Important note: Information in this article was accurate in June 2000. The state of the art may have changed since the publication date.Spring came suddenly to Chicago on April 13 and 14, vouchsafing two preposterously balmy days for what will probably be the last International Workshop on Salvage Therapy for HIV Infection.
No, it's not that ardent research or innovative regimens have solved the salvage riddle. Indeed, version 3.0 of this workshop confirmed yet again the difficulty of reining in a runaway virus that has stampeded through just a couple of combos. Rather, the organizers pretty much agreed in a wrap-up session to strike salvage from the title of future gatherings, in favor of something less grim. "Management of Treatment-Experienced Patients," perhaps.
Who can blame them? Salvage remains moored fast to its first hoary meaning, resurrecting the rotting hulks of sunken ships. Not a very wholesome correlate. For that reason the move to change the meeting's name seems more than semantic preciosity. Language, after all, shapes our realities, if it does not determine them. So saying someone needs "salvage therapy" does not imply--it proclaims--a desperate enterprise. And should desperate measures be invoked when a person starting HIV therapy endures a viral rebound to, say, 1000 copies/mL?
Maybe.
That possibility emerged as a core conundrum at this workshop. To be sure, no one endorsed mega-HAART for an unconfirmed viremic spike. No one evinced wide-eyed terror at a 3-log viral load. No one even echoed guideline advice to switch drugs quickly when virologic response turns wobbly. The point--made time and again in study upon study--is that the first treatment failure all too often heralds a second, and a second a third. And a third failure starts the search for desperate measures.
So perhaps that first sign of a regimen gone wrong should sound a loud alarm--not an alarm that means a person with a first viral rebound has stepped forever beyond rescue's grasp, but an alarm that means failure to pinpoint the reasons for failure will condemn that rebounder to a certain series of botched treatments and bungled opportunities. So proposed meeting cochair John Mellors, MD (University of Pittsburgh) and others at this workshop.
Suddenly salvage--with all its unglamorous associations--sounds hardly scary enough to describe the urgency that attends unsuccessful therapy. And let us not forget that the word has evolved from its baleful Plutonian connotation to a more nuanced and figurative suggestion of a prudently husbanded prize. A latter-day (1881) citation in the Oxford English Dictionary offers this example: "I shall retain a certain salvage of wisdom."1 If the Third International Workshop on Salvage [!] Therapy for HIV Infection offered no immediate solutions to antiretroviral failure, it did drop a few hints on how wisdom follows failure in the reflective mind.
The workshop readily evidenced one notable advance in salvage therapy: Carefully planned controlled salvage trials--nonexistent just a few years back--have begun to swipe at the many-headed Hydra of treatment failure. Four prospective studies, three of them randomized, took different approaches to treating individuals with moderate to heavy antiretroviral experience (Table 1). The studies confirmed that combining more drugs to which people remain susceptible pays virologic dividends. But those dividends were invariably modest for heavily pretreated individuals and, in three of four studies, they depended on sacrificing a nonnucleoside reverse transcriptase inhibitor (NNRTI) in regimens that didn't work for lots of people.
Everyone got the now-withdrawn nucleotide adefovir dipivoxil at 60 mg daily plus standard doses of abacavir and efavirenz. On the basis of PI experience, investigators "selectively" randomized participants to take amprenavir plus
After 24 weeks, in an intent-to-treat (missing-data-equal-failure) analysis, 35 percent in the dual-PI arms had a viral load below 200 copies/mL, compared with 23 percent in the amprenavir/placebo arm (P = 0.002). NNRTI experience proved to be decisive in 24-week virologic success. Whereas 43 percent of NNRTI-naive people in any arm attained a viral load below 200 copies/mL by week 24, only 16 percent of NNRTI-experienced individuals did so (P < 0.001). The inevitable question raised by this trial is whether one should risk the resistance-vulnerable NNRTI class even in nonnuke-naive people when fewer than half are likely to respond if their other treatment experience approximates this group's. Also unsurprising was the finding that previous treatment with only one PI betokened a better 24-week virologic outcome than previous treatment with two or more PIs (P = 0.058).
It's tough to sort out whether saquinavir, indinavir, or nelfinavir made the best mate for amprenavir in a situation like this. Probably the question is moot anyway, since research done since the ACTG planned this trial shows that the only PI not studied, ritonavir, seems the best choice because it moderates amprenavir's tendency to lower efavirenz levels (see below). Saquinavir, indinavir, and nelfinavir all pushed plasma viremia down about 1.1 logs in this study, compared with a 0.8-log drop in the placebo arm.
ACTG 398 also underscored the difficulty of analyzing pharmacokinetic variables in multidrug regimens. Amprenavir levels turned out to be significantly lower in the saquinavir and placebo arms than in the indinavir and nelfinavir arms (P = 0.009). But that difference didn't line up neatly with the virologic and CD4+ outcomes.
In Chicago Gulick told what happened to people who continued treatment through 48 weeks. Among the 30 percent who had a viral load below 500 copies/mL at week 16, 55 percent still had a sub-500 load at week 48 in a missing-data-equal-failure analysis, and 70 percent maintained sub-500 loads in a missing-data-excluded analysis. In other words, by the more stringent of these two analyses, 16.5 percent of original study participants (55 percent of 30 percent) kept plasma viremia in check for 48 weeks.
As ACTG 398 does, ACTG 359 challenges the wisdom of giving a nonnucleoside to nonnuke-naive people who have apparently slender prospects of responding after a PI failure. In a Q&A session after Gulick's presentation, Stephen Deeks, MD (University of California, San Francisco) observed that after 16 weeks of treatment 70 percent of ACTG 359 enrollees graduated from dual-class failure to triple-class failure. Later in the meeting, though, Julio Montaner, MD (Vancouver) argued that NNRTIs substantially boost the potency of megadrug combos tried after multiple treatment failures (see "mega-HAART" below). And, in fairness to the ACTG trial planners, the scant salvage data available when these studies were on the drawing board suggested that turning to a nonnuke was the only salvage tactic that seemed to pay off.
