A Pragmatic Approach to HIV-1 Drug Resistance Determination in Resource-Limited Settings by Use of a Novel Genotyping Assay Targeting the Reverse Transcriptase-Encoding Region Only

Abstract

In resource-limited settings (RLS), reverse transcriptase (RT) inhibitors form the backbone of first-line treatment regimens. We have developed a simplified HIV-1 drug resistance genotyping assay targeting the region of RT harboring all major RT inhibitor resistance mutation positions, thus providing all relevant susceptibility data for first-line failures, coupled with minimal cost and labor. The assay comprises a one-step RT-PCR amplification reaction, followed by sequencing using one forward and one reverse primer, generating double-stranded coverage of RT amino acids (aa) 41 to 238. The assay was optimized for all major HIV-1 group M subtypes in plasma and dried blood spot (DBS) samples using a panel of reference viruses for HIV-1 subtypes A to D, F to H, and circulating recombinant form 01_AE (CRF01_AE) and applied to 212 clinical plasma samples and 25 DBS samples from HIV-1-infected individuals from Africa and Europe. The assay was subsequently transferred to Uganda and applied locally on clinical plasma samples. All major HIV-1 subtypes could be detected with an analytical sensitivity of 5.00E+3 RNA copies/ml for plasma and DBS. Application of the assay on 212 clinical samples from African subjects comprising subtypes A to D, F to H (rare), CRF01_AE, and CRF02_AG at a viral load (VL) range of 6.71E+2 to 1.00E+7 (median, 1.48E+5) RNA copies/ml was 94.8% (n = 201) successful. Application on clinical samples in Uganda demonstrated a comparable success rate. Genotyping of clinical DBS samples, all subtype C with a VL range of 1.02E+3 to 4.49E+5 (median, 1.42E+4) RNA copies/ml, was 84.0% successful. The described assay greatly reduces hands-on time and the costs required for genotyping and is ideal for use in RLS, as demonstrated in a reference laboratory in Uganda and its successful application on DBS samples.

INTRODUCTION

There are currently more than 3.9 million HIV-1-infected individuals receiving highly active antiretroviral treatment (HAART) in sub-Saharan Africa (1), with recent studies showing an estimated prevalence of 5.6% antiretroviral (ARV) drug resistance prior to the initiation of treatment, ranging from 1.1% in South Africa to 12.3% in Uganda (2). The use of HIV-1 drug resistance (HIVDR) genotyping in resource-limited settings (RLS) is limited to clinical research studies and for preauthorized private medical care. Limited routine resistance testing is performed due to the high cost, infrastructure requirements, and the complexity of available commercial assays. The use of HIV-1 drug resistance genotyping is vital for advising policy makers on the status of HIV-1 drug resistance profiles to ensure that optimal HAART options are maintained, but it should also be considered for individualized treatment management.

Treatment monitoring of individuals receiving therapy primarily includes clinical evaluation with or without CD4 testing. When available and affordable, it is recommended to perform viral load (VL) monitoring. Current guidelines do not recommend HIV-1 drug resistance genotyping be included as part of treatment management in RLS (3), primarily due to the cost and complexity of the assays and limited ARV drug options.

Current ARV regimens for the treatment of HIV-1 in adults and adolescents in RLS, as recommended by WHO guidelines, comprise a first-line regimen consisting of two nucleoside reverse transcriptase inhibitors (NRTIs) and one nonnucleoside reverse transcriptase inhibitor (NNRTI) (3). There is limited use of protease inhibitor (PI)-containing regimens, restricted to second-line therapy, and the rate of PI drug resistance is still low (4, 5). According to a systematic review by Barth et al. on the success of ARV treatment programs in sub-Saharan Africa, 94% of African adult patients on ART received a combination of NRTI and NNRTI first-line therapy (6). When therapy failure occurs, more than 83% of subjects failing first-line regimens harbor mutations that confer resistance to both NRTIs and NNRTIs. The most common mutation observed is the M184V, followed by several NNRTI mutations, such as K103N, Y181C, and V106M (7). Complex resistance to NRTIs such as K65R, thymidine analogue mutations (TAMs), and Q151M is also observed, with frequencies increasing when viral load monitoring is not observed (7, 8).

