Introduction
Cardiac troponin (cTn) is the biomarker of choice for the evaluation of patients with symptoms suggestive of acute myocardial ischemia. Increases in cTn are necessary for the diagnosis of acute myocardial infarction (MI). Some even have opined that with the advent of high-sensitivity assays that the diagnosis of acute unstable coronary disease will eventually require an increasing pattern of cTn values, assuming timing permits such an evaluation. The present chapter attempts to provide clinicians with the data and skills necessary to understand and use cTn values optimally in patients with possible unstable ischemic heart disease. The selection of appropriate patients in whom to measure cTn and the interpretation of cTn results in the context of other features of the clinical presentation are discussed in Chapter 6 . The use of cTn in conjunction with other cardiovascular biomarkers is addressed in Chapter 8 .
Basic Considerations about Cardiac Troponin for Clinicians
All cTn assays are different, and values from one assay cannot be extrapolated to another. Even assays that seem to identify a similar number of patients in a given clinical situation are calibrated differently; they often use different antibodies, and thus the values will not be the same between assays ( Tables 7-1 and 7-2 ). There have been attempts to standardize assays, but they are unlikely to be successful in the near term. Thus, clinicians need to learn to use the assay(s) that are employed locally. Physicians should know a variety of characteristics as described in the following. Different assays have different issues in regard to specificity, interferences, and sensitivity.
| Company/Platform/Assay | 99th Percentile (ng/L) | Does 10% CV Fall Above or Below 99th Percentile |
|---|---|---|
| Abbott AxSYM ADV | 0.04 μg/L | > |
| Abbott ARCHITECT | 0.028 μg/L | > |
| Abbott i-STAT | 0.08 μg/L | > |
| Alere Triage | <0.05 μg/L | > |
| Beckman Access AccuTnI | 0.04 μg/L | > |
| bioMérieux Vidas Ultra | 0.01 μg/L | > |
| Mitsubishi Pathfast | 0.029 μg/L | < |
| Ortho Vitros ECi ES | 0.034 μg/L | = |
| Radiometer AQT90 cTnI | 0.023 μg/L | > |
| Radiometer AQT90 cTnT | 0.017 μg/L | > |
| Response RAMP | <0.01 μg/L | > |
| Roche Elecsys TnT Gen 4 | <0.01 μg/L | > |
| Roche Elecsys TnI | 0.16 μg/L | > |
| Siemens Centaur Ultra | 0.04 μg/L | < |
| Siemens Dimension RxL | 0.07 μg/L | > |
| Siemens Immulite 2500 | 0.2 μg/L | > |
| Siemens Stratus CS | 0.07 μg/L | < |
| Siemens Vista | 0.045 μg/L | < |
| Tosoh AIA | <0.06 μg/L | > |
| Company/Platform/Assay | 99th Percentile (ng/L) | Does the 10% CV Fall Above or Below 99th Percentile |
|---|---|---|
| hsTnI | ||
| Abbott ARCHITECT ∗ | 16 | < |
| Beckman Access | 8.6 | = |
| Nanosphere MTP | 2.8 | < |
| Singulex Erenna | 10.1 | < |
| Siemens Vista | 9 | < |
| hsTnT † | ||
| Roche Elecsys ∗ | 14 | < |
∗ Available for use worldwide, but not cleared by the U.S. Food and Drug Administration for use in the United States.
† Some would question the classification of this assay as high sensitivity; see text.
Specificity and Interferences
There may be a small number of patients in whom cTnT has some skeletal muscle interference because the proteins that occur in response to skeletal muscle and injury may be re-expressed, and therefore, cause false positives. The frequency of this phenomenon is unclear, and only a handful of cases have been reported. However, a systematic and scientifically robust evaluation of this issue has not been accomplished. cTnI assays do not have this problem, but these assays are more prone to interference from heterophilic antibodies (see the following) and also from antibodies that can block binding and lead to reduced values. For some assays, this phenomenon is estimated to occur in as high as 0.5% of all positives. All assays can be affected by fibrin interference, which on rare occasions, can cause high values that are not repeatable and do not fit with the clinical picture.