This little study got littler fast. Presenting week 24 results, David Hardy, MD (Pacific Oaks Research, Hollywood) reported that only half the participants remained in the study. He tied most dropouts to virologic failure, side effects, or bad adherence. In a 24-week on-treatment analysis, eight of 13 (62 percent) had a viral load below 500 copies/mL, and seven of 13 (54 percent) had fewer than 50 copies. The week 24 intent-to-treat analysis yielded a sub-500-copy response rate similar to that of other PI rescue studies, 32 percent.
By intent-to-treat analyses, people who combined the PI and the NNRTI did better after 48 weeks than people who added one or the other to two NRTIs. In the nelfinavir/efavirenz group, 67 percent had a viral load below 50 copies/mL, compared with 44 percent in the efavirenz group and 22 percent in the nelfinavir group (P = 0.001 in a three-way comparison). Measuring response durability by seeing how many people rebounded above 200 copies/mL at 48 weeks yielded similar results: 81 percent in the nelfinavir/ efavirenz arm averted failure by this gauge, compared with 61 percent taking efavirenz and 41 percent taking nelfinavir (P < 0.001 in a three-way comparison). But CD4+ gains differed little from arm to arm. The median increase in all groups was 70 cells/mm3 at week 16 and 94 cells/mm3 at weeks 40 to 48.
Taking the nelfinavir-plus-efavirenz group as the response standard, investigators then figured baseline predictors of achieving a 48-week viral load below 50 copies/mL. Nelfinavir alone sliced the chance of success 88 percent, while efavirenz alone knocked 66 percentage points off the success rate. Every 1-log lower viral load at baseline increased the odds of success 2.3 times, and being naive to lamivudine (3TC) at baseline raised the odds 3.3 times. The study also showed that reaching a sub-50 viral load by week 16 significantly predicted a sustained 48-week response (P < 0.001).
In a subset analysis of 146 people with baseline viral loads above 2000 copies/mL, a higher number of baseline reverse transcriptase mutations correlated with 48-week virologic failure (P = 0.016). Results suggested that baseline genotyping can help plan a post-NRTI regimen. The ACTG team determined which study participants had high-level zidovudine (AZT) resistance at baseline, that is, the codon 215 mutation plus three or more mutations including those at positions 41, 67, 70, 210, and 219, or a position 69 insertion mutation, or a position 151 multi-NRTI resistance mutation.
Among 11 people in the nelfinavir/efavirenz arm who began treatment with high-level AZT resistance, 10 (91 percent) had a 48-week viral load under 50 copies/ mL, compared with two of 11 (18 percent) in the efavirenz arm (P = 0.002) and five of 22 (23 percent) in the nelfinavir arm (P < 0.001). So high-level baseline nucleoside resistance could be one factor that calls for a more potent rescue regimen.
But applying lessons learned in ACTG 364 won't be easy because, as Albrecht observed, the cohort is atypical. The average enrollee had taken only NRTIs for more than five years, some for as many as 10 years. Of course there are people who fit that profile, but these days more than a few clinicians would change nothing for a person with numbers matching this cohort's medians of 350 CD4+ cells/mm3 and 7776 HIV RNA copies/mL.
If a new regimen does seem in order, the crucial question again becomes whether to build the new combo around a single PI, a single NNRTI, or drugs from both classes. As John Mellors observed, ACTG 364 resoundingly demonstrates the bigger heft of a triple-class regimen in this context, "but once you play that [NNRTI] card, things are pretty dismal."
The meeting's other cochair, Julio Montaner, proposed resorting to an NNRTI if a person still has a reasonable complement of other antiretrovirals to back up the nonnuke. But Montaner's own multidrug rescue experiments notched better results when he stopped holding back nonnukes in salvage candidates without nonnuke experience (see "mega-HAART" below). And Mellors questioned whether 24- or 48-week gains from combining PIs and NNRTIs will hold up over two or three years. He confessed that he doubted it. It's too bad ACTG 364 didn't look at a third rescue option for people exposed only to NRTIs--dual PIs without an NNRTI.
Because four- and five-drug regimens after a PI failure yield uninspiring virologic results, some investigators have fashioned more exotic strategies. Two examples that have roused recent interest are strategic treatment interruptions (STIs) before salvage and the elephant-gun gambit variously termed mega-HAART or multidrug rescue therapy (MDRT).
Veronica Miller, PhD (JW Goethe University, Frankfurt) almost single- handedly launched the STI phenomenon at the 1999 Salvage Workshop, when she reported that mutant virus reverted to wild-type in 67 percent of people who took a drug holiday before starting a rescue regimen.3 And people who enjoyed such a reversion responded to subsequent treatment better than those whose virus stayed mutant during the holiday. But even at least year's meeting, attendees assumed that drug-resistant virus still lurked at undetectable levels in these so-called reverters. And Miller herself worried about vertiginous CD4+ drops in the reverters, drops not quickly erased after treatment began again.
Miller returned to this year's workshop with the news many had anticipated [abstract 25]. Among 33 wild-type reverters whose viral load dipped below 500 copies/mL with salvage treatment, 24 (73 percent) suffered a virologic rebound within a median 78 days. But because 100 percent of those who did not have a wild-type shift rebounded, Miller argued that the reverters had more durable responses. Some attendees remained unswayed by that logic, though, noting that a 27 percent success rate after fewer than 12 weeks matched or fell below response rates in similar cohorts. And Mike Youle, MB, ChB, reported that in his Royal Free Centre salvage cohort, those who took presalvage holidays gained no apparent treatment advantage over those who did not.
