Drug Resistance Among Newly Diagnosed People Living with HIV-1 in Sichuan, China: A Large-Scale Population-Based Cross-Sectional Study

Introduction

Acquired immune deficiency syndrome (AIDS), as a type of infectious disease, has an important and far-reaching influence on Chinese society. To curb deaths and new infections, China actively responds to the “95–95-95” AIDS prevention and control goal. Since 2016, China’s “universal treatment” program has enabled all people living with HIV (PLWH) to receive timely treatment upon request. Early treatment not only extends lifespan of PLWH but also prevents secondary HIV transmission. Detecting and treating it promptly is the most effective AIDS prevention and treatment approach. However, with all PLWH on antiretroviral therapy (ART), the spread of HIV-resistant strains poses a major challenge in AIDS prevention and treatment.^1–4 From the 2003 World Health Organization (WHO) guidelines advocating the use of nucleoside reverse transcriptase inhibitors (NRTIs) and non-nucleoside reverse transcriptase inhibitors (NNRTIs) to the present day, when more efficacious and better-tolerated integrase strand transfer inhibitors (INSTIs) are in use, it is clear that the evolution of ART drug utilization within the population has a direct bearing on the prevalence of drug resistance.⁵ Transmitted drug resistance (TDR), detected in newly diagnosed PLWH without prior ART,⁶ can affect ART treatment, increase mortality, and cause secondary HIV transmission.⁷ Thus, assessing TDR levels is crucial for informing ART policies and ART services.

Sichuan, situated in Southwest China (Figure 1A), comprises 21 prefecture-level cities (Figure 1B) and 183 counties (Figure 1C). Geographically, it is divided into the Chengdu Plain, Northeast Sichuan, South Sichuan, Panxi and West Sichuan (Figure 1B). With a permanent population of 83.675 million in the seventh census, it serves as Southwest China’s economic hub. In terms of the AIDS epidemic, Sichuan had 220,000 cumulative and 17,000 newly diagnosed PLWH in 2023, indicating a severe epidemic situation that necessitates continuous TDR monitoring. In 2022, the Chinese Center for Disease Control and Prevention(CDC) conducted a drug-resistance survey on newly diagnosed PLWH in six provinces including Sichuan, revealing a 7.4%⁸ resistance rate, higher than sub-Saharan Africa (6.0%) but lower than Greece (7.8%), Hungary (10.7%), Cameroon (10.4%), Namibia (12.7%), Israel (12.1%), Mozambique (14.4%), Mexico (14.8%), the US (22.5%), and Japan (12.5%).^9–18Integrating 2014 and 2019 surveillance results in Sichuan, a significant upward trend in TDR was found, reaching a medium epidemic level in 2019.^19,20 Compared to the 2022 cluster-sampling drug-resistance survey conducted by the Chinese CDC, this study provided excellent population coverage and covered high-epidemic areas, offering a more representative view of HIV drug-resistance prevalence in Sichuan. To establish a scientific and reliable ART program for PLWH, the Sichuan CDC conducted HIV resistance surveillance in all newly diagnosed PLWH from July 2022 to June 2023. This aimed to determine the prevalence, influencing factors, geographical differences and transmission patterns of HIV-resistant strains. These research findings provide a solid and evidence-based scientific basis for AIDS prevention and control initiatives in Sichuan.

Figure 1 (A–C) The geographical distribution of Sichuan in China.

Materials and Methods

Research Subjects

A cross-sectional survey was conducted in Sichuan to evaluate TDR. The study enrolled individuals newly diagnosed PLWH from July 2022 to June 2023. The inclusion criteria included individuals did not accept prior ARV exposure.All respondents needed to have a unique card ID and could provide written oral informed consent, while PLWH<18 years old required written or oral informed consent from their guardians. The Ethics Committee of the Sichuan Provincial Center for Disease Control and Prevention approved both the informed consent process and the research. The study followed the Declaration of Helsinki.

Sample Collection and Preservation

Plasma were collected from HIV antibody screening-positive patients. For each newly diagnosed PLWH, 5 mL of venous whole blood was collected using an 8-mL vacuum blood collection tube containing EDTA-K2. Two 1-mL plasma samples and one 1-mL lymphocyte-enriched sample were separated within 6 hours. One plasma sample was stored at 4°C for HIV supplementary tests. The other plasma sample and the lymphocyte-enriched sample from research subjects meeting the inclusion criteria were stored at −70°C for HIV genotype drug resistance testing. The investigation and sampling were carried out in a standardized manner by municipal and county-level medical institutions or centers for disease control and prevention.

