Pretreatment HIV-1 drug resistance and molecular subtype characteristics among newly diagnosed individuals in Guangxi, China

Abstract

Background

The scale-up of antiretroviral therapy (ART) in China has increased concern regarding HIV-1 pretreatment drug resistance (PDR), particularly in Guangxi due to high subtype diversity and population mobility. However, the epidemiology and transmission dynamics of PDR in the region remain insufficiently characterized.

Methods

We assessed PDR prevalence, subtype distribution, and transmission dynamics among ART-naïve individuals in central Guangxi. Between 2018 and 2023, 1,226 newly diagnosed individuals were screened, of whom 1,214 were included after quality control. Drug resistance mutations and subtypes were determined using the Stanford HIV Drug Resistance Database. Molecular transmission networks were constructed using the TN93 model with a genetic distance threshold of < 1.5%.

Results

The study population was predominantly male (68.8%) and older (57.6% aged > 45 years). Overall, PDR prevalence was 6.0% (73/1,214; 95% CI 4.8–7.4%), ranging annually from 4.3 to 8.3%. CRF01_AE was the dominant subtype (74.3%), followed by CRF07_BC (11.1%) and CRF08_BC (10.3%). Higher PDR prevalence was observed in CRF08_BC (11.1%; 95% CI 6.6–17.9%) and CRF07_BC (8.8%; 95% CI 5.1–15.0%). Network analysis identified 102 transmission clusters comprising 612 sequences (50.4%), including two clusters containing ≥ 2 individuals sharing identical NNRTI resistance mutations, all linked through heterosexual transmission.

Conclusions

In central Guangxi, HIV-1 PDR remains moderate, primarily driven by NNRTI-associated mutations, with limited evidence of clustering within transmission networks. These findings highlight the need for continued resistance surveillance and optimization of first-line ART regimens.

Trial registration Not applicable. This retrospective cohort study was not registered as a clinical trial.

Introduction

Since the global scale-up of antiretroviral therapy (ART), the burden of HIV/AIDS has declined substantially, leading to notable gains in survival and quality of life for people living with HIV and marked reductions in HIV incidence and mortality [1, 2]. However, the effectiveness of ART is critically dependent on viral susceptibility to antiretroviral agents. As ART coverage has expanded, pretreatment drug resistance (PDR) resulting from resistance-associated mutations has emerged as an increasingly important challenge to treatment success and public health [3, 4]. PDR may compromise virologic suppression, increase mortality risk, and contribute to ongoing HIV transmission [5, 6]. Evidence from multiple countries indicates that PDR among ART-naïve individuals with HIV infection has either increased or has stabilized at moderate prevalence levels. In Thailand, for instance, PDR prevalence rose from 0% to 5.2% between 2003 and 2006 and was closely associated with treatment failure among sexual partners and suboptimal adherence [7]. In Mexico, PDR prevalence remained relatively stable at 7.4% from 2002 to 2014 [8]. Similarly, a study from South Africa reported a PDR prevalence of 12.9%, which was associated with a markedly increased risk of first-line efavirenz treatment failure (adjusted hazard ratio [aHR] = 3.7) [9]. Taken together, these findings underscore that the accumulation and onward transmission of PDR represent a growing challenge to public health systems globally.

In China, particularly in the southwestern regions such as Guangxi, HIV-1 PDR has shown a rising trajectory, with moderate resistance observed, especially to non-nucleoside reverse transcriptase inhibitors (NNRTIs) [5, 10, 11]. The prevalence of PDR in this region has been reported to be as high as 18.7%, a trend driven by high HIV-1 subtype diversity, substantial population mobility, and densely interconnected transmission networks [11–13]. Within Guangxi, studies have documented a sustained increase in PDR in Nanning, predominantly driven by NNRTI resistance, with significant clustering of resistance-associated mutations observed in dominant subtypes such as CRF08_BC [14]. These findings suggest the presence of expanding drug-resistant transmission chains, highlighting the urgent need for strengthened surveillance and targeted public health interventions.

During the study period (2018–2023), first-line ART in Guangxi and across China primarily comprised NNRTI-based regimens, typically efavirenz (EFV) or nevirapine (NVP) combined with NRTIs, consistent with national treatment guidelines at that time [15, 16]. However, in 2021, China’s National Health Commission recommended transitioning to integrase strand transfer inhibitor (INSTI)-based regimens, particularly dolutegravir (DTG), as preferred first-line therapy, aligning with World Health Organization (WHO) guidelines [17–19]. This transition was not fully implemented during the study period, which explains the observed predominance of NNRTI resistance in this cohort. Understanding the PDR landscape during this transitional period is essential for informing current treatment optimization strategies.

Given these challenges, robust assessment of PDR levels is essential to inform national antiretroviral drug policies and optimize ART programs. Routine pretreatment genotypic drug resistance testing plays a critical role in identifying drug-resistant HIV strains among newly infected individuals, guiding the selection of individualized ART regimens, and ultimately reducing the risk of early treatment failure [20, 21]. Additionally, molecular epidemiology and transmission network analysis offer valuable insights into the dynamics of resistance-associated mutations and can help identify high-risk transmission clusters [12]. Establishing region-specific drug resistance surveillance frameworks that incorporate local resistance profiles and viral subtype characteristics is essential for maintaining the long-term effectiveness of ART.

