TAGline/Volume 4 Issue 1
Four in Favor, Four Against; Long-Awaited Non-Nuke Receives Mixed Reviews at Agency Hearing
Clinical benefit elusive
In a November 22nd FDA hearing that harkened back to the bad old days of ddC, the Food and Drug Administration's Antiviral Drugs Advisory Committee considered an application for approval of delavirdine (trade name Rescriptor). Delavirdine is a non-nucleoside reverse transcriptase inhibitor (NNRTI), similar to Boehringer-Ingelheim's nevirapine, which was approved by the FDA in June, 1996.
Three randomized studies were presented to support the approval of delavirdine. Study 0021 compared AZT alone to AZT in combination with 200, 300, and 400 mg of delavirdine three times daily. Participants had 200-500 CD4+ cells (mean 325-340). Mean baseline HIV RNA was 5.2 log copies/ml, and about 60% had previously been treated with AZT. Participants taking AZT in combination with the higher doses of delavirdine had significant improvements in T-cell counts (average increase was about 40 cells), and HIV RNA levels (average decline was about 1 log).
In study 0017, 780 patients were treated with ddI or with ddI+delavirdine. These were more advanced patients, with < 300 CD4+ cells (average baseline count was about 140 cells), mean baseline HIV RNA of 5.8 log copies/ml, and average prior ddI experience of 4 months. Mean change in CD4+ count in patients taking ddI+delavirdine was significantly better than ddI monotherapy at weeks 4 and 8. Mean change in viral load was significantly better after week 4 (.9 log decrease). Sixty-one deaths occurred in the ddI group, and 66 deaths occurred in the ddI+delavirdine group: this was not statistically significant. According to Pharmacia & Upjohn (makers of delavirdine), the Data Safety and Monitoring Board (DSMB) recommended terminating the study because the protocol goals of showing a clinical outcome difference between the two regimens was unlikely to be achieved.
A third study compared four groups of patients, who were treated with AZT+delavirdine, ddI+delavirdine, AZT+ddI, or all three drugs together. Mean baseline CD4+ cell count was 285-303. HIV RNA data were not available. 63% of patients were treatment naive, 36% of patients had been treated with < 6 months of AZT monotherapy, and 1% had been treated with 6 months of ddI monotherapy. In an intent-to-treat analysis, patients treated with AZT+ddI+delavirdine and AZT+ddI had a significantly better CD4+ response than patients treated with AZT+delavirdine or with ddI+delavirdine. In an on-treatment analysis, patients treated with the triple therapy had a significantly better CD4 response than patients treated with AZT+ddI.
The major side effect seen in patients taking delavirdine is rash. This rash appears between two and four weeks after beginning treatment and is usually not serious enough to require patients to stop treatment. People taking Bactrim for PCP prophylaxis had an increased risk of a serious rash. Warning signs of a serious rash include high fever and rash on mucous membranes. For most people, treatment with antihistamines was sufficient to allow continuation of treatment through the rash.
Like other NNRTIs, delavirdine monotherapy rapidly leads to the emergence of delavirdine resistant HIV. When used in combination with other drugs, development of resistance may be delayed. The primary amino acid substitutions in the HIV reverse transcriptase gene following long-term in vitro delavirdine exposure appear at codons 103, 181, and 236. Treatment with delavirdine monotherapy or delavirdine+ddI produces mutations at codon 103 or a combination of mutations at codons 103 and 181. Treatment with AZT+delavirdine most frequently leads to mutations at codons 103 and 236, with the 181 mutation occurring in fewer than 2% of samples. Pharmacia & Upjohn asserts that delavirdine can re-sensitize nevirapine-resistant virus to nevirapine, however Boehringer-Ingelheim (nevirapine's manufacturer) believes that the nevirapine-resensitizing mutation in question rarely occurs in vivo, while more common mutations confer complete loss of sensitivity to both nevirapine and delavirdine.
Delavirdine is processed through the CYP3A pathway. Most protease inhibitors are also processed through CYP3A. Consequently, delavirdine probably interacts with most protease inhibitors. Unfortunately, Pharmacia & Upjohn only conducted brief pharmacokinetic and acute safety studies of delavirdine in combination with approved protease inhibitors.
