What is the status of thymidine-associated mutations with regard to AZT vs. d4T?
Moderator: Right. The question is, ‘What is the status of thymidine-associated mutations, in regard to AZT vs. d4T?’ And I edited that a little to say that, to ask the question of whether these are actually thymidine-associated mutations, or AZT mutations?
All right, keep going here. Do the TAM’s really have similar significance in regard to these two drugs? Should we assume that four-fold resistance to the d4T with TAM’s is equivalent to fifty-fold resistance to AZT? And should d4T no longer be considered a newer effective drug after failure of AZT containing regiments? And vice-versa…what about AZT after d4T? Mark, do you want to go for it first?
Mark Wainberg: Yes, I wrote this question. So I have, sometimes you write questions in a way that suggests you know what the answers might be. Of course, we don’t know what the answers are in any definitive way to these questions, and I think, again, it’s an example of clinical trials potentially being useful at some point to help us get the answers.
But I think there is certainly accumulating evidence that the TAM’s are, in fact, common to both of these drugs, and therefore, it would be my bias that d4T should not be considered a new drug after first round therapy with zidovudine and not either in the aftermath of zidovudine treatment failure.
David Katzenstein: Well, I think I’ve actually, to confirm what Mark said, I’ve been doing exactly that in my analysis, which he’s suggesting. These numbers are small, but I wanted to provide for our listeners as well as the group here, at least the strategy we’re using and some of the preliminary results from a small study. And that is, we’ve looked at patients from the 175 study who rolled over to the 302 study, and this is the virology subset of that.
So we have, essentially, 35 patients who had been on long term AZT and this is now talking 3-4 years, who were then switched to d4T. We’ve done a genotypic analysis of their baseline RT ?, and tried to correlate that not with phenotype, but with in fact, the virologic response at 8 weeks, to in fact, the switch to d4T. And this provides very, very clear results, that if you in fact have only a K70R mutation, you get nearly a log response. If you have more than a K70R, in terms of AZT resistance mutations, there’s less than .03 log response. And in fact, a number of patients increased their virus load at 8 weeks.
So I think this is the beginnings of trying to go directly from phenotype to clinical, or I’m sorry, genotype to a "clinical" response. And I think we need more of these kinds of focus studies looking at single drug switches to really isolate activity of specific…again to ??, mutations.
That’s the parenthetical addition to this is, yes, we had and will have phenotype associated with genotypic markers, but I think we all know, that the phenotypic assays, particularly for ddI and d4T, have for many, many years caused a real problem. Because fold resistance changes are in fact, very low. The most highly resistant d4T and ddI viruses, in fact, may have less than four or five-fold resistance in a phenotypic assay. We know that they don’t work under those circumstances, and so, sort of the phenotype, which uses a cutoff, often gives you a false security that in fact a drug is active.
Moderator: Bob, just to bring you up to date. We’re just dealing with the first question with regard to AZT mutations and potential cross-resistance to d4T. I actually have a couple of things to say. I think it’s fair to say from looking at the data presented to date, that the frequency with which d4T selects for mutations previously associated with AZT is probably significantly less than the frequency with which AZT selects for those mutations.
Bob Shafer: I’d like to comment on that. I think we don’t really know the answer to that, and that’s because the data on d4T comes from d4T used in combination with other drugs, and so it’s d4T/ddI or d4T/3TC or d4T as part of heart regiments. I think if you looked at AZT in that setting, AZT in combination, or AZT as part of heart regiments, the rate of AZT resistance is much lower.
So I think, our notion that AZT resistance occurs so commonly, and more than half of the patients getting AZT within let’s say, 6-12 months, that’s from the era when AZT was used as monotherapy. So I think we just, we don’t know for sure. There’s that sense, but you know, our data on the frequency of AZT mutations comes from AZT monotherapy, and there’s really hardly any data on d4T monotherapy. There’s only 6 published sequences of d4T monotherapy.
Moderator: Right, but even in those, Bob, most of those mutations is 215.
David Katzenstein: Well, I think we’ve all been impressed by François Clavel’s data, Charles Boucher has produced a little, and there have been several small studies looking at d4T monotherapy. Where after a period of time, in fact, 215, 41 and some of the AZT-associated mutations begin to appear. The rate at which you develop this kind of resistance may be different, but I would, frankly suggest, the rate at which you develop TAM mutations is probably directly proportional to virus replication that’s ongoing, but potency of drug. So if we say there is a much slower rate of mutations associated with resistance of d4T, we’re in fact acknowledging that there may be, in fact, a lower potency or selective pressure placed on the virus.
