PD-L1 expression in lung non small cell carcinoma: evaluation of heterogeneity and definition of bioptic criteria for improving patients selection

Purpose: PD-L1 expression determination defines eligibility for treatment with pembrolizumab in patients with advanced non-small lung cancer (NSCLC). Conflicting results have been reported in the few studies that have addressed the issue of PD-L1 heterogeneity. This study was designed to better define which value across core biopsies from the same case more closely reflects the PD-L1 expression status on whole sections and how many biopsies are needed for confident classification of tumors in terms of PD-L1 expression. Materials and Methods: we built tissue microarrays as surrogate of biopsies collecting 5 cores per case from 268 cases and compared PD-L1 staining results using clone SP263 with tumor whole sections; moreover, we evaluated PD-L1 expression consistency between 84 paired primitive and relapsed tumors. Results: we found an overall positivity in 39% of cases for 1% cutoff and 10% of cases for 50% cutoff; we observed a discordance rate within tissue cores of 20% and 7.9% for ≥ 1% and ≥ 50% cutoffs, respectively. The maximum value across cores was associated with high concordance between cores and whole sections and the lowest number of false negative cases overall. In order to reach an area under the curve (AUC) and sensitivity > 0.90, 4 and 3 cores were necessary at 1% and 50% cutoff, respectively. Importantly, with 20% as cutoff, less than 3 cores showed high sensitivity and specificity in identifying cases with 50% of tumor cells positive for PD-L1 on whole sections. Finally, we found PD-L1 expression concordance between primary and paired metastatic tumor in 88% and 91% of cases using 1% and 50% cutoff, respectively. Conclusions: an accurate definition of the criteria to determine the PD-L1 status of a given tumor may greatly help to select those patients who could benefit from anti-PD1/PD-L1 treatment.