ICAAC 2007: Key HIV Research on Antiretrovirals in Development

This is part two of a three-part article. For part one, First-Line Therapy, click here. For part three, HIV/HAART Complications and HIV Drug Resistance, click here.

Listening and Viewing Options Full Interview and Slides Podcast of Interview (29MB MP3, 72 min.) Slide Set (11.6MB PowerPoint) Read Full Transcript Interview and Slides Divided by Topic First-Line Antiretroviral Therapy: Podcast (5.5MB, 13.5 min.) Download Slides (1.1MB PowerPoint) Read Transcript Antiretrovirals in Development for Treatment-Experienced Patients: Podcast (10.6MB, 26.5 min.) Download Slides (4.9MB PowerPoint) Read Transcript HIV Drug Resistance and the Complications of HIV/HAART: Podcast (14.1MB, 35 min.) Download Slides (7.4MB PowerPoint) Read Transcript

Table of Contents

• NNRTIs in Development
• Integrase Inhibitors in Development
• CCR5 Inhibitors
• Footnotes

NNRTIs in Development

Antiretrovirals in development, I think, was the real story of ICAAC 2007. Can you discuss some of the highlights?

Sure. We saw new information about novel classes, as well as new drugs in familiar classes. Let's start with one of the big studies in the field, the study called DUET that evaluates the new non-nucleoside [NNRTI] called etravirine [also known as TMC125] in treatment-experienced patients. This study, DUET 1 and DUET 2, was combined here, and so we saw a combined analysis of roughly 1,200 people who participated in these two randomized studies.²

The design is a simple one: it's etravirine or a placebo, with a background regimen containing darunavir, optimized nucleosides, and the option to use enfuvirtide [brand name: Fuzeon; also known as T-20]. The key to this study, of course, is that etravirine is being developed for non-nucleoside resistant virus, and so patients had to have a history of, or, at that time of entry, evidence of, resistance to the non-nucleosides. Otherwise, patients were also eligible because they had at least three or more primary protease mutations, suggesting that this was a pretty heavily pretreated population.

In terms of demographics, this study involved mostly men, 90% males, about 70% of whom were white. Their viral loads were pretty high, 4.8 log, and their CD4s were pretty low, around 100 cells ... not surprising for a triple-class-experienced patient population with at least three protease mutations.

Of interest is that about half of the patients used enfuvirtide on this study, although only about a quarter started enfuvirtide for the first time while on this study. Also of note is that a fair number of patients had a regimen that was pretty compromised, meaning that when we look at the optimized background, about 16 to 17% had no other fully active antivirals in their background regimen, and that it was either etravirine or placebo with a compromised background for that small subset.

Nevertheless, despite these advanced patient populations, etravirine did exactly what it was designed to do. It improved the response rate by about 50%, meaning it was a 41% response rate without etravirine and 59% with it. So, 40 to about 60 means an improvement by about 50%.

Not surprisingly, the change in viral load out to week 24 was also considerably more, statistically more, by about .7 log on the etravirine versus the placebo arm.

Similarly, there was a CD4 benefit of around 20 cells -- a difference that became clinically significant, not just statistically significant, as we'll get into in a couple of moments.

The virologic suppression was actually kind of interesting, particularly with regards to the question around using enfuvirtide. In recent analyses of this study, what was noted was that etravirine clearly demonstrated activity in the patients who did not have fully active enfuvirtide. In the subset [of patients], for example, who either recycled enfuvirtide or didn't use it all, etravirine clearly improved the response rate. It went from about a third to just over a half of the patients who got their viral load suppressed to less than 50 at week 24.

But in the subset of people who used enfuvirtide for the first time, when the data were presented for the first time in the Sydney conference [the 4th International AIDS Society Conference on HIV Pathogenesis, Treatment and Prevention (IAS 2007)], we saw similar numbers: about 65% on average in the two arms who had viral load suppression. And then we didn't see the additional benefit of etravirine, if enfuvirtide was used in the regimen. So the statisticians were asked to help us understand this, since [we were interested in finding out] where is the effect of etravirine. Just because enfuvirtide's there, this doesn't mean etravirine becomes inactive. What explains this?

