Maintenance Therapy in Advanced Non-Small Cell Lung Cancer: Before We Accept This Treatment Paradigm, Let's Critically Evaluate and Interpret the Data

Non—small cell lung cancer (NSCLC) remains the leading cause of cancer-related mortality in the United States. Many patients present with symptomatic advanced metastatic disease and a significant proportion of patients with stage I-III relapse with metastatic spread. The use of platinum-based therapies, which have an acceptable toxicity profile, has been shown to prolong survival in patients who have a good performance status (PS). For many patients, platinum-based therapies are also effective at palliating disease-related symptoms. The benefit of platinum-based therapies occurs early, and several randomized trials have suggested that 3 to 4 cycles of therapy are just as effective as 6 or more cycles when considering overall survival (OS).¹

After years of assuming that all platinum-based doublets including a third-generation agent (ie, paclitaxel, docetaxel [Taxotere], vinorelbine [Navelbine], gemcitabine [Gemzar], or irinotecan) were equally efficacious, studies showed that treating non-squamous NSCLC with a combination of pemetrexed (Alimta) and cisplatin improved OS compared with a doublet of gemcitabine and cisplatin.² Other trials demonstrated that adding bevacizumab (Avastin)³ or cetuximab (Erbitux)⁴ to standard platinum-based chemotherapy in the first-line treatment of stage IV NSCLC prolonged survival. Despite these advances, stage IV NSCLC remains an incurable disease.

Second- and third-line treatments have been shown to improve survival and palliate symptoms,^5,6 with pivotal phase III trials leading the FDA to approve docetaxel,⁵ pemetrexed,⁷ and erlotinib (Tarceva)⁶ for this indication. Approval of these drugs ushered in the era of multiple-line therapy, changing the treatment paradigm for advanced NSCLC.

MAINTENANCE OR IMMEDIATE THERAPY TRIALS

Second-line therapy has traditionally been administered at the time of objective disease progression following first-line therapy. Four recent trials have questioned whether this strategy is prudent in patients with advanced NSCLC. These trials were similarly designed. Each administered 4 cycles of a platinum-based chemotherapy regimen as first-line treatment for advanced stage IV NSCLC. Those patients who were non-progressive after first-line therapy were randomized to immediate or delayed treatment with a single agent approved as second-line therapy (Figure 1). Three of the trials used the term maintenance^8-10 to describe treatment administered immediately after completion of 4 cycles of first-line therapy, while the fourth used the term immediate.¹¹ Delayed treatment was initiated only upon clear evidence of renewed progression. The single agents used after first-line therapy included pemetrexed, erlotinib, and docetaxel, all drugs that have available evidence suggesting they improve survival over best supportive care or placebo.

Three of the trials registered patients at the time of first-line therapy, and approximately 50% of study participants never made it to the randomization phase (Table 1). The reasons for this dropout have not yet been presented, but presumably they related largely to early disease progression, toxicity, declining PS due to disease or treatment, or patient preference. This important observation needs to be considered as one interprets the benefit of maintenance therapy.

Only one trial¹¹ dictated and exposed patients in the delayed cohort to the same agent patients in the maintenance or immediate arm received. The other 3 trials were placebo-controlled^8-10 and allowed physicians to treat the patients in the placebo arms at their discretion, leading to an imbalance in exposure to second-line agents. Patients in the placebo-controlled trials were seen every 3 weeks, underwent scanning every 6 weeks, and had to demonstrate clear objective evidence of disease progression before transitioning from placebo to active treatment. Consequently, 30% to 50% of patients in the placebo arms never received second-line therapy. This has led many to interpret the results from these trials as affirmation of the value of second-line therapy, rather than as evidence that such treatment should begin immediately following 4 cycles of first-line therapy.

There is no consensus about the terminology to use when discussing the phenomenon of maintenance therapy. Other interpretations of the strategy used in these trials are prolonged duration of first-line therapy, sequential therapy, or earlier administration of effective second-line therapy. Because these trials (Table 1) used only FDA-approved agents with acceptable toxicity profiles in this setting, the trial summaries herein focus on efficacy outcomes and do not include specific toxicity data.

Immediate or Delayed Docetaxel

The first trial reported was by Fidias and colleagues.¹¹ This trial registered 566 patients for first-line treatment consisting of 4 cycles of gemcitabine and carboplatin. Only 398 (69.3%) patients completed all 4 prescribed cycles. Of these, 309 (54.6% of the original 562 patients) non-progressing patients were randomized 1:1 to receive immediate docetaxel (n = 153) or delayed docetaxel (n = 156), administered at the time of documented disease progression. In the immediate docetaxel arm, 145 (94.7%) patients received at least 1 cycle of treatment, while only 98 (62.8%) patients in the delayed docetaxel arm were treated with at least 1 cycle.

