DuPont's central trial is Study 006, which compared AZT/3TC/efavirenz to efavirenz/indinavir (Crixivan) and AZT/3TC/indinavir. There were 450 participants, who had no prior experience with 3TC, NNRTIs or protease inhibitors. Overall mean initial viral load was 4.77 log (58,900), and the mean baseline CD4 count was 345. These nearly treatment-naïve individuals have now been followed through 36 weeks. Study 006 was analyzed in several highly conservative ways that took into account the experience of those who dropped out of the trial -- which in any trial tend to be the ones doing most poorly.
At week 36, 64% of those who started on AZT/3TC/efavirenz had attained viral loads of less than 50 copies/ml of plasma, compared to 43% of those on efavirenz/indinavir and 44% of those on AZT/3TC/indinavir. The rest either had viral loads above the 50 copies/ml mark (the limit of the viral load assay's ability to accurately quantify HIV) or had dropped out of the trial.
The use of this "intent-to-treat" analysis is important because about 48% of the people in the AZT/3TC/indinavir arm had dropped out by week 36, compared to about 32% in the AZT/3TC/efavirenz arm. The dropout rate in the efavirenz/indinavir arm was 37% by week 36. Evidently, the combined effects of AZT and indinavir made a lot of people leave Study 006 prematurely. The disappointment of not receiving efavirenz in this open-label trial also may have played a role. Either way, one wonders about those who stayed in the trial -- were those on AZT/3TC/indinavir really taking all of their pills, especially the thrice-daily indinavir?
An on-treatment analysis (which included only those remaining on their assigned treatment) of percent below 400 copies/ml at week 24 found no significant difference between treatments overall, and almost exact equivalence in those with initial viral loads under 100,000 (about 90% of those remaining on their study treatment attained viral loads below 400 in each arm). In the group starting at above 100,000 copies, the proportion below 400 was still about 90% for AZT/3TC/efavirenz and only 75% for both AZT/3TC/indinavir and indinavir/efavirenz. A different protease inhibitor or a different population might have made a difference in the comparative results.
For example, DuPont's first trial, Study 003, was conducted in a similar protease inhibitor- and NNRTI-naïve cohort as 006's. Mean initial viral load was about 100,000 while the baseline CD4 count was about 280. Study 003 compared efavirenz/indinavir to indinavir alone for 12 weeks, with d4T and efavirenz then added to the indinavir monotherapy. By week 60, 74% of the 53 persons starting on efavirenz/indinavir had achieved a viral load below 50 copies/ml, based on an intent-to-treat analysis that counted the last recorded measure as the week 60 result for the nine dropouts. By comparison, the same figure for those on indinavir plus deferred d4T/efavirenz is 63% (including the five dropouts, whose last measured viral load also was carried forward to week 60).
(Indinavir monotherapy is considered substandard treatment that promotes the development of indinavir-resistant HIV. Many critics questioned the ethics of employing such a treatment in the trial [see Treatment Issues, May 1996]. The 003 figures, even though better than those from 006, bear out this criticism.)
Study 020 was another trial whose population varied only slightly from 006's. Included were volunteers without prior treatment with protease inhibitors and NNRTIs but who had had more than eight weeks of treatment with nucleoside analogs. The 282 volunteers could switch their nucleoside analogs as desired at weeks 0 and 16. At week 24, 49% of the cohort receiving efavirenz/indinavir/2 nucleoside analogs had attained a viral load under 50 copies/ml, compared to 36% for the group receiving just indinavir and the two nucleoside analogs, according to an analysis that counted the dropouts as above 50 copies/ml. This result might seem impressive evidence of the extra benefit efavirenz brings to an antiviral combination, but on second thought, the figures just demonstrate that four drugs are more often successful than three. The individual advantage that efavirenz imparts is impossible to gauge. To do that would have required a head-to-head comparison of two four-drug regimens, one with efavirenz and one without.
A similar objection can be raised to ACTG 364, which studied 195 heavily pretreated individuals who still had viral loads above 500. Study populations received a double nucleoside analog combination plus either the protease inhibitor nelfinavir (Viracept) and/or efavirenz. An intent-to-treat analysis that considered the dropouts as treatment failures was carried out at week 24. By that analysis, about 75% of those on the two nucleoside analogs/nelfinavir/efavirenz combination had a viral load below 500 copies/ml (this trial did not use the ultrasensitive PCR viral load assay that goes down to 50 copies/ml). For the study participants receiving efavirenz plus two nucleoside analogs, the corresponding figure was 65%, and for those taking two nucleoside analogs plus indinavir, about 50% were considered below 500.
