INTRODUCTION

During the last decade our understanding of both the biology and the clinical relevance of human immunodeficiency virus (HIV) drug resistance has increased exponentially. There is now evidence that emergence of drug resistance to the initial highly active antiretroviral therapy (HAART) regimen is associated with an increased risk of death.¹ If drug resistance emerges to all three major drug classes (nucleoside reverse transcriptase inhibitors, non-nucleoside reverse transcriptase inhibitors (NNRTIs) and protease inhibitors (PIs)), the risk of death is substantially increased.²

Resistance testing is a laboratory tool that clinicians often use when making decisions about initial drug regimens, and usually use when choosing salvage regimens. In order to use resistance testing effectively in clinical practice, clinicians must understand the biology of resistance, the strengths and weaknesses of each assay, and the clinical implications of the results. Because new resistance mutations evolve, clinicians must be adept at using web-based information systems that are updated frequently.

BIOLOGY OF DRUG RESISTANCE

Evolution of Drug Resistant Viruses

In the past it was generally believed that drug resistance was an unavoidable consequence of antiretroviral (ARV) therapy.³ This assumption was based on the knowledge that HIV replication is error prone (roughly one transcription error per every 4 Kb of transcribed cDNA), the lack of transcription proofreading enzymes, and the production of billions of viral particles every day (see Chapter 4). Thus, all infected individuals have potential to quickly develop an extremely heterogeneous viral population with a seemingly unlimited number of variants. As a consequence, all possible resistance patterns theoretically exist as subpopulations in untreated patients. Exposure to ARV pressure, therefore, has potential to select for the preexisting resistant subpopulation. This concept was derived from modeling based on HIV variability in initial treatment studies using suboptimal regimens such as single or double nucleoside combinations.

Mutations that confer resistance are designated in a shorthand format. The HIV gene is indicated first (for instance PR for protease), followed by a single letter abbreviation for the wild-type amino acid present at a particular location in a protein and encoded by a particular triplet of nucleotides, or codon (for instance L for leucine). The number of the amino acid/codon follows (for instance, 90). The single letter code for the new, mutant amino acid that has replaced the wild-type amino acid is given next (for instance M for methionine). The designation PR L90M, thus, indicates that the wild-type amino acid leucine at position 90 of the protease gene has been replaced with a methionine.

Mutations readily occur in HIV when patients are treated with HAART regimens that are not optimally suppressive. However, the resulting drug-resistant viruses do not necessarily have equal replicative capacity (RC) in vitro when compared to wild-type virus. Thus, resistant virus may not overtake wild-type virus, allowing some resistant viruses to remain as minority variants. This suggests that, in such settings, retaining drug therapy even when viral load value is above 50 copies/μL may be beneficial since the resistant virus may never achieve the viral loads that would be encountered if the wild-type repopulates as the dominant species. The less fit virus populations result in viral load values that are significantly below the pretherapy baseline level created by wild-type virus. An example of this phenomenon is the emergence of the M184V-resistant mutant selected by lamivudine monotherapy. Once this mutation appears in the viral population while patients are on lamivudine monotherapy, the HIV RNA level rebounds toward baseline but remains 0.6–0.8 log below pretherapy levels. This phenomenon of partial suppression is best explained by the classical Darwinian principle of survival of the fittest. Drug-resistant variants, preexisting as minority species owing to a reduced replicative capacity compared to wild-type virus, become dominant under conditions of drug selection pressure while the wild-type virus is suppressed by the antiviral regimen. Thus, the less fit virus predominates during therapy, but might not attain the same level of replicative success as the wild-type virus.

The risk of continuing a partially suppressive ARV regimen is the continued accumulation of additional mutations over time. As further resistance conferring mutations and compensatory mutations (i.e., those that allow the virus to compensate for the presence of the resistance conferring mutation), the replicative capacity of these variants ultimately increase leading to outgrowth of more fit viruses. Thus, poorly replicating resistant viruses can convert into viruses with higher replicative capacity that eventually can attain the original baseline.

HAART and the Genetic Barrier Principle

Clinical experience over the last decade has shown that HAART can suppress viral load to levels <50 copies/μL for many years if adherence is adequate and the original virus is susceptible to the drug regimen used. Several studies suggest that in the majority of these patients, even with very sensitive techniques, no resistant viruses can be detected and no evidence of ongoing viral evolution occurs during long periods of therapy. In these patients, there are two possible explanations for what is being observed: (1) virus is completely suppressed, or (2) resistant virus is being generated in occult reservoirs that are not being measured. The latter scenario seems less likely, however, since resistant virus should spill over from those reservoirs into the circulation and spread to other sites.

