New Literature Since 2007 Relevant to this Sub-question

Our update search identified seven new RCTs relevant to this sub-question.^33–39 Details of these studies, plus those from the CCO review, are in Table F1. Among these seven new studies, none reported significant benefit in outcomes when comparing the newer, to older, doublets. One study actually found worse outcomes with the newer agent, pemetrexed, versus the older agent, etoposide.³⁶ Overall survival was 8.1 months with pemetrexed, versus 10.6 months with etoposide (HR=1.56, 95% CI: 1.27–1.92; log-rank P<.01) and the objective response rate was 31% vs. 52% (P<.001).

Table F1

New Doublets vs. Older Doublets (adapted from Goffin).

Summary of Key Sub-question 1.1

New trials continue to support the conclusion by the CCO that any differences in survival between platinum-based doublets are modest (GRADE=High).

New Literature Since 2007 Relevant to this Sub-question

We identified one new published trial relevant to this Key Sub-question. Reynolds and colleagues randomized 170 patients with performance status of two to either gemcitabine or gemcitabine plus carboplatin.⁴¹ The median survival was 6.7 (4.9–10.0) months in the gemcitabine plus carboplatin group and 5.1 (3.9–6.3) months for gemcitabine. This difference was not statistically significant (p=0.14). Overall survival at one year was 31.3 percent in the doublet therapy group and 21.2 percent in the gemcitabine alone therapy group (statistical testing not performed). There was a statistically significant higher rate of confirmed response in the gemcitabine plus carboplatin (21.1%) group compared to gemcitabine alone (6.3%, p=.01). In addition, patients’ tumors were evaluated for RRM1 and ERCC1 protein levels and found that these were significantly and inversely correlated with disease response.

During peer review, we were directed to a recently presented abstract of a second relevant study⁴². Patients with advanced NSCLC and a performance status of 2 were randomized to receive pemetrexed or carboplatin plus pemetrexed. During the trial, eligibility was later restriced to only patients with nonsquamous histology. Overall survival among 205 patients studied was 9.1 months versus 5.6 months, favoring doublet therapy (p=0.001). We await full publication of these results, which provide additional support favoring doublet therapy, even in patients with a poor performance status.

Summary of Key Sub-question 1.2

This result continues to support the conclusions by the CCO that doublet chemotherapy including a platinum agent has a higher survival rate and a higher response rate than a single agent (GRADE=High).

New Literature Since 2007 Relevant to this Sub-question

Our search identified nine additional studies^53–61 and two subsequent meta-analyses^{62, 63} relevant to this Key Sub-question. These new results did not alter the conclusions reached by the CCO. The combinations studied included gemcitabine, paclitaxel, paclitaxel poliglumex (CT-2013 PPX), S-1, pemetrexed, vinorelbine, and docetaxel. Details of these studies, as well as the studies contained in the CCO review, are in Table F2. No particular combination was shown to be significantly superior in outcomes. Median survival ranged from 7.0 to 15.2 months. Of note, the study comparing paclitaxel plus carboplatin vs. gemcitabine plus oxaliptin had to be terminated after adverse events exceeded the safety threshold set by the data safety and monitoring board.⁶⁰

Table F2

Comparison of Doublets of New Agents in Combination with Platinum Analogues (adapted from Goffin).

We also identified one clinical trial comparing doublet chemotherapy regimens, but neither included a platinum agent. This study compared paclitaxel and gemcitabine to paclitaxel and vinorelbine in 39+ patients with stage IIIB or IV NSCLC. There was no difference in median overall survival (11.1 months vs. 8.6 months, p=0.14); the group treated with vinorelbine had more grade 3/4 toxicities.²⁵

Summary of Key Sub-question 1.3

New trials continue to support the conclusion by the CCO that any differences in outcomes between doublet therapies with platinum-based agents are modest (GRADE=High).