What about the scary CD4+ antics in the Frankfurt group? Cell counts had swooned by about 120 cells/mm3 among the holiday-takers with wild-type shifts. Miller found that CD4+ counts clawed their way back to baseline in 74 percent through 251 days of follow-up, so about one quarter in this cohort hadn't restocked their lost T cells. That trend held true in a second, larger cohort of 163 persons who took holidays or STIs ranging from 63 to 982 days. Three months after they resumed treatment, 75 percent had CD4+ counts within 90 percent of baseline.
The CD4+ results in the two cohorts indicate that a troubling minority of those who take STIs or drug holidays trade some immunologic armor for a still dubious virologic advantage. And those CD4+ plummets must mean something, because Miller charted 17 AIDS diagnoses in 15 people during STIs, including esophageal candidiasis and pulmonary TB in three each, Kaposi's sarcoma, wasting, cytomegalovirus infection, and encephalopathy in two each, and single cases of nonpulmonary TB, herpes, and cryptosporidiosis.
Among people in the expanded cohort with pre- and post-STI viral samples, 64 percent had a shift to wild-type for all three classes of antiretrovirals. But Miller didn't report exactly how many of these triple-class reverters saw their viral loads sink below 500 copies/mL when treatment resumed, or how their responses compared with those of nonreverters. Overall, though, 104 of 163 people in the expanded cohort (64 percent) had a virologic response, and 86 percent had a virologic rebound by the time of Miller's report.
Miller prudently concluded that the "effect of resistance and shift [to wild-type virus] on durability of response remains to be determined." Although she also concluded that the clinical effects of STIs remain unknown, one clinical effect during the STI emerged clearly from her work: People with compromised immunity open the door to opportunists when they stop taking antiretrovirals.
Miller's colleague Schlomo Staszewski, MD, agreed that treatment interruptions cannot be recommended as a therapeutic strategy before rescue therapy, though he maintained that the tactic merits a controlled, randomized study. "I don't see how it could be recommended today," concurred workshop cochair John Mellors, except as a way to determine whether a failing regimen is doing anything at all.
Two updates on mega-HAART cohorts appraised this strategy in four groups of individuals, one studied at London's Royal Free Centre by Mike Youle and colleagues, and three studied by Julio Montaner and his Vancouver team.
Youle's 92 patients all had one or more failed regimens on their charts, but their treatment experience is light compared with that of many salvage candidates [abstract 23]. They've taken an average 3.2 NRTIs and an average 1.7 PIs. All had detectable viral loads while taking a PI, and dual PIs had flunked in about 40 percent. But no one had taken efavirenz, the cornerstone of Youle's multidrug rescue regimen, and only 12 percent had tried nevirapine.
The group began the salvage combo (which includes efavirenz, two PIs, a mouthful of NRTIs, and sometimes hydroxyurea) with a median CD4+ count of 112 cells/mm3 and a median viral load of 309,000 copies/mL. Youle noted that those numbers are misleading, because they were measured after a drug holiday in about half of the cohort. But everyone had a viral load above 5000 copies/mL with their failing regimen.
In an intent-to-treat analysis, about 80 percent had a viral load below 400 copies/mL 40 weeks after starting the megaregimen. And among those treated for 64 weeks, about 70 percent had a sub-50 viral load. Some of these responders have already abandoned parts of their original combos and added others, including Abbott's investigational PI lopinavir (ABT-378) and foscarnet. Ten people had new AIDS diagnoses, and one person died from progressive multifocal leukoencephalopathy with a viral load below 50 copies/mL and after a 100-cell CD4+ gain. Youle said that 37 percent endured at least one drug side effect that led to withdrawal of at least one drug.
That side effects rate reflects Youle's high-octane strategy: Start with every drug that makes sense, then discontinue drugs that make trouble. He encourages candidates for this complicated treatment to think of it as chemotherapy for lung cancer. Sure, there will be side effects, but the no-holds-barred "induction" regimen will be scaled back in time. And he tells them it's like lung cancer treatment in another way--it's probably their last good chance to control the disease.
Why is Youle getting good virologic results so far? One obviously critical factor is that most of these people have sampled fewer antiretrovirals than those in some other salvage cohorts. Specifically, only a few had taken an NNRTI, nevirapine, before starting a powerhouse regimen including efavirenz and two PIs. Indeed, Youle noted that nevirapine experience appeared to correlate with virologic failure, but the number who had tried nevirapine is too low to make that association certain. People in this cohort also benefit from being treated as part of a dedicated "salvage clinic" in which only three experienced clinicians make treatment decisions.
How much hydroxyurea contributes to this regimen--for good or for ill--remains an open question. Youle reported that 16-week CD4+ counts climbed significantly higher among those not taking hydroxyurea (P = 0.048), although people who continued taking this drug had steady T-cell gains after 16 weeks. Certainly use of hydroxyurea in a cohort like this calls for extra-careful monitoring.
Julio Montaner and colleagues now have three megasalvage cohorts--106 who began treatment between August 1997 and June 1998, 75 who started between July 1998 and December 1998, and 69 who bit the megabullet between January 1999 and May 1999 [abstract 24]. Antiretroviral experience averaged around 40 months in all three groups, and all had virologic failures with "several prior regimens." But Montaner didn't report the mean or median numbers of antiretrovirals taken or how many people remained NNRTI virgins before salvage.
Salvage combos included up to 9 drugs--four NRTIs, two NNRTIs, two PIs, and hydroxyurea. Most people, though, took five to seven antiretrovirals picked on the basis of treatment history and lab profiles. Baseline CD4+ counts were 180, 200, and 190 cells/mm3 for groups 1, 2, and 3. Median baseline viral loads ranged from 62,000 to 65,000 copies/mL in these three groups.
Intent-to-treat analyses showed that, after 25 to 35 weeks of multidrug rescue, about 30 percent for group 1 (n = 91), 40 percent for group 2 (n = 58), and 50 percent for group 3 (n = 50) had viral loads below 400 copies/mL. Montaner attributed the incremental success rate to three factors: a greater willingness to prescribe NNRTIs right off the bat for groups 2 and 3, the availability of abacavir for immediate treatment of the later groups, and improving skill in "coaching" the later cohorts about taking their medications.
