Most North American schoolchildren know that Samuel de Champlain founded Québec, the oldest city on the continent. But it wasn't easy.
Champlain had already outlasted three gelid winters in Acadia, though most of his troop succumbed to the cold. By 1608 he'd had enough and moved south -- to Québec. The doughty explorer counted only eight surviving followers when the ice-clad St. Lawrence River yielded to spring.
Persistence paid off. As Montreal virologist Mark Wainberg (McGill University) noted in opening the 6th International Workshop on Clinical Pharmacology of HIV Therapy, Québec became the base of exploration that ended only at the continent's other side.
Antiretroviral pharmacology has survived more than four mean winters, but a workshop like the one in Québec can convince optimists that these are still early days. Research has barely started scratching through a continent-thick list of potential antiretroviral interactions. And new antiretrovirals -- along with other drugs people with HIV take -- are coming all the time.
Pharmacologic science has reached the point that sustains predictions of likely interactions. But this workshop showed that predictions can be wrong.
On top of that, people change. Kids grow up. Adults grow old. They get fat. They get hepatitis. All of these permutations and many more -- workshop attendees learned -- permeate the metabolic machinery in ways overt and subtle.
This workshop review begins with one variable that changes ceaselessly, and one that hardly ever shifts -- age and gender. Other topics considered are:
For the still unsated, a long table lists the many studies that saw no hurtful interactions with antiretrovirals.
Older people had higher lopinavir (LPV) levels in one study, but not in another. In the second study, though, seniors saw higher efavirenz (EFV) concentrations.
Earlier work charted dwindling activity of CYP3A4, the key PI-metabolizing enzyme, in older people.1 Less CYP3A4 means slower PI metabolism, and that should mean higher drug levels. But previous studies in people taking antiretrovirals saw no apparent effect of age on PI metabolism -- perhaps because those studies were too small.
Teresa Parsons (Johns Hopkins University, Baltimore) and AIDS Clinical Trials Group (ACTG) collaborators managed a bigger analysis in a pharmacologic substudy of treatment-naive people starting standard-dose LPV/ritonavir (RTV) with emtricitabine (FTC) and stavudine (d4T) [abstract 40, poster 3.6]. They prospectively measured LPV troughs 10 to 14 hours after dosing in 44 people, half of them 18 to 30 years old (median 26 years) and half 45 years old or older (median 50 years).
Gauging LPV troughs at weeks 24, 36, and 96, Parsons logged significantly higher levels in the older contingent. Though the overall difference fell just short of statistical significance (P = 0.056), the difference at week 24 was significant (P = 0.001) (Table 1).
In a multivariate analysis adjusted for age, hours since last dose, week of blood draw, and gender, only age independently predicted a higher LPV trough (P = 0.047). In a multivariate model reckoning age and gender, age alone predicted higher LPV troughs at week 24 (P = 0.0002), but age lost predictive significance at week 36 (P = 0.18) and week 92 (P = 0.33).
Parsons stressed that LPV levels overlapped considerably between younger folk and their elders, but most of the very high levels turned up in the older group -- including the highest reading in a 79-year-old.
In a retrospective British study EFV troughs proved significantly higher in postmenopausal women than in premenopausal women [abstract 54, poster 4.8]. This study failed to find a significant LPV trough difference in pre- and postmenopausal women, perhaps because the cohort included only seven postmenopausal women taking the PI. Earlier work chalked up 20% lower CYP3A4 content in postmenopausal women,2 but had not eyed the potential effect of age and menopausal status on CYP2D6, the main EFV metabolizer.
Sara Gibbons (University of Liverpool) ransacked records from Liverpool's therapeutic drug monitoring (TDM) service to cull LPV levels measured 10 to 14 hours after dosing and EFV concentrations toted eight to 16 hours after treatment in three groups of women:
The analysis excluded levels below 100 ng/mL -- a likely signal of bad adherence. The only significant difference arose in the comparison of premenopausal and postmenopausal women taking EFV (Table 2).
Gibbons noted two study limitations -- using age as an indicator of menopausal status and lack of data on contraceptive steroid or hormone replacement therapy.
But together these studies suggest that older people may run a higher risk of PI or nonnucleoside reverse transcriptase inhibitor (NNRTI) toxicity.
Lower weight -- but not gender -- predicted higher EFV sums in a retrospective analysis of people taking 800 or 600 mg of the NNRTI with rifampin (rifampicin) [abstract 19, poster 2.12]. Lisa Almond and University of Liverpool colleagues sifted the group's 1999 to 2004 TDM service data, looking for adults taking EFV and rifampin and scoring a drug level above 100 ng/mL to verify adherence.
Efavirenz concentrations tallied eight to 16 hours after dosing did not differ significantly in 111 people taking 800 mg of the NNRTI and 20 taking 600 mg -- both with rifampin. Neither group had a significantly higher number of people outside the 1,000- to 4,000-ng/mL target range. Nor did EFV levels differ between 58 men and 54 women taking 800 mg of EFV with rifampin or in 281 men and 79 women taking EFV without the tuberculosis (TB) drug.
Body weight did make a difference. Among the 111 people taking 800 mg of the NNRTI with rifampin, 42.5% of those weighing under 60 kg had an EFV reading above 4,000 ng/mL compared with 25.3% of those weighing more (odds ratio [OR] 0.46, 95% confidence interval [CI] 0.19 to 1.14, P = 0.09).
Higher weight meant a lower EFV trough in both men (rho = -0.27, P = 0.003) and women (rho = -0.15, P = 0.004). Nested regression analysis linked weight (P = 0.03) but not gender or EFV dose to EFV tabs topping 4,000 ng/mL.
While gender did not affect EFV quotients, one might expect a gender difference in PI levels because research confirms higher hepatic CYP3A4 expression in women. That would speed metabolism of CYP3A4 substrates like saquinavir (SQV), and stepped-up metabolism usually means faster clearance and lower levels. But most PIs these days come with a kick from RTV, a fierce CYP3A4 inhibitor.
Ritonavir seems to be winning the metabolism contest in women taking it to boost SQV -- stunting elimination of the PI and expanding its exposure. Laura Dickinson (University of Liverpool) confirmed earlier studies showing higher SQV levels in women than in men in a retrospective analysis of 28 men and six women taking 1,000/100 mg of SQV/RTV twice daily [abstract 9]. Dickinson extended previous work by weighing drug transporter expression as a possible mechanism behind this gender difference in a separate study of 63 men and 30 women.
This retrospective analysis charted much higher SQV and RTV levels in the six women than in the 26 men. Gender did not affect the terminal half-life of either PI (Table 3).
Looking at drug transporter expression on peripheral blood mononuclear cells (PBMCs), Dickinson found that women had significantly lower levels of P-glycoprotein (0.66 versus 0.99 relative fluorescence units, P = 0.0016) and multidrug-resistance protein (MRP) 1 (0.15 versus 0.35, P = 0.018). Whether transporter expression on PBMCs closely reflects expression in the intestine and other sites is unknown.
With the steep plunge in HIV prevalence among children in developed countries, kids stand likely to remain a pharmacologically understudied contingent until more trials in unluckier lands bear fruit. That gap poses a stern challenge since antiretroviral dosing is vastly more complicated in ever-changeful children than in adults.
Big Western trial groups such as the Paediatric European Network for Treatment of AIDS (PENTA) and ACTG have tried to fill that gap, as have researchers in countries with still woefully flush pediatric cohorts like South Africa and Thailand. Edmund Capparelli (University of California, San Diego) came to the workshop with data drawn from five ACTG trials that cataloged 2,449 nevirapine (NVP) concentrations in 495 infants and children ranging in age from one month to 19.5 years (mean 6.6 years) [abstract 37]. The results suggested a more reliable dose than the US Food and Drug Administration (FDA)-sanctioned draught for children more than a few years old.
