Further Thoughts on Drug Resistance in HIV Infection
Further Thoughts on Drug Resistance in HIV Infection
Last month I discussed the growing importance of multidrug resistance among patients with newly acquired HIV infections as well as in those receiving long-term HAART. There are several additional issues that warrant examination.
First, in terms of the contribution of medication adherence to the resistance equation, some simple, commonsense considerations may apply. A new, retrospective analysis by Bartlett's group at Duke University compared 3 HAART regimens of equivalent clinical efficacy. The researchers found that daily pill burden was a powerful predictor of compliance, with pill count inversely correlated with a plasma viral load below 50 copies/mL at week 48 of treatment (P < .01). This is going to become a potentially greater problem, at least in the near term, as more persons move on to salvage therapies. A recent study of virologic failure among those previously exposed to only 1 class of antiretroviral drugs showed that 2 nucleosides plus nelfinavir and efavirenz were required for maximum viral suppression: 79% showed undetectable virus (below 50 copies/mL) at 48 weeks.
Second, it appears that recognition of intermittent bursts of viremia while receiving a stable HAART regimen -- be they related to "drug holidays," transient immune activation (via infection, allergic reaction, etc), or even lab variations and errors -- is frequent. Yet, these bursts were not a harbinger of virologic failure, nor did they require a change in therapy. In one study, intermittent viremia -- defined as a single viral load greater than 50 copies/mL in a patient with otherwise undetectable virus -- occurred in 40% of a cohort of 343 patients. Thirteen percent of this group had 2 consecutive vRNAs greater than 50 copies/mL. Yet, no long-term adverse effect was noted. The authors concluded that "virologic failure" might best be defined as 2 consecutive measurements of greater than 200 copies/mL, not 50 copies/mL.
Third, HAART administered outside the western epidemic of clade B HIV-1 may vary in clinical efficacy. A group at Johns Hopkins University noted that some of the amino acid polymorphisms that occur among viral proteins in the clades of HIV-1 more common in Africa (types A and C) are associated with drug resistance in type B virus. In studying viruses isolated from drug-naive adults in Uganda, these researchers found a higher biochemical fitness of proteases from clades A and C compared with clade B in the presence of all 4 protease inhibitors (PIs) tested (indinavir, ritonavir, saquinavir, and nelfinavir). This phenotypic resistance was up to 7-fold greater for type A than for type B virus with some PIs.
Whether this in vitro difference will translate into treatment failure is of extreme interest, but it is unclear at this point. Indeed, the pharmacologic drug levels that are usually achieved would overcome this relative difference in inhibitor efficiency.It is also uncertain how the resistance mutations in clade A and C viruses being selected for in Africa under the pressure of HAART will interact with these naturally occurring differences in HIV-1 proteases. Normally, PIs cause compensatory mutations in clade B viruses to overcome reductions in viral fitness associated with early drug-resistant breakthroughs,thus delaying the appearance of large drug-resistant populations. Whether this will occur with equal facility among clade A and C isolates is unknown.
Fourth, many clinicians believe that use of genotypic resistance testing will help retard the development of multidrug resistance. But controlled trials demonstrating this for prolonged periods are lacking. Scientists working at the Los Alamos National Laboratory have published a model suggesting that reliance on such assays, which currently do not detect "minor" viral subpopulations -- those with frequencies less than 10% to 25% -- may lead to
erroneous decisions in choice of therapy. Their mathematic simulations predict that only by simultaneously reducing the basic reproductive number of all the strains initially present in a given patient would consistent reductions in the probability of treatment failure occur. That is not particularly surprising. But what was unexpected was the finding that a drug regimen that is very effective against wild-type strains but only marginally effective against intermediately sensitive strains was associated with a much higher probability of emergence of a new resistance mutation than was a drug regimen marginally effective against all strains present.
A final issue is treatment failure among patients who swear strict adherence to potent HAART regimens to which their viral populations, assessed by resistance testing (with its limited accuracy, as discussed above), should be sensitive. Some host-dependent factors can be relatively easily explored; differential drug metabolism, on a genetic basis or through drug-drug interactions, may be ruled out by pharmacokinetic studies. Yet, clinical drug resistance still occurs. In this instance, multidrug transporter effects may be involved.
