Background: Currently, four FDA-approved biomarker assays are available to screen for PD-L1 to enrich for patient response to checkpoint inhibitors (CPI). Each are immunohistochemical (IHC) assays that approximate the percentage of immune or tumor cells expressing PD-L1 using various antibodies, staining and scoring systems. Given this inherent variability, there are concerns whether any single PD-L1 IHC assay, or IHC in general, can be used as a companion or complimentary diagnostic. This is highlighted by the number of individuals, regardless of histology or antibody used, who score below the IHC scoring threshold but respond to PD-L1 inhibitors.

Methods: In this study, we compared PD-L1 protein expression (IHC) to PD-L1 gene expression (CD274) in 436 tumors. PD-L1 IHC was assessed in melanoma using the 28-8 antibody, with kidney, HNSCC, and lung cancer assessed with 22C3. All were scored as per published guidelines. CD274 gene expression was determined by targeted RNA-seq, with each sample's expression level compared and ranked to a reference population.

Results: ANOVA demonstrated a significant relationship between RNA-seq and IHC PD-L1 measurements (p.value < 2e-16). Tukey's HSD comparisons of mean TPS at <1%, 1-4%, and >5% demonstrate significant differences between the three groups that are consistent with gene expression rankings (p adj <0.002). Additionally, for metastatic melanomas with CPI response data, a strong association of objective response rate (ORR) to high RNA-seq expression exists, regardless of IHC result (Table 1). Conclusion: In 400+ tumors, PD-L1 demonstrates correlated mean expression values when assessing protein by IHC and gene expression by RNA-seq. For the CPI treated melanomas with outcomes, PD-L1 IHC ≥1% had a 56% ORR, which improved to >71% ORR when combined with high PD-L1 gene expression. With the need to better predict CPI response, this data suggests that combination PD-L1 testing is an improvement over the FDA approved IHC assays.

Table 1: ORR for 37 melanoma patients based on combination PD-L1 expression results

PD-L1 Method Combinations CR PR SD PD ORR 95% CI 
RNAseq High + IHC ≥1% 71.4% 29.04%-96.33% 
RNAseq High + IHC <1% 75.0% 19.41%-99.37% 
RNAseq Low + IHC ≥1% 50.0% 1.26%-98.74% 
RNAseq Low + IHC <1% 15 20.8% 7.13%-42.15% 
PD-L1 Method Combinations CR PR SD PD ORR 95% CI 
RNAseq High + IHC ≥1% 71.4% 29.04%-96.33% 
RNAseq High + IHC <1% 75.0% 19.41%-99.37% 
RNAseq Low + IHC ≥1% 50.0% 1.26%-98.74% 
RNAseq Low + IHC <1% 15 20.8% 7.13%-42.15% 

Citation Format: Jeffrey Conroy, Sarabjot Pabla, Mary Nesline, Sean Glenn, Blake Burgher, Maochun Qin, Jonathan Andreas, Vincent Giamo, Felicia L. Lenzo, Angela Omilian, Wiam Bshara, Antonios Papanicolau-Sengos, Yirong Wang, Marc Ernstoff, Mark Gardner, Carl Morrison. Predicting response: PD-L1 biomarker testing by IHC and RNA-seq [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2018; 2018 Apr 14-18; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2018;78(13 Suppl):Abstract nr 4526.