Bevacizumab requires some unique eligibility criteria, such as absence of hemoptysis and major blood vessel invasion by the tumor. The prognostic impact of these bevacizumab-specific criteria has not been evaluated.
Patients with stage IIIB/IV, non-squamous non-small cell lung cancer who started chemotherapy before the approval of bevacizumab were reviewed. Patients with impaired organ function, poor performance status or untreated/symptomatic brain metastasis were excluded before the evaluation of bevacizumab eligibility. We compared overall survival and time to treatment failure among patients who were eligible (Group A) or ineligible (Group B) to receive bevacizumab.
Among 283 patients with stage IIIB/IV non-squamous non-small cell lung cancer, eligibility for bevacizumab was evaluated in 154 patients. Fifty-seven patients were considered ineligible (Group B) based on one or more of a history of hemoptysis (n = 20), major blood vessel invasion (n = 43) and cardiovascular disease (n = 8). The remaining 97 patients were classified into Group A. Overall survival was significantly better in Group A (median, 14.6 months) than in Group B (median, 7.1 months; p<0.0001). Time to treatment failure was also significantly longer in Group A (median, 6.9 months) than in Group B (median, 3.0 months; p<0.0001). Adjusted hazard ratios of bevacizumab eligibility for overall survival and time to treatment failure were 0.48 and 0.38 (95% confidence intervals, 0.33–0.70 and 0.25–0.58), respectively.
Eligibility for bevacizumab itself represents a powerful prognostic factor for patients with non-squamous non-small cell lung cancer. The proportion of patients who underwent first-line chemotherapy without disease progression or unacceptable toxicity can also be biased by bevacizumab eligibility. Selection bias can be large in clinical trials of bevacizumab, so findings from such trials should be interpreted with extreme caution.
Citation: Takagi Y, Toriihara A, Nakahara Y, Yomota M, Okuma Y, Hosomi Y, et al. (2013) Eligibility for Bevacizumab as an Independent Prognostic Factor for Patients with Advanced Non-Squamous Non-Small Cell Lung Cancer: A Retrospective Cohort Study. PLoS ONE 8(3): e59700. https://doi.org/10.1371/journal.pone.0059700
Editor: Ramon Andrade de Mello, University of Porto, Portugal
Received: December 18, 2012; Accepted: February 17, 2013; Published: March 26, 2013
Copyright: © 2013 Takagi, et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: The authors have no support or funding to report.
Competing interests: The authors have declared that no competing interests exist.
Eligibility is often narrowed in clinical trials of targeted drugs because of specific adverse effects . This is intended to exclude patients who might be at high risk of developing severe adverse events and to maximize the overall efficacy of the drug of interest. As a result, modified eligibility criteria can affect endpoints such as overall survival (OS) independently of the actual effect of an investigational drug.
Bevacizumab (BV), an anti-vascular endothelial growth factor antibody, requires modified eligibility criteria such as absence of hemoptysis and major blood vessel invasion (MVI) in clinical trials –. Some studies have indicated that patients who meet the eligibility criteria for BV are in the minority in the real world , but the impacts of these criteria on survival and treatment efficacy have not been evaluated. Understanding the potential selection bias derived from BV-specific eligibility criteria is important for clinicians, so that the results of key clinical trials can be interpreted appropriately.
We investigated whether the eligibility criteria characteristically applied for BV lead to selection bias. This retrospective cohort study examined the relationship between eligibility for BV and prognosis among patients with non-squamous non-small cell lung cancer (NSCLC), by enrolling patients who started chemotherapy before BV gained approval for use in Japan.
This study was approved by the institutional review board of Tokyo Metropolitan Cancer and Infectious Diseases Center Komagome Hospital (Tokyo, Japan). We used routinely collected data and anonymized data for all analyses, and individual patient consent was not required. The waiver of need for written informed consent was also approved by the institutional review board of Tokyo Metropolitan Cancer and Infectious Diseases Center.
Patients were identified from the database at Tokyo Metropolitan Cancer and Infectious Diseases Center and included those who had undergone systemic chemotherapy for the treatment of lung cancer at the Department of Thoracic Oncology and Respiratory Medicine.
