IAPAC Monthly - Vol. 8, No. 2, February 2002
Mark Mascolini

Perched atop the Acropolis, looking south and directly down, one can see the very ruined Theater of Dionysos, where Euripides and that crowd held forth. To the right lies the Theater of Herodes Atticus, very much restored and entertaining Athenians to this day, if they can endure the stone seats. Between these ancient amphitheaters, the eagle-eyed will spot the Asclepion, a monument built atop a sacred spring and dedicated to Apollo's physician son, Asclepius.

That's right. At the very foot of the most astonishing acropolis in Greece, the Athenians erected a memorial to a doctor. And with good reason. Asclepius may have been the first physician but, by all accounts, he was the best, inspiring his own cult of temple builders.¹ Asclepius intimidated even the gods with his canny knack for egging mere mortals toward immortality. Finally, fed up, Zeus dispatched Asclepius with a thunderbolt. But even that dramatic intervention failed to dissuade the sick. Merely sleeping in a temple of Asclepius could effect a cure, if the doc deity deigned to make a house call in your dreams.

The HIV docs who came to Athens for the 8th European Conference on Clinical Aspects and Treatment of HIV Infection aren't ducking thunderbolts yet. So far, the gods betray no envy of their prowess. Or maybe the Olympians figure they still easily hold the winning hand in the AIDS contest. Although HIV medicine makers and the clinicians who joined those drugs in life-prolonging strategies helped thousands cheat death, millions will die before they get to visit any antiretroviral Asclepions.

But the news from Athens was good for people who can afford antiretrovirals. Indeed, the strongest message to emerge from a handful of Athens studies is that first-time HAART has fulfilled its promise. No, it can't wipe out HIV and, yes, its side effects plague people who've made pill taking a 21st century religion. But these nicely analyzed cohort studies show that 70 to 80 percent rates of undetectable viremia aren't always clinical trial artifacts. They're real rates from hundreds of people treated for years across Europe. But there's a catch--and more on that soon.

This success story from latter-day Asclepii went nearly unnoticed in the buzz stirred by a 12-week study of structured treatment interruption (STI). Preliminary results of that trial, though good news for 34 people with deep--and unsuccessful--treatment experience, won't immediately undo the failures chronicled in other presalvage STI studies and essays in mega-HAART. So let's leave that multidrug rescue headline till later and look first at the multiyear story of standard HAART success. That story started in the poster tent:

How's HAART been doing lately?

At the Institute of Tropical Medicine in Antwerp, Christa Dreezen and colleagues have been tracking 134 people who began a three-drug regimen between February 1995 and May 1997 and continued treatment for at least four years [abstract P349^*]. Most of them (96 percent) had a protease inhibitor (PI) in their first triple regimen, and most (81 percent) had antiretroviral experience before the PI. After four years of therapy, 94 (70 percent) had a viral load below 400 copies/mL. The average CD4+ count climbed from 181 cells/mm³ before HAART to 488 cells/mm³ at month 48.

Compared with other HIV cohorts, these long-term responders don't seem extraordinary. Their average age is 44 years and they've been infected with HIV for an average 9.6 years. About three quarters are Belgian, though 17.5 percent are African. About three quarters are men. As in other random cohorts, most have switched their antiretrovirals more than once. In fact, they average five switches. Not surprisingly, though, those with the fewest changes have done the best. Among people who revised their regimen one to four times, 91 percent had a 48-week viral load under 400 copies/mL. Among those with five to eight regimens on their charts, 49 percent had a sub-400 load at 48 weeks.

Certainly this cohort doesn't represent a broad cross-section of people who start antiretroviral therapy. Dreezen limited the study to people who took HAART for four years, so the analysis excludes everyone who may have quit earlier because of drug toxicity, or because the drugs kept failing, or because they died. Still, the study shows that a large majority of people who respond to antiretrovirals, and find a regimen they tolerate, can keep the lid on replication.

The always-provocative Andrew Phillips (Royal Free Centre for HIV Medicine, London) contributed two intriguing studies to the Athens meeting, both indicating that most people who respond well to HAART in the first half year or so keep on responding. The larger study set out to weigh the effect of nucleoside reverse transcriptase inhibitor (NRTI) experience on the durability of a HAART regimen in 1,433 people at the Frankfurt clinic and the Royal Free Centre in London [abstract 027].

Phillips found that even one or two months of pre-HAART treatment with NRTIs significantly reduced the chance of responding to a three-or-more-drug regimen including a PI, a nonnucleoside RT inhibitor (NNRTI), or abacavir. But, just as importantly, he showed that every additional year of controlled replication increased the chance of continued suppression, regardless of NRTI experience.

The analysis involved people who reached a viral load below 400 copies/mL and kept it there through 24 weeks while taking their first HAART. Then Phillips charted how long replication stayed under wraps, defining a rebound as two consecutive RNAs above 400 copies/mL. People with rebounds due to treatment breaks got censored from the analysis. After follow-ups stretching from one year to more than four, 307 people (21 percent) endured a rebound, including 175 of 409 (43 percent) with nucleoside experience and 132 of 1,024 (13 percent) who started HAART drug naïve.

In a Cox proportional hazards model, two factors predicted a rebound: Any length of pre-HAART nuke experience raised the odds of rebounding two to four times depending on months of pretreatment (fewer than two, two to six, six to 12, 12 to 24, 24 to 36, and 36 to 60). And every year of under-400 suppression halved the chance of a subsequent rebound.

"The rate of rebound declines substantially over increasing time with viral suppression on HAART," Phillips concluded. And that finding held true for people with and without NRTI experience before HAART. Yet significantly more rebounders came from the NRTI-experienced group, and their rebound rate remained higher than the naïve group's for up to three years after starting HAART.

The worse response among people who sampled NRTIs before HAART supports the hypothesis that NRTI resistance mutations archived during suboptimal nucleoside therapy come back to haunt some people after they begin more potent therapy. A smaller French study published soon after the Athens meeting reached the same conclusion about archived mutations.² Phillips also found that rebound rates didn't depend on whether people had taken one or two NRTIs before HAART; which NRTIs they took didn't matter either.

Although very different from Dreezen's study, and much larger, Phillips' rebound analysis reached one similar conclusion: A large proportion of people who push their viral load under 400 copies/mL with HAART and keep it there for several months can maintain that suppression, even if (as in Phillips' study) they took NRTIs before HAART, or if (as in Dreezen's) they try as many as four HAART combos.

