The pattern and magnitude of T cell subsets reconstitution during ten years of ART with viral suppression in HIV-infected patients

Clinical and immunology characteristics at baseline

In 662 HIV- infected patients, 88.2% (584) of them were male (Table 1). The mean age was 36.4 ± 11.2 years. The sexual transmission was the most frequent in transmission routes, especially in patients having sex between men (63.0%). 65.1% of patients has one or more signs or symptoms of HIV infection. 212 participants (32.0%) had an opportunistic infection and 43 (6.5%) of enrolled patients were HBsAg positive or 24 (3.6%) were HCV-Ab positive. Mean log₁₀ (viral load) of these patients was 4.74 ± 0.77. 36.3% of patients’ baseline viral load is higher than 10⁵ cps/ml, which represents a relatively high level of viral load. 225 patients (34.0%) did not achieve viral suppression at the first six months, most patients (74.8%) initiated ART with two Nucleoside Reverse Transcriptase Inhibitor (NRTIs) with non-Nucleoside Reverse Transcriptase Inhibitors (NNRTIs). The median follow-up time was 5.67 (3.17, 8.33) years.

The demographic and T cell subpopulation characteristics of patients were presented in Table 1, based on their follow-up periods (Y = 1 (n = 652), Y = 3 (n = 515), Y = 5 (n = 342), and Y = 10 (n = 89)). In enrolled HIV-infected patients, the median number of CD4+T cells and median number of CD8+T cells at baseline were 226 cells/μl (interquartile range, hereafter IQR 83–336) and 789 cells/μl (IQR 532–1045), respectively. The median percentages of naive CD4+T and median percentages of memory CD4+T cells at baseline in overall patients were 28.6% (15.9%, 38.5%) and 71.6% (61.5%, 84.2%), respectively. Median baseline percentages of CD28+CD4+T cells and CD28+CD8+T cells, CD38+CD8+T cells, HLA-DR+CD8+T cells, CD4+/CD8+T in overall patients were abnormal than the reference range (shown in Table 1). For patients followed at least 3-year, 209 (40.5%) patients achieved CD4+T cells >500 cells/μl, the percentage climbed at 44.2%, and 50.6% in patients followed for 5 and 10 years. Also, there were 7.4% patients whose baseline CD4+T cell over 500 cells/μl. Besides, the percentage of CD8+T cells in 500–999 cells/μl and 1000–1499 cells/μl changed from 45.3%, 22.8% till 62.5%, 9.1% after 10-year of ART. Whereas, the percentage of CD4/CD8> = 1 kept increasing from 1.2% to 31.5% until 10 years.

Dynamics of absolute counts of CD4+ T cell, naive CD4+ T cells, and memory CD4+T cells

The median number of CD4+T cell counts was 226 cells/μl (IQR 83–336) before treatment and increased to 449 cells/μl (IQR 303–604) after 3-year of ART (Table 1). Through fitting the increasing curve of CD4+T cells, an initial surge in CD4+T cell counts within 3 months was evident, but relatively slower accumulation in CD4+T cell count until the fourth year after treatment (Table 2 and Figure 1A). After the fourth year of steady increment, the significant change period of CD4+T cell counts prolonged to 2-year until the seventh year while there was no significant difference of CD4+T cells between the next three year (Figure 1A and Supplementary Table 1). Given the covariate factors that may affect the increase in CD4+T cell counts, it was showed that the ART scheme including Protease Inhibitors (PI), Integrase Inhibitor (INSTI) and the CD4+T cell counts (>50 cells/μl) at baseline were beneficial factors.

Absolute CD45RA+Naïve CD4+T cells started from 62 cells/μl (IQR, 14–117) at baseline to 135 cells/μl (IQR 69–218), 133 cells/μl (IQR,79,210), 130 cells/μl (IQR,76,198), 120 cells/μl (IQR,66,203) after 3 years, 5 years, 8 years and 10 years of ART, respectively (Table 1). A significant climbing on the Naïve CD4+T cells in HIV patients in every-year at first 3-year of ART but every two year of significant change for 3–8 year (Figure 1B).

The number of absolute CD45RA-Memory CD4+T cells was 150 cells/μl (IQR, 73–226) before treatment. After 3 years,5 years, 8 years, and 10 years of ART, it increased to a median of 300 cells/μl (IQR, 217–390), 326 cells/μl (IQR, 246, 417), 349 cells/μl (IQR, 265–441), 374 cells/μl (IQR, 290–456), respectively (Table 1). Unlike CD45RA+ naïve CD4+T cells, there was a significant change in CD45RA- CD4+T cells within 3 months of treatment initiation, but there was no evident change between first 3-month and 6-month. The significant increasing span went from one year during 0–3 years after ART and two years in 3–6 years (Figure 1C), and there was no significant increase thereafter through multiple comparisons (Seen in Table 2 and Supplementary Table 2).

