Retro Part 1: A Suite of New Therapies (or, My Funny Valentine)
10th Conference on Retroviruses and Opportunistic Infections, February 10-14, 2003, Boston
Some 20 years ago Françoise Barré-Sinoussi spotted RNA in the biopsied lymph node of a 33-year-old man with an apparently acquired immune deficiency. Ten years after that U.S. clinicians and researchers organized "The First National Conference on Human Retroviruses and Related Infections." And 10 years on most of those same clinicians -- lacking too many of their 1993 patients and gaining too many more -- gathered again for the tenth refrain of what has become the Conference on Retroviruses and Opportunistic Infections (CROI). This year's CROI (rhymes with Troy) had some novel features:
This tender lovers' fete -- celebrated in defiance of dark winter's remorseless slog -- might seem a cruelly ironic capstone for a meeting about a lethal virus spread in bed. But, in truth, there has always been something a little sad and sinister about Valentine's Day.
After all, on the liturgical calendar February 14 commemorates one of two people (as with many things liturgical, there is some confusion), both named Valentinus and both slain by the Romans in the unhappy third century (by beheading or "languishing in a dungeon," depending on the source). But no apparent link has yet emerged between these Valentines and the ones traded by schoolchildren.1 Then there was the fabled St. Valentine's Day Massacre, a Chicago gangland slaughter visited upon hirelings of George ("Bugs") Moran by triggermen allied to Al Capone. Even the most touching of Valentine ballads, "My Funny Valentine," makes its mark mainly through Lorenz Hart's rueful lyrics:
Your looks are laughable,
Indeed, when you look at antiretroviral art through Hart's self-mocking lens, Valentine's Day isn't a bad match. Some tries at thwarting the virus could be called pathetic (if not exactly laughable), and some of this therapy's failures ought to be unphotographable (though cruel science demands otherwise). But antiretrovirals -- commonly abbreviated ARTs -- have proved "favorite works of art" to countless individuals and vast populations.
Tracking rates of AIDS and death in nearly 10,000 EuroSIDA cohort members, Amanda Mocroft (Royal Free and University College Medical School, London) traced a continued dwindling of both endpoints through the most recent year of follow-up, 2002 [abstract 180*, see "Of Cohorts and Copies" below]. Reckoning "excess death rates" in Swiss people with HIV infection relative to the whole population, Bernard Hirschel (University Hospital, Geneva) learned that those with HIV (but without hepatitis C) have a lower excess death rate than successfully treated cancer patients, "a group who is able to obtain life insurance" [abstract 917a, see "Of Cohorts and Copies" below].
There's another way this year's CROI at first seemed to resemble Valentine's Day -- in carefree excess. One does not buy one's honey just one rose, but a dozen roses. For one's sweetie, a few well-picked bon-bons are a non-non; one needs a double-decker box of caloric bomblets. Just so, the stacks of tempting new antiretrovirals on review at CROI recalled juvenile fantasies of entrapment in a candy factory.
At the meeting's opening press conference, John Mellors (University of Pittsburgh) alerted scribes that the "pipeline of new antiretrovirals is fuller than it's been for a long time." And the riches didn't stop there. Alluring cellular (CEM15) and genetic (siRNA) strategies outlined by Mario Stevenson (University of Massachusetts, Worcester) bespoke an ever-sharpening focus on novel shortcuts to arrest the retrovirus (see the next section).
Impressed by this honest enthusiasm, and caught up in the 10th CROI anniversary spirit, this reporter thought it would be clever to compare 2003's deluxe bouquet of fresh therapies with the meager, seedy offerings of 1993. Further, a calculation of how many "new drugs" of yesteryear later won regulatory sanction seemed instructive. Entry criteria for the list were simple: at least one meeting abstract for an agent -- or cellular or genetic manipulation -- that constrained viral replication or boosted CD4 cells in people, other animals, or cell cultures.
As anticipated, the list for this year's conference impresses (Table 1). One could count six new protease inhibitors (PIs), four nucleosides (NRTIs), two nonnucleosides (NNRTIs), four integrase inhibitors (still not called IIs), 13 binding, fusion, or entry inhibitors -- even a viral budding inhibitor -- plus a suite of immune-based therapies and the aforementioned genetic étoile du jour, small interfering RNA (siRNA). The total of vernal viral nemeses reached a lavish 45.
Then there was 1993. The list of hopefuls included some future stalwarts -- lamivudine (3TC), nevirapine, and indinavir. But who, besides the presenters, remembers cyclopiazonic acid (another Merck PI), RO24-7429 (Roche's tat inhibitor), or the organosilicon compound from the University of Bordeaux? Toting up this 10-year-old list, one counts -- 45 vernal viral nemeses!
Of course one could argue that the 1993 roster includes some off-the-wall offerings (even back then lots of docs looked askance at infusing people with plasma rich in anti-p24 antibody) and some dandy shockers (hypericin, the prime mover in St. John's wort, studied in ACTG protocol 150). But the list also includes a few agents still hotly debated, and used, today, viz hydroxyurea (in a Franco Lori poster) and interleukin-2 (with Cliff Lane as the senior author). IL-2 isn't the only therapy to appear on both lists -- the others are FTC, now called emtricitabine, the embattled immune stimulant Remune, and interferon alfa2B. And who's to say the failures of yore -- for example, antisense -- made less sense then than today's genetic pole sitter, siRNA? To be sure, many a gene therapy and antisense savant from 1993 resurfaced in 2003 espousing siRNA.
So here we are 10 years later, without the benefit of hindsight that smarty pants of 2013 will enjoy, and the job of everyone with a stake in stopping HIV remains the same: Without passion (ardent Valentines begone), fathom the cold data coolly gathered by the world's top HIV researchers and decide which will blossom in one year or two, which will flourish as the decade matures, and which will wither neath the insensate harrow of scientific scrutiny.