There was a significant benefit with regard to median progression-free survival (PFS) in the immediate docetaxel arm compared with the delayed cohort of treated and untreated patients (5.7 mo vs 2.7 mo, respectively; P = .001). OS was the primary endpoint of this trial, and the immediate docetaxel group demonstrated a trend toward improved median OS compared with the delayed arm, but this was not statistically significant (12.3 mo vs 9.7 mo, respectively; P = .0853). The difference in 1-year survival between the groups was approximately 8%, with 51.1% (95% confidence interval [CI], 43.0%-59.3%) of patients in the docetaxel arm and 43.5% (95% CI, 35.5%-51.4%) of patients in the delayed docetaxel arm surviving 1 year. Of interest, median OS was identical at 12.5 months for the cohort of patients who received immediate docetaxel and the cohort of patients randomized to receive delayed docetaxel who actually received the treatment. This suggests that the key point is if docetaxel is administered, not when. The primary reasons for patients in the delayed arm not receiving docetaxel were relatively rapid disease progression and PS decline. No new safety signals with regard to the use of docetaxel were seen in this trial.

Pemetrexed or Placebo

The second trial was reported by Ciuleanu and colleagues.⁸ Patients in this trial received 4 cycles of a platinum-based doublet consisting of cisplatin or carboplatin plus gemcitabine or a taxane. Those with non-progressing disease were randomized 2:1 to pemetrexed (n = 441) or placebo (n = 222).

The primary endpoint was PFS. Median PFS in the pemetrexed cohort was 4.0 months compared with 2.0 months for the placebo arm (hazard ratio [HR], 0.60; P <.00001). Median OS was 13.4 months for patients taking pemetrexed versus 10.6 months for those on placebo (Figure 2; HR, 0.79; P = .012). The survival benefit was restricted to patients with non-squamous histology; no benefit over placebo was seen in patients with squamous histology. Considering only the patients with non-squamous histology, median OS in the pemetrexed arm was 15.5 months compared with 10.3 months in the placebo arm (HR, 0.70; P = .002). One concern about this trial is that only 67% of the placebo arm received subsequent therapy. Of these patients, only 19% received pemetrexed; another 69% received treatment with another agent approved for use in the second-line setting. There were no new safety signals for pemetrexed in this trial.

SATURN and ATLAS

Two trials have evaluated the role of maintenance erlotinib following 4 cycles of first-line platinum-based therapy.^9,10 The SATURN trial9 registered 1949 patients, and 889 (45.6%) of these patients were randomized to receive 4 cycles of erlotinib or placebo. The primary endpoints were overall PFS and PFS in patients with epidermal growth factor receptor (EGFR) positivity by immunohistochemistry (IHC). Erlotinib significantly improved PFS (HR, 0.71; P <.0001; absolute difference in median PFS, 1 wk) and OS (HR, 0.81; P = .0088) in all patients (Figure 3). Positivity for EGFR by IHC did not affect the impact of erlotinib (HR, 0.69 in EGFR IHC-positive patients vs HR, 0.77 in EGFR IHC-negative patients). Of note, patients with an EGFR mutation who received maintenance erlotinib had a huge PFS benefit (HR, 0.10; P <.0001), but they saw no benefit in OS (HR, 0.83; P = .68). This is likely because the majority of these patients received an EGFR tyrosine kinase inhibitor (TKI) upon disease progression. Also of note, the benefit of erlotinib in wild-type EGFR was significant for both PFS (HR, 0.78; P = .0185) and OS (HR, 0.77; P = .0243). As seen with the delayed treatment arms in Fidias, et al and Ciuleanu, et al, only 72% of patients in the placebo arm received second-line therapy, and only 21% of these patients received an EGFR TKI.

The ATLAS trial was unique in that it included a bevacizumab (Avastin)-eligible population of first-line patients, who received 4 cycles of bevacizumab combined with a choice of platinum-based doublets that included a taxane or gemcitabine.¹⁰ Following 4 cycles of therapy, patients were randomized to continue the bevacizumab with placebo or erlotinib. Of the 1160 patients initially registered, only 768 (66%) were randomized. The primary endpoint was median PFS, which was met. Erlotinib combined with bevacizumab significantly improved PFS compared with bevacizumab plus placebo (4.8 mo vs 3.7 mo, respectively; HR, 0.72; 95% CI, 0.59-0.88; P = .0012). Data were cut off in June 2009, and median OS for patients in the bevacizumab plus erlotinib arm was reported as 15.9 months compared with 13.9 months for patients in the bevacizumab plus placebo group (HR, 0.90; 95% CI, 0.74-1.09; P = .2686).¹² This 10% improvement over the duration of treatment was not statistically significant. Only 55% of the placebo arm received second-line therapy, and only 40% of these patients received a regimen containing an EGFR TKI. Neither SATURN nor ATLAS revealed new safety signals for erlotinib alone or in combination with bevacizumab.

DISCUSSION

In looking at these 4 trials together (Table 1), we see that a significant number of patients dropped out during the 4 cycles of first-line therapy. This occurrence was greatest in the SATURN trial, where 54% of registered patients did not get randomized to either cohort. Though no explanations have been given, this dilutes the applicability of the data to the potential effect of maintenance therapy on the advanced NSCLC population as a whole. As expected, maintenance therapy produced responses; however, the response rates were low, which is typical of second-line treatment with the agents used. The impact of maintenance chemotherapy with docetaxel or pemetrexed on PFS and OS seemed to be greater than that seen with erlotinib, though one must be very careful in making this comparison across trials. All 4 trials affirmed that these agents can be tolerated well following first-line chemotherapy, at least when they are limited to 4 cycles.

