Published in the Bulletin of Experimental Treatments for AIDS January 1998 issue, by the San Francisco AIDS Foundation.

January 1998 Table of Contents

Main Page

beta@sfaf.org

Resistance to Antiretroviral Drugs

William O'Brien, MD

What is Drug Resistance?

The usefulness of treatment for HIV infection is limited by the emergence of virus strains that resist antiretroviral therapy. This means that mutations in the virus genome allow HIV to reproduce even in the presence of therapeutic concentrations of drugs (concentrations that usually stop viral reproduction). Resistance is possible because HIV reproduces at a very rapid rate and mutates very often during the process of reverse transcription.

The viral enzyme reverse transcriptase (RT) begins the process of reverse transcription in order to make a double-stranded DNA copy of the HIV RNA genome. This DNA form is identical in structure to the genetic material in human chromosomes. HIV integrates its viral DNA genome into the human chromosome to form a structure called the provirus. Unless HIV's genetic material is converted from RNA to DNA, the virus cannot use human cells to reproduce. During reverse transcription, the viral RT enzyme makes mistakes (mutations) in copying the HIV genome, on average about 1 mistake for each replication cycle. This means that new virus particles are just a little bit different from the parent virus.

Mutation appears to be random. Most mutations create a disadvantage for the virus. In some cases, however, a mutation may actually result in a virus protein that can still function and allow reproduction of the virus, but that is no longer inhibited by current antiviral drugs. The RT inhibitors and the protease inhibitors both target enzymes encoded by the HIV genome.

While reverse transcription is an early event leading to formation of the provirus, protease is a late-acting enzyme that cuts long chains of proteins into their proper size and form so that they are ready to be assembled into new viruses just before they are released from the cell.

In the early days of single-drug anti-HIV treatment, resistance developed rather quickly, since the virus only had to develop resistance to 1 drug, and monotherapies still permitted a fairly substantial level of virus replication in most people. At higher levels of virus replication, mutations occur more often and resistance develops more quickly. This is why current therapeutic strategies that combine 3 or more drugs have been so effective in many people. These "cocktails" can decrease virus replication to "undetectable" levels, which results in the much slower appearance of mutated viruses. When combination therapies are used, it is likely that emergence of clinically important drug resistance will be delayed further because several different mutations will be required.

Table 1. Amino Acids Commonly Found in Proteins and Their Abbreviations

What Mutations are Associated with Resistance?

HIV is made up of proteins. Proteins are in turn made up of basic building blocks called amino acids. Scientists start numbering amino acids at one end of a protein. For example, the184th amino acid in a protein chain is called position 184. A mutation at a certain position simply means that a different amino acid replaces the amino acid that is usually expected at that place in the chain. This is demonstrated by the mutation that makes HIV resistant to the nucleoside analog drug 3TC. The substitution of the amino acid methionine (abbreviated by the letter M) for valine (abbreviated V) in the reverse transcriptase enzyme is written M184V (see Table 1 for the common abbreviations for amino acids). This mutation allows the virus to continue to reproduce, but creates a high level of resistance to 3TC. Single amino acid changes have also been associated with resistance to ddI, ddC and nevirapine. Although at least 5 different mutations contribute to AZT resistance, and resistance seems to increase with multiple mutations, even a single mutation can decrease the sensitivity of a virus to AZT.

Resistance to several of the RT inhibitors can be conferred by single mutations, as shown in Table 2.

Amino acid mutations have been described that lead to simultaneous resistance to several RT inhibitor drugs. Recently, a mutation called G333E was found to make HIV simultaneously resistant to AZT and 3TC. Some mutations in RT appear to be associated with multidrug resistance. The most important of these mutations seems to be Q151M, which may confer at least partial resistance to several of the RT inhibitor drugs. Typically, other mutations such as F77L and F116Y must occur in association with the Q151M mutation to confer the full multi-drug-resistant phenotype. The presence of mutations poses a big problem for selection of drugs in combination therapies.

