Heavy Lifting in Glasgow (Wherein HIV Researchers Uncrate Some Clydeside Surprises)
Sixth International Congress on Drug Therapy in HIV Infection
November 17-21, 2002, Glasgow
That last nugget surfaced in Glasgow's pharmacology session when French researcher Jacques Grassi stunned the uninitiated and the initiated as well with evidence that some AZT turns into d4T inside people's cells -- maybe enough to have clinical consequences (see the next section). His results -- already stirring debate, speculation, and more study -- surely require confirmation. But, if verified, they demonstrate again how outrageously ironic, confounding, and unpredictable HIV medicine can be.
A stock line in reports on recent HIV meetings is that they lack "breakthrough news." But it's not true. Breakthroughs happen all the time. It's just that reporters and reviewers got blinded by the incandescent dawn of potent combinations that turned HIV infection into a 30 Years' War against drug toxicity instead of a fast, messy battle against invincible opportunists. No, the breakthroughs are still there, even if they fail to announce themselves in neon. To find them, one need only track the literature, follow the meeting news, and connect the dots.
Glasgow -- coming after headline-siphoning summer and fall meetings -- can offer the ideal forum for linking some data dots and seeing whether what emerges makes sense. Grassi's little shocker is only the most obvious example. Sparks arcing from his serendipitous trouvaille may illumine still-dark corners of mitochondrial toxicity and nucleoside (NRTI) cross-resistance. For example, David Nolan, who closely studies mitochondrial mischief in Simon Mallal's Perth group,1, 2 told IAPAC Monthly that, "for me, this may be a crucial piece of the puzzle."
Yet other Glasgow reports extended the research skein with less startling -- though already less tenuous -- threads:
Can these advances be labeled "breakthroughs"? One could argue yes. For someone who has kept HIV under wraps with indinavir/ritonavir -- at the cost of sore sides and cracked lips -- a lower yet still effective dose allows a breakthrough to an easier life and fewer temptations to skip a dose. For someone toying with trying treatment breaks, a physician dispensing hard data can break through the web of hypotheses and prevent irremediable mistakes. For someone who got the adherence lecture -- once, a while ago -- new insights on making adherence routine (plus smaller, fewer, friendlier pills) can make antiretrovirals work more than a year or two.
"Breakthroughs" like these don't make headlines. Sometimes they don't even get noticed. And they all require heavy lifting -- by the people who make the pills, the people who take the pills, and the people who do the research to make the taking simultaneously easier and more effective. Everyone who walked or rode to the Glasgow meeting hall each penumbral November morn got a daily reminder of heavy lifting. A mere parking lot away from the Scottish Exhibition and Conference Center stands the hulking Clydeport crane (above), greeting the surprising prospect of a wintry sunrise, but ready for work in weather good and bad.
Because of this poor conversion -- or phosphorylation -- to its triphosphate form, and for other reasons, Perth researcher David Nolan told IAPAC Monthly, "I have always had trouble coming to grips with the mechanism by which AZT leads to [mitochondrial DNA] depletion," the proposed mechanism of mitochondrial toxicity. Work by Jacques Grassi (CEA, Gifsur-Yvette) may help Nolan get a better grip.
note 3), Grassi and colleagues at the University Hospital of Bicêtre were busy looking for d4T-negative controls. Their strategy seemed logical enough: measure triphosphates in peripheral blood mononuclear cells (PBMCs) of people taking NRTIs other than d4T. To their surprise, the assay spotted d4T-TP in PBMCs from two people taking AZT, but not d4T [abstract PL8.3*].
Running the assay in 15 people taking AZT, they could not detect d4T-TP in five of them. But in the other 10 d4T-TP concentrations ranged from 3.0 to 10.4 fmol per 1 million PBMCs. Those levels represented 4.5 percent to 17.1 percent of the AZT-TP measured. The median d4T-TP: AZT-TP ratio stood at 12.6 percent in these 10 people. The assay never found d4T-TP in cells from people not taking AZT or d4T. Do these petite d4T-TP footprints fit into the triphosphate tracks laid down by d4T itself? Not quite, but maybe enough to matter. Comparing d4T-TP levels in people taking AZT and people taking d4T in another study,4 Grassi showed that levels in AZT takers lie in the lower range of concentrations found in d4T takers.
The bigger question is whether this new assay3 can be trusted. The University of Liverpool's David Back told IAPAC Monthly that Grassi's "methodology seems to be very sound," though Back can't figure where the d4T-TP came from. He suspects it may be a breakdown product of AZT-TP rather than a result of AZT phosphorylation. Whether this d4T-TP residue has some clinical impact, Back believes, depends on the relative concentrations of the two triphosphates and on their relative affinity for reverse transcriptase. Back is collecting the data he needs to suggest an answer.
Nijmegen pharmacologist David Burger (University Medical Center St. Radboud) proposed several grounds for caution in interpreting Grassi's results. First, he wrote to IAPAC Monthly, "the technique of measuring triphosphates is extremely difficult. Therefore, such techniques are highly prone to errors in outcomes." Second, Grassi's findings come from only 10 people, so no one can say now whether the results apply to everyone taking AZT. Third, there is no external quality control for Grassi's assay. Without such a control, the results are a matter of "believe-it-or-not."
But if Grassi is right, his findings could throw new light on two nagging issues surrounding AZT and d4T -- mitochondrial toxicity and cross-resistance. David Nolan explained to IAPAC Monthly that his difficulty in "coming to grips" with how AZT depletes mitochondrial DNA (mtDNA) rests on two concerns: "All of the literature, based on in vitro data, suggests that (1) AZT is a poor inhibitor of polymerase-gamma [which mediates synthesis and repair of mtDNA] and (2) not much free AZT gets converted to active AZT-TP (about 7 percent), because it gets 'stuck' at the AZT-monophosphate stage" (see reference 2). Because d4T has high affinity for polymerase-gamma, the production of d4T-TP in cells exposed to AZT "may provide a very neat explanation for the mtDNA-depleting effects of AZT."
What could be the clinical fallout of these mitochondrial machinations? It's easy to imagine at least one. Three studies show that replacing d4T with AZT slowly reverses peripheral fat atrophy,5-7 a purported mitochondrial toxicity. Could concentrations of d4T-TP in people switched to AZT be too low to worsen atrophy, but just enough to slow its reversal to a crawl? "This must be a possibility," David Back allows. Whether traces of d4T-TP in blood cells of people taking AZT contribute to the demonstrated cross-resistance between these thymidine analogs remains even more speculative, but people are thinking about it.
One 10-minute slide talk cannot, in a stroke, solve long-standing riddles of resistance and toxicity. But it can at least suggest that some riddles are soluble. And while Grassi's fresh clues inspire keen researchers like Back and Nolan to approach the sphinx anew, no one should forget that early clues can prove treacherous if interpreted too glibly. A famous example of over-hasty conclusions, David Burger reminded IAPAC Monthly, involved -- yes -- d4T-TP. One study suggesting slower d4T metabolism in people with AZT experience8 launched a heated campaign to prescribe d4T first. That campaign cooled down fast, though, when a second study9 failed to confirm the first.
Jürgen Rockstroh (University of Bonn, Germany) outlined his approach to indinavir/ritonavir dosing in a satellite symposium and backed it up with a randomized, crossover study of 600/100 versus 400/100 mg twice daily in 16 healthy volunteers [abstract P182]. Except in people coinfected with hepatitis B or C virus, he typically begins with an 800/100-mg dose and doesn't bother with drug level monitoring. But if indinavir toxicity starts taking a toll, Rockstroh trims the dose to 600/100 or 400/100 mg, now checking indinavir levels to ensure adequate troughs.
