Fig. 3: Spatial analysis of immune cells using the nearest neighbour distance (NND) function and the cancer-specific survival analysis for PD-1/PD-L1 cluster density. | British Journal of Cancer

Fig. 3: Spatial analysis of immune cells using the nearest neighbour distance (NND) function and the cancer-specific survival analysis for PD-1/PD-L1 cluster density.

From: Spatially resolved multimarker evaluation of CD274 (PD-L1)/PDCD1 (PD-1) immune checkpoint expression and macrophage polarisation in colorectal cancer

Fig. 3: Spatial analysis of immune cells using the nearest neighbour distance (NND) function and the cancer-specific survival analysis for PD-1/PD-L1 cluster density.The alternative text for this image may have been generated using AI.

a pseudo-immunofluorescence image from a tumour site showing all nine markers (i). Cell phenotyping maps with nearest neighbour distance analysis from each immune cell to the closest tumour cell and boxplots visualising the distribution of nearest neighbour distances across all tumour images (N = 3,190). The statistical significance was tested with Wilcoxon rank-sum test. ****P < 0.0001 (ii). The expression levels of various phenotypic markers in macrophages according to the distance to the closest tumour cell. The plots are based on 1,582,095 T cells and 1,788,538 macrophages (iii). b pseudo-immunofluorescence image from a tumour site showing all nine markers (i). Cell phenotyping maps and nearest neighbour distance analyses from each PD-1+ and PD-1 T cell to the closest tumour cell (ii) and to the closest PD-L1+ macrophage (iii). Boxplots visualise the distribution of nearest neighbour distances across all tumour images (N = 3,190). The significance was tested with Wilcoxon rank-sum test. ****P < 0.0001. c Kaplan–Meier survival curves and Cox proportion hazards regression models for cancer-specific survival for PD-1+ T cell/PD-L1+ macrophage cluster densities in 910 patients in the tumour centre and the invasive margin. One cluster is composed of one PD-1+ T cell with at least one PD-L1+ macrophage within a 20 µm radius. The cluster densities were divided into ordinal quartiles from low (Q1) to high (Q4). Statistical significance for Kaplan–Meier survival estimates were determined with Log-rank test. Univariable (blue) and multivariable (red) Cox proportional hazards regression models are represented as forest plots with HRs along with their 95% CIs as whiskers. Multivariable Cox proportional hazards regression models were adjusted for sex (male, female), age (<65, 65–75, >75), year of operation (2000–2005, 2006–2010, 2011–2015), tumour location (proximal colon, distal colon, rectum), stage (I–II, III, IV), tumour grade (well/moderately differentiated, poorly differentiated), lymphovascular invasion (negative, positive), MMR status (proficient, deficient), and BRAF status (wild-type, mutant).

Back to article page