Overview of HIV drug resistance testing assays

INTRODUCTION — The use of drug resistance testing has become an integral part of HIV clinical care. The first clinical description of HIV resistance to an antiretroviral agent was published in 1989, when patients taking zidovudine monotherapy accumulated mutations within the reverse transcriptase gene, resulting in a marked increase in drug resistance [1]. Subsequently, HIV variants resistant to every available antiretroviral agent have been identified. The evolution of drug resistance has significant clinical implications for choosing effective antiretroviral regimens.

This topic will provide an overview of HIV drug resistance testing. The interpretation of these tests and the approach to selecting an antiretroviral therapy regimen for patients with drug resistance mutations are discussed elsewhere. (See "Interpretation of HIV drug resistance testing" and "Selecting an antiretroviral regimen for treatment-experienced patients with HIV who are failing therapy" and "Evaluation of the treatment-experienced patient failing HIV therapy".)

RATIONALE FOR HIV SUSCEPTIBILITY TESTING — The use of HIV susceptibility testing to guide therapy leads to better viral suppression and has been associated with improved survival [2-5]. As an example, in an observational study of more than 2699 HIV-infected patients who were eligible for genotypic and phenotypic testing between 1999 and 2005, resistance testing was associated with improved survival (adjusted hazard ratio, 0.69 [95% CI 0.51-0.94]) after controlling for demographics, CD4 cell count, HIV RNA level, and intensity of clinical follow-up [3].

GENOTYPIC VERSUS PHENOTYPIC ASSAYS — This section will provide an overview of the commonly used resistance assays. The use of resistance testing when selecting an antiretroviral regimen is discussed elsewhere. (See "Selecting antiretroviral regimens for treatment-naïve persons with HIV-1: General approach", section on 'Considerations prior to initiating treatment' and "Evaluation of the treatment-experienced patient failing HIV therapy".)

Overview of the assays — Resistance assays can be categorized as either genotypic or phenotypic. These assays detect resistance in fundamentally different ways, although the results generally correlate with each other.

●Genotypic resistance assays detect the presence of specific drug resistance mutations in the regions of the HIV genome encoding protease, reverse transcriptase, and integrase. Results are reported as the individual mutations (eg, M184V, the signature mutation for lamivudine resistance) with comments such as "susceptible," "possibly resistant," or "resistant" for each antiretroviral agent. (See "Interpretation of HIV drug resistance testing".)

●Phenotypic resistance assays measure the extent to which an antiretroviral drug inhibits virus replication in vitro. The susceptibility that is measured with phenotypic assays is the aggregate of the acquired drug mutations in a patient's viral strain [6]. Similar to bacteriologic methods, this is typically performed by demonstrating an increase in the inhibitory concentration (IC) that is required to inhibit in vitro growth by 50 percent (IC50) compared with virus replication in the absence of drug. Results are reported as a fold-change in drug susceptibility of the patient sample compared with a laboratory reference strain without resistance.

Advantages and limitations — In most patients, genotypic assays are preferred. Genotypic assays provide information on the specific resistance mutations present within the virus. Additional advantages of genotypic relative to phenotypic testing include lower cost and shorter turnaround time. These assays may also be more sensitive for detection of "mixtures" of resistant and wild-type viruses. HIV treatment guidelines recommend a genotypic assay be the preferred testing method in treatment-naïve patients [4,7-10]. (See 'Genotypic resistance assays' below.)

However, in treatment-experienced patients with multiple resistance mutations, interpreting complex genotypes can be challenging since it can be difficult to interpret the clinical significance of a specific mutation when many mutations are present. As an example, some mutations cause resistance to certain drugs but increase susceptibility to others, while other mutations may impact viral fitness (see "Interpretation of HIV drug resistance testing", section on 'Viral fitness'). For such patients, phenotypic resistance testing may have an advantage over genotypic testing, as it measures resistance more directly and can assess relative susceptibility and interactions among mutations. This includes mutations that may not have been recognized to date as causing resistance, or combinations of mutations that may have a different effect on susceptibility compared with each individual mutation alone.

Both genotypic and phenotypic assays are limited by insensitivity to minority variants comprising 1 to 20 percent of the virus population [11,12]. This limitation is important because once antiretroviral therapy (ART) is discontinued, a wild-type virus may reemerge to replace the drug-resistant virus as soon as four to six weeks after therapy is stopped. Thus, absence of a detectable resistance mutation must be interpreted with caution in a patient who has recently discontinued ART. Other techniques have been developed to improve the detection of these minority species compared with standard drug resistance assays. (See 'Low-abundance drug-resistant variants' below.)