In the mixed linear model, no significance of naïve and memory CD4+T cell counts between gender was found. Age was independently inversely linked to the CD45RA-Memory and CD45RA+CD4 T cell counts. CD4+CD45RA-T cells increased with age, while CD45RA+Naïve CD4+T cell counts decreased with age group. ART regimens (2NRTIs+PI and 2NRTIs+INSTI) have marked changes of naïve CD4+T cell and Memory CD4+T cells, which are consistent with results from CD4+T cell counts (Table 2). Since CD4+T cell counts were composed of CD4+naïve and CD4+memory T cells, it is unreasonable to put this indicator in the mixed model, so we calculated the trends of naïve and memory T cell ratio stratified by baseline CD4+T cell counts. It is shown that the kinetics of naïve/memory CD4+T cells decreased at first three months in patients whose CD4+T cell counts <100 cells/μl. However, in patients with CD4+T cell counts > = 100 cells/μl, the naïve/memory CD4+T cells would not transiently decrease at first three months (Figure 2).

Dynamics of counts of CD8+ T cells and CD4/CD8 ratio

A significant decreased effect occurred on the absolute counts of CD8+ T cells during the first years. CD8+T cells decreased from 789 cells/μl (IQR, 532–1045) to 699 cells/μl (IQR, 527–925) after 3 years of treatment (Table 1). Hereafter, the CD8+T cell counts decreased slightly to 647 cells/μl (IQR, 515–843) after 10 years of ART. Baseline CD8+T cell counts of <500 cells/μl, 500–999 cells/μl, 1000–1499 cells/μl and >1500 cells/μl were 22.2%, 45.3%, 22.8% and 9.7% became 22.7%, 62.5%, 9.1% and 5.7% after 10-year of follow-up, respectively (The trends were seen in Figure 1D). In the mixed linear model, baseline CD8+T counts significantly influenced the CD8+T cells after ART. Higher CD4+T cell counts, delayed viral suppression may interfere with CD8+T cells descent. Patients who used ART in PI-based and INSTI-based achieved higher CD8+T cell counts than NNRTI-based patients after ART treatment (Table 3).

In our patients, CD4+/CD8+T ratio achieved from 0.25 (IQR 0.12–0.40) pre-treatment to 0.66 (IQR 0.41–0.92), 0.72 (IQR 0.47–1.02), 0.76 (IQR 0.57–1.04) after 3, 5, 10 years of ART, respectively (Table 1). The CD4/CD8 continued to be a marked year-over-year increase but maintained a substantial two-year increase until the seventh year and no significant increase until the tenth year (Figure 1E and Supplementary Table 3). In our mixed linear model, baseline CD8+T cells, baseline CD4+T cell counts were independent with CD4/CD8 ratio normalization, but the ART regimen did not (Table 3).

Other lymphocyte subpopulations

There was decreasing kinetics of CD38+CD8+/CD8+, HLA-DR+CD8+/CD8+T cells, increasing kinetics of CD28+CD4+/CD4+T cells and CD28+CD8+CD8+T cells after initiation of ART. The percentage of activated CD38+CD8+T cells pretreatment was 79.8% (IQR 68.5–89.3%). ART-induced viral suppression made the percentage decreased to 44.7% (IQR 32.1–57.3%), 41.6% (IQR 30.2–54.1%), 35.2% (IQR 27.2–44.7%), 31.0% (IQR 21.5–44.8%) after 3 year, 5 year, 8 year and 10 year (Table 1). The most dramatic change occurred within the 3 months of ART, and the major decrement of CD38+CD8+T cells % occurred during the 1–5 year of treatment (Table 3 and Supplementary Table 3). Baseline CD8+T cells was not significant with CD8+CD38+T cells, but baseline CD4+T cells were independent with CD8+CD38+T cells. ART regimens seem to have little effect on CD38+CD8+T cells kinetics, ART based on PI-backbone achieved higher CD38+ expression on CD8+T cells than NNRTI-backbone.

Activated HLA-DR+CD8+/CD8+ T cells at baseline in HIV-infected patients was 63.2% (IQR 48.3–74.2%) and it decreased to 34.9% (IQR 27.2–49.4%) after 10-years ART (Table 1). After one year of ART, the rapid decrease of HLA-DR+CD8+/CD8+T cells had no significant change between the next years. In the mixed linear model, older patients were higher than younger patients in related to percentage of HLA-DR+ CD8+T cells. Unlike the CD38+CD8+/CD8+ T cells, the percentage of HLA-DR+ expression on CD8+T cells was independently related with CD8+T cells instead of CD4+T cells. Delayed viral suppression had no significant effect on CD38+ expression on CD8+T cells and HLA-DR+ expression on CD8+T cells.

The expressing molecule CD28 on CD4+T cells and CD8+T cells were 89.2% (IQR 75.0–95.5%) and 30.5% (IQR 21.7–40.1%) before treatment, respectively. After 3-year of follow-up, the percentage inclined to 93.3% (IQR 86.5–97.4%), 48.5% (IQR 38.6, 60.1), respectively (Table 1). In our results, CD28+CD4+T cells significantly mounted within the second year, while CD28+CD8+T cells ascended significantly at the first five years after ART (seen in Supplementary Table 4).