Laboratory Detection and Drug Resistance Analysis

HIV complementation tests were completed by an HIV confirmation laboratory approved by the Sichuan Provincial Health Commission or a registered nucleic acid laboratory. The CD4+ T lymphocyte test was completed in a laboratory that passed the provincial or national assessment of quality control. The Center of HIV Confirmation Laboratory of Sichuan Province was responsible for HIV nucleic acid extraction, PCR amplification and HIV-1 pol gene sequencing of PLWH plasma samples. Nested PCR was used to amplify the PR-RT region of the HIV-1pol gene. The process of resistance testing, PCR amplification, primer sequence, and data analysis referred to the HIV Resistance Monitoring Strategies and Detection Technologies⁶ compiled by Shao Yiming and the Guide for Resistance Testing and Quality Assurance of HIV-1 genotyping²¹ (2013 edition) compiled by the Chinese CDC.

Sequence and Genotyping

The Sequencer 5.1 was used to edit, clean and splice the sequences, and then these sequences were corrected with the BioEdit 7.0 and compared with the reference sequences in the HIV database.²² The neighbor-joining method of Mega7.0 was used to construct the phylogenetic tree with the Kimura2-parameter model, and the phylogenetic tree was constructed 1000 times. The accuracy of the genotyping was verified by a bootstrapvalue >70%. Finally, genotyping was identified in combination with the HIV BLAST tool in the HIV Database.²² All possible sample cross-contamination, laboratory contamination, and sequence quality control methods were validated by constructing adjacent phylogenetic trees and using QC tools in the HIV Database.²²

Resistance Analysis

The results were interpreted by using the HIV dbdatabase (https://hivdb.stanford.edu) developed by Stanford University in the United States for the resistance analysis among PLWH.²³ Twenty antiviral drugs, including NNRTIs such asdoravirine(DOR), efavirenz(EFV), etravirine(ETR), nevirapine(NVP), and rilpivirine(RPV), were used in the HIV dbdatabase; NRTIs, such asabacavir(ABC), zidovudine (AZT), emtricitabine(FTC), lamivudine(3TC), tenofovir(TDF), stavudine(D4T), and dehydroxyinosine(DDI); and PIs, such asazanavir(ATV), darunavir(DRV), lopinavir(LPV), fosanavir(FPV), indinavir(IDV), nelfinavir(NFV), sadequinavir(SQV), and tipranavir (TPV), were used. If resistance to one or more drugs reaches a low-level or higher degree, it is classified as drug resistance. Based on this criterion, the overall and categorized (NNRTI, NRTI and PI) as well as various types of drug resistance were calculated. According to the WHO’s definitions of drug resistance prevalence levels, a resistance rate of <5% indicates low prevalence, 5–15% indicates moderate prevalence, and >15% indicates high prevalence.

Data Collection

The clinical and demographic data was obtained and exported using the unique card number of each PLWH from the AIDS Prevention and Control Information System of the Chinese CDC. Among the analytical variables, the regional viral suppression failure rates over the past three years were calculated as the proportion of treated PLWH with an HIV-1 viral load(VL) >1000 copies/mL, based on test results collected by all nucleic acid testing labs under the Sichuan CDC.

Statistical Analysis

SPSS 19.0 was used for the statistical analysis. Univariate and multivariate logistic regression were used to screen for factors correlated with resistance. Variables with P<0.05 in the univariate analysis were further entered into a multivariate stepwise logistic regression model, and variables significant at P<0.05 were retained in the final model. The detection rate of mutation sites among each HIV subtype was tested by the chi-square test (test level α=0.05).

Construction of the Molecular Network of HIV-1

Gene distances between genes were calculated via HIV-TRACE under the model of Tamura Nei 93. According to the Technical Guidelines for Monitoring and Intervention of the HIV Transmission Network (2021PilotEdition),²⁴ a 1% gene distance was selected as the cutoff for the construction of the molecular network. The transmission network was subsequently visualized via Cytoscape 3.6.1,²⁵ and several molecular clusters with potential connections were identified. The network characteristics include nodes (individuals in the network), edges (links between two nodes that represent potential transmission relationships between two individuals), degrees (the number of edges connecting one node to other nodes), network size (the number of individuals in a cluster), and clusters (groups of link sequences). The gene sequences with drug resistance mutation sites were screened to construct a molecular network that was composed of only resistance gene sequences.