Genotypic resistance assays

Available assays — Commercially available tests include United States Food and Drug Administration (FDA)-approved kits, as well as a variety of in-house assays performed by reference laboratories. These assays sequence regions of the HIV genome (eg protease, reverse transcriptase, and integrase genes) that have been polymerase chain reaction-amplified from the viral quasispecies circulating in a patient's plasma. Based on when genotype tests were first introduced, standard tests frequently include testing for nucleoside reverse transcriptase inhibitors (NRTI), non-nucleoside reverse transcriptase inhibitor (NNRTI), and protease inhibitor (PI) resistance. If a clinician is interested in identifying resistance to integrase inhibitors or the fusion inhibitor enfuvirtide, these tests usually must be requested separately.

Genotypic definitions of resistance — Defining resistance using genotypic assays is generally more complex than for phenotypic assays. A mutation is generally identified as playing a role in resistance to a given drug if one or more of the following conditions are met:

●The mutation confers phenotypic resistance when introduced into a drug-sensitive laboratory strain of HIV.

●The mutation is selected for during serial in vitro passage of virus in the presence of drug.

●The mutation is selected for during clinical therapy with that drug.

●The presence of the mutation in clinical isolates is associated with phenotypic resistance and virologic failure.

Interpreting genotypic assays — There are currently two general approaches to interpreting genotypic resistance testing:

●Expert interpretation, in which data concerning the association of a mutation with drug resistance are evaluated and synthesized. Over 20 rules-based genotypic interpretation systems have been proposed [13]; however, there is good correlation among the various genotypic interpretation systems [14].

●Databases, in which genotype is correlated with phenotype [4,14].

Collaborative groups have been formed to ensure interpretation of genotypic mutational patterns remains up to date as new data emerge and the information reflects agents used in the clinic [14].

There are insufficient data to recommend one type of interpretation over another. Evidence suggests that many of these systems have a similar ability to predict virologic outcome [15,16]. The following summarizes some of the specific genotypic resistance testing interpretations that are available to clinicians and investigators. This summary provides only a few examples and is not intended to be a comprehensive list of the different approaches that are available.

●Genotypic resistance assay reports typically list the specific drug resistance mutations detected in the test sample, compared with a reference sequence. In addition, each report summarizes which drugs may have reduced activity based upon the viral sequence. These interpretations are made using algorithms defined by experts in the field of resistance testing, based upon a synthesis of in vitro and clinical data. No direct prospective comparisons of the genotypic interpretations using these two methods have been made.

●The International AIDS Society-USA provides an updated list of mutations associated with HIV drug resistance to all FDA-approved drugs, based on evaluations by a panel of experts (www.iasusa.org) [17].

●The Stanford drug resistance database (hivdb.stanford.edu) provides access to a genotypic resistance algorithm, which assigns a drug penalty score for each resistance mutation, based on published studies of clinical outcome and expert opinion [18]. Some information on expected fold-change in IC50 can be obtained by searching the database, which is drawn from published studies and some unpublished clinical trials data.

●Agence Nationale de Recherches sur le SIDA (ANRS) [19], and Rega [20], in addition to the Stanford HIV database [15,21], are the publicly available drug resistance interpretation systems that are most commonly used by HIV providers. The algorithms for these systems are transparent, updated regularly, and are free to the clinician. These systems are comparable in their ability to predict virologic outcome [15,16].

Caution should be used when interpreting resistance testing to investigational or recently approved antiretroviral agents. In addition, if a genotype was interpreted in the past using older algorithms, it should be reinterpreted using updated algorithms with new information as resistance interpretations can change over time [22].

Phenotypic resistance assays — HIV phenotypic resistance testing is only offered through reference laboratories in the United States, and only the PhenoSense assay (LabCorp, Burlington, NC) is available. An older assay, entitled Antivirogram (Janssen Diagnostics, Belgium), was withdrawn from the market due to a decline in demand. However, a clinician may still encounter an old Antivirogram test result when reviewing a patient's prior drug resistance history.

Recombinant virus assay — Phenotype testing is performed using recombinant virus assays (RVAs) [23]. These replaced the slow, labor-intensive cell culture method initially used [24] and has allowed phenotyping to be done on a larger scale.