Every HIV meeting worth sitting through starts with the same question: What's new? CROI's organizers give reporters a head start by stapling together a sheaf of abstracts with trenchant analyses by scientific committee members. Some of those Solons show up at a first-day press conference to construe their best-of-show picks and to twine the strands of discrete studies into signals of important change. Hence, John Mellors, as just mentioned, nudged the new-therapies bandwagon to the starting line, inviting willing reporters to push. Eyeing vaccine research and some budding benchwork, David Ho (Aaron Diamond Center, New York) and Mario Stevenson also espied possibly pivotal developments. There were three, and a fourth caught the fancy of the mainstream press:
HIV vaccine research began -- and has continued for many a year -- with attempts to summon antibodies that will neutralize a newly transmitted retrovirus. Early on, this worked well in the lab, but not on the street or in the field -- breeding grounds for unimpressed "field isolates." So vaccine research reversed course, prodding troops of viral proteins to excite cell-mediated responses that would stymie HIV after infection but not prevent infection itself.
Bad news on that front, said David Ho. Daniel Barouch, working in Norman Letvin's Beth Israel Deaconess lab in Boston, reported that three of four monkeys given a vaccine that dragoons cytotoxic T lymphocytes (CTLs) against the virus had fallen sick after only three years of follow-up [abstract 76]. The infecting simian immunodeficiency virus, or SIV, had wriggled free from strong-arm CTLs through "a stereotypic pattern of viral escape." This doleful outcome does not spell the end of so-called prime-boost CTL strategies because the animals got only the priming shots and newer prime-boost medleys may prove more potent.
But this study and others, Ho commented, make it clear that "protection against disease progression may be lost with time" if one riles only cell-mediated immunity. So, it's back to antibodies, in Ho's opinion. He cited "significant progress" in finally tracking down spots on HIV's envelope that may prove vulnerable to antibody attack. "The new information," he told the press corps, "is already generating novel ideas for creating 'immunogens' that could be tested in animals for the ability to induce HIV neutralizing antibodies."
The tenth CROI had no word on one antibody vaccine -- in fact, the only vaccine -- to confront HIV in big efficacy trials, VaxGen's AIDSVAX. But that news came a few weeks later, confirming majority opinion that the gp120 vaccine would not work and rousing controversy about whether the small numbers of African Americans and Asians enrolled supported a claim that AIDSVAX may protect them.
Tufts University's John Coffin, in the first talk at the Boston gathering [abstract 1], displayed two stark and still inescapable statistics:
Number cured of HIV: <1
And, Coffin added, these null results "will probably not change for a considerable time." VaxGen later claimed that Coffin's second dictum no longer holds -- in African Americans and Asians -- but many demurred pending closer scrutiny of the AIDSVAX results and their statistical analysis.
If it has nothing else going for it, small interfering RNA has a sleek, digital-age moniker: siRNA. But, then, antisense and even gene therapy exuded linguistic cachet in their prime.
Here's an amazing tale. "Because of proteins like APOBEC," according to Mario Stevenson, "humans are innately resistant to infection by viruses like HIV." So what goes wrong? Most human cells carry an APOBEC protein called CEM15, which slams the door on HIV replication. But HIV, never at a loss, has a gene whose sole function seems to be knocking out CEM15. The gene is vif, the appropriately named viral infectivity factor.
Michael Malim (King's College, London) amplified on this classic tug of war in a plenary talk [abstract 5]. The vif gene and its protein product Vif, he explained, show up in most lentiviruses and in all mammalian lentiviruses, such as HIV-1, HIV-2, SIV, and FIV. Vif makes budding virions infectious. Without Vif, virions still bud from infected cells, but they can't infect fresh cells. Malim offered some details:
To get a better handle on how APOBEC/CEM15 stymies HIV, Roberto Mariani and colleagues at the Salk Institute in La Jolla tracked the action of this protein's mouse equivalent [abstract 72]. They found it in lymphoid tissues including spleen, thymus, and lymph nodes, and in heart and lung. Mouse APOBEC shut down HIV replication even better than human APOBEC. As little as 1 ng, the Salk team learned, "effectively blocked HIV-1 infectivity." And HIV-1 Vif failed to reverse this action unless APOBEC levels were low.
Can science exploit CEM15's innate antiviral moxie in humans? It should be "considered," Malim opined, but it won't be easy. Eviscerating vif may be preferable to boosting CEM15, because this protein's panoply of activities remains poorly understood. In animal models, Malim noted, meddling with CEM15 makes tumors grow.
Can you fight a virus with a virus? If one is GBV-C and the other HIV-1, the answer may be yes. GBV-C, once labeled hepatitis G virus, turned out not to cause hepatitis or any other known disease. But when GBV-C actively replicates in people infected with HIV, it seems to blunt HIV's virulence. If a GBV-C-infected person clears that virus, however, HIV runs amok.
Studying 230 people with HIV infection, Per Björkman (Malmö University Hospital, Sweden) found that harboring active GBV-C at HIV diagnosis does not affect progression to AIDS or death [abstract 157]. But the viruses seem to interact. Whereas two of 31 people (6 percent) who had AIDS when diagnosed with HIV also had GBV-C in their blood, 54 of 175 (31 percent) diagnosed with asymptomatic HIV disease carried GBV-C (P=0.008).
During a median 4.33 years of follow-up, 11 of 44 people originally infected with GBV-C cleared that virus without evidence of GBV-C antibody seroconversion. Compared with others in the study, those people had:
Scrutinizing stored plasma samples from gay men enrolled in the Multicenter AIDS Cohort Study (MACS), Carolyn Williams (National Institute of Allergy and Infectious Diseases, Bethesda) and MACS colleagues confirmed two of Björkman's findings: Early GBV-C status does not affect survival (after five to six years in this study). And clearing GBV-C has nasty consequences [abstract 159lb].