In these studies, co-administration of delavirdine and saquinavir raised saquinavir levels by five times, producing blood levels of saquinavir that were comparable to those achieved in the Stanford high-dose saquinavir study. Co-administration of delavirdine and indinavir raised indinavir levels two-fold. Based on this observation, Pharmacia & Upjohn recommend that indinavir be dosed at 400-600 mg every eight hours, from the current recommended dose of 800 mg every eight hours. In addition, co-administration of ddI and delavirdine lowered ddI levels by 20%, and increased clarithromycin levels by two-fold. The following drugs are contraindicated in patients using delavirdine: alprazolam, astemizole, barbamazepine, cisapride, phenobarbitol, phenytoin, rifabutin, rifampin, terfenadine and triazolam.
In a consensus statement regarding approval, a group of five AIDS organizations, including the AIDS Action Council, the AIDS Treatment Data Network, the American Foundation for AIDS Research, Gay Men's Health Crisis and TAG, identified four key problems with the application:
- Surrogate Marker Claim. Based on a meta-analysis of studies 0017 and 0021, Pharmacia & Upjohn asserted that a half-log reduction in viral RNA sustained for at least eight weeks would result in clinical benefit for at least a year. However, in study 0017, patients treated with delavirdine+ddI experienced a half-log drop in RNA (compared to patients treated with ddI monotherapy), and yet experienced no clinical benefit. Therefore, its clinical trial undermined the strength of this surrogate claim.
- Indication. The proposed indication read "for use in combination with other antiretroviral agents in the treatment of HIV infection whenever antiretroviral therapy is warranted." The only data available about delavirdine in combination with protease inhibitors, however, was pharmacokinetic and acute safety data in HIV-negative patients. The data were sufficient to identify major interactions with saquinavir and indinavir, but not enough to allow for an assessment of the safety and efficacy of the combinations.
- Confirmatory Trial. Pharmacia & Upjohn identified its on-going study of delavirdine+ AZT+ either ddI, ddC or 3TC compared to AZT+ either ddI, ddC, or 3TC as its pivotal trial intended to provide evidence of clinical benefit. This study is being conducted in Europe, Australia and South Africa. These countries, however, are just now beginning wide-spread use of protease inhibitors, which are prohibited in the study. The consensus group was concerned about the viability of this trial. They recommended that the company make every attempt to complete their planned clinical endpoint study, and that the company should undertake a clinical endpoint study using delavirdine and a protease inhibitor.
- Resistance and Cross-Resistance. The consensus group was greatly concerned that Pharmacia & Upjohn had not adequately characterized the effects of resistance to delavirdine on sensitivity to other non-nucleoside analogue reverse transcriptase inhibitors (NNRTIs), such as nevirapine and DuPont-Merck's up-coming DMP-266. The group recommended that Pharmacia & Upjohn work rapidly to develop further data on cross-resistance with NNRTIs, including bi-directional in vivo studies.
Based on the presentation by Pharmacia & Upjohn, four members of the committee voted to recommend approval of delavirdine, and four voted against. The FDA will have to review the application in the coming weeks, and make a decision.
At NIH-Sponsored 'Principles' Panel, A First Peek at Second Phase Decay Data -- and at Lymphoid Architecture
FDCs as storage depots
In mid-November, an NIH panel convened under the auspices of the Office of AIDS Research (OAR) to discuss scientific and clinical data which might help produce new state-of-the-art guidelines for the treatment of HIV infection. Mark Harrington represented TAG at the meeting. Other AIDS research luminaries who shared new data ran the gamut from Ashley Haase (University of Minnesota) and David Ho (Aaron Diamond) to David Chernoff (Chiron Corp.) and Steven Herman (Roche Molecular Systems).