The difference between these last two arms was not considered statistically significant. Although there was a small trend toward superiority for efavirenz over nelfinavir when looking at the two three-drug regimens, by week 16, both seemed to be regressing in terms of percent below 500 copies/ml, indicating viral rebound in some study participants. The four-drug arm, on the other hand, was holding stable at about 80% below 500 copies from week eight on. Once again, the main message of this trial is that four drugs have more antiviral activity than three.
NNRTIs have always had a problem finding their proper role in HIV therapy because of the ease with which HIV develops resistance to them. In the original test tube studies, efavirenz seemed different than the other two, nevirapine (Viramune) and delavirdine (Rescriptor). But such was not to be: As with the other NNRTIs, one point mutation in HIV confers extensive resistance to efavirenz, although the magnitude of this resistance is somewhat less. About 70% of the viral rebounds in people on efavirenz contain a mutation at amino acid 103 on HIV's reverse transcriptase. This mutation reduces efavirenz's potency by 10- to 30-fold. This is enough to render efavirenz completely ineffective in those who absorb the drug into their blood relatively poorly. Further treatment failure is observed as the virus accumulates one or two other resistance mutations. Nevirapine and delavirdine are even more sensitive to the 103 mutation and will never be usable as salvage therapy in persons with prior efavirenz therapy. Efavirenz, despite its high antiviral activity, therefore cannot be used alone, as monotherapy. In the few people in whom this was attempted, the 103 mutation appeared within a few weeks. The drug needs to be supported by powerful concurrent drugs that will repress any efavirenz-resistant HIV as it appears.
There are three ways to add efavirenz that are supported by the clinical trial data: One can either combine it with just a protease inhibitor, with a protease inhibitor plus a nucleoside analog or two, or with two nucleoside analogs alone. Actually, the simplest regimen is to spare the nucleoside analogs and use efavirenz plus a protease inhibitor only. The problem with that type of regimen is that if someone's viral load rebounds on it, then their HIV probably is resistant to not only efavirenz and the protease inhibitor, but quite possibly all the other members of the NNRTI and protease inhibitor classes as well. All that person would have to fall back on are the nucleoside analogs. He or she would be in the unstable situation everyone was in prior to 1994.
Efavirenz added on to protease inhibitor plus nucleoside analogs provide extra insurance as first-line therapy, and succeed in many instances as salvage therapy. They may be efavirenz's most popular immediate use, but they are very onerous in terms of dosing schedule and side effects. The simplified regimen employed in Study 006, which is limited to reverse transcriptase inhibitors, has gathered a lot of interest, although as noted above, its durability and overall success rate have yet to be determined.
Efavirenz plus AZT/3TC combined in the Combivir formulation will require patients to take only three pills a day once the 600 mg efavirenz pill is released. Such a regimen offers the hope of avoiding the metabolic problems (high blood lipids, concentration of fatty tissue on the abdomen and upper back) and digestive side effects associated with protease inhibitors. And if that regimen fails, protease inhibitors and other nucleoside analogs can be prescribed to overcome the resistant virus. (The protease inhibitors would have no cross-resistance issues to confront, but the nucleoside analogs would have to be chosen carefully. For this reason, it might be better to start with efavirenz/d4T/ddI -- see Treatment Issues, Feb. 1998)
An alternative to using efavirenz up front is to apply it in salvage therapies for people failing protease inhibitors, whether because of viral rebound or side effects. Again, there would be no cross-resistance issuesassuming that the person was NNRTI-naïve -- and doctors' hesitations about abandoning protease inhibitors would not be an obstacle. Individuals who do not obtain adequate HIV suppression with protease inhibitors may have no choice but to move to a protease-sparing regimen.
Before replacing a protease inhibitor with efavirenz, one should remember that efavirenz can have its own toxicities. In the first place, it is not yet clear whether the lipid abnormalities observed with protease inhibitors are due to these compounds' effect on the liver, some more occult side effect, or a lingering consequence of long-term HIV infection. Already there are reports of increased cholesterol levels in people taking efavirenz but not protease inhibitors, and the effect may be reinforced when efavirenz is combined with the protease inhibitors.