Interfering Proteins
The most common analytical source of spurious cTn elevations, which are more common with cTnI assays, are related to cross-reacting antibodies, the heterophilic antibodies, which are antibodies to the protein from which the assay is developed. These phenomena normally cause high values that do not change over time. They can be easily unmasked in the laboratory, where additional blocking antibodies can be added and/or with dilution studies. When there is such interference, the sample will not dilute at all until such time as the interfering substance is totally gone, and then the value will become undetectable. There also are rare cases of macrotroponemias, which are troponin–immunoglobulin complexes that can cause analytical false positive results. These have rarely been a problem with conventional assays, but could become more common with high-sensitivity assays for troponin I (hsTnI).
Impact of Blood Sampling
With the increased sensitivity of modern cTn assays, quality control of sampling is important. For example, hemolysis will decrease cTnT levels and may increase levels of cTnI with some assays. Thus, avoiding blood draws from indwelling catheters (the most common cause of hemolysis) is advised as much as possible. It is not unreasonable to ask good laboratories to quality assure values that appear peculiar to a given clinical setting or circumstance, or violate what is known about the kinetics of cTn.
Sensitivity
It is critical to understand the different levels of sensitivity of cTn assays. The best available approach to classification of the sensitivity of cTn assays is based on the number of normal individuals detected with a given assay. In this framework, assays that detect values in more than 50% of apparently normal individuals are termed high-sensitivity assays ( Figure 7-1 ). This classification framework also indicates that assays with less imprecision, that have a level of a 10% coefficient of variation (CV) ≤99th percentile should be preferred; these are called “clinically acceptable.” Assays with a CV between 10% and 20% are called usable, and assays with a CV at the 99th percentile of more than 20% are not acceptable. Assays that deliver a CV of less than 20% at the values of interest have been shown not to increase the frequency of false-positive results. Nevertheless, even greater precision of the assay improves the ability to recognize a changing pattern.
Although this framework for classification is the most commonly used, it is not perfect. There is an apparent discordance between the proportion of normal individuals with detectable levels of cTn and those with disease who have detectable concentrations. In comparative studies, the hsTnT assay that detects cTn in a relatively low proportion of normal individuals compared with other hsTn assays (see Figure 7-1 ). However, the hsTnT assay appears to detect more elevations in patients with cardiac disease than a cTnI assay that delivers detectable values in a far greater number of normal individuals. This observation suggests that clinical sensitivity and the proportion of normal individuals detected with the cTn assay are not the same. Alternative reasons for this apparent discordance include the possibility of false-positive cTnT values (e.g., due to skeletal muscle damage or false-negative cTnI values, which are caused by the previously mentioned anti-cTnI antibodies). Despite these caveats, in general, the proportion of normal individuals with detectable cTn can be used to decide the relative sensitivity of a particular assay (see Figure 7-1 ). From that perspective, the hsTnT assay appears more similar to standard “sensitive” assays now in contemporary practice than to other assays classified as high sensitivity.
The 99th Percentile Reference Limit
It is important to know the 99th percentile of a normal reference population for your local assay (see Table 7-1 ). This upper reference limit (URL) value, rather than the 97.5 percentile, which was the typical convention with most laboratory tests, was recommended when cTn was first codified in the Universal Definition of MI. This higher URL was selected to minimize the frequency of elevations that were not associated with cardiovascular pathology. In many hospital systems, cut points at concentrations higher than the 99th percentile have been arbitrarily assigned or are based on outdated benchmarks and reported as definite MI. However, use of such a higher cut point will reduce the clinical sensitivity for identifying patients with MI. For this reason, clinical providers should be familiar with the 99th percentile decision limit for the assay used locally. In addition, in knowing the 99th percentile URL for your assay, the clinician must also understand the importance of identifying a rising or falling pattern of cTn values (see the section on Definition of a Changing Pattern of Cardiac Troponin Values ).
