Print this page    •   Back to Web version of article

Making Right Turns in the HIV Pharmacology Maze
5th International Workshop on Clinical Pharmacology of HIV Therapy, April 1-3, 2004, Rome

By Mark Mascolini

July 2004


Making Right Turns in the HIV Pharmacology Maze

All roads lead to Rome, they say. But once you get there, none of them leads anywhere. At least it can seem that way to the accidental tourist or the purposeful pilgrim in town for an HIV pharmacology meeting.

The trouble is that Rome has no straight streets. That's what happens when you assemble a city on bumpy terrain over the course of several centuries and never call in a Baron Haussmann to straighten things out. As a result, it's almost impossible to find a direct route from any point A to any point B.

Take, for example, the driving directions proffered by the Cavalieri Hilton, one of the meeting hotels for the 5th International Workshop on Clinical Pharmacology of HIV Therapy, held April 1-3, 2004, in Rome. These scrupulous specs starting at the ring road include five vias, three piazzas, two piazzales, two circonvallaziones, one autostrada, two plain old roads, an array of right and left turns, and one U-turn. Online directions to another meeting hotel, the nearby Mövenpick Central Park, try to pretend you can get there from the ring road by traversing only four vias, one piazza, one circonvallazione, and one road. But we know better.

The roads to Rome may be straight and simple, but the roads in Rome are a maze.

HIV pharmacology can seem straight and simple too: "Drug A" impedes elimination of "drug B." People with crippled kidneys should get half the usual dose of "drug Z." A single nucleotide polymorphism (SNP) at gene site such-and-such favors a better response to "drug D." Or does it? As soon as pharmacology research moves from the straight-and-narrow effects of drug interactions -- and sometimes even sooner -- parsing results can get as perilous as plotting your course from one circonvallazione to another. U-turns are allowed but -- once results are stated -- rarely taken. If one learns anything from attending pharmacology workshops for a half decade, it is this: Interpretation is everything, and it's not easy.

From SNPs apparent only to real-time PCR (research has cataloged about a million) to real big differences apparent to a child -- race, weight, gender, pregnancy -- the human host revels in diversity. The final product, for better or worse, holds more helter and more skelter than a Roman street map, and plying those bio-byways takes the savvy, pluck, and sass of a taxi driver at rush hour.

How Much Does the Host Matter?

How much do these host factors matter in how well antiretrovirals behave (the science of pharmacokinetiquette)? A lot and a little, plenty and hardly at all, depending on the variable invoked and the question ventured. And no more important question confronts HIV medicine today than how antiretrovirals -- either branded or generic -- will behave in millions of untreated people who could not be more diverse in size, shape, habit, habitat, and genetic heritage. This year's Pharmacology Workshop started with the good news that some of these differences matter little, if at all.

Generics and Adherence in Rwanda

Rwanda, the tiny nation in East Africa's lake district that will long be remembered for the 20th century's last great (and the first televised) genocide, still faces a more insidious killer, HIV. Serge Schneider (Laboratoire National de Santé, Luxembourg) [abstract 1] listed some of the grim numbers:

Recalling the Biblical heroine who forestalled a massacre, ESTHER, a collaborative effort by four European and several African and Asian countries, aims to train healthcare workers, procure and maintain equipment, acquire antiretrovirals and monitor their quality, and conduct therapeutic drug monitoring (TDM). With colleagues in Luxembourg and Kigali, Schneider set out to rate the quality of generics being used in the program and reckon how well 70 people adhered to their regimens.

Testing four generic syrups, four tablets, and three capsules to see how closely antiretroviral content matched label claims, they found that levels of 13 of the 14 constituent agents lay within 10 percent of the stated amount. Among those agents with on-target drug contents were three fixed-dose combination tablets -- Triviro (stavudine [d4T]/lamivudine [3TC]/nevirapine [NVP]) and Coviro (3TC plus 30 or 40 mg of d4T). One syrup, Nevimune (NVP), had only 81 percent of the claimed content.

Workers collected unannounced blood samples from 27 people taking efavirenz (EFV) and 43 taking NVP, froze the plasma, and shipped it to Luxembourg for TDM to get a snapshot of regimen adherence. Schneider reported that nine of 70 people (13 percent) had nonnucleoside levels below the low end of the therapeutic range, while the others had concentrations within or above the therapeutic range (Table 1).

Table 1. Drug Levels as a Measure of Adherence in 70 Rwandans

Table 1. Drug Levels as a Measure of Adherence in 70 Rwandans

The ESTHER team is measuring virologic responses in these people, but Schneider did not yet have those numbers.

GIQ Predicts Salvage Response in the Caribbean

Earlier work by Gilles Peytavin (Hôpital Bichat-Claude Bernard, Paris) showed that a genotypic inhibitory quotient (GIQ) can predict virologic response to amprenavir/ritonavir (APV/RTV) after failure of other protease inhibitor (PI) regimens.1 At this year's Pharmacology Workshop he showed the same is true for lopinavir/ritonavir (LPV/RTV) in heavily pretreated Caribbean patients [abstract 2].

The study involved 50 men and 24 women who had taken from one to six PIs and who had one to eight PI mutations (median 3), a median viral load of 4.6 logs, and a median CD4 count of 97 cells/µL. Study participants had LPV levels checked at week 2 and doses adjusted to aim for a target trough range of 3,000 to 7,000 ng/mL.

Fourteen people had LPV minimum concentrations (Cmins) below 3,000 ng/mL at week 2 and had a dose change or adherence counseling. Seven people had dose changes for troughs above 7,000 ng/mL. The proportion of people with on-target LPV levels rose from 36 percent before these adjustments to 64 percent afterwards. After the dose change or counseling, 85 percent had an LPV trough above 3,000 ng/mL compared with 68 percent before those changes.

A six-month intent-to-treat analysis determined that 50 percent reached a viral load below 50 copies/mL. The median load dropped 2.79 logs in that time. Peytavin found no correlation between LPV Cmin or number of mutations and change in viral load at month 6. But the GIQ, defined as "LPV Cmin/number of LPV mutations," did predict virologic success. Seventy-two percent of people with a GIQ above the median cutoff of 2,133 had a six-month viral load under 50 copies/mL. People with a GIQ above the median had a 2.54 times higher chance of a sub-50-copy load at month 6.

Rich, Poor; Female, Male

Results from the studies by Schneider and Peytavin confirm that antiretrovirals -- even generic antiretrovirals -- follow similar pharmacologic footpaths across the world. That does not mean those pathways won't diverge once antiretrovirals reach more than a few people in poor countries and researchers plot their diverse responses. In a review lecture, Francesca Aweeka (University of California, San Francisco) listed several reasons for potential concern in poor countries:

One such race-based genetic difference blipped rudely onto the pharmacologic radar at the 11th Conference on Retroviruses and Opportunistic Infections (11th CROI) earlier this year. Analysis of 190 people enrolled in ACTG A5095 found that non-Hispanic whites cleared EFV 32 percent faster (95 percent confidence interval [CI] 15 percent to 51 percent, P<0.001) than blacks or Hispanics.2 A second study of people in that same trial tracked changes at position 516 of the gene that codes for CYP2B6, the metabolic crankshaft for EFV.3 A TT genotype, which turned up in 20 percent of blacks and 3 percent of whites, correlated with slower clearance and higher levels of EFV, as well as with more central nervous system (CNS) side effects.

At the Pharmacology Workshop, Charles Flexner (Johns Hopkins University, Baltimore) and Courtney Fletcher (University of Colorado, Denver) argued that such findings should not inspire a quick call for dose changes or even TDM in certain populations. If one figured the positive predictive value of a T516T genotype, Flexner noted, it would not pass muster as a diagnostic test. Efavirenz concentrations in blacks and whites in this study overlapped so much, he added, that anyone proposing TDM to all blacks taking EFV should be ready to offer TDM to all whites as well.

Fletcher called for closer collaboration between pharmacologists and statisticians to decipher the relative importance of genetic or other differences that may affect drug levels and response, as in his recent study of saquinavir (SQV) levels in women versus men.4

Flexner and Fletcher agreed that the question raised by such results should not be "How should this change the way I prescribe?" but "Do these results have biologic plausibility?" and "What further research will help us understand this issue better?" That research furnace is fully stoked and already sparking febrile debate. Aweeka tabulated potentially telling ethnicity data from the past few years (Table 2).

Table 2. Effects of Ethnicity on PKs and Response

Table 2. Effects of Ethnicity on PKs and Response

The Pharmacology Workshop offered one study of potential ethnic (and body weight) effects on antiretroviral pharmacokinetics -- in Cambodians -- plus several reports on another host factor that can affect antiretroviral action: coinfection with hepatitis B virus (HBV) or hepatitis C virus (HCV).