It turns out that, by chance, there was an explanation, in that in those patients who got etravirine, they also had, just by chance, more compromised darunavir. Again, recall that everybody took darunavir in this study. By chance, more of the people who were assigned to get etravirine had higher phenotypic resistance to the darunavir. When that difference is adjusted for, what we again see is the statistical advantage for using etravirine versus the placebo, in that what we saw was about a 10% improvement, 11% improvement, in the response rate of etravirine, or no etravirine, in those patients who had enfuvirtide in their background regimen. This is consistent with what we've learned for, again, several years now -- which is, it's not just picking active drugs; it's picking active regimens.

Overall, what we saw is that the use of etravirine was clearly, dramatically important in the subset of people who had no fully active drugs -- active is, again, determined by phenotypic score, suggesting that the drug was not fully active. It's actually pretty impressive that, for the subset of people who had no fully active drugs and who received the placebo arm, only 8% got their viral load to less than 50. But for those on etravirine and that same compromised background regimen, 45% got their viral loads to less than 50 ... not something we would have expected from a non-nucleoside in a non-nucleoside resistant population, and a dramatic difference illustrating that this is not just the typical non-nucleoside with a fragility in a compromised background, but certainly a much more durable response.

Nevertheless, as we added additionally active drugs, we got a better response. We're up to a 60% [response] with one additional active drug, and we're up to 74% for people who have two or more active drugs.

This issue with, of course, new drugs in existing classes is that there can be resistance mutations that confer cross-resistance to new drugs, like etravirine. Etravirine's no exception. There are, in fact, 13 non-nucleoside resistance mutations that can confer resistance to etravirine. The impact of those was again reviewed in this presentation. In the subset of people who had none of these etravirine-associated mutations, 75% of them got their viral load to less than 50. In contrast, if somebody had three or more etravirine mutations, their response rate went down to 41%, or less. Again, consistent with what we understand, which is: This is a great drug for the right person, but, unfortunately, there are resistance mutations that can impact its response.

Thankfully, in the population, at least, that enrolled in DUET, only about 16% of the patients had three or more etravirine relevant resistance mutations, suggesting that these mutations are not common in the population.

Now, I mentioned earlier that the CD4 count was clinically significant, not just statistically significant. How we learned that was a pretty unique presentation that we don't see with most of these studies, but with 1,200 patients you can learn some additional findings.

One of the things that, of course, we're trying to do is to not just change numbers, but help people stay healthier. And in as little as 24 weeks, there was evidence that people were, in fact, healthier. In the overall population, there was a trend of reduced numbers of AIDS-defining illnesses or death in those who got etravirine versus no etravirine. When they looked at the subset of people who were in some ways the most in trouble -- the subset who either did not use enfuvirtide or had recycled enfuvirtide, the subset of people who were the most in trouble clinically -- then there were statistically fewer clinical events in the subset who got etravirine versus placebo. It was 3.8% on etravirine and 8.3% on the placebo arm who developed some clinical illness, even in as little as 24 weeks. It was a highly significant finding, not just for statistics, but for patients.

What are the downsides to these drugs? What are the cautions? Well, really, the only one that was identified is [the incidence of] rash in about 8% of patients. There are no nervous system disorders, as we see with other drugs like efavirenz. There were certainly no abnormal dreams reported in any difference versus placebo, and no difference in hepatic adverse events, as we see with other non-nucleosides, like nevirapine [brand name: Viramune; also known as NVP]. The rash is worth some attention, since it happened at a rate of 8% greater on drug versus placebo. As was true with most of these drugs, the rash typically happens in week 2, and typically lasts for about two weeks. Thankfully, the rashes were grades 1, 2 and 3. There were no grade 4 rashes, meaning no Stevens-Johnson. Two percent of patients did discontinue permanently, but most of the other patients could treat through their rash.

Of interest is that the rash incidence occurred at a rate that was higher in women, but the severity or discontinuations are similar by gender. For whatever reason, it's just more common to have a rash in women than in men. The rash was not associated with CD4 counts or other characteristics of our patients, nor was there evidence that it was a higher rate in people who had a history of non-nucleoside rash.

So, etravirine looks like a very important drug. While it's not yet approved, it is available in expanded access protocols. And clearly, the use of etravirine certainly improved the response rates and, depending on the activity of the regimen, what we can see is consistent relevant activity with etravirine, of course, realizing that there are, unfortunately, some percentage of patients who will have too many resistance mutations to take advantage of this drug. Thankfully, the majority of patients -- as I mentioned, 86% -- have less than three mutations conferring resistance. So this drug looks like it will be an important drug for people with a history of non-nucleoside resistance.

