William Check, PhD
From the start of the AIDS epidemic, physicians who care for HIV-infected persons have been early adopters. Soon after viral load-the amount of HIV RNA in the blood-was demonstrated to be associated with prognosis, assays to measure HIV RNA concentrations became widely used to monitor disease stage and response to therapy. And it was only a few years after multidrug combinations against the virus became available that significant reductions in progression to AIDS-defining illnesses and in mortality from HIV disease were demonstrated on a population basis, reflecting broad deployment of these highly active regimens.
Most recently, assays to detect resistance to antiviral drugs have become de rigueur for optimal management of many HIV-infected persons. An International AIDS Society-USA Panel on antiretroviral drug resistance testing (JAMA. 2000;283:2417- 2426) and other groups recommend them for use in several clinical situations. Emphasis on the utility of such testing is much more emphatic in the panel’s current statement than in its first report, published less than two years ago. Says panel chair Martin Hirsch, MD, professor of medicine at Harvard Medical School and director of clinical AIDS research at Massachusetts General Hospital, "Over time new data became available so that we could refine the suggestions with more confidence than in the first go-around." Data have been published, he says, "that validate the utility of genotypic analysis, and also, but to a lesser extent, phenotypic analysis, to make decisions in clinical situations that we face daily."
Adds panel member John Mellors, MD, chief of the Infectious Disease Division and director, HIV/AIDS Programs, at the University of Pittsburgh Medical Center, "There is no question that in the last year there has been a steady accumulation of data supporting the use of resistance testing." Retrospective analyses had consistently shown that, in treatment-experienced patients, HIV drug resistance is predictive of response to therapy. "That is unequivocal," Dr. Mellors says.
"Now," he adds, "there are several prospective studies showing that if physicians are provided with resistance test results, their patients respond better virologically to the prescribed antiretroviral regimen compared to patients of physicians who don’t have such information."
Calling himself a "big fan of resistance testing," Mark Wainberg, PhD, professor of medicine at McGill University, Montreal, director of McGill’s AIDS Center, and immediate past president of the International AIDS Society, agrees. "There is absolutely no doubt that resistance testing is useful in the majority of cases. If we are judicious in how we use this technique," he says, "it will improve patient outcomes and save money."
Practice patterns already reflect study data to some degree, says Tim Alcorn, PhD, director of infectious diseases at the Laboratory Corporation of America Center for Molecular Biology and Pathology in North Carolina. "There has been a significant increase in volume" in resistance assays, Dr. Alcorn says. "We are seeing usage that appears to be in compliance with the guidelines."
At Specialty Laboratories in Santa Monica, Calif., the volume of HIV resistance tests also has grown in the past five years. "The demand reflects clinical utility and a growing understanding of the need for individualized treatment regimens," says Andreas Bakker, PhD, Specialty’s director of genotyping and molecular technology.
But using the resistance assays optimally is not simple. "Certainly adoption of resistance testing is going to increase the complexity of HIV care," says panel member Daniel Kuritzkes, MD, associate professor of medicine at the University of Colorado Health Sciences Center, Denver, and co-director of the Colorado AIDS Clinical Trials Unit. For example, the panel recommended "expert interpretation" of assay results.
And the degree of benefit from resistance assays is limited by cross-resistance among classes of antiretroviral drugs. Says Graeme Moyle, MD, PhD, director of HIV Research at Chelsea and Westminster Hospital, London, "You can only get returns if you have drugs available that are effective in that situation." Dr. Kuritzkes adds, "As new drugs are developed and come on line, and assuming that full understanding of resistance to these drugs is part of the package submitted to FDA, that will drive the utility of these tests in more advanced or experienced patients."
Data from three major prospective studies-GART, VIRADAPT, and VIRA 3001-support the value of HIV resistance assays in patients who fail their current regimens. In all three studies, virologic response was superior among patients whose new therapy was selected in accord with results from either genotypic or phenotypic resistance assays.
Guiding therapy with the results of resistance assays reduced HIV RNA by about a half log on average, which translates to about a 30 percent reduction in the risk of disease progression. Is this benefit clinically important? Yes, says Dr. Kuritzkes.
