Introduction

Here are some questions asked at the 40th ICAAC. (This is a quiz.)

1. What treatment strategy do 58 percent of people with HIV infection ask for, only to find that a mere 31 percent of clinicians agree to try it?
2. What do abacavir, stavudine (d4T), and nelfinavir have in common?
3. Who's more likely to have a lymphoproliferative response to HIV p24 antigen--people with viral loads below 50 copies/mL for more than six months, or people with viral loads between 50 and 10,000 copies/mL for more than six months?
4. Which drug can outgun nelfinavir in treatment-naive people--lopinavir (ABT-378) or nevirapine?
5. Is the accent in pyrophosphorolysis on the fifth or sixth syllable?

Right. These are trick questions. Unless you went to the same ICAAC sessions as this reporter and have a flawless memory, don't count on acing this test. And anyone who is not Victoria Johnson and gets number 2 right has been cursed with total recall.

Yet all of these questions make the same point, even while suggesting key themes of this year's ICAAC edition. The point--already appreciated by HIV physicians with more than two minutes of experience--is that the unexpected became routine when Michael Gottlieb realized a few young men had hopelessly wrecked immune systems for no apparent reason. And that truism about expectations has held since Gottlieb described this immunodeficiency syndrome 20 years ago.

The five themes are:

1. Structured treatment interruptions (STIs). Which is also the answer to the first question.
2. The vagaries of viral resistance and resistance testing. A survey of resistance experts found the worst agreement on genotypic calls involving abacavir, d4T, and nelfinavir.
3. The mysterious ways in which the body responds to HIV and to anti-HIV drugs. In a small but provocative study, a higher proportion of people with viral loads under 50 copies/mL for at least six months had lymphoproliferative responses to p24. But the difference in response rates between this group and people with viral loads between 50 and 10,000 copies/mL was not statistically significant.
4. The advent of the much heralded protease inhibitor (PI) lopinavir, and the antiviral pluck of nonnucleoside reverse transcriptase inhibitors (NNRTIs) in some recent studies. Both lopinavir and nevirapine beat nelfinavir in trials reported at the meeting.
5. Better understanding of the mechanisms of antiretroviral activity and viral resistance and development of better drugs. The best evidence suggests the fifth syllable in py-ro-phos-phor-OL-y-sis gets the heaviest stress. Regardless, it appears that pyrophosphorolysis does not pluck the nucleotide analog tenofovir from the growing viral DNA chain as avidly as it does zidovudine (AZT). This difference between tenofovir and AZT could explain why AZT-resistant mutants remain susceptible to tenofovir.

Of course researchers who flashed slides or unfurled posters at the 40th ICAAC raised dozens of other issues critical to the health of people with HIV infection. Perhaps the most important of these other concerns--the fast-evolving drama of antiretroviral side effects--got the spotlight in November's IAPAC Monthly. This article will stick to antiretroviral pressure and its immunologic consequences, to resistance, and to STIs.

Return to Article Contents

First phase III results as lopinavir goes to market

Perhaps the biggest headline to emerge during ICAAC came not from the Toronto convention center but from Rockville, Md., where the US Food and Drug Administration (FDA) gave its seal of approval to Abbott Laboratories' second PI, lopinavir. Packed into the same capsule with 100 mg of ritonavir, the drug sells as Kaletra. Abbott hadn't brought any Kaletra pens, Kaletra coffee cups, or Kaletra canvas bags to ICAAC. But it did show up with three slide reports, plenty of update posters, and the package insert.

Sharon Walmsley (Toronto General Hospital) detailed findings from the key lopinavir study, a one-on-one joust with nelfinavir [abstract 693^*]. "One-on-one" may lean toward inaccuracy because the 326 people randomized to take 400 mg of lopinavir twice daily automatically got 100 mg of ritonavir twice daily as a pharmacokinetic kicker. The 327 people randomized to take nelfinavir at the three-times-daily dose got no other PI. Everyone in this randomized, double-blind trial also took d4T and lamivudine (3TC), and no one had tried any antiretrovirals before. Average baseline viral load measured 4.9 logs (about 79,000 copies/mL) in both groups, and average starting CD4⁺ counts hovered around 260 cells/mm³.

After 40 weeks people in the lopinavir arm had significantly better virologic responses by intent-to-treat (missing-data-equal-failure) and on-treatment analyses (Table 1). Forty-week CD4⁺ gains measured 190 cells/mm³ with lopinavir and 177 cells/mm³ with nelfinavir. The good 40-week viral load numbers in the lopinavir group actually fell a tad short of 48-week results in a smaller, earlier study.¹ In that phase II trial, 100 treatment-naive people started various doses of lopinavir/ritonavir,² also with d4T/3TC. Average baseline viral loads matched those in the newer trial. In the phase II study, a 48-week missing-data-equal-failure analysis found that between 75 and 79 percent notched a sub-50 viral load at 48 weeks, depending on whether they started d4T/3TC with lopinavir or three weeks later.

And 48-week virologic results with lopinavir seem to last and last. Constance Benson (University of Colorado, Denver) offered a 96-week follow-up of the phase II study at ICAAC [abstract 546]. In the intent-to-treat analysis, 78 percent still had a viral load under the 50-copy mark. And 92 percent stayed sub-50 in the on-treatment analysis.

So no matter how you cut it, the licensed 400/100-mg package of lopinavir/ritonavir muffles viral replication in a big proportion of treatment-naive people. How much should be made of the nelfinavir comparison is another question. Graeme Moyle (Chelsea and Westminster Hospital, London) observed that the double-blind trial design meant everyone had to take pills or placebos three times a day, since the study used the three-times dose of nelfinavir. But a midday dose, Moyle argued, is the one people miss most. If that habit held true in this study, it would mean the nelfinavir group got less active drug, because they would be taking nelfinavir in the mid-day dose and the lopinavir group would be taking placebo. Walmsley said adherence is being analyzed now. She noted, though, that the virologic response to nelfinavir in this trial resembled the response in the twice- versus thrice-daily nelfinavir study run by Agouron Pharmaceuticals, Inc. So she didn't think adherence would prove an issue.

When the lopinavir study began, nelfinavir was the logical comparative drug because it had emerged as the PI of choice for most clinicians (although later work suggested it did no better than indinavir in challenging efavirenz).³ These days, though, many would argue that double

PIs have become the standard of care. Martin Hirsch (Massachusetts General Hospital, Boston) made the point in his ICAAC overview of antiretroviral therapy. "Whether [lopinavir/ritonavir] is any better than indinavir/ritonavir remains to be seen," he said [presentation 611].

HIV had a tough time outmaneuvering lopinavir during the first 40 weeks of the phase III trial. Among people with RNA loads above 400 copies/mL despite lopinavir therapy, Walmsley and colleagues genotyped virus from 18 individuals. They found no primary site protease mutation in any of them. Among 40 nelfinavir nonresponders who were genotyped, four had the 30N or 90M protease mutation.

How did lopinavir do in the side effects department? Here, nelfinavir may have won by a nose, preserving its track record as a relatively tolerable and safe protease inhibitor. Slightly more people taking nelfinavir did have a PI-related side effect--3 percent versus 2 percent for lopinavir. Fifteen people taking lopinavir and 16 taking nelfinavir had moderate to severe diarrhea. Moderate to severe nausea bothered seven taking lopinavir and four taking nelfinavir. Perhaps the most interesting difference, and the only statistically significant one, involved lipids. Seven percent in the lopinavir group had (nonfasting) triglycerides above 750 mg/dL, compared with 1 percent for nelfinavir (P < 0.001). At 24 weeks, 7 percent taking lopinavir had grade 3/4 cholesterol jumps, compared with 3 percent taking nelfinavir.

Benson's 96-week update on the phase II study affords a good look at what happens when previously untreated people take lopinavir for nearly two years [abstract 546]. Diarrhea does appear to be the main physical complaint: 23 percent had more than three loose stools daily and another 8 percent had three or fewer loose stools. Nausea bedeviled 15 percent.

Like other PIs, lopinavir with ritonavir riles markers of lipid metabolism and liver function. Benson reported that 14 percent had cholesterol readings above 300 mg/dL, 12 percent had triglycerides above 750 mg/dL, and 10 percent has an AST or ALT more than five times the upper limit of normal. Still, only two people dumped lopinavir by 96 weeks because of side effects or lab abnormalities, one with diarrhea and one with a high AST/ALT.

Oddly, Benson's poster mentioned nothing about body shape changes in this 100-person cohort, though lipodystrophy would certainly be expected after 96 weeks of PI therapy. Earlier this year, discussing a 72-week report on this same study at the July 9-14 XIII International AIDS Conference, Abbott's Eugene Sun told IAPAC Monthly that five people had abnormal fat distribution at that point. Among 70 people who began lopinavir after taking one other PI, six had lipodystrophy at 72 weeks. The package insert does not specifically mention fat changes in the list of side effects, though it cites "obesity" and "wasting" as possible problems.

