Maintenance treatment after induction therapy in non-small cell lung cancer: latest evidence and clinical implications

Continuation maintenance

The platinum component of platinum-based doublet therapy is responsible for the majority of reports of serious treatment-related toxicity. For this reason, most patients cannot tolerate prolonged administration of platinum-based doublets. However, this is not true for third-generation single-agent treatment and targeted therapies, in which some patients can get multiple cycles of second-line therapy with minimal toxicity [Hanna et al. 2004]. Because the tolerability of these newer drugs, specifically gemcitabine, docetaxel, pemetrexed, and erlotinib, is significantly better than many second-generation chemotherapeutics, a recent strategy has been to investigate the value of continued treatment of responding and stable patients with the nonplatinum component of their induction regimen, ‘continuation maintenance’ therapy.

The first study to use continuation maintenance with newer agents was ECOG 4599 [Sandler et al. 2006]. In this study, 878 patients with advanced nonsquamous NSCLC were randomized to carboplatin, paclitaxel, and bevacizumab induction followed by bevacizumab maintenance or carboplatin and paclitaxel induction alone. There was a 2-month improvement in median OS for patients who received bevacizumab (12.3 months versus 10.3 months; hazard ratio [HR] 0.79, p = 0.003). Similar results were noted in FLEX, a randomized phase III trial that added cetuximab to induction chemotherapy and continued cetuximab maintenance for those with nonprogressive disease [Pirker et al. 2009]. In this study, the addition of cetuximab to cisplatin and vinorelbine chemotherapy improved OS from 10.1 to 11.3 months (HR 0.87, p = 0.044). Although these studies were not randomized maintenance studies to specifically evaluate the effect of continuation maintenance, they provided a proof-of-concept that additional therapy prolongs survival and that some of the benefit could be attributable to maintenance therapy after induction.

Two phase III European trials, CECOG [Brodowicz et al. 2006] and IFCT-GFPC 0502 [Perol et al. 2012], have investigated gemcitabine as a maintenance agent. These studies demonstrated improvements in PFS of 1.5 to 2 months with maintenance gemcitabine in patients without progression after induction with cisplatin and gemcitabine. A third study, presented only in abstract form, failed to show an improvement in PFS or OS with continuation maintenance gemcitabine after carboplatin and gemcitabine induction [Belani et al. 2010]. The difference between this study and the two European studies has been attributed to a patient population with worse performance status (PS). The majority, 64%, of patients had a PS of ≥2 at the time of randomization. However, it is worth noting that the median PFS was 3.8 and 3.9 months in the control and experimental groups, respectively, which was similar to median PFS in the experimental arms of the European studies. These two positive gemcitabine maintenance trials included patients with all NSCLC histologies, were all similarly sized with between approximately 120 and 150 patients per maintenance arm, and had primary endpoints of PFS. Neither achieved a statistically significant improvement in OS, although they were not powered to detect this difference.

PARAMOUNT [Paz-Ares et al. 2013] was a phase III study that evaluated pemetrexed as continuation maintenance therapy, having been used previously only as switch maintenance (see the description of the JMEN study in the following section). This study, the largest maintenance study conducted at the time, enrolled over 900 patients with nonsquamous NSCLC and randomized those with nonprogressive disease (539 patients) after four cycles of cisplatin and pemetrexed to maintenance pemetrexed versus best supportive care (BSC) in a 2:1 randomization in favor of pemetrexed. The final analysis confirmed that maintenance pemetrexed prolonged OS by 3 months compared with BSC (16.9 months versus 14 months; HR 0.78, p = 0.019).

Switch maintenance

The central premise of the switch maintenance strategy is the Goldie–Coldman hypothesis [Goldie and Coldman, 1979], which states that even small cancers harbor at least one drug-resistant clone. To overcome resistance, the use of different, non-cross-resistant chemotherapy regimens has been investigated. This hypothesis is the basis of maintenance trials in which patients with NSCLC are switched to a new, potentially non-cross-resistant agent if they respond or remain stable after initial chemotherapy. Many see this strategy as early second-line treatment, and this treatment paradigm is likely the most debated.

