Immune reconstitution efficacy and the risk factors of immune non-responsiveness after combined antiretroviral therapy in HIV-1 positive MSM and heterosexual population – a prospective cohort study

Abstract

Background

Immune reconstitution following combined antiretroviral therapy (cART) varies across populations and remains suboptimal in certain subgroups. This study assessed the trajectory and predictors of CD4⁺ T cell recovery among heterosexual (HET) and men who have sex with men (MSM) living with HIV-1 in Guangxi, China.

Methods

A prospective cohort of 458 newly diagnosed HIV-1 positive individuals (244 HET and 214 MSM) was followed from 2015 to 2024. Generalized additive models (GAM) were applied to evaluate CD4⁺ T cell trends. A Bayesian Markov Chain Monte Carlo (MCMC) generalized linear model (glm) estimated the posterior probability of immune recovery (Complete immune recovery (CIR), defined as two consecutive CD4⁺ T cell counts > 500 cells/µL after cART initiation.). Multivariable logistic regression identified risk factors for immune non-responsiveness.

Results

Median CD4⁺ T cell counts increased from 346 to 700 cells/µL in the HET group and from 404 to 778 cells/µL in the MSM group. However, CD4⁺ T cells declined during the first year among HET, and counts remained below 500 cells/µL after four years, while MSM surpassed this threshold earlier. Those with baseline CD4⁺ T cell counts < 150 cells/µL failed to achieve complete immune reconstitution even after six years. The risk of immune non-response among those with CD4⁺ T cell percentage ≤ 20%, 20%-30%, and 30%-40% were 10.82 (95% confidential interval (CI): 7.54–15.64, P < 0.001), 0.68 (95%CI: 0.49–0.94, P < 0.023), and 0.06 (95%CI: 0.02–0.12, P < 0.0001) times higher than other groups. Longer cART duration was correlated with improved recovery. Key factors associated with lower probability of immune recovery included being age ≥ 50 years, absence of cotrimoxazole prophylaxis (SMZ), heterosexual transmission, baseline CD4⁺ T cell < 350 cells/µL, and CD4⁺ T cell percentage ≤ 20%. Those who were divorced and those with shorter cART duration also exhibited reduced immune reconstitution rates.

Conclusions

Immune recovery after cART was slower and less complete in HET individuals and those with low baseline CD4⁺ T cell counts and low baseline CD4⁺ T cell percentage. Early diagnosis, prompt initiation of cART, long-term treatment adherence, and SMZ prophylaxis were critical to optimize immune reconstitution, particularly in high-risk subgroups.

Introduction

Acquired immunodeficiency syndrome (AIDS), caused by the human immunodeficiency virus (HIV), leads to varying degrees of immune deficiency, resulting in opportunistic infections (OIs) that can be life-threatening. Although the number of HIV newly reported in 2023 was lower than any year since the late 1980s, the AIDS epidemic remains a major global public health concern. According to an UNAIDS report, 1.3 million (1–1.7 million) people were newly infected with HIV in 2023, with a total of 39.9 million people living with HIV (PLHIV) [1]. By the end of 2023, China had 1.29 million PLHIV. Among them, 110,500 were newly reported and 98.50% were transmitted through sexual contact [2]. Men who have sex with men (MSM) have the highest risk of HIV infection; in 2022, 25.60% of newly reported HIV cases in China were occurred among MSM [3]. The MSM population is the most vulnerable group to HIV infection in East and Central China, accounting for 62.4% to 77.9% of newly reported HIV cases [4]. HIV transmission related to MSM is also a public health concern in major metropolitan areas such as Shanghai, Beijing, and Hangzhou, accounting for more than half of newly reported HIV cases [5]. Moreover, newly reported HIV cases among MSM in Southern China, such as Guangdong, Guangxi and other provinces, has increased recently [6, 7].

Sexual transmission accounts for the majority of newly reported HIV cases in mainland China [8]. The proportion of newly reported HIV cases through heterosexual transmission (HET) and MSM have risen from 48.30% to 9.10% in 2009 to 72.80% and 25.70% in 2023, respectively [2]. In contrast, the proportion of HIV transmission via injection drug use has declined significantly from 25.20% in 2009 to less than 2.50% in 2020 [9]. A previous study covering 2001–2018 reported an average HIV prevalence of 5.70% [95% confidential interval (95% CI): 5.40%-6.10%] among the Chinese MSM population, with the proportion of newly diagnosed cases increasing from 0.30% in 2005 to 23.30% in 2018 [10].

Since 2010, the MSM population has become the group with the highest HIV/AIDS prevalence in China [11]. After the first HIV/AIDS case was detected among injection drugs users (IDUs) in Guangxi in 1996, HIV spread rapidly throughout the province. The disease burden caused by HIV has remained high in Guangxi, which has consistently ranked among the top three in provinces nationwide in reported cases [12, 13]. Although IDU was once the primary transmission route in Guangxi, sexual transmission has become predominant route of transmission since 2006 [14]. In recent years, HIV/AIDS prevalence among the MSM population in Guangxi has increased from 3.90% in 2010 to 9.53% in 2020 [15]. A community-based HIV testing campaign conducted on and around university campuses in 2023 revealed that 43% of newly reported HIV cases were MSM-related (unpublished data), highlighting the rapid spread of MSM-associated HIV infection in urban areas.