While the ACTG has tested NVP doses of 120 or 200 mg/m2 twice daily, the European PENTA team favors 150 mg/m2 twice daily while recognizing that some children may need 200 mg. The FDA licensed doses of 7 mg/kg twice daily for children from two months to eight years old, and 4 mg/kg twice daily starting at eight years.
The ACTG cohort represented a close split between boys (54%) and girls (46%). Reflecting HIV epidemiology in the United States, the group consisted mainly of blacks (54%) and Hispanics (31%). The analysis rested on large numbers of infants and children in each age bracket -- 40 under six months old, 46 between six months and two years, 152 between two and six, 206 between six and 12, and 51 older than 12.
The ACTG ran two population pharmacokinetic analyses to hone a model that correlated dose with a target area under the concentration-time curve (AUC) of 63.6 µg·h/mL (Table 4). The FDA dose and 150 mg/m2 yielded similar NVP exposure in the first years of life. But the 150-mg dose resulted in more consistent NVP AUCs across age groups:
Ritonavir lowered NVP clearance by 23% across age groups, and nelfinavir (NFV) trimmed clearance of the NNRTI 28% in infants. Nevirapine clearance dwindled 30% from age five to 18.
Capparelli concluded that 150 mg/m2 "is less likely to underdose older children than the FDA weight-based dosing." Whatever the dose, he added, high interindividual variability in NVP exposure suggests dose monitoring may be helpful.
Discovering the interactions between PIs and other drugs (including other PIs) is an odyssey more trying and less terminable than the fraught adventures of Homer's hero.
Development of every new PI, every new formulation, and every new dose demands another set of studies to steer the drug between the Scylla of side effects and the Charybdis of incompetence. And every time someone takes a PI with another drug, the fretful pharmacologist must ensure that other drug is no Circe set to transmute PIs into swinish distraction.
Workshop attendees heard news on potentially pernicious interactions between PIs and lamotrigine, tacrolimus, prednisone, rifampin, and paroxetine. Other studies explored PI-PI liaisons and how liver disease affects boosted PI levels.
Epileptics with HIV take lamotrigine, as do some people with peripheral neuropathy. High-dose RTV induces glucuronidation of certain drugs, and lamotrigine's metabolism depends on hepatic glucuronidation. But no one assessed the effects of low-dose RTV on lamotrigine until Manon van der Lee (Radboud University Nijmegen Medical Center, The Netherlands) tested standard-dose LPV/RTV in 24 healthy volunteers taking lamotrigine [abstract 12]. Results suggested the need to double lamotrigine's dose with these PIs.
Twelve women and 12 men started lamotrigine at 50 mg daily for two days, then upped the dose to 100 mg twice daily on study days three through 10. At that point they added 400/100 mg of LPV/RTV twice daily for the next 10 days.
Comparing day 10 and 20 lamotrigine troughs, the Nijmegen team planned to stop the study or adjust the dose depending on results:
In the 18 volunteers who completed 20 days of treatment, van der Lee found lamotrigine troughs halved compared with day 10 levels (Table 5). With the doubled dose of 200 mg twice daily, 15 people completed the study at day 31. The inflated dose brought total lamotrigine exposure and peak levels back to day 10 levels, while the trough improved markedly.
Lopinavir and RTV AUCs and troughs on day 20 matched those of historical controls. The study also assessed levels of lamotrigine's main metabolite, but those results had not been analyzed at the time of the workshop.
The Nijmegen group concluded that clinicians should bump lamotrigine to 200 mg twice daily in people taking the antiepileptic with LPV/RTV. Although van der Lee suggested low-dose RTV may have the same effect when boosting other PIs, Richard Bertz (now at Bristol-Myers Squibb, but formerly at Abbott Laboratories) cautioned that the effects of low-dose RTV are not always consistent from one drug to the next.
With liver transplantation a life-saving option for many people with well-controlled HIV infection, the interactions between the antirejection agent tacrolimus and antiretrovirals remain largely undefined. Anne-Marie Taburet (Bicêtre Hospital, Paris) learned that EFV and nucleoside reverse transcriptase inhibitors (NRTIs) have little effect on tacrolimus metabolism, but LPV/RTV and NFV slow elimination of the drug to a glacial pace [abstract 26, poster 2.24]. This small study showed, however, that well-planned monitoring permits accurate dose adjustment.
Taburet tracked tacrolimus and antiretroviral levels in eight people coinfected with hepatitis C virus (HCV) who had orthotopic liver transplantation with a CD4 count above 150 cells/mm3 and an HIV load below 50 copies/mL. Three were taking LPV/RTV, two NFV, two EFV, and one a triple-NRTI regimen. They stopped their antiretrovirals on the transplant day and restarted 10 days later. The immunosuppressive regimen included tacrolimus and prednisone.
Measuring tacrolimus levels when liver function stabilized (about 10 days after transplantation), Taburet recorded oral clearances ranging from 5.3 L/h to 19.4 L/h. After 10 days of NFV, clearance of the transplant drug plunged to a range of 1.1 to 3.0 L/h, and after 10 days of LPV/RTV to 0.5 to 0.9 L/h. With LPV/RTV the half-life of tacrolimus stretched up to 234 hours, and Taburet had to slow dosing of the drug to once every five to 10 days. With all patients she aimed for tacrolimus targets of 8 to 20 ng/mL from day 0 to week 6 and from 5 to 15 ng/mL afterwards.
Tacrolimus had no effect on antiretroviral pharmacokinetics, and everyone maintained a viral load below 50 copies/mL. The Bicêtre team concluded that managing tacrolimus-PI interactions is feasible with attentive monitoring.
Taburet did not report interactions between antiretrovirals and prednisone, but a study in healthy volunteers found that low-dose RTV jacks up exposure of prednisone's metabolite prednisolone in four days [abstract 14]. This study by Scott Penzak (US National Institutes of Health, Bethesda, Maryland) looked at 200 mg of RTV twice daily rather than the more frequently used 100-mg twice-daily dose.
Rare organ transplants are not the only reason people with HIV take this corticosteroid, Penzak reminded colleagues. It also has indications for Pneumocystis pneumonia, asthma, inflammatory arthritis, and some cancers. Research links corticosteroids to osteonecrosis in people with HIV.3,4 Corticosteroid metabolism relies on CYP3A4, so potent inhibitors of this enzyme -- like RTV -- could boost prednisolone exposure.
Ten healthy volunteers took single doses of prednisone before starting RTV, then four and 14 days after they began a twice-daily RTV dose of 200 mg. Penzak measured prednisolone levels before RTV dosing and on day four in 10 people, and on day 14 in nine people.
Ritonavir raised prednisolone exposure in everyone studied, while clearance of the corticosteroid slowed and its half-life lengthened (Table 6). Troughs of RTV on day 14 reflected those seen in healthy controls.
Penzak believes this change in prednisolone exposure may be clinically relevant, especially in people taking prednisone over the long term. Whether 100 mg of RTV twice daily would have the same impact is anyone's guess. When Penzak submitted his study plan for ethical review, 200 mg twice daily was more common than it is today.
Earlier this year Roche Laboratories alerted clinicians to the risk of liver toxicity in people taking SQV/RTV with the TB drug rifampin (rifampicin). The startling result compelled Roche to contraindicate rifampin with SQV/RTV twice daily, even though some work suggests the toxicity arises only in people without HIV infection.5 In people with HIV, on the other hand, cohort studies saw no toxic threat with the combination.6
Roche Laboratories' Malte Schutz detailed results of the 28-man study at the workshop [abstract 35]. Rifampin, a potent CYP3A4 inducer, trimmed exposure of unboosted SQV 70% in earlier work. Schutz and colleagues planned this trial in healthy volunteers to see if a 100-mg RTV boost would erase that deficit.
In a crossover design they randomized volunteers to take 1,000/100 mg of SQV/RTV twice daily or 600 mg of rifampin daily for 14 days. Then the PI group added rifampin and the rifampin group added the PIs.