Return to Article Contents

Lopinavir drug interactions, and concern over low-dose ritonavir

Other lopinavir presentations involved resistance studies presented earlier this year and described at length in this publication.⁴ But some new details emerged:

• Raise lopinavir dose with efavirenz. An open-label study of lopinavir/ritonavir plus efavirenz in 57 PI-experienced but NNRTI-naive people suggested greater potency with a 533/133-mg dose of the PIs than with 400/100 mg [abstract 697]. Stephen Becker (Pacific Horizon Medical Group, San Francisco) reported an on-treatment analysis in which 92 percent taking the higher dose had a 24-week viral load below 400 copies/mL, compared with 80 percent taking the lower dose. In a missing-data-equal-failure analysis, percentages with sub-400 viral loads were 82 percent for 533/133 mg and 69 percent for 400/100 mg. Pharmacokinetic studies show that efavirenz trims concentrations of lopinavir, so Abbott recommends that clinicians "consider" the 533-mg dose when giving the PI with this nonnucleoside. The same advice applies to nevirapine.
• Watch for atorvastatin side effects. Abbott's Robert Carr summarized a study of lopinavir/ritonavir plus atorvastatin or pravastatin in eight healthy women and 16 healthy men [abstract 1644]. Neither statin substantially affected concentrations of lopinavir/ritonavir. The PIs boosted pravastatin levels 30 percent--not enough to require a dosage adjustment. But lopinavir/ritonavir pushed atorvastatin's maximum concentration up 4.7-fold, and its 24-hour area under the curve climbed 5.9-fold. Abbott recommends close monitoring for possible atorvastatin side effects if the drug is used with lopinavir/ritonavir, especially at a dose above 10 mg daily.
• Saquinavir, tipranavir, or amprenavir for lopinavir rescue? A resistance study of three isolates from PI-experienced people with viral rebounds while taking lopinavir showed that virus from two saquinavir-naive individuals remained susceptible to saquinavir [abstract 2112]. All other isolates from people with rebounds remained fully susceptible to tipranavir and fully or nearly fully susceptible to amprenavir. Abbott's Scott Brun suggested saquinavir, tipranavir, and amprenavir may be appropriate rescue PIs after virologic failure of lopinavir.
• High resistance threshold with lopinavir. Analyzing a study of 57 multi-PI-exposed individuals, Abbott's Dale Kempf found that the number of protease mutations when lopinavir begins predicted response to this PI better than specific individual mutations [abstract 1264]. Kempf counted the familiar mutations at protease positions 10, 20, 24, 46, 53, 54, 63, 71, 82, 84, and 90. Virus with up to five of those changes remained sensitive to lopinavir. Six or seven mutations left a virus with "intermediate" sensitivity. And eight or more mutations made virus resistant to lopinavir.

But a few yards down from Kempf's poster, an ugly fly landed in the low-dose ritonavir ointment--or seemed to, depending on your interpretation. It certainly grabbed the attention of an ever-present swarm of attendees who buzzed about this poster and read it through to its unhappy conclusion [abstract 1267]:

"In ritonavir/indinavir patients, 'baby dose' [100-mg ritonavir] selective pressures could lead to re-emergence of resistance mutations associated with these drugs," contended S. Chaillou and colleagues (Nice University Hospital, France). "Use of 'baby dose' ritonavir could be a threat to naive patients as it might select for ritonavir/indinavir resistance-associated mutations and those of the boosted PI. To avoid this it would be reasonable to use 'baby dose' [ritonavir] only in patients who already have ritonavir mutations. In naive patients it should only be used in association with a PI sharing these mutations."

What's the evidence? Chaillou looked retrospectively at 34 PI-experienced people who took saquinavir (600 mg of the hard-gel formulation three times daily) plus ritonavir (100 mg twice daily) as part of the VIRADAPT genotyping trial. Before VIRADAPT, three of the 34 had used saquinavir, nine had used ritonavir, and 21 had used indinavir. The number of people with saquinavir-associated mutations (48V and 90M) or ritonavir/indinavir-associated mutations (46I/L and 82A/F/T) rose steadily while they took saquinavir/ritonavir for 12 months (Table 2).

Now here's the crux of the argument. The median plasma concentrations of "baby dose" ritonavir in these 34 people stood at 0.47 µg/mL (range 0.08 to 1.1 µg/mL). Those concentrations, the Nice researchers maintained, "could favor emergence of specific mutations as demonstrated in a monotherapy study."⁵ In the 34 VIRADAPT patients, the investigators contended, these seemingly low ritonavir concentrations could have been enough to beckon the mutations observed.

It seems plausible that low-dose ritonavir elicited those position 46 and 82 mutations, but it seems just as plausible that ritonavir did not. Whether 100-mg shots of ritonavir had anything at all to do with the accumulation of saquinavir-linked 48V and 90M mutations seems even more questionable.

A problem with the analysis, observed Abbott's Dale Kempf, is that the saquinavir/ritonavir regimen studied--600 mg of saquinavir three times daily, but 100 mg of ritonavir only twice daily--is suboptimal. To boost another PI best, ritonavir should always be dosed at the same time. The low proportion of study participants who got their viral loads under 200 copies/mL with saquinavir/ritonavir (Table 2) attests to the inadequate dosing. Because virus was replicating freely in most of these people, it stands to reason that the inadequate saquinavir dose would arouse more and more of the typical saquinavir mutations, 48V and 90M, with or without low-dose ritonavir.

Did 100 mg of ritonavir twice daily select for the indinavir- and ritonavir-associated 46 and 82 mutations? Maybe. Those mutants may have been archived or circulating at low levels in study participants who had taken indinavir or ritonavir before saquinavir/ritonavir. And maybe not. When Mark Winters (Stanford University, Stanford, Calif., USA) studied people in whom saquinavir alone failed, he found not only 48V and 90M (in 58 percent), but also 36I, 46I, 82V, and 84V (in 35 percent).⁶ So the suboptimal saquinavir dose in VIRADAPT may have driven the re-emergence or accumulation of the 46 and 82 mutations in that study.

Kempf noted, though, that boosting a PI with low-dose ritonavir "may produce a different pattern than that PI usually does, even if ritonavir is not actually directly contributing to selective pressure. Because the blood levels of that PI are higher in the enhanced regimen, there is a different barrier for the virus to overcome." So the virus may "choose" an atypical genetic pathway that works better against the boosted PI. In any case, as Chaillou and colleagues observed, possible resistance risks with 100 mg of ritonavir twice daily don't matter to PI-naive people beginning lopinavir/ritonavir, since the two PIs select the same mutations.

Return to Article Contents

Some good lipid news, and fair antiviral numbers, with a new PI

With lopinavir now on pharmacy shelves, attention will focus more acutely on one of the next PIs in the pipeline, Bristol-Myers Squibb's BMS 232632. Even without an easily remembered alphanumeric moniker, this drug has aroused interest because it could be the first once-daily PI. Early work also suggests that it may handle virus resistant to one, two, or possibly three other PIs. At ICAAC, Bristol-Myers researcher Richard Colonno reviewed this resistance study, which he first presented a few months ago.⁴ Colonno hasn't been able to pin down a specific mutation pattern that makes virus shrug off BMS 232632. As with lopinavir, already familiar PI mutations can consort to hamstring the BMS drug.

Bristol-Myers faces pressing questions about how to develop BMS 232632. After Colonno's presentation, Steven Deeks (University of California, San Francisco) noted that the drug's trough levels with once-daily dosing dangle perilously close to the concentration needed to suppress nonmutant (wild-type) virus. Deeks wondered whether Bristol-Myers may move to a twice-daily schedule when studying the drug for salvage. Colonno replied that the company is plotting the best phase III tacks right now, and that a ritonavir boost is not out of the question.

In fact in the first trial that throws BMS 232632 at PI-resistant virus, Bristol-Myers has taken a once-daily approach some might call cautious. The study will randomize 300 to 400 PI-experienced people to combine 400 mg of BMS 232632 with 1200 mg of saquinavir, or to take ritonavir/saquinavir at the twice-daily 400/400-mg dose. Caution, of course, is in the eye of the trial planner. When Abbott first gave lopinavir to single-PI veterans, it not only boosted the drug with ritonavir but also added nevirapine as the first NNRTI anyone had ever taken. Indeed, setting low early hurdles for a new drug is a hallowed tradition in antiretroviral research.

The principal study of BMS 232632 so far involves treatment-naive people randomized to take 200, 400, or 500 mg of the drug once daily or 750 mg of nelfinavir three times daily [abstract 691]. Ian Sanne (Johannesburg Hospital, South Africa) noted that the trial was double-blind for BMS 232632 doses but open-label for nelfinavir. First people took only one of the PIs for two weeks, then they added didanosine (ddI) and d4T. Sanne reported results for the 98 study participants who had been treated for 24 weeks.

BMS 232632 didn't outmuscle nelfinavir, as lopinavir/ritonavir had [abstract 693, above], but it held its own, especially at the highest dose. During the two-week monotherapy phase, the average viral load fell about 1.5 logs in all four treatment arms. RNA levels slipped another 1 log lower after people started taking ddI/d4T (Table 3). But when Sanne and colleagues tallied proportions whose viral loads sank below 400 or 50 copies/mL by 24 weeks, nelfinavir held a nonsignificant edge, matched most closely by 500 mg of BMS 232632 (Table 3).