New Literature Since 2007 Relevant to this Sub-question

The additional three trials we identified did not find a significant difference between agents and did not alter the conclusions reached by the CCO.^{59, 71, 72} Details of these three and the four RCTs identified by the CCO are presented in Table F3. In one of these trials, platinum-based regimens of gemcitabine or paclitaxel with carboplatin led to a median survival of 7.9 (7.1–9.2, P=0.693) and 8.7 (7.7–9.9) months, compared to gemcitabine with paclitaxel which led to 8.5 (7.6–10.0) months.⁵⁹ In another trial, the combination of gemcitabine with paclitaxel led to a median survival of 9.97 (8.74–12.0) months compared to 10.49 months with gemcitabine and carboplatin (9.04–11.94).⁷² Though the final study appeared to favor gemcitabine and epirubicin with a median survival of 21.5 months versus 13.2 months for gemcitabine-cisplatin, the confidence intervals were wide (9.4–33.6 and 10.4–16.0) and the number of patients small (80 and 85).⁷¹

Table F3

New Platinum-Based Doublets vs. New Nonplatinum-Based Doublets (adapted from Goffin).

Summary of Key Sub-question 1.4

New trials continue to support the conclusion by the CCO that doublet chemotherapy including a platinum agent probably has a slight advantage over nonplatinum doublets (GRADE=moderate).

Literature Relevant to this Sub-question Not Included in the CCO Review

We found one additional trial.⁶⁴ It compared vinorelbine plus cisplatin and paclitaxel plus carboplatin finding no significant difference between the two regimens in median survival 8.1 (6.7–9.6) months for the cisplatin containing regimen compared to 8.6 (7.2–10.7) for the carboplatin containing regimen. The overall response rates were 28 and 25%, respectively. However, the difference in second agents in the doublet precludes reaching a conclusion about differences between carboplatin and cisplatin.

Summary of Key Sub-question 1.5

One new trial does not alter the conclusion by the CCO that cisplatin combinations may have a slight advantage over carboplatin combinations in terms of survival and response rate. However, carboplatin generally has a milder toxicity than cisplatin (GRADE=moderate).

New Literature Since 2007 Relevant to this Sub-question

We identified a systematic review not in the review by the CCO that was relevant to this question.²⁴ Azim and colleagues identified six trials including 1,932 patients who received either third generation triplet therapy or standard doublet therapy. Pooled analyses showed that triplet therapy resulted in a statistically significant increase in response rate (pooled OR=1.33, 95% CI: 1.50–2.23), however overall survival was improved by less than six weeks (42.8 vs. 37.4 weeks) and this was not statistically significant. Patients receiving triplet therapy had statistically significantly more grade III/IV toxicities than patients receiving doublet therapy. The authors concluded that triplet therapy “is associated with higher tumor response rate at the expense of increased toxicity.”²⁴

We found four new trials,^{83, 84} one of which was included in the review by Azim and colleagues.⁸⁴ The first trial randomized 433 patients into 4 arms of treatment- gemcitabine-cisplatin, gemcitabine-vinorelbine, gemcitabine-ifosfamide-cisplatin or gemcitabine-ifosfamide-vinorelbine. ⁸³ They found no benefit with the addition of ifosfamide to a gemcitabine based doublet. Overall survival was 10.4 (9.4–12.2) months for the doublet and 10.3 (9.2–11.8) months for the triplet and the response rate was 29% (23–35) for the doublet and 26% (21–33) for the triplet. The second trial had 433 patients randomly assigned to gemcitabine-vinorelbine, gemcitabine-paclitaxel, gemcitabine-vinorelbine-cisplatin, gemcitabine-paclitaxel-cisplatin.⁸⁴ They found an increase in the response rate with triplets over doublets, 48% (42–54) vs. 35% (32–38). However, there was no benefit in overall survival with triplets having a median OS of 10.7 months and doublets having an overall survival of 10.5 months (P=0.379).

We also found one trial that compared a platinum-based triplet (paclitaxel, carboplatin, gemcitabine) versus a nonplatinum doublet (gemcitabine + vinorelbine). There were no statistically significant differences in overall survival between groups (10.7 vs. 10.3 months), and certain toxicities were greater in the triplet therapy group.⁸⁵ Another “triplet vs. doublet” study compared a nonplatinum triplet (vinorelbine, gemcitabine, docetaxel) to carboplatin plus paclitaxel and showed no statistically significantly difference in overall survival (13.6 vs. 14.1 months), with toxicity greater in the platinum doublet treated patients.⁸⁶

Summary of Key Sub-question 1.6

New trials continue to support the conclusion by the CCO that triplet cytotoxic therapy might have some slight advantages in terms of response rate but at an increased risk of toxicity (GRADE=high).

