Primary drug resistance and transmission analysis of HIV-1 in acute and recent drug-naïve seroconverters in Singapore


  • An abstract based on this work was presented at the National Health Group (NHG) Annual Scientific Congress 2008, 7–8 November 2008, Suntec Singapore International Convention and Exhibition Centre, Singapore.

Dr Yong-Jiang Sun, Infectious Disease Research Laboratory, Level 2, TTSH Medical Centre, Tan Tock Seng Hospital, 11 Jalan Tan Tock Seng, Singapore 308433. Tel: 65-6357-8461; fax: 65-6256-7023; e-mail:



The aim of the study was to elucidate primary drug resistance and transmission of HIV-1 in acute and recent drug-naïve seroconverters in Singapore.


Acute and recent HIV-1 seroconverters were enrolled in the study. The HIV-1 polymerase (pol) gene was sequenced and used for genotypic drug resistance analysis and phylogenetic analysis. HIV-1 transmission clusters were inferred from phylogenetic clustering analysis.


Of the 60 subjects analysed, 95% were men, and 73.3% were men who have sex with men (MSM). Six HIV-1 subtypes were identified, including CRF01_AE (46.7%), subtypes B (30%), B′ (15%) and G (1.7%), CRF33_01B (1.7%) and CRF34_01B (5%). Primary genotypic resistance was detected in only one (1.7%) subtype B variant. Thirty-one patients (51.7%) were phylogenetically clustered, of whom 90% reported having local risk exposure, compared with 59% of the patients who were not phylogenetically clustered [odds ratio (OR) 6.35, 95% confidence interval (CI) 1.65–23.95]. MSM (OR 5.63, 95% CI 1.17–27.15), high viral load (OR 4.28, 95% CI 1.37–13.36) and young age (OR 0.92, 95% CI 0.85–0.99) were independently associated with clustered individuals.


In Singapore, HIV-1 primary resistance is insignificant; individuals with seroconversion account for about half of onward transmission among recently infected seroconverters. MSM, high viral load and young age are factors that facilitate transmission. Early detection of these individuals is of paramount importance for the prevention of HIV-1 transmission.


Primary or transmitted drug resistance in HIV-1 has been a significant clinical and public health concern with the widespread use of antiretroviral therapy (ART) world-wide, particularly in the developed world [1]. A number of recent studies conducted in the USA and Western European countries have shown that the prevalence rates of HIV-1 primary resistance range from approximately 8 to 16% in recently infected, drug-naïve individuals [2–6]. Although HIV-1 primary resistance is less prevalent in Asian countries, a recent report from Thailand has demonstrated a comparable prevalence rate (12.4%) to those reported in developed countries, presumably reflecting increased access to ART in Thailand [7].

In Singapore, antiretrovirals have been available since the early 1990s and ART has been readily accessible from the year 2000 [8]. However, a previous study conducted in our centre showed that none of the 35 variants of CRF01_AE (which is the predominant subtype in both Thailand and Singapore) obtained from newly diagnosed, drug-naïve patients was drug-resistant [9]. As certain drug resistance mutations of HIV-1 can persist in transmitted viruses for only about 2 years [10,11] and the durations of HIV-1 infection for the subjects in the previous Singapore study [9] are unknown, it was necessary to further validate this finding and to obtain a more recent picture of primary resistance in the HIV-1 CRF01_AE subtype and other viral subtypes in Singapore.

Studies in European countries have shown that the transmission of drug-resistant HIV-1 is more common in patients with primary HIV-1 infection (PHI) or seroconversion than in chronically infected patients [3]. Paraskevis et al. [3] reported that the overall prevalence of HIV-1 primary resistance in newly diagnosed individuals was 9%, but when only recent seroconverters were considered, the rate increased to 22%. These findings suggest that patients with PHI/seroconversion are a more sensitive study population for the detection of HIV-1 primary resistance.

