The Limitations of Drug Resistance Testing
Many Factors Influence the Results
In someone not taking antiviral therapy, HIV replicates at a rate of 1010-1011 copies -- or 10 to 100 billion -- per day. As the virus seeks to construct a DNA version of itself from its RNA origins, prior to combining with the DNA of the host human cell, it makes every mistake possible every single day in the arrangement of its nucleotide bases. This is because the HIV enzyme reverse transcriptase that makes this change from RNA to DNA has no internal proofreader to make sure these mistakes don't happen.
In the presence of an only partially suppressive drug regimen, the mutations that allow the virus to replicate in the presence of the drug offers HIV an obvious benefit. The virus that contains these mutations, "resistant virus," becomes the predominant species. This resistant virus may be passed on to another person and the number of people being infected with already resistant virus is rising in some areas.
In order to characterize this resistance, as discussed in earlier articles, two types of tests have been developed, called a genotype and a phenotype. They may be done separately or together, an example of the latter being the Phenosense GT from Monogram Biosciences.
The genotype looks at a population of viruses from the patient and determines the genetic sequence in regions of the pol gene of the virus that codes for reverse transcriptase and protease enzymes, as well as those from its env gene that codes for proteins critical to the entry of the virus into the cell.
Specific mutations in these areas are known to confer resistance to certain drugs, and algorithms are developed to predict resistance or susceptibility. These algorithms are only as good as our knowledge of what those mutations are, something that is being constantly updated. Sometimes it takes months or even years for a new drug's resistance mutations to be clearly identified.
The phenotype measures the susceptibility of recombinant virus (HIV is one type of recombinant virus) from the patient directly to different drugs in cell culture. This is a direct test of resistance and can be performed even for drugs still in development. For both these tests, the expanding number of antiretroviral drugs, now 20, means that the tests get more complex as time passes.
Both of these methods have been shown to be useful in constructing drug regimens to treat resistant virus and have greatly increased our understanding of HIV and its natural evolution. However, there are important limitations to both of these tests.
1. Viral Load
Both viral load tests need a viral load (HIV-RNA) of at least 500 copies/ml of virus and preferably more than 1,000 copies/ml. This can create a problem when someone has a persistent low viral load of greater than 200 copies/ml but less than 1,000 copies/ml. Should the regimen be changed or intensified? Is it poor adherence? If we do nothing, will the virus develop more mutations, more resistance, and make it more difficult to construct a new regimen?
2. Minority Species
The tests always miss virus that is present at less than 10-20% of the total viral load. They may even miss minority species that comprise up to 30% of the viral population. This means that mutations may be hidden and not be found, so-called archived mutations. These mutations may appear rapidly when a drug they offer resistance to is added to a new regimen. This is why a careful history of which drugs the patient has taken in the past is so important, especially in failing regimens that have resulted in viral rebound (an increase in viral load).
Methods are available to test for these minority species, such as single genome sequencing, where instead of populations of viruses being tested, single viruses are sequenced. The more viruses that are looked at, the more sensitive the test, even down to as little as 1% of the species. These tests are very expensive and time consuming and are reserved for research labs.
3. Mixtures of Viral Species
When a patient's virus is in the process of developing resistance, there may be a time when the viral population consists of wild type virus, the original infection (which could be wild type or a resistant strain), and virus with drug resistance mutations. Eventually, the resistant species become dominant, but a test done in the transition stage may show that the total population is "sensitive" on the phenotype, while the genotype shows the presence of both wild type and resistant virus. This can also occur in reverse when a patient stops taking drugs because of virologic rebound. While off medication, over time the predominant strain of HIV becomes the better growing wild type and there may be mixtures of both species if the tests are done at this stage. In this situation, the genotype is the best test to determine which drugs to use.
4. Re-Sensitizing Mutations
Combinations of mutations in a viral species may behave differently than they do when they occur separately. For example, the K65R mutation causes the virus to be resistant to Viread (tenofovir). However, if the M184V mutation is also present, the combination of the two mutations means that the virus is usually still sensitive to Viread. The two mutations together also make the virus hypersensitive (more sensitive) to Retrovir (zidovudine or AZT). A genotype would show both of the mutations noted above but only an experienced clinician would know how to interpret the combination. On the other hand, a phenotype would show the effect of the combination of mutations.
Other examples of hypersusceptibility include the Y181C and the L100I mutations that cause resistance to the NNRTI class of drugs. They give HIV increased susceptibility to AZT. HIV variants with multiple NRTI mutations such as K65R appear to be more susceptible to NNRTIs than wild type virus. Some studies have suggested that this hypersusceptibility improves treatment response to NNRTIs. In the protease inhibitor (PI) class, the N88S mutation generated by pressure from Viracept and others causes the virus to be more susceptible to Lexiva (fosamprenavir), while the I50L mutation seen in Reyataz failures causes hypersusceptibility to most of the PIs. In these cases, the phenotype is the best test as it reflects these interactions.
5. Incomplete Algorithms
The genotype predicts resistance or susceptibility based on a set of rules about which mutations cause resistance to which drugs. It does not reflect interactions like those described above, nor does it reflect mutations that might occur in other parts of the viral genome like the gag gene. It is also always playing catch up as we identify new mutations for old and new drugs that we did not know about before.
We can rely on current algorithms to accurately predict drug susceptibility for single mutations like M184V, a mutation commonly seen in failing regimens containing Epivir (3TC) or Emtriva (emtricitabine, FTC), and the D30N seen with Viracept (nelfinavir) failures.
However, where there are complex patterns of mutations such as those seen with boosted protease inhibitors or most NRTIs, and because there is such a huge variety of HIV species, these predictions become a lot more difficult.