One of the first trials to demonstrate a benefit of maintenance with cytotoxic chemotherapy was a phase III study by Fidias and colleagues [Fidias et al. 2009] that randomized over 300 patients with advanced NSCLC to docetaxel immediately following four cycles of carboplatin and gemcitabine induction versus delayed docetaxel at the time of disease progression. Early docetaxel therapy resulted in a significant 3-month improvement in PFS (p = 0.0001). An OS improvement of approximately 2½ months did not reach statistical significance, possibly because the study was underpowered to detect an OS benefit shorter than 4 months. Toxicity and QOL were similar in both groups. Notably, approximately one-third of patients randomized to delayed docetaxel never received the intended second-line therapy due to disease progression, patient decision, or death. The patients who did receive docetaxel in the delayed treatment arm had similar OS as those treated in the immediate docetaxel arm.

Pemetrexed has been shown to be noninferior and better tolerated than docetaxel in the second-line setting [Hanna et al. 2004]. Pemetrexed has also gained approval for the first-line setting of nonsquamous NSCLC based upon a phase III noninferiority study of cisplatin and pemetrexed versus cisplatin and gemcitabine (10.3 months OS in both arms) [Scagliotti et al. 2008]. However, in a preplanned subset analysis, survival in the 847 patients with adenocarcinoma was significantly prolonged with cisplatin plus pemetrexed compared to cisplatin plus gemcitabine (median 12.6 versus 10.9 months). Conversely, cisplatin plus gemcitabine was superior to cisplatin plus pemetrexed in the 473 patients with squamous cell carcinoma (SCC; median 10.8 versus 9.4 months).

JMEN, a landmark phase III study that evaluated maintenance pemetrexed and that began prior to the publication of the study by Scagliotti and colleagues, was the first to show an improvement in OS with maintenance therapy [Ciuleanu et al. 2009]. In this study, 663 patients with NSCLC who had completed induction chemotherapy with a platinum doublet without progression were enrolled and randomized to pemetrexed switch maintenance or BSC. No patients in this study received pemetrexed as part of their induction regimen because pemetrexed had not yet been shown to be effective in the first-line setting at the time of patient enrollment. These patients, approximately 25–30% of whom had SCC, were randomized in a 2:1 fashion to pemetrexed or BSC. OS was significantly longer in those receiving pemetrexed (13.4 versus 10.6 months; HR 0.79, p = 0.012). In subgroup analysis by histology, there was no difference in PFS or OS in those with SCC. The effect on those with nonsquamous NSCLC was even more dramatic, with an OS prolonged by nearly 5 months (15.5 versus 10.3 months; HR 0.70, p = 0.002). This study has been criticized because 82% of the patients on the BSC arm never received second-line pemetrexed, and many (33%) were not treated with any post-discontinuation therapy.

Erlotinib is appealing as a maintenance agent due to its oral administration and tolerability. Subsequent to the BR.21 study that identified an OS benefit for erlotinib in the second- or third-line setting for patients with NSCLC [Shepherd et al. 2005], two clinical trials have studied erlotinib as a single-agent maintenance strategy. SATURN was a phase III study that randomized 889 patients with nonprogressive disease after four cycles of chemotherapy to maintenance erlotinib or placebo [Cappuzzo et al. 2010]. The use of erlotinib maintenance was associated with approximately 1-week and 1-month improvements in PFS and OS, respectively, and was well tolerated. These results were replicated in the IFCT-GFPC 0502 study, in which patients were randomized to maintenance gemcitabine (see the continuation maintenance section above), erlotinib, or placebo after four cycles of cisplatin and gemcitabine induction chemotherapy [Perol et al. 2012]. Those randomized to the erlotinib arm had a 1-month improvement in PFS when compared with placebo; an improved OS did not reach statistical significance. The erlotinib arm was approximately one-third the size of those receiving erlotinib in the SATURN trial, which could account for the reduced power to detect a statistically significant OS benefit.

Histology

Since the recognition that the benefit of pemetrexed is limited to nonsquamous NSCLC [Scagliotti et al. 2008], histology has become an important consideration in all patients with NSCLC. In JMEN, where unselected patients were treated with pemetrexed, patients with SCC had no benefit from maintenance therapy in subset analysis [Ciuleanu et al. 2009]. All subsequent pemetrexed maintenance trials have included only patients with nonsquamous NSCLC. Therefore, a pemetrexed maintenance strategy, whether with continuation or switch maintenance, is only applicable to patients with nonsquamous NSCLC. The efficacy of maintenance with gemcitabine or erlotinib may be less contingent upon histology and have broader applicability.