HIV PIs interact with 2 multidrug efflux transporters, P-glycoprotein (Pgp, also known as the MDR1 gene product) and multidrug resistance protein 1 (MRP1). These proteins are expressed on several types of cells, including CD4 T lymphocytes and CD34 hematopoietic stem cells. Pgp presumably functions in normal tissue as a defense mechanism against toxic substances. It is an energy-dependent pump that moves hydrophobic molecules and peptides from the intracellular to the extracellular compartments.
In vitro, the initial rates of PI uptake and the steady-state intracellular concentrations of the PIs are reduced substantially in T-cell lines with augmented expression of Pgp or MRP1. Inhibitors of the transporter proteins, GF 120918 and MK 571, respectively, reverse these effects. (GF 120918 has been used in clinical trials with anticancer agents.) But the involvement of these systems in HIV disease is far from straightforward. Pgp is abundantly expressed on endothelial and other cells that form the blood-brain and blood-testis barriers, as well as cells responsible for GI absorption of drug. However, overexpression of Pgp itself can inhibit the infectivity of HIV, probably by affecting viral fusion. And PIs can either induce or partially suppress Pgp, depending on the cell type investigated.
The MDR1 gene encoding Pgp is polymorphic, and one of these changes, known as C3435T, correlates with its intestinal expression. Those homozygous for the T allele have substantially lower Pgp expression in the GI tract compared with C-allele homozygotes, leading to genetic differences in bioavailability of certain drugs -- typically, enhanced absorption. Of interest in terms of potential racial and ethnic differences in the efficacy of anti-HIV drugs is the fact that 83% of West Africans and 61% of African Americans are C-homozygous, compared with only 26% of whites (P < .001). The consequences of this increased frequency of the C/C genotype in blacks could be lowered bioavailability of certain HAART components.
In peripheral blood lymphocytes derived from HIV-positive patients receiving HAART, PIs inhibited Pgp function, with ritonavir being the strongest inhibitor and indinavir the weakest. (Lopinavir and amprenavir were not tested.) In contrast, zidovudine and other nucleoside reverse transcriptase inhibitors are substrates for Pgp yet do not inhibit its function; resistance to these agents has been linked to Pgp expression in vitro. But the suggestion that designing PIs that are not substrates for Pgp or MRP1 might be valuable is premature. Clinical, immunologic, and virologic correlations with expression and function of these drug transporters would be most welcome.h
Last month I discussed the growing importance of multidrug resistance among patients with newly acquired HIV infections as well as in those receiving long-term HAART. There are several additional issues that warrant examination.
First, in terms of the contribution of medication adherence to the resistance equation, some simple, commonsense considerations may apply. A new, retrospective analysis by Bartlett's group at Duke University compared 3 HAART regimens of equivalent clinical efficacy. The researchers found that daily pill burden was a powerful predictor of compliance, with pill count inversely correlated with a plasma viral load below 50 copies/mL at week 48 of treatment (P < .01). This is going to become a potentially greater problem, at least in the near term, as more persons move on to salvage therapies. A recent study of virologic failure among those previously exposed to only 1 class of antiretroviral drugs showed that 2 nucleosides plus nelfinavir and efavirenz were required for maximum viral suppression: 79% showed undetectable virus (below 50 copies/mL) at 48 weeks.
Second, it appears that recognition of intermittent bursts of viremia while receiving a stable HAART regimen -- be they related to "drug holidays," transient immune activation (via infection, allergic reaction, etc), or even lab variations and errors -- is frequent. Yet, these bursts were not a harbinger of virologic failure, nor did they require a change in therapy. In one study, intermittent viremia -- defined as a single viral load greater than 50 copies/mL in a patient with otherwise undetectable virus -- occurred in 40% of a cohort of 343 patients. Thirteen percent of this group had 2 consecutive vRNAs greater than 50 copies/mL. Yet, no long-term adverse effect was noted. The authors concluded that "virologic failure" might best be defined as 2 consecutive measurements of greater than 200 copies/mL, not 50 copies/mL.
Third, HAART administered outside the western epidemic of clade B HIV-1 may vary in clinical efficacy. A group at Johns Hopkins University noted that some of the amino acid polymorphisms that occur among viral proteins in the clades of HIV-1 more common in Africa (types A and C) are associated with drug resistance in type B virus. In studying viruses isolated from drug-naive adults in Uganda, these researchers found a higher biochemical fitness of proteases from clades A and C compared with clade B in the presence of all 4 protease inhibitors (PIs) tested (indinavir, ritonavir, saquinavir, and nelfinavir). This phenotypic resistance was up to 7-fold greater for type A than for type B virus with some PIs.
