Abstract

Background

Previous studies have demonstrated the efficacy and safety of bevacizumab in the treatment of non-small-cell lung cancer (NSCLC).

Methods

Summary data from randomised trials comparing first-line bevacizumab plus platinum-based chemotherapy with chemotherapy alone for inoperable locally advanced, recurrent or metastatic NSCLC were meta-analysed. Pooled hazard ratios (HRs) for overall survival (OS) and progression-free survival (PFS), and pooled odds ratio (OR) for adverse events were calculated. The chi-squared tests evaluated interactions between treatment effects, and prognostic factors and patient characteristics.

Results

Data of 2194 patients (1313 bevacizumab; 881 controls) from four phase II and III trials: AVF-0757g, JO19907, ECOG 4599 and AVAiL, were analysed. Compared with chemotherapy alone, bevacizumab significantly prolonged OS (HR 0.90; 95% confidence interval [CI] 0.81, 0.99; P = 0.03), and PFS (0.72; 95% CI 0.66, 0.79; P < 0.001). Bevacizumab showed a significantly greater effect on OS in patients with adenocarcinoma versus other histologies (P = 0.02), and patients with body weight loss ≤5% versus >5% (P = 0.03). Bevacizumab significantly increased the risk of grade ≥3 proteinuria, hypertension, haemorrhagic events, neutropenia, and febrile neutropenia.

Conclusions

Bevacizumab significantly prolonged OS and PFS when added to first-line platinum-based chemotherapy in patients with advanced NSCLC; no unexpected toxicity was evident.

introduction

Lung cancer is the leading cause of cancer-related mortality, killing almost 1.4 million people worldwide each year [1]. Non-small-cell lung cancer (NSCLC) accounts for most cases of lung cancer, and carries a poor prognosis when locally advanced or metastatic. The standard platinum-based chemotherapy regimens recommended for the first-line treatment of advanced NSCLC [2] provide limited clinical benefit [3–8]. More efficacious and less toxic drugs are needed.

Tumour angiogenesis is critical for tumour progression. The vascular endothelial growth factor (VEGF) promotes angiogenesis [9, 10] and over expression of the VEGF has been correlated with poor prognosis in various malignancies, including NSCLC [11]. Bevacizumab is a recombinant, humanised, monoclonal antibody against the VEGF [12]. Bevacizumab has demonstrated significant clinical benefit in first- and second-line metastatic colorectal cancer [13, 14], and in the first-line treatment of metastatic breast cancer [15, 16] and renal cell cancer [17–19].

Bevacizumab (Avastin®, Roche) is approved for the first-line treatment of NSCLC [20] based on the data from two randomised, phase III clinical trials. In the open-label Eastern Cooperative Oncology Group (ECOG) 4599 study, the addition of bevacizumab to carboplatin–paclitaxel (Bristol-Myers Squibb) (Taxol®) significantly extended overall survival (OS), progression-free survival (PFS) and improved the objective response rate, versus chemotherapy alone [21]. In the double-blind Avastin in the Lung (AVAiL) trial (BO17704), bevacizumab significantly extended PFS and improved the response rate versus placebo when added to cisplatin plus gemcitabine, but did not significantly extend OS [22–23]. Adverse events associated with bevacizumab include hypertension, proteinuria, bleeding and neutropenia [21, 22, 24]. There are no predictive biomarkers of bevacizumab efficacy. Currently, the patient selection criteria are based on the risk factors associated with bevacizumab toxicity. For example, patients with predominantly squamous tumours, those with clinically significant haemoptysis and (in AVAiL) those with tumours invading or abutting major blood vessels were excluded from phase III and IV trials in order to reduce the risk of pulmonary haemorrhage during bevacizumab therapy [21, 22, 25]. Some evidence suggests that cavitation may also increase the risk of pulmonary haemorrhage [26], but the data are conflicting [27].

This paper reports a summary data meta-analysis from randomised, phase II and III trials, which was conducted to further assess the efficacy (in terms of OS and PFS) and toxicity of bevacizumab used in combination with platinum-based chemotherapy, compared with chemotherapy alone, in the first-line treatment of patients with advanced NSCLC. This analysis was initially intended as the first stage before an analysis of individual patient data from relevant trials.

methods

The protocol for this analysis is available at http://www.igr.fr/index.php?p_id=1349.

trials

Eligible studies were randomised, clinical trials comparing bevacizumab plus platinum-based chemotherapy with chemotherapy alone as first-line therapy for inoperable locally advanced (stage IIIB), recurrent or metastatic (stage IV) NSCLC. Eligible trials were un-confounded, i.e. having treatment arms that differed only with regard to bevacizumab administration.

Trials were identified on 24 April 2009 from electronic publication databases (Pubmed/MEDLINE, SCOPUS, Cochrane Library), trial registers (www.clinicaltrials.gov), scientific conference proceedings (American Society of Clinical Oncology abstract database and World Conference on Lung Cancer proceedings 2005 and 2007) and review articles (see the online Appendix available at Annals of Oncology online for search strategy). Both the published and unpublished trials were sought and were eligible for inclusion, in order to minimise publication bias [28]. Eligible trials were independently selected by two persons with discussions with a third person in case of disagreement. Statistical reports and individual patient data were requested from trial investigators and the manufacturer of bevacizumab.

data extraction

Data were extracted from publications or trial reports independently by two persons using a specially designed form. All data were checked for consistency with the trial protocol, statistical report and publication. The following data were extracted: treatments and doses; patient characteristics/prognostic factors; hazard ratios (HRs) for OS and PFS (overall and by patient subgroup) and rates of major toxic effects (i.e. haematological, haemorrhagic, gastrointestinal, renal, cardiovascular, thromboembolic and neurological adverse events).

To assess trial quality, the following items were extracted: blinding, method of randomisation—including stratification factor—number of patients randomly assigned excluded from analysis by arm, patients follow-up time if possible by arm, number of patients lost to follow-up by arm. The quality of trials and the risk of bias were assessed by considering randomisation methods, stratification factors, blinding, follow-up and intention-to-treat analysis.

statistical analysis

Where possible, data were analysed from all randomly assigned patients on an intention-to-treat basis. However, when patients were excluded from summary data in trials, only the analysed patients were included. The primary analysis compared bevacizumab plus chemotherapy with chemotherapy-only/placebo in terms of OS. Secondary analyses were carried out on PFS, rates of major toxic effects, the effect of bevacizumab dose on efficacy and toxicity, and the interaction between the treatment effect and patient characteristics.

Fixed-effect models were used to compute pooled HRs for OS and PFS data, and pooled odds ratios (ORs) for adverse events, with two-sided 95% confidence intervals (CI) and P values. Random-effect models were used in cases of significant and unexplained heterogeneity. Trials that evaluated two bevacizumab doses used the same control arm for the two comparisons. Thus, for these trials with a double count of the placebo arm, the weight of each comparison was reduced according to a correction factor equal to the number of events observed in the trial (three arms) divided by the number of events taken into account in the analysis (four arms). This correction resulted in an increase in the width of the CI for the estimated HR.

