Maintenance therapy in NSCLC: why? To whom? Which agent?

Lung cancer is the leading cause of cancer mortality in USA and worldwide more than one million people die from this disease every year: the overall 5-year relative survival rate measured by the Surveillance Epidemiology and End Results program in USA is 15.8% [1]. Approximately 87% of lung cancer cases are Non Small Cell Lung Cancer (NSCLC) and the majority of patients presents with advanced stage disease at diagnosis [2, 3]. In two independent phase III trials the addition of bevacizumab to standard first-line therapy was shown to improve both overall response rate (ORR) and PFS, although OS advantage was demonstrated in only one of these studies [4, 5]. In combination with platinum-based chemotherapy, cetuximab has also demonstrated a small statistically significant OS advantage as compared to chemotherapy alone [6]. Second-line treatment has been shown to improve survival and to palliate symptoms: approved treatment options include cytotoxic chemotherapy (docetaxel or pemetrexed) or epidermal growth factor - EGFR tyrosine kinase inhibitors (erlotinib or gefitinib) [7, 8]. However, only approximately 50% of the patients will be able to receive second-line therapy, mainly because of the worsening of clinical conditions [9].

The U.S. National Cancer Institute's medical dictionary defines maintenance therapy as "any treatment that is given to keep cancer from progressing after it has been successfully controlled by the appropriate front-line therapy; it may include treatment with drugs, vaccines or antibodies, and it should be given for a long time". Maintenance therapy has also been referred to as "consolidation therapy" or "early second-line therapy", depending on treatment type and timing of the specific therapeutic agent employed [10]. The latter definition is probably the least appropriate, because "second-line" implies a disease progression event, which, by definition, is not the case for the maintenance setting and the term "switch maintenance" (used in the National Comprehensive Cancer Network - NCCN - Clinical Practice Guidelines) appears more precise[11].

Currently, for advanced NSCLC the options to continue treatment after first-line induction include: 1) continuing induction therapy for a fixed number of additional cycles over the standard or, when possible, until progression; 2) continuing only the third-generation non-platinum compound used in the induction regimen; 3) switching to a different agent after induction therapy.

The first American Cancer Society of Clinical Oncology (ASCO) guidelines, published in 1997, addressed the appropriate duration of therapy in advanced NSCLC recommending no more than eight cycles, even if in most clinical trials the median number of delivered cycles is typically three or four [12]. Four trials clarified that were no response, survival or QoL differences between short versus longer treatments in advanced NSCLC but an increased risk for cumulative toxicity only (Table 1) [13‐16]. As consequence ASCO changed recommendations regarding the appropriate duration of therapy in 2003, stating that treatment should have been stopped at four cycles for non responders patients and no more than six cycles should have been administered for any patient; no major changes for this specific issue were reported in the ASCO guideline update in 2009 [17, 18].

In patients responding or stable after the induction, a maintenance strategy should be to continue the same therapy withholding platinum, in an attempt at consolidating disease control and increasing survival, maintaining tolerability within acceptable limits.