The crossover study enrolled nine men and seven women who took 600/100 mg or 400/100 mg twice a day for two weeks before 12-hour sampling for drug levels. Then they switched to the other dose for two weeks. Geometric mean minimum concentrations, maximum concentrations, and total exposure proved significantly lower during 400-mg dosing (Table 1). While taking 600 mg of indinavir, no one had a trough below the recommended threshold of 0.10 mg/L. But three people saw their troughs slip below that mark while taking the 400-mg dose, a finding that underscores the need for therapeutic drug monitoring with this dose.
A 24-week study of 16 men and 14 women beginning antiretroviral therapy with 400/100 mg of indinavir/ritonavir twice daily charted good RNA and CD4 responses [abstract P18]. But Claudine Duvivier and colleagues at two Paris hospitals found that indinavir lite did not always go down easy and that indinavir troughs dropped as follow-up continued.
The cohort included 20 Caucasians, 17 people from northern Africa, and three from sub-Saharan Africa. They had a worrisome median CD4 count of 84 cells/mm3 (range 3 to 558 cells/mm3) and a lofty median viral load of 230,975 copies/mL (range 5,000 to 750,000 copies/mL). Six people (15 percent) had to quit treatment before week 24 because of side effects, and two others (5 percent) stopped keeping appointments. Grade 3 or 4 side effects included GI problems and diabetes (in two each) and xerosis, arthralgias, "renal toxicity," and hemolysis in one each.
While the median CD4 count climbed from 84 to 188 cells/mm3 through 24 weeks, the median viral load fell from 5.36 to 2.28 logs (about 229,000 to 190 copies/mL). By intent-to-treat analysis, 27 of 40 people (67.5 percent) had a viral load below 400 copies/mL at week 24. The 24-week on-treatment sub-400 rate measured 87 percent. The only resistance mutation that arose when treatment failed was the lamivudine (3TC)-linked M184V change.
Again this study showed that low-dose indinavir/ritonavir requires routine drug level checks, at least in the early months. Whereas 31 of 34 people (91 percent) assessed at week 4 had an indinavir trough above a threshold of 150 ng/mL, by week 24 only 13 of 19 (68 percent) still had an adequate trough.
Clinicians from Saint-Pierre University Hospital in Brussels offered the biggest study of low-dose indinavir/ritonavir in a nonrandomized, prospective analysis of 181 people starting twice-daily antiretrovirals [abstract P16]. D. Konopnicki and colleagues compared 71 people starting nelfinavir, 56 starting 800/100 mg of indinavir/ritonavir, and 54 starting the PIs at 400/100 mg. The three groups matched well in median age (34, 34, and 33.5 years), viral load (4.84, 5.02, and 5.02 logs), and CD4 count (193, 207, and 250 cells/mm3).
Intent-to-treat and on-treatment analyses discerned no significant differences between the groups in 48-week CD4 gains or proportions with a viral load under 50 copies/mL. But when it came to tolerability, the 800/100-mg group significantly lagged those taking nelfinavir or 400/100 mg of the PIs. After 48 weeks 61 percent had stopped 800/100 mg of indinavir/ritonavir because of side effects compared with 15 percent in the 400/100-mg group and 6 percent in the nelfinavir group (P<0.001). The probability of treatment interruption proved significantly higher in the high-dose double PI group than in the other groups at 12, 24, and 48 weeks (P<0.0001). While 37 people (66 percent) who started 800/100 mg of indinavir/ritonavir had at least one treatment-related side effect and 12 (21 percent) had two or more, 13 (24 percent) starting 400/100 mg had at least one side effect (P<0.0001) and four (7 percent) had two or more (P=0.06). Among 21 people who switched from 800/100 mg to 400/100 mg because of toxicity, side effects resolved or improved in 17 (81 percent).
Week 4 drug level monitoring found dangerously low indinavir troughs (<150 ng/mL) in only three of 30 (10 percent) taking 400/100 mg of indinavir/ritonavir, including one trough below 80 ng/mL. Among seven people taking 800/100 mg who had drug levels checked at week 4, two (28 percent) had a trough under 80 ng/mL. But everyone with low indinavir troughs had a viral load under 50 copies/mL. It would be interesting to see if more 24-week troughs sagged in this cohort, as they did in the French study. The higher proportion of low indinavir levels with 800/100 mg may reflect toxicity-driven lapses in adherence or the small sample.
Measuring drug levels between two and 10 weeks after starting the regimens, Harris charted a median indinavir trough around 400 ng/mL, but the range stretched from 0 to 1,600 ng/mL. Historically, she noted, indinavir troughs average about 900 ng/mL with an 800/100-mg dose. Indinavir peak levels (median 3,200 ng/mL) also stood substantially lower than historical peaks with 800/100 mg (about 7,200 ng/mL). Lopinavir peaks and troughs approximated those recorded when given without other PIs or nonnucleosides.
Among 12 men with presalvage resistance tests, six had protease mutations that translated into a 19-fold change in indinavir 50 percent inhibitory concentration (IC 50) compared with wild-type virus, a 37-fold lopinavir change, and a 22-fold saquinavir change. One of these six had nearly a 4-log climb in viral load after two weeks of salvage, but the other five had a median 4.4-log drop after a median 4 weeks. Among six people without protease mutations at baseline, viral loads dropped a median 4.8 logs after 4.5 weeks.
Harris reported that the salvage regimens were "well tolerated" by everyone, but she did not offer specifics.
And what nuke-sparing regimens merit study? Cooper had these suggestions:
Wait a minute. A PI-only regimen? Didn't that tactic come up short in the sobering induction-maintenance trials of yesteryear? Yes, but Pietro Vernazza (Cantonal Hospital, St. Gallen, Switzerland) is trying it again and -- in a highly selected group -- early results are good. Cooper outlined 20-week trends in Vernazza's study of people given 400 to 800 mg of indinavir twice daily plus 100 mg of ritonavir. Everybody had a viral load below 50 copies/mL for at least three months, no one had a virologic failure on their chart, and (to ensure a rescue option) no one had nonnucleoside experience. Vernazza picked indinavir because it penetrates the central nervous system, and he adjusted the dose with drug monitoring. Through week 20 seven of eight people maintained a viral load under 50 copies/mL; the eighth had one blip above 50 copies.
Elke Lauenroth-Mai and colleagues in a Berlin practice took a more complex nonnucleoside route in 18 people -- all with PI experience and a median antiretroviral history of 5.9 years [abstract P235]. The Berlin clinicians arched their rescue regimen around a keystone comprising standard-dose lopinavir/ritonavir and 666 mg of indinavir twice daily, reasoning that lopinavir can subdue virus already burdened by multiple protease mutations and that indinavir penetrates sanctuary sites.
Four people without nonnucleoside experience started efavirenz or nevirapine, and eight started the nucleotide analog tenofovir. None starting tenofovir had more than three TAMs (including M41L and L210W) or the multinucleoside-resistant mutation Q151M or T69SS. They began salvage with a median CD4 count of 235 cells/mm3 (range 12 to 543 cells/mm3) and a median viral load of 135,000 copies/mL.
All 18 people completed 24 weeks of treatment with the new regimen, and nine had made it to week 48. The Berlin group quelled GI toxicity in three people by cutting the lopinavir/ritonavir dose to 266/66 mg twice daily. They trimmed the indinavir dose to 400 mg twice daily in two people because of side effects. No one in the study had indinavir-related renal complaints.