Advanced Considerations
For assays that are not high-sensitivity assays, the 99th percentile tends to work well in practice as the diagnostic cut point for MI. However, some experts have questioned whether the 99th percentile URL is ideal or if the value should be lowered to the 97.5% value in the high-sensitivity assays currently available in Europe, and which are expected to come to the United States. This issue is a complicated one and in part depends on how the 99th percentile is defined.
The details of methods for determining the 99th percentile in the normal population are beyond the scope of this chapter, but it should be understood that the more rigorously a clinician screens for occult cardiovascular disease by history, physical examination, measurement of other biomarkers (e.g., natriuretic peptides), assessment of renal function, and cardiovascular imaging, the more apt a clinician will be able to identify a truly normal population, absent any underlying cardiovascular abnormalities. With more rigorous screening, the value for the 99th percentile of the population distribution becomes lower and lower. Thus, with high-sensitivity assays, because few manufacturers have used such rigorous approaches to screening, some confidence boundaries around the estimated 99th percentile URL will likely be necessary. As such, the reported 99th percentile values in package inserts for hsTn assays may be higher than those observed in a more intensively screened population.
For patients with unstable ischemic heart disease, the values of hsTn are likely to be high, so this issue is unlikely to influence the diagnosis of MI significantly. For more chronic disease states, it may well be that a value lower than the 99th percentile URL would be optimal for clinical decision-making. However, regardless of how these issues related to determination of the URL are viewed, the key to interpretation of hsTn is to use not only the 99th percentile value but also determine whether a changing pattern is present along with the clinical circumstances of the patient’s presentation (see the sections on Definition of a Changing Pattern of Cardiac Troponin Values and Importance of Clinical Context ).
Sex-Specific Cutoffs
Sex-specific cutoff values will be needed with hsTn assays. It has been known for some time that the frequency of elevations of cTn with conventional assays in patients with acute coronary syndrome (ACS) are much greater in men than in women ( Figure 7-2 ). With hsTn assays, it is now clear that women have lower 99th percentile URL values. The clinical relevance can be debated for the diagnosis of MI. Nevertheless, because of differences in the pathogenesis of atherosclerotic coronary artery disease between the sexes, it is likely that sex-specific cut points for cTn will be important across all clinical settings, especially in emerging clinical applications in chronic heart disease.
Definition of a Changing Pattern of Cardiac Troponin Values
The concentration of cTn is now measurable in most of the population with high-sensitivity assays (by definition), and chronic elevations of cTn are apparent in many patients with underlying structural heart disease. Therefore, the ability to define and identify changing concentrations of cTn (a delta) is important to discriminating acute myocardial injury, including MI, from chronic disease or normal. The optimal definition of a changing value continues to be debated and is a focus of ongoing investigation. However, at present, some best available practical approaches can be outlined for clinical practice.
With conventional cTn assays, a delta criterion of at least three SDs of the variation around the value has been advocated to be sure that any two values are analytically different from each other. It is from this concept that the criterion of a 20% relative change was developed. This criterion works well with conventional assays when values are elevated. However, a 20% relative change is likely too small as a delta criterion when the values are near the 99th percentile URL (see the following). For high-sensitivity assays, I recommend using a criterion of a 50% relative change when near the 99th percentile URL, which is a lower relative change criterion if the baseline value is elevated. An absolute value that is similar in magnitude to a 50% change also appears to work well near the 99th percentile, but may be superior especially when baseline values are elevated ( Figure 7-3 ).
The clinician must recognize that with imprecise assays, noise in the results may exceed these change criteria. This limitation highlights why it is important to have good precision at the 99th percentile URL. Moreover, some patients with acute MI will have a changing pattern that is smaller in magnitude. Additional important caveats to interpreting changes in cTn are described in the next section.
Challenges in Defining a Change Criterion
With conventional assays for cTn, when values are not elevated, imprecision is much greater and the 20% relative change criterion will be too small and must be individualized for each assay. Nevertheless, assays that do not have an imprecision at less than 10% at the 99th percentile URL can still be used and do not cause false-positive elevations.
