Development of a Qualitative Quantitative Polymerase Chain Reaction Test to Identify Patients Failing First-Line Therapy to Non-Nucleotide Reverse Transcriptase Inhibitor

Abstract

Antiretroviral therapy (ART) can be compromised by selection of drug resistance strains, which can be promoted by lack of adherence during therapy and drug tolerance, and some of these drug-resistant strains can persist for years as minority populations. The K103N drug resistance mutation is selected by the use of non-nucleotide reverse transcriptase inhibitors, including nevirapine or efavirenz (EFV), used in low-income countries. Here we describe the use of a less expensive qualitative point mutation polymerase chain reaction (PMqPCR_K103N) targeting K103N mutation. To validate the use of this methodology, we tested previously sequenced samples from patients treated with highly active ART with viral loads above 2,000 copies/ml and compared the results of our assay with Illumina deep sequencing. Due to its low cost and high specificity, this test is particularly suitable for low-income countries to screen for pretreatment resistance in patients either initiating ART or failing first-line regimens containing EFV.

Introduction

Antiretroviral therapy (ART) decreases the morbidity and mortality caused by the human immunodeficiency virus (HIV). However, it can become ineffective by selection of drug-resistance strains.^1
–3 Some patients have problems with adherence or tolerance that may lead to the selection of a drug-resistant virus and compromise treatment increasing viral load, allowing the establishment of other infections. Some virus-resistant strains can persist for years as minority subpopulations and suddenly emerge affecting future patient treatment. Early detection of resistant viruses is important for management of HIV-infected patients.^4,5 In addition, recent studies have shown that up to 10% of the newly diagnosed infections have evidence of transmitted drug resistance.^6

–12 Therefore, identification of a key mutation in the viral population is crucial for surveillance of primary and acquired resistance to select the proper treatment involving other drug interventions.¹³

The WHO recommends a first-line regimen based on two nucleotide reverse transcriptase inhibitors (NRTI) and a non-nucleotide reverse transcriptase inhibitor (NNRTI) in resource-limited settings. When patients taking this first line start to fail therapy, the first mutations to appear are NNRTI related, such as K103N and Y181C. K103N deserves special attention since it does not substantially affect viral fitness, can be transmitted with high efficiency, and is retained in the viral genome for more than 3 years postinfection.^3,7,14

In addition, K103N is one of the most common drug resistance mutations (DRMs) in most low-income countries as nevirapine (NVP) or efavirenz (EFV) are the first drugs adopted.¹⁵ NVP has been extensively utilized as monotherapy during labor to prevent mother-to-child transmission,^13,16 accompanied by a week of zidovudine (AZT) and lamivudine (3TC) treatment postpartum. Nonbreastfeeding newborns usually receive (NVP) or (AZT) for 6 weeks after birth.¹⁵ These treatments select for low-frequency HIV drug resistance variants that are not easily detected by Sanger-based methods.¹³

Despite the relative low sensitivity and high cost of Sanger-based sequencing, it still remains the dominant platform for drug resistance studies. This method has good specificity and can accurately detect critical DRM codons. However, it is unable to detect the low-frequency variants that are present in <5%–30% of the virus population in the sample,^{13,17

–23} and its sensitivity to identify DRM is <70%.^24
–26 Sanger sequencing is also unable to discover complex combinations of mutations^24,27 present in quasispecies from a single sample. These limitations can compromise genotyping, affecting treatment decisions.^20,28 In 2014, both Bellecave et al. ²⁰ and Lipscomb et al. ²⁸ presented patients with low-frequency DRM previously detected to antiretroviral (ARV) treatment, compromised by the emergence of those viruses.

Quantitative polymerase chain reaction (qPCR) is an alternative method to detect low-frequency mutations in virus populations.^13,16 qPCR genotyping is potentially more sensitive, specific, and easy to perform than Sanger sequencing. In fact, qPCR is especially useful for screening large numbers of patient samples, having been incorporated since the early 1990s for the quantification of HIV-1, a predictor of AIDS progression.^17,29 Currently, qPCR is considered the gold standard for virus detection and quantification,³⁰ also accurate and reliable for detection of point mutations in human genome as well as in infectious agents.^31
–33 Although qPCR methodology has the sensitivity and specificity to detect point mutations in HIV-1, its performance can be inhibited by sequence variation in the virus present in primer- and probe-binding regions. This is critical for genotyping HIV-1 group M viruses since the nine major subtypes and at least 71 Circulating Recombinant Forms (http://hiv.lanl.gov/content/sequence/HIV/CRFs/CRFs.html) vary from 25% to 35% in nucleotide diversity from each other.^3,34
–36

Herein, we targeted K103N, one of the main DRMs selected during NNRTI treatment,^13,15,16 and developed a point mutation qPCR method targeting K103N (PMqPCR_K103N) with SYBR Green. This method can be adopted in low-income countries without losing sensitivity and specificity when compared to Sanger-based methods.^13,16 We have defined PMqPCR_K103N specificity and sensitivity in a well-characterized HIV-1 isolate panel composed of drug-experienced individuals infected with subtypes B, C, and F by comparing this assay to sequencing data generated by Illumina deep sequencing, a methodology able to detect mutations as low as 2%.^37

–41

Materials and Methods

We used 103 plasma samples (87 collected from 2003 to 2006 and 16 in 2015) from Brazilian patients infected with HIV-1 subtypes B, BF, F, and C, on antiretroviral treatment. All samples were selected from patients with previous Sanger sequencing analysis and had viral loads greater than 2,000 copies/ml, indicating treatment failure or loss of immune control. Twenty-eight of the seventy-five patients failing highly active antiretroviral therapy (HAART) had no K103N mutation identified by Sanger. We designed a sensitive qPCR test to detect a K103N mutation associated with NNRTI, for which we adapted primers already described^13,16 to better match HIV-1 sequences circulating in Brazil (see Supplementary Table S1 for details; Supplementary Data are available online at www.liebertpub.com/aid). We included some nucleotides providing alternatives in the primers according to the polymorphism patterns of HIV-1 circulating in Brazil (data not shown). This strategy was also described previously by Pérez et al. ¹⁶ who carried out the study of resistance mutations in Argentina. We adopted the SYBR Green detection system, which allowed us to minimize detection costs and ensure identification of the questioned mutation.

