Peripheral adaptive immune cell counts and mortality in adult sepsis: a systematic review and meta-analysis.

Lower peripheral lymphocyte counts are associated with mortality in sepsis, but evidence across related adaptive immune-cell compartments has not been systematically quantified. Nine databases were searched from inception through 7 August 2026 using expanded controlled-vocabulary and free-text terms for sepsis, lymphocyte/T-cell subsets, absolute cell counts, and mortality. Observational studies comparing absolute peripheral CD4+ T-cell, CD8+ T-cell, CD3+ T-cell, or absolute lymphocyte counts (ALCs) between adult non-survivors and survivors were included. Random-effects mean differences (MDs) were estimated using restricted maximum likelihood with a modified Hartung-Knapp adjustment. Prediction intervals, restricted and leave-one-out analyses, ratio-of-means analyses, Egger tests, and exploratory meta-regression were also performed. Results: Thirteen studies were included. Counts were lower in non-survivors for CD4+ T cells (13 studies; MD = -86.06 cells/µL, 95% CI -127.29 to -44.83; I² = 77.0%), CD8+ T cells (11 studies; MD = -41.83, 95% CI -68.37 to -15.29; I² = 61.9%), CD3+ T cells (10 studies; MD = -133.99, 95% CI -220.41 to -47.56; I² = 88.3%), and ALCs (10 studies; MD = -201.46, 95% CI -323.04 to -79.88; I² = 71.7%). All prediction intervals crossed zero. CD4 remained lower after exclusion of studies classified as high risk in the descriptive QUIPS synthesis; corresponding CD8, CD3, and ALC estimates remained negative but were imprecise because only four, two, and three lower-risk studies remained. After excluding statistically converted data, CD8 was no longer significant and ALC evidence was based on only two studies. Adjusted analyses varied across studies and did not consistently support independent prognostic associations. Evidence robustness was greatest for CD4; CD8, CD3, and ALC were more sensitive to data conversion, heterogeneity, or sparse lower-risk evidence. Adult sepsis non-survivors had lower absolute counts across lymphocyte and T-cell compartments. The findings are most consistent for CD4, whereas CD8 and ALC estimates are more dependent on converted data. Heterogeneity, geographic concentration, and prediction intervals crossing zero limit direct clinical generalization and support the need for standardized longitudinal studies with severity-adjusted analyses. https://www.crd.york.ac.uk/PROSPERO/, identifier CRD420261446443.