Soon after the Athens meeting, Phillips published a different analysis of the Frankfurt cohort in which people who get their viral load under 50 copies/mL within 24 weeks of starting antiretrovirals rarely suffered virologic failure over 3.3 years of follow-up.³ Although 61 of 336 people had a rebound above 50 copies/mL, 47 of those breakthroughs resulted from treatment interruptions usually mandated by concurrent illness. In only 14 people did rebounds result from drug failure, and that rate translated into a 5.2 percent risk of failure in 3.3 years. Phillips again found that the rebound risk dwindled as time with undetectable viremia grew (P = 0.01), this time using 50 copies as a cutoff instead of 400.

A study of 988 people starting their first PI regimen in France suggested, on the other hand, that maintaining a viral load under 500 copies/mL for 16 weeks offers less assurance of continued virologic control [abstract 019] compared with Phillips' sub-400 [abstract 027] or sub-50³ cutoff at 24 weeks. François Raffi (Hôtel Dieu, Nantes) reported that 786 (79 percent) achieved a 16-week sub-500 response, but 248 of them suffered a rebound in 23 months of follow-up. That put the 23-month response rate at 54 percent.

Raffi and colleagues correlated baseline CD4+ counts and initial viral load responses to short- and long-term CD4+ responses. Virologic responders trying their first PI with a T-cell count under 200 cells/mm³ gained an average 15 cells per month during the first four months, compared with 28 cells per month among those who started with 200 to 500 cells/mm³ (P < 0.001). Among responders who hadn't rebounded, the monthly CD4+ gain after month four averaged 8 cells/mm³, compared with 5 cells/mm³ among people with a rebound between 500 and 5,000 copies/mL, and 3 cells/mm³ among those with a rebound between 5,000 and 10,000 copies/mL. People whose rebounds topped 10,000 copies/mL lost 2 cells/mm³ monthly.

Phillips' second Athens study itemized the benefits of keeping a treatment-induced viral load under 50 copies/mL [abstract P372]. With colleagues at London's Royal Free Centre, he kept tabs on people who notched a HAART-induced viral load below 50 copies/mL and two consecutive CD4+ counts above 500 cells/mm³. The 166 people who hit those marks and never rebounded above 50 copies/mL had practically no change in their CD4+ count through a median 47 weeks of follow-up. In that time the median CD4+ change was +2 cells/mm³ (intraquartile range -9 to +12 cells/mm³).

Only five of these 166 people saw their CD4+ count sink below 350 cells/mm³--to 349, 347, 331, 330, and 262 cells/mm³. The CD4+ percents in these five remained higher than percents normally associated with such CD4+ counts, ranging 22 to 30 percent. Only one person endured an AIDS-defining illness through 162 person-years of follow-up, a lymphoma diagnosed at 635 cells/mm³.

Phillips concluded that CD4+ drops below 350 cells/mm³ are "extremely rare" in people who maintain sub-50-copy viral suppression. Factors not related to HIV infection, such as toting CD4+ cells after rest and in the early morning, could account for some low readings.

What happens to people whose viral load becomes undetectable, rebounds into detectable territory, then goes back down? One Athens study confirmed that an isolated blip above 50 copies/mL does not signal an enduring rebound, while another study showed that blips over 200 copies/ mL spell trouble mostly for people with heavy treatment experience. In 102 people taking HAART for at least 48 weeks, Juan Marcías (Valme University Hospital, Seville) found similar rates of virologic failure (two consecutive viral loads above 200 copies/mL) among those who had a single blip above 50 copies/mL (6 percent) and among those who stayed under 50 copies/mL (8 percent) [abstract P380]. Follow-up in the two groups stretched to about 168 weeks.

Among 585 people at London's Chelsea and Westminster Hospital who had an isolated blip above 200 copies/mL, Sundhiya Mandalia tied a later rebound above 500 copies/mL to experience with all three antiretroviral classes (P < 0.001). People who had taken either PIs or NNRTIs, but not both, had significantly lower risks of virologic failure (relative hazard 0.59 for NNRTIs only [P = 0.028] and 0.74 for PIs only [P = 0.049]). People in this cohort had taken a median of three antiretroviral regimens.

ATHENA confirms Athens findings, and Geneva's primary infection gamble

Treatment results in one large prospective study of Europeans starting HAART, the Dutch ATHENA cohort, did not get presented at the Athens meeting, despite the shared namesake. But earlier in 2001 ATHENA investigators confirmed two trends reported in Athens--the overall good response to first-time HAART, and the advantage of starting a potent regimen with no antiretroviral experience.⁴

Among 3,348 people beginning HAART, 80 percent of those without earlier antiretroviral experience reached a viral load below 500 copies/mL in 24 weeks. Nearly all of them maintained that suppression through 144 weeks, despite regimen changes. Among those who started HAART with treatment experience, 57 percent had a sub-500 load by week 24. Compared with treatment-experienced people, treatment-naïve individuals had nearly twice the chance of a sub-500 response in 24 weeks.

Since the second half of 1996, a consistent percentage of the ATHENA cohort changed their regimen during any six months. Toxicity, not virologic failure, accounted for most of these switches (Figure 1). Changes due to virologic failures fell from 13 percent in 1996 to 4 percent in the second half of 1999, while toxicity-driven revisions rose from 21 percent in 1996 to 43 percent at the end of 1999. As in other developed countries, AIDS-defining diagnoses in the Netherlands plummeted during this period, from 2.63 per person-year in the first half of 1996 to 0.04 per person-year in the first half of 2000. To put it another way, when HAART arrived Dutch people with HIV infection suffered two to three new AIDS diseases every year; four years later, the per-person rate had shrunk to one new diagnosis every 25 years.

Seven years ago, at the Fourth European Conference, Sabine Kinloch from Luc Perrin's group at Geneva's University Hospital shocked many with a placebocontrolled study of zidovudine (AZT) for primary HIV infection.' This year, Samir Vora from Perrin's team drew only approving nods with a five-year follow-up of people who began treatment for primary infection with AZT plus didanosine (ddI) or ddI plus stavudine (d4T) [abstract 025]. Eleven of these 15 later switched to triple therapy including a PI or NNRTI, while four stuck with their dual nukes. Vora and colleagues can't be accused of age discrimination. Though the median age was 45 years, the range stretched from 30 to 69 years. The median time between the acute retroviral syndrome and treatment was eight days (range one to 42 days).