(You all know that cross-study comparisons are an abomination in the eyes of the Supreme Statistician, but no one will tell if your eyes drift between Tables 1 and 3 for a few seconds. The studies represented in those tables used the same dose of nelfinavir.)

Two people had to leave the BMS 232632-nelfinavir trial during the monotherapy phase because of nausea, the only side effect that could be attributed to either PI. While 79 percent taking nelfinavir complained of diarrhea, that problem bothered 29 percent taking BMS 232632. On the lipid front, there may be some good news with this new PI. Total cholesterol, LDL cholesterol, and triglycerides all rose in the nelfinavir arm, but stayed flat in the BMS 232632 arm. Those nice numbers with the BMS PI are not novel. A few studies show that indinavir doesn't boost lipids in the first weeks of treatment (see "Nevirapine versus indinavir" below). So the clinical correlates of this early trend with BMS 232632 will be eagerly anticipated.

The main lab abnormality with the BMS PI, unconjugated hyperbilirubinemia, struck 62 percent taking the drug in this trial. Researchers rated most cases as grade 1 or 2, and grade 4 elevations could be reversed by reducing the PI dose. This study did not link high bilirubins with soaring liver enzymes, but the side effect arouses concern because it is dose dependent and the highest dose of BMS 232632 seemed to work best in this trial.

Bristol-Myers Squibb's Edward O'Mara probed this problem more closely in a gene study involving 20 people taking the BMS drug plus saquinavir [abstract 1645]. He singled out a certain genetic make-up that appears to triple the risk that bilirubin will climb over 43 µM (2.5 mg/dL), the level associated with jaundice.

This little study is not conclusive, O'Mara observed, because genotyping alone says nothing about the actual phenotypic expression of the genes studied. He concluded, though, that "this preliminary observation provides a scientific rationale for a population-based study" to see whether the implicated genotype makes a big difference. Those population studies are under way as part of phase II trials.

Return to Article Contents

Tenofovir and DAPD: tenacious mutant maulers?

Three reports showcased the potential merits of tenofovir, the nucleotide RT inhibitor from Gilead Sciences. Robert Schooley (University of Colorado, Denver) rolled out the 48-week results of a double-blind, placebo-controlled trial that added tenofovir to stable regimens people had taken for at least eight weeks [abstract 692]. Investigators randomized 54 people to add 300 mg of tenofovir daily, 51 to add 150 mg daily, 53 to add 75 mg daily, and 28 to add placebo. After 24 weeks everyone in the placebo group could trade the dummy pill for tenofovir. People taking 75 or 150 mg of tenofovir switched to 300 mg at 40 weeks.

Study participants had taken antiretrovirals for an average 4.6 years, and it showed. Nearly all of them, 94 percent, had nucleoside reverse transcriptase inhibitor (NRTI)-related mutations, 57 percent had PI mutations, and 32 percent had NNRTI mutations. Average baseline viral loads ranged from 3.6 to 3.8 logs (about 4000 to 6000 copies/mL) in the four treatment arms. Everyone had detectable viremia when the study began. The average starting CD4⁺ count stood at 375 cells/mm³.

After 48 weeks mean viral loads dipped 0.7 log in the 300-mg group and 0.7 log in people who began with placebo and traded up to tenofovir at 24 weeks. Average viral load drops measured 0.6 log in the 150-to-300-mg group and 0.4 log in the 75-to-300-mg group. But CD4⁺ counts hardly budged. Gains averaged 11 cells/mm³ for 300 mg, 16 cells/mm³ for 150 mg, 11 cells/mm³ for 75 mg, and 25 cells/mm³ for the placebo/300-mg group.

Those scanty T-cell spurts surprised Richard Haubrich (University of California, San Diego), who said he'd expect bigger bounds in people whose viral load fell more than a half log. Schooley noted, though, that viral loads remained detectable in 72 percent of study participants throughout the 48 weeks. He suggested the mingy CD4⁺ response may reflect findings of a study in which people with low but detectable viremia had lower CD4⁺ gains than people with undetectable virus. But as a group the partial responders in that study had a clinical course equivalent to the full responders over three years [abstract 565, discussed below].

A comparison between the placebo group and the 300-mg group showed a better virologic response with tenofovir regardless of certain baseline resistance patterns. Among people with no AZT resistance mutations, the mean time-averaged difference for viral load area under the curve (DAVG) for placebo measured +0.19 versus -0.61 for 300 mg of tenofovir (P = 0.019). For people who began the trial with AZT mutations, mean DAVGs were -0.08 for placebo and -0.57 for tenofovir (P = 0.003). And for those who started with the 3TC mutation, 184V, mean DAVGs were -0.20 for placebo and -0.65 for tenofovir (P = 0.025).

Those findings buttress Gilead's contention that tenofovir can grapple with virus that has evolved to resist current NRTIs. Michael Miller, Gilead's resistance specialist, explored this tenofovir trait in a study of 72 clinical isolates with common clusters of NRTI mutations [abstract 2115]. He gauged the susceptibility of these isolates to tenofovir by using Virco's phenotyper, which rates a viral sample as sensitive, "intermediate," or resistant depending on how much a drug's 50 percent inhibitory concentration (IC₅₀) climbs when exposed to that virus.

The only isolates that completely slipped tenofovir's grasp were multinucleoside-resistant isolates with 69S insertion mutations. Another notable multi-NRTI-resistant virus, featuring a 151M mutation and others, remained fully susceptible to tenofovir. Virus with the 184V mutation proved hypersensitive to tenofovir, and 184V bolstered the susceptibility of virus with high-level resistance to AZT. Even without 184V, most samples with high-level AZT resistance bowed to tenofovir. Tenofovir calls forth the 65R mutation in vitro, as do didanosine (ddI), zalcitabine (ddC), and abacavir. But patient isolates marked by 65R proved susceptible to tenofovir in the Virco system, and even more susceptible when the isolate carried 184V.

The question remains how all this will play out clinically. Gilead's first nucleotide, adefovir, had this same fondness for 184V mutants. But that virtue proved insufficient to convince the FDA to license adefovir as a worthwhile addition to HIV salvage regimens. Early studies of tenofovir show that it packs a stronger antiviral punch than adefovir, and it appears not to share adefovir's troubling renal side effects. But these studies have yet to prove that tenofovir's 184V trick will pay off clinically in NRTI-experienced people.

After Miller's presentation, Joseph Eron (University of North Carolina, Chapel Hill, USA) noted that people with 184V mutants in the trial Schooley described didn't have a better viral load response than people without 184V. That may not be terrible news for tenofovir, given the small size of the study, the participants' multifarious mutation patterns, and the grab-bag of regimens they were taking. But it sure would have been nice if those people with 184V mutants had done decidedly better than those without it. Miller said it's too early to tell whether 184V will make a clinical difference. A phase III study in treatment-experienced people has a better shot at answering this question.

Another study, presented by Gilead's Lisa Naeger, suggested why AZT-resistant virus remains susceptible to tenofovir [abstract 1265]. Work in the past few years discerned two tactics HIV uses to outmaneuver AZT--pyrophosphorolysis and ATP-dependent primer blocking.⁷ Don't worry. Unless you're already a pyrophosphorolysis pooh-bah, you'll probably never be called on to explain these mechanisms. The simplest way to understand them is to remember that AZT and the other nucleosides are dubbed "chain terminators" because they insinuate themselves onto the growing viral DNA chain being stamped out by reverse transcriptase, and they stop that process dead. But just a dash of pyrophosphate can lyse AZT right off that DNA chain. So reverse transcriptase gets back to work turning viral RNA, link by link, into viral DNA.

In a clever experiment that actually reckoned how adeptly these mechanisms yank a chain terminator off the chain, Naeger found that tenofovir clings to its target more tightly than AZT. And that holds true whether the virus tested is wild-type, a position 215 mutant, a 67/70 mutant, or a 67/70/215 mutant. The "efficiency of tenofovir removal," Naeger calculated, "is 2- to 3-fold less efficient than [AZT] removal by pyrophosphorolysis and 10- to 30-fold less efficient than [AZT removal] by the ATP-dependent unblocking mechanism." This study impressed resistance maven Mark Wainberg (McGill University, Montreal), who suggested in an overview lecture that Naeger's findings "could explain why tenofovir may have excellent potential" [presentation 2119].

Another RT inhibitor, the nucleoside DAPD, can also go toe-to-toe with virus resistant to other drugs in the class. It complements tenofovir in tangling with multinucleoside-resistant virus, stifling 69S insertion mutants but not 151M mutants. Joseph Eron spelled out results of two studies, one involving treatment-naive individuals and one in people treated with nucleosides and often with PIs or NNRTIs [abstract 690].

Eron and colleagues tested 25, 100, 200, 300, and 500 mg of DAPD, given twice daily to coveys of six or seven people with group viral load averages ranging from 4.3 to 4.7 logs (about 20,000 to 50,000 copies/mL) and group CD4⁺ counts from 307 to 560 cells/mm³. After 15 days of DAPD monotherapy, viral loads fell 1.5 logs in the 300- and 500-mg groups and 1.14 logs or less in lower-dose groups. No resistance mutations emerged during the study, and no one quit because of side effects.