Williams found that GBV-C status at the five- to six-year follow-up visit had a dramatic impact on mortality. Compared with men who had circulating GBV-C at both visits:
Men with GBV-C viremia at both visits lost an average 26 CD4 cells/mm3 yearly; men negative for GBV-C at both visits lost 37 cells/mm3 yearly; and men who cleared GBV-C between visits lost 107 cells/mm3 yearly. These three groups did not differ in median duration of follow-up, date of seroconversion, antiretroviral use, median HIV load, or prevalence of the protective CCR5 delta32 mutation.
What's the link between GBV-C and HIV? Coreceptors and chemokines, according to a cell study by Jack Stapleton (University of Iowa, Iowa City) [abstract 156]. He found curtailed expression of the HIV coreceptors CCR5 and CXCR4 on peripheral blood mononuclear cells (PBMCs) exposed to GBV-C isolates or infectious clones. Inhibition of HIV replication in these cells required ongoing GBV-C replication. Compared with mock-infected PBMCs, cells spiked with GBV-C produced more of the cytokines IL-2 and IL-8 and more of the chemokines MIP-1 alpha, MIP-1 beta, SDF-1, and RANTES.
Stapleton believes results of his experiments and the two cohort studies mean GBV-C infection may have "a direct inhibitory effect" on HIV replication. The Swedes aren't so sure. They suggest that "GBV-C status in HIV-1 infection is probably a secondary phenomenon during disease progression rather than an independent prognostic factor." But Stapleton thinks researchers should even consider injecting GBV-C into people already burdened by HIV.3 Yet the notion of exposing people to a virus discovered only nine years ago -- even though it seems innocuous now -- sounds scary to some, because the immune system could not be allowed to clear GBV-C. All three studies suggest that, once infected with GBV-C, getting rid of that virus would be worse than never having it at all.
Only a few years ago, drug industry merger mania occasioned concern over the future of antiretroviral development. How many HIV drugs could one reasonably expect from one company where once there had been two, three, or even four? Indeed, one heard nothing at the 10th CROI about the second-generation NNRTIs once being developed by DuPont but now the property of Bristol-Myers Squibb. And if GlaxoSmithKline buys Bristol-Myers Squibb, what then? Yet the bleakest versions of this doomsday scenario remain dark imaginings. Instead, old hands long absent from the fray (Pfizer, Schering-Plough) and fresh faces with new wares (Tanox, Panacos) peopled the new antiretroviral slide session and poster boards.
But wait. Before auditioning CROI's cavalcade of new candidates, consider two mainstays of the antiretroviral repertoire -- nevirapine and efavirenz -- on stage together for the first time. Well, at least for the first time in a big, multicontinental, randomized trial: 2NN.
Before there was 2NN, the 1,216-person head-to-head showdown between nevirapine and efavirenz, there was SENC, the Spanish Efavirenz-Nevirapine Comparison trial, which also randomized treatment-naive people to begin one or the other NNRTI plus two nucleosides.4 This little-noted trial differed from 2NN in two important ways: it enrolled only 67 people, and it excluded people with viral loads above 100,000 copies/mL. After 48 weeks SENC's intent-to-treat analysis determined that 64 percent assigned to nevirapine and 74 percent to efavirenz had a viral load under 50 copies/mL. But that difference lacked statistical significance. The authors appropriately noted that such a difference may have emerged in a bigger trial.
2NN was 18 times bigger than SENC, and nevirapine still kept statistical pace with efavirenz through 48 weeks, even among people who began treatment with a viral load above 100,000 copies/mL. Joep Lange, who heads the IATEC trials group at the University of Amsterdam, and IATEC's Frank van Leth spelled out the details in a slide talk [abstract 176] and poster [abstract 752].
Lange and colleagues on six continents randomized untreated adults to take stavudine (d4T) and 3TC plus 400 mg of nevirapine once daily, 200 mg of nevirapine twice daily, standard once-daily efavirenz, or nevirapine plus efavirenz at 400/800 mg once daily. About 400 people got assigned to arms two and three to ensure statistical power to discriminate between those standard doses, while about 200 got assigned to the unconventional dosage arms, one and four. Baseline traits matched well from arm to arm, with (for all groups) 63 percent men, a median age of 34 years, a median CD4 count of 190 cells/mm3 (range 70 to 330 cells/mm3), and a median viral load of 4.7 logs (range 4.4 to 5.5 logs). Equivalent proportions had hepatitis B virus (HBV) coinfection (5.3 percent overall) or hepatitis C virus (HCV) coinfection (9.5 percent overall).
The researchers defined failure by four criteria:
A week 48 missing-data-equal-failure analysis found a significant response difference only between the efavirenz arm and the nevirapine/efavirenz arm (Table 2). Nor did the single-nevirapine arms differ from the single-efavirenz arm in intent-to-treat or on-treatment analyses of proportions with viral loads below 50 copies/mL at 48 weeks. For people starting therapy with a viral load above 100,000 copies/mL, those taking efavirenz met the trial's success criteria slightly more often (61.3 percent) than those taking nevirapine/efavirenz (57.1 percent), twice-daily nevirapine (57.3 percent), or once-daily nevirapine (51.5 percent), but between-arm differences fell shy of statistical significance. CD4 gains proved equivalent across arms.
The 2NN team did tease out some differences in side effect rates. Grade 3 or 4 hepatobiliary side effects trended higher in the twice-daily (2.6 percent) and once-daily (1.8 percent) nevirapine groups than in the double-NNRTI arm (1.0 percent) or the efavirenz arm (0.5 percent) (P=0.082 across arms). Clinical hepatitis also cropped up more in the nevirapine groups (2.1 and 1.4 percent) than with nevirapine/efavirenz (1.0 percent) or efavirenz (0.3 percent), but these differences lacked statistical significance. (One woman without evidence of HBV or HCV coinfection died of hepatitis while taking nevirapine.) The once-daily nevirapine group had a significantly higher rate of grade 3 or 4 hepatobiliary lab toxicities (13.2 percent) than did the efavirenz arm (4.5 percent,PGrade 3 or 4 rash proved more common with once-daily (4.1 percent) and twice-daily (3.1 percent) nevirapine than with efavirenz (1.8 percent), though again these differences lacked statistical significance. (A man taking nevirapine died of septicemia while recovering in the hospital from Stevens-Johnson syndrome.) Grade 3 or 4 central nervous system or psychiatric side effects troubled more people in the efavirenz arms (5.7 percent with nevirapine/efavirenz and 5.5 percent with efavirenz) than in the nevirapine arms (3.6 percent twice daily and 1.4 percent once daily) (P=0.001 across arms).
