Introduction of HIV drug-resistance testing in clinical practice.

Abstract

Introduction The therapeutic management of HIV-1 infection has improved significantly in the past 2 years. Diseases such as AIDS, with a prolonged asymptomatic period (up to 10 years), allow the initiation of treatment before symptoms appear. The possibility of monitoring the progression of HIV infection by means of the CD4 cell count or, most recently, of predicting the evolution to AIDS by measuring the plasma viral load [1], has provided the opportunity for designing therapeutic and preventive guidelines [2-4]. Despite these goals, and given that highly active antiretroviral therapy (HAART) does not achieve the eradication of HIV [5], although effective control of its replication is possible for a variable period, new parameters are necessary to enable clinicians to recognize quickly when a regimen begins to fail and to choose the most useful drugs after therapeutic failure has occurred. Laboratory tests for assessing the development of drug resistance have been used in some clinical trials and their results proved to be correlated with the loss of antiviral efficacy. However, the predictive value of the drug-resistance profile on the subsequent response to antiviral drugs, as well as their use as an early marker of virological failure, before a rebound in the viral load is apparent, has only recently been reported [6-9]. Molecular basis of drug resistance HIV, like other RNA viruses (i.e. hepatitis C virus), shows a high degree of genetic variability. Reverse transcriptase (RT), one of the main enzymes involved in the replication of HIV, has not the proof-reading ability of other DNA-dependent DNA polymerases and is highly error-prone. This explains why the newly synthesized RNA molecules show frequent mutations compared with the initial HIV genomes. This high variability could compromise the viability of the virus and a slightly higher mutation rate could lead to catastrophic error [10]. In contrast, the provision of genetic diversity constitutes a wonderful mechanism for biological adaptability, when the enormously high viral replication rate of HIV (approximately 1010 new viral particles are produced per day) is taken into account. Among the viral particles generated per day, some might have some adaptive advantage compared with those from which they evolve, whereas others are less efficient or even defective. This replicative advantage (‚fitness‚) implies, for instance, the ability to escape the host‚s immune system, or to become less sensitive to antiretroviral drugs. The rapid generation of HIV variants in an infected person, which coexist at the same time, provides HIV with its ‚quasispecies‚ characteristics. The representation of a genetic variant within the total virus population in a single individual (approximately 1012 viral particles) will depend on the relative degree of compromise or adaptive advantage of each variant in different environments and at different times. When a drug is administered, resistant viruses are selected over sensitive ones, and they subsequently fill the ‚viral universe‚ of the host [11-13]. Genotypic, phenotypic, and cellular resistance Lower sensitivity (or resistance) to an antiretroviral drug usually depends on the presence of some genetic changes in the viral genome, which implies a change in the enzyme targeted by the drug. Some changes may prevent the binding of the drug. This is the mechanism of resistance that operates with the substitution of a tyrosine for a cysteine at codon 181 of RT, which confers a high level of resistance to nevirapine. The loss of sensitivity to a drug is sometimes difficult to explain exclusively by genetic changes, involving one or a small number of mutations. An important loss of sensitivity to protease inhibitors usually requires the accumulation of three or more mutations in the related gene or even in the proteolytic cleavage site in its substrate, the polyprotein gag-pol [14]. The genetic barrier against the development of resistance is variable for each antiviral drug; although some drugs (e.g. lamivudine (3TC) or nevirapine) lose their activity when a small number of mutations appear, the sensitivity of other drugs is compromised only after the accumulation of many mutations. The phenotypic resistance is recognizable in vitro measuring the drug concentration necessary to inhibit the viral growth half-life (IC50) or 90% (IC90) in a cell culture. The mechanism of resistance to some antiviral drugs, such as stavudine (d4T), may be partly dependent on a reduction in the concentration of cellular phosphokinases (thymidin-kinase), necessary to activate d4T to its triphosphate form. A mechanism of ‚downregulation‚ causes this cellular resistance, and explains why in patients previously treated with zidovudine (ZDV), which also requires thymidin-kinase to be metabolized to its triphosphate active form, the antiviral activity of d4T seems to be lower than in patients who have never received ZDV [15]. In contrast, the induction in cellular efflux transporters for nucleosides or protease inhibitors, such as P-glycoprotein, can also interfere with the antiretroviral activity of these compounds [16-18]. Mutations in the viral genome, which directly affect the sensitivity of the enzyme to an antiviral drug, are known as primary mutations. These are usually relatively specific for each drug, and appear soon after the selective pressure begins to operate, tending to minimize the antiviral action of the drug. However, such mutations often produce some kinetic disadvantages for the mutated enzyme. The secondary mutations accumulate in viral genomes already harbouring some primary substitutions, in an attempt to compensate for the