A Genetic Test to End Angiography?
A Genetic Test to End Angiography?
Medscape: First, I would like to congratulate you and your colleagues for having had this paper selected by Time magazine as one of the top 10 medical breakthroughs in 2010.
Dr. Topol: Yes, thank you. We were all quite pleased about that. They got it wrong, though; they said the test was for heart attack. It is not for heart attack; obviously, it is for coronary disease. That was unfortunate, but at least they recognized that it was a big step forward, because we have never before had a blood test that could help in any way.
Medscape: Is the Corus CAD test still the only one of its kind available in the United States?
Dr. Topol: Yes, it is the only one. I was skeptical at first that it would really pan out, but it actually performed far better than even I expected.
Medscape: You are referring to validation of the test by the PREDICT results?
Dr. Topol: Yes. We put it to the "acid test," a very large study in over 500 patients, and we had the angiograms for all the participants. The only thing we did not know was what their coronary disease was going to be when the blood test was done. It was all done on a blinded basis, and the gene expression work was done independently of the angiogram, and when they were matched up, they matched up quite well in their ability to predict whether or not there was any significant narrowing in the main coronary arteries. I think it is a very useful test in certain circumstances.
There are a couple of points that frame this. If a patient comes in and says they have some type of chest or neck discomfort with exercise, these symptoms may be angina. Typically what such a patient gets in the United States is a nuclear scan in conjunction with an exercise stress test. About 9 million of these are done each year, and each nuclear scan exposes the patient to 41 mSv, which is roughly equivalent to the adult effective dose from 2000 chest x-rays, which is a lot of radiation. In addition, the results from many of these scans are either ambiguous or they prove to be false-positive; that is, they give a patient an angiogram, but then the patient does not have significant disease. Over 2 million patients have an angiogram each year in the United States. It is one of the most common, if not the most common, medical procedure in US hospitals. A critical study published in March 2010 looked at data from almost 400,000 patients at 663 hospitals who had an angiogram.Approximately 84% had had noninvasive testing such as resting electrocardiography, echocardiography, CT, or an exercise nuclear scan before angiography, but interestingly only 38% were found to have significant obstructive CAD. That meant that 62% did not even have significant coronary disease, but they received a double exposure of the nuclear scan, equivalent to 2000 chest x-rays, and then they got an angiogram, another 400 chest x-rays, so they got a huge dose of radiation when they did not even have disease.
Medscape: There is a lot of discussion nowadays about studying "real world" patients in clinical trials. Were the patients "real world" in your validation study? Were they the types of people a physician would see day to day?
Dr. Topol: Yes, these were people who were having a regular angiogram at 39 centers. There was nothing special about them, except we took out those with diabetes because we thought that it was much more difficult to say that they would not have coronary disease.
Medscape: But would these patients have the typical risk factors for CAD, like hypertension and high cholesterol levels?
Dr. Topol: Yes, those are the type of patients who usually wind up with the angiograms.
Medscape: So you already know that they are at high risk?
Dr. Topol: The threshold for doing a heart catheterization is very low. Much of the reason for this is because of incorrect or ambiguous results. So, if I see a patient who may have angina and I do a nuclear scan and it shows uncertain ischemia, typically I would not want to miss it, so I will do the angiogram, and then, of course, it is negative. A lot of this goes on, resulting in hundreds of thousands of patients each year getting angiograms unnecessarily.
Medscape: So this is one of the main reasons you have welcomed this simple blood test.
Dr. Topol: The test is skewed to not miss anything, so it errs on the side of "do the angiogram because the score is high." But if the score is low, based on the gene expression, then it is very unlikely that the patient is going to have any significant coronary disease. So, in my view, if a patient is involved with the decision process -- or the insurance company, because this blood test could be done pretty inexpensively compared with an angiogram or even a nuclear scan -- many patients would rather have a blood test than go through the alternative testing or procedures. And when it is ambiguous, when it is unsettled, this helps put in another key data point to resolve the situation or shed some light on whether you should do the angiogram or not. Currently, it is as if the patient gets on a train and there are no stops; the train keeps going on right into stenting. They have symptoms that are not clear; they have a scan and it is also unclear; they have an angiogram and they find a narrowing that may be completely innocent or ornamental; then they have a stent, or many stents even, which do not even have anything to do with the patient's initial symptoms. This is American cardiology today.
