The U.S. FDA granted accelerated approval to amprenavir (Agenerase) in April 1999. In the intervening four years, clinical management questions have not been studied by the sponsors, and the drug has languished at around the 5% mark of the total protease inhibitor market share. This may have to do with unresolved uncertainties about how best to integrate amprenavir into clinical practice—what with its menacing pill burden (16 pills per day), less than ideal side effects (primarily GI and rash), unelucidated interactions with other commonly used agents and its less than overwhelming efficacy profile.

The only two important interaction studies completed since amprenavir's 1999 licensure (both of which took more than two years to complete) are for methadone and oral contraceptives.

Now a reformulated, "fos" amprenavir, arrives on the scene. Known among wonkish types by bits of its Vertex code name, "908," the less messy PI gives GlaxoSmithKline what it sees as a fighting chance to whittle away some of the protease market share from the likes of Abbott, Agouron and Merck. Not unlike the hasty development and approval of its predecessor, the clinical research on fos-amprenavir (Lexiva, as it will be marketed) is similarly lacking—at least according to the authors of TAG's official tract on this the 19th antiretroviral, expected to be waved in to home by the time we hit the newsstand. For Rob Camp and Heidi Nass, the new drug begs the usual questions.

How should fos-amprenavir be used?
In the July 2003 Public Health Service Guidelines for the Use of Antiretroviral Agents in HIV-1 Infected Adults and Adolescents, amprenavir is recommended (and only when boosted with ritonavir) as only an alternative protease inhibitor-based regimen.

Two of the three open label studies done—Solo and Context—have clearly demonstrated that fos-amprenavir, like amprenavir itself, is more potent when boosted with ritonavir. We see no reason to approve another mediocre protease inhibitor without the requirement for co-administered ritonavir.

Is fos-amprenavir effective?
In all studies to date, this drug has been seen as a mediocre protease inhibitor. In the language of the trial design, it has been seen as non-inferior to (as potent as?) nelfinavir in the Neat study, although not non-inferior (inferior?) to lopinavir/ritonavir in the Context study. The sponsor's development strategy utilized non-inferiority designs, with an AAUCMB statistical analysis [biostat-ese for "Average Area Under the Curve Minus Baseline": see your bio-stat wonk manual, v 6.0].

Does the AAUCMB analysis "forgive" a large dropout rate?
All studies had significant rates of drop-outs (approximately 30% in the better arms).

Fos-amprenavir has been analyzed in an advanced yet naïve or less advanced but experienced population, but not in a non-advanced population, nor has there been head-to-head comparison with amprenavir or an expanded access program of any type.

Co-administered with ritonavir, fos-amprenavir is more potent both clinically and virologically. (In Context, of the few who failed virologically in the fos-amprenavir/r arms, it was due to NRTI resistance—and not PI resistance.)

What are the benefits of fos-amprenavir?
Unlike amprenavir, fos-amprenavir does not have major GI tract side effects, reducing from approximately 70% to 5-9% (grades 2-4). Fos-amprenavir also gets better marks than its predecessor regarding rash, with incidence at 2-7% as compared to approximately 27% with amprenavir. Lab abnormalities do not reach 2% at 24 weeks.

The pill burden for fos-amprenavir (one pill BID) is greatly reduced from amprenavir (eight pills BID), because it is more water-soluble and is now in line with the pill count of the majority of protease inhibitors. (One mustn't forget the one extra pill—ritonavir—with each administration!)

What are the risks of fos-amprenavir?

• The absence of an expanded access program for fos-amprenavir may leave the community at a disadvantage in assessing the product in real life. The lack of data comparing fos-amprenavir to amprenavir raises many questions regarding bio-equivalence, not least in the panoply of drug-drug interactions.
• The community is greatly concerned that as a sulfa drug, the use of fos-amprenavir needs to be monitored when administered with other sulfa drugs and is contraindicated in people with sulfa allergies.
• Depression and "mood problems" hover around 5% and are of particular concern. Headache and sleep disorders are also higher than with amprenavir.

