Background In retrospective studies, differential expression of immunological biomarkers in tissue biopsies has shown cancer-type and therapy-specific associations with patient outcomes. A prospective study has not been performed to identify independent sources of variation in tumor-immune microenvironments (TIMEs) across tumor types and explore their associations with specific genomic alterations and patient survival.
Methods We validated and prospectively employed ImmunoPROFILE, an immune profiling assay designed to identify and quantify individual tumor cells and CD8+, FOXP3+, PD-1+, and PD-L1+ immune cells in fixed tissue biopsies with spatial resolution in a semi-automated manner. Immune cell densities, tumor proportion scores (TPS), and cell-to-cell proximity scores were automatically calculated. Unsupervised principal component analysis (PCA) identified independent factors of TIME variation. We explored the association between these principal components (PCs) and genomics and fit Cox model to predict overall survival controlled for major risk factors, including tumor type.
Results We successfully performed ImmunoPROFILE on 2,023 clinical specimens from 27 tumor types over three years (figure 1A , top). PCA identified distinct TIME phenotypes which align with biological intuition while incorporating novel spatial characteristics (PC 1-PC 9; figure 1A , bottom). Tumors with specific genomic alterations were associated with specific PCs. For example, tumors with mismatch repair, DNA-damage repair, and RAS/MAPK pathway mutations were significantly associated with a high density of intra-tumor immune cells in general (PC 1, figure 1B). In contrast, tumors with Wnt/β-catenin pathway mutations were specifically associated with a high ratio of FOXP3+ cells to CD8+ cells (PC 5, figure 1B). After adjusting for risk factors, including tumor type, tumors with a high density of immune cells (PC 1) or with an ‘immune cell-excluded’ phenotype (PC 2) were significantly associated with longer OS (HR 0.76, P=0.003 and HR 0.72, P=0.0002, respectively, figure 1C). Conversely, tumors with PD-L1+ immune cells very near tumor cells (PC 3) or a high ratio of FOXP3+ cells to CD8+ cells (PC 5) were significantly associated with shorter OS (HR 0.58, P=7.9e-10, and HR 0.75, P=0.001, respectively, figure 1C).
Conclusions This study represents a significant milestone as the first enterprise-level immune biomarker assay to use multiplexed staining, digital imaging, machine learning, and genomics in a large-scale prospective analysis of clinical specimens. Through this comprehensive approach, we have demonstrated the value of integrating cell density measurements and spatial immune cell profiling across various cancer types for identifying genomic alterations and prognostic features.
This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See http://creativecommons.org/licenses/by-nc/4.0/.
Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.