After a median 58 months of follow-up (range 51 to 66 months), 12 of 15 have a viral load below 100 copies/mL, and nine of those 12 are under 10 copies/mL. Their CD4+ counts have stayed between 500 and 1,000 cells/mm³. Lymph node mononuclear cells sampled from eight people all had detectable cell-associated RNA, while three of the eight (two on double NRTIs) had undetectable RNA in peripheral blood mononuclear cells. All eight had cellassociated HIV DNA in both lymph node and peripheral cells. Despite the long follow-up only four of 15 have signs of lipodystrophy. Vora didn't report metabolic values. Everyone in the group continued treatment throughout this period; two later interrupted therapy and had prompt viral rebounds.

Is there a "best" HAART? (And does it include an E-drug?)

When people talk about HAART these days, they usually mean a medley of at least three or four drugs, at least one of them a PI, an NNRTI, or abacavir. And, when talking about PI regimens, they usually mean two PIs, excepting perhaps stand-alone nelfinavir. But few so-called opinion leaders will stand in front of a room of clinicians and venture an opinion on what the best HAART may be. The usual dodge is that different combinations are better suited to different individuals. But when dispensing the results of a randomized. double-blind trial with P values all pointing to the superiority of drug X, no one has trouble saying "drug X was virologically superior to drug Y," even though the study population was a bunch of individuals with different wants and different needs.

So rephrase the question: At the population level, is one HAART better than another? Even without randomized, doubleblind trials, a case can sometimes be made for the superiority of one regimen over others. Two large cohort studies showed, for example, that people whose first HAART included efavirenz did better virologically after six to eight months than people starting single or even double PIs.⁶, 7 And at the Athens meeting, two randomized, open-label studies found that switching from a PI to efavirenz yielded better RNA results than sticking with the PI. Christine Katlama (Pitié Salpôtrière Hospital, Paris) summarized both studies in a single talk [abstract 06].

In both trials efavirenz soundly thrashed continued PI therapy in nearly every contest that statisticians devised (Table 1). Everyone in the two studies had reached a viral load below 50 copies/mL with PI therapy, and everyone kept their NRTIs at randomization; most had about 24 months of antiretroviral experience. A US study randomized 226 people to trade their PI for efavirenz and 120 to keep their PI. Respective numbers in a European study were 58 and 60. Mean CD4+ counts at randomization measured 582 cells/mm³ in the efavirenz group and 517 cells/mm³ in the PI group of the US study, and 498 cells/mm³ for efavirenz and 500 cells/mm³ for PIs in the European study. After randomization CD4+ counts continued to rise in both treatment arms of both studies.

At the 48-week mark the efavirenz groups included fewer people with rebounds above 50 copies/mL, more people with sub-50 loads in a noncompleterequals-failure analysis, and fewer dropouts for any reason (Table 1). Time to virologic failure or dropout was significantly longer in the US (P = 0.009) and European (P = 0.008) efavirenz arms. In the US study three people taking PIs had lipodystrophy after 48 weeks, compared with none taking efavirenz, a nonsignificant difference. But that difference proved significant in the European study, which recorded 15 cases of lipodystrophy with continued PIs and four with efavirenz (P < 0.05). After 48 weeks the groups didn't differ in average cholesterol, low-density lipoprotein (LDL) cholesterol, or triglycerides. High-density lipoprotein (HDL) cholesterol rose significantly among people who switched to efavirenz, but Katlama doubted this increase in "good" cholesterol was clinically meaningful.

One reason efavirenz came out ahead was better adherence, calculated in the US study. A higher proportion of people keeping their PI (43 percent) than switching to efavirenz (29 percent) said they missed one dose during any follow-up visit (P = 0.011). And a higher proportion in the continued PI group (29 percent) than in the efavirenz group (12 percent) reported missing a dose during multiple visits (P < 0.001).

Two cohort studies reported earlier in 2001 found that efavirenz outperforms nevirapine as a first-time NNRTI regimen.⁷, 8 A third study from three US HIV clinics confirmed those results in Athens [abstract LB/P22]. Philip Keiser (University of Texas, Dallas) reported that 555 people who launched their antiretroviral odyssey with efavirenz did better by several measures than 523 who set sail with nevirapine, even though the nevirapine group started with a lower average viral load (3.87 versus 4.37 logs, P < 0.01). The average starting CD4+ count was lower in the efavirenz group, but not significantly so (264 versus 310 cells/mm³).

After the first year of treatment, the average viral load fell significantly less with nevirapine (0.51 log) than with efavirenz (1.32 logs, P < 0.001). After about two years of follow-up, more people taking nevirapine (59 percent) than efavirenz (37 percent) had two consecutive viral loads above 400 copies/mL (P < 0.001). In a multivariate analysis, efavirenz halved the risk of two consecutive viral loads above 400 copies/ mL compared with nevirapine (P < 0.001).

The biggest advantage nevirapine has over efavirenz--an advantage not examined in Keiser's study--is probably the attractive lipid profile demonstrated in the Atlantic study.⁹ And comely lipid profiles are certainly not a characteristic of the other prime contender for first-line cornerstones--double PIs. Two studies reported at the Athens Lipodystrophy Workshop showed that two PIs push lipids higher than one.¹⁰, 11 At the 8th European meeting, two studies of dual PIs looked only at the antiviral prowess of such regimens, and one of those studies showed that two PIs don't always beat one.

The BEST study left efavirenz on the sidelines and pitted standard-dose indinavir against indinavir plus ritonavir at 800 and 100 mg twice daily [abstract 08]. The ritonavir-boosted regimen came up on the short end after 48 weeks because significantly more people dropped out of that study arm with side effects. People randomized to switch from solo indinavir to indinavir/ritonavir in this open-label trial probably started at a disadvantage because the BEST formulation of ritonavir wasn't available at the time; they got liquid ritonavir instead.

Everyone began the study with a sustained sub-500-copy response to three-times-a-day indinavir. Jan Gerstoft (Rigshospitalet, Copenhagen) and coworkers randomized 162 people to continue that regimen while 161 switched to indinavir/ritonavir. After 48 weeks 86 percent on indinavir and 88 percent on indinavir/ritonavir still had a viral load under 500 copies/mL in a missing-data-equal-failure analysis. But when statisticians counted noncompleters as failures, 74 percent taking indinavir and 58 percent taking indinavir/ritonavir maintained viral suppression (P < 0.01).