All major mutations that affect the efficacy of RTI therapy, as defined by the International AIDS Society (IAS) drug resistance mutation list, are located between reverse transcriptase (RT) amino acids (aa) 41 and 238 (9). Furthermore, in silico analysis of genotyping profiles of this specific RT region demonstrated that the predicted drug susceptibilities were as informative as sequences that more broadly cover the RT gene (aa 1 to 400) (10). The studies therefore indicate that for HIV-1 drug resistance genotyping for individuals experiencing therapy failure while receiving the current ARV regimens in RLS, an assay that targets the RT region only would be effective for monitoring.

As part of the Affordable Resistance Test for Africa (ART-A) initiative to develop affordable resistance testing for use in Africa (www.arta-africa.org), we have used this knowledge base to design an RT-specific assay for a simplified genotyping screen. For application in RLS, the potential use of the assay on dried blood spots (DBS) was considered an important addition. This choice, in combination with a smaller fragment for PCR amplification, has enabled the development of a test that fulfils the criteria with an analytical sensitivity of 5.00E+3 RNA copies/ml at a more affordable cost. We evaluated the assay on both plasma and DBS in a WHO reference laboratory in The Netherlands and subsequently evaluated its application in a reference laboratory in Uganda.

MATERIALS AND METHODS

Samples.

A selection of reference viruses for subtypes A, B, C, D, F, G, H, and circulating recombinant form 01_AE (CRF01_AE) (Table 1) from the BBI panel (BBI Biotech Research Laboratories Inc., Gaithersburg, MD) was used to evaluate the sensitivity and specificity of the assay. Virus stocks for each subtype were used to prepare dilution series of plasma and spiked whole blood for DBS preparation in concentrations of 1.00E+4, 5.00E+3, 1.00E+3, and 5.00E+2 RNA copies/ml, which were used to optimize the assay.

Table 1.

Summary of HIV-1 isolates in the subtype reference panel

Subtype	Strain	Country of origin	GenBank accession no.
A	UG275	Uganda	AB485632
B	BK132	Thailand	AY173951
C	ZB18	Zambia	AB485641
D	SE365	Senegal	AB485648
CRF01_AE	CM240	Thailand	AF067154
F	BZ126	Brazil	AY173957
G	BCF-DIOUM	Zaire	AB485661
H	BCP-KITA	Zaire	AB485665

The assay was subsequently evaluated on a selection of clinical samples from Africa and The Netherlands that were available at the University Medical Centre (UMC), Utrecht. A total of 212 samples with a VL range of 6.71E+2 to 1.00E+7 (median, 1.35E+4) RNA copies/ml, comprising subtypes A (n = 47; 22.1%), B (n = 20; 9.4%), C (n = 72; 33.9%), D (n = 18; 8.4%), CRF01_AE (n = 35; 16.5%), and CRF02_AG (n = 10; 4.7%) and rare subtypes H, G, and F (n = 10; 4.7%), were used. For the clinical samples, plasma was separated from EDTA tube blood and stored at −80°C prior to analysis.

The clinical samples available at the UMC Utrecht were included to assess the assay application for various African subtypes. For these samples, the VL was determined using the COBAS AmpliPrep/COBAS TaqMan System v2 (Roche, Penzberg, Germany). These samples were obtained from two separate sources, RNA isolates previously genotyped using the ViroSeq HIV-1 genotyping system (Celera Diagnostics, Alameda, CA, USA) or with a laboratory-developed genotyping assay (n = 88) (11) and plasma samples from a study to monitor for the development of drug resistance mutations in children failing therapy in Rwanda (n = 67), which were only genotyped with the described method. In addition, clinical samples from South Africa (n = 57) sent for routine VL testing were included. VLs were determined in South Africa using NucliSENS EasyQ HIV-1 version 1.2 (bioMérieux, Boxtel, The Netherlands). Plasma samples were stored at −80°C and shipped on dry ice to The Netherlands.

