Dried plasma/blood spots for monitoring antiretroviral treatment efficacy and pharmacokinetics: a cross-sectional study in rural Burundi

Abstract

Aims

In limited resource settings monitoring antiretroviral (ARV) treatment efficacy is restrained by the lack of access to technological equipment. The aim of the study was to assess the use of dried plasma (DPS) and blood spots (DBS) to facilitate ARV monitoring in remote settings where clinical monitoring is the primary strategy.

Methods

A cross-sectional study in HIV-positive ARV-treated patients in Kiremba, Burundi was performed. DBS were used for HIV-1 viral load (limit of the assay 250 copies ml⁻¹) and genotypic drug resistance tests and dried plasma spots were used for concentration measurements.

Results

Three hundred and seven patients [201 female (88.6%), 14 children (4.5%)] were enrolled. HIV-1 viral load was <250, 250–1000 and >1000 copies ml⁻¹ in 250 (81.7%), 33 (10.8%) and 23 patients (7.5%). Eleven samples out of 23 were successfully amplified revealing nucleoside reverse transcriptase inhibitor (NRTI) and non-nucleoside reverse transcriptase inhibitor (NNRTI)-resistance associated mutations [in seven (58.3%) and six patients (50%)]. Nevirapine trough concentrations were <3000 ng ml⁻¹ in 28/189 patients (14.8%) and efavirenz 12 h concentrations were <1000 ng ml⁻¹ in 2/16 patients (12.5%). Children and patients with nevirapine exposure <3000 ng ml⁻¹ presented a higher risk of viral replication.

Conclusions

Viral loads <250 copies ml⁻¹ were observed in 81.7% of patients (83.6% adults and 42.9% children). Children and patients with low nevirapine concentrations had higher risk of viral replication. Dried blood and plasma spots may be useful for monitoring HIV-positive patients including viral load and drug level measurement as part of treatment management in remote areas.

What is Already Know about this Subject

• In resource limited countries it is hard to monitor efficacy of antiretroviral therapy due to the lack of facilities.

• Therapeutic Drug Monitoring (TDM) is useful to predict possible virological failures.

• Scarce data are available on the genotypic resistance test in Sub-Saharan Africa.

What this Study Adds

• Dried blood and plasma spots are useful tools to perform PK and virologic analyses in a rural setting.

• Nevirapine C_trough < 3000 ng ml⁻¹ was found to be associated with higher viral load.

• In 47.8% of samples with HIV RNA > 1000 copies ml⁻¹ we succeeded to perform genotypic resistance tests.

Introduction

By the end of 2012, 35 million people were living with HIV/AIDS, the majority of whom reside in Sub-Saharan Africa and more than 8 million people were receiving antiretroviral treatment (ART) in low and middle income countries 1. Several reports have documented beneficial effects of ART on HIV-associated morbidity and mortality in resource-limited settings 2, and ART effectively prevents HIV transmission from pregnant women to their children 3.

Burundi is an east-African country in which HIV prevalence in adults is estimated to be 4.1–8.8%, with great differences among rural, semi-urban and urban settings (respectively 2–4%, 8–9% and 6–7%). According to the last edition of national guidelines (October 2010) the first line treatment consists of two nucleoside reverse transcriptase inhibitors (NRTIs) plus a non-nucleoside reverse transcriptase inhibitor (NNRTI) 4. Limited data are available on the virological efficacy of ART regimens in Burundi even if evidence from neighbouring countries may probably be applied. A systematic review on virological follow-up in Sub-Saharan African countries showed that the proportions of patients with on-treatment virological success after 6–24 months of first line therapy were comparable with those from the developed world, but intention-to-treat data were less optimistic 5. A substudy of the DART trial in Uganda, investigated virological response in HIV patients randomized to receive three NRTIs vs. two NRTIs plus nevirapine (NVP). HIV RNA levels were lower in the NVP group than in the NRTI only group at 24 and 48 weeks 6.

The relationship between NVP plasma concentrations and virological response has been evaluated in several studies and a minimum trough concentration (C_trough) of 3000 ng ml⁻¹ has been proposed as a target to guarantee a considerable prevalence of virological success 7. Recent data suggest that a lower concentration may be needed in patients previously unexposed to ARV drugs (1000 ng ml⁻¹) 8. Moreover a study defined a range of NVP concentrations in which the selection of resistant mutants was the highest (mutant selection window, between 3000 ng ml⁻¹ and 4300 ng ml⁻¹) paralleling what is observed with some antibiotic drugs 9.