Spatial Analysis

The prevalence of total TDR, NNRTI, NRTI, and PI in183 counties in Sichuan was visualized via ArcGIS10.3.²⁶ Spatial autocorrelation analysis was used to determine the spatial distribution by combining the Z value, P value and Moran’s I index. When Moran’sI is positive, the spatial data are positively correlated, indicating a clustering trend. Conversely, when the Moran I index is negative, there is a negative correlation in the spatial data, suggesting a discrete trend. When the global Moran’sI index is 0, the correlation of spatial data is not significant, indicating that the spatial data are randomly distributed. Local Moran’s I and Getis‒OrdGi* statistics were used for local autocorrelation analysis, and local spatial8uassociation index (LISA) clusters and cold and hot spot maps were drawn. Five patterns were included in the LISA cluster map. “High-high” means that areas with high TDR prevalence are surrounded by areas with high TDR prevalence; “low-low” means that areas with low TDR prevalence are surrounded by areas with low TDR prevalence; “low-high” means that areas with low TDR prevalence are surrounded by areas with high TDR prevalence; “high-low” means that areas with high TDR prevalence are surrounded by areas with relatively low TDR prevalence; and a value of zero means that TDR prevalence is randomly distributed across regions,namely, there is no spatial clustering pattern. Individual links in the molecular network clusters were used to reflect whether the TDR of HIV was more prevalent within or between the regions. To show more granular transmission of TDR between cities, each transmission network was visualized, and the links between regions are represented by lines of different colors; the darker the color is, the thicker the lines are, indicating that the relationship between the regions of TDR transmission is stronger. Links within a region are indicated by blue circles, and the larger the circle is, the stronger the relationship between the TDR of HIV transmission within regions. The transport diagram was created in QGIS (version 3.10).²⁷

Results

Demographic and Epidemiological Characteristics of the Subjects

The survey covered 13,619 qualified sequences. Of these, 9,963 (73.16%) were male, 6,781 (49.79%) were aged 50--69, and 8,394 (61.63%) were farmers. In terms of education, 6,366 (46.74%) had a primary school education. Marital status revealed that 6,889 (50.58%) were married or had spouses. With respect to transmission routes, 12,385 (90.94%) were heterosexual transmission cases. Additionally, 8,996 (66.05%) were HIV-infected patients, whereas 10,057 (73.85%) had no history of sexually transmitted diseases. Before treatment, 47.10% (6,415) of the patients had CD4+ T lymphocyte counts ranging from 200 to 499/mm³. The failure rate of HIV ART-induced viral suppression in the areas where 8673 (63.68%) PLWH reside ranged from 5% to 14% in the past three years. A total of 21 genotypes were detected in this study, including 7,986 cases (58.64%) of CRF07_BC, 2,955 cases (21.70%) of CRF01_AE, 1,259 cases (9.24%) of CRF08_BC, 602 cases (4.42%) of CRF85_BC, 189 cases (1.39%) of the B subtype, 165 cases (1.21%) of CRF55_01B, 57 cases (0.42%) of CRF105_108, 109 cases (0.80%) of other subtypes, and 297 cases (2.18%) of the unclassified recombinant form (URF) (Table 1).

Table 1 The Demographic and Clinical Characteristics of Newly Diagnosed HIV-Infected Individuals in Sichuan From July 2022 to June 2023

Analysis of TDR and Mutation

Among 13,619 individuals, 1,486 were found to be drug resistant, with a total TDR prevalence of 10.91%. A total of 59 sites with drug resistance mutations were identified. In terms of ARV, the prevalence of TDR among NNRTIs was 8.08%, whereas the prevalence of TDR, specifically for NVP, EFV, RPV, DOR, and ETR, was 6.50%, 6.26%, 3.69%, 1.82%, and 1.42%, respectively. The primary drug resistance mutations were V179D/E/X/Q/N/A/I/T/L/K (8.38%) and K103N/S/R/Q (4.32%). For NRTI, the TDR incidence was 1.34%, and the prevalence of ABC, D4T, FTC, 3TC, DDI, and AZT were 0.84%, 0.81%, 0.67%, 0.67%, 0.50%, 0.47%, and 0.35%, respectively. The primary drug resistance mutations were S68G/N (3.73%) and M184V/I (0.58%). For PI, the TDR prevalence rate was 2.22%, and the prevalence rates of TPV, NFV, FPV, IDV, LPV, ATV, SQV, and DRV were 1.49%, 0.81%, 0.31%, 0.14%, 0.12%, 0.09%, 0.07%, and 0.04%, respectively. The major drug resistance mutation was M46I/L (0.37%), and the accessory mutation was Q58E (1.42%). Regionally, the total TDR prevalence rates in all regions exceeded 5%, except for BZ and AB. Among them, the TDR prevalence rates in MY, SN, TA, and ZY of the Chengdu Plain; ZG, LZ, and YB of southern Sichuan; GY and GA of northeastern Sichuan; and PZH of the Panxi area exceeded 10%. DY in the Chengdu Plain and LSZ in the Panxi area exceeded 15%. For NNRTI, the TDR prevalence rate in all regions exceeded 5%, except AB and BZ. The TDR prevalence rates in LZ (southern Sichuan), SN (Chengdu Plain), and LSZ (Panxi Area) exceeded 10%. For EFV, the TDR incidence rates in CD, DY, MY, SN, YA, ZY, and MS of the Chengdu Plain; ZG, LZ, NJ, and YB of southern Sichuan; and PZH of the Panxi area exceeded 5%. The LSZ (Panxi area) exceeded 10%. The TDR incidence of NVP in various regions was similar to that of EFV. For RPV, the TDR incidence rates in DY, SN, and YA of the Chengdu Plain, LZ (southern Sichuan), and GA (northeastern Sichuan) exceeded 5%. Among the common drug-resistant mutation sites of various virus subtypes, except for the detection rates of T215, D67, and K65 in NRTI and G190, A98, and Y188 in NNRTI, which showed no statistically significant differences (P≥0.05), other drug-resistant mutation sites with higher detection rates were all statistically significant.(Tables 2–4).