EFV Concentrations and Toxicity in Lower-Weight Cambodians

Measuring EFV levels, weight, and toxicity in 21 Cambodians and 51 French patients, Anne-Marie Taburet (Hospital Bicêtre, Paris) found higher concentrations in the lighter Cambodians but no consequent toxicities requiring lower doses [abstract 63]. The cohorts matched closely in age (median 35 years in Cambodia and 39 years in France), but the Cambodians weighed significantly less (P<0.05):

The antiretroviral-naive Cambodians started EFV plus d4T/3TC with a median CD4 count of 62 cells/µL (range 11 to 116 cells/µL). Steady-state EFV levels in a 12-hour post-dose sample averaged 3,580 ng/mL (median 3,133 ng/mL) in the Cambodians and 2,344 ng/mL (median 2,042 ng/mL) in the French. Two Cambodians had readings below the 1,100- to 5,000-ng/mL therapeutic range (one <50 ng/mL and one 1,073 ng/mL), and three had levels above that range (5,286; 8,559; and 9,232 ng/mL). Higher EFV concentrations correlated with lower body weight in both groups, though the slope describing that correlation proved much steeper among Cambodians (r2 = 0.1934 versus 0.0256). Nine Cambodians were taking prophylactic fluconazole, which did not affect EFV levels.

The Cambodians' low weight reflected not only their genetic endowment but also their advanced disease. Four months of therapy boosted their weight an average 2 kg (16 kg in one 24-kg woman). Five people suffered from vertigo, which disappeared in a median of five days. Two people had transient liver cytolysis, and three had immune reconstitution flares (two tuberculosis [TB] and one Cryptococcus meningitis). None of the Cambodians needed to have their EFV dose trimmed.

Antiretrovirals, Hepatitis, and Cirrhosis

While tuberculosis, malaria, and other endemic diseases can complicate antiretroviral therapy in poor countries, in lands of plenty the co-morbidity of greatest currency is hepatitis. Several Pharmacology Workshop studies chronicled antiretroviral vicissitudes in people with hepatitis C virus (HCV) or hepatitis B virus (HBV).

High NVP Levels Do Not Drive Toxicity in People With HCV

Higher pretreatment alanine or aspartate aminotransferase (ALT or AST), coinfection with HBV or HCV, higher pretreatment CD4 count, and female gender all raise the risk of symptomatic hepatitis in people beginning NVP. But high NVP levels in people with HCV appear not to raise the toxic risk, according to results of a small comparative study by Thibaut Lavrut (University Hospital, Nice) [abstract 8].

The study involved 10 people without HCV infection and 10 with HCV and mild chronic hepatitis. The groups were similar in median age (41.5 years with HCV and 39 years without) and median CD4 count (689 cells/µL with HCV and 589 cells/µL without). Median duration of NVP therapy exceeded three years in both groups. The HCV group had a significantly higher median ALT (47 versus 25.5 UI/L) and AST (41 versus 22.5 UI/L) than the non-HCV group (P<0.05 for both comparisons). But ALTs and ASTs measured less than two times the upper limit of normal in all 10 coinfected people.

Measuring NVP levels before dosing and 1, 2, 3, 4, 6, 8, and 12 hours afterwards, Lavrut discerned no significant differences between people with and without HCV:

Lavrut concluded that the higher incidence of liver toxicity among NVP takers with HCV cannot be blamed on higher NVP levels. He proposed that clinicians can prescribe NVP at the usual dose for people with mild chronic hepatitis, as long as they consider the other precautions in prescribing this drug.

Ribavirin Does Not Alter PKs of AZT, d4T, or 3TC

Although pegylated interferon (PegIFN) plus ribavirin yields lower sustained response rates in HCV/HIV-coinfected people than in those with HCV alone, it is the best option for HIV-infected people who need anti-HCV therapy. Ribavirin cannot be prescribed with didanosine (ddI) because it dangerously boosts ddI's intracellular triphosphate. On the other hand, ribavirin cut intracellular phosphorylation of zidovudine (AZT), d4T, and 3TC in cell studies. But a substudy of the international APRICOT trial found that ribavirin doesn't alter plasma or intracellular levels of those three nucleosides in people.

APRICOT randomized enrollees to one of three arms: 180 µg of PegIFN weekly plus placebo, 180 µg of PegIFN weekly plus 800 mg of ribavirin daily, or standard interferon plus ribavirin. The substudy presented by Jean-Michel Gries (Roche Laboratories) compared 31 people in the PegIFN/placebo group with 25 in the PegIFN/ribavirin group who were also taking AZT, d4T, and/or 3TC [abstract 4].

Comparing plasma levels of AZT, d4T, and 3TC before ribavirin or placebo and 12 weeks later, Gries traced virtually overlapping curves for peak plasma concentration and plasma area under the concentration-time curve (AUC). He found no significant changes in intracellular triphosphate levels of the three nucleosides after 12 weeks of either ribavirin or placebo. Nor did triphosphate levels differ much between the ribavirin group and the placebo group. Intracellular concentration-time profiles of the nucleoside triphosphates and corresponding endogenous nucleotide triphosphates varied little during 12 weeks of ribavirin therapy.

Low-Dose Nelfinavir Effective in People With HCV and Cirrhosis

Because liver impairment can slow elimination of PIs, dose reductions of indinavir (IDV), nelfinavir (NFV), LPV/RTV, and APV have been advocated. A study of 20 HCV-coinfected people with cirrhosis found that lower doses of NFV can maintain concentrations of the PI high enough to control HIV replication.

Renato Maserati (IRCCS Policlinico San Matteo, Pavia) tracked NFV levels in four people on long-term NFV therapy and 16 just starting the PI [abstract 12]. Twelve took the full twice-daily dose, while eight started lower doses (250, 500, 750, or 1,000 mg twice daily). Measuring NFV AUC after at least 15 days of NFV therapy, Maserati adjusted the dose to aim for an AUC that fell between 45 and 75 µg•h/mL.

Eight of 12 people in the full-dose group had their NFV dose trimmed, and the group's median AUC fell from 149.5 µg•h/mL before the adjustment to 87 µg•h/mL afterwards. The median AUC measured 91.2 µg•h/mL (range 40 to 106.4 µg•h/mL) in people who started with lowered doses, and none of them had a further dose change.

After a median follow-up of 45 weeks (range eight to 208 weeks), Maserati recorded no significant changes in median:

All 10 people who started the study with a viral load below 50 copies/mL maintained that level of suppression, and four people who began with a detectable load joined the undetectable group. Slicing the NFV dose did not push up viral loads in the others. Except for three people lost to follow-up, everyone was taking NFV when Maserati made this analysis.

Lower EFV Levels in People With HBV or HCV

Efavirenz levels measured eight to 14 hours after dosing proved significantly lower in people coinfected with HBV or HCV than in those with HIV alone, according to results of a 43-person study by Rita Côrte-Real (Hospital of Desterro, Lisbon) [abstract 57]. She monitored 34 men and nine women who had taken EFV for an average 19 months (range 17 to 21); 27 of them (60 percent) were taking EFV as part of their first regimen.

Among eight people coinfected with HBV or HCV, the geometric mean EFV level measured 1.37 mg/L (95 percent CI 1.08 to 1.74 mg/L), compared with 2.06 mg/L (95 percent CI 1.87 to 2.26 mg/L) in those without HBV or HCV (P = 0.003). Côrte-Real also found higher EFV levels in people taking tenofovir disoproxil fumarate (TDF).

PIs in Children

Three Pharmacology Workshop studies scaled PI peaks and plumbed troughs in those most mercurial pharmacologic cauldrons: children.

Keeping NFV Trough Above 0.8 mg/L Improves Response

The ATHENA study showed that TDM improved virologic response in adults taking NFV,5 but two studies of NFV levels in children yielded contradictory results. One linked AUC with virologic response,6 but a second failed to confirm that finding.7 David Burger (University Medical Center, Nijmegen, The Netherlands) came to the Pharmacology Workshop with tie-breaking results that linked higher troughs to virologic response at treatment week 48 in 32 previously naive children [abstract 10].

The children were taking 25 to 30 mg/kg of NFV three times daily or 45 to 55 mg/kg twice daily in the PENTA 5 trial.8 Burger and coworkers measured morning NFV troughs after an unobserved evening dose. They set 0.8 mg/L as the lower trough threshold.

Nelfinavir troughs averaged 2.1 mg/L (coefficient of variation [CV] 65.8 percent) in the 18 children taking twice-daily NFV and 1.7 mg/L (CV 90.0 percent) in the 14 taking NFV three times daily. The two dosing groups did not differ significantly in the percentage with levels below 0.8 mg/L. Seven children (22 percent) had troughs under that threshold, with levels ranging from 0.10 to 0.57 mg/L. Fewer children with sub-therapeutic troughs had a viral load below 50 copies/mL at weeks 24 and 48 (Table 3).

Table 3. NFV Troughs and Sub-50-Copy Responses in 32 Children

Table 3. NFV Troughs and Sub-50-Copy Responses in 32 Children

Children with sub-therapeutic NFV troughs were significantly younger and weighed significantly less than children with higher troughs. Two of five children with low troughs and virologic failure at week 48 had genotyping. One had the NFV-associated L90M mutation, while the other had only secondary PI mutations -- K20R, M36I, and M46L.

Burger surmised that his results may differ from those in the study that found no link between NFV levels and response7 for one or more reasons: a larger sample in the PENTA pharmacokinetic (PK) study, longer follow-up in PENTA, a different trough threshold (0.8 versus 1.0 mg/L), different virologic outcomes (percent below 50 copies/mL versus viral load decline), and exclusion of non-adherent children from the PENTA analysis. He proposed that 0.8 mg/L is an appropriate NFV trough for children.

Lipodystrophy and LPV/RTV Troughs in Kids

Children taking PIs share other similarities with adults, according to results of a 14-child study by Josep Mallolas (Hospital Clinic, University of Barcelona): LPV/RTV makes a good rescue regimen, but it pushes up lipids fast [abstract 11].

