Arora K, Mehrotra H, Saikia K, Abou Shaar R, Alhamar M, Chitale DA. Correlation of PD-L1 expression, clinicopathologic and molecular characteristics in an array of solid tumors: A large-scale real world study. Laboratory Investigation 2021; 101(Suppl 1):499-501.
Background: Programmed death ligand-1 (PD-L1) is a predictive marker of anti-programmed death protein 1 (PD- 1)/PD-L1 therapies for solid tumors. Limited literature exists correlating PD-L1 expression, clinicopathological & molecular profiles. We aimed to 1) correlate PD-L1 immunohistochemistry (IHC) results with these profile across multiple solid tumors & 2) assess clinical outcomes (overall survival (OS) & disease-free survival (DFS)) of PD-L1 status with / without anti-PD-L1 immunotherapy (IT).
Design: All cases tested for PD-L1 IHC over 2 years (Aug 2019-Sep 2020) were retrieved for this study. Clinicopathological variables recorded included age, race, tumor type, type of PD-L1 clone, PD-L1 status (Tumor Proportion Score (TPS): negative: <1%, low:1-49%, high: >50%), Combined Positive Score (CPS): negative <1, low 1-10, high > 10), clinical stage, anti-PD-L1 IT. Microsatellite instability (MSI) status using IHC & Ploymerase chain reaction (PCR) assays was recorded. High PD-L1 was defined as PD-L1 expression of TPS >50%/CPS>10. Outcome studies included OS and DFS after generating Kaplan-Meier curves & compared using log rank test. Univariate analysis using Cox regression models were also used.
Results: There were 205 cases tested for PD-L1 by IHC. Cohort included non-small cell lung cancers (127), head & neck carcinomas (37), gastric or gastroesophageal carcinoma (20), kidney or urothelial carcinoma (16), cervical carcinomas (5). Median age was 70 years (range 28-90). Most were high stage cancers [stage 1: 5/205, stage 2: 5/205, stage 3: 30/205, stage 4 165/205]. PD-L1 IHC clones included: 22C3 (152/205), 28-8 (21/205) & both (32/205). High PD-L1 expression was observed in 52/205 (25.3%), out of which [37/127 (29.1%) were adenocarcinoma, 13/54 (24%) were squamous cell carcinoma, 2/24 (4.1%) others]. Anti PD-L1 IT was given in 65/205 (31.7%) patients. Anti PD-L1 IT was significantly associated with longer median survival OS (p=0.015) & DFS (p=0.004) (Figure 1). PD-L1 status was significantly associated with OS (p=0. 034) but not DFS (p=0. 076) (Figure 1). High PD-L1 had shorter median survival and higher hazards of death in OS (HR=5.4, CI-1.3-23.1) irrespective of IT. Association between three groups of PD-L1 status when compared with IT was statistically significant (p=0.048, Figure 2). PD-L1 & MSI testing was available in 29 patients & did not show any statistical correlation in this small cohort. No significant difference in survival for those received IT (4/29) vs no IT (25/29) & tested for both PD-L1 & MSI (OS: p= 0.277, DFS: p= 0.107).
Conclusions: This study supports the rational approach for PD-L1 therapy. High PD-L1 expression is more commonly seen in adenocarcinoma. Expression of high PD-L1 is associated with worse OS but not DFS. PD-L1 IT is significantly associated with longer median survival, OS & DFS. Larger, prospective studies are needed to support our findings.