Response to questions asked online about: “Outcomes in the ISCHEMIA Trial Based on Coronary Artery Disease and Ischemia Severity” by Reynolds et al. Circulation September 2021
Some of the questions below relate to differences in various CCTA data across manuscripts. Differences may be due to varying selection criteria across secondary analyses, as is typical in secondary analyses of large trials. Exclusion criteria pertinent to one analysis may not be necessary in another. Also, there are different numbers of CCTA images that are evaluable for specific analyses, leading to different denominators. For example, as outlined in the manuscript, some CCTA images could not be evaluated for the number of diseased vessels because one or more vessels could not be scored for percent stenosis due to artifact.
Reader Comment: In supplement Tables 1 and 2, only 40 patients are in the modified Duke Prognostic Index Group 6 (3-vessel severe stenosis (>70%) or 2-vessel severe stenosis with proximal LAD), not 659 as in the paper. In addition, in the supplement Tables 1 and 2, there are 659 patients in modified Duke Prognostic Index group 5, not 894 as in the manuscript; there are 894 patients in modified Duke prognostic index group 4, not 743; and, there are 743 patients in modified Duke Prognostic Index group 3, not 179. It appears that there may have been an upgrading of 1 grade in the modified Duke Prognostic Index from the Supplemental Tables to the actual manuscript in all categories of coronary artery disease analyzed. We ask the authors to clarify if our observations are correct and if there is a discrepancy between the Supplemental Tables and the manuscript, which is correct.
- ISCHEMIA Clarification: The numbers in the manuscript were correct and the indicated Supplemental Tables were incorrect. The error occurred when cutting and pasting data from a statistical report into the Supplement. We are grateful that Circulation has allowed us to make necessary updates to Supplemental Tables 1 and 2.(https://www.ahajournals.org/doi/10.1161/CIR.0000000000001080)
Reader Comment: Careful examination of the corrected numbers in the context of the totality of the published ISCHEMIA data identifies significant discrepancies that persist. For example, in the corrected Supplemental Tables I and II (pages 3 and 5), the number of patients with left main (LM) stenosis ≥50% is 40. This number (40) is surprisingly identical to the number cited for LM ≥50% stenosis in the Supplement (Table S5, page 107) of the primary ISCHEMIA manuscript published in the New England Journal of Medicine (NEJM). Logically, one would expect a drop-off in the number of patients with LM≥50% when moving from a denominator of 3845 participants (Table S5, primary NEJM manuscript) to 2475 in participants with evaluable modified Duke Prognostic Score Index in the Reynolds et al. manuscript.
- ISCHEMIA Clarification: Both publications are counting the same 40 participants with left main ≥50% stenosis. The NEJM left main calculation is based on participants who have CCTA scans that are interpretable for presence/absence of left main ≥50% stenosis whereas Reynolds et al. is based on participants who have CCTA scans that are interpretable for the Duke score. The latter is a relatively smaller subset. The number of left main cases is the same because, by definition, patients with left main stenosis ≥50% have an interpretable Duke score, i.e., category 7.
Reader Comment: If 2475 is the correct number as stated in the abstract and body of the manuscript, then why is it different from the 2518 in the corresponding Supplemental Figure I, page 35, of the corrected manuscript?
- ISCHEMIA Clarification: 2518 is the number of participants with an evaluable stress test result and evaluable Duke score. 2475 is the number remaining after restricting to Duke scores 3 thru 6. As noted in the revised manuscript: "Duke categories 1 and 2 (nonobstructive coronary artery disease or normal arteries) and 7 (left main stenosis ≥50%) were excluded from analysis because these subgroups were small."
- Additional detail: The 40 randomized participants with CCTA core laboratory confirmed left main stenosis ≥50% in general had a) protected left main with prior CABG; b) pre-randomization invasive coronary angiography documenting <50% stenosis in the left main; or c) a non-study CCTA transferred to the core laboratory near the end of the study that was interpreted by the core laboratory as showing stenosis ≥50% but interpreted locally before randomization as showing <50% left main stenosis. The 4 randomized participants who had CCTA interpreted by the core laboratory as showing no obstructive CAD had pre-randomization invasive coronary angiography documenting obstructive CAD; 1 of these was excluded from the Reynolds et al. analysis because stress data were not interpretable.
Reader Comment: In addition, the percentage of participants with one coronary vessel having at least moderate (≥50%) stenosis is 7.2% (179/2475) in the corrected manuscript (Supplemental Tables I and II) vs. 23.3% (697/2986) in the primary NEJM manuscript (Table S5).
- ISCHEMIA Clarification: Duke score categories are assigned hierarchically, meaning that participants are assigned to the highest category for which criteria are met. Among 697 participants randomized with only one vessel exhibiting stenosis ≥50% as reported in NEJM Table S5, 135 met criteria for single vessel stenosis ≥70% affecting the proximal LAD (Duke category 5), 359 met criteria for single vessel stenosis ≥70% not affecting the proximal LAD (Duke category 4), and 24 did not have an interpretable Duke score. The remaining 179 participants were assigned to Duke category 3 (1 vessel with at least ≥50% stenosis).