The bottom line in the toxicity match-ups is a distinct disadvantage for nevirapine/efavirenz, with 29.7 percent in that group stopping therapy temporarily or permanently for any side effect, compared with 24.1 percent for once-daily nevirapine, 21.2 percent for twice-daily nevirapine, and 15.5 percent for efavirenz (P<0.001 across arms). Are those differences between the single-nevirapine arms and the single-efavirenz arm statistically significant? By the strictest interpretation, Frank van Leth told IAPAC Monthly, the answer is no. Blunt pairwise comparisons between the once-daily and twice-daily nevirapine arms and the efavirenz arm did suggest significantly more dropouts with nevirapine (P=0.009 once-daily versus efavirenz and P= 0.039 twice-daily versus efavirenz). But after statisticians scrupulously adjusted those figures to account for multiple testing in four preplanned comparison arms, the only significant difference remained between efavirenz and nevirapine/efavirenz.
This is probably more than most readers want to know -- or need to know -- about side effect statistics. As so often proves true when statistical hair-splitting requires electron microscopy to discern differences, the proper question becomes not "Are these differences statistically significant?" but "Are these differences clinically meaningful?" Most seasoned HIV clinicians will be content to walk away from 2NN with the primary endpoint message -- equivalent failure rates in the three single-NNRTI arms and more failures with the more toxic double-NNRTI tactic. From there, picking between these two drugs for first-line regimens will not be a roll of the dice, but an already set appreciation of corollary risks and benefits.
Among nevirapine's possible benefits, earlier work suggested, may be its effect on lipids. 2NN confirmed findings of an Atlantic trial substudy that saw an apparently antiatherogenic profile with nevirapine but not with indinavir.5 Considering only people who stayed with their assigned 2NN regimen for 48 weeks and lumping the two nevirapine arms to compare them with the single efavirenz group, van Leth charted three significant benefits for nevirapine [abstract 752]:
Whereas the total-to-HDL-C ratio stayed fairly flat among people taking efavirenz (5.9 at baseline and 6.3 at week 48), it narrowed with nevirapine (5.3 at baseline and 4.5 at week 48).
Again, the savvy clinician will ask whether these statistically significant lipid differences translate into clinical benefit. The context is complicated -- a chronic disease that will require treatment for two, three, or more decades with drugs that will be as different 10 years from now as today's best antiretrovirals are from 1993's crop. Add to that the still-muzzy datastream from cohort studies metering antiretroviral fallout on cardiovascular quiddities.
For now, 2NN confirms what many prescribers have long sensed: In the peerless treatment-naive person with world-class liver function, no propensity to rash, zero cardiovascular risk factors, exemplary psyche scores, and a fondness for vivid dreams, nevirapine and efavirenz are both swell picks. If you see imperfect patients, more thought will be needed. On average, nevirapine looks more inclined to cause rashes and is more hepatotoxic. Careful liver function monitoring guidelines are on the books for people starting this drug.6 Efavirenz can play mean tricks on the central nervous system, and it's a bad choice for women who may become pregnant. And, oh yes, nevirapine costs less.
Researchers are still puzzling through the poor results of the atazanavir-versus-efavirenz trial presented last year.7 A 48-week intent-to-treat analysis of that multinational study counted only 32 percent of treatment-naive people randomized to atazanavir with a viral load under 50 copies/mL, and only 37 percent assigned to efavirenz reached that mark. The trial's official conclusion underlined atazanavir's "equivalence" with efavirenz. But at the Retrovirus meeting Martin Hirsch (Massachusetts General Hospital, Boston) [abstract 187] called that analysis "difficult to reconcile" with results of an atazanavir-versus-nelfinavir trial in treatment-naive people, which also showed 48-week virologic equivalence.8 Yet nelfinavir did not keep pace with efavirenz in a study Hirsch planned, ACTG 384,9 and no one would consider nelfinavir today's standard of comparison for the phase III trial of a new drug.
The 2NN results (preceding section) contradict the poor showing of efavirenz in the atazanavir-efavirenz trial and confirm earlier studies, like ACTG 384, attesting to this NNRTI's antiviral vigor. Even though 2NN, like the atazanavir-efavirenz study, counted consecutive viral loads above 50 copies/mL as a virologic failure, the 48-week intent-to-treat rate of sub-50-copy responses measured 70 percent with efavirenz (as well as with once-daily nevirapine). But atazanavir scored some virologic points at CROI in a long-term analysis of the atazanavir-nelfinavir trial.
After 48 weeks of random assignment to atazanavir or nelfinavir (plus d4T/3TC), study participants could continue one of the two atazanavir doses (400 or 600 mg once daily) or switch from nelfinavir to 400 mg of atazanavir [abstract 555]. Robert Murphy (Northwestern University, Chicago) reported that 63 people made the switch, while 283 continued atazanavir. At the switch point 70 percent in the nelfinavir-to-atazanavir arm had a viral load below 400 copies/mL, and 48 percent had fewer than 50 copies/mL. After 24 weeks on atazanavir those rates improved to 86 percent under 400 copies/mL and 59 percent under 50 copies/mL. No one in the switch arm stopped atazanavir because of treatment failure. At 108 weeks after randomization in the initial study, an intent-to-treat analysis counted 51 percent in the 600-mg atazanavir arm, 47 percent in the continuous 400-mg arm, and 49 percent in the nelfinavir-to-atazanavir arm with a viral load under 50 copies/mL.
