Predictive advantage of age and baseline CD4+T cell count in immune status

A multiple regression model was employed to assess the explanatory power of four clinical categories on both the CD4+T cell count and the CD4/CD8 ratio. The most suitable model was identified based on a ΔAICc ≤ 2 [21]. The findings indicated that clinical factors exerted a more substantial influence on the CD4+T cell count (R² = 0.53) compared to the CD4/CD8 ratio (R² = 0.18). Age and baseline measurements accounted for 61.9% and 24.8% of the variance in the CD4+T cell count, respectively, which was greater than the contributions of hematological parameters (10.6%) and treatment details (2.7%). Among these factors, age at examination and baseline CD4+T cell count were identified as the most significant contributors to the variance in the CD4+T cell count (p < 0.01) (Fig. 1a). Our findings were consistent with previous studies, highlighting the critical role of baseline CD4+T cell counts and age in immune recovery [22,23,24]. The ability of baseline CD4+T cell count and age at examination to predict the risk of low CD4+T cell count and low CD4/CD8 was compared separately by the ROC analysis. A larger area under the curve (AUC) indicates better discriminant power [25]. Figure 1b shows that the baseline CD4+T cell count had a significantly better ability to distinguish low CD4+T cell count and low CD4/CD8 than did age at examination, with an AUC above 0.7 after 4 years of HAART, whereas the AUC for age at examination never exceeded 0.6 after 7 years of HAART. The results indicated that the baseline CD4+T cell count was a more reliable predictor of immune status than the age at examination. A closer inspection of the results revealed that the ROC curves of the baseline CD4+T cell count in predicting the risk of low CD4+T cell count and a low CD4/CD8 ratio were significantly influenced by age. Specifically, only PLWH over the age of 60 years exhibited an AUC greater than 0.9, indicating a strong predictive value. However, as the age of the PLWH decreased, the AUC decreased, dropping to an AUC lower than 0.6 for individuals under the age of 40 years (Fig. 1c). Similarly, the ROC curves for HGB, PLT, and Cr in predicting low CD4+T cell and low CD4/CD8 risk were significantly influenced by age at examination to various degrees (Figure S1). These observations further indicated that the considerable variability in the CD4+T cell count caused by age at examination is primarily mediated by synergistic effects with other clinical indicators.

Fig. 1

Effect of age on immune reconstitution in PLWH. (a) Relative effect of clinical factors on CD4+T cell count and CD4/CD8 ratio. Variables are grouped into 4 components: red = age characteristics (age at diagnosis, age at treatment initiation, and age at examination), green = baseline immune outcomes (CD4+T, CD8+T cell counts, and CD4/CD8 ratio), blue = hematological parameters (WBC, HGB, PLT, T.BIL, ALT, AST, Cr, Glu, TG, TC), purple = treatment details (treatment interval, treatment years, initial and current treatment plan). The averaged parameter estimates (standardized regression coefficients) for the model predictors are presented alongside their corresponding 95% confidence intervals and the relative importance of each predictor, expressed as a percentage of the explained variance. The graph illustrates the best model selected based on the AICc criterion. The relative effects of the predictors and their interactions can be calculated by taking the ratio of the parameter estimate of each predictor to the sum of all parameter estimates, with the result expressed as a percentage. (b) Time-dependent AUCs for baseline CD4+T cell count and age to predict the risk of CD4+T cell count <350 cells/µL and CD4/CD8 < 0.4. Shaded region indicates the 95% confidence interval. (c) ROC curve analysis for predicting the risk of low CD4+T cell or low CD4/CD8 based on baseline CD4+T cell counts across different age groups (18–30, 30–40, 40–50, 50–60 and > 60 years old). (d) Dynamics of CD4+T cell counts, CD4/CD8 ratios, relative CD4+T cell counts, and relative CD4/CD8 ratios over the 7-year treatment period in the adult group (age < 35 years, n = 392) and the older group (age > 55 years, n = 396). Relative CD4+T cell count or relative CD4/CD8, represents the increase in the number of CD4+T cells or CD4/CD8 relative to baseline after HAART. Data are shown as means ± SEs. The p values are calculated by two-way ANOVA. * p < 0.05, ** p < 0.01; *** p < 0.001. WBC: white blood cells, HGB: hemoglobin, PLT: platelets, T.BIL: total bilirubin, ALT: alanine aminotransferase, AST: aspartate aminotransferase, Cr: creatinine, Glu: glucose, TG: triglycerides, TC: total cholesterol, AICc: akaike information criterion corrected, AUC: area under the curve, ROC: receiver operating characteristic curve, HAART: highly active antiretroviral therapy