Table 2. Resistance Mutations Associated with Reverse Transcriptase Inhibitor Use

Drug Resistance Mutations

• AZT: M41L, D67N, K70R, T215 Y/F, K219Q/E
• ddI: K65R, L74V, V75T, M184V
• ddC: K65R, T69D, L74V, V75T, M184V
• d4T: V75T
• 3TC: M184 V/T/I
• abacavir: K65R, L74V, M184V
• nevirapine: V106A, Y181 C/I, Y188C
• multidrug resistance: F77L, F116Y, Q151M, M184V, G333E

In contrast to the RT inhibitors, most single amino acid substitutions resulting from mutations in the protease enzyme do not on their own make the virus resistant to protease inhibitor drugs (see Table 3). In general, more than one mutation is needed to create protease inhibitor resistance.

The one exception may be a mutation at position 90 (L90M), which may contribute to viral resistance to all 4 of the protease inhibitor drugs approved so far. In contrast, the D30N mutation appears to be specific for nelfinavir and does not seem to cause problems with the other drugs in this class. The gene that encodes protease is small -- only 99 amino acids long -- and there are at least 11 mutations that contribute to resistance. Resistance to protease inhibitors in most cases seems to require more than 2 or 3 of these changes in combination. Unfortunately, when one protease inhibitor fails, resistance mutations have usually developed that make the path to resistance for the second protease inhibitor (and the durability of virus suppression) shorter.

The mutations described above involve changes directly to the gene that serves as a blueprint for the reverse transcriptase and protease enzymes, but there are other mechanisms for developing resistance. The HIV protease enzyme cleaves the long chain of viral protein at specific sites that are coded by specific amino acid sequences. Recent studies have shown that mutations within these cleavage sites in proteins other than protease may lead to resistance to protease inhibitors. With this type of resistance, the viral protease may be wild-type with none of the typical resistance mutations present. The reverse transcriptase enzyme is much larger than the protease enzyme, and it appears that there must be a greater complexity of changes to give rise to resistance.

It is reasonable to think that mutations in this critical gene might be a problem for the virus and would result in virus strains that do not reproduce as well, referred to as diminished viral fitness. From an evolutionary perspective, the virus compromises a little bit of enzyme efficiency in order to have a structure that is not affected by inhibitor drugs. Although slight decreases in viral fitness can be demonstrated, the differences are not great, and most of the virus mutations that allow HIV to resist the action of drugs still allow these enzymes to function quite well. Despite small differences in enzyme efficiency and viral fitness, the high rate of replication and the long duration of infection allow these resistant HIV strains to damage the immune system and replicate in a way that differs little from wild-type strains. The early hope that the development of resistance mutations would cause the virus to evolve into a pitifully weak pathogen is probably unfounded.

Table 3. Resistance Mutations Associated with Protease Inhibitor Use

Drug Resistance Mutations

• overall includes accessory mutations L10I, K20M, L24I, V32I, M46I, I44V, L63P, A71V, V82A/F, I84V, L90M
• saquinavir: G48V, L90M
• ritonavir: M36I, M46I, I54V/L, A71V, V82F, I84V, L90M
• indinavir: L10I, V32I, M46I, I54V, L63P, A71V, V82A/F, I84V, L90M
• nelfinavir: D30N, A71V, N88D, L90M

The Utility of Resistance Testing

There may be substantial benefits from measuring resistance in people who take antiretroviral drugs. One possible advantage would be to distinguish between actual resistance to the drugs that are being taken and inadequate drug levels that occur in people who do not take their drugs on schedule, who do not absorb them well or who metabolize them very quickly. In the latter case, one would expect that failure of the drugs, as shown by increasing viral load, would not be associated with the detection of any resistance mutations. On the other hand, when the drugs fail because of antiretroviral resistance, amino acid mutations associated with resistance should be detectable in the virus in the blood. Currently there is a limitation to our ability to detect mutations and an inability to actually measure drug levels in the body. Perhaps routine measurement of antiretroviral drug levels in the blood, which is already done for a number of antibiotic drugs, will become a part of future monitoring.

Two types of laboratory tests can determine whether resistant HIV is present in a person using anti-HIV drugs. Genotypic testing identifies mutations in the genetic structure of HIV in a blood sample. Phenotypic testing measures the amount of drug required to completely stop HIV replication in a blood sample.

Complications of Antiretroviral Treatment

Many factors contribute to determining whether or not a particular anti-HIV therapeutic regimen will be successful. Obviously, one of the most important considerations is whether individuals are taking the drugs properly, at the correct times and in the correct dosages. Since many approved drugs for HIV infections require administration 2 or 3 times a day, and some need to be taken with food while others need to be taken without food, it can be a challenge to strictly adhere to a complicated 3-drug regimen. The best way to deal with this thorny issue is to make taking the medications as easy as possible. Both the physician and patient may work together to solve this problem.

