Antiretroviral treatment outcome in HIV-1-infected patients routinely followed up in capital cities and remote areas of Senegal, Mali and Guinea-Conakry

Introduction Access to antiretroviral treatment (ART) becomes more and more effective in resource-limited settings (RLS). However, this global effort would be even more profitable if the access to laboratory services especially in decentralized settings was strengthened. We report the virological outcome and HIV-1 drug resistance in three West African countries using dried blood spots (DBS) samples. Methods We included HIV-1-infected adults on ART ≥6 months and followed up in capital cities and decentralized sites in Senegal, Mali and Guinea-Conakry. Patients were consecutively enrolled and DBS were collected in field conditions and kept at ambient temperature before transfer to the reference laboratory. Viral load (VL) was quantified using the NucliSENS EasyQ HIV-1 v1.2. Genotyping of HIV-1 pol gene was performed using in-house protocol. Results Of the 407 participants, 119, 152 and 136 were from Senegal, Mali and Guinea-Conakry, respectively. The median treatment duration was 36 months [IQR: 6–136]. Virological failure (VF) (VL≥3log10 copies/mL) was observed in 26% (95% confidence interval (CI), 18–35; n=31), 11% (95% CI, 6–17; n=16) and 24% (95% CI, 17–32; n=33) of patients in Senegal, Mali and Guinea-Conakry, respectively (p=0.001). Of samples presenting VL≥3log10 copies/mL (n=80), 70 were successfully genotyped. At least one drug resistance mutation (DRM) was detected in the following proportions: 70% (95% CI, 50–86; n=19), 93% (95% CI, 68–100; n=14) and 68% (95% CI, 48–84; n=19) in Senegal, Mali and Guinea-Conakry, respectively (p=0.22). Twenty-six per cent (26%; 95% CI, 16–38; n=18) of patients in VF harboured wild-type viruses, which is likely indicative of weak adherence. Phylogenetic analysis showed the predominance of CRF02_AG subtype (73%; 95% CI, 61–83; n=51). Conclusions We describe the ART outcome in capital and rural settings of Senegal, Mali and Guinea-Conakry. Our results in all of the three countries highlight the need to reinforce the ART adherence in order to minimize the occurrence of drug resistance. In addition, these findings provide additional evidence that the use of DBS as a sampling support could assist virological monitoring of patients on ART in remote areas.


Introduction
In the latest UNAIDS report, the number of people on antiretroviral therapy (ART) has dramatically increased in the past few years [1]. However, ART expansion was not sufficiently accompanied by access to laboratory services and diagnostics, especially in remote areas of developing countries. It is also established that now decentralizing HIV care at the community level is an essential link for the retention of patients in the healthcare system [2,3], which, in addition, promotes good treatment adherence. Consequently, it determines successful long-term viral load (VL) suppression [4]. HIV VL and resistance testing are essentials for monitoring the response to treatment, diagnosing and confirming treatment failure (TF) and surveillance of drug resistance. However, in remote areas of developing countries lacking appropriate equipment for plasma processing and transportation, virological monitoring is still a challenge. Several studies have demonstrated that the filter paper is a suitable tool for blood sample collection, transportation and storage [5Á11] and may facilitate HIV virological monitoring. However, few studies have reported the use of dried blood spots (DBS) in routine field conditions to account with variable environmental conditions [11Á14]. The aim of this study was to document the virological outcome and HIV-1 drug resistance in adult patients followed in the capital cities and decentralized settings in three West African countries, namely Senegal, Mali and Guinea-Conakry, using DBS samples.

Study design and settings
We conducted a multi-site study on patients on ART in capital cities and decentralized medical centres of three West African countries (Senegal, Mali and Guinea-Conakry). We included patients who were at least 18 years old, under ART treatment for at least six months and consenting to participate in the study. Women who received the protocol for the prevention of mother-to-child transmission (PMTCT) as well as patients positive for HIV-2 or co-infected (HIV-1 and HIV-2) were not included. Patients were consecutively enrolled between February 2010 and December 2011 in 17 sites distributed as follows: Senegal (n 07), Mali (n 06) and Guinea-Conakry (n 04). In Mali and Guinea-Conakry, recruitment sites were chosen depending on the active file of patients. In Senegal, all samples received for routine virological monitoring, from patients who met the inclusion criteria were considered.