Not surprisingly, response rates began to slip with continued treatment. Among 128 people with two consecutive viral loads below 400 copies/mL, 81 (63 percent) maintained a sub-400 load for 24 weeks after their first undetectable reading. But it is interesting that group 3 did worst by this measure. Whereas 63 percent of group 1 responders and 76 percent of group 2 responders maintained that sub-400 reading at least 24 weeks, only 49 percent of group 3 responders (17 of 35) did so. It seems, then, that whatever advantages helped cohort 3 score the highest initial response often deserted them in the long run.
Side effect rates were not overwhelming for people taking so many drugs, but they weren't trifling either. All told, 25 percent had severe laboratory or clinical changes, 28 percent modified their regimen because of side effects, and 15 percent stopped treatment as a result.
Montaner argued that his results support recycling already used antiretrovirals for salvage candidates. But evidence from work in the laboratory of Robert Siliciano, PhD, suggests recycling won't work for long.4 Studying 21 perinatally infected children, Siliciano's group detected drug-resistance mutations in virus inhabiting the latent T-cell reservoir of children whose plasma viremia had remained suppressed for more than a year. The tenacious mutant virus apparently evolved during early nonsuppressive regimens. That finding, Siliciano maintained, renders "untenable the idea of recycling drugs that were part of failed regimens."
At the workshop John Mellors advanced another argument against megasalvage. Why increase the cost and risk of toxicity with mega-HAART when resistance testing could help pick fewer drugs that may do just as well? "How much ‘mega' do we need in mega-HAART?" he wondered. Indeed, Montaner himself noted that treatment response typically correlates with the number of active drugs in a regimen, and this "active drug effect" plateaus at around three drugs. But he proposed that the fourth, fifth, and sixth drugs--though perhaps rated "inactive" by phenotyping--may contribute a modest additive effect that gives the megaregimen more muscle.
Richard Hoetelmans, PharmD, PhD (Slotervaart Hospital, Amsterdam) suggested another mega-HAART mitigator--ever more complex drug-drug interactions. A few analyses of multiagent combos, he said, suggest that the concentrations of some components are too low to be meaningful. The savvy prescriber may well be better off leaving such drugs out of a megacombo, Hoetelmans proposed, if only that prescriber knew which drug levels fell in which combos. But that prescriber is out of luck, because so far the pharmacologists of the world have neither the time nor the temperament to catalog every conceivable drug-drug interaction in every conceivable salvage mix of five or more drugs.
Still, Montaner's group made a small step in that direction by figuring out what happens when they partner two nonnukes in a multidrug rescue combo. Alicia Tesiorowski, MD, compared historical drug level data on nevirapine, delavirdine, and efavirenz with trough levels of those agents when combined in megasalvage [abstract 14]. She analyzed several timed samples from 26 people taking nevirapine plus delavirdine and from seven taking nevirapine plus efavirenz.
Although neither delavirdine nor efavirenz substantially lowered nevirapine troughs, nevirapine did not return the favor. The mean delavirdine trough in this study measured 1927 ng/mL, compared with a historical trough of 8200 ng/mL when delavirdine is the sole NNRTI in a regimen. The mean efavirenz trough when given with nevirapine was 592 ng/mL, compared with 1767 ng/mL for solo efavirenz. Tesiorowski noted, though, that the lower delavirdine and efavirenz troughs "far exceed the IC90 for these drugs."
But who knows what these numbers mean clinically? Not Richard Hoetelmans, whose own crossover study of nevirapine plus efavirenz suggests to him that efavirenz doses should be raised. And not Julio Montaner, who said after Tesiorowski's presentation that no one at this point can weigh the putative additive merits of dual NNRTIs against these squat troughs. At this point though, it's certain that a second PI can be a PK enhancer, while a second NNRTI can be a PK impeder.
What role will amprenavir play in salvage therapy? The answer to that question seems to keep changing. At the first Salvage Workshop a study of amprenavir, efavirenz, and abacavir found that the combination brought viral loads below 400 copies/mL in only 26 percent of people after 16 weeks, while only 39 percent had even a half-log drop in plasma viremia.5 Reasons for the poor result included nonnuke experience by more than half of the study participants and an efavirenz-induced lowering of amprenavir concentrations.
More recent in vitro work showed, though, that people with PI experience have virus less cross-resistant to amprenavir than to other protease drugs, although cross-resistance to amprenavir looked like a problem for those with the I84V mutation6,7 or with high-level resistance to indinavir.7 Glaxo tried to verify these test-tube talismans of post-PI amprenavir activity in a pediatric study presented at the Salvage Workshop.* The results proved yet again that in vitro and in vivo differ in more ways than spelling. As in ACTG 398, amprenavir came up short in the pediatric salvage study, probably not surprisingly. The pediatric trial used amprenavir as the only salvage PI instead of combining it with a second PI, as the ACTG investigators had done.
Glaxo scrutinized 12-week virologic results in 60 children switched to amprenavir from another PI as part of a larger study [abstract 3]. A. Klein, MD, and colleagues then painstakingly sifted through PI and NRTI mutations in 51 of these children to see if they could divine any differences between responders and nonresponders. Sad to say, the response rate in these kids fell in line with woeful rates in adult PI salvage studies. After 12 weeks only 13 of 51 (25 percent) had a viral load under 400 copies/mL or at least a 0.7-log (5-fold) drop in viremia. That poor result, the investigators proposed, may reflect the extensive nucleoside experience in this cohort; 38 percent of PI-experienced children in this trial had already taken the two NRTIs that they combined with amprenavir.
