Pathological Proliferation of CD4⁺ T Cells in Late Presentation of HIV Infection After Antiretroviral Therapy

Introduction

The application of antiretroviral therapy (ART) has greatly reduced human immunodeficiency virus (HIV) related morbidity, mortality and lowered the risk of onward transmission worldwide. But the high prevalence of “late-presentation” of HIV infection remains a major challenge in the prevention and treatment of acquired immune deficiency syndrome (AIDS).¹ Several international working groups have defined Late Presentation as either a CD4⁺ T cell count of fewer than 350 cells/μL at diagnosis or the occurrence of an AIDS-defining event, regardless of CD4⁺ T cell count.^2–4 According to these definitions, about 50% of people living with HIV (PLWH) in developed countries are late presenters (LP),⁵ with even higher proportions observed in low- and middle-income countries.^4,6 Studies showed that LP have more complications, a higher risk of hospitalization, and worse long-term outcomes.^7,8

Some studies have found that the rebound of CD4 cells is an independent predictor of the long-term prognosis for LP, suggesting that restoring immune function is key to better outcomes.^9–11 However, while LP achieved similar viral suppression compared to n-LP after ART, they exhibited poorer recovery of multiple T-cell markers and faced a higher risk of becoming immunological non-responders.^9,12 Despite substantial evidence showing that immune reconstitution is more difficult in LP, their specific immunological features and the mechanisms underlying immune dysfunction remain insufficiently understood.

Our previous study showed that early ART in people with acute HIV infection normalizes CD4⁺ T cell activation, while effectively suppressing T cell proliferation and systemic inflammatory responses.¹³ The higher proliferative CD4⁺ T cells are co-expressed with markers for cell exhaustion (PD-1), persist during ART treatment, and are inversely correlated with CD4⁺ T cell counts after ART.¹⁴ Later, we demonstrated in immunological non-responders that excessive CD4⁺ T cell proliferation accompanied by abnormal activation and senescence constitutes “pathological proliferation” and is associated with poor functional recovery of CD4⁺ T cells.¹⁵ Since LP has lower numbers of CD4⁺ T cells, and is susceptible to poor immune reconstitution, therefore, we hypothesize that the pathological proliferation may exist in LP, and ultimately leads to immune dysregulation, that can not be restored by ART.

In the present study, we accessed the phenotypes and function of CD4⁺ T cells in both LP and n-LP, and verified the existence of pathological proliferation in LP. We observed a higher proportion of CD4⁺ T cell subpopulations with characteristics of proliferation, activation, exhaustion, and senescence in LP. Moreover, the proportion of CD4⁺ T cells with pathological proliferation traits was significantly higher in LP. Our findings provide a new perspective for the immunotherapy and help to find factors that improve the long-term prognosis of LP.

Methods

Participants

From April 1, 2022 to September 30, 2022, a total of 109 HIV-infected adults were enrolled at Beijing Youan Hospital, Capital Medical University. Inclusion criteria included: (1) HIV-Positive men who have sex with men; (2) at least 18 years of age. Patients with documented autoimmune diseases, current immunosuppressive therapy, or other severe concurrent conditions considered by the investigators to substantially affect immune status, such as active opportunistic infections, active tuberculosis coinfection, or malignancy, were excluded. Plasma and peripheral blood mononuclear cell (PBMC) samples from all eligible participants were collected and stored at −80°C and in liquid nitrogen tanks. Clinical data, including demographic characteristics, CD4⁺ T cell counts, CD8⁺ T cell counts, HIV RNA load, the date of HIV diagnosis, the duration of HIV infection, and the duration of ART, were obtained from the medical records of the AIDS research cohort at Beijing Youan Hospital. Patients were divided into LP (CD4⁺ T cells < 350 cells/μL or having an AIDS event at initial diagnosis) or n-LP. Some potential confounders related to immune activation or T-cell function (such as age, body mass index [BMI], etc.) were assessed and reported for comparability between groups. The study was approved by the Ethics Committee of Beijing Youan Hospital (2023/057). All participants were informed of the purpose of the study, and written informed consent was obtained.