Who will benefit from fos-amprenavir?
There are many ill-defined aspects of this drug, including interactions, side effects and resistance, all of which would have been helped through an expanded access program. The generous expanded access programs for amprenavir, which tried to look at questions of lipodystrophy and double PIs, were abandoned in the fos-amprenavir development program.

We urge GSK and Vertex to price fos-amprenavir cost neutral with Kaletra (lopinavir/r) to afford the greatest number of people the option of using the drug.

What is unknown about (or missing from) the fos-amprenavir application?

• Influence of gender on efficacy, side effects and toxicities
• Influence of race on efficacy, side effects and toxicities
• Pediatric dosing
• Liquid formulation
• The clinical relevance of the I50V/L mutation
• Once daily vs. twice daily dosing

These results suggest that insertions in the p6 region of HIV-1 gag gene can affect the virological response in highly pre-treated patients receiving an unboosted amprenavir-containing regimen.

In the Neat and Solo studies, emergence of resistance was examined by ViroLogics via genotypic and phenotypic analyses of virus from all individuals with HIV RNA >1,000 copies/mL at two consecutive visits between weeks 12 and the end of study.

In the Neat study, mutations characteristic of development of fos-amprenavir resistance were detected in virus from 5/29 (17%) fos-amprenavir treated individuals analyzed and included I54L/M, V32I + I47V and M46I. Mutations observed with other protease inhibitors (D30N, I54V, V82, L90M) were not observed with fos-amprenavir. Nelfinavir-selected mutations (D30N, N88D/S, or L90M) were detected in 6/26 (23%) nelfinavir-treated individuals analyzed.

In the Solo study, no selection of protease inhibitor resistance by fos-amprenavir/r was observed in virus from 31 individuals analyzed. Emergence of resistance with nelfinavir was significantly greater, with D30N and/or L90M detected in 20/55 (36%) nelfinavir-treated individuals analyzed. The absence of resistance selection at 48 weeks in the group treated with fos-amprenavir in combination with ritonavir contrasts strikingly with the resistance picture when fos-amprenavir is used as a single protease (as in the Solo study). Clearly, this is another reason to recommend its use only when boosted with ritonavir.

Cross Resistance
A poster from the 2003 Retrovirus conference (poster 598) suggests that fos-amprenavir virologic failures are still susceptible to the gamut of protease inhibitors (i.e., no cross-resistance). Experience tells us that no matter how good the in vitro data, real life can't be beat. We do not know what fos-amprenavir users switch to—and how well the subsequent therapy works—if they are one of the approximately 30% of people in whom it fails.

Novel early mutations to amprenavir may not initially confer cross-resistance to other PIs, but the subsequent accumulation of additional mutations confers broad cross-resistance to the entire protease inhibitor class.

Fos-amprenavir-resistant isolates of HIV have been selected in vitro. Genotypic analysis of these isolates do not show much although the drug has been around for four years! This is simply unacceptable.

What's unknown?

1. Amprenavir is defined as a Class C antiretroviral for pregnancy; that is, it has unknown placental passage. A full four years after approval, GSK has not even completed long-term animal carcinogenicity studies on amprenavir—let alone fos-amprenavir. (Results of animal teratogen studies were negative, but deficient ossification and thymic elongation were observed in rats and rabbits). We firmly believe that fos-amprenavir should be studied and categorized for use in pregnant women.
2. Three ACTG studies were in place at the time of amprenavir's accelerated approval. One of those had to close because of important interaction issues, while another could not recruit sufficient numbers of study volunteers. Results of the third one have never been fully published. The community hopes that transparency and better definition will be the hallmarks of all subsequent Phase IV studies—both those already begun as well as those requested in our full report.
3. The pediatric formulations need to be devised and studied in newborns and infants.
4. Fos-amprenavir/r vs. atazanavir/r (QD) in different populations to help better define the use of fos-amprenavir/r in the real world.
5. Another study in advanced patients—fos-amprenavir/r vs. tipranavir/r vs. lopinavir/r (all BID)—would be very useful.
6. A pharmacokinetic (PK) assessment in people with different stages of hepatic impairment should be done.
7. What is the significance of the I50L and hypersusceptibility?
8. ACTG studies 5015 (accelerated disease and aging) and 5073 (a DOT study) are fully enrolled and should be reporting results soon.