The chi-squared heterogeneity test was used to test for gross statistical heterogeneity between the trials [29]. The I2 statistic (0%–100%) was used to assess the proportion of variability in the results that was attributable to heterogeneity between the trials [30]. Chi-squared tests were used to test for any interactions between the treatment effect and the following prognostic factors and patient characteristics: age, sex, tumour histology, cancer stage, performance status, race, body weight loss and smoking status. All analyses were stratified by trial. Computations were carried out using SAS 9.1 software (SAS Institute Inc., Cary).

results

trials and patients

Fourteen trials were screened for eligibility (supplementary Figure S1, available at Annals of Oncology online). Of these, four eligible randomised trials were included in the meta-analysis: the phase II AVF-0757g [24] and JO19907 [31] trials, and the phase III ECOG 4599 [21] and AVAiL trials (Table 1) [22, 23]. The detailed statistical reports were available for all trials, except JO19907, and were used in this analysis.

Table 1.

Characteristics of the four analysed trials of bevacizumab added to standard chemotherapy as the first-line treatment in patients with advanced NSCLC

Trial Period of inclusion Design, main inclusion/exclusion criteria, primary end point Random-isation Treatment armsa N analysed /randomly assigned patients 
AVF-0757g [24May 1998–Sep 1999 Design: open-label, parallel-group, multicentre, blinded assessment phase II
Inclusion critera: histologically confirmed stage IIIB (with pleural effusion), stage IV or recurrent NSCLC; ECOG PS ≤2; life expectancy ≥3 months; no previous chemotherapy, biological therapy or radiotherapy. Exclusions included: CNS metastasis, therapeutic anticoagulation, uncontrolled hypertension
Primary end point: PFS
Tumour assessment: every three cycles (i.e. 9 weeks) for the first six cycles and every four cycles (12 weeks) thereafter 
1:1:1 Bevacizumab 7.5 mg/kg + carboplatin + paclitaxel 32/32 
Bevacizumab 15 mg/kg + carboplatin + paclitaxel 34/35 
Carboplatin + paclitaxel 32/32 
ECOG 4599 [21Jul 2001–Dec 2005 Design: open-label, parallel-group, multicentre, phase III
Inclusion criteria: histologically or cytologically confirmed, predominantly non-squamous stage IIIB (with pleural effusion), stage IV or recurrent NSCLC; ECOG PS 0–1; no previous chemotherapy. Exclusions included: haemoptysis (≥2.5 ml per episode), tumours invading or abutting major blood vessels, CNS metastasis, therapeutic anticoagulation, uncontrolled hypertension
Primary end point: OS
Tumour assessment: every two cycles (i.e. every 6 weeks) for 24 weeks and then every three cycles thereafter 
1:1 Bevacizumab 15 mg/kg + carboplatin + paclitaxel 434/434 
Carboplatin + paclitaxel 444/444 
AVAiL [22Feb 2005–Aug 2006 Design: double-blind, parallel-group, multicentre, international, phase III
Inclusion criteria: histologically or cytologically confirmed, stage IIIB (with supraventricular lymph node metastasis, or malignant pleural or pericardial effusion), stage IV or recurrent non-squamous NSCLC; ECOG PS 0–1; no previous chemotherapy. Exclusions included: haemoptysis (≥2·5 ml per episode), CNS metastasis, therapeutic anticoagulation, uncontrolled hypertension
Primary end point: PFS
Tumour assessment: every three cycles 
1:1:1 Bevacizumab 7.5 mg/kg + cisplatin + gemcitabine 345/345 
Bevacizumab 15 mg/kg + cisplatin + gemcitabine 351/351 
Cisplatin + gemcitabine + placebo (low or high dose) 347/347 
JO19907 [31Apr 2007– Mar 2008 Design: open-label, parallel-group, multicentre, phase II
Inclusion criteria: previously untreated stage IIIB (with pleural and/or pericardial effusion and/or pleural dissemination), IV or recurrent non-squamous NSCLC; ECOG PS 0–1. Exclusions included haemoptysis and CNS metastasis, uncontrolled hypertension
Primary end point: PFS
Tumour assessment: every 6 weeks for the first 18 weeks and every 9 weeks thereafter 
2b:1 Bevacizumab 15 mg/kg + carboplatin + paclitaxel 117/121 
Carboplatin + paclitaxel 58/59 
Trial Period of inclusion Design, main inclusion/exclusion criteria, primary end point Random-isation Treatment armsa N analysed /randomly assigned patients 
AVF-0757g [24May 1998–Sep 1999 Design: open-label, parallel-group, multicentre, blinded assessment phase II
Inclusion critera: histologically confirmed stage IIIB (with pleural effusion), stage IV or recurrent NSCLC; ECOG PS ≤2; life expectancy ≥3 months; no previous chemotherapy, biological therapy or radiotherapy. Exclusions included: CNS metastasis, therapeutic anticoagulation, uncontrolled hypertension
Primary end point: PFS
Tumour assessment: every three cycles (i.e. 9 weeks) for the first six cycles and every four cycles (12 weeks) thereafter 
1:1:1 Bevacizumab 7.5 mg/kg + carboplatin + paclitaxel 32/32 
Bevacizumab 15 mg/kg + carboplatin + paclitaxel 34/35 
Carboplatin + paclitaxel 32/32 
ECOG 4599 [21Jul 2001–Dec 2005 Design: open-label, parallel-group, multicentre, phase III
Inclusion criteria: histologically or cytologically confirmed, predominantly non-squamous stage IIIB (with pleural effusion), stage IV or recurrent NSCLC; ECOG PS 0–1; no previous chemotherapy. Exclusions included: haemoptysis (≥2.5 ml per episode), tumours invading or abutting major blood vessels, CNS metastasis, therapeutic anticoagulation, uncontrolled hypertension
Primary end point: OS
Tumour assessment: every two cycles (i.e. every 6 weeks) for 24 weeks and then every three cycles thereafter 
1:1 Bevacizumab 15 mg/kg + carboplatin + paclitaxel 434/434 
Carboplatin + paclitaxel 444/444 
AVAiL [22Feb 2005–Aug 2006 Design: double-blind, parallel-group, multicentre, international, phase III
Inclusion criteria: histologically or cytologically confirmed, stage IIIB (with supraventricular lymph node metastasis, or malignant pleural or pericardial effusion), stage IV or recurrent non-squamous NSCLC; ECOG PS 0–1; no previous chemotherapy. Exclusions included: haemoptysis (≥2·5 ml per episode), CNS metastasis, therapeutic anticoagulation, uncontrolled hypertension
Primary end point: PFS
Tumour assessment: every three cycles 
1:1:1 Bevacizumab 7.5 mg/kg + cisplatin + gemcitabine 345/345 
Bevacizumab 15 mg/kg + cisplatin + gemcitabine 351/351 
Cisplatin + gemcitabine + placebo (low or high dose) 347/347 
JO19907 [31Apr 2007– Mar 2008 Design: open-label, parallel-group, multicentre, phase II
Inclusion criteria: previously untreated stage IIIB (with pleural and/or pericardial effusion and/or pleural dissemination), IV or recurrent non-squamous NSCLC; ECOG PS 0–1. Exclusions included haemoptysis and CNS metastasis, uncontrolled hypertension
Primary end point: PFS
Tumour assessment: every 6 weeks for the first 18 weeks and every 9 weeks thereafter 
2b:1 Bevacizumab 15 mg/kg + carboplatin + paclitaxel 117/121 
Carboplatin + paclitaxel 58/59 