Whether this in vitro difference will translate into treatment failure is of extreme interest, but it is unclear at this point. Indeed, the pharmacologic drug levels that are usually achieved would overcome this relative difference in inhibitor efficiency.It is also uncertain how the resistance mutations in clade A and C viruses being selected for in Africa under the pressure of HAART will interact with these naturally occurring differences in HIV-1 proteases. Normally, PIs cause compensatory mutations in clade B viruses to overcome reductions in viral fitness associated with early drug-resistant breakthroughs,thus delaying the appearance of large drug-resistant populations. Whether this will occur with equal facility among clade A and C isolates is unknown.
Fourth, many clinicians believe that use of genotypic resistance testing will help retard the development of multidrug resistance. But controlled trials demonstrating this for prolonged periods are lacking. Scientists working at the Los Alamos National Laboratory have published a model suggesting that reliance on such assays, which currently do not detect "minor" viral subpopulations -- those with frequencies less than 10% to 25% -- may lead to
erroneous decisions in choice of therapy. Their mathematic simulations predict that only by simultaneously reducing the basic reproductive number of all the strains initially present in a given patient would consistent reductions in the probability of treatment failure occur. That is not particularly surprising. But what was unexpected was the finding that a drug regimen that is very effective against wild-type strains but only marginally effective against intermediately sensitive strains was associated with a much higher probability of emergence of a new resistance mutation than was a drug regimen marginally effective against all strains present.
A final issue is treatment failure among patients who swear strict adherence to potent HAART regimens to which their viral populations, assessed by resistance testing (with its limited accuracy, as discussed above), should be sensitive. Some host-dependent factors can be relatively easily explored; differential drug metabolism, on a genetic basis or through drug-drug interactions, may be ruled out by pharmacokinetic studies. Yet, clinical drug resistance still occurs. In this instance, multidrug transporter effects may be involved.
HIV PIs interact with 2 multidrug efflux transporters, P-glycoprotein (Pgp, also known as the MDR1 gene product) and multidrug resistance protein 1 (MRP1). These proteins are expressed on several types of cells, including CD4 T lymphocytes and CD34 hematopoietic stem cells. Pgp presumably functions in normal tissue as a defense mechanism against toxic substances. It is an energy-dependent pump that moves hydrophobic molecules and peptides from the intracellular to the extracellular compartments.
In vitro, the initial rates of PI uptake and the steady-state intracellular concentrations of the PIs are reduced substantially in T-cell lines with augmented expression of Pgp or MRP1. Inhibitors of the transporter proteins, GF 120918 and MK 571, respectively, reverse these effects. (GF 120918 has been used in clinical trials with anticancer agents.) But the involvement of these systems in HIV disease is far from straightforward. Pgp is abundantly expressed on endothelial and other cells that form the blood-brain and blood-testis barriers, as well as cells responsible for GI absorption of drug. However, overexpression of Pgp itself can inhibit the infectivity of HIV, probably by affecting viral fusion. And PIs can either induce or partially suppress Pgp, depending on the cell type investigated.
The MDR1 gene encoding Pgp is polymorphic, and one of these changes, known as C3435T, correlates with its intestinal expression. Those homozygous for the T allele have substantially lower Pgp expression in the GI tract compared with C-allele homozygotes, leading to genetic differences in bioavailability of certain drugs -- typically, enhanced absorption. Of interest in terms of potential racial and ethnic differences in the efficacy of anti-HIV drugs is the fact that 83% of West Africans and 61% of African Americans are C-homozygous, compared with only 26% of whites (P < .001). The consequences of this increased frequency of the C/C genotype in blacks could be lowered bioavailability of certain HAART components.
In peripheral blood lymphocytes derived from HIV-positive patients receiving HAART, PIs inhibited Pgp function, with ritonavir being the strongest inhibitor and indinavir the weakest. (Lopinavir and amprenavir were not tested.) In contrast, zidovudine and other nucleoside reverse transcriptase inhibitors are substrates for Pgp yet do not inhibit its function; resistance to these agents has been linked to Pgp expression in vitro. But the suggestion that designing PIs that are not substrates for Pgp or MRP1 might be valuable is premature. Clinical, immunologic, and virologic correlations with expression and function of these drug transporters would be most welcome.h