Table 2 Antiviral Medications and Drug Resistance Mutation Sites of Newly Diagnosed PLWH in Sichuan Province From July 2022 to June 2023

Table 3 Drug Resistance Rate of Newly Diagnosed PLWH in Sichuan Province From July 2022 to June 2023

Table 4 Detection Rate of Drug-Resistant Mutation Sites of HIV Subtype in Sichuan Province From July 2022 to June 2023

Factors Influencing TDR

The results of the multi-factor logistic regression revealed that PLWH under 15 years of age had a higher TDR incidence than those aged 15--29 years did (OR=1.64, 95% CI=1.01~2.66). Additionally, compared with PLWH infected with CRF01_AE, individuals infected with CRF08_BC (OR=1.65, 95% CI=1.36~2.01), CRF55_01B (OR=2.06, 95% CI=1.37~3.12), the B subtype (OR=2.23, 95% CI=1.53~3.26), or other subtype viruses (OR=1.90, 95% CI=1.14~3.16) had a greater risk of TDR. Moreover, compared with regions with antiretroviral therapy viral suppression failure rates less than 5% in the past three years, areas with failure rates ranging from 5% to 14% and 15% or higher had a greater prevalence of TDR (Table 5).

Table 5 TDR Factors of Newly Diagnosed PLWH in Sichuan Province From July 2022 to June 2023

Spatial Autocorrelation Analysis and Getis‒Ord Gi Analysis of TDR

There were 183 county-level administrative districts in 21 municipal administrative regions in Sichuan. The overall, NNRTI, NRTI, and PI drug resistance rates in each county are presented in Figures 2A–D. In the spatial autocorrelation analysis of the TDR incidence rate, the overall and PI results were significant, indicating that their spatial autocorrelation was clustered (Figure 3A and B). The NRTI results were not significant and showed a random distribution without spatial clustering (Figure 3C). In addition, the NNRTI results suggested that there was a 90% possibility of a clustered distribution (Figure 3D).

Figure 2 (A–D) The TDR incidence of newly diagnosed PLWH in Sichuan Province from July 2022 to June 2023.

Figure 3 (A–D) Spatial autocorrelation analysis for the TDR incidence of newly diagnosed PLWH in Sichuan Province from July 2022 to June 2023.

Local Moran’s I analysis revealed that the total TDR prevalence in five regions (two in the Chengdu Plain, one in the Panxi area, one in southern Sichuan, and one in northeastern Sichuan) was high-high (H-H), whereas three areas (two in western Sichuan and one in northeastern Sichuan) were low-low (L-L) (Figure 4A). For the TDR incidence of PI, three H-H clusters (one in the Panxi area, one in the adjacent areas to the Chengdu Plain and Northeast Sichuan, and one in Northeast Sichuan) and one L-L cluster (Panxi area) were detected (Figure 4B). The NRTI results revealed four H-H cluster areas (two in the Chengdu Plain, one in southern Sichuan, and one in the Panxi area) and one L-L cluster area (northeastern Sichuan) (Figure 4C). Apart from the Chengdu Plain and northeastern Sichuan, which had no H-H cluster areas of NNRTI, other areas had similar results to the total TDR rate (Figure 4D).

Figure 4 (A‒D) Cluster by local Moran’s I analysis for the TDR incidence of newly diagnosed PLWH in Sichuan Province from July 2022--June 2023.

The local Getis Ord Gi* index depends on the differences between the observations and expectations of spatial data.spots and cold spots with different confidence intervals (CIs) were identified and marked. The hot spots indicate significant clusters of TDR in the region, whereas the cold spots indicate the opposite. Specifically, the hot spots of total TDR were mainly concentrated in the west, southwest, south, and northeast, whereas the cold spots were primarily concentrated in the north and east. The hot spot areas of PI were mainly concentrated in the southwest and northeast regions, and NRTIs were mainly concentrated in the central and southwest regions, both of which had no cold spots. The hot spots of NNRTI were broadly similar to the total TDR, except for the central and northeast regions (Figure 5A–D).

Figure 5 (A–D) Getis Ord Gi* analysis for the TDR incidence of newly diagnosed PLWH in Sichuan Province from July 2022 to June 2023.

