Despite the availability of 23 approved antiretroviral drugs (Table 1), there is continued interest in developing new agents for the treatment of HIV/AIDS for the following reasons:
- An effective vaccine or comparable preventative is unlikely to become available for years, and the epidemic will therefore be sustained in endemic areas and increase in new regions where the virus is introduced.
- There will be a need for better tolerated, more convenient and less expensive treatments.
- Increasing resistance to existing drugs will prompt the search for agents in classes that do not share cross-resistance, and for new classes of drugs that would not be affected by such resistance.
For the foreseeable future, HIV is likely to be one of the most common chronic infectious diseases on the planet. As antiretroviral drugs improve in tolerability, safety and long-term efficacy, there will be continued impetus to put more patients who are infected on treatment. This, in turn, will expand the market for existing and new products, and encourage innovation in discovery and development. Although future prospects are bright, important questions remain about the forces likely to drive HIV pharmacotherapy for the next 25 years.
Overview of present treatment options. Published guidelines for developed countries now recommend that treatment offered to all persons infected with HIV with a CD4 lymphocyte count of <350 cells per mm3, regardless of their plasma HIV RNA concentration (also known as viral load )1. All recommended regimens include a minimum of three active drugs; starting regimens generally consist of a non-nucleoside reverse-transcriptase inhibitor (NNRTI) or HIV protease inhibitor combined with two nucleoside or nucleotide reverse-transcriptase inhibitors (NRTIs).
Possible regimens for treatment-experienced patients are more numerous, and include consideration of all potential agents to which the patient's virus is sensitive. One recent review of the records of a single large HIV clinic in the United States found that more than 800 unique regimens had been prescribed to several thousand patients within a decade (M. Saag, University of Alabama at Birmingham, personal communication).
As many patients modify their antiretroviral regimen at least once during the first 3 years of therapy2, a substantial fraction of the antiretroviral drugs prescribed in the developed world are used by treatment-experienced patients. For various reasons, treatment-experienced patients are themselves at increased risk of regimen failure3. Resistance therefore continues to serve as a major impetus for new drug development, as patients harbouring drug-resistant strains of virus are a large group in need of new treatment options.
Of note, present guidelines already place millions of patients worldwide in the treatable category. This number is likely to grow as regimens become better tolerated, more convenient and more widely available, especially in developing countries. Whether starting regimens that are easier to take will produce fewer long-term failures, and therefore reduce the need for new drugs with better resistance profiles, is an important unanswered question.
Opportunities for drug discovery
As shown in Fig. 1, there are many steps for potential pharmacological intervention in the replication cycle of HIV. Despite this, approved drugs attack only three targets: reverse transcriptase; protease; and viral entry (Table 1; Fig. 1). Table 2 provides a list of promising investigational oral agents in clinical development. Two new drugs — maraviroc (Celsentri/Selzentry; Pfizer), a chemokine (C-C motif) receptor 5 (CCR5) chemokine receptor antagonist, and raltegravir, a HIV intergrase inhibitor — have novel mechanisms of action. Maraviroc was FDA-approved in August 2007 and raltegravir is likely to be approved in the United States before the end of 2007.
Figure 1 | Replication cycle of HIV with current and possible targets for antiviral intervention.
Proteins that are targets for approved drugs are coloured green: gp160, reverse transcriptase, HIV protease and chemokine (C-C motif) receptor 5 (CCR5). Integrase (yellow) is a target in advanced clinical development. Proteins that are more speculative drug targets are coloured red: chemokine (C-X-C motif ) receptor 4 (CXCR4), RNase H, tat, rev, nef and vif.
Although new integrase and entry inhibitors have generated a great deal of excitement (see below), many of the drugs that are currently in advanced clinical development belong to existing classes. For example, two new NNRTIs — rilpivirine (also known as TMC-278) and etravirine (also known as TMC-125) — are expected candidates for approval within the next few years because there is no cross-resistance development to existing drugs in this class and they may be better tolerated. New HIV protease inhibitors with improved resistance profiles, better tolerability and convenience may also be approved in the near future. Motivation to develop new NRTIs could diminish, as existing agents in this class already possess better long-term safety and tolerability than older agents.