In addition, according to the Joint United Nations Programme on HIV/AIDS (UNAIDS)/World Health Organization (WHO) (December 2007) report, the globally increasing trend of HIV-1 infection has been slowing down in recent years, with a decreasing incidence of new infection [12]. However, the incidence of new HIV-1 infection in Singapore has increased in the last few decades, from less than 1 case per million of population in 1985 (when the first case of HIV infection was reported in this country) to 101.3 cases per million of population in 2006 (HIV statistics of Singapore, available at This situation calls for more research to elucidate the transmission dynamics and patterns of HIV-1 infection in order to enable the design of more effective strategies for HIV-1 prevention in Singapore.

In the present study, we aimed to analyse HIV-1 primary resistance in acute and recent drug-naïve seroconverters in Singapore. In addition, the HIV-1 polymerase (pol) gene sequences generated from drug resistance genotyping have served as very useful, convenient data for the phylogenetic investigation of the spread of HIV-1 in a community [13,14]; we therefore also aimed to study the dynamics and patterns of HIV-1 transmission among seroconverters in Singapore.


Subjects and setting

Seroconverters were defined in this study as HIV-1-seropositive patients who had at least one of the following: (1) a previous negative HIV antibody test [by enzyme-linked immunosorbent assay (ELISA) or Western blot] within 24 months of the first positive HIV test; (2) an evolving Western blot (i.e. indeterminate to positive); or (3) a consistent clinical seroconversion illness associated with a significant recent HIV exposure, with a positive Western blot at presentation. All recently infected patients who were diagnosed between May 2006 and December 2007 and satisfied the above-mentioned inclusion criteria were prospectively and consecutively recruited from the out-patient clinic at the Communicable Diseases Centre (CDC), Tan Tock Seng Hospital, which provides clinical care for approximately 95% of all HIV-1-infected individuals in Singapore [8]. In addition, 14 seroconverters diagnosed between October 2002 and April 2006 who had plasma samples stored at −80 °C were also included in this analysis to produce an appropriate sample size. All plasma samples were collected within 24 months of the date of the last negative HIV test/indeterminate Western blot or the estimated date of seroconversion based on clinical seroconversion illness, and before ART initiation. Demographic and clinical data were collected from medical records and by interview. The Ethics Committee of the National Healthcare Group (NHG), Singapore, granted approval for this study. Written informed consent was obtained from all study subjects.

Genotypic resistance analysis

Population-based nucleotide sequence analysis of the HIV-1 pol gene was performed on the plasma samples using the Celera Diagnostics ViroSeq HIV-1 Genotyping System (version 2.0) (Celera Diagnostics, Alameda, CA, USA) according to the manufacturer's instructions, except that less RNA diluent was used for three samples which had a viral load <2000 HIV-1 RNA copies/mL. Capillary electrophoresis sequencing was performed on an ABI PRISM 3100-avant Genetic Analyzer (Applied Biosystems, Foster City, CA, USA). Sequence data were analysed using the Celera Diagnostics ViroSeq HIV-1 Genotyping System software (version 2.6), which assembles DNA sequences from sequencing primers into a 1302-nucleotide contiguous sequence covering the whole protease gene and two-thirds of the reverse transcriptase gene (corresponding to nucleotides 2253 to 3554 in the HXB2 reference sequence). The amino acid mutations of the sequenced pol gene were identified using the online tool of the Stanford University HIV Drug Resistance Database (version 5.0.0; The identification of drug resistance-associated mutations was based on the updated (August/September 2007) International AIDS Society-USA (IAS-USA) Drug Resistance Mutations Figures [15].