Preparation of virus template

HIV viral RNA was extracted from plasma with the QIAamp^® Viral RNA kit (QIAGEN^®). The cDNA template for our experiment was obtained with Superscript III (Life Technologies) with random primers according to the manufacturer's instructions. We then performed an external PCR, to ensure enough DNA for repeat testing, with primers (P1 and P2), previously described by Johnson et al.,¹³ which amplifies from nucleotide 1,799 to nucleotide 2,850 relative to the HXB2 sequence (Supplementary Table S1). Approximately 30 ng of the cDNA was added to the following PCR mastermix in a first round of PCR amplification: Taq Gold buffer 1 × , MgCl₂ 0.5 mM, dNTP 250 nM, Primers P1 Fw, P2 Rev 250 nM each, and Taq Gold Invitrogen 1 U. The cycling conditions were as follows: 1 cycle of 94°C 5 min; 6 cycles of 94°C 1 min, 55°C 1 min, 72°C 1.20 min followed by 35 cycles of 94°C 1.20 min, 55°C 1 min, 72°C 1.20 min, and 1 cycle of 72°C 7 min. The approximate 1,000 bp amplicon was observed by gel electrophoresis in a 1% agarose gel stained with ethidium bromide. The amplified product was purified with Montage PCR centrifugal filter devices (Millipore Corp.) and quantified on a NanoDrop spectrophotometer (Thermo Fisher Scientific, Inc.) before joining the second round on the PMqPCR_K103N.

PMqPCR_K103N

The objective of the test requires the amplification of the same region with two different sets of primers. The first set of primers (general target amplification for mutant and nonmutant detection) anneals in a constant region of the RT gene, exhibiting low diversity between resistant and wild-type viruses. General target amplification is necessary to obtain the baseline Ct value corresponding to the Ct value that represents all templates regardless of mutation status. For the general amplification, we used one forward primer (K103N Fw) and two reverse primers (K103N Rev Tot1 and K103N Rev Tot2) to amplify all HIV-1 variants circulating in Brazil. The cycle threshold detected with these general primers generates Ct₁. For the questioned K103N DRM (specific target amplification), we used the same forward (K103N Fw) and four reverse primers (K103N Rev AAT1, K103N Rev AAT2, K103N Rev AAC1, and K103N Rev AAC2) similar to the first set but with one located nucleotide downstream (5′→3′) of the general reverse primer (Fig. 1). With this second set of primers, we obtained the Ct₂ representing the presence of the K103N mutation. It is necessary to calculate the ΔCt value (Ct₂ − Ct₁) to determine the presence or absence of the K103N mutation. All samples were tested in triplicate reactions with the Ct average used to calculate the final ΔCt. We expected to have low ΔCt values for the samples with the questioned DRM and higher values for those without.

FIG. 1.

Scheme representing primers annealing with coordinates relative to HXB2. P1 Fw and P2 Rev—forward primer and reverse primer, respectively, K103N Fw—forward primer used in qPCR, K103N Rev Tot1, K103N Rev Tot2—reverse primers used with K103N Fw for total detection; K103N AAC1, K103N AAC2, K103N AAT1, K103N AAT2 second set of primers used with K103N Fw for allele-specific mutation. qPCR, quantitative polymerase chain reaction.

Individual reactions were carried out in a total volume of 20 μl/well in 96-well PCR plates on an Exicycler TM Real Time PCR system (Bioneer Corporation) with SYBR Green reaction mix (Applied Biosystems). The reaction conditions were as follows: 10.0 μl of SYBR mix (Applied Biosystems), primer K103NFw 300 nM pM, primers RevTot1 and RevTot2 250 nM for all DNA templates or primers RevAAC1, RevAAC2, RevAAT1, RevAAT2 25 pM for K103N DRM, and 0.1 ng of target DNA per reaction. The cycling conditions were as follows: 95°C for 11 min followed by 35 cycles of 95°C 30 s, 57°C 1 min. and the melting curve was recorded from 55°C to 94°C increasing 1°C per minute to check the identity of amplified material.

PMqPCR_K103N validation

To establish the sensitivity and specificity of the PMqPCR_K103N, we used two clonal samples, one representing the wild-type virus (NL4-3) and the other carrying the K103N (clone 154) mutation. Ten-fold dilutions were produced starting with 1 ng/μl, ∼8.63 × 10⁸ to 8.63 × 10⁴ copies, of purified amplicons generated in the first amplification (P1 and P2 primers) for each clone to determine the best annealing temperature, primer concentrations, proper amount of sample for each assay, and to also establish the standard curve (Fig. 2).

FIG. 2.

Ct × DNA dilutions. Results obtained NL4-3 and 154 clone samples in different concentrations. Average of total Ct is the value obtained with primers K103N Fw and K103N Rev Tot1 and K103N Rev Tot2; K103N Ct was obtained with K103N Fw and K103N Rev AAT1, K103N Rev AAT2, K103N Rev AAC1, and K103N Rev AAC2 in triplicate experiments.

