J Clin Oncol 31: by American Society of Clinical Oncology INTRODUCTION

Transcription

1 VOLUME 31 NUMBER 31 NOVEMBER JOURNAL OF CLINICAL ONCOLOGY O R I G I N A L R E P O R T ATLAS: Randomized, Double-Blind, Placebo-Controlled, Phase IIIB Trial Comparing Bevacizumab Therapy With or Without Erlotinib, After Completion of Chemotherapy, With Bevacizumab for First-Line Treatment of Advanced Non Small-Cell Lung Cancer Bruce E. Johnson, Fairooz Kabbinavar, Louis Fehrenbacher, John Hainsworth, Saifuddin Kasubhai, Bruce Kressel, Chin-Yu Lin, Thomas Marsland, Taral Patel, Jonathan Polikoff, Mark Rubin, Leonard White, James Chih-Hsin Yang, Chris Bowden, and Vincent Miller See accompanying article on page 3987 Author affiliations appear at the end of this article. Published online ahead of print at on October 7, 213. Supported by Genentech, South San Francisco, CA. B.E.J. and F.K. are both primary authors. Presented in part at the 45th Annual Meeting of the American Society of Clinical Oncology, May 29-June 2, 29, Orlando, FL. Authors disclosures of potential conflicts of interest and author contributions are found at the end of this article. Clinical trial information: NCT Corresponding author: Bruce E. Johnson, MD, Dana-Farber Cancer Institute, 45 Brookline Ave, Dana 1234, Boston, MA 2215; BEJohnson@ Partners.org. 213 by American Society of Clinical Oncology X/13/3131w-3926w/$2. DOI: 1.12/JCO A B S T R A C T Purpose This phase III trial was performed to assess the potential benefit of adding maintenance erlotinib to bevacizumab after a first-line chemotherapy regimen with bevacizumab for advanced non small-cell lung cancer (NSCLC). Patients and Methods One thousand one hundred forty-five patients with histologically or cytologically confirmed NSCLC (stage IIIB with malignant pleural effusion, stage IV, or recurrent) received four cycles of chemotherapy plus bevacizumab. Seven hundred forty-three patients without disease progression or significant toxicity were then randomly assigned (1:1) to bevacizumab (15 mg/kg, day 1, 21-day cycle) plus either placebo or erlotinib (15 mg per day). The primary end point was progression-free survival (PFS). Results Median PFS from time of random assignment was 3.7 months with bevacizumab/placebo and 4.8 months with bevacizumab/erlotinib (hazard ratio [HR],.71; 95% CI,.58 to.86; P.1). Median overall survival (OS) times from random assignment were 13.3 and 14.4 months with bevacizumab/placebo and bevacizumab/erlotinib, respectively (HR,.92; 95% CI,.7 to 1.21; P.5341). During the postchemotherapy phase, there were more adverse events (AEs) overall, more grade 3 and 4 AEs (mainly rash and diarrhea), more serious AEs, and more AEs leading to erlotinib/placebo discontinuation in the bevacizumab/erlotinib arm versus the bevacizumab/ placebo arm. The incidence of AEs leading to bevacizumab discontinuation was similar in both treatment arms. Conclusion The addition of erlotinib to bevacizumab significantly improved PFS but not OS. Although generally well tolerated, the modest impact on survival and increased toxicity associated with the addition of erlotinib to bevacizumab maintenance mean that this two-drug maintenance regimen will not lead to a new postchemotherapy standard of care. J Clin Oncol 31: by American Society of Clinical Oncology INTRODUCTION Bevacizumab (Avastin; F. Hoffmann-La Roche, Basel, Switzerland), a vascular endothelial growth factor specific antibody, significantly prolonged progression-free survival (PFS) and overall survival (OS) when added to doublet chemotherapy in the Eastern Cooperative Oncology Group (ECOG) study E4599 and prolonged PFS in the Avastin in Lung (AVAiL) study. This approach is approved by the US Food and Drug Administration and European Medicines Agency for patients with advanced nonsquamous non small-cell lung cancer (NSCLC) without hemoptysis. 1-3 The BR.21 trial demonstrated an OS benefit for the epidermal growth factor receptor (EGFR) tyrosine kinase inhibitor erlotinib (Tarceva; F. Hoffmann-La Roche) compared with placebo in second- and third-line by American Society of Clinical Oncology

2 Erlotinib Plus Bevacizumab Maintenance Therapy in Advanced NSCLC Chemonaїve advanced/recurrent NSCLC (N = 1,145) Chemotherapy Phase Four cycles of first-line chemotherapy + bevacizumab (15 mg/kg) PD or significant toxicity No PD or significant toxicity (n = 743; 65%) Randomly allocated 1:1 Postchemotherapy Phase Bevacizumab (15 mg/kg) + Erlotinib (15 mg/day) to PD Bevacizumab (15 mg/kg) + Placebo to PD Fig 1. Study design. NSCLC, non smallcell lung cancer; PD, progressive disease. Study treatment discontinuation Unblind As a result of PD Reasons other than PD Termination visit within 3 days of study treatment discontinuation Optional postprogression therapy (at investigators discretion) for patients discontinuing as a result of PD (randomly allocated patients only) Survival follow-up every 3 months advanced NSCLC, 4 leading to approval by the US Food and Drug Administration and European Medicines Agency. However, the long-term prognosis for patients with lung cancer remains disappointing, 5 and maintenance therapy, whereby further active treatment is administered to delay progression after successful tumor control with first-line chemotherapy, has been the subject of recent research. Because bevacizumab and erlotinib target different molecular pathways involved in tumorigenesis (vascular endothelial growth factor and EGFR, respectively), a phase I/II trial examined their combined use in patients with previously treated NSCLC. No unexpected toxicities or pharmacokinetic interactions were observed between the two drugs at their recommended doses. 