The European Cooperative Oncology Group conducted a phase III trial testing gemcitabine maintenance versus best supportive care (BSC) in 350 patients with complete/partial response or stable disease after four cycles of gemcitabine/cisplatin induction, randomized in a 2:1 ratio. Sixty one percent of patients (among 73% of responders after the induction) were randomized: during the maintenance period, patients received a median of three cycles of gemcitabine (range: 0-38 cycles). Median TTP was significantly longer in the gemcitabine arm both throughout the study (6.6 versus 5 months, p < 0.001) and during the maintenance period (3.6 versus 2 months, p < 0.001). Median OS in the gemcitabine arm was 13 months, compared to 11 months in the BSC arm (p = 0.195). In terms of toxicity, the most important difference between the two arms during the maintenance phase was the need for red blood cells transfusions (20% in the gemcitabine arm versus 6.3% in the BSC arm, p = 0.018) [19]. Another phase III trial comparing gemcitabine versus BSC as maintenance therapy for patients not progressing after 4 cycles of gemcitabine/carboplatin induction was recently presented. Two hundred and fifty five patients (among 519 enrolled) were randomized; median PFS was 3.9 months (95% CI: 3.3-5.6) for the experimental arm and 3.8 months (95% CI: 2.6-5.5) for the BSC arm; median OS (primary end point) was 8 months (95% CI: 6.0-10.2) for the gemcitabine maintenance arm and 9.3 months (95% CI: 7.7-12.7) for the BSC arm, without any statistical difference [20]. In a third trial employing gemcitabine or erlotinib maintenance after 4 cycles of gemcitabine/cisplatin induction and with a preplanned II-line treatment option (pemetrexed), PFS (primary end point) by independent review was significantly prolonged by both G (HR 0.51, 95% CI 0.39-0.66) and E (HR 0.83, 95% CI 0.73-0.94), as compared to O. OS data are not yet mature [21]. Belani et al. treated 401 patients with carboplatin and paclitaxel for 16 weeks; responding patients were then randomly assigned to receive weekly paclitaxel maintenance or BSC. Response was seen in 130/390 evaluable patients, who were deemed eligible for randomization into the maintenance phase, during which only 23% completed four cycles. Median TTP (primary endpoint) was 38 weeks in the paclitaxel arm versus 29 weeks in the BSC arm (p not reported); median OS was 75 and 60 weeks in the paclitaxel and BSC arm, with 1-year survival rates of 72% and 60%, respectively. During maintenance therapy, 86% of patients in the chemotherapy arm experienced at least one adverse event and 45% reported at least one grade 3 or 4 adverse event [22].

According to the Goldie-Coldman hypothesis, showing that even the smallest detectable cancers contain at least one drug resistant clone and that increasing numbers of resistant clones emerge as tumors grow and progress, a rational strategy would be to use all effective drugs as early as possible in the treatment program [23, 24]. Different non-cross-resistant agents have been used as a maintenance strategy after a defined number of induction cycles with a platinum-based regimen in several randomized clinical trials (Table 2).

As highlighted in the previous paragraphs, evidence on the continued (maintenance) use of the same third-generation agent employed in the induction regimen remains inconclusive with respect to gemcitabine and frankly negative in terms of cost/benefit ratio with respect to weekly paclitaxel [20‐22, 34]. Nowadays, available data about pemetrexed in maintenance setting do not answer to the question if this approach could be useful in those patients responding to a first line with platinum compound and pemetrexed and the answer will be available soon from a randomized trial comparing pemetrexed versus placebo in patients who do not progress following four cycles of pemetrexed plus cisplatin [35]. Positive data in terms of cost-effectiveness switching to pemetrexed, which employment in non-squamous NSCLC is really cost-effective, are driven by its impact on PFS and OS [36]. This is indeed a crucial point: resources use and costs involved with this new paradigm in the clinic, would all argue for a meaningful improvement in survival as a critical necessity from a practical standpoint. As a consequence, the usefulness of maintenance therapy has to be based on a clearly defined, reproducible and measurable endpoint. Using PFS as the basis for the adoption of a new therapeutic approach, may be considered as a limitation due to the variability in the definition of progression and frequency of response assessment across studies; in this context, it seems very relevant to standardize PFS measurement in definitive phase III trials. For example, in the Fidias trial, patients on the immediate docetaxel arm underwent radiologic assessment after cycles two, four and six, while patients in the delayed docetaxel arm the evaluation was performed every three months. Timing and the type of imaging studies used in the control arm has been considered one of the main limitations of this study, as unfavorably delaying detection of possible disease progression [37]. As it happens in routine daily practice, only about two thirds of patients on the control arm was able to receive second-line docetaxel, as opposed to 95% of patients who received the study drug in the immediate, maintenance arm; thus, the true benefit with "immediate" docetaxel in this study could be entirely attributed to the higher proportion of patients receiving active therapy in the maintenance setting. Indeed, a post-hoc analysis documented an identical OS duration of 12.5 months for patients who received docetaxel on either arm of the study, clearly indicating that when patients stop first-line chemotherapy, they should be followed closely to detect progression early and at a time when they remain fit for further treatment [24].

