Transmitted drug-resistance in human immunodeficiency virus-infected adult population in El Salvador, Central America

Authors


Abstract

El Salvador harbours one of the largest Central American human immunodeficiency virus (HIV) epidemics, but few studies have analysed it in depth. Here, we describe the presence of transmitted drug resistance (TDR) and HIV variants in the HIV-infected adult population in El Salvador. Dried blood spots from 119 HIV-infected antiretroviral-naive adults attended in El Salvador were collected in 2011. The TDR was assessed according to the list recommended by the WHO. HIV-1 variants were described using phylogeny. Pol sequences could be amplified in 88 patients (50.6% men), with a mean age of 35 years. Almost all (96.7%) were infected with HIV through sexual practice and 58.7% were recently diagnosed. The mean CD4+ count was 474 cells/mm3 and 43.1% and 15.5% of patients showed moderate (<500 CD4 cells) or severe (<200) immune suppression, respectively. HIV-1 viral load was >100 000 copies/mL in 24.7% of patients and <2000 copies/mL in 9.1%. Five samples (5.7%) harboured any TDR mutation: 2.3% for nucleoside reverse transcriptase inhibitor (NRTI) and non-nucleoside reverse transcriptase inhibitor (NNRTI), and 1.4% for protease inhibitor (PI). All showed only one TDR single-class resistance mutation: M184I (two cases) for NRTI, K101E and K103N for NNRTI and L23I for PI. All viruses excepting one (URF_BG) belonged to subtype B. No phylogenetic TDR networks were found. In conclusion, we report a TDR prevalence of 5.7% in El Salvador, lower than in other Central American studies. Periodical studies are essential to monitor and prevent TDR emergence in low-income and middle-income regions. Also, more efforts are needed to promote early diagnosis and prevention of infection in El Salvador.

Introduction

Central America is among the regions with highest human immunodeficiency virus (HIV-1) prevalence worldwide, with 1.5 million people infected by HIV [1]. The HIV prevalence ranges from 0.3% in Mexico to 2.3% in Belize [1]. In 2009, 34 000 people were living with HIV in El Salvador, with an estimated HIV prevalence of 0.8% in the adult population [1]. The national antiretroviral treatment (ART) programme started to offer therapy in 2001, and began to scale-up access in 2004. It established as a first-line regimen the combination of two nucleoside reverse transcriptase inhibitors (NRTI, zidovudine or stavudine plus lamivudine) and one non-nucleoside reverse transcriptase inhibitor (NNRTI, nevirapine or efavirenz). In 2010, ART coverage was 59% of the 16 000 HIV-infected people eligible for treatment [2].

Effectiveness of ART could be compromised by the presence of transmitted HIV drug-resistance mutations (TDR), i.e. those found in previously untreated HIV-infected individuals, which increase the rates of virological failure [3]. Recent studies report an increasing prevalence of HIV drug resistance in low-income and middle-income countries following the scaling -up of ART [4, 5]. Certainly, TDR rates in Latin America and the Caribbean are on the rise [6]. In El Salvador's neighbouring countries, TDR is 7–12% [7-10] among the HIV-infected adult population and recent studies found a TDR prevalences of 10% in the Salvadorean HIV-infected paediatric population [11] and of 9.4% in sex workers and men who have sex with men [12]. However, no studies have so far described the TDR rate among the general HIV-infected population in El Salvador.

To prevent the widespread transmission of drug resistance in low-income and middle-income countries, systematic population-based surveillance studies among untreated individuals are crucial. For that purpose, the use of dried blood spots is a practical and safe method of sample recollection, storage and shipment [13, 14]. In this study, promoted by the National STD-HIV/AIDS Programme of the Health Ministry of El Salvador, we describe the prevalence of transmitted drug resistance and the distribution of HIV variants in the HIV-infected adult population in El Salvador using dried blood spots collected in December 2011.