6 The combination demonstrated encouraging clinical activity in a randomized, phase II, secondline study, with a disease control rate of 85% and modest prolongation of PFS, although the prespecified primary end point of a clinically meaningful difference in PFS for bevacizumab/erlotinib versus chemotherapy (pemetrexed or docetaxel) was not achieved (hazard ratio [HR],.72; 95% CI,.42 to 1.23). 7 More recently, the phase III Bevacizumab Plus Tarceva (BeTa) study in advanced NSCLC failed to meet its primary end point of improved OS with the addition of bevacizumab to second-line erlotinib but showed promising PFS benefit with the combination versus erlotinib alone (median, 3.4 v 1.7 months, respectively; HR,.62; 95% CI,.52 to.75). 8 Overall, these studies suggested good tolerability and promising antitumor activity for the erlotinib and bevacizumab combination, making it a logical regimen for evaluation as maintenance therapy in advanced NSCLC. PATIENTS AND METHODS Study Design The Avastin Tarceva Lung Adenocarcinoma Study (ATLAS) was a phase III, multicenter, randomized, placebo-controlled trial conducted at 187 sites worldwide between January 1, 26, and June 19, 29 (ClinicalTrials.gov identifier: NCT25768). The primary objective was to compare the PFS of patients with advanced NSCLC randomly assigned to bevacizumab/erlotinib versus bevacizumab/placebo after completion of four cycles of chemotherapy plus bevacizumab without disease progression or significant toxicity. Secondary objectives included safety and OS. The study comprised two treatment phases, an open-label chemotherapy plus bevacizumab phase and a randomized, double-blind, placebo-controlled postchemotherapy phase (study design, Fig 1; CONSORT diagram, Fig 2). The control regimen was based on results from E Six platinum-based chemotherapy options were permitted, at the choice of the investigators. No bevacizumab dose reductions were permitted. If an adverse event (AE) occurred that necessitated interruption of bevacizumab, the dose was to remain unchanged once treatment resumed, or the drug was discontinued. Erlotinib dose reductions in decrements of 5 mg per day and standard chemotherapy dose modifications were permitted for toxicity. The study was conducted in accordance with Good Clinical Practice guidelines and was approved by the review board/local ethics committees at each participating center. All patients provided written informed consent. Eligibility Criteria Eligible patients were 18 years of age, with histologically or cytologically confirmed NSCLC (stage IIIB with malignant pleural effusion or stage IV) or recurrent disease and an ECOG performance status of or 1. Patients with a history of brain metastases were eligible provided their brain metastases had been treated and they did not require ongoing dexamethasone. Patients by American Society of Clinical Oncology 3927

3 Johnson et al Patients enrolled (N = 1,145) Chemotherapy phase Bevacizumab plus Carboplatin/paclitaxel Carboplatin/gemcitabine Carboplatin/docetaxel Cisplatin/gemcitabine chemotherapy (N = 1,145)* Random allocation of eligible patients Postchemotherapy phase (n = 743) (n = 524) (n = 326) (n = 162) (n = 14) (n = 29) Discontinued bevacizumab Disease progression Adverse event Treatment required that was not permitted in study Patient decision Investigator decision (n = 376) (n = 139) (n = 138) (n = 27) (n = 35) (n = 28) (n = 9) Bevacizumab/placebo (n = 373) Bevacizumab/erlotinib (n = 37) July 18 data cutoff Bevacizumab treated Not known to have discontinued bevacizumab Discontinued bevacizumab Disease progression Adverse event Not treated Received second-line therapy Erlotinib Bevacizumab (n = 369) (n = 19) (n = 26) (n = 24) (n = 34) (n = 22) (n = 4) (n = 213) (n = 152) (n = 152) July 18 data cutoff Bevacizumab treated Not known to have discontinued bevacizumab Discontinued bevacizumab Disease progression Adverse event Not treated Erlotinib treated Not known to have discontinued erlotinib Discontinued erlotinib Disease progression Adverse event Not treated (n = 363) (n = 126) (n = 237) (n = 168) (n = 38) (n = 31) (n = 7) (n = 364) (n = 124) (n = 24) (n = 158) (n = 48) (n = 34) (n = 6) Received second-line therapy Erlotinib Bevacizumab (n = 188) (n = 15) (n = 93) Intent-to-treat efficacy population Bevacizumab-treated efficacy population Safety-evaluable randomly allocated patients (n = 373) (n = 1,14) (n = 367) Intent-to-treat efficacy population Bevacizumab-treated efficacy population Safety-evaluable randomly allocated patients (n = 37) (n = 1,14) (n = 368) Fig 2. CONSORT diagram for 1,145 accrued patients. (*) One patient in the other chemotherapy group did not receive the protocol-specified