Declaration of conflict of interest.
Comparative effectiveness of bevacizumab plus cisplatin-based chemotherapy versus pemetrexed plus cisplatin treatment in East Asian non-squamous non-small cell lung cancer patients applying real-life outcomes
Article first published online: 18 MAY 2011
© 2011 Blackwell Publishing Asia Pty Ltd
Asia-Pacific Journal of Clinical Oncology
Special Issue: Bevacizumab in Non Small-cell Lung Cancer in Asia
Volume 7, Issue Supplement s2, pages 34–40, June 2011
How to Cite
CHANG, G.-C., AHN, M.-J., WRIGHT, E., KIM, H. T., KIM, J.-H., KANG, J. H., KIM, S.-W., SHERMAN, S. and WALZER, S. (2011), Comparative effectiveness of bevacizumab plus cisplatin-based chemotherapy versus pemetrexed plus cisplatin treatment in East Asian non-squamous non-small cell lung cancer patients applying real-life outcomes. Asia-Pacific Journal of Clinical Oncology, 7: 34–40. doi: 10.1111/j.1743-7563.2011.01400.x
E Wright and S Walzer are employed by F. Hoffmann-La Roche AG, Basel, Switzerland and S Walzer holds shares in F. Hoffmann-La Roche Ltd. Ahn M-J is on the bevacizumab advisory board advising on its pharmacoeconomic study in NSCLC. J-H Kim has previously received a grant for research from Roche, Eli Lilly, Astra Zeneca and GSK. S Sherman is a consultant in health economics employed by Analytica International Inc. and have received consulting fees by F. Hoffman-La Roche AG.
The authors do not report any conflict of interest with regard to the contents of this study other than those stated.
- Issue published online: 18 MAY 2011
- Article first published online: 18 MAY 2011
- Accepted for publication 3 April 2011.
- antineoplastic combined chemotherapy protocol;
- comparative study;
- non-small cell lung carcinoma
Aim: To indirectly compare real-life clinical effectiveness of bevacizumab + cisplatin-based therapy from the Safety of Avastin in Lung (SAiL) phase IV clinical trial with published evidence from the phase III clinical trial for pemetrexed + cisplatin among East Asian patients with non-squamous metastatic or recurrent non-small cell lung cancer (NSCLC).
Methods: Survival outcomes were compared between subgroups of East Asian patients receiving treatments of either bevacizumab + cisplatin-based chemotherapy or pemetrexed + cisplatin using a matching-adjusted indirect comparison approach. Patient-level data were used to derive a new group with similar characteristics compared to those reported in a phase III clinical trial evaluating pemetrexed + cisplatin therapy. Exclusions to the SAiL data included those with an Eastern Cooperative Oncology Group (ECOG) performance status (PS) of 2, those with mixed cell histology, non-East Asians and those who did not receive cisplatin-based chemotherapy. In total 1000 samples of the pre-matched analysis set of the SAiL data were selected that resulted in equal distributions of the ECOG PS and gender matching variables selected and evaluated for a progression-free survival (PFS) outcome.
Results: Median PFS was longer for patients treated with bevacizumab-based therapy (7.4 months; 95% confidence interval [CI]: 6.7–8.2) versus pemetrexed + cisplatin (6.4 months; 95% CI N/A) among non-squamous East Asian NSCLC patients.
Conclusion: The results suggest that East Asian non-squamous NSCLC patients treated with bevacizumab-based therapy have a trend toward improved PFS outcomes compared to those treated with pemetrexed + cisplatin, even after adjusting for differences between the two trial groups.