In the double PI arm, 47 (29 percent) dropped out because of toxicity, compared with 20 (12 percent) who continued thricedaily indinavir (P < 0.001). Significantly more people taking indinavir/ritonavir had kidney stones (9 versus 4 percent, P = 0.044) or blood-tinged urine (27 versus 13 percent, P = 0.001). Gastrointestinal problems, some probably due to the liquid ritonavir, accounted for 18 percent of dropouts in the indinavir/ritonavir group and 1 percent in the standard-dose indinavir group. More people in the double PI group switched regimens because of side effects (29 versus 12 percent, P < 0.001). Fasting triglycerides and cholesterol were higher in the indinavir/ritonavir group, but lipodystrophy rates were similar in the two study arms.

Eric Sandström (Karolinska Institute, Stockholm) suggested that liquid ritonavir wasn't the only drawback in the double PI arm. He maintained that therapeutic drug monitoring can help lower the indinavir dose--and indinavir toxicity--when boosting with ritonavir. A head-to-head comparison of indinavir/ritonavir (again at 800/100 mg twice daily) and saquinavir/ ritonavir (1000/100 mg twice daily) suggested Sandström may be right about 800 mg of ritonavir-boosted indinavir being too much for some people.

This randomized, open-label trial (which used ritonavir capsules) charted a 20 percent 24-week toxicity dropout rate in the indinavir arm versus 8 percent in the saquinavir group [abstract 010]. Among 116 people who continued indinavir/ritonavir, 75 had grade 3 or 4 side effects (65 percent), compared with 45 of 123 (37 percent) who continued saquinavir/ritonavir. But Ulrik Dragsted (Hvidovre University Hospital, Denmark) said these differences lacked statistical significance. And an intent-to-treat analysis that included everyone who took their assigned regimen at least once showed that about 75 percent in both groups had a viral load under 400 copies/ mL after 24 weeks. Baseline numbers were similar in the two treatment arms; 60 percent entered the trial with some PI experience, and 24 percent started with a clean antiretroviral slate.

A salvage solution or too early to tell?

The GIGHAART presalvage treatment interruption trial probably settled one key question about this risky strategy. But the most important question--whether a treatment break before mega-regimen salvage is worth the risk--awaits longer follow-up and reporting of CD4 results, which weren't disclosed in Athens.

The question that GIGHAART probably answers is whether reversion of resistance mutations to susceptible wild-type virus during the break favors viral control once salvage begins. The answer is no. Although people randomized to take a break had a significantly better RNA response than people randomized to immediate salvage, wild-type virus didn't reappear in most break takers during their eight-week drug holiday.

Christine Katlama (Pitié Salpêtrière Hospital, Paris) and colleagues from across France recruited 68 people who had taken at least two NRTIs, two PIs, and one NNRTI [abstract 016]. Ninety percent had at least one major nonnucleoside-induced mutation, and 84 percent had more than two key protease mutations. The group's median viral load was above 150,000 copies/mL, and their median CD4+ count was below 30 cells/mm³. Half got randomized to an eight-week treatment break before starting a multiple-drug regimen, and the other half began the new regimen immediately. That heavy-duty combination included three or four NRTIs (usually with hydroxyurea), one NNRTI, and ritonavir (400 mg twice daily), amprenavir (600 mg twice daily), plus a third PI.

Thirty-two people in the immediate treatment arm and 30 in the interruption arm completed 12 weeks of treatment. At that point, in an intent-to-treat analysis, 59 percent of the interrupters and 26 percent of the noninterrupters met the primary endpoint of at least a 1-log drop in viral load (P = 0.007). Respective percentages with viral loads under 400 copies/mL were 35 percent and 15 percent (P = 0.05). In an on-treatment analysis, the difference between groups in sub-400 response was not statistically significant (40 versus 16 percent, P = 0.32). On-treatment analysis showed an average 1.9-log viral load drop in the treatment break group and a 0.4-log drop in the other group (P = 0.01).

Reversion of resistance mutations didn't explain the better RNA response by the break takers. Nineteen of the 34 interrupters had no reversions, while another five lost mutations to only one drug class. So about 70 percent had so few reversions that their better 12-week RNA response can't be explained by control of a newly drug-susceptible viral population.

A separate analysis by Katlama and colleagues at her center confirmed that reversions don't improve the RNA response when mega-HAART salvage follows a break.¹² This study involved 20 people with a median viral load of 160,000 copies/mL, treatment experience with every antiretroviral class, and plenty of resistance mutations to prove it. During breaks ranging from four to 24 weeks (median 8 weeks), 11 of 20 had reversions of mutations linked to either PIs or NRTIs. After six months taking a rescue regimen including five to nine drugs, median viral loads dropped about 2 logs among both the reverters and the nonreverters. During that time, resistant virus replaced susceptible virus that emerged during treatment breaks in eight of the 11 reverters.

A 32-person study of presalvage breaks presented in Athens by Christine Rouzioux (Necker Hospital, Paris) found no differences in viral load or CD4 responses between nine people with no reversions, 13 with partial reversions, and 10 with complete reversions [abstract P268]. This group began their breaks, which lasted a median of 4.5 months, with a lower average viral load (4.85 logs) and a much higher average CD4+ count (229 cells/mm³) than the GIGHAART population. The prebreak CD4+ count was the only variable that correlated with a good response to a postbreak regimen of four or more drugs (P = 0.03).

Together these three studies appear to sound the death knell for the wild-type reversion hypothesis. That's probably a good thing, because other studies show that resistant virus persists at undetectably low levels in people who seem to have reversions during drug breaks.¹³, 14 And those low-level mutants re-emerge quickly under renewed drug pressure, as Katlama's published study shows.¹² So something else--something more durable than apparent reversions, one hopes--must explain the better response in GIGHAART's treatment break group.

The explanation appears to be that GIGHAART's good responders had higher drug concentrations than the poor responders. The average viral load dropped 0.4 log among people with low drug levels, regardless of whether they took a break before salvage. Among people with adequate concentrations and no reversions, the average viral load fell 2 logs. People with decent drug levels and reversions did even better, averaging a 2.6-log viral load drop. But that extra 0.6-log drop can't mean much in a study this small, especially since so few people--two of 34--had the complete genotypic and phenotypic reversions that Veronica Miller originally tied to a better RNA response.¹⁵

There can be no surprise that adequate drug levels led to better responses than low levels in GIGHAART. If they didn't, you could start tearing up the pharmacology textbooks. Although Katlama didn't list how many people in each study arm had adequate drug levels, more interrupters than noninterrupters must have had good drug levels, simply because good levels meant better responses and the interrupters had significantly better responses. Probably not coincidentally, the 0.4-log drop in people with inadequate drug concentrations equals the 0.4-log drop in the noninterrupters. The question that must be answered, then, is why the break takers had higher drug concentrations. Some possible answers aren't mutually exclusive:

• They took more of their drugs on time.
  