The study involved seven girls and seven boys taking 300 to 375 mg/m2 of LPV/RTV twice daily after experience with more than one other PI. Their ages ranged from two to 13 years and they began LPV/RTV with a median CD4 percent of 16 percent (range 2 percent to 32 percent) and a median viral load of 5.5 logs (range 4.3 to 5.8 logs). No one had a lipid disorder before starting LPV/RTV. One girl dropped out of the study because of gastrointestinal side effects.

By week 12, 10 of 13 children (76 percent) had a viral load below 200 copies/mL. But most paid a price for that fine response:

Troughs measured at least three weeks after treatment began proved significantly higher in children with a lipid disorder (8.75 µg/mL) than in those without lipid problems (3.35 µg/mL) (P = 0.0043). Adherence measured by questionnaire stood above 90 percent in all 13 children.

Noting that these results should be confirmed in a larger study, Mallolas suggested that TDM may be warranted to trim high LPV/RTV levels in children responding to these PIs. But this study is too small to pinpoint a trough that is safe and still effective for children with PI experience.

Once-Daily LPV/RTV for Children

Lower LPV/RTV doses than those used in Mallolas's study are possible in PI-naive children. And in those who control HIV well, even a once-daily dose may be feasible, reported Gwenda Verweel (Erasmus University Medical Center, Rotterdam) [abstract 60].

Verweel tested an LPV/RTV dose of 460/115 mg/m2 once daily in seven boys and seven girls who had kept their viral load below 50 copies/mL for at least six months. Their median age stood at 4.5 years (interquartile range [IQR] 3.3 to 9.5 years), median weight at 19.9 kg (IQR 14.6 to 33.5 kg), and body surface area at 0.77 m2 (IQR 0.62 to 1.11 m2). All were also taking AZT and 3TC.

The median absolute LPV/RTV dose reached 400 mg (IQR 282 to 533 mg), while the median dose per square meter reached 461 mg (IQR 448 to 483 mg). Lopinavir levels measured at steady state varied widely, but median values compared closely with those in adults taking 800/200 mg of LPV/RTV once daily,9 except for a somewhat lower Cmin:

Three of 14 children (21 percent) had troughs below the threshold of 1.0 mg/L and needed a dose boost; two of them were under four years old. Twelve of the 13 children maintained a viral load below 50 copies/mL through three months of treatment. The one child with a detectable load had 52 copies/mL.

Gender, Pregnancy, and PKs

Happily, mother-to-child transmission of HIV has become a rarity in the industrialized world. But a few million children are already infected elsewhere. As antiretroviral access continues to improve in those HIV epicenters, mapping out the PKs of antiretrovirals in children will assume even greater urgency.

The race is already on to fill the pharmacokinetic void left by surplus enrollment of men in most trials of current antiretrovirals. And for good reason, the University of California, San Francisco's Francesca Aweeka observed: 58 percent of infected Africans are female, as are 36 percent of infected Southeast Asians. Research already shows that SQV, NVP, LPV, and enfuvirtide (ENF) levels climb higher in women than in men, Aweeka said, while women attain lower quotients of EFV, AZT monophosphate, and AZT triphosphate. But weight, she added, can be a stronger correlate of antiretroviral exposure than gender. Pregnancy tends to taper exposure to SQV, IDV, RTV, NFV, and LPV.

A study published just before the Pharmacology Workshop detailed the effects of higher SQV levels among women than men also taking RTV or NFV in a rescue regimen.4 Saquinavir concentrations proved higher with RTV than with NFV. Women had higher SQV AUCs and higher Cmins than men did. And those higher levels made a virologic difference: 42 percent of women reached a viral load of 500 copies/mL or less by week 16, compared with 28 percent of men. Higher SQV AUC and Cmin raised the chance of getting the viral load under 500 copies/mL (P = 0.008).

Lower NFV and M8 Metabolite Exposure in Pregnancy

In his report on NFV and M8 metabolite levels during the third trimester and after delivery, Rolf van Heeswijk (Ottawa Health Research Institute) outlined the potential effects of pregnancy on antiretroviral absorption, distribution, and clearance [abstract 9]. Pregnancy may alter:

The CYP3A4 and CYP2C19 enzymes metabolize NFV, he noted, and CYP2C19 drives the formation of M8, NFV's active metabolite. Once formed, M8 is further metabolized by CYP3A4.

The study involved one Hispanic, one Caucasian, and nine black women taking 1,250 mg of NFV twice daily during pregnancy and after delivery. None were taking agents that may alter NFV's metabolism. Women had NFV and M8 levels measured before and after taking an observed dose of NFV with food at a median of 32 weeks gestation (range 31 to 36 weeks), then again a median of eight weeks after delivery (range six to 15 weeks).

Nelfinavir levels proved lower during pregnancy than after delivery, but significantly so only for the 12-hour concentration (C12h):

M8 levels were all significantly lower during the third trimester:

Induction of CYP3A4 or inhibition of CYP2C19 could explain the low M8 values during pregnancy, van Heeswijk suggested. Although none of the women in this study had a virologic breakthrough, wide interindividual variations in NFV and M8 levels could put some women at risk of virologic failure during pregnancy. Although M8 matches NFV in antiviral potency in vitro, van Heeswijk noted, the clinical implications of low M8 levels are unknown.

The Fault Is in Our Genes

Had Shakespeare been our chronological as well as our psychological contemporary, he might have written in Julius Caesar that, "The fault, dear Brutus, is not in our stars, but in our genes." And if the Bard came to HIV meetings, he might pen a yet more trenchant trope after noting that our stars lie in our genes when one cites mutations such as CYP2B6*5.

Higher NFV Levels Linked to a Solitary SNP

A SNP apparently snipped away at P-glycoprotein (P-gp) expression in people taking NFV, resulting in higher plasma and intracellular concentrations of the PI. Whether people whose lucky stars include this C-to-T SNP at position 3435 on exon 26 of the MDR1 gene thereby benefit in CD4 gains remains controversial.

Sara Colombo (University Hospital, Lausanne) snooped for SNPs in 153 people taking NFV and 203 taking EFV [abstract 50]. Suspect SNPs at positions 3435 and 2677 did not affect EFV levels. But a shift from a CC to a TT genotype at 3435 raised plasma and intracellular levels of NFV, as well as NFV accumulation measured as intracellular-to-plasma ratio:

The results make genetic sense, Colombo argued, because NFV is a P-gp substrate and EFV is not. Expression of the P-gp drug pump on pharmacologically critical tissues and T cells is lower in people with the 3435TT genotype, so more NFV can pile up in their T cells.

But does the measured difference in NFV concentrations explain the better CD4 response seen in people with the TT genotype?10 Not everyone is ready to buy that logic, including Charles Boucher (University of Utrecht, The Netherlands), who claimed he had "a hard time convincing myself" that small differences in NFV exposure would translate into big gains in CD4 cells.

Pharmacology News on New Antiretrovirals

Making Right Turns in the HIV Pharmacology Maze

There's nothing like a new drug to get a pharmacologist's creative juices flowing, especially when that drug leaves the cosseted confines of controlled trials and starts getting used by thousands of people who may claim exactly those traits that kept them out of trials. This year's Pharmacology Workshop featured dozens of reports on recently approved antiretrovirals including TDF, ENF, and atazanavir (ATV). Among the investigational agents spotlighted were the PIs tipranavir (TPV) and TMC114, the co-receptor antagonists AMD070 and UK-427,857, and the NRTI elvucitabine. These headlines emerged:

Tenofovir Disoproxil Fumarate

Tenofovir won the Pharmacology Workshop's sweepstakes as the most-studied drug, beginning with the news that it may not consistently lower levels of ATV, as previously reported, and ending with the first good look at its much anticipated coformulation with emtricitabine (FTC).

Are ATV Troughs Lower With TDF?

Earlier work found lower troughs of the once-daily PI ATV when given with TDF, whether boosted by RTV11 or not.12 With RTV and TDF, ATV's Cmin was 26 percent lower than normal,11 and with TDF but without RTV the trough was 39 percent lower than normal.12 But a study of people prescribed various doses of ATV at 31 German clinics did not confirm those findings. A smaller Italian study found substantially (but not significantly) lower ATV levels with TDF than without it.

Guido Kruse (HIV-Lab, Berlin) evaluated 178 plasma samples from people taking ATV with or without TDF in one of seven doses [abstract 49]:

Kruse used samples collected between 23 and 25 hours after ATV dosing to figure ATV troughs and samples collected between 11 and 13 hours after dosing to figure troughs of other PIs.

First, the analysis showed that German clinicians usually give ATV with an RTV boost and often with a boost plus a second PI, either SQV or LPV. Only 11 of the 178 samples studied (6 percent) came from people taking unboosted ATV. Those 11 people all took ATV with TDF, and their mean ATV trough measured 148 ng/mL (95 percent CI 75 to 291 ng/mL). That mean exceeded the trough of 120 ng/mL without TDF listed in ATV product information and more than doubled the 70-ng/mL trough found with TDF in healthy volunteers.13

Among 53 people taking 300/100 mg of ATV/RTV plus TDF, the ATV trough averaged 576 ng/mL (95 percent CI 469 to 713 ng/mL). In 26 people taking 300/100 mg of ATV/RTV without TDF, the average ATV trough stood at 567 ng/mL (95 percent CI 416 to 773 ng/mL). Both of those mean troughs lie above the trough of 491 ng/mL reported earlier for ATV/RTV with TDF.11 Kruse argued that the fairly large number of samples in his analysis of people taking boosted ATV with or without TDF inspire confidence in the reliability of these results. With either regimen, though, the troughs varied substantially from one person to the next.