Medscape: We have all heard about unnecessary stenting lately.
Dr. Topol: There is a lot of good stenting, and there is some bad stenting too. Some of it is unnecessary, and some of it is propelled by current practice.
Medscape: So at what point could the gene expression test be applied in these patients?
Dr. Topol: It could be applied at many different points. It could be applied without even doing a nuclear test, or it could be done after the nuclear test, to resolve a question. This is the point where there are still a lot of choices, and this is one of the new expanded choices to factor in. You can do Framingham score, which is just a very basic epidemiologic probability method that does not take into account specific patient symptoms and the specific patient; it is very primitive. Much more useful is the individualized assessment, and this genetic test can add to that. I believe the main function of the new test that we validated -- and there will be more studies of it over time -- will be to get rid of some unnecessary angiograms. That is the key. It is built for that, and I think it is going to emerge as a nice tool, particularly if it becomes quite inexpensive over time. Then I think it would be useful.
Medscape: The PREDICT report did not provide many details about the genes and how the algorithm itself was developed, but I understand more data will be published soon?
Dr. Topol: The algorithm is going to be published soon in another journal. More details about the genes in the test will be published in that paper.
Medscape: How were these genes selected for the algorithm? Were you involved in those studies?
Dr. Topol: I was not involved in that part. We just did the major validation of the test. It was done in the same way as other gene expression tests, similar to the approach that Genomic Health (Redwood City, California) used for their Oncotype DX® breast cancer assay, which has been validated for its ability to predict the likelihood of chemotherapy benefit and recurrence in early-stage breast cancer, and also the approach by XDx (Brisbane, California) for AlloMap®, which is used to predict low probability of acute cellular rejection in heart transplant recipients. Basically, you start out with a whole genome, and you look for signals, transcripts that are hyper- or hypo-expressed, and then you whittle it down to what are the 15, 20, 30, or whatever the number is that drives the difference in the people who have coronary disease in this case vs those who do not. There was a precedent study done in hundreds of patients at a single site, Duke University (Durham, North Carolina), where the investigators started with an unbiased approach, let the genome talk, and found out which of the transcripts are correlated with coronary disease; then computation was used to find the minimum number of genes. In this case it was 23 genes that drive the signature of coronary disease through white blood cell gene expression. But it is unbiased to get to that point, and then it is basically a genome-wide approach of expression for white cells to get down to the details of the biostatistical determinants of the signature.
Medscape: Are the genes expressed in the algorithm all established in association with coronary disease? In the editorial it was suggested that for at least one of them, there was insufficient evidence for its inclusion.
Dr. Topol: There is always some controversy. Because of that, I am not as big a fan of gene expression as I am about sequencing. Gene expression is a dynamic process. It is time-related; if you do it again a few hours later, you get different results, so it is a tricky area to work in. However, it has clearly changed the way that breast cancer is treated.
Medscape: Gene expression profiling has been well validated in breast cancer, and prospective, randomized comparisons are underway between genomic testing, such as Oncotype DX, and clinical factors in making clinical decisions. So the process in your study is basically the same?
Dr. Topol: It is the exact same process, although breast cancer gene expression is temporally related, as is the gene expression assay for transplant rejection, which is now being used instead of doing biopsies of the heart. So it has turned out better than expected, and it has been very useful. The validation started with breast cancer, but Genomic Health is now doing this with assays for other cancers, including prostate cancer, non-small cell lung cancer, renal cancer, and melanoma, and it will soon be or is already available for colon cancer. I do not know whether 5-10 years from now this will be the "go-to" strategy, but it is making a difference now.
Medscape: Some commentators, including the author of the editorial in the Annals of Internal Medicine, have said that more work needs to be done on the Corus CAD test and implied that it isn't quite ready for prime time yet.
Dr. Topol: The test is ready for use now, and I am so tired of people saying that it needs more work. We have patients today whom we want to help and it can help, absolutely.
Medscape: So you wouldn't agree with the conclusions of the editorial?
Dr. Topol: The editorial was not written by a cardiologist or a clinician, so the author does not know about the things that we have talked about in terms of how to make decisions. I certainly respect the editorialist's views, but they were not clinically anchored.