Lung cancer represents a significant disease burden in East Asia as these patients account for nearly a quarter of all cancer diagnoses in the region.1 Approximately 85% of all diagnosed patients have non-small cell lung cancer (NSCLC) histology, of which 77% are of a non-squamous subtype present as either adenocarcinoma, large cell or otherwise unspecified.2
Although platinum-based chemotherapy remains the standard of care for the first-line treatment of all NSCLC patients, there has been a recent push to develop more targeted approaches to therapy due to prognostic differences between the various histological subtypes of NSCLC.3 Bevacizumab, in combination with chemotherapy, has shown survival gains over chemotherapy alone among non-squamous NSCLC patients.4–9 A statistically significant overall survival (OS) advantage was demonstrated in a phase III trial (E4599, N = 878) in favor of treatment with bevacizumab + cisplatin and gemcitabine versus cisplatin and gemcitabine alone (hazards ratio [HR]: 0.79; P = 0.003).8 In another phase III trial (Avastin in Lung [AVAiL]) including 1043 randomized patients, those who received bevacizumab + chemotherapy saw a significant improvement in progression free survival (PFS) in treatment arms receiving doses of bevacizumab of either 7.5 mg/kg (HR: 0.75; 95% confidence interval (CI): 0.64–0.87, P = 0.0003) or 15 mg/kg (HR: 0.85; 95% CI: 0.73–1.00, P = 0.0456).7 Among the East Asian subgroup in this study (N = 71), those treated with 7.5 mg/kg of bevacizumab also experienced an OS improvement versus placebo (HR: 0.46; 95% CI: 0.22–0.97), although a PFS survival improvement was not seen in either dosage group (7.5 mg/kg HR: 0.6; 95% CI: 0.32–1.13; 15 mg/kg HR: 0.65; 95% CI: 0.35–1.22).10 A phase IV single arm non-comparative safety study (Safety of Avastin in Lung [SAiL]) examining outcomes from a real-life oncology practice setting was conducted. 2,172 patients were followed and demonstrated an OS of 14.6 months (95% CI: 13.8–15.3) and a median time to progression of 7.8 months (95% CI: 7.5–8.1) among patients receiving either 7.5 or 15 mg/kg of bevacizumab + chemotherapy.4 A subgroup of Chinese patients from the SAiL trial had a median PFS of 8.8 months (95% CI: 8.1–10.0) and a median OS of 18.5 months (95% CI: 16.3–19.6), further demonstrating the real-life survival benefit of bevacizumab treatment11
A recent phase III clinical trial performed by Scagliotti et al. comparing a third-generation chemotherapy, pemetrexed, in combination with cisplatin versus gemcitabine + cisplatin for NSCLC patients, showed non-inferiority for OS among the intent to treat group (HR: 0.94; 95% CI: 0.84–1.05; median OS: 10.3 months vs 10.3 months), which included both squamous and non-squamous patients; however, among the subgroup of non-squamous patients, statistical superiority for OS was demonstrated (HR: 0.84; 95% CI: 0.74–0.96; median OS: 11.0 months vs 10.1 months) in favor of the pemetrexed + cisplatin treatment group.12,13 Patients in this study treated with pemetrexed + cisplatin experienced a 5.3-month median PFS (HR: 0.95; 95% CI: 0.84–1.06). A further analysis was performed examining East Asian patients. Among non-squamous East Asian patients a 21.2 months median OS (HR: 0.70; 95% CI: 0.39–1.24) and 6.4 months PFS (HR: 0.61; 95% CI: 0.39–0.96) were found.13
To date, a single head-to-head clinical trial examining the comparative effectiveness of bevacizumab-based therapy versus pemetrexed + cisplatin has not been conducted. Matching-adjusted indirect treatment comparison methodologies have been developed to estimate the comparative effectiveness of two treatments when patient-level data are available from one study while only aggregate data summarizing patients' characteristics and outcomes are available for another.14,15 The aim of this study is therefore to employ a matching-adjusted indirect analysis to compare the real-life clinical effectiveness of bevacizumab + cisplatin-based chemotherapy from the SAiL clinical trial with published evidence from the phase III clinical trial for pemetrexed + cisplatin among East Asian patients with non-squamous metastatic or recurrent NSCLC, while controlling for prognosis-related differences between the two trial groups.
A matching-adjusted approach was employed to indirectly compare the treatment effects of bevacizumab + cisplatin-based chemotherapy versus pemetrexed + cisplatin among East Asian patients with advanced or metastatic non-squamous NSCLC. Although patient-level data were available from the SAiL clinical trial, an appropriate study and accompanying paper providing relevant patient characteristics and survival outcomes for the pemetrexed + cisplatin comparator were researched and selected based on the similarity of their inclusion and exclusion criteria with those of the SAiL trial. A phase III study evaluating pemetrexed + cisplatin among several racial groups was chosen as the basis of the indirect comparison, of which an article authored by Yang et al.13 provided all patient characteristics and efficacy data for the East Asian subgroup in the trial.
