Troponin testing: Analytical considerations

INTRODUCTION — Cardiac injury occurs when there is disruption of normal cardiac myocyte membrane integrity. This results in the loss into the extracellular space (including blood) of intracellular constituents, including detectable levels of a variety of biologically active cytosolic enzymes and structural proteins, referred to as biomarkers, such as troponin, creatine kinase, myoglobin, and lactate dehydrogenase. When a sufficient number of myocytes have died, be it due to necrosis, apoptosis, or some other mechanism, troponin (and other biomarkers) is detectable in the blood [1]. (See "Troponin testing: Clinical use", section on 'Diagnosis of acute MI'.)

Injury may be acute or chronic, and it is usually considered irreversible (cell death). The causes of cardiac injury are numerous. Ischemia, consequent to an imbalance between the supply of and demand for oxygen (and nutrients), is one common cause. Other causes of myocardial injury are included in a table (table 1).

This topic will provide a basis for understanding troponin and aspects of measuring troponin in the blood. The clinical use of troponin testing is discussed separately. (See "Troponin testing: Clinical use".)

Other relevant topics include:

●(See "Use of creatine kinase to detect myocardial injury".)

●(See "Elevated cardiac troponin concentration in the absence of an acute coronary syndrome".)

WHAT IS TROPONIN — Cardiac troponin I (cTnI) and T (cTnT) are cardiac regulatory proteins that control the calcium-mediated interaction of actin and myosin [2]. Both have early releasable and structural pools, with most troponin being present in the structural pool [3,4]. (See "Excitation-contraction coupling in myocardium", section on 'Role of tropomyosin and troponins'.)

Troponins are products of specific genes and therefore have the potential to be unique to the heart (ie, specific cardiac isoforms). Studies performed with cTnI have failed to find any cTnI outside of the heart at any stage of neonatal development [5,6]. In contrast, cTnT is expressed to a minor extent in skeletal muscle. Data indicate that there are at least some patients with chronic skeletal muscle disease who have proteins that are detected by the antibodies in the cTnT and high-sensitivity cTnT assay. This implies that skeletal muscle can, in some patients, be the source for elevations of cTnT detected in the blood [7]. Some data suggest this may be more common than was originally thought [8]. In most (but not all) of these patients, the typical rise and fall of cTnT with myocardial infarction is not seen. In addition, chronic skeletal muscle disease should be suspected as a cause for the cTnT increase if a cTnI measurement for the same patient is within the normal range.

For most clinical purposes, it appears that cTnT and cTnI have equal utility except in patients with renal failure where there are more elevations of cTnT than of cTnI [9]. With high-sensitivity assays, almost 100 percent of dialysis patients will have an increased cTnT; a lesser percentage will have increases in cTnI [9]. Such increases are associated with an adverse clinical prognosis [9].

HOW IS TROPONIN MEASURED? — Almost all assays for cTn are enzyme-linked immunosorbent assays in which there is an antibody that captures the material and then a tag antibody that labels it. In most assays, the capture antibodies are monoclonal antibodies that are specific for the cTn being measured, either cardiac troponin I (cTnI) or cardiac troponin T (cTnT) [10,11]. Often, more than two antibodies are used in order to increase the amount of protein captured and labelled. Each assay is different because the antibodies used in the assays are different. For this reason, as well as differing detection methods and calibration, the values from one assay are not harmonized with other assays. One cannot substitute the value from one assay for another. This issue is particularly evident for cTnI assays because of the large number of assays available commercially. In contrast, because of patent protections, only a single assay is available for cTnT in the United States. When the patent protection ends, it is anticipated that additional and different assays for high-sensitivity cTnT (hs-cTNT) will become available.

The detailed analytical characteristics of all cTn assays can be found on the International Federation of Clinical Chemistry web page at http://www.ifcc.org/ifcc-education-division/emd-committees/committee-on-clinical-applications-of-cardiac-bio-markers-c-cb/.

TERMINOLOGY — The following terms are used in discussions of the analytical aspects of troponin testing:

●Coefficient of variation is a measure of reproducibility (or precision) of the assay and is calculated as the ratio of the standard deviation over the mean value for repeated testing of the same sample. Ideally, this value should be measurable with a coefficient of variation of 10 percent or less [12]. Importantly, good analytical precision allows for detection of a smaller magnitude of changing values that enhance diagnostic specificity.

●Limit of detection is the lowest value of cardiac troponin (cTn) that can be measured as one progressively dilutes a sample. The limit of detection has been used in a variety of studies as a threshold for ruling out myocardial infarction [13]. Stated another way, it is the lowest detectable cTn concentration reliably distinguished in a sample containing low cTn concentration [14]. This value has taken on additional importance because criteria for a high-sensitivity (hs) assay now require that at least 50 percent of men and 50 percent of women have a detectable value above this level [11].