aDoses: carboplatin, dosed to a target area under the curve of 6 mg/ml/min; paclitaxel, 200 mg/m2; cisplatin, 80 mg/m2; gemcitabine, 1250 mg/m2. In all trials, treatment was administered in 3-week cycles for up to six cycles, or until disease progression or unacceptable toxicity. Patients who completed six cycles of bevacizumab-containing therapy in ECOG 4599, AVAiL and JO19907 then received bevacizumab monotherapy until disease progression or unacceptable toxicity. In AVF-0757g, non-progressing patients randomly assigned to bevacizumab could receive up to 18 doses of bevacizumab following the initial six cycles. Patients in the control arms were permitted to receive bevacizumab (15 mg/kg) on disease progression.

bExperimental arm.

CNS, central nervous system; NSCLC, non-small-cell lung cancer; ECOG, Eastern Cooperative Oncology Group; PS, performance status; OS, overall survival; PFS, progression-free survival.

In two trials (AVF-0757g and ECOG 4599), stage IIIB NSCLC was defined according to the presence of pleural effusion. The JO19907 trial required pleural and/or pericardial effusion and/or pleural dissemination, while AVAiL specified the presence of supraventricular lymph node metastasis, or malignant pleural or pericardial effusion.

In all trials, treatment was administered in 3-week cycles for up to six cycles, or until disease progression, or unacceptable toxicity. The AVF-0757g and AVAiL trials each evaluated two doses of bevacizumab (7.5 mg/kg and 15 mg/kg). Summary data for the two experimental arms versus the control arm were not available. The ECOG 4599 and JO19907 trials evaluated only one bevacizumab dose (15 mg/kg).The backbone chemotherapy regimen was carboplatin plus paclitaxel in AVF-0757g, JO19907 and ECOG 4599, and cisplatin plus gemcitabine in AVAiL. The AVF-0757g, JO19907 and ECOG 4599 trials compared bevacizumab plus chemotherapy with chemotherapy alone, i.e. with no placebo treatment. The AVAiL trial had two placebo groups, one for each bevacizumab dose arm. However, only pooled data for the two placebo groups were available. The trials varied in the timing of tumour assessment (Table 1). Supplementary Table S1, available at Annals of Oncology online summarises the treatment received after first-line therapy in each trial.

All trials used central randomisation stratified using between one and four factors (Table 2). Only one trial was double-blind. For the main end point of this study, OS, an objective end point, the absence of blinding was not a problem. The proportion of randomly assigned patients excluded from the analysis by trial ranged from 0% to <3% and overall was 0.3%. Follow-up was good without clear imbalance between arms (Tables 1 and 2).

Table 2.

Quality of the four analysed trials

Trial Randomisation methods Stratification factors Double-blind Follow-up Intent to treat 
AVF-0757g [24Centralised ECOG PS No NA No: 1 of 99 excluded from analysis 
ECOG 4599 [21Centralised Measurable disease, prior radiotherapy, weight lost No Median follow-up 19 months in both the arms in the 440/427 patients treated in control/bevacizumab armsa
Rate of loss to follow-up for at least 6 months (statistical report)
-9% in the control arm
-7.8% in the bevacizumab arm 
Yes 
AVAiL [22Centralised ECOG PS, sex, stage, region of the world Yes Median follow-up (rate of loss to follow-up at cut-off date):
-Control arm: 12.5 months (4%)
-Bevacizumab 7.5 mg/kg arm 12.9 months (3%)
-Bevacizumab 15 mg/kg arm: 12.5 months (3%) 
Yes 
JO19907 [31Centralised ECOG PS, sex, stage No Median follow-up of 26 months in both the arms among the 175 included in the analysis
Lost to follow-up: one patient 
No: 5 of 180 excluded from analysis 
Trial Randomisation methods Stratification factors Double-blind Follow-up Intent to treat 
AVF-0757g [24Centralised ECOG PS No NA No: 1 of 99 excluded from analysis 
ECOG 4599 [21Centralised Measurable disease, prior radiotherapy, weight lost No Median follow-up 19 months in both the arms in the 440/427 patients treated in control/bevacizumab armsa
Rate of loss to follow-up for at least 6 months (statistical report)
-9% in the control arm
-7.8% in the bevacizumab arm 
Yes 
AVAiL [22Centralised ECOG PS, sex, stage, region of the world Yes Median follow-up (rate of loss to follow-up at cut-off date):
-Control arm: 12.5 months (4%)
-Bevacizumab 7.5 mg/kg arm 12.9 months (3%)
-Bevacizumab 15 mg/kg arm: 12.5 months (3%) 
Yes 
JO19907 [31Centralised ECOG PS, sex, stage No Median follow-up of 26 months in both the arms among the 175 included in the analysis
Lost to follow-up: one patient 
No: 5 of 180 excluded from analysis 

aThe analysis in the statistical report was updated, compared with the one by Sandler et al [21]. In this paper, among the 867 who started treatment, 3 were not followed and the median follow-up was the same in both the arms (19 months).

ECOG PS, Eastern Cooperative Oncology Group performance status; NA, not available.

The analysis population included 2194 patients out of 2200 randomly assigned patients: 1313 patients in the bevacizumab arm and 881 patients in the control arm (Table 3). Overall, approximately three-quarters of patients had adenocarcinoma histology. A similar proportion had non-recurrent stage IV NSCLC, with 15% having stage IIIB disease.

Table 3.