Moderator: Does the group think that the start trials will help us in this regard?
Man: In terms of their crossover design and the fact that they’re using different up front regimens, or potentially, what is it, ACTG 384 will help us in this regard?
David Katzenstein: I think Bob and I are convinced that will be a great data set from which to examine AZT/3TC vs. d4T potency.
Bob Shafer: Right, well, we’re going to be looking at, with regards to the nucleosides, we’ll be looking at two things. We’re going to look at d4T/ddI vs. AZT/3TC. I think, which one is more potent may be obscured by the fact that there’s also a comparison of nefinavir vs. efavirenz vs. both. So, but I think we’ll have a fairly good idea if there’s a tremendous difference between the potency of those two, we’ll have an idea.
The next thing is though, that patients that don’t do well on one of those regimens, will be switched to the opposite. So if somebody is, let’s say, began with two nucleosides and nefinavir, and doesn’t do well, they’ll go to a different two nucleosides and efavirenz. We’ll be able to test the activity of one dual-nucleoside combination following the use of the other dual-nucleoside combination. And you know, hopefully, we’ll get some useful information from that.
Moderator: Mark, you mentioned that you really don’t think that d4T should be considered a new drug after AZT. What about the reverse? Do you think AZT should be considered a new drug after d4T or do you think that’s only the case as David suggested, in the absence of AZT mutations?
Jonathan Schapiro: Can I just make a brief comment. Just to get back to the resistance part of it. I think just to highlight something that David which was important of course, is that, going back to our first discussion, we were talking about will there be a virtual clinical outcome as opposed to virtual phenotype. I think what David presented earlier is the first inkling of that type of data where we can go directly from a genotype or phenotype to clinical outcome. I think that’s the kind of tools we would need to refine our interpretation systems. Be able to say there’s data to say that there’s a difference you have a just of 70, or a variety of other AZT associated mutations when you want to give d4T.
And I would also say that the questions that you proposed to Mark, I think here, resistance assays do help us very much. I think that if we see mutations present, these TAM’s present, it does impact what confidence we have if the next regimen will work. So when we’re trying to decide how many drugs, in addition to the d4T or AZT we’re going to use, I think considering how many mutations are present helps us to decide how much bang we’re going to get out of those drugs.
And I think here, probably, is one of the places where resistance assays do help us, cause we don’t have definitive data, but I do think, from what David said and some other data, if you use d4T and you don’t have mutations for AZT, I don’t think there’s convincing data to say that AZT won’t work. Of course, we happened to hear other’s impressions, but I think here, resistance assays do help us.
Moderator: Right, well the, I mean the, attempt to do clinical analysis of this, the data that Glaxo presented, has been really problematic. In their retrospect or reviews of patients clinical charts, in that, there’s no consideration and analysis given for what other variables and other drugs the patient's are actually on.
Jonathan Schapiro: Beyond that, I think we can be pretty sure that there are contaminants there that aren’t taken into consideration because of the nature of those studies, I’d agree.
David Katzenstein: Steve and Jonathan, this is where I think we have to recognize that the only place we’re going to be able to get this kind of information in a way that will be really helpful for planning much more aggressive and hopefully, successful combination therapies, is in fact to go back to stored samples from carefully conducted trials of nucleoside therapy, where in fact the other variable, the NNRTI or PI, is not part of the regimen. Because too often, then, and I think all of us have seen this, we’re trying to infer the resistance or susceptibility of viruses and viral genotypes and even phenotypes based on, in fact, current studies in which patients are receiving 3, 4, or even 5 drugs.
So, I think the, not to advocate for you know, continued nucleoside therapy as a strategy, although there may be evidence that could be a good approach, but I think its, by looking at the days in which we did, and the patients in whom we gave essentially sequential nucleoside monotherapies, that we can learn the most about the clinical virtual phenotype and genotype.
Moderator: What do you make of Brendan Larder and Rich Colonno’s presentation in Sitges using Virco data set? Where they showed that the patients who had four-fold or greater resistance to d4T actually had a very low level of cross-resistance to AZT, and most of those patients actually had either 151 or 69 insertions in a relatively small portion of them actually had either 70, or 215, or 41.