First, the regimen should be simplified as much as possible. Instructions for taking medications must be very clear. For example, if ritonavir and saquinavir should be taken with a high-fat meal, foods that constitute a high-fat meal should be listed and given to the patient. Drug therapy should be tailored to individual lifestyles. For example, if a person usually is away from home in the middle of the day and requires a mid-day drug dose, an extra bottle of pills should be provided so that medications can be kept where the dose will be taken. Another tool is to establish a regular pattern for pill taking that includes a reminder for medications, such as a television show or a meal.

Physicians must be available to answer questions and to follow-up with additional information after a clinic visit. Because clinical interactions are often brief, important comments about care may soon be forgotten. Therefore, handouts should be provided during clinic visits to explain more about the medications, how they work, and how people with HIV infection are monitored using viral load tests and other means.

Behavioral and educational issues may be the most important determinants of success. Still, many other factors relate more to the virology of infection and to the biologic response to therapies. People who have never taken antiretroviral therapy before will have a better response to therapy than those who are experienced with other drugs. Prior therapy selects for viruses that are able to grow in the presence of these drugs. Often, there is some cross-resistance to new agents already present.

Thus, the first antiretroviral regimen has the best chance to achieve a durable virologic response, which is the current goal of therapy. If the patient and physician agree, the initial therapy should be the most potent regimen possible, since any regimen will work better when used first than when employed as "salvage therapy."

An additional limitation to the ability to suppress virus is the lack of possible combinations that can be employed sequentially. Although there are thousands of theoretical combinations possible with 11 approved antiretroviral drugs, certain drugs should never be used together (such as AZT and d4T). Some combinations should be avoided because of overlapping toxicities or resistance patterns.

In other cases, the use of these drugs may be limited by contraindications. For example, individuals receiving coumadin for anticoagulation cannot take ritonavir at the same time. In addition, use of other protease inhibitors may be problematic because they delay the metabolism of coumadin, leading to potentially dangerous levels of anticoagulation. Some nucleoside analogs may not be appropriate for patients with existing severe peripheral neuropathy. These individuals may have difficulty taking ddC or d4T.

Individualization of Therapy

The complexities discussed above make long-term HIV suppression a challenge for many patients. It may be difficult to select a combination regimen that can be taken easily, does not cause too many side effects, is well-absorbed to achieve therapeutic levels and is effective against all the different virus subpopulations that may be present in an infected individual. For this reason, there must be great flexibility in determining the components of a regimen for each person. The treatment philosophy of both the care provider and the patient -- whether aggressive or more conservative -- is probably the overriding consideration. Regimens must be tailored to be as easy as possible for patients to take, and people must be motivated to stick with a regimen, sometimes for a period of many years.

Guidelines for Antiretroviral Therapy

Various guidelines have been proposed for antiretroviral therapy, including the Guidelines for the Use of Antiretroviral Agents in HIV-Infected Adults and Adolescents released by the U.S. Department of Health and Human Services and the Treatment Guidelines for HIV/AIDS released by the International AIDS Society-USA. It should be kept in mind that these are suggestions, not rigorously mandated instructions. Nonetheless, these guidelines emphasize the importance of maximal virus suppression. Use of resistance tests is generally not included in any of the current recommendations.

The principal components of these guidelines relate to viral load. The early studies used to validate viral load monitoring demonstrated that a 0.5 log decrease (67% reduction or reduction by 3-fold) is the minimum change at 1 month that is associated with a clinical benefit. These were monotherapy trials. Since more effective treatment is now standard, the current guidelines appropriately suggest a 1.0 log (90% or 10-fold) reduction in viral load by 8 weeks. The goal of therapy is to achieve "undetectable" virus levels by 6 months. Failure of therapy is indicated by failure to achieve either of these goals, or by sustained viral load increases of greater than 0.5 log in patients with early stable suppression or in whom detectable virus was initially absent.