To address these gaps, we performed a retrospective analysis of individuals newly diagnosed with HIV infection at the Guangxi Zhuang Autonomous Region Chest Hospital between 2018 and 2023. The objectives of this study were to assess the prevalence of PDR and HIV-1 subtype distribution and to explore the clustering patterns and transmission dynamics of resistance-associated mutations using molecular transmission network analysis. Our findings are aimed at informing more targeted regional HIV prevention and optimizing the initial ART regimen selection to reduce PDR-associated treatment failure.

Materials and methods

Study design and subjects

This retrospective cohort included individuals who were newly diagnosed with HIV-1 infection at the Guangxi Zhuang Autonomous Region Chest Hospital between January 2018 and December 2023. The Guangxi Zhuang Autonomous Region Chest Hospital is a major referral center for HIV/AIDS in central Guangxi, specializing in infectious disease management. The hospital receives patients from across the region, particularly those with advanced disease or complications requiring specialized care, which may introduce selection bias toward more clinically advanced cases. All participants were ART-naïve at the time of diagnosis. To be eligible, participants had to be ≥ 18 years of age, have laboratory-confirmed HIV-1 infection, and have no prior exposure to ART. Plasma samples were collected for HIV-1 genotypic drug resistance testing, with successful retrieval of target sequences. Only individuals with complete clinical data were included in the analysis.

Ethical approval

Written informed consent was obtained from all participants before enrollment. The study protocol was reviewed and approved by the Medical Ethics Committee of the Guangxi Zhuang Autonomous Region Chest Hospital [Approval No.: 2024-S027-01]. All procedures were conducted in accordance with the Declaration of Helsinki and its subsequent amendments. Participant confidentiality was maintained throughout the study, and all personal data were anonymized before analysis.

Sample collection

For each participant, a total of 10 mL of peripheral venous blood was collected into K2-EDTA anticoagulated tubes. The samples were centrifuged at 500 × g for 10 min to separate plasma, which was aliquoted into microcentrifuge tubes and stored at − 20 °C until further analysis.

HIV-1 RNA quantification

HIV-1 RNA levels were quantified using the Roche cobas AmpliPrep/cobas TaqMan fully automated PCR platform, following the manufacturer’s protocol.

HIV-1 genotypic drug resistance testing and subtype determination

Plasma samples were transported at − 20 °C to Beijing Anpu Biochemical Technology Co., Ltd. for genotypic drug resistance testing. HIV-1 RNA was extracted using a commercial nucleic acid extraction and purification kit. The extracted RNA was reverse-transcribed and amplified by polymerase chain reaction (PCR) to obtain HIV-1 protease (PR) and reverse transcriptase (RT) gene fragments corresponding to positions 2253–3553 of the HXB2 reference strain. PCR amplicons were visualized via 1% agarose gel electrophoresis and then subjected to Sanger sequencing. Sequence data were analyzed using the Stanford HIV Drug Resistance Database (HIVdb; version 9.0) for genotypic resistance interpretation and HIV-1 subtype classification. Resistance levels were classified as susceptible (S), potential resistance (P), low-level resistance (L), intermediate resistance (I), or high-level resistance (H), with viral strains classified as L, I, or H considered drug-resistant.

PCR amplification conditions PCR amplification of the HIV-1 pol gene fragments (protease and reverse transcriptase regions, HXB2 positions 2253–3553) was performed using a nested PCR approach. The first-round reverse transcription PCR (RT-PCR) was performed under the following conditions: reverse transcription at 50 °C for 30 min; initial denaturation at 94 °C for 2 min; followed by 40 cycles of 94 °C for 30 s, 55 °C for 30 s, and 68 °C for 90 s; and a final extension at 68 °C for 10 min. Second-round nested PCR was performed under the following cycling conditions: initial denaturation at 94 °C for 5 min; 30 cycles of 94℃ for 30 s, 63℃ for 30 s, and 72℃ for 2.5 min; and a final extension at 72 °C for 10 min. All amplifications were carried out on a GeneAmp PCR system 9700 thermocycler (Applied Biosystems HIV-1 Genotyping Kit, USA) according to the manufacturer’s instructions.

Quality control criteria Only sequences that met the following criteria were included in downstream analyses: (1) sequence length ≥ 1000 bp; (2) average Phred quality score (QS) ≥ 25 (with an estimated sequencing error rate of < 0.3%); (3) < 5% mixed bases/ambiguous nucleotides (N) per sequence; (4) successful amplification of the entire pol region (protease and reverse transcriptase, HXB2 positions 2253–3553); (5) no evidence of APOBEC-mediated hypermutation (checked by examining G-to-A hypermutation patterns); (6) no evidence of laboratory contamination or sample mix-up, confirmed by phylogenetic analysis with reference sequences. Sequences with insertions or deletions (indels) that caused frame shifts were manually reviewed and excluded if they affected drug resistance interpretation.

Duplicate sequence removal All sequences were screened for duplicates using pairwise comparisons. Duplicate sequences from the same individual were removed. No duplicate sequences from different individuals were identified that could indicate laboratory contamination or sample mix-up.

Sequencing success rate Of the 1,226 samples initially screened, 1,214 (99.0%) yielded successful sequencing results. The 12 excluded samples (1.0%) were excluded due to insertions/deletions causing frameshifts that affected drug resistance interpretation.

Subtype determination The subtype was determined using the Stanford HIVdb subtyping tool (version 9.0). For sequences with ambiguous assignments, the REGA HIV-1 subtyping tool (version 3.0) and the HIV BLAST tool from the Los Alamos National Laboratory HIV Sequence Database were used for confirmation. Recombination analyses were performed for sequences with discordant subtype assignments using jpHMM and the REGA HIV-1 Subtyping Tool.

