Age-associated remodeling of NK cell subsets, B cell subsets, and T cell surface markers in peripheral blood of healthy Chinese adults: a prospective multicolor flow cytometry study

1 Introduction

Aging is one of the most consistent drivers of immune dysfunction across species. In humans, the progressive remodeling of both innate and adaptive immunity that accompanies aging—collectively termed immunosenescence—underlies much of the increased susceptibility to infection, poor vaccine immunogenicity, and impaired tumor surveillance seen in older individuals (Montecino-Rodriguez et al., 2013; Nikolich-Žugich, 2018). These changes are not merely quantitative; they reflect deep shifts in the composition and functional capacity of circulating lymphocyte populations, and are now recognized as a contributing factor in a range of age-related diseases and geriatric syndromes (Franceschi et al., 2018; GBD 2017 Mortality Collaborators, 2018).

Multicolor flow cytometry has made it possible to capture these shifts in considerable detail. Rather than simply counting T, B, and NK cells, it is now routine to interrogate differentiation state, activation status, and functional marker expression all at once within the same sample (Maecker et al., 2012). From this work, a fairly consistent picture has emerged for T cells: the loss of CD28 co-stimulation, the accumulation of CD57-expressing terminally differentiated cells, and the build-up of TEMRA (terminally differentiated effector memory cells re-expressing CD45RA) are now considered hallmarks of T cell aging (Fülöp et al., 2013; Weng, 2006). CD31 (PECAM-1) deserves particular mention as a marker of recent thymic emigrants (RTEs) — its decline with age is a direct readout of thymic involution and reduced naive T cell output (Ferrando-Martínez et al., 2011). For NK cells, the picture is one of functional rebalancing: the immunoregulatory CD56^hiCD16⁻ subset contracts with age, while the cytotoxic CD56^dimCD16⁺ fraction becomes proportionally dominant (Solana et al., 2012; Hazeldine and Lord, 2013). The atypical CD56⁻CD16⁺ subset has attracted attention in the context of chronic infection and may also expand during normal aging (Solana et al., 2012). In the B cell compartment, the IgD/CD27 four-subset framework has been useful for capturing age-related contraction of switched memory B cells and shifts in the double-negative (IgD⁻CD27⁻) population, sometimes referred to as age-associated B cells (ABCs) (Frasca and Blomberg, 2014; Rubtsov et al., 2011).

Despite the richness of this literature, systematic data from healthy Chinese adults are still relatively sparse (Qin et al., 2016; Chang et al., 2024), and an integrated analysis covering NK cell trisubsets, the IgD/CD27 B cell framework, and terminal T cell differentiation markers (CD57 and TEMRA) in this population has not previously been reported. We therefore designed two complementary multicolor flow cytometry panels and applied them to 146 healthy adults spanning 20–92 years of age, with the goal of building an age-stratified reference dataset for peripheral blood immunophenotyping in the Chinese population.

2 Materials and methods

2.1 Participants

Participants were recruited prospectively between 2023 and 2024 from the annual health examination center of Peking Union Medical College Hospital (PUMCH). Eligibility required age ≥18 years, no clinically significant abnormalities on routine examination, and written informed consent. We excluded individuals with known autoimmune disease, chronic infections (including HIV, HBV, or HCV), hematological disorders, recent use of immunosuppressive agents, glucocorticoids, or biologics (within 3 months), acute infection in the preceding 4 weeks, pregnancy or lactation, or evidence of malignancy or severe organ dysfunction. The study was approved by the PUMCH Institutional Review Board (ethics number: I-24PJ0474).

A total of 146 participants were enrolled across two non-overlapping cohorts. Panel 1 (n = 103) comprised 86 men and 17 women, age range 20–82 years (mean ± SD: 45.3 ± 14.0 years). Panel 2 (n = 43) comprised 29 men and 14 women, age range 25–92 years (mean ± SD: 63.6 ± 19.8 years); this cohort was intentionally enriched for older participants to improve statistical power for detecting age effects at the upper end of the age spectrum. Demographic characteristics are summarized in Table 1.

TABLE 1. Participant demographics.

SD, standard deviation.