Illumina MiSeq sequencing

The 1,067 bp PCR product generated with the P1 and P2 primers was subjected to Illumina deep sequencing. The DNA was either purified with AMPure XP beads (Beckman Coulter, Inc.) or gel purified with a MinElute Gel Extraction Kit when there were PCR products in addition to the expected 1,067 bp PCR product (Qiagen). Paired-end DNA libraries were prepared with the Nextera XT DNA sample prep kit (Illumina) according to the manufacturer's protocol. Briefly, ∼1.0 ng of PCR product was subjected to tagmentation with Nextera amplicon tagment enzyme mix at 55°C for 5 min followed by a twelve-cycle PCR with the Nextera PCR mix to add Illumina indexes to the PCR products. The reaction conditions were 72°C for 3 min, 95°C for 30 s followed by 12 cycles of 95°C 10 s, 55°C 30 s, 72°C 30 s and finally one cycle of 72°C for 5 min. The average size of the fragmented libraries was 683 bp, later purified with AMPure XP beads (Beckman Coulter, Inc.) at a 5:3 ratio of PCR product to bead. The product of each sample was normalized to 2 nM and pooled together in preparation for loading into the Illumina MiSeq Sequencing cartridge for sequencing. Samples were sequenced with an Illumina MiSeq Sequencing 500-cycle version 2 kit (Illumina).

Illumina MiSeq data analysis

The next-generation sequencing (NGS) data generated with Illumina for each sample were analyzed with Geneious Pro Version 7.1 (Biomatters). From a total of 103 sequenced samples, 21 had low coverage (<100 reads) and were excluded from the study. Next, we randomly subsampled 15,400 sequences from each sample with Seqtk tool from the Geneious software. This downsampling reduced computational load while still providing 3,589 reads of average coverage of each base in the 1,060 bp amplicons. Reads were then mapped to the HIV-1 reference sequence HXB2 (GenBank NC_001802) annotated with DRMs in accordance with the “medium” preset sensitivity mapping parameters in Geneious. Polymorphisms (defined as variants found in at least 2% of reads, supported by at least five reads, and with a p-value of at least 10e-6) at the K103N resistance site were identified with the Geneious genotyping tool. A workflow automating these steps is available on request. Samples with the K103N mutation found below 2% were considered negative.

Statistical analysis

The data from Geneious identified which samples contained the K103N mutation and at what frequency were exported to Microsoft Excel 2013, and the plots were generated with R v3.02 in RStudio v0.98. We established sensitivity and specificity of PMqPCR_K103N with the Clinical Research Calculator on VassarStats statistical computation website (http://vassarstats.net) based on the Newcombe.⁴¹

Results

Initially, we designed standard curves with five different concentrations ranging from 1.0 to 0.0001 ng/μl of NL4-3 or 154 cloned strain samples, and the efficiency of the assay was 99.768% (slope of −3.328 and r ² = 0.993) and 99.581% (slope of −3.332 and r ² = 0.99), respectively, as determined by Life Technologies 7500 software V2.3 (Fig. 2). However, when 0.1 ng of DNA input was considered, a more evident difference in Ct was observed between the mutant and wild-type samples with total and mutant primers. Finally, we established a fixed concentration of 0.1 ng/μl of purified DNA from all samples amplified with the P1 and P2 primers as the ideal input for our PMqPCR_K103N assay.

We initially separated the samples into two groups: those with K103N DRM and those with mutations other than K103N (NR-DRM), based on the genotyping generated by Sanger sequencing (Tables 1 and 2). As mentioned before, the Sanger sequencing method has a detection limit of minority populations around 5%–30%,^{13,17

–23} and so, we decided to enlist Illumina MiSeq to investigate the limitations of our PMqPCR_K103N. From the initial 103 samples, we obtained good Illumina sequence data in 82 samples and only these sequences were adopted for further analysis. We used data generated with Illumina MiSeq for each sample as our gold standard to determine the presence or absence of the K103N mutation and the total of mix populations to evaluate our PMqPCR_K103N performance. All 47 samples carrying the K103N mutation based on both methodologies displayed an average of 82% of reads carrying K103N by Illumina sequencing (Range 5%–99.8%). The cutoff calculated obtaining ΔCt average of all positive samples was 2.5. To calculate the cutoff value for ΔCt, we added two standard deviations (2.12) resulting in a cutoff of 6.742 Cts (see Fig. 3 for details). There was a good correlation between the percentage of reads carrying K103N by Illumina sequencing and the observed ΔCt. However, we witnessed outliers with low and high amounts of reads and ΔCt. When the 11 specimens originated from patients failing ARV regimens containing previous NNRTI and no K103N mutation by Sanger methodology, we realized that two of them, 044/05 and 012/06, contained the K103N mutation frequency at 4.3% and 31.3% by Illumina MiSeq, respectively. We considered one sample, 059/05, as undetermined, although the ΔCt value was below the cutoff value. The reason for this is because the K103N frequency achieved with Illumina MiSeq was under 2.0%, this value was considered low for reliability. When all NR-DRM samples were analyzed, we found all of them had ΔCt values below the cutoff value, thus suggesting the presence of the K103N mutation. Remarkably, all these samples were from subtype C isolates circulating in Brazil, and the percentage of reading carrying K103N was below the cutoff value of Sanger methodology except for samples 457 and 462, which had 33.3% and 96.2% of readings with K103N DRM, respectively. Considering Illumina sequencing as the gold standard, the specificity of the PMqPCR_K103N assay was 94.53% and 91.98% for sensitivity (http://vassarstats.net).

FIG. 3.

Mean and distribution of ΔCt (ordinate) results obtained for samples with K103N mutation and NRDRM. N is for number of samples; NDRM, no related drug mutation.

Table 1.