It should be understood that at this concentration value, there is substantial analytical variability, ie, noise that makes it problematic to discriminate small differences or changes.

●Limit of quantitation is the concentration at which the coefficient of variation for the assay is <20 percent. The more sensitive and precise an assay is, the lower the limit of quantitation will be. However, for all hs-cTn assays, it will be higher than the limit of detection. Presently, in the United States, available high-sensitivity assays are approved by the US Food and Drug Administration only to report values at or above the limit of quantitation.

●Limit of the blank is an assessment of noise in the system. It is the lowest value that one can distinguish from zero and is estimated as the upper 95^th percentile of observed values from repeated measurements of a blank sample. Stated another way, it is the highest apparent cTn concentration expected when multiple tests of a sample that has no cTn are performed.

●Upper reference limit (URL) is the upper boundary of a normal population. For cTn, that is defined as the 99^th percentile of the normal distribution, which is roughly 3 standard deviations from the mean value. Thus, 1 percent of normal subjects may have a value that is above the 99^th percentile URL (see below).

INTERPRETING TROPONIN RESULTS — The original cardiac troponin (cTn) assays were relatively insensitive with regard to their ability to detect the presence of troponin. Typically, results were reported in ng/mL or mcg/mL. As sensitivity was increased, the number of leading zeros that would be necessary to report very low detectable values has increased substantially. Accordingly, it is recommended that values be reported in ng/L for high-sensitivity assays (1000 ng/mL = 1 ng/L) [11].

Although, in an ideal world, hs-cTn assays would simply have the same cut-off values as standard assays and provide similar values, this is not the case. For hs-cardiac troponin T (hs-cTnT), as an example, a value with the fourth-generation assay of 0.01 ng/mL corresponds to a value of 30 ng/L with the hs-cTnT assay, and a value of 0.03 ng/mL to a value of 52 ng/L [15,16]. However, values above 100 ng/L with the hs-cTnT assay are well correlated with values with the fourth generation except for the change in units from ng/L with hs-cTnT compared with ng/mL for the standard assay. Thus, a value of 0.1 ng/ml is equivalent to a value of 100 ng/L with the hs-cTnT assays. For most of the hs-cardiac troponin I (hs-cTnI) assays, new cut-off values that differ from the prior generation assays must be used. The only known exception is the Abbott hs-cTnI assay for which the 99^th percentile upper reference limit (URL) remains the same as the prior iteration [17].

Defining a meaningful delta — Knowing whether one value is different from another depends on the ability to be sure that the apparent change in values is not simply a consequence of analytic and biologic variation. Biological variation can be assessed accurately only with high-sensitivity assays, and the assessment presumes normal physiology [18]. Abnormal physiology might markedly influence the amount of change present. Conjoint analytic and biologic variation can be measured and when done so, short-term (hours) biologic and analytic variability range somewhere between 40 and 60 percent for hs-cTn assays. Longer-term analytical variation can be higher [19]. Thus, in order to know with certainty that an individual has had a true change in cTn values, a difference exceeding 40 to 60 percent is necessary.

Clinically, the situation is more complex, particularly in relation to possible acute myocardial infarction [20]. Use of such criteria for relative (percentage) change among patients with possible acute myocardial infarction does not provide optimal sensitivity for detection of acute cardiac injury. In that circumstance, fixed absolute thresholds for change criteria are preferable and preserve sensitivity when baseline values are already elevated [20]. Once values are elevated, they are not likely to increase by large percentages, and less marked changes are required to preserve the sensitivity for diagnosis. Some have suggested lowering the percentage change to 20 percent. This may be the reason why absolute values have been shown to be better because the higher the value, the lower the percent change with an absolute cut-off. Obviously, this becomes excessive once values are significantly elevated (assay dependent) [20,21].

Taking these issues into account, we recommend the use of a fixed absolute threshold that will vary for each assay. Evidence-based, assay-specific values should be used when available. In most cases, a delta value that is approximately 50 percent of the 99^th percentile URL has been shown to be a reasonable absolute delta criterion at low values (below the 99^th percentile URL). At higher baseline values, when one uses an absolute threshold, the percentage change becomes less. This is necessary to preserve sensitivity, as indicated above. For example, with the hs-cTnT assay, this would yield an absolute delta criterion of 7 ng/L. Investigation of the optimal delta thresholds remains intense and thus this approach may evolve with new data.

Normal range — With the use of hs-cTn assays, all healthy ("normal") individuals have small but detectable levels of troponin in their blood [22]. (See 'Defining a high-sensitivity assay' below.)

