How fourteen anti-HIV drugs could provide more bang for the buck.

Background. Controlling HIV, at least in the short run, requires that persons infected with the virus be fully suppressed or so goes the theory. Yet the point of suppression keeps moving with time. In the early days of viral load using bDNA (branched DNA), patients were told that a viral load of less than 10,000 copies/mL constituted an undetectable viral load. Translated to PCR (polymerase chain reaction), the test more commonly used today, that figure would be a viral load of 20,000 copies/mL--high enough to give any patient and doctor pause.

Currently, a viral load below 400 copies/mL is not low enough to reassure the more rigorous clinician that the virus is fully suppressed. Evidence keeps mounting that lower and lower viral load counts are necessary to keep viral rebounds at bay. What then, is the ultimate line in the sand for viral loads? The answer to that question remains unclear and is further complicated by side effects and long-term complications of anti-HIV therapies.

Even if we could for the moment put aside the question of side effects and complications, the rate of viral rebound among patients using highly active antiretroviral therapies (HAART) remains high. To the outside world this rebound rate may seem surprising given the number of anti-HIV drugs available and their possible combinations. Yet the seemingly large number of approved antiviral therapies is deceptive because of resistance and cross-resistance, and because of the recommended practice of changing at least 2 and preferably 3 drugs following virologic rebound. This mandate to switch multiple drugs has greatly diminished the individual value of each drug.

What if, instead of this wholesale switching of drugs, each drug could be switched individually as its failure was detected? The value of such single switches is apparent, but only now are scientists and clinicians beginning to learn the intricacies of how drug regimens fail and only now is there some early evidence that it may be possible to switch some drugs one at a time.

What is drug failure? Drug failure is generally characterized as some change in the virus that renders it less susceptible to a drug. Scientists have found that certain changes in the genetic make-up of a virus can render a drug useless. For example, a change in position 184 from methionine to valine in the reverse transcriptase enzyme gene of HIV renders that virus completely resistant to the drug lamivudine (Epivir). Patients with that genetic change or mutation (referred to as M184V) in their HIV can take lamivudine, but the virus will continue to replicate as if the drug was not present. This type of resistance, where mutations in the genetic composition of a virus are detected and associated with the diminished efficacy or total failure of a drug, is generally referred to as "genotypic" resistance. Some drugs stop working with a single viral mutation, while other drugs become ineffective only after the virus has acquired several mutations. Resistance is not always an all-or-nothing proposition: some mutations merely hinder a drug's efficacy, while a single mutation can completely cripple other drugs. However, given enough time HIV replication can render any drug inefficacious.

Phenotypic testing is simply another way to measure the viral changes leading to resistance to a particular drug. This form of testing does not rely on detecting changes in the genetic make-up of the virus, but instead places the virus in the presence of the drug and measures the ability of the virus to grow in culture. The drug fails if the virus grows well despite the presence of the drug. Thus, by some current phenotypic tests, a 2.5 to 4.0 fold loss in susceptibility is considered to render the virus resistant to the tested drug. There is no set standard fold loss that establishes phenotypic resistance and this value can vary from test to test.

Resistance occurs when HIV is allowed to replicate in the presence of a drug. The higher the level of replication, the greater the chances that the virus will mutate and render the drug ineffective. Conversely, the lower the level of viral replication, the smaller the chance of developing resistance to any particular drug. The goal of achieving lower and lower viral loads arises from this basic premise of resistance.

Is any detectable viral load in the presence of a drug evidence of resistance? Experience has shown that any drug or combination of drugs that does not fully suppress HIV will eventually lead to resistance. Does this then imply that any detectable viral load in the presence of a drug is evidence of resistance? This is not so clear. For example, zidovudine (Retrovir) monotherapy can partially suppress HIV for some time before evidence of viral resistance develops. Thus, a detectable viral load in the case of zidovudine monotherapy may be first predictive of the potency of the drug and only later become evidence of resistance. Simply stated, zidovudine monotherapy may not by itself have the power to suppress all measurable viral replication. Thus, the replication that is detected, at least at an early stage, may not be caused by viral mutations normally associated with resistance but simply a result of a lack of potency. Lack of potency may allow for viral replication in the presence of a drug, but it might not in and of itself denote resistance.

In the test tube scientists have sorted out with some certainty the potency and resistance patterns of individual antiretroviral drugs. Still, the unique contribution of each drug to the potency of a multidrug regimen and the pattern, order and interrelations of the development of resistance to each component of a multidrug regimen in a patient are still a mystery.

Early studies of protease inhibitors as monotherapy demonstrated that resistance could develop quickly. Subsequent studies of protease inhibitors in combination with other drugs, where the drugs were added sequentially, led many to assume that it was the protease inhibitor that failed first and was principally responsible for viral rebound since these were the mutations first detected after viral rebound. In cases where protease inhibitor mutations could not be found, it was presumed that the patient had failed due to lack of adherence (JAMA, 283:2, p.232, 2000). The question of the overall potency of the regimen was not considered and thus a failed regimen required the automatic substitution of the protease inhibitor and as many of the other drugs as possible.