Elena Seminari (San Raffaele Hospital, Milan) did find lower ATV troughs when people took the PI with TDF [abstract 22]. No one in this study used an RTV boost, but some took APV (600 mg twice daily or 1,200 mg once daily) as a second PI. These 32 heavily pretreated people received 400 mg of ATV as part of an expanded access program. No one took an NNRTI.

Measuring ATV levels after at least two weeks of treatment, Seminari found lower median troughs with TDF than without it, and the troughs with TDF were much lower than the 148 ng/mL average Kruse found in people taking unboosted ATV with TDF:

The trough differences with and without TDF did not reach statistical significance, perhaps because of the small number of people studied. Intrapatient ATV trough variability proved higher with TDF and APV (coefficient of variation [CV] 51 percent, range 11 percent to 141 percent) than with TDF but without APV (CV 30 percent, range 2 percent to 91 percent). Amprenavir lowered ATV's peak and AUC but had little effect on ATV's trough concentration.

In Kruse's study, ATV troughs averaged 818 ng/mL (95 percent CI 692 to 967 ng/mL) in 56 people taking the PI with 100 mg of RTV and 2,000 mg of SQV, and 824 ng/mL (95 percent CI 537 to 1,263 ng/mL) in seven people taking the same doses of ATV and SQV with 200 mg of RTV. Those troughs came close to the 767 ng/mL reported earlier with a once-daily 300/100/1,600-mg dose of the three PIs.14 Saquinavir's trough with 300/100 mg of ATV/RTV averaged 448 ng/mL (95 percent CI 338 to 593 ng/mL).

Mean LPV troughs stood at 3,155 ng/mL (95 percent CI 2,310 to 4,310 ng/mL) in 16 people taking 300/800/200 mg of ATV/LPV/RTV and 3,707 ng/mL (95 percent CI 2,482 to 5,537 ng/mL) in nine people taking the same doses of those PIs plus 2,000 mg of SQV. Those LPV levels fell short of the 5,500 ng/mL reported with 400/100 mg of LPV/RTV twice daily,15 but Kruse noted that no one has reported evidence of an interaction between LPV and ATV.

No Interactions Between SQV/RTV and TDF

Jintanat Ananworanich (HIV-NAT, Bangkok) found no SQV Cmin change after 14 people switched from ddI/d4T to 3TC/TDF [abstract 27]. Indeed, fewer people had suboptimal SQV Cmins after they started TDF.

Using the SQV hard-gel capsule (HGC), five men and nine women had begun once-daily SQV/RTV (1,600/100 mg) with ddI (250 or 400 mg once daily depending on weight) and d4T (30 or 40 mg twice daily depending on weight). The HIV-NAT team measured SQV troughs after people took that regimen for at least eight weeks. Everyone later traded ddI/d4T for once-daily 3TC/TDF (300/300 mg) and had repeat trough testing after at least eight weeks of the new regimen. Body weight did not differ significantly from the time of the first trough reading (median 55.5 kg, interquartile range [IQR] 47.8 to 60.8 kg) to the time of the second reading (median 55.1 kg, IQR 49.2 to 61.7 kg). The median time on SQV measured 14 months (IQR seven to 18 months) when they had their second trough measure.

Mean and median SQV troughs both rose non-significantly after the switch to TDF (Table 4).

Table 4. SQV Troughs With ddI/d4T Versus 3TC/TDF

Table 4. SQV Troughs With ddI/d4T Versus 3TC/TDF

TDF did not alter 24-hour concentration-time profiles of SQV HGC at a twice-daily RTV-boosted dose of 1,000/100 mg in a 14-day study involving 18 people, reported Marta Boffito (Chelsea and Westminster Hospital, London). Nor did SQV change two-day or 13-day levels of TDF [abstract 31].

Boffito studied 14 men and four women with tight viral control while taking an SQV/RTV regimen. They had PK sampling on study day 1 (before adding TDF), day 3 (one day after adding TDF), and day 14 (13 days after adding TDF). Tenofovir levels proved similar to those recorded in a single-dose TDF study on both day 3 and day 14. Boffito detected no change in TDF AUC, Cmax, trough, half-life, or time to maximum concentration from day 3 to day 14.

No Intracellular ABC-TDF Antagonism

Surely the biggest antiretroviral disappointment of this young century (aside from skimpy access in most poor countries) is the triple failure of triple regimens containing TDF, ABC, and 3TC taken once daily,16-18 along with suboptimal potency19,20 or out-and-out inanition21 of other nuke troikas. Many experts fingered a low genetic barrier to resistance via the mutations M184V and K65R as a prime suspect in the untimely demise of regimens combining TDF, ABC, and/or ddI. But other possibilities -- including intracellular antagonism between TDF and ABC -- had to be explored. After all, inside-the-cell incompatibility undid AZT/d4T. But two Pharmacology Workshop reports ruled out the TDF-ABC antagonism hypothesis.

Arnold Fridland (Gilead Sciences) measured phosphorylation of ABC to carbovir triphosphate and of TDF to tenofovir diphosphate in peripheral blood mononuclear cells (PBMCs) from healthy volunteers and in CEM cells [abstract 5]. At concentrations of 10 µM and 100 µM (much higher than plasma concentrations in humans), neither TDF nor ABC inhibited intracellular phosphorylation of the other drug. Using the same CEM cell model, Fridland showed that 3TC slices ddC triphosphate levels by 75 percent.

Trevor Hawkins (Southwest CARE Center, Santa Fe, New Mexico) measured tenofovir diphosphate and carbovir triphosphate directly in PBMCs from 15 people taking ABC/TDF with 3TC or d4T [abstract 2.4]. He collected samples while people were taking the ABC/TDF regimen, then for 28 days after eight people traded TDF for a new drug and seven switched from ABC to another antiretroviral. Withdrawing either reverse transcriptase inhibitor had no effect on intracellular sums of the other agent's metabolite over the 28 days of study.

The intracellular half-life of carbovir triphosphate ranged from 12 to 19 hours, while the half-life of tenofovir diphosphate exceeded 60 hours. Hawkins found less interpatient and intrapatient variability with tenofovir diphosphate than with carbovir triphosphate.

Do these findings clinch the case against resistance as the sole culprit in failures of TDF, ABC, and 3TC? Not completely, the University of Utrecht's resistance sage Charles Boucher argued in a review lecture. Among people genotyped after failure of that triple regimen, 64 percent had both the M184V and K65R mutations. Together, Boucher reasoned, those two resistance rousers are enough to cripple all three drugs.

But 36 percent of people in whom the regimen foundered had only the M184V (or I) mutation -- without K65R. Phenotypic studies show that this virus should remain susceptible to TDF. Indeed, position 184 mutants should be hypersusceptible to TDF. And TDF monotherapy can peel 1.5 logs off a high viral load. But the regimen flopped anyway. To Boucher, that must mean these people had too little tenofovir diphosphate in their cells to counter the M184V/I mutant.

The same rationale may explain some failures of other regimens -- AZT/3TC/ABC, AZT/3TC/NFV, AZT/3TC/NNRTI -- in which genotyped virus had mutations that confer resistance to some, but not all drugs in the regimen. In cases like these, Boucher argued, drugs to which the mutant virus remains susceptible may fail to reach inhibitory levels in cells or some other critical compartment.

High TDF Levels in a Person With Renal Insufficiency

Although rare, renal insufficiency with renal tubular dysfunction has troubled people taking TDF,22-24 and TDF combined with LPV/RTV may explain one case of Fanconi syndrome.25 A Pharmacology Workshop case report by Milly Hillebrand-Haverkort (Onze Lieve Vrouwe Gasthuis, Amsterdam) appears to be the first linking high plasma TDF to renal insufficiency, and again the patient was also taking LPV/RTV [abstract 72].

With a CD4 count of 60 cells/mm3, a 54-year-old man was taking TDF, d4T, ABC, EFV, and APV. A relentlessly rising viral load prompted his clinician to add four capsules of LPV/RTV twice daily. That prescription rose to five capsules twice daily when the man failed to achieve adequate LPV levels with the four-capsule dose. Three months later he traded d4T for 3TC, and two weeks after that he was admitted to the hospital with abdominal pain and dyspnea diagnosed as renal insufficiency (serum creatinine 374 µmol/L, hyperphosphatemia at 2.2 mmol/L, and metabolic acidosis). Serum creatinine rose to 447 µmol/L as the patient continued his antiretrovirals during the hospital evaluation. At that point clinicians stopped all antiretrovirals and creatinine fell to 129 µmol/L. When the same regimen (minus 3TC) resumed, creatinine climbed back to 172 µmol/L. When 3TC replaced TDF, creatinine fell to normal levels.

This man was also taking seven other potentially nephrotoxic drugs, but he had taken them for years at the same dose with no problem. Because LPV/RTV boosts TDF exposure, Hillebrand-Haverkort measured TDF in frozen plasma samples from this patient. Tenofovir levels measured six to seven hours after dosing rose inexorably with LPV/RTV:

Renal clearance of TDF fell from 115 to 55 mL/min after LPV/RTV got added to the mix.

