Human Immunodeficiency Virus Resistance Testing Technologies and Their Applicability in Resource-Limited Settings of Africa

Clinical indications for HIV drug resistance testing

Investigating drug resistance is routinely prescribed by the physician as part of the primary care received by newly enrolled persons for clinical care of HIV infection. ⁷ Resistance testing should be done irrespective of the patient’s choice to either commence therapy with ARVs immediately or delay treatment. ⁸ If the choice is to delay treatment, resistance testing may be repeated consequently when consumption of ARVs is resumed. The results of primary HIV resistance testing can comparatively detect whether initial infection was with a drug-resistant strain of HIV or not. Drug resistance testing results are used to decide which ARVs to include in a person’s first treatment regimen. ⁸ After treatment is resumed, secondary drug resistance testing is required if HIV VL results are indicative that the HIV regimen is poorly controlling viral proliferation and informs the physicians choice to prescribe second-line ARV drugs. Drug resistance testing is also recommended for all pregnant women living with HIV prior to commencing therapy on and also in some pregnant women already on HIV medications. ⁸

Resistance testing is requested in cases of virologic failure when a treatment change is being considered. Laboratory investigations for such cases should be characterized by (1) a confirmed detectable VL from 2 independent consecutive samples collected after an undetectable measurement, (2) a steady or increasing VL after the initiation of medications (HIV RNA levels >1000 copies/mL), and (3) detectable VL 6 months after treatment initiation. In persons with HIV RNA levels >500 copies/mL, but <1000 copies/mL, drug resistance testing is usually unnecessary because of low success rates recorded but should still be considered. ⁹ In such cases of treatment failure, resistance testing is requested on the premise that the patient is consistently taking his medication at the time of the sampling. If the sample is collected without the presence of selective pressure, the chances are high that wild-type strain of the virus will be detected instead of the mutant. ⁹

Separate studies conducted in Canada and Italy developed a study design aimed at examining the reliability and clinical utility of standard GRTs in cases of low-level viraemia – a VL between 50 and 999 copies/mL.^10,11 The retrospective study involving 4915 samples collected from 2492 people who received care in British Columbia between 1996 and 2012 reported an 88% occurrence of drug resistance in tests conducted when VL was below 1000 copies/mL. ¹⁰ Successful results were obtained from three-quarters of samples with VLs below 250 copies/mL and from 90% of samples where VL was above 250 copies/mL. ¹⁰

The study also examined the use of HIV resistance tests in a subset of 196 people taking their first ARV combination that had experienced low-level viraemia and had no record of a baseline resistance. At the time of resistance testing, the patients had a median CD4 count of 415 cells/mm³, with a median VL of 374 copies/mL. Approximately a fifth (19%) of the patients under study developed resistance during their time with low-level viraemia. Only 5% of patients with a VL between 50 and 250 copies/mL developed resistance, compared with 30% of individuals with VLs between 750 and 999 copies/mL. ¹⁰

Utility and efficiency of dried blood spots as an alternative method of specimen collection for HIV genotyping

Dried blood spots (DBSs) are increasingly becoming appealing as alternative specimen collection format for HIV genotyping especially in areas where low-cost diagnostic techniques are desired. They are easy to obtain, less stressful on the patient, easy to process and transport, and require simplified storage conditions.^12,13 Plasma HIV genotyping detects the RNA contained within the cell-free circulating virus. However, template material present in DBSs consists of RNA from circulating virus and DNA from the cell-associated, integrated provirus. ¹⁴

Studies have shown that genotyping results from DBS are highly informative and efficiently equivalent to the results obtained when using plasma for conventional sequencing methods. ¹⁴ Recent studies have demonstrated good correlation of results obtained from DBS and plasma samples in resource-limited settings such as Malawi, Kenya, Nigeria, and Senegal. ¹⁵ In contrast, some studies have demonstrated that concordance of genotyping results in plasma and DBS is not absolute and variation may be accorded to certain factors such as the VL, the duration an individual has been infected with HIV, and ART status. It has been noted in a previous study that the plasma/DBS concordance is the highest when VL is ≥5000 copies/mL and/or the patient has not been treated with ARTs and/or the person got infected with HIV within ≤2 years. ¹⁴

