Impact of Specimen Characteristics on PD-L1 Testing in Non-Small Cell Lung Cancer: Validation of the IASLC PD-L1 Testing Recommendations.

INTRODUCTION:Molecules targeting programmed cell death 1 or its ligand programmed death ligand 1 (PD-L1) revolutionized the treatment of patients with NSCLC. The only approved biomarker for predicting treatment response is the PD-L1 tumor proportion score (TPS) determined by immunohistochemistry. According to International Association for the Study of Lung Cancer recommendations, specimens that include fewer than 100 tumor cells or are older than 3 years should not be used for PD-L1 testing and the reliability of cell blocks has yet to be validated. METHODS:This retrospective study included 1249 consecutive patients with NSCLC who were tested for PD-L1 (using the clone 22C3) between September 2016 and April 2017. The associations between the presence of suboptimal characteristics (specimens with <100 tumor cells, specimens older than 3 years, or cell blocks) and PD-L1 TPS were examined by using a multinomial logistic regression. RESULTS:Specimens from 35.5% of the patients had at least one suboptimal characteristic. For patients with a PD-L1 TPS of higher than 50%, there was a significantly higher probability that they had a specimen with more than 100 tumor cells (OR = 1.97, p = 0.008) and a more recent block (within 30 days versus after >3 years) (OR = 2.46, p = 0.023). There was no statistical difference in PD-L1 TPS between cell blocks and tissue specimens (biopsy OR = 0.99 [p = 0.996] and surgery OR = 0.73 [p = 0.302]). CONCLUSIONS:Our results suggest that specimens containing fewer than 100 tumor cells or older than 3 years may lead to an underestimation of PD-L1 status. Our findings also provide support for the use of cell blocks for PD-L1 testing, although further research is needed.