"Given that resistance testing will be applied in settings where patients are already failing therapy," he says, "achieving complete suppression of HIV replication is not likely. But a half-log reduction, although not impressive compared to what we can achieve in treatment-naive patients, confers a significant benefit."
Moreover, averages hide individual benefits. In GART and VIRADAPT, about one-third of patients treated according to resistance data achieved undetectable levels of HIV RNA (<200 or <500 copies/mL). "And patients in VIRA 3001 did much better," Dr. Kuritzkes says. About 60 percent of patients in VIRA 3001 who were being treated based on resistance data had HIV RNA <400 copies/mL. "These were first-time failures," Dr. Kuritzkes notes. "As assays are used more often and earlier, they may enable greater benefit."
Dr. Mellors expresses a similar appreciation. "We are asking an awful lot from a single test that only detects resistance in the dominant viral species present in a patient to make a major difference in treatment response over 24 weeks," he says. Other important variables influence treatment response, such as disease stage, adherence to medication, pharmacokinetics of individual drugs, and availability of agents with activity against resistant strains. "Resistance tests don’t improve adherence or enhance the drugs’ pharmacokinetics or substitute for new drugs," Dr. Mellors says. "I think it is remarkable that a single test result shows a difference in response. But we also need to optimally manage patients, such as with medication adherence support systems."
An additional trial from France, called NARVAL, found no advantage of resistance testing at three months and a minimal advantage for genotyping at six months. Dr. Wainberg calls NARVAL "an outlier."
"NARVAL is a very important study and seems to have been well done," Dr. Wainberg acknowledges. But the majority of patients in NARVAL had more advanced HIV disease than patients enrolled in GART, VIRADAPT, or VIRA 3001. "So they may have had such complicated genomic resistance patterns that expert interpretation and advice might not have made as big a difference," Dr. Wainberg says. "In fact," he adds, "there is some feeling that phenotype has its best place in populations where genotype is so complicated that you can’t tell for sure what is going on."
Providing another perspective, Dr. Moyle points out that the group in NARVAL that didn’t receive resistance testing took more "new" drugs-drugs they had never taken before-than the group randomized to testing. "Resistance testing enabled clinicians to recycle drugs into the patient’s regimen and achieve a similar virological response without using as many treatment options," he explains. So resistance testing may buy time without using up effective drugs with which the physician might be able to formulate a better regimen later when newer drugs emerge.
Genotyping assays have been thought to be less useful in persons who have failed multiple regimens, as Dr. Wainberg noted. But, says Charles Boucher, MD, PhD, clinical virologist at the University Hospital, Utrecht, the Netherlands, "The HAVANNA trial [conducted in Europe and involving genotype and expert advice by software] showed that genotyping gave a better result throughout all the different patient categories. We formerly thought if a patient had failed their third regimen, phenotyping would be better. But in HAVANNA we observed that genotyping still had an impact even in patients who had cycled through one or two regimens already." Further work will clarify these issues.
Based on these studies, HIV resistance assays are now recommended for patients who experience treatment failure on their first or subsequent regimen or who do not respond to a potent regimen. The assays help to determine whether resistance is contributing to treatment failure and what drugs to use in the next regimen to optimize response. "In clinical trials, response to treatment is clearly and consistently related to the number of new drugs prescribed to which the patient’s virus is sensitive," Dr. Mellors says. And studies have shown that physicians given results of a resistance test choose with greater accuracy drugs that are active against a patient’s virus.
In addition, the panel’s statement says clinicians should "consider testing" for HIV resistance in two groups of patients: those with established infection that has not yet been treated and those with primary or acute infection. In both types of patients, the value of resistance assays remains to be proved but is a source of intense discussion.
Whether to test for resistance in untreated patients with established infection of unknown duration is a difficult decision. If such patients harbor a resistant virus strain, either acquired at primary infection or generated by mutation during viral replication, it is likely to be present as a minor quasi-species, probably below the sensitivity of a resistance assay. Dr. Kuritzkes notes that many HIV-positive persons not yet treated have been infected for a long time. Many acquired their virus prior to the current therapeutic era, so they are less likely to have a (primary) resistant virus. But some data indicate that in cities where the epidemic has gone on longest and treatment has been available longest, such as New York and Los Angeles, there may be a much higher prevalence of primary resistance. "Eventually we may need to establish guidelines based on local prevalence rates," he says.
