Heterogeneity assessment of two PD-L1 clones: 22C3 and SP142 in patients with non-small cell lung cancer (NSCLC).

Abstract

e14515 Background: Pembrolizumab proves to be effective for clinical treatment of non-small cell lung cancer (NSCLC). The clinical application of pembrolizumab should be hinged on the magnitude of PDL1 expression justified by 22C3 assay on Dako platform. However, whether another diagnostic assay, SP142, can replace 22C3 in pembrolizumab-based immunotherapy thus far remains largely unknown. Methods: Forty-nine cases were randomly collected from 207 NSCLC patients with known PD-L1 expression detected by 22C3. Each case was stained with SP142 on ventana platform. The scores obtained with SP142 were compared with 22C3 , according to the interpretation of 22C3 standard. The concordance of PD-L1 expression between 22C3 and SP142 were assessed by weighted Kappa Coefficient and McNemar-Bowker test, respectively. Results: Based on Dako 22C3-IHC platform, 5 (10.2%), 9 (18.4%) and 35 (71.4%) of 49 cases were classified into groups of strongly positive, weakly positive, and negative for PD-L1, respectively. Yet, 37 (75.5%) of 49 cases had the same results with SP142 protocol (Kappa value: 0.472, p < 0.001). Six cases with weakly positive PD-L1, 3 cases with strongly positive PD-L1 and 1 case with negative PD-L1 according to the 22C3 were misdiagnosed as negative, weak positive and weak positive PD-L1 by SP142, respectively (McNemar-Bowker analysis: p = 0.037). When using the 50% tumor cell PD-L1 expression as the cut-off point, overall sensitivity and specificity of SP142 were 40% (95% [CI], 5.3%-85.3%) and 100% (95% CI, 87.1%-100%), respectively (Kappa value: 0.545, p = 0.009; McNemar-Bowker analysis: p = 0.125). When using the 1% tumor cell PD-L1 expression as the cut-off point, adoption of SP142 yielded the sensitivity of 57.1% (95% CI, 28.9%-82.3%) and the specificity of 97.1% (95% CI, 83.3%-99.9%), respectively (Kappa value: 0.608, p < 0.001; McNemar-Bowker analysis: p = 0.063). Conclusions: Compared with 22C3 assay, PD-L1 expression scores were usually underestimated by the SP142, and it has a higher specificity but lower sensitivity. Our findings collectively suggest that it is not applicable to adopt the SP142-PD-L1 IHC Ventana platform to stratify NSCLC patients for pembrolizumab-based immunotherapy.