High-sensitivity assays for cardiac troponins – continued

Two years ago I wrote an editorial on a series of articles published in CCLM dealing with the novel high-sensitivity assays for cardiac troponins (cTn) [1]. The articles covered a broad spectrum of critical topics which were debated in the scientific community and they addressed gaps of knowledge associated with the use of these novel assays. In this issue we have again a series of articles dealing with various aspects of the novel assays for cTn showing clearly that even though there has been substantial progress there are still unanswered questions and unmet needs in the field [2], [3], [4], [5], [6], [7].

A large number of high quality clinical studies have been published in the meantime and guidelines and recommendations have been adapted. However, the question of what is the optimal strategy for rule-out and rule-in of myocardial infarction when the novel more sensitive assays are used has not been finally resolved. Accordingly, there are numerous similar but slightly different recommendations and guidelines on how to triage patients with symptoms of acute coronary syndrome on the basis of cTn plasma concentrations in the emergency room [8], [9], [10], [11], [12], [13]. This somewhat confusing situation is discussed by Ferraro et al. [2] who also present their synthesis of available data as a proposed procedure in this situation. By implementing their in-house consensus procedure this group removed uncertainties in interpretation which is in some way similar to the approach favored by Greene and Tate [3].

One obvious trend in all the approaches for rapid decision making is that cutoffs related to assay characteristics like limit of detection (LoD) or the 99th percentile are gradually abandoned. It has been widely accepted that the major issue is related to negative predictive values (NPV) (rule-out) and positive predictive values (PPV) (rule-in) for myocardial infarction but also for the individual patient’s risk of future major adverse cardiac events. While optimal cutoffs for these diagnostic purposes need to be defined for every assay system, it is obvious that traditionally used cutoffs like the 99th percentile provide some guidance but are not necessarily optimal. This immediately raises the question, whether we should spend much effort on determination of the 99th percentile or rather on clinical studies providing individual NPV and PPV. As Clerico et al. [4] discuss in detail, determination of the 99th percentile is strongly dependent on the selection of the reference cohort. Age, sex, underlying cardiac and non-cardiac diseases all affect individual cTn concentrations. Thus, even if only age and sex shall be taken into account, reference cohorts must be very large [14]. In this situation one may argue that it is not worthwhile to spend much effort on the 99th percentile but rather invest in clinical studies which will provide NPV and PPV in cohorts of patients with symptoms suggestive of acute coronary syndrome. In doing so one should not forget that predictive values and in the case of myocardial infarction in particular PPV strongly depend on pretest probability, i.e. the proportion of patients admitted to the emergency room who have myocardial infarction. Admittedly, studies aiming at predictive values will be even more time consuming and expensive. On the other hand, we know that cTn are stable in frozen samples so that high quality biobanks can be used to assess different cTn assays on the same study material over prolonged periods of time.

Another issue relates to the sex dependent interpretation of cTn results. While there is ample evidence that women have lower cTn plasma concentrations than men and the 99th percentiles of the sexes are significantly different, the use of sex specific cutoffs is not uniform [4]. Interestingly, most rapid rule-out and rule-in strategies proposed so far do not take this sex difference into account [8], [9], [10], [11], [12]. On the other hand, there are indications that women who would be reclassified as myocardial infarction if a lower, sex specific cutoff was used may have a much higher risk of future major adverse cardiac events than women classified as no myocardial infarction using this lower cutoff. In fact, their risk was comparable to women classified as myocardial infarction with the higher cutoff for both sexes [15]. As described by Greene and Tate [3], an Australian/New Zealand expert group recommended the use of sex specific cutoffs for cTnI in the two countries.

One topic which becomes very obvious from these considerations is the difference between analytical and diagnostic performance of an assay. In laboratory diagnostics we tend to concentrate on parameters of analytical performance, e.g. limit of blank, LoD, limit of quantitation, trueness, imprecision. While these are undoubtedly of importance, diagnostic performance is determined by many additional factors. The discussion on analytical sensitivity and data censoring [5], [6] highlights an important facet of this problem showing that the difference in diagnostic performance between higher and lower sensitivity troponin assays may have been artificially inflated by the common practice of data censoring with the lower sensitivity assays as Ungerer and Pretorius argue [5]. While their argument is valid, the proposed solution, i.e. reporting of all measured results together with their uncertainty appears impractical in clinical routine as outlined by Jarolim [6]. An obvious limitation of all assays with high imprecision at low cTn concentrations is their inability to reliably detect minor changes in cTn concentration within 1–3 h. Most protocols for rapid rule-out or rule-in use absolute changes of cTn concentration in the range of few ng/L within 1–3 h [8], [9], [10], [11], [12]. Assays with high imprecision will not be suitable for this purpose, because of the high uncertainty of the observed delta.

And finally, currently all protocols for rapid triage are based on cTn assays run on traditional central laboratory immunology analyzers. However, we have seen the development of novel point of care assays with significantly improved analytical sensitivity and precision. Lopez-Calle et al. [7] present another novel method suitable for point of care analyzers with analytical performance comparable to the currently available highly sensitive cTn assays. This is further evidence that we will soon see a new generation of true point of care assays designed for handling by non-laboratory staff with analytical performance equivalent to central laboratory systems. As turn-around time is supposedly a critical issue this will likely affect our way of measuring cTn in the emergency setting. However, point of care testing will be more costly and likely increases the workload of emergency room staff. Therefore, we will need cost-benefit analyses of this approach. Currently available data do not support an immediate compared to a delayed intervention in patients with non-ST elevation myocardial infarction or acute coronary syndromes with the exception of hemodynamically unstable patients [8]. Therefore, more rapid rule-in perhaps will not improve patient outcome. If there is no benefit from immediate intervention, the major potential improvement in clinical decision making will be rapid rule-out of myocardial infarction and earlier discharge from the emergency room. In this situation the decision for or against point of care testing will be driven purely by economic and organizational aspects. These again may depend on local peculiarities of a given hospital.

An interesting future perspective of point of care testing might be rapid rule-out with 1-h algorithms in the doctor’s office. The outdated 6-h algorithms that in theory must be used with the less sensitive point of care assays were not a realistic approach in the outpatient setting. However, if rapid rule-out strategies can be adopted at the point of care this may change profoundly the management of patients with suspected myocardial infarction reducing the number of patients admitted to a hospital emergency room or chest-pain unit. A recent article based on current evidence suggests that empowering practitioners to rapidly and reliably identify patients at low risk will also reduce health expenses for this patient group significantly [16], [17].

Overall cTn is a remarkable example of how technical improvement of laboratory tests alone affects medical practice. I am sure that we will see further rapid improvement of our knowledge regarding the most efficient use of markers of myocardial injury in the years to come.