The current recommendations seem to be appropriate for people who have not taken anti-HIV therapy before. They may be problematic for drug-experienced patients, since the best response achievable may not be a reduction of virus to an "undetectable" level. The philosophy of both the patient and the care provider also influences assessment of success. The guidelines reflect a more aggressive treatment stance that is popular now that drugs are better able to achieve virus suppression, but a more conservative approach to treatment may be reasonable. Some recent retrospective studies have demonstrated that individuals with fewer than 5,000 copies/mL of HIV RNA are very unlikely to progress for at least several years. Moreover, the viral load parameters commonly cited for response to therapy and for prognosis are based on old plasma samples that were stored for many years in the freezer. The values obtained on fresh samples are perhaps 2Ð5 times higher, and the "undetectable" level in stored samples may actually have been closer to 5,000 copies/mL if the same plasma sample had been tested before freezing.

Viral load monitoring is the principal measure for determining when to begin therapy and when to change therapy that is failing. The current issue of greatest importance is to develop ways to measure virus resistance so that patients and physicians can know which therapies are less likely to be successful because of pre-existing resistance. This will help to more intelligently select combinations that will provide the most durable suppression of virus replication. Although a variety of HIV resistance assays are available commercially, it is not yet clear how to use the substantial information that can be derived from these tests, and their role in clinical management remains to be determined.

Available Genotyping Services

There are at least 3 commercial assays for assessing genotypic changes associated with HIV resistance. Table 4 shows the advantages and disadvantages of the different tests. The most straightforward genotypic assay is direct sequencing following polymerase chain reaction (PCR) amplification. Mutations known to be associated with resistance to a given drug are identified by analysis of the sequence of the reverse transcriptase or protease gene. Affymetrix takes a different approach using different permutations of short stretches of HIV RNA arrayed on a microchip. Analysis of hybridization patterns of sequences for a particular person compared with these random arrays identifies mutations based on binding to the short sequences representing resistance mutations. The LIPA test, made by Innogenetics, immobilizes fewer than 50 short stretches of HIV RNA in order to identify the major mutations that cause resistance to AZT. The company plans to include tests for other mutations in the future.

These tests are straightforward and reproducible and can be rapidly processed. The great disadvantage is that, in most cases, only one sequence is detected. In people, however, many different sequences may coexist. It is hard to know whether the single sequence analyzed by these assays represents the dominant sequence type, or even if it has importance in determining resistance in the person. Although fairly sensitive, these genotypic assays do not detect mutations unless they are present in more than 25% of the HIV present in the body. Thus, if drug therapy fails and a large amount of the virus is resistant to a given drug, related resistance mutations are likely to be detected. Unfortunately, the sequences available are unlikely to tell us which drugs to use next, since resistance mutations will probably only be detected under the selective pressure of therapy. It will probably not be possible to detect rare resistance mutations that will very quickly lead to failure of a subsequent drug if the person being tested is not experiencing selective pressure at the time of the assay.

An even bigger problem may be the issue of interpreting the sequence pattern. Many mutations associated with drug resistance when the drugs are used as monotherapy are known, but the pattern of resistance is much more complex in people taking combination therapy. More experience and information will allow better interpretation of genotypic sequence data.

Table 4. Genotypic HIV Drug Resistance Tests

Available Phenotyping Services

Classical resistance testing for antimicrobial drugs is performed in culture assays that put infectious organisms into various concentrations of drug. Using this type of test, the sensitivities of organisms to various agents can be measured. Table 5 shows the advantages and disadvantages of the different tests.

Initial attempts to assess phenotypic resistance to antiretroviral drugs relied on use of existing virus stocks in culture gathered prior to assessment of drug inhibition. This process eliminated viruses that did not grow well in test tubes and favored viruses that grew better in culture. The process may have decreased the number of resistant viruses which may be slightly impaired in their ability to grow in culture.

Recently, phenotypic assays have been developed that are linked to the genotypic tests. Stretches of HIV genetic material are amplified and used to construct virus clones in the laboratory. From these, virus can be derived for phenotypic testing. This avoids the step of expanding HIV in culture. The problems with phenotypic assays include the potential danger of propagating live HIV in hospital laboratories. The tests are also expensive and time-consuming. In addition, phenotypic assays suffer from the same problem seen with genetic assays -- only one variant can be assessed at a time. It would be easy to miss clinically relevant HIV strains by analysis of only one variant. Furthermore, the amount of drug needed to inhibit HIV replication in the test tube does not directly relate to sensitivity or resistance to drugs in people with HIV. Therefore, resistance testing is not yet a routine part of clinical monitoring.

Table 5. Phenotypic HIV Drug Resistance Tests