Samples with K103N Drug Resistance Mutation Identified by Sanger Technology and Processed by PMqPCR_K103N and Illumina MiSeq Sequencing

Sample ID	ΔCt 103	K103N % MiSeq	qPCR × MiSeq agreement
027/06	0.05	99.4	YES
104/06	0.29	99.6	YES
124/06	0.5	26.6	YES
045/05	0.56	97	YES
052/06	0.72	99.2	YES
177/05	0.79	62.5	YES
286/03	0.91	23.5	YES
042/05	1	99.1	YES
185/05	1.05	89.1	YES
011/05	1.08	89.9	YES
073/05	1.42	93.8	YES
002/06	1.5	97.6	YES
183/05	1.61	56.9	YES
069/06	1.64	65.1	YES
067/05	1.85	98.6	YES
043/06	1.87	99	YES
077/05	1.95	59.9	YES
126/05	2.07	79.4	YES
021/05	2.18	99.6	YES
133/06	2.23	99.1	YES
139/06	2.23	99.4	YES
087/05	2.32	99.8	YES
192/05	2.34	97.7	YES
078/06	2.48	41.8	YES
097/05	2.49	96.5	YES
009/06	2.68	99.5	YES
103/06	2.7	87	YES
035/05	2.71	99.8	YES
029/05	2.79	99	YES
143/06	2.88	93.9	YES
030/05	2.95	46.5	YES
076/06	2.97	99.8	YES
166/05	3.16	99.2	YES
280/03	3.42	99.9	YES
069/05	3.79	98.6	YES
013/05	4	67.1	YES
119/06	4.02	91.3	YES
138/06	4.09	96.9	YES
205/05	4.13	99.9	YES
093/05	4.71	8.5	YES
114/06	4.91	99.3	YES
193/05	5.37	79.4	YES
058/06	5.49	98	YES
011/06	5.55	99.7	YES
072/06	5.58	99.4	YES
046/05	5.71	14	YES
061/06	6.31	12.1	YES

qPCR, quantitative polymerase chain reaction.

Table 2.

Samples from Drug-Naive Treatment Patients or Patients Failing Highly Active Antiretroviral Therapy and Processed by PMqPCR_K103N and Illumina MiSeq Sequencing

Sample ID	ΔCt 103	K103N% MiSeq	DRM Sanger	qPCR × MiSeq agreement
457	0.13	33.3	NR-DRM	YES
65	1.04	5	NR-DRM	YES
394	1.25	16.7	NR-DRM	YES
462	1.39	96.2	NR-DRM	YES
044/05	2.74	31.3	NR-DRM	YES
012/06	2.76	4.3	NR-DRM	YES
468	3.79	17.1	NR-DRM	YES
62	5.18	18.2	NR-DRM	YES
455	5.88	2.4	NR-DRM	YES
059/05	6. 43	1.5	NR-DRM	IND
RS11	9.00	0	NR-DRM	YES
123/04	9.26	0	NR-DRM	YES
079/05	11.53	0	NR-DRM	YES
RJ01	13.01	0	NR-DRM	YES
SP14	13.06	0	NR-DRM	YES
PR20	13.25	0	NR-DRM	YES
SP17	13.59	0	NR-DRM	YES
002/05	13.67	0	NR-DRM	YES
SC07	13.78	0	NR-DRM	YES
147/04	13.97	0	NR-DRM	YES
008/05	14.07	1.2	NR-DRM	YES
MG07	14.79	0	NR-DRM	YES
PR04	15.11	0	NR-DRM	YES
RS03	15.36	0	NR-DRM	YES
RS07	15.76	0	NR-DRM	YES
RJ05	17.27	0	NR-DRM	YES
RJ15	17.69	0	NR-DRM	YES
007/06	18.2	0	NR-DRM	YES
RS08	19.00	0	NR-DRM	YES
137/06	19.01	0	NR-DRM	YES
024/06	20.93	0	NR-DRM	YES
078/05	21.98	0	NR-DRM	YES
RS12	22.61	0	NR-DRM	YES
MG17	25.35	0	NR-DRM	YES
SP45	>35.0	0	NR-DRM	YES

Bold values indicate indeterminate and negative results for K103N mutation.

IND is for inconclusive result if compared to Illumina MiSeq. NR-DRMs are samples from patients failing HAART carrying mutations other than K103N.

DRM, drug resistance mutation; HAART, highly active antiretroviral therapy.

To analyze the reasons for the discrepancy between the amount of reads containing K103N and the ΔCt values, we aligned the sequences with different ΔCt values exhibiting the region of forward and reverse primer anchorage (Table 3). Of note, there were much more mismatches, in 3′ primer positions, between the forward and reverse primers in samples with ΔCt value discrepancies when compared to the concordant counterpart.

Table 3.