Application of the assay with clinical DBS samples was evaluated with a selection of 25 clinical DBS samples from South Africa sent for genotyping with a laboratory-developed genotyping assay at the UMC Utrecht. Selected samples had a VL range of 1.02E+3 to 4.49E+5 (median, 1.42E+4) RNA copies/ml, and all samples were HIV-1 subtype C.

Following development and evaluation in The Netherlands, the assay was transferred to, and applied in, a reference laboratory in Kampala, Uganda. A total of 132 pretreatment baseline plasma samples from the PASER-M cohort (12) were used for further evaluation. These samples were chosen to represent a distribution of subtypes A (n = 60; 45.4%) and D (n = 43; 32.6%) and unassigned subtypes (n = 29; 22.0%), with a VL range of 1.05E+3 to 1.00E+7 (median, 6.92E+4) RNA copies/ml. All of these samples were isolated and amplified at the local laboratory. A selection of 50 amplicons was subsequently sequenced in order to confirm compatibility with the local sequencing method.

Nucleic acid purification. (i) UMC Utrecht, The Netherlands.

Viral RNA was isolated using the manual NucliSENS miniMAG (bioMérieux). For the subtype reference panel plasma dilutions and the clinical samples, an input volume of 100 μl was used. In the case of DBS samples, two 50-μl spots were used as input material. Prior to processing, the excised spots were incubated in 2 ml NucliSENS lysis buffer (bioMérieux) at room temperature for 30 min with gentle shaking, after which the filter papers were removed and the isolation proceeded according to the manufacturer's instructions. Isolated RNA was eluted in 50 μl elution buffer and was either used immediately for RT-PCR (n = 67) or stored at −20°C for 50 to 120 weeks (n = 145) prior to amplification. For each isolation, a positive and a negative control were used, consisting of subtype C virus from the subtype panel (2.00E+4 RNA copies/ml) and negative plasma, respectively.

(ii) Joint Clinical Research Centre, Kampala, Uganda.

Viral RNA was isolated using the QIAamp viral RNA minikit (Qiagen GmbH, Germany). For the clinical samples, 100 μl of plasma was used as the input volume. Isolated RNA was eluted in 50 μl elution buffer and was either used immediately for RT-PCR amplification or stored at −20°C for up to 4 weeks. For each isolation, a positive and a negative control were used, consisting of HIV-1 subtype C virus from the subtype panel and nuclease-free water, respectively.

Amplification (The Netherlands and Uganda).

Isolated viral RNA was reverse transcribed and amplified using a single-round RT-PCR encompassing the RT gene from aa 41 to 238. The RT-PCR was performed using the SuperScript III-one-step RT-PCR system with Platinum Taq high-fidelity polymerase (Life Technologies, Foster City, CA) as follows. Ten microliters of isolated RNA was amplified in a reaction mixture consisting of 0.4 μM forward primer, 0.4 μM reverse primer (Table 2), 2× reaction buffer, 0.5 μl SuperScript III RT/Platinum Taq high-fidelity enzyme mix, and nuclease-free water in a final volume of 25 μl. The combined cDNA and amplification reactions were carried out in a GeneAmp 2720 thermal cycler (Life Technologies) under the following conditions: 1 hold at 50°C for 30 min, 1 hold at 94°C for 2 min, 2 cycles at 94°C for 15 s, 61°C for 30 s, and 68°C for 1 min, 14 cycles at 94°C for 15 s, 60°C (with a decrease of 0.5°C with every cycle) for 30 s, and 68°C for 1 min, 34 cycles at 94°C for 15 s, 53°C for 30 s, and 68°C for 1 min, and a final extension step at 68°C for 7 min.

Table 2.

Amplification and sequencing primers

Primer	Positiona	Length (ntb)	Sequence	Application
ART-A_2611_M13-40	2611–2635	42	5′-GTTTTCCCAGTCACGACTTAAACAATGGCCATTGACAGAAGA-3′	Forward amplification
ART-A_3349_SR	3370–3348	23	5′-ATCCCTGSRTAAATCTGACTTGC-3′	Reverse amplification/sequencing
M13-40		17	5′-GTTTTCCCAGTCACGAC-3′	Forward sequencing

^aAccording to HIV-1 HXB2.