In resource limited settings monitoring for treatment failure is based on clinical and immunological criteria, due to a yet scarce availability of viral load measurements and the almost absence of genotypic resistance testing 10. Nevertheless this approach has poor sensitivity and specificity in detecting true virological failure until advanced immune deficiency 11,12. Several data suggest that this delay in identifying patients with virological failures may be associated with a high rate of selection of resistance associated mutations and of impairment in future treatment efficacy 13. International guidelines have therefore suggested the implementation of viral load monitoring as well as the introduction of higher genetic barrier drugs in first line regimens (such as tenofovir) 14.

Several logistic and economic barriers are challenging the opportunity to collect and analyze samples (plasma or serum) both for virological and pharmacological tests. Alternative specimen types are dried blood spots (DBS) or dried plasma spots (DPS). DBS have been extensively used in limited resource countries for storing specimens for virological purposes (for instance in the management of mother to child transmission analysis or for resistance testing). These devices can be stored and shipped at ambient temperature, thus avoiding the requirement for a freezer and carriage on dry ice. Lower biohazard risks associated with shipment and handling may be an advantage of such tools.

Our group previously conducted a study in rural Burundi using both DBS and DPS showing the feasibility of a pharmacokinetic and pharmacogenetic study 15,16. DPS performed quite well in this clinical scenario showing a good correlation with plasma samples revealing approximately 18% lower concentrations. This moderate underestimation was considered of marginal impact on clinical decisions since the misjudged samples (5/60) were near the clinical threshold (2447 to 2880 ng ml⁻¹). Undetectable and low concentrations (below 3000 ng ml⁻¹) were uncommon, being found, respectively, in 3.8% and 12.1% of patients. Nevertheless 50 patients (27.5%) were at risk of selecting resistance associated mutations showing NVP C_trough concentrations below 4300 ng ml⁻¹ 17.

The aim of this study was to evaluate the monitoring of antiretroviral (ARV) treatment using DPS and DBS in rural Burundi.

Methods

Study participants

The study was conducted as a cross-sectional analysis on unselected HIV positive patients attending the outpatient clinic at the Kiremba Hospital, Kiremba, in the north of Burundi between 2009 and 2011. To be considered for this study, candidates had to be HIV positive, in good general health, on highly active antiretroviral therapy (HAART) for at least 6 months and currently not on any other interacting drugs (such as rifampicin or fluconazole). Visual scale measured self-reported adherence had to be above 95% of doses.

The study was conducted according to the Declaration of Helsinki and the European Guidelines on Good Clinical Practice. The protocol was approved by the Local Review Board and written informed consent was obtained from all participants or their legal tutors.

Sampling and transport

Venous blood samples were obtained from all subjects in 7 ml lithium heparin tubes and whole blood (50 μl per spot) was spotted on Whatman 903 protein saver cards (VWR International, Milan, Italy) for virological analysis (DBS).

After centrifugation (3000 rev min⁻¹ at 4°C for 10 min) 100 μl plasma was spotted onto each glass filter (purchased from Laboratori Biomicron srl, Italy) and used for concentration measurement (DPS). DPS were allowed to dry at room temperature for 10 min and then heated at 58°C for 30 min (for HIV inactivation and complete drying).

DBS and DPS were separately inserted in a foil bag with desiccant (Foil Bag and Dessiccant Packs, purchased from Laboratori Biomicron srl, Italy) and then stored at room temperature. After a maximum of 30 days samples were delivered to the Laboratory of Pharmacokinetics and Pharmacogenetics of the University of Torino and to the Laboratory of Molecular Biology, Amedeo di Savoia Hospital, ASLTO2, Torino, Italy. They were stored at −20°C until analysis was performed (within 3 months from collection).

NVP and efavirenz concentrations on DPS were measured as previously described 15,16. Samples obtained 9 to 17 h after drug intake were considered trough concentrations (C_trough) for NVP and mid-dose concentrations (C₁₂) for efavirenz.

Nucleic acid isolation

Nucleic acid was extracted from two DBS for each patients (corresponding to 100 μl of blood) using the Boom's method with the semi-automatic extraction system NucliSENS® easyMAG® (Biomerieux, Fi, Italy). Blood was eluted from the filter papers (two spots for each patient) with 2 ml of Nuclisens Lysis buffer for 2 h in a 15 ml vial with gentle rotation. Filter papers were then removed and the solution containing the eluted blood was directly transferred into one of the vessels of the NucliSENS® easyMAG® extraction disposable devices and further processed according to the manufacturer's instructions. Total nucleic acid was reverse transcribed and amplified using GeneAmp Gold RNA PCR Core Kit (Life Technologies Italia, Milan, Italy).