Because this case report and others suggest that renal dysfunction may be a cumulative dose-dependent effect of TDF, especially when taken with LPV/RTV, Hillebrand-Haverkort believes continuous monitoring of renal function makes sense for people taking the drug. Tenofovir drug monitoring may help single out people with a high risk of kidney problems, she added. "A significant increase in TDF serum concentrations, even in patients with normal levels, should warn the physician," she proposed. "To prevent [TDF] accumulation, dose reduction should be considered in such cases, even if renal clearance still exceeds 50 mL/min."

Higher EFV Concentrations With TDF

As summarized earlier in this article, a study presented at this year's Pharmacology Workshop found significantly lower EFV levels in people coinfected with HBV or HCV. The same study, presented by Rita Côrte-Real (Hospital of Desterro, Lisbon), found higher EFV concentrations in people taking the nonnucleoside with TDF than in those taking it with nucleosides or PIs [abstract 57].

The analysis involved 45 people taking EFV for six months or more; 19 of them also had TDF prescriptions. Collecting two to seven samples from each person eight to 14 hours after EFV dosing, Côrte-Real measured the following geometric mean EFV levels (and 95 percent CIs):

The difference between the TDF group and the nucleoside-only group was statistically significant (P = 0.029).

TDF/FTC: One Pill Once a Day

Fixed-dose conflations of two or three antiretrovirals have become mainstays of generic makers' product lines. But so far competition between brand name houses has limited antiretroviral coformulations to two all-in-the-family combos -- Combivir and Trizivir -- while 3TC/ABC and TDF/FTC pick their way across the pharmacologic proving grounds. Brian Kearney (Gilead Sciences) came to the Pharmacology Workshop with a progress report on TDF/FTC, and the news was good [abstract 62]. The once-a-day pill, about as big as a standard hefty multivitamin, yielded concentrations of its constituents equivalent to those of the single-drug pills now on the market.

The single-dose crossover study measured levels of the individual drugs dosed together and, one week later, of the coformulated capsule. The 44 healthy volunteers -- 26 men and 18 women -- had an average age of 43.6 years (+10.8 SD) and weight of 76.5 kg (+10.1 SD). Five people dropped out of the study, none with grade 3 or 4 toxicity and only one with a problem attributed to study drugs -- a mild rash.

In blood samples collected 48 hours after dosing, Kearney found nearly identical TDF and FTC peaks, AUCs, and half-lives with the individual drugs and the fixed-dose combination.

HIV Sparks a Key Nucleoside Activator

Levels of the nucleoside activator thymidine kinase (TK) are significantly higher and vary more widely in people with HIV infection than in uninfected individuals. Guido Antonelli (University La Sapienza, Rome) reached those conclusions after gauging TK levels in PBMCs from 46 antiretroviral-naive people with HIV infection and from 10 healthy volunteers without HIV or other infections and taking no drugs [abstract 7].

Thymidine kinase activity proved significantly higher in the HIV group whether Antonelli used AZT or 3TC as a substrate for the enzyme. Thymidine kinase activity also ranged widely in people with HIV infection and much less so in the uninfected controls. Deoxycytidine kinase activity varied more widely in people with HIV than in healthy controls, but mean activity was not significantly higher in the HIV group (Table 5).

Table 5. Thymidine or Deoxycytidine Kinase Activity and Variability in People With or Without HIV

Table 5. Thymidine or Deoxycytidine Kinase Activity and Variability in People With or Without HIV

Antonelli urged colleagues to probe for viral or cellular factors that may contribute to these differences in people with HIV and to explore the effects of such factors on response to antiretroviral therapy.


Despite its steep price and multistep, twice-daily injection schedule, ENF has gained adherents among those who can add it to a regimen including one or more active drugs. Now that this pioneer fusion inhibitor has reached a wider clinic population, however, two studies suggest that ENF levels are not measuring up to those reported in clinical trials. And one of these studies turned up evidence that nodules formed at injection sites could contribute to low blood levels. Whether such results support TDM for ENF is another question.

Lower ENF Concentrations Seen in the Clinic

Stefano Bonora (University of Torino, Italy) studied 22 men and four women who started ENF with a median CD4 count of 77 cells/µL and a median viral load of 4.8 logs while taking a median of three other antiretrovirals (range two to four) [abstract 46]. Comparing ENF concentrations on treatment day 2 and treatment day 14, Bonora found a significantly lower mean trough and a marginally lower mean AUC on the second reading:

The day 2 trough comes close to the 3,300 ng/mL listed in product literature, while the day 14 trough reflects a significant falloff. Bonora suggested these findings mean several days of ENF therapy must pass before levels reach a steady state.

Ten people (38 percent) had a minimum effective concentration below a target of 1,000 ng/mL, and 13 percent of all troughs measured fell below that threshold. The coefficient of variation for interindividual variability in ENF troughs was 71 percent, while intraindividual variability measured 38 percent.

About 65 percent of people taking ENF get injection site nodules or cysts, and earlier immunoperoxidase staining studies showed that the drug can pool in these lesions.26 Wondering whether nodules explained ebbing ENF levels in these people, Bonora compared concentrations in people with and without nodules. Median and mean troughs proved lower in 65 samples from people with nodules than in 44 samples from those without them, but these differences fell short of statistical significance (P = 0.18 for medians and P = 0.15 for means).

Analysis of ENF quotients in 19 people on day 14 found no significant difference in troughs or peaks in seven with nodules versus 12 without them. But the mean AUC stood significantly lower in people with nodules (39,450 + 14,433 versus 50,041 + 14,052 ng•h/mL, P = 0.04).

Hartmut Stocker (Vivantes Auguste-Viktoria Klinikum, Berlin) also found a lower mean trough in 39 people taking ENF than the trough listed in product literature (2,030 versus 3,300 ng/mL) [abstract 45]. Comparing 15 people with the highest troughs (above 2,410 ng/mL) and 15 with the lowest (below 1,750 ng/mL), he found a substantially lower median CD4 count (50 versus 182 cells/µL) and higher viral load (5.4 versus 4.5 logs) in those with the low troughs. The coefficient of variation for interindividual variability in troughs proved much lower in Stocker's study (8.5 percent) than in Bonora's (71 percent) but still comparable with interindividual variability in troughs of LPV (8.1 percent) and NFV (8.8 percent). Stocker suggested several possible reasons for lower ENF readings in the clinic than in trials:

Because of interindividual variability in troughs and scant data on the effect of drug interactions or comorbidities, Stocker advocated TDM for ENF. Bonora also endorsed TDM for the fusion inhibitor, citing interpatient variability and suboptimal troughs seen in some people. Others, such as John Gerber (University of Colorado, Denver), evinced less enthusiasm for ENF TDM at this point. Gerber called for more study of how ENF samples should be collected and stored and for ENF assay proficiency testing in TDM labs.


The most surprising Pharmacology Workshop news on ATV came in the 31-clinic study that found good troughs of the PI when given with TDF -- a finding at odds with earlier reports (see "Tenofovir Disoproxil Fumarate" section above). Other Pharmacology Workshop studies suggested a role for TDM when prescribing ATV and described one ATV combination that did not work well.

Lower Once-Daily SQV Levels With ATV

A small study discussed above found that APV at a dose of 600 mg twice daily or 1,200 mg once daily lowered ATV's peak concentration (median 2,990 ng/mL with APV and 5,170 ng/mL without APV) and AUC (median 23,980 ng•h/mL with APV and 33,400 ng•h/mL without APV) but had little effect on ATV troughs or clearance (see "Tenofovir Disoproxil Fumarate" section above). A 400- or 600-mg dose of ATV plus 1,200 mg of SQV HGC proved safe and comparable to RTV/SQV (400/400 mg twice daily) as a rescue regimen in one published trial.27 But SQV troughs proved low with this double-PI regimen in two other studies.28,29 At this year's Pharmacology Workshop, Francesca Canta (University of Torino, Italy) confirmed low SQV exposure with ATV and a trend to low ATV troughs as well [abstract 33].

The study involved 21 people with an average age of 49 years enrolled in an ATV expanded access program. Seven took 400 mg of ATV plus 1,200 mg of SQV HGC once daily plus one nucleoside, while 14 took ATV plus two nucleosides without SQV. No one in either group took a nonnucleoside or other drugs that may muddle metabolism of the two PIs. The 14 people in the ATV-only group included five women, while all seven people in the ATV/SQV group were men. The median baseline viral load stood at 1.98 logs in the ATV-only group versus 4.49 logs in the dual-PI group. Neither of these differences reached statistical significance. About half in each group had HCV coinfection.

Steady-state drug level readings showed a non-significantly lower mean ATV trough in the ATV/SQV group (86 ng/mL, range 5 to 316 ng/mL) than in the ATV-only group (205 ng/mL, range 5 to 920 ng/mL, P = 0.25). Four of seven (57 percent) in the double-PI group and two of 14 (14 percent) in the single-PI group had an ATV trough below 14 ng/mL, the 90 percent inhibitory concentration for wild-type virus, but this difference also lacked statistical significance (P = 0.12). The SQV trough in the dual-PI group averaged 46 ng/mL (range 5 to 310 ng/mL), and six of those seven people (86 percent) had an SQV trough below a minimum effective concentration of 100 ng/mL.