By week 24 the median CD4 count added 175 cells/mm3 (range 117 to 198 cells/mm3). Among those with 48 weeks of follow-up, the median CD4 gain measured 197 cells/mm3. Nine of the 18 had a viral load under 50 copies/mL by week 24, and six of nine had a sub-50 load at week 48.
What separated week 24 responders from nonresponders? Baseline protease mutations proved the clearest discriminator. Among the seven nonresponders who had baseline genotypes, all had more than five PI-puncturing mutations, including L10I, D30N, M46L, I54V, L63P, V82A, and L90M. Among six people who started salvage with lopinavir experience, five fell into the nonresponse group. But among seven with baseline indinavir experience, only three failed to notch a sub-50-copy load at week 24.
Tenofovir added little to the antiviral activity of these regimens. Of the nine people who used the nucleotide, only one had a 24-week viral load under 50 copies/mL. Not surprisingly, starting a nonnucleoside did bolster the regimen. Three of the four people starting their first nonnucleoside were 24-week responders.
Median fasting triglycerides rose 110 mg/dL through week 24, and two of 18 people had a triglyceride level above 500 mg/dL. Total cholesterol rose by a median 67 mg/dL at week 24; protective high-density lipoprotein (HDL) cholesterol accounted for 11.5 mg/dL of that gain. Meanwhile, venous lactates fell by a median 0.41 mmol/L through week 24. Whereas nine people had lactates above 2.2 mmol/L at baseline, only three did at week 24.
Yet mitochondrial-friendly nukeless regimens do not always sit on the stomach as kindly as chamomile tea. Daniela Gey (University of Heidelberg) and colleagues found that nine of 18 people who tested the no-nucleoside waters came away scalded [abstract P122]. All 18 suffered from a mitochondrial toxicity, lipoatrophy, or both when they started saquinavir/ritonavir (1,000/100 mg twice daily) plus standard-dose efavirenz and nevirapine. Their median 3.93-log viral load (about 8,500 copies/mL) stayed stable through 24 weeks, as did their CD4 counts. Median lactates dipped from 1.8 to 1.5 mg/dL.
But the full-dose double NNRTIs probably proved this regimen's undoing. Six people quit treatment because of rash, while nausea or diarrhea forced two more to stop. Another person dropped out because of the onerous 19-pill regimen. Gey proposed that starting with 600 mg of efavirenz daily for two weeks may improve this regimen's tolerability. But whether double nonnukes ever make sense remains an open question. Gey did not report the treatment experience in these people.
A study of 198 people starting a rescue regimen after AZT or d4T experience bolstered the rationale for more study of NRTI-sparing tactics in people with plentiful thymidine analog mutations [abstract P228]. Mauro Zaccarelli (National Institute for Infectious Diseases "Lazzaro Spallanzani," Rome) tracked people who started new regimens based on genotyping and expert advice in 1999 and 2000. All new combinations included two or three NRTIs.
After a median 60 weeks of follow-up, only 111 (56 percent) had one or more undetectable viral load readings. Beginning the rescue regimen with two TAMs lowered the chance of having an undetectable viral load by 7 percent, and starting with three TAMs lowered the chance by 24 percent. That downtrend became statistically significant only with four or more TAMs, which lowered the chance of a virologic response 65 percent (P=0.01).
10, 11 does that faster action pay clinical dividends? Some suggest that the extra antiviral pop from the fourth drug makes most sense for people with ultralow CD4 counts who need a quick turnaround. And a study by Ed Wilkins (North Manchester General Hospital) and colleagues in London showed good 24-week results with a simple four-drug combo of efavirenz plus Trizivir (AZT, 3TC, and abacavir) [abstract P13].
The treatment-naive participants in the TETRA study all started with CD4 counts under 200 cells/mm3, and 82 percent had fewer than 100 cells/mm3. The average baseline count measured 66 cells/mm3, and the average viral load 554,548 copies/mL. In an intent-to-treat analysis that included people who switched to nevirapine because of efavirenz side effects but excluded people who withdrew, 53 percent had a 24-week viral load below 50 copies/mL. In a 24-week as-treated analysis, 72 percent were under the 50-copy mark. As Wilkins observed, these proportions are likely to climb as the 48-week study continues. The mean CD4 count rose to 211 cells/mm3 by week 24.
But would these people do as well on three drugs, efavirenz plus Combivir (AZT/3TC), for example? Clinicians as different as Diane Havlir (University of California, San Francisco) and Brian Gazzard (Chelsea and Westminster Hospital, London) think they may. At the 2002 ICAAC Havlir argued that effective therapy depends not so much on the number of drugs in the regimen as on how potent those drugs are.12 And because four drugs may kick up more toxicity than three, she advised sticking with stalwart triple regimens for now (again counting a boosted PI as one drug).
Gazzard, a coinvestigator in the Wilkins four-drug study, sounded similar cautions in a Glasgow satellite symposium. So far he sees "no evidence at all that we should be using more [than three] drugs." A Chelsea and Westminster study of Trizivir plus tenofovir left him "impressed by the extra difficulty of taking just one extra -- even though 'tolerable' -- drug."
At least one study bolsters these opinions. ACTG 384 compared two four-drug medleys (efavirenz/nelfinavir plus AZT/3TC or ddI/d4T) with four three-drug combos (efavirenz or nelfinavir plus one or the other NRTI pair) in 980 treatment-naive people.13 After a median 28 months of follow-up, any of the quad therapies halved the time to first regimen failure or first virologic failure when compared with nelfinavir/AZT/3TC, nelfinavir/ddI/d4T, or efavirenz/ddI/d4T. But efavirenz/AZT/3TC was as good as either quad, a result supporting Havlir's point that potency matters more than mere numbers.
14, 15 It seems to be coming true. People no longer die of HIV infection, said Bernard Hirschel (University of Geneva) in a Glasgow keynote talk [presentation KL3], at least not if they've cleared three iffy hurdles:
Of course that second contingency leaves out most people in Africa, India, China, Russia, and other spots where the epidemic festers. But from the West there are lots of numbers to back Hirschel's contention. Three Glasgow studies and one anecdotal report documented consistent, balmy trends:
Why is this happening? Two obvious reasons: the drugs are better now than in 1996, and clinicians are using them better. Hirschel showed data on regimen preferences across Europe from the first half of 1998 through the first half of 2002. In the first period about half of those studied took HAART version 1.0 -- two nucleosides and a single PI. Another 20 percent took only two nucleosides, and mere slivers on the histogram reflected nascent use of two nucleosides with a nonnucleoside or with two PIs. A wafer-thin slice augured the dawn of ritonavir boosting.
By the first half of 2000, ritonavir boosting had grown from a wafer to a wedge, while double PI use stayed flat. Double nucleoside therapy shriveled to 5 percent, and two NRTIs plus a PI shrank to 35 percent. In 2000 more than 20 percent of combos relied on a nonnuke. Those trends continued through the first half of 2002, with nonnucleoside regimens accounting for more than 30 percent and single-PI regimens for less than 30 percent. Triple nukes had all but knocked double nukes off the bar chart, and ritonavir-boosted regimens -- reflecting the advent of lopinavir -- made up about 15 percent of antiretroviral mixes. Throughout the whole period of study, about 20 percent of people took no antiretrovirals.
To Hirschel, these changes say clinicians are tracking the latest studies and -- as a result -- prescribing more efavirenz and lopinavir. In Switzerland, he added, a study done with the insurance industry confirmed that people with HIV infection now have the same life expectancy as those without it.