The amprenavir regimen failed virologically in everyone who had more than four key PI mutations at baseline. Children with three key PI mutations had a response rate of 37 percent. Eight or more changes at any protease position linked with resistance significantly inflated the odds of virologic failure (P = 0.032). As the adult in vitro studies6,7 predicted, all children with the I84V mutation were 12-week nonresponders. That change showed up in six kids, always linked with L90M.
Just about the only possibly favorable news to emerge from this analysis is that four of the 13 responders had the D30N nelfinavir-induced mutation, compared with four of 38 nonresponders. Although that difference did not approach statistical significance (P = 0.179), Klein and coworkers interpret it to mean that nelfinavir failure does not rule out a later response to amprenavir.
Of course a safer bet after nelfinavir failure may be a dual PI regimen, as earlier studies suggest. And, as with other PIs, amprenavir probably should not be the sole protease drug in a regimen. Everyone's favorite PI additive, ritonavir, teamed up with amprenavir in one workshop study. But Glaxo also broached a rationale for a triple-PI regimen--600 mg of amprenavir, 100 mg of ritonavir, and 800 mg of saquinavir twice a day [abstract 16]. That adds up to nine capsules in each of the two daily doses.
The reasoning runs like this: Amprenavir and saquinavir are the Punch and Judy of PI histrionics. Amprenavir-resistant virus wielding the I50V mutation is hypersensitive to saquinavir. Saquinavir-resistant virus brandishing G48V and L90M remains susceptible to amprenavir. Together, amprenavir and saquinavir swing a synergistic wallop against wild-type and RT-resistant HIV.
Then ritonavir gets into the act. Ritonavir boosts amprenavir troughs. Glaxo's Eric Furfine and colleagues figured that the amprenavir trough with 600 mg of amprenavir plus 100 mg of ritonavir will be 6-fold higher than the 1200-mg amprenavir trough and 13-fold above the wild-type 50 percent inhibitory concentration (IC50). Ritonavir also jacks up saquinavir levels. Certainly, the results of ACTG 398 hint that amprenavir and saquinavir should be combined only with low-dose ritonavir, since amprenavir levels in that study lagged significantly in the amprenavir/ saquinavir arm compared with amprenavir/ indinavir or amprenavir/nelfinavir (P = 0.009).
Glaxo bravely suggested that this PI trio "might be considered an effective stand-alone antiviral regimen" or might be dosed along with abacavir or efavirenz. But not yet. Appropriately, Furfine and colleagues encouraged clinicians to await the safety, pharmacokinetic, and efficacy studies that Glaxo now plans.
Meanwhile, the prescriptively ambitious may have to settle for amprenavir, ritonavir, and efavirenz, which Jean-Louis Vildé, MD, studied with coworkers at the Bichat Hospital in Paris [abstract 15]. They gave two NRTIs and standard-dose efavirenz, along with 450 mg of amprenavir and 100 mg of ritonavir twice daily, to seven people who had taken PIs but no NNRTI. Their baseline viral loads ranged from about 725 to 575,000 copies/mL (mean 195,000 copies/mL), while their starting CD4+ counts varied from 3 to 460 cells/mm3 (mean 186 cells/mm3). Vildé started all the salvage drugs at the same time.
Amprenavir trough levels drifted downward at first, to 1190 ng/mL at day 14. Then the troughs bottomed out and started back up through 90 days of follow-up. After day 14, Vildé reported, the average trough was five times higher than the expected trough with 1200 mg of amprenavir twice daily, 30 times the IC50 for wild-type virus, and 10 times the IC50 for "some resistant strains." The intraindividual variation in amprenavir troughs measured about 30 percent, and the interindividual variation 40 percent.
So how did the regimen work? After one month, the average viral load dropped 2 logs, to about 1600 copies/mL. Three of seven people reached a viral load below 500 copies/mL, and the average CD4+ count jumped from 186 to 255 cells/mm3. Vildé did not report later trends, so it's impossible to guess whether a ritonavir arm in ACTG 398 would have outdone the other amprenavir/PI combos. The Bichat team advised giving amprenavir, ritonavir, and efavirenz only when amprenavir levels can be monitored.
The primeval PI pair, ritonavir plus saquinavir, has already seen duty in legions of persons infected with HIV. At recent meetings, however, ritonavir plus indinavir has emerged as the most-parsed PI combo. Does this new twosome merit so much attention? And, more important, is it better than the old rit/saq standby? Two studies presented at the Salvage Workshop suggest the answer to the first question, at least, is yes. And the two studies made a nice pair themselves--an in vitro hypothesis generator, and an in vivo hypothesis tester.
Merck's Jon Condra, PhD, did the in vitro work [abstract 2]. He coupled ritonavir with saquinavir, indinavir, or amprenavir and tested the duos against 20 iced isolates of PI-resistant virus. The isolates came from people in whom resistant virus emerged during indinavir monotherapy, during a first-line nelfinavir regimen, or during an indinavir regimen after nelfinavir failure. So the isolates harbored a small fleet of mutations associated with indinavir resistance, and an accompanying flotilla of nelfinavir-linked mutations.8 Condra wanted to see how 95 percent inhibitory concentrations (IC95s) stacked up against trough concentrations when he pelted these isolates with the three study PIs, alone or in various pairings. First, though, he corrected the drug concentrations for the effect of protein binding, noting that failure to account for this hobbling of PI potency results in overestimates of the drugs' antiviral activity. He based his math on protein binding correction factors figured by Abbott's Ahkter Molla, PhD,9 and on concentration data from the literature and product information.
With wild-type (nonmutant) virus, Condra's calculus indicated that standard doses of indinavir and ritonavir, and 750 mg of nelfinavir three times daily, handily yield trough concentrations that top these drugs' IC95s (Table 2). The call was a little closer with 1250 mg of nelfinavir twice daily. And standard-dose amprenavir and soft-gel saquinavir appeared to fall short of the mark. All the PI combos Condra studied had trough levels above IC95s for wild-type virus, though ritonavir plus saquinavir at 400 mg each twice daily just made it under the wire (Table 2).