All of the treatment-experienced people had tried 3TC with either AZT or d4T. Their viral loads ranged from 4.77 to 5.05 logs (about 59,000 to 112,000 copies/mL), and they had taken a median of 6.5 antiretrovirals for a median of 3.7 years. The 500-mg twice-daily DAPD dose did the best, pushing down viral loads about 1 log. Grade 2 or worse lab abnormalities included creatine kinase jumps in 17 percent and triglyceride upticks in 9 percent. Eron said many people had high triglycerides when they started DAPD. People taking bigger doses of the drug did not appear to have more bad lab numbers than people taking smaller doses.

Return to Article Contents

Nevirapine takes on nelfinavir, then indinavir, then efavirenz

The debate over starting treatment with a PI or an NNRTI continues. And, if several hundred HIV clinicians at an ICAAC interactive session are a representative sample, the nonnukes have gained the first-line upper hand, at least in one situation. In a talk on lipodystrophy and metabolic complications, Judith Currier (University of California, Los Angeles) posited this case report [presentation 1877]. A 48-year-old man wants to start his first antiretrovirals. He's trying to quit smoking and has a family history of myocardial infarction. What would you pick as the keystone of the regimen--efavirenz, nevirapine, indinavir/ritonavir, nelfinavir, amprenavir, or something else?

Nearly two-thirds, 63 percent, went with an NNRTI; 45 percent picked efavirenz and 18 percent nevirapine. The PIs fought it out for the scraps, 16 percent going to nelfinavir, 11 percent to indinavir/ritonavir, and 2 percent to amprenavir. The remaining clinicians took the "something else" option.

Although nevirapine placed far behind efavirenz in this unscientific survey, studies in which people switch from a PI show that nevirapine does more than efavirenz to rein in runaway lipids. And both NNRTIs maintain PI-induced viral suppression. Trials of efavirenz in treatment-naive people suggest to many that it's tougher on HIV than nevirapine, perhaps even tougher than stand-alone PIs. But three ICAAC studies comparing nevirapine with nelfinavir, indinavir, or efavirenz found that nevirapine is no wimp as a first-line drug. A fourth study, though, endorsed the opinion that PIs ensure a longer response than NNRTIs, either as first-line or second-line drugs.

• Nevirapine versus nelfinavir. When nelfinavir pushed indinavir aside as the most prescribed PI, it also knocked indinavir off the standard-of-care stool for comparisons with other antiretrovirals. Now nelfinavir is taking its lumps, coming up short in a 40-week contest with lopinavir/ritonavir [abstract 693, above, Table 1]. And a 36-week Spanish and Argentine comparison with nevirapine also left nelfinavir on the mat, but some remained unconvinced by that contest. Daniel Podzamczer (Hospital Princeps d'Espanya, Barcelona, Spain) delivered the surprising results [abstract 694].
  
  

  The 15-site open-label study assigned 70 treatment-naive people to begin nelfinavir (1250 mg twice daily) plus Combivir (AZT/3TC) and 72 to start nevirapine and Combivir. The median baseline T-cell count was marginally lower in the nelfinavir group (351 versus 361 cells/mm³), and the baseline viral load somewhat higher in the nelfinavir arm (165,000 versus 118,000 copies/mL). But those differences were not statistically significant. About 30 percent of study participants had viral loads above 100,000 copies/mL, and they were evenly distributed between treatment arms.

  After 36 weeks, nevirapine had outdistanced nelfinavir by every virologic measure, though not always significantly (Table 4). What do these numbers say? Nevirapine certainly looks like the clear virologic winner (CD4⁺ gains were equivalent). And unlike the nelfinavir-lopinavir contest, everyone in this study took the drugs on a twice-daily schedule. But nelfinavir response rates are much lower than in other trials, even in the lopinavir contest (Table 1).

  One can cite a few extenuating circumstances favoring nelfinavir. Even though baseline viral loads were not significantly higher in the nelfinavir group, they were on the average about 47,000 copies/mL higher. Even when a regimen works, it typically takes longer to get a higher viral load under a given mark (see "Not all post-24 week responders" below), and the targets were lower than usual in this trial: 200 and 20 copies/mL. But 36 weeks is a pretty long time.

  Table 4. Nevirapine versus nelfinavir (both with Combivir)
  in treatment-naive people for 36 weeks

  Viral load measure (copies/mL), type of analysis	Nelfinavir, 1250 mg twice daily (% of patients)	Nevirapine, 200 mg twice daily (% of patients)	P
  <200, missing = failure	55.7	70.8	0.06
  <200, on treatment	78.0	83.7	0.5
  <20, missing = failure	38.6	66.7	<0.001
  <20, on treatment	56.1	79.6	0.02
  With baseline viral load >100,000 copies/mL
  <200, missing = failure	53.8	71.4	0.22
  <20, missing = failure	15.4	61.9	0.001
  Source: Daniel Podzamczer, abstract 694.

  Perhaps more important is the higher dropout rate in the nelfinavir group. Although 16 people left the nevirapine group because of side effects and 13 left the nelfinavir group, overall dropouts were higher with the PI: 45.7 percent versus 33.3 percent. So all the missing-data-equal-failure analyses favor nelfinavir. The 36-week on-treatment analysis with the 200-copy goal was a statistical dead heat.

  Although Podzamczer didn't present adherence data, he speculated that worse adherence in the nelfinavir group may explain some of the difference in viral load outcomes. He noted that about 40 percent of trial participants are current or former injecting drug users, some of them enrolled in methadone programs. The baseline data presented did not indicate whether one treatment group had more drug injectors than the other. But whatever accounts for the poor nelfinavir showing, nevirapine did well in this study.

• Nevirapine versus indinavir. In a single-site study, Josep Guardiola (Sant Pau Hospital, Barcelona, Spain) compared nevirapine with a different PI, indinavir, in another randomized, open-label trial [abstract 539]. This time, the virologic response to the PI was in line with what one might anticipate. And nevirapine did just as well.

• The Barcelona clinicians randomized 35 people to take standard-dose indinavir and 35 to take standard-dose nevirapine, both with twice-daily d4T and once-daily ddI. As in the nevirapine-nelfinavir study, people in the PI arm started treatment with a slightly lower mean CD4⁺ count (310 versus 335 cells/mm³ for nevirapine) and a higher average viral load (269,362 versus 192,773 copies/mL for nevirapine). The treatment groups matched closely in age and HIV disease stage.

  After nine months of treatment, almost identical proportions in each group had viral loads under 200 or 50 copies/mL by intent-to-treat and on-treatment analyses (Table 5). People taking indinavir gained an average 409 CD4⁺ cells/mm³ over 15 months of follow-up, and people taking nevirapine added an average 487 cells/mm³. Five people left the study because of drug side effects, including two ALT/AST elevations attributed to nevirapine and one case of vomiting blamed on indinavir. Investigators linked two neuropathy dropouts to ddI or d4T. One person died of lactic acidosis.

  Table 5. Nevirapine versus indinavir (both with ddI/d4T) in treatment-naive people for nine months

  Viral load measure (copies/mL), type of analysis	Indinavir (% of patients)	Nevirapine (% of patients)
  <200, intent-to-treat	77	71
  <200, on treatment	81	80
  <50, intent-to-treat	45	44
  <50, on treatment	50	56
  Source: Josep Guardiola, abstract 539.

  These investigators also tracked lipid levels from the beginning of the study through nine months and found no substantial change in total cholesterol or triglycerides with either nevirapine or indinavir. In the nevirapine group, average cholesterol levels edged up from 4.19 to 4.52 mmol/L and average triglycerides from 1.11 to 1.21 mmol/L. Among people taking indinavir, average cholesterol inched from 4.39 to 4.4 mmol/L and triglycerides from 1.36 to 1.52 mmol/L over nine months.

  Those flat lipid trends among treatment-naive people beginning indinavir confirm a 1999 study by Michael Dubé (University of Indiana, Indianapolis). Dubé found, though, that fasting glucose rose and insulin sensitivity fell in the first eight weeks of indinavir therapy.⁸ Another study repeated Dubé's indinavir findings in a four-week study of healthy volunteers.⁹

• Nevirapine versus efavirenz. A second randomized, open-label, single-center study pitted nevirapine against efavirenz (plus d4T and once-daily ddI) in treatment-naive people with baseline viral loads around 20,000 copies/mL [abstract 472]. Maria Nuñez (Instituto de Salud Carlos III, Madrid, Spain) reported that 28 people randomized to take nevirapine had a baseline median CD4⁺ count of 405 cells/mm³, and the 26 starting efavirenz had a median of 242 cells/mm³.

• After a median nine months of follow-up, a noncompleter-equals-failure analysis counted 22 of 28 people (79 percent) with viral loads below 50 copies/mL in the nevirapine group and 22 of 26 (85 percent) in the efavirenz group. In the on-treatment analysis, 22 of 25 (88 percent) taking nevirapine went below 50 copies/mL, compared with 22 of 22 taking efavirenz. None of these differences were statistically significant. Daniel Kuritzkes (University of Colorado, Denver) observed that these results establish the comparability, but not the equivalence, of the two NNRTIs in this population, because confidence limits were not defined.

