Compositional analysis of lymphocytes and their relationship with health outcomes: findings from the health and retirement study.

Abstract

Background: Immunosenescence, the gradual deterioration of the immune system, is critical for aging-related diseases. However, the lack of detailed population-level immune data has limited our understanding, underscoring the need for innovative analytical approaches. The Health and Retirement Study (HRS) in the United States provides a unique opportunity to examine T and B lymphocyte subsets using compositional data analysis and dimension reduction techniques.

Methods: We constructed a hierarchical tree structure to map relationships among T and B subset cells in HRS. Network analysis examined conditional dependence across 16 immune subset cells, while stepwise redundancy analysis (SRDA) identified a subset of pairwise logratio measures that capture main variance in immune composition. We conducted two sets of supervised learning analyses: first, linear penalized log-contrast models to examine the associations between subset cells and three health outcomes (chronic disease index, self-reported health, and frailty level); second, linear regressions to examine the associations between the top selected logratios and health outcomes.

Findings: Our study included 6,250 participants from the HRS with a median age of 68. Network analysis showed some dependence among 16 immune subset cells, including associations between central memory CD4 + T cells and both other CD4 + T cells and other lymphocytes, as well as between central memory CD8 + T cells and other CD8 + T cells. SRDA identified nine key log-ratio measures, explaining over 90% of the variance in immune composition. Linear penalized log-contrast models showed that a lower proportion of naïve CD4 + T cells and higher proportions of other CD4 + and central memory CD8 + T cells were significantly associated with greater chronic disease burden, poorer self-reported health, and higher frailty levels. Linear regression models using log-ratios reinforced these patterns, showing that a higher ratio of other lymphocytes over naïve CD4 + T cells and terminally differentiated effector memory CD4 + T cells over other CD8 + T cells were associated with greater chronic disease burden, poorer self-reported health, and higher frailty levels. In contrast, a higher ratio of other lymphocytes over central memory CD4 + T cells was associated with better health outcomes.

Interpretation: Our findings highlight the value of a systems-based approach and compositional analysis in understanding immunosenescence and its impact on health. The identified subset cells and logratio measures provide meaningful insights into immune aging and warrant further investigation to explore their long-term relationships with health outcomes.