treatment. ( ) Primary efficacy population; all patients randomly assigned during the postchemotherapy phase. ( ) All enrolled patients who received at least one dose of bevacizumab during the chemotherapy phase. with squamous cell carcinoma were eligible provided their disease was extrathoracic or their intrathoracic disease comprised peripheral lesions only. Known EGFR mutation status was not required for inclusion and was not a stratification factor. Random Assignment and Blinding Patients eligible for the postchemotherapy phase were randomly assigned 1:1 to receive bevacizumab plus either erlotinib or placebo and were stratified by sex, smoking history (never v former/current), ECOG performance status ( v 1), and initial chemotherapy regimen. Random assignment was conducted centrally using a hierarchical dynamic random assignment scheme using an interactive voice response system provided by a third party (ClinPhone, Nottingham, United Kingdom) to ensure approximately equal sample sizes between the two treatment arms overall, within each stratification factor, and at each study site. The study sponsor, investigators, and patients were blinded to treatment assignment. Procedures Baseline information was collected, including brain imaging with computed tomography or magnetic resonance imaging, tumor assessment via by American Society of Clinical Oncology JOURNAL OF CLINICAL ONCOLOGY

4 Erlotinib Plus Bevacizumab Maintenance Therapy in Advanced NSCLC RECIST (version 1.), 9 vital signs, physical examination, and blood/urinary laboratory analyses. Tumor assessments were recorded at the end of cycles 2 and 4 (chemotherapy phase) and then every second cycle in the postchemotherapy phase (to cycle 12 and every four cycles thereafter) and at the study treatment termination visit (approximately 3 7 days after study treatment discontinuation). Disease progression was classified according to RECIST, 9 by investigator assessment. AEs were recorded after every treatment cycle (chemotherapy and postchemotherapy phases) and up to 3 days after treatment termination and were mapped to the Medical Dictionary for Regulatory Activities terms and classified by grade according to the National Cancer Institute Common Terminology Criteria for Adverse Events (version 3.). Statistical Analyses The primary end point was PFS in patients randomly assigned to bevacizumab/erlotinib versus bevacizumab/placebo in the postchemotherapy phase. PFS was defined as the length of time from random assignment until disease progression, assessed by RECIST, 9 or death from any cause (when this occurred within 3 days of stopping study treatment). Tumor measurements from the assessment before random assignment served as the baseline for determining progression. Secondary efficacy end points included OS (time from random assignment to death) and safety. Exploratory efficacy analyses included the effects of baseline characteristics on PFS and OS. PFS and OS were tested using a two-sided stratified log-rank test (stratified by all random assignment stratification factors except study site) with an overall.5 controlled by a fixed sequence approach. 1 Median PFS and OS were estimated using the Kaplan-Meier method, and the HR was estimated using a stratified Cox regression model. All analyses were conducted by the study statistician with input from the lead investigators. The primary population for the efficacy analyses included all patients randomly assigned during the postchemotherapy phase (intent-to-treat [ITT] population). A secondary efficacy population of bevacizumab-treated patients was also analyzed, defined as all enrolled patients who received at least one dose of bevacizumab during the chemotherapy phase. The primary safety analysis was in the postchemotherapy phase (ie, from the first dose of randomly assigned treatment until either 3 days after the final dose in the postchemotherapy phase or until the first dose of postprogression phase treatment). The safety-evaluable randomized (SER) population included all randomly assigned patients who received at least one complete or partial dose of bevacizumab/erlotinib or bevacizumab/ placebo. Data on AEs leading to treatment discontinuation, prespecified AEs of special interest (AESIs; ie, grade 3 GI perforation, arterial thromboembolic events, proteinuria, congestive heart failure, and hypertension), and serious AEs (grade 2 pulmonary hemorrhage and symptomatic CNS hemorrhage) were collected during the initial chemotherapy phase, the postchemotherapy phase, and for up to 3 days after discontinuation of study treatment. All other AEs were recorded during the postchemotherapy phase only. The study had 78% power to detect an increase in PFS from 5.4 to 6.8 months (measured from random assignment) with a two-sided overall significance level of 5% (PFS HR of.79 or a 21% risk reduction in PFS). Assuming a 3% drop out of enrolled patients before random assignment, 1,15 patients Table 1. Baseline Demographics and Clinical Characteristics by Treatment Arm (postchemotherapy phase; July 18, 28, cutoff) Bevacizumab Placebo (n 373) Bevacizumab Erlotinib (n 37) All Patients (N 743) Demographic or Clinical Characteristic No. of Patients % No. of Patients % No. of Patients % Age, years Median Range Sex Male Female ECOG PS Clinical stage IIIb IV Recurrent Histology Adenocarcinoma Large-cell carcinoma Squamous cell carcinoma Smoking status Never Former Current Chemotherapy Carboplatin paclitaxel Carboplatin gemcitabine Carboplatin docetaxel Cisplatin gemcitabine Abbreviation: ECOG PS, Eastern Cooperative Oncology Group performance status. Data missing for one patient in the bevacizumab placebo arm by American Society of Clinical Oncology 3929

