Although the technology is fascinating, caution must be exercised when applying coronary computed tomography angiography (CCTA) to emergency department chest pain patients. The only way this diagnostic modality will help us, and our patients, is if it guides us with risk stratification, showing a clear path to discharge for patients we currently don’t know what to do with. Some feel CCTA does just that, and others don’t. So, where do we stand with the data? Has the recent New England Journal of Medicine study, ROMICAT-II, by Hoffman (not Jerry) answered all of the relevant questions?1 Despite the fact that it is a large, multi-center trial enrolling 1,000 patients, I think the answer is no.
On one hand, we are trying to avoid the substantial dose of ionizing radiation seen with the proliferation of CT scanning. On the other hand, some are reporting that CCTA is the new “sliced bread.” Before addressing ROMICAT-II, the initial studies performed should be discussed. They were fraught with issues, from referral bias to technical limitations, which we’ll break down one by one.
With respect to referral bias, patients with high clinical suspicion, already referred for cardiac catheterization, were often used in the early studies. There should be no surprise that the presence of CAD was frequently noted in these cases, making CT identification of plaques very likely.2 However, such patients that present to the ED with chest pain will most likely be admitted and are not candidates for out patient evaluation anyway. So, using patients with relatively high pre-test probability provides us with little useful data with respect to low to intermediate risk ED chest pain patients.
Calcium scores were originally utilized to assess risk. However, it has been proved that 41% of coronary artery plaques are not calcified.3 In addition, the sensitivity of a calcium score of “0” for the absence of a stenosis 50% or greater has been shown to be only 45%.4 Calcium scores are worthless and only serve to confuse the discussion at this point.
Earlier techniques/studies documented high sensitivity but low specificity for obstructive CAD. Thus, many false positives required expensive and potentially risky follow up studies (e.g. cardiac catheterization).5 Although techniques have improved, this is still a limitation with CCTA.
Many of the earlier studies used non-ED patient populations, which has been rectified in more contemporary studies. However, many, including the Hoffman study, still include low risk patients. The argument can, and should be made, that if these patients are low risk, they don’t need such an expensive test to begin with. Furthermore, if they are truly low risk, can’t they be safely discharged for outpatient evaluation? Some authors have suggested that low-pretest probability results in negligible post-test risk of adverse outcomes. So, focusing on intermediate-risk patients is more appropriate.6
The use of a 50% stenosis as the standard for a positive study seems a bit arbitrary. However, it has been, and continues to be, the standard for positive CCTA. Is this the right standard? It is well understood that a lesion of less than 50%, dependent upon the composition of the plaque, may be more unstable, and thus more likely to rupture than one greater than 50%. The original ROMICAT study (Rule Out Myocardial Infarction using Computer Assisted Tomography) had excellent negative predictive value for plaque or stenosis, 100% and 98% respectively. However, the sensitivity for a stenosis of more than 50% was only 77%. In addition, the positive predictive value of plaque for ACS was only 17%. Thus, many patients with plaque didn’t have ACS. So, based on this study, the positive predictive value was excellent for excluding the disease we are looking for. The older the patients got, the worse the specificity of the presence of plaque for ACS became. In those older than 65 years, the specificity was 21%, and for those less than 65, it was 59%. This is due to the increasing prevalence of plaque with advancing age. Although the negative predictive value is compelling when no plaque is present, this study begs the question: should CT be utilized in patients with identified plaque? The answer is probably not. Proving negative predictive value in a low risk population that doesn’t have the disease is a fairly easy task. In this study, only 8.4% of patients had ACS. In ROMICAT, there was enough clinical suspicion for ACS for all of these patients to be admitted. So, were they really low-to-intermediate risk? Thus, the question with ROMICAT and the new ROMICAT-II study is how do we risk stratify, not whether or not CT excludes ACS in patients at very low risk.
