How Accurate is Cardiac CT? The CATSCAN Study

CATSCAN, funded by Philips Medical Systems, included 238 patients who were clinically referred for non-emergency coronary angiography from June 2004 through March 2005 at 11 participating sites. A total of 187 patients underwent contrast-enhanced MDCT and also had conventional angiography performed one to 14 days after MDCT. The results of these two tests were then compared. Dr. Garcia spoke with CLD about the trial and resulting data, as well as his future plans for studying cardiac disease and CT. How does computed tomography (CT) compare to fluoroscopy? Cardiac CT is an x-ray procedure where an x-ray source rotates around the patient with a detector on the opposite side, and by rotating very quickly and moving the table where the patient lies, you get a three-dimensional picture of the organ that is being imaged. In cardiac CT, the three-dimensional picture is acquired together with the electrocardiogram data. Then the two are put together to reconstruct different phases, so we can generate a three-dimensional picture at any time of the cardiac cycle. Quite often, what we try to do is find out at what part of the cardiac cycle the heart moves the least, in order to capture detailed anatomy such as the coronary arteries. As in the cath lab, you also need to give contrast for a CT study. The contrast is given in an antecubital vein. The total amount of contrast usually given is anywhere from 70 to 100 mL, given over a period of 20 to 30 seconds. You just published the results of a multi-center trial looking at 16-row MDCT in CAD.1 What were you hoping to discover? What we wanted to do with the multi-center trial is determine whether the CT technology, which was 16-detector technology at the time when we started the trial, was accurate enough to detect stenosis greater than 50%, as defined by quantitative coronary angiography. The CT was compared with the cath data, which was the gold standard. Whenever the CT said there was no significant disease, we found that the likelihood was very, very high that the CT was correct. In fact, the negative predictive value of CT was close to 99%, meaning it is a good test to exclude significant coronary disease. The positive predictive value was much lower than was previously reported; in fact, when the CT says there is disease, half of the time, at least, the disease is not significant, according to the cath. So a negative CT is useful data, but a positive CT may not necessarily be true. Our results, however, were obviously based on 16-detector data. We anticipate that if the same type of study were to be conducted today, with the 64-detector scanners, the number of false positives would decrease. These newer systems reduce the number of imaging artifacts because they obtain the images in 6-10 seconds, requiring only a few heart beats. Therefore, it is less likely than the patient will move, breathe or change the heart rhythm. Why do you think there was a high rate of false positives? Well, some of them were true false positives, which means we saw an abnormality and called the findings positive. We may have determined there was a 70% blockage on the CT, for example, and the cath said there was only a 40% blockage. However, the cath provides a higher spatial and temporal resolution than the CT, but the cath doesn’t have the three-dimensional information. We can take projections with CT that cannot be done in the cath lab, and therefore, it’s possible that many of the false positives with CT in fact are false negative cases by cath. For example, if the patient have a very eccentric plaque causing stenosis, or an obstructive lesion at the ostium of a branch, the lesion could be missed or underestimated by cath if the adequate angle and projection is not obtained. That’s one explanation. Another explanation may be that whenever we have very severe calcification in the coronaries, they tend to appear larger than they really are. In those cases, we tend to overestimate the severity of a stenosis. The third possible explanation is that during the trial, if we had a segment that was not well-seen, perhaps some sort of an artifact where the patient was breathing or the heart was beating too fast during the test causing the picture to come out blurry, we considered that a positive segment. We elected to do that because we assumed that in the real world, we would dictate that the patient go for a cath if there is incomplete or inadequate visualization of any of the major coronary segments. Accordingly, we censored every non-evaluable segment as a positive. In most previous single-center studies, these were excluded from analysis, but we didn’t think it was fair to do so. If all non-evaluable MDCT segments had been excluded (or considered negative) from the study, 15 patients with stenosis of more than 50% would have been missed. In the real world, if you do a test and the test is non-diagnostic, then you just don’t stop there, you have to go onto another test. You assume that it could be positive. What were patient criteria for trial enrollment? We had a specific criteria for enrolling these patients, who were clinically referred for a cardiac cath. They had to have symptoms, and intermediate or high probability of having coronary disease based on clinical criteria. They were approached and asked to have a CT for a research study. The CT was then done between one and seven days before the cath. The CT studies were sent to a core lab in Germany, where they were reviewed by an investigator who was blinded to the results of the cath. The caths were all sent to a core lab in Cleveland, where