2.2 Sample processing

Fasting peripheral venous blood (5 mL) was collected into EDTA-anticoagulated tubes in the morning and processed within 2 h. For surface staining, 200 μL of whole blood was incubated with fluorochrome-conjugated antibodies at room temperature in the dark for 20 min, followed by red blood cell lysis with BD FACS Lysing Solution (1:10 dilution, 10 min at room temperature), centrifugation, and two PBS washes. Intracellular Ki67 staining was performed using the BD Cytofix/Cytoperm kit per manufacturer’s instructions, with anti-Ki-67 (BV421) added after fixation and permeabilization (30 min incubation), followed by washing and immediate acquisition.

2.3 Flow cytometry panels

All samples were acquired on a BD LSRFortessa, with data collection via BD FACSDiva software and analysis in FlowJo v10 (BD Biosciences).

Panel 1 was a 12-color single-tube assay (n = 103) designed to cover T, NK, and B cell subsets simultaneously. For T cells, CD4⁺ and CD8⁺ fractions were assessed together with CD57, Ki67, and PD-1. NK cells were classified into three subsets by CD56/CD16 intensity: CD56hiCD16⁻ (regulatory), CD56d^imCD16⁺ (cytotoxic), and CD56⁻CD16⁺ (atypical), with Ki67 and PD-1 co-assessed. B cells were classified into four subsets by IgD and CD27: IgD⁺ CD27⁻ (naive), IgD⁺CD27⁺ (marginal zone-like), IgD⁻CD27⁺ (switched memory), and IgD⁻CD27⁻ (double-negative/ABCs), with Ki67 included. The gating hierarchy started from CD45⁺ lymphocytes (by FSC/SSC), with CD3⁺ cells gated as T cells, CD3⁻CD19⁺ as B cells, and CD3⁻CD56⁺/CD16⁺ as NK cells. Representative gating strategies for both panels are provided in Supplementary Figure S1. Antibody details are provided in Table 2.

TABLE 2. Panel 1 antibody panel.

All antibodies from BioLegend.

Panel 2 was a 12-color dual-tube assay (n = 43) focused on T cell differentiation state. Tube 1 (T cell tube) combined CD45RA (FITC), CD62L (APC), CD28 (PE/Cy7), CD38 (BV605), HLA-DR (BV510), CD31 (PE), CD4 (PerCP-Cy5.5), CD8a (APC/Cy7), CD57 (BV711), Ki67 (BV421), PD-1 (PE/Dazzle 594), and CD45 (AF700). Within the CD4⁺ and CD8⁺ gates, CD45RA/CD62L allowed resolution of four subsets: naive (CD45RA⁺ CD62L⁺), central memory (TCM, CD45RA⁻CD62L⁺), effector memory (TEM, CD45RA⁻CD62L⁻), and TEMRA (CD45RA⁺ CD62L⁻); the TEMRA fraction was calculated as total CD45^RA+ minus naive. Tube 2 (B/NK tube) used the same antibody configuration as Panel 1. At least 100,000 lymphocyte events were acquired per tube. Antibody details for both panels are provided in Tables 2 and 3.

TABLE 3. Panel 2 antibody panel (dual-tube).

Tube 1, T cell tube; Tube 2, B/NK, cell tube. Shared markers: CD45-AF700, CD4-PerCP-Cy5.5, CD8a-APC/Cy7, PD-1-PE/Dazzle 594, Ki67-BV421, CD57-BV711. All antibodies from BioLegend.

2.4 Statistical analysis

Statistical analysis was performed in SPSS 26.0 and Python 3.10 (scipy, pandas). Continuous variables are reported as median (interquartile range, IQR). Age associations were assessed by Spearman rank correlation, reported as r and two-tailed p values. Participants were stratified into three age groups: young (20–44 years), middle-aged (45–64 years), and older (≥65 years); between-group comparisons used the Kruskal–Wallis test with Dunn post hoc testing and Bonferroni correction where indicated. Sex differences were tested by Mann-Whitney U test. Multivariate analysis employed partial Spearman correlations controlling for sex, supplemented by ordinary least squares (OLS) regression with standardized age and sex as covariates. A two-tailed p < 0.05 was considered statistically significant.

