The thirteenth HIV drug is about to enter the market: After a day's worth of agonized debate, the FDA Antiviral Advisory Committee recommended approval of Glaxo Wellcome's abacavir (brand name: Ziagen) by a 7-2 vote. At Treatment Issues' press time, official FDA approval was imminent. Abacavir is a nucleoside analog like AZT, 3TC and ddI. It appeared exceptionally potent in pilot studies and promised to open up new roles for nucleoside analogs in HIV therapy.

Glaxo applied to the FDA under the agency's accelerated approval statutes, which allow for conditional approval of lifesaving drugs based on early data when no feasible alternatives exist. Most HIV drugs were first approved by this mechanism. But the approval of other highly active antivirals, especially the four protease inhibitors, is beginning to encourage a reexamination of the appropriateness of accelerated approval.

Jeff Bloom, the advisory committee's consumer representative for the day, went so far as to oppose accelerated approval on the grounds that Glaxo could document no advantage for abacavir over similar-acting drugs. Bloom wanted to hold off on approval for another six months to a year, until more data were available, and abacavir could qualify for approval under standard FDA procedures rather than accelerated ones.

Ironically, abacavir is better than many approved antiviral drugs. The first trials, which enrolled people without prior therapy, used the drug as monotherapy and recorded viral load drops over the first four weeks amounting to 1.5 to 2 logs (97% to 99%). This is more than any single nucleoside analog can accomplish and as good as AZT/3TC together performed during trials in treatment-naive individuals. The abacavir dose, one tablet (300 mg each) taken twice a day, is also comparatively convenient. Unfortunately, Glaxo's emphasis has been on combining the drug with two other company products, AZT (Retrovir) and 3TC (Epivir). Glaxo gambled on conducting pivotal studies that would quickly show abacavir's utility but problems with each trial's design, or just bad luck (depending on your point of view), contributed to results that were less than persuasive. Abacavir's proper role awaits further definition.

We Know It Works, But...

The largest study (CNA3003) tested the triple combination of AZT, 3TC and abacavir against AZT plus 3TC as initial therapy in only 173 persons. Trial participants were all treatment-naive. Their median baseline viral load was about 40,000, and their median baseline CD4 count was 450.

The comparative phase of the trial lasted only 16 weeks, and even allowing that period is questionable since AZT/3TC by itself was already widely considered deficient when the trial began recruiting in 1997. Persons who take this dual combination risk the early evolution of viral drug resistance since HIV is not completely suppressed. The results came out as expected. In an intent-to-treat analysis that included everyone who started the trial, including those who dropped out, 75% of the participants on three drugs had viral loads below 400 copies/ml (54% were below 50). For those on AZT/3TC only, the corresponding figures were much less. Thirty-five percent had viral loads under 400 copies (and only 15% had viral loads under 50). No one was surprised that three drugs are better than two. The real question is, How good is abacavir relative to other drugs? Everyone was perplexed, though, by the CD4 counts at week 16 -- they increased by a meager 47 cells/mm³ on three drugs and 113 on two. No reason could be found for this difference, and Glaxo noted that CD4 counts rose considerably more during the post-16 week extension phase of the trial, in which everyone received the abacavir-containing triple combination.

CNA3003 was an attempt to explore the value of a first-line regimen that delayed the use of protease inhibitors, but it did not compare a protease inhibitor triple combination to a "protease-sparing" one. The ambiguities in CNA3003 may be resolved by CNA3005, a 48-week trial that directly tests such regimens against one another -- AZT/3TC/abacavir versus AZT/ 3TC/indinavir in 562 volunteers without prior treatment. At the last moment, Glaxo came up with preliminary data from CNA3005 to solidify its case for abacavir. CNA3005 is still blinded, meaning that investigators do not yet know which participant is getting which treatment. They just know that one group of people is getting "Treatment A" and one group getting "Treatment B" (aside from the three active drugs, everyone is also receiving a placebo facsimile of either indinavir or abacavir). An intent-to-treat analysis found that at weeks 12 to 24, about 60% of those in either treatment arm had viral loads below 400. CD4 count changes in both groups were close, increasing by 83 cells/mm³ for Treatment A and by 103 for Treatment B.