New Literature Since 2007 Relevant to this Sub-question

Our update search identified an additional 10 trials relevant to this sub-question.^87–96 Details of these trials, in addition to the eight contained in the review by the CCO, are presented in Table F4. Two of the new trials were of the agent PF-3512676. The first of these two trials, of 839 patients, did not find an improvement in OS or PFS and, in fact was halted based on efficacy, futility, and increased toxicity.⁹² The other study of this agent had similar findings in 828 patients and was therefore also halted.⁹⁰ There was one study of endostar which enrolled 126 patients and found improvement in the overall response rate of 39.3% in the treatment group versus 23.0% in the control group (p=0.078).⁸⁹ However, there was no significant difference in the overall survival (17.6 vs. 15.8 months, p=0.696). There were two reports of bevacizumab. One of these had 1043 patients and was included in the review by the CCO,⁹⁵ and the other was the Asian subset of that population, consisting of 105 patients.⁹³ In the overall sample, as noted above, an improvement in progression-free survival was found. However this did not translate into an improvement in overall survival, which was about 13 months in each group. ⁹⁵ In the study of the Asian subset, however, bevacizumab at a dose of 7.5 mg/kg improved the overall survival in Asian patients compared to placebo (HR=0.46, 95% CI: 0.22–0.97).⁹³ The other agents failed to show positive results. One trial of sorafenib found no clinical benefit and, in fact, was halted after an interim analysis found higher mortality in patients with squamous cell histology taking sorafenib than those in the placebo group.⁹⁶ Another study that was halted examined cediranib.⁸⁸ That trial was halted to review imbalances in assigned causes of death. A trial of cetuximab found a slight improvement in overall survival but this was not statistically significant, 9.69 months with cetuximab vs. 8.38 months with standard chemotherapy alone, (HR= 0.890, 95% CI: 0.754–1.051; p=.169).⁹¹ There were two trials of bexarotene with chemotherapy.^{87, 94} The first looked at 612 patients and found little difference in overall survival (8.5 vs. 9.2 months) or overall response (19.3 vs. 23.5%, p=.24).⁸⁷ However, when they examined a subpopulation of those with hypertriglyceridemia, they found that those with hypertriglyceridemia had significantly longer median survival than control patients (12.4 vs. 9.2 months; log-rank, P=.014). The other trial was of 623 patients and found the same, no significant difference (OS: 8.7 vs. 9.9 months, P=.3, ORR: 16.7 vs. 24.4%, P=.0224) except in those with hypertriglyceridemia (OS 12.3 months).⁹⁴ As both of these results were post-hoc subgroup analyses, their results should be interpreted with caution.

Table F4

Targeted Agents (adapted from Goffin).

Key Sub-question 1.7.1. Does targeted monotherapy improve outcomes in selected patient populations?

We also identified seven publications (from six trials) that assessed the use of targeted monotherapy compared to conventional chemotherapy, primarily in the population of patients with the EGFR gene mutation. We consider this to be related but distinct to the above discussion regarding the addition of targeted therapy to conventional chemotherapy, and therefore have created a new key sub-question for this topic. These studies, which compared gefitinib to carboplatin-paclitaxel in a general population and then performed a subgroup analysis on patients with the EGFR mutation,^{107, 108} gefitinib vs. chemotherapy in patients selected for EGFR gene mutations,^{109, 110} gefitinib vs. chemotherapy in a general population and then performed a subgroup analysis on patients with EGFR mutations,¹¹¹ and erlotinib vs. chemotherapy in patients selected for EGFR gene mutations,^{112, 113} consistently show large differences in progression-free survival favoring targeted monotherapy, such as10.1 months vs. 5.4 months(Table F5). Overall survival also tended to favor groups treated with targeted monotherapy as opposed to chemotherapy treatment, but this has not reached statistical significance, and two trials favored chemotherapy over targeted monotherapy, however neither of these contradictory findings reached statistically significance either.^{111, 113} One of these trials was consistent with the support for targeted monotherapy compared to chemotherapy within a subgroup analysis of patients with EGFR mutations.¹¹¹ Other than rash and aminotransferase elevation, erlotinib and gefitinib were also in general better tolerated than cytotoxic chemotherapy.

Table F5

Targeted Monotherapy.

One additional trial is relevant to this topic. Gridelli and colleagues compared a strategy of using erlotinib as first-line therapy in unselected patients with advanced NSCLC, followed by cisplatin and gemcitabine at the first sign of progression, versus the opposite strategy of initial therapy of cisplatin and gemcitabine followed by erlotinib given at the first sign of progression.¹¹⁴ Among nearly 800 patients, the erlotinib first strategy was significantly worse in terms of overall survival.