• Their regimens had fewer deleterious drug-drug interactions.
  
• Chance.
  

It would be nice to eliminate the third possibility, though that could be hard in a 60-person study. The investigators may be able to dispense with the second possibility by looking more closely at the regimens used by each group and reporting individual drug levels. The first possibility could be confirmed, or at least buttressed, if GIGHAART included some adherence measure. Katlama didn't say. But one could fashion a rationale for better adherence in the treatment break group. Perhaps the eight weeks off therapy helped renew their commitment to treatment. And, in an openlabel study, clinicians inclined to believe in presalvage breaks might have bolstered the resolve of people randomized to the interruption group by consciously or unconsciously suggesting that they would have an edge.

This is sheer speculation but, if true, it would lend some logic to what must still be seen as a chancy strategy. Everyone in GIGHAART had fewer than 200 CD4+ cells/mm³, and most had many fewer than that. The median count numbered 26 cells/mm³ in the immediate treatment group and 28 cells/mm³ in the interruption group. Stopping treatment for eight weeks with so few T cells takes some nerve. One person in the interruption group died, though Katlama didn't say exactly when or why. Two interrupters suffered new opportunistic infections, as did three in the immediate treatment group.

Indeed, one shortcoming of Katlama's presentation is her failure to report CD4+ responses, as noted in Athens by an early and staunch supporter of megadrug rescue, Julio Montaner (University of British Columbia, Vancouver). Montaner's own nonrandomized study of "multiple drug rescue therapy"¹⁶ earned criticism because even people who got their viral load under 50 copies/mL twice in a row gained only a median of 2 cells/mm³. Seven of 106 people died during the study. A good viral load response in people with CD4+ counts as low as those in GIGHAART doesn't mean much if people don't add T cells.

A few weeks after the Athens meeting, Steven Deeks (University of California, San Francisco) detailed the dangers of interrupting treatment in people with low CD4+ counts.¹⁷ In his 23-person study, six people endured clinical setbacks during their treatment breaks or soon after restarting treatment. And these weren't minor problems: One person got Pneumocystis carinii pneumonia, one got severe peripheral neuropathy, two were diagnosed with nonHodgkin's lymphoma, one cytomegaloviral infection cropped up, and Kaposi's sarcoma worsened in one person, who died. Deeks also learned that one person lost to followup never restarted antiretrovirals and died of AIDS. All these people interrupted therapy with CD4+ counts under 100 cells/mm³.

Katlama maintained that CD4+ gains have been good so far in GIGHAART and that she would spell them out soon. A good immunologic response would certainly set GIGHAART's results apart from Montaner's--and from Katlama's own published analysis of heavy-duty salvage.¹² That 20-person study found that people lost a median of 15 cells/mm³ during their treatment break. The wild-type reverters returned to their prebreak CD4+ baseline after two months of treatment but added no more CD4+s through month six. People without wild-type reversions had a median count about 20 cells/mm³ above baseline after three months of treatment. A graph in the published article shows that the count plateaued there through month six. All told, the immunologic response in these 20 people was not reassuring, since 16 started with fewer than 200 cells/mm³.

Two other drawbacks to mega-HAART salvage will also have to be considered as more GIGHAART data emerge. The first is the prohibitive cost for people not enrolled in clinical trials or blessed with a munificent third-party payer. The second is toxicity. Katlama reported that people tolerated these seven- or eight-drug regimens reasonably well through the first three months of treatment. Only five had to pare their regimens to fewer than six drugs. Four had to drop ritonavir.

Those rates will almost certainly grow, since any triple-PI regimen is bound to take its toll on many an organ. In Montaner's megasalvage study people took a maximum of two PIs, and most took only five drugs.¹⁶ After a median 15 months of follow-up, 91 of 106 people (86 percent) endured lab abnormalities, characterized as "severe" in 26 (25 percent). Six people quit because of drug side effects, and three others bailed out without explaining why.

Salvage alternatives still slim

If the presalvage drug holiday in GIGHAART turns out to make a longterm difference, it will be critical to find out why so clinicians and people with HIV can weigh this strategy's certain risks and cost with hard data rather than hope. With people like these highly pretreated GIGHAART participants, resistance testing just wastes time and money. So a reasonable goal is keeping the CD4+ count as high as possible until drugs from new classes or with different resistance profiles can be combined. Does it take an eight-week holiday at a 30-cell CD4+ count and eight antiretrovirals to do that? Or can it be done with simpler, more tolerable regimens and without the drug break?

Rationales for continuing a virologically failing regimen, rather than taking a drug break, rest on findings that viral loads jump higher and CD4+ counts fall lower when faltering regimens stop,¹³, 15 and that people who continued taking a PI regimen when their CD4+ count fell under 50 cells/mm³ had a 43 percent lower risk of progression than people with sub-50 T-cell counts not taking a PI.¹⁸ Clinical researchers like Steven Deeks¹⁹ and Joel Gallant²⁰ suggest that incompletely suppressive regimens may be better than no regimen because they partly control replication or favor less fit resistant virus.

But even the most skeptical critic of presalvage break taking must admit that few reassuring deep salvage tactics leap to mind. Two groups combined PIs with decent resistance-fighting credentials, lopinavir and amprenavir, to little avail. Lowered amprenavir concentrations in people taking lopinavir/ritonavir may explain the disappointment, but neither group analyzed drug levels yet.

In a retrospective look at 33 people starting lopinavir with amprenavir for salvage and 35 starting lopinavir alone, Mona Loutfy (University of Toronto) detected no RNA or CD4+ advantage for the combined PIs after six to 11 months of treatment [abstract P66]. Stéphane De Wit (Saint-Pierre University Hospital, Brussels) discerned a trend to better virologic responses among people with fewer than three major mutations to lopinavir and one or no major mutations to amprenavir, but the difference wasn't statistically significant. After six months of lopinavir/ amprenavir, people with six or fewer major mutations to all PIs had a median 1.3-log drop in viral load, compared with a 0.1-log drop in people with more than six major mutations (P = 0.02). Yet the median CD4+ gain at that point measured only 18 cells/mm³. Among people who stayed with the regimen for 12 months, the median CD4+ count had fallen 12 cells/mm³ below baseline.