Today, Margaret Johnson suggested in a satellite symposium, the tables have turned. People in her clinic at London's Royal Free Centre for HIV Medicine are doing better than cosseted clinical trial enrollees. She reckons that 87 percent of people she sees have a viral load under 400 copies/mL, and 82 percent are under 50 copies. "I think we're all doing better than we expected based on clinical trials," she said.
Of course Johnson's sterling success rate did not emerge from a switch-equals-failure analysis. She counted anyone with fewer than 400 copies as a success, regardless of how many times she tempered or tweaked the regimen. But you don't find many mopey "noncompleters" with undetectable viral loads. The yearly virologic failure rate in her cohort stands below 4 percent, and case records tie most of those failures to missed pills.
Work by Johnson's colleague Andrew Phillips shows that the first 24 weeks of treatment are critical to long-term suppression. This study, published just before Glasgow,16 involved 1,433 people in London and Frankfurt beginning their first HAART regimen and reaching a viral load below 400 copies/mL by week 24; 409 of them had already tried one or two nucleosides. The longer a person's RNA stayed out of sight, the lower was that person's chance of a later rebound (Table 2). And that held true regardless of pre-HAART nucleoside experience, although people who took no NRTIs before HAART had consistently lower failure rates. Earlier, Phillips recorded the same fading rebound rate in naive Frankfurt patients starting HAART.17
Among those ever treated in this group, 80 percent have a viral load under 50 copies/mL, another 8 percent have between 50 and 499 copies/mL, and 5 percent have more than 5,000 copies/mL. CD4 counts top 200 cells/mm3 in 91 percent. In a plenary lecture Anders Sönnerborg noted that only 3.5 percent of this cohort got resistance testing in 2002 [presentation PL5.2]. That's not because Sönnerborg has given up on resistance testing; it's because regimens are failing so rarely.
Things aren't much worse over at Helsinki University Hospital, according to Matti Ristola [abstract P51]. In a clinic population of about 200 people taking three or more antiretrovirals, 71 percent had a viral load under 400 copies/mL in 1998, a rate that improved to 72 percent in 1999, 80 percent in 2000, and 87 percent in 2001. In 2000 and 2001, 72 percent and 78 percent had viral loads under 50 copies/mL. In 2001, 69 percent were on triple therapy, 26 percent on quad therapy, and 4 percent on five or six drugs.
At the Southern Alberta HIV Clinic in Calgary, reported Nikola Ostrop-Hanhoff, more and more people are sticking with their first regimen as the HAART era matures [abstract P17]. The proportion of infected people starting therapy has also dropped consistently (from 96 percent in 1996 to 56 percent in 2001), and they're starting at lower T-cell counts (293 cells/mm3 in 1996 and 180 cells/mm3 in 2001).
Sustainability, defined as the percentage staying on their first regimen, has climbed steadily over the years, a gain that Ostrop-Hanhoff attributed partly to efavirenz. She found that 57 percent starting a nonnucleoside (97 percent efavirenz) and 55 percent starting a PI (65 percent indinavir) kept the same regimen for 12 months. But by month 30 only 25 percent had stuck with their PI regimen versus 55 percent in the NNRTI group. In 1996 first-regimen sustainability measured 69 percent at month 6, 39 percent at month 18, and 23 percent at month 24. By 2000 those rates had improved to 67 percent, 62 percent, and 62 percent.
But the critical question about starting therapy, Andrew Phillips proposed in Glasgow, remains whether faster treatment wards off AIDS or death better than beginning later [presentation PL1.1]. To address that question he weighed three variables -- age, CD4 count, and viral load -- on an unusual endpoint: risk of AIDS or death after six months. He argued that these short-term termini make more sense than AIDS or death at two or three years, because most people diagnosed with HIV infection get a checkup every three to six months. And clinicians are more concerned about what may happen before they next see an untreated person than about what may happen three years hence.
The CASCADE cohort that Phillips analyzed included more than 3,000 people with known CD4 and RNA tallies before starting antiretrovirals. For a 25-year-old with a viral load of 10,000 copies/mL and a CD4 count of 200 cells/mm3, the six-month risks of AIDS and death measured only 2.3 percent and 0.3 percent. For 45-year-olds with the same numbers, respective risks measured 3.6 percent and 0.6 percent. Those risks approximately doubled for people in both age groups with a viral load of 100,000 copies/mL.
Six-month risks of AIDS or death were lower, but not zero, at CD4 counts of 350 cells/mm3, for both age groups and both viral load brackets. For example, 45-year-olds with a six-figure viral load and 350 cells/mm3 had a 2.9 percent risk of AIDS in six months and a 0.6 percent risk of death.
Are such risk levels, even though small, high enough to justify the risks of antiretroviral therapy? "If the dominant concern is to avoid AIDS and death over the next few years," Phillips counseled, "the better strategy is to start therapy immediately." But the risk of AIDS is low and the risk of death lower still, even with deferred therapy. On top of that, if an untreated person has 350 CD4 cells/mm3 today, that person probably won't approach 200 cells/mm3 until 2007. By then antiretrovirals will be even more potent and tolerable, and clinicians will know more about monitoring patients and promoting adherence. So a person who defers therapy today -- and survives until 2007 -- may stand a better chance of long-term health than the person who starts antiretrovirals now.
Another way to frame an answer, Phillips offered, is to look at risk cutoffs for other diseases. For example, 1998 British guidelines call for antilipid medications in people with at least a 1.8 percent six-month risk of coronary heart disease. If HIV clinicians applied that benchmark to AIDS, they would start antiretrovirals for both 25-year-olds and 45-year-olds with 350 cells/mm3 and viral loads of 100,000 copies/mL -- though not for either age group with 350 cells/mm3 and a viral load of 10,000 copies/mL. But such analogies are loose, Phillips cautioned, because today's antiretrovirals are more toxic than lipid lowerers.
Multicenter AIDS Cohort Study (MACS) researchers published a similar exercise shortly after Glasgow, looking at one-, two-, and three-year risks of AIDS or sub-200 CD4 counts in 1,504 untreated gay men who enrolled in the cohort from 1985 to 1988.18 Among 231 men who came into MACS with 200 to 350 cells/mm3, none with fewer than 10,000 RNA copies/mL had an AIDS-defining disease or a CD4 count under 200 cells/mm3 within one year. For those in this CD4 bracket who started follow-up with 10,001 to 20,000 copies/mL, none had clinical AIDS and 8 percent had a sub-200 CD4 count within a year. But people with viral loads under 20,000 copies/mL made up only 29 percent of the 200- to 350-CD4 contingent.
For 1,273 MACS members who enrolled with more than 350 cells/mm3, the critical RNA cutoff was 60,000 copies/mL. No one with more than 350 cells/mm3 and a viral load under 60,000 copies/mL (79 percent of the group) had an AIDS diagnosis in one year, although 11 percent in the 40,000 to 50,000 RNA bracket and 8 percent in the 50,000 to 60,000 bracket had fewer than 200 CD4 cells/mm3 in one year.
The MACS team concludes that their findings support current U.S. guidelines on starting antiretrovirals, and they propose a decision tree for starting treatment in people without AIDS:
An online review19 spells out details of the Case Definition Study and preliminary results from FRAM. Both studies see lipoatrophy as a hallmark fat failure in people with HIV infection. According to the Case Definition, central fat buildups distinguish people with lipodystrophy from HIV-infected people without the syndrome. But FRAM found that neither central fat gain nor buffalo hump separated men with HIV from age-matched seronegative controls. (FRAM data on women remain under wraps.)