Construction of molecular transmission networks

All HIV-1 RNA pol region sequences generated in this study were used for molecular transmission network analysis. Sequences exceeding 1,000 bp in length and containing fewer than 5% mixed bases were retained for downstream analysis. Multiple sequence alignment was performed using MAFFT with the HXB2 reference sequence included as the alignment backbone. Molecular transmission networks were inferred using a genetic distance threshold of 1.5% under the TN93 model, as implemented in HIV-TRACE. This threshold was selected based on extensive validation demonstrating that a 1.5% genetic distance captures both recent and intermediate transmission linkages while minimizing false-positive connections from unrelated sequences [22, 23], and has been widely applied in Chinese HIV-1 molecular epidemiology studies [12, 24–26]. While subtype-specific thresholds may improve sensitivity, we used a conservative pan-subtype threshold of 1.5% to ensure comparability across the diverse HIV-1 subtype population in this study. Pairwise genetic distances were calculated using the TN93 model, as implemented in HIV-TRACE, and transmission networks were constructed based on these distances. Ambiguous nucleotides were coded as missing data, and pairwise distances with > 5% ambiguous nucleotides were excluded from network construction. Network metrics, including cluster size distribution, degree distribution, and the proportion of sequences in clusters, were calculated. Importantly, molecular transmission networks represent genetic linkage rather than confirmed epidemiological transmission; networks identify individuals with closely related viral sequences who may share transmission links but do not establish direct person-to-person transmission chains. Sensitivity analyses were performed using alternative genetic distance thresholds (1.0%, 2.0%, and 2.5%) to assess the robustness of network findings. Network visualization and layout were performed using Cytoscape (version 3.8.2).

Statistical analysis

Statistical analyses were performed using SPSS (version 22.0) and GraphPad Prism (version 10.0). Categorical variables were presented as frequencies and percentages, with comparisons between groups conducted using the chi-square test of independence. For all prevalence estimates (overall PDR, annual PDR, subtype-specific, and drug-class resistance), 95% confidence intervals (CIs) were calculated using the Wilson score method.

Multivariable logistic regression was performed to identify independent predictors of PDR. Variables included in the model were age (≤ 45 vs. > 45 years), sex, transmission route (heterosexual vs. other), baseline CD4 count (≤ 200 vs. > 200 cells/µL), baseline viral load (≤ 100,000 vs. > 100,000 copies/mL), and HIV-1 subtype. Results are presented as adjusted odds ratios (aOR) with 95% CIs.

Temporal trend analysis was performed using pairwise chi-square tests with Bonferroni correction for multiple comparisons to control the family-wise error rate.

A two-sided p-value < 0.05 was considered statistically significant.

Sample size consideration This study included all eligible participants diagnosed with HIV-1 infection between 2018 and 2023 (N = 1,214). Based on an expected PDR prevalence of approximately 6% (from preliminary data), the sample size provided sufficient precision to estimate the overall PDR prevalence with a 95% CI width of ± 1.3%. For comparisons of PDR prevalence between subtypes (e.g., CRF08_BC vs. CRF01_AE), the available sample size provided > 80% power to detect a twofold difference in prevalence (e.g., 11% vs. 5.5%) at a two-sided ⍺ = 0.05, assuming equal group sizes.

Results

Demographic and baseline clinical characteristics of the study population, by drug resistance status

Among the 1,226 individuals, a total of 1,214 ART-naïve individuals with newly diagnosed HIV-1 infection were included in the analysis. The overall PDR prevalence was 6.01% (73/1,214; 95% CI 4.80–7.49%). Participants ranged in age from 18 to 93 years, with a median age of 52.5 years. The cohort was predominantly male (68.80%), and the majority were aged ≥ 50 years (57.6%) and married (52.3%). Most participants were of Zhuang ethnicity (51.0%), engaged in agricultural occupations (61.0%), and had relatively low educational attainment, with 85.9% having completed junior high school or less. Heterosexual contact accounted for the vast majority of HIV transmission routes (93.94%) (Fig. 1).

Fig. 1.

Flowchart of participant selection and inclusion in the study. A total of 1,226 individuals newly diagnosed with HIV-1 infection between 2018 and 2023 were screened, of whom 1,214 ART-naïve individuals were included in the final analysis after quality control. 12 individuals were excluded due to sequence insertions or deletions (indels) causing frame shifts that affected drug resistance interpretation. All participants met eligibility criteria and were included in the analysis of pretreatment drug resistance, subtype distribution, and molecular transmission networks

When stratified by PDR status, no statistically significant differences were observed in sociodemographic characteristics or baseline clinical indicators. Specifically, variables including year of diagnosis, sex, age, transmission route, educational level, ethnicity, occupation, marital status, baseline CD4⁺ T-cell count, and baseline viral load were comparable between individuals with and without PDR (p > 0.05; Table 1).

Table 1.