"You could make an argument that everybody should get a genotype before their first regimen is initiated," Dr. Wainberg says. He cites the HIV care program in British Columbia, which he calls "the longest running program in genotyping." In that program, genotyping is done on almost everyone about to embark on their first course of therapy, as well as those with treatment failure, Dr. Wainberg reports.
Clinical researchers have become increasingly interested in the second group-patients in the acute phase of HIV infection, within 30 to 60 days of acquisition and before seroconversion. Use of special "detuned" assays makes detection of acute infection possible in patients who come in with nonspecific symptoms of viral infection and a suspected exposure. Work from Bruce Walker, MD, and colleagues at Massachusetts General Hospital has showed that prompt reduction in viremia to undetectable levels in these patients preserves immune function. "We don’t want to make a mistake in prescribing therapy in that group," says Dr. Mellors, "so we need to know the resistance profile of the strain in those patients. There is enough primary resistance around to warrant it. And the stakes are very high."
Data from Dr. Wainberg and colleagues underscore this point. They genotyped viral isolates from 81 untreated patients who were infected between 1997 and 1999. Mutations associated with resistance to at least one class of antiretroviral drug were found in 30 percent of patients; almost 10 percent had genotypes suggestive of primary multidrug resistance. "Our results are pretty consistent with other work," Dr. Wainberg says. "This is an increasing problem. Drug-resistant viruses are being spread sexually and through injecting drug use. And it is not just in Europe and the U.S., but in developing countries." Transmission of drug-resistant strains may become more problematic in developing countries because of widespread use of less-suppressive one- and two-drug regimens, which Dr. Wainberg calls "a prescription for resistance."
Testing for resistance in primary infection has two purposes, Dr. Wainberg says: It shows how much resistance is present in a population, which is important from a public health standpoint, and it provides crucial treatment information. "You don’t want to use drugs as first-line therapy in patients whose virus is resistant to those compounds," he says.
Dr. Kuritzkes agrees that testing for resistance in patients with acute infection is useful both for epidemiologic purposes and because of "growing interest in providing immediate therapy for such patients." Finding those patients is "challenging," he notes. Acute infection looks like a general viral syndrome, so most patients will be seen by a general internist or an emergency room physician. "More education is needed for primary care practitioners," Dr. Kuritzkes says. "Physicians do monospots at the drop of a hat, even though most come back negative. There needs to be more thinking about HIV when there is a compatible clinical presentation, even when the patient doesn’t seem to be at risk."
Dr. Hirsch is convinced that greater effort will go into detecting persons with acute infection, largely because of Dr. Walker’s data. "His work certainly suggests that treating very early has significant advantages and might allow you to discontinue therapy down the road," Dr. Hirsch says. "These are very provocative observations and, if confirmed by others-and I have no reason to think they won’t be, will make identifying acute HIV infection extremely important." Dr. Hirsch believes professional organizations for HIV care will try to devise ways to detect HIV earlier. "We have been doing that at this institution, as have a number of others, and it’s pretty amazing how many such people you can find if you are looking very hard for them," Dr. Hirsch says. "In that setting, susceptibility patterns can be very helpful."
There may be some difference in the utility of genotypic assays, which detect resistance based on sequencing of the viral genome looking for known resistance mutations, and phenotypic assays, which measure growth of virus in vitro in the presence of antiretroviral drugs. In Dr. Alcorn’s laboratory, genotypic assays are ordered two to three times as often as phenotypic assays, primarily because genotypic assays are faster and less expensive and just as useful for patients failing the first time with uncomplicated therapeutic histories. "Phenotyping may be more useful for complicated therapeutic histories," Dr. Alcorn says.
He believes genotyping may be favored even more with the introduction of virtual phenotyping, a procedure made possible by the creation of a large database containing genetic sequences of a great many clinical HIV isolates along with their known phenotypes. A virtual phenotype is obtained by matching the sequence of a patient’s virus with isolates in the database that have a similar sequence, and extrapolating the phenotype of the isolate in the database to the patient’s virus. Virtual phenotypes strongly correlate to actual phenotypes, Dr. Alcorn says.