Although the major viral targets — envelope, reverse transcriptase, integrase and protease — are already marked by drugs approved or nearing approval, there are a limitless number of new inhibitory mechanisms available for each of these viral proteins. For example, the maturation inhibitor bevirimat (also known as PA-457), which has shown anti-HIV activity in 10-day clinical trials, inhibits proteolysis by binding directly to a specific cleavage site in the gag polyprotein, rather than by binding to protease4. Novel mechanisms can therefore be exploited for existing targets.
Novel antiretroviral targets
Integrase. The integrase inhibitor raltegravir appears to be as potent as any previously developed antiretroviral drug in terms of its short-term5 and 24-week anti-HIV effects6, 7. It is expected to become a mainstay of second-line therapy, if not an eventual candidate for first-line use. The advantage of this drug target is that integrase is an essential and highly conserved enzyme. However, one disadvantage is that moderate-level to high-level resistance to this and other integrase inhibitors can follow after only one or two amino-acid mutations8. A second integrase inhibitor, elvitegravir, is also in advanced clinical development. In September 2007, an independent FDA advisory committee voted for accelerated approval of raltegravir.
CCR5. The CCR5 antagonist maraviroc has excellent short-term anti-HIV activity, and is associated with substantial efficacy after 24 weeks of treatment when combined with antiretroviral nucleosides in treatment-experienced patients9, 10. The attraction of this chemokine co-receptor target is that virtually all individuals are initially infected with the CCR5-trophic virus, and maraviroc is the first approved oral drug in the broad category of entry inhibitors. One drawback, however, is that maraviroc has little or no activity against viruses that use the chemokine (C-X-C motif) receptor 4 (CXCR4) as a co-receptor, or have dual/mixed tropism. Patients will need to be screened for virus co-receptor use with a commercial tropism assay before receiving this drug.
A second CCR5 antagonist, vicriviroc, has efficacy in Phase II trials in treatment-experienced patients11, and is entering Phase III trials.
CXCR4. Two CXCR4 antagonists have demonstrated anti-HIV activity in small clinical studies, although changes in the plasma concentration of CXCR4-tropic viruses have been inconsistent, with some subjects having no measurable response12, 13, 14. Both drugs cause a transient elevation in circulating neutrophil, lymphocyte and monocyte cell counts in treated subjects, which suggests a role for CXCR4 in peripheral trafficking of these cells15. In fact, one of these drugs, AMD3100 (plerixafor), is currently being used for adjunctive mobilization of stem cells in patients with various malignancies. The CXCR4 receptor and its native ligand, stromal cell-derived factor 1 (SDF1), also appear to play a critical role in central and peripheral axon migration and nervous system development16; Cxcr4-knockout mice have severely malformed brains. It is expected that every antagonist of this receptor will be teratogenic in humans. Whether these agents will also deter peripheral nerve regeneration in adults, or accelerate HIV-associated peripheral neuropathy, will require further study.
An important unanswered question about chemokine receptor antagonists is the potential long-term toxicity that might be associated with using an anti-infective drug that targets a host protein. CCR5 and CXCR4 have physiological roles in mediating host defences, and CXCR4 plays an important role in fetal development. Although humans who are homozygous for a 32-amino-acid deletion mutation that disables CCR5 are phenotypically normal under most circumstances, and are protected from infection by HIV-1 (Ref. 17), recent studies indicate that inactivation of this receptor greatly increases the risk of severe encephalitis after West Nile virus infection18. One study with the CCR5 antagonist vicriviroc reported four cases of lymphoma in 86 treated patients during the first 48 weeks of study, although a causative role for the drug in these malignancies has yet to be proven11. So far there appears to be no significant association between maraviroc treatment and lymphoma or other malignancies.