5 Johnson et al Progression-Free Survival (proportion) No. at risk Bev + Er Bev + PI Time (months) = censored value Bev + PI (n = 373, median = 3.71) Bev + Er (n = 37, median = 4.76) HR =.78 (95% CI,.58 to.864) Log-rank P < Fig 3. Kaplan-Meier curve of progressionfree survival for patients receiving bevacizumab/placebo (Bev Pl) or bevacizumab/ erlotinib (Bev Er; stratified analysis, July 18, 28, cutoff). HR, hazard ratio. Time Since Random Allocation 3 months 6 months 9 months 12 months Patients With Event 138 (37.%) 2 (53.6%) 216 (57.9%) 226 (6.6%) Bevacizumab + Placebo (n = 373) Kaplan Meier Percent Event Free (Pointwise 95% CI) 54.1 (48.2 to 59.6) 28.2 (2 to 3) 19.3 (14.3 to 24.9) 12.1 (7.8 to 17.5) Patients With Event 92 (24.9%) 16 (43.2%) 177 (47.8%) 193 (52.2%) Bevacizumab + Erlotinib (n = 37) Kaplan Meier Percent Event Free (Pointwise 95% CI) 68.4 (6 to 73.4) 4.2 (34.1 to 46.2) 3.2 (24.2 to 36.5) 17.8 (12.3 to 24.1) were enrolled to ensure that at least 8 patients were observed and 638 post random assignment PFS events had occurred. Two interim efficacy analyses were prospectively planned, to be performed at 33% and 67% of PFS events. At the second efficacy interim analysis (data cutoff July 18, 28; 69% of PFS events had occurred), the study had met its primary end point of improved PFS, and therefore, this became the primary analysis, with any further analyses of PFS being considered exploratory. An additional exploratory analysis of efficacy and safety was conducted on data collected up to and including January 28, 29, when the positive outcome of the study was announced and the study was unblinded. Furthermore, an additional OS data sweep was performed with a cutoff date of June 19, 29. Given the potential for bias, additional updates for all efficacy analyses except OS were not conducted on data collected after July 18, 28. Unless otherwise stated, the data presented here are for the primary data cutoff (July 18, 28). bevacizumab/placebo arm and 22 patients (54.6%) in the bevacizumab/erlotinib arm had experienced a PFS event (disease progression or death). Median PFS times from time of random assignment were 3.7 and 4.8 months in the bevacizumab/placebo and bevacizumab/erlotinib arms, respectively. This difference was statistically significant for both the stratified (HR,.78; 95% CI,.58 to.864; log-rank P.1; Fig 3) and unstratified analyses (HR,.683; 95% CI,.565 to.824; log-rank P.1). The median OS times from time of random assignment were 13.3 months in the bevacizumab/placebo arm and 14.4 months in the bevacizumab/erlotinib arm (stratified analysis: HR,.917; 95% CI,.698 to 1.25; P.5341; Fig 4). Similar results were also obtained in the unstratified analysis (HR,.941; 95% CI,.726 to 1.22; RESULTS Patient Population A total of 1,145 patients were enrolled onto the chemotherapy phase (Fig 2) from January 26 to June 29. At the time of the July 18, 28, data cutoff, 743 patients (64.9%) had been randomly assigned into the postchemotherapy phase (ITT population) and 735 patients (64.2%) were in the SER population. Because accrual was not halted until after the data monitoring committee recommendation was received (Data Supplement), an additional 25 patients had been randomly assigned by the January 28, 29, data cutoff; the resulting ITT population comprised 768 patients (67.1%) and the SER population comprised 76 patients (66.4%). Baseline characteristics for the ITT population were well balanced for all randomly assigned patients (Table 1). Efficacy The median follow-up times for the bevacizumab/placebo and bevacizumab/erlotinib arms were 8.3 and 8.5 months (July 18, 28, cutoff) and 14.1 and 14.6 months (January 28, 29, cutoff), respectively. At the primary analysis cutoff, 237 patients (63.5%) in the Overall Survival (proportion) No. at risk Bev + Er Bev + PI = censored value Bev + PI (n = 373, median = 13.31) Bev + Er (n = 37, median = 14.39) HR =.917 (95% CI,.698 to 1.25) Log-rank P = Time (months) Fig 4. Kaplan-Meier curve for overall survival for patients receiving bevacizumab/placebo (Bev Pl) or bevacizumab/erlotinib (Bev Er; July 18, 28, cutoff). HR, hazard ratio by American Society of Clinical Oncology JOURNAL OF CLINICAL ONCOLOGY