Mass Cytometry and Data Analysis

To assess the immune cell characteristics and functions of LP, mass cytometry was performed on PBMC samples with 23 custom-designed antibodies to differentiate various immune cells. (Heavy metal isotope-labeled monoclonal antibodies are listed in Supplementary Table 1) The pre-conjugated antibodies, purchased from Fluidigm (South San Francisco, USA), followed established protocols.¹⁶ After washing, PBMCs were stained with cisplatin-195Pt (Fluidigm, 201064) to exclude dead cells. Antibody staining was preceded by blocking Fc receptors using human TruStain FcX. After staining for 30 minutes at a low temperature, the samples were incubated with 125nM Cell-ID Intercalator-Ir (Fluidigm, United States) in phosphate-buffered saline (PBS, SigmaAldrich, United States) at 4°C. Samples were resuspended in double-distilled water with EQ beads (Fluidigm, South San Francisco, USA) at a concentration of 5.5 × 10⁵ cells/mL and analyzed by the CyTOF2 mass cytometry system (Fluidigm, South San Francisco, USA).

The raw data for each sample were de-barcoded using unique mass-tagged barcodes, in a double-filtering scheme. The.fcs files produced by different batches were standardized using the bead normalization technique. Gating was performed with FlowJo software (version 10) to exclude debris and dead cells. CD45⁺CD3⁺CD4⁺ T cells with different characteristics were gated (Supplementary Figure 1) manually for R language analysis. The PhenoGraph clustering technique was applied to separate cells based on surface marker expression levels. A dimensionality reduction technique, t-distributed stochastic neighbor embedding (t-SNE), was used to visualize high-dimensional data. Marker distribution and differences between groups were analyzed using R software (version 3.6.1).

Statistical Analyses

Statistical analysis was conducted with R (version 3.6.1). Categorical variables between LP and n-LP were compared using the χ2 test or Fisher’s exact test, while continuous variables were analyzed using Student’s t-test or Mann–Whitney U-test. Pearson’s correlation analysis was applied for normally distributed data, while Spearman’s correlation analysis was used for non-normally distributed data. GraphPad Prism and R were used to generate diagrams illustrating the statistical results.

Results

Characteristics of the Participants

N-LP (n = 54) and LP (n = 55) were included in the study, and their clinical information was shown in Table 1. The two groups were comparable in age, BMI, time to initiate ART after HIV discovery, duration of HIV infection, and duration of ART (all P > 0.05). Initial CD4⁺ and CD8⁺ T-cell counts, CD4⁺ T cell counts and CD8⁺ T cell counts at the beginning of ART, and recent CD4⁺ T cell counts were significantly lower in LP. However, the initial viral load (VL) and the VL at the beginning of ART were higher in LP.

Table 1 Demographic and Clinical Characteristics of All Subject

LP Showed a Higher Proportion of Double Positive Tem and CD4⁺ Tcm

We performed mass spectrometry to investigate the phenotypes and functions of T cells in both LP and n-LP. By using R software (version 3.6.1), CD4⁺ T cells were categorized into 21 clusters on the basis of marker expression (Figure 1A). The distribution of each cluster in LP and n-LP was depicted in Figure 1B. Among the 21 clusters of CD4⁺ T cells, C3, C13 and C21 represented CD4⁺ naive T cells (Tn) (CD3⁺ CD4⁺ CD8⁻ CD45RA⁺ CCR7⁺); C4, C5, C6, C9, C10, C11, C15, C16 and C18 represented CD4⁺ central memory T cells (Tcm) (CD3⁺ CD4⁺ CD8⁻ CD45RA⁻ CCR7⁺); C1, C2, C7, C12, C14, C17 and C19 represent CD4⁺ effector memory T cells (Tem) (CD3⁺ CD4⁺ CD8⁻ CD45RA⁻ CCR7⁻); C8 and C20 represent double positive Tem (CD3⁺ CD4⁺ CD8⁺ CD45RA⁻ CCR7⁻) (Figure 1C). By analyzing the expression positions of different markers through t-SNE visualization, we found that CD57 was significantly clustered in C1, C12 and C20, and perforin was significantly clustered in C1 and C2, while other markers were scattered in different clusters without significant clustering performance (Supplementary Figure 2). After comparing the proportions of the same clusters in different groups, it was observed that the frequencies of C8 (double-positive Tem) and C11 (CD4⁺ Tcm) in LP were notably higher than those in n-LP (Figure 1D), while no significant differences were observed in the frequencies of other subpopulations of CD4⁺ T cells between n-LP and LP (Supplementary Figure 3).