Characteristics of randomly assigned patients

Characteristics Bevacizumab arm N = 1318a (%) Control arm N = 882a (%) 
Sex (male/female) 59/41 62/38 
Histology 
 Adenocarcinoma 80 75 
 Large cell 
 Otherb 14 16 
Stage 
 IIIB 15 15 
 IV non-recurrent 75 76 
 Recurrent 10 
Performance status (PS, 0/≥1)c 41/59 41/59 
Race (white/non-white)d 79/21 82/18 
Characteristics Bevacizumab arm N = 1318a (%) Control arm N = 882a (%) 
Sex (male/female) 59/41 62/38 
Histology 
 Adenocarcinoma 80 75 
 Large cell 
 Otherb 14 16 
Stage 
 IIIB 15 15 
 IV non-recurrent 75 76 
 Recurrent 10 
Performance status (PS, 0/≥1)c 41/59 41/59 
Race (white/non-white)d 79/21 82/18 

aTable based on randomly assigned patients; data for six patients were not available for this study: one in the AVF-0757g trial (15 mg/kg arm) and five patients in the JO19907 trial (one in the control arm and four patients in the experimental arm).

bIncludes 23 patients with squamous cell tumours in ECOG 4599 and AVF-0757g trials (14 in the bevacizumab arm and 9 in the control arm).

cIncludes seven patients with PS = 2, all in AVF-0757g trials.

dNot available for AVF-0757g. Non-white included black, Hispanic and Asian patients in ECOG 4599 and Asian patients only in AVAiL. All patients in JO19907 were Japanese.

overall survival

The OS analysis was based on 1563 deaths in 2194 patients. The two largest trials (ECOG 4599 and AVAiL) accounted for 91% of the overall variance associated with the estimation of the effect on OS (see the online Appendix). In the AVF-0757g trial, only a P value was available, without information on the direction of the effect; two scenarios were possible: an HR of <1 or >1. The ‘worst scenario’ (i.e. an HR >1) was used in this analysis to give a conservative estimate. OS was significantly longer in patients treated with bevacizumab plus chemotherapy, compared with chemotherapy alone (Figure 1A), with an estimated HR of 0.90 (95% CI 0.81, 0.99; P = 0.03).

Figure 1.

Forest plots of hazard ratios (HRs) for (A) overall survival (OS) and (B) progression-free survival (PFS) from four randomised trials of bevacizumab (7.5 mg/kg or 15 mg/kg) added to standard chemotherapy, compared with chemotherapy alone, as first-line therapy in patients with advanced non-small-cell lung cancer (NSCLC; ECOG 4599, AVAiL, AVF-0757g and JO19907). Pooled HRs were computed using fixed-effect models. The bars indicate 95% confidence intervals (CI). The chi-squared test showed no significant heterogeneity between the trials. (OE, observed-expected).

Figure 1.

Forest plots of hazard ratios (HRs) for (A) overall survival (OS) and (B) progression-free survival (PFS) from four randomised trials of bevacizumab (7.5 mg/kg or 15 mg/kg) added to standard chemotherapy, compared with chemotherapy alone, as first-line therapy in patients with advanced non-small-cell lung cancer (NSCLC; ECOG 4599, AVAiL, AVF-0757g and JO19907). Pooled HRs were computed using fixed-effect models. The bars indicate 95% confidence intervals (CI). The chi-squared test showed no significant heterogeneity between the trials. (OE, observed-expected).

One-year OS rates in the control arm were available from the four trials and were 59% in the AVF-0757g trial, 44% in ECOG 4599, 55% in AVAiL and 74% in JO19907. Based on the variance, the following weights were calculated (six comparisons)—3% for AVF-0757g, 6% for JO199707, 45% for ECOG 4599, 46% for AVAiL—and used to calculate a weighted survival rate in the control arm. The overall weighted OS rate at 1 year was 51%. The absolute benefit of bevacizumab on OS at 1 year was 4% (from 51% to 55%).

There was no significant (P = 0.44) heterogeneity between the trials with regard to the effect of bevacizumab on OS (Figure 1A). There was no significant difference between the two bevacizumab doses (P = 0.64). With regard to AVF-0757g, the rate of deaths and the median OS of the control arm, compared with the experimental arm, favoured a HR >1 for the low-dose comparison (17/32 versus 18/32 deaths; 14.9 versus 11.6 months), and a HR <1 for the high-dose comparison (17/32 versus 15/34 deaths; 14.9 versus 17.7 months). Considering a HR of <1 for the high-dose comparison led to an overall HR of 0.89 (95% CI 0.81, 0.98; P = 0.02). After exclusion of the 20 patients with squamous cell cancer from the AVF-0757g trial, the HR was 0.88 (95% CI 0.80, 0·97; P = 0.01).

progression-free survival

The four trials used different definitions of PFS (supplementary Table S2, available at Annals of Oncology online). The double count of the placebo arm in AVAiL and AVF-0757g trials was corrected in the same manner as for the OS analysis. Estimation of the HR was based on the unadjusted analysis for all trials, except AVF-0757g, for which only adjusted HRs were available.

The analysis of PFS was based on 2194 patients and 1876 events. PFS was significantly longer in patients treated with bevacizumab plus chemotherapy, compared with chemotherapy alone (Figure 1B), with an estimated HR of 0.72 (95% CI 0.66, 0.79; P < 0.001). There was no significant (P = 0.17) heterogeneity between the trials (Figure 1B). There was no significant difference between the two bevacizumab doses (test for interaction, P = 0.55).

interaction between patient characteristics and treatment effect

Subgroup analysis data for OS and PFS were available from the AVAiL, ECOG 4599 and JO19907 trials, representing 2096 patients (95.5% of the total population). There were few missing data (from 0.2%–3% of patients, depending on the variable).

Bevacizumab showed a significantly greater treatment effect on OS in patients with adenocarcinoma than with other histological types (i.e. large cell and other types; P = 0.03), and in patients with a loss in body weight of ≤5%, compared with >5% (P = 0.04) (Figure 2A). There was no significant interaction between the treatment effect and sex, age, histology, performance status or smoking status.

Figure 2.

Forest plots of hazard ratios (HRs) for (A) overall survival (OS) and (B) progression-free survival (PFS) according to patient and tumour characteristics from three randomised trials of bevacizumab (7.5 mg/kg or 15 mg/kg) added to standard chemotherapy, compared with chemotherapy alone, as first-line therapy in patients with advanced non-small-cell lung cancer (NSCLC; ECOG 4599, AVAiL and JO19907).

Figure 2.

Forest plots of hazard ratios (HRs) for (A) overall survival (OS) and (B) progression-free survival (PFS) according to patient and tumour characteristics from three randomised trials of bevacizumab (7.5 mg/kg or 15 mg/kg) added to standard chemotherapy, compared with chemotherapy alone, as first-line therapy in patients with advanced non-small-cell lung cancer (NSCLC; ECOG 4599, AVAiL and JO19907).

Bevacizumab showed a significantly greater treatment effect on PFS in patients with stage IIIB than in non-recurrent stage IV NSCLC (P = 0·007) and in patients with a body weight loss of ≤5%, compared with >5% (P = 0.009) (Figure 2B). There was no significant interaction between the treatment effect and sex, age, histology, performance status, race or smoking status.

toxicity

Data for five of the nine analysed toxic effects (hypertension, thrombosis, haemorrhagic events, neuropathy and neutropenia) were available for all trials. Data for proteinuria and febrile neutropenia were available for all trials, except the AVF-0757g trial; those for thrombocytopenia and anaemia were available for all trials, except the ECOG 4599 trial. For AVAiL and AVF-0757g trials, the number of events in the two experimental arms were pooled and compared with the control arm.