We cannot say if fos-amprenavir is effective in specific populations because the stratified data have not been made available. Fos-amprenavir has not been looked at head-to-head with amprenavir, nor has there been an expanded access program. FDA oversight of Phase IV, as previously implied, is a toothless tiger. Companies agree to do trials with FDA, and are reminded ad infinitum that they need to be done; if they are not, FDA can pull a drug from the market. This has never happened in the history of the HIV pandemic. Fos-amprenavir is the 19th anti-HIV drug to be approved. Is it time to start pulling approved HIV drugs off market when Phase IV commitments have not been honored? ¤

It has been known for some time that an HIV protein called Vif suppresses a human cellular protein that can block viral replication. Last year, this human protein was identified as CEM15, also known as APOBEC3G. It turns out that APOBEC3G targets the mechanism used by HIV to copy its genetic material. A slew of new papers have now homed in on the mechanism underlying the antiviral activity of APOBEC3G. Richard Jefferys files this report under the provocative (if somewhat alarming) title, "Lethal Editor: How A Cellular Protein Scrambles [HIV] Viral DNA." And, it might be added, thwarts HIV replication in the process!

In order to successfully infect its target, HIV must copy its RNA genetic blueprint into DNA, which can then integrate into the DNA of the cell. HIV converts its RNA into DNA in a three step process: first, an enzyme called reverse transcriptase creates a mirror image of the RNA blueprint in the form of DNA. But much like a real mirror image, this DNA copy is back-to-front (it is known as a "minus strand" DNA copy) and must be copied one more time into "plus strand" DNA which can then integrate into the cell's DNA.

APOBEC3G wrecks the conversion of minus strand HIV DNA into plus strand DNA by scrambling the genetic ingredients—known as nucleotides—that make up the minus strand DNA.

Specifically, APOBEC3G converts a nucleotide called cytosine into a different nucleotide, uracil. These nucleotides are strung together in a particular sequence in the minus strand DNA, and act as a code that is read and translated in order to make the plus strand copy of the DNA. The conversion of cytosines into uracils by APOBEC3G scrambles the code of HIV's minus strand DNA, potentially disrupting the viral life cycle in several different ways.

Firstly, because uracils are not normally found in DNA, specialized cellular DNA-repair enzymes are likely to remove these nucleotides, eventually leading to breakages in HIV's minus strand DNA that cause it to be destroyed. Secondly, the initiation of the process that copies the minus strand DNA into plus strand DNA can be inhibited by the presence of the uracil nucleotides. Finally, even if a plus stand copy of HIV DNA is made, the translation process goes awry because a uracil in minus strand DNA is translated into a nucleotide called adenosine in plus strand DNA, whereas a cytosine is translated into a different nucleotide, guanine. Therefore, in the presence of active APOBEC3G, any plus strand HIV DNA that gets made contains an abundance of adenosines instead of guanines.

The genetic blueprint for making new virus thus contains the equivalent of many typographical errors, making it much less likely that functional infectious viruses will be produced. The research teams involved in this discovery hope to find ways of enhancing the anti-HIV activity of APOBEC3G, either by increasing its expression within cells or by blocking the ability of HIV's Vif protein to suppress APOBEC3G activity.

IL-15 Levels and Structured Treatment Interruptions
Cytokines are a family of proteins that are vital to the proper function of the immune system. The best known cytokine is interleukin-2 (IL-2), which has been studied as a potential HIV/AIDS therapy for nearly twenty years. Recently, interest has grown in a related cytokine called IL-15.

Based on studies in animals, IL-15 appears to play a key role in activation and maintenance of CD8 T cell responses and also in the development and maintenance of natural killer cells. A variety of other immune cells may also rely on IL-15 for their proper function.

Early studies have shown that levels of IL-15 appear to be diminished in individuals with HIV infection compared to uninfected controls.

