We thank Dr. Bould for his interest in our article.1Our definition of myocardial infarction was based on the normal values in our specific laboratory.2The threshold for abnormal cardiac troponin I (cTnI) was modified by the laboratory of our hospital, which, after an improved accuracy of the dosage technique of cTnI, considered as abnormal a value above the serum 99th percentile of the normal population.3Because the normal value in our laboratory changed, so did the definition of myocardial necrosis. This is in complete agreement with the definition of myocardial infarction in the literature, which clearly considers cTnI plasma level as the accepted standard to evidence postoperative myocardial infarction.
Dr. Bould suggests a complementary analysis to better describe differences between myocardial damage and delayed myocardial infarction. We believe that no complicated descriptive statistics are needed, because cTnI values of the two groups are very similar, as evidenced in figure 1. A 24-h delay was arbitrarily retained to separate the early myocardial infarction and delayed myocardial infarction groups. From a clinical standpoint, we consider that myocardial damage lasting more than 24 h preceding myocardial infarction provides a unique opportunity to introduce a treatment to improve myocardial oxygen balance. In addition, a 24-h interval has some statistical relevance, considering that a normal distribution cannot be excluded using a Shapiro-Wilk W test for the distribution of the delay to peak cTnI (and 1.5-ng/ml threshold) in the delayed myocardial infarction and postoperative myocardial infarction groups, whereas an abnormal distribution was found in the overall myocardial infarction population. Nevertheless, we assume that normality tests have low statistical power (probability of detecting nonnormal data) in small samples, and so those normality tests are not strong arguments to statistically separate two populations. Dr. Bould emphasizes that the statistical differences observed in table 2 might be due to change. We want to underline that these data are presented as more descriptive than explicative, and that no causal assumption could be made in this study.
Our conclusions are based on a significant difference in the incidence of previous myocardial infarction between early myocardial infarction and myocardial damage groups with a P value lower than 0.002.
A “representative” example of a control group in which by definition cTnI plasma level equals zero, cannot be shown in a figure in which cTnI plasma levels are represented on the vertical axis. In addition, this graph was included in the manuscript to illustrate a “golden period,” explaining why the inclusion of a “representative” patient of the early group would have been confusing for the reader.
As firmly established in the literature, abnormal postoperative cTnI levels are rarely associated with clinical symptoms. Moreover, pain, dyspnea, and hypotension have many extracardiac causes after major vascular surgery and are unacceptable surrogates for myocardial infarction.
In conclusion, there is a very clear temporal distinction between the groups as to both time to first abnormal cTnI value and time to peak cTnI value, whereas the time from the end of surgery to first abnormal value was the same. Consequently, we believe that our conclusions are justified. Numerous previously published articles describe considerably prolonged myocardial damage before postoperative myocardial infarction. This has led to the development of the prevalent theory of cumulative myocardial injury as the main process that leads to postoperative myocardial infarction. Our early myocardial infarction group clearly does not fit into that category. Hence, we dared to offer the hypothesis that there may be two types of postoperative myocardial infarction. Dr. Bould’s arguments have not convinced us that our hypothesis is wrong.
*Centre Hospitalo-Universitaire Pitié-Salpêtrière, Paris, France. email@example.com