With conventional (older) assays, apparently healthy individuals with values above the 99^th percentile URL (see 'Terminology' above) are at increased cardiovascular risk; such individuals have higher cardiovascular event rates than those with values <99^th percentile. Aggregate studies of "predictors" of troponin elevation in this population suggest that such individuals harbor cardiac comorbid conditions (eg, left ventricular hypertrophy) that may or may not have been detected. For example, it has been shown that having a detectable level of troponin with a conventional assay defined a group of stable patients likely to have either significant coronary artery disease or elevated filling pressures [23]. Even within the normal range of these hs assays, it appears that the higher the value, the greater the risk [24-27]. This suggests that each individual has his/her own normal baseline and that elevations above that baseline occur as cardiac disease ensues and thus signify increased risk.

Conventional (older) assays are not capable of accurately defining such values below the 99^th percentile value due to their significant imprecision. With hs-cTn assays, there will be more gradation in risk as values rise, even within the normal range [26]. As with most risks, there will be a gradient that integrates other characteristics of the individual (eg, age, diabetes mellitus, renal function, ventricular hypertrophy).

One of the major issues with hs-cTn assays is how to best determine the 99^th percent URL. Manufacturers often do not perform studies under ideal circumstances. For example, the hs-cTnT assay was initially reported to detect values in roughly 80 percent of normal subjects [15]. However, in a study based on a population including abnormal subjects where one might anticipate fewer undetectable values [26], only 25 percent of the population had detectable values and in another study, only 34 percent were detected [22]. These differences in the proportion of patients with measurable cTn likely reflect differences in the age of the samples, the specific equipment used for the analysis, and the populations studied [28]. Some studies screen only with questionnaires, some with additional biomarker testing, and some with imaging. The more extensive the screening for underlying cardiac abnormalities, the lower the 99^th percent URL [29-31].

Studies have confirmed the early diagnostic and therapeutic value of using the 99^th percentile URL for cTn, as well as the diagnostic value of using high-sensitivity assays [32-35]. (See "Non-ST-elevation acute coronary syndromes: Selecting an approach to revascularization" and "Initial evaluation and management of suspected acute coronary syndrome (myocardial infarction, unstable angina) in the emergency department" and "Acute ST-elevation myocardial infarction: Selecting a reperfusion strategy".)

In 1999, a decision was made that the 99^th percentile of normal values would be used to define an abnormal cTn value [36]. This 99^th percentile URL is higher than the 97.5^th percentile that is used for most laboratory tests. However, given the greater sensitivity of cTn assays than their predecessor, creatine kinase MB, there was concern about including 2.5 percent of the population as abnormal, which would have been the case with a URL at the 97.5^th percentile. Eventually, with hs-cTn assays, it has become clear that very few of the values detected by most standard assays are in the normal range when established in apparently healthy individuals using an hs-cTn assay. Accordingly, for most non-high-sensitivity cTn assays, the finding of any detectable value is abnormal [22].

It is also now clear that there are significant differences in the 99^th percentiles between men and women. It has been hypothesized that this difference is explained by a lower cardiac mass in women; however, the reason for this difference is not yet clear. In addition, there is debate as to whether sex-specific cut-offs improve clinical performance [37]. Most studies of this issue have focused on diagnosis of acute myocardial infarction. In that setting with hs-cTnT, it has been difficult to see a clear signal of superiority using sex-specific cut-offs [38], though other data have challenged those findings [39]. In other studies with hs-cTnI with a focus on this issue, use of sex-specific cut-offs improved the identification of women with disease [40], although outcomes have not been improved, perhaps because the newly diagnosed patients were not evaluated and treated. Thus, there is a need for outcome data in this area. In addition, when one begins to consider the use of cTn for chronic indications (which is one of the areas for potential emerging applications of hs assays), sex-specific cut-offs are important. For that reason, and because of anticipation of additional data from better-designed studies in this area, most guidelines recommend the use of sex-specific cut-offs for the 99^th percentile URL for cTn assays [11,37].

Defining a high-sensitivity assay — We prefer highly sensitive (newer) to sensitive (older) tests in most clinical situations (see "Troponin testing: Clinical use"). This section will provide the background for that recommendation.