Materials and Methods

Study cohort

The sample size was determined according to the number of eligible patients (HIV-infected, ART-naive individuals with detectable HIV viral load in the last sample and followed at Salvadorean HIV/AIDS clinics) and an expected TDR rate of 9% as reported in neighbouring countries [9]. To determine the study sample size the following formula was used,

display math

where n is the sample size, N is the population size, Z is the value corresponding to the Gaussian distribution 1.96 for α = 0.05, p is the expected prevalence, q is 1–p and i represents the expected error, which is 10% for α = 0.05. Considering 2300 eligible HIV-infected patients in El Salvador according to the system of health monitoring and evaluation that collects information from epidemiologically relevant diseases in El Salvador (SUMEVE March 2010, personal communication), we estimated the size of the study sample to be 119 patients. The Ministry of Health requested that patients of all health centres were chosen by random selection. A number was assigned to each patient and new cases were incorporated consecutively at the time of diagnosis until the number of cases was completed. Demographic (age, gender, origin, diagnosis date and mode of HIV infection) and clinical (infection stage, HIV-1 viral load, CD4+ cell count) information was collected by the physicians.

This study was part of a project approved by the Health Ministry of El Salvador Ethics Committee and by the Hospital Ramón y Cajal Clinical Research Ethics Committee. It was designed to protect the rights of all subjects involved and to maintain their confidentiality. A signed informed consent was obtained for each patient.

RNA extraction and sequencing

Dried blood spots were collected by adding two blood drops to 903 filter paper cards (Schleicher & Schuell, Bio-Science GmbH, Barcelona, Spain) in the Central Laboratory of the Health Ministry of El Salvador in December 2011. Cards were dried overnight at room temperature and stored at −20°C as recommended [13] before they were shipped on dry ice to the Hospital Ramón y Cajal (Madrid, Spain) for testing. Then, blood from two dried blood spot circles per patient was eluted in 2 mL of NucliSENS lysis buffer (BioMérieux, Madrid, Spain) for 2 h with gentle rotation and processed to extract RNA according to the manufacturer's instructions using NucliSENS EasyMAG instrumentation (BioMérieux). HIV-1 RNA was amplified by RT-PCR followed by nested PCR in partial HIV-1 pol gene (including the protease (PR, 297 nucleotides) and/or reverse transcriptase (RT) coding regions, 775 nucleotides) using conditions based on recommendations from the WHO [13]. The nested PCR-purified products obtained from HIV-1 RNA extracted from dried blood spots were sent to Macrogen (Gasan-dong, Geumchun-gu, Seoul, Korea) for direct sequencing.

Drug resistance

The pol sequences were uploaded to the Calibrated Population Resistance tool (http://cpr.stanford.edu/cpr.cgi) [15] to identify TDR changes defined according to the list of mutations for TDR surveillance recommended by the WHO and described in Bennett et al. [16]. Additional mutations considered in the 2011 International AIDS Society-USA (IAS-USA) mutation list [17] were manually identified to increase the information available for the clinicians about potential emergence of resistance when facing the first use of ART.

HIV-1 subtyping

HIV-1 subtypes and circulating recombinant forms were identified by phylogenetic analysis of the pol sequences. At least two representative sequences of each of the nine subtypes and the 49 circulating recombinant forms of HIV-1 group M available at the time of analysis were taken as references. Neighbour-joining trees were built using MEGA5 [18] under the Kimura two-parameter model, with 1000 bootstrap re-sampling. Unclassified sequences were tested for recombination using the program RDP3 [19].

HIV-1 clustering

To evaluate the presence of networks involving resistance transmission, additional approximate maximum likelihood and Bayesian phylogenetic trees (see Supplementary material, Figures S1 and S2, respectively) were constructed. These trees included all the pol sequences available at GenBank from El Salvador (n = 124) and representative sequences from neighbouring countries (Honduras, n = 28; Mexico, n = 5; Belize, n = 5; Panama, n = 4; Costa Rica, n = 1). The nucleotide substitution model GTR+Γ was used in the phylogenetic reconstruction for both methods, according to the dataset analysis using jModeltest (http://darwin.uvigo.es/). The maximum likelihood tree was obtained with FastTree v2.1.3 (http://www.microbesonline.org/fasttree) and using Shimodaira–Hasegawa-like local branch support to test its robustness. The Bayesian tree was estimated using Bayesian Markov chain Monte Carlo inference with the program BEAST v1.6.1 (http://beast.bio.ed.ac.uk). The Markov chain Monte Carlo chain ran for 4 × 107 generations, sampling estimates every 10 000th generation. The tree samples were used to generate a maximum clade credibility tree, after a 25% burn-in, in which the clusters' robustness was determined by the posterior probability.