In recent years, access to combination antiretroviral therapy (cART) has increases significantly worldwide. By the end of 2023, 77% of PLHIV globally were on cART, and 72% had achieved viral load (VL) suppression [16]. The introduction of cART has greatly reduced morbidity and mortality associated with opportunistic infections. cART partially or completely restores the damaged immune function, suppressing HIV-1 replication, and improving CD4⁺ T cell counts, a process known as immune reconstitution. However, some individuals, known as immunological non-responders (INRs), fail to achieve normal CD4⁺ T cell recovery despite achieving VL suppression (define as HIV-1 RNA < 50 copies/ml) following cART, putting them at higher risk of progression to AIDS, opportunistic infections, and death [17, 18]. Previous studies showed that 10–40% of PLHIV with sustained VL suppression did not fully recover their CD4⁺ T cell counts, resulting in increased morbidity and mortality, and raising concerns about immune dysregulation and its impact on long-term health outcomes [19, 20].

Different modes of sexual transmission may have varying impacts on immune function recovery in PLHIV. Studies have shown that the rate of CD4⁺ T cell recovery was significantly lower in HIV-1 positive HET individuals compared to MSM [4], leading to higher rate of virological failure and increased mortality in HET after cART. These differences may stem from variations in viral strains [21], or delayed diagnosis [22]. However, the long-term effects of transmission model on immune constitution, CD4⁺ T cell dynamics, and the risk factors for incomplete immune reconstitution or INRs in PLHIV have not been fully elucidated. Therefore, this study aims to compare the immune reconstitution efficacy and associated risk factors in virologically suppressed MSM and HET individuals newly diagnosed with HIV-1 between 2015 and 2024 at a HIV-treatment center in Guangxi, a province with a high HIV/AIDS burden and a rising prevalence among MSM.

Methods

Study design and participants

This cohort study included 458 newly diagnosed HIV-1–positive individuals enrolled between January 2015 and December 2024 at a designated HIV/AIDS treatment center in urban Nanning, Guangxi, southern China. Of the total cohort, 244 (53.3%) identified as heterosexual (HET) and 214 (46.7%) as men who have sex with men (MSM). All participants were antiretroviral therapy (cART) naïve at the time of HIV diagnosis.

Inclusion and exclusion criteria

Recruited individuals were required to match the following inclusion criteria: confirmed HIV-1 infection, aged ≥ 18 years, cART naïve at the time of recruitment but on an ART regimen during the 96 weeks of the study, and undetectable plasma HIV-1 VL at 96 weeks. Exclusion criteria were: the presence of active opportunistic infections, changes in the ART regimen, the use of drugs known to affect/modify CD4⁺ T cell counts, and did not achieve the VL suppression (Fig. 1).

Fig. 1.

Study flow chart. OIs: opportunity infections; ART: antiretroviral therapy; VL: viral load; HET: heterosexual transmission; MSM: men have sex with men; GAM: generalized additive models; CIR: complete immune reconstitution; MCMCglm: Markov Chain Monte Carlo generalized linear model; INR: immune non-responder. Legend: Fig. 1 indicated the study cohort enrollment and exclusion criteria

Data collection

Socio-demographic data, including gender, date of birth, age, marital status, transmission route, and WHO clinical stage were collected from electronic medical records. Clinical data included date of HIV-1 diagnosis, cART initiation, follow-up visits, CD4⁺ T cell testing, CD4⁺ T cell counts, and plasma HIV-1 VL measurements. Only those with confirmed virological suppression (defined as HIV-1 RNA < 50 copies/mL) and with at least two years of follow-up were included in the analysis.

Study setting

The study was conducted at a HIV/AIDS treatment center which was established in 2015 in urban Nanning. This center serves more than 100,000 residents, is part of the National HIV Program (NHP) and provides comprehensive services including voluntary counselling and testing (VCT), HIV diagnosis, community-based testing, cART, CD4⁺ T cell measurement, and routine follow-up care for PLHIV.

Definitions

This study used INRs, CD4⁺ T cell recovery trends, and the probability of immune recovery (IR) to evaluate the effect of immune reconstitution in both HET and MSM treatment cohorts.

INRs: INRs were defined as PLHIV with plasma HIV RNA<50 copies/ml, and total CD4⁺ T cell counts <350 cells/µL after two years of cART initiation. Immune responders were defined as PLHIV with plasma HIV RNA <50 copies/ml, and total CD4⁺ T cell counts ≥ 350 cells/µL after two years of cART initiation [20, 23].

Complete Immune Reconstitution (CIR): CIR is defined as the restoration or approximation in an impaired immune function to a normal immune system in PLHIV through cART. In our study, CIR was defined as two consecutive CD4⁺ T cell counts of > 500 cells/µL after cART initiation [24], regardless of baseline CD4⁺ T cell count level. Those with early baseline-high participants were also included them in the analysis, and were categorized as CIR if they were experienced two consecutive CD4 counts > 500 after cART initiation.

Immune reconstitution failure (IRF): IRF refers a CD4⁺ T cell counts less than 250 cells/µL, or a CD4⁺ T cell counts persistently below 100 cells/µL after 6 months on cART [25].

Research methodology

This prospective cohort study was conducted between January 2015 and December 2024 in an urban HIV-treatment center in Guangxi, Southern China. A total of 458 newly diagnosed, cART-naïve HET and MSM were included in this study. To evaluate immune reconstitution, participants underwent biannual follow-up visits during which CD4⁺ T cell counts and VL were measured. Longitudinal trends in CD4⁺ T cell recovery were analyzed over a ten-year period.

To model the trajectory of CD4⁺ T cell recovery, a generalized additive model (GAM) was used for both HET and MSM group to fit CD4⁺ T cell counts over a ten-year period.

Subsequently, a Bayesian generalized linear model (GLM) utilizing Markov Chain Monte Carlo (MCMC) simulation was applied to explore the association between VL, CD4⁺ T cell counts and sexual transmission route. The Bayesian MCMCglm approach was used to estimate the posterior probability of immune recovery across stratified subgroups.