A study published in the Journal of Infectious Diseases now suggests that levels of IL-15 may predict a favorable response to structured treatment interruptions (STIs). A group of Italian researchers led by Massimo Amicosante evaluated 25 individuals participating in an uncontrolled pilot STI trial. Criteria for study participation were >2yrs on HAART with a CD4 count over 500 and viral load less than 50 copies for the past year.

After treatment interruption, 18 participants experienced a rapid rebound in HIV viral load whereas 7 showed a delayed or absent rebound (four of these individuals remain off therapy after more than a year of follow-up). These two groups were classified as "non-responders" and "responders" for the purposes of further analyses. Factors that did not seem to influence the divergent outcomes included age, sex, pre-HAART viral load and CD4 and CD8 T cell counts at the time of treatment interruption.

The researchers decided to look at plasma levels of various cytokines (IL-2, IL-7, IL-15, alpha interferon and TNF alpha) to see if they could identify any relationship with post-STI control of viral load. This analysis revealed a significant difference in the levels of plasma IL-15—at baseline and at all times measured—between the responder and non-responder groups. Both prior to and during the STI, levels of IL-15 were around four-fold higher in the participants with a blunted or absent rebound in viral load. In contrast, TNF alpha levels were significantly elevated in non-responders during the STI, but remained unchanged in responders. Levels of the other cytokines studied did not show any significant variation during the study, and did not appear to affect the post-STI outcome.

The researchers suggest that IL-15 may play an important role in supporting the function of HIV-specific CD8 T cells, thus contributing to control of viral load in the absence of HAART. (Alternatively, elevated levels of IL-15 may be merely a marker for a more intact and functional HIV-specific immune response. Future trials involving the therapeutic administration of IL-15 should help distinguish between these possibilities.) Strategies involving HIV-specific immunization combined with IL-15 are also under investigation in animal models (by David Weiner at the University of Pennsylvania in collaboration with Wyeth Ayerst), and human studies are being planned.

HIV Assembly in Macrophages: Hiding in the Endosome
Macrophages are scavenger cells of the immune system, possessing the ability to engulf and destroy pathogens in a process called phagocytosis. They also play an important role in triggering secondary or "memory" T cell responses to previously encountered pathogens by processing and presenting parts of the pathogen (antigens) to memory T cells, a function known as antigen presentation. HIV can infect some macrophages due to the expression of CD4 and CCR5 molecules on their surface (the virus uses these two molecules as docking points on the cell).

The mechanisms by which newly-produced HIV exits infected cells, including infected macrophages, are incompletely understood. In a new study published in the Journal of Cell Biology, a group of researchers at University College in London provide evidence that HIV particles assemble in compartments called late endosomes in infected macrophages (endosomes are small pockets within cells that selectively take in materials for transport into or out of the cell). This finding contrasts to the situation seen in infected CD4 T cells, where newly-produced virus appears to assemble at the cell membrane.

The researchers speculate that the release of virus from the endosomal compartment in macrophages may be facilitated during the presentation of antigen to T cells: "Thus, macrophage transmission of HIV may be mechanistically similar to the proposed sequestration and release of virus from dendritic cells. These cells can sequester virus, possibly in an endosome compartment, without themselves becoming infected, and subsequently release virus during their interactions with T cells." In other words, HIV may have the ability to use both major types of antigen-presenting cell (dendritic cells and macrophages) as Trojan horses in order to bring the virus into contact with its preferred target, the CD4 T cell.

A Boost for Merck's Adenovirus Vaccine
In a talk at this year's Keystone conference, Merck's Emilio Emini revealed one of the reasons for the disappointing results with their DNA vaccine, as a sequé into his presentation of a new prime-boost regimen utilizing the Ad5 adenovirus vaccine platform in combination with Aventis-Pasteur's ancient ALVAC vector.