Why the difference? The two studies' divergent goals and methods explain much, Powderly proposed. Although both made cross-sectional comparisons, the comparison groups and baseline assumptions differed (Table 3). Both studies, Powderly noted, have their limits. Because both are cross-sectional, neither can gauge progression of fat abnormalities. Also, the Case Definition Study does not address degrees of lipodystrophy and may underestimate or miss milder expressions of the syndrome. The Case Definition Study began with a subjective identification of cases and controls -- based on an assumption of the syndrome's main features -- so it may suffer from selection bias.
Powderly suggested that criteria used to define buffalo hump in FRAM may explain the similar rates of that fat buildup in cases and controls. He also stressed, as Carl Grunfeld has, that FRAM does not say that people with HIV don't have excess visceral fat. Instead it shows that, compared with CARDIA controls, visceral adiposity is not more common in people with HIV infection and is not statistically linked to lipoatrophy.
FRAM's most important lesson, Powderly believes, is that lipoatrophy is more common than anyone thought in people with HIV infection. But he cautioned that when you measure fat atrophy can make a big difference in a cross-sectional study. Two recent trials show that limb fat increases during the first few months of antiretroviral therapy, then begins to wane.20, 21 Measuring arm and leg fat in people who started antiretrovirals eight weeks ago will paint a plumper picture than measuring limb fat in people treated for three years.
Grunfeld's key conclusion from FRAM -- a conclusion credited by experts including Powderly -- is that lipoatrophy and lipohypertrophy are separate syndromes and should not be lumped into a single "lipodystrophy" syndrome.
Whatever the merits and demerits of these two studies, the ultimate arbiter of their worth will not be William Powderly or any other analyst. HIV docs will decide whether either offers a clinical edge, or at least a sharper sense of what lipodystrophy (or lipoatrophy and lipohypertrophy) mean. By the time of the Glasgow meeting, Carl Grunfeld had laid out results from some of the men, and none of the women, enrolled in FRAM. Even when the first complete data set appears, this rich trove of evidence will require more time for full analysis.
Clinicians won't have to wait as long to size up the Case Definition Study. Andrew Carr has already put it in publication shape, and it should appear in a refereed journal around the same time as this article. When it does, Carr and colleagues will also post several online interactive calculators (www.med.unsw.edu.au/nchecr) that will allow clinicians with varying sets of clinical, laboratory, and body scan data to see if the Case Definition makes sense for them.
Surveying 394 outpatients (including 334 men) at London's Royal Free Hospital, Collette Smith [abstract P156] found that the most consistent risk factors had nothing to do with HIV infection:
The risk picture grew scarier still in people taking HAART. Compared with those not on HAART, higher proportions of HAART-treated people had a total cholesterol above 6.3 mmol/L (23 versus 0 percent), triglycerides above 2.9 mmol/L (43 versus 4 percent), HDL cholesterol below 0.9 mmol/L (11 versus 6 percent), high blood pressure (12 versus 10 percent), and diabetes mellitus (3 versus 0 percent). But fewer HAART takers were older than 40 (40 versus 89 percent), smoked (43 versus 47 percent), or drank alcohol above "recommended limits" (6 versus 10 percent). The survey also suggested one hidden benefit of starting HAART: Although 72 percent of this cohort's smokers had tried and failed to quit, 65 percent of ex-smokers stopped when they started HAART.
What type of HAART one takes matters in MI risk calculations, according to a multicohort D:A:D*** analysis presented by Jens Lundgren (University of Copenhagen) [abstract PL9.2]. Applying the Framingham risk equation to 13,326 men and 4,278 women with HIV infection, he found a consistently low three-year MI risk among women (0.1 percent or lower) regardless of treatment experience or current therapy. But for men the three-year risk clambered from 0.4 percent in the treatment naive, to 0.7 percent among those taking only nucleosides, to 0.9 percent among those taking a nonnucleoside, to more than 1 percent among those on a PI, and to more than 1.2 percent for those taking both a PI and an NNRTI.
Which PI or PIs one takes also alter the risk equation, according to another D:A:D analysis presented by Christian Pradier of Nice [abstract PL12.1]. Ritonavir had the worst lipid scores and nelfinavir the best in this analysis of 7,729 people (79 percent men) taking licensed PIs but not nonnucleosides. The study did not pin down lipid numbers in people taking lopinavir, which got grouped with other ritonavir-boosted PIs. A multivariate analysis adjusted for age, gender, cardiovascular risk factors, CD4 count, viral load, current nucleosides, previous nonnucleosides, treatment experience when starting a PI, and year of HIV infection yielded the following findings:
The D:A:D team plans to report cardiovascular endpoints in cohort members next year.
Paolo Maggi (University of Bari) charted progression of vascular wall lesions in vessels near the aorta in a longitudinal study of 39 people starting a first-line PI [abstract P157]. Two people who switched from a PI to an NNRTI had stabilization or possible reversal of lesions. Maggi used color Doppler ultrasonography to check periaortic vessels 12 months after people started taking PIs and 12 months after the first scan (but unfortunately not before therapy). Figure 1 outlines scan results at 12 and 24 months. Maggi linked cigarette smoking, high triglycerides, and CDC disease stage with a higher risk of lesions, but he tied the highest risk to PI therapy. The poster did not spell out these statistical analyses.
For pravastatin (n=77)
For atorvastatin (n=102)
Smith recorded no serious statin side effects.
In 20 consecutive patients who started atorvastatin with an LDL above 160 mg/dL, Rosario Palacios (Hospital Virgen de la Victoria, Málaga) found that the average reading fell significantly and into the reference range, from 203.8 to 127.5 mg/dL (P<0.0001) [abstract P141]. The study involved 16 men and four women who had normal lipids before starting HAART and who did not respond to diet and exercise. Their average age was 47 years, nine had an AIDS diagnosis, 10 smoked, and five were obese. After 24 weeks of atorvastatin at 10 mg daily, total cholesterol and triglyceride averages returned to "borderline high" or normal in this group (Table 4).
For 19 men and one woman with high triglycerides that did not respond to diet or exercise, Palacios tried 200 mg of fenofibrate daily for 24 weeks [abstract P140]. This group's age averaged 40.5 years, 10 had AIDS, 10 smoked, and three were obese. Again average total cholesterol retreated to the "borderline high" range (Table 4). Triglycerides dropped significantly but remained well above normal. Neither atorvastatin nor fenofibrate affected CD4 counts or viral loads in these people.
The bottom line comes at the top of the article, in its title: "Stimulation of HIV-specific cellular immunity by structured treatment interruption fails to enhance viral control of chronic infection." Although HIV-specific CD4 cells burgeoned during the study's treatment breaks, only people with low pretreatment viral loads (below the group median of 30,976 copies/mL) maintained those gains. (Another study found that the retrovirus preferentially infects HIV-specific CD4 cells during drug breaks.23) The immune triggers recognized by cytotoxic T lymphocytes (CTLs) during and after TIs nearly mirrored those recognized before the TIs, a finding "indicating that [treatment interruptions] largely restored the pretherapy response and did not significantly alter the recognition profile within a patient." As Oxenius notes, this limp CTL retort in people with chronic infection contrasts with broadening responses in people who try TIs after treatment of primary HIV infection.24
After sifting these results and listing TI risks (acute retroviral syndrome, reseeded viral reservoirs, drops in CD4 cells, HIV transmission, resistance, poor adherence to later therapy), the University of Pittsburgh's Ume Abbas and John Mellors conclude that "risk should not be taken without reward."25
The Oxenius study takes a lot of steam out of the argument that puffed-up immune activation during TIs will abet viral control during chronic infection. Indeed, another study found that the best way to rebuild HIV-specific responses is to keep HIV under the tightest wraps possible. Comparing double PIs with single PI regimens in 49 people with acute or chronic infection, Australian researchers learned that "minimal immune activation, resulting from maximal suppression of viral replication, was required for long-term restoration and maintenance of Gag-specific T-cell responses."26 The 27 people who maintained better viral control with dual PIs had more than twice the chance of mounting HIV-specific CD4 responses than did 22 people with poorer viral control while taking one PI (P=0.0222).