Comparison of demographic and baseline clinical characteristics of ART-naïve individuals newly diagnosed with HIV-1 infection, stratified by PDR status (N = 1,214)

Characteristic	Category	Drug-resistant (N = 73)	Non-drug-resistant (N = 1,141)	p-value
Year				0.516
	2018	9 (6.57)	128 (93.43)
	2019	13 (6.81)	178 (93.19)
	2020	18 (4.96)	345 (95.04)
	2021	6 (5.41)	105 (94.59)
	2022	8 (4.32)	177 (95.68)
	2023	19 (8.37)	208 (91.63)
Sex
	Male (reference: female)	49 (5.87)	786 (94.13)	0.753
Age				0..820
	≤ 45	25 (6.23)	376 (93.77)
	> 45	48 (5.90)	765 (94.10)
Route of Infection				0.758
	Heterosexual	70 (6.06)	1085 (93.94)
	Others	3 (5.08)	56 (94.92)
Education level				0.735
	Primary school or below	37 (6.25)	555 (93.75)
	Junior high school or above	36 (5.79)	586 (94.21)
Ethnicity				0.908
	Han	29 (5.75)	475 (94.25)
	Zhuang	39 (6.30)	580 (93.70)
	Other/unknown	5 (5.49)	86 (94.51)
Occupation				0.913
	Farmer	45 (6.07)	696 (93.93)
	Other/unknown	28 (5.92)	445 (94.08)
Marital status				0.280
	Single	20 (7.70)	240 (92.30)
	Married	32 (5.03)	603 (94.97)
	Divorced/widowed/unknown	21 (6.58)	298 (93.42)
Baseline CD4 cell count
	≤ 200	51 (6.08)	788 (93.92)	0.886
	> 200	22 (5.87)	353 (94.13)
Baseline viral load				0.672
	≤ 100,000	24 (5.62)	403 (94.38)
	> 100,000	49 (6.23)	738 (93.77)
HIV-1 subtype
	CRF01_AE	44 (4.88)	858 (95.12)	0.012*
	CRF07_BC	12 (8.89)	123 (91.11)
	CRF08_BC	14 (11.11)	112 (88.89)
	CRF55_01B	2 (6.67)	28 (93.33)
	Other	1 (3.45)	28 (96.55)
Pre-ART symptoms
	Yes (reference: No)	37 (5.59)	625 (94.41)	0.496

Data are presented as counts and percentages. Comparisons were performed using the chi-square test. The percentages may not total 100% due to rounding. Baseline CD4⁺ T-cell count was categorized as ≤ 200 and > 200 cells/µL, and viral load (VL) as ≤ 100,000 and > 100,000 copies/mL. Transmission routes were grouped as heterosexual vs. other routes

*p = 0.012 for CRF08_BC vs CRF01_AE (pairwise comparison with Bonferroni correction). No other variables reached statistical significance

Multivariable logistic regression analysis was performed to identify independent predictors of PDR, including age, sex, transmission route, baseline CD4 count, baseline viral load, and HIV-1 subtype. The analysis revealed no significant independent predictors of PDR (Table 2). aOR with 95% Cis for all variables included in the model are presented in Table 2.

Table 2.

Multivariable logistic regression analysis of HIV-1 subtypes associated with PDR among ART-naïve individuals newly diagnosed with HIV-1 infection

Variable	Category	aOR	95% CI	p-value
HIV-1 subtype	CRF01_AE	1.00	–	–
	CRF07_BC	1.90	0.97–3.73	0.06
	CRF08_BC	2.45	1.29–4.67	0.006**
	CRF55_01B	1.40	0.32–6.06	0.66
	Other subtypes	3.73	0.53–14.06	0.23

The outcome variable was PDR status (presence vs. absence of low-, intermediate-, or high-level resistance mutations according to the Stanford HIV drug resistance database). CRF01_AE was used as the reference subtype, as it was the most prevalent. Subtype CRF08_BC was significantly associated with higher odds of PDR compared to CRF01_AE (aOR = 2.45; 95% CI 1.29–4.67; **p = 0.006). No other variables (age, sex, transmission route, baseline CD4 count, or baseline viral load) reached statistical significance and were therefore excluded from the final model

Distribution of HIV-1 subtypes and PDR by subtype

A diverse distribution of HIV-1 subtypes was observed. CRF01_AE was the dominant subtype, accounting for 74.30% of cases, followed by CRF07_BC (11.12%) and CRF08_BC (10.38%). Other subtypes, including subtype B, subtype C, and several rare circulating recombinant forms, were detected at low frequencies (Table 3).

Table 3.

Distribution of HIV-1 subtypes and associated PDR among ART-naïve individuals (N = 1,214)

Subtype	Count	Percentage (%)	PDR prevalence (%, 95% CI)	p-value	aOR (95% CI)
CRF01_AE	902	74.30	4.88% (3.64–6.50%)	Reference	1.00
CRF07_BC	135	11.12	8.89% (5.12–15.03%)	0.061	1.90 (0.97–3.73)
CRF08_BC	126	10.38	11.11% (6.65–17.98%)	0.012*	2.45 (1.29–4.67)
CRF55_01B	30	2.47	6.67% (1.82–21.31%)	0.662	1.40 (0.32–6.06)
Other subtypes†	21	1.73	4.76% (0.82–22.77%)	0.979	0.97 (0.12–7.49)

Data are presented as counts and percentages. PDR was defined by the presence of low-, intermediate-, or high-level resistance mutations according to the Stanford HIV Drug Resistance Database. 95% CIs for PDR prevalence were calculated using the Wilson score method. P-values from pairwise chi-square tests with Bonferroni correction comparing each subtype to CRF01_AE. aOR from multivariable logistic regression adjusted for age, sex, transmission route, baseline CD4 count, and baseline viral load. CRF08_BC was significantly associated with higher odds of PDR compared to CRF01_AE (aOR = 2.45; 95% CI 1.29–4.67; **p = 0.006)

^†Other subtypes include B (n = 8), C (n = 4), CRF59_01B (n = 3), CRF52_01B (n = 2), CRF53_01B (n = 2), A1 (n = 1), and CRF67_01B (n = 1)

PDR prevalence varied across HIV-1 subtypes. A single case of PDR was detected in subtype A1. However, this proportion should be interpreted with caution due to the extremely small sample size. Excluding subtype A1, PDR prevalence was highest among individuals infected with CRF08_BC (11.11%; 95% CI 6.65–17.98%), followed by CRF07_BC (8.89%; 95% CI 5.12–15.03%), CRF55_01B (6.67%, 95% CI 1.82–21.31%), and CRF01_AE (4.88%; 95% CI 3.64–6.50%). Pairwise comparisons with Bonferroni correction revealed that the higher PDR prevalence in CRF08_BC compared to CRF01_AE was statistically significant (p = 0.012). No other pairwise subtype comparisons reached statistical significance. These findings suggest a possible association between CRF08_BC and higher PDR prevalence, although the small sample sizes of some subtypes warrant cautious interpretation.