Available RVAs use recombinant (chimeric) viruses composed of protease, reverse transcriptase, and integrase gene sequences from viruses circulating in a patient's plasma, which are inserted into the genetic backbone of a laboratory reference strain of HIV. The resulting chimeric DNA clone is transfected into mammalian cells to produce an infectious virus that expresses the protease and reverse transcriptase enzymes from the patient HIV strain. The drug susceptibility of this chimeric virus can then be measured in a phenotypic resistance assay.

RVAs accurately reflect the drug susceptibility of the original clinical isolate only if the region of the viral genome being amplified contains the genetic determinants (mutant sequences) of drug resistance. Modifications of the RVA are also available for fusion and CCR5 inhibitors [25-29].

Phenotypic definitions of resistance — Current phenotypic assays measure the ability of a laboratory virus that contains a specific viral gene from the patient sample to grow in the presence of different concentrations of an antiretroviral drug. The ability to define what constitutes the threshold for clinical resistance is a difficult process as it includes both assay and biologic factors.

The cutoffs for clinical resistance would be best verified by clinical trials; however, this has not been performed for many agents. Thus, what constitutes significant clinical phenotypic resistance must often be inferred from decreases in susceptibility of the patient sample from that of a known wild-type HIV strain (usually the drug concentration it takes to inhibit viral replication by 50 percent, expressed in fold-increase in IC50). Clinical cutoffs are often calculated by examining the fold-change in susceptibility of a patient's virus and the clinical treatment response. When these studies are performed they also correlate the fold-change in susceptibility with the mutations present in the virus.

Assay variability — When validating an assay, studies are done to define the inter-assay and inter-laboratory assessment of assay repeatability and robustness. Earlier definitions of resistance for phenotypic assays were based on the technical reproducibility of the assay and did not necessarily reflect clinically relevant definitions of resistance [30].

Biologic variability — Resistance thresholds based upon biologic variability are usually defined as a level that exceeds the mean fold-change in IC50 of HIV isolates obtained from treatment-naive patients by a certain factor. This definition of resistance is dependent upon the population chosen for study and assumes that all treatment-naive patients will respond equally well to that drug.

Virologic outcome — Resistance thresholds based upon virologic outcome are considered to be a "gold standard," but can be difficult to define. There is no consensus yet on the appropriate magnitude or timing of the virologic response that should be used to define these thresholds; such thresholds can be influenced by the activity of the background regimen (other antiretroviral agents used in the combination). The PhenoSense assay lists suggested clinical cutoffs for the antiretroviral agents based on clinical outcome data.

Studies performed to date suggest that virologic response may vary continuously with baseline resistance. For example, in one study (JAGUAR study), didanosine (ddI) or placebo was added to a failing ART regimen, and virologic outcomes at four weeks were compared with phenotypic testing results using the PhenoSense assay [31]. The proportion of virologic responders was 83 percent (15 of 18) for patients with a ddI fold-change <1.3, 50 percent (33 of 66) for patients with a fold-change of 1.3 to 2.2, and 29 percent (4 of 14) for patients with a fold-change of >2.2. Although these data need to be validated in larger populations, they suggest that a single resistance cutoff that sensitively and specifically predicts virologic failure in all patients may be difficult to define.

ADDITIONAL RESISTANCE ASSAYS

Low-abundance drug-resistant variants — Genotypic and phenotypic assays are typically unable to detect low-abundance drug-resistant variants that are present at less than 20 percent of the viral quasispecies in a given sample. Other techniques have been developed to improve the detection of these minority species compared with standard drug resistance assays [12,32]. The presence of low frequency-resistant variants can be associated with inferior treatment responses [12,33-35].

Point mutation assays — Point mutation assays depend on the differential hybridization of oligonucleotide probes to the wild-type virus and mutant variants [36]. These assays must be individually tailored for each mutation they are designed to detect. False-positive or false-negative test results can occur due to binding site variability [32]. However, the low cost of this type of assay makes it a good candidate for drug resistance epidemiology studies, where information may be needed only on the most common resistance mutations.

Clonal sequencing assays — Clonal sequencing assays can detect minority species, but are labor intensive; approximately 100 to 300 clones must be sequenced in order to identify a viral variant present in only 1 or 2 percent of the viral quasispecies [37].