3 Results

3.1 Study population

The two cohorts together comprised 146 healthy adults with no participant overlap. Panel 1 covered ages 20–82 years (mean 45.3 ± 14.0), with 52 young (50.5%), 39 middle-aged (37.9%), and 12 older participants (11.7%); 83.5% were male. Panel 2 spanned ages 25–92 years (mean 63.6 ± 19.8), with 62.8% older participants and a somewhat higher proportion of women (32.6%). Detailed demographic data appear in Table 1.

3.2 T cell changes in panel 1 (n = 103)

The two strongest age correlations in Panel 1 were both CD57-related: CD4⁺CD57⁺ (r = +0.338, p < 0.001) and CD8⁺CD57⁺ (r = +0.308, p = 0.002) rose consistently with age, reflecting progressive accumulation of terminally differentiated T cells across both subsets (Figure 1).

Beyond CD57, the CD4⁺ fraction (as a proportion of T cells) declined with age (r = −0.218, p = 0.027), while the CD8⁺ proportion rose correspondingly (r = +0.155, p = 0.117), driving a downward shift in the CD4/CD8 ratio. Overall T cell frequency also declined (r = −0.256, p = 0.009). CD4⁺Ki67⁺ showed a modest but significant increase (r = +0.201, p = 0.042), consistent with compensatory homeostatic proliferation. CD4⁺PD-1⁺ did not reach significance within Panel 1 alone (p = 0.629); CD8⁺PD-1⁺ however showed a significant positive correlation (r = +0.230, p = 0.019). When all 146 participants were pooled for the shared markers, both CD4⁺PD-1⁺ (r = +0.291, p < 0.001) and CD8⁺PD-1⁺ (r = +0.367, p < 0.001) became significantly positive, suggesting that statistical power rather than a true null effect explains the Panel 1 CD4⁺PD-1⁺ finding (see Table 4).

TABLE 4. Spearman correlations with age for Panel 1 parameters (n = 103 or n = 146).

ABCs, age-associated B cells; * p < 0.05; p < 0.01; * p < 0.001; ns, not significant.

CD3⁺ T cells (% lymphocytes) and IgD⁺CD27⁺ marginal zone-like B cells were available only in Panel 1 (n = 103); all other parameters were analyzed in the full 146-participant dataset (Panel 1 n = 103 + Panel 2 n = 43) for shared baseline markers.

3.3 T cell changes in panel 2

3.3.1 CD8⁺ T cells

The Panel 2 data told a particularly clear story for CD8⁺ T cells (Figure 2). TEMRA accumulation (r = +0.532, p < 0.001), CD28 loss (r = −0.552, p < 0.001), and naive cell depletion (r = −0.534, p < 0.001) formed a tight triad of age-associated changes, all reaching p < 0.001 (Table 5). Together they paint a picture of progressive, co-stimulation-independent terminal differentiation that dominates the CD8⁺ compartment in older individuals. Activation and proliferation markers reinforced this: CD8⁺HLA-DR⁺ (r = +0.370, p = 0.015), CD8⁺CD38⁺HLA-DR⁺ (r = +0.363, p = 0.017), and CD8⁺Ki67⁺ (r = +0.375, p = 0.013) all rose with age, indicating that despite their differentiated phenotype, CD8⁺ T cells in older adults remain in a state of chronic activation. Total CD8⁺ memory frequency (CD45RA⁻) showed no net change, nor did CD8⁺CD57⁺ or CD8⁺PD-1⁺ within this smaller cohort.

TABLE 5. Spearman correlations with age for Panel 2 T cell parameters (n = 43).

TEMRA, terminally differentiated effector memory T cells re-expressing CD45RA; RTE, recent thymic emigrants. * p < 0.05; p < 0.01; * p < 0.001; ns, not significant.

3.3.2 CD4⁺ T cells

Two significant age-related changes were identified in the CD4⁺ compartment (Figure 3). First, CD4⁺CD31⁺ cells declined with age (r = −0.361, p = 0.017); as a marker of recent thymic emigrants, this drop is consistent with thymic involution progressively curtailing new naive T cell output in older individuals. Second, CD4⁺ naive cells (CD45^RA+ CD62L⁺) also showed a significant age-related decline (r = −0.303, p = 0.048), reinforcing the picture of progressive naive T cell attrition. None of the other CD4⁺ parameters—including TEMRA, memory, CD28⁺, CD38⁺, HLA-DR⁺, CD57⁺, Ki67⁺, or PD-1⁺ fractions—showed a statistically significant age relationship in this cohort (Figures 4, 5).