PDR prevalence and temporal trends

From 2018 to 2023, the prevalence of PDR showed modest year-to-year variation, ranging from 4.32% (95% CI 2.16–8.37%) in 2020 to 8.37% (95% CI 5.39–12.72%) in 2023 (Table 1). A Cochran-Armitage trend test confirmed that there was no significant linear trend in overall PDR prevalence from 2018 to 2023 (p = 0.38).

Temporal variation was observed across antiretroviral drug classes. NNRTI-associated resistance remained relatively stable, ranging from 2.16% (95% CI 1.06–4.35%) to 4.19% (95% CI 2.31–7.47%), with no significant trend (p for trend = 0.52). NRTI-associated resistance ranged from 0.73% (95% CI 0.25–2.10%) to 3.96% (95% CI 2.14–7.23%), with a notable increase in 2023 (Fig. 2). A Cochran-Armitage trend test for NRTI resistance from 2018 to 2023 showed a borderline significant trend (p = 0.04). Protease inhibitor (PI) resistance remained consistently low, ranging from 0 to 2.92%, with no significant trend (p = 0.61). These findings suggest that the modest increase in NRTI resistance in 2023 warrants continued monitoring.

Fig. 2.

Annual prevalence of drug-class resistance among ART-naïve individuals (2018–2023). NNRTI-associated resistance remained relatively stable (2.2–4.2%) throughout the study period. Notably, NRTI-associated resistance increased modestly in 2023 (3.96%; 95% CI 2.14–7.23%), while PI resistance remained consistently low (< 3% in all years). Error bars represent 95% CIs calculated using the Wilson score method. Sample sizes (n) for each year are shown below the x-axis

Distribution of HIV-1 drug resistance–associated mutations and antiretroviral resistance profiles

Drug resistance–associated mutations were identified across NRTIs, NNRTIs, and PIs (Fig. 3). Among NRTI-associated mutations, frequencies ranged from 0.08% to 0.25%, with M184MV being the most prevalent mutation (0.25%). NNRTI-associated mutations were more frequent, ranging from 0.08% to 0.99%, with K103N identified as the dominant NNRTI resistance mutation (0.99%). In contrast, PI-associated mutations were relatively rare, with frequencies between 0.08% and 0.16%; I84IV and M46L were the most commonly observed mutations.

Fig. 3.

Distribution of HIV-1 drug resistance-associated mutations and antiretroviral resistance profiles. A Prevalence of NNRTI-associated drug resistance mutation sites; B Prevalence of NRTI-associated drug resistance mutation sites; C Prevalence of PI-associated drug resistance mutation sites; D Drug resistance rates by antiretroviral drug. EFV efavirenz, ETR etravirine, NVP nevirapine, RPV rilpivirine, 3TC lamivudine, ABC abacavir, AZT zidovudine, d4T stavudine, ddI didanosine, FTC emtricitabine, TDF tenofovir, ATV/r atazanavir plus ritonavir, DRV/r darunavir plus ritonavir, FPV/r fosamprenavir plus ritonavir, IDV/r indinavir plus ritonavir, LPV/r lopinavir plus ritonavir, NFV nelfinavir, SQV/r saquinavir plus ritonavir, TPV/r tipranavir plus ritonavir

Substantial variability was observed in resistance across individual antiretroviral drugs. NVP showed the highest resistance prevalence (2.55%), followed by EFV (2.22%) and rilpivirine (RPV) (1.81%). Drug-level resistance reflects the combined effect of multiple resistance-associated mutations; therefore, resistance prevalence for individual drugs may exceed the frequency of any single mutation.

Molecular transmission network analysis

Using a TN93 genetic distance threshold of < 1.5%, a total of 102 transmission clusters (including clusters of size ≥ 2) were identified, comprising 612 of 1214 sequences (50.4%). The 102 clusters ranged in size from 2 to 19 individuals, with a mean cluster size of 6.0 ± 4.2 (median: 4.0), and 50 clusters (49.0%) contained three or more individuals. The mean degree within the network was 5.1 ± 7.4 (range: 1–43).

PDR was identified in two clusters, each containing at least two individuals with resistant strains. Of the 73 sequences with PDR mutations, 12 (16.4%) were found within clusters. One cluster included sequences carrying both K103N and V179E, while the other cluster comprised sequences harboring K103N only. Individuals within these clusters containing PDR were predominantly connected through heterosexual transmission (Fig. 4). These clusters provide suggestive evidence of localized transmission of resistant strains, although the small number limits broader inference.

Fig. 4.