Results

Study population

HIV-1 pol sequences could be obtained for 88 patients. They were deposited in GenBank with numbers JX466948JX467035. In 17 cases, only the RT coding region (775 nucleotides) was amplified. Samples were taken in the following hospitals: Santa Ana (11 patients), Ahuachapán (nine patients), Saldaña (nine patients), Zacamil (eight patients), General y de Psiquiatría (six patients), San Vicente (five patients), Sonsonate (five patients), San Bartolo (four patients), San Rafael (four patients), Maternidad (three patients), Usulután (three patients), Bloom (two patients), Chalchuapa (two patients), Cojutepeque (two patients), Sensuntepeque (two patients), Zacatecoluca (two patients), Chalatenango (one patient) and San Miguel (one patient). For nine patients, the corresponding data were not available.

Table 1 shows the patients' main demographic and clinical characteristics at sampling time of these 88 patients. Distribution according to gender was equal, with a mean age of 35 years. In fact, 40.9% of the patients were <30 years old. The main HIV transmission route was unprotected sexual practice. The probable place of infection was known for 69 subjects, being in most cases (95.6%) El Salvador. Almost all patients (96.6%) were diagnosed with HIV after 2006, and 58.7% were diagnosed in the year of sampling. The mean CD4+ count was 474 cells/mm3 and more than half of patients showed moderate (43.1%) or severe (15.5%) immune suppression (Table 1). In a quarter of patients (24.7%) the HIV-1 plasma viral load was >100 000 RNA copies/mL and in 9.1% it was <2000 copies/mL.

Table 1. Demographic and clinical characteristics of the human immunodeficiency virus (HIV) -infected study population at sampling time
Characteristic N % (95% CI)
Number of patients88
Gender
Male4150.6 (39.9–61.2)
Female4049.4 (38.8–60.0)
Unknown7
Mean age (years)35.5
Age group (years)
<2089.1 (4.7–16.9)
21–302831.8 (23.0–42.1)
31–402123.8 (16.2–33.7)
41–501921.6 (14.3–31.3)
51–601011.4 (6.3–19.7)
>6022.3 (0.6–7.9)
Route of HIV infection
Sexual8596.6 (90.4–98.8)
Vertical11.1 (0.2–6.2)
Others22.3 (0.6–7.9)
Date of HIV diagnosis
1997–200011.1 (0.2–6.2)
2001–200522.3 (0.6–8.0)
2006–20103337.9 (28.4–48.4)
20115158.6 (48.1–68.4)
Unknown1
CDC stage
A5568.7 (57.9–77.8)
B1113.7 (7.8–2.3)
C1417.5 (10.7–27.3)
Unknown8
CD4+ count interval (cells/mm3)
≥5002441.4 (29.6–54.2)
200–4992543.1 (31.2–55.9)
<200915.5 (8.4–26.9)
Unknown30
Viral load interval (copies HIV-RNA/mL)
<100033.9 (1.3–10.8)
1000–10 0001519.5 (12.2–29.7)
10 000–100 0004051.9 (41.0–62.7)
>100 0001924.7 (16.4–35.3)
Unknown11

Transmitted HIV drug resistance

Five out of the 88 antiretroviral-naive patients (5.7%) were infected by HIV-1 variants carrying any drug-resistance mutation, with resistance to RT in four cases and to protease inhibitor (PI) in one patient. The five patients (four women and one man, with ages between 25 and 44 years old) acquired the HIV-1-infection by a sexual route and were infected by subtype B. They had viral loads between 19 900 and 435 000 HIV-RNA copies/mL at time of sampling. One had an advanced disease stage (74 CD4+ cells/mm3), and the remaining four patients presented between 231 and 669 CD4+ cells/mm3. Four were diagnosed in 2011 and one in 2010 in five different Salvadorean clinical centres.

Regarding the drug class (Table 2), two patients (2.3%) were infected with virus carrying NRTI-resistance mutations (M184I in both cases). Two other patients (2.3%) presented NNRTI-resistance mutations (K101E and K103N, respectively). Finally, one patient 1.4% was infected with a virus carrying the PI-resistance mutation L23I. All patients showed only one TDR mutation, therefore affecting only one drug class in all cases. Notably, neither thymidine-analogue mutations nor K65R, NRTI-resistance substitutions, were found in this study.