HIV regulatory proteins. HIV encodes several regulatory proteins whose functions range from transcriptional transactivation of the virus genome to antagonism of host defences (Fig. 1). As critical virus-encoded proteins, these represent potentially important drug targets; however they have failed to attract as much interest as virus elements already mentioned for several reasons. First, although these regulatory proteins exist to maximize replicative efficiency, most are not essential for the virus to reproduce. For example, the nef protein, which inhibits apoptosis in infected cells, can be deleted to produce a mutant virus of reduced virulence in primates. Still, nef deletion mutants are perfectly capable of propagating and infecting other animals, and can cause disease in humans19. A chemical nef antagonist is likely to quickly select for resistance, as its target is non-essential, and might not produce detectable reductions in plasma HIV RNA to the same extent as existing drugs.
A second possible difficulty faced by inhibitors of regulatory proteins is mechanistic. As the targets are usually protein–protein or protein–nucleic acid interfaces, rather than enzymes, there are pharmacological hurdles to impeding interactions involving such large surfaces. Parallel models for successful drug development in other diseases are few.
Perhaps as a case in point, two drugs targeting HIV regulatory proteins had no significant antiretroviral activity in previous clinical studies. A selective HIV tat antagonist had no anti-HIV activity in 96 patients treated for 12 weeks20, and mifepristone, a progesterone analogue that interferes with vpr function in vitro, had no detectable anti-HIV activity at doses up to 225 mg per day when given to 56 HIV-infected subjects for up to 28 days21.
Other host proteins. Successful replication of HIV requires the participation of several host proteins22. Any of these could become potential antiviral targets. However, most cellular processes that are essential to the virus are also essential to the host, and targeting these pathways in a way that is selectively toxic to the virus may be difficult.
More promising are strategies to enhance naturally occurring host defences. For example, APOBEC3G, a cellular enzyme of the cytidine deaminase family, is incorporated into virus particles and is capable of inactivating HIV RNA or DNA. The HIV vif protein facilitates APOBEC degradation and thus counteracts its effects23. Pharmacological strategies that neutralize vif or upregulate APOBEC expression might therefore protect cells from infection.
Strategies for boosting host immunity have also been tested. Results of therapeutic vaccination trials have been disappointing to date; whether new, more sophisticated vaccine approaches will have antiviral efficacy in established infection is speculative. Cytokine therapy remains of interest to some investigators. Intermittent therapy with interleukin 2 (IL2) certainly boosts CD4+ lymphocyte counts, but has no beneficial impact on viral RNA concentrations24. Given its expense, toxicity and need for parenteral administration, IL2 is unlikely to prove cost-effective for most patients. Other nonspecific immune-based therapies, including systemic corticosteroids and cyclosporine A, have failed to produce consistent benefit in short-term trials, and possess long-term side effects that render them unattractive.
The next 25 years
Will some drug classes be favoured over others? As convenience and tolerability improve, and the number of highly active agents expands, long-term drug safety becomes increasingly important. One must assume that any agent — or combination — sustaining plasma HIV RNA concentrations below the detection limit will have an equivalent mortality and morbidity benefit for subjects infected with HIV25. The major factors distinguishing one regimen from another are likely to be toxicity and tolerability. Several existing agents — for example, lamivudine (Epivir; GlaxoSmithKline/Shire Pharmaceuticals), emtricitabine (Emtriva; Gilead) and tenofovir (Viread; Gilead) — have absent, or nearly absent, long-term side effects, and this list is expected to grow in the future. An important obstacle to creating safer drugs is the lack of cell lines or animal models that precisely predict long-term drug safety in humans.