The number of mutations required to produce detectable resistance to a drug or a regimen is defined as the genetic barrier. For many drugs, such as the original NNRTI drugs or lamivudine, the genetic barrier is one to two mutations. As a consequence, the development of resistant variants occurred rapidly when these drugs were used by themselves (monotherapy) or with only one additional agent (dual therapy). Therefore, the combination of less than complete suppression of viral replication and a low genetic barrier leads to virologic failure with rapid emergence of resistant virus.

With other drugs with a higher genetic barrier, more mutations (typically three or more) are usually required to produce clinically detectable breakthrough. When these drugs are used in a typical three-drug combination, breakthrough does not occur readily, suggesting that viruses with sufficient mutations to lead to breakthrough do not ordinarily preexist in these patients. However, if viruses with one or two mutations are acquired at the time of initial transmission, or if such mutations develop due to poor adherence, further mutations can occur that will lead to regimen failure. The presence of preexisting resistance mutations, in essence, lowers the genetic barrier at the outset resulting in higher rates of regimen failure. This is also the reason why second-line, third-line, and subsequent regimens do not work as well as initial regimens and require more potent agents or those with novel mechanisms of action to more reliably achieve undetectable levels of HIV RNA.

Because regimen failure occurs more readily if the viral isolate has preexisting resistance mutations, most guidelines recommend the use of resistance testing prior to initial therapy and prior to any change in therapy. Similarly, multidrug regimens are universally recommended to create a higher genetic barrier and more completely suppress viral replication in order to prevent resistance and help assure long-term virologic success.

Resistance Mechanisms

Mutations in reverse transcriptase and protease genes produce reduced drug susceptibility through several different mechanisms. Substitutions in the amino acid sequence of binding proteins can lead to reduced binding of drug to the target protein. Such mutations can cause cross-resistance to other drugs in the class, as is observed with nucleoside and nucleotide reverse transcriptase inhibitors (NRTIs), NNRTIs, PIs, and fusion inhibitors.

Other more complex mechanisms also exist. Several common resistance-conferring mutations to NRTI agents can improve the enzymatic efficiency of the reverse transcriptase (RT)^4,⁵ enzyme, thereby enabling the RT to more efficiently remove the terminal nucleoside triphosphate incorporated into a growing DNA strand (excision).^6,⁷ In other words, the RT enzyme with the resistance-conferring mutations more efficiently “throws out” any unnatural nucleoside, i.e., the drug. This mechanism results in resistance to a broad number of nucleoside agents. An example of this type of resistance mechanism is the development of the K65R mutation.

Novel mechanisms leading to resistance to PI drugs have been identified. Changes in the PI substrate, e.g., the cleavage site, can lead to resistance to the PIs in the absence of changes in the protease itself.⁸ The clinical importance of this mutation is under investigation.

Mutational Patterns and Hypersusceptibility

For most drugs, the development of resistance occurs through accumulation of mutations that lead to progressively reduced drug susceptibility. The mutations have an additive or synergistic resistance effect.⁹ Generally the chance of cross-resistance to other drugs in the same class increases as the number of mutations grows.

“Thymidine-associated mutations”, or TAMs pathways, are the best described pattern of mutation accumulation relevant to the thymidine analog nucleoside drugs. The most common TAMs are M41L, D67N, K70R, T215Y/F, L210W, and K219Q/E. These TAMs confer resistance not only to the drugs that select for them (zidovudine and stavudine) but also confer cross resistance to most other NRTIs, such as didanosine, tenofovir, and abacavir. Cross resistance to TAMs is critically dependent on the number of mutations. The more TAMs, the more resistance there is to more nucleoside drugs.

While more mutations usually result in more resistance, there are some exceptions to this rule. When M184V is selected by lamivudine or emtricitabine, or when Y181C is selected by the NNRTIs, or L74V selected by didanosine, the virus becomes more susceptible in vitro to zidovudine and stavudine. This “hypersensitization” effect is seen both when these mutations are introduced in wild-type virus and when they occur in the presence of one or more TAMs. Similarly, the M184V mutation often results in hypersusceptibility of the virus to NNRTI agents.