Genetic mutations in the genome of HIV

DNA viruses unlike RNA viruses exhibit proofreading mechanisms which enable them retain their original genetic composition during replication. However, an RNA virus such as HIV, due to the lack of a proofreading exonuclease, will have each new copied genome differing from the parent virus, by at least a single nucleotide. ¹⁶ This repair inability of HIV RT confers a high mutational rate to the virus (approximately 5-10 nucleotide mis-incorporations per genome per viral cycle). ¹⁷ For some ARV drugs, such as lamivudine and certain NNRTIs (eg, nevirapine), a single mutation can confer high-level resistance. When these drugs are given in combinations to partially suppress viral replication, drug-resistant mutants predominate within weeks. ¹⁸

HIV unchecked by ARV drugs in an infected person

Due to its large progeny, HIV can replicate up to a billion copies of new viral particles per day. ¹⁹ When combined, both factors (genetic mutations in the genome of HIV and HIV unchecked by ARV drugs in an infected person) promote increased viral plasticity, leading to the presence of viral quasi-species within patients, frequent escape from the host’s immune system, and the emergence of resistant viral strains under drug selective pressure. ²⁰

Some other factors that may hinder the efficacy of ARTs in patients include individual differences in pharmacokinetics, several adverse drug reactions, limited drug absorption and penetration into reservoirs of viral multiplication, and coinfections with other pathogens. In a study, multidrug-resistant HIV-1 highlighted a 3-fold increased risk of death following the diagnosis of multidrug resistance, compared with the risk observed overall in HIV-infected individuals. ²¹

Increased provision of ART has led to the emergence of therapeutic failure and acquired drug-resistant HIV in patients who will require more expensive advanced lines of ART to control the disease. Along with increased and quicker access to laboratory-based VL monitoring, feasible and cost-effective technologies to detect drug-resistant HIV could help maximize the duration of use of available drug regimens. Current existing affordable technologies include in-house Sanger and NGS in centralized laboratories, as well as PMAs and genotype-free systems that predict response to ART at point of care. ⁶

In sub-Saharan Africa, which is home to some of the world’s poorest countries, a substantial number of adults and children receiving first-line ART are likely to develop treatment failure within the first 5 years on the drug regimen. ²² Among those patients on first-line drugs, an estimated 70% to 90% acquire drug-resistant HIV. ²³ In sub-Saharan Africa, a fewer than 1 million people are on second-line ART regimens but numbers are estimated to increase to about 4 to 6 million people by 2030. ²⁴ In future, there will be a demand for third-line and other advanced therapies. Low-income and middle-income countries are currently facing challenges managing the growing numbers of patients requiring second, third, and more advanced lines of therapy, driven by the limited availability of ARV drugs, diagnostic technologies, and clinical expertise. In contrast, high-income countries adopt efficient guidelines which ensure the availability of GRTs which promote profound response of adequate ART regimen in patients with failure to ARV drugs. ²⁵

Sanger sequencing

This technique, also known as dideoxy sequencing or chain termination, is based on the use of dideoxynucleotides (ddNTPs) which act as DNA chain terminators in addition to the usual nucleotides (NTPs) found in DNA. Dideoxynucleotides share similar features with NTPs; except that instead of having a hydroxyl group (OH), they possess a hydrogen group on the 3′ carbon. These modified NTPs, when integrated into a sequence, avoid the addition of further NTPs. This occurs because a phosphodiester bond cannot form between the ddNTP and the next incoming nucleotide, and thus the DNA chain is terminated. ²⁹

This approach has been modified using automated sequencing to accommodate the sequencing of more DNA within a shorter period. This advancement enables all 4 ddNTPs to react in a tube with different colour dyes. Consequently, the DNA is allowed to separate only on 1 lane. Because the 4 dyes fluoresce at different wavelengths, a laser is used to read the gel to determine the identity of each band based on its unique wavelength. Data are displayed in the form of a chromatogram which indicates coloured peaks that correspond to the nucleotide in that sequence location. ³⁰