Phylogenetic analysis

HIV-1 M group subtype and circulating recombinant form (CRF) reference strain sequences, including the pre-grouped set of year 2005 in the Los Alamos HIV Sequence Database (55 sequences, comprising subtypes A to K and CRFs 01 to 19) and an additional three reference sequences (B.CN.RL42 for sub-subtype B′, 05MYKL007 for CRF33_01B, and OUR2275P for CRF34_01B), were obtained from the Los Alamos HIV Sequence Database ( Alignment of these reference sequences and the sequences generated from this study was performed using ClustalW implemented in Bioedit software (version 7.0.9; The best-fit model of nucleotide substitution was selected using Modeltest [16]. The model chosen for this set of sequences was the general time-reversible model with gamma-distributed rates across sites (GTR+G). A maximum-likelihood (ML) phylogenetic tree was reconstructed under the above-mentioned evolution model using Treefinder (January 2008 version) [17] and statistical support for the ML tree topology was assessed by bootstrap analysis with 1000 replicates. A bootstrap value of 70% or greater was used to define a phylogeny cluster. The study sequence(s) was assigned the same subtype or CRF as the reference sequence(s) in the same phylogeny cluster. The tree was rooted using mid-point rooting and presented using the program Figtree (version 1.1.2;

Identification of transmission clusters

The ML tree clades that consisted of only study sequences and were supported by 95% or greater bootstrap values were identified as transmission clusters. Patients in each of these clusters were considered to have been infected as a result of local network transmissions.

Statistical analysis

Statistical comparisons of the characteristics and viral subtypes of patients in transmission clusters with those of patients not in transmission clusters were performed using the χ2 test, the Mann–Whitney U-test, or Student's t-test, as appropriate. A P-value of <0.05 was considered statistically significant. Odds ratios (ORs) and their 95% confidence intervals (CIs) were also calculated, as appropriate. Multivariate logistic regression analysis was used to adjust for potential confounding factors.

Sequence accession numbers

The sequences analysed in this study have been deposited in GenBank under accession numbers EU715177–EU715236.


Patient characteristics

Forty-seven recently infected seroconverters were enrolled in the study between May 2006 and December 2007, accounting for approximately 8% of all newly diagnosed HIV-infected patients (n=578) in this period; together with the 14 subjects identified between October 2002 and April 2006, a total of 61 seroconverters were recruited. The sample from one patient (which was one of the three samples with a viral load <2000 copies/mL) did not yield a polymerase chain reaction (PCR) product for sequencing and was excluded from further analysis. The study analysis therefore concerns 60 patients for whose samples genotyping analysis was successfully performed.

Of the 60 patients, 54 (90%) had a history of having either an HIV-1-negative test or an evolving Western blot before the first positive HIV-1 test, and 42 of these patients also showed clinical seroconversion illness; the remaining six patients (10%) who did not have a prior HIV-1-negative test presented with a syndrome consistent with HIV seroconversion illness correlated with recent high-risk exposure(s). The median time from the date of the last negative HIV-1 test/indeterminate Western blot before the first positive HIV-1 test or the estimated date of seroconversion to the genotyping sample collection date was 5 months (range <1–24 months; 90th percentile 18 months).

The risk factors for HIV-1 infection in the 60 subjects were MSM (n=44; 73.3%), heterosexual contacts (n=14; 23.3%), and injecting drug users (IDUs) who shared injection equipment (n=1; 1.7%); risk factors could not be elicited for the one remaining patient. Twenty-nine patients (48.3%) reported having local risk exposure only, 16 (26.7%) reported having overseas risk exposure only, and 14 (23.3%) had both local and overseas risk exposures; this information was not available for the remaining one patient. Almost all of the patients (57; 95%) were men and of Chinese ethnicity. The mean age was 34.9 years [standard deviation (SD) 10.5 years], the median viral load was 5.88 log10 copies/mL (range 2.51–6.21 log10 copies/mL), and the median CD4 cell count was 352 cells/μL (range 44–1148 cells/μL) (Table 1).

Table 1.   Demographic and clinical characteristics of the 60 HIV-1 seroconverters at the time of diagnosis
  • *

    Viral load was unknown for one patient.