Delayed and limited immune reconstitution of CD4+T cells in older PLWH

This study investigates the impact of age on immune reconstitution by evaluating the dynamics of CD4 + T cell counts and CD4/CD8 recovery in adults aged < 35 years (n = 392) and older PLWH aged > 50 years (n = 396) over a 7-year follow-up period. The participant’s characteristics are detailed in Table S2. Our findings indicate that old PLWH had a significantly lower baseline CD4+T cell count (246.2 cells/µL) compared to adult PLWH (402.8 cells/µL, p < 0.0001). While HAART effectively restored CD4+T cell counts across the different age groups, more pronounced changes were observed in the adult cohort compared to the old PLWH. Notably, CD4+T cell counts were consistently higher in the adult than in the old throughout the 7 years of follow-up (p < 0.0001). To account for differences in baseline CD4+T cell counts, we calculated relative CD4 + T cell counts, representing the increase in CD4+T cells following HAART relative to baseline. Results showed that relative CD4+T cell counts in the adult group consistently outperformed those in the old group. Further analyses revealed that the differences between the two groups reached statistically significant levels at the 1st, 2nd, 4th, and 5th year follow-up time points. Additionally, over the 5-year HAART period, the old PLWH exhibited a slower growth rate (211.8 cells/µL/year) compared to the adult group (266.3 cells/µL/year, p < 0.0001). It took 5 years for the overall CD4+T cell count in old PLWH to exceed 500 cells/µL, whereas adult PLWH reached this milestone after only 1 year of HAART. Similarly, the baseline CD4/CD8 ratio was significantly lower in the old (0.29) than in the adult (0.41, p = 0.00869); With the use of HAART, there was a gradual increase in the CD4/CD8 ratio, which remained consistently higher in the younger group than in the older group, achieving statistical significance at 2, 3, and 4 years after HAART. However, the relative CD4/CD8 were not significantly different, and neither group achieved a mean CD4/CD8 ratio exceeding 0.8 after HAART (Fig. 1d). These results revealed that older PLWH tend to have lower CD4+T count at baseline and poorer recovery of CD4+T cell count following HAART than younger counterparts, and the differences in CD4+T cell recovery between age groups increased over time, indicating that the negative impact of older age on CD4+T cell recovery becomes more pronounced with extended follow-up.