As part of an update search, we identified two recently published meta-analyses of tyrosine kinase inhibitors.^{13, 22} One meta-analysis was of monotherapy with either erlotinib or gefitinib compared to cytotoxic therapy in patients with advanced NSCLC and EGFR mutations.¹³ It included six trials, all of which are included in Table F5. The pooled analysis showed a statistically significant benefit for progression-free survival (HR=0.37; 95% CI: 0.27–0.52), and a nonstatistically significant benefit in overall survival (HR=0.94; 95% CI: 0.77–1.15). The second meta-analysis was restricted to gefitinib, and included studies of it as monotherapy, combined with systemic chemotherapy, and then given after systemic chemotherapy (only one trial of this).²² The monotherapy portion included four trials, three of which are included in Table F5, and also found improvement in progression-free survival for patients with EGFR mutations.

Summary of Key Sub-question 1.7

New trials of a number of novel targeted agents have so far failed to find results equivalent to the increases in progression-free survival seen with erlotinib (mostly in patients who have never smoked) and bevacizumab (in an Asian population subgroup analysis) in the CCO review (GRADE=moderate).

Erlotinib or gefitinib monotherapy is in general superior in terms of beneficial outcomes and adverse events than cytotoxic chemotherapy in patients with EGFR mutations (GRADE=high).

New Literature Since 2007 Relevant to this Sub-question

We identified four new trials that compared single agents to doublet regimens in the elderly population.^{108, 113, 115, 116} The first two trials were focused on elderly patients.^{108, 113} The second two trials examined the elderly as a subpopulation.^{115, 116}

The first trial focused specifically on the elderly population, comparing treatment with docetaxel to treatment with both docetaxel and gemcitabine.¹¹⁶ Three hundred and fifty patients were randomized, with the median age of 74 years. ITT analysis showed median survival of 5.5 months in the group receiving the doublet chemotherapy, compared to 5.1 months in those receiving a single agent (P=.65). The median time-to-progression was longer in those receiving the doublet regimen (4.8 months) compared to those receiving a single agent (2.9 months; P=.004).

The second trial compared individuals aged 70–89 receiving either the doublet regimen of carboplatin and paclitaxel or therapy with a single agent of vinorelbine or gemcitabine.¹¹⁵ The 451 patients were randomly assigned to the two groups. Median overall survival was 10.3 months for the doublet chemotherapy compared to 6.2 months for therapy with a single agent (HR 0.64, 95% CI 0.52–0.78; p<0.0001). One year survival was also greater in the doublet group with 44.5% (95% CI 37.9–50.9) alive at one year compared to 25.5% in the single agent group (95% CI 19.9–31.3). There were greater toxic effects in the doublet chemotherapy group but this was still considered a superior regimen to single agent therapy.

The first subgroup trial assessed elderly patients within a group of adults treated for NSCLC compared erlotinib to standard chemotherapy of cisplatin or carboplatin plus docetaxel or gemcitabine.¹¹³ The study was halted after interim analysis revealed that the study had met its primary endpoint. In the final analysis, the median progression-free survival for those treated with erlotinib was 9.7 months (95% CI 8.4–12.3) compared to 5.2 months for those treated with the doublet regimen (95% CI 4.5–5.8) (HR 0.37, 95% CI 0.25–0.54; p<0.0001). A multivariable analysis of progression-free survival found that age was not a significant factor.

Another subgroup study assessed differences in progression-free survival in patients treated with gefitinib compared to carboplatin plus paclitaxel.¹⁰⁸ They found that progression-free survival was longer in the gefitinib group but that this was affected by age. For those patients younger than 65, the hazard ratio was 0.81; 95% CI, 0.70 – 0.95, and for those patients 65 years or older, the hazard ratio was 0.58; 95% CI, 0.45–0.76; P<0.001. There was a statistically significant interaction between age and treatment effect (p=0.03).

We add here one additional trial, which compared singlet vs. doublet therapy, although the mean age of patients was “elderly” (mean age = 63), and not all clinical experts would judge this study as relevant to the elderly population.¹¹⁷ This study compared docetaxel alone to docetaxel plus gemcitabine in 312 patients with stage IIIB or IV NSCLC. There was a slight, statistically significant advantage for overall survival in the doublet-treated group (9.4 months vs. 8.3 months). There was a greater proportion of patients in the doublet-treated group with grade 2–4 anemia, plus two treatment-related deaths.