Samples that had K103N Drug Resistance Mutation Detected by PMqPCR_K103N

Primers	ΔCt	% found by MiSeq	Consensus obtained by MiSeq 5′→3′	Consensus obtained by MiSeq 3′←5′
K103N Fw.			AGR GAA CTT AAT AAR AGA ACT CAA GA
REV1				AYY TCA GTY ACA GTA CTG GAT GTG GG
REV2				AAA TCA GTG ACA GTA CTG GAT GT
AAT1				TAY YTC AGT YAC AGT ACT GGA TGT GGT
AAT2				TAA ATC AGT GAC AGT ACT GGA TGT GGT
AAC1				CAY YTC AGT YAC AGT ACT GGA TGT GGT
AAC2				CAA ATC AGT GAC AGT ACT GGA TGT GGT
Samples ID.
027/06	0.05	99.5	AGA GAG CTT AAC AAG AGA ACT CAA GA	CAA ATC AAT AAC AGT ACT AGA TGT GGG
104/06	0.29	99.6	AGA GAA CTT AAT AAG AGG ACT CAA GA	CAA ATC AGT AAC AGT GCT AGA CGT GGG
124/06	0.5	26.6	AGA GAA CTT AAT AAG AGA ACT CAA GA	CAA ATC AGT AAC AGT ACT GGA TGT GGG
045/05	0.56	99.6	AGA GAA CTT AAT AAG AGA ACT CAA GA	CAA ATC AGT AAC AGT ACT GGA TGT GGG
177/05	0.79	99.8	AGA GAA CTT AAT AAA AGA ACT CAA GA	CAA GTC AAT AAC AGT ACT GGA TGT GGG
286/03	0.91	26.6	AGA GAA CTT AAT AAA AGA ACT CAA GA	CAA GTC AGT AAC AGT ACT GGA YGT GGG
185/05	1.05	89.2	AGA GAA CTT AAT AAA AGA ACT CAA GA	CAA GTC AGT AAC AGT ACT GGA TGT GGG
011/05	1.08	92.2	AGA GAA CTT AAT AAG AGA ACT CAA GA	TAA ATC AGT AAC AGT ACT GGA TGT GGG
073/05	1.42	99.8	AGA GAG CTT AAT AAG AGG ACT CAA GA	CAA ATC AGT AAC AGT ACT AGA TGT GGG
002/06	1.50	97.6	AGA GAA CTT AAT AAG AGA ACT CAA GA	CAA ATC AGT AAC AGT ATT GGA TGT GGG
093/05	4.71	8.5	AGA GAA CTT AAT AAG AGA ACT CAA GA	CAA AAC AGT AAC GGT ACT GGA TTA GGG
114-06	4.91	95.9	AGA GAA CTT AAT AAG AAA ACT CAA GA	CGA AAA AGT AAC AGT ATT GGA TGT GGG
119/06	4.02	91.3	AGA GAA CTT AAT AAA AGA ACT CAA GA	CAA ATC AGT AAC AGT ATA GAT TGT GGG
138/06	4.09	96.9	AGA GAA CTT AAT AAG AGA ACT CAA TA	CAA ATC AGT AAC AGT ACT GGA TGT GGG
205/05	4.13	99.9	AGA GAG CTT AAT AAA AGA ACT CAA GA	CAA ATC AGT AAC AGT ACT AGA TGT GGG
193/05	5.37	79.4	AGA GAA CTT AAT AAA AGG ACT CAG GA	YAA ATC AGT AAC AGT ACT AGA TGT GGG
058/06	5.49	98.0	AAT GGA GGA AAT AGT AGA TTT TAG GA	CAA GTC AGT AAC AGT ATT GGA TCT GGG
011/06	5.55	99.7	AAT GGA GAA AAT TAG TAG ATT TCA GA	CAA ATC AGT AAC CGT ATT GGA TGT GGG
072/06	5.58	99.7	AGA GAA CTC AAT AAG AGA ACT CAA GA	CAA ATC AGT AAC AGT GTT GGA TGT GGG
455	5.88	2.4	AGA GAA CTT AAT AAG AAA ACW CAA GA	SAA ATC AGT GAT AGT ATT AGA CGT GAG
059/05	6.43	1.5	AAA GAA CTT AAA AAG AGT ACC TAA GA	CCA ATC AGT AAC AGT ACT GGA TGT GGG

Based on our cutoff, samples with ΔCt <8.814 have K103N mutation present. “Bold” nucleotides show the discrepancies against the primer. Sample 093/05 was not detected by Sanger methodology.

Discussion

Herein, we designed a low-cost, highly sensitive qPCR test based on previous works^13,16 to detect K103N, an important point mutation associated with (EFV) and (NVP) resistance. The estimated cost of our assay here in Brazil is US$2.00 to US$5.00 per sample, about US$10.00 less than Sanger sequencing. The cost difference, in part, is attributed to embedded taxes. In addition, the time spent in real-time execution is lower by at least 8 h. Due to the sensitivity limitations of Sanger-based sequencing, we decided to use Illumina sequencing to better investigate the amount of mixed populations to check the performance of our PMqPCR_K103N. In brief, we evaluated HIV RT sequences from 66 Brazilians harboring subtypes B, BF, F, and C. We designed PCR primers accommodating all polymorphic sites located at codon position 103 with a Brazilian sequence database. Utilizing this strategy, we were able to generate suitable primer sets to detect K103N mutations in the background of Brazilian HIV-1 isolates. In our hands, this methodology was better performed with amplicon concentrations above 0.01 ng/μl (around 10⁷ copies). We tried to evaluate our cutoff value for ΔCt considering 47 selected samples, where Illumina sequencing yielded results presenting K103N proportions above 2%. Although Illumina NGS technology has the lowest error rate compared to the other available NGS platforms, we prefer to use conservative profiles for demand polymorphisms. In this case, we used 2% based on the assay sensitivity of the NGS technology. Afterward, an assay cutoff was generated and utilized to check unknown samples from other individuals.

Since we utilized the product of PCRs from first-round PCR with external primers as DNA input on our PMqPCR_K103N, this amount of DNA is easily obtained starting with viral RNA extracted from plasma from HIV-1 patients above 2,000 copies/ml, following the Brazilian Ministry of Health protocol.⁴² Unfortunately, we do not have information regarding the drug regimen of patients, only that they have been treated according to the Brazilian Ministry of Heath⁴² protocol with TDF, 3TC, and EFV (Supplementary Table S2).

Analyzing the 82 samples that had good coverage data through Illumina sequencing, we were able to identify isolates carrying K103N DRM applying an independent method. We also adopted Illumina data as our gold standard to calculate our sensitivity and specificity. Overall, the prevalence between PMqPCR_K103N and Illumina MiSeq was 99.92% for the presence of K103 N DRM.

In our analysis, we treated ΔCt as a qualitative variable that is proportional to identify the presence of the K103N mutation. We calculated the ΔCt test cutoff as 6.742, allowing us to discriminate samples carrying K103N from the wild-type counterpart. We established that one sample (059/05) was undetermined for the questioned mutation as the K103N frequency was 1.5% in Illumina Deep Sequencing with a ΔCt lower than the cutoff of our test. Of note, this sample had only lysine in RT position 103 when sequenced by Sanger methodology.

We have selected some samples showing concordant and discordant ΔCt values with the proportion of K103N reads in our NGS methodology as well as designing an alignment with the HXB2 reference sequence and our forward and reverse primers (Table 3). We have observed a high number of mismatches in 3′ primer positions, between the forward and reverse primers in samples demonstrating a discrepancy in ΔCt values when compared to the concordant counterpart. Mismatches at primer 3′ ends are likely to modify the initial PCR causing lags in the amplification chain reaction yielding a higher Ct. This fact may explain the disparity in results where samples carrying a high proportion of K103N reads had high ΔCt values such as isolates 193/05, 058/06, 011/06, and 072/06 as opposed to samples 027/06, 104/06, 045/05, and 177/05.

