Causal association between genetically predicted circulating immune cell counts and frailty: a two-sample Mendelian randomization study

Despite the recognized link between immune responses and frailty, the association between immune cell counts and frailty based on previous observational studies remains disputed, with uncertain causal nexus. This study aimed to elucidate causal association between genetically predicted circulating immune cell counts and frailty. We conducted the two-sample Mendelian randomization (MR) study with independent genetic variants associated with six immune cell subtype counts from genome-wide association studies in 563,946 European individuals. Frailty summary data, assessed via frailty index (FI), was obtained from study comprising 175,226 subjects. Univariate MR, reverse MR and multivariate MR were conducted to comprehensive investigate the association between immune cell counts and FI, with two-step MR analysis for mediation analysis. Univariate MR evidence indicated that among six leukocyte subtype counts, only elevated eosinophil count was significantly correlated with higher FI (β = 0.059, 95% confidence interval [CI], 0.042-0.078, =5.63E-11), with no reverse causal relationship identified in reverse MR. In multivariate MR, the causal effect of eosinophil count retained statistical significance (β = 0.063, 95% CI, 0.021-0.104, = 0.003). Ultimately, the two-step MR analysis demonstrated two mediators in this causal pathway: asthma (β= 0.019, 95% CI, 0.013-0.025, = 35.84E-10, mediated proportion, 31.732%) and rheumatoid arthritis (β= 0.004, 95% CI, 0.001-0.006, =1.75E-03, mediated proportion, 6.411%). Within immune cell subtypes, MR evidence indicated only genetically predicted circulating eosinophil count had irreversible and independent causal effect on frailty, with asthma and rheumatoid arthritis possibly serving as partial mediators. The finding stressed the need for further exploring physiological functions of eosinophils in order to develop effective strategies against frailty.