In a meta-analysis of the three maintenance trials that studied erlotinib, SATURN, ATLAS, and IFCT-GFPC 0502, all subgroups of patients, including those with SCC, derive a benefit in PFS (all histology: HR 0.76, p < 0.00001; nonsquamous NSCLC: HR 0.71, p < 0.00001; SCC: HR 0.83, p = 0.02) [Petrelli et al. 2011]. However, data from SATURN and IFCT-GFPC 0502 suggest the benefit may be smaller in those with squamous histology. In SATURN, there was an OS benefit only in patients with adenocarcinoma, and the PFS benefit appears to have been more profound in those with nonsquamous NSCLC (HR 0.60; 95% confidence interval [CI] 0.48–0.75) compared with those with SCC (HR 0.76; 95% CI 0.60–0.95), although the study was not powered to compare these two groups [Cappuzzo et al. 2010]. The PFS difference between these two subgroups (HR 0.60 versus 0.76), however, is likely not clinically significant because a HR of 0.71 for the larger cohort correlated to a difference in PFS of only 1 week. Unlike the SATURN trial, in the IFCT-GFPC 0502 trial, subgroup univariate analysis demonstrated that patients with SCC did not have a significant benefit in PFS when treated with erlotinib maintenance compared to placebo (HR 0.79; 95% CI 0.53–1.19) [Perol et al. 2012]. Neither patients with adenocarcinoma nor SCC had improved OS in this analysis.

Molecular characteristics

Mutation status has been shown to impact effectiveness of epidermal growth factor receptor (EGFR)-targeted therapies, but erlotinib was initially studied and gained approval for second-line and maintenance treatment in unselected patients. The INTEREST and TITAN studies demonstrated no difference in outcomes between gefitinib or erlotinib and chemotherapy in the second-line setting [Kim et al. 2008; Ciuleanu et al. 2012]; however, these studies did not require EGFR mutation testing, which could have skewed results by having EGFR-mutated patients in the EGFR directed therapy arms. Similarly, the SATURN trial demonstrated that PFS and OS were prolonged for all patients receiving maintenance erlotinib compared with placebo, regardless of the presence or absence of EGFR-activating mutations. However, it is worth noting that nearly 50% of the patients enrolled in SATURN had missing or indeterminate EGFR mutation testing, making the results difficult to interpret [Cappuzzo et al. 2010]. It is now recommended that all patients with nonsquamous NSCLC have their tumor tested for EGFR-activating mutations at the time of diagnosis, so treating patients with unknown EGFR mutation status is uncommon.

More recent studies have highlighted the lack of benefit of erlotinib for patients without EGFR mutations. The TAILOR trial, which was presented in abstract form at ASCO 2012, compared erlotinib and docetaxel as options for second-line therapy in EGFR-wild-type patients [Garassino et al. 2012]. In this study, those receiving erlotinib had a significantly lower response rate (RR) and PFS than those receiving docetaxel (RR, 2.2% versus 13.9%, p = 0.004; 6-month survival, 16% versus 28%, HR, 0.70, p = 0.016). Overall survival data have not been reported. This trial has been criticized because it showed unprecedented response rates to second-line docetaxel with 4% of patients achieving a complete response (CR), which suggests the results should not be generalized to other patient populations. A phase III noninferiority trial, TORCH, evaluated the use of erlotinib in the first-line setting before cisplatin–gemcitabine in an unselected population and found that first-line erlotinib resulted in inferior median OS (11.6 versus 8.7 months; HR 1.24) to chemotherapy followed by erlotinib [Gridelli et al. 2012a].