Table 2. Rates of transmitted HIV drug resistance in the study population according to drug class
Drug class (available sequences)TDR rate (95% CI)TDR mutations (cases)a
  1. TDR, transmitted drug resistance; NRTI, nucleoside reverse transcriptase inhibitors; NNRTI, non-nucleoside reverse transcriptase inhibitors; PI, protease inhibitors.

  2. a

    According to the mutation list described in Bennett et al. [16].

NRTI (88)2.3% (0.6–7.9)M184I (2)
NNRTI (88)2.3% (0.6–7.9)K101E (1), K103N (1)
PI (71)1.4% (0.2–7.6)L23I (1)
Any (88)5.7% (2.4–12.6) 

Other mutations at residues associated to drug resistance

Besides the substitutions mentioned before, additional drug-resistance mutations included in the 2011 IAS-USA list were detected, as well as minor resistance mutations to PI (Table 3). In addition, five patients were infected with viruses carrying substitutions not included in any mutation list but located in positions related to NNRTI-resistance: four samples included the K103R and one the K101Q.

Table 3. Mutations associated with antiretroviral drug resistance found in 31 isolates of the 88 human immunodeficiency virus (HIV) -infected patients attended in El Salvador
PatientWHO 2009aIAS-USA 2011bOther changesc
TDR to PITDR to NRTITDR to NNRTIDRM to PIDRM to NRTIDRM to NNRTIAt PRAt RT
  1. TDR, transmitted drug resistance; PI, protease inhibitors; NRTI, nucleoside reverse transcriptase inhibitors; NNRTI, non-nucleoside reverse transcriptase inhibitors; DRM, drug-resistance mutations; PR, protease; RT, reverse transcriptase; –, no DRM was found; n.a., viral regions that could not be amplified.

  2. a

    Mutation list described in Bennett et al. [16].

  3. b

    Mutation list described in Johnson et al. [17].

  4. c

    At PR, minor resistance mutations to PI according to Johnson et al. [17]. At RT, changes in positions associated with NNRTI resistance.

SAL40K103NK103N
SAL86K101EK101E, E138KA71V
SAL94n.a.M184In.a.M184I
SAL107n.a.M184In.a.M184I
SAL138L23I
SAL87V90I
SAL146V90IL10I
SAL153V90IL10I
SAL108V106IK103R
SAL109V106IK103R
SAL134V106I
SAL136V106I
SAL145V106IA71V
SAL42n.a.n.a.V179Dn.a.
SAL118V179DA71T
SAL135V179D
SAL166n.a.n.a.n.a.K103R
SAL174K103R
SAL181K101Q
SAL18A71T
SAL85A71T
SAL91I50F, I54F
SAL95L10I
SAL112L10I, I54F
SAL114A71T
SAL142L10V
SAL144A71T
SAL149L10V
SAL150L10F
SAL165L10I
SAL172L10V

HIV-1 variants' distribution

Eighty-seven of the 88 patients (98.9%) were infected by HIV-1 subtype B viruses. Only one (SAL111) was infected with an HIV-1 non-B variant, close to the recombinant CRF20_BG described in Cuba [20] (Fig. 1). The recombination analysis showed that SAL111 was a unique recombinant between subtypes B and G (URF_BG). In addition, the maximum likelihood tree (Fig. 1) showed four robust phylogenetic clusters (more than two sequences) and two sequence pairs of subtype B with 100% statistical support. These transmission networks were confirmed in the subsequent maximum likelihood and Bayesian trees that included all the publicly available sequences.

Figure 1.

Maximum likelihood tree of the 88 human immunodeficiency virus type 1 (HIV-1) pol sequences from naive population in El Salvador. Tree constructed under the GTR+Γ model of nucleotide substitution using FastTree v2.1.3 (see Materials and methods) from a 1300-nucleotide alignment. Sequence labels from this study are coloured in grey, other sequences are in black and labelled indicating subtype, accession number and sampling country. The arrow points to the unique recombinant form specimen (SAL111), whose recombinant structure is shown in the diagram. The asterisks denote nodes with 100% statistical support.

The four clusters included 30 patients: ten in clusters I and II, and five in clusters III and IV, as labelled in Fig. 1. All involved both men and women (in total, 18 men and 12 women). All but one patient included in the clusters were infected with HIV through a sexual route. None of the clusters involved transmission of drug-resistance mutations.