At Specialty Laboratories, too, genotyping is the more commonly requested test. "Cost, time to result, and efficacy are the main determinants of the test’s continued popularity," says Specialty’s Dr. Bakker. Evidence is still lacking, he adds, that "phenotyping guides therapy to superior patient outcome" when compared to treatment based on appropriately interpreted genotypes.
Results from phenotypic testing may seem easier to read, especially since they look like an antibacterial susceptibility test, Dr. Moyle says. "Anything above 10-fold change in sensitivity is reported from the laboratory as resistance," Dr. Moyle says. But, he points out, the problem is that phenotypic results are not always clinically relevant. For example, phenotypic resistance to stavudine (D4T) is not always apparent, even when mutations are present and viral load has rebounded. And with protease inhibitors (PIs) a 10-fold increase in sensitivity may not impact efficacy when clinical exposure to the drug is 70- to 80-fold above the effective concentration-which may be true for booster PI regimens such as ritonavir 100 mg bid with saquinavir, indinavir, or the recently approved PI lopinavir. For the nonnucleoside reverse transcriptase inhibitor efavirenz, too, clinical exposure may be much more than 10-fold higher than the effective concentration, Dr. Moyle says.
Both types of tests are improving. Clinically relevant cutoffs for drugs are being defined. "This is critical," Dr. Mellors says.
Dr. Kuritzkes and Richard D’Aquila, MD, of Harvard Medical School are co-chairing a trial just getting underway in which both types of tests will be compared in patients failing therapy.
Where the two types of tests will be done is another difference. "Assays for phenotyping will continue to be done in specialized laboratories," Dr. Kuritzkes says. Phenotyping relies on a technique in which amplified protease and reverse transcriptase gene segments from a patient’s isolate can be easily and quickly inserted into a standard HIV vector to generate infectious viruses that are assayed in an automated system. "Automation allows high throughput and makes the assay economically feasible," Dr. Kuritzkes says. His laboratory used to do "quite a number" of phenotypic assays, mostly for research purposes. "We took six weeks from the time of getting the sample. Now commercial assays can be done with an average two- to three-week turnaround," he says.
Genotypic assays, on the other hand, can be done in a more decentralized fashion, Dr. Kuritzkes says. Whether a laboratory does the genotypic assay itself will depend to a great extent on economics, local demand, and available expertise. "Applied Bioscience/Perkin-Elmer and Visible Genetics are developing kits and systems that can be placed in a hospital-based molecular biology laboratory to produce high-throughput genotypes and generate reports useful for clinicians," he says. Hospital laboratories and centers that see a large number of HIV-infected individuals may find it more cost-effective to do assays in-house rather than as sendouts, much like HIV RNA testing.
Before genotypic assays become routine in clinical laboratories, quality control needs to be addressed, Dr. Boucher cautions. "QC is an issue that people have largely ignored, because most genotyping has been done in research laboratories, which normally are not as focused on QA and QC as clinical laboratories," he says. A group of investigators led by virologist Rob Schuurman, MD, of Utrecht, did a QC probe of HIV resistance genotyping. They sent a panel of coded samples containing various mixtures of clinically relevant resistance mutations to about 60 laboratories-"all the laboratories in the world doing this testing," Dr. Boucher says. "We observed that quite a number of laboratories missed the correct interpretation if a mutation was present as a mixture with wild type virus. We also observed that the more experienced the laboratory and the better trained its people, the better were the results."
Missing a mutant virus present as a minor species would be relevant to patients who have been on therapy for a short time and those who have stopped therapy. "Generally, if a patient has been on a stable regimen for a long time, you would expect most mutations to become dominant in the viral population," Dr. Boucher says. A minor mutant strain might also be present in a patient who has been newly infected.
Dr. Boucher advises laboratories that are starting to do genotypic HIV resistance testing to build in run controls and QC controls and to adhere to an independent QC program. "I think it is good that more clinically oriented laboratories are starting to do this assay," Dr. Boucher says. "They are more used to thinking in terms of QA and QC. But they must realize that it is a difficult assay to do, probably the most error-prone assay in clinical virology."
Asked about the panel’s call for "expert interpretation" of resistance assay results, Dr. Hirsch says, "In an ideal world we could have not only the commercial laboratory’s interpretation of the results it provides but also someone local with expertise in interpreting HIV resistance test results to give practicing clinicians useful guidance." Physicians in clinical practice who see hundreds of patients don’t always have the time to keep up with the latest advances in this complex area. "I have no doubt that [the amount of expert advice] could be improved," Dr. Hirsch says. "In metropolitan areas with major medical centers there are people with that kind of expertise, but it is not available everywhere."
