Background: Aging is a key risk factor for oncogenesis as cellular and genomic changes accumulate over the course of the lifespan; however, some individuals remain cancer-free even at advanced age. Considerable molecular variation exists even among aged cancer-free individuals, which can be captured by gene expression patterns in blood and could contribute to differences in health risk within this population. Mapping the mutational and transcriptional landscape of healthy aging thus provides a unique model of cancer avoidance and could point towards actionable targets for cancer prevention.

Objectives: We sought to uncover heterogeneity within the blood compartment both across and within age groups among prevalent cancer-free participants, but who in some cases have gone on to develop incident cancers, to define axes of variation in healthy states which may be prognostic for future cancer outcomes.

Methods: We conducted single cell RNA sequencing (scRNAseq), whole genome sequencing, and epigenomic sequencing (ATAC-seq) of whole blood from ~400 individuals in the Ontario Health Study (OHS), who were stratified by age (young <45 years, aged >65 years) and the Intermountain Risk Score (low-risk, high-risk), which is correlated with all-cause mortality. These stratifications defined four quadrants of participants, including young low-risk, young high-risk, aged low-risk, and aged high-risk, with no participants having prevalent conditions. We defined cell type-specific gene expression modules using Hotspot and scored pseudobulk profiles across age and risk in both OHS and replicated these findings in a separate longitudinal scRNAseq cohort - the Sound Life Cohort. Finally, we associated gene module enrichment with the number of somatic point variants called within each cell type.

Results: We discovered differentially expressed genes (DEGs) in a cell type-specific manner, comparing low-risk vs high-risk groups in both young and aged participants. 21 cell type-specific gene expression modules were identified across the hematopoietic hierarchy, of which a metabolism-related module classified a subtype of Naive T cells and included genes such as PHGDH, ANTKMT, and NT5E, was upregulated in young samples (p=2.23e-30, Wilcoxon rank-sum test). In the Sound Life cohort, the same Naive T cell metabolism-related module enrichment trended towards an increase in young compared to aged individuals across the sampling years 2020-2022, with a significant difference in 2021 (p=0.020, Wilcoxon rank-sum test). A separate module in B cells involved in transcriptional regulation with genes such as TLE4, CPEB4, and GATAD1 was significantly upregulated in low-risk individuals (p=0.014, Wilcoxon rank-sum test). Gene module enrichment was positively associated with cell type-specific somatic mutation burden, for certain B cell, CD14 monocyte, Naive T cell, and NK cell modules, while mutation burden was negatively associated with other CD14 monocyte, Naive T cell, and NK cell modules. 

Conclusions: Here, we uncovered signatures defining abnormal blood profiles as well as aging across immune cell types in a pre-disease setting, and present molecular vignettes from which parallel hypotheses could be drawn about cancer initiation. These signatures may be leveraged for transcriptome-based risk stratification for cancer incidence using blood, an easily accessible, dynamic tissue amenable to population-wide sampling.