HIV RNA quantification

Viral load measurement was performed with a NASBA™-based real-time amplification, the NucliSENS EasyQ® HIV-1 v2.0, according to the manufacturer's instructions. The linear dynamic range on DBS, is between 250 and 7.1 10⁷ copies ml⁻¹.

HIV genotypic drug resistance on DBSs

A HIV genotypic drug resistance test was performed only in patients with viral load >1000 copies ml⁻¹ with two separate nested RT-PCR amplifying fragments of the pol gene encompassing the protease (PR), nucleotide positions 2137–2650 (reference strain HXB2) and the reverse transcriptase (RT), nucleotide positions 2531–3334 (reference strain HXB2). Internal nested primers of both PR and RT were constructed with a 5′ tailed M13 universal sequence. Amplicons generated from separate nested RT-PCR were purified using QIAquick PCR Purification Kit (Qiagen, Hilden, Germany) and bidirectionally sequenced with a BigDye Terminator v1.1 reaction cycle sequencing kit on the ABI Prism 3130XL DNA Sequencer (Applied Biosystems), using primer M13-F 5′-TGTAAAACGACGGCCAGTT-3′ and primer M13-R 5′-CAGGAAACAGCTATGACC-3′. Sequence fragments were edited and aligned with the SeqScape® Software v2.7 (Life Technologies Italia, MI, Italy). Consensus pol sequences were submitted to Stanford University HIV Drug Resistance HIVdb program version 6.2.0 (http://hivdb.stanford.edu) for genotypic resistance interpretation according to the IAS USA Resistance Testing Guidelines (https://www.iasusa.org/).

Statistical analysis

Normally distributed variables are described as means (± SD) and analyzed with parametric testing while not normally distributed ones are described as medians (interquartile ranges) and analyzed with non-parametric tests. Categorical variables are described as number (percentage). Spearman's correlation was used to test the association between NVP C_trough and viral load measurements. Analysis of variance was used to test for differences in viral load categories and chi-squared was used to compare viral load suppression and categorical variables. A binomial logistic regression was performed to test factors associated with viral load <250 copies ml⁻¹ including variables with bivariate P values below 0.10. STATA version 11.2 for Macintosh (STATA Corp College Station, TX, USA) and SPSS version 18.0 for Macintosh (SPSS Inc., Chicago, IL, USA) were used for statistical analysis.

Results

Baseline and immunovirological characteristics

A total of 307 patients were included. Fourteen were children aged from 4 to 14 years old. Baseline and immunovirological characteristics are reported in Table 1. HIV-1 viral load was <250 copies ml⁻¹, 250–1000 and >1000 copies ml⁻¹ in 250 (81.7%), 33 (10.8%) and 23 patients (7.5%). In one patient the amplification process was not successful and the viral load was undetermined.

Table 1.

Demographic and immunovirological characteristics of study population

		Adults (n = 293)	Children (n = 14)
Age (years)	Median (IQR)	38 (32.5–46)	11 (10–13)
Gender (female)	n (%)	201 (68.6%)	6 (42.8%)
BMI (kg m−2)	Mean (±DS)	20.8 (±2.53)	15.7 (±2.71)
Clinical stage (WHO stage 3/4)	n (%)	229 (78.1%)	10 (71.4%)
CD4 + T lymphocytes (cell ul−1)	Median (IQR)	435 (260–707)	528.5 (357–849.5)
ARV backbone
Zidovudine/lamivudine	n (%)	16 (5.5%)	4 (28.6%)
Stavudine/lamivudine	n (%)	276 (95.2%)	10 (78.5%)
ARV third drug
Nevirapine	n (%)	264 (90.1%)	13 (92.8%)
Efavirenz	n (%)	23 (7.8%)	1 (7.1%)
Lopinavir/ritonavir	n (%)	5 (1.7%)	0 (0%)
Duration of antiretroviral treatment (months)	Median (IQR)	25 (17–39)	23.5 (15–29)
On cotrimoxazole	n (%)	73 (24.9%)	3 (21.4%)

In one patient viral load determination was not possible.