Antiretroviral treatment and delayed viral suppression showed little influence on the CD28+ expression on CD4+ and CD8+T cells (seen in Supplementary Table 5). In patients with CD4>100 cells/μl pretreatment whose percentage of CD4+CD28+T cells increased significantly more than patients with CD4<100 cells/μl before treatment. Age and gender do not affect CD4+CD28+T cells but a significantly slower increase of CD8+CD28+T cells in older people and males. Higher CD4+T cell and lower CD8+T cell pretreatment contributed to a quicker climb of the CD8+CD28+/CD8+T cells.

Patients

This cohort enrolled HIV–infected adults visiting the HIV/AIDS Clinic Center of Peking Union Medical College Hospital (PUMCH) and keeping follow-up since 2003. The cohort has been recruited for almost 20 years based on actual treatments, most of them ensuring regular follow-up. The criteria for enrollment were: (1) HIV-infection based on serology and HIV-1 RNA, (2) ART-naive, (3) age 18–65, (4) both genders (4) with regular follow-up documents. Patients with an AIDS-defining disease or CD4+T cell counts below 200 cells/μl were initiated with a combination of antiretroviral therapy prior to 2015, in accordance with national HIV/AIDS treatment guidelines. After 2015, if HIV infection was confirmed, patients were recommended for therapy. After meeting these criteria, we retrospectively recruited 662 HIV-infected patients followed by at least 2 years who had eliminated the HIV-1 RNA plasma load <50 copies/ml. The flow chart of participants selection was seen in Supplementary Figure 1. The study was approved by an independent ethics committee and the institutional review board of PUMCH (Peking Union Medical College Hospital, JS-1431). The trial was carried out in accordance with the principles of Good Clinical Practice and the Declaration of Helsinki. Written informed consent was obtained from all the participants.

Main laboratory analysis (including T cell subsets and HIV viral load) were conducted at baseline, followed by 3 months, 6 months, 12 months, and every 6 months thereafter. Plasma samples for laboratory tests were taken during routine diagnosis.

Laboratory test

In order to assess the T cell immune response in HIV infected patients, we compared lymphocyte subpopulations in subjects before and during treatment to reference range, which examined 56 healthy blood donors (36 male, mean age was 36 ± 10 years old) in PUMCH and established reference value range since 2002 [32, 33]. Immune profiles were analyzed by three-color flow cytometry (Epics XL flow cytometry; Beckman Coulter, USA) as previously described [34]. Freshly EDTA-anticoagulated whole blood was incubated, lysed, and tested. The combination of monoclonal antibodies panels is as following: CD3/CD8/CD4, CD3/CD16CD56/ CD19, HLA-DR/CD38/CD8, CD28/CD8/CD4, CD45RA/CD4 and isotype controls (Immunotech, France). By measuring white blood cell counts and lymphocyte differentials from routine blood tests, cell counts of lymphocyte subsets were calculated of the same sample. HIV-RNA plasma viral load was examined by COBAS Ampliprep/TaqMan 48 RT-PCR real-time test (Roche, CA, USA).

Statistical analysis

Kolmogorov-Smirnov test was used to examine data distribution. The mean standard deviation with mean were used for parametric data and 25% and 75% for non-parametric data. two groups and parametric data comparison need student-test, non-parametric data demands for Mann-Whit test, χ² test for categorical variables.

Analysis of lymphocyte subsets was performed in individuals with an initial CD4+T count and available CD4+T cell records for at least twice visits. The linear mixed model (MIXED) with repeated measurements intended for identifying the possible covariates and time effect on repeated measurements of the immune profile. The mixed linear model was used to represent T cell subpopulation trajectories. Let y_it (i = 1, 2, … , n; t = 1, 2, … , ni) be the immune indicators of HIV infected patient i at follow-up visit time t. The longitudinal immune indicators dynamics y_it depicted the following model:

$$y_i\text{(}t\text{)}=x_i^{\prime }\text{(}t\text{)}\beta +z_i^{\prime }\text{(}t\text{)}{b}_i+{\epsilon }_i\text{(}t\text{)}$$

where $x_i^{\prime }\text{(}t\text{)}$ and $z_i^{\prime }\text{(}t\text{)}$ represents time-varying covariates with fixed effects β and random effects b_i, respectively. Random effects represent the effects of each subject that cannot be explained by the observed covariates. The errors in our assumption are independent and follow a normal distribution, ε_i(t) ~N(0, σ), and that b_i and ε_i(t) are independent. The unstructured covariance of fixed effect was assumed for the covariance structure. Model fit statistics (e.g., AIC, BIC, etc) were used to choose the best covariance structure. To check the effect of time on CD4, we used the Scheffe-Test as post-hoc comparison. P-value < 0.05 (two-side) was regarded as significant.

Statistical analysis was performed with SAS software version 9.4 and SPSS software (SPSS^® for Windows^™ version 22.0, SPSS Inc., Chicago, IL, USA). Plots were drawn by GraphPad Prism 8.