Molecular transmission network of HIV-1 and distribution of pretreatment PDR mutation sites. Nodes represent individual sequences, and edges indicate genetic linkage inferred using a TN93 genetic distance threshold of < 1.5%. Red nodes indicate sequences harboring PDR mutations, while gray nodes represent sequences without PDR. Clusters represent groups of genetically linked individuals within the transmission networks. A PDR transmission cluster was defined as a cluster in which sequences shared identical drug resistance-associated mutations

Sensitivity analyses using alternative genetic distance thresholds (1.0%, 2.0%, and 2.5%) demonstrated that the identification of PDR-containing clusters was robust across thresholds, although the total number of clusters and proportion of sequences in clusters varied (Table 4). At the 1.0% threshold, 47 clusters comprising 216 sequences (17.8%) were identified; at 2.0%, 156 clusters comprising 942 sequences (77.6%) were identified; and at 2.5%, 183 clusters comprising 1058 sequences (87.2%) were identified.

Table 4.

Sensitivity analysis of molecular transmission network characteristics across alternative genetic distance thresholds

Threshold (%)	Clusters (≥ 2)	Sequences in clusters	Proportion of total sequences (%)	PDR-containing clusters
1.0	47	216	17.8	2
1.5	102	612	50.4	2
2.0	156	942	77.6	3
2.5	183	1058	87.2	3

PDR-containing clusters were defined as clusters with ≥ 2 sequences harboring pretreatment drug resistance mutations (low-, intermediate-, or high-level resistance according to the Stanford HIV Drug Resistance Database). The 1.5% threshold was selected as the primary threshold based on published validation studies and its balance of sensitivity and specificity [22, 23]. Alternative thresholds (1.0%, 2.0%, and 2.5%) were evaluated to assess the robustness of network findings. All networks were constructed using the TN93 genetic distance model as implemented in HIV-TRACE

Discussion

In this study, we analyzed HIV-1 pol gene sequences from 1,214 ART-naïve individuals newly diagnosed with HIV-1 infection at the Guangxi Zhuang Autonomous Region Chest Hospital between 2018 and 2023. Using a molecular epidemiological approach, we provide a comprehensive characterization of PDR patterns and their transmission dynamics in the central Guangxi region.

The study population was predominantly composed of middle-aged and older men, married individuals, persons of Zhuang ethnicity, and those engaged in agricultural occupations, with generally low educational attainment. The Zhuang ethnic group is the largest minority population in Guangxi and predominantly resides in rural areas with limited healthcare infrastructure. Agricultural occupations, which constituted 61.0% of our cohort, are associated with lower socioeconomic status, limited health literacy, and reduced access to healthcare services compared to urban populations. These factors may contribute to delayed HIV diagnosis, as reflected in the high proportion of participants with baseline CD4 ≤ 200 cells/µL (69.1%), and may influence engagement in HIV testing and treatment programs. However, the predominance of Zhuang ethnicity and farming occupations in this cohort is consistent with the demographic profile of central Guangxi and may reflect regional population characteristics rather than healthcare access disparities per se.

Heterosexual contact accounted for the vast majority of infections, suggesting that HIV transmission in this setting remains largely driven by general population dynamics rather than concentrated within traditional high-risk groups. PDR prevalence ranged from 4.32 to 8.37%, remaining within the moderate range defined by the World Health Organization (WHO) and generally comparable to, or slightly higher than, the national estimate of 4.4% reported for China [27, 28]. Although year-to-year variation was observed, no consistent temporal trend was identified. The increase in 2023 may reflect the indirect effects of COVID-19-related disruptions, including changes in risky behavior and variability in testing coverage [29]. This suggests that broader public health events may influence HIV drug resistance dynamics.

From a therapeutic perspective, PDR in this region was predominantly driven by NNRTI resistance, which consistently accounted for the largest proportion of resistance across all study years. This pattern likely reflects both the historical reliance on NNRTI-based first-line regimens and the onward transmission of resistant strains. In contrast, resistance to NRTIs remained comparatively low, although a modest increase was observed after 2021, reaching 3.96% in 2023. This increase was mainly associated with mutations such as M184V and T215A/S mutations, which may compromise the effectiveness of NRTI backbones in first-line regimens [30].

Resistance associated with PI remained rare throughout the study period, consistent with the high genetic barrier of this drug class. At the drug level, resistance to EFV and NVP remained comparatively higher (2.22% and 2.55%, respectively), whereas resistance to second-generation NNRTIs such as etravirine and rilpivirine remained below 2% [31]. These findings support cautious use of EFV- and NVP-based regimens in settings without routine baseline resistance testing. In contrast, the very low prevalence of resistance to tenofovir (0.08%) supports its continued role as a key component of first-line therapy [32, 33].

A diverse distribution of HIV-1 subtypes was observed, with CRF01_AE predominant (74.30%), alongside CRF07_BC (11.12%) and CRF08_BC (10.38%), consistent with previous reports from Guangxi [8]. Notably, higher PDR prevalence was observed in CRF08_BC and CRF07_BC (11.11% and 8.89%, respectively). Pairwise comparisons with Bonferroni correction showed that the higher PDR prevalence in CRF08_BC compared to CRF01_AE was statistically significant (p = 0.012), and this association was confirmed in multivariable logistic regression (aOR = 2.45; 95% CI 1.29–4.67; p = 0.006). These findings suggest that CRF08_BC may be associated with a higher likelihood of PDR, although the relatively small sample sizes of non-CRF01_AE subtypes warrant cautious interpretation. The coexistence of multiple subtypes reflects the complexity of local transmission dynamics, likely influenced by sustained heterosexual transmission and regional population mobility. The identification of CRF52_01B cases, a subtype previously reported mainly in Southeast Asia [34, 35], suggests possible cross-border introduction and highlights the importance of continued molecular surveillance in this region. Given the substantial population mobility between Guangxi and Southeast Asia, the emergence of rare recombinant forms underscores the importance of strengthened cross-regional molecular epidemiological surveillance to evaluate their potential for further local spread.