Patients with stage IIIB/IV non-squamous NSCLC who started chemotherapy between 2005 and 2009 were reviewed. After receiving approval as a therapeutic drug for treating lung cancer in Japan in November 2009, BV was first applied to treat lung cancer at our institution in 2010. Lung cancer was staged according to the 7th edition of the TNM Classification of Malignant Tumors  by the International Union Against Cancer (UICC). Tumors with mixed histological subtypes of NSCLC were categorized into a subtype according to the predominant component.
We excluded patients with indications for combined chemoradiotherapy, Eastern Cooperative Oncology Group (ECOG) performance status (PS) 3 or 4, untreated or symptomatic brain metastasis or impaired bone marrow, hepatic or renal function at the start of chemotherapy, because these patients are excluded from most clinical trials of first-line chemotherapy for lung cancer , . Patients with PS 2 were included because the AVAPERL study did not exclude these patients . We included patients without information about PS in survival analyses, because patients with PS 3/4 rarely start chemotherapy without this being specifically mentioned in the medical records.
Evaluation of eligibility for bevacizumab
Patients were considered ineligible for BV if they had one or more of a history of hemoptysis, MVI by the tumor and clinically significant cardiovascular disease (CVD).
A history of hemoptysis was defined as an episode of hemoptysis within 3 months prior to starting chemotherapy. Since investigating the volume of expectorated blood is difficult, all episodes of hemoptysis were included, irrespective of severity. A radiologist who was blinded to clinical outcomes evaluated MVI. Major blood vessels included the aorta, superior vena cava, inferior vena cava, main pulmonary arteries and main branches of the pulmonary arteries and pulmonary veins within the pericardial sac. Invasion was defined as contact of >180°with these vessels  or an irregular bump in the vessels. We defined CVD as chronic congestive heart failure, ischemic heart disease or vascular disease (including thrombotic events) requiring medication.
If any one of these three elements was identified, the patient was considered ineligible for BV (Group B). When all elements were negative, the patient was assigned to Group A. If data about any elements were unobtainable and all other elements were negative, eligibility could not be determined and the patient was excluded from survival analysis.
The primary endpoint was OS, defined as the number of months between starting the first chemotherapy regimen until the date of death. Patients alive at the end of follow-up were censored, except for those with disease progression who were unwilling to undergo further anticancer therapy. To diminish the influence of informative censoring, the last follow-up was regarded as an event in these patients (for example, lost to follow-up after transfer to hospice).
Time to treatment failure
We defined the time to treatment failure (TTF) of patients who underwent first-line chemotherapy containing platinum as the number of months that elapsed between the initiation of chemotherapy until the date of starting any type of subsequent therapy, request from the patient to terminate anticancer therapy, or death. Follow-up duration was defined as the time elapsed from initiating chemotherapy until the date of death or last follow-up. Subsequent therapy included chemotherapy, radiotherapy and tumor resection. Patients who remained alive without subsequent therapy at the end of follow-up were censored.
We collected information about age, sex, histological type of NSCLC, stage of lung cancer, laboratory findings and PS at the time of chemotherapy initiation from medical records. Tumor histology categories were defined as adenocarcinoma, NSCLC not otherwise specified (NOS), large cell neuroendocrine carcinoma (LCNEC) and others. We also surveyed the first chemotherapeutic regimen and subsequent therapies.
Sex, disease stage (IIIB or IV), PS (0, 1 or 2) and platinum as first-line chemotherapy were included as explanatory variables, because these factors are known to affect the survival of patients with NSCLC , . Whether platinum is beneficial for elderly patients remains unclear , so we also included age (<70 vs. ≥70 years) among the variables. A history of hemoptysis, MVI and CVD were also included among the variables. Number of metastatic sites, weight loss and laboratory markers were not included because the relevance of the prognostic impact of these markers has not appeared constant through a number of studies –. Patients with insufficient information about any variable were excluded from multivariate analysis. Since epidermal growth factor receptor (EGFR) mutation status was known for only a small number of patients, EGFR mutations were not included as variables in this analysis.