Response to first-line therapy

Response to first-line therapy may be an indicator for how much benefit a patient will derive from maintenance therapy. In the SATURN trial, patients with SD after first-line chemotherapy had a significant improvement in OS with maintenance erlotinib, whereas those with partial response (PR) or CR did not [Cappuzzo et al. 2010]. Interestingly, data from a meta-analysis combining patients from SATURN and IFCT-GFPC 0502 show that both subsets of patients had improved PFS with erlotinib [Petrelli et al. 2011]. It is difficult to understand how patients with SD could have an improved OS without improved PFS without a confounding factor, such as a difference in disease characteristics or second-line treatments. Exploratory subgroup analysis from the IFCT-GFPC 0502 trials suggested that there was no difference in PFS or OS based on disease response to first-line chemotherapy for those receiving erlotinib, but there may have been an OS advantage for patients receiving gemcitabine who had an objective response to first-line chemotherapy [Perol et al. 2012]. In PARAMOUNT, a similar subgroup analysis suggests there may be a larger benefit for patients with responsive disease after first-line chemotherapy [Paz-Ares et al. 2013]. Unfortunately, these subgroup analyses from SATURN, IFCT-GFPC 0502, and PARAMOUNT are simply exploratory, and the 95% CIs of the subgroups are overlapping in all cases. In AVAPERL, there were no differences based on response to first-line chemotherapy [Barlesi et al. 2013]. Complete data from PointBreak have not yet been reported, but the importance of response to first line chemotherapy is unlikely to be determined unless a maintenance trial is designed to answer this question. In keeping with these conflicting results, the United States Food and Drug Administration (FDA) has approved maintenance erlotinib for all patients who have not progressed after induction chemotherapy. In contrast, the European Medicines Agency (EMA) has only approved maintenance erlotinib for patients with SD.

Performance status

PS is a strong prognostic factor and is predictive of survival benefit and toxicity from chemotherapy [Schiller et al. 2002]. Several studies have evaluated the effect of PS on outcomes for maintenance therapy. In the CECOG trial, the improvement in OS with maintenance gemcitabine did not reach statistical significance, but subgroup analysis based on PS revealed that patients with a Karnofsky Performance Status (KPS) >80 had a significantly improved OS on the gemcitabine maintenance arm [Brodowicz et al. 2006]. In the trial from Fidias and colleagues [Fidias et al. 2009], patients with PS 2 were included, but made up only about 10% of the patients enrolled. One of the major reasons for failure to receive docetaxel in the delayed dose arm was clinical deterioration. Although there was no subgroup analysis reported, a higher proportion of patients with PS 2 (10.3% versus 5.9%) were in the delayed docetaxel arm, which may have contributed to differences between the two groups. Patients with PS 0 in the IFCT-GFPC 0502 trial had the greatest benefit from gemcitabine or erlotinib maintenance [Perol et al. 2012]. Those with PS 1–3 did not have a statistically significant improvement in OS. Most other maintenance studies have excluded patients with PS 2. Based on the subgroup analyses of these studies, patients with worse PS appear to have less significant survival benefits, and those with PS 2 should not receive maintenance therapy.

Quality of life

Maintenance chemotherapy has been shown to improve PFS or OS in many clinical trials, but most studies have demonstrated improvements of only 1–3 months. Given only modest improvements in PFS or OS, QOL during this time becomes critical for evaluating the utility of maintenance therapy. On the one hand, giving continual therapy provides an opportunity for additional treatment exposure and prevention of a more rapid deterioration in health that could preclude some patients from receiving second-line therapy. On the other hand, tying patients to continual therapy, whether intravenous infusions every 3 weeks or daily oral medications, prevents patients from enjoying a chemotherapy holiday they might otherwise have had if they chose to forgo maintenance therapy.

Several maintenance trials have incorporated QOL analysis into their design, often using the Lung Cancer Symptoms Scale (LCSS), a validated instrument used to assess symptom burden and impact on QOL in patients with lung cancer [Hollen et al. 1994]. The Fidias and colleagues study compared LCSS scores and concluded that there was no difference in QOL between those receiving immediate docetaxel versus delayed docetaxel treatment [Fidias et al. 2009]. Patients treated on the JMEN study also completed a symptom assessment with each cycle of chemotherapy from the time of randomization. On this study, however, instead of comparing LCSS scores directly, the time to worsening symptoms, defined as a 15 mm decrease on the visual scale, was compared between the two arms [Belani et al. 2012]. For most symptoms, there were no differences in time to worsening of symptoms or overall QOL between the pemetrexed and placebo groups. However, those receiving maintenance pemetrexed had significantly longer time before worsening of pain and hemoptysis symptoms. Although there was no difference in time to developing loss of appetite, patients receiving pemetrexed had a larger degree of loss of appetite while receiving chemotherapy.