^bnt, nucleotides.

Amplification products were visualized on an agarose gel, and positive reactions were purified using QIAquick PCR purification columns (Qiagen GmbH, Germany) according to the manufacturer's instructions.

Sequence analysis. (i) Applied Biosystems-based sequencing, UMC Utrecht, The Netherlands.

The cycle sequencing master mix for each primer consisted of 0.2 μM primer (Table 2), 1× BigDye sequencing buffer, 10 to 40 ng of PCR product, 2 μl BigDye v3.1, and nuclease-free water in a final volume of 20 μl. The cycle sequencing reactions were performed using a GeneAmp 2720 thermal cycler and consisted of 25 cycles at 96°C for 10 s, 50°C for 5 s, and 60°C for 4 min. The reaction products were purified using an EDTA-ethanol precipitation, subsequently run on an ABI 3730 automated sequencer (Life Technologies), and analyzed using SeqScape data analysis software v2.6 (Life Technologies).

(ii) Beckman Coulter-based sequencing, Joint Clinical Research Centre, Kampala, Uganda.

The cycle sequencing master mix for each primer consisted of 0.2 μM primer (Table 2), GenomeLab DTCS quick start kit mix (Beckman Coulter, Inc., Brea, CA), 10 to 40 ng of PCR product, and nuclease-free water in a final volume of 20 μl. The cycle sequencing reactions were performed using a 9800 Fast thermocycler (Life Technologies) and consisted of 30 cycles at 96°C for 20 s, 50°C for 20 s, and 60°C for 4 min. Reaction products were purified using an EDTA-ethanol precipitation, subsequently read on a CEQ800 genetic analysis system (Beckman Coulter), and analyzed using BioEdit v7.0 (13).

(iii) Phylogenetic analysis.

Consensus sequences were aligned using MEGA 4.1 software (http://www.megasoftware.net/) to control for contamination. To determine the subtype and the HIV-1 drug resistance (HIVDR) profile, the NCBI subtyping tool (http://www.ncbi.nlm.nih.gov/projects/genotyping/formpage.cgi) and GRADE analysis programs (http://www.hiv-grade.de/grade/deployed/grade.pl?program=hivalg) were used, respectively. Consensus sequences, obtained as described above, were compared at the nucleotide level to sequences for the same samples generated using both the ViroSeq and an in-house genotyping assay (11) for quality control purposes.

RESULTS

Assay design.

Based on the optimal RT amplicon to contain all the RT resistance mutations that are relevant after an RT-based first-line failure, a single-round RT-PCR assay targeting RT aa 41 to 238 was developed. Primers were designed using a database of 267,841 sequences from nine subtypes and seven CRFs in order to have optimal sensitivity and specificity for all major HIV-1 subtypes and CRFs.

Nucleic acid amplification (RT-PCR sensitivity).

The assay was optimized for the amplification of plasma samples containing ≥5.00E+3 RNA copies/ml. The initial amplification of a dilution series of the subtype panel indicated that amplification from plasma virus was achievable from 5.00E+2 RNA copies/ml and was reproducibly achievable for samples with ≥5.00E+3 RNA copies/ml (Table 3). The results of the dilution series of DBS samples for the subtype panel indicated reliable amplification of samples with ≥5.00E+3 RNA copies/ml. All subtypes were amplified and sequenced with equal sensitivities and success rates.

Table 3.

Assay sensitivity determined using a dilution series performed in duplicate using isolates from the subtype panel

Subtype	Sensitivitya at a dilution (RNA copies/ml) of:
	10,000	5,000	1,000	500
A	++	++	++	+−
B	++	++	++	++
C	++	++	++	−+
D	++	++	++	−+
F	++	++	++	++
G	++	++	++	++
H	++	++	++	++
CRF01_AE	++	++	++	−+

^a+, positive amplification; −, negative amplification.

Clinical samples from UMC Utrecht, The Netherlands.