HIV-1 Molecular Network of TDR and Transmission Characteristics

The HIV-1 molecular network contained 13,619 sequences. At a genetic distance threshold of 0.01, 7696 sequences (56.5%) were incorporated into the network, whereas among the 1486 drug-resistant PLWH, 784 (52.76%) were included (Figure 6A). There was no statistically significant difference in inclusion rates between drug-resistant patients and overall patients. Subsequently, among the 784 drug-resistant PLWH, the molecular network was constructed again via a rigorous screening process involving interconnected drug resistance patterns. There were 605 PLWH (77.17%) in the network, which comprised 181 distinct molecular clusters. The largest cluster contained 52 individuals, mainly from CD and DY, accounting for 61.54% (32/52) and 34.62% (18/52), respectively. There were 7 clusters of more than 10 individuals, mainly from the CD, DY, GA, LSZ, and other regions. Furthermore, 174 clusters with 2–9 individuals were identified, which were distributed mainly in the CD, LZ, LSZ, ZG, YB, and other regions (Figure 6B). Heterosexual transmission was the primary mode in the network, accounting for 90.08% (545/605), followed by homosexual transmission, accounting for 8.10% (49/605). In terms of drug types, 466 (77.02%) patients were resistant to NNRTIs, 138 (22.81%) were resistant to PIs, and 37 (6.12%) were resistant to NRTIs. Across the network, 140 NNRTI-resistant clusters were identified, with the largest cluster containing 24 individuals. For NRTI resistance, 12 clusters were found, the largest with 6 individuals. For PI resistance, 30 clusters were observed, the largest comprising 52 individuals (Figure 6C). For drug resistance mutations, the K103N, E138A, V179D, and Y181C clusters formed 39.14%, 25.36%, 10.14%, and 2.90% of the NNRTI clusters, respectively. The M184V cluster accounts for 41.67% of the NRTI clusters, whereas the Q58E cluster constitutes 63.33% of the PI clusters (Figure 6D).

Figure 6 (A–D) HIV-1 molecular network for the TDR incidence of newly diagnosed PLWH in Sichuan Province from July 2022 to June 2023.

Regional correlation analysis revealed a high proportion of TDR transmission between regions, reaching 54.12%. Cities such as LS (93.22%), MS (85.06%), DZ (82.76%), GY (80%), CD (67.16%), MY (66.98%), NC (66.67%), NJ (63.89%), and ZY (63.08%) presented strong interregional TDR, whereas GA (96.01%), LSZ (82.69%), ZG (78.95%), SN (75.61%), and YA (75%) presented strong intraregional TDR. LZ, PZH, YB, and DY had similar proportions of intraregional and interregional TDR. The spatial distance transmission analysis revealed that CD and DY had the highest correlation, radiating strongly southward. Remarkably, CD had strong TDR with MS, ZY, and YB, whereas DY had strong TDR with MY, LS, LZ, MS, and YB. In the southwestern region of Sichuan, the LS, in addition to intraregional connections, had low-intensity connections (below 5) with multiple regions but formed a strong connection with the CD. Similarly, the DZ also had connections with multiple regions, forming a strong connection with the MY (Figure 7 and 8).

Figure 7 TDR incidence of newly diagnosed PLWH between and within regions in Sichuan.

Figure 8 TDR matrix diagram of HIV transmission within and between regions for newly diagnosed PLWH in Sichuan.

Discussion

The study conducted a large-scale cross-sectional drug resistance surveillance among all newly diagnosed PLWH from July 2022 to June 2023. The results revealed a moderate TDR prevalence of 10.91%, significantly higher than the 4.9% reported in the 2022 Chinese CDC sampling survey. Combined with the 2014 and 2019 Sichuan drug resistance surveillance data, a distinct upward trend was observed.^20,28 Regionally, except for BZ (2.2%) and AB (0%), other regions exhibited moderate to high TDR prevalence, with LS (16.45%) reaching a high level. NNRTIs were the main contributor to drug resistance (8.08%), with EFV and NVP having TDR prevalence rates of 5.54% and 5.73% respectively. Key resistance mutations included V179 and K103. Further analysis revealed significant differences in the detection rates of drug resistance mutation sites among different HIV-1 subtypes. Multifactorial logistic regression analysis revealed that individuals under 15 years old, infected with CRF08_BC, CRF55_01B, or subtype B viruses, and those in areas with HIV antiviral treatment viral suppression rates ≥5% in the past three years were at a significantly elevated risk of TDR. Spatial autocorrelation analysis revealed clustered distributions of TDR and PIs, identifying geographic hotspots and high-high clusters. Molecular network analysis identified transmission clusters (52.76%) and interregional versus intraregional dynamics. Collectively, these findings emphasize the escalating threat of TDR, necessitating continuous TDR surveillance in Sichuan to monitor the prevalence of TDR in real-time, curb resistance transmission, and implement pretreatment drug resistance testing for newly diagnosed PLWH when necessary to guide clinical treatment.