People who continued atazanavir in the extension study had little change from initial baseline levels of total cholesterol, ominous low-density lipoprotein cholesterol (LDL-C), or triglycerides. HDL-C rose from 40 mg/dL at baseline to 46 mg/dL at week 48 of the first study and stayed there for another 24 weeks of atazanavir. People who traded nelfinavir for atazanavir enjoyed four significant lipid benefits 12 weeks after the switch:
For total cholesterol, LDL-C, and triglycerides, the significant improvements held through another 12 weeks. About 10 percent in each study arm had clinician-reported lipodystrophy, while 26 percent taking continuous 400- mg atazanavir, 44 percent taking continuous 600- mg atazanavir, and 13 percent switching from nelfinavir to atazanavir had grade 3 or 4 bilirubin elevations. Respective jaundice rates were 3 percent, 3 percent, and 6 percent.
As with atazanavir, U.S. Food and Drug Administration (FDA) analysts are even now threshing through trial data on another PI, GlaxoSmithKline's GW433908, the amprenavir prodrug (fosamprenavir) also called 908. CROI attendees scanned an ample load of those data in two slide talks, one involving a comparison with nelfinavir in treatment-naive people and one pitting ritonavir-boosted 908 against lopinavir in people with PI experience. The GlaxoSmithKline PI shares at least one trait with rival atazanavir: It can be taken once daily (but only when boosted by ritonavir). Unlike atazanavir, 908 outdistanced twice-daily nelfinavir in treatment-naive people. But it may lag lopinavir as a second- or third-line PI.
The naive trial randomized 251 people in a 2-to-1 ratio to 1,400 mg of 908 or 1,250 mg of nelfinavir twice daily [abstract 177]. Everyone also took abacavir and 3TC. Though most study participants lived in the U.S. (n=153), sizable proportions signed up in Panama (n=52), Puerto Rico (n=25), and South Africa (n=21). About one third were women. The median baseline viral load measured a little over 4.8 logs (about 63,000 copies/mL) in both treatment groups, and more than 40 percent in both groups had a viral load above 100,000 copies/mL. The median CD4 count stood at 214 cells/mm3 in the 908 arm and 212 cells/mm3 in the nelfinavir arm, and close to half had fewer than 200 cells/mm3.
After 48 weeks 66 percent taking 908 and 51 percent taking nelfinavir reached the primary endpoint, a viral load below 400 copies/mL in an analysis that defined failure as never going below 400 copies/mL, rebounding from below 400 copies/mL, or stopping treatment. Jeffrey Nadler (University of South Florida College of Medicine, Tampa) did not report whether this (or any) difference reached statistical significance. By the same type of analysis, 55 percent taking 908 and 41 percent taking nelfinavir had a 48-week viral load below 50 copies/mL. In the 908 arm nearly equal proportions starting treatment above or below the 100,000-copy mark reached a sub-50 viral load. Not so in the nelfinavir group, where 54 percent starting below 100,000 copies/mL ended up under 50 copies/mL compared with 24 percent starting above 100,000 copies/mL. The median CD4-cell gain measured 201 cells/mm3 with 908 and 216 cells/mm3 with nelfinavir.
Side effect rates proved similar in the two groups, except that 18 percent taking nelfinavir versus 5 percent taking 908 endured diarrhea (P=0.002). While 9 percent taking 908 had a hypersensitivity reaction or rash, 5 percent taking nelfinavir did, a nonsignificant difference. Mean fasting triglycerides and total-to-HDL-C ratio were below U.S. National Cholesterol Education Program cutoffs (200 mg/dL and 6.5) in both treatment groups.
The second 908 study randomized 320 people who had taken one or two PIs (a handful had taken more) to genotype-selected NRTIs (usually including tenofovir) plus 908/ritonavir (1,400/200 mg once daily or 700/100 mg twice daily) or standard-dose lopinavir [abstract 178]. No one could take a nonnucleoside. Median baseline viral loads stood over 4.1 logs (about 12,500 copies/mL), and the median starting CD4 count was 263 cells/mm3.
The study's primary endpoint is time-averaged change in viral load from baseline (AAUCMB) at 48 weeks. The trial lacks the power to show statistically significant between-group differences in rates of sub-50 or sub-400 viral loads. Edwin DeJesus (Infectious Disease Clinic, Altamonte Springs, Florida) reported 24-week findings, which did not include a reckoning of statistical significance for AAUCMB. At 24 weeks the numbers showed "noninferiority" of either 908 arm to lopinavir, although virologic trends all ran in lopinavir's favor (Table 3).
Some attendees wondered whether AAUCMB is the best endpoint for a study population like this. Such an area-under-the-curve calculation makes sense for study groups with highly drug-resistant virus and faint chance of reaching a viral load under 50 copies/mL, explained session cochair Julio Montaner (University of British Columbia). But because respectable proportions in this trial did notch sub-50 loads (Table 3), AAUCMB may not be the most telling yardstick. Another way to size up 908 in PI-experienced people would be to chart virologic responses against baseline resistance mutations and viral susceptibility. DeJesus said that analysis will come with the 48-week results. Fewer people taking once-daily 908 (19 percent) than twice-daily 908 (35 percent) or lopinavir (34 percent) had a grade 2 to 4 side effect.