Deep sequencing — Deep sequencing (DS) can detect minority species present at ≥1 percent using a technique whereby hundreds of thousands of individual molecules are sequenced in a single assay run [12,38]. The clonal sequences are aligned and analyzed to evaluate the prevalence of low-abundance drug-resistant HIV strains. Data suggest that using DS may lead to a lower risk of virologic failure due to improved detection of minority species, which may be missed by standard genotypic testing [12,32]. However, the significance of finding low-level resistant variants depends upon the specific antiretroviral agent and its genetic barrier to resistance.

DS techniques were retrospectively applied to a subset of 264 treatment-naive patients within a large clinical trial on baseline samples prior to initiating antiretroviral therapy (ART) [12]. A significantly higher proportion of mutations were detected by DS than by standard genotypic sequencing (28 versus 14 percent, respectively). Among patients who initiated treatment with an ART regimen that combined nucleoside and non-nucleoside reverse-transcriptase inhibitors (NNRTI), all individuals who had an NNRTI-resistance mutation identified by DS experienced virologic failure over a prolonged follow-up period.

Subsequent data have shown that low-level variants with isolated protease inhibitor (PI) mutations were not predictive of virologic failure; this is likely due to the higher genetic barrier to resistance for ritonavir-boosted PI-based regimens [39]. In contrast, subjects with low-level resistant variants with extensive nucleoside analog mutations had a high rate of virologic failure, as would be expected.

One study of DS found that the detection of low-level resistant variants was more closely correlated with the patient's prior treatment history than with results of DS. These results confirm the importance of obtaining a thorough medical history [40]. (See "Evaluation of the treatment-experienced patient failing HIV therapy".)

HIV-1 proviral DNA assays — Some genotypic resistance assays are designed to analyze HIV-1 proviral DNA located in host cells. These tests can detect drug resistance mutations within proviral HIV DNA archived within peripheral blood mononuclear cells (Genosure Archive, Monogram Biosciences, South San Francisco, California) or determine viral tropism (HIV-1 Coreceptor Tropism, Proviral DNA, QUEST Diagnostics). (See 'Tropism assays' below.)

Proviral DNA resistance testing may be useful when switching a patient's ART regimen when the HIV viral load is nondetectable and there are no prior resistance results available, or if there is low-level viremia and a plasma HIV RNA genotypic assay is unlikely to be successful [41]. (See 'Genotypic resistance assays' above.)

However, the clinical utility of proviral DNA resistance testing has yet to be fully determined [41], and the results must be interpreted within the clinical context.

Tropism assays

General background — HIV requires both the CD4 receptor and a coreceptor in order to enter CD4+ T-cells. This coreceptor can be CCR5 or CXCR4. Viruses that use CCR5 as their coreceptor are referred to as R5 viruses; those that use CXCR4 are X4 viruses. An individual may have R5 viruses, X4 viruses, or a combination of both [42]. Most transmitted viruses are R5 viruses; X4 viruses usually emerge later in infection.

The frequency of exclusive R5 virus varies depending on the patient's treatment history; in treatment-naive patients, exclusive R5 virus is found in approximately 80 to 85 percent [43,44]; in treatment-experienced patients with late-stage disease it is found in only 50 to 56 percent [45-47].

Testing for viral tropism is recommended when considering the use of a CCR5 antagonist (maraviroc), which only inhibits R5 virus [7]. Coreceptor tropism testing may also be helpful for patients who demonstrate virologic failure on a CCR5 inhibitor [7].

Types of tropism assays — Both phenotypic and genotypic based assays have been developed to determine viral tropism, although there are much more clinical trial data regarding the use of phenotypic tropism assays [48,49]. The United States Department of Health and Human Services (DHHS) HIV treatment guidelines recommend phenotypic assays as the preferred method for tropism testing; a genotypic assay can be considered as an alternative [7].

Phenotypic tropism assays — There are two types of phenotypic assays (Trofile assay, LabCorp, Burlington, NC; Phenoscript assay, VIRAlliance, Paris, France) [48,50-55]. Similar to the previously mentioned phenotypic resistance methods, these assays use laboratory viruses that express patient-derived viral envelope proteins [48]. The laboratory-generated pseudovirus is used to infect cell lines that express either CCR5 or CXCR4; results of these infectivity assays determine the tropism of the patient's isolate.