Besides the preliminary GIGHAART findings, the only bit of good news for treatment-experienced people came in a study of tenofovir added to a stable but incompletely suppressive regimen [abstract 017]. This double-blind trial randomized 368 people to add tenofovir at 300 mg once daily and 184 to add placebo. People in this study weren't in the dire straits of the GIGHAART enrollees: 58 percent had PI mutations and 48 percent had NNRTI mutations. The median RNA stood at 4,502 copies/mL in the tenofovir group and 4,366 copies/mL in the placebo arm. Respective CD4+ counts were 417 and 447 cells/mm³.

Anton Pozniak (Chelsea and Westminster Hospital, London) reported that 45 percent taking tenofovir and 13 percent taking placebo got their viral load under 400 copies/mL by week 24. Respective proportions with fewer than 50 copies/mL were 22 percent and 1 percent. The timeweighted average RNA change from baseline measured -0.61 log with tenofovir versus virtually no change with placebo (P < 0.0001 for all three comparisons). Side effect rates were similar and low in the two study arms.

T-cell shifts during treatment breaks

Three other studies looked at CD4+ cell and viral load changes in people who stopped their antiretrovirals in clinical trials or on their own. The CD4+ drops were rarely precipitous, especially in well controlled studies. But there are exceptions. And whether the drops were speedy or slow, the loss of T cells shows that treatment breaks do little to rally the immune system during chronic infection.

Davide Bertelh (Spedali-Civili, Brescia, Italy) retrospectively analyzed CD4+ changes in 33 people who stopped treatment for four to 52 weeks (median 21 weeks) [abstract P68]. The median CD4+ count dropped from 636 cells/mm³ (range 200 to 1,218 cells/mm³) to 417 cells/mm³ (range 89 to 829 cells/mm³, P < 0.01). In seven of the 33, CD4+ counts remained stable or actually increased, but in everyone else they fell. Nineteen (58 percent) suffered immunologic and virologic deterioration within a month of stopping their antiretrovirals. Seven (12 percent) lost more than half of their CD4+ cells within three months of stopping. These seven included two with an undetectable viral load when they started their drug holiday. In one the CD4+ count crashed from 1,139 to 369 cells/mm³, and in the other from 505 to 145 cells/mm³. The latter person suffered an acute mononucleosislike syndrome. Another person got sent to the hospital because of cryptosporidiosis.

A T-cell analysis of 91 people who completed four two-week treatment breaks then stopped antiretrovirals under close observation charted an initial sharp drop in CD4+ cells, but then a much slower decline-perhaps a plateau [abstract 026]. During the four cycles of two weeks off treatment and eight weeks on, most CD4+ counts remained stable, as they did in a published study of 10 people going off and on treatment every week.²¹

People in both of these studies had viral loads under 50 copies/mL at screening and robust T-cell tallies. In the Swiss Spanish Intermittent Treatment Trial (SSITT), presented by Catherine Fagard in Athens, the median starting CD4+ count stood at a giddy 740 cells/mm³. The 91 people who began their long breaks at week 40 represent the best responders among the 133 enrollees; 42 (32 percent) had already been eliminated from the study, usually because their viral load didn't fall back below 50 copies/mL after a two-week break. But four people had to drop out because their CD4+ counts fell under 400 cells/mm³.²²

Among 82 people who stopped treatment at week 40 and went 12 weeks without restarting, the average CD4+ count slipped from 775 to 600 cells/mm³ (P < 0.001). This decrease correlated with the viral load rebound after week 40 (r = 0.3, P = 0.019). But from week 52 to the last followup, the average CD4+ count shed only another 17 cells/mm³, a nonsignificant drop.

These findings indicate that most people who can maintain a healthy baseline CD4+ count through four two-week treatment breaks could lose a couple hundred T cells when they then take a 12-week break. But that decrease, though statistically significant, probably means nothing clinically to people with 700 cells/mm³. As the treatment break continues, the CD4+ drop seems to slow.

Other researchers have also reported this two-phase CD4+ swoon in people taking drug holidays. A study of 16 people with a prebreak count of 775 cells/mm³ measured a 44 cell per month drop in the first six months without antiretrovirals, and a 17 cell per month fall after that.²³ Another study of 72 people who stopped treatment with an average 571 cells/mm³ lost an average 16 cells per month over a mean 45 weeks of follow-up.²⁴ But again most of the drop came in the first few months.

Of course these are all averages. Some people with good CD4+ numbers, as in Bertelli's study, can see them plunge during drug breaks. Generally, though, people who start drug breaks with more T cells are safer than those who start with fewer, as one would expect. In fact, pretreatment CD4+ counts predicted better sustained counts after the long break following week 40 in SSITT. Among 48 people with pretreatment baselines above 400 cells/ mm³, 44 (92 percent) kept their count above 400 cells/mm³ during the long treatment interruption. But among 34 with pretreatment baselines under 400 cells/mm³, only 18 (53 percent) maintained a count above 400 cells/mm³ during the post-40-week break (P = 0.0003). Since treatment guidelines now uniformly recommend delaying antiretroviral therapy until CD4+ counts fall to 350 cells/mm³, or less, clinicians could see a dwindling pool of people who respond to SSITT-like interruptions with steady CD4+ counts.

Even if scrupulously monitored treatment breaks are safe for most people with high pre-HAART CD4+ nadirs and HAARTboosted counts of 600 or 700 cells/mm³, the question remains whether such breaks do any good immunologically. A detailed survey of immune markers in 15 people with well-controlled viremia suggests the answer is no [abstract P69]. These 15 all had viral loads below 25 copies/ml, for at least 13 months before stopping antiretrovirals during a clinical trial. They had no virus detectable in cerebrospinal fluid and no or minimal residual replication in lymphoid tissue. But after four to eight weeks without antiretrovirals, reported Jan van Lunzen (University Hospital Eppendorf, Hamburg), all had viral rebounds to pre-HAART levels.

During the drug breaks, activation markers popped up on CD4+ and CD8+ cells, and a rising fraction of cycling CD4+s and CD8+s indicated increased T-cell turnover. T-cell responses to mitogen were mixed after the interruption, and van Lunzen saw no clear-cut pattern of response to stimulation with recall antigens. Although HIV specific cytotoxic T lymphocytes emerged during the interruption, they did nothing to forestall rebounds and disappeared when treatment resumed.