  • SD, standard deviation; MSM, men who have sex with men.

Age (years) (mean ± SD)34.9 ± 10.5
Male sex [n (%)]57 (95.0)
Ethnicity [n (%)]
 Chinese57 (95.0)
 Malay2 (3.3)
 Other1 (1.7)
Risk factor [n (%)]
 MSM44 (73.3)
 Heterosexual contact14 (23.3)
 Injecting drug use1 (1.7)
 Unknown1 (1.7)
Geographic exposure [n (%)]
 Singapore only29 (48.3)
 Overseas only16 (26.7)
 Singapore and overseas14 (23.3)
 Unknown1 (1.7)
Viral load (log10 copies/mL) [median (range)]*5.88 (2.51–6.21)
CD4 count (cells/μL) [median (range)]352 (44–1148)

HIV-1 subtypes

Figure 1 shows the phylogenetic tree of the study sequences and the reference sequences constructed using the ML method. Based on the criteria for HIV-1 subtyping described above, the 60 pol sequences were assigned to each of six HIV-1 subtypes/CRFs, including CRF01_AE (n=28; 46.7%), subtype B (n=18; 30%), sub-subtype B′ (n=9; 15%), subtype G (n=1; 1.7%), and the recently identified CRF33_01B (n=1; 1.7%) as well as CRF34_01B (n=3; 5%) [18,19].

Figure 1.

 Maximum likelihood phylogenetic tree of HIV-1 polymerase (pol) sequences. The maximum likelihood tree was constructed based on 60 pol sequences of HIV-1 from seroconverters in Singapore and 58 subtype/circulating recombinant form (CRF) pol reference sequences. Bootstrap analysis was performed with 1000 replicates. Bootstrap values higher than 70% are shown on internal branches. The study sequences are indicated using coloured branches. The taxon name of the study sequences represents the subject code (e.g. S07001) followed by HIV-1 risk factors (MSM, men who have sex with men; HS, heterosexual exposure; IDU, injecting drug user; question mark, denies any risk factor) and the year of HIV-1 diagnosis. The seven phylogenetic clusters supported by a bootstrap value ≥95% are labelled C1 to C7. HIV-1 subtypes and CRFs are indicated.

HIV-1 primary resistance

Genotypic evidence of drug resistance was detected in only one subtype B variant (S07027-HS-07 in Fig. 1) which had mutations K103N and Y181C in the reverse transcriptase (RT) gene, which confer resistance to nonnucleoside reverse transcriptase inhibitors (NNRTIs). Therefore, the overall prevalence rate of primary resistance was 1.7% (1/60). On the ML tree, this variant had a long branch and presented as a unique strain (Fig. 1). The tree topology showed no change when we replaced the mutant codons with wild-type codons at the two resistance mutation positions. The patient who carried this drug-resistant variant reported that he was probably infected in China by heterosexual contact.

Clustering analysis

Based on the tree topology and the definition of a transmission cluster, seven transmission clusters were identified which comprised two sub-subtype B′ clusters, three subtype B clusters, and two CRF01_AE clusters. These clusters corresponded to 31 (51.7%) clustered sequences (Fig. 1).

In order to identify risk factors for transmission clusters, we compared the characteristics and HIV-1 subtypes of the clustered patients with those of the nonclustered patients using both univariate (χ2 test, Mann–Whitney U-test or Student's t-test) and multivariate logistic regression analyses (Table 2). Univariate analyses revealed that clustered patients were more likely to have acquired HIV-1 infection by MSM (OR 6.35; 95% CI 1.65–23.95; P=0.006), to have had local risk exposure in Singapore (OR 6.35; 95% CI 1.65–23.95; P=0.006) and to have higher viral loads (median viral load 5.03 vs. 4.58 log10 copies/mL, respectively; P=0.02) than nonclustered individuals. In addition, clustered patients appeared to be younger (mean age 32.5 vs. 37.5 years, respectively; P=0.066) and more likely to be infected with CRF01_AE (OR 2.63; 95% CI 0.94–7.4; P=0.067) than the nonclustered patients, but these differences were not statistically significant in univariate tests. No significant association was seen with sex, ethnicity or CD4 cell count.