Canta advised clinician colleagues to avoid once-daily ATV/SQV (400/1,200 mg), though the interaction between these PIs probably needs further study. In a 20-person trial, adding 300 mg of ATV to 1,600/100 mg of once-daily SQV/RTV significantly boosted SQV's AUC by 60 percent, its peak by 42 percent, and its trough by 112 percent (P<0.05).14

High ATV Exposure Predicts High Bilirubin

Two studies tied higher ATV concentrations to a greater risk of hyperbilirubinemia, and both study teams suggested that TDM may help predict this risk.

Daniel González de Requena (University of Torino, Italy) tracked ATV and bilirubin levels in 31 men and 14 women taking 400 mg of the PI without RTV in an expanded access program [abstract 58]. Their median age was 42 years, and 19 (42 percent) had coinfection with HCV or HBV.

After four weeks of treatment, 36 (80 percent) had a viral load below 50 copies/mL or more than a 1-log drop in viral load. Pretreatment ATV-linked mutations did not affect virologic response, but ATV concentrations did. The four-week drop in log viral load correlated inversely with Cmax (r = -0.547, P = 0.023) and 24-hour AUC (r = -0.5, P = 0.049) but not with Cmin. Area under the curve also correlated with percent gain in CD4 cells. The percentage of people with total and unconjugated bilirubin levels above 1 mg/dL rose substantially after four weeks of ATV, but no one had a conjugated bilirubin level above that threshold at either point (Table 6).

Table 6. Bilirubin Changes After Four Weeks of Unboosted ATV (n = 45)

Table 6. Bilirubin Changes After Four Weeks of Unboosted ATV (<i>n</i> = 45)

Atazanavir Cmin correlated positively with the percent of total bilirubin change from baseline (r = 0.47, P = 0.013). And Cmin (r = 0.44, P = 0.02), Cmax (r = 0.43, P = 0.04), and AUC (r = 0.42, P = 0.05) correlated with percent change in unconjugated bilirubin. Linear regression analysis including baseline total and unconjugated bilirubin, HCV coinfection, and ATV Cmin, Cmax, and AUC singled out AUC as the only independent predictor of percent gain in total and unconjugated bilirubin.

González proposed that TDM could help improve the early response to ATV while enhancing the drug's safety. Whether the same proves true with RTV-boosted ATV -- the way most people will take this PI -- awaits further study.

Sonia Rodriguez-Novoa (Carlos III Hospital, Madrid) also mapped bilirubin changes in people taking 400 mg of unboosted ATV [abstract 55]. Studying 48 men and 20 women, 44 percent of whom had HCV coinfection, she found higher 12-hour concentrations of the PI after 12 weeks of therapy in people with hyperbilirubinemia.

By week 12, the median AST rose 15.2 U/L and the median ALT climbed 2.4 U/L. Neither change was significant and neither correlated with HCV status. Median total cholesterol fell 9.8 mg/dL by week 12 (P = 0.007), while median triglycerides dropped 20.6 mg/dL (P<0.001). Atazanavir concentrations did not correlate with lipid changes.

Bilirubin rose in 70 percent of study participants, and 11 percent had a grade 4 elevation (above 3.9 mg/dL). Median 12-hour ATV levels proved significantly higher in people with hyperbilirubinemia (0.82 µg/mL) than in those without it (0.22 µg/mL, P = 0.006), though confidence intervals in this comparison overlapped substantially.

Tipranavir and TMC114

Two protease inhibitors within sight of the clinic, TPV and TMC114, both stifle PI-resistant virus. One Pharmacology Workshop study rated TPV/RTV in deep salvage, while another gauged its interactions with atorvastatin and antacids. A third study plotted the PKs of TMC114/RTV.

Rebound After Good Early Response to TPV/RTV Plus another PI

Tipranavir/RTV at a dose of 500/200 mg twice daily has reached phase 3 trials. That dose may fall short for people rich in PI experience, even when they add a third PI, suggest preliminary results presented by Kevin Curry (Boehringer Ingelheim) [abstract 34].

This trial enrolled 296 people, mostly white men, whose charts included triple-class failure, failure of two or more PI regimens, three or more protease mutations at positions 33, 82, 84, or 90, and a viral load above 1,000 copies/mL. In fact, the median viral load approached 100,000 copies/mL, while the median CD4 count stood at 140 cells/µL (range 0 to 867 cells/µL). Experience with PIs does not get much deeper than this.

Curry and colleagues randomized these people to one of four regimens:

After two weeks, everyone not taking TPV added TPV/RTV 500/100 mg twice daily. Investigators picked the other drugs to go with the PIs. Similar proportions in each arm (from 11 percent to 18 percent) added ENF. After two weeks, 55 percent to 60 percent in each group had some side effect, usually diarrhea (5.3 percent with SQV, 10.1 percent with LPV, 13.6 percent with TPV, and 15.8 percent with APV).

At the two-week point people taking TPV/RTV enjoyed the biggest viral load drop -- a median of 1.2 logs compared with 0.2 with APV/RTV, 0.3 with SQV/RTV, and 0.4 with LPV/RTV. When people taking APV, SQV, or LPV added 500 mg of TPV and an extra 100 mg of RTV, the median viral load dipped to 1.2 logs below baseline in the next two weeks. But viral loads began to rebound after that. By week 8, the median drop from baseline stood at about 0.5 log with TPV/RTV, about 0.8 log with SQV/RTV/TPV and LPV/RTV/TPV, and about 0.9 log with APV/RTV/TPV.

Dwindling PI concentrations may have helped propel these rebounds in the triple-PI groups. Adding TPV at week 2 cut LPV's trough by about 45 percent, APV's by about 50 percent, and SQV's by about 80 percent. Peak concentrations and AUCs of the third PIs also fell when they teamed up with TPV. The CYP3A4-inducing effect of TPV may have undercut the inhibiting effect of RTV, although in a study of healthy volunteers a single dose of TPV/RTV (500/200 mg) nearly completely inhibited hepatic CYP3A4 activity (see next section). Further analysis of this salvage trial should pick out factors that favored rebounds or a more durable response.

Interactions Between TPV/RTV and Atorvastatin or Antacids

TPV/RTV (500/200 mg) jacked up exposure of atorvastatin nearly 9-fold, while the antacid Maalox lowered TPV concentrations by up to 29 percent in a study of 23 healthy volunteers by Ottawa Hospital's Rolf van Heeswijk [abstract 35]. He gauged the effect of the three drugs in 12 women and 11 men with a median age of 32 years (range 18 to 55 years) and a median weight of 72 kg (range 55 to 99 kg). The intricate 22-day study design measured drug concentrations at the following points:

The geometric mean single-dose AUC of atorvastatin rose from 89.3 h • ng/mL on day 1 (40 mg without TPV/RTV) to 209 h • ng/mL on day 20 (10 mg with steady-state TPV/RTV). Atorvastatin's dose-normalized day-20 AUC (figured by multiplying the day 20 value by 40/10 mg) was 836 h • ng/mL (range 478 to 1,850 h • ng/mL). The geometric mean Cmax climbed from 17.6 ng/mL on day 1 to a dose-normalized 151 ng/mL (range 80.0 to 345 ng/mL) on day 20 with the steady-state PIs. The statin's 12-hour level jumped from 2.25 ng/mL on day 1 to a dose-normalized 11.7 ng/mL on day 20 with the PIs. Atorvastatin had no apparent effect on TPV levels. Van Heeswijk rated the atorvastatin surges "clinically relevant" and recommended close monitoring of people who take the statin with TPV/RTV.

Geometric mean ratios of TPV/RTV plus antacid to TPV/RTV alone showed substantial drops in TPV exposure:

Because the antacid lowers TPV levels, clinicians prescribing these drugs with TPV/RTV will have to consider the timing of the antacid dose.

These volunteers must have been glad when the study ended because all but one of them suffered some kind of gastrointestinal distress. During treatment with TPV/RTV alone, 17 (74 percent) had diarrhea, 11 (48 percent) had nausea, and nine (39 percent) had abdominal pain. Sixteen people (70 percent) reported some central nervous system complaint, including headache and loss of taste. All of these problems were mild, and no one quit the study because of them. There was only one clinically relevant lab abnormality, a grade 3 ALT gain.

This study also showed that single-dose TPV/RTV reduced hepatic CYP3A4 activity by 96 percent. Tipranavir induces CYP3A4 but clearly not nearly as much as 200 mg of RTV inhibits that drug-metabolizing enzyme.

Higher TMC114 Tablet Exposure With Food

In a study of 50 people with plentiful PI experience and taking a failing PI regimen, TMC114/RTV doses of 300/100 or 600/100 twice daily or 900/100 mg once daily lowered median viral loads from 1.24 to 1.50 logs over eight weeks.30 In that period no one changed the nucleosides in their baseline regimen. At the Pharmacology Workshop, Richard Hoetelmans (Tibotec) offered food interaction findings on a new 400-mg solid formulation of TMC114 and an oral solution, both given with RTV [abstract 39].

The four-way crossover study design called for 16 healthy volunteers to take the tablet or oral solution with or without a standard breakfast. They also took 100 mg of RTV twice daily from two days before to one day after dosing of TMC114. One person withdrew consent during the washout after the first dosing period. Others reported mild or moderate nausea, vomiting, dyspepsia, and headache.