If the immunologic rationale for TIs during chronic infection looks like a sand castle after high tide, drug breaks may still make sense as a way to ease or avert side effects, especially in people who began taking antiretrovirals at CD4 counts and viral loads that no longer sound the start siren in recent guidelines. At least three studies now show that TIs are safest in people who began therapy with decent T-cell tallies and enviable viral loads.27-29 Two Glasgow studies confirmed a bigger risk of bad TI results in people with higher loads or fewer T cells.
In a retrospective analysis of 56 people who took at least a two-month TI and then resumed treatment for at least 48 weeks, Nicola Gianotti (San Rafaele Scientific Institute, Milan) found that pre-TI viral load and CD4 count and CD4 cells at the time of the TI separated virologic responders from nonresponders [abstract PL6.3]. Most people, 87.5 percent, had a viral load above 400 copies/mL when they stopped therapy. Their pre-TI CD4 counts and viral loads ranged from 162 to 543 cells/mm3 and from undetectable to 5.0 logs. Using a 48-week viral load below 400 copies/mL to define a good response, Gianotti counted 19 responders (34 percent) and 37 nonresponders (66 percent).
Nonresponders had a significant CD4 drop during their TI, and 48 weeks after restarting treatment their CD4 gain lagged that of responders by 100 cells/mm3 (111 versus 11 cells/mm3, P=0.02). Gianotti could separate responders from nonresponders by their higher pre-TI CD4 nadir (226 versus 128 cells/mm3, P=0.0186), their higher CD4 count when they stopped antiretrovirals (515 versus 285 cells/mm3, P=0.0195), and their lower viral load before the TI (2.28 versus 2.93 logs, P=0.05). In a multivariate analysis, a higher peak viral load before the TI (P=0.006) and a shorter drug break (P=0.05) favored a poor post-TI virologic response. Four people, all of whom had suspended treatment with fewer than 200 cells/mm3, endured HIV-related "events" during their drug break.
In a Montreal cohort with much better virologic control when they stopped antiretrovirals, pretreatment nadir CD4 emerged as the only independent predictor of a CD4 drop below 200 cells/mm3 during the TI [abstract P25]. Nimâ Machouf (Clinique Médicale L'Actuel) prospectively tracked 83 people (75 of them gay men) who took at least a two-month TI after maintaining a sub-50-copy viral load with treatment. Their median CD4 count when they quit was 660 cells/mm3 and their nadir CD4 count 360 cells/mm3.
They lost an average 32 cells/mm3 monthly during the TIs, and the CD4 count when they started the TIs correlated highly with how many T cells they lost (r=0.394, P<0.01). Fourteen people (17 percent) saw their CD4 counts slide below 200 cells/mm3 during the TI, and only the nadir count predicted this outcome (r2=0.57). Among 28 people who resumed therapy and had at least six months of follow-up, 26 regained a sub-50 viral load (after a median of 15 weeks). The two who didn't get back under 50 copies/mL had stopped treatment again after restarting.
Compared with the Italian study, this analysis confirms the relative safety of stopping treatment during tight virologic control in people who started therapy at higher CD4 counts. Despite the four CD4 dips below 200 cells/mm3, no one suffered a new AIDS diagnosis.
A study of mutant viral populations in 30 people taking a 14-week TI confirmed the (apparent) complete return of drug-sensitive virus in a minority of such groups [abstract P207]. Jacques Izopet (Hôpital Purpan, Toulouse) and coworkers at other sites studied 22 men and eight women with a median viral load of 4.21 logs (about 16,000 copies/mL) and CD4 counts ranging from 212 to 733 cells/mm3 when they set sail on drug holidays.
In 10 people the mutant viral population did not change at all during the TI. Sixteen others had partial reversions to wild-type virus, maintaining mixed mutant populations. Only eight people (27 percent) had no mutant virus detectable in peripheral blood mononuclear cells (PBMCs) after 12 weeks without drugs. As other studies show, though, mutant species may persist at undetectable levels during prolonged drug holidays. Jonathan Schapiro (Stanford University and Tel Aviv University) suggested in Glasgow that resistance testing during a TI tests the assay's sensitivity better than it charts the waning of mutant virus. "You're not seeing disappearance of resistance mutations," he explained. "You're seeing the inability of the assay to detect [mutants] in small populations."
In the French study only people with some shift in mutant virus began a rescue regimen of efavirenz plus Trizivir (AZT, 3TC, and abacavir). In a switch-or-missing-data-equal failure analysis, nine of 17 people (53 percent) had a viral load below 400 copies/mL after 24 weeks of treatment. Among 10 people with a 24-week viral load under 50 copies/mL, seven had mutant virus in PBMCs during the TI. It's difficult, though, to rate the strategy's success in this group. Although everyone had key reverse transcriptase mutations (87 percent with T215Y and 50 percent with M184V), only 14 of the original 30 enrollees (47 percent) had a major nonnucleoside mutation. And Izopet did not report how many of the 24-week responders -- if any -- had mutations conferring resistance to efavirenz.
Two savvy HIV clinicians who addressed the first question agree on the answer: Although many people with virus highly resistant to all current classes will surely want a crack at T-20, people starting "first salvage" stand to gain more. Chelsea and Westminster's Graeme Moyle uses the term "first salvage" to denote the last combination with a good chance to work for any given person. "Encouraging earlier initiation of enfuvirtide," he counsels, "may help diminish the risk of losing current options."30 Moyle suggests colleagues may position a T-20 regimen as taking "something a bit more complex now to avoid something really tough next time."
New York clinician Howard Grossman echoes these sentiments more bluntly in an article by Mike Barr: "The best use will be when you're starting on what looks like your last good highly active regimen, not a crappy, cobbled-together salvage job."31 But because Roche expects to make only enough T-20 to treat 15,000 people in 2003, the drug may be rationed to those with the most resistant virus.
One reason T-20 looked good in two phase 3 trials32, 33 involving hundreds of people with multiresistant virus is the drug many took with T-20: lopinavir. Multiple regression analysis of the North-and-South American study showed that people starting lopinavir with T-20 carved an extra quarter-log off their viral load (P=0.0348).34 But people who had tried lopinavir before the T-20 study added 0.83 log copies to their RNA tallies (P<0.0001). The same analysis showed, not surprisingly, that people with less resistant virus and higher CD4 counts got a bigger antiviral kick from T-20.
The certain advantage of starting T-20 with other novel agents raises a third question, Bernard Hirschel observed in Glasgow: No one knows how well it will meld with drugs like the PI tipranavir, the nonnucleoside TMC 125, or integrase inhibitors. Hirschel added another unknown -- how long it will work. The phase 3 trials lasted 24 weeks.
Besides its Beverly Hills price tag, the biggest bugbear with T-20 will be taking it. Except for people who have sampled recombinant growth hormone or done penance with IL-2, few folks with HIV have had to give themselves regular injections. Can they cope? A comprehensive survey of 600 phase 3 trial enrollees suggests that most can [abstract P48]. Roche's Jesse Green reported that substantial majorities claimed T-20 wasn't that hard to take and didn't interfere much with quotidian pursuits like going to work and having sex (Table 5). And the good response rate didn't slip between study weeks 4 and 24.
