Spatial heterogeneity of PD-L1 expression influence its assessment in esophageal squamous cell carcinoma

Abstract

Immune checkpoint inhibitors are a promising treatment for esophageal squamous cell carcinoma (ESCC). However, the predictive value of programmed death-ligand 1 (PD-L1) expression, the most common biomarker for immunotherapy, remains controversial, particularly in the neoadjuvant setting. We hypothesized that the spatial heterogeneous of PD-L1 expression within tumors might render limited biopsy samples unrepresentative of the bulk tumor. In this study, we assessed the spatial heterogeneity in PD-L1 expression within ESCC by sampling four distinct regions using endoscopic biopsy forceps and the largest longitudinal sections on complete resected tumor from treatment-naïve patients. Our findings demonstrated the insufficiency of using limited biopsy tumor tissue to accurately determine the combined positive score (CPS) within the tumor. Notably, spatial heterogeneity was reduced when tumor's CPS was sufficiently high. Multi-region sampling assessment revealed that the maximum CPS derived from three regions provided a more accurate approximation of the bulk tumor's PD-L1 status. Additionally, the densities of CD8⁺/CD4⁺T cells were positively correlated with CPS. These findings emphasize the clinical need for standardized and modified biopsy assessment strategies to improve the accuracy of PD-L1 evaluation, thereby guiding therapeutic decision-making in ESCC.

Keywords: Endoscopic biopsy; Esophageal squamous cell carcinoma; Immunotherapy; PD-L1; Spatial heterogeneity.

Fig. 1

Overview of study design.

Fig. 2

Higher discordance and lower correlation of CPS across regions. (A) Heatmap of CPS in different regions and surgical specimens in Cohort 1. (B-G) Line charts of CPS between two regions. (H) Heatmap of correlation of CPS between four regions and surgical specimens. CPS, combined positive score. *,p < 0.05; **,p < 0.01; ***,p < 0.001; ****,p < 0.0001. Red pentagrams represent patients with positive and regionally consistent PD-L1 expression, and blue triangles represent patients with negative and regionally concordant PD-L1 expression.

Fig. 3

The degree of spatial heterogeneity may decrease when CPS was high enough. (A)Schematic showing the partitioning of Cohort 2 following IHC staining. (B) Distribution of CPS in Cohort 2 (data shown as median with 95 % confidence interval). (C) Coefficient of variation for all patients in Cohort 2. (D) Number of concordant and discordant regions in Cohort 2. (E) Simple linear regression of CPS versus CV for Cohort 2. (F) Simple linear regression of CPS versus overall inconsistency rate for Cohort 2.

Fig. 4

Multi-region biopsy assessment improves the accuracy of PD-L1 expression status using limited tissue samples.

Fig. 5

The densities of CD8⁺T cells and CD4⁺T cells was positively correlated with CPS in regions. (A) Representative pictures of CD8⁺T cells and CD4⁺T cells stained in the same field of view for CPS<1 and CPS≥10. (B) Distribution of CPS under regions with low and high CD8⁺T cell infiltration levels in Cohort 1. (C) Distribution of CPS under regions with low and high CD4⁺T cell infiltration levels in cohort 1. (D) Distribution of CPS under areas with low and high CD4⁺T cell infiltration levels in areas with low CD8⁺T cell infiltration in Cohort 1. (E) Distribution of CPS under regions with low and high CD8⁺T cell infiltration levels in regions with low CD4⁺T cell infiltration in Cohort 1. (F) Distribution of CPS under areas with low and high CD4⁺T cell infiltration levels in areas with high CD8⁺T cell infiltration in Cohort 1. (G) Distribution of CPS under areas with low and high CD8⁺T cell infiltration levels in areas with high CD4⁺T cell infiltration in Cohort 1. (H) Correlation of CPS with CD8⁺T cells and CD4⁺T cells in the subgroup with CPS <10 in Cohort 1. (I) Correlation of CPS with CD8⁺T cells and CD4⁺T cells in the subgroup of CPS ≥10 in Cohort 1. (J) Distribution of CD8⁺T cell infiltration density in Cohort 2. (K) Distribution of CD4⁺T cell infiltration density in Cohort 2. (L) Distribution of CPS under regions with low and high CD8⁺T cell infiltration levels in Cohort 2. (M) Distribution of CPS under regions with low and high CD4⁺T cell infiltration levels in Cohort 2. (N) Distribution of CPS within different types of immune microenvironments in Cohort 2.

Fig. 6

Higher CPS Thresholds and Multi-Region Biopsy Sampling May Enhance the Predictive Value of CPS for Immunotherapy (A) Forest plot of MPR rates at a CPS threshold of 1. (B) Forest plot of MPR rates at a CPS threshold of 10. (C) Forest plot of MPR rates at a CPS threshold of 20. (D) Heatmaps of CPS from proximal and distal biopsy samples alongside pathological response outcomes. (E) Comparison of CD8⁺T cell infiltration density between MPR and non-MPR patients with a CPS maximum >10. (F) Comparison of CD4⁺T cell infiltration density between MPR and non-MPR patients with a CPS maximum >10.