The benefit of maintenance therapy can be confounded by the absence of a predefined post-study treatment. Indeed, in JMEN trial (as well as in other ones) the discretion given to investigators in the choice of second-line therapy has been addressed as a major limitation, because it fails to provide any insight into the possibility that the benefit of maintenance therapy may be obtained also by the appropriate use of the same agent as salvage therapy at the time of disease progression. In that respect, the design of the Fidias' trial, with all patients receiving docetaxel as either maintenance or second-line treatment, appears to be a methodologically more correct study design to test the efficacy of a strategy introducing a non cross-resistant agent before progression. In the SATURN trial only a minority of patients assigned to placebo actually received an EGFR-TKI: with the current evidence, we do not know if the improvement in OS observed with maintenance erlotinib would have been the same, or reduced, if the study protocol had imposed cross-over after disease progression. Importantly, the adoption of a pre-specified, built-in second-line treatment option offers the advantage of reducing the proportion of patients who do not get access to further treatment, as demonstrated in the recently reported trial from Perol, in which more than 80% of patients in the observation arm received second-line pemetrexed [21, 30, 31].

Even if a bevacizumab maintenance in patients receiving bevacizumab combined with chemotherapy in the context of their first-line regimen is considered common practice on the basis of the registration trials, both of which maintained bevacizumab until progression after the completion of the assigned first-line regimen, with the notable exception of the recently-presented ovarian cancer trial clearly supporting the use of maintenance bevacizumab, this specific issue has never been assessed in ad hoc designed randomized trials [4, 5, 38]. Currently there are at least two trials designed to clarify its role in maintenance: the ECOG three-arm, phase III study of Paclitaxel/Carboplatin/Bevacizumab followed by randomization to pemetrexed versus bevacizumab versus pemetrexed/bevacizumab in non-squamous carcinoma and a study with Pemetrexed/Cisplatin/Bevacizumab followed by Pemetrexed/Bevacizumab versus Bevacizumab alone [39]. The approximately 4-month median PFS with single-agent erlotinib maintenance in the SATURN trial and 4.76 months with the combination of erlotinib and bevacizumab in the ATLAS trial, highlights the importance of establishing the relative contribution of each agent when a combination therapy strategy is being evaluated in the maintenance setting [31, 32]. Another related question is whether subgroups of patients with specific clinico-pathological and/or molecular characteristics would especially benefit from the choice of a particular maintenance agent, among those currently available. Within the limits imposed by such methodological considerations, the only biomarker that clearly showed a statistically significant, quantitative interaction with the treatment assigned (erlotinib or placebo) was the presence of sensitizing EGFR mutations (p for interaction <.001); indeed, although even EGFR M- patients derive a small, but statistically significant, benefit in PFS from erlotinib maintenance (HR: 0.78, 95% CI: 0.63-0.96, p = 0.0185), the PFS gain of EGFR M+ patients is exceptionally wide (HR: 0.10, 95% CI: 0.04-0.25, p < 0.0001). The potential benefits of the inclusion of erlotinib in the maintenance treatment of EGFR M+ patients were consistent in the ATLAS trial, where erlotinib was combined with bevacizumab. However, at the moment there are no survival data and no further analyses of OS are planned, due to loss of patients to follow up [32]. In routine clinical practice obtaining information on EGFR mutational status is not always easy and time-consuming, being not exceptional that such information becomes available only when the patient is already receiving a standard first-line chemotherapy treatment: should this be the case, EGFR M+ patients have now the option to receive TKI right after the induction. The impact of erlotinib maintenance on OS of EGFR M+ patients, however, is currently uncertain. Survival data in EGFR M+ patients included in SATURN trial are not yet mature although the low number of EGFR M+ patients and the shape of the survival curves, make it unlikely that a statistically significant benefit will become apparent with longer follow up. It is true that EGFR TKI are effective in advanced NSCLC even when administered late in the course of the disease, but recent data document that about 50% of NSCLC patients treated with EGFR-TKIs will develop resistance-inducing EGFR mutations (such as the T790M) implying the possibility that resistant clones may expand as disease progresses [40‐42]. Talking about costs in this specific context a recent retrospective cost-effectiveness analysis by Bradbury et al. reported the cost per year of life gained being not the most favorable in patients with sensitizing mutations in the EGFR gene. This was because these patients derived relatively greater benefit and stayed on treatment longer, thereby incurring considerably higher drug acquisition costs [43]. Besides EGFR mutations, histology represents a potentially crucial decision factor for the choice of specific maintenance agents. Currently, no direct comparisons between different agents in histology-selected subgroups of patients have been reported. In the JMEN trial, the benefit of maintenance pemetrexed is clearly confined to patients with non-squamous histology: indeed, in patients with squamous histology OS on pemetrexed maintenance was indistinguishable from that on placebo; conversely, in non-squamous patients pemetrexed maintenance resulted in a reduction of the risk of death of approximately 30% and prolonged median survival from 10.3 to 15.5 months [27]. In the SATURN trial, non-squamous patients on erlotinib maintenance experienced a 21% reduction in the risk of death and a prolongation of median survival from 10.5 to 13.7 months [31]. Similar results were obtained in the IFCT-GFPC trial (for which only PFS data are available), where the benefit for erlotinib maintenance was also confined to adenocarcinoma patients [21]. Conversely, in the ATLAS trial the benefit in OS gained from the addition of erlotinib to bevacizumab is very limited in both the adenocarcinoma and non-adenocarcinoma groups of patients (HR 0.91, 95% CI 0.74-1.12 and HR 0.98, 95% CI 0.64-1.49, respectively) [32]. Overall, in patients with non-squamous histology pemetrexed maintenance appears to provide the greatest benefit in terms of both PFS (HR 0.44) and OS (HR 0.70). Erlotinib also represents a reasonable choice (HR 0.60 and 0.79 for PFS and OS respectively) and may possibly be preferable in selected subgroups, such as females (HR 0.64 for erlotinib vs. HR 0.83 for pemetrexed) and east Asians patients (HR 0.66 for erlotinib vs. HR 1.05 for pemetrexed). An improvement in PFS was obtained with either erlotinib in patients with squamous histology in the SATURN trial (HR 0.76, 95% CI 0.60-0.95) or gemcitabine in patients with non-adenocarcinoma histology in the IFCT-GFPC trial (HR 0.56, 95% CI 0.37-0.85)[21, 32]. Many other phase II and III trials are currently ongoing looking at maintenance therapy in NSCLC (Tables 3 and 4) [35, 39, 44, 45]. Modulating the immune response in lung cancer is a strategy that is being actively investigated also in maintenance approach. The L-BLP25 (Stimuvax; Biomira Alberta, CA) is a liposome vaccine targeted to the extracellular core peptide of mucine 1 (MUC 1), a transmembrane protein expressed on epithelial cells. In a phase IIb trial, patients in stage III NSCLC, who had disease control after induction therapy, were randomized to receive vaccination weekly for 8 weeks and then they had the option to proceed to maintenance therapy, consisting in vaccination every 6 weeks or BSC. The median OS (primary endpoint) was 17.4 months for the vaccinated patients versus 13.0 months for those on BSC arm (p = 0.66)[46].

