Author + information
- Edward A. Hulten, MD, MPH⁎ (, )
- Sara P. Petrillo, MD and
- Todd C. Villines, MD
- ↵⁎Cardiology Service, Walter Reed Army Medical Center, Washington, DC 20307
We appreciate the important points of clarification of our meta-analysis (1) raised by Dr. Genders and colleagues. Combined endpoints have inherent advantages and limitations both in single studies and in pooled analyses (2). Because not all studies included in our analysis had complete follow-up for revascularization, we first sought to summarize the raw data and then provide a pooled estimate of individual endpoints and a combined endpoint. We should emphasize again that revascularization is a “soft” endpoint, subject to biases such as verification and referral bias more so than death or myocardial infarction (MI). Use of a combined endpoint of major adverse cardiovascular events (MACE) (including death, MI, or revascularization) presents limitations that we evaluated with subgroup analysis using individual endpoints and a “hard” MACE (death or MI) endpoint. We have presented the raw data in an effort at full disclosure and to allow for readers to appraise each endpoint of their own accord. We agree with Dr. Genders and colleagues that this limitation should be stated explicitly.
With regard to studies that did not report revascularization, these were included in the pooled combined endpoint, although as Genders and colleagues have noted, this underestimates a combined endpoint (MACE) most significantly in the patients with obstructive coronary disease by cardiac computed tomography angiography (CCTA). Pooling only studies that reported revascularization (n = 12) results in a sensitivity, specificity, and negative likelihood ratio of 99.6%, 70%, and 0.006, respectively (for future MACE). The annual event rates for MACE analyzed in this manner are 0.08% for normal coronary arteries, 2.5% for nonobstructive disease, and 28.2% for any obstructive disease on CCTA, in contrast to our originally reported 0.17%, 1.4%, and 8.8% annual event rates among these groups, respectively. As discussed by Dr. Genders and colleagues, restricting the study population to only those reporting all MACE endpoints results in a higher estimated pooled annual event rate, but does not qualitatively change the sensitivity and other summary statistics. With regard to the Fagan's nomogram, we agree that for the term “pre-test probability of disease,” the use of “disease” is unnecessarily vague in that context and could be misinterpreted as angiographic obstructive disease. However, consistent with the context of our paper, we are referring to “disease” as clinical adverse cardiovascular outcomes (death, MI, revascularization). In that regard, a hypothetical pre-test probability of 20% would connote a patient with high baseline risk, considering, for example, a 24% revascularization rate in patients with obstructive disease.
The current literature consistently demonstrates that CCTA has significant prognostic usefulness; however, because of the highlighted limitations to the available published data, additional studies that include longer follow-up durations and detailed outcome reporting will further clarify the prognostic test parameters of CCTA.
Please note: Dr. Villines has received moderate speaker honoraria from Novartis pharmaceuticals. There are no other disclosures or conflicts of interest.
- American College of Cardiology Foundation
- Hulten E.A.,
- Carbonaro S.,
- Petrillo S.P.,
- Mitchell J.D.,
- Villines T.C.
- Kip K.E.,
- Hollabaugh K.,
- Marroquin O.C.,
- Williams D.O.