Although there were problems with the sensitivity of earlier assays, subsequent tropism assays are able to detect CXCR4-utilizing viruses at very low levels (ie, 0.3 percent of the viral population) [54,56]. These new phenotypic tropism assays compare well to older established methods that used cell lines (eg, MT-2 cells) that do not express CCR5 to determine tropism [48,53].

The main limitations of these assays are the turnaround time needed for results (approximately three to four weeks from specimen collection) and the requirement for a plasma HIV RNA level ≥1000 copies/mL. Even with adequate viremia, tropism cannot be determined on a small number of patient samples. (See "Overview of antiretroviral agents used to treat HIV", section on 'CCR5 antagonists'.)

Genotypic tropism assays — Genotypic assays sequence specific regions in the env gene [49,57]. Env, a major protein of HIV, determines HIV tropism. As with other genotypic methods, sequence algorithms are used to predict tropism.

When compared with early generation phenotypic assays, population-based genotypic assays are specific but have lower sensitivity for the detection of CXCR4-using viruses [58]. However, when used retrospectively to determine tropism in patients enrolled in clinical trials of maraviroc, newer genotypic assays performed as well as or better than the original Trofile assay in predicting virologic response and clinical outcomes [57-59].

The use of deep sequencing of the envelope protein can detect and quantify low prevalence subpopulations of CXCR4-using HIV within a large set of clinical isolates [60,61]. This method surpassed the ability of the original Trofile assay to predict viral suppression and the likelihood of switching tropism after maraviroc exposure.

European guidelines favor genotypic testing for determining coreceptor usage. In the United States, DHHS HIV treatment guidelines still recommend that a phenotype be used to determine coreceptor tropism. Although there are more data correlating phenotypic tropism results to outcome than with genotypes, the latter are less expensive and results return much more rapidly. As a result, genotype tropism testing is now an alternative option for clinical practice [7].

In patients with an undetectable viral load or low-level viremia (eg, HIV RNA <1,000 copies/mL), HIV-1 proviral DNA tropism testing can be used to help determine co-receptor usage [7,27]. As an example, this assay may be helpful when changing a patient with undetectable HIV RNA to a new regimen that contains maraviroc [62,63]. If only R5 virus is detected, maraviroc could be considered for use in the new regimen. However, caution should be used when interpreting the results from these assays, as the clinical utility of this approach is still under investigation [7,64].

SUMMARY AND RECOMMENDATIONS

●Resistance assays can be categorized as either genotypic or phenotypic. Genotypic resistance assays detect the presence of specific drug resistance mutations in the regions of the HIV genome encoding protease, reverse transcriptase, and integrase. Phenotypic resistance assays measure the extent to which an antiretroviral drug inhibits virus replication in vitro. (See 'Overview of the assays' above.)

●In most settings, genotypic assays are preferred due to lower cost, faster turnaround time, and greater sensitivity for detecting mixtures of resistant and wild-type virus. However, in treatment-experienced patients with multiple resistance mutations, interpreting complex genotypes can be challenging since it can be difficult to interpret the clinical significance of a specific mutation when many mutations are present. For such patients, phenotypic resistance testing may have an advantage over genotypic testing, as it measures resistance more directly and can assess relative susceptibility and interactions among mutations. (See 'Advantages and limitations' above.)

●Genotypic and phenotypic assays are typically unable to detect low-abundance drug-resistant variants that are present at less than 20 percent of the viral quasispecies in a given sample. Other techniques have been developed to improve the detection of these minority species compared with standard drug resistance assays. (See 'Low-abundance drug-resistant variants' above.)

●Some genotypic resistance assays are designed to analyze HIV-1 proviral DNA located in host cells. These tests can detect drug resistance mutations within proviral HIV DNA archived within peripheral blood mononuclear cells or determine viral tropism. This test may be useful when switching a patient's antiretroviral regimen when the HIV viral load is nondetectable and/or there are no prior resistance results available. However, the clinical utility of proviral DNA resistance testing or proviral DNA tropism detection has yet to be fully determined. (See 'HIV-1 proviral DNA assays' above.)

●Testing for viral tropism should be performed when considering the use of the CCR5 antagonist maraviroc, since this agent only inhibits virus that uses the CCR5 coreceptor to enter the CD4 cells. Coreceptor tropism testing may also be helpful for patients who demonstrate virologic failure on a CCR5 inhibitor. Both phenotypic- and genotypic-based assays have been developed to determine viral tropism, although there are more clinical trial data regarding the use of phenotypic tropism assays. (See 'Tropism assays' above.)