation. Dr. Belani reported that despite the fact that more patients responded (28% vs. 6%) and fewer progressed (9% vs. 17%) on maintenance gemcitabine vs. observation, respectively, there was no hint of improvement in progression-free survival or overall survival with the continuation maintenance with gemcitabine: These negative results could be due to any of a number of possible explanations, or a combination of them. First, perhaps switching to a different drug and attacking the cancer from a new angle may be critical. Perhaps continuation maintenance therapy is just inferior to switch maintenance therapy. Second, continuation therapy may be effective with the right agent, but perhaps gemcitabine isn't an effective agent in this setting. Third, the trial results may have largely been a by-product of enrolling so many patients with a marginal performance status, who may be the patients for whom ongoing treatment may be as harmful to the patient as it is beneficial in treating the cancer. We also know that only a minority of patients (less than 20%) received additional therapy, a lower proportion than we'd generally expect in a population that hadn't progressed after four cycles of first line therapy. We know that carboplatin/gemcitabine, despite its many favorable features, can be a very myelosuppressive chemo regimen (meaning that it leads to markedly decreased blood counts). It may well be that the disproportionately large population of frail patients enrolled on this trial were the exact people least likely to be able to tolerate ongoing myelosuppressive treatment after this regimen. But another provocative trial provides additional insight. A French trial presented by Dr. Perol administered first line cisplatin/gemcitabine for four cycles to 834 patients with advanced NSCLC (all with a good performance status of 0-1), then randomized the 464 without progression or prohibitive toxicity 1:1:1 to one of three options: observation alone, continuation maintenance therapy with gemcitabine, or switch maintenance therapy with the oral EGFR inhibitor Tarceva (erlotinib). Importantly, the trial was designed to compare each of the maintenance therapies with observation alone, not really to compare the two maintenance therapies to each other directly. The enrolled population reportedly included 38% never-smokers, which is higher than you'd expect, and Dr. Perol noted that it's possible that some (many?) former smokers were misclassified as never-smokers. About 2/3 of patients had an adenocarcinoma. These factors are relevant because they are associated with a higher probability of patients having an EGFR mutation, which is highly associated with a very significant clinical benefit from Tarceva. The investigators collected tissue for mutation testing, but these results weren't yet available for presentation. There were no real surprises in side effects: gemcitabine was associated with more problems with low blood counts and higher risk for needing a transfusion or growth factor support (erythropoietin or G-CSF). Tarceva was associated with more rash and diarrhea. These findings are just what we would expect. PFS was the primary endpoint, and the trial demonstrated that PFS was significantly better with either of the two maintenance therapies compared with observation alone, though the numbers were more impressive with continuation gemcitabine. Gemcitabine was associated with a highly significant hazard ratio of 0.55, meaning a 45% improvement in the time before patients developed progression. The chance of being progression-free at three months was 55% with maintenance therapy, vs. 30% with observation alone; at 6 months, the difference was 22% vs. 9%. The benefit was really seen across all patient subgroups (smokers or never-smokers, all histologies, men vs. women, etc.), though there was a tendency for the greatest benefit to be in the patients who had a significant response to the cisplatin/gemcitabine initially. This shouldn't be surprising, but it highlights that for patients who respond robustly in the first line setting, they don't necessarily exhaust all of the benefit of a chemo agent after four cycles. But it also highlights that patients don't necessarily "max out" on their benefit from some first line therapies after four cycles if they achieved a good initial response. For Tarceva, there was a benefit as well, but it did look less pronounced. Here, the hazard ratio for PFS was 0.81, significantly favoring Tarceva (p - 0.002), with PFS at 3 months of 35.3% vs. 30.3%, and PFS at 6 months of 16.3% vs. 8.6%. As with gemcitabine, the benefit in progression-free survival was similar across all patient subgroups, showing no clear differences based on smoking status, tumor histology, or, unlike gemcitabine, response to prior chemotherapy. This latter difference makes sense to me, because Tarceva is not only a new treatment but also represents more of a "wild card". People who have a chemo-sensitive tumor often respond similarly well when they receive other chemo agents, while those who have a chemo-resistant cancer tend to have more disappointing results with various chemo agents and regimens. But while patients with an EGFR mutation seem to have a higher probability of responding to chemo as well as an EGFR inhibitor, overall it is a treatment that is different enough from standard chemo that prior chemo results are not very predictive of how someone will do on an EGFR inhibitor. Finally, we saw some preliminary survival data that indicated a modestly higher overall survival with both maintenance strategies compared with observation alone. The HR for OS was 0.86, or about a 14% improvement, with gemcitabine compared with observation, while for Tarceva, the HR for OS was 0.91 (9% improvement). In absolute terms, the median OS was 12.1 vs. 11.8 vs. 10.7 months for gemcitabine vs. Tarceva vs. observation, respectively. It should be emphasized that this trial, unlike most others testing maintenance therapy, built in subsequent therapy with Alimta (pemetrexed) (though it enrolled all NSCLC histologies, this trial was developed before it was known that Alimta is active only in non-squamous NSCLC subtypes). About 2/3 of the patients received subsequent Alimta, eliminating questions about timing of treatment vs. overall access, but also obscuring any effects on overall survival. So what can we say, putting this all together? First, it appears that unlike the aforementioned trial of continuation maintenance therapy presented by Dr. Belani that was also with gemcitabine, the French study showed benefits with continuation therapy that were every bit as good, and even numerically a little better than with a switch maintenance therapy that is already FDA approved. I think it's fair to conclude that the negative results from the Belani trial are likely related to the fact that nearly two thirds of the enrolled had a compromised performance status. With the more favorable results from the French trial with a very similar treatment plan, it appears that patients who aren't in as robust a shape after first line chemo receive the side effects but no real improvement in outcomes, while the more fit patients appear to do better with the same continuation therapy. The French trial highlights that an ongoing treatment can be beneficial in patients who have already completed four cycles of first line chemotherapy, perhaps especially in those who had a very good response early on. The other difference between the Belani trial and the Perol study is that the former started with carboplatin/gemcitabine, the latter with cisplatin/gemcitabine. It's possible that the more myelosuppressive (blood count dropping) regimen with carboplatin could have made it more problematic, or at least less effective, to deliver additional gemcitabine. Though the studies that led to the recent FDA approvals of both Alimta and Tarceva both used a strategy of switch maintenance, we now have some evidence that continuing one or more agents from the front line setting may be at least comparably effective, even if it's with an agent that hasn't been as well studied as a maintenance therapy. These studies all have limitations, but they add to our growing body of information helping us understand that some patients but not all benefit from maintenance therapy, and also that continuing on an effective therapy from the first line setting may be at least as helpful as coming at the cancer from a new angle.