  What surprised some in the audience were the cholesterol leaps Nuñez found in 13 individuals in each treatment group. Overall, the 21 percent of study participants who had hypercholesterolemia at baseline ballooned to 55 percent after nine months of follow-up. Earlier studies charted cholesterol increases with efavirenz, sometimes attributed to gains in "good" high-density lipoprotein (HDL) cholesterol. But nevirapine has not been tied to surging cholesterol, as the just-described Guardiola study confirmed. Nuñez did not say whether other factors, such as changes in diet, may have accounted for some of the increase.

  Two people in each treatment arm had signs of grade 3/4 liver toxicity, while 13 taking nevirapine and 10 taking efavirenz had grade 1/2 liver toxicity.

  This small study surely doesn't close the book on the relative merits of the two NNRTIs. But a big trial organized by the Amsterdam-based International Antiviral Therapy Evaluation Center (IATEC) may. The 2NN Study hopes to recruit 1200 people and randomize them to take d4T/3TC plus nevirapine twice daily, nevirapine once daily, efavirenz, or nevirapine plus efavirenz. So far 400 people have signed up.

• A vote for first- (and second-) line PIs. At this point in antiretroviral history, clinicians have two good NNRTIs they might use in a first-line regimen and several good PIs, all of which can be boosted by ritonavir. And most clinicians surely have their favorite PI-NNRTI sequencing schemes, depending on the needs of the individual being treated. But that hasn't stopped the inquisitive from poking through the entrails of outcome data to see if they can divine the best way to juggle drug classes.

• An interesting record review by David Butcher (Chase Brexton Health Services, Baltimore) and colleagues at Abbott found that first-line PI regimens last longer than first-line NNRTI regimens [abstract 1542]. And second-line PI regimens following an NNRTI last longer than NNRTIs following a PI. Butcher used the 1705-person database of Clinical Partners, a San Francisco outfit that feeds HIV data to managed care groups. He picked out 65 treatment-naive people who started with a PI and switched to an NNRTI, and 68 naive people who started with an NNRTI and switched to a PI. Pretreatment viral loads were equivalent in the two groups at about 12,000 copies/mL, and Butcher excluded people who took verboten combos like AZT/d4T.

  People who began with a PI stayed with the regimen for a median of 378 days, compared with 197 days for folks starting with an NNRTI (P = 0.001). Those who switched from an NNRTI to a PI stayed with the PI combo for a median of 384 days, compared with 188 days for those taking an NNRTI after a PI (P = 0.008). Among 45 people who began a first-line PI with a viral load above 400 copies/mL, 20 (44 percent) pushed their viral load below 400 copies, and two of those 20 (10 percent) had a rebound. Among 53 people with viral loads over 400 copies/mL when they began a first-line NNRTI, 14 (26 percent) scored a sub-400 reading, and 10 of those 14 (71 percent) rebounded.

  Butcher said his results suggest that clinicians should use PIs first when starting antiretrovirals. Of course an NNRTI advocate could find grounds to quibble. A small but meaningful proportion of the PI-first group, 9 percent, started therapy not with one protease drug, but with two. So this dual-PI group could have given at least a small virologic edge to the total PI group, because everyone starting with an NNRTI took only one.

  Nearly everyone who begin treatment with an NNRTI , 93 percent, took nevirapine. Results of the nevirapine studies described above suggest this drug can at least hold its own against efavirenz, indinavir, and nelfinavir, but blinded trials or even big trials have not proved that. Regardless of what one makes of these nevirapine studies, it would be interesting to see what would happen if Butcher redid this kind of a chart comparison focusing on people who began treatment with only single PIs versus efavirenz. It must be said, though, that efavirenz got plenty of use as a second-line NNRTI in Butcher's analysis (43 percent took it versus 57 percent on nevirapine), yet Team NNRTI didn't do great in the second-round contest either.

  Finally, this kind of analysis steers clear of the patchwork of factors that can skew results. Who were these patients anyway? What kind of support did they get from their health providers? How many of those providers were managed care physicians who see five or six people with HIV infection a year and don't particularly want to see more? Certainly the virologic success rate in the NNRTI group--26 percent--is abysmally atypical and raises questions about whether all those people starting nevirapine were eased through early rashes, for example, or whether their physicians opted for a quick-and-easy switch to a PI instead.

Return to Article Contents

Three treatment response riddles

In ICAAC's bustling poster halls, four groups addressed three perennial questions in antiretroviral management:

• Do complete responders differ clinically from partial responders?
• Should treatment start during primary infection?
• Should people respond to a potent regimen in 24 weeks?

The answers proffered were not much, yes, and not necessarily.

• The pluck of the plateau responder. All HIV clinicians care for people who get their viral load under 10,000 copies/mL but never reach the charmed kingdom of undetectability. How do those people do over the years compared with complete virologic responders? With colleagues at Rush Medical College, Chicago, and the University of Colorado, Denver, A. Tenorio planned a study to find out [abstract 565].
  
  

  The Chicago-Denver team compared 25 plateau responders with 10 full responders and with 12 people who suffered virologic failure. People qualified for the plateau group if their viral load lingered between 50 and 10,000 copies/mL for more than six months. Full responders had a viral load below 50 copies/mL for more than six months. And people in the virologic failure group had viral loads above 10,000 copies/mL for more than six months. In fact plasma viral loads remained stable in these three brackets through 42 months of follow-up. The plateau responders had pretreatment and nadir CD4⁺ counts similar to the full responders, but they had been infected and treated longer (Table 6). The virologic failure group had much lower baseline and nadir CD4⁺ counts.

  Table 6. Baseline numbers for three groups with different responses to antiretrovirals

  Group	n	Duration of HIV infection (m)	CD4+ baseline (cells/mm3)	CD4+ nadir (cells/mm3)	Antiretroviral duration (m)
  Virologic failure (RNA >10,000 >6 m)	12	84	43	15	59
  Plateau responder (RNA 50-10,000 >6 m)	25	98	384	235	65
  Full responder (RNA <50 >6 m)	10	67	438	293	49
  Source: A. Tenorio, abstract 565.

  During 42 months of follow-up, Tenorio found that all 10 full responders remained free of opportunistic diseases, as did 22 of 25 plateau responders, a nonsignificant difference (P = 0.54). By this measure both groups differed significantly from the virologic failure group, all of whom had a new opportunistic disease (P = 0.02 versus full responders and P = 0.04 versus plateau responders). The plateau and full responders also changed regimens significantly less often than the failure group (P < 0.01 for both comparisons). And the two responder groups had significantly higher CD4⁺ jumps from baseline than the failure group (+295 cells/mm³ for full responders, +207 cells/mm³ for plateau responders, and +46 cells/mm³ for the failure group; P < 0.01 for either responder group versus the failure group).

  Five of 10 full responders had a lymphoproliferative response to p24 antigen (stimulation index >10), as did eight of 25 plateau responders, a nonsignificant difference (P = 0.44). Again, both responder groups did significantly better than the failure group, none of whose members churned up a response to p24 (P = 0.01 for full responders and P = 0.04 for plateau responders versus the failure group). In a surprise, the plateau responders acquired significantly fewer new primary or secondary resistance mutations during follow-up than the full responders (mean 1.0 versus 2.33, P = 0.03).

  Tenorio and co-workers concluded that "complete virus suppression may not be necessary for HIV-specific immune reconstitution and/or preservation." The intriguingly slow evolution of resistance mutations in the plateau group may be explained, these researchers suggested, by the group's maintenance of low-level viremia. At the same time, though, the plateau responders had significantly fewer primary and secondary resistance mutations when first genotyped (mean 6.5 versus 13 for full responders, P< 0.001).

• Gains from treating primary infection. Robert Geise (University of Washington, Seattle) drew a big crowd to his poster on immediate treatment during acute infection versus delayed intervention [abstract 808]. Apparently many clinicians are still digging for clues to help solve the mystery of when to start. Geise offered many such clues, and most favored early treatment.
  
  

  Of course this was not a randomized study. Geise compared 15 people diagnosed within 120 days of infection who wanted antiretrovirals fast and 30 people also diagnosed early who did not start treatment promptly. The early group began their antiretrovirals within a median of 50.5 days of infection, compared with 793 days for the delayed-intervention group (P < 0.0001).

  During two years of follow-up after treatment began, the early group scored significantly better than the delayed group on several measures. Their average absolute CD4⁺ count rose to 953 cells/mm³, compared with 701 cells/mm³ in the delayed group (P = 0.016). The early group had a higher CD4⁺ percent (44.3 versus 33.0 percent, P = 0.0005) and a better CD4⁺/CD8⁺ ratio (1.55 versus 0.88, P = 0.0021).

  It's true that the early group began treatment with a higher average CD4⁺ count (628 versus 523 cells/mm³, P = 0.076). But that early lead let them push their absolute counts higher, because the two groups had equivalent CD4⁺ gains from baseline. Geise noted that "immune replenishment seems to plateau at similar levels" above baseline regardless of when antiretroviral therapy started.