The Interview
Medscape: First, I would like to congratulate you and your colleagues for having had this paper selected by Time magazine as one of the top 10 medical breakthroughs in 2010.
Dr. Topol: Yes, thank you. We were all quite pleased about that. They got it wrong, though; they said the test was for heart attack. It is not for heart attack; obviously, it is for coronary disease. That was unfortunate, but at least they recognized that it was a big step forward, because we have never before had a blood test that could help in any way.
Medscape: Is the Corus CAD test still the only one of its kind available in the United States?
Dr. Topol: Yes, it is the only one. I was skeptical at first that it would really pan out, but it actually performed far better than even I expected.
Medscape: You are referring to validation of the test by the PREDICT results?
Dr. Topol: Yes. We put it to the "acid test," a very large study in over 500 patients, and we had the angiograms for all the participants. The only thing we did not know was what their coronary disease was going to be when the blood test was done. It was all done on a blinded basis, and the gene expression work was done independently of the angiogram, and when they were matched up, they matched up quite well in their ability to predict whether or not there was any significant narrowing in the main coronary arteries. I think it is a very useful test in certain circumstances.
There are a couple of points that frame this. If a patient comes in and says they have some type of chest or neck discomfort with exercise, these symptoms may be angina. Typically what such a patient gets in the United States is a nuclear scan in conjunction with an exercise stress test. About 9 million of these are done each year, and each nuclear scan exposes the patient to 41 mSv, which is roughly equivalent to the adult effective dose from 2000 chest x-rays, which is a lot of radiation. In addition, the results from many of these scans are either ambiguous or they prove to be false-positive; that is, they give a patient an angiogram, but then the patient does not have significant disease. Over 2 million patients have an angiogram each year in the United States. It is one of the most common, if not the most common, medical procedure in US hospitals. A critical study published in March 2010 looked at data from almost 400,000 patients at 663 hospitals who had an angiogram.Approximately 84% had had noninvasive testing such as resting electrocardiography, echocardiography, CT, or an exercise nuclear scan before angiography, but interestingly only 38% were found to have significant obstructive CAD. That meant that 62% did not even have significant coronary disease, but they received a double exposure of the nuclear scan, equivalent to 2000 chest x-rays, and then they got an angiogram, another 400 chest x-rays, so they got a huge dose of radiation when they did not even have disease.
Medscape: There is a lot of discussion nowadays about studying "real world" patients in clinical trials. Were the patients "real world" in your validation study? Were they the types of people a physician would see day to day?
Dr. Topol: Yes, these were people who were having a regular angiogram at 39 centers. There was nothing special about them, except we took out those with diabetes because we thought that it was much more difficult to say that they would not have coronary disease.
Medscape: But would these patients have the typical risk factors for CAD, like hypertension and high cholesterol levels?
Dr. Topol: Yes, those are the type of patients who usually wind up with the angiograms.
Medscape: So you already know that they are at high risk?
Dr. Topol: The threshold for doing a heart catheterization is very low. Much of the reason for this is because of incorrect or ambiguous results. So, if I see a patient who may have angina and I do a nuclear scan and it shows uncertain ischemia, typically I would not want to miss it, so I will do the angiogram, and then, of course, it is negative. A lot of this goes on, resulting in hundreds of thousands of patients each year getting angiograms unnecessarily.
Medscape: So this is one of the main reasons you have welcomed this simple blood test.
Dr. Topol: The test is skewed to not miss anything, so it errs on the side of "do the angiogram because the score is high." But if the score is low, based on the gene expression, then it is very unlikely that the patient is going to have any significant coronary disease. So, in my view, if a patient is involved with the decision process -- or the insurance company, because this blood test could be done pretty inexpensively compared with an angiogram or even a nuclear scan -- many patients would rather have a blood test than go through the alternative testing or procedures. And when it is ambiguous, when it is unsettled, this helps put in another key data point to resolve the situation or shed some light on whether you should do the angiogram or not. Currently, it is as if the patient gets on a train and there are no stops; the train keeps going on right into stenting. They have symptoms that are not clear; they have a scan and it is also unclear; they have an angiogram and they find a narrowing that may be completely innocent or ornamental; then they have a stent, or many stents even, which do not even have anything to do with the patient's initial symptoms. This is American cardiology today.