The Ad5 vector is susceptible to being blocked by neutralizing antibodies, and because many individuals have been exposed to naturally occurring Ad5 (which causes severe colds), about a third of the participants in Merck's phase I trials possessed high titers of anti-Ad5 antibodies. Emini had originally hoped that the initial use of the DNA vaccine to prime the immune response—before boosting with the Ad5 vector—might overcome any inhibitory effects of such pre-existing antibody responses. The results obtained with this approach, however, have proven disappointing. Emini presented a slide that broke down the data from a phase I DNA prime/Ad5 boost trial based on the pre-enrollment titer of anti-Ad5 antibody.

These results showed that only a minority of individuals with antibody titers over 200 developed a detectable T cell response to the HIV Gag protein, and there appeared to be no difference between those primed with the DNA vaccine compared to those primed with the Ad5 vector. In addition to studying the potential of ALVAC to boost immune responses induced by the Ad5 vaccine, Emini also reported that Merck is developing adenovirus vectors based on rarer serotypes that most people have never been exposed to.

In a poster session at Keystone, Merck presented the first look at immune responses induced in macaques by a prime-boost regimen utilizing the Ad5 adenovirus vaccine platform combined with Aventis-Pasteur's ALVAC vector.

The Ad5 vector was administered as a prime at weeks 0, 4 and 26, followed by a booster shot with ALVAC at week 56. Both vectors carried the HIV gag gene only. The combination proved surprisingly potent, inducing Gag-specific T cells in the range of 400-700 spot-forming cells or SFC (as measured by an ELISpot assay that captures T cells based on their ability to produce the cytokine interferon-gamma in response to stimulation with the HIV Gag protein—a reading over 50 SFC is typically considered significant).

Current candidates that Merck is studying in macaques are Ad24, Ad34 and Ad35 (antibodies directed against Ad5 do not cross-neutralize these adenovirus serotypes). Once more data from these pre-clinical studies have been evaluated, the company will select which vectors to use in future human trials.

Novel Adjuvants for DNA Vaccines
Over the past few years, the number of potential advjuvants (substances intended to boost the immune response to vaccines) being evaluated in research studies has expanded at a dizzying rate. They include cytokines (e.g., IL-2, IL-12, IL-15), chemokines (MIP-1 alpha), co-stimulatory molecules (CD40 ligand), heat shock proteins, oligonucleotides (called CpG motifs) and many, many others. At Keystone, a collaboration between researchers at the National Institutes of Allergy and Infectious Diseases and the National Cancer Institute headed up by George Pavlakis offered intriguing data on yet another novel adjuvant approach.

Pavlakis and colleagues have chosen to use DNA vaccine constructs that encode forms of the SIV proteins Gag and Env that are fused to substances with potential adjuvant effects. One set of constructs (the researchers created one DNA vaccine for each protein) fuses the SIV proteins to a chemokine called MCP-3, with the aim of improving the secretion of the proteins from cells that take up the DNA vaccine (MCP-3 is able to exit cells particularly easily, using a route called the secretory pathway).

The second set of constructs fuses the SIV proteins to a protein fragment (peptide) called β-catenin which chaperones the proteins out of the cell via a different route called the proteasomal degradation pathway. The third pair of DNA constructs employed in the study simply encoded the SIV Gag and Env proteins without adjuvants.

The investigators tested the vaccines in a cohort of rhesus macaques. Four groups of four animals each were assigned to receive DNA constructs containing:

1. Gag and Env alone
2. Gag and Env alone + Gag and Env fused to β-catenin
3. Gag and Env alone + Gag and Env fused to ß-catenin + Gag and Env fused to MCP-3
4. No immunization (controls)

Immunizations were given at weeks 0, 4, 12, 24 and 48, followed by a challenge with the highly pathogenic SIVmac251 at week 54.

The somewhat surprising result—given the virulence of the challenge virus and the fact that only DNA vaccines were used—was that animals in group 3 were able to maintain significantly lower viral loads than controls over 30 weeks of follow up (the study is still ongoing). Although the viral load levels in these immunized animals were around 10,000—100,000 copies, SIVmac251 typically replicates to extremely high levels in naïve animals and the outcome of this experiment suggests that the adjuvant potential of ß-catenin and MCP-3 should be further investigated. ¤