Table 2.   Comparison of the characteristics and HIV-1 subtypes of clustered and the nonclustered individuals
Characteristic and viral subtypeClustered (n=31)Nonclustered (n=29)Univariate*Multivariate
OR (95% CI)POR (95% CI)P
  • *

    χ2 test, Mann–Whitney U-test, or Student's t-test.

  • Logistic regression.

  • Risk factor was unknown for one clustered patient.

  • §

    § Geographic exposure was unknown for one clustered patient.

  • Viral load was unknown for one clustered patient.

  • CD4 cell count was unknown for one nonclustered patient.

  • OR, odds ratio; CI, confidence interval; IDU, injecting drug user; SD, standard deviation; MSM, men who have sex with men.

Male sex [n/total (%)]31/31 (100)26/29 (89.7) 0.107  
Chinese ethnicity [n/total (%)]31/31 (100)26/29 (89.7) 0.107  
Age (years) (mean ± SD)32.5 ± 7.837.5 ± 12.4 0.0660.92 (0.85–0.99)0.033
Risk factor [n/total (%)]
 MSM/bisexual27/30 (90.0)17/29 (58.6)6.35 (1.65–23.95)0.0065.63 (1.17–27.15)0.031
 Heterosexual+IDU3/30 (10.0)12/29 (41.4)    
Geographic exposure [n/total (%)]§
 At least Singapore27/30 (90.0)16/29 (55.2)7.31 (1.90–27.49)0.003  
 Overseas only3/30 (10.0)13/29 (44.8)    
Viral load (log10 copies/mL) [median (range)]5.03 (3.78–6.21)4.58 (2.51–6.18) 0.024.28 (1.37–13.36)0.012
CD4 count (cells/μL) [median (range)]345 (44–1145)367 (60–900) 0.70  
HIV-1 subtype [n/total (%)]
 CRF01_AE18/31 (58.1)10/29 (34.5)2.63 (0.94–7.40)0.0673.73 (0.96–14.50)0.058
 All other subtypes13/31 (41.9)19/29 (65.5)    

We further performed multivariate logistic regression analysis using HIV-1 risk factor (dichotomous as MSM and other risks), viral load (log-transformed), age, and HIV-1 subtype (dichotomous as CRF01_AE and other subtypes/CRFs) as predictor variables; the last two were included because the P-values of their univariate analyses were very close to the 0.05 level of significance. Geographic exposure was not included in logistic regression analysis, despite it being a significant factor in univariate analysis, because having local risk exposure is a prerequisite for being involved in a local transmission cluster. Multivariate analysis confirmed that MSM (OR 5.63; 95% CI 1.17–27.15; P=0.031) and higher viral loads (OR 4.28; 95% CI 1.37–13.36; P=0.012) were independently associated with transmission clusters. In addition, the negative association between the age of the patient and transmission clusters became statistically significant (OR 0.92; 95% CI 0.85–0.99; P=0.033) after multivariate adjustment by logistic regression analysis. The association between CRF01_AE variants and transmission clusters was still not statistically significant after the multivariate adjustment.


We detected drug resistance mutations in only one subtype B variant (1.7%). Patient contact data suggested that this drug-resistant variant was probably introduced from China rather than being circulated locally, and this suggestion was supported by phylogenetic evidence, which showed that this variant had a long and unique branch on the phylogenetic tree (Fig. 1). In agreement with the previous Singapore study [9], we did not find resistance mutations in any of the 28 CRF01_AE variants in the present study, despite the facts that this study was conducted approximately 4 years after the previous one and a more sensitive study population (acute and recent seroconverters) was used. These data suggest that HIV-1 primary resistance is insignificant in Singapore and therefore routine monitoring of primary drug resistance is currently not especially necessary.

