A manuscript by Pérol and colleagues just being published in the Journal of Clinical Oncology describes a trial I’ve found intriguing but inexplicably ignored by most lung cancer leaders since it was presented at ASCO 2010, and that I reviewed in preliminary form then.   A French study with the decidly un-sexy name IFCT-GFPC 0502, it is the first and only one we can turn to for direct insight on the comparison of continuous to switch maintenance therapy, and I think it provides several helpful points for clinical decision-making.

The Pérol trial enrolled an initial cohort of 834 patients with previously untreated advanced NSCLC of any histologic subtype, who were then treated with first line cisplatin/Gemzar (gemcitabine). The 464 (56%) who proceeded through 4 cycles without progression or particularly nasty side effects were then randomized to either ongoing continuation maintenance therapy with gemcitabine alone until progression (N = 154), switch maintenance with erlotinib (N = 155), or a control arm that received a treatment break and no further therapy until disease progression (N = 155).  At the time of progression, all of the three groups were assigned to transition to second line Alimta (pemetrexed) after progression (including the approximately 20% with squamous histology, since this trial was developed prior to the recognition that pemetrexed is not active in patients with squamous NSCLC).  The primary endpoint of the study was progression-free survival (PFS).

The study was designed to compare each of the maintenance arms independently to the control arm of observation/treatment break alone, but it does lend itself to a comparison of the two maintenance strategies, at least for some general if not strictly statistically kosher conclusions.  PFS was significantly improved compared with no maintenance therapy for both the continuation maintenance (HR 0.56, P < 0.001) and switch maintenance (HR 0.69, P = 0.003) approaches, as shown in the figures below.

Interestingly, there was a trend toward a more striking improvement in PFS in the continuation maintenance therapy arm with Gemzar among patients who had demonstrated a prior objective response than in patients who had achieved stable disease as a best response to cisplatin/Gemzar, while there was no such trend toward more favorable PFS benefit in the subset of patients who had experienced more tumor shrinkage from first line chemotherapy.  I can certainly understand this difference: the patients who derive the greatest benefit from continuation maintenance therapy are those who had demonstrated the greatest benefit from the same therapy earlier.  This same concept that was also demonstrated in the PFS but not overall survival (OS) results from the PARAMOUNT trial of continuation maintenance therapy with Alimta.

In contrast with the significant difference in PFS in both arms of this study, there were only weak, non-significant trends in OS (HR about 0.9) in both of the maintenance therapy arms compared with the control arm.  I believe this was related to the beneficial effect of second line therapy that was received by the vast majority of the trial participants (91% of the patients in the control arm and 77% and 80% of the patients in the continuation and switch maintenance therapy arms, respectively), which likely led to an equalizing effect among the three arms.  Side effects were greater in the maintenance therapy arms but were not not excessive or unexpected.

There are certainly criticisms of this trial that can be made. It isn’t particularly large, with just over 150 patients per arm, and it is fair to criticize the fact that about 20% of the patients had a squamous NSCLC and therefore likely didn’t benefit from second line Alimta.  Despite these limitations, I note four key take-away points: 

1)   Though the trial technically doesn’t compare the two strategies for maintenance strategies directly, the fact that each is compared to a common control arm leads to a conclusion that continuation maintenance therapy is in definitely not inferior to switch maintenance, despite the fact that the latter introduces a new active therapy after four cycles of first line therapy.  This is the case even with continuation maintenance Gemzar, an agent for which there is little enthusiasm and no FDA approval as maintenance therapy, compared with switch maintenance with Tarceva, a drug for which there is an FDA approval in this setting.

2)   In patients who haven’t progressed, the benefit of the non-platinum agent in the first line chemotherapy regimen is not necessarily exhausted after four cycles of treatment, as was also illustrated in the PARAMOUNT trial. Gemzar demonstrated a trend toward greater ongoing incremental benefit in patients with a prior response.

3)   I would contend that the absence of a survival benefit in either maintenance therapy arm is related to the fact that the majority of patients received an active agent (Alimta) at progression, which blunted the benefit of maintenance therapy.  It is not a coincidence that the trials that have demonstrated a survival benefit have also demonstrated an imbalance in delivery of active salvage therapy between the maintenance therapy and control arms in various other trials.  I strongly suspect that if all patients who received a treatment break after first line therapy were to be followed closely and started promptly on the same treatment that was given to the other arm in the maintenance therapy trials that have been positive for OS benefit, that benefit would be compromised. 

4)   Gemzar held its own against erlotinib here, and we could also look to the trial of immediate vs. delayed Taxotere (docetaxel) after first line carboplatin/Gemzar, with a 2.6 month difference in OS that barely missed statistical significance at P = 0.08, as another maintenance therapy strategy that looks very comparable to some of our FDA approved and heavily marketed strategies.   Our practice patterns may be shaped largely by FDA approvals, but the actual data should lead us to recognize that the treatment options for which FDA approvals have been sought are related to marketing interests as much as or more than the strength of the data.  Clinical judgment could very well lead a thoughtful oncologist to consider maintenance Gemzar or Taxotere.

Putting it all together, I conclude that the principles applied for maintenance therapy are likely generalizable across several agents and likely comparably between continuation and switch maintenance therapy.  Moreover, I see that the effects of maintenance therapy will be blunted if patients receive subsequent effective therapy.  The leading benefit of a maintenance approach is the only reliable way to ensure that the patients most likely to benefit from further treatment – the patientss who demonstrated a response or stable disease with first line therapy – actually receive it.