We designed the study assuming that median OS of the BV-eligible cohort would be 1.5 years with a hazard ratio of 0.5, based on previous findings , . To detect a difference in OS with 90% power for a two-sided significance value of 0.05, 121 patients were required. Our database accumulates 100–120 patients with lung cancer annually, with about 50% having presumed advanced non-squamous NSCLC, and almost half meeting the eligibility criteria of ordinary clinical trials. We therefore analyzed data from 2005 to 2009 to accumulate sufficient subjects.
Differences in characteristics except for age between Groups A and B were evaluated using the χ2 test. Differences in age were compared using Student's t-test. Survival and TTF were estimated using the Kaplan-Meier method and the log-rank test was used for inter-group comparisons. We also examined the prognostic impact of the variables described above on OS and TTF using Cox proportional hazards modeling. All tests were two-sided with a significance level of 0.05. All data were analyzed using JMP version 9.0 software (SAS Institute, Cary, NC, USA).
Among 576 patients with lung cancer treated at our hospital between 2005 and 2009, a total of 283 had stage IIIB/IV non-squamous NSCLC. BV was not administered to any of these patients throughout the course of treatment. After excluding 129 patients showing indications for concurrent chemoradiotherapy, PS 3 or 4, impaired organ function, untreated or symptomatic brain metastasis or insufficient information, we evaluated the eligibility of the remaining 154 patients for BV. Fifty-seven patients were considered ineligible for BV (Group B), based on one or more of a history of hemoptysis (n = 20), MVI (n = 43) or CVD (n = 8). The remaining 97 patients were assigned to Group A (Figure 1).
BM, brain metastasis; BV, bevacizumab; CRT, chemoradiotherapy; NSCLC, non-small cell lung cancer.
The median age of Groups A and B combined (n = 154) was 67 years (range, 41–84 years), and 54 patients (35%) were female. Most patients had adenocarcinoma (84%) and stage IV disease (92%). The mutation status of EGFR was determined in 39 (25%) patients. The proportions of patients with EGFR mutations among those for whom EGFR status was determined did not differ significantly between groups, at 10 of 30 (33%) in Group A and 4 of 9 (44%) in Group B.
Baseline characteristics of both groups were similar, although the proportion of patients with PS 1 was rather higher in Group B (Table 1). Proportions of PS 2 patients among Groups A and B were almost equal. Platinum-based chemotherapy was administered to 77 patients (79%) in Group A and 44 patients (77%) in Group B, showing no significant difference between groups.
Median OS was significantly better in Group A (14.6 months) than in Group B (7.1 months; p<0.0001) (Figure 2). The crude hazard ratio of BV eligibility for OS was 0.50 (95% confidence interval (CI), 0.36–0.72). One-year survival rates for Groups A and B were 62% and 28%, respectively. Differences in OS between groups remained even after censoring patients with disease progression and those who declined further therapy at the end of follow-up (median OS, 16.8 and 7.8 months, respectively; p = 0.0001). Median OS was significantly longer in Group A than in Group B among 121 patients who had received first-line chemotherapy with platinum (18.8 vs. 9.2 months; p = 0.0006).
Kaplan-Meier curves for overall survival (A) and time to treatment failure (B). BV, bevacizumab; TTF, time to treatment failure.
Multivariate analysis indicated PS 2, use of platinum, history of hemoptysis and MVI as significant prognostic factors (Table 2). The adjusted hazard ratio of BV eligibility for other variables was 0.48 (95%CI, 0.33–0.70, p = 0.0001), indicating that BV eligibility itself represents an independent prognostic factor for patients with advanced non-squamous NSCLC. Prognostic impact of BV eligibility consists of MVI and history of hemoptysis, with CVD exerting no influence (Figure 3).
Kaplan-Meier curves for overall survival by MVI (A), history of hemoptysis (B) and CVD (C) and time to treatment failure by MVI (D), history of hemoptysis (E) and CVD (F). CVD, cardiovascular disease; Hemop, history of hemoptysis; MVI, major blood vessel invasion; TTF, time to treatment failure.