A total of 205 of 212 clinical samples (96.7%) were successfully amplified using the single-round RT-PCR. Of the 205 successfully amplified samples, full bidirectional sequencing of RT aa 41 to 238 (Table 4) was obtained 98% of the time (201 samples), giving an overall genotyping success rate of 94.8% (201/212). Of the seven samples that did not amplify, five had a VL of <5.00E+3 RNA copies/ml, all subtype C, and the remaining two were subtypes C and D, with VLs of 6.30E+3 and 8.78E+3 RNA copies/ml, respectively. Four samples resulted in incomplete sequences for subtypes A (n = 1), C (n = 2), and D (n = 1).

Table 4.

Amplification and sequencing results of clinical samples tested at the UMC Utrecht, The Netherlands

VLa range (RNA copies/ml)	No. of samples	No. of samples of the following subtypesb:							Success rate (no. of successes/total no. [%]) of:
		A	B	C	D	AE	AG	Rare	Amplification	Bidirectional sequencing
>125,000	35	10	6	1	2	12	2	2	35/35 (100)	35/35 (100)
25,000–125,000	45	12	2	8	7	9	3	4	45/45 (100)	44/45 (97.8)
5,000–25,000	89	14	8	47	3	10	5	2	87/89 (97.8)	84/87 (96.6)
1,000–5,000	43	11	4	16	6	4	0	4	38/43 (88.4)	38/38 (100)
Total	212	47	20	72	18	35	10	12	205/212 (96.7)	201/205 (98.1)

^aVL, viral load.

^bIndicated subtypes as assigned by the NCBI and GRADE.

The sequences generated with the described method were compared to the sequences previously generated (n = 30), using either ViroSeq or the in-house assay, and demonstrated average homologies of 98.9% and 99.3% at the nucleotide level, respectively. Sequences from a total of 73 samples were compared to those from the in-house assay alone and demonstrated a 99.2% homology. The differences observed were almost entirely due to mixture calling (347/43,362 positions), with only six positions at which the nucleotides were different, all at nonresistance positions.

For the selection of 25 clinical DBS samples genotyped with the described method, the amplification success rate was 95.0% for samples with >5.00E+3 RNA copies/ml (19/20) and 88.0% for samples with >1.00E+3 RNA copies/ml (22/25). The sequencing was 95.5% successful for the amplified samples (21/22). The overall genotyping success rate was 84.0% (21/25). The generated sequences showed 97.0% nucleotide homology to the previously generated sequences from plasma.

Field implementation in Kampala, Uganda.

A total of 130 of 132 local plasma samples (98.5%) were amplified using the described method (Table 5) with the alternative viral RNA isolation method. The samples that did not provide a positive amplification result had <5.00E+3 RNA copies/ml; however, amplification of lower-VL samples ranging from 1.05E+3 to 4.92E+3 (median, 2.12E+3) RNA copies/ml was also successful (15/17 samples; 88.2%). Using the Beckman Coulter sequencing method was successful, with full bidirectional sequences obtained for the 50 amplicons processed.

Table 5.

Amplification results of clinical samples tested at the Joint Clinical Research Centre, Kampala, Uganda

VLa range (RNA copies/ml)	No. of samples	No. of samples			Amplification success rate (no. of successes/total no. [%])
		Ab	Db	Xc
>125,000	48	23	14	11	48/48 (100)
25,000–125,000	36	14	13	9	36/36 (100)
5,000–25,000	31	19	11	1	31/31 (100)
1,000–5,000	17	4	5	8	15/17 (88.2)
Total	132	60	43	29	130/132 (98.5)

^aVL, viral load.

^bIndicated subtype previously assigned.

^cUnknown subtype, not previously sequenced.

Cost.