Discussion

Monitoring TDR in low-income and middle-income countries where ART programmes are being implemented is essential to prevent the poorer virological outcomes associated with TDR [3]. Rates of TDR in these regions tend to increase following the scaling up of ART [4-6], so periodical molecular surveillance studies of resistance mutations among the ART-naive population are needed. Particularly, there is an urgent need for TDR prevalence data in the Central and South American regions. Here, we report the first study of HIV-transmitted resistance supported by the National STD-HIV/AIDS Programme of the Health Ministry of El Salvador (Central America) in the general, adult population.

Transmitted drug resistance in El Salvador

Rates of TDR from 10% to 17% have been reported in naive adult subjects in Western Europe and the USA [6] but few TDR studies have been conducted in other regions such as Latin America. Table 4 reviews HIV drug resistance studies in Central America and Mexico. In this region, the TDR rates among adults ranged from 0% in Panama [21] to 12.5% in certain areas of Mexico [22]. The variability across these values can be the result of the restricted size of some study populations, the target collective and the use of different mutations lists (and/or different versions of these lists) when calculating the TDR prevalence. For instance, TDR rate was higher (above 12% in all cases) in studies focused on the paediatric HIV-infected population [7, 11, 23], highlighting the need for more studies assessing TDR in this population and even of the revision of both the first-line regimen aimed at paediatric patients as well as the antiretroviral prophylaxis used to prevent mother-to-child transmission.

Table 4. Transmitted drug resistance studies in naive subjects from Central America and Mexico
StudyCountryHIV prevalenceStudy periodPopulationSpecimen N TDR (%)Resistance mutations (cases)HIV-1 variants (%)
AnyPINRTINNRTIPINRTINNRTI 
  1. HIV, human immunodeficiency virus; N, number of naive patients; TDR, transmitted drug resistance; PI, protease inhibitors; NRTI, nucleoside reverse transcriptase inhibitors; NNRTI, non-nucleoside reverse transcriptase inhibitors; DBS, dried blood spots; FSW, female sex worker; MSM, men who have sex with men; n. d., not determined; CRF, circulating recombinant forms; URF, unique recombinant forms.

  2. a

    HIV prevalences in high-risk collectives in El Salvador according to a report from the Health Ministry [25].

  3. b

    Mutation rates were re-calculated to adapt them to the WHO mutation list [16] from the published data.

  4. c

    Only the most frequently found TDR mutations are shown.

Present studyEl Salvador0.8%2011AdultDBS885.71.42.32.3L23I (1)M184I (2)

K101E (1)

K103N (1)

B (98.9)

BG (1.1)

Murillo et al. [12]El Salvador5%a2008FSWPlasma3910.302.67.7NoneL210W (1)

K103N (2)

G190S (1)

B (100)
Murillo et al. [12]El Salvador10%a2008MSMPlasma789.01.35.15.1

M46I (1)

I54V (1)

L76V (1)

V82T (1)

I84V (1)

M184V (3)

D67GN (2)

T215Y (2)

M41L (1)

L210W (1)

K219N (1)

K103N (4)

L100I (1)

P225H (1)

B (100)
Holguín et al. [11]El Salvador0.8%2009PaediatricDBS1127.3b18.2b10.0b20.0b

M46I (2)

I54V (2)

V82A (2)

L24I (1)

F53L (1)

L76V (1)

L90M (1)

D67G (1)

K70R (1)

K103S (1)

Y181C (1)

G190A (1)

B (100)
Holguín et al. [11]Honduras0.8%2009PaediatricDBS616.6b16.6b0b0bM46I (1)NoneNoneB (100)
Parham et al. [23]Honduras0.8%2001–04PaediatricDBS5512.73.63.65.4

D30N (1)

M46I (1)

D67N (1)

K70R (1)

M184I (1)

Y181C (2)

K101E (1)

B (100)
Murillo et al. [10]Honduras0.8%2004–07AdultPlasma2007.00.53.05.0I85V (1)

K219E (2)

M184V (2)

K70E (1)

T215S (1)

K103N (5)

Y181C (3)

Y188C (2)

K101E (2)

B (99)

URF (1)

Lloyd et al. [9]Honduras0.8%2002–03AdultPlasma3369.5b2.7b7.7b6.8b

N88D (3)