Roche halted the trial before the crossover because the first nine men who added SQV/RTV to rifampin all had grade 3 or 4 transaminase elevations. At that point two of eight people in the other arm had grade 2 or 3 transaminase spurts. In the group that added the PIs after 14 days of rifampin, the transaminase jumps were dramatic, ranging from 12 to 70 times the upper limit of normal.
These changes came quickly. On day 14 everyone had rifampin or PI levels in the normal range. The liver enzyme leaps all happened in the next one to three days. One man in the group that added SQV/RTV to rifampin landed in the intensive care unit, but everyone regained normal liver function when treatment stopped.
The limited data that could be gathered before the study ended suggested that SQV/RTV slowed the breakdown of rifampin and its metabolite desacetylrifampin. That could account for the mounting transaminase readings, Schutz proposed.
The contraindication of rifampin with SQV/RTV comes as a keen blow to countries with a high prevalence of HIV/TB coinfection. Since clinicians usually try to control TB before attacking HIV, the apparently more dangerous rifampin-plus-PI sequence would be common.
Because this toxicity did not emerge in HIV-infected people combining rifampin with SQV/RTV -- but only in healthy volunteers -- some workshop attendees were reluctant to accept a permanent ban on combining these therapies. And Schutz said his company will not abandon study of the combination. Coincidentally, Argentine researchers found themselves in the midst of a rifampin/SQV/RTV trial -- in people with HIV -- when Roche Laboratories announced its results. Blood samples from that study are now analyzed.
Another shocker on a smaller scale came in a study combining fosamprenavir (FPV)/RTV with the antidepressant paroxetine [abstract 13]. One might expect RTV-boosted PIs to inhibit metabolism of drugs (like paroxetine) whose metabolism depends on CYP2D6, explained David Burger (Radboud University Nijmegen Medical Center). But expectation did not conform to reality in this study of 26 healthy volunteers, 18 of them women.
The crossover design called for half of the volunteers to take 20 mg of this selective serotonin reuptake inhibitor (SSRI) once daily for 10 days and for the other half to take it with 700/100 mg of FPV/RTV twice daily. After a 16-day washout, volunteers jumped to the alternate regimen for another 10 days.
Paroxetine did not affect levels of amprenavir (APV) or RTV, but the PIs sliced paroxetine's AUC by 55%. The SSRI's maximum concentration proved 51% lower with FPV/RTV than when given alone. Paroxetine's half-life dwindled 25%.
What happened? Burger had a list of possibilities (Table 7).
Although this SSRI looks safe with FPV/RTV, Burger concluded, clinicians may have to raise its dose to achieve the desired antidepressant effect.
Good absorption of atazanavir (ATV) depends on an acidic gastric stew, so drugs that quench stomach acid may lower levels of this PI. That's just what happened in a study combining 40 mg of the proton pump inhibitor omeprazole with 300/100 mg of ATV/RTV once daily, and Bristol-Myers Squibb advised clinicians not to mix their PI with proton pump drugs.
But what about H2 receptor antagonists, the other major class of acid quellers? Data presented by Bristol-Myers Squibb's Sangeeta Agarwala suggested that the H2 antagonist famotidine and ATV can be taken safely with or without RTV [abstract 11]. But the studies involved healthy volunteers, whose gastric acidity differs from that of people with HIV infection.
In the first of two studies, 64 volunteers took either 400 mg of ATV in the morning or 400 mg in the evening for six days. Over the next week the morning-dose group added either (1) 40 mg of famotidine twice daily, (2) 40 mg of famotidine twice daily plus 8 ounces of cola, or (3) 40 mg of famotidine twice daily and 100 mg of RTV while trimming the ATV dose to 300 mg. The evening-dose group added 40 mg of famotidine, with the first famotidine dose taken 10 hours before ATV and the second two hours after ATV.
For people who took 400 mg of ATV in the morning, adding 40 mg of famotidine twice daily lowered ATV levels 40% to 50% compared with ATV alone. Quaffing a cola with the H2 blocker didn't help. Compared with 400 mg of ATV in the morning, 300/100 mg of ATV/RTV plus famotidine yielded a 79% higher ATV AUC, a 346% higher ATV trough, and an equivalent ATV peak. Separating the evening ATV dose from the two famotidine doses yielded ATV peaks and AUCs equivalent to those with ATV alone. In this arm the ATV trough with famotidine was 21% lower than with 400 mg of ATV alone.
The second protocol started with 300/100 mg of ATV/RTV in the morning for 10 days. Over the next 10 days the 48 volunteers added (1) 40 mg of famotidine twice daily, (2) 40 mg of famotidine twice daily plus 8 ounces of cola, or (3) 40 mg of famotidine twice daily plus an extra 100 mg of ATV.
Compared with ATV concentrations at the 300/100-mg dose, adding famotidine clipped the AUC 18%, the trough 28%, and the peak 14%. Adding famotidine with a cola chaser shaved ATV readings even more. The 400/100-mg dose plus famotidine yielded an ATV AUC and peak concentration similar to that attained with 300/100 mg without famotidine. The trough with 400/100 mg plus famotidine was 14% lower than with the 300/100-mg dose without famotidine.
Putting these pieces together, Bristol-Myers Squibb concluded that either (1) a 300/100-mg dose of ATV/RTV plus famotidine or (2) a 400-mg evening ATV dose 10 hours after the morning famotidine and two hours before the evening famotidine yields ATV levels equivalent to those reached with 400 mg once daily. Most clinicians probably hope for the ATV exposure they get with RTV-boosted 300/100 mg once daily. To reach those levels with famotidine, the ATV/RTV dose should be 400/100 mg.
A study of 28 heavily pretreated people switching to standard doses of ATV plus LPV/RTV (and other antiretrovirals) found adequate levels of both boosted PIs after an average 29 weeks of follow-up [abstract 56, poster 4.10]. Fasting lipids and bilirubin rose slightly.
Michael Zilly (University of Würzburg, Germany) and colleagues checked 171 plasma PI levels, planning to raise doses if they saw consecutive LPV troughs below 3,500 ng/mL or consecutive ATV troughs below 250 ng/mL. They never did. In fact they cut the LPV dose in two people with high troughs and nagging diarrhea.
The median LPV trough measured 4,472 + 2,226 ng/mL and the median ATV trough 695 + 450 ng/mL. Lopinavir levels never fell under 1,400 ng/mL and ATV levels never dipped under 150 ng/mL. Both drugs peaked four hours after dosing at a median 10,201 + 3,472 ng/mL for LPV and 2,544 + 638 ng/mL for ATV. Zilly rated virologic suppression "adequate" but did not define responses.
A 12-person study presented at this year's European HIV Drug Resistance Workshop charted higher LPV levels with 300 mg of ATV than in historical controls taking only LPV/RTV or LPV/RTV plus SQV.7
Hepatitis virus-coinfected people with cirrhosis had higher APV levels than coinfected people with chronic hepatitis or people infected only with HIV in a study reported by Elena Seminari (San Raffaele Scientific Institute, Milan) [abstract 66, poster 5.8]. She ran two analyses of APV exposure in people taking 700/100 mg of FPV/RTV twice daily in an expanded access program.
The first study involved 15 controls infected only with HIV, 12 coinfected people with chronic hepatitis (10 with HCV and two with hepatitis B virus [HBV]), and six coinfected people with histologically or clinically diagnosed cirrhosis (all with HCV).
Amprenavir troughs measured before the morning dose of FPV/RTV averaged 2,066 ng/mL in controls, 1,431 ng/mL in the chronic hepatitis group, and 4,678 ng/mL in the cirrhotic group. The geometric mean ratio for troughs was 1.90 (95% CI 1.07 to 3.01) when comparing the cirrhotic group with controls and 2.46 (95% CI 1.9 to 4.58) when comparing the cirrhotic group with the chronic hepatitis group.
Seminari also collected blood samples over 12 hours in 14 men -- six controls, four with chronic hepatitis, and four with cirrhosis. Again the group with cirrhosis had substantially higher APV exposure than the other two groups (Table 8).