With high-sensitivity assays, determination of the optimal criteria for a changing value has become much more complicated. With the ability to measure values in normal individuals, biologic variation has an impact on this evaluation. When biologic variation is taken into account, it is clear that the reference change value (RCV), which is the value where one can be sure that the values are different, increases from 50% to 85%. For some assays, to increase complexity, imprecision depends not only on the reagents but on the piece of equipment used to make the measurement. Although a criterion of a 50% relative change or an absolute value that is similar appears to work fairly well around the upper range of normal, when cTn values are substantially elevated, which usually means patients have presented late after the onset of acute MI, a delta may not be found at all if the concentrations are around the peak values or on the downslope of the curve. In addition, once cTn values are elevated, the relative (percent) rise is not nearly as robust because values may not be able to increase further. Therefore, in those circumstances, I recommend use of an absolute criteria or a lower percentage (e.g., 20%). In my opinion, it is likely that a change criterion based on the absolute concentration will turn out to be optimal for diagnosis when baseline cTn values are elevated.
As indicated previously, there will be a group of patients in whom some clinical judgment will be necessary. At present, many of the approaches advocated for defining changing values have been predicated on data collected by convenience registries, which are less rigorous than ideal. In addition, some of the criteria are based on a flawed approach of using validated changes over 6 hours and extrapolating those changes over shorter periods of time. cTn release is blood flow–dependent, and the assumption that cTn release is consistent assumes that blood flow does not change over time. That assumption is not proven and likely incorrect. In addition, some of the delta criteria are based on such small values that even the best assays may not be capable of providing sufficiently accurate values. Thus, caution is suggested with the use of these approaches, especially in patients who present early after the onset of symptoms. Nonetheless, with hsTn assays, the preceding suggestions are likely to be valid even in more rigorous studies. However, it is likely that the optimal criteria will be assay-dependent. Finally, the use of any criteria less than the RCV will result in a tension between sensitivity and specificity. Using delta criteria less than the RCV will include some patients who do not have acute ischemic heart disease, but who may have other acute diseases that can cause elevations or more chronic disease with myocyte injury (see the section on Situations That May Be Confused with Myocardial Infarction ). Decisions about how to manage these issues in practice should be made conjointly by emergency department (ED) physicians, cardiologists, and laboratorians.
Sampling and Reporting
To provide a consistent approach to measurement of cTn, a set sampling interval is recommended. The optimal sampling interval between the first 1 and 3 hours differs from one that might be advocated for the period between 3 and 24 hours after presentation with suspected MI. I favor a strategy of 0, 3, and 6 hours, an approach that is recommended in current professional guidelines. Moreover, it is very helpful to have laboratory reports indicate whether a changing pattern of cTn values is present. In addition, to avoid difficulties associated with large numbers of zeroes, as assay sensitivity improves, whole numbers, (generally, in units of nanograms per liter or picograms per milliliter) should be used rather than complex decimal type values for hsTn assays.
The Importance of Clinical Context
Elevations in cTn, although indicative of myocyte injury, are not solely caused by ischemic heart disease. There are a huge number of patients with increased values that reflect other cardiac pathophysiological disturbances ( Table 7-3 ; see also Chapter 6 ). For example, left ventricular hypertrophy (LVH) increases the amount of troponin per gram of myocardium, and therefore, is associated with higher values. Metabolic perturbations, such as those that occur with renal failure, often in association with LVH, are also associated with elevations in cTn that are not necessarily caused by ischemic heart disease. Similarly, in acute circumstances such as sepsis, the elaboration of a variety of cytokines and tumor necrosis factor can cause elevations. In addition, drug toxicities, such as carbon monoxide poisoning and cardiotoxic chemotherapy (e.g., adriamycin and herceptin) cause elevations that are likely caused by injury to myocytes. Therefore, proper interpretation requires that the cause of any given increase of cTn be considered within the clinical context in which it occurs.