Finally, to identify independent risk factors associated with INRs, a multivariate logistic regression model was constructed.

CD4⁺ T cell subgroups

We classified CD4⁺ T cell counts into six subgroups: ≤50, 51–150, 151–250, 251–350, 351–500, and > 500 cells/µL.

GAM modelling

Changes in CD4⁺ T cell counts after cART in the HET and MSM were fitted with GAM model by “mgcv” package in R. Cyclic splines was used as the smoother and a Poisson distribution was identified as the best conditional distribution to estimate the CD4⁺ T cell counts, with the Log function chosen as the link function in the model. The k parameter was set as 12 to capture the monthly cyclic effects and as 120 to account for monthly cumulative effects. The model explained 99.40% of the variance in CD4⁺ T cell counts. This optimal model was subsequently used to predict the trends in CD4⁺ T cell counts over the following twelve months.

Bayesian MCMCglm modelling

The posterior probabilities of immune reconstitution cross stratifications were estimated with Bayesian MCMCglm model.

For the HET and MSM groups, the posterior probabilities of CIR were estimated by using CD4⁺ T cell measurement from each follow-up. Covariables included baseline CD4⁺ T cell subgroups, age, sex (for the HET group), marital status, and whether or not taking cotrimoxazole prophylaxis for opportunistic infections. The number of warm-ups was set at 10,000, and the mean value was saved 1,000 posterior samples. In total, 501,000 posterior samples were drawn to estimate the probability of CIR.

Statistical analysis

To examine the differences of socio-demographic parameters between HET and MSM population, Chi-square tests were used. Long-term trends in immune reconstitution were assessed by modeling CD4⁺ T cell counts trajectories for HET and MSM cohorts over a ten-year period using a generalized additive model (GAM)implemented with the mgcv package in R. To further examine the relationship between CD4⁺ T cell counts, VL, and transmission routes, a Bayesian glm was employed using the MCMC sampling method. Analyses were conducted using the ‘MCMCglmm’ package in R. The posterior distribution of predicted probabilities was used to estimate immune recovery outcomes across HET and MSM strata. To identify determinants of INR, we constructed a multivariate logistic regression model. Variables included age, baseline CD4⁺ T cell counts, sex, transmission route, WHO stage, and time to cART initiation.

All statistical analyses were carried out by R software (version 4.5.0) and Rstudio (version 2025.05.0-496). The level of significance was set at 0.05.

Results

Socio-demographic profile in HET and MSM groups

A total of 458 PLHIV were included in the analysis. Demography characteristics of the study population were summarized in Table 1. The majority of individuals in the MSM group were under 30 years of age (78.50%) and single (88.79%), whereas the HET group was predominantly aged 30–50 years (47.95%), married (46.31%), and male (60.66%). Statistically significant differences were observed between HET and MSM groups in age, gender, marital status, education, occupation, and cotrimoxazole prophylaxis use. The range of follow-up duration in HET and MSM was 0-12.51 and 0-11.03 person-years, respectively. The median of follow-up in HET was 5.48 (IQR: 1.25–7.23) person-years, while median in MSM was 4.97 (IQR: 2.03–6.31) person-years.

Table 1.

Socio-demographic characteristics between HET and MSM

Variable	HET		MSM		χ2	P-value
	N	%	N	%
Age (years)					129.17	< 0.000
< 30	65	26.64	168	78.50
30 ~ 50	117	47.95	41	19.16
> 50	62	25.41	5	2.34
Total	244	100.00	214	100.00
Gender					104.16	< 0.000
Male	148	60.66	214	100.00
Female	96	39.34	0	0.00
Ethnic groups					3.15	0.207
Han	129	52.87	128	59.81
Zhuang	104	42.62	74	34.58
Others	11	4.51	12	5.61
Education					118.84	< 0.001
Junior high school and below	153	62.71	30	14.01
Technical secondary school or high school	43	17.62	57	26.64
College or above	48	19.67	127	59.35
Occupations					42.48	< 0.001
Farmer	45	18.44	3	1.40
Unemployment	52	21.31	45	21.03
Freelancer	51	20.90	46	21.50
Student	6	2.46	18	8.41
Other occupations	90	36.89	102	47.66
Marital Status					146.41	< 0.000
Single	83	34.02	190	88.79
Married	113	46.31	12	5.61
Divorced or Widowed	48	19.67	12	5.61
Cotrimoxazole					9.69	< 0.001
Yes	35	14.34	11	5.14
No	209	85.66	203	94.86
WHO stage					2.99	0.392
1	215	88.11	198	92.52
2	19	7.79	12	5.62
3	6	2.46	2	0.93
4	4	1.64	2	0.93
Baseline CD4± T Cell Counts	M:302.50	--	M:352.50	--	2.84	0.092
	(IQR:201.00, 454.50)		(IQR262.00, 499.00)
< 350 Cells/µL	140	57.38	105	49.07
≥ 350 Cells/µL	104	42.62	109	50.93

M median, IQR inter-quantile

CD4⁺ T cell counts and percentage recovery after cART initiation in HET and MSM

The median baseline CD4⁺ T cell counts were 302 (IRQ: 201, 454) cells/µL in the HET population and 352 (IRQ: 262, 499) cells/µL in MSM (P = 0.00063) (Fig. 2). After ten years of cART, the median CD4⁺ T cell counts increased to 700 (700 ± 947) cells/µL in HET and to 778 (778 ± 324 cells/µL in MSM (P < 0.001) (Fig. 2A and B). The median baseline and the most recent CD4⁺ T cell percentage were 9.69% and 27.89% in HET, respectively. While median baseline and the most recent CD4⁺ T cell percentage were 11.94% and 27.74% in HET (Fig. 2C).