In the analysis of drug types, Sichuan’s NNRTI drug resistance rate rose steadily from 0% in 2014 to 8.08% in 2023, while NRTI’s increased from 0% to 1.34% and PI’s climbed from 2.13%to 2.22% over the same period.^8,20,28 The prevalence of NNRTI was moderate. Specifically, LZ, SN and LSZ exceeded 10%, reaching 10.85%, 10.37%, and 13.18%, respectively. These findings underscore the necessity for further investigations to identify the specific causes of the elevated NNRTI resistance rates in these regions. If the rates remain above 10%, the 2016 WHO ART guidelines recommend considering supplementary first-line regimens for PLWH initiating ART with NNRTI TDR ≥10%.²⁹ From 2014 to 2020, among 30 countries reporting to the WHO, 21 exhibited NVP or EFV TDR prevalence rates exceeding 10% in PLWH commencing first-line ART.³⁰ In this study, the rates for EFV and NVP were 5.54% and 5.73%, respectively, aligning with earlier surveys conducted in China in 2015 and 2018.^19,31 However, in regions such as PZH, LZ, and LSZ, the resistance rates to both drugs surpassed 7%, with LSZ exceeding 10%, higher than previous surveys in Sichuan and LS.^20,32,33 The primary resistance mutations identified in this study were V179, K103, E138, and V106. Studies from Zimbabwe, Kenya, and Argentina have identified K103 as the most common NNRTI resistance mutation associated with first-line antiviral drugs.^34–36 Given the high prevalence of NNRTI resistance, a reevaluation of national HIV treatment strategies and regimens is warranted. In 2023, China released the “National Free HIV Antiretroviral Therapy Drug Treatment Manual” (Fifth Edition),³⁷ incorporating RPV into the first-line treatment regimen.³⁸ In this study, the RPV resistance rate was 4.08%, with rates exceeding 5% in LZ, DY, SN, GA, and YA. However, less than 1% of PLWH are currently using RPV, suggesting that resistance may primarily stem from cross-resistance among PLWH receiving ART,²² potentially contributing to the spread of RPV resistance. Therefore, RPV should be incorporated into routine drug resistance surveillance.

Furthermore, This study revealed significant differences in the detection rates of drug resistance mutations among different subtypes, such as S68 (12.79%) in CRF01_AE, Q58 (2.29%) in CRF07_BC, M184 (4.24%), K103 (4.98%), V179 (93.33%) in CRF55_01B, and V179 (87.72%) in CRF105_108, which were significantly greater than those in the other subtypes. The most common drug resistance mutation at S68 was S68G, which accounted for a relatively high proportion of NRTI mutations in CRF01_AE. It has no effect on drug sensitivity, but when combined with the major drug resistance mutation K65R, it can increase the replication fitness of K65R mutant resistant viruses.³⁹ Q58E does not affect sensitivity to LPV/r (a second-line treatment drug used only in China), so it can also be used as a specific marker for the spread of CRF07_BC.The most common mutation in V179 was V179D/E, which alone did not cause drug resistance, but when it occurred simultaneously with other mutation sites, such as K103R and V106I, it can cause moderate resistance to EFV and NVP, as well as low-level resistance to RPV. Some studies have also shown that, except for V179F, which was included in the 2009 version of SDRMs, other mutation sites have been excluded due to their high prevalence in certain viruses. Moreover, as a result of low genetic barrier of NNRTI, this site can be used as a specific marker for CRF55_01B.^40,41 In addition, this mutation site accounted for a relatively high proportion of CRF105_108, and whether it can be used as a specific marker for CRF105_108 requires further research. Moreover, the most prevalent drug resistance mutation observed in K103 was K103N, accounting for 93.20% (548/588). This mutation comprises more than 3% of the major subtypes in Sichuan and exceeds 14% in subtype B. K103N is highly resistant to EFV and NVP, both of which are free antiretroviral drugs in China.⁴² According to studies by the WHO and China, K103N is the most prevalent mutation in countries reporting HIV drug resistance data.^42,43 Therefore, considering the variations in drug resistance mutation sites across diverse HIV-1 subtypes, incorporating molecular epidemiological findings into the development of targeted treatment strategies can significantly increase the efficacy of ART.