Since no comparative trials of maintenance with different chemotherapy drugs or targeted agents have been conducted, no conclusive data are available yet of an advantage of maintenance therapy. As consequence, the choice and the duration of maintenance treatment remains largely empirical and needs to be explained and discussed with each patients in terms of current trials, different toxicity profiles (fatigue and myelosuppression on chemotherapy versus rash and diarrhea on EGFR TKis) or intra venous versus oral treatment options [47]. On the basis of the previous data in patients for whom maintenance therapy is deemed appropriate and who accepted to prolong treatment: i) switching to a non cross-resistant agent appears to provide greater benefit than continuing on one of the agents employed in the induction regimen (although critical information on this issue will be provided by the PARAMOUNT S124 trial, which investigated pemetrexed maintenance after pemetrexed/cisplatin induction and recently concluded enrollment) ii) in patients harboring sensitizing EGFR mutations, erlotinib (either alone or combined with bevacizumab for patients who have received bevacizumab as part of their primary treatment, if future data will state a benefit) currently appears to be the agent of choice iii) in patients with non-squamous/adenocarcinoma histology and with wt-EGFR or unknown mutational status, pemetrexed appears to provide the greatest advantage, although erlotinib, and to a lesser extent gefitinib (where available), may be reasonable alternatives for selected patients, taking into account the possible patient preference for an oral treatment option iv) patients with squamous histology and patients with KRAS mutations have limited treatment options and should be enrolled in specific clinical trials whenever possible.