Fig. 2.

Comparison of the baseline and most recent CD4⁺ T cell counts between HET and MSM. A Baseline CD4⁺ T cell counts. B Most recent CD4⁺ T cell counts. (C) Baseline and most recent CD4⁺ T cell percentage in HET and MSM. Legend: The figure indicates that the average baseline CD4+ T cell counts and percentage among MSM is significantly higher than HET. After ten years of cART, CD4+ T cell counts and the percentage improved significantly in both groups

Longer recovery period of CD4⁺ T cell counts after cART initiation in HET

Over the ten-year follow-up, both the HET and MSM groups demonstrated increases in CD4⁺ T cell counts following cART initiation. However, in the first year, the HET group experienced a decline, with CD4⁺ T cell levels remaining below 500 cells/µL, indicating delayed immune reconstitution (Fig. 3A). In contrast, the MSM group showed a consistent upward trend, maintaining CD4⁺ T cell counts above 500 cells/µL across the ten years period (Fig. 3B). Additionally, the time to achieve immune recovery was significantly shorter among MSM compared to HET (Table 2). GAM simulations confirmed a gradual recovery trajectory in both groups, showing that HET required approximately four years to exceed 500 cells/µL, while MSM reached this threshold in six months (Fig. 3).

Fig. 3.

Trends of CD4⁺ T cell counts after cART initiation in HET and MSM. A Trend of CD4⁺ T cell recover after cART initiation in HET. B Trend of CD4⁺ T cell recovery after cART initiation among MSM. Legend: Increasing CD4+ T cell counts in the ten-years after cART initiation in HET and MSM. A longer recovery period of CD4+ T cell counts after cART initiation in HET than in MSM. The black circle and triangle in (A) and (B) represent the CD4+ T cell measurements of HET and MSM. The red lines in A and B indicated the median trends. The black lines demonstrated the CD4+ T cell counts trend fitted by GAM. The black lines with shadows in (A) and (B) represented the trends in next twelve months predicted by GAM. The GAM model explained 99.4% of variation among HET and 98.6% among MSM

Table 2.

Immune reconstitution process in baseline CD4+ T cell subgroups

Transmission route	Follow-up years	CD4+ T cell subgroups (Median, cells/µL)
		≤ 50 (%)	51–150 (%)	151–250 (%)	251–350 (%)	351–500 (%)	> 500 (%)
HET	1 year	135.88 (1.65)	272.12 (8.42)	301.86 (17.45)	427.35 (17.58)	519.58 (23.21)	696.02 (31.87)
	2 years	--	270.50 (2.11)	347.62 (8.42)	483.43 (20.53)	566.68 (29.47)	724.09 (39.47)
	3 years	--	308.81 (1.23)	367.56 (9.26)	527.28 (11.73)	564.73 (27.78)	654.08 (50.00)
	4 years	--	283.63 (1.36)	413.75 (5.44)	570.86 (12.93)	594.08 (27.89)	742.43 (52.38)
	5 years	--	310.20 (0.75)	433.36 (6.82)	614.95 (7.58)	680.06 (27.27)	754.74 (57.58)
	≥ 6 years	--	449.00 (0.99)	514.00 (3.45)	583.00 (9.36)	821.00 (21.67)	916.00 (64.53)
MSM	1 year	654.50**** (0.01)	259.20 (2.55)	406.76*** (7.79)	448.76 (16.72)	552.20 (30.25)	678.85 (42.68)
	2 years	--	338.18 (0.56)	464.32*** (1.12)	570.06* (11.80)	625.50 (24.16)	710.78 (62.36)
	3 years	--	404.25 (0.76)	488.29** (3.05)	542.23 (10.69)	656.58 (31.3)	671.47 (54.20)
	4 years	--	455.27 (0.53)	489.88 (3.08)	546.11 (6.92)	628.94 (23.32)	730.81 (66.15)
	5 years	--	396.40 (0.72)	544.46* (1.14)	591.71 (3.41)	654.96 (17.46)	740.70 (77.27)
	≥ 6 years	--	490.70 (0.81)	580.49 (0.91)	674.53* (3.67)	799.59 (14.79)	824.23 (79.82)

1.--: data is not available

2. Significance level: * <0.05; ** <0.01; ***<0.001; ****<0.0001

CIR baseline CD4⁺ T cell counts among subgroups

PLHIV in the HET and MSM population with baseline CD4⁺ T cell counts of 51–150 cells/µL and ≤ 50 cells/µL, exhibited slower immune reconstitution following cART initiation, with CD4⁺ T cell counts remaining below 500 cells/µL even after six years of treatment. In individuals with baseline CD4⁺ T cell counts of 151–250 and 251–350 cells/µL, it took six and three years among HET, and five and two years among MSM to reach 500 cells/µL. In contrast, those with baseline CD4⁺ T cell counts > 350 cells/µL achieved CIR within the first year of cART (Fig. 4; Table 2).

Fig. 4.

Baseline CD4⁺ T cell counts impacted the immune reconstitution among HET and MSM. Legend: Figures indicated that HET in the same subgroup of baseline CD4+ T cell counts had a slower, longer immune reconstitution than MSM

GAM was used to estimate the annual CD4⁺ T cell recovery rates after cART initiation for HET and MSM groups. Results indicated a sustained increase in CD4⁺ T cell counts over time for both HET and MSM groups (Fig. 5).

Fig. 5.