The characteristics of study population

A total of 146 PLWH were recruited and the participant’s characteristics are detailed in Table 1. Participants were categorized into distinct groups based on their age and HAART status. The study included PLWH aged between 18 and 83 years. Among those receiving HAART, the ages for the adult, middle-aged, and old groups were 28, 41, and 58 years, respectively. In contrast, for the untreated PLWH, the mean ages for the adult, middle-aged, and old groups were 23, 44, and 54 years, respectively. The participants were predominantly male (67.8%, N = 99), with no significant inter-group differences in sex distribution observed (p = 0.339). Marital status exhibited a significant difference among the groups, with comparable values (p < 0.001); 48.6% (N = 71) of the participants were unmarried or single. Among adult PLWH receiving HAART, most were unmarried or single, comprising 92.1% (N = 35). In contrast, old PLWH were predominantly married or cohabiting, accounting for 82.8% (N = 24). Sexual transmission was identified as the primary route of HIV infection, with heterosexual transmission responsible for 58.2% (N = 85) and homosexual transmission for 24.0% (N = 35) of all participants. In terms of treatment, NRTIs combined with NNRTIs were the predominant therapies, representing 93.2% (N = 136) for the initial regimen and 75.3% (N = 110) for the current regimen, respectively. Previous evidence has shown that co-infection status induces broad and strong T-cell responses in HIV-infected individuals, particularly in those who are positive for CMV and EBV [26]. In light of this, we conducted a comprehensive screening of the co-infection status among the enrolled patients for pathogens including HCV, HBV, CMV, and EBV. The results indicated a low overall percentage of co-infections among the 146 participants. Specifically, 2.7% (N = 4) were positive for HBV, 6.2% (N = 9) for HCV, 12.8% (N = 18) for CMV, and 15.6% (N = 22) for EBV. Comparisons between groups revealed that HBV- and HCV-positive participants were most prevalent in the HAART-treated old group, with 6.9% (N = 2) and 13.8% (N = 4), respectively. In contrast, the HAART-treated middle-aged group exhibited the highest positivity rates for both CMV and EBV, with 33.3% (N = 5) for each. Although there were differences in positivity rates among the groups, these differences were not statistically significant. All participants underwent a comprehensive health check-up before the commencement of the study, which included routine laboratory tests for blood counts and blood biochemistry. The results indicated no significant differences in WBC, Cre, TG, CHO, and ALT when comparing adult, middle-aged, and old participants on HAART with their untreated counterparts in a two-by-two analysis. Notably, PLT in the HAART-treated old groups were 195.0 (143.0, 249.0)×109/L, which was significantly lower than those in the treated adult group at 255.5 (225.0, 277.8)×109/L and the middle-aged group at 254.0 (222.0, 287.0)×109/L. Glu in HAART-treated old groups were recorded at 5.6 (5.2, 6.3) mmol/L, significantly higher than the 5.2 (4.9, 5.4) mmol/L observed in the HAART-adult group. Additionally, Hb, TBIL, and AST were significantly different between the groups. According to the Standard reference intervals for blood cell analysis of the People’s Republic of China’s health industry. (WS/T 405–2012, WS/T 404–2012), although numerical differences were observed among the three groups, all values remained within the normal range.

Table 1 Demographic and clinical baseline characteristics of study participantsAbsolute counts of peripheral lymphocytes and chronic inflammatory state in participants

To achieve a comprehensive understanding of the differences in T-cell immunophenotype and immune function among PLWH of varying ages, we conducted a specific experimental procedure, as illustrated in Fig. 2a. PBMCs were analyzed using a 16-parameter flow cytometer to identify the phenotype and function of subsets within CD4+T and CD8+T cells. Plasma samples were collected to assess the levels of various inflammatory cytokines. We proceeded to evaluate the peripheral lymphocytes of the enrolled participants. As shown in Fig. 2b, there were no statistically significant differences in the percentage and absolute numbers of CD4+T cells, CD8+T cells, CD4/CD8 ratio, or viral load among the untreated groups. HAART successfully restored the absolute numbers of CD4+T cells and the percentage of CD8+T cells, resulting in a higher CD4/CD8 ratio and a dramatic reduction in the plasma viral load for PLWH who had received HAART (p < 0.001). The CD4+T cell counts in the treated adults were notably higher than the older, whereas no significant differences were observed in CD8+T cell count, percentage, the CD4/CD8 ratio, or viral load. The levels of inflammatory factors were subsequently assessed. Our observations indicated that HAART significantly diminishes IL-10 levels in the plasma of PLWH when compared to those who remain untreated. Previous reports indicate that successful HAART reduces plasma IL-10 levels, which corresponds with a decrease in viral load among PLWH. A lack of decline in plasma IL-10 levels after HAART administration has been associated with virologic treatment failure [27, 28]. Furthermore, HAART-induced reduction in the number of Th2 cells may be directly responsible for the significant decrease in IL-10 levels [29,30,31]. Additionally, we discovered that following HAART, CXCL2 levels were notably lower in adults compared to their middle-aged and older counterparts (Fig. 2c). The overall similarity in immunological characteristics indicated that all PLWH appeared to derive benefits from HAART without any discernible impact of age on immune recovery. This consistency prompted further analysis of T cell subsets and rendered our subsequent analyses more comparable.