L90M (3)

M46I (3)c

M184V (20)

T215FY (10)

D67NG (10)c

K103N (10)

Y181CI (6)

G190AES (6)c

B (99.1)

F1 (0.3)

C (0.3)

AD (0.3)

Castillo et al. [7]Panama0.9%2008–10AdultPlasma4712.808.54.2None

M41L (4)

T215FL (1)

K103N (2)

P225H (2)

Y181C (1)

n. d.
Castillo et al. [7]Panama0.9%2007–09PaediatricPlasma2512.008.04.0None

T215Y (1)

K219Q (1)

K103N (1)n. d.
Ahumada-Ruiz et al. [21]Panama0.9%2004–05AdultPlasma530000NoneNoneNoneB (100)
Ávila-Ríos et al. [24]Guatemala0.8%2010–11AdultPlasma1458.30.70.76.9L23I (1)K219Q (1)

K101E (5)

K103N (3)

E138K (3)

V106M (1)

Y118H (1)

B (96.6)

BF (2.8)

C (0.7)

Parham et al. [23]Belize2.3%2001–04PaediatricDBS1127.30027.3NoneNone

K101E (1)

K103N (1)

V106AV (1)

Y181C (1)

B (81.8)
Ávila-Ríos et al. [8]Mexico0.3%2005–10AdultPlasma16556.81.84.21.9M46IL (12) V82A (3)c

T215rev (36)

M41L (19)

K219QENR (11)c

K103NS (18)

G190SA (4)

K101E (4)c

B (99.9)

CRF12_BF (0.06)

CRF06_cpx(0.06)

Bertagnolio et al. [27]Mexico0.3%2004AdultPlasma476.406.40None

T215S (2)

L210W (1)

M184V (1)

T69ADNT (1)

NoneB (100)
Viani et al. [28]Mexico0.3%2003–05Pregnant women/adultsPlasma412.402.40NoneK219Q (1)Nonen. d.
Escoto-Delgadillo et al. [22]Mexico0.3%2002–03AdultPlasma9612.5b2.1b10.4b6.2b

D30N (1)

I54L (1)

I84V (1)

N88D (1)

L90M (1)

F77L (3)

M184V (2)

L201W (2)

T215Y (2)

K103N (3)

Y181C (3)

G190A (2)

n. d.

We found a TDR rate of 5.7% in the general HIV-infected population in El Salvador, which is included in the range found in Central America (see above) but is lower than in the neighbouring countries Honduras and Guatemala [9, 10, 24]. Recent studies in El Salvador reported a higher TDR prevalence (although the confidence intervals overlap) among female sex workers and men who have sex with men (10.3% and 9%, respectively) [12]. It could be explained by the higher HIV prevalence among both high-risk groups (5% and 10%, respectively) [25], than in the general, adult population (0.8%) [1]. In addition, these collectives might have more access to ART, or be more likely to be recent infections, factors associated with higher TDR levels; although in this case, Murillo et al. [12] reported recent infections in only 16% of patients. The age of infection was unknown in our cohort, although the high rate of immune depression suggests that patients were chronically infected, therefore the presence of TDR was possibly underestimated. Moreover, in Murillo et al. [12], only 16% of patients had a documented recent infection, so once more the two study populations are not so widely different. However, other reasons for the different findings cannot be ruled out, including the limited sample size in our study.

Clinical implications of TDR on first therapy

The prevalence of NRTI-resistance mutations was low (2.3%), and due to the presence of the M184I substitution in isolates from two patients. Nevertheless, treatment including lamivudine or embricitabine is not contraindicated in the presence of this mutation, because M184I increases the susceptibility to other NRTIs and decreases HIV-1 replication. In all the study population, zidovudine and stavudine would be active, and >97% would be susceptible to a first antiretroviral regimen including the remaining NRTI. Therefore, the use of this family as the backbone of the first regimen is still strongly indicated in El Salvador.

For NNRTIs, the same TDR rate was found (2.3%). Two isolates harboured the mutations K101E and K103N in the RT gene. Both can cause resistance to nevirapine and efavirenz, but their low presence supports their use as third drug in the first-line ART regimen, as recommended by the Health Ministry of El Salvador. However, we cannot exclude the presence of other NRTI or NNRTI-related mutations in sanctuary sites or not being detected by a population sequencing that does not reveal minority variants.