Changing trajectories of CD4⁺ T cell rate in HET and MSM between 2015 and 2024. Legend: Figures indicated that variate trajectory of CD4+ T cell among HET and MSM population

Estimated probability of CIR

We applied a Bayesian MCMCglm model to estimate the posterior probability (PP) of CIR across different stratifications. Results showed that individuals age > 50 years had a significantly lower probability of CIR compared to those aged < 50 years (PP = -0.119, P < 0.01). Divorced individuals also exhibited a reduced likelihood of CIR compared to those were single (PP = -0.065, P < 0.05). Absence of cotrimoxazole prophylaxis was associated with a markedly lower probability of CIR (PP = -0.332, P < 0.01). In contract, MSM had a significant higher probability of CIR compared to HET (PP = 0.075, P < 0.01). Both baseline CD4⁺ T cell counts and cART duration were independently associated with immune recovery. Compared with baseline CD4⁺ T cell counts <=50 cells/µL, those in the 51–150, 151–250, 251–350, 351–500, and > 500 cells/µL subgroups demonstrated progressively higher probabilities of achieving CIR. Additionally, longer cART duration was independently associated with increased CIR likelihood (Table 3).

Table 3.

Estimation of posterior probabilities on CIR across stratifications using a Bayesian MCMCglm model

Variable	Posterior	Low-95% CI	Up-95% CI	P-value*
Age (years)
< 30
30 ~ 50	0.001	-0.041	0.046	0.941
> 50	-0.119	-0.173	-0.056	< 0.002 **
Gender
Male
Female	0.012	-0.030	0.048	0.538
Marital Status
Single
Married	-0.033	-0.083	0.011	0.232
Divorced or Widowed	-0.065	-0.116	0.001	0.029 *
Unknown	-0.264	-1.006	0.477	0.473
Cotrimoxazole
Yes
No	-0.332	-0.383	-0.265	< 0.002 **
Transmission route
HET
MSM	0.075	0.037	0.107	< 0.002 **
CD4+ T Cell Counts at cART Initiation (Cells/µL)
≤ 50
51–150	0.491	0.381	0.596	< 0.002 **
151–250	0.643	0.557	0.759	< 0.002 **
251–350	0.813	0.719	0.918	< 0.002 **
351–500	0.964	0.868	1.067	< 0.002 **
> 500	1.204	1.115	1.327	< 0.002 **
Duration of cART (years)
0~
1~	0.254	0.209	0.305	< 0.002 **
2~	0.291	0.247	0.337	< 0.002 **
3~	0.349	0.300	0.402	< 0.002 **
4~	0.394	0.344	0.442	< 0.002 **
5~	0.507	0.455	0.562	< 0.002 **
6~	0.576	0.530	0.625	< 0.002 **

*(1) Significance level: ‘***’: P < 0.001; ‘**’: P < 0.01; ‘*’: P < 0.05; ‘.’: P < 0.1; ‘ ’: P = 1

* (2) “-” indicated a decreased probability

Risk factors associated with INRs

Among participants in this study, 326 completed at least two years of cART. Multivariate logistic regression analysis identified several factors significantly associated with INRs. Those aged > 50 years had 3.54-flod (adjusted OR = 3.54, 95% CI: 1.56–8.01, P = 0.002), those aged 30–50 years had 1.91-fold (adjusted OR = 1.91, 95%CI: 1.03–3.58, P = 0.042) higher risk of INR compared to those < 30 years, respectively. Lacking cotrimoxazole prophylaxis was associated with a 7.52-fold increased risk of INR (OR = 7.52, 95% CI: 3.71–15.24, P < 0.001). Those with baseline CD4⁺ T cell counts ≥ 350 cells/uL had 86% lower risk of INR compared to those with counts < 350 cells/µL (OR = 0.14, 95% CI: 0.06–0.30, P < 0.001). those with CD4⁺ T cell percentage ≤ 20%, 20%-30%, and 30%-40% were 10.82 (95%CI: 7.54–15.64, P < 0.001), 0.68 (95%CI: 0.49–0.94, P < 0.023), and 0.06 (95%CI: 0.02–0.12, P < 0.0001) times higher than other groups. The relatively wide CIs of aged > 50 years, Lacking cotrimoxazole prophylaxis, and CD4⁺ T cell percentage ≤ 20% suggested that these effect sizes should be interpreted with caution due to the limited sample size in these specific subgroups. Additionally, MSM had a 47% lower risk of INR than the HET (OR = 0.53, 95% CI: 0.30–0.95, P < 0.05). No statistically significant associations were observed for gender, marital status, and WHO disease stage (Table 4). CD4/CD8 ratio was also a robust biomarker for predicting immune recovery. The longitudinal evolution of CD4/CD8 ratio analysis showed the average CD4/CD8 ratio was 0.44 in INR population and 0.80 in IR population (Fig. 6A).

Table 4.

Risk factors associated with INRs across stratifications

Variable	Frequency	INRs (%)	OR (95% CI)	P-value
Age (years)
< 30	156	20(12.82)	--	--
30–50	132	29(21.97)	1.91(1.03,3.58)	0.042
> 50	38	13(34.21)	3.54(1.56,8.01)	0.002**
Gender
Male	257	51(19.84)
Female	69	11(15.94)	0.77(0.38,1.56)	0.464
Marital Status
Single	191	31(16.23)	--	--
Married	95	21(22.11)	1.46(0.79,2.72)	0.227
Divorced Or Widowed	40	10(25.00)	1.72(0.76,3.88)	0.191
Cotrimoxazole
Yes	286	40(13.99)	--	--
No	40	22(55.00)	7.52(3.71,15.24)	< 0.001***
Transmission route
HET	170	40(23.53)	--	--
MSM	156	22(14.10)	0.53(0.30,0.95)	0.032*
Baseline CD4+ T Cell Counts (Cells/uL)
< 350	182	54(29.67)	--	--
≥ 350	144	8(5.56)	0.14(0.06,0.30)	< 0.001***
CD4+ T Cell
Percentage (%)
≤ 10	15	15(100.00)	--
≤ 20	205	121(59.02)	10.82(7.54,15.64)	< 0.0001****
≤ 30	431	75(14.82)	0.68(0.49,0.94)	< 0.023*
≤ 40	321	8(2.49)	0.06(0.02,0.12)	< 0.0001****
> 40	70	0(0.00)	--	< 0.0001****
WHO stage
1	297	54(18.18)	--	--
2	19	6(31.58)	2.08(0.76,5.71)	0.76
3	5	1(20.00)	1.12(0.12,10.27)	0.12
4	5	1(20.00)	1.12(0.12,10.27)	0.12