Whether this amprenavir prodrug can find a following as an unboosted or a boosted first-line agent, or a boosted rescue PI, cannot be surmised from findings so far. What can be said is that boosted 908 given to treatment-naive people spawned no protease mutations, while unboosted 908 did [abstract 598]. This analysis, presented by GlaxoSmithkline's Sarah Macmanus, involved people with consecutive viral loads above 1,000 copies/mL in two 908 studies -- the nelfinavir comparison outlined by Jeffrey Nadler (above) and a contest between once-daily 908/ritonavir (1,400/200 mg) and twice-daily nelfinavir, both with abacavir and 3TC. Comparing baseline genotypes with on-therapy genotypes during viral rebounds, Macmanus consistently mapped more protease and nucleoside mutations in Nadler's unboosted 908 group:
The GlaxoSmithKline team suggested the results support using boosted 908 as "first-line [therapy] or early in the treatment continuum."
The much-heralded PI tipranavir had a head start over atazanavir and 908 in the race to regulatory sanction. But it became the laggard hare in this marathon, only recently reaching the phase III leg because of difficulties defining the best dose and delays due to a shift in developers. A phase II trial detailed by Joseph Gathe (Therapeutic Concepts, Houston) confirmed tipranavir's activity against virus laden with some infamous protease mutations [abstracts 179 and 596].
Because the study aimed to pick the phase III dose of tipranavir in a cohort with treatment experience, the primary endpoints were viral load drop at two weeks (before other drugs in the regimen could be changed) and side effects at four weeks. Study entry thresholds included treatment with all three antiretroviral classes and one or more mutations at protease sites 30, 46, 48, 50, 82, 84, and 90. Researchers randomized 216 people to one of three twice-daily tipranavir/ritonavir doses: 500/100 mg, 500/200 mg, or 750/200 mg. The median starting CD4 count stood at 153 cell/mm3 and the median viral load at 4.53 logs (about 33,900 copies/mL).
Whereas half the people assigned to 500/100 mg failed to hit a target trough concentration above 20 µM, more than three quarters in the other two arms did. By a last-observation-carried-forward analysis, week-two viral load drops measured 0.87 log with 500/100 mg, 0.97 log with 500/200 mg, and 1.18 logs with 750/200 mg; differences between arms lacked statistical significance. Boehringer Ingelheim, tipranavir's developer, chose the 500/200-mg dose for phase III studies because it yielded better troughs than 500/100 mg, with less interpatient variability than 750/200 mg. Fewer people taking 500/200 mg than 750/200 mg had to suspend treatment because of side effects. Gathe reported that 12.5 percent taking 500/200 mg and 16.9 percent taking 750/200 mg had one or more severe side effects. Overall, about 15 percent suffered grade 2 diarrhea and 12 percent vomiting.
Genotypic analysis focused on four changes that Boehringer Ingelheim calls universal PI-associated mutations or UPAMs -- L33I/V/F, V82A/F/L/T, I84V, and L90M. Virus with one or two UPAMs at baseline hardly tarnished susceptibility to tipranavir but proved resistant to other PIs. Two UPAMs, for example, yielded a median 1.3-fold change in tipranavir IC50 (50 percent inhibitory concentration) relative to wild-type virus, compared with a 9.6-fold change for saquinavir, 13.1-fold for amprenavir, 21.1-fold for indinavir, 32.8-fold for nelfinavir, 79.6-fold for lopinavir, and 97.9-fold for ritonavir. Only people with three UPAMs, and as many as 16 to 20 mutations in all, had more than a 2-fold change in susceptibility to tipranavir.
A 2-fold change in IC50 may mark the cutoff for virologic response to tipranavir, the researchers propose. Across the three study arms, the median viral load fell over 1 log in people with a 1- to 2-fold change in viral susceptibility to tipranavir. But responses nose-dived to a median 0.2 log or less when virus had a 2- to 4-fold change in susceptibility. Yet even in this PI-experienced population (93 percent had more than 10 protease mutations at baseline), 69 percent began the study with less than a 2-fold change in IC50 to tipranavir.
Although the once-a-day nucleoside emtricitabine (FTC) did not earn a slide slot at the 10th CROI, four posters sated the inquisitive. Interest in this long-studied drug leapt a quantum or two when Gilead bought its developer, Triangle. This union raised the winsome possibility of a once-daily double nuke in a single pill -- FTC and tenofovir -- heady competition for twice-daily Combivir (ZDV/3TC).
The longest FTC study presented, by Triangle's Charles Wakeford, involved 215 people originally randomized to take FTC (200 mg once daily) in a 48-week comparison with continued 3TC [abstract 550]. All 215 had a viral load under 400 copies/mL at week 48 and agreed to continue FTC in an extension study, along with 74 other people. A Kaplan-Meier estimate of time to virologic failure (consecutive viral loads above 400 copies/mL) stood at 10 percent 250 days after people started FTC. At 1,450 days that estimate had inched to 11 percent. The Kaplan-Meier probably of tolerability failure (permanent discontinuation because of death or toxicity) measured 13 percent after four years of FTC.
In a placebo-controlled comparison of FTC with d4T (plus didanosine [ddI] and efavirenz) involving 571 treatment-naive people, FTC outperformed d4T by every measure [abstract 606]. Resistance patterns in people who suffered virologic failure also painted an interesting picture. After 48 weeks of treatment, Pedro Cahn (Fundacion Huesped, Buenos Aires) reported, FTC won the primary endpoint battle -- proportion with a viral load under 50 copies/mL -- and several other contests (Table 4).
Among people who had a virologic failure, 71.4 percent taking FTC and 97.1 percent taking d4T had at least one resistance mutation (P=0.019). In this virologic failure group, 42.9 percent taking FTC and none taking d4T had the M184V mutation (P<0.001), 7.1 percent taking FTC and 20.0 percent taking d4T had a thymidine analog mutation (at positions 41, 67, 70, 210, 215, or 219) (not significant), none taking FTC and 11.4 percent taking d4T had the ddI-associated L74V mutation (not significant), 71.4 percent taking FTC and 88.6 percent taking d4T had an NNRTI mutation (not significant), and 28.6 percent taking FTC versus 2.9 percent taking d4T had wild-type virus (P=0.019).