Figure 1 Higher proportion of double positive T cells and CD4+ Tcm cells in LP group. (A) t-SNE plots of CD4+ T cells in all participants and the serial numbers of each CD4+ T cells cluster. (B) t-SNE plots of CD4+ T cells in group n-LP and LP. The two plots share the same y-axis, and the x-axis range in both plots is from -20 to 20. (C) Heatmap of markers expression for 21 CD4+ T cell clusters. Clusters were ordered by hierarchical clustering based on median expression of all markers. (D) Comparison of the proportion of double positive T cells and CD4+ Tcm cells between n-LP and LP. Mann–Whitney U-test was used in (D). *P < 0.05; **P < 0.01. Abbreviations: n-LP, non-late-presenters; LP, late-presenters; Tn, naïve T cells; Tcm, central memory T cells; Tem, effector memory T cells.

Certain CD4⁺ T Cells in LP Group Exhibited High Levels of Proliferation, Activation, Exhaustion and Senescence

After observing that some markers were concentrated in certain subpopulations, we intended to find whether there were differences in marker expression in CD4⁺ T cells between n-LP and LP. We compared the expression of markers related to activation (CD38, HLA-DR), senescence (CD57), cytotoxicity (CD107A, perforin), programmed cell death (PD-1, PD-L1) and others (CD31) in the same CD4⁺ T cell clusters of n-LP and LP. Compared with n-LP, CD4⁺ Tcm in the LP showed increased expression levels of PD-1 (C5; Figure 2C) and CD57 (C4, C10, C15; Figure 2E). CD4⁺ Tem in the LP exhibited high expression levels of Ki67 (C17; Figure 2A), HLA-DR (C1; Figure 2B), CD31 (C12, C19; Figure 2D) and CD57 (C7, C19; Figure 2E). CD4⁺ Tn in LP exhibited high expression levels of Ki67 (C21; Figure 2A) and CD57 (C13, C21; Figure 2E). These findings suggest that T cells at different stages in LP had showed higher levels of proliferation, activation, exhaustion and senescence compared to n-LP. Although we expected to observe differences in multiple markers in a specific cell subpopulation, these differences were evenly distributed in different clusters, except for C21 (CD4⁺ Tn) in LP, which highly expressed both Ki67 and CD57.

Figure 2 The CD4+ T cells in LP group exhibited higher levels of proliferation, activation, apoptosis and senescence. (A) Comparison of Ki67 expression levels between two groups. (B) Comparison of HLA-DR expression levels between two groups. (C) Comparison of PD −1 expression levels between two groups. (D) Comparison of CD31 expression levels between two groups. (E) Comparison of CD57 expression levels between two groups. Mann–Whitney U-test was used to test group difference. *P < 0.05; **P < 0.01. Abbreviations: n-LP, non-late-presenters; LP, late-presenters; C, Cluster; Tcm, central memory T cells; Tem, effector memory T cells.

Pathological Proliferation of CD4⁺ T Cells Was Observed in LP Group

Considering that C21 highly expressed both Ki67 and CD57 (Figure 2A and E), we hypothesized that there were T cells in the LP co-expressing Ki67 with other markers. We performed further gating analysis using FlowJo (Supplementary Figure 1) to verify the existence of this co-expression phenomenon. Specifically, we compared the differences in the proportion of T cells expressing different markers between the two groups respectively, and then gated co-expressing cells in ki67⁺ CD4⁺ T cells for analysis. The results showed that the percentage of CD4⁺ T cells expressing Ki67, PD-1, HLA-DR, and CD31 was higher in LP than in n-LP (Figure 3A–D). In addition, in line with our hypothesis, there were CD4⁺ T cells co-expressing Ki67 and CD57, CD107a, HLA-DR, CD38 in LP, and the proportion of these co-expressed cells was higher than that in n-LP (Figure 3E–H). These results suggest that T cells in LP exhibit higher levels of exhaustion and activation, implying a greater degree of cell depletion, compared to n-LP. In addition, we observed that a large part of the proliferation of T cells in LP is accompanied by aging and over-activation, which is at an abnormal proliferation status. Our results suggest that there is a “pathological proliferation” phenomenon in LP, and pathological proliferation may fail to promote the recovery of the number or function of T cells.