Some current regimens are associated with an increased risk of cardiovascular events such as myocardial infarction or stroke26. To what extent this is a consequence of direct toxic effects of the agents involved, a reflection of the ageing population under study, or an unavoidable metabolic complication of reversing established HIV infection, will require further study. Several different drugs have been associated with increased cardiovascular risks, which suggests a generalized metabolic mechanism. However, certain agents, including the widely used protease inhibitor ritonavir (Norvir; Abbott), have a greater propensity to elevate blood lipids such as cholesterol and triglycerides26. As these markers represent increased risk of morbidity and mortality, agents associated with such problems are likely to be shunned once alternatives become available.
The possible global impact of HIV-2 and its comparative resistance to some classes of antiretroviral drugs is an issue that has been underappreciated. All approved NNRTIs are inactive against HIV-2, whereas NRTIs and protease inhibitors appear to be equally effective against both families of the virus. Although HIV-2 is currently restricted in distribution to parts of West Africa27, increasing prevalence of this virus might sway drug development decisions towards more broadly active agents. Drug resistance issues related to co-infection with HIV and hepatitis B virus are also likely to assume greater importance (Box 1).
Is drug resistance unavoidable? Current dogma dictates that all effective antiretroviral agents can and will select for drug-resistant virus. This view is driven by an understanding of the huge reservoir of replicating virus (1 
1010 infected cells in an average patient), and its mutational propensity (about three mutations per virion per round of replication)28. Combining agents without cross-resistance promotes long-term efficacy. No treatment strategy has violated this paradigm, although some agents with measurable anti-HIV effects in patients, such as interferon-
(IFN), have not yet been shown to promote drug-specific resistance29. In the case of IFN, this may simply reflect the nonselective or pleuripotent mechanism of action of the drug.
One paradoxical drug development strategy involves promotion of mutagenesis during HIV replication, using ribonucleoside analogues, so that no progeny virus can survive30. In effect, the virus mutates itself out of existence. Whether HIV can develop resistance to such a mechanism remains unclear.
Approaches targeting host proteins, for example chemokine receptors, might seem less prone to select for drug resistance. Experience so far suggests otherwise. In one instructive example, HIV can become resistant to the CCR5 antagonist maraviroc by modifying its envelope glycoprotein, gp160, so that it can bind to the host chemokine receptor even with inhibitor in place31.
Although drug resistance may be inevitable, treatment failure due to resistance is not. Strategies in which single active drugs are added to failing regimens are no longer acceptable because they predictably produce drug resistance and another round of treatment failure. The increasing number of approved and investigational antiretroviral drugs should help eliminate this scenario. Future trials for heavily treatment-experienced patients will come with the expectation that every regimen will contain at least two active drugs32 (Box 2).
Ultimately, the best way to avoid resistance is to treat the patient with an effective combination of drugs, regardless of the individual agent's barrier to resistance. It should not be forgotten that efavirenz (Sustiva/Stocrin; Bristol–Myers Squibb), which is only a single point-mutation removed from complete resistance, is one of the most widely used and effective of antiretroviral agents. Its susceptibility to resistance is offset by high antiviral potency, an excellent pharmacokinetic profile — including a long elimination half-life — and a reasonable side-effect profile in most patients.
Which strategies are most likely to succeed? For regimens without substantial toxicity, treatment failure is the consequence of non-adherence or pre-existing resistance. Baseline resistance testing reduces the chance that patients will receive drugs to which they are resistant. But the most successful treatment strategies minimize or tolerate occasional non-adherence. Imperfect adherence is widespread and nearly impossible to prevent, especially in cases where a drug combination will be taken for years. Figure 2a indicates the prevalence of missed and late doses in patients taking clarithromycin every 12 hours for life-threatening Mycobacterium avium infection. As shown, the range of adherence is wide and unpredictable.