When is it expected for this drug to be approved?

As we understand it, it will be sometime -- obviously the FDA [U.S. Food and Drug Administration] has to review the data, but if the data continue to be as good as the data shown here, it would have to be sometime next year, since certainly the FDA has not had a committee meeting about this drug. So, odds are good it's going to be sometime in 2008.

Integrase Inhibitors in Development

We will now focus on the integrase inhibitor data for treatment-experienced patients, if for no other reason than we certainly are seeing a lot of enthusiasm for this class, based on the results of Merck's BENCHMRK trial, illustrating just how interesting these integrase inhibitors can be in the treatment-experienced populations.^3,4

But the interest in these drugs actually started about a year ago at ICAAC, at ICAAC 2006 [the 46th Interscience Conference on Antimicrobial Agents and Chemotherapy], when we saw the first 24-week results of raltegravir [also known as MK-0518] in a highly treatment-experienced population from a study called Protocol 005.⁵ Here at this meeting, we saw the 48-week results of that population.⁶

The results were essentially the standard dose finding result in a phase 2 design. It was randomized and double blind. Patients got either placebo or one of three doses of raltegravir with an optimized background regimen. To be in this study, people had to be, again, triple-class experienced and use an optimized background regimen in this study.

There was a 24-week, double-blind phase. Then afterwards, there was an open-label phase, in which patients could -- if they did well on raltegravir -- remain on an open label. There was also a crossover for people who were on the placebo arm if they had virologic failure -- or in the raltegravir arms, certainly had access to raltegravir, as well.

It's a phase 2 study, which means that there are about 43 to 45 people per arm. This was again a dose finding study. What we learned from it was consistent with what we saw in BENCHMRK. Those who enrolled into this study were, again, the typical patients who were triple-class experienced. Their CD4 is at 200. Their viral loads are around 4.6, 4.7 log -- around 50,000 [copies/mL]. These patients had a lot of resistance. For example, in terms of their phenotypic susceptibility score -- again, demonstrating a lack of complete activity -- about half of the patients had an optimized background regimen with no fully active drug in the regimen other than raltegravir, or the placebo, suggesting that most of these patients really had a lot of resistance to contend with.

Nevertheless, raltegravir did dramatically improve the response rates, certainly to less than 50. Somewhere around 50% to 60% of patients, even a year later, had viral loads less than 50, in contrast with the placebo arm, in which about 10% had their viral loads suppressed to less than 50. So, a dramatic illustration of this drug.

In order to also understand the durability of response, they presented a Kaplan-Meier. The Kaplan-Meier curves illustrate quite nicely that, if we look at who failed on raltegravir, most of the virologic failure seems to occur in the first 24 weeks. Meaning, if your viral load got to less than 50 by week 24, you typically seemed to stay there, although they certainly didn't break it down as precisely as we might, to answer that question of: How predictive is less than 50 at any given week?

Nevertheless, if there was virologic failure on raltegravir -- and there was in about 30% to 40% of the patients -- it seemed to happen usually in the first 24 weeks, after which failure was much less likely. Not surprisingly, the CD4 response was also much better on raltegravir, about a 100-cell bump, versus almost no change on the placebo arm.

If somebody didn't respond to raltegravir with virologic suppression, as about 30% didn't, there was raltegravir resistance. Indeed, of the 38 people who had virologic failure, almost all of them had integrase resistance. There's one of two pathways that will tell you where resistance has -- it's what's called the 155-pathway or the 148-pathway.

In fact, at the time the samples were analyzed, most patients had two or more resistance mutations, suggesting that this drug is either fully active in a good regimen, or can select for resistance in nonsuppressive regimens. Not surprisingly, the factors that predict resistance or suppression are: how active is your regimen -- the PSS [phenotypic susceptibility score]; did you use enfuvirtide or not, for example; and your viral load -- low viral load patients typically, once again, have a better chance of achieving [suppression] than do people with high viral loads in this and other studies.