Figure 3 Pathological proliferation of CD4+ T cells was observed in LP group. (A-D) The frequencies of Ki67+, PD-1+, HLA-DR+ and CD31+ T cells (CD4+ T cells) in LP group and n-LP group. (E-H) The frequencies of CD57+, CD107a+, HLA-DR+ and CD38+ T cells (Ki67+ CD4+ T cells) in LP group and n-LP group. Mann–Whitney U-test was used to test group difference. *P < 0.05; **P < 0.01; ***P < 0.001. Abbreviations: n-LP, non-late-presenters; LP, late-presenters.

Correlations Between T Cell Characteristics and Clinical Parameters in LP Compared to n-LP

Correlation analyses were conducted to investigate the associations between different CD4⁺ T cell subsets and clinical features in the two groups (Figure 4A and B). In LP (Figure 4A), subsets C1 and C8 showed a positive correlation with recent CD8⁺ T cell counts. Conversely, subsets C4, C19 and C21 demonstrated a negative correlation with CD8⁺ T cell counts at both ART initiation and recent measurements. Subset C5 was inversely correlated with recent CD4⁺ T cell counts. C7 had a negative association with CD4⁺ T cell counts at baseline, ART initiation, and in recent tests. C10 was positively related to the age of participants. C15 was positively correlated with baseline, ART initiation, and recent CD4⁺ T cell counts, and negatively correlated with baseline viral load. In contrast, C17 was negatively correlated with CD8⁺ T cell counts at baseline and ART initiation, while being positively correlated with baseline and ART initiation viral loads. In n-LP (Figure 4B), the heat map of correlation differed from that of LP, especially in C7 and C19, both of which were positively correlated with recent CD8⁺ T cell counts. These findings suggest that the correlations between T cell subsets and clinical features are more complex in LP than in n-LP, indicating the need for further studies to understand the underlying mechanisms of these differences.

Figure 4 Correlation between T cell characteristics and clinical parameters in LP group was different from that in n-LP group. (A) Correlation between some CD4+ T cell subpopulations and clinical parameters in LP group. (B) Correlation between some CD4+ T cell subpopulations and clinical parameters in n-LP group. (C) Correlation between CD4+ T cells with proliferation, activation, apoptosis, and senescence properties and clinical parameters in LP group. (D) Correlation between CD4+ T cells with proliferation, activation, apoptosis and senescence properties and clinical parameters in n-LP group. Correlations were evaluated using Spearman’s rank correlation coefficient. *P < 0.05; **P < 0.01. Abbreviations: BMI, body mass index; VL, Viral Load; ART, antiretroviral therapy.

We further analyzed the correlation between CD4⁺ T cells with various marker expressions and clinical features in the two groups (Figure 4C and D). Notably, in the LP group (Figure 4C), the proportion of Ki67⁺ CD4⁺ T cells was inversely related to baseline viral load and positively related to CD8⁺ T cell counts at ART initiation. The proportion of PD-1⁺ CD4⁺ T cells was positively correlated with the number of CD8⁺ T cells at ART initiation. Moreover, the percentage of Ki67⁺ CD57⁺ CD4⁺ T cells was positively correlated with CD8⁺ T cell counts at baseline and ART initiation, and negatively correlated with viral load. Additionally, the percentage of Ki67⁺ CD38⁺ CD4⁺ T cells was positively correlated with the number of CD8⁺ T cells at the initiation of ART. However, these correlations were not found in n-LP (Figure 4D). Our findings indicate that a variety of T cell immune dysfunction, including abnormal proliferation, activation, exhaustion, and senescence is present in LP, which may be associated with specific clinical course during HIV infection, especially T cells with pathological proliferation characteristics.