3.4 NK cell changes

Panel 1 (n = 103) showed a clear age-associated rise in total NK cell frequency (r = +0.259, p = 0.008) (Figure 6). Within that expanding NK pool, however, the CD56hiCD16⁻ regulatory subset contracted (r = −0.263, p = 0.007), while CD56dimCD16⁺ and CD56⁻CD16⁺ showed no significant change. NK PD-1 expression also rose with age (r = +0.256, p = 0.009). NK Ki67⁺ was unchanged (p = 0.942).

In the combined dataset (n = 146), segmented Spearman analysis with age 65 as the breakpoint showed that CD56⁻CD16⁺ NK cells were non-significant overall (r = +0.133, p = 0.109) but reached significance in participants aged ≥65 years (n = 39, r = +0.482, 95% CI [0.196, 0.692], p = 0.002), compared with non-significance in those under 65 (n = 107, r = +0.123, p = 0.206). The formal slope-change test did not reach significance (p = 0.094), so this pattern is best described as a late-life tendency requiring replication rather than a confirmed threshold effect. For CD56^dimCD16⁺ NK cells, neither age segment showed a significant association in the combined analysis. In Panel 2, NK Ki67⁺ showed a significant positive correlation (r = +0.420, 95% CI [0.137, 0.640], p = 0.005). Segmented regression results are shown in Figure 7.

3.5 B Cell changes

Total B cell frequency was stable across the Panel 1 age range (r = +0.003, p = 0.973) (Figures 8, 9). Among the four IgD/CD27 subsets, IgD⁺CD27⁺ marginal zone-like B cells showed a significant age-related decline in Panel 1 (r = −0.257, p = 0.009). None of the other three subsets—naive, switched memory, or double-negative/ABCs—reached significance in Panel 1, nor did B Ki67⁺.

Panel 2 presented a more pronounced picture at older ages: total B cell frequency declined significantly (r = −0.381, p = 0.012) and switched memory B cells contracted (r = −0.379, p = 0.012). That these changes appeared in an older-enriched cohort (median age 70, maximum 92) but not in the broader Panel 1 sample suggests that substantial B cell and memory contraction may be a relatively late phenomenon, becoming apparent mainly in the oldest age ranges, consistent with functional data from Frasca and colleagues (Frasca and Blomberg, 2014).

3.6 Sex effects and multivariate analysis

In Panel 1 (n = 103), five parameters differed significantly by sex (Mann-Whitney U, p < 0.05). Women had lower CD4⁺ fractions (p = 0.011) and higher CD8⁺ fractions (p = 0.003) than men, suggesting a modestly more differentiated T cell phenotype in women at a given age. Men had higher CD8⁺PD-1⁺ (p = 0.040). B cell sex differences were also apparent: men had higher naive B (p = 0.028) and ABC proportions (p < 0.001) than women. In Panel 2, five parameters differed by sex: women had higher total CD8⁺ (p = 0.010), CD4⁺CD31⁺ (p = 0.015), and CD8⁺CD31⁺ (p<0.001) fractions, while men had higher NK frequency (p = 0.023) and CD8⁺ memory fractions (p = 0.013).

Multivariate analysis (partial Spearman correlations controlling for sex, plus OLS regression) confirmed that the vast majority of age-related findings were independent of sex. In Panel 1, all ten significant univariate associations held after sex adjustment, including CD4⁺CD57⁺ (partial r = +0.337, p < 0.001), CD8⁺CD57⁺ (partial r = +0.315, p = 0.001), CD4 T% (partial r = −0.217, p = 0.028), NK% (partial r = +0.265, p = 0.007), CD56^hiCD16⁻ NK% (partial r = −0.267, p = 0.006), total T% (partial r = −0.258, p = 0.009), NK PD-1⁺ (partial r = +0.266, p = 0.007), CD8⁺PD-1⁺ (partial r = +0.243, p = 0.013), CD4⁺Ki67⁺ (partial r = +0.202, p = 0.041), and MZ-like B (partial r = −0.258, p = 0.009). In Panel 2, all three core CD8⁺ metrics—TEMRA (partial r = +0.537), naive (partial r = −0.547), and CD28⁺ (partial r = −0.551) — remained strongly significant (all p < 0.001) after controlling for sex.