The safety of the drug looks very good, however. There were very few patients who had increases in transaminases. There was one or two overall in the raltegravir arm, with very few other toxicities identified in the overall study. So the drug, certainly for 38 weeks in these patients, looked very well tolerated. Indeed, certainly as well tolerated as placebo for a lot of the toxicities that we look at, such as GI [gastrointestinal]-related, headache, fatigue, and so on. It looked pretty similar to what we see on placebo.

There were also no signals of problems of malignancies, as was seen in the BENCHMRK trials. Certainly sick patients can develop malignancies, but there was no evidence from this trial that there's any important rate of malignancies, and certainly nothing different than what you might expect from this patient population.

What we can easily conclude is that this is an important drug, and if used in combination with a good regimen, the majority of patients can achieve virologic suppression to less than 50. In this study, about half the patients did, but certainly that half was based on how compromised their regimens were. As we define more and more successful combinations, we will see more patients having virologic suppression.

That was the theme of the Gilead integrase. Their drug, called elvitegravir [also known as GS 9137], was presented to us for the first time at the retrovirus conference in early 2007 [the 14th Conference on retroviruses and Opportunistic Infections (CROI 2007)].⁷ They delved into the study data for us one more time to help us understand this drug, as well as the importance of a good background regimen.⁸

The design of this trial is very different; again, it's a dose finding study, with 60 to 70 people per arm. The design of this is very different. These are triple-class experienced patients who are randomized to either elvitegravir, one of three doses, or a comparative PI. There was no protease inhibitor used in the elvitegravir arms. In this case, the background regimen consisted of nucleosides with or without T-20; non-nucleosides weren't even allowed. This was an assessment of elvitegravir versus comparative PIs.

Interestingly, about half the patients chose darunavir, a quarter took tipranavir [brand name: Aptivus; also known as TPV], and the other quarter took other PIs that were available, such as lopinavir.

So what did we see in this study? What we saw in this study is exactly what we'd expect to see. The first finding, however, was unexpected, which is that the use of elvitegravir for the first 24 weeks was actually more successful than protease inhibitors. Indeed, the elvitegravir regimen got the viral load about a log lower than did the protease inhibitors. So, whether you looked at the viral load out to week 24 or to week 16, the overall results were -- particularly in the subset of people who had the use of T-20 during this study -- patients had virologic suppression that was a log better on elvitegravir versus a comparative PI. Again, not surprising: a new drug in a new class doing better than a new drug in an existing class.

Probably the single analysis that had the easiest way to summarize this data was an analysis by the background regimen. When we presented the analysis of the best dose, or at least the highest dose (the 125 mg dose), out to week 24, the percentage who got to less than 50 was clearly determined by how many active drugs were given with the raltegravir. For example, if raltegravir was given as essentially functional monotherapy -- no other active drugs -- 15% achieved less than 50 copies at week 16. If they had an active nucleoside, 30% achieved less than 50. Two or more nucleosides achieved up to 44%. If somebody got enfuvirtide for the first time -- a fully active enfuvirtide plus a fully active elvitegravir, with or without nucleosides -- we're up to 74% who achieved less than 50.

Interestingly, that 74% response rate shown here is remarkably similar to what was seen in the BENCHMRK trials,^3,4 in which the combination of raltegravir and enfuvirtide certainly got the vast majority of patients to less than 400. Certainly, if we kind of subtracted the numbers that might be different between 400 and 50, we see remarkably consistent findings here in which the vast majority of patients can get to less than 50 copies with the use of two new drugs in two novel classes.

So I think what we've got now is evidence that we've got two active protease inhibitors, one of which has obviously completed its phase 3 studies, and the other of which a dose has been identified that could now be brought forward into phase 2/3 testing.

CCR5 Inhibitors

So we've got yet another class of drugs now to work with in our field: the CCR5 inhibitors, maraviroc [brand name: Selzentry; also known as MVC] being the first FDA-approved drug in this class. Now, at this meeting, we saw the 48-week results of one of the two pivotal studies for this drug, called MOTIVATE.⁹

MOTIVATE studied maraviroc or placebo in treatment-experienced patient populations. The design is typical for treatment-experienced studies; it's maraviroc (in this study, either once a day or twice a day was tested) versus placebo in a triple-class experienced population. Of note is that this study was also perhaps a bit hampered by the fact that darunavir was not used in this study because of how long ago this study was done versus the availability of darunavir, and tipranavir was used in only 10% of patients. So the background was perhaps not as active as it might be at the current time, given changes in background regimens.