Among 61 patients younger than 65 years, OS was also significantly better in Group A (14.8 months) than in Group B (7.8 months; p = 0.003). Median OS was 26.1 months for patients with mutated EGFR and 16.0 months for patients with wild-type EGFR. This difference was not significant, because the EGFR status of most patients was unknown. Median OS was 16.0 months in Group A and 9.2 months in Group B, respectively, among the 25 patients with wild-type EGFR, and 22.1 months and 35.4 months, respectively, in the 14 patients with mutated EGFR. These preliminary findings did not reflect a significant difference, probably due to the small size of each subgroup.
Time to treatment failure
The median duration of follow-up among the 121 patients who underwent platinum-based first-line chemotherapy was 12.6 months. TTF was significantly better in Group A (6.9 months) than in Group B (3.0 months; p<0.0001) (Figure 2). The crude hazard ratio of BV eligibility to TTF was 0.39 (95%CI, 0.25–0.58). The proportions of Groups A and B with a TTF >3 months were 82% and 50%, respectively. Multivariate analysis indicated that stage IV disease, history of hemoptysis, MVI and CVD are associated with shorter TTF (Table 3). The adjusted hazard ratio of BV eligibility to other baseline variables was 0.38 (95%CI, 0.25–0.58; p<0.0001). More patients underwent consequent chemotherapy in Group A (74%) than in Group B (50%, p = 0.008), and the median number of chemotherapy regimens was three in group A and two in group B (p = 0.003).
This study showed that eligibility for BV itself represents a significant prognostic factor for patients with non-squamous NSCLC. Patients unsuitable for participating in ordinary clinical trials were excluded ahead of the evaluation of BV eligibility, therefore our findings suggest that eligibility for BV lead to substantial selection bias for OS in clinical trials of NSCLC. Multivariate analysis also revealed that two of the three factors that define eligibility for BV strongly impacted OS. Although more patients in Group B tended to have PS 1, OS did not differ significantly between patients with PS 0 and 1. The difference in baseline PS between groups thus had minimal effect on the present results.
We also found that BV eligibility is associated with a longer TTF. Patients with a TTF longer than a certain time period indicate individuals who underwent chemotherapy without disease progression or unacceptable toxicity. When maintenance therapy was preplanned, longer TTF results in higher proportions of patients who actually receive maintenance therapy. Progression-free survival (PFS) is generally favorable for evaluating the impact on treatment efficacy, since TTF is affected not only by efficacy, but also by toxic effects. However, we did not use PFS in this retrospective study because intervals between radiographic evaluations varied considerably among patients, which would compromise the accuracy of the results . Our results support the notion that BV eligibility confers a positive impact upon chemotherapeutic outcomes, but this speculation requires validation.
As far as we can determine, this is the first study to directly compare the prognosis of patients with advanced non-squamous NSCLC in terms of eligibility for BV. Patients participating in clinical trials generally have relatively better prognosis because of strict criteria that exclude patients showing established risk factors for early death, but the prognostic impact of BV-specific eligibility criteria among these selected patients was unknown. Our findings that BV-specific eligibility leads to better prognosis may explain why the OS of the control arm (carboplatin-paclitaxel) in ECOG 4599  was relatively longer than that in other contemporary trials . Similarly, a randomized phase II trial of the same design as ECOG 4599 in Japan found that median OS of the control arm was 23.4 months , considerably longer than the median 13.3 months for the carboplatin-paclitaxel arm in a Japanese phase III trial . The concern about selection bias derived from BV eligibility has also been raised in studies of patients with colorectal cancer . These and the present findings indicate that imposing eligibility criteria with the intention of avoiding severe adverse events also helps to improve survival.
The key limitation of this retrospective study is that the findings were derived from data generated at a single institution. We attempted to minimize bias by establishing objective criteria to define eligibility for BV based on large, completed phase III trials , , and having a blinded radiologist assess MVI. Taking the large magnitude of difference in OS and TTF between the two groups into account, the probability of this study is considerable. Second, some of the eligibility criteria for BV applied herein are no longer being applied in some ongoing phase III studies . However, whether patients who were ineligible for BV based on previous pervasive criteria  and then eligible under the new criteria will necessarily enter into current trials is questionable. Although the exclusion criteria for BV were relaxed, resolving the prognostic impact of eligibility for BV might take some time. Third, we did not fully evaluate the impact of EGFR mutations in this study. Mutated EGFR is associated with longer OS in the era of targeted therapies , . The proportions of patients with determined EGFR mutations were similar between groups, but the actual relationship between BV eligibility and EGFR mutation status has yet to be clarified.