According to the pooled analysis, bevacizumab significantly increased the risk of grade ≥3 events of proteinuria (OR 4.81; 95% CI 2.28, 10.1), hypertension (OR 3.69; 95% CI 2.49, 5.47), haemorrhagic events (OR 2.67; 95% CI 1.63, 4.39), neutropenia (OR 1.53; 95% CI 1.25, 1.87) and febrile neutropenia (OR 1.72; 95% CI 1.01, 2.95), compared with chemotherapy alone (Figure 3). There was no significant heterogeneity between the trials in the toxicity analysis.

Figure 3.

Incidence of adverse events and associated odds ratios (ORs) in four randomised trials of bevacizumab (7.5 mg/kg or 15 mg/kg) added to standard chemotherapy, compared with chemotherapy alone, as first-line therapy in patients with advanced non-small-cell lung cancer (ECOG 4599, AVAiL, AVF0757g and JO19907). A pooled HR was computed using fixed-effect models. The bars indicate 95% confidence intervals (CI).

Figure 3.

Incidence of adverse events and associated odds ratios (ORs) in four randomised trials of bevacizumab (7.5 mg/kg or 15 mg/kg) added to standard chemotherapy, compared with chemotherapy alone, as first-line therapy in patients with advanced non-small-cell lung cancer (ECOG 4599, AVAiL, AVF0757g and JO19907). A pooled HR was computed using fixed-effect models. The bars indicate 95% confidence intervals (CI).

discussion

In two phase III trials, ECOG 4599 and AVAiL, bevacizumab significantly improved PFS when added to standard first-line chemotherapy in patients with advanced NSCLC [21, 22], but only ECOG 4599 showed a statistically significant benefit on OS [21]. This analysis pooled summary data from phase II and III trials to increase the number of events and patients, and hence to provide greater statistical power. This analysis has several strengths: it took into account all available data, including data from the statistical reports from trials, from a large population with advanced NSCLC; it also included an analysis of the interaction between patient characteristics and treatment effects, and corrected for duplication of the control arm.

According to this analysis, the use of bevacizumab (doses pooled) in combination with chemotherapy significantly prolonged both OS and PFS, compared with chemotherapy alone. The absolute benefit of bevacizumab addition on OS at 1 year was 4% (from 51% to 55%). There was no significant heterogeneity between the trials. The effect of bevacizumab on PFS was significantly (P = 0.007) greater in patients with stage IIIB than in those with non-recurrent stage IV NSCLC (P = 0.08 for OS). The use of bevacizumab in combination with radiotherapy in stage III patients appears toxic and inappropriate [32]. However, the use of bevacizumab in combination with chemotherapy for stage IIIB (defined as such by pleural or pericardial effusion or supra-clavicular involvement) is not associated with an increased risk of toxicity.

The effect on OS was greater in adenocarcinoma, compared with other histological types, while that on OS and PFS was greater in patients with a loss in body weight of ≤5%, compared with >5%. The observed results on histology concur with those of the analysis carried out on ECOG 4599 [33]. In both analyses, the number of patients with other histological types was small. This analysis evaluated the treatment effect according to the bevacizumab dose but the absence of a significant difference should be interpreted with caution, owing to insufficient statistical power. A limitation of OS as a primary end point in clinical trials of new therapies in the first-line treatment of NSCLC is the potential confounding effect of second- and third-line therapies administered upon disease progression, e.g. epidermal growth factor receptor tyrosine kinase inhibitors [34]. In the present analysis, almost 60% of patients in the control group in the AVF-0557g trial received cross-over therapy with bevacizumab. In ECOG 4599, approximately half of the patients in both the bevacizumab and control arms received second-line chemotherapy, and almost 20% received second-line therapy with a tyrosine kinase inhibitor. In AVAiL and JO19907, the proportion of patients who received subsequent chemotherapy was slightly higher in the control arms than in the bevacizumab arms (supplementary Table S1, available at Annals of Oncology online). The JO19907 trial had the highest rate of second-line treatments. In contrast to data from the ECOG 4599 trial [21], our analysis found no interaction between the patient's sex and the bevacizumab effect. Our analysis did not show any difference in the treatment effect of bevacizumab (OS or PFS) according to patient's age (<65 years versus ≥65 years). Data were not available to study the effect of treatment and its toxicity in patients ≥70 years of age to confirm the results of a subgroup analysis of the ECOG 4599 trial that showed higher toxic effects in the bevacizumab arm compared with the control arm, with no obvious improvement in survival [35]. However, data from the registry studies SAiL and ARIES do not suggest an increased toxicity of bevacizumab in the elderly population, nor a lower level of efficacy [25, 36, 37].

In all studies, bevacizumab was administered as monotherapy until disease progression or unacceptable toxicity in those patients who did not progress during the protocol of the six cycles of combination therapy [21, 22, 24, 25, 31]. However, no definitive evaluation of the usefulness of bevacizumab beyond chemotherapy can be made within the current meta-analysis. The use of concurrent plus maintenance bevacizumab (but not the addition of concurrent bevacizumab only) has significantly prolonged PFS when added to chemotherapy in the first-line treatment of ovarian, peritoneal or fallopian tube cancer [38].

The present analysis adds to other recent meta-analyses of studies of the efficacy and safety of bevacizumab in the treatment of patients with NSCLC [39–46]. Supplementary Table S3, available at Annals of Oncology online, summarises the results of the three meta-analyses regarding efficacy. Yang et al. [43] pooled summary data from four trials; they did not include data from the JO19907 trial [31], but they did include a phase II study of second-line treatment [47]. Yang et al. stratified their analysis according to bevacizumab dose. Unlike our analysis, these authors found no significant improvement in 1-year OS rate among patients treated with high- or low-dose bevacizumab, although high-dose bevacizumab did increase the 2-year OS rate. Yang et al. estimated the risk ratio of survival rate at a fixed time, whereas the calculation of HR (as used in our analysis) is recommended to reduce the risk of bias [48]. Botrel et al. meta-analysed summary level data from the same four studies included in our analysis, although around 10% of the randomly assigned patients were not included in their analysis (Figure 4 in their publication) [44]. These researchers found that the higher bevacizumab dose also significantly extended OS when a fixed-effects model was used (HR = 0.89; 95% CI 0.80–1.00; P = 0.04), but not when a random-effect method was used to account for apparent heterogeneity seen in the fixed-effects analysis. In our study, there was no heterogeneity within the high-dose group (P = 0.23, I² = 30%). Furthermore, the intent-to-treat analysis led to a decrease in heterogeneity. The sensitivity analyses (considering the HR of the AVF-0757g to be <1 and excluding the patients with squamous cell tumours in this trial) led to a higher treatment effect. Only the current study shows an overall effect on OS and PFS. Moreover, our analysis was corrected for the double counting of the control arm, which increased by 17% the number of patients. The three meta-analyses found similar results for the effect of bevacizumab on PFS.