Fig. 2

Study design and characteristics of study participants. (a) Schematic showing the overall study design. The study involved 146 PLWH who were grouped based on their age and whether they received HAART. Flow cytometry and multiple inflammatory cytokines were applied to their PBMCs and plasma, respectively. (b) The percentages and absolute numbers of CD4+T and CD8+T cells, the CD4/CD8 ratio, and viral load in PLWH stratified by age and HAART status. Flow cytometry was employed to assess the T cell count in PBMCs, while viral load was measured in plasma by qPCR. The data distribution is visually presented through violin plots. Statistical analyses were conducted using the nonparametric Kruskal-Wallis test, with significance levels denoted as: * p < 0.05; ** p < 0.01; *** p < 0.001; **** p < 0.0001. (c) The quantitative results of multiple inflammatory mediators in plasma samples determined by cytometric bead array. The data are expressed as means ± SDs and analyses were performed using Two-way ANOVA test. The heatmap utilizes color intensity to convey the p-values for pairwise comparisons. A green color indicates statistical significance at P < 0.05, while red signifies no significant difference (refer to the color bar). PBMCs: peripheral blood mononuclear Cells, sST2: soluble suppression of tumorigenicity 2, sRAGE: soluble receptor for advanced glycation end products, sCD40L: soluble CD40 ligand, sFlt-1: soluble fms-like tyrosine kinase-1, TNF-α: tumor necrosis factor-alpha, IL-6: interleukin-6, IL-18: interleukin-18, IL-10: interleukin-10, CCL2: chemokine ligand 2

Significant restoration of T cell subsets by HAART with distinct patterns in older PLWH

T cell subsets have been characterized in detail as previously reported by Sven Koch. et al. [17]. A representative flow cytometry gating strategy for immune cell subsets was illustrated in Figure S2. As shown in Fig. 3a and c, the CD4+T cells of untreated PLWH were predominantly composed of effector memory (EM) cells, which accounted for approximately 49.1-50.9% of the overall CD4+T cell pool. CCR7+CD4+T cells constituted roughly 37.0%-47.9% of the CD4+T cell composition. Compared with untreated middle-aged and older PLWH, untreated adults exhibited the highest proportion of Naïve CD4+T cells (29.4% vs. 15.5% vs. 18.3%, p < 0.01), and the lowest proportion of EM1 CD4+T cells (14.0% vs. 19.8% vs. 18.3%, p < 0.01). Conversely, untreated older PLWH demonstrated a significantly higher proportion of the terminally differentiated effector memory (TEMRA) subset than untreated middle-aged and young PLWH (3.0% vs. 3.0% vs. 7.3%, p < 0.01), with particularly notable differences observed between the CD4 + E and CD4 + PE2 subsets (P < 0.01). HAART effectively reversed the decline in the CCR7+CD4 subset, and there was a significant increase in Naïve CD4+T cells after HAART in both adults (from 29.4 to 47.6%, p < 0.0001) and middle-aged individuals (from 15.5 to 39.1%, p < 0.001). However, no significant change was observed among the older(from 18.3 to 25.4%, p = 0.086). There was also a decrease in the EM subset in treated PLWH, especially for the EM3 subset (p < 0.001), which exhibited significantly lower levels than untreated PLWH. Notably, the proportions of central memory (CM) subsets were significantly higher in the treated older PLWH than in the treated younger and middle-aged groups (23.5% vs. 18.3% vs. 30.2%, p < 0.001).