Medscape: We have all heard about unnecessary stenting lately.
Dr. Topol: There is a lot of good stenting, and there is some bad stenting too. Some of it is unnecessary, and some of it is propelled by current practice.
Medscape: So at what point could the gene expression test be applied in these patients?
Dr. Topol: It could be applied at many different points. It could be applied without even doing a nuclear test, or it could be done after the nuclear test, to resolve a question. This is the point where there are still a lot of choices, and this is one of the new expanded choices to factor in. You can do Framingham score, which is just a very basic epidemiologic probability method that does not take into account specific patient symptoms and the specific patient; it is very primitive. Much more useful is the individualized assessment, and this genetic test can add to that. I believe the main function of the new test that we validated -- and there will be more studies of it over time -- will be to get rid of some unnecessary angiograms. That is the key. It is built for that, and I think it is going to emerge as a nice tool, particularly if it becomes quite inexpensive over time. Then I think it would be useful.
Medscape: The PREDICT report did not provide many details about the genes and how the algorithm itself was developed, but I understand more data will be published soon?
Dr. Topol: The algorithm is going to be published soon in another journal. More details about the genes in the test will be published in that paper.
Medscape: How were these genes selected for the algorithm? Were you involved in those studies?
Dr. Topol: I was not involved in that part. We just did the major validation of the test. It was done in the same way as other gene expression tests, similar to the approach that Genomic Health (Redwood City, California) used for their Oncotype DX® breast cancer assay, which has been validated for its ability to predict the likelihood of chemotherapy benefit and recurrence in early-stage breast cancer, and also the approach by XDx (Brisbane, California) for AlloMap®, which is used to predict low probability of acute cellular rejection in heart transplant recipients. Basically, you start out with a whole genome, and you look for signals, transcripts that are hyper- or hypo-expressed, and then you whittle it down to what are the 15, 20, 30, or whatever the number is that drives the difference in the people who have coronary disease in this case vs those who do not. There was a precedent study done in hundreds of patients at a single site, Duke University (Durham, North Carolina), where the investigators started with an unbiased approach, let the genome talk, and found out which of the transcripts are correlated with coronary disease; then computation was used to find the minimum number of genes. In this case it was 23 genes that drive the signature of coronary disease through white blood cell gene expression. But it is unbiased to get to that point, and then it is basically a genome-wide approach of expression for white cells to get down to the details of the biostatistical determinants of the signature.
Medscape: Are the genes expressed in the algorithm all established in association with coronary disease? In the editorial it was suggested that for at least one of them, there was insufficient evidence for its inclusion.
Dr. Topol: There is always some controversy. Because of that, I am not as big a fan of gene expression as I am about sequencing. Gene expression is a dynamic process. It is time-related; if you do it again a few hours later, you get different results, so it is a tricky area to work in. However, it has clearly changed the way that breast cancer is treated.
Medscape: Gene expression profiling has been well validated in breast cancer, and prospective, randomized comparisons are underway between genomic testing, such as Oncotype DX, and clinical factors in making clinical decisions. So the process in your study is basically the same?
Dr. Topol: It is the exact same process, although breast cancer gene expression is temporally related, as is the gene expression assay for transplant rejection, which is now being used instead of doing biopsies of the heart. So it has turned out better than expected, and it has been very useful. The validation started with breast cancer, but Genomic Health is now doing this with assays for other cancers, including prostate cancer, non-small cell lung cancer, renal cancer, and melanoma, and it will soon be or is already available for colon cancer. I do not know whether 5-10 years from now this will be the "go-to" strategy, but it is making a difference now.
Medscape: Some commentators, including the author of the editorial in the Annals of Internal Medicine, have said that more work needs to be done on the Corus CAD test and implied that it isn't quite ready for prime time yet.
Dr. Topol: The test is ready for use now, and I am so tired of people saying that it needs more work. We have patients today whom we want to help and it can help, absolutely.
Medscape: So you wouldn't agree with the conclusions of the editorial?
Dr. Topol: The editorial was not written by a cardiologist or a clinician, so the author does not know about the things that we have talked about in terms of how to make decisions. I certainly respect the editorialist's views, but they were not clinically anchored.