ECOG 4599 demonstrated a significant survival advantage by adding BV to carboplatin-paclitaxel . However, whether adding BV to cisplatin-based chemotherapy also prolongs survival remains controversial . A recent large observational study of over 4,000 patients found no significant benefit of adding BV to carboplatin-paclitaxel in patients ≥65 years old . Suitable patient subgroups or concomitant chemotherapies that will maximize the benefit of BV should be identified, since addition of BV is associated with a higher incidence of treatment-related death . However, few studies are designed for randomization based on the presence or absence of BV. Clinicians must therefore design therapeutic strategies by extrapolating the outcomes of trials conducted under various settings. This study showed that selection bias can be substantial in clinical trials of BV. As a result, relatively better OS and success rate of induction chemotherapy observed in trials not based on randomization in terms of administration of BV are not sufficient grounds for using BV with cisplatin-based standard chemotherapy.
In conclusion, eligibility for BV represents a powerful prognostic factor for patients with non-squamous NSCLC, regardless of BV use. The impact of selection bias should be carefully considered when interpreting the results of trials with modified eligibility criteria. Further study is warranted to validate our findings.
Drafting of the manuscript: YT. Critical revision of the manuscript for important intellectual content: AT YN MY YO YH MS TO. Study supervision: MS TO. Conceived and designed the experiments: YT AT YH. Performed the experiments: YT AT YN MY YO. Analyzed the data: YT MS TO. Contributed reagents/materials/analysis tools: AT. Wrote the paper: YT AT YN MY YO YH MS TO.
- 1. Johnson DH, Fehrenbacher L, Novotny WF, Herbst RS, Nemunaitis JJ, et al. (2004) 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 22: 2184–2191.
- 2. Sandler A, Gray R, Perry MC, Brahmer J, Schiller JH, et al. (2006) Paclitaxel-carboplatin alone or with bevacizumab for non-small-cell lung cancer. N Engl J Med 355: 2542–2550.
- 3. Reck M, von Pawel J, Zatloukal P, Ramlau R, Gorbounova V, et al. (2010) 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: 1804–1809.
- 4. Barlesi F, de Castro J, Dvornichenko V, Kim JH, Pazzola A, et al.. (2011) AVAPERL (MO22089): Final efficacy outcomes for patients (pts) with advanced non-squamous non-small cell lung cancer (nsNSCLC) randomised to continuation maintenance (mtc) with bevacizumab (bev) or bev+pemetrexed(pem) after first-line (1L) bev-cisplatin (cis)-pem treatment (Tx). Eur J Cancer (suppl 2): 16.
- 5. Clarey J, Kao SC, Clarke SJ, Vardy J (2012) The eligibility of advanced non-small-cell lung cancer patients for targeted therapy clinical trials. Ann Oncol 23: 1229–1233.
- 6. Detterbeck FC, Boffa DJ, Tanoue LT (2009) The new lung cancer staging system. Chest 136: 260–271.
- 7. Scagliotti GV, Parikh P, von Pawel J, Biesma B, Vansteenkiste J, et al. (2008) Phase III study comparing cisplatin plus gemcitabine with cisplatin plus pemetrexed in chemotherapy-naive patients with advanced-stage non-small-cell lung cancer. J Clin Oncol 26: 3543–3551.
- 8. Paz-Ares L, de Marinis F, Dediu M, Thomas M, Pujol JL, et al. (2012) 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: 247–255.
- 9. Reck M, Barlesi F, Crinò L, Henschke CI, Isla D, et al. (2012) Predicting and managing the risk of pulmonary haemorrhage in patients with NSCLC treated with bevacizumab: a consensus report from a panel of experts. Ann Oncol 23: 1111–1120.
- 10. Brundage MD, Davies D, Mackillop WJ (2002) Prognostic factors in non-small cell lung cancer: a decade of progress. Chest 122: 1037–1057.
- 11. Non-small Cell Lung Cancer Collaborative Group (1995) Chemotherapy in non-small cell lung cancer: a meta-analysis using updated data on individual patients from 52 randomised clinical trials. BMJ 311: 899–909.