3.7 Immunological Composite Score

To integrate the multidimensional age signal into a single summary measure, we constructed an Immunological Composite Score (ICS) from nine parameters that showed consistent, sex-independent age associations across both panels: CD4⁺CD57⁺, CD8⁺CD57⁺, CD4⁺PD-1⁺, CD8⁺PD-1⁺, NK%, and NK PD-1⁺ (each contributing positively) and CD4% T cells, CD56hiCD16⁻ NK%, and switched memory B% (each contributing negatively). Each parameter was z-scored and combined with directional signs aligned to the age correlation direction.

Across all 146 participants, the ICS correlated strongly with chronological age (Spearman r = +0.575, p < 0.001). Kruskal–Wallis testing demonstrated highly significant group separation (H = 49.78, p < 0.001). Dunn post hoc comparisons showed that the older group (≥65 years; median ICS +4.85, IQR +2.45 to +6.53) differed significantly from both the young group (20–44 years; median −2.39, IQR −5.27 to +0.34; p < 0.001) and the middle-aged group (45–64 years; median −0.78, IQR −2.50 to +1.82; p < 0.001), while the young and middle-aged groups did not differ significantly from each other (p = 0.486). The ICS is presented as a proof-of-concept instrument; external prospective validation is required before clinical application (Figure 10).

3.8 Cross-lineage senescence correlations

To assess whether age-associated immune changes were coordinated across lineages within the same individuals, we performed pairwise Spearman correlations between senescence markers from different compartments. In Panel 1 (n = 103), the strongest cross-lineage association was between CD4⁺CD57⁺ and CD8⁺CD57⁺ (r = +0.469, p < 0.001), indicating that terminal T cell differentiation proceeds in parallel across both subsets. CD4⁺CD57⁺ was also inversely correlated with the CD56^hiCD16⁻ regulatory NK subset (r = −0.320, p = 0.001). Positive associations were observed between CD8⁺PD-1⁺ and DN B/ABCs (r = +0.294, p = 0.003), between CD8⁺CD57⁺ and NK PD-1⁺ (r = +0.288, p = 0.003), and between NK PD-1⁺ and DN B/ABCs (r = +0.255, p = 0.009). In Panel 2 (n = 43), CD8⁺ TEMRA accumulation was tightly coupled with CD8⁺CD28 loss (r = −0.775, p < 0.001) and with CD8⁺CD57⁺ accumulation (r = +0.557, p < 0.001), and NK PD-1⁺ correlated positively with CD56⁻CD16⁺ NK expansion (r = +0.327, p = 0.032). Together, these findings suggest that immune aging is partially coordinated across T cell, NK cell, and B cell compartments within individuals (Figure 11; Supplementary Tables S5 and S6).

4 Discussion

This study characterizes age-associated changes in peripheral blood NK cell subsets, B cell subsets, and T cell surface markers across a broad age spectrum (20–92 years) in 146 healthy Chinese adults, providing a multiparameter immunophenotyping reference dataset for this population.

The T cell data are arguably the most informative. In Panel 2, the shift of CD8⁺ T cells toward TEMRA (r = +0.532) was the single strongest age correlation we observed, occurring in concert with CD28 loss (r = −0.552) and naive cell depletion (r = −0.534). The three changes form a coherent triad—the hallmarks of replicative senescence in the CD8⁺ compartment—consistent with established features of T cell aging and with the epigenetic senescence signature of TEMRA cells documented in population-level studies (Fülöp et al., 2013; Salumets et al., 2022; Goronzy and Weyand, 2013). Layered onto this, both CD4⁺CD57⁺ and CD8⁺CD57⁺ rose significantly in Panel 1, reinforcing terminal differentiation as a dominant and lineage-wide feature of immune aging in this cohort. The concurrent decline of CD4⁺ naive cells (r = −0.303) and CD4⁺CD31⁺ RTE-like cells (r = −0.361) points to a parallel process: progressive thymic involution curtailing new naive T cell output, as anticipated from the known age-related loss of thymic function (Ferrando-Martínez et al., 2011). When the full 146-participant sample was analyzed for shared markers, both CD4⁺PD-1⁺ and CD8⁺PD-1⁺ showed robust age correlations (r = +0.291 and +0.367 respectively), consistent with the view that exhaustion-like checkpoint upregulation accompanies terminal differentiation as a co-occurring feature of immune aging in this population (Chang et al., 2024).