  None of the 15 people in the early group endured a viral rebound above 500 cells/ mm³, whereas eight of 30 in the delayed group did (risk difference 26.7 percent, P = 0.0274). But that difference reflects the different points in the epidemic when these groups became infected. People in the delayed group were generally infected earlier, and some of them began treatment with only one or two antiretrovirals. All of them ended up taking highly active antiretroviral therapy (HAART), but everyone in the early group began with PI-based HAART. As Geise pointed out, this so-called cohort effect favors the early treatment group because they all started with robust regimens. When Geise eliminated the monotherapy and dual therapy patients from the delayed intervention group, only two of 12 suffered a virologic rebound, and the difference between this subgroup and the early group was not statistically significant.

  What about clinical differences between the groups? There weren't any big ones, as might be expected in a cohort with such high CD4⁺ counts. In fact, the early-treatment group did a little worse, with significantly more GI symptoms than the delayed cohort (P = 0.028), perhaps a treatment side effect.

  Readers of this journal all know the downside of immediate treatment. Some people who start antiretrovirals during acute infection do have virologic failures. If they don't, they can probably look forward to decades of treatment at a high cost and a high risk of long-term side effects. The flip side of those arguments, advanced by Bruce Walker and Eric Rosenberg at Massachusetts General Hospital in Boston, is also well known: Immediate treatment offers a shot at preserving a highly tuned immune system and possibly of prolonged viral control without drugs (see "STI news" below).

  More evidence favoring treatment during primary infection came on a poster that wins hands down for the most glittering list of authors [abstract 804]. Headed up by the eminent microbiologists Klara Tenner-Racz and Paul Racz (Bernhard-Nocht Institute, Hamburg, Germany), this cast also featured Hans-Jürgen Stellbrink, Mike Youle, Brian Gazzard, David Cooper, Schlomo Staszewski, Paolo Vernazza, and Luc Perrin. No, they weren't all on hand to discuss their findings, but Li-Ean Goh (Glaxo Wellcome) was.

  The poster described a substudy of QUEST, the multinational effort to identify infected people in the first days and weeks of infection and to treat them with AZT, 3TC, abacavir, and amprenavir. Later they get an HIV vaccine and then may stop treatment under close observation. This study compared lymph node biopsies from 15 people with acute infection and eight people with chronic infection. The investigators wanted to learn how quickly HIV nests down in lymphoid tissue, HIV's primary roost and breeding site.

  In situ hybridization showed "no or only minimal storage of HIV" in lymph nodes during primary infection. Lymph nodes from people with chronic infection had an average 3.42 RNA-positive cells/mm², compared with 1.48 cells/mm² in acutely infected individuals (P = 0.098). Goh and colleagues measured significantly lower quotients of intracellular HIV DNA and RNA in lymphomononuclear cells (P < 0.001 for both) of the individuals with primary infection.

  The authors concluded that their findings "support HAART intervention at the earliest stage (primary infection), when viral trapping is still marginal." Stopping HIV from overwhelming lymphoid tissues, they proposed, could limit viral spread to other sites.

• Not all post-24-week responses tardy. "When to switch" follows "when to start" as a treatment puzzler only by a hair's breadth. Researchers from Chelsea and Westminster Hospital and St. Stephen's Centre in London posted a study that suggested most guidelines are too quick on the switch trigger when people start treatment with high viral loads [abstract 547].

• Graeme Moyle and colleagues marshaled data from the 1266-person open-label contest between efavirenz and indinavir plus AZT/3TC. A third group took efavirenz plus indinavir. All will recall that efavirenz/AZT/3TC corralled HIV best through 72 weeks of treatment. But a little-known result, plucked from the data haystack by Moyle, is that a fair proportion of sub-50-copy responders in every treatment arm did not reach that milestone in the guideline-endorsed 24 weeks. Among people assigned to efavirenz/ AZT/3TC, 6 percent needed more than 24 weeks to go sub-50, as did 13 percent taking efavirenz/indinavir and 17 percent taking indinavir/AZT/3TC.

  What factor predicted who needed more time to earn their "undetectable" badge? Not age, gender, race, prior AZT, or baseline CD4⁺ or CD8⁺ count. The only independent predictor was baseline viral load (P < 0.001).

  Moyle concluded that guidelines recommending a sub-50 viral load by week 24 "appear reasonable" for most people starting treatment with fewer than 100,000 RNA copies/mL. But many people who begin antiretrovirals with higher viral loads "may need at least 36 weeks before a viral load below 50 copies/mL is achieved."

  Return to Article Contents

  New mysteries with AZT and d4T mutations

  Three studies found that typical AZT mutations (which may also arise in AZT-naive people taking d4T) do not invariably hamper virologic response to certain new regimens. On the contrary, a study by Victoria Johnson (University of Alabama, Birmingham, USA) found that people with AZT mutations did better virologically than people without those mutations when starting an indinavir combination [abstract 2111].

  Johnson reached that conclusion by analyzing virologic responses in ACTG 370, a randomized comparison of continued 3TC or a switch to the NNRTI delavirdine in people taking AZT/3TC, d4T/3TC, or ddI/3TC. Everyone in the trial also started taking indinavir. Among people who had been taking AZT/3TC, 50 percent had one or more AZT-associated mutations. But 38 percent of those who had been on d4T/3TC also had one or more mutations traditionally associated with AZT, which many now call thymidine analog mutations (TAMs). The TAMs rate difference between the AZT/3TC group and d4T/3TC group was not statistically significant (P = 0.34). More people in the AZT/3TC group than in the d4T/3TC group had two or more TAMs, and this difference approached statistical significance (20 percent for AZT/3TC, 9 percent for d4T/3TC, P = 0.069). With 306 patients, this is the biggest study so far to confirm TAMs in d4T-treated people with no AZT experience.

  After 48 week of treatment with the new regimen, people randomized to take AZT, indinavir, and delavirdine did best virologically, a finding published earlier this year.¹⁰ That was not a surprise, but Johnson's results were. In a multivariate analysis to figure what factors correlate with failure (viral load > 200 copies/mL) at 48 weeks, Johnson found two: Every 100 more CD4⁺ cells/mm³ at baseline cut the risk of failure 40 percent (P = 0.014), and one or more TAMs at randomization in ACTG 370 sliced the risk 64 percent (P = 0.048). That mutation association remained significant when statisticians controlled for baseline viral load and CD4⁺ count (odds ratio 0.23, P = 0.01).

  What explains the apparently protective effect of TAMs in this study population? Johnson could only speculate. One possibility involves the recent discovery that abacavir- or AZT-resistant virus is hypersusceptible to NNRTIs.^11,12 As it turns out, people in ACTG 370 who had baseline AZT mutations and were randomized to the delavirdine arm enjoyed better virologic control than those randomized to 3TC. That difference was not statistically significant, but it could explain some of the benefit Johnson found with these NRTI mutations.

  But NNRTI hypersusceptibility must not explain this phenomenon entirely, because a French group found good 80-week virologic responses among 170 NRTI-experienced people who started a regimen not including a nonnucleoside [abstract 696]. People in the open-label NOVAVIR study were randomized to take indinavir with either AZT/3TC or d4T/3TC. Everyone had taken AZT, ddI, and/or ddC for more than six months (median 19.3 months), but all were naive to d4T, 3TC, and PIs, reported Veronique Joly for ANRS, the French clinical trials group. When randomized to the triple regimen, 69 percent had more than two AZT mutations and 17 percent had one or two. The median baseline CD4⁺ counts stood at 291 cells/mm³ and the median viral load at 22,900 copies/mL.

  After 80 weeks, 76 percent in the d4T arm had a viral load below 500 copies/mL, as did 77 percent in the AZT arm. Fourteen taking d4T and 15 taking AZT met the study definition of virologic failure, a viral load above 5000 copies/mL after eight weeks of treatment. CD4⁺ gains were similar in the two arms, as were rates of toxicity. Joly and colleagues concluded that d4T offers no advantage over AZT when people with heavy AZT experience start a PI regimen. And AZT mutations didn't dim the response in either treatment arm.

  Another study found that virus with three or more AZT mutations, plus the 184V 3TC mutation, often remains susceptible to d4T and abacavir [abstract 1271]. Christian Michelet (Pontchaillou Hospital, Rennes, France) and colleagues scrutinized 21 viral isolates from people treated for more than 36 months with AZT and more than six months with 3TC. All five isolates with two or fewer AZT mutations plus 184V remained fully susceptible to d4T, and four of the five remained fully susceptible to abacavir. The fifth had "intermediate" sensitivity to abacavir. Among 14 isolates with three or more AZT mutations plus 184V, seven remained fully susceptible to d4T while five had "intermediate" susceptibility, and nine remained fully susceptible to abacavir while four had "intermediate" susceptibility.

  Michelet used these results to argue for phenotypic testing when planning a new regimen. "Usually it is considered that the presence of three or more zidovudine mutations plus a lamivudine mutation induces a decrease in abacavir efficacy," he noted. But not in this study. Knowing the genotype of these study participants would have led many to shun abacavir in the next regimen, even though 63 percent of isolates remained sensitive to that drug.