Summary of Key Sub-question 1.12

With the exception of studies of gefitinib and erlotinib monotherapy (in patients with EGFR mutations), doublet chemotherapy probably has a slight benefit in terms of survival compared to singlet therapy, but causes more toxicity (GRADE=moderate). Also, there now has been one trial of platinum therapy in the elderly taken to completion that found a near-doubling of the proportion of patients alive at one year in the doublet therapy group compared to monotherapy.

Trials Included in Existing Systematic Reviews

Two of the trials, by Gebbia and colleagues¹³³ and Takeda and colleagues,¹³⁴ were included in the review by Di Maio and colleagues discussed previously. ¹³² One trial by Herbst and colleagues¹³⁶ was in the review by Yang and colleagues, ¹²⁷ also discussed previously. The remaining three trials were only included in reviews by Qi and colleagues, which were not discussed in detail in the prior section. We therefore summarize these three trials here.

The first trial¹³⁵ was included in Qi and colleagues (2012).¹²⁶ In this RCT, patients who had been pretreated with front-line platinum-free regimens were randomized to receive either docetaxel-carboplatin or docetaxel alone. There was no significant difference in overall survival between the two groups with 10.27 (95% CI: 7.07–13.47) months achieved in the combination arm and 7.70 (95% CI: 3.39–12.01) months in the docetaxel alone arm (p=0.550). The response rate was similar 10.4% for the combination versus 7.7% (p=0.764). Likewise, the one-year survival rate was similar 43.8% versus 40.3%.

The next trial¹³⁷ was included in the systematic reviews by Qi and colleagues (2011)¹²⁵ and Qi and colleagues (2012).¹²⁴ In this study, 534 patients were randomized to receive vandetanib added to pemetrexed or pemetrexed plus placebo. There was no significant difference in overall survival. Median overall survival was 10.5 months for vandetanib and 9.2 months for placebo (HR=0.86; 97.54% CI, 0.65–1.13; p=.219). There were statistically significant improvements in overall response rate with 19% in the vandetanib group and 8% in the placebo group (p<.001). Vandetanib is not currently approved for use in the United States.

The last of the trials¹³⁸ was found in Qi and colleagues (2011).¹²⁵ In this trial 1,240 patients were randomized to receive vandetanib or erlotinib. There was no significant difference in the superiority analysis as the median overall survival was 6.9 months for vandetanib and 7.8 months with erlotinib (HR=1.01; p=.830). Likewise, the overall response rate was 12% in both arms (two-sided P=.98). A preplanned non-inferiority analysis supported non-inferiority, defined as whether vandetanib “retained at least 50%” of the efficacy of erlotinib.

The remaining 15 trials were not found in any of the systematic reviews.^{123, 139–153} Table S3 has the details of these trials. We discuss these here in the order in which they emerged into clinical practice: docetaxel, followed by pemetrexed, then tyrosine kinase inhibitors (TKI), and finally newer agents.

Table S3

Trials Not Included in Existing Systematic Reviews.

Docetaxel

We identified one new trial of docetaxel. Fossella and coleagues 373 patients were randomized to receive either docetaxel 100mg/m2, docetaxel 75mg/m2 versus a regimen of either vinorelbine or ifosfamide.¹⁵¹ Overall survival was not significantly different between the three groups, 5.5 months for docetaxel 100mg/m2, 5.7 months for docetaxel 75mg/m2 and 5.6 months for vinorelbine/ifosfamide. However, the 1-year survival was significantly greater with docetaxel 75mg 32% (95% CI: 23–40) than with vinorelbine/ifosfamide 19% (12–26). The overall response rates for both of the docetaxel arms were significantly higher than that of the vinorelbine/ifosfamide arm with 10.8% for docetaxel 100mg/m2, 6.7% for docetaxel 75mg/m2 and 0.8% with vinorelbine/ifosfamide (p=.001 and p=.036, respectively).

Pemetrexed

Cullen et al, randomized 588 patients to either standard or high-dose pemetrexed.¹⁴⁵ The goal was to accrue 600 patients, however accrual was terminated after an interim analysis indicated a low probability of improved survival and numerically greater toxicity in the higher dose arm. Of the patients randomized, there was no statistically significant difference between the two groups in median survival reaching 6.7 months with the lower dose and 6.9 months with the higher dose (HR=1.0132, 95% CI: 0.837–1.226). Likewise there was no significant difference in overall response with 7.1% in the lower dose and 4.3% in the higher dose (P=0.1616).