The NK cell findings are more nuanced. The overall rise in NK frequency with age, while the CD56hiCD16⁻ regulatory subset contracts, is consistent with the shift toward cytotoxic NK predominance described in other populations (Hazeldine and Lord, 2015). In the combined dataset (n = 146), segmented analysis revealed that CD56⁻CD16⁺ NK cells showed no significant overall age correlation but became significantly positively associated with age in participants aged ≥65 years (r = +0.482, 95% CI [0.196, 0.692], p = 0.002), compared with a non-significant association in those under 65 (r = +0.123, p = 0.206). The formal slope-change test did not reach significance (p = 0.094), so this pattern is best interpreted as a late-life tendency requiring replication in larger cohorts. This atypical subset was first characterized in the context of chronic HIV infection (Solana et al., 2012), and its selective expansion in the oldest participants likely reflects decades of cumulative antigenic exposure rather than frank immunodeficiency. The age-related rise in NK PD-1 expression—which emerged clearly in Panel 1 (r = +0.256) — is a less commonly reported observation and suggests that NK cells, like T cells, may undergo progressive phenotypic exhaustion with age. The mechanisms underlying this rise remain to be established, but likely include chronic low-grade inflammation (inflammaging) upregulating inhibitory receptors, cumulative stimulation by latent herpesviruses such as CMV driving an exhaustion-like NK phenotype, and replicative exhaustion following decades of immune activity. PD-1-expressing NK cells exhibit reduced cytotoxic capacity and impaired cytokine production; accordingly, age-related accumulation of PD-1⁺ NK cells may contribute to the diminished antitumor NK surveillance reported in older adults. Our NK subset findings are broadly consistent with observations from European cohorts (Le Garff-Tavernier et al., 2010), suggesting that the key features of NK cell aging may be conserved across populations.

B cell aging in this cohort followed a somewhat different timeline. In Panel 1, total B cell frequency was stable and the IgD/CD27 subsets were mostly unchanged, with only IgD⁺CD27⁺ marginal zone-like B cells showing a significant decline (r = −0.257). This subset represents antigen-experienced but unswitched B cells, and its contraction may reflect reduced T-independent responses with aging. The more dramatic changes—total B cell contraction and switched memory decline—emerged in Panel 2, in the oldest participants. Taken together, these observations suggest that B cell aging may be a relatively late-stage process, with meaningful contraction of the humoral memory compartment concentrated in the seventh decade and beyond, consistent with functional data from Frasca and colleagues (Frasca and Blomberg, 2014).

Sex was a meaningful source of variation throughout. Women in our cohort tended to have higher CD8⁺ fractions and lower CD4/CD8 ratios than men of the same age, a pattern in line with well-established sex differences in immune aging (Klein and Flanagan, 2016). Both panels showed higher CD31⁺ fractions in women, consistent with greater thymic output or slower naive T cell attrition in women. Men had higher NK frequency in Panel 2. Critically, however, multivariate analysis showed that sex did not explain the primary age effects; all ten significant age correlations in Panel 1 survived adjustment for sex, and the three core CD8⁺ markers (TEMRA, naive, CD28⁺) remained strongly significant regardless. The male-heavy composition of Panel 1 (83.5%) does introduce some interpretive caution, and we would encourage sex-stratified reporting when these data are used as clinical reference ranges. Specifically, the reference values reported here should be applied with caution to female adults, and sex-stratified reference intervals should be established in future studies with more balanced cohort recruitment.

This dataset sits alongside existing reference work from the same institution (Qin et al., 2016; Chang et al., 2024), which established total lymphocyte subset counts and described age trends for T cell markers including PD-1, Ki67, and CD31. The present data add NK cell three-subset classification, IgD/CD27 B cell four-subset classification, CD57 quantification, and TEMRA profiling—elements that together enable a more complete picture of peripheral immune aging in the Chinese adult population.