"We do need someone between the test result and the treating physician," Dr. Mellors says. "But I think what will happen is that the number of people caring for HIV-infected patients will shrink and their collective expertise will increase. Just as you don’t treat cancer in general practice," he continues, "you shouldn’t be treating HIV in general practice unless you do it enough so that you can interpret the test results." Treating HIV disease "is not for the faint of heart," Dr. Mellors says. In the meantime, he recommends consultation with an expert. "If anybody calls me, I would speak with them," he says.
Obtaining such expert consultation depends on who is available locally. "So-called experts don’t really have time to sit down and deal with calls all day," Dr. Kuritzkes says. "They are busy doing their own clinical or research work." He foresees systems that may improve this situation, such as a move to set up a uniform rules-based interpretation system. Experts would regularly review recent data and make tables for suggested interpretation. Dr. Kuritzkes’ sense from talking to companies making kits is that they would be happier if an external body reviewed the data and formulated clinical significance for various mutations. "We are trying to establish some standards across industry so that automated reports would build in these interpretative systems," Dr. Kuritzkes says, "but there are some regulatory issues."
He, too, acknowledges Virco’s (of Mechelen, Belgium) virtual phenotype, a computer-based algorithm that relies on Virco’s large database of linked phenotypic and genotypic data for individual isolates. "It shows promise but clearly requires prospective validation," Dr. Kuritzkes says. "And anyway, our goal should be to determine how a given genotype predicts the outcome with drugs A, B, and C, not how well can I correlate genotype to phenotype."
Of the limited number of experts, Dr. Wainberg says, "I don’t see a dilemma at all. I see an opportunity." It will be impossible to instruct every practitioner on the intricacies of genotyping, of course, and experts are busy people who cannot answer calls all day about interpreting genotypes. "What we should be doing," he says, "is training limited numbers of people to act as resource individuals for physicians who can come to them with questions about mutational profiles. And we can offer this as a reimbursable service."
Along this line, Dr. Hirsch notes that the Infectious Disease Society of America has formed a subset of physicians who specialize in HIV care. This group includes infectious disease specialists but also those from other specialties who see a lot of HIV patients, such as physicians in oncology and even general internal medicine and pediatrics. "HIV care is a sub-branch of medicine now, and it will be increasingly recognized as a specialty as things get more complex over time," he says.
With the large number of assays done by a commercial laboratory, the demand for expert interpretation would be expected to be high. Answering such questions can be complicated, Dr. Alcorn says. For phenotypic assays, what is a relevant clinical cutoff? For genotypic assays, there are so many combinations of mutations that need to be interpreted. "Having said that," he points out, "evidence from clinical trials, VIRADAPT in particular, where a rules-based algorithm was used, demonstrate that even with complexity you can devise a set of rules that will lead to clinically relevant results. Those rules are being established," Dr. Alcorn adds. Physicians are becoming aware of Web sites that help them interpret the results of resistance assays, he notes: http://hivdb.stanford.edu and http:// idiotype.lanl.gov (Los Alamos maintains the latter site). In any event, Dr. Alcorn reports, "I don’t get a whole lot of calls asking about interpretation of mutations."
An important milestone could be reached soon: The first FDA approval of an HIV resistance assay. Visible Genetics has submitted its TruGene genotyping kit for FDA approval. "It would be the first 510(k) HIV resistance assay," Dr. Kuritzkes notes, "and it would make a difference." A number of states are holding back on funding of these assays because they are not approved. "When FDA approval arrives, it will drive a number of states to fund these tests," Dr. Kuritzkes says. "I don’t know how it will shift usage patterns," he adds. Many reference laboratories are now using an in-house kit; how many will shift to the approved kit is unclear.
Dr. Alcorn agrees that FDA approval of a genotyping kit would speed up reimbursement policies. "We are now at the point where it requires communication back and forth to payers about these tests," he says. Making them routinely reimbursable will go a long way toward making them more widely used.
William Check is a freelance medical writer in Wilmette, IL