Considering the adult patient HIV viral load was below the limit of detection in 244 subjects (83.6%), between 250 copies ml⁻¹ and 1000 copies ml⁻¹ in 30 patients (10.3%) and above 1000 copies ml⁻¹ in 18 patients (6.1%). In detectable samples median HIV RNA was 950 copies ml⁻¹ (410–1735) and 2.98 Log₁₀ copies ml⁻¹ (2.61–3.24). In children HIV viral load was below the limit of detection in six subjects (42.9%), between 250 copies ml⁻¹ and 1000 copies ml⁻¹ in three patients (21.4%) and above 1000 copies ml⁻¹ in five patients (35.7%). In detectable samples median HIV RNA was 3900 copies ml⁻¹ (705–49250) and 3.52 log₁₀ copies ml⁻¹ (2.85–4.63).

Genotypic resistance tests

We succeeded in genotyping 11 out of 23 samples in patients with HIV RNA >1000 copies ml⁻¹ (47.8%). Patient and virus characteristics are reported in Table 2. A high prevalence of non-B subtypes (9/11 C subtype, 2/11 A subtype) was found. Mutations associated with resistance to NRTIs (mostly M184V and Type 2 timydine analogue mutations) and to NNRTIs (K103N, G190S/A, Y188L) were selected. Four sequences presented no major mutation. Two NVP recipients showed trough concentrations of 8687 ng ml⁻¹ and 1602 ng ml⁻¹ (patients 2 and 3, Table 2, respectively) while in two other subjects (one on efavirenz and one on lopinavir) plasma measurements were not available.

Table 2.

Patients' characteristics in those with successful genotyping

								Mutations
Patientt	Age (years)	Gender	Viral load	CD4	ARV regimen	ARV duration (months)	St	RT	PR
1	11	F	11000	717	EFV+AZT+3TC	39	A	M184V, K101A/E, G190S	WT
2	45	M	4000	456	NVP+d4T+3TC	47	A	WT	L10I
3	34	F	5200	338	NVP+d4T+3TC	69	C	WT	WT
4	37	F	260000	449	NVP+d4T+3TC	42	C	T96A, K70K/R, M184V, Y188L	WT
5	11	F	6000	982	NVP+d4T+3TC	41	C	K70KN	WT
6	36	F	130000	315	NVP+d4T+3TC	39	C	M184V, K103N, P225H	L10I/L
7	40	F	20000	NA	NVP+d4T+3TC	26	C	M184V, L210LW, T215Y, A98G, Y181V	WT
8	28	M	760000	NA	EFV+AZT+3TC	55	C	WT	WT
9	54	M	21000	266	LPV/r+TDF+ABC	29	C	WT	L10V
10	36	F	36000	NA	LPV/r+TDF+ABC	20	C	M41L, T69AT, M184V, T215SY, V179DV, Y188L	WT
11	NA	F	5300	183	NVP+d4T+3TC	56	C	M184V, K103S, G190A	WT

ARV antiretroviral; St subtype; RT retro-transcriptase; PR protease; NA not available; WT wild type; NVP nevirapine; EFV efavirenz; LPV/r lopinavir/ritonavir; d4T stavudine; 3TC lamivudine; AZT zidovudine; ABC abacavir; TDF tenofovir disoproxil fumarate.

Antiretroviral pharmacokinetics

NVP plasma concentrations were available in 269 patients (256 adults and 13 children) with samples collected 0.5 to 18 h after drug intake. Plasma concentrations ranged from below the limit of detection (n = 3) to 19532 ng ml⁻¹ (Figure 1A). Samples obtained 9 to 17 h after drug intake were considered trough concentrations (C_trough). Median trough samples were 5793 ng ml⁻¹ (3949–7653) in children (n = 11) and 6068 ng ml⁻¹ (4310–8342) in adults (n = 179). Trough concentrations were below 3000 ng ml⁻¹, between 3000 and 4300 ng ml⁻¹ or above 4300 ng ml⁻¹, respectively, in two (18.2%), one (9.1%), eight (72.7%) children and in 26 (14.6%), 18 (10.1%), 134 (75.3%) adults (Figure 1B). Patients with higher body mass indexes showed lower NVP concentrations (rho = −0.13, P = 0.033).

Figure 1.

