• acute myocardial infarction

Symptoms suggestive of myocardial infarction (MI) represent one of the most common conditions in patients presenting to the emergency department. For the diagnostic evaluation of these patients, algorithms based on cardiac troponin are recommended by guidelines and used in daily practice worldwide.1 In the past 10 years, substantial improvements of these diagnostic strategies have been achieved. This is mainly based on a higher assay sensitivity, which allows the detection of very low troponin concentrations with a high accuracy.2 Using low cut-off concentrations, rapid strategies for rule out of MI after 3 hour or only 1 hour are feasible, efficient and safe. Even a single baseline rule out using a very low troponin cut-off concentration, measured by a high-sensitivity troponin (hs-Tn) assay, provides a high safety to rule out MI.3–6 Currently, up to 20 different troponin assays are commercially available.2 This adds substantially to the confusion around interpretation of the results of the assays. These assays differ in most of their characteristics, such as the limit of detection, the concentration with a 10% coefficient of variation and the sex-specific 99th percentile. This makes interassay comparisons nearly impossible. Hence, findings from diagnostic studies are always specific for the investigated troponin assay and specific informations as cut-offs cannot be transferred to other assays.

This topic is addressed in the present manuscript by Chapman et al.3 The authors from the High-STEACS research group investigated the diagnostic performance of a novel hs-TnI assay, which is run on the Siemens Atellica system. Hs-TnI concentrations were measured in 1920 patients with suspected MI directly at admission and after 3 hours. For the rule-out of MI, the authors suggested a single admission hs-TnI cut-off concentration of 5 ng/L, when symptoms were lasting for more than 2 hours. Alternatively, a 0/3 hour change of less than 3 ng/L in combination with a hs-TnI concentration below the assay-specific 99th percentile was used to rule-out MI. Importantly, the cut-offs of 3 and 5 ng/L were transferred from a different study using hs-TnI concentrations measured by the Abbott Architect systems and were not derived in this present study population. This rule-out strategy resulted in high negative predictive value of 99.5% and included 63.4% of the overall study population. For the rule-in of MI, a hs-TnI concentration above the 99th percentile or a 0/3 hour change of more than 3 ng/L was applied. The according positive predictive value was only 38.8% and included the remaining 36.6% of the study population. Finally, the authors compared their findings to the European Society of Cardiology  (ESC) 0/1 and 0/3 hour algorithm. Again, values were utilised that are normally applied for other assays. For the ESC 0/1 hour algorithm, they report a high negative and positive predictive value, while 29% of all patients were categorised in the observe group, which is not present in the high-sensitivity troponin in the evaluation of patients with suspected acute coronary syndrome (high-STEACS) pathway. The ESC 0/3 hour algorithm, which is based on the 99th percentile as a rule-out cut-off, provided much lower negative and positive predictive values.

The authors must be congratulated for undertaking the efforts to establish more evidence in this important field. So what have we learnt from this new diagnostic study?

First, the high-STEACS pathway for patients with suspected MI provides a high diagnostic accuracy to rule out MI. These findings are in line with data from the Advantageous Predictors of Acute Coronary Syndromes Evaluation (APACE) group, who derived and validated an algorithm using the same hs-TnI assay.7 In the APACE manuscript by Boeddinghaus et al, a single baseline hs-TnI cut-off of 3 ng/L was used when symptoms lasted for more than 3 hours, or a baseline hs-TnI cut-off of 6 ng/L in combination with a 0/1 hour delta of less than 3 ng/L to rule out MI.

Second, when using hs-Tn assays, interassay transfer of distinct very low cut-off concentrations might theoretically be possible. This approach could potentially improve comparison between different assays at low concentrations. Nevertheless, this needs to be interpreted carefully. Indeed, it is important to notice that this approach is only feasible when very low changes are taken into account on a high-sensitivity platform. Hence, when the changes are small and cut-off values are at the very lower end, it is of no further importance how low they really are. For example, using values near the limit of detection will always be feasible on modern high-sensitivity assays. This is obvious knowing that very low troponin values are generally considered to represent lowest cardiovascular risk. Nevertheless, if a physician might still want to use higher values around the 99th percentile, this general cut-offs will be dangerous and misleading.

In this regard, application of the 99th percentile as cut-off for rule out using high-sensitivity assays, as recommended in the current ESC guidelines, appears way too high and some patients might be classified as false negatives. Therefore, we feel this is obsolete today.

Finally, strategies for rule in of patients with MI remain highly challenging. The proposed high-STEACS pathway does not allow classification of patients in the observe zone and follows the Fourth Universal Definition of MI to label all patients with hs-Tn concentrations above the 99th percentile as ‘myocardial injury’.8 This results in a low positive predictive value for the actual diagnosis of MI. The APACE algorithm on the other hand uses much higher cut-off concentrations to rule in (120 ng/L at admission or 0/1 hour delta of at least 12 ng/L), which increased the positive predictive value to 72.5%. The main question is: which patients do we aim to identify? Those with an acute MI requiring urgent revascularisation or those with myocardial injury, who are at high risk and require more careful evaluation. Troponin concentrations are influenced by numerous confounders, such as sex, age, kidney function and many more. Therefore, in future, a more individualised approach including troponin concentrations, their dynamic change, but also adjustment for confounders might improve our understanding of elevated troponin concentrations.