Besides profoundly inhibiting HIV, efavirenz does affect liver metabolism, in this case by stimulating the CYP3A enzyme pathway that metabolizes many drugs. For this reason, the standard indinavir dose should be increased from 800 mg to 1,000 mg three times a day. DuPont, in a controversial move, is recommending that saquinavir soft gel capsules (Fortovase) not be taken as the sole protease inhibitor with efavirenz because the latter drug reduces Fortovase levels in the bloodstream by about 60%. A trial combining efavirenz with the protease inhibitors Fortovase and ritonavir (Norvir, which is a strong CYP3A inhibitor) has been delayed because of the shortage of ritonavir capsules (see Treatment Issues, July/August 1998). Similar problems may arise when taking efavirenz and nelfinavir (Viracept) concurrently, although DuPont has issued no recommendation on that score. The company has advised people on efavirenz to find a substitute for the antibiotic clarithromycin (Biaxin), though.
Aside from the drug interactions, other side effects are mainly neurological, in particular dizziness, vivid dreams and euphoria. These symptoms usually, but not always, recede after the first month. They can be ameliorated by taking efavirenz at bedtime. One odd neurological side effect concerns reports that those on efavirenz can test positive for marijuana. This could pose serious problems for individuals in certain jobs. Other major efavirenz side effects involve skin rashes, which are almost always mild to moderate and transient.
A big fear with efavirenz is its tendency to cause birth defects (see Treatment Issues, March 1998). Out of 20 monkeys who gave birth after receiving efavirenz throughout pregnancy, three had babies with severe head deformities (one had no brain). These deformities, if they are due to efavirenz (there were no malformations in the control group), are considered to occur from exposure during the first phase of pregnancy -- the first trimester in humans. In addition, an increase in fetal "resorption" was observed in pregnant rats on efavirenz, although no problems were noted with rabbits.
As a result of these findings, DuPont is issuing warnings that women on efavirenz use two methods of birth control (one a barrier method) to avoid getting pregnant. It remains to be seen how prominent these warnings will be or how well they will be heeded. DuPont is also considering further testing. To monitor fetal outcomes of pregnant women exposed to efavirenz, DuPont has joined the Antiretroviral Pregnancy Registry maintained by several companies with anti-HIV medications. Physicians may register patients by calling 800/258-4263.
It is worth noting that there are several protease-sparing alternatives to efavirenz. The other two NNRTIs, nevirapine and delavirdine, have both shown good results in such regimens (see accompanying article, "NNRTIs: A Neglected Class"). Nevirapine plus protease inhibitor trials have also yielded promising results, but nevirapine has not been tested as extensively as efavirenz. Still, the nevirapine data is encouraging enough that many people are interested in head-to-head trials of the two.
There is also abacavir (Ziagen or 1592U89), Glaxo Wellcome's new nucleoside analog. Abacavir is very potent when used as first-line therapy, but its activity falls off in people whose HIV contains multiple mutations conferring resistance to nucleoside analogs. In a trial announced at the World AIDS Conference, 173 treatment-naïve volunteers were administered either AZT/3TC or AZT/3TC/abacavir. Median baseline viral load was about 40,000 and median baseline CD4 count was about 450. At 16 weeks, an intent-to-treat analysis found that 54% of those on AZT/3TC/abacavir now had viral loads below 50 copies/ml, compared to 15% on AZT/3TC alone. These very short-term results are suggestive, but they are subject to the same criticism noted in some of the efavirenz trials: This trial demonstrates yet again that three active drugs are better than two without evaluating abacavir's relative worth. This summer, Glaxo commenced a year-long trial directly comparing AZT/3TC/ abacavir with AZT/3TC/indinavir. The new trial should provide more decisive data.
Other impressive "protease-sparing" regimens include one or two nucleoside analogs (usually ddI with or without d4T) plus hydroxyurea to boost their activity and make the regimen "resistance-proof" (see Treatment Issues, July/August 1998). For that matter, two nucleoside analogs alone can be successful at stabilizing HIV progression for several years before viral resistance overwhelms it. Backup therapies could still be chosen at that point from all the NNRTIs, all the protease inhibitors and a range of nucleoside analogs.
But all these approaches, with a paucity of long-term data to support them, remain speculative and risky. More research is needed into narrow regimens that provide not only simplicity in dosing but also high-level viral suppression and minimization of cross-resistance so that there is plenty of room for successive therapies.