Nevirapine DPS concentrations. In the left graph (A) nevirapine concentrations according to time post-dose. In the right graph (B) nevirapine trough concentrations in children and adults; central line represents median values while box and whiskers represent interquartile range and 10–90 percentile, respectively. Dotted lines represent 3000 ng ml⁻¹ and 4300 ng ml⁻¹. , children; , adults

Efavirenz plasma concentration were available in 21 patients (20 adults and one child) with samples collected 2 to 18 h after drug intake. Plasma concentrations ranged from below the limit of detection (n = 1) to 21327 ng ml⁻¹ (Figure 2A). The only paediatric patient (11 years old, BMI of 14.4 kg m⁻²) showed a 6 h post-dose concentration of 125 ng ml⁻¹. Samples obtained 9 to 17 h after drug intake were considered mid-dose concentrations (C₁₂) for efavirenz. Median mid-dose concentration was 2593 ng ml⁻¹ (1193–3765) in 18 adult patients. Mid-dose levels were below 1000 ng ml⁻¹, between 1000 and 4000 ng ml⁻¹ or above 4000 ng ml⁻¹, respectively, in one (5.5%), 11 (61.1%) and four (22.2%) patients (Figure 2B).

Figure 2.

Efavirenz DPS concentrations. In the left graph (A) efavirenz concentrations according to time post-dose. In the right graph (B) nevirapine trough concentrations; central line represents median values while box and whiskers represent interquartile range and 10–90 percentile respectively. Dotted lines represent 1000 ng ml⁻¹ and 4000 ng ml⁻¹. , children; , adults

Lopinavir plasma concentrations were not measured (five adults) and 12 samples were not analyzed due to reported damage (open bag/incomplete spot on filter).

Determinants of virological outcome

Viral load was significantly associated with gender (lower in female, P = 0.028) and with NVP trough concentrations (lower in patients with higher NVP concentrations, rho = −0.26 with P < 0.001). In adults the gender was associated with plasma viral load >250 copies ml⁻¹ (13.1% female vs. 23.6% male, Chi-square P = 0.027), but not with viral load >1000 copies ml⁻¹. No effect was noted for age, body mass index, CD4 cell count, WHO stage, ARV use (NVP vs. efavirenz, stavudine vs. zidovudine) or duration of HAART. In children a high proportion of viral load >1000 copies ml⁻¹ and >250 copies ml⁻¹ was observed (35.7% and 57.1%, respectively).

In NVP recipients the risk of viral load above 1000 copies ml⁻¹ in patients with available trough determinations (n = 190) was significantly higher in children (Chi² = 16.1, P = 0.002, OR 9.09, 95% CI 2.57, 32.14) and in patients with nevirapine trough concentrations below 3000 ng ml⁻¹ (Chi² = 5.792, P = 0.028, OR 3.56, 95% CI 1.20, 10.55) (Figure 3). No other significant association was noted. Similar results were noted using the 250 copies ml⁻¹ threshold: higher risk of HIV RNA above 250 copies ml⁻¹ in children (Chi² = 11.5, P = 0.003, OR 6.39, 95% CI 1.92, 21.3) and in patients with NVP trough concentrations below 3000 ng ml⁻¹ (Chi² = 5.41, P = 0.027, OR 2.67, 95% CI 1.14, 6.23).

Figure 3.

Prevalence of viral load above 1000 copies ml⁻¹ according to nevirapine trough concentrations below or above 3000 ng ml⁻¹. □, HIV RNA < 1000 copies ml⁻¹; ▪, HIV RNA > 1000 copies ml⁻¹

Considering the different definitions of virological or immunological failure we studied the relationship between CD4 count, viral load and time on treatment. A significant correlation was noted. With increasing duration of ARV treatment higher CD4 cell counts were noted in patients with viral load below 1000 copies ml⁻¹ (rho = 0.369, P < 0.001) but not in patients with viral load above 1000 copies ml⁻¹ (rho = 0.09, P = 0.68). Similar results were observed using the 250 copies ml⁻¹ threshold: CD4 cell count correlated with duration of ARV treatment in patients with HIV RNA below 250 copies ml⁻¹ (rho = 0.392, P < 0.001) but not in failing patients (rho = 0.082, P = 0.591).

Discussion

In this study we used DBS and DPS to assess HIV viral loads, resistance associated mutations and ARV drug concentrations in a rural setting in Burundi. We found that most of the patients had viral load <1000 copies ml⁻¹ (92.5%). In 11 failing patients in whom genotype was successful we found NRTI and NNRTI resistance associated mutations in seven samples. Low plasma concentrations were recorded in 14.7% of patients receiving NVP and in 12.5% of patients receiving efavirenz.