The cost of this assay is lower than that of other laboratory-developed assays and commercially available genotyping assays due to the limitation of the amplification and sequencing regions to the minimal region required for sequencing of first-line RTI therapy in RLS. The use of a single-round RT-PCR and reduced amplification and elution volumes and the need for only two sequencing reactions equate to a reduction in reagents required. From a reagent perspective, using the described assay would result in a >75% savings compared to using a commercial assay such as the ViroSeq genotyping system version 2.0 (Celera Diagnostics, USA) and an approximate 40% savings compared to using our current in-house assay (11). Furthermore, the quicker laboratory protocol and shorter sequence to be analyzed result in a decrease in the labor required compared to currently available methods (11, 14, 15).

DISCUSSION

We have developed a simplified, specific, lower-cost assay for the determination of HIV-1 drug resistance associated with first-line therapy that can be performed at reference laboratories in RLS and is suitable for use with DBS.

The unique feature of this assay which sets it apart from currently available commercial and laboratory-developed assays is the one-step RT-PCR specifically focusing on the analysis of the most relevant part of RT. The size of the amplicon is roughly half the length of those generated using commercial and in-house HIV-1 drug resistance genotyping assays (11, 15, 16). By amplifying as little as possible of the RT using highly fine-tuned primer combinations, focusing on the region encompassing all of the relevant HIV-1 drug resistance mutations, it was possible to achieve a genotyping success rate of 94.8% for clinical plasma samples with ≥1.00E+3 RNA copies/ml in a single-round RT-PCR. Subsequent sequencing requiring only a single forward and a single reverse primer, compared to the four to six primers needed for commercial and in-house assays (11, 14–16), increases the throughput for processing and decreases the analysis time per sample. In turn, decreasing the number of reactions required decreases the overall cost of the assay and minimizes the hands-on time, contamination risk, and turnaround time.

The described assay has been specifically designed with several key features for use in RLS. The primers were carefully designed and selected to cover all major HIV-1 group M subtypes and CRFs, as shown in Tables 3, 4, and 5. The nonnested approach to amplification strongly reduces the risk for sample contamination. The reduced number of reactions required for amplification and sequencing ensures the efficient use of reagents and allows for greater sample throughput. To demonstrate these principles, the assay was transferred to, and applied in, a reference laboratory in Kampala, Uganda. The results showed a very high success rate for amplification, with 99% (n = 130/132) amplification success for samples with a VL of >1.00E+3 RNA copies/ml.

Genotyping from DBS has become a popular sampling method used to overcome the impediments associated with the cost and logistics of transport and storage of plasma in RLS, and it has been shown to give comparable results from plasma (16–18). Recently, the WHO identified DBS as the primary sample type for genotyping in RLS (19). The described RT-specific genotyping assay using a DBS dilution series and application with clinical DBS samples show that this method displays adequate sensitivity for use with DBS samples, with 95% and 84% genotyping success rates for samples with VLs of ≥5.00E+03 and ≥1.00E+3 RNA copies/ml, respectively. Research is under way in Uganda and South Africa as part of the ART-A project to demonstrate the scalability of HIV-1 drug resistance genotyping using DBS sampling. In this approach, DBS samples are collected from HIV-1-infected individuals and sent to a reference laboratory to screen for virological failure. Samples that test positive for treatment failure, classified as having a VL of >5.00E+3 RNA copies/ml (3), are selected for HIV-1 drug resistance genotyping from the same DBS sample using the described assay.

HIV-1 drug resistance testing for individual patient management is currently not recommended in RLS, mostly due to cost limitations. However, targeted monitoring and surveillance of HIV-1 drug resistance on sentinel populations are increasingly mentioned as a necessity for guiding national ART programs (2). The potential use of this test in DBS-supported applications would markedly increase its application in remote settings. This, in combination with the ability to transport specimens at an ambient temperature to a centralized reference laboratory where actual HIV-1 drug resistance genotyping is performed, will contribute to more affordable population-level HIV-1 drug resistance data collection, which is necessary for keeping national ART programs effective in the long term.

In conclusion, we have designed and tested a simplified method for HIV-1 RT drug resistance genotyping. The sensitivity, broad subtype inclusivity, and compact nature of this assay make it ideal for HIV-1 drug resistance testing in RLS. The described assay generates the most vital information necessary at a lower cost and in a shorter time compared to currently available HIV-1 drug resistance genotyping assays.