What are the perceived advantages and disadvantages of non‐nucleoside reverse transcriptase inhibitors as first‐line therapy?

Over the last 3 years there has been a revolution in the ability to treat HIV infection successfully. This has coincided with the introduction of protease inhibitors (PIs) taken with two nucleoside reverse transcriptase inhibitors (NRTIs). This article discusses whether this remains preferred initial treatment or whether two NRTIs and a non-nucleoside reverse transcriptase inhibitor (NNRTI) may be as good, or better. The first PIs showed remarkable reductions in viral load of between one and two logs in phase 1 and II monotherapy studies[1]. Although such viral load drops have been evanescently seen with some NRTIs, particularly lamivudine[2], viral rebound was extremely rapid because of the selection of resistant mutants. While viral rebound was also seen with PIs this took longer to occur, presumably because a more complex set of mutations was required to reduce the drug's susceptibility, and these were less likely to be pre-existing within the viral swarm. Thus it was appreciated from an early stage that the PIs were extremely potent but were only likely to produce sustained virological suppression if combined with other agents In large-scale clinical endpoint studies, combinations of nucleosides and PIs demonstrated that saquinavir[3], ritonavir[4] and indinavir[5] delayed clinical events and/or improved survival. Rises in the CD4 count and falls in the viral load were surrogate markers for this effect. Thus, some of the newer PIs (e.g. nelfinavir) have been licensed on the basis of changes in surrogate markers rather than because of improvements in clinical endpoints. NNRTIs are a diverse group of chemical structures which inhibit the replication of HIV-1 but not of HIV-2. They all bind to a pocket adjacent to the catalytic site of reverse transcriptase and a variety of mutations within this pocket produce viruses with reduced susceptibility to the drugs. Probably a combination of mutations will produce cross-resistance to all the currently available compounds. The early development of NNRTIs was hampered by evanescent effects on surrogate markers in monotherapy studies because of selection of resistant mutations[6]. Again, their further development has been enabled by combining them with a nucleoside analogue backbone to provide a potent regimen, which appears to be able to suppress viral replication sufficiently to delay or prevent the development of mutants with reduced sensitivity. Nevirapine, when combined with zidovudine (ZDV) and didanosine (ddI), was shown to produce superior results in terms of surrogate markers compared with use of the two NRTIs alone[7]. Efavirenz, when combined with ZDV and lamivudine (3TC), produced similar or even superior results to the same nucleoside backbone and indinavir[8]. This open-label study had a higher discontinuation rate in the indinavir arm and fierce arguments have taken place about whether this represents ‘real life’ or whether the true potency of an efavirenz-containing regimen would have been better studied in a double-blind, placebo-controlled trial. One of the reasons why an efavirenz-containing regimen may be superior to a PI-containing one in clinical practice is the ease of adherence to the once daily schedule. In addition, the high protein binding of this compound and its subsequent long plasma half-life will increase the degree of ‘forgiveness’, i.e. the property that allows variability in the time of dosing without diminishing potency, as the plasma levels remain well above the concentration required to inhibit most viral replication. The third NNRTI, delavirdine, has been licensed in the United States but not in Europe on the basis that combining this with ZDV and 3TC produced superior surrogate marker results to ZDV and 3TC alone[9]. Delavirdine is the only currently licensed NNRTI (at least in some countries), which is an inhibitor of cytochrome P450, and this factor might be used to advantage when NNRTIs and PIs are combined together. One of the hot debates currently in HIV treatment is whether PIs or NNRTIs are the preferred initial therapy. A short while ago there was a strong feeling that all currently available classes of drugs need to be combined together to increase potency. This is now a less strongly held view, partly because it is clear that patients find it difficult to comply with medication in the long term, which will be exacerbated by pill burden and toxicity. Virological suppression can be maintained in the plasma using two classes of drugs, e.g. nucleosides and NNRTIs over the long-term but initial hopes that such therapy might result in complete eradication of the virus have been dashed[10]. Most of the reservations about NNRTIs as first-line therapy concern the limitations of the current dataset available to evaluate their use. In contrast, the disadvantages of using PIs as first-line therapy relate to the difficulties of compliance and toxicity established as a result of clinical therapy. Plasma lipid abnormalities were recognized in the initial phase I and II studies of PIs, but it has only recently been more appreciated that this may be a major toxicity of this class of drugs[11]. The abnormalities in cholesterol and triglyceride are likely to increase the risks of atherosclerosis in the long term although, in the short term, there is little evidence to suggest that cardiovascular damage has occurred[12] and the predictions are that any rise in incidence of vascular disease will eventually be modest[13]. Efavirenz, coupled with two nucleoside analogues, also produces increases in total cholesterol, although a part of this increase is due to cholesterol associated with high-density lipoprotein[14]. In some patients