The Spatial Proximity of CD8+ FoxP3+PD-1+ Cells to Tumor Cells: A More Accurate Predictor of Immunotherapy Outcomes in Advanced Non-Small-Cell Lung Cancer

Abstract

Background: To optimize precision immunotherapy for advanced NSCLC, comprehensive tumor immune microenvironment (TIME) characterization is crucial for efficacy prediction.

Methods: Pretreatment tumor samples from 46 advanced NSCLC patients treated with PD-1/PD-L1 inhibitors were analyzed. The subregional abundance and spatial proximity scores of TIME cell subpopulations in 27 samples were assessed via multiplex immunohistochemistry (mIHC) targeting pan-CK, CD163, CD8, FoxP3, PD-1, and PD-L1. Correlations between the TIME features, clinicopathologic factors, treatment response, and prognosis were evaluated.

Results: CD8⁺FoxP3⁺ cells were identified in NSCLC tissues, predominantly expressing PD-1/PD-L1. The PD-L1 TPS subgroups showed significant immune cell density/proximity differences, but CD8⁺FoxP3⁺PD-1⁺ infiltration was PD-L1 TPS-independent. Responders had higher CD8⁺FoxP3⁺PD-1^high density (p = 0.0497) and proximity scores (p = 0.0099) than non-responders. The CD8⁺FoxP3⁺PD-1⁺ presence and tumor proximity were essential for favorable outcomes. In low-PD-L1 TPS patients, the CD8⁺FoxP3⁺PD-1⁺ abundance and proximity scores strongly predicted the response (AUC: 0.79 and 0.75 vs. PD-L1 TPS AUC = 0.58). A survival analysis linked the presence and proximity score of CD8⁺FoxP3⁺PD-1⁺ cells to prolonged overall survival (OS) and progression-free survival (PFS). Notably, a low proximity score of CD8⁺FoxP3⁺PD-1⁺ cells emerged as an independent risk factor for a shorter PFS (HR = 6.16, 95% CI: 2.12-17.93, p = 0.001).

Conclusion: The CD8⁺FoxP3⁺PD-1⁺ spatial proximity to tumor cells robustly predicts improved immunotherapy outcomes in advanced NSCLC.

Keywords: CD8+FoxP3+PD-1+ cells; PD-1/PD-L1 blockade; advanced non-small-cell lung cancer; spatial proximity; therapeutic outcomes.

Figure 1

Overview of the TIME of advanced NSCLC: (a) diagram summarizing the study design; (b) representative mIHC images of the overall TIME, and the cell phenotypes of all possible copositive combinations of the 6 targets, as well as the subsets of other_PD-1+ and other_PD-L1+; (c) immunocyte subpopulations that are ranked in the top 10 based on median density in the tumor area, stroma area, and whole slide; and (d) immunocyte subpopulations sorted according to the median proximity score.

Figure 2

The correlation of CD8⁺FoxP3⁺ cells with clinicopathologic traits and their PD-1/PD-L1 expression features: (a) significant associations of immune cell subregion densities and proximity scores with clinicopathologic features, displaying -log₁₀P-values for p < 0.05 (from left to right: tumor area density, stromal area density, overall density, and proximity score); (b) densities of immune cell subsets showing significant variation among PD-L1 TPS subgroups (* p < 0.05, ** p < 0.01, *** p < 0.001, and the color black indicated significant Kruskal–Wallis test p-values for the three groups, while green showed significant p-values from pairwise Dunn’s Test despite non-significant overall differences); (c) proximity scores of immune cell subsets showing significant variation among PD-L1 TPS subgroups (* p < 0.05, and ** p < 0.01); and (d) comparison of proportions of T-cell subsets with different PD-1/PD-L1 levels.

Figure 3

The predictive role of CD8⁺FoxP3⁺PD-1⁺ cells in response to PD-1/PD-L1 inhibitors: (a) variations in the subregional abundance of CD8⁺FoxP3⁺PD-1⁺ cell subsets between immunotherapy responders and non-responders; (b) variations in the proximity score of CD8⁺FoxP3⁺PD-1⁺ cells between immunotherapy responders and non-responders; (c) comparison of immunotherapy efficacy between subgroups of patients with present and absent CD8⁺FoxP3⁺PD-1⁺ cells in tissues; (d) comparison of immunotherapy efficacy between subgroups of patients with proximity-high and -low CD8⁺FoxP3⁺PD-1⁺ cells; (e) differentials in the density and proximity score of CD8⁺FoxP3⁺PD-1⁺ cells in PD-L1 TPS subgroups between responders and non-responders (with a PD-L1 TPS cutoff value of 50%); (f) ROC curves for predicting objective response to immunotherapy using CD8⁺FoxP3⁺PD-1⁺-cell proximity score, density, and PD-L1 TPS. (* p < 0.05, and ** p < 0.01).

Figure 4

The prognostic value of the proximity score of CD8⁺FoxP3⁺PD-1⁺ cells in immunotherapy in advanced NSCLC: (a) survival analysis of the proximity score of CD8⁺FoxP3⁺PD-1⁺ cells for OS (high: patients with a high proximity score; low: patients with a low proximity score; patients were grouped by the median value); (b) survival analysis of the proximity score of CD8⁺FoxP3⁺PD-1⁺ cells for PFS (high: patients with a high proximity score; low: patients with a low proximity score; patients were grouped by the median value); (c) forest plot of the results from the univariate Cox regression analysis for immunocyte proximity scores associated with OS; (d) forest plot of the results from the multivariate COX regression analysis of immunocyte proximity scores significant in univariate analysis; and (e) forest plot of the results from the multivariate COX regression analysis of PFS considering clinicopathological characteristics and immunotherapeutic regimens (only covariates demonstrating p < 0.1 in univariate analyses were included; patient subgroups with high and low CD8⁺FoxP3⁺PD-1⁺-cell proximity score were defined by the median value).