This modification has paved way for generation of long individual reads. These assays have been widely validated as some have been developed as kit-based for the detection of HIV subtypes. This technology is commonly available in several African countries where kits are used with the DBS. Unfortunately, this technology is limited in throughput and requires complex and expensive equipment which is sparsely available. Furthermore, this technique is not suited for parallel investigation, lacks the ability to detect minor variants (<20%), requires high capital cost for operation, and is impractical for massive sequencing projects. ⁶

The high cost can be reduced using open-source reagents, decreasing the number of primers, 1-step amplification,^31,32 negotiations with suppliers of the technology, collaboration with non-profit manufacturers, sequencing just reverse transcriptase for failed treatment with first-line ART drugs, and sample pooling. ³³

Next-generation sequencing

Due to the high demand for affordable sequencing, the NGS technology was developed with higher high throughput and the ability to parallelize the sequencing process of millions of DNA fragments in a run, which can be used in assaying for resistant strains of HIV to ART. High-throughput sequencing technologies were intended to drastically reduce sequencing costs per genome. Companies involved in the production of NGS include Pacific Biosciences, Oxford Nanopore Technologies, Illumina, Helicos Biosciences, Danaher Motion Cooperation, Applied Biosystems, and 454 Life Sciences. ²⁹

Although equipment costs are still high (>US $100 000), prices are estimated to fall as technological simplifications occur. Despite the high cost per run, the cost of each test is affordable. Higher levels of automation can be achieved through multiplex sequencing, where several primer-barcoded samples are pooled using 24-multiplex (Illumina MiSeq 50; Illumina, San Diego, CA, USA) or 48-multiplex (Roche 454, Penzberg, Germany) systems. Moreover, the potential for full-genome sequencing with minimal additional cost could make NGS a preferred option for evaluation of resistance in patients on PI-based therapy because polymorphisms in group-specific antigen (gag) protein and envelope (env) gene regions could induce PI resistance. ³⁴

Full-genome sequencing also minimizes the need for separate assays for the integrase gene region as the use of integrase inhibitors is anticipated to increase. Further advantages of NGS include the opportunities for simultaneous detection of other pathogens and use with DBSs and broad subtype primers. ³⁵

Unlike Sanger sequencing, multiplex sequencing which is adopted by Illumina MiSeq and Roche 454 systems is more affordable and can generate high throughput. ³⁶ These systems enable full-genome sequencing to be achieved with minimal cost for the assessment of patients failing PI-based ART. ³⁷ Besides HIV, other pathogens can be detected in DBSs using broad subtype primers in NGS. A previous study which compared Illumina and Roche 454 in patients failing raltegravir-based ART observed that Illumina demonstrated larger depth of coverage and increased sensitivity for HIV minority variants as well as lesser false-positive variant calls. ³⁵

The demerits of NGS include high level of expertise and high cost of bioinformatics support needed for data analysis. Ion torrent sequencing technology has been observed to be prone to insertion-deletion errors at homopolymer regions, whereas Illumina sequencing technology is prone to substitution errors and sequence coverage that is not equal. ³⁸

Point mutation assays

These assays provide a means for the investigation of the genetic composition of human as it combines laboratory medicine with molecular genetics for the development of DNA/RNA-based techniques for assessing pathologic conditions. Several methods have been used in the detection of mutation and these could be classified into methods for known, unknown mutations, and both known and unknown mutations. ³⁹

To successfully select a suitable method for the assessment of the point mutation, certain criteria are considered which include the type of nucleic acid, sample type, number of mutations, and the reliability of the method. ³⁹ Methods used for detecting known mutation include restriction fragment length polymorphism, amplification refractory mutation system polymerase chain reaction (PCR), multiplex PCR, nested PCR, real-time PCR, allele-specific amplification, allele-specific oligonucleotide, and single-nucleotide primer extension. Methods for detecting unknown mutation include denaturing gradient gel electrophoresis as well as temperature gradient gel electrophoresis, single-strand conformational polymorphism, hetero-duplex analysis, protein truncation test, and DNA chips. Methods for detecting both known and unknown mutations include reverse transcription-polymerase chain reaction, multiplex ligation-dependent probe amplification, chemical cleavage of mismatch, and oligonucleotide ligation assay. ³⁹

