Before going on to discuss specific recommendations for therapy, it is of value to review the drugs that are presently available for the treatment of HIV. As noted earlier, these drugs fall into three broad categories: NRTIs, NNRTIs, and PIs (

see Table 418-1) . NRTIs and NNRTIs target reverse transcriptase and thus prevent the formation of the DNA provirus. These drugs thus inhibit early steps in HIV infection but do not prevent the production of infectious virions by cells already infected. PIs target HIV protease and prevent the production of mature HIV virions by cells that are already infected. Combination regimens that use both reverse transcriptase inhibitors and PIs thus have the advantage of simultaneously targeting both early and late steps in HIV replication.

Nucleoside Reverse Transcriptase Inhibitors
The first drugs to be developed and approved for the treatment of HIV infection were members of a class of compounds called dideoxynucleosides, special analogues of nucleosides that have a sugar and a purine or pyrimidine base. Included in this class of drugs are zidovudine (formerly azidothymidine [AZT]), didanosine (ddI), zalcitabine (ddC), stavudine (d4T), lamivudine (3TC), and abacavir. In these drugs, the 3 -hydroxy (-OH) group in the sugar is replaced by another group that does not form phosphodiester linkages. These drugs are not active in themselves but must be phosphorylated in target cells to form active 5 -triphosphate moieties. In this form, they block reverse transcription and thus inhibit the formation of a (double-stranded) DNA proviral copy of the viral RNA. The activation of dideoxynucleosides, called anabolic phosphorylation, involves a series of host-cell enzymes (kinases) that usually serve to phosphorylate physiologic deoxynucleosides.

For many purposes, each drug in this broad family is unique. Even a one-atom shift in the sugar or the base of the parent compound can radically change activity and toxicity. Also, there are substantial differences in the rates at which various human cells phosphorylate these compounds and in their enzymatic pathways, and these differences may be important in their antiretroviral activity and differing toxicity profiles (

Table 418-2). For example, thymidine kinase, the enzyme responsible for the initial step in the phosphorylation of thymidine analogues such as zidovudine or stavudine, is a cell-cycle dependent enzyme.

As a result, the activity of these drugs is relatively greater in replicating lymphocytes or cytokine-stimulated monocytes than in resting cells of the same lineage. By contrast, the activity of didanosine, zalcitabine, or lamivudine is not substantially affected by the state of activation of the cells. Therefore, these two groups of drugs can preferentially target different cell populations, and combination regimens combining a member of each group can be especially effective.

As triphosphates, the dideoxynucleosides are believed to inhibit reverse transcriptase in two ways: (1) as DNA chain terminators and (2) as competitive inhibitors for the binding of physiologic deoxynucleoside-5 -triphosphates to relevant sites within reverse transcriptase. Reverse transcriptase, but not mammalian DNA polymerase-alpha, preferentially uses dideoxynucleoside-5 -triphosphates in place of the respective physiologic deoxynucleoside-5 -triphosphates, and this is an important basis for their selective antiretroviral activity. Human mitochondrial DNA polymerase-gamma is also relatively sensitive to inhibition by these drugs as 5 -triphosphates and this may be an important basis for clinical toxicities including myopathy and rare cases of hepatic steatosis in patients receiving dideoxynucleoside therapy.

The first NRTI to be developed clinically was zidovudine. Clinical trials of zidovudine conducted during 1985 and 1986 convincingly showed that the drug was effective at reducing morbidity and mortality in patients with advanced HIV infection. Patients receiving zidovudine were noted to have increased numbers of CD4 cells, improved immunologic function, and clinical improvement. However, if zidovudine is used as a single drug, these benefits are usually transient, lasting from 3 to 6 months in patients with advanced AIDS to a year or longer in patients with earlier disease. It is now appreciated that this drug, like other NRTIs given as single agents, generally induces only a moderate decline in the viral load of HIV (a decrease to about a third to a tenth of the starting value, or of 0.5 to 1.0 log10 viral particles/mL).

Also, the clinical activity of zidovudine as a single agent, like that of other anti-HIV drugs, is limited by the emergence of viral resistance. As described earlier, the development of resistance can be substantially delayed by the use of highly active combination regimens combining two NRTIs with a PI.

There are some important differences among the NRTIs that physicians should be aware of as they build combination regimens. As noted earlier, different enzymes catalyze the intracellular phosphorylation of these agents; and in part because of this, they have different toxicity profiles (

see Table 418-2) . The most frequent dose-limiting toxicity of zidovudine is bone marrow suppression, especially macrocytic anemia. Patients starting on this drug often experience malaise, nausea, and headaches.

Also, patients receiving zidovudine for several months sometimes develop myositis associated with “ragged-red” fibers on biopsy. Occasional HIV-infected patients receiving zidovudine alone or in combination with other dideoxynucleosides have been reported to develop a poorly understood syndrome involving severe macrovesicular hepatic steatosis (a condition related to Reye’s syndrome) and lactic acidosis. A high proportion of these patients have died of this complication. Most of these patients had relatively early HIV infection and were well nourished or even obese. A disproportionate number were female. There is evidence to suggest that this condition, like zidovudine-induced myositis, is caused by mitochondrial toxicity.

By contrast to zidovudine, a principal toxicity of zalcitabine, didanosine, stavudine, and, to a lesser extent, lamivudine is painful peripheral neuropathy, primarily involving the feet. This is generally reversible on discontinuing the drug, but the resolution can take weeks. Some patients with this condition have a decrease in their vibratory sense or ability to discriminate temperatures, but these objective findings are generally less pronounced in relation to the pain than in patients with HIV-induced neuropathy. As with many of the specific organ toxicities induced by NRTIs, neuropathy generally does not appear until after at least 10 weeks of therapy.