A second observational study of coinfected people taking standard-dose LPV/RTV reached a similar conclusion: José Moltó (Germans Trias i Pujol University Hospital, Badalona, Spain) found that HCV barely upset LPV levels in coinfected people without impaired liver function [abstract 36].
This analysis involved 22 coinfected people with a Child-Pugh score below 6 and 18 people infected only with HIV. Everyone had taken standard-dose LPV/RTV for at least four weeks. The groups compared well in age and weight. As one would expect, the HCV group had higher levels of aspartate aminotransferase (53.5 U/L versus 23.0 U/L) and alanine aminotransferase (56.0 U/L versus 23.0 U/L). Measuring LPV levels at steady state, Moltó found no significant differences in mean values between the two groups (Table 9).
Ritonavir troughs did prove significantly higher in the coinfected group (1.12 versus 0.67 mg/L, P = 0.027), and Moltó saw a trend toward higher RTV AUC (6.90 versus 4.82 mg·h/L) in the people with HCV (P = 0.067). But these higher levels apparently had only a modest impact on LPV. Gender, age, body mass index, and transaminase readings did not affect LPV pharmacokinetics.
Workshop attendees observed that studies like these may suffer from selection bias: Coinfected people whose PI levels climb quickly may stop their protease drugs before they can be selected for study.
Atazanavir has won the interest of pharmacologists curious to define its therapeutic range and to predict how people will respond to it. Other PI research presented at the workshop involved potential links between LPV levels and lofty lipids.
Stefano Bonora (University of Turin, Italy) confirmed his preliminary finding that the therapeutic range of ATV stretches from 150 to 850 ng/mL [abstract 60]. Bonora's analysis involved 51 people, 34 of them men, enrolled in an ATV expanded access program. Eleven already had a viral load below 50 copies/mL; in the others median viremia stood at 3.5 logs (interquartile range [IQR] 1.79 to 4.9 logs). Seventeen people had never tried a PI, and the median number of previous PIs was 1 (range 0 to 6). Thirty people (59%) started ATV/RTV (300/100 mg).
Defining virologic response as fewer than 50 copies/mL or more than a 2-log drop in viral load, Bonora counted 40 responders among 50 people at week 12 (80%) and 34 among 45 people at week 24 (75.5%). At week 24 the ATV trough averaged 208 ng/mL among people taking unboosted ATV and 1,046 ng/mL among those getting a lift from RTV.
Comparing 35 week-24 responders and 10 nonresponders, Bonora found three significant differences:
A GIQ based on ATV-specific mutations did not significantly discriminate responders from nonresponders (412 versus 7, P = 0.08).
Among people with a trough above 150 ng/mL, 81% had a 24-week virologic response, versus 37.5% with a trough below that cutoff (P = 0.027). As a response predictor a trough above 150 ng/mL had 88% sensitivity but only 50% specificity. People with a trough under 150 ng/mL had a 4.1 times higher risk of failure.
Among people with a PI-based GIQ above 100, 88.8% had a virologic response at 24 weeks versus 33.3% with a GIQ under 100 (P = 0.006). A GIQ of 100 had 89.5% sensitivity and 85.7% specificity in predicting response. People whose GIQ languished below 100 had a 4.7 times higher risk of failure.
Foraging for the high end of the therapeutic range, Bonora found that an ATV trough atop 850 ng/mL predicted total bilirubin readings above 2.5 mg/dL with 60% sensitivity and 78% specificity. The same trough predicted an unconjugated bilirubin of 2.0 mg/dL with 66.7% sensitivity and 78.6% specificity.
Among study participants whose ATV trough fell within the 150- to 850-ng/mL target range, 75% had a virologic response and 17% had elevated bilirubin (Table 10). Although everyone with a trough above 850 ng/mL responded, 40% of them had high bilirubin.
Bonora concluded that 150 to 850 ng/mL represents a reliable therapeutic range for ATV. But a Bristol-Myers Squibb representative wondered whether clinicians would sacrifice too many responses with a top trough cutoff of 850 ng/mL. Everyone with a trough above that mark responded virologically. Although 40% of them had high bilirubins, the Bristol-Myers Squibb representative argued, some may see that as only a "cosmetic" side effect. Only two people in Bonora's group stopped ATV because of hyperbilirubinemia.
Bruno Lacarelle (Timone Hospital, Marseille, France) also found that GIQ predicts virologic response to ATV [abstract 1]. But unlike Bonora he determined that both a GIQ based on ATV-specific mutations and a GIQ based on all protease mutations predicted response. The different result may reflect (1) disparate response definitions, (2) more RTV boosting in Lacarelle's cohort, and (3) the different populations studied: More people in Lacarelle's group had PI experience. Lacarelle's findings also differed from Bonora's in that he could not define a trough cutoff that predicted response.
The Marseille cohort included 35 people with treatment experience and one naive to antiretrovirals. While 32 took RTV-boosted ATV, four took the unboosted PI. The group had tried a median of two PIs (range zero to five). They had a median of three (range zero to 15) PI mutations and one (zero to 10) ATV-related mutation.
After two months of treatment 26 people (72%) had a virologic response, defined as a viral load below 400 copies/mL or at least a 1-log drop in viremia. Atazanavir trough and a GIQ based on all protease mutations did not predict the month-3 response, but number of PI mutations, number of ATV-specific mutations, GIQ based on ATV mutations, and CD4 gain did correlate with response (Table 11).
In a multivariate analysis ATV trough, PI mutations, and ATV-specific mutations did not predict response, but both the PI mutation-based GIQ and the ATV mutation-based GIQ did. Lacarelle reckoned the ATV GIQ response cutoff at 2.3 because 25% of people with a GIQ under 2.3 responded versus 75% with a higher GIQ (P = 0.014).
Yet in a study presented earlier this year, Bonora's colleague Daniel Gonzalez de Requena set a PI-based GIQ cutoff at 60.8 Lacarelle explained that the wide difference between that result and his reflects the log transformation of his GIQ calculation. One attendee observed that if HIV pharmacologists expect clinicians to start using GIQ, they will have to agree on methods.
Authorities in British Columbia recommend the 300/100-mg dose of ATV/RTV -- and TDM -- for everyone starting this once-daily PI. But do some people need even higher doses?
To find out, Chris Alexander (University of British Columbia) sorted TDM records of people starting ATV/RTV between September 2003 and December 2004 [abstract 91, poster 8.3]. Alexander found that 34 people (25%) had a PI dose adjustment. As a result the Vancouver team made 42 pairwise comparisons:
Looking at steady-state ATV and RTV levels in people not changing other antiretrovirals around sampling time, Alexander found that the first 28 dose changes bolstered the percentage of people with a detectable RTV level (>80 ng/mL) from 21% to 58% (P<0.01). The eight switches from 300/100 to 400/200 mg lifted the percentage of people with a detectable RTV level from 12.5% to 75% (P<0.01). Although 35% of study participants also had at least a 30% jump in their ATV trough with the dose adjustment, overall ATV exposure did not climb significantly.
The British Columbia team concluded that the recommended 300/100-mg dose yields adequate ATV concentrations for most people. While raising the dose to 300/200 mg once daily improves ATV exposure in a subset of people, the 400/200-mg dose "may more reliably increase ATV levels without further increasing the cost."
Research to date disagrees on whether high LPV exposure correlates with higher lipid and glucose levels.9-12 The latest look at this question, a retrospective analysis involving 20 people, found that LPV troughs above 7 mg/L may drive dangerous jumps in total cholesterol and triglycerides [abstract 44, poster 3.10].
Of the 20 people studied, five were starting LPV/RTV as part of their first regimen. The group had a mean baseline CD4 count of 181 cells/mm3 (+ 141 standard deviations [SD]) and a mean viral load of 4.5 log (+ 1.1 SD). For the whole group, fasting glucose and triglycerides did not change significantly during six months of LPV therapy, but total cholesterol climbed significantly by treatment month one and remained elevated at month six (Table 12).