Dr. Ho reviewed the current state of knowledge regarding the kinetics of HIV replication: the viral life-cycle lasts about 2.6; the half-life of a productively infected CD4+ T cell is about 1.6 days; the half-life of virus production is approximately 5.7 hours. As has been shown many times over the year, on potent treatment viral load falls some 2 logs (99%) within 3 weeks (first-phase viral decay) of initiation of treatment and then falls more slowly as other infected cellular components (macrophages and latently-infected T cells) become activated, produce virus and are killed or die of apoptosis (cell suicide). The second-phase decay half-life is believed to be between 8-28 days. Using these numbers as guides, Ho estimates that 93-99% of virus is produced by actively-infected CD4+ T lymphocytes with a half-life of one day, 1-7% is produced by actively-infected macrophages with a half-life of 14.4 days and < 1% is produced by latently-infected T cells with a half-life of 8.5 days. Much of the second-phase decay, then, reflects loss of infected macrophages.
Proviral DNA (that is, reverse transcribed and integrated viral RNA which has incorporated itself into the nucleus of an infected cell), however, is unlikely to disappear as quickly. In fact, at 120 days 1,000 provirus copies could be detected per 1 million cells. The possible existence of unforeseen slower third-phase decay or viral persistence in sanctuary sites inaccessible to treatments cannot be ruled out. Many fixed tissue macrophages (e.g., in the spleen) which are HIV DNA+ are HIV RNA-negative. And while many of these latently infected cells probably contain defective proviral DNA, even defective provirus could still result in cell death if cells bearing them were detected and killed by CTLs (cytotoxic T lymphocytes).
Follicular dendritic cells (FDCs) in lymphoid tissue act as a storage depot for infectious HIV in the lymph nodes, where viral particles are bound to FDCs by antibodies, but may remain infectious. Lymphoid tissue amounts to about 1% of the total body weight -- 700 grams in a 70 kilogram person. University of Minnesota's Dr. Ashley Haase, using new imaging techniques, showed that the FDC viral pool is much larger than the pool of newly-produced RNA inside of infected cells. Productively infected cells amount to only a fraction of total lymphocyte turnover (just 200 million per day), implying that other, indirect mechanisms of lymphocyte destruction must be at work -- but what? Haase explains, "We're looking at proliferation and apoptosis in HIV-positive and negative persons," he answers. "We're seeing a lot of additional proliferation and apoptosis in CD4s from infected persons, which drop in accordance with treatment. Replenishment of the CD4 reservoir is complicated; it includes cell expansion, retrafficking and other factors."
Dr. Haase showed that after six months of potent triple-drug therapy, the CD4 cell count begins to rise in lymphoid tissue, but what's going on with the lymphoid architecture? In Haase's experiments, not only are the CD4s returning in lymphoid tissue (and FDC-trapped antigen levels falling) but the lymphoid architecture "appears to be returning to a healthier, more functional state," he explains.
In contrast to disappointing antiviral effects in lymphoid tissue seen with AZT and ddI (Cohen et al.), researchers are seeing encouraging reductions in viral RNA in all compartments of lymphoid tissue. At 24 weeks they observed a two-fold decrease in the magnitude of the FDC-associated HIV pool. David Ho commented that this suggests that the half-life of FDC-trapped HIV is about three weeks. Four (unpublished) studies at Johns Hopkins, Boston City Hospital, Chiron Corp. and New Jersey Medical Center, have shown that, on potent therapy, viral load is dropping precipitously in lymphoid tissues. While potent combination antiretroviral therapy clearly affects these reservoirs, viral residues remain. Moreover, virus could emerge from immunologically privileged sites such as the central nervous system to reinfect peripheral cells.
In a separate presentation, Chiron's David Chernoff discussed improvements in the sensitivity of the bDNA (Quantiplex) assay. The newer-generation (version 3.0) bDNA assay has greater photoluminescence, providing greater sensitivity, down to about 50 copies/ml of plasma.
To resolve (only recently appreciated) uncertainties about the Mellors numbers (published in mid-1996 to great fanfare) and to help laboratories prepare specimens properly, Chiron tested a known amount of virus in three different standard tubes: EDTA, ACD, and heparin. In EDTA tubes the average RNA value was 171,000 copies/ml; in ACD tubes it was 150,000 copies/ml; and in heparin, 106,000. EDTA tubes are the most sensitive, and so the highest viral copy numbers are are recorded when blood is stored in them. Heparinized tubes are always the least sensitive. ACD tubes are somewhere in between. Using heparin as the anticoagulant (as was done in John Mellors' acclaimed study) reduces the RNA levels to 65% of EDTA levels after 2 hours-and to 45% after 30 hours. (If Mellors' plasma samples had been stored in EDTA or ACD tubes instead of heparin, his cut-off plasma RNA values would have been significantly higher.) The relationship between the Mellors numbers and PCR and NASBA remains unknown.