The ICS introduced in this study demonstrates that the multidimensional age signal captured by a routine 12-color clinical flow cytometry panel can be condensed into a single interpretable composite index. This approach requires no additional reagents or instrumentation beyond what is already used for standard lymphocyte subset analysis, making it practically accessible in clinical settings where multiparameter immunophenotyping is routinely performed. The ICS showed strong age correlation (r = +0.575) and clear separation between young and older groups, consistent with an acceleration of immune aging in the sixth decade and beyond. We emphasize, however, that the ICS was derived from and validated in the same dataset, introducing circularity that limits its current clinical utility. External prospective validation in an independent cohort—ideally one that includes longitudinal follow-up and clinical endpoints such as infection susceptibility or vaccine immunogenicity—is required before the ICS can be used as a clinical instrument.

Several limitations should be acknowledged. Panel 2 was small (n = 43) and skewed toward older participants, which improves sensitivity for age-related signals but limits generalizability of some findings. CMV serostatus was not collected, which is an acknowledged limitation. However, CMV is best understood as a mediating variable in the age→immune remodeling pathway rather than a simple external confounder. In Chinese adults, CMV seroprevalence exceeds 90% by the sixth decade (Noppert et al., 2024), meaning that latent CMV infection is itself an age-dependent biological reality rather than an independent exposure that can be cleanly removed. Evidence from both serological and epigenetic studies indicates that CMV-driven immune remodeling operates through the same senescence pathways activated by intrinsic aging (Salumets et al., 2022; Pawelec et al., 2012; Rodríguez and Parra-López, 2024; Sharpe et al., 2022), and the two processes are deeply intertwined in a seropositive population. For a reference dataset intended to reflect the immune phenotype as it actually exists in healthy Chinese adults, characterizing the landscape inclusive of CMV-associated remodeling is therefore appropriate. That said, future studies incorporating CMV serostatus data would allow stratified analyses and more precise attribution of individual immune changes. The cross-sectional design also means we cannot formally distinguish age effects from cohort effects. Notably, however, all participants underwent systematic health screening and none showed clinical signs of accelerated aging or pathological immune activation; the phenotypic distributions observed are consistent with the expected trajectory of physiological immune remodeling, which provides some reassurance that the signals captured here reflect normal aging rather than disease-driven effects. The sex imbalance in Panel 1 remains a caveat for some of the sex-related analyses. Additionally, this study assessed surface phenotype and proliferation markers only; functional readouts such as cytokine secretion (IFN-γ, TNF-α) and cytotoxicity (CD107a, granzyme B) were not included. The functional consequences of the phenotypic changes described here therefore remain to be directly established. Finally, serum immunoglobulin levels were not measured, which prevents us from determining whether the late-stage B cell phenotypic changes observed in our oldest participants are accompanied by corresponding shifts in antibody class distribution or total immunoglobulin output.

5 Conclusion

Across 146 healthy Chinese adults spanning two to 9 decades of life, we found consistent and largely sex-independent evidence of immune remodeling in all three major lymphocyte compartments. In the T cell pool, CD8⁺ TEMRA accumulation was the dominant signal, accompanied by CD28 loss, naive depletion, and CD57 accumulation in both CD4⁺ and CD8⁺ cells; the parallel decline of CD4⁺ naive cells and CD4⁺CD31⁺ RTE-like cells points to progressive involution of naive T cell generation. NK cells showed age-related contraction of the CD56hiCD16⁻ regulatory subset, expansion of atypical CD56⁻CD16⁺ cells in the oldest participants, and rising PD-1 expression. B cell changes were more modest at younger ages but included IgD⁺CD27⁺ marginal zone-like B cell decline in Panel 1 and switched memory contraction in the older cohort. Together these findings define a multidimensional aging trajectory of the peripheral immune system in healthy Chinese adults, and provide a reference framework for clinical immunophenotyping and vaccine response research.

Generative AI statement

The author(s) declared that generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Supplementary material

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fragi.2026.1847033/full#supplementary-material

Supplementary material Supplementary material at the source

Supplementary material Supplementary material at the source

Supplementary material Supplementary material at the source

Footnotes

Edited by: Keon-Il Im, LucasBio, Republic of Korea

Reviewed by: Peijie Yu, The Hong Kong Polytechnic University, Hong Kong SAR, China

Masakatsu Yamashita, Ehime University, Japan