An open-label study randomized people with a viral load below 400 copies/mL while taking one or two PIs to continue the PIs or switch to a once-daily medley of FTC, ddI, and efavirenz [abstract 551]. Defining virologic failure as consecutive rebounds above 400 copies/mL, Jean-Michel Molina (Saint-Louis Hospital, Paris) found equivalent 48-week failure rates in the two groups by a missing-data-equal failure analysis: 22 of 177 (12 percent) for continued PI therapy and 18 of 178 (10 percent) for the once-daily FTC regimen. In an on-treatment analysis 95 percent taking once-daily therapy and 87 percent continuing PIs had a 48-week viral load below 50 copies/mL (P=0.01). Kaplan-Meier estimates of serious side effects or treatment discontinuations disclosed no significant differences between groups.
Clinicians from Panama, South Africa, Mexico, and the United States studied FTC in 51 treatment-naive children and 31 experienced children, evenly divided between girls and boys and 72 percent of them black [abstract 872]. Ages ranged from three months to 17 years, and all took FTC at a dose of 6 mg/kg once daily to a maximum of 200 mg daily (the adult dose). Treatment-experienced children had viral loads below 400 copies/mL while taking a regimen including 3TC, which they replaced with FTC. Clinicians could also switch other drugs in the regimen.
In a week-24 noncompleter-equals-failure analysis, 63 percent in the naive group and 71 percent in the experienced group had a viral load under 50 copies/mL. Respective proportions below 400 copies/mL were 92 percent and 84 percent. Median follow-up extended to 31 weeks in the naive group and 48 weeks in the experienced group. Xavier Sáez-Llorens (Panama City) and colleagues attributed five severe toxic episodes to study drugs, including pancreatitis, vomiting, leukopenia, anemia, and pleural effusion. One treatment-naive child (2 percent) and four treatment-experienced children (13 percent) had a grade 3 or 4 lab abnormality. The researchers determined that the 6 mg/kg FTC dose yielded exposure similar to 200 mg once daily in adults.
Atazanavir, GW433908, and tipranavir weren't the only new antiretrovirals -- or even the only new protease inhibitors -- to get PowerPoint treatment at the 10th CROI. Table 5 outlines 10 others, including two more PIs, a trio of nonnukes, a bevy of entry inhibitors, and a budding inhibitor. A few claims made for these hopefuls seem especially intriguing -- or at least worthy of note:
Do people with HIV infection really need all these new antiretrovirals? That is probably not the apposite question, since only a handful of these candidates will survive the treacherous gauntlet that ends in the regulator's den. Some fast math shows that 18 percent of the antiretroviral philters in focus at the 1993 Retro gathering found spots on the pharmacy shelf, and one of those (delavirdine) pretty much stayed there. If a similar proportion of the 2003 meeting candidates wins a license, clinicians would have eight more antiretrovirals to add to today's 16.
With luck the next batch of anti-HIV meds will not merely inflate the current total, but will include agents of unique value. And -- if the genie grants a second wish and a third -- these drugs will be easy to take and will not require a home equity loan to pay for a year's dosing. Looking at the list of 45 debutantes on parade at the 10th CROI, and imagining the best of Panglossian worlds, one could envision:
The amazing thing is that one can propose that list with a straight face.
Even as the promise of new, stronger, simpler, kinder antiretrovirals tantalizes, evidence suggests that the drugs at hand have flattened the epidemic -- at least in Western Europe, North America, Brazil, and Australia. But whether lean hints of replication at sub-50 loads portend a load of trouble, or mere spillage, remains controversial.
The strongest sign that today's antiretrovirals can make HIV infection a chronic disease came from a EuroSIDA analysis of 3,793 people in the pre-HAART epoch (1994-1995), 3,425 in the early HAART era (1996-1997), and 2,585 in the late HAART age (1998-2002) [abstract 180]. The early HAART-propelled drop in new AIDS diagnoses and deaths has not bottomed out, Amanda Mocroft reported.
This incessant decrescendo reflects overall trends in CD4 counts of this cohort, which is now about three quarters male. Whereas people with 50 or fewer cells/mm3 made up about 30 percent of the cohort in late 1994 and early 1995, they now account for fewer than 5 percent. The proportion with 51 to 200 cells/mm3 ebbed from about 30 percent in 1994-1995 to about 15 percent today. Meanwhile, the group with more than 200 cells/mm3 grew from 40 percent before HAART to over 80 percent today.
The risk of AIDS or death -- adjusted for age, prior HAART, AIDS status, and CD4 count at cohort entry -- fell from near 10 percent in 1994 and 1995, to 1 percent in late 1998 and early 1999, to less than 1 percent after September 2001. After September 1998, Mocroft figured the risk of AIDS or death slipped by 10 percent per six-month period (P<0.0001). The incidence of death alone fell significantly during that period only among people whose latest CD4 count languished below 50 cells/mm3. But the incidence of AIDS proved 50 percent lower in the late HAART era than in pre-HAART days regardless of CD4 count (P<0.0001). In a multivariate model considering baseline differences, the risk of AIDS (P=0.0004) or death (P=0.0013) dropped by about one third from the early HAART era to the late HAART era.
Bernard Hirschel and Swiss HIV Cohort Study colleagues took a different tack in tracking the ongoing effect of HAART [abstract 917a]. They charted the "excess death rate" among HIV-infected people versus the population at large. Swiss HIV Cohort members still die at a faster rate than the whole population, but not faster than successfully treated cancer patients, "a group who is able to obtain life insurance."
The study involved more than 2,000 HIV-infected people (29 percent of them women) treated after January 1, 1997 and followed through the last day of 2001. People with HIV but without HCV had a substantially lower excess death rate than did those coinfected with HCV. Among people who reached a CD4 count above 250 cells/mm3 with treatment, nadir CD4 count seemed not to affect mortality. Excess death rates range from 5 to 20 per 1,000 people per year among successfully treated cancer patients. Hirschel reported these excess death rates per 1,000 per year for HIV-infected people with or without HCV:
The Swiss team noted that keeping the viral load below 400 copies/mL appears to confer an extra survival benefit among people whose CD4 tally climbs above 250 cells/mm3 with HAART.