kinetic disadvantage produced by the primary mutation itself. Most secondary mutations scarcely increase the resistance to the drug, but are selected by mutant viruses because they are able to restore the replicative capacity of the virus. Despite the minimal overlapping among the primary mutations (Table 1), many secondary mutations, also called compensatory mutations, are shared by drugs from the same family (nucleoside analogues, non-nucleoside analogues, and protease inhibitors). This explains why crossresistance among drugs of the same group is more often recognized when a previous HAART has been maintained for long periods in a patient with virological failure (persistent high viral replication). The clinical implication of this is that drugs should be switched soon after the recognition of plasma viraemia rebound, if therapeutic benefit is to be maintained (Fig. 1). Figure 2 summarizes the positions reported to be involved as secondary mutations in the protease gene. Some of them appear as natural polymorphisms in untreated patients [19], and tend to be selected and accumulate under treatment.Table 1: Primary mutations associated with resistance to antiretroviral drugsFig. 1: Different antiviral activity of second-line treatments depends on their early or late introduction after virological failure of the first treatment.Fig. 2: Consensus sequence of the protease gene (HIV-1 subtype B) and amino acid substitutions at critical codons associated with resistance to protease inhibitors. Key substitutions are underlined, the rest being compensatory.The amino acid positions involved in the development of drug resistance overall seem to be conserved among the different HIV variants, including the different HIV-1 group M subtypes as well as HIV-1 group O and HIV-2 [20,21]. Although differences of up to 40% exist at the nucleotide level comparing the HIV pol gene of these variants, their amino acid sequence is relatively conserved, and consequently their tertiary protein structure. This can explain why drug susceptibility is relatively homogeneous among the different HIV variants [22], except for non-nucleoside RT inhibitors. The predominance of mutations at critical positions, such as codon 181 of the RT gene, in HIV-2 and HIV-1 group O explains their natural resistance to these compounds [23-25]. Interactions and new drug-resistant mutations The selection of mutations in the HIV genome as a consequence of failing under a specific compound confers loss of susceptibility to that drug, but can restore or increase the susceptibility to other antiviral drugs. For instance, codon 74 and codon 184 mutations, respectively, appearing under didanosine (ddI) and 3TC therapies, can initially restore to some extent the sensitivity to ZDV in ZDV-resistant isolates [26,27]. It has been suggested that this antagonistic interaction between drugs could be exploited to potentiate antiretroviral combinations. In a similar fashion, susceptibility to adefovir seems to be increased for isolates carrying the codon 184 mutation after failing on 3TC [28]. Hypothetically, this mechanism could confer a synergistic effect to the adefovir-3TC combination. Drug-resistant mutations emerging in patients receiving combination therapy may not represent just the addition of mutations that appear under those drugs used separately. New mutations can develop when drugs are used either simultaneously or sequentially. In fact, individuals who start their first antiretroviral combination and those who receive an alternative regimen after suffering a virological failure, show some different characteristics. In the case of a patient who first begins an antiretroviral combination, the emergence of mutations is a direct consequence of the pressure of the separate drugs included in the regimen. For instance, the administration of ZDV and 3TC together may induce the development of a mutation at codon 333 (GÆD), which confers resistance to both drugs simultaneously [29]. In the situation of a patient who receives an alternative regimen after a virological failure, the new mutations appear as a consequence of the pharmacological pressure over a previously mutated enzyme. Patients who fail under ZDV, for instance, and switch to ddI, may develop a mutation at codon 196 (GÆE) [30]. This mutation does not appear in antiretroviral-naive individuals who are treated with ddI, in which a mutation at codon 74, which confers a moderate level of resistance to ddI, is observed. It is interesting to note that the mutation at codon 196 alone does not confer any resistance to ddI; however, in the presence of ZDV-associated mutations, it confers a high level of resistance to ddI (IC50>100). Crossresistance and multidrug resistance Although monotherapy usually selects for HIV-1 variants with specific mutations for each antiviral molecule, some mutations induce a loss of sensitivity to several drugs. A substitution at the RT codon 103 (KÆN) produces a loss of sensitivity to all known non- nucleoside RT inhibitors (nevirapine, delavirdine, loviride, efavirenz). In a similar way, a mutation at codon 184 (MÆV) induces a high level of resistance to 3TC, but also a low-moderate level of resistance to ddI, zalcitabine (ddC) and abacavir (ABC). Finally, a mutation at the protease codon 82 (VÆA/F/T), selected by the administration of indinavir, induces crossresistance to ritonavir, and vice versa. Combination therapy may eventually force the acquisition of mutations distinct from those observed with monotherapy, and with a broad spectrum. Patients who receive ZDV plus ddI or ZDV plus ddC, are prone to develop a mutation at RT position 151 (QÆM/L), which produces resistance to all nucleoside analogues (although in a lesser degree to 3TC). This