these lipid abnormalities are associated with changes in the body habitus, with loss of fat from the face and from the extremities and an increase in visceral adipose tissue in the abdomen. Such changes are stigmatizing for the patient and can potentially lead to poor compliance. While such changes have been occasionally described with dual NRTIs and also with regimens containing an NNRTI, these events appear to be uncommon and these changes have been seen mainly with PI-containing regimens. Of course, if such changes are slow in development, an association of PI regimens would be expected at the moment, as these have been given to patients for far longer than NNRTI-containing ones. The major toxicities of the NNRTIs are, in the case of nevirapine, sometimes associated with a fatal outcome. There is a small incidence of both severe rash of the Stevens–Johnson type and a fatal fulminant hepatitis. Efavirenz seems much less likely to cause either of these side effects although it is associated with central nervous system (CNS) disturbances which make the drug intolerable in a small percentage of patients. Minor toxicities of both the NNRTI and PI classes of drugs may, in the short term, be almost equally important in determining which treatments are to be preferred, as such toxicities are likely to be particularly important in determining the ability of the patient to adhere to the regimes. These are mainly gastrointestinal side effects in the case of PIs and, again, rash and CNS side effects are important with individual NNRTIs. It is not clear whether new agents in the PI or NNRTI class can be developed which will be free of these side effects. Most clinicians would consider that randomized clinical controlled trials with clinical endpoints as the outcome provide the best possible evidence of drug efficacy. By this criterion a PI-containing regimen is to be preferred to one containing NNRTIs. However, most clinicians believe that the excellent surrogate marker responses produced with two NRTIs and an NNRTI will translate into improved survival. The currently available PIs have to be given two or three times per day as they have relatively short plasma half-lives. Adherence may be a problem with drugs taken three times a day and with exacting food requirements (either a fatty meal or being taken on an empty stomach). The currently available PIs either inhibit the cytochrome P450 system or are metabolized by it, and thus a potentially large number of drug interactions may also adversely affect plasma levels and reduce potency. In contrast, efavirenz can be taken once a day and the relatively long plasma half-life means that the timing of the dose is less crucial. There is considerable interest at present in the possibility that HIV may replicate independently in different sites of the body, which are penetrated to a variable extent by different drugs in a particular regimen. The two sites where this is most likely to occur are the CNS and the testes. Poor penetration of one class of drugs into these sites may encourage the development of resistant mutants to the other drugs that are present. Although indinavir may penetrate the CNS relatively well[15], the other currently available PIs penetrate the CNS very poorly. It has not, however, been shown that the development of resistance is related to this poor penetration or that there is a high incidence of AIDS-associated dementia in individuals taking such treatment. Penetration of NNRTIs into the brain is related to the degree of protein binding, with only low levels of efavirenz present in the cerebrospinal fluid (although, as this is free drug, it may be sufficient to inhibit viral replication) and higher levels of nevirapine (see elsewhere in this supplement). There are a number of other areas where the benefits of PIs are confirmed and where data are lacking about NNRTI regimens. Kaposi's sarcoma has been reported to regress on PI treatment. Few data have been presented for NNRTIs, although one study does indicate that the time to requiring further chemotherapy is prolonged in individuals on highly active antiretroviral therapy (HAART) compared with historical controls and that this time is equally prolonged in patients taking NNRTIs or PIs[16]. Relatively few data are available on the value of NNRTIs in advanced HIV infection with very low CD4 counts, as most of the clinical controlled trials have been carried out in individuals with CD4 counts above 200 cells/μL. Recent data from the 10th European Congress of Clinical Microbiology and Infectious Diseases (ECCMID) have shown that an efavirenz-containing regimen is equally effective in patients with low-base line CD4 counts as in those with less advanced disease[17]. It has been well demonstrated that PI-containing regimens are able to inhibit HIV replication in lymph nodes. The data are less complete with NNRTI regimens, although one small study does indicate that such regimens are successful in this regard[17]. One of the encouraging developments associated with the treatment of relatively advanced HIV disease with HAART is a striking improvement in immune function as well as numbers of CD4 cells. Most of the studies involve individuals taking PI-containing regimens, although some individuals in Professor Autran's important studies were also taking NNRTIs[18]. Recent data have also indicated that when people are switched from PI-containing regimens to NNRTI-containing (efavirenz) ones because of toxicity, there is improvement in the immune repertoire following such a switch[19]. The question of which is the optimum therapy to begin with when treating HIV will only be answered finally when the results of ongoing long-term comparative studies, such as the European/Australian INITIO study and ACTG384, become available.