According to WHO HIV guidelines 14, viral load determination is the preferred monitoring approach to diagnose and confirm ARV treatment failure. Virological failure is defined as ‘Plasma viral load above 1000 copies ml⁻¹ based on two consecutive viral load measurements after 3 months, with adherence support’. This approach, strongly recommended, is supported by low quality evidence, due to the lack of well-designed trials on virological failures in developing countries. Our analysis confirms previous findings reporting a high efficacy of ARV regimens (with 81.7% and 10.8% of patients, respectively, presenting suppressed or low level viraemias) after a median follow up of 2 years 6. The cohort here presented is mainly composed of young female HIV-positive patients treated with stavudine/lamivudine plus NVP in a rural setting. These data are supporting the efforts made by the Burundese Government and by some non-governmental organizations in implementing the availability of ARV treatment as well as trained workers. We found a relationship between NVP exposure and the virological outcome and we confirmed previously proposed cut offs in the Caucasian population 7. However a high proportion of patients presented high NVP concentrations (75.3% of adults and 72.3% of children with trough values above 4300 ng ml⁻¹) thus overcoming the mutant selection window. Previously observed pharmacogenetic variants can explain such observations 15. Nevertheless children had the worst virological outcome with 57.1% of patients presenting detectable viral loads suggesting that even adequate NVP concentrations are not sufficient in patients with previous long term treatments and adherence issues.

The significant consequences of virological failures in low income countries may suggest that the optimization of first line regimens could be a cost effective approach. Unfortunately routine viral load monitoring and therapeutic drug monitoring (TDM) are widely unavailable in such setting due to money expenditures and logistic difficulties. These data may be useful in the ongoing discussion on whether to apply Department of Health and Human Services guidelines to limited resource countries 18. NVP and efavirenz exposure was successfully assessed in DPS. Efavirenz concentrations on DPS are reported here for the first time. Only two patients (one adult and one child) presented suboptimal mid-dose levels (below 1000 ng ml⁻¹) supporting adherence interventions or dosage increases. In agreement with these findings a Tanzanian study on 152 HIV patients showed sub-therapeutic NVP and efavirenz plasma concentrations in 28.3% of patients with immunological failure. The proportion increased with poor adherence, NRTI backbone comprising stavudine and lamivudine, increasing viral loads and advanced HIV stage 19. Even if we advocate the implementation of local nation-based laboratories able to perform such analysis we recognize that dried spot devices may be useful in collecting samples from remote and rural areas.

The observed correlation between immunological recovery over time and viral replication is of particular interest as it may suggest that immunological (and clinical) events occur at a significant time-lag after viral replication. In patients with a viral load above 1000 ng ml⁻¹ (and even 250 copies ml⁻¹) no correlation was found between duration of treatment and CD4 cell count thus confirming that CD4 decline follows this event 20. Late failure diagnosis and unnecessary therapy switches are of concern since few data are available on the amount of primary and acquired resistance associated mutations 21,22. We successfully genotyped 11 out of 23 samples with high viral loads. Several thymidine analogue mutations and NNRTI-resistance associated mutations were found in the retro-transcriptase while only minor mutations in the protease gene in this group of patients harbouring C and A HIV-1 subtypes. Four out of 11 patients showed no major resistance associated mutations supporting incomplete adherence in two of those (low NVP concentration in one and very high viral load in another one). More than half of the sequences were not adequate due to RNA fragmentation. Our samples were stored at room temperature as required for HIV RNA quantification with NucliSENS EasyQ, but this approach may explain the low recovery of good quality sequences for HIV drug resistance determination especially when long nucleotide segments are required. A handbook for handling DBS for HIV drug resistance genotyping determination was published by WHO in March 2010, that is after our study was conducted, stating that only short term storage (≤14 days) at room temperature is acceptable and for longer periods of time DBS must be frozen at −20°C or below (up to 2 years) 23.

Apart from the aforementioned technical issue some other limitations should be stated: the cross-sectional design (harbouring channelling biases), the limited sample size of subgroups (efavirenz and lopinavir-receiving patients as well as children) and the limited availability of laboratory tests (including CD4 cell count).

In conclusion our study suggests the feasibility of monitoring HIV viral load and genotype resistance test using DBS in rural settings and the need to implement such tests in order to limit the emergence of resistance associated mutations. The pharmacokinetic/pharmacodynamic relationship observed in NVP intakers supports the optimization of ARV regimens and the need for studies investigating the cost effectiveness of TDM in limited resource countries.

Competing Interests

All authors have completed the Unified Competing Interest form at http://www.icmje.org/coi_disclosure.pdf (available on request from the corresponding author) and declare AC, SB, GDP have received speakers' honoraria and research grants from Boehringer and Bristol Myers Squibb in the previous 3 years. There are no other relationships or activities that could appear to have influenced the submitted work.