A relatively infrequent but serious toxicity seen with several of these drugs is pancreatitis. This complication is best associated with didanosine but is also reported with the use of lamivudine (especially in children), zalcitabine, or stavudine. The incidence of pancreatitis is higher in patients with more advanced disease or with higher doses of the drugs. Some patients receiving these drugs have asymptomatic hyperamylasemia, which may be of either salivary or pancreatic origin. Although it is prudent to temporarily discontinue didanosine (or the other drugs whose use is associated with pancreatitis) in patients with elevated levels of pancreatic amylase, the drugs may be continued in patients who have only elevated salivary amylase levels.

Patients taking didanosine should be counseled to avoid alcohol, and this drug should be avoided in patients with a previous history of pancreatitis. Also, these drugs should be used with caution or stopped if patients are receiving other drugs that cause pancreatitis (e.g., systemic pentamidine). Other toxicities of NRTIs that physicians should be aware of include aphthous ulcers (zalcitabine), arthralgias (stavudine), rash (zalcitabine and stavudine), and diabetes mellitus (didanosine).

About 2 to 5% of patients receiving abacavir develop a hypersensitivity reaction with rash, fever, nausea, and vomiting. The drug should be stopped in such patients and they should not be rechallenged.

All of the NRTIs can be given orally and most have good absorption (60 to 86%) when taken by mouth. However, didanosine is unstable in the acid environment of the stomach, and for this reason it is formulated with buffers as either a tablet or powder. In these forms, it has an oral bioavailability of 30 to 40%. It should be noted that the buffers used with didanosine sometimes cause diarrhea and can interfere with the absorption of drugs such as delavirdine or indinavir that require a low stomach pH. If didanosine is used together with either of these drugs, they should be spaced at least an hour (delavirdine) or 2 hours (indinavir) apart. With the exception of lamivudine, the serum half-life of the NRTI is generally short, on the order of 1 to 1.6 hours. However, the intracellular half-life of most of these compounds is somewhat longer, and for this reason they can be effective when given two to three times daily. In this regard, the intracellular half-life of the active moiety of didanosine is quite long (25 to 40 hours), and this compound is active even when administered twice or even once daily. Efforts are now underway to develop a formulation of didanosine for once-daily dosing. Zidovudine penetrates well into the central nervous system; and of all the dideoxynucleosides, this has the best documented activity in patients with HIV-induced cognitive impairment. Stavudine can also penetrate reasonably well. However, non-thymidine-based NRTIs have better in vitro activity in resting cells, and there is evidence that some (such as didanosine) can have activity in the brain.

The requirement for NRTIs to undergo intracellular phosphorylation can also lead to drug interactions. There is evidence that the 5 -triphosphate of zidovudine interferes with the phosphorylation of stavudine, and these two drugs should not be used together. As noted earlier, there is evidence that thymidine-based and non-thymidine-based NRTIs preferentially target different cell populations, and there are thus advantages to combination regimens that use one drug from each class.

Resistance to most dideoxynucleosides develops relatively slowly, and this is one reason that they are important components of combination regimens. Strains resistant to zidovudine generally have two or more mutations in the gene encoding reverse transcriptase. Of these, substitution of tyrosine (or phenylalanine) for threonine at codon 215 appears to be the most important. High-level (over 100-fold) resistance can develop to zidovudine; however, because it requires several mutations, it generally emerges only after several months or more of therapy. Resistance to most of the other NRTIs can develop with a single mutation; however, the level of resistance attained is generally small (on the order of 10-fold or so) and in part for that reason it also emerges relatively slowly.

A notable exception to this pattern is lamivudine. A single base substitution of valine for methionine at codon 184 can induce highlevel (1000-fold or more) resistance to this drug, and clinical resistance can emerge within 2 to 4 weeks in patients receiving lamivudine as a single agent. If one examines the sugar ring of lamivudine, one notices that it is flipped with respect to physiologic nucleosides (and the other NRTI), and it is possible that this structural difference facilitates the development of resistance. Interestingly, the mutation at codon 184 that confers resistance to lamivudine also partially reverses resistance to zidovudine in HIV strains with a mutation at codon 215, and in part for this reason the combination of zidovudine and lamivudine is associated with long-term activity.

A similar pattern of antagonistic resistance occurs with zidovudine and the mutation at codon 74 that is induced by didanosine, and this combination is also associated with long-term activity. Up to 15% of patients who have received long-term sequential or combination treatment with NRTI have been observed to develop a unique pattern of broad HIV resistance to this class of drugs associated with a substitution of methionine for glutamine at codon 151 along with several other mutations. Because of the importance of NRTI in combination regimens, such patients have very limited treatment options at present. However, two NRTIs under development, lodenosine (F-ddA) and adefovir dipivoxil, have some activity against these strains of HIV and may prove to be useful in these patients. Lodenosine is a fluorinated analogue of didanosine that has a unique resistance pattern and, because of the fluorine substitution, is resistant to acid degradation. It may be suitable for once-daily dosing.

Adefovir dipivoxil is now available under an expanded access program; the principal toxicities associated with this drug are proximal renal tubular dysfunction, nausea, and elevated liver function tests. Other NRTIs currently under development include PMPA and FTC. PMPA is a phosphorylated nucleotide phosphate that can thus bypass the initial phosphorylation step. This drug has been shown to be very active against simian immunodeficiency virus.