Fig. 3

Distribution of CD4+T and CD8+T cell subsets in PLWH of different ages. (a-b) Distribution of the CD4+T and CD8+T cell subsets at different ages of PLWH. Percentage data in the graphs were accurately determined by flow cytometry and reflect the average proportion of each cell subset in the total CD4+T cells or CD8+T cells poor. (c-d) Differences in the frequencies of CD4+T and CD8+T cell subsets were compared between the HAART and untreated PLWH across the three age groups. The groups are represented in different colors. The data are presented as means ± SDs, and analyses were conducted using the Kruskal-Wallis test. The heatmap illustrates the p-values for pairwise comparisons. A green color indicates statistical significance at P < 0.05, while red denotes no significant difference (see color bar). CM: central memory cells, EM: effector memory cells, CCR7: C-C Chemokine Receptor 7

In contrast to CD4+T cell subsets, CCR7+T cells comprised the smallest subset (4.8-12.0%) of CD8+T cells in untreated PLWH, while EM exhibited the highest frequency (64.0-69.0%), closely followed by TEMRA (21.8-31.0%). Similar to the CD4+T cells, the untreated older PLWH showed significantly lower percentages of Naïve CD8+T cells (11.4% vs. 6.5% vs. 3.7%, p < 0.001) and higher percentages of TEMRA CD8+T cells (24.0% vs. 21.8% vs. 31.0%, p < 0.001) than the untreated adult and middle-aged PLWH. HAART-induced changes in the CD8+T cell compartment appear to be more pronounced. Compared with untreated PLWH, all treated PLWH experienced a notable increase in the proportion of Naïve, EM1, and PE1 subsets, whereas the proportion of EM2 subsets significantly decreased (p < 0.001). Nevertheless, Naïve CD8+T cells were consistently the lowest in older PLWH (27.9% vs. 21.7% vs. 13.6%, p < 0.001) after HAART (Fig. 3b and d).

Cytokine dysregulation and immune dysfunction of T cells in older PLWH

Cytokine dysregulation and inflammation have been recognized as the primary drivers of CD4+T cell exhaustion and immune dysfunction during the progression of HIV-1 disease [32]. Subsequently, T cells were stimulated with PMA and ionomycin, enabling the investigation of age-related variations in cytokine profiles. Our findings demonstrated comparable expression levels of IL-2, Ki67, and MX1 across untreated age groups. Notably, untreated older PLWH exhibited significantly elevated levels of CD57, CXCR3, HLA-DR, IFN-γ, T-bet, and TIGIT compared to untreated younger PLWH (p < 0.05). The administration of HAART resulted in a significant reduction in the levels of HLA-DR, Ki67, and MX1 across all PLWH. In both adult and middle-aged PLWH, HAART resulted in a reduction of CD57 and PD-1 expression. However, when comparing older PLWH who received treatment with those who did not, no significant differences were observed (p > 0.05). Similar to the untreated status, markers of cell activation and exhaustion, such as CXCR3, HLADR, and TIGIT, continue to exhibit higher expression in the treated older PLWH compared to treated adult and middle-aged controllers (Fig. 4a).

Fig. 4

Characterisation of T cell subsets and their correlation with clinical characteristics. (a, d) Expression of activation, senescence and exhaustion molecules by CD4 (a) and CD8+T (d) lymphocytes from the PLWHs by flow cytometry. Statistical analysis was performed using Kruskal-Wallis. The heatmap illustrates the p-values for pairwise comparisons. A green color indicates statistical significance at P < 0.05, while red denotes no significant difference (see color bar). (b, e) The correlation matrix shows the associations between the clinical characteristics and immune markers within the CD4+T subsets (b) and CD8+T subsets (e). The colors signify negative correlations (red) and positive correlations (blue). The size of the squares and the saturation of the color denote the absolute values of the Spearman coefficients. (c, f) Multidimensional correlation analysis between T cell subsets and parameters of immune function within CD4+T subsets (c) and CD8+Tsubsets (f). The rectangular boxes represent Spearman’s correlations between the different immune markers, with red indicating negative correlations and blue indicating positive correlations, with darker colors and larger color blocks indicating stronger correlations. The relationship between each subset of T cells and each immune variable was determined via Mantel’s test, with lines of different thicknesses related to Mantel’s r statistic, while different ranges of P values are indicated by lines of different colors. VL: viral load, WBC: white blood cells, HGB: hemoglobin, PLT: platelets, T.BIL: total bilirubin, ALT: alanine aminotransferase, AST: aspartate aminotransferase, GLU: glucose, TG: triglycerides, CHOL: total cholesterol, CM: central memory cells, EM: effector memory cells, TNF-α: tumor necrosis factor-alpha, IL-2: interleukin-2, CXCR3: C-X-C motif chemokine receptor 3, MX1: murine myxovirus resistance 1, PD-1: programmed cell death protein 1, TIGIT: T cell Immunoreceptor with Ig and ITIM domains, HLA-DR: human leukocyte antigen DR, T-bet: T-box expressed in T cells, GrB: Granzyme B, IFNg: IFN-γ

