What do each of you think of the relative value of genotypic/phenotypic assays with respect to assay sensitivity, turn around time, expense, and ease of interpretation for a clinician?
Moderator: And actually, what I’m going to do is I’m going to truncate the last question to basically just ask you to give your view on the following: What do each of you individually, think of a relative value of genotypic/phenotypic assays are, with respect to assay sensitivity, turn around time, expense, and ease of interpretation for a clinician? Keeping in mind that this is a clinician who’s going to get these reports back. What do you think each of these are, relative to each other?
Charles Boucher: Can I start?
Man: Yes, absolutely, Charles.
Charles Boucher: OK, I think if you use the assays to basically rule out drugs, if that’s your objective, to rule out drugs we cannot count on to work, the assays have the greatest sensitivity because they can identify single mutations, if there’s low fold resistance which is difficult to take in the phenotype, and they can even detect mixtures. And my feeling is, and that’s really corroborated by all the trials where the genotypic trials show advantage when being used, where phenotypic data is really very, very poor. So I say the genotype over phenotype. As long as you have a good data set.
Moderator: OK. Bob what do you think? You’re fairly heavily involved in this.
Bob Shafer: I personally think that the phenotype is the secondary test. In fact, I think a phenotype without a genotype is dangerous, because there’s so many interactions with the different mutations, and this came up a lot at times, for instance, you could have the K103N, which we know is a very poor prognostic factor from the literature, and yet it can only be, it may be associated with low levels of phenotypic resistance. If they’re antagonistic mutations. The same thing occurs with 184V and T215Y and L90M and A82A, possibly.
So the ideas that the whole literature of HIV resistance is based on correlation’s between mutations and generally people not doing well. Not based on phenotypic, because every phenotypic assay is a little bit different. The whole literature from early 90’s on, is based on known people with the T215Y mutation don’t do well with AZT. Or people with K103N, don’t do well with non-nucleosides. So that’s why, because the whole literature is based on these mutations, I think you have to have the mutations as the primary result. I think the phenotype should be for those phenotypes that are not easy to interpret, which probably are in the range of about 10% of genotype results.
Now, in terms of what’s easy for the clinicians to deal with, I think the phenotype has that big advantage, that the reports just say, sensitive, intermediate, resistance. I mean, they are a much more appealing type of report, but I think they may, although they may be appealing, they may be giving misleading information frequently. So the genotype, unfortunately, the way the reports come across, and that’s what I’m trying to work on, is that they come across as being very confusing.
Man: But Bob, I agree…
Charles Boucher: I’m so confused. Because with phenotype, people transfer genotype into phenotype, but there’s trial showing that knowing phenotype has a clinical effect. Especially if not, 2, 3, 4, we chose that good ___ with genotype works. So, I’m getting confused here.
Man: But Bob, would you say it’s an advantage of the phenotype that the result is easy to understand because it’s misleading. That’s bad, I prefer to…
Bob Shafer: No, its…
Man: That’s a disadvantage, I think that might be the reason that in some of the studies it does poorly because sometimes it’s better to be confused than to be misled.
Charles Boucher: That was Bob’s point.
Bob Shafer: The idea is though, the genotype, and this is what I’m really trying to do with my work, the genotype is rich and has a lot of information, and often the information consists of mutations which cancel each other out phenotypically, but you want to know that they’re there, because they are very relevant as to how the patient will do when treated with a new regimen. So the genotype has a lot of information and I think the people who are involved in interpreting the genotype and providing the results to clinicians have to, that includes me, have to do a much better job at using all the data the genotype tells you, but also of presenting to clinicians in ways that are more useful than we’ve been doing in the past.
David Katzenstein: Wait I want to try one paradigm here which is something both Jonathan and Bob have mentioned. Which is that ultimately we’re all concerned, and particularly those of us who occasionally see patients are concerned about "how patients will do". And Bob was talking about an end point. I think Jonathan has been instrumental in designing many of the best trials and how they will do.
But in fact, our clinical trial data, on both phenotype and genotype, and the question Charles was posing about has phenotype really been demonstrated to improve patient outcomes, are really limited by in fact, what those outcomes are and how well and how long we’re able to…
Charles Boucher: But even the short outcome. If you take 53002, an abstract in San Francisco which you will never see in presentations…
Man: I just saw it presented actually, I thought, at the ACTG meeting.
Charles Boucher: Good. Did you see the results? It’s frightening.
Man: Phenotype did not do as well as genotype, it was my recollection.
Charles Boucher: No, it was not genotype, but phenotype compared to placebo. And 75% of the patients were back to baseline at 8 weeks or so. Only with the 53001, it also was not very impressive. Then we have the French study. Give me one study which shows even a hunch that phenotype is working.
Jonathan Schapiro: I tell you what I think, I think a key to answering that question is that there is a difference between the results that an assay gives, and you can measure that with sensitivity and specificity, reliability, accuracy. And the interpretation of those results. I think that we’re seeing that the technology for both genotyping and phenotyping is greatly improving. I think we’re just over the last 5-6 years, the assays have gotten much, much better and I think there’s a lot of data to show that the results of a genotypic test, which is the generation of a list of the mutations, and the accuracy of a phenotypic test, which is the generation of fold changes, are very accurate. The problem is, we don’t know to interpret either of these correctly to give the clinical benefit that David is talking about.
So my two bits would be that these great technologic advances we have, and all the companies that do phenotyping or genotyping have done a lot and they’re much more accurate. And we’ll continue to follow this and hopefully, I think, the one limitation might be that the cutoffs for the fold changes have to go a little bit lower. But in general, when people compare the different genotypic technologies or different phenotypic technologies, we get very nice, very nice results. The problem is we do not know how to interpret them.
That goes back to what Bob said, what David is saying is that the limitation of these assays and the fact that it will show us how wonderfully they’re all correlating the tech mutations are nice, but they’re not the main point. And until we know what these fold changes mean, and these mutations mean in clinical outcomes, we’ll continue to have poor results.
And I think the biggest limitation is, and I think this is probably why phenotypic assays do poorly, I think this is something we teach our medical students. It’s good for a doctor to be confused and worried and run around and call everyone they know and try to get a good solution. That is a good situation. The worst thing we want is a physician to think they have an answer when they don’t and that’s why I think that it’s a very difficult to use a phenotypic assay that does not have clinical relevance, and that’s why I think some of the problems have been that people have been misled into thinking they’re using drugs which still work, when they haven’t. And that’s why I think the interpretation is the problem, not the assays themselves.
Moderator: OK, I think we’ve certainly used up plenty of time today. I thank all of you for participating and we’ll hear from you again on another call.
