Director of the Nucleotide Polymorphism Laboratory
Name: Dennis O'Kane
Position: Director of the Nucleotide Polymorphism Laboratory; Associate professor of laboratory medicine and pathology at the Mayo Clinic in Rochester, Minn.
Education: PhD in Botany cellular and developmental biology, State University of New York at Stony Brook
The American College of Medical Genetics invited Dennis O'Kane to speak at its conference last week about clinical assay validation. Researchers have been concerned recently about the equivalence of platforms and assays, and efforts such as the US National Institute of Standards and Technology's Microarray Quality Control project have been trying to clear up those problems in order to allow new genomic and other molecular technologies to get to the clinic.
Pharmacogenomics Reporter spoke to O'Kane recently to learn about his perspective on assay validation and the barriers to pharmacogenomics in the clinic.
What criteria are required to establish a genomic test for clinical use?
First off, are you measuring what you think you're measuring? How do you verify that what you're measuring is correct?
You need to have an independent measure. So, if you set up, let's say, a genotyping assay where you're looking at two alleles how do you know that the results of what you're testing, let's say it's homozygous wild type how do you know that's correct? You have to go in and verify by an independent, acceptable methodology.
What we've found over the years is that some people will do this, but they'll use the same primers in both applications, so it's merely a measure of concordance between different platforms, it's not an independent measure of whether the result is correct or not. So, one of our requirements when we develop an assay here, internally, is that we have independent primers available so that we can do an independent verification of the sequence.
That's one thing. You have to be able to go in and, in our case, demonstrate that you get the same result multiple times using controls, for example. So we'll set up, say, five different controls that contain different polymorphisms, and we'll do five independent amplifications and testings and make certain that everything is concordant. That is one of our basic minimum requirements that you have consistency in both the results and the robustness of the test over time.
Are microarrays suited for clinical diagnostic use at all?
It would depend upon the use. If you're saying they're not useful for genotyping, I think individuals who've used the AmpliChip, for example, might argue against that. It is useful in that format.
If, on the other hand, you're talking about detecting deletions, or you're talking about gene expression, then I'd say there's a great deal more concern both from lack of reproducibility between platforms and also in some basic concepts of how one [assesses the quality of] a microarray that may have 30,000 genes on it.
Now, each of these issues is probably being addressed by different groups. There's the [US Food and Drug Administration's] MAQC [MicroArray Quality Control] project, which is looking at different microarray platforms and the data on that should be out sometime shortly. Then there is a group that is developing guidelines for the quality-control aspects of microarrays.
So, at the present time, I think the FDA's stance on this is that they're going to be fairly rigorous in reviewing an application for anything that involves gene expression, for example, and it probably would not get through at this time, because not everything is in place in terms of the [quality control] at this time; not everything is in place in terms of whether you're measuring what you think you're measuring.
And so, there you probably have a legitimate case for saying microarrays would not be suited for the clinical laboratory, but it's at this time. And it also has to do with the scale. If we're talking about whole-genome assays, where you're looking at all the genes on a microarray at the same time, then it's probably true. I don't know how I would actually analyze the data in sufficient time to be able to use it in the lab.
On the other hand, if you're looking at 60 different genes, which all have quality control, and this can be performed in a two- or three-day period and analyzed in the next day, then it very well may be that there will be suitable microarray applications in the very near future.
Roche has a project in Europe looking at leukemias on microarrays. I don't know what the final outcomes of that are I'm not sure if they're at that stage but if you're looking at 100 genes or [fewer], it may be quite useful for looking at leukemias quite rapidly. These can take some period of time to actually go through a differential diagnosis using classical methods, and it may be possible in a three-day period to arrive at what the leukemia really is, how it is subdivided molecularly, and how you would go about best treating the patient based upon the individual expression of genes or non-expression of genes, in these particular cases.
So, I think it's really up in the air, but I would agree that at the current time, applications other than in genotyping would be pretty sparse, I would think.
Are technical issues a major barrier to getting genomic and related tests into the clinic for pharmacogenomic treatments?
No. Technical aspects are the easy part that's why I can do this.
The hard part [is] physician education. Physicians do not understand pharmacogenetics: one, what it is; and two, how to use it. And so, you've got a full generation-plus of new physicians to train in this area, plus you've got all the previous generations who are in practice who have to be brought up to a certain level of professional understanding on this.
That's a major barrier. If you don't understand how to use the test, you're not going to order it. That's number one.
Number two we're really just scratching the surface on pharmacogenomics. The genotyping part is easy I think it is. Correlating it to a phenotype is difficult, and that's where I think a lot of work is going to have to be done in the next few years, to really further develop the phenotypes that go with a genotype. In particular, there are some drugs that may be metabolized very poorly by one genotype say, for CYP450 2D6 and yet the same genotype may metabolize other drugs on an extensive-metabolizer basis.
So, an individual, given a certain genotype, may be an extensive metabolizer for some drugs, but a poor metabolizer for other drugs. And that hasn't been sorted out, as it turns out. This is something that's been emerging in the last couple of years, and so that's going to have to be dealt with. How do you deal with these variable phenotypes given a constant genotype for an individual?
So that's number two we need further phenotype development and correlation with the genotypes.
Number three we're talking about pharmacogenetics as if it's a monogenic trait, and that's not the case for many drugs. Many different genes may be involved in producing enzymes that metabolize drugs in different ways, and each of these may be variable in terms of their activities in terms of their expression, as well as in terms of having polymorphisms that inactivate or partially inactivate the gene product. So that needs to be worked out in some cases to refine our interpretations.
It's not terribly surprising, in retrospect, I guess, that pharmacogenetics has just not taken off as rapidly as everyone would have liked. But I think each of these areas can be worked upon and we can move forward on that. I'd say the key issue here what we really have a good grip on is how to educate physicians in being able to understand how to use the tests in their practice. That, I think is the major problem.
Following on that, you've got reimbursement issues. Right now, pharmacogenetics is not getting reimbursed by third-party payors at least as far as I'm aware at really significant rates. And so, I think physicians would have to be really selective in the patients that they want the testing performed on. And it means that some patients who would benefit from testing may not be receiving it because they may not be able to pay for it. That's what it comes down to. And that's going to be an ongoing discussion for, I'd estimate, the next five to ten years, as to where pharmacogenetic and genetic testing fits in and how we pay for it and try to make this a cost-effective technology that we're actually using in the lab.