What happened when the combos took on the PI-resistant virus from Condra's collection of 20 isolates? With ritonavir/saquinavir (400 mg each twice daily), the IC95 exceeded the trough for nearly every isolate. In other words, Condra proposed, this standard PI duo "may not have adequate potency to maintain suppression of most of these PI-resistant viruses." Amprenavir/ritonavir, at doses of 1200 mg and 200 mg twice daily, did better. That combo yielded a trough above the IC95 for all but three of the 20 isolates. So Condra concluded that this dual-PI regimen could make sense after failure of indinavir or nelfinavir, although the study does not address its potency after amprenavir wilts as a solo PI.
The best PI partners in this study were indinavir and ritonavir at doses of 800 mg and 200 mg twice daily. The indinavir trough with this regimen remained above the IC95s for at least 18 of the 20 isolates. The two remaining isolates may also be susceptible to this duo, but Condra couldn't say for sure because the IC95s exceeded the assay limit. Indinavir at 800 mg twice daily plus ritonavir at 100 mg twice daily could probably suppress replication in 15 of the 20 resistant isolates, whereas 400 mg of indinavir and ritonavir twice daily got the job done only half the time. All indinavir/ritonavir results were calculated for dosing with a high-fat meal.
On the basis of these results, Condra posited the important conclusion that "for sufficiently potent protease inhibitor regimens . . . neither ‘genotypic resistance' nor low-to-moderate ‘phenotypic resistance' to PIs may predict therapeutic failure." This is the same conclusion Abbott's Dale Kempf, PhD, reached after studying the effects of lopinavir (ABT-378) plus low-dose ritonavir after a single PI failure.10 Kempf's study, unlike Condra's, involved people actually taking the two PIs after an earlier PI fizzled. In fact, the folks in Kempf's study did well with the new PI regimen, although they also began taking a nonnucleoside for the first time. So how does indinavir/ritonavir stand up in the clinical arena?
Pretty well, according to a small chart review by Raphael Campo, MD, and colleagues at the University of Miami and Merck [abstract 7]. Campo plucked the records of 27 people who began a regimen including 800 mg of indinavir and 200 mg of ritonavir twice daily--the star couple in Condra's analysis. All had failed to push their viral load below 400 copies/mL after at least six months of PI treatment or had a viral load above 1000 copies/mL after scoring a sub-400 viral load with an earlier PI combination.
This small cohort differs from the larger group of 60 PI-experienced people studied by Dale Kempf in that they had already taken, on average, at least two protease drugs. In Kempf's study people had burned through only one PI before starting lopinavir/ritonavir. Eight of Campo's 15 patients began an NNRTI for the first time with indinavir/ritonavir, whereas everyone in Kempf's study started an NNRTI, nevirapine, for the first time.
After switching to indinavir/ritonavir, 15 of 27 people in Campo's study (56 percent) could be called responders because their viral load dipped beneath the 400-copy mark at least once. (More than 80 percent in the less experienced lopinavir/ritonavir cohort had a sub-400 viral load through 48 weeks.) What's intriguing about Campo's cohort is that an indinavir- and ritonavir-resistant genotype (P = 0.05) and at least a four-fold drop in susceptibility to indinavir and ritonavir (P = 0.03) actually correlated with a virologic response. As Jon Condra had predicted, indinavir and ritonavir at doses of 800 mg and 200 mg twice a day can overcome resistant virus in PI-experienced people.
But if indinavir/ritonavir beats up resistant virus, why didn't it control replication in all the people Campo treated? The deciding factor was adherence. The University of Miami clinicians assessed adherence by questioning patients at each visit. "Adequate adherence" meant taking 85 percent of the doses in a regimen; "inadequate" meant taking fewer than that. Among 17 adequate adherers, 13 (76 percent) had a sub-400 response, compared with two of 10 inadequate adherers (P = 0.007). Even among the 11 adequate adherers with a PI-resistant genotype at baseline, eight (73 percent) responded to indinavir/ritonavir. Similar proportions with decreased viral susceptibility to indinavir and ritonavir responded to the two PIs if they adhered adequately.
Although Campo's analysis underscores--yet again--the vital importance of adherence and demonstrates that indinavir plus ritonavir can rein in resistant virus, virologic response in this study can be traced to other factors as well. Campo counted five responders among six people (83 percent) who began at least two new drugs with indinavir/ritonavir, but in only one of six (17 percent) who did not add two new antiretrovirals. Taking a nonnucleoside for the first time also helped people earn their sub-400 stripes. Five of eight NNRTI-naive individuals (63 percent) who began an NNRTI with indinavir/ritonavir responded. Although these differences did not approach statistical significance because of the small numbers involved, they corroborate a clutch of other studies demonstrating the advantage of using new drugs, especially nonnukes, in a salvage regimen. The question, once more, is when to use that nonnuke.
One could argue that the standard for a response in this study is low--a single viral load below 400 copies/mL. But Campo showed that the 15 responders kept responding, as long as they kept taking their drugs. Eight of the 15 had maintained their sub-400 reading through their most recent visit. Five of 15 with transient rebounds linked to episodes of poor adherence regained their sub-400 status when their adherence improved. And two other shaky adherers with bigger rebounds, to 2696 and 151,929 copies/mL, lowered their viral loads to 873 and 451 copies/mL, respectively, when they started taking their drugs on time again.
The indinavir/ritonavir studies give some grounds for optimism--not because indinavir/ritonavir suddenly looks like the antiresistance elixir, but because it tenders one more rescue option. Jon Condra's work suggests another--amprenavir/ritonavir--and other work shows the merits of lopinavir/ritonavir and tipranavir/ritonavir in battling PI-resistant virus. Another Workshop study, presented by Jeffrey Fessel, MD, and colleagues from Kaiser Permanente in San Francisco, charted sub-50-copy viral loads with saquinavir (>2400 mg/day) plus ritonavir (200 to 400 mg/day) in four people with the saquinavir-resistant L90M mutation [abstract 8].