Yet these macroviews of mortality may not reflect the microreality of individual clinics, reported W. Christopher Mathews from the Owen Clinic at the University of California, San Diego [abstract 911]. As elsewhere in Western Europe and North America, death rates began to plunge at that clinic in 1995, when better therapies for HIV and opportunists arrived. That trend continued downward until 1998 -- then turned back up. Analyzing data from nearly 5,000 people who entered the clinic from 1991 on, Mathews traced the post-1998 upswing to two groups -- those who first sought care with a CD4 count below 200 cells/mm3, and those with no CD4 data on file within 90 days of their first visit.
Looking more closely at 2,278 people who came to the clinic after 1997, Mathews noted nine factors that changed the risk of death stratified by year of entry:
Gender did not affect the risk of death in the post-1997 cohort. An entry CD4 count between 200 and 349 (versus more than 350) cells/mm3 raised the risk 1.7 times, but that difference fell shy of statistical significance (P=0.089).
Despite bringing the Western epidemic to a virtual standstill, today's antiretrovirals still suffer the steely-eyed scrutiny of scientists who can show that these drugs do not stamp out every ember of replication. So the question becomes whether those smolderings yield only smoke, or also fire.
Jan van Lunzen (University Hospital Eppendorf, Hamburg) saw some ominous flickerings in lymphoid tissue of people taking apparently suppressive HAART [abstract 187]. All 32 study participants had viral loads below 25 copies/mL for more than nine months and had taken the same regimen for more than 12 months. Fourteen used a PI combination (only one with ritonavir boosting), 11 used a nonnucleoside, and seven used two or three nucleosides (including one taking two nucleosides and hard-gel saquinavir).
Axillary lymph node biopsies turned up productively infected cells in everyone, but the people taking only nucleosides had a much higher rate of viral trapping on follicular dendritic cells. Three of six in the nucleoside group, including one taking Combivir/abacavir and one taking two nukes plus hard-formula saquinavir, had HIV ensnared in dendritic tendrils. Only one of 14 taking a PI and none of eight taking an NNRTI had similar evidence. The M184V mutation arose during viremic blips in lymph nodes of the person taking Combivir/abacavir, but that mutation did not appear in peripheral blood mononuclear cells. Yet the nucleoside group didn't do much worse than the other groups when rated for HIV-infected germinal centers (16 of 26 samples overall) or HIV in extrafollicular tissue (21 of 26 overall).
Using a high-octane viral load assay with a dynamic range of 1 to 1,000,000 RNA copies/mL, Frank Maldarelli (National Cancer Institute [NCI], Frederick, Maryland) turned up viral traces in 14 of 15 people in whom a standard assay clocked viral loads below 50 copies/mL for at least 131 days [abstract 466]. In the 14 with these midgey loads, RNA numbers ranged from 1 to 40 copies/mL and averaged 9.3 copies/mL. Five people with serial measures had stubby but utterly stable loads for seven to 12 months. While higher pretreatment viral load correlated with higher low-level viremia, baseline CD4 count, treatment-induced CD4 change, and treatment duration did not.
Maldarelli also found that people taking four or more antiretrovirals had lower levels of Lilliputian viremia than did people taking only three drugs. "The apparent relation between regimen potency and level of viremia," the NCI team proposed, suggests sub-50-copy viremia "may be sustained by ongoing HIV-1 replication."
Does this runty replication portend full-fledged rebounds? Not according to results of a study of 12 children with viral loads below 50 copies/mL for one to six years [abstract 619]. Using an assay that spots 5 RNA copies/mL, Deborah Persaud and colleagues in Robert Siliciano's Johns Hopkins laboratory logged viral loads every three months in children who had kept an undetectable load since starting therapy or since switching to a PI regimen after taking suboptimal combinations. Persaud also managed to amplify and genotype viral samples from these children.
In 18 of 21 samples analyzed, the viral load always remained under 50 copies/mL. In three samples from three children, Persaud recorded blips to 69, 124, and 140 copies/mL. Sequencing 199 viral clones, she found that HIV spilling into plasma during suppressive HAART almost always proved wild-type or housed mutants that arose during earlier suboptimal therapy and hung on in the absence of drug pressure. There were two exceptions: The V82I protease mutation sullied two of six clones derived during a blip in a child taking nelfinavir. Another protease change, N88S, popped up in one of six clones from a nelfinavir-treated child with fewer than 20 copies/mL of circulating virus. Both children maintained viral suppression.
The Hopkins team concluded that "ongoing [low-level] viremia ... during effective HAART in children represents release of largely archival drug-sensitive or drug-resistant HIV-1 rather than recently generated, drug-resistant mutations." Viremia below the 50-copy mark or blips below 200 copies/mL, they added, "does not necessarily represent impending therapeutic failure or the evolution of resistance and it therefore may not require a change in therapy." Earlier, Siliciano and colleagues failed to detect genotypic evidence of viral evolution in 11 adults with sub-50 viremia.10
If compelled to posit a single conclusion from the last six studies -- the cohort analyses by Mocroft, Hirschel, and Mathews, and the sub-50 fathomings of van Lunzen, Maldarelli, and Persaud -- one might say this:
In people without a dire treatment history or untreated but dangerously advanced disease, improved regimens of recent years do a credible job of crippling HIV.
So why, as the 10th CROI amply demonstrated, do drugmeisters maintain their frenzied pace of discovery and development? This is an easy question to answer:
Those are the topics for Part 2 of this review, due next month.
Mark Mascolini writes about HIV infection (firstname.lastname@example.org).
This article was provided by International Association of Physicians in AIDS Care. It is a part of the publication IAPAC Monthly.