We do, however, see evidence that maraviroc is clearly an important drug. Particularly the twice-a-day, which was FDA-approved. Even after week 48, there was an additional log suppression associated with the use of maraviroc versus placebo, and we've got two to three times as many people whose viral load is less than 50 with maraviroc than without. For example, at week 48, 16% on placebo had a viral load less than 50 while 47% (46.8 to be exact) on maraviroc had a viral load less than 50 copies. Again, supporting that this is an important drug, perhaps a two- to three-fold improvement in response rates, even with a relatively compromised background in terms of the protease inhibitors that were available in this study.

Again, not surprisingly, there's a good CD4 response: about a 70-cell difference favoring maraviroc at the twice-a-day regimen.

But of course the baseline viral loads do influence response. What they saw was for patients whose viral load baseline was below 100,000, 60% got to less than 50. In contrast, if somebody had a viral load over 100,000, only 30% did, once again illustrating how important the entire regimen is going to be for these very high viral load patients, and certainly the reason why it's so good that we have so many active drugs in our field at this point in time. For example, they illustrated this point by looking at the enfuvirtide data. When patients used enfuvirtide for the first time in this study, 60% to 64% of them got their viral loads to less than 50. In contrast, if someone didn't used enfuvirtide or recycled it, then only 30% got to less than 50, illustrating once again just how important it is to have an active regimen, not just single active drugs.

The great news for maraviroc continues to be not just how well it works, but how well it's tolerated. There doesn't seem to be all that much difference from placebo overall, in terms of adverse events, in terms of subjective toxicities. There [was a slightly larger percentage] of people who had some upper respiratory tract infections and a cough, but certainly not more than that, and [it was only a difference of a few percent]. There was no evidence of an increase of AIDS-related events, and certainly very little evidence of differences in LFT increases. There were a couple of grade 3 ALT [alanine transaminase] increases on maraviroc versus the control, although we were reassured by the presenter, Jay Lalezari, that in MOTIVATE 2, the companion trial, that the results were in fact the opposite, that there weren't increases in transaminases.

The last point, which was the focus of not just one presentation, but in fact two, is that an important concern for this class is not just response rates, but what are the implications of not responding.

First of all, if you're using an R5 inhibitor, what is going to happen in patients who don't respond? We've learned for some time now that what we're going to see typically are viruses that are dual/mixed or X4 tropic.^10,11 Indeed, in this study in the maraviroc twice-a-day arm, about two thirds of the patients who rebounded did so with a dual/mixed-tropic virus population.¹² Again, recall that everybody in this study had to have CCR5-tropic virus to be eligible at baseline, but at the time of failure, two thirds of the patients emerged with dual/mixed virus. We understand this to be essentially because at baseline people have dual/mixed virus. Even in patients who are screened, they may have some small amount of dual/mixed virus that is hidden among their viral population. If you use an R5 inhibitor and it's not fully suppressant, it's the dual/mixed virus that has the advantage.

Then the question is, so what, does it matter? The good news so far is that at least at the time of failure, people who rebound with R5 populations seem to have a pretty good CD4 response as a result of their participation in the study. In one analysis, they have about a 100-cell bump on maraviroc, even at the time of virologic failure. If somebody emerged with dual/mixed virus, they still have a pretty good CD4 response, though the CD4 response in that case was not different from placebo. Nevertheless they certainly didn't have any evidence of harm as a result of rebounding with a dual/mixed population. It was just no better than the placebo group, in that particular case.

That's been a concern for a while: If we unmask the dual/mixed virus, will we then cause some destruction of CD4s? So far, the news is reassuring in that we don't; we may not get the benefit of R5s, but we certainly don't seem to be creating harm, at least in the short-term of these studies.

So what we've learned so far is that maraviroc looks like an important drug. In these treatment-experienced populations, it provides substantial activity in an R5-screened population. Once again, we have another opportunity to take a pretty easy drug to work with and create a good regimen around it.

So we're getting reassuring news of some of the drugs that are in phase 2/3, and some good news that there continues to be a pipeline behind these drugs.

This is part two of a three-part article. For part one, First-Line Therapy, click here. For part three, HIV/HAART Complications and HIV Drug Resistance, click here.