If these harbingers hold true in majorities of people laden with PI-resistant virus, research will have plugged one big hole in the ship of salvage. But as Raphael Campo showed, plugging that hole ultimately depends on good adherence.
The pivotal role of adherence in the success of a salvage regimen, a first-line regimen, any regimen, can play like a dusty old drama after too many iterations. Probably a better metaphor for the much-retold adherence story is the epic poem in the oral tradition. The bards of yore, one assumes, reshaped their tales with each telling, perhaps when memory faltered, or perhaps when an apter trope, a fitter image, fell from mind to tongue. Raphael Campo's discovery that PI-resistant virus crumples beneath the aegis of adherence and a potent PI pairing (see preceding section) offers one example of a subtly reshaped adherence saga. And Salvage Workshop attendees heard two other retellings of this evolving idyll.
Jean-Louis Vildé and colleagues in APROCO, the multicenter French cohort study, found that good adherence correlated independently with virologic response in 1283 experienced or naive individuals [abstract 5]. Study participants, 56 percent of them beginning a second or later regimen, rated their adherence on a self-administered questionnaire as total (31 percent), almost total (40 percent), uncertain (26 percent), or poor (3 percent). Their median starting CD4+ count measured 304 cells/mm3, and their median viral load was about 25,000 copies/mL. The APROCO team defined response as a viral load below 500 copies/mL after four months of the new regimen, and they defined rebound as a postresponse bounce above 500 copies/mL.
Among total adherers, 85 percent had a sub-500 response at four months, compared with 84 percent of almost total adherers, 70 percent of uncertain adherers, and 23 percent of poor adherers. In a multivariate analysis total or almost total adherence increased the chance of a four-month response 17 times compared with poor adherence (odds ratio [OR] 17.1, P < 0.0001). Uncertain adherence multiplied the chance of responding almost six times compared with poor adherence (OR 5.85, P = 0.002).
Other factors that correlated with a sub-500 viral load at month four were a baseline CD4+ count above 300 cells/mm3 (OR 1.83, P = 0.002), being antiretroviral naive (OR 2.23, P < 0.0001), and a baseline viral load below 1000 copies/mL (OR 4.27, P < 0.0001). In other words good adherence was by far the most important response correlate. But of course a naive status is almost always an advantage.
Of the 670 four-month responders with a median follow-up of 9.7 additional months, 198 (30 percent) strayed back into 500-plus territory. Again, total or almost total adherence independently predicted continued suppression. But this time total adherence had an edge over almost total adherence. Total adherence cut the chance of rebound 64 percent compared with poor or uncertain adherence (OR 0.36, P < 0.001), while almost total adherence trimmed 39 percentage points off the chance of rebound compared with poor or uncertain adherence (OR 0.61, P = 0.005). Younger age, a baseline naive status, lower baseline viral load, and higher baseline CD4+ count also correlated significantly with a durable response.
Vildé and coworkers concluded that either perfect or nearly perfect adherence can corral viral replication when a new regimen begins, but total adherence seems necessary to sustain that response.
ARGENTA, a single-center randomized study of 168 people with viral loads above 2000 copies/mL while taking HAART, confirmed that genotyping can help pick an effective rescue regimen, but only among people who practice good adherence and switch at a low viral load [abstract 30]. Median baselines values came close those in the APROCO cohort, at 264 CD4+ cells/mm3 and 17,000 HIV RNA copies/mL. One regimen had failed in 46 percent, two regimens in 28 percent, and three or more in 26 percent. Andrea DeLuca, MD, and colleagues from Catholic University in Rome randomized participants to a standard-of-care group whose rescue regimens were determined by history alone, or to a group whose new regimen was based on genotypic results dissected by a panel of clinicians.
After three months of treatment, the genotyping and expert advice hadn't helped much. Only 22 percent in the genotyping arm logged a viral load below 500 copies/mL, compared with 19 percent in the standard-of-care arm, a nonsignificant difference. But genotyped individuals with good adherence (determined by a validated questionnaire) and a baseline viral load below 10,000 copies/mL shaved 1.51 logs off their viral loads, compared with a 0.23-log drop among adherent people with low baseline loads in the standard-of-care arm (P = 0.015).
In a univariate analysis, poor adherence lowered the odds of a sub-500-copy three-month response by 13 percent (P = 0.05), as did a history of three or more failed regimens (OR 0.08, P = 0.001) and every additional log of baseline viral load (OR 0.39 per log, P = 0.02). Every extra active drug in the regimen increased the chance of responding at three months by 69 percent (P = 0.05), and a history of undetectable viremia (a likely marker of good adherence) inflated the chance of success more than three time (P = 0.002). In a multivariate analysis, more than two failed regimens remained a significant predictor of failure (OR 0.11, P = 0.03).
DeLuca and colleagues concluded that the "benefit of monitoring HAART-failing patients by genotypic resistance testing is limited by patient adherence and available treatment options."
Defining success and failure remains a problem in appraising salvage study results. Raphael Campo's study of indinavir plus ritonavir, for example, used a liberal definition of success--a single viral load below 400 copies/mL (see "Indinavir and ritonavir" above). In the poster right next to Campo's, Jeffrey Fessel set a much higher success standard--a viral load below 50 copies/mL among people beginning their third, fourth, or fifth HAART regimen [abstract 8]. With genotyping to help pick the salvage treatment, Fessel found that nine of 15 people (60 percent) taking their third HAART combo for a median of 15 months met the sub-50 benchmark. But he noted that this on-treatment analysis excluded individuals who stopped treatment because they couldn't stomach the drugs. And genotyping couldn't help pick a good fourth or fifth HAART combo.