We also found NNRTI-resistance substitutions excluded from the WHO mutation list because they are common polymorphisms: V90I (in three patients), V106I (in five patients) and V179D (in three patients). However, they are included in the IAS-USA list because they have been associated with reduced NNRTI susceptibility [17]. In other patients we found substitutions (Table 3) not included in either list because, by themselves, they have no effect on NNRTI efficacy, but may contribute to reduced susceptibility in the presence of other NNRTI-resistance mutations: K101Q, V179D and K103R. Longitudinal studies to assess the virological response in these patients are needed to evaluate the effect of these polymorphisms in the treatment efficacy.

Therefore, despite the low prevalence of major NNRTI resistance mutations, the relatively high frequency of the mentioned polymorphisms in the resistance positions suggest not only that these patients whould be carefully monitored but also that efforts should be multiplied to increase the frequency of genotypic resistance testing before starting ART in the El Salvador population, particularly as low genetic barrier regimens (as NNRTI) are recommended as preferred options.

Finally, only one transmitted PI-resistance mutation (L23I) was found. It is a rare substitution not previously reported in the naive population in Central America (Table 4) that causes low-level nelfinavir resistance only when it occurs in combination with other protease mutations [26]. This makes PI a suitable ART class to use in first-line treatment in those cases with a contraindication to NNRTI, such as pregnancy, NNRTI intolerance or pharmacological intolerances.

Late HIV diagnosis in El Salvador

Although most patients were young (40% being <30 years old) and had been diagnosed <5 years before sampling, 58.6% also showed moderate or severe immune depression (<500 CD4+ cells/mm3 in plasma), indicating a late HIV diagnosis in the country. Accordingly, a previous study reported that nearly 60% of naive HIV-infected children in El Salvador had CD4 cells counts <500 when they started ART [11]. More efforts are needed to promote early diagnosis and prevention of infection in El Salvador.

HIV variants in El Salvador

This work confirms that HIV-1 subtype B, as in the rest of Central America (Table 4), is the prevalent variant in El Salvador, accounting for almost 99% of cases. Only one infection with HIV-1 non-B variants was reported, caused by a unique recombinant URF_BG close to the BG recombinants first described in Cuba [20]. This patient was probably infected with HIV through a heterosexual relationship with a lorry driver.

Our study highlights that blood sampling on filter papers (dried blood spots) is an inexpensive and practical alternative to plasma for antiretroviral treatment monitoring in countries with limited resources, especially when samples need to be analysed in specialized laboratories abroad because of a lack of resources and/or material [14].

The sample size was limited by the high rate of failures when amplifying RNA from dried blood spots (31/119), especially for the protease coding region (48/119). Some of these dried blood spot samples were accidentally kept at room temperature instead of at −20°C as planned. Another limitation was the lack of more specific data about the infection route. In addition, although patients were chosen at random, the different HIV-1 prevalence rates in the areas where the clinics were located may affect the resistance results. Finally, the fact that most patients were chronically infected with HIV can make the comparison of TDR results with cohorts consisting of recently infected patients difficult, although it is a representative sample of the naive HIV-infected population living in El Salvador.

In conclusion, in this work we report a moderate TDR rate in El Salvador, Central America. Our results support the adequacy of the Salvadorean ART guidelines. In all cases the mutations affected only one antiretroviral class and we found no transmission network involving TDR. However, periodic surveillance studies are essential to monitor and prevent the emergence of HIV drug-resistance mutations (especially to NNRTI class) in low-income and middle-income regions.

Acknowledgements

We are grateful to the members of Proyecto Esther (‘Red de Solidaridad Hospitalaria contra el Sida’, Health Ministry of Spain) for their valuable support. This work was supported by grants from Ministerio de Salud Pública y Asistencia Social of El Salvador; Global Fund Project/MSPAS RCC: ‘Estrategia de lucha contra el VIH-sida en poblaciones vulnerables como coadyuvante a la reducción de pobreza en El Salvador 2009–2014′; and Fondo de Investigaciones Sanitarias (FIS 09/00284). A. H. is supported by Agencia Laín Entralgo. This study is included in the Subprograma de Inmigración y Salud, CIBERESP (Spain). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Transparency Declaration

The authors declare no conflicts of interest.

Ancillary