We suggest that before adopting this methodology, the ΔCt cutoff value with the samples circulating in the country carrying the K103N mutation, as well as wild-type sequences from drug-naive individuals to better fit the primers with circulating viral variants, should be calculated. In our case, we think it may have superior results if we develop new forward primers with some more degenerate nucleotides to better accommodate polymorphism present in these sequences (Table 3).

We did not compare our allele specific polymerase chain reaction with other groups, but Jonhson et al. ¹³ reported sensitivity variation from 96% to 99% with specificity up to 99%. Pérez et al. ¹⁶ inform only that the lower limit of K103N DRM was 0.1%. Our test incorporated SYBR Green instead of TaqMan to minimize costs as well as primers, different from those developed by other groups, so as to match with HIV-1 circulating in Brazil (data not shown). Another point to be considered is that our assay was developed only to identify low-frequency mutations, but not to quantify as was described by other groups.^13,16

Some studies have reported that the importance of low-frequency DRMs and K103N is implicated to produce high-level resistance to EFV,^43

–46 both EFV and NVP known to select K103N DRM.^9,47 This DRM can also be found in children born from HIV-positive mothers under prevention of mother to child transmission interventions, as reported in a study in Kenya where it was the major DRM identified.⁴⁸ Due to its low viral fitness cost,⁴⁹ K103N can be present in DRM transmission cases. It is reported that virological failure may occur ∼6 months after being detected.²⁰ It has also been reported that patients with low-frequency DRMs ranging from 0.1% to 1.0% and with virus load above 2,000 copies/ml could lead to DRMs with stronger impact in treatment outcome.^8,19,44,50 This level of minor mutant detection is achievable by allele-specific real-time PCR technology.⁵⁰

Conclusions

We have developed an inexpensive qPCR incorporating SYBR Green to detect a K103N low-frequency mutation for the HIV-1 circulating in Brazil. We have used 103 samples of Brazilian patients infected with HIV-1 subtypes B, BF, F, and C and compared our results to those of both the Sanger-based method and MiSeq deep sequencing. Our results confirm that PMqPCR_K103N can be useful as a qualitative test, especially as the first-line drug regimen, adopted in low-income countries, includes EFV or NVP plus Stavudine (D4T) or AZT.⁵¹ Owing to the low negative predictive value of our PMqPCR_K103N, we think that it is critical to associate new qPCRs targeting other key NRTI or NNRTI DRMs, such as M184V or Y181C, to increase the likelihood of identifying patients failing first-line regimens. As the test was developed as a proof of concept, other targets can always be developed taking into account the diversity of the circulating viruses in the region. Since the cost of the HIV genotyping test based on Sanger sequencing is very high, about 80% more expensive than qPCR and not accessible to low-income countries, we propose the use of our PMqPCR_K103N for first-time screening in naive treatment HIV-1 individuals or in patients with virological failure in first-line antiretroviral regimens in low-income countries to minimize expensive HIV genotyping. This test will also reduce the early emergence of the DRM virus and increase the selection of patients for second-line regimens to suppress again patient viremia and diminish the costs of sequencing and identification.

Footnotes

Acknowledgments

This work was sponsored by CNPq, FAPERJ, CAPES, Department of AIDS/STD and Hepatitis, Brazilian Ministry of Health. English review and revision by Mitchell Raymond Lishan, native of Chicago, Illinois, USA-UCLA 1969.

Author Disclosure Statement

No competing financial interests exist.

References

Araujo

, Brites

, Monteiro-Cunha

, Santos

, Galvao-Castro

, Alcantara

LCJ

: Lower prevalence of human immunodeficiency virus type 1 Brazilian subtype B found in northeastern Brazil with slower progression to AIDS. AIDS Res Hum Retroviruses, 2010; 26:1249–1254.

Archer

, Weber

, Henry

, Winner

, Gibson

, Lee

, et al.: Use of four next-generation sequencing platforms to determine HIV-1 coreceptor tropism. PLoS One, 2012; 7:e49602.

Santoro

, Perno

: Review article. HIV-1 genetic variability and clinical implications. ISRN Microbiol, 2013; 2013;article ID 481314.

Booth

, Garcia-diaz

, Youle

, Johnson

, Phillips

, Geretti

: Prevalence and predictors of antiretroviral drug resistance in newly diagnosed HIV-1 infection. J Antimicrob Chemother, 2007; 59:517–524.

Bulteel

, Bansi-Matharu

, Churchill

, Dunn

, Bibby

, Hill

, et al.: The emergence of drug resistant HIV variants at virological failure of HAART combinations containing efavirenz, tenofovir and lamivudine or emtricitabine within the UK collaborative HIV cohort. J Infect, 2014; 68:77–84.

Weinstock

, Zaidi

, Heneine

, Bennett

, Garcia-lerma

, Douglas

, et al.: The epidemiology of antiretroviral drug resistance among drug-naive HIV-1—infected persons in 10 US cities. J Infect Dis, 2004; 189:2174–2180.

Vairo

, Nicastri

, Liuzzi

, Chaula

, Nguhuni

, Bevilacqua

, et al.: HIV-1 drug resistance in recently HIV-infected pregnant mother's naïve to antiretroviral therapy in Dodoma urban, Tanzania. BMC Infect Dis, 2013; 13:439.

: HIV-1 drug-resistant minority variants: Sweating the small stuff. J Infect Dis, 2014; 209:639–641.

Kim

, Song

, Ahn

, Kim

, Oh

, Choi

, et al.: HIV antiretroviral resistance mutations among antiretroviral treatment-naive and -experienced patients in South Korea. AIDS Res Hum Retroviruses, 2013; 29:1617–1620.