ROMICAT-II utilized CCTA to determine who was safe for discharge from the emergency department. 1000 patients aged 40-74 years, without ischemic ECG changes and with a negative initial troponin were ultimately randomized to either standard ED treatment or CCTA. If the attending emergency physician felt the patient needed further risk stratification, they were eligible for enrollment.
Their primary endpoint was length of stay. Again, the study population’s rate of ACS was 8%. The hospital length of stay was reduced by 7.6 hours and more patients were able to be discharged from the ED in the CCTA group than the standard evaluation group (47% v. 12%).
Effective risk stratification is a critical piece of this equation. In this study, the clinical gestalt of the attending emergency physician, along with a non-ischemic ECG and negative troponin, was the risk stratification strategy. Adding an expensive test that exposes patients to significant amounts of ionizing radiation (although newer techniques have reduced the dose) is probably overkill in the low risk patients and not far enough in those with intermediate risk. From a clinical perspective, the distance between those with low risk for ACS and those with intermediate risk is too wide a gap to lump them all together. In other words, evaluation strategies used for low risk patients may be totally inappropriate in intermediate risk patients. Quite honestly, in low risk patients, very few actually have the disease. So, many diagnostic strategies would work. What they often need is good outpatient follow-up in a couple of days, not an expensive test performed in the ED. However, the intermediate group patients have a much greater likelihood of having the disease. So, a more careful approach may be needed. The real question that ROMICAT-II does not answer is how we determine which patients are truly low risk. Instead, they defer to the result of the CT to answer this important question.
It’s also important to note that in ROMICAT-II, despite the reported reduced length of stay and reduced number of admissions, the CCTA group was exposed to higher doses of radiation and was subjected to more “downstream” tests than the standard evaluation group. In addition, the costs for both strategies were about the same, $4,289 and $4,060 for CCTA and the standard groups, respectively.
A companion piece, or rebuttal was published in the New England Journal as well.7 Although we really shouldn’t pick and choose which parts of an article we like, in this case I’ll make an exception. I think the rebuttal was spot on with some of its criticisms, but way off the mark in the final recommendation. In short, Dr. Redberg stated that there is no benefit proved for those who had CCTA performed and thus, when the low-to-intermediate risk patients have a negative ECG and a negative troponin, they can be discharged from the ED without further evaluation. I agree with the criticisms of the study. However, her recommendations oversimplify a very complicated diagnostic dilemma. Just because this article doesn’t prove that CCTA was beneficial, it can’t be interpreted to mean that no diagnostic evaluation is necessary. In addition, just because you’re a cardiologist, doesn’t mean you’re qualified to determine the standards of practice in emergency medicine.
The downside risk of missing ACS is a much different proposition than missing a rib or ankle fracture. When ACS is missed, bad things usually happen. Thus, a relatively conservative approach to patients suspected of ACS is warranted. We have to be cautious to not adopt technology without clear rules for its application. In truly low-risk patients, I agree that emergent imaging isn’t necessary most of the time. The question isn’t whether or not CCTA can prove that low risk patients don’t have coronary artery disease. It can. The question is who is truly low risk and who has greater risk.
The technology is intriguing and ROMICAT-II is well designed. However, it may be answering the wrong question. I agree that a negative CCTA excludes disease and we are done. However, I’m still not certain how, when, where and on whom to apply this evolving technology.
1. Hoffman. N Engl J Med 367;4 nejm.org july 26, 2012
2. Ravipati, G., et al, Am J Card 101(6):774, March 15, 2008
3. Scholte, A.J.H.A., et al, Heart 94(3):290, March 2008
4. Gottlieb, I., et al, J Am Coll Card 55(7):627, February 16, 2010
5. Circulation. 2000 Jul 4;102(1):126-40.
6. Van Werkhoven, J.M., et al, Heart 95:1607, October 2009.
7. Redberg. n engl j med 367;4 nejm.org july 26, 2012
Kevin Klauer, DO, EJD Editor-in-chief of Emergency Physicians Monthly, CMO of Emergency Medicine Physicians, Vice Speaker of the ACEP Council.