Expected results from induction/maintenance studies. In the late 1990s several groups of scientists conducted trials to determine if it would be possible to start patients on fully suppressive HAART regimens, maintain this suppression for a period of time and then treat with fewer drugs while still maintaining full suppression. This concept was referred to as induction/maintenance. One of the 3 major studies conducted was AIDS Clinical Trials Group (ACTG) 343.

ACTG 343 started patients on the regimen of indinavir (Crixivan)/zidovudine/lamivudine. The induction period lasted 6 months, at which time participants with viral loads below 200 copies/mL were randomized to 1 of 3 maintenance arms: a) indinavir alone, b) zidovudine/lamivudine or c) indinavir/zidovudine/lamivudine. The theory behind this design was that if patients were first brought to a certain level of suppression, fewer drugs might subsequently be required to maintain that level of suppression. In essence this was an attempt to simplify drug regimens and curb potential side effects.

The results of ACTG 343, as well as of other induction/maintenance trials, were disappointing. In ACTG 343, a full 23% of patients randomized to the indinavir arm and the zidovudine/lamivudine arm had viral rebounds. Patients who experienced rebounds in these arms were allowed to return to the induction three-drug combination once their viral loads exceeded 200 copies/mL. By comparison only 3% of patients remaining on the three-drug combination during the maintenance phase of the trial had viral rebounds.

Unexpected results from induction/maintenance trials. After the induction/maintenance trials were stopped, further investigation into a subset of patients continued. In a paper published in the Journal of the American Medical Association (283:2, p. 229, 2000), Diane Havlir, MD, et al. reported on the loss of drug susceptibility by 9 patients who had been randomized to the indinavir arm and 17 patients who had failed on the three-drug arm.

In the paper, Havlir et al. report that the 9 patients who had been randomized to the indinavir-only arm had rebounded between 2 and 8 weeks after switching to the maintenance regimen. Despite the quick rebound in these patients, no phenotypic loss of susceptibility was detected to indinavir, nelfinavir (Viracept), ritonavir (Norvir) or saquinavir (Fortovase). Further, no primary indinavir mutation could be found in the plasma of these patients. Five of the 9 patients discontinued the study, but 4 chose to stay on the study and returned to the three-drug combination. As of the publication of the article, 3 of 4 patients returning to the three-drug combination had again achieved viral loads of less than 200 copies/mL and maintained them for 7 to 10 months. The fourth patient had achieved a viral load of less than 200 copies/mL but then rebounded at 4 months.

Among the 17 patients who had rebounded while on the three-drug combination, none had any phenotypic loss of susceptibility to indinavir despite viral rebounds ranging from 1864 to 138,989 copies/mL. Fourteen patients had genotypic (as confirmed by the presence of M184V) and phenotypic resistance to lamivudine, while 3 retained phenotypic susceptibility to lamivudine. The lack of evidence for resistance to indinavir is surprising given the viral rebound levels in these patients.

What conclusions can be drawn? Among the patients who received a maintenance regimen of indinavir only and rebounded, one possible explanation may be that indinavir monotherapy has suboptimal potency and that viral rebound can occur without resistance to indinavir being present. Further, if that is the case, resistance to indinavir theoretically could have been delayed for months. Among the rebounders taking the three-drug combination, the situation may be more complex. The triple-drug combination may not have been potent enough to provide full suppression, in which case Havlir et al. point to lamivudine resistance as accounting for the viral outgrowth. The single mutation that causes resistance to lamivudine makes the resulting virus more fit and able to outgrow other mutant viruses. Given the continuing pressure of lamivudine, the outgrowth of the more fit lamivudine resistant virus may be evident more quickly in genotypic and phenotypic tests than other mutant viruses.

Havlir et al. point to lamivudine as a "weak link" in a suboptimal treatment regimen. To further make their point they cite a study of efavirenz (Sustiva) in combination with indinavir where patients with rebound were more likely to have the mutation required for efavirenz resistance before any indinavir mutations occurred (Ibid., p. 233). Efavirenz, like lamivudine, is crippled by the appearance of a single viral mutation.

The tantalizing suggestion of this substudy is that in early viral rebound only these weak-link therapies are lost to resistance and substituting only these drugs could result in a fully suppressive regimen once again. This theory of weak links is yet unproven and it is possible that resistance to other drugs in the regimen could be hiding among the minority variant populations that genotypic and phenotypic tests are not equipped to measure.

However, if this theory proves to be true and drugs can be substituted one by one, this would create a great boon for persons starting therapy. One way to verify these findings would be to do single drug substitutions and see what results. This, of course, presents a danger to the patient. Another strategy of some clinicians is to change more than one drug but reserve these other drugs to be reintroduced if the patient starts running out of drug options later. Genotypic and phenotypic tests that could accurately measure small populations of viral variants would be most helpful. Clearly, the present phenotypic and genotypic tests with their inherent inability to measure virus subpopulations that constitute less than 10 or 20% of the viral population are not up to the job.

Design of Induction/Maintenance Trail of ACTG 343

[ILLUSTRATION OMITTED]

The theory behind the induction/maintenance design was that if patients were first brought to a certain level of suppression, fewer drugs might be necessary to subsequently maintain that level of suppression. Many patients failed the monotherapy and dual therapy maintenance arms of this study. Later other patients failed the three-drug arm of the study. The question posed is whether patients rebounded due to resistance or the suboptimal potency of the maintenance regimen.