This study found that individuals under 15 years old, those infected with CRF08_BC, CRF55_01B, or the B subtype, and residents in regions with an HIV ART viral suppression rate of 5% or higher in the past three years face a significantly increased risk of TDR. In the study, most PLWH under 15 were from LS (86.8%, 99/114), which had the highest TDR prevalence. Despite LS’s successful AIDS prevention and control project since 2017,⁴⁴ several issues remain: school students, the majority in this age group, may show poor medication adherence due to privacy concerns; pediatric patients struggle to access consistent medical services due to schooling; local medical resources and healthcare quality are limited; and overall treatment adherence in the region is poor, with guardians’ adherence indirectly affecting their children’s treatment. The high TDR prevalence among young PLWH may also stem from a large proportion of female PLWH infected heterosexually, mostly young and middle-aged adults,⁴⁵ and the region’s high viral suppression failure rate,⁴⁶ which increases mother-to-child transmission risk. Even if mother-to-child transmission is blocked, daily blood contact may still transmit drug-resistant viruses to children. Therefore, it is vital to ensure healthcare facilities and trained healthcare workers in rural or resource-constrained settings like LS. In addition, point-of-care resistance testing has the potential to provide rapid results, enabling clinicians to make timely treatment decisions in high risk populations. This capability is particularly valuable, and we urgently need the development of this technology. This highlights the need for ongoing vigilance over young PLWH in specific regions to prevent a decline in their quality of life, limit treatment options, and exacerbate drug resistance.^47–49

Compared with CRF01_AE, PLWH infected with CRF55_01B had 2.07 times higher TDR (OR: 2.07, 95% CI=1.37–3.13), aligning with Chen team’s research.⁸ This may stem from V179, a drug resistance mutation site in CRF55_01B, often linked with E138G and causing resistance to EFV and NVP.⁵⁰ CRF55_01B’s prevalence has risen in Sichuan (from 0.38% in 2019 to 1.21% now),²⁰ so controlling its spread may curb drug resistance. Similarly, TDR prevalence among PLWH with CRF08_BC is 1.64 times higher than those with CRF01_AE, consistent with a 2019 Sichuan survey. The E138 mutation site (8.90%) detection rate was higher than in other subtypes, mainly conferring low-level RPV resistance, necessitating strengthened RPV resistance monitoring for CRF08_BC-infected PLWH. Additionally, the drug resistance rate among PLWH with the B subtype was 2.23 times higher than those with CRF01_AE (OR: 2.23, 95% CI = 1.52–3.25). Despite recombinant viruses’ prevalence in Sichuan, some studies found a short-term B subtype outbreak in certain regions (5.79% with an 83.3% detection rate),⁵¹ mainly among those aged 50 and above and heterosexually infected. The survey showed higher K103 (14.81%) and V106 (7.41%) mutation detection rates in the B subtype, indicating a need for further drug resistance surveillance.

The higher HIV viral suppression failure rate in regions where PLWH have lived over the past three years positively correlates with a higher TDR prevalence, highlighting the need to improve ART quality. The viral suppression failure rate among PLWH has stayed around 10% in recent years, mainly due to drug resistance.^52,53 Persistent high viral loads spread drug-resistant strains, increasing resistance prevalence.⁵⁴ Poor treatment quality and drug resistance are mutually causative. Thus, it’s essential to standardize ART management in high TDR prevalence areas, especially by improving healthcare personnel training and patient adherence education. This will boost antiviral treatment success and curb drug-resistant strain generation. Additionally, drug resistance monitoring should be strengthened in areas with poor ART quality, with timely feedback to guide clinicians in regimen selection.

At a genetic distance threshold of 0.01, clustered drug resistance mutations were significant, encompassing 605 PLWH (77.17%, 605/784). NNRTI mutations were predominant (77.02%), followed by PI (22.81%) and NRTI (6.12%), indicating severe TDR mutation transmission among newly diagnosed PLWH. K103N (40.83%, 247/605) was the most frequent TDR mutation, which is consistent with the findings of the 2022 national survey and Kenyan coastal research.^8,55 Owing to the good viral fitness of K103N-containing strains, which can establish stable TDR genetics, this mutation likely has circulated and spread as new infections regionally.⁵⁴ A Q58E-dominated PI resistance cluster (52 individuals) emerged in CD and DY, primarily via heterosexual and MSM((men who have sex with men). Conversely, only five (2.76%) NRTI clusters with M184V mutations were found, far fewer than in ART PLWH, underscoring viruses with the M184V mutation have limited transmission potential due to reduced replicative fitness.⁵⁴ These findings suggest integrating drug resistance surveillance with molecular network analysis in Sichuan to pinpoint TDR clusters. The molecular clusters provide a valuable framework for prioritizing contact tracing efforts. Combining this information and epidemiological data can guide targeted contact tracing, enabling health authorities to focus their resources on those most at risk of further transmission. The molecular findings can also inform targeted counseling strategies. Healthcare professionals can better educate patients about their individual risks, adherence to treatment, and preventive measures by understanding the genetic characteristics of different clusters. Furthermore, researchers and clinicians can design ART regimens that are more effective against the prevalent strains in those areas. By molecular networks with epidemiological data, precise interventions for high-risk groups and targeted measures to curb TDR can be implemented.