Ethical considerations
National Ethics Committees of participating countries approved this study. Patients were recruited on a voluntary basis. To keep confidential data of the participants, a unique identifier was assigned to each sample and used throughout the study, ascertaining anonymity. Only the attending physician could establish the correspondence between this identifier and the patient.

DBS preparation and shipping
For each patient, two DBS cards (Whatman 903 filter paper, Dassel, Germany) were spotted: 50 mL/spot, five spots/card from whole blood EDTA tube. Cards were left to dry overnight at ambient temperature (range 22Á378C) before being packed in an individual sealed bag with desiccants and a humidity indicator card. These individual bags were in turn packed in zip-lock plastic bag and sent at ambient temperature to the reference laboratory in Dakar, Senegal, for testing. On site, collected DBS specimens were checked frequently for humidity and sent within one month after sampling. At the reference laboratory, upon reception, the conformity (in respect of delivery times, presence of humidity indicator card, desiccant packets and integrity of blood spots [no moisture]) was checked and DBS were stored at (808C until testing. The samples were discarded if they did not meet the criteria mentioned above.

Laboratory procedures
Total HIV-1 nucleic acids were extracted from two spots with NucliSENS miniMAG (bioMérieux, Craponne, France) with magnetic silica as previously described [5]. Briefly, two spots from each sample were punched and placed in a tube containing 2 mL of lysis buffer. After 30 min of gentle rocking at room temperature, the supernatant was clarified by centrifugation at 2500 rpm during two minutes and then transferred to a clean 15 mL conical tube. Extracted nucleic acids were eluted in 25 mL of elution buffer and stored at 48C for immediate use (VL quantitation or PCR amplification) or at (808C for longer storage. HIV-1 VL was quantified using the NucliSENS EasyQ HIV-1 v1.2 (bioMérieux, Marcy l'Etoile, France) according to manufacturers' instructions. The VL cut-off was 800 copies/mL with this assay for DBS [15]. In the present study, we set the VF to 3log 10 (1000) copies/mL as recently recommended by WHO [4].
HIV-1 drug resistance test was performed according to the ANRS AC11 protocol (http://www.hivfrenchresistance.org/) by amplifying separately the entire Protease (PR) gene and first 240 codons of Reverse Transcriptase (RT) using, respectively, 5?Prot1/3?Prot1 and MJ3/MJ4 as outer primers and 5?Prot2/3?Prot2 and A35/NE35 as inner primers. Second round PCR products were purified with QIAquick Gel Extraction Kit † (Qiagen, Courtaboeuf, France) according to the manufacturers' instructions. Purified DNA was sequenced directly on ABI 3100 Avant Genetic Analyzer using Big Dye Terminator Technology † v3.1 (Applied Biosystems, Carlsbad, CA) and their respective inner primers. Sequences obtained were assembled and edited manually using SeqMan TM II 5.08 from DNAstar † software (Lasergene, Konstanz, Germany). Drug resistance analysis and interpretation were performed using the Stanford University HIV database version 6.0.8 (http://hivdb.stanford.edu/). HIV-1 subtypes were determined by phylogeny. Nucleotide sequences were aligned with a set of reference sequences of HIV-1 group M subtype and circulating recombinant forms (CRFs) downloaded from Los Alamos HIV database (http://www.hiv.lanl.gov/content/index). Each subtype was represented by at least three reference sequences. Sequences were aligned with MUSCLE (and gap positions removed by using Gblocks program on SEAVIEW v4.4.1). Maximum Likelihood phylogeny was inferred online using the PhyML software (http://www.atgc-montpellier.fr/phyml) with branch supports determined by the approximate likelihood ratio test method (aLRT) SH-like option, and the substitution model was GTR'I'G. The recombinant strains analysis (similarity and bootscanning) were performed on Simplot software v3.5.1 [16].
The new PR and RT generated sequences were deposited in EMBL with the following accession Numbers: HG380024 to HG380051, HG380054 to HG380063, HG380065 to HG380069, HG424394 to HG424413, HG424415, HG424417, HG424418 and HG424420 to HG424423.