Discussion

This study updates our understanding of CD4⁺ T-cell features in LP after long-term ART. CD4⁺ T-cell counts remained lower in LP than in n-LP, indicating incomplete immune recovery. We further identified a CD4⁺ T-cell pattern in LP in which proliferation (Ki67) occurred together with activation and/or senescence markers. We refer to this pattern as “pathological proliferation”. This finding suggests ongoing immune dysregulation in LP despite ART and may be linked to poorer recovery of a functional CD4⁺ T-cell pool. Key results are summarised in Supplementary Table 2. Future longitudinal and mechanistic studies are needed to confirm this link, define the underlying pathways, and identify targets to improve immune recovery in LP.

Despite the poor recovery of the total number of CD4⁺ T cells in the LP, we observed the frequency of some specific cells was higher in LP than those in n-LP. In our study, C8 was a group of double-positive Tem with aging properties (higher CD57 expression). Double positive T cells are induced by the activation of single positive T cells stimulated by antigen.¹⁷ In patients with advanced HIV, the number of double positive cells significantly increases, is inversely correlated with CD4⁺ T cell counts, and exhibits higher levels of activation and exhaustion.¹⁸ We speculate that due to the prolonged stimulation of chronic HIV infection, double-positive T cells are induced to increase in order to exert better immune function.¹⁹ However, these cells are also the target of viral attack,¹⁹ and therefore, they are more activated and depleted, and they are unable to exert stronger antiviral function. In addition, it has been shown that double-positive T cells can home to the brain and serve as an HIV reservoir to support efficient HIV replication and bring about neural invasion.²⁰ In conclusion, little is known about double-positive T cells, and further studies on the function of this cell subset could help us understand their role in the progression of AIDS. We also observed an increased frequency of CD4⁺ Tcm (C11) with relatively high PD-1 expression in LP. The up-regulation of PD-1 in CD4⁺ Tcm is a sign of chronic antigen stimulation, which reflects the continuous immunosuppression and immune system damage under long-term ART, and can partly explain the cause of immune dysfunction in LP.

We found that multiple CD4⁺ T cell subsets in the LP highly expressed activation marker HLA-DR, exhaustion marker PD-1 and senescence marker CD57 after ART. In treated HIV-infected individuals with poor CD4 recovery, activated memory CD4+ T cells increase, and this phenotype is associated with inflammatory exposure (including elevated sCD14 and interleukin-6).²¹ Thus, the persistence of HLA-DR, PD-1, and CD57 in the LP may reflect the presence of ongoing inflammation and homeostasis imbalance, driving the differentiation of CD4⁺ T cells into activated, depleted, and senescent states rather than durable functional recovery. This implies that although the CD4⁺ T cell counts in LP has increased compared to that before ART, a large number of these cells are of poor quality, showing a high degree of activation, exhaustion and senescence, and may not be capable of performing a perfect immune function. We also found that CD4⁺ Tem in the LP group expressed higher CD31 than in the n-LP group. CD31 is generally expressed on the surface of Tn and is considered an indicator of thymic function and immune ability.^22,23 In addition, CD31 has a role as a negative regulator of T cell function,²⁴ and its expression in memory T cells is associated with higher PD-1 and CD38/HLA-DR co-expression.²⁵ This suggests that the high expression of CD31 in LP may be caused by the excessive activation and depletion of T cells in LP, which requires negative regulatory factors to play a “brake” role.