10.

Castro

, Pillay

, Cane

, Asboe

, Cambiano

, Phillips

, et al.: Persistence of HIV-1 transmitted drug resistance mutations. J Infect Dis, 2013; 208:1459–1463.

11.

Descamps

, Assoumou

, Chaix

, Chaillon

, Pakianather

, De

, et al.: National sentinel surveillance of transmitted drug resistance in antiretroviral-naive chronically HIV-infected patients in France over a decade: 2001–2011. J Antimicrob Chemother, 2013; 68:2626–2631.

12.

Reis

, Alca

: Polymorphisms in the HIV-1 gp41 env gene, natural resistance to enfuvirtide (T-20) and pol resistance among pregnant Brazilian women. J Med Virol, 2014; 86:8–17.

13.

Johnson

, Li

, Wei

, Lipscomb

, Bennett

, Brant

, et al.: Simple PCR assays improve the sensitivity of HIV-1 subtype B drug resistance testing and allow linking of resistance mutations. PLoS One, 2007; 2:e638.

14.

Panos

, Charatsis

, Paparizos

, Kazantzi

, Falagas

: Prevalence of genotypic resistance to nucleoside analogues, and protease inhibitors in HIV-infected persons in Athens, Greece. AIDS Res Hum Retroviruses, 2008; 24:43–51.

15.

Karim

, Banegura

, Cahn

, Christie

CDC

, Dintruff

, Distel

, et al.: Asking the right questions: Developing evidence- based strategies for treating HIV in women and children. BMC Public Health, 2011; 11:388.

16.

Pérez

, Vignoles

, Laufer

, Gómez

, Coll

, Lattner

, Rolón

, Salomón

, Cahn

: Short communication. Low rate of emergence of nevirapine and lamivudine resistance after post-partum interruption of a triple-drug regimen. Antivir Ther, 2008; 13:135–139.

17.

Johnson

, Li

, Wei

, Lipscomb

, Irlbeck

, Craig

, et al.: Minority HIV-1 drug resistance mutations are present in antiretroviral treatment—naive populations and associate with reduced treatment efficacy. PLos Med, 2008; 5:e158.

18.

Deeks

: Transmitted minority drug-resistant HIV variants: A new epidemic?. PLoS Med, 2008; 5:e164.

19.

Gianella

, Richman

: Minority variants of drug-resistant HIV. J Infect Dis, 2010; 202:657–666.

20.

Bellecave

, Recordon-Pinson

, Papuchon

, Vandenhende

M-A

, Reigadas

, Tauzin

, et al.: Detection of low-frequency HIV type 1 reverse transcriptase drug resistance mutations by ultradeep sequencing in naive HIV type 1-infected individuals. AIDS Res Hum Retroviruses, 2014; 30:170–173.

21.

Iyer

, Casey

, Bouzek

, Kim

, Deng

, Larsen

, et al.: Comparison of major and minor viral SNPs identified through single template sequencing and pyrosequencing in acute HIV-1 infection. J Clin Virol, 2015; 70:95–100.

22.

Fonager

, Larsson

, Hussing

, Engsig

, Nielsen

, Fischer

: Identification of minority resistance mutations in the HIV-1 integrase coding region using next generation sequencing. J Clin Virol, 2015; 73:95–100.

23.

Todesco

, Rodriguez

, Morand-Joubert

, Mercier-Darty

, Desire

, Wirden

, et al.: Improved detection of resistance at failure to a tenofovir, emtricitabine and efavirenz regimen by ultradeep sequencing. J Antimicrob Chemother, 2014; 70:1503–1506.

24.

Lin

, Kuritzkes

: NIH public access. Curr Opin HIV AIDS, 2009; 4:481–487.

25.

Sánchez

, Masiá

, Robledano

, Padilla

, Ramos

, Gutiérrez

: Performance of genotypic algorithms for predicting HIV-1 tropism measured against the enhanced-sensitivity Trofile coreceptor tropism assay. J Clin Microbiol, 2010; 48:4135–4139.

26.

Svicher

, Arrigo

, Alteri

, Andreoni

, Angarano

, Antinori

, et al.: Performance of genotypic tropism testing in clinical practice using the enhanced sensitivity version of Trofile as reference assay: Results from the OSCAR study group. New Microbiol, 2010; 33:195–206.

27.

Shapiro

, Biezuner

, Linnarsson

: Single-cell sequencing-based technologies will revolutionize whole-organism science. Nat Rev Genet, 2013; 14:618–630.

28.

Lipscomb

, Switzer

, Li

, Masciotra

, Michele

, Johnson

, et al.: HIV reverse transcriptase drug resistance mutations during early infection reveal greater transmission diversity than in Env. J Infect Dis, 2014; 210:1827–1837.

29.

Antoniadou

, Hezka

, Kousiappa

, Mamais

, Skoura

: Short communication. Cellular HIV-1 DNA levels are equivalent among drug sensitive and drug resistant strains in newly diagnosed and antiretroviral naive patients. AIDS Res Hum Retroviruses, 2014; 30:266–271.

30.

Paryan

, Mohammadi-Yeganeh

, Mirab Samiee

, Rezvan

: Design and development of a multiplex real-time PCR assay for detection of HIV-1 and HCV using molecular beacons. Indian J Microbiol, 2012; 52:456–463.

31.

Jalava

, Rintala

, Ollgren

, Maunula

, Gomez-Alvarez

, Revez

, et al.: Novel microbiological and spatial statistical methods to improve strength of epidemiological evidence in a community-wide waterborne outbreak. PLoS One, 2014; 9:e104713.

32.

Harris

, Marston

, Sangrujee

, DuPlessis

, Park

: Cost-effectiveness analysis of diagnostic options for pneumocystis pneumonia (PCP). PLoS One, 2011; 6:e23158.

33.