Total cholesterol at month six correlated significantly with LPV's trough (r = 0.583, P<0.001). In four people whose LPV trough stood above 7 mg/L, triglycerides climbed 111% and cholesterol 56%. In comparison eight people with a trough between 3 and 7 mg/L averaged a 22% gain in cholesterol and a 35% triglyceride spurt.
Tenofovir has had its ups and downs since winning approval as an antiretroviral nucleotide analog. No one doubts the potency or convenience of this once-daily pill, now combined in a single tablet with FTC and perhaps soon with FTC and EFV as a once-daily troika. Tenofovir boasts a handsome short-term toxicity profile, though its long-term impact on kidneys13 and bones14 may prove more problematic.
But the worst news about TDF came from studies of people taking it with two other purine analogs -- didanosine (ddI) and abacavir (ABC). In a cohort coupling TDF with 400 mg of ddI, more than half of 302 people lost more than 100 CD4 cells/mm3 over 48 weeks, and a third lost more than 200 cells/mm3 -- even though they responded virologically.15 Tenofovir combined with ddI, ABC, and/or lamivudine (3TC) -- without a PI or NNRTI -- fell far short virologically in a spate of studies.16-20
When TDF plus ddI also flopped virologically in people starting it with EFV,21-23 the drugs' two makers warned European clinicians away from combining TDF and ddI.24 At the 6th International Workshop on Clinical Pharmacology of HIV Therapy, Giovanni Di Perri (University of Torino, Italy) wondered why regulators had not banned TDF/ddI combinations from the start, since both drugs compete for adenosine in binding to reverse transcriptase. But he admitted he overlooked that inconvenient fact and prescribed the twosome as avidly as anyone, looking to build tolerable once-daily regimens.
Much head scratching fostered some hypothesizing on the internecine nuances between TDF and other reverse transcriptase inhibitors. The workshop's keynote speaker, Daniel Kuritzkes (Harvard Medical School, Boston), proposed that certain nucleosides (-tides) team up to stifle evolution of resistant virus, while other combinations appear to promote resistance (Table 13).
Notably, TDF/3TC turns up in both columns of this table -- and no one knows why. In lab dishes TDF handcuffs the 3TC- and FTC-resistant M184V mutant. Yet in trials that combined TDF and 3TC with ABC in treatment-naive people, M184V arose and the regimen crashed.16,17 One way to explain this seeming antinomy, Kuritzkes proposed, may be that different resistance mutations can emerge from different cellular compartments, as happened in the GlaxoSmithKline study of TDF, 3TC, and ABC.16,25
Kuritzkes urged HIV pharmacologists to look harder at NRTI effects in different cell subsets, though no one calls that assignment easy. Only eight of 98 workshop studies (8%), for example, sized up intracellular quotients of NRTI triphosphates. But Séverine Compain (SPIBO/CEA Salay, Gif-sur-Yvette, France) did propose a method for measuring triphosphates in different cell populations (combined high-performance liquid chromatography and tandem mass spectrometry) [abstract 68, poster 5.15].
Late in 2004 Thomas Kakuda (Roche Laboratories) and colleagues proposed that both the CD4 decays with TDF/ddI and the antiviral frangibility of TDF/ddI/ABC rest in TDF's inhibition of purine nucleoside phosphorylase (PNP).26 Because PNP breaks down naturally occurring purines, thwarting PNP would promote purine pileups -- especially deoxyadenosine triphosphate (dATP) or deoxyguanosine triphosphate (dGTP). That could do two things, Kakuda explains:
The second effect would enfeeble purine analogs like ddI and ABC when combined with TDF.
Kakuda suggests clinicians may want to limit TDF's nucleoside teammates to the pyrimidine analogs zidovudine (AZT), d4T, 3TC, and FTC. But that may not be a failsafe strategy, according to one workshop report:
At this year's workshop, Andrew Luber (University of Pennsylvania, Philadelphia) extended Kakuda's hypothesis, suggesting that CD4s may also tumble in people taking TDF without ddI -- though five of nine people studied took TDF with ABC [abstract 53, poster 4.7].
Luber looked at CD4 changes in seven men and two women who had at least two consecutive RNA readings below 400 copies/mL yet lost CD4 cells on at least two measures separated by at least two months. He eliminated anyone who might have T-cell deficits due to other drugs or illnesses. The regimens involved a PI boosted by RTV in five people, quadruple NRTIs in two, triple NRTIs in one, and TDF plus ABC and NVP in one.
These nine people had taken their current combination for a median of 21 months (range six to 40 months). Over a median of 11 months (range four to 25 months), their CD4 counts plunged by a median 297 cells/mm3 (range 88 to 615 cells/mm3).
What explains these CD4 drops in people taking TDF without ddI? For the five people also taking ABC, Luber invoked Kakuda's hypothesis that TDF-provoked purine spirals could tax the ability of ABC's triphosphate to bind to reverse transcriptase. But that would not explain why ABC apparently worked well virologically in these people.
For the five people taking TDF with a boosted PI, Luber observed that low-dose RTV may inhibit expression of the drug transporter multidrug-resistance protein 4 (MRP4) on T cells. That would favor intracellular stockpiling of TDF's active metabolite. And more intracellular TDF would mean more PNP inhibition -- the mechanism Kakuda elects to explain CD4 ebbs. (Another workshop study found that RTV inhibits MRP2 in a renal cell line, an effect that may explain TDF's renal toxicity [abstract 23, poster 2.16].) According to Luber, an ongoing study of LPV/RTV plus TDF/FTC saw no CD4 reversals through 48 weeks of treatment, but a few downturns after that.
As Luber cautions, an uncontrolled analysis like this cannot eliminate other possible explanations of CD4 swoons in people taking TDF without ddI. Perhaps natural variability in CD4 tallies accounts for these nine cases; only three of them also had drops in CD4 percent. Or CD4 slumps could occur as often with non-TDF regimens but go unreported in trials dominated by good CD4 responders.
Tenofovir's active metabolite lingers in PBMCs for more than a week after people stop the drug, according to results of a study by Jacques Grassi (CEA, Gif-sur-Yvette, France) [abstract 20, poster 2.13]. The analysis, published just after the workshop,27 involved eight people taking TDF without ddI, 16 taking 400 mg of ddI daily without TDF, and 14 taking TDF plus ddI at 250 mg daily. A longitudinal substudy assessed intracellular TDF and ddI levels in three people quitting TDF and raising their ddI dose from 250 to 400 mg daily.
Comparing people taking each drug alone with those combining the NRTIs, Grassi found no difference in intracellular concentrations of the active metabolites:
Neither did Grassi see significant variation between intracellular TDF readings from one TDF dose to the next. But TDF's active diphosphate metabolite could be measured for three weeks in all three people who stopped the nucleotide analog. Grassi figured the intracellular half-life at 7.5 days.
Tenofovir's perdurable intracellular half-life probably helps explain the drug's potency. But it also poses a risk of resistance if people stop TDF along with all other drugs in their regimen, or if they stop TDF and continue a poorly suppressive regimen. That would allow HIV to replicate in the face of dwindling -- but persistent -- TDF levels.
Because neither plasma nor intracellular NRTI half-lives predict responses to regimens containing these drugs, Trevor Scott and GlaxoSmithKline colleagues explored a new inhibitory quotient -- the intracellular inhibitory quotient (IIQ) [abstract 7, poster 1.7]. As with the traditional IQ and the GIQ, a higher IIQ should mean a better response because it accounts for both drug exposure (usually measured as a trough) and viral susceptibility.
Scott tried two IIQ formulas:
IIQ = Intracellular NRTI tri- or diphosphate trough/50% inhibitory concentration (IC50)
IIQ = Intracellular NRTI tri- or diphosphate trough/Ki
Ki is the concentration of NRTI needed to overcome competition by endogenous triphosphate for incorporation into cellular DNA by HIV reverse transcriptase.
