%0 Journal Article
%T Clinical and Molecular Predictors of PD-L1 Expression in Non-Small-Cell Lung Cancer: Systematic Review and Meta-analysis.
%A Petrelli F
%A Maltese M
%A Tomasello G
%A Conti B
%A Borgonovo K
%A Cabiddu M
%A Ghilardi M
%A Ghidini M
%A Passalacqua R
%A Barni S
%A Brighenti M
%J Clin Lung Cancer
%V 19
%N 4
%D 07 2018
%M 29530732
%F 4.84
%R 10.1016/j.cllc.2018.02.006
%X Clinicopathologic and molecular characteristics of non-small-cell lung cancers (NSCLCs) associated with a strong expression of programmed death ligand 1 (PD-L1+ in > 5% of cells) have not been well elucidated. Expression of PD-L1 is a poor prognostic factor, but NSCLCs with higher levels of PD-L1 have greater benefit when treated with immunotherapy. We have performed a systematic review to synthesize the available evidence regarding clinicopathologic and molecular variables associated with PD-L1 expression in NSCLC. PubMed, EMBASE, SCOPUS, Web of Science and Cochrane Library databases were searched for relevant articles assessing predictors of PD-L1 expression in > 5% cells. Data were reported as odds ratio (OR) of events. Fifty-two studies (for a total of 5066 PD-L1+ out of 13,279 NSCLC patients) were included in this meta-analysis. Factors associated with PD-L1 expression were: smoking status (OR 5.48; 95% confidence interval (CI) 2.8-10.4; P < .001), male gender (OR 4.8; 95% CI 3.2-7.2; P < .001), adenocarcinoma histology (OR 2.75; 95% CI, 1.5-4.8; P < .001), Epidermal growth factor receptor (EGFR) wild type (OR 4.83; 95% CI, 2.1-11.1; P < .001), ALK mutation negative (OR 388.6; 95% CI, 222.5-678.7; P < .001), ROS mutation negative (OR 1904.8; 95% CI, 630-5757; P < .001), and KRAS wild type (OR 19.8; 95% CI, 7.6-51.6; P < .001). Conversely higher pT stages (OR 0.16; 95% CI, 0.04-0.7; P = .01), pN+ stages (OR 0.29; 95% CI, 0.17-0.5; P < .001) are inversely associated with PD-L1 expression in > 5% cells. Expression of PD-L1 is more common in male smokers, with adenocarcinoma histology and not carriers of EGFR/ALK/ROS/KRAS mutations. These data could be useful to screening of PD-L1 expression and to select patients for immunotherapy.