The PARAMOUNT study reported QOL using the EQ-5D instrument, a two-part questionnaire that gauges a patient’s health status with descriptive questions and a visual analog scale [Gridelli et al. 2012b]. The investigators found no statistical difference in EQ-5D scores between the maintenance pemetrexed and BSC groups.

Patients on the SATURN study were given Functional Assessment of Cancer Therapy-Lung (FACT-L) questionnaires, an instrument validated to assess physical function and well-being, every 6–12 weeks or until progression while on study [Juhasz et al. 2013]. There were no differences in time to symptom progression, deterioration, or time to deterioration in trial outcome index between the erlotinib and placebo maintenance groups. There was a significant delay in the time to pain and analgesic use in those receiving erlotinib maintenance.

Based on these multiple QOL analyses accompanying their respective trials, it appears that QOL is not significantly worse with maintenance therapy and may delay the time to pain or other disease-related symptoms. Unlike earlier studies that show an improvement in QOL with chemotherapy for metastatic NSCLC [Non-Small Cell Lung Cancer Collaborative Group, 1995; Cullen et al. 1999; Spiro et al. 2004], there is no evidence that additional maintenance therapy improves QOL. It is likely, however, that our QOL tools are crude at best, and subtle toxicities are not reflected as major deterioration.

Cost and healthcare utilization

Prior to the use of maintenance therapy in lung cancer, patients with SD after four to six cycles of chemotherapy were routinely observed until progression. The additional therapy during this period has been shown to have modest improvements in PFS and OS, but significantly increased financial cost. In the United States, US$2.8 trillion per year is spent on healthcare, and growth of healthcare spending is expected to outpace the gross domestic product (GDP) and increase to as much as 40% of federal healthcare spending in the next 25 years [Emanuel et al. 2012]. There has been significant political focus on ways to contain healthcare spending, including increases in the transparency of cost and quality of care.

Several cost-effective analyses of maintenance pemetrexed have been conducted using the JMEN trial as a model in the United States, United Kingdom, Switzerland, and Japan. The US study used maximum Medicare reimbursement rates for pemetrexed in 2009 and other direct NSCLC costs from a claims database to calculate the cost of maintenance pemetrexed without dose reductions or delays [Klein et al. 2010]. The investigators concluded that for patients with nonsquamous NSCLC, the incremental cost-effectiveness ratio (ICER) per life-year gained of maintenance pemetrexed was US$122,371, just below the accepted US standard of renal hemodialysis, which has an ICER of US$129,090 [Lee et al. 2009]. A similar cost-effective analysis was conducted using the JMEN results and the Swiss healthcare system as a model and determined the ICER was €106,202 (approximately US$139,000), above the nationally accepted willingness to pay (WTP) threshold in Switzerland of €72,000 [Matter-Walstra et al. 2012]. Interestingly, maintenance pemetrexed was approved by the United Kingdom’s National Institute for Health and Clinical Excellence (NICE) based on an estimated ICER of £47,000 (approximately US$72,000) [Learn and Bach, 2010]. Finally, a Japanese analysis concluded the ICER for maintenance pemetrexed was US$150,115. This ratio was above the Japanese threshold of US$43,478. The differences between the ICER in the UK and in other countries may be accounted for by the difference in the cost of pemetrexed in the UK. The Swiss cost-effectiveness study found the main driver of the high ratio was cost of the drug; using the cost of pemetrexed as available in the UK, they found the ICER was reduced to €68,886, which is below the Swiss WTP threshold of €72,000 [Matter-Walstra et al. 2012].

One cost-effective analysis has been conducted for continuation maintenance pemetrexed based on the PARAMOUNT results in a Chinese healthcare system model [Zeng et al. 2013]. This study found the ICER to be US$124,793, similar to the US and Swiss studies for switch maintenance pemetrexed. The study concluded that the ICER was significantly higher than the World Health Organization’s (WHO) recommendation for a threshold of three times the country’s per-capita GDP, which ranges from approximately US$7500 to US$40,000, depending on geography. Therefore, maintenance pemetrexed was determined not to be cost effective for use in China. In conclusion, the variance in ICER for maintenance pemetrexed between different countries is largely attributed to differences in drug costs. The threshold to determine cost-effectiveness may also vary from country to country depending on the national standards and relative wealth of its citizens.