Figure 2 | Adherence to clarithromycin (for Mycobacterium avium infection) in HIV-infected patients, and variability in pharmacokinetics of lopinavir (for HIV infection).
a | Patterns of adherence in HIV-infected patients taking a twice-daily drug. Box (25th–75th percentile plus median) and whisker (5th–95th percentile) plots of measured dose intervals for 69 patients infected with HIV taking clarithromycin every 12 hours for the treatment of Mycobacterium avium infection. Dose intervals were measured using Medication Event Monitoring System (MEMS) devices. Figure courtesy of D. Noe and C.F.; unpublished observations from a randomized prospective treatment trial47. b | Measured intra-subject variability in concentrations of lopinavir dosed 400 mg every 12 hours. Participants had undetectable plasma HIV RNA on treatment for at least 3 months, and were seen in the clinic 3 times a week for up to 4 months. Blood for lopinavir concentration analysis was collected at approximately the same time of day at each visit, as described previously47. Figure adapted with permission from Ref. 48 © (2006) The University of Chicago Press.
Past treatment failure is often a surrogate marker for past non-adherence. A non-adherent patient is more likely to fail a new regimen regardless of the number of active agents3. Unfortunately, second-line agents often lack the convenience and tolerability of first-line drugs. For example, the entry inhibitor enfuvirtide (Fuzeon; Trimeris/Roche) is generally active in heavily treatment-experienced patients, but must be injected subcutaneously twice a day. These drug traits exacerbate non-adherence, and this is an agent that would never be given as part of initial therapy to a treatment-naive patient.
As past non-adherence may be a marker for future non-adherence, at least as much attention needs to be paid to improving adherence in such patients as to monitoring resistance and finding active drugs. Over the next several decades, drug development should focus on agents for treatment-experienced patients that are as safe and convenient as agents meant for treatment-naive patients.
Favourable pharmacokinetics, including predictable oral bioavailability and long elimination half-life, are characteristic of the most useful antiretroviral drugs. However, inter-individual and intra-individual pharmacokinetic variability for many of these agents is high (Fig. 2b). Possible explanations include variable food effects, intercurrent medications and drug interactions, and other genetic and environmental influences (see below). The most successful agents must be able to tolerate this degree of variability in drug concentrations, or risk an unacceptably high rate of failure.
As a practical matter, all current antiretrovirals are administered once or twice-daily. Three approved drugs — dideoxycytidine (zalcitabine/HIVID; Roche), delavirdine (Rescriptor; Agouron) and indinavir (Crixivan; Merck) without pharmacokinetic enhancement by ritonavir (Norvir; Abbott) — must be given three times a day, and are rarely used as a consequence. Although once-daily and twice-daily regimens perform equivalently in most randomized clinical trials33, 34, there is a simplicity advantage to once-a-day drugs, especially in patients who require other medications for intercurrent conditions.
Two antiretroviral drugs — zidovudine and didanosine — are already available as generics in the United States. Generic copies of several other antiretrovirals are available in developing countries through accepted circumvention of intellectual property law. The impact of generic antiretrovirals represents a major unknown in future prescribing practices. To date, most treatment decisions in the developed world take place without regard to cost. As effective generic combination regimens and co-formulations (Boxes 3,4) become available in the next 25 years, prescribers may be forced to accept more inconvenient generic drugs (for example a twice-daily, multiple pill regimen) in place of more expensive but more attractive alternatives.
Pharmacogenomics and individualized therapy. Although genetic constitution is an important determinant of drug disposition and response, there are few examples today of treatment decisions being driven by results of genetic tests. In HIV pharmacotherapy, one important and life-threatening toxicity, abacavir hypersensitivity syndrome (HSS), is strongly associated with HLA-B*5701, an uncommon human leukocyte antigen (HLA) genotype, with possible additional contributions from an ancestral haplotype that includes HSP70-HOM (also known as HSPA1L), a locus encoding a heat shock protein35. In a homogenous Caucasian population, the presence of these two markers had a positive predictive value for HSS of 93.8% and negative predictive value of 99.5%. Because of this strong association, many patients are now pre-screened for HLA-B57 genotype before prescribing abacavir. One older survey reported that HLA-B57 was present in only 46% of HSS cases36. However, several recent studies using abacavir sensitivity skin testing indicate that many of the B57-negative cases may represent misdiagnosis or a distinct, milder form of the syndrome.