TKI versus Cytotoxic Therapy

There were six trials comparing TKI to cytotoxic therapy.^{139–144, 152} Ciuleanu et al, included 424 patients who were randomized to receive either erlotonib or chemotherapy.¹³⁹ About two-thirds of patients were positive for EGFR mutations. Median overall survival was 5.3 months (4.0–6.0) in the erlotinib group and 5.5 months (4.4–7.1) with chemotherapy (HR=0.96, 95% CI: 0.78–1.19; log-rank p=0.73). The response rate was also similar between the 2 groups with a RR of 7.9% (4.6–12.5) in the erlotinib group and 6.3% (3.5–10.4) in the chemotherapy group. The one-year survival was 26% (19–32) in the erlotinib group versus 24% (18–30) in the chemotherapy group.

Maruyama and colleagues randomized 489 patients to receive either gefitinib or docetaxel.¹⁴⁰ Patients were enrolled irrespective of EGFR receptor mutation status. There was no significant difference in overall survival, which was 11.5 months (95% CI: 9.8–14.0) in the gefitinib arm and 14.0 months (95% CI: 11.7–16.5)in the docetaxel arm (P=.330). However, for ORR gefitinib was statistically superior to docetaxel (22.5% v 12.8%, odds ratio, 2.14; 95% CI: 1.21–3.78; P=.009).

Sekine and colleagues assessed quality of life outcomes in the preceding study reported by Maruyama and colleagues.¹⁴² This studied evaluated the quality of life differences between these two regimens and found gefitinib to have statistically significant benefits over docetaxel. The Functional Assessment of Cancer Therapy-Lung analysis was 23% for gefitinib versus 14% for docetaxel (P=0.023). The Trial Outcome Index was 21% vs 9% (P=0.002). There was no significant differences between treatments in Lung Cancer Subscale improvement rates (23% vs 20%, P=0.562). This study is the quality of life assessment outcome of a trial by Maruyama and colleagues, discussed above.¹⁴⁰

Kim and colleagues randomized 1466 patients to either gefitinib or docetaxel.¹⁴¹ Patients were enrolled irrespective of EGFR mutation status. Median overall survival was 7.6 months in the gefitinib group and 8.0 months in the docetaxel group. One-year survival was 32% for gefitinib and 34% for docetaxel.

Lee and colleagues randomized 161 patients to either gefitinib or docetaxel.¹⁴³ Patients were enrolled irrespective of EGFR mutation status. The overall survival was 14.1 months for gefitinib and 12.2 months for docetaxel (HR=0.870, 95% CI: 0.613–1.236, 2-sided p=0.4370). The objective response rate was statistically superior in the gefitinib group (28.1%) compared to the docetaxel group (7.6%) (P=0.0007).

Lastly, during peer review an abstract presented at the 2012 ASCO Annual Meeting was identified.¹⁵² This study randomized 221 patients to erlotinib or docetaxel; all patients had wild type EGFR status. At a median follow-up of 20 months, progression-free survival favored treatment with docetaxel. Overall survival was not yet analyzed. We await full publication of these results, which will provide more complete analyses.

TKI As an Addition to Cytotoxic Therapy

This study of pemetrexed was added during peer review.¹⁵³ The study has been presented at the the 2011 ASCO Annual Meeting, but full publication is still pending. This was a phase II study in which 165 patients were randomized to receive either pemetrexed or pemetrexed plus erlotinib. The median progression-free survival was 2.9 months versus 3.2 months favoring combination therapy (HR=0.63, 95% CI 0.44–0.90). Overall survival also favored combination therapy (11.8 months vs. 7.8 months, HR=0.68, 95% CI 0.47–0.98). We await full publication of these results, which will provide more complete analyses, and confirmation in a phase III study.

TKI Plus Bevacizumab versus TKI Alone

Herbst and colleagues randomized 636 patients to either bevacizumab plus erlotinib or erlotinib alone.¹⁴⁴ Approximately 90 percent of patients had wild type EGFR receptor status. The overall survival did not differ between the two groups with the median overall survival of 9.3 months (IQR 4.1–21.6) in the bevacizumab arm compared to 9.2 months (IQR 3.8–20.2) in the erlotinib alone arm (HR=0.97, 95% CI: 0.80–1.18, p=0.7583). Objective response rate was much higher in the bevacizumab group (13% vs 6%).

