What is the level of phenotypic resistance that is clinically significant for individual drugs?
Moderator: OK, why don’t we move on to the next question. Which is one that’s very good use for the clinicians and Mark, you obviously have some opinions here, and I’ll try my best as a Canadian accent for Brian here. But the question is, what is the level of phenotypic resistance that’s clinically significant? Obviously the answers will vary from drug to drug, and may also be compounded by other factors, but this is a recurring question. Is there data to support some rules of thumb for individual agents, say, d4T and ddI where a low cut-off is important. And in RTI’s where there are high cutoffs that are important, etc.
Let me just read Brian’s comments here. Apologies to the Canadians here. The relevant levels of phenotypic resistance will vary according to the drug or drug categories. For 3TC and the NNRTI’s, it’s very clear that only high-grade phenotypic resistance is really clinically relevant. Greater than 20-50 fold or so.
Charles Boucher: How did you know? How did you know that because we don’t have viruses with 10-fold resistance to 3TC?
Moderator: That’s correct, you’ve got a good point there, Charles. The only unanswered question is whether they may be a graduation of resistance for efavirenz, where the 103N mutation alone may not be sufficient to predict clinical failure. Although anecdotal data suggests the presence of K103N without mutations at 100, 108, or 225, may be sufficient to make the drug fail. For the NNRTI’s, the low level resistance, 4-10 fold increases is not of relevance. I believe this can be now said with good confidence.
Charles Boucher: But I think it’s all ridiculous, because you’re referring to, it’s always in a combination setting. And I’m sure if you have four-fold or ten-fold increased for nonnucleosides, actually we know, we presumably know the selects for these…???
Moderator: Right, but I guess what he’s saying is he would never anticipate anybody using these in monotherapy.
Charles Boucher: OK, but then you’re talking about combination? So that’s one important variable. The other variable is to ??? in the patient.
Moderator: Right, right. OK.
Charles Boucher: I think to make statements, this level, this level may be important. It’s very difficult to do.
Moderator: Right. Charles, I can’t defend him.
David Katzenstein: I think that you’re killing the messenger here. I think the key question that Charles is bringing up is how are we applying fold resistance and cutoff, certainly to different drugs? And the answer is that each drug is a completely different genotype/phenotype clinical response algorithm so the use of microbiology approach, which is to say, we can statistically demonstrate greater than 2.5 fold resistance as being something that we can detect, therefore, we’ll call that resistance, may or may not have meaning for certain drugs.
Moderator: Well, I mean, I would go one step further and I know this comes back to something that Charles likes to talk about as well. We’re almost never using these drugs as monotherapy and you’re really talking about pools of viruses with varying degrees of resistance to one drug or the other. My problem, my fundamental problem with the phenotypic assays, is that they’re talking about isolated drugs alone.
Charles Boucher: Yes, so, they talk about isolated drugs. So it’s very difficult. And in general, I think what you do, is you don’t do with absolute levels. You get a resistance profile and then you select the most favorable groups, right? From the available drugs you have?
Moderator: Well, you do, but wouldn’t you rather know what a combination did to that virus isolate? I mean, wouldn’t you really rather know what d4T/3TC and indinavir did, rather than what d4T did alone?
And I think that’s really the issue here. It isn’t really what does one drug does by itself. It’s what do all three drugs together do?
Charles Boucher: And you’re question is really relating in the end to the relationship between drug levels. And phenotype, right? And we don’t have any data, we may have some data from the new, but even that is not rock solid.
Jonathan Schapiro: And that’s a good example, Charles, because that again is in a statistic clinical scenario. You can give some information when you give in this combination, in these patients, with this drug, then these cutoffs appear to be relevant for 24-48 week data. That’s what you can say. That’s a far cry.
Charles Boucher : Even in patients who then got the new nucleosides and…so that’s how far it goes.
Moderator: But let’s bring it back to the clinicians. In other words, I think we even when we made our charts up, we said, we think that four-fold resistance of d4T and ddI is significant because you almost rarely, if ever, see an isolate that has more than ten-fold resistance, correct?
Man: You don’t even see four-fold.
Moderator: Yes, so if you see four to ten-fold resistance, we think, we equate that with high level resistance. On the other hand, you almost never see resistance to, you know, to take Charles’ point, you almost never see 100-fold resistance of 3TC because it’s almost always 184 and it’s 10,000 fold resistance to 3TC. So, you know, in that particular case you’d say, well, we really like to see high level resistance. And I guess that’s the question, that Brian is trying to raise here, is ‘Can we say to clinicians, look, you know, low level resistance is really important for d4T/ddI, high level resistance is really important for the NNRTI’s and 3TC. And then everything else is a model, or somewhere in the middle, in terms of protease and others. Is that a fair statement?
Jonathan Schapiro: Well, I think for protease we could elaborate a little bit. We saw some interesting data going back to the 90 mutation that we had discussed before. I think, dependent again, as Charles mentioned, on the drug levels, but with the saquinavir, when given as a single PI, be it as Invirase or Fortovase, 90 mutation is bad news. And we saw a large data set showing that it confers somewhere around three-fold change if it’s by itself.
Now I think that has been shown to be a very poor marker when using saquinavir by itself, and we don’t really have good data about when we augment it with ritonavir, but if the assays give us three-fold change from 90 to saquinavir, I would caution, that for many of the protease inhibitors, we may also find that levels below four-fold have a lot of clinical significance. And I think a lot of times when we’re evaluating new protease’s, we see presentations from the states showing, oh, it was only low level resistance so you can use it, and they’re talking about anything between four and ten fold…
Moderator: Right, but if you look at lopinavir, for example, it’s eight-fold resistance to most of them.
Jonathan Schapiro: That’s right, and I think there again, we have to be very cautious with these cutoffs. And probably we’re going about it, you know, the wrong way. I remember which resistance, I think Charles, you mentioned it, but the question was, if we should go the other way around? Look at what folds we had when the patients failed, and then determine that as the level.
I think for protease inhibitors, to arbitrarily use low, medium and high, has no relevance, and the thing is going to be for amprenavir, where, do we really know what it means to have 2 indinavir mutations and we say, well, it was less than four-fold cutoff for amprenavir, so you can probably use it. That’s based, I think, on nothing.
Moderator: OK, now we’re going to get into sheer opinion. As if any of what we previously said was on fact.
Charles Boucher: Everything we said before was all rocket science!