It is important to note that molecular transmission networks represent genetic linkage rather than confirmed epidemiological transmission. While the identification of individuals with closely related viral sequences suggests shared transmission linkages, these networks do not establish direct-to-person transmission chains. The clustering patterns observed in this study are therefore indicative of potential transmission relationships that warrant further epidemiological investigation.

Molecular transmission network analysis based on a TN93 genetic distance threshold of < 1.5% identified 102 transmission clusters, with half comprising three or more individuals, indicating the presence of localized transmission networks. Two clusters contained individuals with shared PDR mutations, specifically K103N with or without V179E. Although limited in number, these clusters provide suggestive evidence that resistant strains can be transmitted within local networks, particularly in a predominantly heterosexual epidemic. These findings suggest the possibility that NNRTI resistance mutations, especially K103N, may be capable of localized transmission, although the limited number of resistant clusters identified in this study precludes firm conclusions about sustained transmission. Their presence represents a potential clinical and public health concern in settings where NNRTI-based first-line regimens remain in use. Although limited by the small number of resistant clusters, these observations raise the possibility that NNRTI-associated resistance mutations may be transmitted within local networks, underscoring the need for continued molecular surveillance to monitor resistance transmission dynamics.

Our findings, therefore, indicate that PDR in central Guangxi remains at a moderate level, is predominantly driven by NNRTI-associated mutations, and shows evidence of clustering within specific viral subtypes and transmission networks. These results highlight the importance of continued resistance surveillance, the use of molecular network analysis to identify high-risk clusters, and the optimization of initial ART strategies. These approaches provide a robust molecular epidemiological foundation to inform more targeted regional HIV prevention, treatment, and drug resistance control efforts.

From a clinical perspective, the persistence of moderate PDR levels, particularly for NNRTIs, supports consideration of baseline resistance testing where feasible. In settings without such testing, the observed resistance patterns provide further support for the transition from NNRTI-based regimens to those with higher genetic barriers, such as a dolutegravir-based regimen, in line with WHO guidelines. Additionally, the identification of clustered resistance suggests that targeted interventions, including partner notification and focused prevention strategies, may help interrupt transmission chains.

The findings of this study have important implications for ART regimen selection in the context of China’s evolving treatment guidelines. During 2018–2023, first-line regimens in Guangxi were predominantly NNRTI-based (EFV/NVP plus two NRTIs) [36]. The moderate NNRTI resistance observed in this cohort (2.16–4.19% annually) provides empirical support for the transition from EFV/NVP-based regimens toward DTG-based regimens, which have a higher genetic barrier to resistance and are now recommended as preferred first-line therapy in China [19, 36]. The very low resistance to tenofovir (TDF, 0.08%) and other NRTIs observed in this study supports the continued use of tenofovir-containing backbones in first-line regimens [37]. These findings are reassuring given that TDF remains a cornerstone of both NNRTI-based and INSTI-based first-line combinations. However, the absence of integrase resistance testing in this study represents an important limitation, as transmitted DTG resistance, while currently rare globally, has been reported and could impact treatment outcomes if integrase inhibitor-based regimens are introduced without baseline resistance screening. Given the moderate PDR levels observed, particularly for NNRTIs, baseline resistance testing, including integrase genotyping where feasible, should be considered to guide individualized ART selection. In settings without such testing capacity, the observed resistance patterns provide further support for the use of DTG-based regimens with high genetic barriers, in line with WHO recommendations. Furthermore, continued molecular surveillance should be expanded to include integrase resistance monitoring to detect emerging INSTI resistance and inform future guideline updates.

Several factors should be considered when interpreting the representativeness of this cohort. The Guangxi Zhuang Autonomous Region Chest Hospital serves as a major referral center for HIV/AIDS in central Guangxi, and the study population may therefore overrepresent individuals with advanced disease or complications requiring specialized care. This is reflected in the high proportion of participants with baseline CD4 ≤ 200 cells/µL (69.1%) and WHO clinical stage III/IV (32.5%), which may not fully represent the broader population of people living with HIV in Guangxi who are diagnosed through routine screening programs. The median age of the cohort was 52.5 years, which is higher than the national average for newly diagnosed HIV infections in China [38, 39], and the predominance of heterosexual transmission (93.94%) and agricultural occupations (61.0%) reflects the demographic characteristics of central Guangxi. However, the hospital’s catchment area primarily covers central Guangxi. Patients from remote or rural areas with limited healthcare access may be underrepresented, potentially introducing geographic selection bias. The high proportion of participants with low educational attainment (85.9% with ≤ junior high school) and agricultural occupations may reflect regional socioeconomic demographics but could also indicate barriers to healthcare access and engagement. Patients who present for diagnosis at this referral hospital may have different healthcare-seeking behaviors, testing histories, and engagement patterns compared to those diagnosed through community-based or routine screening programs. Consequently, while our findings provide valuable insights into PDR patterns in this referral population, they may not be fully generalizable to all regions of Guangxi or to the broader HIV epidemic in China. Multicenter studies incorporating diverse populations and healthcare settings are needed to validate these observations and provide a more comprehensive understanding of PDR epidemiology in the region.