The two groups look like they responded similarly to treatment, but when the trial is unblinded, it might turn out that certain factors are skewing the results. For example, there is already a 25% dropout rate in CNA3005, and these noncompleters may not be evenly distributed between the two treatments. In the recent past, a high rate of dropouts among those receiving indinavir was one of the problems in DuPont's groundbreaking 006 trial, an open-label comparison of AZT/3TC/efavirenz, efavirenz/Rindinavir and AZT/3TC/indinavir.

A third pivotal abacavir trial, CNA3006, looked at the drug's performance in people with a history of past treatments. The trial enrolled 205 treatment-experienced children who again received either AZT/3TC/abacavir or just AZT/3TC. At 16 weeks, it found a small response with the triple-drug regimen (18% with viral loads below 400 by intent-to-treat analysis) versus an even smaller one with AZT plus 3TC (2%). Another trial testing abacavir's effect in treatment-experienced persons enrolled 99 volunteers with AIDS dementia. When added to prior treatment, abacavir had no apparent benefit for neuropsychologic function -- both the abacavir and control groups improved almost equally over the course of the 12-week trial. The people receiving abacavir did end up with lower viral loads than the others, but they also happened to have started with much lower viral loads (a baseline median of 5,300 copies/ml versus 32,000 copies/ml for those not receiving abacavir).

The results in people with prior treatment were to be expected given the recently reported results from a cross-trial analysis relating response to abacavir to HIV drug resistance in adults. (See abstract 32289 from last summer's 12th World AIDS Conference.) Response to abacavir, as measured by percentage with viral loads below 400 copies/ml at weeks 12 or 16, was greatly reduced in persons whose HIV contained three or more mutations associated with resistance to nucleoside analogs. In the AIDS dementia trial, 66% of participants were in this category. A wide variety of accumulated mutations will attenuate the response to abacavir, with the ones that create dual high-level resistance to AZT and 3TC being the most significant.

The initial phase of the pre-approval expanded access program for abacavir, conducted in the second half of 1997, also illustrated abacavir's lack of effect in individuals with a record of poor response to antiviral therapy. The program enrolled 2,200 people who had CD4 counts below 100 cells/mm³ and viral loads over 30,000 copies/ml. These advanced patients previously had failed at least two nucleoside analogs and one protease inhibitor or were intolerant to all marketed drugs. Only about 25% of enrollees experienced a 0.5 log (70%) or greater viral load reduction after adding abacavir to their other therapy.

Safety Concerns

Abacavir is a comparatively safe drug. In the trials, it seemed to add little toxicity to the AZT and 3TC with which it was usually combined. Nausea, malaise and fatigue and headaches, all frequent occurrences with nucleoside analogs, were the side effects most associated with abacavir.

There was one big exception to this benign picture, however. Out of all the people who have now taken abacavir, about 3% have suffered severe allergic, or hypersensitivity, reactions within the first few weeks of starting the drug. The constellation of hypersensitive symptoms includes fever, rash, nausea, trouble breathing and muscle pain, which gradually increase until the affected individuals go off abacavir. Symptoms then rapidly diminish, but hypersensitive individuals can go into potentially fatal shock if they restart abacavir. At least one death has resulted. Several members of the advisory committee remarked half-seriously that it would have been better if abacavir came up for approval in the spring, after the flu season was over. Then the rather nebulous hypersensitivity reaction would be easier to recognize by unpracticed healthcare providers.

As yet, Glaxo has no way to predict who is at risk and ensure that they are informed of the dangers. Educating the public on recognizing this flu-like syndrome and avoiding its inherent dangers will be a major task whose details are still a matter of discussion between the company and the FDA.

Abacavir's Price

On December 1, World AIDS Day, Glaxo took the unusual step of announcing abacavir's price before FDA approval had become official. The company's price to wholesalers, $3,540 for a year's supply, is about a fifth greater than AZT, which is a much weaker and generally more toxic drug, and about an eighth less than DuPont's efavirenz, which had better data to support its approval. (See Treatment Issues, September 1998, for the controversy over efavirenz's exceptionally high price.) The original rumors had abacavir priced much higher, even more than the combined AZT/3TC tablet Combivir. Community members attending a November 4 meeting with Glaxo officials report that the company was impressed by the efavirenz dispute and wished to avoid the same sort of public debacle.

Back to GMHC Treatment Issues November 1998 contents page.