(1) The frequency of CD4⁺ T Cell percentage in each sub-group referred to the person-time that PLHIV received follow-up during 120 months

*(2) Significance level: ‘****’: P < 0.0001, ‘***’: P < 0.001, ‘**’: P < 0.01, ‘*’: P < 0.05

Fig. 6.

CD4_CD8 ratio and CD4% longitudinal evolution in INR and IR. Legend: A: The red line indicated that most CD4/CD8 ratios in the INR group were < 0.5, with no significant temporal increase. CD4/CD8 ratios in the immunological responder (IR) group were predominantly higher than those in the INR group. B: The majority of CD4+ T cell percentage in the INR group were below 20%, whereas those in the IR group ranged from 20% to 40%. The CD4+ T cell percentages in IR and INR groups remained stable throughout the observation period

Discussion

In this study, we evaluated the efficacy and determinants of CIR following cART in an HIV-1–positive cohort focused on sexually transmitted cases (HET and MSM) at an AIDS treatment center in urban Guangxi, Southern China, from 2015 to 2024. To our knowledge, this represents the longest longitudinal cohort on HIV/AIDS immune recovery reported in mainland China. Our ten-year analysis demonstrated that cART led to CD4⁺ T cell recovery in both HET and MSM populations, regardless of baseline CD4⁺ T cell counts. However, immune reconstitution was significantly more robust among MSM compared to HET. A longer duration of cART was associated with higher CIR rates across all baseline CD4⁺ T cell subgroups.

Factors associated with a lower probability of CIR included being age > 50 years, divorced, an absence of cotrimoxazole prophylaxis, HET, baseline CD4⁺ T cell counts < 350 cells/µL, and shorter cART duration. Similarly, increased risk of immune non-response was observed among older individuals, those not receiving cotrimoxazole prophylaxis, individuals with lower baseline CD4⁺ T cell counts, and members of the HET group.

Initiation of cART leads to rapid inhibition of intracellular HIV replication and a marked reduction in plasma viral load, promoting immune reconstitution and increased CD4⁺ T cell counts [26]. Consistent with previous studies [4], our findings indicate that immune recovery was slow among both HET and MSM. However, MSM demonstrated significantly more efficient immune reconstitution, with mean CD4⁺ T cell counts exceeding 500 cells/µL within six months of cART initiation. In contrast, the HET group showed delayed recovery, with CD4⁺ T cell levels remaining below 500 cells/µL after one year and requiring up to four years to surpass this threshold. These findings suggest that HET may require a longer duration of cART to achieve complete immune reconstitution. Clinical treatment centers should identify HET as a priority population and allocate resources to promote adherence to their cART regimens.

It’s well-documented that variations in viral pathogenicity of HIV are, in part, attributed to its subtypes and circulating recombinant forms. Recently, we performed a HIV-1 molecular epidemiological investigation among individuals with viral suppression failure in the north of Guangxi. 841 Pol genes were sequenced, and the majority of HIV-1 subtype among those with virus suppression failure was CRF_01AE (66.11%), followed by CRF_08BC (24.14%) and CRF_07BC (6.67%). Our unpublished data indicate that individuals with CRF_01AE exhibit a higher risk of immune not-response. These findings align with prior studies conducted in Asian populations, which have confirmed that CRF_01AE is more pathogenic than CRF_07BC strains [4, 24, 27]. Specifically, CRF_01AE infection is associated with accelerated disease progression, elevated mortality, and poorer immune reconstruction outcomes.

Superior immune recovery among MSM was also evident across baseline CD4⁺ T cell strata. Several factors likely contribute to this difference. First, the “Testing as Prevention” strategy has prioritized high-risk populations such as MSM, resulting in earlier HIV diagnosis and prompt initiation of treatment [28, 29]. Second, HIV strains transmitted via heterosexual contact are often more virulent [21], leading to more extensive CD4⁺ T cell depletion and a higher likelihood of immune reconstitution failure [30, 31]. Third, delayed diagnosis is more prevalent among the HET population. Our previous research showed that 89% of late diagnoses in China occurred in HET individuals [32], and HET transmission has been identified as an independent risk factor for delayed HIV detection [33, 34]. Regions with HET as the majority in HIV transmission should consider community-based testing strategies to detect HIV-positive HET individuals earlier, thereby avoiding subsequent immune reconstitution failure.

Earlier diagnosis among MSM contributes to more favorable immune outcomes [22]. Conversely, advanced-stage HIV case (e.g., CD4⁺ T cell counts < 200 cells/µL or WHO stage III/IV) is associated with prolonged viral replication, diminished CD4⁺ recovery, and reduced likelihood of full immune reconstitution despite cART [35, 36].

Moreover, late diagnosis increases vulnerability to opportunistic infections—such as tuberculosis, cryptococcosis, candidiasis, Pneumocystis pneumonia, and toxoplasmosis—which further compromise immune recovery and survival [37–40]. These factors may help to explain the observed differences in immune reconstitution between HET and MSM populations and are consistent with our study findings.