RESISTANCE TESTING

Two types of resistance tests are currently available. The genotypic approach looks directly for mutations. The phenotypic approach determines changes in drug sensitivity of viral constructs in vitro. Each technology has strengths and limitations.

For genotype testing, the most frequently used approach in assessing for resistance is population sequencing. Two tests have been approved by the FDA and the European CE Notification Body: Trugene HIV-1 genotyping assay (Bayer Healthcare, Tarrytown, NY, USA) and Viroseq genotyping assay (Abbott Laboratories, Chicago, IL, USA). The availability of FDA-approved genotype kits makes testing feasible in local and regional laboratories. Some laboratories are still using their own unique (“home-brew”) assays that are not FDA approved. Less is known about the accuracy or reproducibility of such assays.

The most frequently used phenotypic assays are the Virco Vircotype (Brussels, Belgium) and the Monogram Biosciences Phenosense assay (South San Francisco, California). Several other phenotypic assays are being performed on a smaller scale in research laboratories. All of these assays are based on the creation of recombinant virus through polymerase chain reaction (PCR) amplification of the relevant viral genes from patient viruses and shuttling them into a reference virus lacking those particular genes. This approach is referred to as the recombinant virus approach. Phenotypic tests are not likely to be performed in local laboratories because of the requirement of using live virus and the associated biohazards. To safely use such virus the laboratory must have extensive expertise and sophisticated, expensive technology that is difficult to scale-up to high capacity with sufficient accuracy and reproducibility.

Currently both phenotypic and genotypic assays focus on four specific genes: protease, reverse transcriptase, integrase, and gp41 (for fusion inhibitors). Tropism assays employ phenotype-like technology using gp 120 genes. In order to perform either assay, the gene of interest must be PCR amplified first.

Amplifying the Gene of Interest

Amplification of the gene currently utilizes PCR and employs primers from highly conserved regions of the virus that anneal to sequences in the viral genome. As part of the PCR the viral RNA is transcribed into DNA in high copy number to allow sequencing (genotype) or shuttling into a recombinant expression vector (phenotype). Therefore, for each technology, the testing evaluates a representative sample of the population of circulating virus from the patient at the time the specimen was obtained. As a result, minor variants may not be detected. Adequate levels of virus must be present to be sampled (>500–1000 copies/mL of HIV RNA) in order to generate a result. For the test to be clinically useful, the patient must be taking their current regimen to accurately assess the resistance pattern, owing to the high rate of viral production and rapid overgrowth (within days) of wild-type virus once drug therapy is discontinued.

GENOTYPIC ASSAYS

Genotype assays detect changes in the cDNA sequence of the targeted HIV genes. The coding regions of genes are organized into nucleotide triplets, or codons. Each codon encodes a single amino acid of the gene product (protein). Genotypic tests indirectly measure resistance by detecting mutations in the HIV-1 genome that lead to one or more specific amino acid substitutions in proteins. The specific changes in the protein may or may not cause drug resistance. If the mutated virus has reduced susceptibility to the drug, then administration of the drug will suppress wild-type virus and select for the resistant strains. Some mutations result in “silent mutations”, which are nucleic acid changes that do not alter the amino acid sequence because of the redundancy of the genetic code. While some amino acid changes result in reduction of susceptibility, others may lead to, in hypersusceptiblity, as noted above.

Specific mutations known to be associated with altered activity of a specific drug are sought in genotypic assays. These changes have been previously associated with drug resistance or hypersusceptibility. In most cases the mutations appear when virus escapes under selective pressure following exposure to that drug. Introduction of these mutations into a reference virus may produce a change in phenotype in vitro. In some cases these mutations may have been associated with a lack of clinical response in vivo, as measured by an increase in the plasma HIV RNA levels. If a large data base correlates the presence of the mutation with a poor virologic response, then that specific mutation(s) is considered to predict a lack of drug effect. The most convincing evidence that a mutation causes drug resistance occurs when both in vitro genotypic data and viral load response data from several independent studies confirm the association.

Given the complexity of these assays, the sensitivity and specificity of assays performed by individual laboratories must be carefully and regularly assessed.^10,¹¹ Genotypic assays are not easy to interpret when multiple mutations are present simultaneously. Summary tables and on-line resources are useful for interpreting complex patterns. Most laboratories provide interpretation in their patient result reports (Fig. 28-1).⁹ Despite this complexity, many clinicians are learning how to incorporate genotypic information into their treatment strategies.