Chernoff showed that samples can vary by as much as two- to three-fold due to operator and assay differences, as well as intrapatient biological variability, although there does not appear to be a diurnal periodicity to HIV plasma viral load as there is with CD4 counts.
When Roche's Steven Herman looked at HIV RNA plasma viral load variability using the Roche (Amplicor) assay in 20 asymptomatic antiretroviral naive patients with CD4 > 400, he found the maximum variation to be three-fold or less in 15 patients, three- to four-fold in 18 patients and four- to six-fold in one. David Chernoff noted that many people become concerned if their viral load increases from, for example, 5,000 to 10,000. "But that's well within the normal variation of the assay," he cautions. Roche's Herman concurred, "A two-fold change in HIV RNA is not significant. A significant change is larger than a three-to-five-fold change in RNA, e.g., a rise from 10,000 to over 50,000 copies." And this sort of variability is especially high in patients with very low numbers.
Recent Research in Brief
At NIH-Sponsored 'Principles' Panel, A First Peek at Second Phase Decay Data -- and at Lymphoid Architecture
FDCs as storage depots
"It's the Faucet, Stupid." Comparing something called telomere lengths of CD4+ and CD8+ T lymphocytes in HIV-infected and uninfected persons, Amsterdam's Frank Miedema et al. claim "there is no evidence for increased turnover of CD4+ T cells in HIV infection." Miedema's provocative paper (Science 274,1543) concludes that, rather than destroying the body's CD4+ T cells (as Ho would have it), HIV instead prevents the production of T cells -- the "tap," not the "drain." The Aaron Diamond team insists that the preponderance of evidence favors the "drain" model.
"Genomic Resistance Doesn't Matter." Contrary to the conventional wisdom, R. Tedder, member of the Delta virology team, reports that resistance to AZT actually developed more quickly on the AZT+ddI and AZT+ddC combination arms than it did in the AZT monotherapy arm -- and that the development of drug resistant mutants didn't affect clinical outcome. "Genomic resistance doesn't matter," he asserted, "instead, it is the amount of resistant virus that matters." "Let's use resistance information in the most general terms," he challenged his U.K. audience, "that is, as it is reflected in rises in plasma RNA."
T-cell Tropic (SI) Virus Independently Predicts Progression to AIDS or Death. Meanwhile, the ACTG 175 virology team, led by Stanford's David Katzenstein, reports that the HIV viral phenotype (T-cell tropic ("SI") vs. macrophage tropic ("NSI")) at baseline was independently predictive of clinical outcome in the ACTG's AZT/ddI/ddC mega-trial. With a relative risk ratio (SI present at study entry/SI absent at study entry) of .40, the risk of progressing to new disease or death is reduced 60% for persons with an NSI viral phenotype at baseline versus an SI.
"Undetectable" on Delta. Also from the Delta virology team, investigators report that in study volunteers with no previous antiretroviral treatment ("naive"), the drug regimens AZT, AZT+ddI and AZT+ddC reduced HIV RNA plasma viral load to below the lower limit of detection of the assay (quaintly referred to as "undetectable") in 10%, 40% and 20%, respectively, of study participants. Maybe "undetectable" is not such a big deal after all. (Seems it's all about durability...)
Hit it Early, Hit it -- with nevirapine? Dr. Brian Conway, principal investigator of the INCAS study, upon presenting his AZT+ddI+nevirapine data which reduced plasma viral load to < 500 copies in 60% of patients at 6 months (AZT+ddI did the same thing in 30% of patients), asks, "Do we really want to hit early and hard -- or just early with a regimen that will get the job done?" Oddly, after presenting follow-up data showing that, even in compliant patients, resistance to nevirapine and AZT begins to emerge at 12 months, he seems to have answered his own question.