6 Erlotinib Plus Bevacizumab Maintenance Therapy in Advanced NSCLC P.6473). The 1-year survival rates were 58.6% and 6.1% in the bevacizumab/placebo and bevacizumab/erlotinib arms, respectively. At the updated data analysis cutoff (January 28, 29), median OS times from time of random assignment were 13.6 and 14.4 months in the bevacizumab/placebo and bevacizumab/erlotinib arms, respectively (HR,.94; 95% CI,.729 to 1.12; P.3534; Appendix Fig A1, online only). Final OS data from the June 19, 29, cutoff are included in Appendix Figure A2 (online only). The reduction in the risk of disease progression or death for clinically important subgroups was consistent with the overall treatment effect as showninfigure5.posthocanalysisofoverallresponserateinthepostchemotherapy phase is included in the Data Supplement. Biomarker analyses (cutoff July 18, 28) found that patients whose tumors harbored an activating EGFR mutation (n 52) had a greater improvement in PFS with bevacizumab/erlotinib (n 27) versus bevacizumab/placebo (n 25; HR,.44; 95% CI,.22 to.86) Bevacizumab + Placebo Bevacizumab + Erlotinib Hazard Ratio Total No. of Patients Median (months) Total No. of Patients Median (months) Hazard Ratio 95% CI Age group, years < to to.9 Age group, years < to to 1.3 Bevacizumab + Erlotinib Better Bevacizumab + Placebo Better Race/ethnicity White Asian or Pacific Islander to.92.6 to to 1.13 Sex Male Female to to.83 Smoking history Never Former Current to to to 1.1 ECOG PS at random allocation to to.89 ECOG PS at enrollment to to.99 Chemotherapy Carboplatin + paclitaxel Carboplatin + gemcitabine Carboplatin + docetaxel Cisplatin + gemcitabine to to to to to 1.61 Squamous cell carcinoma No Yes to to 4.8 History of brain metastases No Yes to to 1.31 Histology Adenocarcinoma Large cell carcinoma Squamous cell carcinoma to.8.36 to to to 1.96 Current cancer stage Stage IIIB Stage IV Recurrent NSCLC to to to 1.76 Prior radiotherapy No Yes to to Fig 5. Progression-free survival (PFS) by baseline characteristics for all randomly assigned patients (July 18, 28, cutoff). ECOG PS, Eastern Cooperative Oncology Group performance status; NSCLC, non small-cell lung cancer by American Society of Clinical Oncology 3931

7 Johnson et al compared with the wild-type subgroup (total, n 295; n 15 for bevacizumab/erlotinib and n 145 for bevacizumab/placebo; HR,.85; 95% CI,.64 to 1.13). OS results at the July 19, 29, cutoff showed a similar difference in outcome by EGFR mutation status (HR,.46; 95% CI,.21 to 1.2 for the EGFR mutation-positive subgroup; and HR,.86; 95% CI,.65 to 1.15 for the EGFR wild-type subgroup). Safety Treatment exposure. At data cutoff (July 18, 28), in the postchemotherapy phase, 732 patients (98.5%) received at least one dose of bevacizumab (bevacizumab/placebo, n 365; bevacizumab/erlotinib, n 367), and 734 patients (98.8%) received erlotinib or placebo (bevacizumab/placebo, n 366; bevacizumab/erlotinib n 368). A total of 19 patients (29.2%) in the bevacizumab/placebo arm and 126 patients (34.1%) in the bevacizumab/erlotinib arm remained on bevacizumab, and 124 patients (33.5%) in the bevacizumab/erlotinib arm were still receiving erlotinib as of July 18, 28. The median durations of postchemotherapy bevacizumab were 64 and 72 days in the bevacizumab/placebo and bevacizumab/erlotinib arms, respectively. The median durations of erlotinib and placebo were 85 and 78 days, respectively. More patients missed a dose or had a dose reduction of erlotinib/placebo in the bevacizumab/erlotinib arm than in the bevacizumab/placebo arm (missing dose: 21 patients [57.1%] v 121 patients [33.1%], respectively; dose reduction: 95 patients [25.8%] v seven patients [1.9%], respectively). AEs. At the data cutoff (July 18, 28), during the postchemotherapy phase only, more AEs were reported in the bevacizumab/ erlotinib arm than in the bevacizumab/placebo arm (Table 2). Table 2. Overview of Safety Results During the Postchemotherapy Phase: SER Population (July 18, 28, cutoff) Patients Receiving Bevacizumab Placebo (n 367) Patients Receiving Bevacizumab Erlotinib (n 368) All Grades Grade 3 or 4 Grade 5 All Grades Grade 3 or 4 Grade 5 Adverse Event No. % No. % No. % No. % No. % No. % All AESIs Hemorrhage Proteinuria Hypertension Cardiovascular event other than hypertension Rash Diarrhea Infection AESIs leading to bevacizumab discontinuation Any Hemorrhage Pulmonary hemorrhage GI perforation 2.5 Proteinuria RPLS 2.5 ATE VTE Rash 2.5 Diarrhea Infection ILD-like event Renal failure/deficiency 2.5 AESIs leading to erlotinib/placebo discontinuation Any Hemorrhage Pulmonary hemorrhage GI perforation 2.5 Cardiovascular events other than hypertension ATE VTE Rash Diarrhea Infection ILD-like event Abbreviations: AESI, adverse event of special interest; ATE, arterial thromboembolic event; ILD, interstitial lung disease; RPLS, reversible posterior leukoencephalopathy syndrome; SER, safety-evaluable randomly assigned; VTE, venous thromboembolic event. Incidence of 5% for all grades in at least one treatment arm. Discontinuation of study drug in at least two patients in at least one study arm. Of the nine patients with pulmonary hemorrhage, seven had adenocarcinoma and two had other histology by American Society of Clinical Oncology JOURNAL OF CLINICAL ONCOLOGY