According to the present analysis, bevacizumab significantly increased the risk of grade ≥3 proteinuria, hypertension, haemorrhagic events, neutropenia and febrile neutropenia. The toxicity profile of bevacizumab was consistent with previous reports and no unexpected patterns of toxicity were identified. Supplementary Table S4, available at Annals of Oncology online, summarises the results of the different meta-analyses regarding toxicity. The differences in methods and populations are important to consider when comparing these results. This analysis provides a more accurate estimation of the frequency of these toxic effects of bevacizumab than the Stony Brook University studies [39–42, 45, 46, 49] because its patient population was larger and more homogenous, as it included only patients receiving first-line treatment. However, the present analysis did not take dose effects into account, as the Botrel et al. [44] and Stony Brook University analyses did. The incidence of all grade ≥3 haemorrhagic events in patients treated with bevacizumab plus chemotherapy was 4.6% (58/1272), compared with 1.4% (12/857) with chemotherapy alone (OR 2.67; 95% CI 1.63, 4.39). This incidence is lower than that reported by Hapani et al. in NSCLC patients [39], but similar to that reported by Botrel et al. [44].

In common with other analyses [39–44, 49], our analysis is limited by its use of summary data rather than data from the individual patients from each trial [50–52]. However, an evaluation of the interaction between covariates and treatment effects was possible in our study. Individual patient data are needed to better account for the control arm in the three-arm trials, to standardise the analysis of the secondary end points, to include all trials in the different analyses, to perform an intent-to-treat analysis, to draw survival curves, to perform a more complete analysis of the variation of treatment effects according to patient characteristics and to perform sensitivity analysis after excluding the patients with squamous cell carcinoma. It was originally intended that this summary level analysis would be followed by an analysis of individual patient data from the eligible trials but, ultimately, this has not been possible.

In conclusion, this meta-analysis of randomised studies indicates that bevacizumab significantly prolonged OS and PFS when added to standard platinum-based chemotherapy as first-line therapy in patients with advanced NSCLC, with no unexpected toxicity patterns being evident.

funding

This work was supported by an unrestricted grant from Hoffmann-La Roche Ltd (Basel, Switzerland) and Ligue National Contre le Cancer (there were no grant numbers). The sponsors did not contribute to the analysis, interpretation and reporting of the data. On behalf of his co-authors, JPP had full access to all the study data, and had final responsibility for the decision to submit for publication.

disclosures

JCS has acted as a consultant for Abbott, Amgen, AstraZeneca, Bristol-Myers Squibb, Boehringer-Ingelheim, glaxosmithkiline, Lilly, Merck-Serono, Merck Sharp & Dohme, Pfizer, Roche-Genentech, Servier and sanofi-aventis. MR has acted as a consultant (compensated) for F. Hoffmann-La Roche, Lilly, Pfizer, AstraZeneca, Daiichi Sankyo and BMS and has accepted honoraria for academic presentations from F. Hoffmann-La Roche, Lilly, Daiichi Sankyo and AstraZeneca. ABS has acted as an advisory board member, consultant and speakers bureau member, has undertaken legal consulting and has received honoraria from Genentech in the last 12 months. MF has received honoraria from Chugai Pharmaceutical Co., Ltd. BB has received travel and research grants from F. Hoffmann-La Roche. JPP received an unrestricted grant from F. Hoffmann-La Roche for the present project and declares no other conflicts of interest. AM, NS, DHJ and DB declare no conflicts of interest. JPP acted as a co-secretariat of the group, co-prepared the protocol, discussed discrepancies in the extraction of the data, led the analysis, reviewed the statistical report, led the drafting of the manuscript and approved the final version. JCS acted as a co-secretariat of the group, reviewed the protocol, reviewed the manuscript at each stage and approved the final version. AM co-prepared the protocol, extracted the data from the literature/reports, carried out the analyses and prepared the statistical report, drafted the manuscript and approved the final version. BB reviewed the protocol, reviewed the manuscript at each stage and approved the final version. DB helped in the preparation of the protocol, extracted the data from the literature/reports, reviewed the manuscript and approved the final version. MR, ABS, NS, DHJ and MF acted as a representative of an analysed trial, provided additional data, reviewed the manuscript and approved the final version. These data have previously been presented in abstract and poster form [53].

acknowledgements

This analysis and publication were supported by an unrestricted grant from F. Hoffmann-La Roche Ltd (Basel, Switzerland) and Ligue National Contre le Cancer. We are grateful to Chugai Pharmaceutical Co., Ltd for providing supplementary data on the JO19907 trial. The authors acknowledge the medical writing assistance of Lee Baker of Prism Ideas, Cheshire, UK, supported by F. Hoffman-La Roche. JPP acted as a co-secretariat of the group, co-prepared the protocol, discussed discrepancies in the extraction of the data, led the analysis, reviewed the statistical report, led the drafting of the manuscript and approved the final version. JCS acted as a co-secretariat of the group, reviewed the protocol, reviewed the manuscript at each stage and approved the final version. AM co-prepared the protocol, extracted the data from the literature/reports, carried out the analyses and prepared the statistical report, drafted the manuscript and approved the final version. BB reviewed the protocol, reviewed the manuscript at each stage and approved the final version. DB helped in the preparation of the protocol, extracted the data from the literature/reports, reviewed the manuscript and approved the final version. MR, ABS, NS, DHJ and MF acted as a representative of an analysed trial, provided additional data, reviewed the manuscript and approved the final version.

references

1
American Cancer Society
Global Cancer Facts & Figures
 , 
2011
2nd Edition
Atlanta
American Cancer Society
2
D'Addario
G
Früh
M
Reck
M
, et al.  . 
ESMO Guidelines Working Group
Metastatic non-small-cell lung cancer: ESMO clinical practice guidelines for diagnosis, treatment and follow-up
Ann Oncol
 , 
2010
, vol. 
21
 