Gumede

, Lentsoane

, Burns

, Pallansch

, de Gourville

, Yogolelo

, et al.: Emergence of vaccine-derived polioviruses, Democratic Republic of Congo, 2004–2011. Emerg Infect Dis, 2013; 19:1583–1589.

34.

Bhargava

, Cajas

, Wainberg

, Klein

, Pai

: Review article. Do HIV-1 non-B subtypes differentially impact resistance mutations and clinical disease progression in treated populations? Evidence from a systematic review. J Int AIDS Soc, 2014; 17:18944.

35.

Boullosa

, Bachu

, Bila

, Ranga

, Süffert

, Sasazawa

, et al.: Genetic diversity in HIV-1 subtype C LTR from Brazil and Mozambique generates new transcription factor-binding sites. Viruses, 2014; 6:2495–2504.

36.

Siemieniuk

RAC

, Beckthold

, Gill

, Chb

, Siemieniuk

RAC

, Beckthold

, et al.: Increasing HIV subtype diversity and its clinical implications in a sentinel North American population. Can J Infect Dis Med Microbiol, 2013; 24:69–73.

37.

Cushing

, Flaherty

, Hopmans

, Bell

, Ji

: RVD: A command-line program for ultrasensitive rare single nucleotide variant detection using targeted next-generation DNA resequencing. BMC Res Notes, 2013; 6:206.

38.

Ekici

, Rao

, Sönnerborg

, Ramprasad

, Gupta

, Neogi

: Cost-efficient HIV-1 drug resistance surveillance using multiplexed high-throughput amplicon sequencing: Implications for use in low- and middle-income countries. J Antimicrob Chemother, 2014; 69:3349–3355.

39.

Luthra

, Patel

, Reddy

, Haghshenas

, Routbort

, Harmon

, et al.: Next-generation sequencing-based multigene mutational screening for acute myeloid leukemia using MiSeq: Applicability for diagnostics and disease monitoring. Haematologica, 2014; 99:465–473.

40.

Quail

, Smith

, Coupland

, Otto

, Harris

, Connor

, et al.: A tale of three next generation sequencing platforms: Comparison of Ion Torrent, Pacific Biosciences and Illumina MiSeq sequencers. BMC Genomics, 2012; 13:341.

41.

Newcombe

: Interval estimation for the difference between independent proportions: Comparison of eleven methods. Stat Med, 1998; 17:873–890.

42.

Brasil Ministério da Saúde. Secretaria de Vigilância em Saúde. Departamento de Vigilância Epidemiológica. Recomendações para Terapia Anti–retroviral em Adultos Infectados pelo. HIV, 2008; 7Ed:34–51.

43.

Hauser

, Mugenyi

, Rose

, Bluethgen

, Harms

, Kunz

, et al.: Detection and quantification of minor HIV-1 variants harbouring K103N and Y181C resistance mutations in subtype A and D isolates by allele-specific real-time PCR. Antimicrob Agents Chemother, 2009; 53:2965–2973.

44.

Delobel

, Saliou

, Nicot

, Dubois

, Trancart

, Tangre

, et al.: Minor HIV-1 variants with the K103N resistance mutation during intermittent efavirenz-containing antiretroviral therapy and virological failure. PLoS One, 2011; 6:e21655.

45.

Persaud

, Bedri

, Ziemniak

, Moorthy

, Gudetta

, Abashawl

, et al.: Slower clearance of nevirapine resistant virus in infants failing extended nevirapine prophylaxis for prevention of mother-to-child HIV transmission. AIDS Res Hum Retroviruses, 2011; 27:823–829.

46.

Kuhn

, Hunt

, Technau

K-G

, Coovadia

, Ledwaba

, Pickerill

, et al.: Drug resistance among newly diagnosed HIV-infected children in the era of more efficacious antiretroviral prophylaxis. AIDS, 2014; 28:1673–1678.

47.

Holguín

, Yebra

, Martín

, de Pineda

, Ruiz

, Quezada

, Nieto

, Escobar

: Transmitted drug-resistance in human immunodeficiency virus-infected adult population in El Salvador, Central America. Clin Microbiol Infect, 2013; 19:E523–E532.

48.

Lel

, Ngaira

, Lihana

, Khamadi

: HIV-1 drug resistance mutations among infants born to HIV-positive mothers in Busia, Kenya. AIDS Res Hum Retroviruses, 2014; 30:11–13.

49.

Johnson

, Li

, Morris

, Martinson

, Gray

, Mcintyre

, et al.: Emergence of drug-resistant HIV-1 after intrapartum administration of single-dose nevirapine is substantially underestimated. J Infect Dsi, 2005; 192:16–23.

50.

Vandenhende

M-A

, Bellecave

, Recordon-Pinson

, Reigadas

, Bidet

, Bruyand

, et al.: Prevalence and evolution of low frequency HIV drug resistance mutations detected by ultra deep sequencing in patients experiencing first line antiretroviral therapy failure. PLoS One, 2014; 9:e86771.

51.

Aghokeng

, Monleau

, Eymard-Duvernay

, Dagnra

, Kania

, Ngo-Giang-Huong

, et al.: Extraordinary heterogeneity of virological outcomes in patients receiving highly antiretroviral therapy and monitored with the World Health Organization public health approach in sub-saharan Africa and southeast Asia. Clin Infect Dis, 2014; 58:99–109.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.02 MB

Development of a Qualitative Quantitative Polymerase Chain Reaction Test to Identify Patients Failing First-Line Therapy to Non-Nucleotide Reverse Transcriptase Inhibitor

Abstract

Introduction

Materials and Methods

Preparation of virus template

PMqPCRK103N

PMqPCRK103N validation

Illumina MiSeq sequencing

Illumina MiSeq data analysis

Statistical analysis

Results

Discussion

Conclusions

Footnotes

Acknowledgments

Author Disclosure Statement

References

Supplementary Material

PMqPCR_K103N

PMqPCR_K103N validation