8 Erlotinib Plus Bevacizumab Maintenance Therapy in Advanced NSCLC Prespecified AESIs were predominantly grade 1 or 2, and the incidence was similar between treatment arms, with the exception of rash, diarrhea, and infection, which were more common with bevacizumab/ erlotinib. No grade 3 or 4 interstitial lung disease like events were reported in the bevacizumab/placebo arm, but three patients (.8%) in the bevacizumab/erlotinib arm experienced a grade 3 interstitial lung disease like event (after 15, 41, and 143 days of bevacizumab/ erlotinib treatment). Rash and diarrhea were the only grade 3 or 4 AEs reported at a 5% higher incidence in the bevacizumab/erlotinib arm than the bevacizumab/placebo arm (rash: 6.8% v.5%, respectively; diarrhea: 9.8% v 1.9%, respectively). Six grade 5 AESIs were reported, two with bevacizumab/placebo (congestive heart failure and lobar pneumonia) and four with bevacizumab/erlotinib (cardiac arrest, n 2; cerebellar infarction, n 1; and deep vein thrombosis, n 1). One additional grade 5 event (GI perforation) was reported in the bevacizumab/erlotinib arm after the July 18, 28, cutoff. DISCUSSION After four cycles of chemotherapy plus bevacizumab, the addition of erlotinib to maintenance bevacizumab significantly improved PFS in patients with advanced NSCLC. This PFS benefit was consistently observed in prespecified subgroups defined by stratification factors and other baseline characteristics. As expected after the results of other recent studies, EGFR mutation status was strongly associated with exceptional PFS benefit from erlotinib in this study. These results are comparable with recent data from other advanced NSCLC studies showing a significant PFS benefit after maintenance therapy. 2,3,14-18 In AVAiL, the addition of bevacizumab (7.5 or 15. mg/kg) to platinum-based doublet chemotherapy achieved a significant but modest PFS prolongation versus placebo (6.5 to 6.7 v 6.1 months, respectively; P.3) when administered until disease progression in advanced nonsquamous NSCLC. 2,3 The Sequential Tarceva in Unresectable NSCLC (SATURN) study reported significantly longer PFS and OS for patients receiving erlotinib maintenance versus placebo (median PFS, 12.3 v 11.1 weeks, respectively; P.1; median OS, 12. v 11. months, respectively; P.88), with the greatest benefit being observed in the EGFR mutation-positive subgroup (HR,.1; P.1). 11,16 Both docetaxel and pemetrexed maintenance therapy demonstrated prolonged PFS versus waiting for second-line treatment, 14,15 whereas data from the AVAPERL study suggest that the addition of pemetrexed to bevacizumab maintenance therapy reduces the risk of progression compared with bevacizumab alone, adding up to 4 months in PFS benefit. 17 In addition, the PARAMOUNT study of continuation maintenance with pemetrexed after a cisplatin/pemetrexed first-line regimen found that maintenance pemetrexed significantly reduced the risk of progression (P.1) 18 and prolonged survival by approximately 3 months versus placebo (P.2). 19 Examining the results of this study versus the SATURN study of erlotinib monotherapy in the maintenance setting, the PFS benefit was similar; median PFS in the erlotinib arm of SATURN was 2.8 months, 16 compared with 3.7 months (bevacizumab/placebo) and 4.8 months (bevacizumab/erlotinib) in ATLAS. In both studies, EGFR mutation-positive status was a strong predictor of increased PFS benefit with erlotinib. The point estimate for OS HR in this study was only marginally in favor of the combination regimen (HR,.897; 95% CI,.74 to 1.87) and not statistically significant up to the last follow-up (June 19, 29). It should be noted that the study was not powered to detect differences in OS, it was unblinded after the interim analysis, and further survival follow-up was not pursued based on the low likelihood of observing significant differences between arms. More than 5% of patients in each arm had subsequent second-line therapy; 41% of patients in each arm received second-line erlotinib, whereas 41% and 25% of patients in the bevacizumab/placebo and bevacizumab/erlotinib arms, respectively, received subsequent bevacizumab. This crossover highlights the difficulties of using OS as an end point for studies in advanced NSCLC, particularly for first-line studies where patients may go on to receive multiple additional lines of therapy. Although a significant OS benefit was observed in the E4599 study (12.3 months with chemotherapy/ bevacizumab v 1.3 months with chemotherapy alone; P.3), 1 no survival difference was seen between treatment arms in AVAiL where patients received gemcitabine/cisplatin plus bevacizumab (7.5 or 15 mg/kg) to progression or gemcitabine/cisplatin alone. 3 studies have also failed to convert a PFS benefit into an OS benefit in the maintenance 15 or second-line settings, 8 although the JMEN study reported a significant OS advantage with pemetrexed versus placebo in the overall (13.4 v 1.6 months, respectively; P.12) and nonsquamous populations (15.5 v 1.3 months, respectively; P.2). 14 The choice of end point is clearly an important factor in whether a study is positive or negative, and this has implications for future trials in NSCLC. It could be argued that PFS treatment effects should be much larger, in which case trials would not need to be sized to detect a survival difference (and a greater degree of crossover could be permitted). Alternatively, in unselected populations, OS could be used as the primary end point, although it is possible that larger treatment effects should also be considered for this end point. Safety data during the chemotherapy phase of the study indicate an acceptable safety profile for bevacizumab in combination with a variety of chemotherapy regimens, with results similar to those reported in previous phase III trials. 