This study, however, has several limitations that merit consideration when interpreting these findings. First, this was a single-center study conducted at a major referral hospital in central Guangxi. Despite the larger sample size, the patient population may overrepresent individuals with advanced disease or complications requiring specialized care, as reflected in the high proportion of participants with baseline CD4 ≤ 200 cells/µL (69.1%) and WHO clinical stage III/IV (32.5%). Additionally, the high median age (52.5 years) may not fully reflect the broader HIV epidemic in Guangxi or China. Therefore, the findings may not be fully generalizable to other regions of Guangxi, to individuals diagnosed through routine screening programs, or to other populations in China. Multicenter studies incorporating diverse populations and healthcare settings are needed to validate these observations. Second, the analysis was limited to the pol gene, which may underestimate recombinant diversity and overlook resistance mutations in other genomic regions, including integrase. Integrase resistance testing was not available in this cohort, and transmitted INSTI resistance, while currently rare, warrants continued monitoring as DTG-based regimens are scaled up in China. Third, the use of a < 1.5% genetic distance threshold, while standard, may underestimate network connectivity. Fourth, the lack of detailed behavioral data limits the ability to distinguish between transmitted and acquired resistance. Fifth, the cross-sectional design precludes assessment of long-term clinical outcomes. Sixth, the study period (2018–2023) overlapped with a transitional period in China’s ART guidelines. Therefore, the resistance patterns observed primarily reflect the NNRTI-based treatment era and may not fully represent the resistance landscape under current DTG-based regimens. Seventh, the hospital’s catchment area primarily covers central Guangxi, and patients from remote or rural areas with limited healthcare access may be underrepresented, potentially introducing geographic selection bias. Finally, the disruption during the COVID-19 pandemic may have influenced testing patterns and data completeness (2020–2022). This may have introduced ascertainment bias through reduced testing and healthcare access, potentially affecting the representativeness of these years’ data. Future studies incorporating multicenter data and prospective follow-up are needed to better characterize the clinical impact and transmission dynamics of PDR.

Despite these limitations, our study provides important baseline data on PDR patterns in central Guangxi during a critical transitional period in ART policy. The findings highlight the need for continued surveillance and the importance of considering local epidemiological contexts when interpreting PDR data. Future research should incorporate multicenter collaborations, standardized sampling strategies, and longitudinal follow-up to better characterize PDR transmission dynamics and their clinical impact across diverse populations in China.

Conclusion

This study provides a comprehensive characterization of the epidemiological and molecular landscape of pretreatment HIV-1 drug resistance in the central Guangxi region during a period when NNRTI-based regimes predominated. PDR remained at a moderate level and was predominantly driven by NNRTI-associated mutations, with limited evidence of clustering within transmission networks. However, the small number of resistant clusters identified highlights the need for continued surveillance to confirm these patterns. The moderate NNRTI resistance observed supports the transition to integrase inhibitor-based regimens, particularly DTG, which offer a higher genetic barrier to resistance. The findings underscore the need for strengthened resistance surveillance, including integrase resistance monitoring, optimization of first-line antiretroviral regimens, and the integration of molecular transmission network analyses to support targeted prevention and control strategies. However, given the single-center referral hospital setting, these findings should be interpreted with caution regarding their generalizability to the broader HIV population in Guangxi or China, and multicenter validation studies are warranted.

Abbreviations

ABC

Abacavir

ART

Antiretroviral therapy

ATV/r

Atazanavir boosted with ritonavir

AZT

Zidovudine

CD4 + T cells

Cluster of differentiation 4 positive T lymphocytes

CRF

Circulating recombinant form

D4T

Stavudine

ddI

Didanosine

DRV/r

Darunavir boosted with ritonavir

EFV

Efavirenz

ETR

Etravirine

FTC

Emtricitabine

FPV/r

Fosamprenavir boosted with ritonavir

HIV-1

Human immunodeficiency virus type 1

HIVdb

Stanford HIV drug resistance database

INSTI

Integrase strand transfer inhibitor

IDV/r

Indinavir boosted with ritonavir

LPV/r

Lopinavir boosted with ritonavir

MAFFT

Multiple alignment using fast Fourier transform

NFV

Nelfinavir

NRTIs

Nucleoside reverse transcriptase inhibitors

NNRTIs

Non-nucleoside reverse transcriptase inhibitors

NVP

Nevirapine

PCR

Polymerase chain reaction

PDR

Pretreatment drug resistance

PIs

Protease inhibitors

RPV

Rilpivirine

SPSS

Statistical package for the social sciences

SQV/r

Saquinavir boosted with ritonavir

TDF

Tenofovir Disoproxil fumarate

TPV/r

Tipranavir boosted with ritonavir

TN93

Tamura-Nei 93 genetic distance model

VL

Viral load

WHO

World Health Organization

Author contributions

GC and MH contributed equally to the conceptualization, methodology, data curation, formal analysis, and writing of the original draft. YP, LH, YD, WS, and YT were involved in the investigation and data collection. JC and AL contributed equally to conceptualization, supervision, project administration, and writing—review & editing. All authors read and approved the final manuscript.

Funding

This study was funded by Major Science and Technology Projects in Guangxi (AB25069013) and the Liuzhou Science and Technology Plan Project (2020NBAD0802). The funder had no role in study design, data collection, data analysis, interpretation of results, or manuscript writing.

Data availability

The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.

Declarations

Ethics approval and consent to participate

The study protocol was reviewed and approved by the Medical Ethics Committee of the Guangxi Zhuang Autonomous Region Chest Hospital [Approval No.: 2024-S027-01]. All procedures were conducted in accordance with the Declaration of Helsinki and its subsequent amendments. Participant confidentiality was maintained throughout the study, and all personal data were anonymized before analysis.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.