Whereas food lowered exposure to TMC114 taken as an oral solution, food bumped up exposure to the tablet (Table 7).

Table 7. Effect of Food on TMC114 Exposure

Table 7. Effect of Food on TMC114 Exposure

Hoetelmans concluded that the TMC114 tablet formulation should be taken with food.

Two Entry Inhibitors and a Nucleoside

Dose Response Effect of AMD070, a CXCR4 Antagonist

A phase 1 study of AMD070, an orally bioavailable CXCR4 antagonist, showed a dose response effect, with most 12-hour concentrations of a 400-mg dose above the protein binding-adjusted 90 percent effective concentration (EC 90) of 125 nM. AMD070 has a 50 percent inhibitory concentration similar to that of its forerunner, AMD3100, which could not be taken orally (6.6 and 4.3 nM respectively against HIV-1 NL4.3 in MT-4 cells).

Nimalie Stone (Johns Hopkins University, Baltimore) charted AMD070 levels in healthy men taking single doses of 50, 100, 200, or 400 mg, twice-daily doses of 100 or 200 mg, and a single 400-mg dose with food [abstract 36]. Food had no impact on the drug's bioavailability.

The Cmax and AUC proved dose-proportional across doses, though interindividual variation was wide. White blood cell gains, a possible marker of CXCR4 inhibition, ranged from 1.3- to 1.6-fold with the 50-mg dose and from 1.5- to 2.9-fold with 400 mg. AMD070 concentrations 12 hours after a single 400-mg dose ranged from 79 to 155 nM, though nearly everyone taking that dose had a concentration above the protein binding-adjusted EC 90. Multiple dosing did not result in drug accumulation. Some volunteers suffered mild and reversible headaches.

CYP3A4 Inducers Lower Exposure of UK-427,857, a CCR5 Antagonist

The CYP3A4 inducers rifampin and EFV lower levels of the CCR5 antagonist UK-427,857, reported Gary Muirhead (Pfizer Global Research and Development) [abstract 37]. Doubling the dose of UK-427,857 compensates for this inducing effect.

The placebo-controlled, parallel-group study involved 36 healthy men who took 100 mg of the CCR5 antagonist twice daily on study days 1 through 21. On days 8 through 21 they also took 600 mg of rifampin once daily, 600 mg of EFV once daily, or placebo. By comparing UK-427,857 troughs on days 7 and 17, the Pfizer team gauged the effect of the CYP3A4 inducers and adjusted the dose of the antagonist if necessary on days 22 through 28.

Muirhead calculated a 6.6-fold increase in CYP3A4 induction with rifampin and a 2.4-fold increase with EFV. Both drugs substantially lowered UK-427,857 exposure compared with placebo. By day 28 the dose adjustment had lifted exposure of the antagonist back to levels comparable with placebo (Table 8).

Table 8. Dose Adjustment Compensates for Low UK-427,857 Levels With CYP3A4 Inducers

Table 8. Dose Adjustment Compensates for Low UK-427,857 Levels With CYP3A4 Inducers

Elvucitabine, an NRTI With a Difference

The sundry once-daily antiretroviral regimens now feasible make life easier for people starting down the road of lifelong therapy. Dare one anticipate once-weekly therapy? Pharmacology Workshop attendees got a glimmer of that hope in a review of the nucleoside cytosine analog elvucitabine by John Pottage (Achillion Pharmaceuticals) [abstract 38].

in vitro work indicated that elvucitabine has 5- to 10-fold greater antiviral activity than 3TC against wild-type HIV and HBV at respective 50 percent inhibitory concentrations of 4.8 and 1.0 nM. At doses of 50 mg or more daily, the nucleoside also slowed replication of 3TC-resistant M184V mutant virus. But those doses caused bone marrow toxicity.

Using data from studies of 29 healthy volunteers and 64 people with HBV infection, Pottage arrived at a "conservative" therapeutic window defined as a steady-state 24-hour AUC of 300 µg•h/mL and a Cmin of at least 2 µg/L. To prevent bone marrow toxicity he calculated that the Cmin must stay below 23 µg/L, which would keep peak concentrations under 40 µg/L in bone marrow.

With an estimated elimination half-life of 175 hours, elvucitabine could be a candidate for once-weekly therapy, and software-driven simulation settled on three once-weekly doses that yielded a Cmin, Cmax, and AUC within the therapeutic window -- 40, 50, and 75 mg. Other doses that fit the window were 5 mg twice daily; 5, 7.5, and 10 mg once daily; and 10, 15, and 20 mg every 48 hours.

Drug-Drug Interactions -- Good, Bad, or Neither

Making Right Turns in the HIV Pharmacology Maze

Antiretroviral drug-drug interactions can be good, or bad, or neither good nor bad. But they always take time to figure out and may require precise prescription planning, drug level monitoring, or dose adjustments. And, after all that, the person prescribed these drugs had better take them all and on time.

If only monotherapy had worked, things would be much simpler. And maybe it will yet, at least if you count an RTV-boosted PI as monotherapy. When the news arrived that some had started studying boosted PI monotherapy, Andrew Hill (Roche Laboratories) confessed he found the concept hard to swallow, since so many had choked on it the first time around. But in the Pharmacology Workshop's first review lecture -- "MicroHAART" -- Hill noted that boosted monotherapy has found its boosters.

Swiss clinicians studied IDV/RTV maintenance therapy after induction with IDV/RTV plus two nucleosides.31 Adjusting the IDV/RTV dose to keep IDV's trough between 500 and 2,000 nM/L, these workers saw no consecutive rebounds above 400 copies/mL through 48 weeks. In the United States, clinicians looked at LPV/RTV as first-line therapy for 30 people at 400/100 mg twice daily for those weighing less than 70 kg and 533/133 mg twice daily for those weighing more.32 All but one reached a viral load below 400 copies/mL, but 50-copy results were not reported. Eight people dropped out of the study or added another drug by week 24. Hill added that early results of a solo SQV/RTV trial may appear this month at the XV International AIDS Conference in Bangkok.

But these are "early days" for recidivist monotherapists, Hill cautioned. Indeed, with once-daily double NRTIs like 3TC/ABC and TDF/FTC on the horizon, boosted-PI monotherapy will have to prove highly safe and effective before its proponents can argue against adding a couple of nukes. In a stimulating talk reviewed at the end of this article, Hill himself called on colleagues to study lower doses of current combinations, after marshalling evidence that microHAART may sometimes prove feasible.

For at least the immediate future, though, clinicians and their pharmacologist allies will have to grapple with the interactions possible when they combine "drug X" with "drug Y," with or without "drug Z." And continued study of those interactions shows that continuing is a good idea, because early impressions can change -- or at least come into question. A prime example of such shifts came in Pharmacology Workshop studies (reviewed in the second section of this article) challenging the conclusion that TDF consistently undercuts ATV. And a dozen studies discussed below affirmed that mixing antiretrovirals can often yield a productive match for reasons not always apparent a priori.

It's hard enough for HIV pharmacology experts to keep up with all this, so what's a harried clinician to do? Seek out a reliable pharmacologist, for starters, and a German group has just made that easier.

German Drug-Drug Interaction Hotline

Leonie Meemken (ifi-Institut, Hamburg) reported early experience with a ripe idea -- a telephone/Internet hotline to answer questions about antiretroviral interactions and other pharmacologic issues [abstract 24]. Both physicians and pharmacists have tapped the hotline, which is managed by HIV pharmacologists. When necessary, questions get referred to clinical experts in other disciplines. Answers are returned within 24 hours.

So far about two thirds of the questions involve co-medications given with antiretrovirals, about one quarter involve only antiretrovirals, and the rest involve herbal agents. (In a Pharmacology Workshop study of 437 HIV-infected people in Barcelona, 25 percent reported using some complementary or alternative agent in 2003.33) Co-medications have resulted in three quarters of reported toxic antiretroviral levels. Meemken gave hotline users high marks for their knowledge of antiretroviral pharmacology, but they often need help understanding the impact of other agents that they prescribe less often.

Hotline workers catalog questions and answers in a database that will be routinely analyzed. Groups in other countries who study and adapt this model to local needs will render a valuable service to HIV clinicians and the people they treat.

Lopinavir and Ritonavir

Separate studies surveyed the effect of LPV/RTV on CYP2C isoenzymes and its interactions with APV and EFV. Giving RTV with minidose cyclosporine A looks feasible in people who have liver transplants.

LPV/RTV Boosts Activity of CYP2C19 and CYP2C9

Drugs metabolized via CYP2C19, and possibly those metabolized via CYP2C9, may need their doses dialed upwards when given with LPV/RTV, which increases activity of both isoenzymes. Rosa Yeh (University of North Carolina, Chapel Hill) [abstract 13] reached that conclusion in a study inspired by earlier work showing drops in phenytoin, LPV, and RTV exposure when the three drugs are mixed.34 Because CYP2C9 is the prime route of phenytoin metabolism and CYP2C19 a minor route, she rated activity of the two isoenzymes in 10 healthy volunteers given LPV/RTV.