Other host genetic markers that are predictive of antiretroviral treatment response or toxicity are more weakly associated with outcomes. In one study, patients harbouring a single base-pair polymorphism at position 516 of the cytochrome P450 2B6 gene (CYP2B6; G516T) had median efavirenz concentrations that were twofold to threefold higher than other variant genotypes37. Individuals who were homozygous T/T at this locus had higher median efavirenz concentrations, but also had median central nervous system toxicity questionnaire scores that were more than twice those of G/G homozygotes after the first week of treatment36. Despite these group differences, the range of drug concentrations and side effects overlapped substantially between patients of different genotypes, and the CNS toxicity difference did not persist beyond week one. This interesting association is therefore unlikely to lead to genetically guided treatment decisions.
Genomic screening holds promise in the diagnosis and understanding of some important pharmacological effects, as shown with abacavir hypersensitivity. Additionally, advances in cardiovascular genetics could help guide regimen selection for some groups of patients, for example by avoiding those drugs most strongly associated with cardiac risk. Common adverse drug effects and likelihood of response to any given regimen are likely to be genetically complex and susceptible to environmental influences. Even if access to multi-allelic haplotypes becomes routine over the next 25 years, it is unlikely that many HIV treatment decisions will be guided by genetics. Drug approval tied to a requirement for prior genetic screening could even be a disadvantage for some future antiretrovirals, given the wide array of agents already on the market without such restrictions.
Will special populations drive drug development? Several patient groups have characteristics that merit unique approaches to treatment. This includes infants and young children, pregnant women and, increasingly, the elderly. To some extent, HIV treatment decisions are already being tailored to the special needs of these groups. Efavirenz, which is teratogenic, is avoided in women who intend to become pregnant. Special formulations have been developed to allow individualized dosing of antiretrovirals in paediatric patients.
Whether any of these special populations will ever be large enough to warrant development of drugs mainly targeted to their needs is an open question. Certainly, the number of paediatric patients infected with HIV is large in developing countries. However, this number is expected to shrink as effective prevention of mother-to-child transmission is practiced more uniformly, as has happened in the United States, Europe and Australia.
Will global demographics drive drug discovery? As the majority of HIV-infected patients will continue to live in some of the poorest parts of the world, it is logical to consider whether this market will become more attractive for those who discover and develop new antiretroviral drugs. One immediate implication of these changing demographics is that new parenteral antiretrovirals — unless their properties are extraordinary — will probably attract few proponents. Early screening for oral bioavailability in human subjects is already a pivotal step for most investigational agents, and will remain so in the future. Human microdosing studies with radiolabelled candidates can rapidly identify compounds with optimal pharmacokinetic properties such as oral bioavailability, making this process more efficient38.
Although some drugs will continue to be marketed in the developed world regardless of cost, other agents could be selected on the basis of decreased manufacturing expense and reduced sales price. For example, there could be renewed interest in peptide-based HIV protease inhibitors that could be produced cheaply in recombinant bacteria or plants, especially if problems of oral bioavailability can be solved.
Last, it should be pointed out that response rates to antiretroviral therapy have improved dramatically in the past 5 years, probably as a consequence of better drugs and better knowledge of how to use them2, 39 (Box 4). As treatments improve, so will the impetus to treat more infected individuals. Prospective studies could identify those at greatest risk of HIV-associated morbidity and mortality, for example due to tuberculosis co-infection, leading to targeted campaigns to deliver antiretrovirals to those likely to experience the greatest benefit.
With 40 million people infected with HIV alive today, and a higher number expected in the future, the first prospective study to demonstrate the benefits of earlier intervention, say at a CD4-count threshold of 500 cells per mm3, will increase the number of patients eligible for treatment by millions. This kind of statistic will encourage the development of new antiretroviral drugs for decades to come.