Once the appropriate publication was identified, a subset of the SAiL group was selected to align the inclusion and exclusion criteria between the two studies. Data from the SAiL trial were limited to East Asian patients with cisplatin-based chemotherapy as their treatment and those with an Eastern Cooperative Oncology Group (ECOG) performance status (PS) of 0 or 1. After these exclusions, 72 patients remained in the pre-matched analysis set out of the original 2172 in the overall SAiL study group (Fig. 1).
Distributions of patients' characteristics were compared between the pre-matched SAiL analysis set and those reported in the article published by Yang et al. to identify imbalances in the study groups. Moderate to large differences persisted in the pre-matched distributions for the bevacizumab versus pemetrexed treatment group, with a median age 6.7 years higher, an ECOG PS of 1 8.9% higher and current or former smokers 3.6% higher, while additional discrepancies were observed in histology subtypes, including a 21.6% higher proportion of adenocarcinoma patients, a 2.4% higher proportion of other non-squamous histologies and the absence of any squamous or large cell carcinoma patients (Table 1). Potentially prognostic factors that did not present approximately equal distributions between the two groups were considered as candidate-matching variables. Based on these criteria, ECOG PS and gender were chosen as variables by which to match the bevacizumab and pemetrexed study groups.
|Baseline characteristics||SAiL Phase IV study (before exclusions and matching) (N = 2,172)||SAiL Phase IV study East Asian subset (after exclusions and before matching) (n = 72)||SAiL Phase IV study East Asian subset (final group after exclusions and matching) (n = 46 for each matched sample)||Phase III pemetrexed study, Yang et al.13 (N = 67)|
|Median (range)||60.0 (24–86)||55.0 (29–76)||54.0 (29–76)||61.7|
|ECOG performance status (%)|
|Disease stage (%)|
|Smoking status (%)|
|Histology subgroup (%)|
A matching algorithm was developed whereby 1000 samples of the pre-matched analysis set of the SAiL data were selected that resulted in equal distributions of the ECOG PS and gender-matching variables selected. A distribution of survival outcomes was derived by calculating the median PFS estimate from each of the 1000 matched sets. The mean of this distribution provided the PFS point estimates, while a 95% CI was defined by the 2.5 and 97.5 percentiles.
Two separate sensitivity analyses were performed by additional subgroup matches using the distributions of age alone and also matching on smoking status alone. Both sensitivity matches were performed as described above to derive median PFS estimates with corresponding 95% CIs. However, as only the median values were presented for the pemetrexed + cisplatin baseline age, the SAiL group was first categorized by the median value reported in the Yang et al. article. The SAiL group was then adjusted so that equal proportions were present above and below the median values reported for the pemetrexed group. All analyses were performed using SAS version 9.2 (SAS, Cary, NC, USA).16
Subsequent to taking a subset of the SAiL group to match the inclusion and exclusion criteria of the published overall phase III pemetrexed study, all patient characteristics were aligned, with the exception of age, smoking status and histology (Table 1). The median age after matching for patients receiving bevacizumab treatment was 7.7 years younger than those treated with pemetrexed, while the proportion of current or former smokers was 4.5% higher in the bevacizumab treatment arm. With respect to the histology subtype, no squamous patients were included in the SAiL study while 29.9% of East Asian patients described in Yang et al. were reported as having a squamous histology; 90.2% of bevacizumab patients presented with adenocarcinoma histology versus 70.5% receiving pemetrexed. Furthermore, 1.5% of pemetrexed patients had large cell carcinomas versus none in the SAiL matched group, and those with bronchioalveolar carcinomas represented 1.5% the total matched bevacizumab group versus none for the pemetrexed study group. The proportion of patients with other non-squamous histology was 3.7% higher in the bevacizumab patient group than in the observed proportion for the pemetrexed group.