David Katzenstein: Well I think the extraordinary power of the Virco database, I didn’t see their presentation, is the phenotype/genotype relationship. The weakness of their database is that it doesn’t extend to what else the patients were getting and what they actually did.
Bob Shafer: I know, but I think what you’re, the problem was that the AZT resistance mutations don’t cause four-fold resistance to d4T. So in that study, d4T resistance was only associated with 151, 69 insertion.
Man: That’s correct.
Man: And occasionally I think, the 69 without the insertion plus the AZT mutations, I think.
Man: That’s right.
Bob Shafer: So, but most patients who develop virologic rebound on d4T don’t have one of those. I think it’s one of the problems that are unexplained, that these AZT, what we call AZT mutations, which maybe are better to be called pyrophosphorolysis mutations, seem to affect, you know, perhaps all the drugs.
Mark Wainberg: I want to comment, Mark, again, rejoining. I don’t think we can call them pyrophosphorolysis mutations, because the relationship between pyrophosphorolysis and each one of these mutations on its own and in combination with the other mutations in the context that we’re calling RT enzymes is still being worked out. So we don’t have all that information as yet.
Bob Shafer: OK, but I’m not sure that the term ‘AZT mutations’ is, tells the whole story. We may find it affects more of the drugs…
Mark Wainberg: The additional story that needs to be told is still one in which we have to remember that the phenotypes that we’re dealing with, when we talk about either recombinant viruses in which mutations have been introduced, or in fact, even clinical isolates from patients or patient plasma, so far, our analysis has really been restricted to differences between HXB10 or PNL4-3 and in fact the particular isolate we’re looking at, there’s a whole world of another 200 RT codons, all of which can be, both fitness and resistance mutations in the context of very specific changes in AZT or TAM mutations.
So you also are obliged, I think Andrew Leigh Brown has begun to do this in his JID paper, although it’s a difficult one to get through, to look at non-conical sites that may be in fact, associated with failure, clinical failure of RT drugs in patients with multiple conical mutations.
Bob Shafer: But you know, Mark, from your work, and tell me if I’m right, the 41 and 215 cause increased processivity of the enzyme and perhaps affect all the drugs. What do you think, is there evidence that they’re really, from a biochemical basis that those 2 mutations are specific to causing AZT resistance?
Mark Wainberg: Yes, I think you’re absolutely right that those 2 mutations can increase processivity and that is something that could then impact on other drugs as well. If we have a virus that may have a more efficient enzyme as an example. But, I think what you’re getting at is suggesting that there need always be a linkage between increased processivity and increased fitness, and I think that is also information that we are still gathering for certain of the mutations and viruses that we work with. There’s at least one mutation that some of you will recall, it’s the E89 gene mutation, and it is associated with increased processivity but it is certainly not a mutation that increases fitness and that’s, in fact, it seems to diminish fitness, which is probably one of the reasons that E89G is not a mutation that seems at this point in time, at least, to have any kind of clinical significance.
Moderator: OK, I want to just wrap this up with one question for the group, and that is, let’s assume that these mutations are associated with the thymidine analog. Then the converse question is, we’re really all challenged with the fact that we have large numbers of patients who have failed the d4T containing regimens, who don’t have any of these mutations. Correct?
Man: Right, and the question, it comes to the last part of the question. Do we consider d4T after zidovudine to be a new drug? And vice versa.
Moderator: Right, and the flip side to that is, actually for patients that fail AZT containing regimens, then AZT/3TC is today’s modern art, there’s relatively few of those patients who don’t have more than one of these mutations. So there must be some quantitative, if you will, or time dependent difference between these agents.
Bob Shafer: I’m not sure yet, because if you take people who fail AZT/3TC and indinavir, it depends what you mean by fail? If it’s just develop a virus load above 200, the most common mutation, the most common scenario is just simply a mutation of 184 with no AZT mutations.
Moderator: Right, we’re going to get to that.
Bob Shafer: On the other hand, if you have somebody who fails and comes back to baseline, whether its AZT or d4T, I bet you’re going to see a lot of these mutations. We just don’t have the data, but you know…
Man: But the frequency is lower, though, with d4T.
Bob Shafer: When we speak about failing now, failing means just developing detectable virus. But years ago, when it was AZT monotherapy, it was really coming back to baseline, so I think we need to mention, when we say, what do we see when someone fails, I think we have to ask do we mean failing, just developing detectable virus or failing all the way back to baseline, which might take one or two additional years of therapy.