ArcGIS-based spatial clustering and hotspot analysis of TDR prevalence showed high-high clusters and hotspots in adjacent counties, especially in the southwest, south, central, and northeast, indicating spatial correlation likely due to similar lifestyles, convenient transport, and economic ties. More medical resources should be directed to these areas, particularly key epidemic regions like LS, CD, ZG, YB, DY, and MY, for enhanced AIDS prevention and control, such as increasing the number of laboratories, adding laboratory facilities, and raising the frequency of viral load testing as well as follow-up visits. In contrast, cold spots may require a different surveillance strategy, namely a more cost-effective and less intensive surveillance model, such as periodic sampling and data analysis. Compared with cold spots, hot spots have a more urgent need to update first-line treatment regimens to achieve virological suppression.

Geographical analysis of TDR among PLWH revealed strong ties between CD, DY, and other cities, contrasting with a 2018 Sichuan survey on PLWH with viral suppression failure.⁵⁶ Key cross-regional TDR transmitters to DY were LS(31), YB(34), LZ(32), and MS(26), despite long distances, while only neighboring CD(444) and MY(46) showed high transmission intensity, indicating distant cities can also be closely linked. Within-city TDR linkages were prominent in GA(96.01%), LS(82.69%), ZG(78.95%), and SN(75.61%), suggesting regional transmission. Deeper-rooted societal determinants are likely the driving forces behind the spatial movement patterns of PLWH harboring TDR-associated HIV strains. Prior studies show increased population exposure and mobility raise disease transmission risks, with cities acting as outbreak incubators, and urbanization may strengthen intercity disease spread.⁵⁷

There are still limitations in our research. Firstly, variations in sample preservation conditions and delivery times across regions in 2022 resulted in a detection success rate of only 79.49% (13,619/17,132), potentially introducing bias into the findings. Secondly, as a cross-sectional study, it could not accurately capture the trends of TDR in recent years across regions or properly evaluate the impact of ART on TDR in Sichuan. Thirdly, we used the reported location of PLWH in the AIDS prevention and treatment information system as the geographic parameter, but due to population mobility, the actual regional TDR prevalence and geographic characteristics may differ. Lastly, this study did not collect behavioral risk data which may limit interpretations of transmission dynamics. And While current guidelines advocate for INSTI-based regimens, the absence of INSTI resistance testing in our study resulted in a critical gap in data on transmitted INSTI resistance.

The findings from our 2023 TDR survey in Sichuan underscore the necessity to review and reconsider some current first-line ART regimens. Pre-treatment resistance testing allows for the selection of targeted antiretroviral therapy, which is crucial for achieving the goal of 95% viral suppression among treated individuals, within 95–95-95 target set by UNAIDS (UNAIDS, 2023) and China’s Plan for Containing and Preventing AIDS (2024–2030). This approach aligns with WHO recommendations for genotype-guided therapy initiation Regions in Sichuan with higher TDR prevalence should formulate targeted strategies and allocate additional resources. Feasible interventions include expanding genotypic TDR surveillance pilot programs (enabling HIV genotypic resistance testing in medical institutions), promptly conducting resistance testing for all newly diagnosed PLWH, regularly performing viral load testing to facilitate rapid ART regimen adjustments. Additionally, enhancing the resistance interpretation capabilities of frontline healthcare workers and implementing tailored measures to improve patient adherence are crucial. Interregional joint prevention and control efforts should be strengthened, leveraging molecular network technologies and fostering enhanced communication and collaboration to curb the emergence of transmitted drug resistance.

Conclusion

The overall prevalence of transmitted HIV drug resistance in Sichuan exhibits an upward trend, reaching a moderate epidemic level, with certain cities even experiencing high-level prevalence. Resistance to NNRTIs was predominant. Through molecular epidemiology, spatial cluster analysis, subtype-specific mutation profiling, and interregional transmission network mapping, we identified elevated drug resistance among infections with CRF55_01B, CRF08_BC, and subtype B. Additionally, we pinpointed key mutations (K103N, V179D, M184V, Q58E) that are critical for understanding both the emergence of drug resistance and its transmission dynamics.Further analysis revealed an increasing rate of RPV resistance driven by cross-resistance. In regions with high HIV prevalence and strong transmission linkages, such as LS, DY, and CD, baseline drug resistance testing should be integrated into first-line ART decision-making. Municipal and county-level health authorities should enhance surveillance capacity through mobile sequencing platforms. Additionally, frontline healthcare workers require training in drug resistance analysis and molecular network techniques to implement precision interventions and adherence education for high-risk individuals, leveraging molecular network data combined with epidemiological investigations. Resistance data must also be submitted to the National CDC to inform treatment guideline updates.