they were analyzed by quantitative coronary angiography (QCA). Once the cath analysis and the CT analysis were completed, then an independent data management and statistical group took the two sets of data and compared them. In summary, there was a very subjective and rigorous analysis that was undertaken in this study. What were your conclusions as to the type of patient that would benefit most from a CT scan? Since we utilized a 16-detector scanner in the trial, we have to take that into consideration. Our recommendations are that if you are using a 16-detector scanner, you should be very selective. Physicians should not take every patient with suspected coronary disease and chest pain symptoms, do a 16-detector study, and then base their decision on the results as to whether the patient should go for a cath, because there will be too many cases where the results will be a false positive. On the other hand, if you were to apply the study to patients that, for example, have an equivocal stress test, then the CT test in that situation could be useful, because it would eliminate many patients that have normal coronaries where the stress tests are equivocal or are false positives. In that scenario, a negative CT pretty much excludes coronary disease. In general terms, we didn’t feel a 16-detector scanner should be the first test of choice for evaluating a patient, but rather the second test of choice. Are there economic benefits to a CT scan versus fluoroscopy? Certainly. If you apply it wisely, and you use it in those patients in whom you suspect the diagnostic stress test is incorrect, you will reduce the number of patients that go onto cath and have normal coronaries. The cost of a cath is significantly higher than the cost of a CT, depending on whether it is a Medicare or a primary insurance payer. It could be considerably higher, maybe four or five times as high. Also, there are more complications with cath, and these are also costly. The economic implications are positive for using CT on a controlled, selective basis. What type of coronary disease do you think might best be viewed with CT? We did not address that in the study. Sensitivity and specificity were based on whether we could detect 50% or more coronary stenosis. In fact, CT is an extremely good test to pick up early coronary disease. You can see plaques in the walls of coronary vessels, even before you see any in the cath film. CT is a combination of cath and intravascular ultrasound, so it has great potential to not only to see the plaque, but determine whether the plaque is calcified or non-calcified. We don’t know yet whether a non-calcified plaque means a vulnerable plaque. There are no studies published as yet that have shown a different prognosis with a non-calcified versus a calcified plaque. Some studies are going to come out soon, but so far they haven’t been published. I do believe the potential exists to identify the type of plaque and potentially determine which patient is at greater risk for a cardiac event. Since contrast is utilized with a CT scan, did any patients have issues with contrast-induced nephropathy in the trial? It is a concern that causes us to be selective as to the type of patient we recommend for a CT scan. In our trial, we excluded any patient with serum creatinine over 1.5 mg/dl (in other words, anyone with significant renal insufficiency). In the trial, we did not have a single episode of renal insufficiency or any complications related to the CT procedure. What are your plans for future studies? We are going to be looking in greater detail at the type of plaques that we can see with CT and the different contrast agents that could perhaps help to identify the more vulnerable plaque. We will be looking at prognosis, whether the findings on CT are predictive of cardiac events, as well as looking more closely at the potential role of CT in the emergency room to decide which patients can be safely discharged from the hospital when they come with chest pain. A word of caution: The bottom line is that no test replaces the need for clinical judgment. CT is revolutionary because it allows us to look at the coronary anatomy non-invasively, which is something we couldn’t do before. However, the idea that we are going to use a test, whether CT or something else, and apply it to an entire population of adults in the U.S., is impractical. Application of CT will never be practical without introducing or selecting out the patients who will most benefit from the test, based on clinical judgment. The CATSCAN trial has shown us that if we apply this test to the right people, we can exclude those with coronary disease, but if we apply it incorrectly, we will detect too many false positives. At the time of this interview, Dr. Garcia was Director of Cardiac Imaging at The Cleveland Clinic Foundation (CCF). He was involved with the CATSCAN study while at CCF. He has since taken a position at Mount Sinai Medical Center. Regarding his position at CCF, Dr. Garcia noted, I [was] primarily involved in the interpretation of cardiac CT and MRI. We share that function with the department of radiology, and we also are involved in other imaging modalities such as echocardiography. In our center, we have a Philips Brilliance 64-slice and a Siemens Definition dual-source scanners dedicated almost exclusively for cardiac imaging. Dr. Garcia can be contacted at rkapur @ hmpcommunications. com

1. Garcia MJ, Lessick J, Hoffmann MH; CATSCAN Study Investigators. Accuracy of 16-row multidetector computed tomography for the assessment of coronary artery stenosis. JAMA 2006 Jul 26;296(4):403-411.