primary mutation is usually accompanied by other compensatory mutations at codons 62, 75, 77 and 116. The prevalence of this multidrug-resistant mutation is currently low; approximately 2-4% in individuals previously pretreated with nucleoside analogues [31,32]. Recently, the insertion of two amino acids (usually two serines) between those at positions 69 and 70 of the RT, together with an amino acid change at position 69 (TÆS), seem to induce a critical structural change at the RT. The enzyme became multiresistant to all known nucleoside analogues [33-36]. Although this insertion had been reported previously [37,38], its ability to induce multiresistance has only recently been underlined. The mechanism by which this insertion appears is not well known, but most likely supposes a phenomenon of genetic duplication of codons 68 and 69. Among the required conditions for the development of this insertion seems to be a previous exposure to ZDV. In Spain, four patients (prevalence: 0.8%) have been reported with the T69S-SS insertion at the Instituto de Salud Carlos III in Madrid, after testing 475 individuals who failed on a previous antiretroviral combination that included at least two nucleosides [39]. Drug-resistance assays Both genotypic and phenotypic assays are currently available for the recognition of drug-resistant viruses in the clinical setting. Genotypic tests examine the presence of mutations in the viral genome population, which have previously been demonstrated to produce drug resistance (i.e. M184V for 3TC, T215Y/F for ZDV, L74V for ddI or Y181C for nevirapine). Phenotypic assays measure the degree of sensitivity to a drug on a cell culture, usually quantifying the IC50 value. All currently available genotypic assays use the polymerase chain reaction method for amplifying the genetic material, although there are three different strategies to identify the mutant virus population: (i) hybridization of the amplified product with specific probes, which discriminate the mutant and wild-type viruses, checking at specific positions (i.e. the LiPA assay); (ii) sequence analysis using chips examining multiple codons by genetic hybridization (i.e. the Affimetrix assay); and (iii) automatic cyclic sequence analysis of the target genomic region (i.e. with a PE-Applied Biosystems sequencer). A sequence analysis of individual clones needs to be done to prove that several mutations coexist in a single viral genome. Phenotypic resistance analyses are usually performed using viruses isolated from patients, after growing them in donor lymphocyte co-culture, or using a laboratory T-cell line. Alternatively, the use of recombinant viruses (recombinant virus assay), obtained after the conjugation of a fragment of a patient‚s plasma RT or protease with a standard complementary defective genome provides the opportunity to examine functionally the effect of changes in the enzymes. This method provides more comparable results and avoids the problems of most phenotypic studies, which require the growing of a patient‚s isolated viruses. Two assays using recombinant virus assay (Virco and ViroLogic) are currently available, but not yet commercially. The measurement of plasma RT activity by Amp-RT [40] in the presence or absence of antiretroviral drugs has been used recently with success to detect drug resistance in vivo. Viruses with mutations conferring a high level of drug resistance (i.e. harbouring a mutation at codon 181 for nevirapine, or at codon 184 for 3TC) have been identified using this new methodology [41,42]. The development of mutations causing a loss of sensitivity to any antiretroviral drug usually precedes its recognition using phenotypic assays. Therefore, when mutant viruses are present even in small amounts, genotypic assays may be more sensitive than phenotypic tests. However, phenotypic results provide more direct and reliable information on the sensitivity of a virus to any drug. Clinical applications Not all virological failures are caused by the development of drug resistance (see Table 2) [43]. Other factors can also produce a therapeutic benefit lower than expected; such as: (i) inadequate adherence to the prescribed antiretroviral regimen (forgetting doses, incorrect schedule or doses, etc.); (ii) gastrointestinal malabsorption of drugs (i.e. due to the presence of diarrhoea, or because ddI is administered after meals, or ritonavir fasting, etc.); (iii) interaction with other drugs (i.e. the association of rifampin and a protease inhibitor decreases the plasmatic levels of the latter). After excluding other possible causes, the lack of benefit of any HAART might be attributed to the presence of drug resistance. When this circumstance occurs in patients without a past history of antiviral medication (primary resistance), resistant viruses could have been acquired at the time of primary HIV-1 infection, from individuals undergoing HAART carrying those mutant viruses. Alternatively, naturally resistant viruses need to be excluded (i.e. HIV-1 group O and HIV-2 isolates show primary resistance to all non-nucleoside RT inhibitors).Table 2: Main causes of failure of antiretroviral drugsIn June 1998, at the Second Workshop on HIV Drug Resistance and Therapeutic Strategies (Lake Maggiore, Italy) and at the XIIth World AIDS Conference (Geneve, Switzerland), the first results from retrospective clinical studies [6-9] demonstrating that the drug-resistance profile predicts the response rate to a particular antiretroviral regimen were reported. A therapeutic failure or a response lower than expected can be predicted when drug-resistant viruses are demonstrated at baseline, before initiating treatment. Both genotypic and phenotypic assays are equally useful in this respect. In