Various Other Agents

There were seven trials of various other agents or doses.^{123, 146–151, 153} Ready and colleagues randomized 106 patients to either the proapoptotic agent AT-101 or placebo in combination with docetaxel.¹⁴⁶ There was no significant difference in the median overall survival which was 7.8 months in the AT-101 arm and 5.9 months in the placebo arm (HR=0.82, 95% CI: 0.5–1.3, p=0.21). According to an independent review, the response rate and median PFS were not different between the arms, with the PFS for the AT-101 group of 7.5 weeks and 7.1 weeks for placebo (HR=1.04, p=0.57).

Ramlau and colleagues randomized 829 patients to either oral topotecan or IV docetaxel.¹⁴⁷ The median survival was higher with docetaxel at 30.7 weeks (95% CI: 28–34) compared to 27.9 weeks (95% CI: 24–31) with topotecan (p=.0568). Overall response rate was 5% (95% CI: 3–7%) for both docetaxel and topotecan. One-year survival was 29% (24–33%) for docetaxel and 25% (21–29%) for topotecan.

Paz-Ares and colleagues randomized 849 patients to either paclitaxel poliglumex or docetaxel.¹⁴⁸ The median survival was not different, at 6.9 months in each arm (HR 1.09, p=0.257). The 1-year survival was also similar with 25% in the paclitaxel poliglumex arm and 29% in the docetaxel arm (P=0.006).

Hermes and colleagues randomized 220 patients to either irinotecan plus carboplatin or etoposide plus carboplatin.^{39, 154} Overall survival was better in the irinotecan group with a median survival of 8.5 months compared with 7.1 months in the etoposide group (etoposide relative to irinotecan HR=1.41; 95% CI: 1.06–1.87). The one-year survival rate was 34% for irinotecan and 24% for etoposide.

Krzakowski and colleagues randomized 551 patients to either vinflunine or docetaxel.¹⁴⁹ The median overall survival was 6.7 months (95% CI: 5.9–7.9) in the vinflunine group and 7.2 months (6.0–8.5) in the docetaxel group (HR=0.973, 95% CI: 0.805–1.176). The overall response rate was 4.4% for vinflunine and 5.5% for docetaxel treatment.

Finally, Schiller and colleagues randomized 150 patients to receive either pemetrexed alone, pemetrexed plus matuzumab 800mg/week, or pemetrexed plus matuzumab 1600mg/3 weeks.¹⁵⁰ There was a trend for improved OS in patients receiving matuzumab weekly compared to the every 3 weeks group with the former of 12.4 months and the latter 5.9 months. This was also greater than the 7.9 months of the pemetrexed group. Pooling the matuzumab arms gave an ORR of 11% compared to 5% for pemetrexed alone. Of note, all responses, except for one patient in the pemetrexed group, occurred in patients whose tumors expressed EGFR.

Summary of Second-line therapy Trials Not in Existing Systematic Reviews

The summary of these trials not included in existing systematic reviews is:

• Considering data from first line and maintenance therapy studies in addition to second line studies, there are sufficient data to support the conclusion that histology type influences the effectiveness of potential treatments. Pemetrexed is more effective in nonsquamous NSCLC, while docetaxel is more effective in squamous NSCLC (GRADE=moderate).
• Tyrosine kinase inhibitors, when used as second-line therapy in patients unselected for EGFR mutation status, produce overall survival similar to docetaxel (GRADE=strong).
• There is insufficient data to support effectiveness of other drugs, or drugs in combinations, in second-line therapy (GRADE=moderate).
• The above second line studies are typically undertaken after evidence of disease progression, and should be distinguished from mainenance therapy, which is undertaken when a patient has at least stable disease during treatment (typically four cycles).

Summary of Key Question 4

There are a large number of published cost-effectiveness analyses, but approximately two-thirds of such studies are supported by the makers of the drugs being assessed. Invariably, studies supported by the makers concluded that their drug was cost-effective. Of the cost-effectiveness analyses not supported by industry, the addition of bevacizumab to first-line therapy was found in one study to be not cost-effective, erlotinib was found in one study to be marginally cost-effective, and the differences between erlotinib and docetaxel maintenance therapy were slight in another study (GRADE=low).