• Pui San Tan, Pedro Aguiar, Benjamin Haaland, and Gilberto Lopes.
• Nuffield Department of Primary Care Health Sciences, University of Oxford, UK. Electronic address: puisan.tan@gmail.com.
• Lung Cancer. 2018 Jan 1; 115: 84-88.

IntroductionRole of PD-L1 expression to guide immunotherapies in previously treated advanced NSCLC remains unclear and there is a lack of data comparing immune checkpoint inhibitors (ICIs) with each other. This network meta-analysis (NMA) aims to compare survival with ICIs to docetaxel and perform indirect comparisons between ICIs in the PD-L1 unselected population and by PD-L1 expression levels.MethodsPubMed was searched and study screening was performed by two independent reviewers. NMA of survival outcomes in the PD-L1 unselected population and by PD-L1 expression levels <1%, >=1%,>=5%,>=10%, and >=50% was performed. Head-to-head indirect comparisons were constructed and treatment rankings were provided. Potential survival benefits by PD-L1 expression level as compared to a PD-L1 unselected population were estimated.Results5 trials with 3024 total patients were included for meta-analysis. Overall, ICIs improved survival across PD-L1 expression levels compared to docetaxel, although there was only weak evidence of benefit for individual ICI nivolumab or atezolizumab in PD-L1<1%. PD-L1 subgroups suggested positive dose-response relationship between PD-L1 expression levels with survival benefits. In addition, there were also survival benefits due to selecting for PD-L1 in the PD-L1>=10% and >=50% subgroups as compared to the PD-L1 unselected population. Indirect comparisons of ICIs showed little evidence of differences between nivolumab, pembrolizumab and atezolizumab.DiscussionICIs improve survival in previously treated advanced NSCLC patients across PD-L1 expression levels compared to docetaxel. There is a positive dose-response relationship between PD-L1 expression and survival benefits, and little evidence of survival differences between nivolumab, pembrolizumab and atezolizumab.Copyright © 2017 Elsevier B.V. All rights reserved.

29 keywords...

hide…