contrast, the lack of recognition of resistant viruses at baseline does not preclude a therapeutic failure. In November 1998, at the Fourth International Congress on Drug Therapy in HIV Infection (Glasgow, UK) the first preliminary data from VIRADAPT, a French prospective study [44], demonstrating that drug-resistance information provided to clinicians translates into a better outcome for patients were presented. Currently, there is no longer any doubt about the advantage for clinical care provided by drug-resistance testing, mainly in choosing the most appropriate drugs in pretreated individuals switching to an alternative regimen. Other issues, such as the cost and technical facility of assays, need to be addressed urgently before their implementation. The detection of resistant mutations may be carried out by examining viral RNA isolated from a patient‚s plasma, or looking at DNA extracted from the peripheral blood mononuclear cells. Although the analysis of plasma RNA reflects the genotype of currently replicating viruses, the proviral genotype may give information about a former virus population in that patient. Therefore, the lack of recognition of mutant viruses in plasma from a pretreated patient does not exclude the presence of drug-resistant genotypes hidden as provirus in the lymphocytes. The recognition of a mutant genotype before a viral rebound occurs in patients undergoing HAART could be used as an early marker of virological failure, providing the opportunity for an early intervention, switching drugs selectively. However, information from drug-resistance assays can only be obtained when a minimal number of HIV-1 genomes are present in the study sample. For instance, in patients with low or undetectable viral loads, the difficulty of identifying HIV-1 genomes has made it impossible to analyse the sequence. This has precluded the clinical application of the examination of a few characteristic and crucial point mutations that confer a high level of resistance to some drugs (i.e. at the 181 codon for nevirapine, or the 184 codon for 3TC). Although hypothetically their recognition could be used as an early marker of virological failure in patients receiving these drugs, even before a rebound in plasma viraemia occurs, the lack of sensitivity of the current drug-resistance assays limits their applicability. New procedures are currently under development to avoid this problem. There is no doubt that drug-resistance assays have achieved sufficient technological maturity to be introduced into clinical practice. However, the information they provide seems to be useful at present in only a few circumstances (see Table 3). Among ‚naive‚ patients, the presence of drug-resistant viruses should be discounted in two situations: (i) individuals with acute primary HIV-1 infection; and (ii) in seropositive pregnant women. In the first case, the early initiation of HAART has been shown to improve the outcome in those individuals; however, the risk of acquiring a resistant virus is currently not negligible. Several studies in the USA and western Europe [11,13,31,45] have reported that 10-15% of new infections could be caused by viruses carrying drug-resistant mutations. Moreover, concern about the transmission of multidrug-resistant viruses has recently emerged [45-47]. In the case of pregnant women, given that HAART significantly decreases the risk of vertical transmission, knowledge of the drug-resistant profile, either in naive or pretreated women, may allow clinicians to choose the most appropriate combination of drugs.Table 3: Clinical situations for which drug-resistance testing might be usefulAnother circumstance in which the study of the drug-resistant profile could be of interest is when a high-risk exposure has occurred, in order to prevent infection or, if it occurs, to improve the outcome. Accidents in healthcare workers, with exposure to blood from HIV-1-infected patients, or after unprotected sexual contact with an HIV-1-infected person, are situations for which HAART is recommended as early as possible, in an attempt to decrease the risk of infection. Although treatment should be introduced as soon as possible (within a few hours post-exposure), information on the drug genotype in virus from the index case could allow modification of the drug combination as soon as results become available. Among pretreated patients, drug-resistance genotyping might be useful in three circumstances. First, to select an alternative regimen when a previous treatment fails, excluding drugs with cross-resistance profiles. Second, to choose the most appropriate rescue therapies in heavily pretreated patients under current therapeutic failure. In such cases, the recycling of antiretroviral drugs may be done more accurately if the resistance pattern is taken into account. Third, in heavily pretreated patients in whom a new regimen does not provide the expected benefit, the exclusion of a multidrug-resistant genotype (i.e. the 151 codon mutation complex, or the T69S-SS insertion) may allow the selection of alternative drugs for salvage therapy. In pretreated patients without evidence of failure, that is with undetectable plasma viraemia or a sustained increase in the CD4 cell count, the investigation of the drug-resistance profile is currently not justified. Conclusion HIV-1 drug-resistance assays have reached maturity, and it is predictable (and desirable) that they will achieve widespread use in clinical practice. Together with the information provided by the CD4 cell count and the viral load, these tests will provide to HIV/AIDS practitioners a more complete picture of their patients‚ prognoses and current status, as well as the opportunity for improving antiretroviral decision-making.