Notably, we found that more cells in LP co-express proliferation and senescence (CD57) or activation (HLA-DR, CD38) properties than in n-LP. Ki67 represents the proliferation ability of cells.²⁶ Previous studies have found that a balanced proliferation of T cells occurs in the presence of lymphopenia, including naive and memory T cells.²⁷ In treated HIV infection, lower numbers of central and effector memory CD4⁺ T cells are associated with higher proportions of Ki67⁺ cells in these subsets, also suggesting that CD4⁺ T cell depletion may drive compensatory proliferation.²⁸ In addition, individuals with poor CD4 recovery display increased expression of homeostatic proliferation-related markers such as OX40 and α4β7 even before ART, and these abnormalities may persist despite suppressive treatment.²⁹ These results are consistent with our findings in LP, where LP exhibit greater CD4⁺ T cell depletion, as well as a greater propensity for proliferation, compared to n-LP. However, we found that the frequency of the C21 (CD4⁺ Tn), which has high proliferative capacity but simultaneously exhibits senescent properties, was lower in the LP than in the n-LP. We suggested that this phenomenon was caused by the expression of CD57. Then we analyzed the co-expressing cells and found that a subset of T cells in the LP had simultaneously proliferative, activated, or senescent properties. Senescence is a state characterized by impaired T cell function³⁰ and is prevalent in HIV infection.³¹ HIV infection induces sustained and active proliferation, leading to the accumulation of senescent T cells.^8,15,32 CD38 is a marker of T cell immune activation, and the expression of CD38 can promote proliferation of CD4⁺ T cells in PLWH.³³ Therefore, we suggest that the excessive proliferation of T cells in LP is accompanied by aging and an increased degree of activation. In other words, the proliferating cells were of poor quality, thus did not promote the recovery of T cell function in LP, and this phenomenon did not improve with the progress of ART. This phenomenon is consistent with our previous finding of pathological proliferation in INRs.¹⁵

Our findings emphasize the complexity of the immune profile of LP by revealing the correlation of specific CD4⁺ T cell subsets with clinical parameters. Notably, certain T cell clusters in LP, especially those cell subsets that express markers associated with activation, apoptosis, and senescence, were not suppressed by ART as they were in n-LP. In LP, two CD4⁺ Tem clusters, C7 and C19, showed a negative correlation with CD4 cell count. We speculated that the low CD4 count in LP group led to a passive increase in the composition ratio of Tem. However, the conversion of naive cells to effector cells was accelerated under continuous antigen stimulation.³⁴ Thus, lower CD4 counts in LP induce Tem amplification and may also be a marker of immune imbalance. It is suggested that ART is not sufficient to restore the immune response of LP to the normal level or the same level as that of n-LP. Similarly, LP co-expressing Ki67 and CD38/CD57/PD-1 were also significantly correlated with immune status, and the correlation profile was significantly different from that of n-LP. However, the underlying mechanism is unclear and requires further investigation. The persistence of these T cells with aberrant immune function highlights the need for long-term monitoring to improve the immune dysregulation of LP and the development of new targeted therapies to enhance the effectiveness of ART.

Limitations

This study has several limitations. First, as a cross-sectional study, it lacked longitudinal observations from baseline to multiple time points after ART, and therefore could not assess the dynamic changes in T cell characteristics or their sustained impact on immune recovery in PLWH. Second, we focused only on the CD4⁺ T cell compartment; broader analyses of other immune cell populations, including CD8⁺ T cells, monocytes, B cells, and their interactions, are needed to more comprehensively characterize immune dysregulation in LP. Third, all participants were men who have sex with men, which may limit the generalizability of our findings to women and the broader population of PLWH. Finally, we did not perform treatment-stratified analyses by ART regimen class because the sample size was too limited to generate stable estimates after further stratification; thus, residual confounding related to treatment category cannot be excluded. Future studies should include longitudinal, multicenter cohorts with more diverse populations and more detailed analyses of treatment heterogeneity and multiple immune cell subsets.

Conclusions

Our findings indicate that LP experience persistently low CD4⁺ T-cell counts and impaired CD4⁺ T-cell function even after ART. Notably, pathological proliferation has been observed in the T cells of LP, characterized by excessive proliferation accompanied by cellular senescence and overactivation, which ultimately hindering the recovery of T-cell counts and impairing the T-cells function. Therefore, a thorough understanding of the mechanism underlying pathological proliferation is essential for developing innovative therapeutic strategies aimed at mitigating such abnormal proliferation phenomenon, enhancing the immune restoration and thereby improving the long-term prognosis and overall health outcomes for LP.