These results, van Lunzen proposed, argue that treatment breaks during wellcontrolled chronic infection do not induce "immunologic control of viral replication by induction of HIV specific immunologic responses." One might argue that repeated breaks, rather than the one in this study, may eventually kindle an HIV-specific response. But that didn't happen in two published studies of repeatedly interrupted HAART²⁵, 26 HIV specific immune responses in SSITT have yet to be reported.

Paying the debt to Asclepius

Plato reports that Socrates prepared for death with an arching summary of his philosophy and an eloquent assertion of the soul's immortality. But on the last page of Phaedo, after Socrates has quaffed the fatal hemlock, something funny happens. Lying on his back as death steals coldly toward his heart, Socrates covers his face to spare his friends the spectacle. Then, in a scene straight from opera buffa, he yanks down the cloth and addresses a young colleague.

"Crito, I owe a cock to Asclepius; will you remember to pay the debt?"²⁷

Philosophy's exegetes have had a field day with that one. Is Socrates solemnly paying his respects to Asclepius, the original healer, for freeing him from life's sorrows? Or is this yet another example of wry Socratic wit, as the master punctures the scene's solemnity by congratulating Asclepius for brewing an immediately effective medicine?

Without presuming to contribute anything profound to this highbrow hypothesizing, one could venture a third opinion. Maybe Plato sought to cap this erudite dialog with a down-to-earth authenticating detail--the chance recollection of a half-forgotten promise, an obeisance owed to the medical profession for a long and healthy life. When handed the hemlock, Socrates was in his 70s and going strong--not bad for the 5th Century BC.

That last interpretation might find the most favor with physicians, maybe especially HIV docs. Those who have been at this more than a while started out by honing their palliative skills, soothing the incurable with hands-on care, and going to funerals. And often these sincere but ultimately futile ministrations earned clinicians--if not a sacrificial cock--a hug and a thanks.

Then technology started gaining on HIV, equipping clinicians not only with potent drugs, but also with viral load assays, resistance kits, drug level monitoring, and--maybe soon--viral fitness tests. Along the way, some fear,²⁸ the laying on of hands got lost. So when a late-breaking poster at the Athens meeting argued the value of yet another monitoring tool, one could be forgiven for fleeing the poster tent to gaze across the sparkling Saronic Gulf. But that would have been a mistake, because the poster by Rita Murri (Catholic University of Rome) suggested that clinicians refocus, in a structured way, on that simple question, How are you feeling?

In a prospective, multicenter cohort study, Murri and coworkers sized up the predictive prowess of a physical health score [see note 29], a mental health score, a symptom score, and other more traditional talismans of HIV disease progression and death in 807 people [abstract LB/P24]. In a multivariate Cox model adjusting for age, gender, RNA during follow-up, CDC disease stage, symptom score, and mental and physical status scores, three factors signaled a change in the risk of disease progression or death: CDC grade C disease raised the risk 3.63 times (P < 0.0001), a CD4+ count under 200 cells/mm³ raised the risk 2.1 times (P = 0.01), and every five-point increment in the physical health score lowered the risk 15 percent (P = 0.001).

The physical health score, Murri concluded, predicts clinical progression. "Health status assessment should be incorporated into clinical practice for routine follow-up as well as for clinical decision making," she argued.

Perhaps more than one reader, weighing this conclusion, is thinking, No, I really would rather be gazing at the sparkling Saronic Gulf. To be sure, at first the study's results seem tautological, almost trite. People who are feeling sick very often are sick, and sick people are more likely to get worse and die than people who don't feel sick. Socrates would not stoop to such simple-minded syllogizing. The point, though, is not that the argument is obvious and irrefutable, but that the tyranny of technomedicine and time constraints can blur the obvious and irrefutable.

A delightfully simple conclusion of another Athens poster suggests a specific instance of how the obvious can be overlooked. This study of 35 people with DEXA scans done at least a year apart showed little decline in bone mineral density [abstract P138]. In a multivariate analysis Antonia Moore (Royal Free and University College Medical School, London) found only two factors that predicted stable or worsening bone mass. A higher CD4+ count lowered the chance of thinning bones, but that chance rose in people who complained of backaches or joint pain. Even without having people fill out a health status questionnaire, Moore found that asking about achy bones reliably identifies people with osteopenia.

"The body," Socrates says in Phaedo, "is in the very likeness of the human, and mortal, and unintellectual, and multiform, and dissoluble, and changeable."³⁰ T-cell counts, and viral loads, and resistance tests are good ways to chart those changes. Murri and Moore remind everyone that there are others.

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

References and Notes

* Abstracts from the 8th European Conference begin with a P for poster or an O for oral presentation.

1.. Asclepius. Encyclopaedia Britannica CD 98.

2.. Masquelier B, Neau D, Chene G, et al. Mechanism of virologic failure after substitution of a protease inhibitor by nevirapine in patients with suppressed plasma HIV-1 RNA. J Acquir Immune Defic Syndr 2001 Dec 1;28(4):309-12.

3.. Phillips AN, Miller V, Sabin C, et al. Durability of HIV-1 viral suppression over 3.3 years with multi-drug antiretroviral therapy in previously drug-naïve individuals. AIDS 2001 Dec 7;15(18):2379-84.

4.. de Wolf F, Lange JIMA, Bossuyt PMM, et al. Monitoring of human immunodeficiency virus type 1 (HIV-1) infection in the Netherlands: ATHENA. First edition. July 2001.

5.. Kinloch S, Hoen B, Cooper D, et al. Zidovudine versus placebo in primary HIV infection. Fourth European Conference on Clinical Aspects and Treatment of HIV Infection. March 16-18,1994. Milan. Abstract 039.

6.. Lucas GM, Chaisson RE, Moore RD. Comparison of initial combination antiretroviral therapy with a single protease inhibitor, ritonavir and saquinavir, or efavirenz. AIDS 2001 Sep 7;15(13):1679-86.

7.. Matthews GV, Sabin CA, Mandalia S, et al. Virological suppression at 6 months is related to choice of initial regimen in antinetroviral-naïve patients: a cohort study. AIDS 2002 Jan 4;16(1):53-61.