The important question now is how these trials should influence management of patients with advanced NSCLC. Is maintenance therapy now a new standard of care? The answer is no, in my opinion, but the results of these 4 trials should be considered in the overall management strategy for patients with advanced NSCLC.

The best-designed trial was that of Fidias and colleagues,¹¹ where patients were supposed to receive docetaxel either immediately following 4 cycles of carboplatin-gemcitabine or at the time of objective progression. The 3 placebo-controlled trials8-10 allowed physicians to treat at their discretion, which produced imbalanced exposure to the maintenance agent. In Fidias,¹¹ the survival of patients who received delayed docetaxel was identical to that of patients getting immediate docetaxel. This suggests treatment with an effective second-line agent is important, but the exact timing of dose administration may be less significant. Delaying treatment until objective disease progression creates major risks because many patients have disease- or comorbidity-related events that rapidly erode PS and preclude further therapy.

One way to increase exposure is using a maintenance strategy that ensures treatment-sensitive patients (those with a lack of progression during the 4 cycles of first-line therapy) are given effective therapy more often. Table 2 shows the rates of exposure to multiple lines of therapy in the 3 placebo-controlled trials.^8-10 Second-line therapy is synonymous with maintenance therapy in Table 2, which shows that patients randomized to the maintenance arm of these 3 trials were more often exposed to an agent in the “second-line” setting.

Based on the results of BR.216 and the Hanna trial,⁷ it is known or believed that both erlotinib and pemetrexed improve survival, and they have been approved by the FDA for second-line treatment. Considering the increased exposure to these agents in the SATURN9 trial and in the study by Ciuleanu and colleagues,⁸ it is not surprising, therefore, that these trials showed improved survival.

Most of the patients randomized to the maintenance arms of these trials also received third-line therapy, but these data were not presented for patients randomized to the placebo arms (Table 2). Given the drop-off in patients from the placebo arms who went on to receive second-line therapy, it is likely a similar drop-off occurred between second- and third-line therapies, resulting in lower exposure to effective therapies in these arms.

These trials do establish maintenance therapy as an option for patients, but I do not believe it is the correct strategy for every patient with advanced NSCLC. Allowing patients to have time off therapy or a treatment break is very therapeutic, particularly given the incurable nature and relatively poor survival outcomes we still see in this disease. The danger of a treatment break, however, is the risk of rapid disease progression and declining PS, which may preclude subsequent therapy. One must therefore exercise good clinical judgment with regard to how patients are managed. Many patients can enjoy the quality-of-life benefits a treatment break provides and still go on to receive second-line therapy with its resultant benefits.

On the other hand, patients who remain symptomatic or who have minimal responses to first-line therapy may benefit from a switch in therapy. These patients may also be at risk of rapid disease progression off therapy and may benefit from a maintenance strategy.

How patients should be followed during a treatment break has not been well studied, but I would be conservative and recommend not over-testing patients in this setting. I believe with a good clinical history, physical examination, and chest radiograph, monthly clinical visits are adequate. CT scanning should be done every 3 months or immediately for patients in whom there is a suspicion of disease progression based on prior analysis. One of the pitfalls of the design of the 3 placebo-controlled trials^8-10 was the need to wait for objective radiographic evidence of disease progression before administering second-line therapy. This might have decreased the ability to deliver second-line therapy. In practice, one should not wait if symptoms are suggestive, as the risk in waiting too long may be the rapid decline of a patient and the preclusion of further therapies.

If we accept maintenance therapy as standard and prescribe it as treatment, what do we do when patients progress? Three trials showed that maintenance therapy did not decrease the likelihood of additional therapy, with 50% to 55% of patients receiving “third-line therapy or therapy after maintenance” (Table 2). In practice, one would have the opportunity to use treatments not used in the maintenance setting; however, is this traditional second-line therapy or actually third-line therapy? We can debate definitions, but these arguments are semantic in nature and quite uninteresting. The point is that advanced NSCLC has become a disease in which multiple lines of therapy can affect the OS of patients. Adopting maintenance therapy as a standard would likely increase the exposure of more patients to active agents, particularly if we sequence to more therapy once patients progress on their maintenance agent.

CONCLUSIONS

Maintenance therapy with docetaxel, pemetrexed, or erlotinib is an option for patients with advanced NSCLC who exhibit non-progressive disease following 4 cycles of first-line platinum-based therapy. Based on the results of the ATLAS trial, this includes patients who receive bevacizumab-based therapies in the first-line setting.¹⁰ I do not believe maintenance therapy is the standard of care, but it is certainly a standard of care. Selection of the specific agent in this setting remains controversial, and ongoing studies are addressing this issue. Maintenance therapy should be discussed with patients, but a treatment break in certain patients is also an appropriate strategy based on individual preference.