In this cohort, participants were stratified by baseline CD4⁺ T cell counts to prospectively assess the impact of initial immunologic status on immune reconstitution during cART. Our findings reveal a strong association between lower baseline CD4⁺ T cell counts and suboptimal immune recovery. Notably, individuals with baseline counts below 150 cells/µL failed to achieve complete immune reconstitution over the 120-month follow-up, despite sustained adherence to cART, indicating persistent immune dysfunction. Similarly, a previous study has indicated that a baseline CD4⁺ T cell counts below 200 cells/µL increased mortality risk in MSM living with HIV on cART [41]. These results are consistent with findings from a Dutch cohort study [42], though our longer follow-up period and finer stratification provide more detailed insights. Additionally, study showed that the CD4⁺ T cell count increase did not vary in patients who control viremia with HAART despite starting therapy at different CD4⁺ cell count levels [43]. It means that different levels of baseline CD4⁺ T cell share a common increasing pattern during the cART period. Therefore, PLHIV with higher baseline CD4 count systematically reach 500 in a shorter period of time after initiated cART. Bring all together, we conclude that baseline CD4 differences can drive recovery timing for immune reconstitution to CD4⁺ >500 cells/µL under the achievement of viral suppression. In the HET group, individuals with baseline CD4⁺ T cell counts of 150–250 and 250–350 cells/µL required more than six and three years, respectively, to reach immune recovery, while in the MSM group, the corresponding durations were five and two years. These observations align with prior studies reporting that individuals with baseline CD4⁺ counts ≥ 350 cells/µL generally achieve immune reconstitution within 36 months of therapy initiation [44]. Collectively, these findings highlight that the baseline CD4 differences can drive recovery timing and may partly explain group differences furthermore, it is emphasized that the critical importance of early cART initiation to restore CD4⁺ T cell levels and improve long-term immune outcomes.

A key finding of this study is that prolonged cART duration is significantly associated with improved immune reconstitution. This association was consistent across both HET and MSM populations, as well as among subgroups stratified by baseline CD4⁺ T cell counts. Our data provide a detailed timeline of CD4⁺ T cell recovery, with GAM modeling illustrating distinct trajectories based on baseline immunologic status. Consistent with prior studies, including a five-year cohort analysis that reported the most rapid CD4⁺ T cell gains during the first year followed by a plateau [45], our findings confirm this pattern among MSM.

However, among the HET population, those with baseline CD4⁺ T cell counts below 150 cells/µL did not experience a marked increase in CD4⁺ T cell counts within the first year of treatment, instead showing delayed or minimal recovery. This highlights a critical distinction in immune response by transmission category that previous studies, which lacked stratification by transmission mode, did not address.

Our ten-year follow-up provides more granular insights into long-term immune recovery, especially among those with advanced immunosuppression at baseline. Findings from other cohorts, including those from the UK [46], support the notion that extended cART durations—particularly over three to five years—are associated with sustained immune restoration. Our results reinforce the need for continued adherence to long-term cART, particularly for individuals with severe baseline immunodeficiency, even in the absence of early immune gains. Sustained treatment remains essential for achieving optimal long-term immune outcomes in people living with HIV.

We further examined the determinants of poor immune responders (PIRs) and INRs after ten years of cART, stratified by demographic and clinical factors. The analysis revealed overlapping risk factors for both outcomes. Specifically, those aged ≥ 50 years, those not receiving cotrimoxazole prophylaxis, individuals with baseline CD4⁺ T cell counts < 350 cells/µL, CD4⁺ T cell percentage ≤ 20%, HET, being divorced, and shorter cART duration were all associated with reduced likelihood of immune reconstitution. Similarly, those age ≥ 50 years, absence of cotrimoxazole prophylaxis, HET, and baseline CD4⁺ T cell counts < 350 cells/µL were significantly associated with increased risk of immune non-response. Prior studies have shown accelerated naïve T cell depletion with aging [46, 47], indicating that older age may significantly impair long-term CD4⁺ T cell recovery. These findings suggested the importance of early diagnosis, timely treatment initiation, and tailored intervention strategies. However, despite the factors of age ≥ 50 years, absence of cotrimoxazole prophylaxis, and CD4⁺ T cell percentage ≤ 20% were statistically significant with the immune non-response, the wide 95% CIs for these variables suggest a degree of statistical imprecision. The point estimates (ORs values) should be viewed as indicator of a strong trend rather than precise effect values. Therefore, those potential association require validation in larger, multi-center cohorts in the future.

Our analysis revealed that divorced and married individuals had a lower likelihood of achieving immune recovery compared to single individuals. This may be partially attributed to age, as single individuals tended to be younger and therefore more likely to experience effective immune reconstitution. These findings are consistent with previous research [33]. Additionally, cotrimoxazole prophylaxis was significantly associated with improved immune recovery in our cohort. HIV co-infections with pathogens such as hepatitis B virus, hepatitis C virus, Epstein-Barr virus, cytomegalovirus, and Mycobacterium tuberculosis (MTB) have been shown to enhance HIV replication and sustain viral reservoirs [48]. These co-infections trigger chronic immune activation, which contributes to CD4⁺ T cell depletion and incomplete immune reconstitution, even after cART initiation. In this context, cotrimoxazole prophylaxis plays a critical role in reducing opportunistic infections and mitigating immune activation. For instance, in sub-Saharan Africa—where HIV-MTB co-infection is prevalent—cotrimoxazole has significantly reduced AIDS-related mortality due to MTB [49]. In Guangxi, the region with the highest incidence of HIV-MTB co-infection in China [50], the National AIDS Control Program has promoted cotrimoxazole use among PLHIV. This intervention likely contributed to the improved immune recovery observed among those receiving prophylaxis in our study.