1-3 During the postchemotherapy phase, the bevacizumab/erlotinib regimen was well tolerated, although toxicity was increased compared with the safety profile of either agent alone. 1-4 In conclusion, ATLAS met its primary end point; maintenance treatment with bevacizumab/erlotinib improved PFS compared with bevacizumab maintenance alone. However, OS was not improved, and treatment with bevacizumab/erlotinib resulted in increased toxicity. Therefore, the results of this study do not provide enough evidence to support a new standard of care, particularly with regard to the benefit-risk profile of this regimen. AUTHORS DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST Although all authors completed the disclosure declaration, the following author(s) and/or an author s immediate family member(s) indicated a financial or other interest that is relevant to the subject matter under consideration in this article. Certain relationships marked with a U are those for which no compensation was received; those relationships marked with a C were compensated. For a detailed description of the disclosure categories, or for more information about ASCO s conflict of interest policy, please refer to the Author Disclosure Declaration and the Disclosures of Potential Conflicts of Interest section in Information for Contributors by American Society of Clinical Oncology 3933

9 Johnson et al Employment or Leadership Position: Bruce E. Johnson, KEW Group (C); Chin-Yu Lin, Genentech (C); Chris Bowden, Genentech (C) Consultant or Advisory Role: Bruce E. Johnson, Genentech (C), AstraZeneca (C), Chugai Pharmaceutical (C), sanofi-aventis (C), Transgenomics (C); Thomas Marsland, Celgene (C), Eli Lilly (C), Spectrum GHC (C); Taral Patel, Genentech (C); James Chih-Hsin Yang, Eli Lilly (U), Boehringher Ingelheim (U), Novartis (C), Roche (C), OSI Astellas (C), Pharmagene (C), Pfizer (C), Takeda Pharmaceuticals (C), Clovis Oncology (C), TTY Biopharm (C), InnoPharma (C), MSD (C), Merck (C); Vincent Miller, Genentech (C), sanofi-aventis (C) Stock Ownership: Bruce E. Johnson, KEW Group; Chin-Yu Lin, Genentech; Thomas Marsland, Johnson & Johnson, Novartis, RainTree Oncology Services, Seattle Genetics, Pfizer; Chris Bowden, Roche Honoraria: Fairooz Kabbinavar, Roche, Genentech; Thomas Marsland, Oncology Business Review; Taral Patel, Genentech; James Chih-Hsin Yang, AstraZeneca, Roche, Bayer AG Research Funding: None Expert Testimony: None Patents: None Remuneration: Bruce E. Johnson, Postmarketing Royalties for EGFR Mutation Testing AUTHOR CONTRIBUTIONS Conception and design: Bruce E. Johnson, Fairooz Kabbinavar Provision of study materials or patients: Bruce E. Johnson, Fairooz Kabbinavar, Louis Fehrenbacher, John Hainsworth, Thomas Marsland, Jonathan Polikoff, Mark Rubin, Leonard White, Vincent Miller Collection and assembly of data: Bruce E. Johnson, Fairooz Kabbinavar, Louis Fehrenbacher, John Hainsworth, Saifuddin Kasubhai, Bruce Kressel, Thomas Marsland, Taral Patel, Jonathan Polikoff, Mark Rubin, Leonard White, James Chih-Hsin Yang Data analysis and interpretation: Bruce E. Johnson, Fairooz Kabbinavar, Louis Fehrenbacher, Chin-Yu Lin, Jonathan Polikoff, Chris Bowden, Vincent Miller Manuscript writing: All authors Final approval of manuscript: All authors REFERENCES 1. Sandler A, Gray R, Perry MC, et al: Paclitaxelcarboplatin alone or with bevacizumab for non-smallcell lung cancer. N Engl J Med 355: , Reck M, von Pawel J, Zatloukal P, et al: Phase III trial of cisplatin plus gemcitabine with either placebo or bevacizumab as first-line therapy for nonsquamous non-small-cell lung cancer: AVAiL. J Clin Oncol 27: , Reck M, von Pawel J, Zatloukal P, et al: Overall survival with cisplatin-gemcitabine and bevacizumab or placebo as first-line therapy for nonsquamous non-small-cell lung cancer: Results from a randomised phase III trial (AVAiL). Ann Oncol 21: , Shepherd FA, Rodrigues Pereira J, Ciuleanu T, et al: Erlotinib in previously treated non-small-cell lung cancer. N Engl J Med 353: , Detterbeck FC, Boffa DJ, Tanoue LT: The new lung cancer staging system. Chest 136:26-271, Herbst RS, Johnson DH, Mininberg E, et al: Phase I/II trial evaluating the anti-vascular endothelial growth factor monoclonal antibody bevacizumab in combination with the HER-1/epidermal growth factor receptor tyrosine kinase inhibitor erlotinib for patients with recurrent non small-cell lung cancer. J Clin Oncol 23: , Herbst RS, O Neill VJ, Fehrenbacher L, et al: Phase II study of efficacy and safety of bevacizumab in combination with chemotherapy or erlotinib compared with chemotherapy alone for treatment of recurrent or refractory non small-cell lung cancer. J Clin Oncol 25: , Herbst RS, Ansari R, Bustin F, et al: Efficacy of bevacizumab plus erlotinib versus erlotinib alone in advanced non-small-cell lung cancer after failure of standard first-line chemotherapy (BeTa): A