With either equation the 24-hour IIQ for ABC easily topped the IIQ of TDF. With Ki as the denominator, the IIQ for ABC's active metabolite (carbovir triphosphate) measured 4, compared with 2 for TDF's active diphosphate metabolite. With IC50 as the denominator the IIQs were 2 for ABC and 0.1 for TDF.
Scott thinks the Ki equation works better because the low TDF IIQ with the IC50 formula belies TDF's potency in clinical trials. But the bigger question is whether the IIQ truly predicts potency or virologic response.
Scott made one cross-study comparison suggesting no difference in response to ABC or TDF in treatment-naive people taking either drug with 3TC plus EFV for 48 weeks. In Gilead's GS-903 trial 76% taking TDF plus 3TC/EFV had a viral load below 50 copies/mL at 48 weeks.14 In GlaxoSmithKline's CNA30024 study 70% taking ABC with 3TC/EFV reached the same mark.28 Tenofovir and ABC proved virologically equivalent as replacements for AZT or d4T in a randomized 48-week trial.29
At the very least this inquest confirms that responses to NRTIs remain notoriously difficult to predict. One variable in that difficulty is variability itself, according to a study of plasma AZT, d4T, and 3TC levels in 88 people taking those drugs with indinavir (IDV) or NFV in the French Cophar 1 trial [abstract 58, poster 4.12].
France Mentré (Bichat Hospital, Paris) toted NRTI levels in 88 people who had a viral load below 200 copies/mL after eight months of treatment with IDV or NFV plus AZT/3TC or d4T/3TC. The study group included 54 taking 3TC, 39 taking d4T, 27 taking AZT, 42 taking IDV, and 46 taking NFV.
NRTI absorption, oral clearance, and volume of distribution varied greatly from one person to the next (Table 14).
These parameters also varied with the PI taken. Compared with IDV, NFV lowered absorption of 3TC 1.4-fold (P = 0.108), boosted AZT's distribution volume 1.5-fold (P<0.0001), and hoisted oral clearance of d4T 1.6-fold (P = 0.001) and of AZT 1.5-fold (P = 0.001).
How such changes may affect intracellular triphosphate levels -- and response -- remains to be determined.
Earlier studies disagree on whether TDF lowers levels of ATV. Two studies recorded lower ATV troughs with TDF, whether ATV got the RTV boost30 or not.31 But a cross-sectional study of 79 people found no ATV trough difference in people taking the RTV-boosted PI with or without TDF.32 At this year's workshop, Sangeeta Agarwala and Bristol-Myers Squibb colleagues offered data confirming reasonable ATV exposure with simultaneous TDF and ATV/RTV at 300/100 mg [abstract 16, poster 2.9].
The study involved 28 healthy adults, 18 of them men and 10 women. For 10 days they took 300 mg of TDF daily. After a washout everyone took 300/100 mg of ATV/RTV in the morning for 10 days followed by either:
Atazanavir's AUC proved 38% higher with the 400/100-mg dose plus TDF than with the 300/100-mg dose without TDF. The PI's peak concentration was 31% higher with 400/100 mg plus TDF than with 300/100 mg alone. Tenofovir's AUC was 55% higher and its peak 39% higher with 400/100 mg of ATV/RTV than without the PIs. Because of the higher exposure of both drugs, Bristol-Myers Squibb does not recommend 400/100 mg of ATV/RTV with TDF.
Taking 300/100 mg of ATV/RTV in the morning and 300 mg of TDF in the evening did not yield ATV or TDF exposures substantially different from those attained with the drugs taken alone. Separating the doses lowered ATV's AUC 11%, its peak 14%, and its trough 20%. Tenofovir's AUC, peak, and trough were 37%, 34%, and 29% higher with separated dosing than when taken without ATV/RTV.
Agarwala proposed that separating the doses "may not provide a significant clinical benefit compared with simultaneous dosing." Bristol-Myers Squibb endorses the 300/100-mg dose of ATV/RTV given simultaneously with TDF, guidance that will make life easier for people taking these drugs.
In a 10-person pharmacologic substudy of the ANRS Puzzle 2 trial, Anne-Marie Taburet (Bicêtre Hospital, Paris) found TDF exposure with ATV/RTV 300/100 mg equivalent to TDF levels seen historically in healthy and HIV-infected people taking the drug [abstract 55, poster 4.9].
The study involved 10 people in whom at least two PI regimens and one NNRTI regimen had failed. They switched to ATV/RTV plus TDF and other NRTIs with a median viral load of 5.1 log (range 4.1 to 5.7 log) and a median CD4 count of 117 cells/mm3 (range 19 to 328 cells/mm3).
Steady-state median TDF AUC measured 2,253 ng·h/mL (range 1,305 to 11,285 ng·h/mL), peak 234 ng/mL (range 121 to 855 ng/mL), and trough 45 ng/mL (range 23 to 215 ng/mL). Over the 26 weeks of treatment, the median TDF trough varied from 58 to 73 ng/mL. The within-patient coefficient of variation was 67%.
Intracellular levels of TDF's active diphosphate metabolite proved higher with LPV/RTV than without [abstract 34, poster 2.27], a finding confirming studies that measured TDF plasma concentrations.33,34
Henri Benech (SPIBIO/CEA Saclay, Gif-sur-Yvette, France) compared TDF diphosphate quotients in 15 people taking standard doses of TDF and LPV/RTV and in 14 taking TDF without LPV/RTV. No one was taking ddI.
The study showed little change in intracellular TDF levels one, two, and four hours after dosing, a result reflecting the drug's long intracellular half-life. Averaging intracellular concentrations from the three readings in each person, Benech recorded mean (and median) levels of 181.4 (157.0) fmol/106 cells without LPV and 280.3 (199.2) fmol/106 with LPV, a 56% difference that fell short of statistical significance (P = 0.144). Intrapatient variability was high -- 286.3% with and 44.1% without LPV.
Because no one has defined a therapeutic index for TDF, the clinical consequences of this TDF increase with LPV/RTV remain uncertain. Benech called for further study of the interaction in more people monitored over time.
An intricate study of 30 healthy men by Michael Kurowski (Therapia, Berlin) found that TDF had no substantial effect on FPV/RTV, and the PIs had little effect on TDF [abstract 10]. Half of the men started 1,400/100 mg of FPV/RTV once daily with or without 300 mg of TDF. On day 14 they had 24-hour sampling for drug levels. Then the men taking TDF stopped the drug, and the men not taking TDF started it for another 14 days. The other half of the study group followed the same crossover pattern with 1,400/200 mg of FPV/RTV once daily.
Tenofovir had practically no impact on APV levels with either dose. With the 1,400/100-mg dose, APV's trough was 24% higher with TDF than without it, a nonsignificant difference. With the 1,400/200-mg dose, APV's peak was 11% higher with TDF than without it.
Tenofovir monophosphate levels differed little in men taking 1,400/100 mg of FPV/RTV versus those taking 1,400/200 mg. Tenofovir's AUC and maximum concentration were slightly lower with 1,400/200 mg, and the trough was slightly higher with the 1,400/200-mg dose.
Three people dropped out of the study, two with skin reactions and one with gastrointestinal intolerance. Kurowski concluded that no dose modifications are needed when combining FPV/RTV with TDF.
Gilles Peytavin (Bichat-Claude Bernard Hospital, Paris) looked for potential interactions between twice-daily 700/100-mg FPV/RTV and once-daily TDF in 21 people with HIV infection [abstract 32, poster 2.25]. All of these treatment-experienced people had taken boosted FPV with TDF and another NRTI for at least one month. None were taking an NNRTI or a second PI. Peytavin measured APV, RTV, and TDF levels once in each of these 14 men and seven women and genotyped virus for resistance mutations.
The study group had a median viral load of 3.79 log (range 2.30 to 6.12 log); four people (19%) had a viral load below 200 copies/mL. The median CD4 count was 174 cells/mm3 (range 2 to 497 cells/mm3). Genotyping spotted a median of two major protease mutations (range zero to four).