8.. Phillips AN, Pradier C, Lazzarin A, et al. Virological and clinical outcome of NNRTI-containing regimens for 1932 patients in EuroSIDA. Conf Retroviruses Opportunistic Infect. 2001 Feb 4-8;8th:Abstract No. 324.

9.. van der Valk M, Kastelein JJP, Murphy RL, et al. Nevirapine-containing antiretroviral therapy in HIV-1 infected patients results in an anti-atherogenic lipid profile. AIDS 2001 Dec 7;15(18):2407-14.

10.. Burnside A, Bush L, DeJesus E, et al. Effect of low-dose ritonavir on metabolic parameters in therapy-naïve subjects started on amprenavir/ lamivudine/abacavir (GSK COL30325). Antiviral Ther 2001;6(suppl 4):32. Abstract 41.

11.. Hulse SE, Bleichner KM, Keiser P, Nassar N. A comparative, retrospective analysis comparing lipid levels in individuals treated with ritonavir versus non-ritonavir-containing antiretroviral regimens. Antiviral Thor 2001;6(suppl 4):50. Abstract 72.

12.. Delaugerre C, Valantin MA, Mouroux M, et al. Re-occurrence of HIV1 drug mutations after treatment re-initiation following interruption in patients with multiple treatment failure. AIDS 2001 Nov 9;15(16):2189-91.

13.. Deeks SG, Wrin T, Liegler T, et al. Virologic and immunologic consequences of discontinuing combination antiretroviral-drug therapy in HIV-infected patients with detectable viremia. N Engl J Med 2001 Feb 15;344(7):472-80.

14.. Hance AJ, Lemiale V, Izopet J, et al. Changes in human immunodeficiency virus type 1 populations after treatment interruption in patients failing antiretroviral therapy. J Virol 2001 Jul;75(14):6410-7.

15.. Miller V, Sabin C, Hertogs K, et al. Virological and immunological effects of treatment interruptions in HIV-1 infected patients with treatment failure. AIDS 2000 Dec 22;14(18):2857-67.

16.. Montaner JS, Harrigan PR, Jahnke N, et al. Multiple drug rescue therapy for HIV-infected individuals with prior virologic failure to multiple regimens. AIDS 2001 Jan 5;15(1):61-9.

17.. Deeks SG. STI in patients with incomplete viral suppression. 41st Interscience Conference on Antimicrobial Agents and Chemotherapy. December 16-19, 2001. Chicago. Abstract 422.

18.. Miller V, Mocroft A, Clotet B, et al. Association of Viral Load, CD4 Cell Count, and Treatment with Clinical Progression in HIV Patients with Very Low CD4 Cell Counts: The EuroSIDA Cohort. Conf Retroviruses Opportunistic Infect 2000 Jan 30-Feb 2; 7:161 (Abstract 454).

19.. Deeks SG, Martin JN. Reassessing the goal of antiretroviral therapy in the heavily pre-treated HIV-infected patient. AIDS 2001 Jan 5;15(1):117-9.

20.. Gallant JE. Antiretroviral therapy: starting, continuing, and fixing. Medscape HIV/AIDS Annual Update 2001. Online at http://hiv.medscape.com/Medscape/HIV/AnnualUpdate/2001/mha.update06.03.gall/mha04.gall-01.html. Accessed December 13, 2001.

21.. Dybul M, Chun TW, Yoder C, et al. Short-cycle structured intermittent treatment of chronic HIV infection with highly active antiretroviral therapy: effects on virologic, immunologic, and toxicity parameters. Proc Natl Acad Sci U S A 2001 Dec 18;98(26):15161-6.

22.. Fagard C, Lebraz M, Gunthard H, et al. SSITT: a prospective trial of strategic treatment interruption in 128 patients. Conf Retroviruses Opportunistic Infect. 2001 Feb 4-8;8th:Abstract No. 357.

23.. Mussini C, Bugarini R, Perno C, et al. Virological and immunological effects of discontinuation of antiretroviral therapy in HIV-positive patients with virological failure and a CD4 count above 500 cells/ml. 1st IAS Conference on HIV Pathogenesis and Treatment. July 8-11, 2001. Buenos Aires. Abstract 110.

24.. Tebas P, Henry K, Mondy K, et al. Effects of prolonged discontinuation of successful antiretroviral therapy. IAS Conf HIV Pathog Treat 2002 Jul 8-11;1st: Abstract No. 31.

25.. Ruiz L, Carcelain G, Martinez-Picado J, et al. HIV dynamics and T-cell immunity after three structured treatment interruptions in chronic HIV-1 infection. AIDS 2001 Jun 15;15(9):F19-27.

26.. Garcia F, Plana M, Ortiz GM, et al. The virological and immunological consequences of structured treatment interruptions in chronic HIV-1 infection. AIDS 2001 Jun 15;15(9):F29-40.

27.. Dialogues of Plato. Jowett Translation. New York: Washington Square Press. 1967:160.

28.. Selwyn PA, Arnold R. From fate to tragedy: the changing meanings of life, death, and AIDS. Ann Intern Med 1998 Dec 1;129(11):899-902. "Before antiretroviral therapy," Selwyn and Arnold write, "physicians learned how to accompany patients through their illness; to bear witness to sickness and dying; and to help patients and their families with suffering, closure, and legacy. Since we have became better at treating the virus, a new temptation has emerged to dwell on quantitative aspects of HIV management and monitoring, although the skills that we learned earlier in the epidemic are no less necessary for providing good care."

29.. Murri and colleagues derived physical health and mental health scores from an authenticated Italian version of the MOS-HIV Health Survey, which evaluates 11 factors: pain, physical functioning, role functioning, social functioning, mental health, vitality, health distress, cognitive functioning, general health perception, health transition, and overall quality of life.

30.. Dialogues of Plato. Jowett Translation. New York: Washington Square Press. 1967:103,

020210
IA020204

Copyright © 2002 - International Association of Physician's in AIDS Care (IAPAC). Reproduction of this article (other than one copy for personal reference) must be cleared through the IAPAC.

AEGiS is made possible through unrestricted grants from Boehringer Ingelheim, iMetrikus, Inc., John M. Lloyd Foundation, the National Library of Medicine, and donations from users like you. Always watch for outdated information. This article first appeared in 2002. This material is designed to support, not replace, the relationship that exists between you and your doctor.

AEGiS 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 ©1980, 2002. AEGiS. All materials appearing on AEGiS are protected by copyright as a collective work or compilation under U.S. copyright and other laws and are the property of AEGiS, or the party credited as the provider of the content. .