Since clinical trial volunteers tend to be a tough-it-out breed, T-20 complaints from rank-and-file HIV contingents will likely gain frequency and volume. The survey population consisted largely of whites (89.3 percent) and gay men (65.4 percent), who typically enjoy greater support and access to information than others with HIV.
Wisely anticipating difficulties in taking T-20, Roche has launched a how-to campaign. The curtain-closer at Roche's Glasgow symposium was Nicky Perry, an HIV nurse from Brighton already well versed in T-20 tips and technicalities. Among the points she made:
So using T-20 won't be a snap. But will it be a snarl? Graeme Moyle doesn't think so, noting that young people with type 1 diabetes -- faced with lifelong insulin shots -- generally trade the burden for the benefits.30 His own experience with growth hormone, IL-2, and T-20 indicates that "injectables are readily accepted, even in healthy persons, when the patient is well educated about the therapy -- its favorable as well as its adverse effects -- and the therapy has effects that are easily observed." For growth hormone and IL-2, those effects are gains in weight and CD4 cells; for T-20, it should be a lower viral load.
Northwestern University's Robert Murphy voiced the unhappy irony that, at a time when drug developers aim for simpler antiretrovirals, T-20 breaks the mold. But a greater irony -- and this time a happy one -- emerged in a Glasgow talk by Elly Katabira from Makerere University in Kampala [abstract PL4.1]. Without mentioning T-20, he noted that many Ugandans would probably be happier with a daily antiretroviral shot than with a hodgepodge of pills. Moyle makes the same point about his London cachement: "For some patients, a simple and well-tolerated injection may be more agreeable than yet more pills."30
Even if those twice-daily needles prove too prickly for some, they may not prove intolerable. As Mike Barr astutely observes, "HIVers are famous for putting up with a lot to stay healthy."31
35 This startling placebo-controlled comparison with efavirenz (plus Combivir in both arms) also found that the nonnucleoside notched a feeble 48-week sub-50 score, 37 percent. The 48-week sub-400-copy measures came closer to the realm of reason, 70 percent for atazanavir and 64 percent for efavirenz.
"This is really very strange," suggested the University of Geneva's Bernard Hirschel while reviewing the results in Glasgow [presentation KL3]. Everyone knows something screwy happened in this study -- because efavirenz has done so much better in a half-dozen previous trials -- but no one knows exactly what. Efavirenz has so loaded its randomized trial trophy shelf that clinicians aren't likely to abandon the drug just because of this study. But the atazanavir results should, at the very least, foment some stimulating forensics at the drug's approval hearing.
In his Glasgow keynote talk, Hirschel became the first clinician of note to suggest in public that there's something funny about this emperor's new clothes. Though atazanavir's antiviral fabric may be tangible, the fit will look loose until researchers can explain these results. And that didn't happen in a detailed Glasgow poster hung by Jean-François Delfraissy (Hôpital di Bicêtre, Paris) [abstract P36]. As in earlier reviews of this study, the unsatisfactory conclusion stressed the "similar efficacy" of once-daily atazanavir and once-daily efavirenz.
Hirschel acknowledged one factor that could contribute to the poor virologic responses -- different versions of the Roche assay used to tote circulating virus. Researchers in North and South America relied on version 1.0, while Europe, Africa, and Asia used version 1.5, which spots non-B subtype viruses better than version 1.0. A 24-week analysis that Bristol-Myers gave Hirschel shows about a 55 percent sub-50 rate with version 1.0 versus about 45 percent with version 1.5.
But that difference can't completely explain the low 48-week scores, Hirschel said, resorting to an automotive analogy. When atazanavir matched nelfinavir in an earlier randomized study, proponents could claim that "our car is as good as a VW." Now, after taking on high-octane efavirenz, they can say "our car is as good as a Mercedes ... but both went only 32 miles per hour." To understand why, Hirschel offered, one must do more than blame the speedometer.
Poor adherence by some people at the 91 sites (on every continent except Australia and Antarctica) seems a more credible explanation at this point. Delfraissy reported the following mutation rates in people who suffered a virologic failure:
Atazanavir's signature mutation, I50L, turned up in only three of 25 genotyped people (12 percent) taking the PI. But even that rate may seem high given the slow accrual of protease mutations as a regimen fails, compared with mutations to 3TC or a nonnucleoside. More detailed analyses of resistance in this study's 800 participants should prove informative. And most everyone hopes that information will prove exculpatory because the efavirenz trial also confirmed atazanavir's sleek lipid profile.
The trial enrolled nearly 650 treatment-naive people from Germany, South Africa, and the United States [abstract PL14.4]. Dirk Schürmann (Charité University Hospital, Berlin) reported median baseline numbers of 166 cells/mm3 in the 908 arm and 177 cells/mm3 in the nelfinavir arm, and respective viral loads of 4.78 and 4.83 logs (about 63,000 copies/mL). Everyone also took abacavir and 3TC. In a missing-data-equal-failure analysis, 68 percent taking 908 and 65 percent taking nelfinavir had a 48-week viral load under 400 copies/mL. The sub-50-copy rates with the same analysis were 56 percent for 908 and 52 percent for nelfinavir. The median CD4 count rose about 200 cells/mm3 in both arms.
While 15 percent taking nelfinavir never got their viral load under 400 copies/mL or did and rebounded, only 4 percent taking 908 had a virologic failure defined this way. But 25 percent quit the 908 arm compared with 15 percent in the nelfinavir arm. The raw numbers don't explain this gap. Toxicity made 8 percent abandon 908 and 5 percent stop nelfinavir. And 908 edged nelfinavir in all those gray-zone grounds for quitting: lost to follow-up, consent withdrawn, protocol violation, and "other." Joep Lange (Academic Medical Center, Amsterdam) contested Schürmann's assertion that dropouts in the 908 arm could not be tied to the drug.
Lange and others also puzzled over the resistance mutation analysis among 32 people taking 908 and 54 taking nelfinavir who had ongoing viral replication during the study. Whereas no primary or secondary protease mutations arose in the 908 group, 30 of 54 taking nelfinavir (56 percent) saw such mutations emerge (P<0.001). One may assume, then, that reverse transcriptase (RT) mutations must explain some of the 908 failures. But RT changes appeared in only three of 32 (9 percent) in the 908 arm compared with 31 of 54 (57 percent) in the nelfinavir arm (P<0.001). Lange labeled these findings in the 908 group "impossible" unless researchers ran the resistance tests at the wrong time.
While 6 percent taking 908 had grade 3 or 4 triglyceride elevations, 2 percent taking nelfinavir had high triglycerides. Only one person in either arm had seriously elevated cholesterol or glucose. Proportions with levels of protective HDL groups.
Another randomized, open-label trial presented earlier in the year also compared 908 (1,400 mg twice daily without ritonavir) with standard-dose nelfinavir plus abacavir and 3TC in treatment-naive people.36 After 24 weeks a missing-data-equal-failure analysis counted 73 percent taking 908 and 54 percent taking nelfinavir with viral loads under 400 copies/mL. Respective sub-50-copy rates were 54 percent and 40 percent. In this study, unlike the one presented in Glasgow, dropouts proved more common with nelfinavir (28 percent by week 24) than with 908 (19 percent). Lipid elevations were rare in either arm through 24 weeks. As David Cooper noted in Glasgow, however, clinicians will want to see how 908 stacks up against one of today's top drugs, efavirenz or lopinavir.
