An immune score reflecting pro- and anti-tumoural balance of tumour microenvironment has major prognostic impact and predicts immunotherapy response in solid cancers

Abstract

Background: Cancer immunity is based on the interaction of a multitude of cells in the spatial context of the tumour tissue. Clinically relevant immune signatures are therefore anticipated to fundamentally improve the accuracy in predicting disease progression.

Methods: Through a multiplex in situ analysis we evaluated 15 immune cell classes in 1481 tumour samples. Single-cell and bulk RNAseq data sets were used for functional analysis and validation of prognostic and predictive associations.

Findings: By combining the prognostic information of anti-tumoural CD8⁺ lymphocytes and tumour supportive CD68⁺CD163⁺ macrophages in colorectal cancer we generated a signature of immune activation (SIA). The prognostic impact of SIA was independent of conventional parameters and comparable with the state-of-art immune score. The SIA was also associated with patient survival in oesophageal adenocarcinoma, bladder cancer, lung adenocarcinoma and melanoma, but not in endometrial, ovarian and squamous cell lung carcinoma. We identified CD68⁺CD163⁺ macrophages as the major producers of complement C1q, which could serve as a surrogate marker of this macrophage subset. Consequently, the RNA-based version of SIA (ratio of CD8A to C1QA) was predictive for survival in independent RNAseq data sets from these six cancer types. Finally, the CD8A/C1QA mRNA ratio was also predictive for the response to checkpoint inhibitor therapy.

Interpretation: Our findings extend current concepts to procure prognostic information from the tumour immune microenvironment and provide an immune activation signature with high clinical potential in common human cancer types.

Funding: Swedish Cancer Society, Lions Cancer Foundation, Selanders Foundation, P.O. Zetterling Foundation, U-CAN supported by SRA CancerUU, Uppsala University and Region Uppsala.

Keywords: Immunoscore; Macrophages; Tumour immunology.

Fig. 1

Prognostic value of CD8⁺T cells and CD68⁺CD163⁺macrophages revealed by comprehensive characterization of immune cell subsets in 286 therapy-naïve colon cancers (See also Figures S1–S4). (a) Immune marker combinations in IHC panels define classes and subclasses of immune cells (See also Supplementary Table S1). (b) Forest plot of univariable associations of tissue immune cell densities translated into three-level categorised values, with OS in patients of stage I-III. Filled squares, hazard ratios (HR); whiskers, 95% confidence intervals (CI), ∗p < 0.050 (Cox regression). (c) Representative multiplex macrophage marker staining of colon cancer tissue. Expression of two markers, CD68 (red) and CD163 (green) with nuclear DAPI staining (white), visualised in pseudocolours, identified three cell types (insets), M1-like macrophages, M2-like macrophages and CD68⁻CD163⁺ cells. (d) Venn diagram of the counts of cells in the entire cohort expressing CD68 only (red, n = 9.0 × 10⁵), CD163 only (green, n = 1.9 × 10⁵) or both markers (gold, n = 4.4 × 10⁴). (e) Density of three macrophage subsets in patient tumours. Boxes, median and interquartile range (IQR) of the ratios; whiskers, 1.5 IQR. (f) Signature of immune activation (SIA), defined as the ratio of CD8⁺ cell density to the sum of CD8⁺ and M2-like cell densities.

Fig. 2

The SIA is an independent prognostic predictor with performance superior to established clinical and immunological predictors for overall (OS) and recurrence-free (RFS) survival in therapy-naïve colon cancer stage I-III patients (See also Supplementary Fig. S5 and Supplementary Table S2). (a) OS (upper panel) and RFS (lower panel) for the patients (n = 286), stratified into SIA-low, -intermediate and -high groups, with SIA-low used as reference group. (b) OS (upper panel) and RFS (lower panel) for the patients stratified by trichotomized IS. Relative hazards were estimated by Cox proportional hazards model in (a) and (b). (c) Predictive accuracy of SIA, IS and clinical parameters for OS (upper panel) and RFS (lower panel) using AUC analysis with 1000-fold bootstrap resampling, and the distribution of achieved median values shown in a box plot: horizontal lines indicate 50 percentage, boxes show 95% confidence interval (between 2.5 and 97.5 percentages) and whiskers show upper and lower AUC values. Univariable Cox proportional hazards models were applied to each of the analysed factors separately and multivariable models used to evaluate the impact of factor combinations. The performance of Cox proportional hazards models was compared using the likelihood ratio p value. (d) Relative contribution to the prediction of OS of SIA and clinical parameters (upper) or SIA, IS and clinical parameters (lower) determined using the χ² proportion test. SIA, signature of immune activation; IS, immunoscore.

Fig. 3

The SIA is a prognostic predictor in bladder cancer, gastroesophageal cancer, lung adenocarcinoma and melanoma (See also Supplementary Table S3). (a) Overall survival stratified by SIA in seven tumour types. Tissue microarrays encompassing 94–295 cases of the respective tumour type were stained and the patients in each cohort stratified in terciles according to SIA score, except melanoma, which was stratified in two groups split by the median. Statistical analysis by log-rank test for three groups, and/or Cox regression for pairwise comparison (high vs low and intermediate vs low). Insert tables demonstrate significance of SIA in female or male patients. (b) Overall survival stratified by IS in 7 tumour types. Statistical analysis performed by log-rank test for three groups. (c) Comparison of the predictive accuracy of IS and SIA for OS in 7 tumour types, generated using tAUC analysis. Statistical analysis performed for the evaluation of the difference between the survival models of AUC and IS: the statistically significant time-points are indicated by asterisks; (d) The OS predictive ability of SIA and IS in 8 analysed tumour types.

Fig. 4

Complement complex C1q expression is a hallmark of CD68⁺CD168⁺macrophages and can be used to generate the SIA signature from bulk RNA sequencing of tumours. (a) Heatmap of scaled gene expression values for the 29 genes with highest fold difference between CD68⁺CD163⁻, CD68⁺CD163⁺ and CD68⁻CD163⁺ macrophage classes in colorectal cancers (See also Supplementary Figs. S6 and S7, Supplementary Tables S5–S7). (b and c) Expression level distributions of the macrophage associated genes C1QA-C and APOE in lung cancer (B) and uveal melanoma (See also Supplementary Figs. S8 and S9). (d) Gene expression level distributions of C1QA-C and APOE in three subsets of macrophages in 15 non-diseased organs (See also Supplementary Table S8). (e) Overall survival stratified by dichotomized ratio between the bulk RNA expression levels of CD8A and C1QA in seven tumour types using gene expression data from the KM plotter database (See also Supplementary Fig. S10).

Fig. 5

The SIA predicts response to immune therapy. (a) SIA values generated from bulk RNA data by computing the ratio between counts of CD8A and C1QA-C expression in 26 ICI-treated melanomas from patients grouped by response. Spearman-rank correlation was used to test the associations between SIA levels and response. The regression line is visualised by a red dashed line. (b) Upper panel demonstrates the difference in SIA values, computed from single cell RNA sequencing data and based on the ratio between counts of CD8⁺ cells macrophages in responder and non-responder lesions of 48 melanoma patients who received ICI treatment. M2 like macrophages were defined by double-positivity of CD68 and either CD163 or C1QA-C. Lower panel demonstrates the receiver operating characteristics (red line) and 95% confidence intervals (pink areas), calculated for SIA and therapy response (c) The difference in SIA values, computed from single cell RNA sequencing data as the ratio between counts of CD8⁺ cells macrophages in four renal cell carcinoma patients, with different therapy. Horizontal lines indicate median values and boxes show interquartile range. Mann–Whitney U-test was used for statistical analysis.