Suppl 5
(pg. 
v116
-
v119
)
3
Delbaldo
C
Michiels
S
Syz
N
, et al.  . 
Benefits of adding a drug to a single-agent or a 2-agent chemotherapy regimen in advanced non-small cell lung cancer. A meta-analysis
JAMA
 , 
2004
, vol. 
292
 (pg. 
470
-
484
)
4
Hotta
K
Fujiwara
Y
Matsuo
K
, et al.  . 
Recent improvement in the survival of patients with advanced nonsmall cell lung cancer enrolled in phase III trials of first-line, systemic chemotherapy
Cancer
 , 
2007
, vol. 
109
 (pg. 
939
-
948
)
5
Jiang
J
Liang
X
Zhou
X
, et al.  . 
A meta-analysis of randomized controlled trials comparing carboplatin-based to cisplatin-based chemotherapy in advanced non-small cell lung cancer
Lung Cancer
 , 
2007
, vol. 
57
 (pg. 
348
-
358
)
6
NSCLC Meta-Analyses Collaborative Group
Chemotherapy in addition to supportive care improves survival in advanced non-small-cell lung cancer: a systematic review and meta-analysis of individual patient data from 16 randomized controlled trials
J Clin Oncol
 , 
2008
, vol. 
26
 (pg. 
4617
-
4625
)
7
Soon
YY
Stockler
MR
Askie
LM
, et al.  . 
Duration of chemotherapy for advanced non-small-cell lung cancer: a systematic review and meta-analysis of randomized trials
J Clin Oncol
 , 
2009
, vol. 
27
 (pg. 
3277
-
3283
)
8
Goffin
J
Lacchetti
C
Ellis
PM
, et al.  . 
Lung Cancer Disease Site Group of Cancer Care Ontario's Program in Evidence-Based Care
First-line systemic chemotherapy in the treatment of advanced non-small cell lung cancer: a systematic review
J Thorac Oncol
 , 
2010
, vol. 
5
 (pg. 
260
-
274
)
9
Ferrara
N
Vascular endothelial growth factor: basic science and clinical progress
Endocr Rev
 , 
2004
, vol. 
25
 (pg. 
581
-
611
)
10
Kerbel
RS
Tumor angiogenesis
N Engl J Med
 , 
2008
, vol. 
358
 (pg. 
2039
-
2049
)
11
Zhan
P
Wang
J
Lv
XJ
, et al.  . 
Prognostic value of vascular endothelial growth factor expression in patients with lung cancer: a systematic review with meta-analysis
J Thorac Oncol
 , 
2009
, vol. 
4
 (pg. 
1094
-
1103
)
12
Ferrara
N
Hillan
KJ
Gerber
HP
, et al.  . 
Discovery and development of bevacizumab, an anti-VEGF antibody for treating cancer
Nat Rev Drug Discov
 , 
2004
, vol. 
3
 (pg. 
391
-
400
)
13
Hurwitz
H
Fehrenbacher
L
Novotny
W
, et al.  . 
Bevacizumab plus irinotecan, fluorouracil, and leucovorin for metastatic colorectal cancer
N Engl J Med
 , 
2004
, vol. 
350
 (pg. 
2335
-
2342
)
14
Giantonio
BJ
Catalano
PJ
Meropol
NJ
, et al.  . 
Bevacizumab in combination with oxaliplatin, fluorouracil, and leucovorin (FOLFOX4) for previously treated metastatic colorectal cancer: results from the Eastern Cooperative Oncology Group Study E3200
J Clin Oncol
 , 
2007
, vol. 
25
 (pg. 
1539
-
1544
)
15
Miller
K
Wang
M
Gralow
J
, et al.  . 
Paclitaxel plus bevacizumab versus paclitaxel alone for metastatic breast cancer
N Engl J Med
 , 
2007
, vol. 
357
 (pg. 
2666
-
2676
)
16
Miles
DW
Chan
A
Dirix
LY
, et al.  . 
Phase III study of bevacizumab plus docetaxel compared with placebo plus docetaxel for the first-line treatment of human epidermal growth factor receptor 2-negative metastatic breast cancer
J Clin Oncol
 , 
2010
, vol. 
28
 (pg. 
3239
-
3247
)
17
Rini
BI
Halabi
S
Rosenberg
JE
, et al.  . 
Bevacizumab plus interferon alfa compared with interferon alfa monotherapy in patients with metastatic renal cell carcinoma: CALGB 90206
J Clin Oncol
 , 
2008
, vol. 
26
 (pg. 
5422
-
5428
)
18
Escudier
B
Bellmunt
J
Négrier
S
, et al.  . 
Phase III trial of bevacizumab plus interferon alfa-2a in patients with metastatic renal cell carcinoma (AVOREN): final analysis of overall survival
J Clin Oncol
 , 
2010
, vol. 
28
 (pg. 
2144
-
2150
)
19
Rini
BI
Halabi
S
Rosenberg
JE
, et al.  . 
Phase III trial of bevacizumab plus interferon alfa versus interferon alfa monotherapy in patients with metastatic renal cell carcinoma: final results of CALGB 90206
J Clin Oncol
 , 
2010
, vol. 
28
 (pg. 
2137
-
2143
)
20
Avastin Summary of Product Characteristics
2010
 
Hoffmann-La Roche Ltd
21
Sandler
A
Gray
R
Perry
MC
, et al.  . 
Paclitaxel-carboplatin alone or with bevacizumab for non-small-cell lung cancer
N Engl J Med
 , 
2006
, vol. 
355
 (pg. 
2542
-
2550
)
22
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
 , 
2009
, vol. 
27
 (pg. 
1227
-
1234
)
23
Reck
M
von Pawel
J
Zatloukal
P
, et al.  . 
BO17704 Study Group
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
 , 
2010
, vol. 
21
 (pg. 
1804
-
1809
)
24
Johnson
DH
Fehrenbacher
L
Novotny
WF
, et al.  . 
Randomized phase II trial comparing bevacizumab plus carboplatin and paclitaxel with carboplatin and paclitaxel alone in previously untreated locally advanced or metastatic non-small-cell lung cancer
J Clin Oncol
 , 
2004
, vol. 
22
 (pg. 
2184
-
2191
)
25
Crinò
L
Dansin
E
Garrido
P
, et al.  . 
Safety and efficacy of first-line bevacizumab-based therapy in advanced non-squamous non-small-cell lung cancer (SAiL, MO19390): a phase 4 study
Lancet Oncol
 , 
2010
, vol. 
11
 (pg. 
733
-
740
)
26
Sandler
AB
Schiller
JH
Gray
R
, et al.  . 
Retrospective evaluation of the clinical and radiographic risk factors associated with severe pulmonary hemorrhage in first-line advanced, unresectable non-small-cell lung cancer treated with carboplatin and paclitaxel plus bevacizumab
J Clin Oncol
 , 
2009
, vol. 
27
 (pg. 
1405
-
1412
)
27
Spigel
DR
Brahmer
JR
Fischbach
NA
, et al.  . 
Risk of severe pulmonary hemorrhage (SPH) associated with development of cavitation while on bevacizumab (BV) therapy in patients (PTS) with non-small cell lung cancer (NSCLC): results from ARIES, a BV treatment observational cohort study (OCS)
J Thorac Oncol
 , 
2010
, vol. 
5
 pg. 
S84
 