Amanda Mocroft, PhD (Royal Free Centre, London) used several definitions of therapeutic response to paint a bigger salvage picture in 167 members of the EuroSIDA cohort [abstract 4]. These people were true treatment pros, having taken at least one drug from each antiretroviral class. The median numbers of drugs tried were four NRTIs, three PIs, and one NNRTI. Their viral loads had rebounded above 1000 copies/mL after taking their latest regimen for 16 weeks or more. In fact, the median viral load stood at about 78,000 copies/mL when they started the salvage combo tracked in Mocroft's analysis. The median CD4+ count was 131 cells/mm3.
How well they did after six months of follow-up depended on what yardstick Mocroft used to measure response. Only 28 percent reached a viral load below 500 copies/mL six months after starting their salvage therapy, a number in line with results of prospective, randomized trials. When Mocroft slackened the success signal to 1000 copies/mL, she could still call only 32 percent responders. But 43 percent attained at least a 1-log drop in viral load after six months of salvage, and 58 percent eked out a half-log decline. Nearly 40 percent gained at least 50 CD4+ cells/mm3.
These modest returns can be explained partly by the EuroSIDA clinicians' idea of how to craft a salvage combo. Only 22 percent of the regimens included two protease inhibitors; 39 percent used a single PI, and 25 percent used a single NNRTI. Mocroft didn't report how many regimens combined a PI and an NNRTI, but it sure looks like mega-HAART, or even dual PIs, haven't caught on at the EuroSIDA sites.
Perhaps the most interesting byproducts of Mocroft's study are the implications of a multivariate analysis of factors predicting response. Two variables correlated with attaining a six-month viral load below 1000 copies/mL: Every prior 12 months of NRTI experience trimmed 8 percentage points off the chance of having a sub-1000 viral load (relative hazard [RH] 0.92, P = 0.013). And chalking up a viral load below 500 copies/mL at any time in the treatment history boosted the chance of success 3.15 times (P < 0.0001). A history of a sub-500 load at any time also independently predicted at least a 50-cell gain in CD4+ cells (RH 2.74, P < 0.0001).
That a longer chronicle of NRTI treatment dimmed the chance of salvage success will surprise no one. People who began treatment with one or two NRTIs are today generally the most-scarred veterans of the antiretroviral wars--and therefore the least likely to have good treatment options left.
As in the ARGENTA study of genotyping and adherence (see "A certain salvage" above), a record of treatment-induced undetectable viremia in the EuroSIDA analysis favored a better response to salvage. That should come as no surprise either. People who respond to one regimen may be more likely to respond to a rescue regimen for several reasons: Perhaps they began their regimens with lower viral loads and higher CD4+ counts, factors that almost invariably foster success. Perhaps they missed the NRTI-only era and started treatment with more potent three- or four-drug confections. Perhaps they tolerated their antiretrovirals better, or learned to live with side effects that poor responders found intolerable.
Almost certainly, though, a previous undetectable viral load predicted salvage success because the EuroSIDA responders were good adherers, just as responders were in Raphael Campo's, Jean-Louis Vildé's, and Andrea DeLuca's cohorts. As John Mellors stressed after Mocroft's presentation, success breeds success, and failure begets failure. Whenever antiretrovirals fail, he argued, the major question must be Why? Failure to find the cause of failure favors yet another round of failure. The Royal Free Centre's Mike Youle proposed a more pungent version of that dictum: "Failure begets failure because it's the same physician."
Mark Mascolini writes about HIV infection (mailmark@ptd.net).
1. The Compact Edition of the Oxford English Dictionary, Vol 2. Oxford: Oxford University Press, 1971:S69.
2. Mascolini M. Salvage solitaire (or, HAART takes a holiday). JIAPAC 1999;5(2):10-23.
3. Miller V, Rottmann C, Hertogs K, et al. Mega-HAART, resistance and drug holidays. Antiviral Ther 1999;4(suppl 1):27-28.
4. Persaud D, Pierson T, Ruff C, et al. A stable latent reservoir for HIV-1 in resting CD4+ T lymphocytes in infected children. J Clin Invest 2000;105: 995-1003.
5. Falloon J. Salvage therapy with abacavir plus amprenavir and efavirenz. Antiviral Ther 1998;3(suppl 2):16.
6. Calvez V, Tamalet C, Molina JM, et al. Amprenavir (APV) shows less protease inhibitor (PI) cross resistance as compared with other PIs in a multiexperienced HIV cohort (CNAB3008, French virological substudy). Presented at: 39th Interscience Conference on Antimicrobial Agents and Chemotherapy; September 26-29, 1999; San Francisco. Abstract I-442.
7. Race E, Dam E, Obry V, et al. "Analysis of HIV cross-resistance to protease inhibitors using a rapid single-cycle recombinant virus assay for patients failing on combination therapies." AIDS 1999 Oct 22;13(15):2061-8.
8. Of the 20 PI-resistant isolates studied, four had mutations at protease position 30, 12 at position 46, nine at position 82, four at position 84, and nine at position 90. None had amprenavir's hallmark codon 50 mutation, since none of the isolates came from people exposed to amprenavir.
9. Molla A, Vasavanonda S, Kumar G, et al. "Human serum attenuates the activity of protease inhibitors toward wild-type and mutant human immunodeficiency virus." Virology 1998 Oct 25;250(2):255-62.
10. Kempf D, Xu Y, Brun S, et al. Baseline genotype and phenotype do not predict response to ABT-378/ritonavir in PI-experienced patients at 24 and 48 weeks. Presented at: 7th Conference on Retroviruses and Opportunistic Infections; January 30-February 2, 2000; San Francisco. Abstract 731
* Most readers will be aware that amprenavir liquid should not be given to children under four years old, or to pregnant women, people with hepatic or renal failure, or people taking disulfiram or metronidazole.
000610
IA000601
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.