- 12. Des Guetz G, Uzzan B, Nicolas P, Valeyre D, Sebbane G, et al. (2012) Comparison of the efficacy and safety of single-agent and doublet chemotherapy in advanced non-small cell lung cancer in the elderly: a meta-analysis. Crit Rev Oncol Hematol 84: 340–349.
- 13. Finkelstein DM, Ettinger DS, Ruckdeschel JC (1986) Long-term survivors in metastatic non-small-cell lung cancer: an Eastern Cooperative Oncology Group Study. J Clin Oncol 4: 702–9.
- 14. Albain KS, Crowley JJ, LeBlanc M, Livingston RB (1991) Survival determinants in extensive-stage non-small-cell lung cancer: the Southwest Oncology Group experience. J Clin Oncol 9: 1618–26.
- 15. Paesmans M, Sculier JP, Libert P, Bureau G, Dabouis G, et al. (1995) Prognostic factors for survival in advanced non-small-cell lung cancer: univariate and multivariate analyses including recursive partitioning and amalgamation algorithms in 1,052 patients. The European Lung Cancer Working Party. J Clin Oncol 13: 1221–30.
- 16. Hoang T, Dahlberg SE, Sandler AB, Brahmer JR, Schiller JH, et al. (2012) Prognostic models to predict survival in non-small-cell lung cancer patients treated with first-line paclitaxel and carboplatin with or without bevacizumab. J Thorac Oncol 7: 1361–8.
- 17. Niho S, Kunitoh H, Nokihara H, Horai T, Ichinose Y, et al. (2012) 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 76: 362–367.
- 18. Qi Y, Allen Ziegler KL, Hillman SL, Redman MW, Schild SE, et al. (2012) Impact of disease progression date determination on progression-free survival estimates in advanced lung cancer. Cancer 118: 5358–5365.
- 19. Treat JA, Gonin R, Socinski MA, Edelman MJ, Catalano RB, et al. (2010) A randomized, phase III multicenter trial of gemcitabine in combination with carboplatin or paclitaxel versus paclitaxel plus carboplatin in patients with advanced or metastatic non-small-cell lung cancer. Ann Oncol 21: 540–547.
- 20. Okamoto I, Yoshioka H, Morita S, Ando M, Takeda K, et al. (2010) Phase III trial comparing oral S-1 plus carboplatin with paclitaxel plus carboplatin in chemotherapy-naïve patients with advanced non-small-cell lung cancer: results of a west Japan oncology group study. J Clin Oncol 28: 5240–5246.
- 21. Hartmann H, Muller J, Marschner N (2012) Is there difference in demography and clinical characteristics in patients treated with and without bevacizumab? J Clin Oncol 30: 3317–3318.
- 22. Gridelli C, Bennouna J, de Castro J, Dingemans AM, Griesinger F, et al. (2011) Randomized phase IIIb trial evaluating the continuation of bevacizumab beyond disease progression in patients with advanced non-squamous non-small-cell lung cancer after first-line treatment with bevacizumab plus platinum-based chemotherapy: treatment rationale and protocol dynamics of the AvaALL (MO22097) trial. Clin Lung Cancer 12: 407–411.
- 23. Mok TS, Wu YL, Thongprasert S, Yang CH, Chu DT, et al. (2009) Gefitinib or carboplatin-paclitaxel in pulmonary adenocarcinoma. N Engl J Med 361: 947–957.
- 24. Maemondo M, Inoue A, Kobayashi K, Sugawara S, Oizumi S, et al. (2010) Gefitinib or chemotherapy for non-small-cell lung cancer with mutated EGFR. N Engl J Med 362: 2380–2388.
- 25. Zhu J, Sharma DB, Gray SW, Chen AB, Weeks JC, et al. (2012) Carboplatin and paclitaxel with vs. without bevacizumab in older patients with advanced non-small cell lung cancer. JAMA 307: 1593–1601.
- 26. Lima AB, Macedo LT, Sasse AD (2011) Addition of bevacizumab to chemotherapy in advanced non-small cell lung cancer: a systematic review and meta-analysis. PLoS One 6: e22681.