  A survey of several hundred HIV clinicians at an ICAAC interactive session showed, however, that most still rely on genotyping (Table 7). Indeed, comparing results of the 2000 ICAAC survey with answers in a similar 1999 survey suggests only marginal increases in numbers of clinicians who use phenotyping exclusively or together with genotyping. But the number who use genotyping alone rose 11 percent over the past year, while the number who use neither assay fell 16 percent. Of course these numbers come from an unscientific survey, but they suggest few new phenotyping enthusiasts in the past 12 months.

  Table 7. Resistance test preferences among HIV clinicians surveyed at the 1999 and 2000 ICAACs*

  Preferred resistance test	1999 ICAAC survey	2000 ICAAC survey
  Genotyping	57	68
  Phenotyping	4	6
  Both	15	19
  Neither	24	8
  *The 1999 sample included 598 clinicians at an interactive session. The number responding to the 2000 survey during a similar session was not available. But the audience looked to be about as big as the one in 1999. Sources: Charles Boucher, 1999 ICAAC presentation 1368, and Douglas Richman, 2000 ICAAC presentation 1876.

  Return to Article Contents

  Who's better at calling resistance: experts or the virtual phenotyper?

  Despite the popularity of genotyping, its reliability--even when interpreted by resistance savants--remains the focus of worried scrutiny. Just how accurate and consistent are these experts? Victoria Johnson took aim at that loaded question in a clever study comparing "expert interpretation" with the cool calculation of Virco's virtual phenotyping system [abstract 2118].

  Johnson recruited three other University of Alabama resistance wonks, and this foursome rated genotypes of 69 patient isolates for susceptibility to 15 antiretrovirals. Two weeks later this virologic quadrumvirate again scored the same isolates, without referring to their earlier decisions, to see if their answers changed. Then Johnson compared the experts' calls with the virtual phenotype, that is, the susceptibility of each isolate to the 15 drugs determined by culling similar isolates from Virco's database and reckoning their actual phenotypes against those drugs.

  The good news is that experts--at least University of Alabama experts--agree most of the time. They agreed completely 76 percent of the time when determining whether a specific genotype meant resistance to the 15 drugs. The best concordance naturally came on calls for resistance to 3TC (91 percent) or the NNRTIs efavirenz and delavirdine (94 percent for both). But the overall score got dragged down by wobbly agreement on resistance to d4T (57 percent), abacavir (52 percent), and nelfinavir (42 percent), the worst scores recorded.

  How did these resistance whizzes do when rating the same viral samples two weeks later? Once more, pretty well. Agreement measured by Cohen's kappa statistic () ranged from a low of 0.81 to a high of 0.91 for the four experts. Perfect agreement is 1.0. In this "intraexpert" analysis, the drugs that scored the lowest with each expert were ddI (0.7), nelfinavir again (0.55), and amprenavir (0.53 for one judge and -0.04 for another). That negative score means this expert completely reversed course on whether certain genotypes implied resistance to amprenavir.

  Johnson's resistance panel also did well in the faceoff with Virco's virtual phenotype, with a of 0.71 (range 0.66 to 0.76). In a univariate analysis, the virtual phenotype and the genotypic consensus of the Birmingham quartet predicted time to virologic failure among 58 people at hazard ratios of 0.69 for the phenotyper (P = 0.03) and 0.67 for the quartet (P = 0.008). The individual resistance calls of three of the four experts also predicted time to failure in a statistically significant fashion, and the fourth expert came close (P = 0.07). (Yes, that was the expert who blew the amprenavir call.) In contrast, antiretroviral history was not a significant predictor of time to failure in this analysis.

  Long lines snaked from the audience mikes as soon as Johnson finished. So, wondered Stephen Deeks, does the time-to-failure analysis mean virtual phenotyping has no advantage over expert opinion? Not necessarily, Johnson replied, since it's only a univariate analysis and since one could use the same results to argue that virtual phenotyping gets it right as often as expert opinion and will be more accessible to clinicians without an expert to tap.

  Calvin Cohen (Community Research Initiative of New England, Boston) registered surprise that treatment history did so poorly in predicting time to failure. Johnson said that was probably because "history" here simply meant a list of drugs taken, not the thorough history a good clinician would elicit.

  Andrew Zolopa (Stanford University, Stanford, California, USA) wondered whether Johnson used some tactic like shuffling the genotype records before the experts went through them the second time, just to make it harder for them to remember their first responses. She assured him that no shuffling was necessary, noting from her experience that it's impossible to remember how you scored 69 isolates for susceptibility to 15 drugs after two weeks.

  Johnson's frank conclusion in her printed abstract was that "interpretation of genotypic resistance testing results varies by interpreter and method of interpretation." It's probably safe to assume that clinicians who don't have the latest findings on K65R and Y181C tattooed onto their temporal lobes would produce even more variable results in this kind of test and might be soundly thrashed by the virtual phenotyper.

  Several studies catalog the difficulties in interpreting both genotypes and phenotypes.¹³ Johnson's findings bolster the notion that clinicians who plan therapy with resistance tests need all the help they can get. McGill University's Mark Wainberg considered just one confounder of accurate interpretation in his overview on resistance [presentation 2119]. "What is the interaction between mutations going to mean for our ability to interpret genotype as experts," he wondered, "or indeed as computers?"

Return to Article Contents

STI news--from good, to cautionary, to sobering

"In my end is my beginning," T.S. Eliot professed oxymoronically in The Four Quartets. Of course Eliot had big ideas in mind--time, being, death, "that the future is a faded song"--all the big ideas, in fact, that might go through the mind of a reflective person living with a dangerous disease. Lots of those people, living with HIV, would like to put Eliot's transcendent notion to practical use. They'd like to see an end to those things that sustain their life as the beginning of a better life. They'd like to stop taking antiretrovirals. Or at least interrupt them, in a structured way.

It's a fantastic idea whose allure has proved irresistible to some. But it's a hard thing to study, and so far it's been harder to make it work. Two important STI studies went into print right after ICAAC, and two capped the meeting's late-breaker session on HIV. The results ranged from good, so far (in a highly selected population), to cautionary (in monkeys), to sobering (in the eyes of one researcher).

The good news came in the first peer-reviewed report on Bruce Walker's and Eric Rosenberg's primary infection cohort at Massachusetts General Hospital in Boston.¹⁴ They published early results on eight people who began HAART during primary infection, which Walker believes will save enough HIV-specific CD4⁺ cells to keep replication in check, at low levels of viremia, without drugs. The eight stopped treatment once or twice. Despite viral rebounds, all achieved "at least a transient steady state off therapy with [a] viral load below 5000" copies/mL. Five of the eight elected not to restart their meds and had maintained sub-500 viral loads for five to 8.7 months at last report. All eight had evidence of HIV-specific cytotoxic T lymphocytes and good CD4⁺-cell responses.

"Our data indicate that functional immune responses can be augmented in a chronic viral infection," Walker and colleagues concluded, "and provide rationale for immunotherapy in HIV-1 infection." Walker also wants to make sure people don't distort the meaning of these results. "I've taken care of patients who've had a lot of false hopes" in the course of the epidemic, he told The Wall Street Journal. "I don't want to contribute to that."¹⁵

The obvious caveat is that everyone in this cohort began treatment right away, most within one month of infection, the Boston team believes. Even people treated in the first few months of infection, for example, may not be able to mount this kind of immune response with treatment breaks. So far studies of STIs in people with chronic infection have turned up few who controlled HIV for months without medicine.

Franco Lori (RIGHT, Pavia, Italy) underscored this distinction between acute and chronic infection in a study of SIV-infected monkeys [abstract L-17]. The RIGHT team looked at STIs in two groups of monkeys: animals who had been infected for more than 14 months, and monkeys who had been infected for only six weeks--macaque counterparts of the Mass General cohort. Some of the acutely infected animals got steady treatment with PMPA, ddI, and hydroxyurea, while others had four three-week treatment breaks sandwiched between three weeks of therapy. The chronically infected monkeys followed the same drug interruption schedule.

The acute STI group mimicked certain traits of Walker and Rosenberg's human acute STI cadre. Their viral rebounds waned after each interruption until all six animals in the group effectively stifled replication after the fourth treatment break (Table 8). That control of viremia correlated with a zesty surge in SIV-specific CD8⁺ T cells. The chronically infected monkeys had no such response. Virus rebounded in all four chronically infected animals at essentially the same rate after the second, third, and fourth drug breaks. The acutely infected STI group and the acutely infected continuous therapy group had equivalent SIV-specific T-cell responses. But CD4⁺ percents stayed steady in the acute STI group while dropping in the continuous acute group.

Lori's comparison of STIs in acutely and chronically infected macaques shows what STI studies in humans have already demonstrated: Reaping immunologic benefits from treatment breaks gets much tougher after HIV or SIV infection has settled into a groove.