Our study excluded PLHIV who had changed in their ART regimen from the cohort. This approach may help to reduce the immune-reconstitution confounding due to treatment failure or toxicity and ensured homogeneity of antiretroviral exposure during the study period. However, it may have introduced selection bias. Specifically, the individual excluded represented a population with greater clinical complexity, increased treatment difficulty, or poorer drug tolerance (e.g., those with a history of multiple treatment failures, drug resistance or multidrug resistance, or severe adverse drug reactions). Consequently, the conclusions of this study may be more applicable to HIV-infected individuals with good tolerance to their current ART regimen and relatively stable disease, and may not fully represent all PLHIV who received ART, particularly those required frequent regimen adjustments.

This study has several limitations. First, we did not assess absolute CD4⁺ T cell gains across baseline subgroups, limiting our ability to evaluate the magnitude of immune reconstitution relative to initial immune status. Second, the sample size of individuals with baseline CD4⁺ T cell counts ≤ 50 cells/µL, absence of cotrimoxazole prophylaxis, and CD4 + T cell percentage ≤ 20% were insufficient to reliably analyze immune recovery in this subgroup. These wide intervals reflected sampling fluctuations and suggested that while the associations are statistically robust, the exact magnitude of the risk requires further validation in larger, multi-center cohorts to improve precision. Third, MSM were younger than HET in our cohort, and age was not included as a potential confounding factor on CD4⁺ T-cell recovery in PLHIV under cART. Fourth, due to the observational nature of this study and the routine clinical data collection protocol in the study setting, this study could not include measurements of immune senescence markers (i.e., CD57, CD28), T cell differentiation markers (i.e., CCR7, CD45RA), and cytokine polyfunctionality assays in our study.

Despite these limitations, the study has notable strengths. We utilized ten years of longitudinal data from an HIV-positive cohort to model CD4⁺ T cell dynamics following cART, stratified by transmission route (HET and MSM), using generalized additive models (GAM). To our knowledge, this represents the longest observational dataset examining immune reconstitution post-cART in mainland China. Our analysis elucidated immune recovery trajectories across baseline CD4⁺ T cell strata and provided clear timelines for achieving complete immune reconstitution (CIR). Additionally, we applied a Bayesian MCMCglm approach to estimate the posterior probability of immune recovery and identified key predictors of suboptimal reconstitution. Multivariate logistic regression further revealed factors associated with immune non-response, offering comprehensive insights into long-term immune recovery outcomes. While most of our findings align with previous reports on cART-associated immune reconstitution, we observed a unique trend: a decline in CD4⁺ T cells during the first year of cART among the HET population, which contrasts with earlier studies. These results contribute valuable insights into the heterogeneity of immune recovery and support the need for tailored treatment strategies in stratified HIV-positive population.

Conclusions

Heterosexual individuals exhibited slower CD4⁺ T cell recovery and a lower likelihood of immune reconstitution compared to MSM. CD4⁺ T cell decline persisted over 12 months, particularly in those with baseline counts < 150 cells/µL. Lower CD4⁺ T cell percentage indicated a high risk of immune non-response. Early cART initiation and prolonged treatment were associated with improved immune outcomes. Key risk factors for immune non-response included older age, lack of cotrimoxazole prophylaxis, heterosexual transmission, and low baseline CD4⁺ counts. These findings underscore the importance of early diagnosis, timely treatment, and targeted interventions for high-risk populations to enhance immune recovery.

Abbreviations

cART

Combined Antiretroviral Therapy

HET

Heterosexual

MSM

Men who have sex with men

GAM

Generalized additive models

MCMC

Markov Chain Monte Carlo

GLM

Generalized linear model

CIR

Complete immune reconstitution

AIDS

Acquired Immunodeficiency Syndrome

HIV

Human Immunodeficiency Virus

PLHIV

People living with HIV

IDUs

Injection drugs users

VL

Viral Load

INR

Immunological non-responders

NHP

National HIV Program

VCT

Voluntary counselling and testing

IR

Immune recovery

IRF

Immune reconstitution failure

PIRs

Poor immune responders

MTB

Mycobacterium tuberculosis

Authors’ contributions

Jiangshan Wang: Formal Analysis, Writing – original draft. Wancha Huang: Data curation, Writing – review & editing. Ying Liu: Project administration, Resources, Writing – review &editing. Yanan Hou: Data curation, Investigation, Writing – review & editing.Jinda He: Visualization, Writing –review & editing . Liuqin Chen: Data curation, Investigation, Writing–review & editing . Haoge Lin: Investigation, Writing – review & editing.Hai Li: Resources, Writing – review & editing. Zhaohua Lu: Project administration, Resources, Supervision, Writing- review & editing. Jian Xiao: Conceptualization, Resources, Writing – review & editing.Zhigang Zheng: Conceptualization, Funding acquisition, Methodology, Writing – review & editing.

Funding

The author(s) declare that financial support was received for the research and/or publication of this article. This work was supported by: China National Natural Scientific Fund (NO. 82460390); Research Initial Fund for Introduced Doctor of Guangxi University of Chinese Medicine (NO.2022BS019); Category A High-Level Talent Cultivation and Innovation Team of Guangxi University of Chinese Medicine (2022A006).

Data availability

All data in this study are available upon request by contacting the corresponding author.

Declarations

Ethics approval and consent to participate

The study protocol was reviewed and approved by Medical Ethics Management Committee of The Sixth People's Hospital of Nanning, Ethical approval number:2025070901. Written informed consent was obtained from all participants at enrollment, and all procedures adhered to the ethical standards outlined in the Declaration of Helsinki.

Competing interests

The authors declare no competing interests.