Statistical analysis
Data were analyzed using Epi Info TM Version 3.5.3. Ninety-five per cent confidence intervals (CIs) were used for all estimates. The chi-square test was used with Yate's correction to search the link between the variables. Comparisons between median values of VL of amplified samples or not were performed using MannÁWhitney test. P values less than 0.05 were considered statistically significant.

Discussion
The aim of this work was to document the virological outcome and HIV-1 drug resistance in adult patients followed in capital cities and decentralized settings in Senegal, Mali and Guinea-Conakry using DBS samples. Most of the patients  in this study were on first-line ART regimen (93%; 95% CI, 90Á95; n0379), with a wide predominance of AZT'3TC' NVP/EFV combination followed by d4T'3TC'NVP/EFV according to WHO 2006 recommendations for resource-limited settings (RLS) [17]. Similar observations were also reported in previous studies conducted in Senegal [18], Mali [19Á21] and other sub-Saharan African countries [12,22,23]. Drug substitutions consisted mainly of d4T phasing out, which was replaced by AZT (36%; 95% CI, 30Á42; n095). Furthermore, substitutions occurred between AZT and TDF (9%; 95% CI, 6Á13; n024) according to WHO guidelines in 2009 [24].
Despite the difference in VL thresholds (800 copies/mL for ''DBS'' vs. 50 copies/mL for ''Plasma''), the rate of viral suppression in our study (80%; 95% CI, 76Á84; n0327) is comparable to those reported in multicentre studies in Burkina Faso and Mali (77.2%, n0467/606) [20], (74.9%, n0598/798) [25], (p 00.13). However, our findings show differences between the VF (VL]3log 10 copies/mL) rate noted in Senegal (26%; 95% CI, 18Á35; n 031), Mali (11%; 95% CI, 6Á17; n016) and Guinea-Conakry (24%; 95% CI, 17Á32; n033), (p 00.01). This difference could be due partly to the size of the study populations between the three countries. In Senegal, all samples received routine virological monitoring and those who met the inclusion criteria were considered. Therefore, the number of patients failing treatment is not representative of the actual situation in the sub-population of persons on ART. Recently, a study outlining virological outcome in patients receiving HAART, and monitored with the World Health Organization (WHO) Public Health Approach, showed a VF of 10.7% among patients at 24 months' treatment duration and recruited consecutively [26], which is significantly lower than 26% observed in the present study. Furthermore, without comparing health systems in order to correlate it to the virological outcome and resistance viruses, the differences observed could find explanation partly in the level of health care services between settings as reported by Pere et al. [23].
In Senegal, the study showed a similar VF rate (26%; 95% CI, 18Á35; n 031) as previously reported by Diouara et al. (23.8%, n 055/231), (p00.74) [13]. The differences seem linked to the size of the study populations, which is more important in this study. Another possible reason is the improvement of health care services and the expansion of virological monitoring (e.g. VL test) to decentralized level [27].
For Guinea-Conakry, to our knowledge, this is the first data of VL and antiretroviral drug resistance in HIV-1-treated patients. So, our results cannot be compared to the national level. However, VF is comparable to what was previously obtained in the neighbouring countries, namely Senegal and Mali [13,19]. Moreover, our findings show that there were no significant differences in the occurrence of VF between patients followed in capital cities and those in regional centres.
To the best of our knowledge, this is the first work in West Africa using DBS samples in VL and drug resistance testing. Overall, we had a good rate of PCR amplification (88%; 95% CI, 78Á94; n 070), which is comparable to that obtained previously in Senegal [13] and those reported in other studies [6,11,14,28]. However, in Kenya, Arnedo et al. observed a lower PCR amplification rate (32.7%, n018/55) [12]. Also, low PCR amplification rate (46%) was reported in a study comparing drug resistance profiles obtained from paired plasma and DBS samples from HIV-1-treated children living in Bangui, Central African Republic [29]. In addition, several studies showed a weak successful PCR amplification rate if VL B3.7log 10 copies/mL [30]. This is in line with our results as it shows significant difference between medians VL of genotyped samples (n 070) and those not amplified (n 010), (p00.03). It has also been reported in several studies that many