Correlation analysis revealed positive associations between the expression of CXCR3, HLADR, and TIGIT with age and viral load. Conversely, a negative correlation was observed with CD4+T cell count, CD4/CD8 ratio, and duration of treatment(Fig. 4b). Furthermore, HLA-DR, CXCR3, and TIGIT were strongly positively correlated with each other and positively correlated with PD-1 and Ki67. Mantel’s tests revealed that the expression of HLA-DR, CXCR3, and TIGIT was closely associated with EM subsets. Specifically, CXCR3 was significantly associated with EM1 subsets (Mantel’s r = 0.22, p < 0.01), while TIGIT was significantly associated with EM2 subsets (Mantel’s r = 0.33, p < 0.01). The markers that were positively correlated with HLA-DR, including PD-1, T-bet, and GranzymeB, were closely associated with the EM3 subset (Fig. 4c). These results indicate that old PLWH exhibit elevated levels of cytokines and immune markers, such as CXCR3, HLA-DR, and TIGIT when stimulated in vitro, compared to their adult counterparts. While these molecules are implicated in the immune response, they are also associated with immune activation, inflammation, and T-cell exhaustion. The increased levels observed in old PLWH suggest a chronic state of immune activation and inflammation, which may be detrimental to immune function and could contribute to accelerated immune senescence and impaired CD4+T cell reconstitution, and HAART does not alleviate the persistent heightened activation of CD4+T cells in old PLWH. Moreover, the CD4+T EM subset is characterized by elevated levels of exhaustion markers (TIGIT, PD-1), an activation marker (HLA-DR), and inflammation markers (CXCR3, TNF-α). The accumulation of the EM CD4+T subset in older PLWH may impede CD4+T cell reconstitution.

The same strategy was employed for CD8+T cells. As shown in Fig. 4d, the expression levels of Granzyme B, HLA-DR, MX1, PD-1, and T-bet did not show significant differences among all untreated PLWH. However, CD57, perforin, and TNF-α were expressed at significantly higher levels in untreated older PLWH compared to untreated young and middle-aged PLWH. Following HAART, the expression levels of CD38, CD57, Granzyme B, HLA-DR, MX1, PD-1, T-bet, TIGIT, and TNF-α were significantly reduced across all PLWH. Particularly in the young and middle-aged PLWHs, there was also a more pronounced decrease in perforin expression. Comparisons of PLWH across different age groups after treatment revealed no significant differences in the expression of CD38, HLA-DR, perforin, or T-bet. However, the expression levels of CD57, Granzyme B, and TIGIT were significantly higher in older PLWH compared to their younger and middle-aged counterparts. Additionally, the MX1 and TNF-a expression levels were significantly higher than the young PLWH (Fig. 4d). Consistently, immune activation markers (CD38 and HLA-DR), cytotoxic phenotypes (Granzyme B, T-bet, and MX1), exhaustion markers (TIGIT and PD1) and the proliferation marker (Ki67) were positively associated with the viral load and significantly negatively correlated with CD4+T cell count, CD4/CD8 ratio, and duration of treatment (Fig. 4e). These markers were positively correlated with each other and were the strongest correlates of both the Naïve and EM2 subsets (Mantel’s r = 0.19–0.58, p < 0.01), moderately correlated with the E and PE1 subsets (Mantel’s r = 0.04–0.30, p < 0.05) (Fig. 4f).