After one to two months of FPV/RTV, the median viral load measured 2.37 log and nine people had fewer than 200 copies/mL. Seven others had at least a 1-log drop in viral load. The median CD4 count climbed to 224 cells/mm3.
Only three people had a minimum APV concentration below 1,250 ng/mL -- the cutoff for a 10-fold drop in viral load at 12 weeks in the Genophar pharmacologic substudy of TDF plus 600/100 mg of APV/RTV twice daily.35 No one had an unmeasurable APV level. The median APV trough, 1,586 ng/mL, compared well with the median trough among 98 people in the Genophar study. Interpatient variability in APV troughs was about 37%.
The median 24-hour TDF plasma concentration measured 48 ng/mL, close to the 44 ng/mL in a study of TDF with FPV/RTV. Interpatient variability in TDF levels was about 55%.
With this twice-daily dose of FPV/RTV, Peytavin confirmed Kurowski's once-daily finding of no clinically meaningful interaction between TDF and the boosted PI.
Enfuvirtide (ENF) remains the only antiretroviral that keeps HIV out of T cells, but others are racing toward approval. CCR5 antagonists have reached crucial clinical trial stages, and one has even forsaken its faceless number for a name. This year's workshop had news on ENF, CCR5 candidates, and two new drugs from a more familiar class, the NNRTIs.
Despite its niche as the only antiretroviral that frustrates fusion, ENF has its drawbacks. Besides its high price, the twice-daily needles make it intolerable to some people and painful to most. Nothing seems likely to solve the first problem, but a transdermal gas pressure-powered injector could solve the second. Marianne Harris (British Columbia Centre for Excellence in HIV/AIDS, Vancouver) offered details of a 23-person trial in which people taking ENF by needle switched to the Biojector system [abstract 48].
Nurses find it easier to teach people to use Biojector than to use needles, according to Harris. And because people can wield the device with one hand, they can reach more body sites to inject. The transdermal injector causes less pain and less tissue trauma than the ENF needle.
The Vancouver team measured ENF levels 11 to 13 hours after dosing before people switched to Biojector and during a median four weeks of treatment with the new device. Enfuvirtide troughs and peaks proved equivalent with the skin-skimming injector and the needle, while a 0-to-31 injection site reaction score dropped after the switch to Biojector (Table 15).
No one who started using Biojector with a viral load below 50 copies/mL had a rebound, and viral loads dropped in people with detectable loads when they switched to the no-needle jet.
Charles Flexner (Johns Hopkins University, Baltimore) observed that ENF troughs ranged lower with Biojector than with the needle, and that will have to be watched in bigger trials. In the United States, others observed, the FDA would probably not approve such a device for ENF without first testing it in healthy volunteers. Biojector is not the only gas pressure-powered injector currently being tested.
A small study of people starting ENF as part of a salvage regimen linked the drug's week-2 AUC with viral load drops at week 4 [abstract 71, poster 5.10]. This study by Stefano Bonora (University of Turin, Italy) also yielded an ENF trough cutoff that predicts the week-12 response.
The 38 study participants all started ENF with one or two boosted PIs and two or three NRTIs. Seven (18%) also started an NNRTI. This group had advanced HIV infection, with a median viral load of 5.16 log (IQR 4.7 to 5.5 log) and a median CD4 count of 49 cells/mm3 (IQR 19 to 109 cells/mm3).
Median viral load dropped 0.58 log at week 4 and 0.41 log at week 12. Enfuvirtide's week-2 AUC, peak, and trough (+ standard deviation) measured 45,633 + 16,239 ng·h/mL, 5,041 + 1,657 ng/mL, and 2,318 + 1,245 ng/mL. Linear regression analysis isolated three factors that predicted viral load dips at week 4:
Only the number of active drugs in the regimen predicted viral load drops at week 12. At that point 15% had a viral load below 50 copies/mL. Logistic regression analysis again picked out number of active drugs as a predictor of a week-12 sub-50 load (P = 0.04). The sub-50 predictive power of ENF trough at week 12 fell short of statistical significance (P = 0.07).
Bonora determined that an ENF trough above 2,200 ng/mL predicted a 12-week load below 50 copies/mL with a sensitivity of 100% and a specificity of 62.5%.
GlaxoSmithKline's CCR5 frustrater, labeled 873140, inhibits CYP3A4, the isoenzyme famous for metabolizing PIs. The CCR5 drug is also a weak inhibitor of CYP2C9 and a modest inhibitor of CYP2C19.
To gauge potential CYP inhibition more closely, Ivy Song (GlaxoSmithKline) recruited 15 healthy volunteers to quaff a cocktail of caffeine, warfarin, omeprazole, dextromethorphan, and midazolam -- probes used to reckon interactions with (respectively) CYP1A2, CYP2C9, CYP2C19, CYP2D6, and CYP3A4 [abstract 75]. Later, everyone took 40 mg of 873140 every 12 hours for seven days, then swallowed the probe brew again.
Figuring drug to no-drug ratios, Song saw the greatest potential interaction with CYP3A4, which she labeled "weak inhibition" (Table 16).
Although Song maintained that the CYP3A4 inhibition she measured would have little clinical relevance, some pharmacologists in the audience were not so sure. Especially if GlaxoSmithKline ends up seeking approval for a higher dose of their CCR5 drug, they worried that Song's numbers may augur higher PI levels in people taking 873140.
Pfizer's CCR5 antagonist -- formerly tagged UK-427,857 but newly christened maraviroc (muh-RAV-eh-rok) -- neither inhibits nor induces CYP3A4. But it is a CYP3A4 substrate, so drugs that inhibit CYP3A4 (PIs, ketoconazole) could raise maraviroc levels and drugs that induce CYP3A4 (EFV, rifampin) could lower them.
That's what happened in a batch of studies reviewed by Pfizer's Gary Muirhead [abstract 76]. Ketoconazole, SQV, ATV, and ATV/RTV inflated maraviroc's minimum concentration two to three times and its AUC four to five times. Efavirenz lowered maraviroc's maximum concentration and AUC by about 50% and rifampin lowered it by 70%. Efavirenz also cut the CCR5 antagonist's exposure when given with PIs, but CYP3A4 inhibition by PIs yielded a net gain in maraviroc exposure.
GlaxoSmithKline believes maraviroc's 100-mg twice-daily dose will have to be halved when the drug teams up with PIs. When given with EFV (without PIs), maraviroc's dose will probably need doubling.
In a separate study using a single 300-mg dose of maraviroc, Muirhead confirmed the effects of EFV and LPV/RTV on maraviroc exposure [abstract 31, poster 2.19]. Nevirapine boosted maraviroc's peak slightly but had no effect on its AUC.
Tibotec has two candidates in the running as "next-generation" NNRTIs, TMC125 and TMC278. Both neatly corral wild-type virus and pen up certain NNRTI-resistant strains. Although TMC125 leads TMC278 in development, TMC278 has the advantage of once-daily dosing and looks potent at a range of doses studied so far.36 But work on both drugs continues.
One problem with the elder candidate, TMC125, is poor oral bioavailability with the tested formulation. Trying to squeeze more drug into fewer pills, Tibotec's Monika Schöller compared four new formulations with the current tablet in 45 healthy volunteers [abstract 82, poster 6.8]. Everyone had TMC125 levels measured over 96 hours after taking a single 400-mg dose of the current formulation. Then Schöller randomized them to take one of four new formulations and measured drug levels again.
All new formulations had significantly greater oral bioavailability than the current pill. Average AUCs for the four new amalgams ranged from 492% to 937% higher, and average peaks from 418% to 847% higher. The current and new formulations all had similar times to maximum concentration and elimination half-lives.
Schöller concluded that the new TMC125 tablets will allow a "substantial reduction" in pill burden, but she didn't suggest anticipated pill counts.