There have been attempts to correlate genotypic resistance patterns to outcomes by compiling so-called mutation scores. A listing of important mutations affecting a specific drug is developed and the mutations in a given viral sample are added up to calculate the score. Cutoffs are suggested, such as a viral sample with a mutation score of less than four should be sensitive. These scores are specific for each drug and are useful, especially for new drugs, where our understanding of the mutational pattern is still evolving. However, they are only a general guide, and often we need a phenotype as well for clarification.
Also, even when the mutation commonly predicts resistance, the phenotype sometimes shows that the virus is still sensitive. An example is the virus with only the L90M mutation, which the genotype would predict as being resistant to Invirase (saquinavir). In fact, up to 30% of phenotypes show that this virus is still sensitive to Invirase.
6. Phenotypic Cutoffs
Fold change refers to the amount of drug needed to suppress any given virus. This is usually expressed as IC50 (the inhibitory concentration or drug level needed to inhibit 50% of virus replication). The IC50 fold change for wild type virus is defined as 1, so a resistant virus might require 2, 3 or 100 times as much drug -- 2, 3 or 100-fold resistance -- compared to wild type.
By correlating baseline phenotype at entry into clinical trials with virologic outcomes at defined points, we are able to derive "clinical cutoffs," which provide the greatest clinical relevance for phenotype assay interpretation. The lower cutoff or reduced susceptibility cutoff is where the response to that particular drug begins to decline, and the upper cutoff is where the response is considered to be negligible. Below the cutoff the drug is usually most effective. In other words the virus can be sensitive to the drug (fold change below the cutoff), partially resistant (fold change between the lower and upper cutoff), or resistant (above the upper cutoff).
Clinical reduced susceptibility cutoffs have been defined for many drugs: 4.5-fold for Ziagen, 1.7-fold for Videx and Zerit, 10-fold for Kaletra and 1.4-fold for Viread. Attempts are being made to find clinical cutoffs for all drugs.
Where they have not been found or calculated, the cutoffs are based on laboratory testing averages rather than actual clinical outcome data. These biologic cutoffs, which are based on the natural variability of wild type viruses from patients, are the next best cutoff, while reproducibility cutoffs, which are based on assay variability with repeated testing of patient samples, are the least sensitive. Clearly, it can be confusing to know which cutoff you're dealing with and how helpful it might be in selecting an antiretroviral regimen.
Another important point to remember is that boosting the level of protease inhibitors, either by adding a small dose of Norvir to inhibit CYP3A, the liver enzyme that breaks down protease inhibitors, or by taking the drug with food in the case of Viracept and Aptivus (tipranavir), will increase the drug level and hopefully push the level above the IC50 of that virus. The clinical cutoffs do reflect boosting while the other cutoffs do not.
Also, the two main companies in the field, Monogram Biosciences and Virco Lab, each calculate their cutoffs differently and thus they are different. This can be confusing. Clinicians need to be sure that they're referring to the clinical cutoff produced by the company whose assay they're using, and not trying to take one company's cutoff and use it with another company's assay.
A regimen that fails because of very poor adherence (taking less than 50% of the prescribed doses) will often not show any mutations on the genotype or resistance on the phenotype. This is because there is not enough drug pressure on the virus for it to develop mutations. It is people who take their medication "fairly" well (adherence between 60-95%) who become resistant. Bottom line: adherence should be all or nothing to have the greatest chance of avoiding the development of drug resistance.
Both tests are expensive, with ranges for the genotype from $200-$400 and $800-$1,000 for the phenotype/phenotype GT. This adds to the already mounting cost of providing care and medication to people living with HIV. However, the ability of the tests to help us avoid the use of expensive drugs that are not active against a particular patient's virus usually makes them cost effective in the long run.
Unless these tests are done right in the clinic's lab, it can take 1-3 weeks before the results are available. This delay may be significant if a patient is very sick.
These tests have proved themselves very useful to guide treatment decisions in both treatment-naive patients (those starting HIV medication for the first time) who may have been infected with resistant virus, and especially in patients whose regimens are failing. When a genotype done on a patient gives a different interpretation than the phenotype done at the same time, that is called "discordance." Discordance between the two types of tests is common but may improve as we update the genotypic algorithms to include new mutations and the potential for interaction between mutations.
10. Is it Worth it?
As noted above, there are some factors that can overwhelm resistance mutations and make resistance testing irrelevant. First is adherence. If a patient isn't taking enough of their pills, or taking them as prescribed, the regimen might fail even without resistance mutations being present. Doing resistance tests on this patient's virus would probably be a waste of time and money. The most important thing is making sure that the patient understands the importance of taking the medications as prescribed, and working to find a regimen that is effective against their virus and fits their lifestyle.
A second factor is Norvir boosting of protease inhibitor levels. In some cases, the addition of Norvir to a protease inhibitor-containing regimen can result in blood levels that are far above the cutoffs, or far above the amount needed to suppress virus -- even if it has some resistance mutations. Norvir boosting, in some cases, overcomes viral resistance.
In deciding what regimen to choose, multiple individual factors have to be taken into account, including drug history, patient preference and the patient's medical history, side effects, adherence and dosing schedule. Resistance testing has taken a firm role in this decision making process, but the limitations described above need always to be accounted for.
Trevor N. Hawkins, M.D. received his medical training in Birmingham and London. He practiced in India, and for the past 22 years has been in Santa Fe, New Mexico, first in general practice and then specializing in HIV care. He has participated in numerous clinical research studies of HIV medications, and serves on several pharmaceutical advisory boards. Dr. Hawkins has received several awards and recognitions for his service as an HIV care provider.
Got a comment on this article? Write to us at firstname.lastname@example.org.
This article was provided by Positively Aware. It is a part of the publication Positively Aware. Visit Positively Aware's website to find out more about the publication.