Two of Glasgow's keynote talks touched on antiretroviral access -- one directly, and one indirectly. Michel Kazatchkine (Agence Nationale de Recherches sur le SIDA [ANRS], Paris) met arguments against fast access head-on (Table 6), charging that double standards in the West offer cozy -- but specious -- rationales against widespread antiretroviral therapy [presentation KL1]. Fears of rampant resistance once antiretrovirals see wider use in Africa and Asia strike Kazatchkine as particularly duplicitous, since ever-evolving resistance in Europe and North America have sparked no clamor for withholding therapy there. The risk of transmitting resistant virus will be worse in poorer countries, he maintained, with haphazard, piecemeal access than with well-planned and well-monitored programs.
Few expect any good news from efficacy trials of VaxGen's antibody-stimulating vaccine. So today's only viable candidates, Goudsmit maintained, are vaccines that stir T-cell immunity. Progress with these vaccines remains steady but slow. And modeling by Goudsmit's group shows that the later you start giving a T-cell vaccine in the epidemic's course, the fewer long-term nonprogressors you create. He did not have to spell out the salient corollary of this finding: Antiretrovirals can create lots of long-term nonprogressors right away.
Glasgow's pharmacology session featured one talk that bolstered Kazatchkine's argument on antiretroviral affordability. Cipla, the Indian generic manufacturer, dispatched J.A. Gogtay to report that a three-in-one pill embracing d4T, 3TC, and nevirapine matches the pharmacokinetics of branded versions of those drugs given individually [abstract PL8.4]. A randomized, single-dose crossover study in 28 healthy volunteers showed equivalent bioavailability between the Cipla product, called Trioimmune, and standard doses of d4T, 3TC, and nevirapine. Cipla has also devised a version of the product containing 30 mg of d4T instead of 40 mg for people weighing less than 60 kg. But there is no version with 100 mg instead of 200 mg of nevirapine, so people trying Trioimmune will have to take the three drugs separately for the first two weeks. Treatment with Trioimmune will cost less than $1 a day.
Reviewing early results of Botswana's national antiretroviral program, the Ministry of Health's Ernest Darkoh-Ampen buttressed Kazatchkine's contention that the equator does not divide good from bad adherers [abstract PL4.3]. The pilot program has enrolled 2,867 people at four sites and started treating 2,142 of them. So far, appointment-keeping adherence measures 97 percent. The cohort's pre-treatment CD4 count averaged 50 cells/mm3, and because the waiting time between enrollment and the first dose ran from six to eight weeks, 7 percent of enrollees died as staffers struggled to meet their needs. But Darkoh anticipates that the next wave of enrollees will include fewer seriously ill people who will require less intense follow-up.
A government-sponsored study figured that 110,000 people in Botswana qualify for treatment because they have an AIDS illness or fewer than 200 CD4 cells/mm3. But the country could not hope to treat them all. Generous outside funding helped buy antiretrovirals and build four clinics; the bottleneck has been training staff. But as more streamlined training takes hold, Darkoh believes Botswana stands poised to open four more HIV centers in the next year.
There is nothing unique about the resolute commitment of people in Botswana to their treatment program. The ANRS-sponsored effort in Senegal charted an 87.9 percent adherence rate among 58 people starting antiretrovirals.37 In a Cape Town cohort of 289 people, 42 percent of whom live in "informal dwellings or shacks," Catherine Orrell measured an 87.2 percent adherence rate.38 The strongest predictor of poor adherence was not poverty or its byproducts, but taking a three-times-daily regimen.
Except for the "informal dwellings or shacks," how very "Northern" Cape Town sounds. Or, perhaps, adherence will be better in the South than in the North if, as Michael Kazatchkine argued, antiretrovirals reach people through well-planned programs rather than through black markets. Kathleen Squires (University of Southern California, Los Angeles), who presented the atazanavir-efavirenz trial in the United States earlier in 2002,35 told IAPAC Monthly that preliminary evidence suggests better adherence in Africa and Asia than in Western Europe.
In five Western European countries, a survey of 504 people found that those taking once-daily therapy had a much better record of remembering to take their antiretrovirals [abstract P99]. Graeme Moyle (Chelsea and Westminster Hospital, London) found that 66 percent taking a thrice-daily regimen and 63 percent taking a twice-daily regimen admitted forgetting to take their antiretrovirals. Among those taking once-daily therapy, only 40 percent said they missed doses.
Although 81 percent of respondents from France, Germany, Italy, Spain, and the United Kingdom claimed to be "extremely" (62 percent) or "somewhat" (19 percent) interested in once-daily dosing, they were precise about how many pills they would down in one sitting. While 92 percent said they preferred to take a three-pill regimen "all at once" and 84 percent said the same about four pills, only 59 percent wanted to take six pills all at once, 38 percent eight pills, and 31 percent more than eight pills.
And the roster of once-daily antiretrovirals continues to grow, Moyle noted in a satellite symposium. Besides efavirenz, ritonavir-boosted amprenavir, ddI, 3TC, and tenofovir, clinicians may also soon be able to choose from extended-release d4T (d4T-XR), nevirapine, other boosted PIs, atazanavir, and the nucleoside FTC (perhaps combined in one pill with tenofovir). As recently as the turn of the century, he added, the next regimen in a sequence almost always meant a more complex regimen -- with more pills, more frequent dosing, and/or food restrictions. Today, sequential regimens can preserve first-line simplicity, for example:
But pharmacologic facility means little for the perpetually forgetful. Simpler regimens are crucial to better adherence, Jonathan Schapiro agreed in a plenary talk on the topic [presentation PL5.1]. But they don't solve the problem. Part of that problem, he reminded colleagues, is the near-total lack of adherence training for HIV clinicians. He proffered a simple three-step program:
Formal monitoring, Schapiro elaborated, need not mean MEMS Caps, pill counts, or Inquisitional questionnaires. Studies show that adherence gauged by simple self-reports correlates highly with results garnered from more exacting exercises.39, 40 A 294-person study presented in Glasgow by Maria Paola Trotta (National Institute of Infectious Diseases, Rome) found significant correlations between self-reports of adherence (on a 16-item questionnaire), drug concentrations (P=0.02), and sub-500-copy virologic responses (P=0.03).
How much adherence is enough? As with CD4 counts, viral loads, and resistance, Schapiro proposed, there are no absolute cutoffs. Saying that a person must take all doses on time at least 90 to 95 percent of the time is too simplistic, he argued, because regimens differ in how quickly missed doses promote replication of resistant virus.
How often should resistance be monitored? Again, no one has set in stone the optimal interval for measuring CD4 cells, HIV RNA, or viral susceptibility to drugs. And no one knows how often clinicians must replay the adherence rule book. Adherence may not have to be checked at every visit, Schapiro offered. But, as with CD4 counts and viral loads, it has to be checked regularly.
Julio Montaner and colleagues have reaped sheaves of rich data from their closely monitored British Columbian cohort. They have studied the effects of CD4 count, viral load, and double or triple regimens with or without PIs or NNRTIs on HIV disease progression and death. One factor, he stressed in Glasgow, means more than any single drug in separating the sick from symptom free, the quick from the dead, in two or three years. Adherence.
Mark Mascolini writes about HIV infection (firstname.lastname@example.org).
** Mutations shared by the thymidine analogs AZT and d4T are M41L, D67N, K70R, L210W, T215Y/F, and K219Q/E.
*** D:A:D stands for Data Collection on Adverse Events of Anti-HIV Drugs.
This article was provided by International Association of Physicians in AIDS Care. It is a part of the publication IAPAC Monthly.