In total, 46 SAiL patients were evaluated for PFS in each of the 1000 matched samples. Median PFS estimates were longer for patients treated with bevacizumab-based therapy (7.4 months; 95% CI: 6.7–8.2) than those treated with pemetrexed + cisplatin (6.4 months; 95% CI N/A) (Table 2). Statistical significance could not be assessed because of unreported CI for the PFS estimate published in the Yang et al. paper.
|SAiL, phase IV study (n = 46 for each matched sample)||Phase III pemetrexed study† (n = 47)|
|Median progression-free survival time||7.4 months (95% CI: 6.7–8.2)||6.4 months (note: 95% CI not published for pemetrexed study)|
Head-to-head clinical trials still remain the gold standard for comparative effectiveness and health technology assessment research; however, alternatives are necessary when direct treatment comparisons between agents of interest have not yet been performed. A trend has developed recently as more reimbursement and health technology assessment agencies (e.g. National Institute for Health and Clinical Excellence, Canadian Agency for Drugs and Technologies in Health, Australian Government Department for Health and Ageing)17–19 are increasingly recognizing the validity of indirect evidence for the evaluation of comparative effectiveness. After adjusting for inclusion and exclusion criteria and baseline characteristics, the results of this matching-adjusted indirect comparison suggest that there is a trend toward a survival improvement for patients treated with bevacizumab-based therapy versus those treated with pemetrexed + cisplatin. The observed effect is consistent with that seen in a subgroup analysis using the AVAiL clinical trial data and an indirect treatment comparison methodology to evaluate comparing bevacizumab + cisplatin-based therapy versus pemetrexed + cisplatin.6,20
The methods employed in this analysis aim to improve upon alternative approaches that do not adjust patient groups to account for baseline differences.14 Matching reduces the potential for confounding effects by balancing important features of the groups prior to comparing outcomes.
There are limitations to this analysis that are important to note. This study was constrained by the relatively small groups that were compared, with 45 patients in the bevacizumab + cisplatin-based therapy treatment group and a similarly low number of patients in the pemetrexed + cisplatin treatment groups (N = 46). This could limit the analysis's ability to detect differences in survival outcomes in this post hoc analysis between the two treatments.
Although matching on key risk factors can also balance distributions of other prognostic characteristics, both observed and potentially unobserved imbalances persisted between the East Asian groups in the two treatment arms of this analysis. Following matching, all observed variables, with the exception of age and smoking status, were similar. Although no squamous patients were included in the SAiL study, with respect to the distribution the non-squamous histology subtypes, the two study groups were similar. A total of 90.2% of bevacizumab patients presented with adenocarcinoma histology versus 91.5% of the non-squamous patients receiving pemetrexed; however, 2.1% of pemetrexed patients had large cell carcinomas versus none reported in the SAiL matched group, while the opposite was observed for bronchioalveolar carcinoma: Overall 1.5% of bevacizumab patients and none reported for the pemetrexed study arm. The proportion of non-squamous patients classified as having other non-squamous histology was slightly higher for the bevacizumab patients (8.2%) than the pemetrexed patients (6.4%).
Sensitivity analyses were performed to address the observed differences in age and smoking status. Matching on median age alone provided a median PFS (8.7 months; 95% CI: 7.3–10.7), consistent with that seen in this analysis. Similarly, matching on smoking status alone yielded results for PFS (7.6 months; 95% CI: 7.2–8.1) in line with the observed outcome in this analysis. Although adding additional matching variables could have potentially improved the balance between the treatment groups, each additional matching variable added lowers the sample size of the subsequent matched samples. In this study, two variables were chosen to ensure that patients were observed in all overlapping levels of the matching variables. Additionally, although this approach is helpful in determining the association between treatment exposures and outcomes, causality cannot be inferred from this analysis.