factors, mainly the conditions and storage time, transportation, temperature, humidity level and genotyping methods could influence the efficiency of DBS samples PCR amplification [7Á9,11Á14, 31,32]. In addition, based on literature data and on our experience in usage of DBS samples collected in real-life conditions for drug resistance tests, in-house methods by amplifying and sequencing the entire PR and RT (at least codon 41Á236) as described above seem to be most suitable in addition to the prerequisites for the conservation of sample integrity throughout the pre-analytical process [8,32]. In the present study, DBS transportation at an ambient temperature for a duration of 15Á30 days, even when dealing with international shipments, do not seem to have any impact on the performance of amplification success. . Their reports have also mentioned that the international shipment at an ambient temperature did not affect amplification success rates [11,33].
Of a total of 407 patients on ART according to WHO 2006 revised recommendations for RLS, 13% (95% CI, 10Á16; n052) had at least one DRM. Depending on the country, no significant difference was observed between the DRM rate: 70% (95% CI, 50Á86; n019) in Senegal, 93% (95% CI, 68Á100; n014) in Mali and 68% (95% CI, 48Á84; n019) in Guinea-Conakry, (p00.15). The DRM profiles in these three countries were similar and show that TAMs (p00.81), M184V/I (p00.37) and K103N (p 00.45) mutations were by far the most frequently encountered. Their strong representation seems to be associated with the wide use of AZT/D4T' 3TC'EFV/NVP in first-line protocol. Our results are in line with previous studies [22,23] reporting the accumulation of TAMs with treatment duration and frequency of the M184V and K103N mutations causing resistance to 3TC/FTC and NVP/EFV, respectively.
Overall, 26% (95% CI, 16Á38; n018) of genotyped samples were distributed as follows: Senegal (n 08), Mali (n 01) and Guinea-Conakry (n 09) had wild-type viruses ( Table 2). This event could probably be related to poor adherence or a VF due to minority variants that could not be detected by bulk sequencing, as performed in the present study. However, similar observations were reported in a recent prospective cohort study conducted in Abidjan, Côte d'Ivoire, where 25% of patients in VF at month 24 of treatment harboured wild-type viruses [22]. Pere et al. in a study conducted in Bangui, Central African Republic, also reported that 24% of patients in VF showed wild-type viruses [23]. It is known that high VL on treatment is associated with a risk of developing drug resistance. This study highlights the need to improve treatment adherence. In a context of universal access to ART, combined with unavailability of VL and HIV-1 resistance testing particularly in remote areas, it is crucial that national programme managers make advocacy for better psychosocial support allowing treatment adherence with focus on capacity building and staffing of health personnel, especially at a decentralized level.
Phylogenetic analysis shows the predominance of CRF02_ AG for each of the three participating countries. And, HIV-1 genotypes found in our study are concordant with those previously observed in Senegal [13], in Mali [19,25] and in Guinea-Conakry [34].
Our study has some limitations. First, to assess treatment adherence, a questionnaire was not provided to patients during the enrolment process. Second, the CD4 cell count at ART initiation was not available and sometime data relating to the treatment history (combination and substitution) too. Other limitations were the small number of patients recruited in some sites, the heterogeneous treatment duration of patients (ranging from 6 months to 11 years) and the recruitment process. For example, in Senegal, all samples received for routine virological monitoring and who met the inclusion criteria were considered. It was not the case in Mali and Guinea-Conakry, where the selection of recruitment sites was made based on the active file of patients. So, the results are not representative nationwide.

Conclusions
Our data show a relatively high rate of patients in VF while among them an important proportion harboured wild-type viruses. This highlights a real need to reinforce treatment adherence and expand VL and resistance testing in remote areas. Regarding virological monitoring of patients followed in remote areas, these results provide additional evidence that the use of DBS samples could facilitate access to viral load and resistance testing for better care of patients in a context of limited resources.