Volunteers all weighed at least 50 kg and were within 20 percent of their ideal body weight. Their age averaged 26.8 years and ranged from 21 to 40 years. The study group consisted of five African Americans and five Caucasians, seven men and three women. On study days 1 to 3 and 14 to 17, they took 10 mg of warfarin, a substrate for CYP2C9, and 40 mg of omeprazole, a substrate for CYP2C19. They took the standard twice-daily dose of LPV/RTV on days 4 to 17. The PIs significantly lowered exposure of both marker drugs, a result indicating increased activity of the two isoenzymes (Table 9).

Table 9. Effect of LPV/RTV on CYP2C19 and CYP2C9

Table 9. Effect of LPV/RTV on CYP2C19 and CYP2C9

Lopinavir and RTV concentrations stayed within the expected range. The study did not clarify how much LPV or RTV individually contributes to induction of CYP2C19 and CYP2C9.

Drugs metabolized by CYP2C19 include diazepam, phenytoin, omeprazole, propranolol, citalopram, imipramine, and certain barbiturates. Those metabolized by CYP2C9 include most nonsteroidal anti-inflammatories, S-warfarin, phenytoin, tolbutamide, and ibuprofen.

LPV Moderates RTV Boost of APV

A retrospective study by Marie-Claude Gagnieu (Edouard Herriot Hospital, France) found a significant drop in elimination of APV when given with RTV or LPV/RTV [abstract 28]. But the ebb in elimination proved smaller with LPV/RTV than with RTV alone. The study involved three groups who had taken APV with or without LPV and RTV:

The elimination rate constant measured 0.31 + 0.14/h with APV alone, 0.17 + 0.08/h with LPV/RTV, and 0.116 + 0.137/h with RTV. In line with these findings, APV Cmins were significantly lower with APV alone (0.59 + 0.77 mg/L) than with LPV/RTV (0.99 + 0.54 mg/L, P = 0.01) or just RTV (1.65 + 0.74 mg/L, P = 0.001).

Interindividual variability in APV elimination proved less with LPV/RTV than only with RTV, but concentrations of LPV and RTV did not correlate with elimination of the third PI.

Interactions Between RTV and Cyclosporine A After Liver Transplant

Martin Vogel (University of Bonn) had to cut the standard dose of cyclosporine A drastically in three HIV-infected liver transplant patients restarting LPV/RTV [abstract 19]. But the three have not had an acute or chronic rejection 311, 504, and 720 days after the transplant.

Vogel recorded 12-hour PK profiles of cyclosporine A before and after these people restarted treatment with two nucleosides and LPV/RTV. After antiretroviral therapy resumed, elimination half-lives of cyclosporine A rose as high as 38 hours. To return to cyclosporine AUCs equivalent to those seen before PI therapy started again, daily doses of the anti-rejection drug had to be sliced to a range of 5 percent to 20 percent of the original dose. Because of flat cyclosporine A absorption/elimination profiles, Vogel could reliably monitor the drug's level by measuring troughs.

Journal of Negative Results

One can leave a Pharmacology Workshop with the strong impression that almost everything but earlobe length tweaks drug levels up or down. But that's just because results are much more interesting when they show something unexpected, or something dangerous, or something salutary -- so those are the data that get presented and published. But just as often drugs don't interact, genes don't disrupt drug distribution, and gender doesn't mean a thing (pharmacokinetically speaking).

Authors of studies with such equable outcomes have a harder time getting published, unless they submit to The Journal of Negative Results (JNR), an august chronicle imagined by David Burger of the University Medical Center, Nijmegen. He nominated his own work on CYP2B6 genotype and EFV [abstract 54] for JNR, and two other Pharmacology Workshop studies on PIs and EFV easily qualify.

(Some) CYP2B6 Shifts Do Not Explain EFV Variability

Because EFV levels vary greatly from person to person, proving toxic in some, researchers are trying to track down triggers of that variability. Earlier this year, AIDS Clinical Trials Group (ACTG) investigators found that a TT genotype at position 516 of the gene that codes for CYP2B6 -- the isoenzyme engine of EFV metabolism -- turned up in 20 percent of blacks and 3 percent of whites.3 The TT genotype also correlated with slower clearance and high levels of EFV, and with more side effects.

Burger found an easier-to-identify predictor of high EFV exposure and more side effects -- female gender.35 And that correlation held true independently of body weight. Pursuing this research goal, Burger turned from gender to genetics, hypothesizing that a mutation labeled CYP2B6*5 -- already linked to poor metabolism of EFV -- could drive up drug levels [abstract 54].

Burger analyzed samples from people taking EFV who used his lab's TDM service in 2002 and 2003. Of 228 individuals with TDM samples, 189 (82.9 percent) had a wild-type CC genotype, 33 (14.5 percent) had the heterozygote CT variant, and 6 (2.6 percent) had the homozygote TT variant. Defining a toxic EFV level as greater than 4 mg/L, Burger found no gene-based difference in proportions with a toxic concentration (P = 0.591):

The analysis did disclose a significantly greater mutation rate (CT plus TT) in Caucasians (21.6 percent) than in blacks (8.2 percent) or Asians (0 percent) (P = 0.036) -- a reversal of the mutation prevalence found in the ACTG study of another position on the gene coding for CYP2B6. 3

Current PIs Do Not Affect EFV Concentrations

A retrospective study of EFV levels by Jean-Marie Poirier (Saint-Antoine University Hospital, Paris) found that taking a PI with the nonnucleoside did not significantly affect EFV exposure [abstract 25]. The analysis involved 33 people taking EFV alone and 145 taking the standard dose of EFV plus NFV or an RTV-boosted PI. Everyone was also taking two nucleosides. Comparing EFV levels measured at least eight hours after dosing, Poirier discerned no differences between the PI and non-PI groups in percentages with high, low, or therapeutic range EFV quotients (Table 10).

Table 10. EFV Concentrations With or Without a PI

Table 10. EFV Concentrations With or Without a PI

Efavirenz levels proved marginally but not significantly higher with certain PIs than with EFV alone (mean + standard deviation [SD]):

Steady LPV and IDV Troughs With Combination

Lopinavir and IDV troughs in five people taking the PIs with RTV did not differ significantly from troughs of two reference populations taking LPV/RTV (400/100 mg twice daily) or IDV/RTV (400/100 mg twice daily). Saint-Antoine University Hospital's Jean-Marie Poirier did find significantly lower RTV troughs with LPV/RTV than with IDV/RTV or LPV/RTV/IDV [abstract 32].

The five retrospectively studied people were all taking 400 mg of LPV and IDV twice daily plus 100 mg of RTV twice daily and two nucleosides. Poirier measured troughs from samples collected 10 to 14 hours after dosing. The reference populations consisted of 106 people taking IDV/RTV and 185 taking LPV/RTV.

Mean (+ SD) LPV and IDV troughs were equivalent from group to group:

Mean (+ SD) RTV troughs measured 217 + 171 ng/mL with LPV/RTV (P<0.02 versus both other groups), 398 + 159 ng/mL with LPV/RTV/IDV, and 576 + 499 ng/mL with IDV/RTV.

Fruitful Teamings, Novel Tactics, Triple PIs

Several groups reported good results -- often with novel tactics -- in studies of RTV-boosted PIs. And all but one of these studies involved the prototype PI, SQV. Two of the SQV trials paired it with other PIs plus RTV. The outlier study in this cluster sized up fosamprenavir (FPV)/RTV for people with well-aged PI histories.

Once-Daily FPV/RTV After PI Failure

People with six protease mutations had less than a 1-log drop in viral load after one month of once-daily FPV/RTV (1,400/200 mg), reported Anne-Marie Taburet (Kremlin Bicêtre Hospital, Paris) [abstract 15]. People with two baseline protease mutations all shaved more than 2 logs off their viral load at one month. But the one-month response did not correlate with number of baseline mutations in people who had three, four, or five.

The ongoing study involves 26 men and 12 women with two or more APV or RTV mutations after failure of one or more PI regimens. The average number of PIs taken stood at three (range one to seven), while the median starting viral load measured 4.44 logs (range 3.03 to 5.67 logs) and the median CD4 count 228 cells/mm3 (range 18 to 503 cells/mm3). Twelve study participants also took a nonnucleoside, and five of them added another 100 mg of RTV. By treatment month 6, two people had quit because of gastrointestinal complaints, and two had a grade 3 or 4 triglyceride jump.

In a one-month analysis of 33 people, all three with only two baseline protease mutations had more than a 2-log drop in viral load, and all three with six mutations had less than a 1-log decline. Among 26 people with three, four, or five baseline protease mutations, 16 had less than a 1-log viral load drop at month 1 and 10 lost 2 logs or more, a nonsignificant difference. Viral load declines did not correlate with an inhibitory quotient measured as unbound APV/90 percent inhibitory concentration.

Among people with four or fewer baseline protease mutations and an amprenavir trough above 1,000 ng/mL, the median viral load drop measured 1.1 log at month 3. But that median had crept up to a 0.7-log decline by month 6. Median troughs stood well above 1,000 ng/mL at months 1, 3, and 6, but troughs ranged widely:

This article was provided by International Association of Physicians in AIDS Care. It is a part of the publication IAPAC Monthly. You can find this article online by typing this address into your Web browser:

General Disclaimer: is designed for educational purposes only and is not engaged in rendering medical advice or professional services. The information provided through should not be used for diagnosing or treating a health problem or a disease. It is not a substitute for professional care. If you have or suspect you may have a health problem, consult your health care provider.