28
Begg
CB
Berlin
JA
Publication bias and dissemination of clinical research
J Natl Cancer Inst
 , 
1989
, vol. 
81
 (pg. 
107
-
115
)
29
Pignon
JP
Tribodet
H
Scagliotti
GV
, et al.  . 
LACE Collaborative Group
Lung adjuvant cisplatin evaluation: a pooled analysis by the LACE Collaborative Group
J Clin Oncol
 , 
2008
, vol. 
26
 (pg. 
3552
-
3559
)
30
Higgins
JP
Thompson
SG
Quantifying heterogeneity in a meta-analysis
Stat Med
 , 
2002
, vol. 
21
 (pg. 
1539
-
1558
)
31
Niho
S
Kunitoh
H
Nokihara
H
, et al.  . 
Randomized phase II study of first-line carboplatin-paclitaxel with or without bevacizumab in Japanese patients with advanced non-squamous non-small-cell lung cancer
Lung Cancer
 , 
2012
, vol. 
76
 (pg. 
362
-
367
)
32
Lind
JS
Senan
S
Smit
EF
Pulmonary toxicity after bevacizumab and concurrent thoracic radiotherapy observed in a phase I study for inoperable stage III non-small-cell lung cancer
J Clin Oncol
 , 
2012
, vol. 
30
 (pg. 
e104
-
e108
)
33
Sandler
A
Yi
J
Dahlberg
S
, et al.  . 
Treatment outcomes by tumor histology in Eastern Cooperative Group Study E4599 of bevacizumab with paclitaxel/carboplatin for advanced non-small cell lung cancer
J Thorac Oncol
 , 
2010
, vol. 
5
 (pg. 
1416
-
1423
)
34
Soria
JC
Massard
C
Le
CT
Should progression-free survival be the primary measure of efficacy for advanced NSCLC therapy?
Ann Oncol
 , 
2010a
, vol. 
21
 (pg. 
2324
-
2332
)
35
Ramalingam
SS
Dahlberg
SE
Langer
CJ
, et al.  . 
Outcomes for elderly, advanced-stage non-small-cell lung cancer patients treated with bevacizumab in combination with carboplatin and paclitaxel: analysis of Eastern Cooperative Oncology Group trial 4599
J Clin Oncol.
 , 
2008
, vol. 
26
 (pg. 
60
-
65
)
36
Laskin
J
Crinò
L
Felip
E
, et al.  . 
Safety and efficacy of first-line bevacizumab plus chemotherapy in elderly patients with advanced or recurrent nonsquamous non-small cell lung cancer: safety of avastin in lung trial (MO19390)
J Thorac Oncol
 , 
2012
, vol. 
7
 (pg. 
203
-
211
)
37
Wozniak
AJ
Garst
J
Jahanzeb
M
, et al.  . 
and for the ARIES Investigators
Clinical outcomes (CO) for special populations of patients (pts) with advanced non-small cell lung cancer (NSCLC): results from ARIES, a bevacizumab (BV) observational cohort study (OCS)
J Clin Oncol
 , 
2010
, vol. 
28
 
15 suppl.
pg. 
7618
 
38
Burger
RA
Brady
MF
Bookman
MA
, et al.  . 
Incorporation of bevacizumab in the primary treatment of ovarian cancer
N Engl J Med.
 , 
2011
, vol. 
365
 (pg. 
2473
-
2483
)
39
Hapani
S
Chu
D
Wu
S
Risk of gastrointestinal perforation in patients with cancer treated with bevacizumab: a meta-analysis
Lancet Oncol.
 , 
2009
, vol. 
10
 (pg. 
559
-
568
)
40
Hapani
S
Sher
A
Chu
D
, et al.  . 
Increased risk of serious hemorrhage with bevacizumab in cancer patients: a meta-analysis
Oncology
 , 
2010
, vol. 
79
 (pg. 
27
-
38
)
41
Ranpura
V
Pulipati
B
Chu
D
, et al.  . 
Increased risk of high-grade hypertension with bevacizumab in cancer patients: a meta-analysis
Am J Hypertens
 , 
2010
, vol. 
23
 (pg. 
460
-
468
)
42
Wu
S
Kim
C
Baer
L
, et al.  . 
Bevacizumab increases risk for severe proteinuria in cancer patients
J Am Soc Nephrol
 , 
2010
, vol. 
21
 (pg. 
1381
-
1389
)
43
Yang
K
Wang
YJ
Chen
XR
, et al.  . 
Effectiveness and safety of bevacizumab for unresectable non-small-cell lung cancer: a meta-analysis
Clin Drug Investig
 , 
2010
, vol. 
30
 (pg. 
229
-
241
)
44
Botrel
TE
Clark
O
Clark
L
, et al.  . 
Efficacy of bevacizumab (Bev) plus chemotherapy (CT) compared to CT alone in previously untreated locally advanced or metastatic non-small cell lung cancer (NSCLC): systematic review and meta-analysis
Lung Cancer.
 , 
2011
, vol. 
74
 (pg. 
89
-
97
)
45
Ranpura
V
Hapani
S
Wu
S
Treatment-related mortality with bevacizumab in cancer patients: a meta-analysis
JAMA
 , 
2011
, vol. 
305
 (pg. 
487
-
494
Erratum in: JAMA 2011; 305: 2294
46
Sher
A
Wu
S
Anti-vascular endothelial growth factor antibody bevacizumab reduced the risk of anemia associated with chemotherapy-A meta-analysis
Acta Oncol.
 , 
2011
, vol. 
50
 (pg. 
997
-
1005
)
47
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
 , 
2007
, vol. 
25
 (pg. 
4743
-
4750
)
48
Michiels
S
Piedbois
P
Burdett
S
, et al.  . 
Meta-Analysis when only the median survival times are known: a comparison with individual patient data results
Int J Technol Assess Health Care
 , 
2005
, vol. 
21
 (pg. 
119
-
125
)
49
Nalluri
SR
Chu
D
Keresztes
R
, et al.  . 
Risk of venous thromboembolism with the angiogenesis inhibitor bevacizumab in cancer patients: a meta-analysis
JAMA
 , 
2008
, vol. 
300
 (pg. 
2277
-
2285
)
50
Stewart
LA
Parmar
MK
Meta-analysis of the literature or of individual patient data: is there a difference?
Lancet
 , 
1993
, vol. 
341
 (pg. 
418
-
422
)
51
Pignon
JP
Bourhis
J
Meta-analysis of chemotherapy in head and neck cancer: individual patient data vs literature data
Br J Cancer
 , 
1995
, vol. 
72
 (pg. 
1062
-
1063
)
52
Riley
RD
Lambert
PC
Abo-Zaid
G
Meta-analysis of individual participant data: rationale, conduct, and reporting
BMJ
 , 
2010
, vol. 
340
 pg. 
c221
 
53
Soria
J
Mauguen
A
Reck
M
, et al.  . 
Meta-analysis of randomized phase II/III trials adding bevacizumab to platin-based chemotherapy as 1st-line treatment in patients with advanced non-small cell lung cancer (NSCLC). 35th ESMO Congress Milan, Italy October 2010
Ann Oncol
 , 
2010
, vol. 
21
 
suppl 8
pg. 
viii147