Michael Saag (University of Alabama, Birmingham, USA) looked back at his HIV clinic's cohort to see what happened to people who stopped treatment for any reason then started again [abstract L-18]. He limited the analysis to people who abandoned HAART for at least 30 days then resumed treatment for at least 30 days. The clinic had 78 people who met those criteria and whose records included the CD4⁺ and viral load data Saag wanted to study. The median interruption measured 67 days, and median follow-up after restarting HAART was 173 days. Most of these people stopped treatment because of side effects, though 11 stopped because of virologic failure and 11 because they ran out of money to pay for their drugs.

After treatment resumed, 46 of the 78 (59 percent) managed to come within 90 percent of their prebreak CD4⁺ count, and 56 (72 percent) got their viral loads down to or under their prebreak level. In his published abstract, Saag characterized these findings by writing that "the majority" of the treatment interrupters "were able to achieve successful [viral load] and CD4⁺ outcomes when HAART therapy was reinitiated."

Some might give the numbers a darker interpretation: Four of 10 people failed to get close to their earlier CD4⁺ count. Since the median postbreak CD4⁺ count was 230 cells/mm³, that means a fair proportion of those people ended up with cell counts in the sub-200 danger zone. Clinicians caring for people who want to stop their drugs for a few months may want to share that result with them.

A more theoretical, but no less illuminating, STI study came from Sebastian Bonhoeffer and colleagues at the Friedrich Miescher Institut in Basel and the Aaron Diamond AIDS Research Center in New York.¹⁶ They used population dynamics models to see how treatment breaks may affect virus-fighting T cells, latent viral reservoirs, and the emergence of drug resistance.

This approach told them STIs lead to "transient or sustained virus control" only if they engender more immune effector cells than susceptible target cells during therapy. The risk of repopulating latent pools or arousing resistance "may be small" if the viral load stays "considerably below baseline." If it doesn't, "both these risks increase dramatically."

One aspect of treatment breaks that worries some seasoned clinicians is how long different antiretrovirals stay in the body if a person stops taking all drugs in a regimen at once. Efavirenz is a particular concern because its half-life outdistances those of other antiretrovirals. So in theory simultaneously stopping all parts of an efavirenz regimen will yield a short stretch of efavirenz "monotherapy" during which the drug hangs on at subtherapeutic levels that promote resistance. That the maker of efavirenz, DuPont Pharmaceuticals, set out to study this specific question is one sign of the intense interest in STIs.

DuPont's Nancy Ruiz addressed the question by comparing people who interrupted efavirenz with people who didn't during a big trial [abstract 479]. The trial protocol required clinicians to stop all drugs in a regimen at the same time if side effects cropped up. Among those who interrupted at viral loads below 400 copies/mL, 73 percent eventually ranked as sub-50-copy responders. That response matched the 72 percent sub-50 rate among study participants who never interrupted efavirenz. But among people who interrupted the NNRTI at a viral load above 400 copies/mL, only 45 percent hit the sub-50 mark.

Although Ruiz said these results have no implications for STIs, one might be forgiven for assuming DuPont would not have spent time and money on this analysis if efavirenz had not been singled out as a tricky drug when planning STIs.

As this article approaches its end, it also returns to its beginning--that first pop quiz question on how many people have asked their clinicians for an STI and how many clinicians agreed. Roy Gulick (Cornell University, New York) posed the question during an ICAAC interactive session attended by several hundred physicians equipped with digital keypads [presentation 1878].

To save you the trouble of flipping back to the opening page, here's the answer again: 58 percent said they treated at least one person who asked for an STI "for viral or immune benefit." And 31 percent in the audience obliged their patients.

"That's a sobering statistic," said Gulick, who concluded his half-hour STI review by maintaining that "it is too early to recommend routine use of STIs in any clinical setting." But what if your patient insists, or what if you agree an STI is a reasonable strategy for that person? Gulick had this advice:

1. STRONGLY consider a clinical trial.
2. Explain the risks in detail.
3. Get a signed informed consent.
4. Be ready with OI prophylaxis.
5. Rethink that clinical trial.

Return to Article Contents

References and Notes

*Abstracts from the 40th Interscience Conference on Antimicrobial Agents and Chemotherapy (ICAAC) are online at http://www.asmusa.org/mtgscr/40icaac.htm.

1. Hicks C, King M, Brun S, et al. ABT-378/ritonavir (ABT-378/R) in antiretroviral naive HIV+ patients: 48 weeks. Presented at: Seventh European Conference on Clinical Aspects and Treatment of HIV Infection. October 23-27, 1999. Lisbon. Abstract L-585.

2. The twice-daily doses of lopinavir/ritonavir were 200/100 mg, 400/100 mg, and 400/200 mg. About half of the study participants got the 400/100 mg dose used in the phase III trial.

3. Albrecht M, Katzenstein D, Bosch R, et al. ACTG 364: virologic efficacy of nelfinavir (NFV) and/or efavirenz (EFZ) in combination with new nucleoside analogs in nucleoside experienced subjects. Presented at: 6th Conference on Retroviruses and Opportunistic Infections. January 31-February 4, 1999. Chicago. Abstract 489.

4. Mascolini M. Revisiting resistance. IAPAC Monthly 2000;6:221-222. Available at: http://www.iapac.org/conferences/resistancemm008m.html. Accessed October 15, 2000.

5. Durant J, Clevenbergh P, Halfon P, et al. Drug-resistance genotyping in HIV-1 therapy: the VIRADAPT randomised controlled trial. Lancet 1999;353:2195-2199.

6. Winters MA, Schapiro JM, Lawrence J, Merigan TC. Genotypic and phenotypic analysis of the protease gene in HIV-1-infected patients that failed long-term saquinavir therapy and switched to other protease inhibitors. Antiviral Ther 1997;2(suppl 5):25-26. Abstract 17.

7. Meyer PR, Matsuura SE, So AG, Scott WA. Unblocking of chain-terminated primer by HIV-1 reverse transcriptase through a nucleotide-dependent mechanism. Proc Natl Acad Sci USA 1998;95:13471-13476.

8. Dubé M, Aqeel R, Edmondson-Melançon H, et al. Effect of initiating indinavir therapy on glucose metabolism in HIV-infected patients: results of minimal model analysis. Antiviral Ther 1999;4(suppl 2):34. Abstract 28.

9. Noor M, Lo J, Mulligan K, et al. Metabolic effects of indinavir in healthy HIV-seronegative subjects. Antiviral Ther 2000;5(suppl 5):8. Abstract 10.

10. Kuritzkes DR, Bassett RL, Johnson VA, et al. Continued lamivudine versus delavirdine in combination with indinavir and zidovudine or stavudine in lamivudine-experienced patients: results of Adult AIDS Clinical Trials Group protocol 370. AIDS 2000;14:1553-1561.

11. Haubrich R, Whitcomb J, Keiser P, et al. Non-nucleoside reverse transcriptase inhibitor viral hypersensitivity is common and improves short-term virological response. Antiviral Ther 2000;5(suppl 3):69. Abstract 87.

12. Whitcomb J, Deeks S, Huang W, et al. Reduced susceptibility to NRTI is associated with NNRTI hypersensitivity in virus from HIV-1-infected patients. Presented at: 7th Conference on Retroviruses and Opportunistic Infections. January 30-February 2, 2000. San Francisco. Abstract 234.

13. Some recent examples of such studies are reviewed in the August 2000 issue of IAPAC Monthly. See, Revisiting resistance. IAPAC Monthly 2000;6:225-229. Available at: http://www.iapac.org/conferences/resistancemm008m.html. Accessed October 15, 2000.

14. Rosenberg ES, Altfeld M, Poon SH, et al. Immune control of HIV-1 after early treatment of acute infection. Nature 2000;407;523-526.

15. Schoofs M. Newly infected AIDS patients may squelch virus drug-free. Wall Street Journal. September 28, 2000.

16. Bonhoeffer S, Rembiszewski M, Ortiz GM, et al. Risks and benefits of structured antiretroviral drug therapy interruptions in HIV-1 infection. AIDS 2000;14:2313-2322.

Return to Article Contents

Mark Mascolini writes about HIV infection (mailmark@ptd.net).

001210
IA001201

Copyright © 2000 - International Association of Physicians in AIDS Care. All rights reserved. http://www.iapac.org

ÆGiS is made possible through unrestricted grants from Roxane Laboratories, Inc., iMetrikus, Inc., the National Library of Medicine, and donations from users like you. Always watch for outdated information. This article first appeared in 2000. This material is designed to support, not replace, the relationship that exists between you and your doctor.

ÆGiS presents published material, reprinted with permission and neither endorses nor opposes any material. All information contained on this website, including information relating to health conditions, products, and treatments, is for informational purposes only. It is often presented in summary or aggregate form. It is not meant to be a substitute for the advice provided by your own physician or other medical professionals. Always discuss treatment options with a doctor who specializes in treating HIV.

Copyright ©1990, 2000. ÆGiS & the Sisters of Saint Elizabeth of Hungary. All materials appearing on ÆGiS are protected by copyright as a collective work or compilation under U.S. copyright and other laws and are the property of ÆGIS and the Sisters of Saint. Elizabeth of Hungary, or the party credited as the provider of the content.