double-blind, placebo-controlled, phase 3 trial. Lancet 377: , Therasse P, Arbuck SG, Eisenhauer EA, et al: New guidelines to evaluate the response to treatment in solid tumors: European Organization for Research and Treatment of Cancer, National Cancer Institute of the United States, National Cancer Institute of Canada. J Natl Cancer Inst 92:25-216, 2 1. Westfall PH, Krishen A: Optimally weighted, fixed sequence and gatekeeper multiple testing procedures. J Stat Plan Inference 99:25-4, Brugger W, Triller N, Blasinska-Morawiec M, et al: Prospective molecular marker analyses of EGFR and KRAS from a randomized, placebo-controlled study of erlotinib maintenance therapy in advanced non small-cell lung cancer. J Clin Oncol 29: , Zhou C, Wu YL, Chen G, et al: Erlotinib versus chemotherapy as first-line treatment for patients with advanced EGFR mutation-positive non-smallcell lung cancer (OPTIMAL, CTONG-82): A multicentre, open-label, randomised, phase 3 study. Lancet Oncol 12: , Rosell R, Carcereny E, Gervais R, et al: Erlotinib versus standard chemotherapy as first-line treatment for European patients with advanced EGFR mutation-positive non-small-cell lung cancer (EURTAC): A multicentre, open-label, randomised phase 3 trial. Lancet Oncol 13: , Ciuleanu T, Brodowicz T, Zielinski C, et al: Maintenance pemetrexed plus best supportive care versus placebo plus best supportive care for nonsmall-cell lung cancer: A randomised, double-blind, phase 3 study. Lancet 374: , Fidias PM, Dakhil SR, Lyss AP, et al: Phase III study of immediate compared with delayed docetaxel after front-line therapy with gemcitabine plus carboplatin in advanced non-small-cell lung cancer. J Clin Oncol 27: , Cappuzzo F, Ciuleanu T, Stelmakh L, et al: Erlotinib as maintenance treatment in advanced non-small-cell lung cancer: A multicentre, randomised, placebo-controlled phase 3 study. Lancet Oncol 11: , Barlesi F, de Castro J, Dvornichenko V, et al: Final efficacy outcomes for patients (pts) with advanced nonsquamous non-small cell lung cancer (nsnsclc) randomised to continuation maintenance (mtc) with bevacizumab (bev) or bev pemetrexed (pem) after firstline (1l) bev-cisplatin (cis)-pem treatment (Tx). European Multidisciplinary Cancer Congress, Stockholm, Sweden, September 23-27, 211 (abstr 34LBA) 18. Paz-Ares L, de Marinis F, Dediu M, et al: Maintenance therapy with pemetrexed plus best supportive care versus placebo plus best supportive care after induction therapy with pemetrexed plus cisplatin for advanced non-squamous non-small-cell lung cancer (PARAMOUNT): A double-blind, phase 3, randomised controlled trial. Lancet Oncol 13: , Paz-Ares L, de Marinis F, Dediu M, et al: PARAMOUNT: Final overall survival (OS) results of the phase III study of maintenance pemetrexed (pem) plus best supportive care (BSC) versus placebo (plb) plus BSC immediately following induction treatment with pem plus cisplatin (cis) for advanced nonsquamous (NS) nonsmall cell lung cancer (NSCLC). J Clin Oncol 3, 212 (suppl 18s; abstr LBA757) Affiliations Bruce E. Johnson, Dana-Farber Cancer Institute, Boston, MA; Fairooz Kabbinavar, University of California Los Angeles, Translational Oncology Research International, Los Angeles; Louis Fehrenbacher, Kaiser Permanente Northern California, Vallejo; Chin-Yu Lin and Chris Bowden, Genentech, South San Francisco; Jonathan Polikoff, Southern California Permanente Medical Group, San Diego, CA; John Hainsworth, Sarah Cannon Research Institute, Nashville, TN; Saifuddin Kasubhai, Northwest Medical Specialties, Tacoma, WA; Bruce Kressel, Sibley Memorial Hospital, Washington, DC; Thomas Marsland, Integrated Community Oncology Network, Orange Park; Mark Rubin, Florida Cancer Specialists, Fort Myers, FL; Taral Patel, The Mark H. Zangmeister Center, Columbus, OH; Leonard White, Arch Medical Services, The Center for Cancer Care and Research, Saint Louis, MO; Vincent Miller, Weill Cornell Medical College and Thoracic Oncology Service, Memorial Sloan-Kettering Cancer Center, New York, NY; and James Chih-Hsin Yang, National Taiwan University, Taipei, Taiwan by American Society of Clinical Oncology JOURNAL OF CLINICAL ONCOLOGY

10 Erlotinib Plus Bevacizumab Maintenance Therapy in Advanced NSCLC Acknowledgment Third-party writing assistance for this article was provided by Rhiannon Owen (Gardiner-Caldwell Communications) and funded by Genentech. Appendix Overall Survival (proportion) = censored value Bev + PI (n = 385, median = 13.63) Bev + Er (n = 383, median = 14.39) HR =.94 (95% CI,.729 to 1.12) Log-rank P =.3534 No. at risk Bev + Er Bev + PI Time (months) Fig A1. Kaplan-Meier curve for overall survival for patients receiving bevacizumab/placebo (Bev Pl) or bevacizumab/erlotinib (Bev Er; January 28, 29, cutoff). HR, hazard ratio. Overall Survival (proportion) = censored value Bev + PI (n = 385, median = 13) Bev + Er (n = 383, median = 15.93) HR =.897 (95% CI,.74 to 1.87) Log-rank P =.2665 No. at risk Bev + Er Bev + PI Time (months) Fig A2. Kaplan-Meier curve for overall survival for patients receiving bevacizumab/placebo (Bev Pl) or bevacizumab/erlotinib (Bev Er; July 19, 29, cutoff). HR, hazard ratio by American Society of Clinical Oncology