A study of TMC278 found that taking the drug with food significantly improves exposure [abstract 80, poster 6.9]. Tibotec's Richard Hoetelmans gave 100 mg of TMC278 to 12 healthy volunteers with or without food. The food in question was a breakfast packed with 25 to 35 g of fat and 600 to 800 kcal. Fasting blood levels were drawn at least 10 hours after the last meal. With food, TMC278 peak levels were approximately 71% higher than after fasting (P = 0.0067) and TMC278 AUC about 45% higher (P = 0.0016).
In a separate study four groups of healthy volunteers took 25, 50, 100, or 150 mg of TMC278 daily for 14 days [abstract 80, poster 6.9]. This time Hoetelmans measured drug levels over 24 hours after the single day-1 dose and over 216 hours after 14 days of dosing.
The results showed that time to maximum concentration and elimination half-life remained independent of TMC278 dose. But 0-hour concentration, trough, peak, and 24-hour AUC rose in proportion to the dose. Volunteers reached steady-state concentrations of TMC278 within one week of starting once-daily dosing.
A large randomized trial of TDM failed to prove that this strategy improves treatment outcome or shields people from antiretroviral side effects. Yet a separate cross-sectional study determined that nearly one quarter of PI and NNRTI levels in a clinic cohort lay outside the therapeutic range. And a database analysis challenged conventional wisdom in suggesting that twice-daily therapy has a distinct advantage over once-daily dosing.
Measuring highly variable antiretroviral levels in highly variable individuals seems to make eminent sense. But no one has nailed down the virologic or safety value of TDM.
Reviewing results of POPIN -- a 122-person randomized TDM trial -- Saye Khoo (University of Liverpool) maintained that previous TDM trials have proved "persuasive rather than conclusive" [abstract 59]. And studies of adherence support -- another element of POPIN -- so far fail to show sustained benefit. Although national treatment guidelines advocate TDM for certain people with HIV, Khoo noted, none supports the strategy for routine use in unselected individuals.
POPIN randomized 45 people taking PIs and 89 taking NNRTIs to standard of care monitoring or TDM plus adherence support. All study participants had pushed their viral load below 50 copies/mL for more than six months or were starting or switching to a new PI or NNRTI regimen.
TDM consisted of trough measures and two-hour post-dose sampling at baseline and every 12 weeks in people with a sub-50-copy load and at weeks 2, 4, 12, then every 12 weeks in people starting or switching to a new regimen. Nurses used a structured questionnaire to encourage adherence. Everyone also kept a TDM diary and got food advice.
POPIN had three primary endpoints:
The analysis included 85 people in the sub-50 group and 37 in the start/switch group. Khoo and colleagues randomized 63 to TDM and 59 to standard of care. The researchers recorded a marginally higher CD4 count in the standard-of-care group (430 cells/mm3) than in the TDM group (280 cells/mm3) (P = 0.06). Median follow-up measured 72 weeks and ranged from six to 172 weeks.
At the end of follow-up only 19 people met one of the failure criteria, with no difference between the TDM and standard-of-care groups in either a switch-equals-failure analysis or a switch-censored analysis. Neither did the groups differ in toxicity rates, CD4 count change, or proportions reaching an RNA reading below 400 copies/mL. Risk of failure at 96 weeks, based on 427 drug levels in 108 people, measured 40% in the TDM group and 32% in the standard-of-care group.
Interindividual variation in EFV, NVP, and LPV levels proved high, while intraindividual variation was somewhat lower (Table 17).
One reason researchers have a hard time proving TDM benefits may be poor adherence -- by physicians. Among 20 POPIN participants with repeatedly high or low drug levels, physicians of only seven (35%) followed dose-switch advice. The earlier ATHENA study charted 50% physician adherence among 57% of those who answered a questionnaire.37
These low adherence rates do not necessarily signal negligence, Khoo suggested. If a person has a drug level above the therapeutic range -- yet shows no hint of toxicity and has HIV under wraps -- physicians may ignore advice to cut the dose. Or if a person keeps HIV in control with consistently low drug levels, why tinker? In reviewing POPIN clinician adherence records, Khoo discovered his own name among those who declined to heed dose-tweaking advice.
Khoo concluded that TDM plus adherence advice yields no significant virologic or safety benefit. But the study lacked statistical power to exclude such a benefit. Whereas most TDM studies involve PI-treated people, most POPIN participants took an NNRTI, and that drove the results. And because many people already had well-controlled viremia, they had only a modest chance of benefiting from TDM.
A cross-sectional study of 151 people taking a PI or an NNRTI underlined two rationales for drug-level monitoring -- high variation in antiretroviral exposure from person to person, and high rates of subtherapeutic or potentially toxic drug levels [abstract 70, poster 5.11].
José Moltó (Germans Trias i Pujol University Hospital, Badalona, Spain) and colleagues collected samples from 151 adult clinic patients, half of them taking a PI and half an NNRTI. The most frequently prescribed drugs were LPV/RTV in 56, EFV in 45, and NVP in 34.
Clinic workers drew samples from people taking a twice-daily regimen 10 to 13 hours after the last dose and from people taking a once-daily regimen either (1) 21 to 25 hours after the last dose or (2) eight hours after the last dose in people taking their drugs at bedtime. The cohort was 71% male with a mean age of 41.4 years, and 41% had HCV coinfection.
Interindividual variability in PI and NNRTI concentrations approached 50%, with the widest wavering for LPV, NVP, EFV, and SQV (n =6). Defining minimum and maximum drug level targets by the HIVPharmacology.com table, Moltó figured that only 53% of NNRTI concentrations and 85% of PI concentrations lay within the therapeutic range. He scored about 30% of NVP and EFV levels toxic. Subtherapeutic concentrations proved most common with EFV (about 30%), APV (about 20%), and LPV, NFV, and ATV (all less than 20%).
Although Moltó did not warn study participants of the drug monitoring before their clinic visits, poor adherence appeared to explain few of the low drug readings. Most people reported missing no doses in the past week (82%) or missing only one dose (14%). Moltó calculated that poor adherence or drug interactions might explain 44% of subtherapeutic drug levels.
Once-daily PI dosing may be more popular with people taking these drugs. And it certainly offers a marketing angle to PI makers with once-a-day products. But twice-daily PI dosing may be superior in one very important way -- drug exposure -- according to a database analysis by Bernard Vrijens (Aardex Biostatistical Research Centre, Vise, Belgium) [abstract 3, poster 1.3].
The reason, Vrijens figured, is that missing a single once-daily PI dose has the pharmacokinetic consequences of missing three consecutive twice-daily doses.
The analysis included 237 people taking a once-daily PI and 245 taking a PI twice daily, all of them while enrolled in a clinical trial. Using the Electronic Medication Event Monitoring System (MEMS) to gauge adherence, Vrijens confirmed that adherence is better with a once-a-day regimen (93.5%) than with twice-a-day drugs (84.5%) (P<0.0001).
Among people taking a twice-daily PI, adherence faltered more often with the evening dose. MEMS monitoring suggested that memory dimmed early in the week: People tended to take both morning and evening doses on Sunday and Monday. But by Tuesday evening some started forgetting their nightly dram.
Vrijens then determined the time to reach a critical PI concentration threshold after missing a single once-daily dose versus a single twice-daily dose. By this reckoning he figured that missing a solitary once-a-day dose has pharmacokinetic consequences similar to those of missing three twice-daily doses in a row. During one year of treatment 75% taking once-daily PIs reached this dangerous threshold compared with 55% taking their PI twice daily.
"The long-standing promotion of [once-daily] regimens as having 'superior compliance' is based on the modest superiority of percentage of prescribed doses taken with [once-daily] versus [twice-daily] regimens," Vrijens concluded. "The factor critical for superior therapeutics, however, is internal exposure, and there the projected impact on PI concentrations of actual dosing histories in almost 500 patients indicates substantial superiority of the [twice-daily] regimen."
Mark Mascolini writes about HIV infection (email@example.com).
Editor's note: This article first appeared at www.HIVPharmacology.com and is reprinted here with permission from Virology Education, Utrecht, The Netherlands.