We found previously that the prevalence of HIV-1 drug resistance, as evidenced by genotypic resistance mutations, among individuals who had been on ART for at least 2 years was approximately 30% in our setting (unpublished data). It is estimated that about 75% of HIV-1-infected patients in Singapore have been on ART for at least 2 years, and hence the prevalence of drug-resistant HIV-1 in the entire HIV-1-infected population in Singapore would be about 22%, much lower than the prevalences in Western countries and in Thailand [7,11,20,21]. This might contribute to the insignificant prevalence of primary resistance. The sexual behaviour of local patients might have changed substantially after they discovered their HIV-positive status. This could also be a big contributing factor to the insignificant transmission of primary HIV drug-resistant variants seen in our study. It should be pointed out, however, that the high rate of clustering of CRF01_AE (58%; Fig. 1 and Table 2) may diminish the advantage of seroconverters being a more sensitive population for the detection of primary resistance, because the clustered variants were actually from a very limited number of source transmitters.

It has long been realized that HIV-1-infected individuals with PHI/seroconversion are highly contagious and therefore play an important role in the spread of the virus [22,23]; a number of studies have confirmed this [20,24–29]. HIV-1-infected individuals at this disease stage contribute disproportionately to the epidemic, with approximately 29–50% of onward transmissions attributable to these individuals, who represent <10% of patients in an HIV-1-infected population, as demonstrated by phylogenetic clustering and epidemiological analysis [20,24,27]. Two major characteristics of this group of people, i.e. their high viral load and ignorance of their infection status, are believed to account for the high transmission rate [24,26,28,29]. Indeed, in phylogenetic clustering analysis in the present study, we found that approximately 52% of the seroconverters were involved in potential transmission clusters, a very similar rate to that (50%) found in a Canadian study [20], but higher than the those (29–34%) found in Swiss and UK studies [24,25,27]. Our study further underscores the importance of patients with PHI/seroconversion as potential transmitters in HIV-1 transmission. However, we cannot exclude the possible involvement of chronically infected patients in the transmission clusters based on the observations of this study.

We found that HIV-1 infection risk factor, viral load and the patient's age were independently associated with phylogenetically related clusters (Table 2). Individuals with MSM risk, high viral load and/or young age were more likely to be found within the clusters, suggesting that these factors facilitate HIV-1 transmission. It is not surprising that MSM, young age, and local exposure are risk factors for clustering as the whole set-up for meeting for MSM is different from that of other risk groups. It is common for young MSM to congregate in places designated for MSM only, for example bath-houses, allowing more opportunities for interactions. The critical role of plasma HIV-1 RNA levels of sexual transmitters and of pregnant mothers has been elegantly demonstrated previously for heterosexual and perinatal transmission [26,30–32]; and an increased transmission risk for younger transmitters has also been observed for heterosexual transmission [26]. In the present study, we found that plasma viral load and the patient's age played the same roles in male–male transmission of HIV-1 as observed in previous studies.

In summary, individuals with PHI/seroconversion account for about half of onward transmissions among recently infected seroconverters in Singapore; MSM, high viral load and young age are factors that facilitate the transmission. Early detection of these individuals is therefore of paramount importance for the prevention of transmission. Prevention programmes for MSM will also need to be improved. Despite the high onward transmission of HIV-1 among seroconverters, HIV-1 primary resistance remains insignificant in Singapore.


We thank Ms Meng Li Teo and Suet Mei Chun for their technical assistance. We also acknowledge the contribution of the physicians and nursing staff at the out-patient clinic of the Communicable Disease Centre, Tan Tock Seng Hospital. We also thank the patients who made the study possible. This study was supported by a grant from the National Medical Research Council of Singapore.

Conflicts of interest: None.