A PMA that can successfully detect 6 resistance mutations in RT successfully detected resistance in 98.8% of patients failing first-line and 61.2% of previously untreated patients. ⁴⁰ The main problem with the PMA technique is the substantial HIV-1 sequence variability at and surrounding each drug resistance mutation target, which means that multiple experiments are needed per target site. The requirement for multiple tests per site increases the costs of PMAs, especially without affordable multiplexing assays. Therefore, development and validation of low-cost multiplex PMAs might lag behind current clinical practice. ⁴⁰

Web-based genotype-free prediction systems

This technology uses computational models to generate predictions for response of viruses to ART regimen based on archived databases of ART-managed patients. The processes of this approach involve the ranking of a set of suitable drugs and selection of the most suitable combination of these drugs for a particular patient. These predications are based on inputted clinical and demographical data which include most recent VL, CD4 cell count, and history of previous treatment. ⁴¹ This technology has the potential to improve clinical decision making by health care providers who are not required to attain high level of professional expertise. ⁶

The HIV Treatment Response Prediction System (HIV-TRePS) had an overall accuracy of 78% (Table 1) in prediction of virologic response in independent testing ⁴² compared with 57% to 60% achieved by GRT with rules-based interpretation for the same cases. The HIV-TRePS, available as a free online service, has the potential to enhance clinical decision making of ART providers with relatively little training. ⁴¹

OPEN IN VIEWER

Table 1.

Performance of human immunodeficiency virus resistance testing technologies.

Technology		Instrument cost, US $	Read length, bp	Accuracy, %	Sensitivity, %	Specificity, %	Reads per run	Time per run	Through-put, Mb/h	Cost per million bp, US $
Sanger-based GRT		NA	400-900	99.9	>99.59	100	NA	20-180 min	0.166	2400
NGS	SMRTS (Pacific Biosciences)	695 000	2900	99	97.0	98.0	35 000-75 000	30-120 min		2
	Ion torrent sequencing	80 000	200	98	99.61	99.98	Up to 5 million	120 min		1
	Pyrosequencing (454)	500 000	700	99.9	94.3	97.8	1 million	1440 min	20-30	10
	Illumina versions	128 000-654 000	50-250	98	98.12-99.15	85	Up to 3 million	1-10 d	20	0.05 to 0.15
	SOLiD sequencing	495 000	50 + 35 or 50 + 50	99.9	92.5-96.8	NA	1.2-1.4 million	1-2 wk	5-15	0.13
PMA	PCR versions	NA	NA	NA	>99.5	72	NA
WGPS		NA	NA	78	65	80	NA

Abbreviations: BP, base pairs; GRT, genotypic resistance test; NA, not available; NGS, next-generation sequencing; PCR, polymerase chain reaction; PMA, point mutation assay; WGPS, Web-based genotype-free prediction system.

Limitations of HIV resistance testing

The correlation between the development of drug resistance and the occurrence of clinical failure is a complex one. This is because the development drug resistance is not the sole cause of treatment failure. ⁵² Other factors have been identified as leading causes are clinical and virologic, notable among them are non-adherence to treatment regimens, the use of insufficiently potent treatment regimens, pharmacokinetic factors that decrease the efficacy levels of one or more drugs in a treatment regimen, and several others. Conversely, a drug may have some benefits even in the setting of resistance because of some, although limited, residual effects of some drugs and because many drug-resistant variants have a reduced replication capacity compared with drug-susceptible variants. ⁵³

Another important limitation of HIV resistance testing is the unreliability of the tests to detect minor genotypic HIV-1 variants. ⁵⁴ This limitation can complicate the interpretation of resistance tests particularly in patients with complicated treatment histories. Therefore, treatment and resistance history always have to be taken into account. HIV-1 exists in each infected individual as a complex of quasi-species in which many different subpopulations of drug-resistant variants co-exist. ⁵⁵ Also subpopulations that are no longer actively replicating will remain archived in latently infected cells. The complexity of these quasi-species may influence the success of therapy in ways that cannot be predicted by any single drug resistance test. ⁵⁵