Prognosis, Prognosis, Wherefore Art Thou Prognosis?

Friends, fellow triagers, colleagues, lend me your eyes—I come to bury Peacock et al.,1 but also praise them. For this one article has added to the literature and furthered our understanding of the diagnostic utility of high-sensitivity troponin (hsTn), yet leaves us desirous of so much more. Allow me, in the paragraphs that follow, to explain. As this investigative team aptly notes, uptake of hsTn in the United States has lagged compared to the rest of the world. While reasons for this lie primarily with approval delays at the Food and Drug Administration (FDA), there remains a reluctance in many institutions to shift from conventional assays to their more sensitive counterparts. When pressed, decision makers at such facilities point to the absence of U.S. data and feign uncertainty about potential accuracy in “their” patient populations. It is within this void that Peacock et al. present new(ish) data on Beckman Coulter's high-sensitivity troponin I (hsTnI) assay, showing that it can reliably identify emergency department (ED) patients with suspected acute coronary syndrome (ACS) who have not sustained myocardial injury. Such work is important from a validation perspective, though incremental at best, providing information that is really a mirror image of itself (i.e., absence of a protein that reflects myocardial injury indicates absence of myocardial injury). Said another way, publication of data showing that the Beckman Coulter hsTnI assay can effectively rule out myocardial injury may be necessary to convince skeptics of its safety profile but the exercise is merely revealing an inherent truth based on known characteristics of the underlying pathophysiology and laboratory medicine. The fact that this analysis included an all-comer prospective cohort of ED patients does make the findings unique and adds to their applicability, although it does not alter this perspective. However, there are nuances to be gleaned from the presented data that warrant further discussion. Foremost, the authors show that low risk, defined as a negative predictive value (NPV) for myocardial infarction (MI) of 99% or greater, can be identified using either the manufacturer delineated limit of quantification (LoQ) of 2.3 ng/L or the 10% coefficient of variation (CV) cut-point of 5.6 ng/L with the Beckman Coulter hsTnI assay. Given that reporting of hsTn in whole integers (rather than with decimals) is the industry standard, it would have been nice to see data with rounding up to 3 and 6 ng/L, respectively, although it probably would not have made much difference. Nonetheless, the high NPV for both cut-points was preserved on singular and serial measurement, provided for the latter that at least one result was below the LoQ or 10% CV and no repeat value was above the upper reference limit (URL) of 17.9 ng/L. Interestingly, as also shown by Peacock et al (see figures 1 and 2 from the manuscript), NPV was consistently 100% regardless of the pattern of hsTnI change for these patients, suggesting independence from the actual delta value. Such a finding stands in distinction from non-U.S. data where absolute deltas with this assay that exceed predefined thresholds of 4 ng/L at 1 hour2 and 5 ng/L3 at 2 hours, even among those with low initial troponins, were associated with acute MI. This may be attributable to the cohort enrolled by Peacock et al., where the prevalence of MI was lower (11.2% vs. 15.4%2 and 14.5%3) and the time from symptom onset to first blood draw which exceeded 3 hours for the vast majority of enrolled patients. From a care delivery perspective, demonstration that the 10% CV cut-point provides acceptable safety is key as it enables rule-out in more than twice as many people (58% vs. 24%) compared to the LoQ. Relevant to evaluating patients with suspected ACS, the work by Peacock et al. also reinforces a critical paradigm shift, namely, recognition that the URL, which has long been the cut-point for ruling out MI, is no longer sufficient. Regardless of the assay or manufacturer, there is clinical meaning and enhanced risk when hsTn is below the URL but above a lower bound cut-point such as the LoQ or 10% CV.4 While protocols that incorporate deltas can help determine the immediacy of associated risk and the actionability of related information, emergency physicians should appreciate that the broader value proposition of hsTn testing may center more on secondary disease prevention than acute diagnosis. Just how much this matters will depend on the approach to hsTn that is used, which in turn is determined by the desire to emphasize acute versus chronic risk. For instance, if employing the LoQ as the cut-point, fewer people will be ruled out and many more will fall into the group that the authors refer to as “Hi-Hi,” that is, both values on serial measurement exceeding the established cut-point. Most will have a delta that is not indicative of an acute problem but all will be at higher risk for a cardiovascular event within the next few years.5 On the other hand, if a higher cut-point is used like the 10% CV, an acute event can be safely ruled out in many more but the very real chronic risk that exists among those with hsTn values between the LoQ and the 10% CV may remain hidden as none will get flagged for exceeding the established risk threshold. Academic thinking, as with iambic pentameter, can get in the way and it is important that the clinical implications of this work not be obscured. To that end, Peacock et al. show, as a growing number of other investigators have also done, that ruling out myocardial injury is no longer the primary challenge when evaluating patients with suspected ACS. High sensitivity troponins have put that to rest and we can all feel comfortable knowing that in patients without ST segment elevation on their electrocardiogram, hsTn concentrations below the LoQ, regardless of the assay, unambiguously exclude the diagnosis of MI. By using higher cut-points an increasing number may also be ruled out, with the caveat that reliability depends on characteristics of the population to be tested, especially disease prevalence. However, this must be counterbalanced by the potential to identify people who are at greater long-term risk for cardiovascular disease. Antithetical as this may seem to the specialty of emergency medicine, the broadest impact of this biomarker may rest at the population level with an eye toward utilizing hsTn as a means to not only rule out acute MI but also reduce the risk of future events.6, 7 Whether the HEART score or EDACS can assist with this as the authors allude to remains to be seen as these instruments were designed to help guide disposition based on risk for short-term (30 day) events and not protracted outcomes. Although not necessarily mutually exclusive, the ability to modify either may not be directly influenced by hospitalization or further diagnostic testing making the question of “stay or go” perhaps a suboptimal one. Unfortunately, data beyond the initial hospital encounter were not collected, limiting interpretation of results to diagnostic potential . Alas poor Peacock (et al)! Prognosis matters too.