The unavailability of aggregate data summarizing non-squamous East Asian patients from the pemetrexed phase III clinical trial hindered the methods employed in this analysis. This led to a potential imbalance between the matched SAiL group and the actual unreported patient characteristics for the non-squamous pemetrexed + cisplatin treatment arm. After examining the aggregate patient distributions reported on the full study group, some differences between the squamous and non-squamous subgroups persisted. Among squamous patients relative to non-squamous patients there was a slightly higher proportion of men (85.0 vs 64.7%), a lower proportion of never smokers (7.0 vs 18.0%), a lower proportion of stage IV patients (68.0 vs 79.7%), approximately equal proportions of individuals with an ECOG PS of 1 (64.0 vs 64.6%) and a median age 2.2 years older.21 This suggests that relative to the non-squamous patients who served as the basis for the outcome comparisons, the SAiL matched group would have a higher proportion of men, a lower proportion of patients with stage IV disease, patients who were older and a lower proportion of never smokers. To test the validity of our findings a subset of patients, which included both East Asian and non-East Asian patients, was matched to the distributions of the full non-squamous pemetrexed + cisplatin group, and a subset of East Asian patients was taken subsequent to matching on ECOG PS and gender. Survival was consistent with that observed in our study for PFS (median: 7.8 months; 95% CI: 7.4–8.2).
Additional research would be valuable to confirm the results from this analysis. These results suggest that East Asian non-squamous NSCLC patients treated with bevacizumab-based therapy have an improved PFS compared with those treated with pemetrexed + cisplatin, even after adjusting for differences between the two trial groups. Given the available evidence, bevacizumab-based therapy should be considered as the standard of care.
This study was supported by F. Hoffman-La Roche AG. Analytica International Inc. (through funding from F. Hoffmann-La Roche AG) provided writing assistance to the authors in the preparation of this manuscript.
- 1GLOBOCAN 2008, Cancer Incidence and Mortality Worldwide: IARC CancerBase No. 10. International Agency for Research on Cancer, Lyons, France 2008 [cited 14 April 2011]. Available from: http://globocan.iarc.fr., , , , , .
- 2SEER Cancer. Statistics review 1957–2007. National Cancer Institute 2011. [Cited 14 April 2011 Available from: http://seer.cancer.gov/csr/1975_2007/browse_csr.php?section=15&page=sect_15_table.28.html.
- 5Randomized, open-label, multicenter phase II study of bevacizumab in combination with carboplatin and paclitaxel in chemotherapy-naive Japanese patients with advanced or recurrent nonsquamous non-small cell lung cancer (NSCLC): JO19907. J Clin Oncol 2009; 27: abstr 8036., , et al.
- 9An indirect comparison of the efficacy of bevacizumab plus cisplatin and gemcitabine (BCG) or bevacizumab plus carboplatin and paclitaxel (BCP) versus pemetrexed plus cisplatin (PC) and cetuximab plus vinorelbine and cisplatin (CVC) in patients (pts) with advanced or recurrent non-small cell cancer (NSCLC). 2009. European Society for Medical Oncology (ESMO) 20–24 September 2009 Berlin., , et al.
- 11Final safety data for Chinese patients (pts) with non-squamous non-small cell lung cancer (NSCLC) who received first-line bevacizumab (Bv) plus chemotherapy in SAiL (MO10390). 2010. European Lung Cancer Conference 2010., , , .
- 14First-line bevacizumab-based therapy versus pemetrexed + cisplatin for the treatment of advanced adenocarcinoma non-squamous non-small cell lung cancer: indirect comparison applying real-life outcomes. 2010. ISPOR 13th European Congress 6–7 November 2010., , , .
- 15Robust comparative effectiveness without head-to-head trials: leveraging. Primary clinical trial data for indirect comparisons. 2011. iHEA 7th World Congress on Health Economics, 12–15 July Beijing..
- 16SAS (computer program). Version 9.2 SAS Institute Inc, Cary, NC, USA 2008.
- 17Department of Health and Ageing. Guidelines for preparing submissions to the Pharmaceutical Benefits Advisory Committee (version 4.3) 2008. Therapeutic Goods Administration, Australian Government, Canberra 2008.
- 18Indirect comparisons of competing interventions. Health Technol Assess 2005; 9: i–iv., , et al.
- 19Indirect Evidence: Indirect Treatment Comparisons in Meta-Analysis. Canadian Agency for Drugs and Technologies in Health (CADTH), Ottawa 2009., , , , .
- 20An indirect comparison of bevacizumab plus cisplatin-gemcitabine and cisplatin plus pemetrexed treatment for patients with advanced first-line non-squamous non-small cell lung cancer in East Asia. Asia–Pac J Clin Oncol 2011; 7(Suppl. 2): 13–21., , et al.