Is a New Classification of Postoperative Myocardial Infarction Justified?

Abstract

Royal Brompton Hospital, London, United Kingdom. dylanbould@yahoo.co.ukI read with interest the prospective audit of routinely measuring cardiac troponin I (cTnI) levels postoperatively in infrarenal aortic surgery by Le Manach et al. 1The authors have proposed a classification of two separate groups of postoperative myocardial infarction (PMI) based on their findings. However, I think that aspects of both the methodology of the study and the presentation of results are open to criticism and that the validity of their distinction between early and delayed PMI is questionable.It is not clear why the value for cTnI that the authors consider to be abnormal changed in the study and in their participating institution from 0.5 ng/ml from September 1995 to November 1999 to 0.2 ng/ml from November 1999 onward. It seems that the same machine was used for testing cTnI throughout the study period. This parameter is the basis of the definitions of both myocardial damage and delayed PMI and also the proposed difference between the early and delayed PMI groups and is clearly a key issue. Regardless of whether the institution changed its normal values for cTnI, it would seem that the most appropriate methodology would be to analyze the whole population by the same rules. It would be of interest whether the endpoints would differ from that reported if all patients were reanalyzed together at 0.2 ng/ml and at 0.5 ng/ml.The abstract states that the cTnI profiles between the delayed PMI group before the cTnI threshold for PMI was reached and the myocardial damage group were identical. In the main part of the Results, it is only written that the profiles are comparable. Even so, the only evidence presented for this is a graph of two “representative” patients. To describe the two profiles mentioned as identical would require descriptive statistics of cTnI values to be presented against time for each of the different groups, with a mathematical analysis of their difference. I also note that there was not even a “representative” example described of the other two groups.The timing of the 24-h cutoff for the difference between early and delayed PMI was predetermined and arbitrary. The authors state that in the early PMI group, PMI was not preceded by subinfarction myocardial damage, but did any of the early PMI group patients have abnormal cTnI for 18 rather than 24 h? If so, how do we know that this was not significant myocardial damage? It would seem to be a more robust choice of timing if at first the pattern within the population was examined and then the most appropriate cutoff was chosen.The authors note that only the early PMI group had an increased incidence of previous myocardial infarction and hint that this supports their hypothesis of different etiology in the two PMI groups. I note that 57 comparisons are made in table 2. Because the P values are only described as less than 0.05 and not specified, it is hard to know how significant their four P values of less than 0.05 in table 2 really are when you would expect almost three to occur just by statistical probability. Defining two groups on their differing temporal characteristics, proving statistical difference in a clearly related temporal parameter in the absence of any other differences in nonrelated parameters, and then inferring that the two groups are fundamentally different is a weak argument.The opinion that the time when the cTnI is abnormal but not diagnostic of PMI should be considered a “golden period,” and the hypothesis that intensive interventions may influence outcome is attractive. The description of the natural history of cTnI changes would be more useful if we knew whether the ischemia was silent or associated with evidence that would have resulted in appropriate therapy anyway. I would like to know whether the myocardial damage group or the patients with abnormal cTnI before PMI had chest pain, dyspnea, heart failure, arrhythmias, or other electrocardiographic changes suggestive of ischemia.In conclusion, the authors have not shown a bimodal distribution that would justify the classification of PMI as early and delayed, nor have they conclusively shown a fundamental difference between rapid-onset and delayed-onset PMI groups on the evidence as presented. Moreover, I think the suggestion that early PMI is “hitherto unrecognized in the postoperative setting” is unfounded.Royal Brompton Hospital, London, United Kingdom. dylanbould@yahoo.co.uk