Criteria for deciding which assays are high-sensitivity and which ones are not were proposed some years ago [11,41]. There are very few, if any, well-done comparative studies between assays. For that reason, it is almost impossible to compare values from one assay with another. In addition to the fact that the assays are not harmonized, the calibrations of the assays vary as well. One reasonable approach is to label an assay as high-sensitivity if it can detect values, ie, values above the level of detection (see 'Terminology' above) in at least 50 percent of normal individuals (see 'Normal range' above) [42]. There are some controversies about which assays are high-sensitivity assays and which ones are not, based on the number of normal subjects in whom values are detected. Some assays, like hs-cTnT, have been on the margin of the 50 percent benchmark, and have, in most studies, not detected more than 50 percent [43]. Finally, it is not at all clear that the designation of high-sensitivity based on these consensus metrics is a totally accurate reflection of clinical sensitivity. For example, hs-cTnT often detects more elevations in cTn than many of the hs-cTnI assays [31,44,45]. The reasons for this discordance are not clear.

In addition, the mechanism of release is of some debate. It is clear that all comorbidities associated with cardiovascular disease such as hypertension, hyperlipidemia, and diabetes raise high-sensitivity values at least modestly [46]. There is a variety of mechanisms that have been proposed for these effects [47], such as the extrusion of troponin in the absence of cell death, but it is also clear that some physiologic increases of troponin can be due to cell death and release due to apoptosis, and that this can be part of normal physiology [1,48]. Thus, rapid atrial pacing elaborates minor amounts of troponin, even in normal subjects [49]. Infusion of dobutamine also can lead to release of cTn [50]. This finding could be due to one of these alternative mechanisms for cardiac release or due to apoptotic cell death. Recent experimental data suggest that brief periods of ischemia cause the release of cTnI, and the cells die due to apoptosis [51]. The concept that cells were dying was not proposed as a possibility in the past because it was thought that cardiac myocytes could not replenish themselves, and therefore if cells died, one would end up eventually running out of cardiomyocytes. We now know this may not be true and that like skeletal muscle, there may be reparative processes [52].

Interestingly, alcohol is associated with lower values, as is smoking [46].

Effects of adopting a high-sensitivity assay — In most scenarios, adoption of a hs-cTn assay with a cut-off at the 99^th percentile URL will result in detection of more patients with MI when compared with continued use of a conventional assay. However, the increase in positive test results will depend primarily on the sensitivity of your current conventional assay and the cut-off values that you use. Many centers still do not use the recommended 99^th percentile URLs and, instead, use higher threshold cut-off values [53,54]. If one uses insensitive assays or a higher 99^th percentile URL, the increase in MI diagnoses will be far greater than if one has been using a sensitive assay prior to moving to a hs-cTn assay [55,56].

Assay false positives and false negatives — There are a variety of reasons for false-positive and false-negative cTn results.

The most common causes of a false-positive assay are cross-reacting and heterophilic antibodies [57] (see "Infectious mononucleosis", section on 'Pathogenesis'). Heterophilic antibodies are antibodies to the immunoglobulin G material from which the antibody comes. These antibodies are rare, occurring in <0.5 percent of individuals. Interference from heterophilic antibodies is slightly more common with cTnI assays and less common with high-sensitivity assays. In addition, false positive results may be caused by macro troponin complexes (eg, immunoglobulin-cardiac troponin complexes) [57,58]. When one suspects an elevated value due to an interfering substance, the laboratory can employ a variety of approaches to address the issue. The first is the use of additional blocking antibodies that block the cross-reacting and heterophilic antibodies. This approach will not rule out a macrokinase. Exclusion of a macrokinase requires column separation to remove the complex immunoglobulins and troponin. However, both are often unmasked by dilution studies. When one has an interferent, the reported concentration from the sample does not dilute until the interferent is gone, at which point in time the values fall precipitously. This approach can be taken by all laboratories. In addition, there is a rare skeletal muscle false-positive with cTnT [7,57,59] that likely occurs due to the re-expression of fetal isoforms in response to skeletal muscle disease. The frequency of false-positive results due to skeletal muscle disease is controversial but usually should not confound the diagnosis of myocardial infarction because the values are stable rather than dynamic. Additional studies are necessary to determine whether the differences between normal value studies where hs-cTnT detects far fewer normal individuals than hs-cTnI assays, and clinical studies where there are more elevations of hs-cTnT, are due to this particular cross-reactivity. Alternatively, it could be due to high values for the 99^th percentile of hs-cTnI assays or cross-reacting antibodies that reduce values, so-called anti-TnI antibodies [57,60].

It should also be appreciated that since many antibody-based troponin assays use biotinylation, the ingestion of large amounts of biotin can interfere with the assays themselves [61,62]. In general, troponin values are reduced, but there have been instances in which they were elevated. Biotin is cleared rapidly from blood and is usually not detectable for six to eight hours in the absence of renal failure. Biotin is increasingly being consumed by patients for a variety of reasons and has become part of an increasing number of vitamins and dietary supplements (see "Overview of water-soluble vitamins", section on 'Biotin'). Often, patients are unaware of how much biotin they may be taking.

One published systematic evaluation of the effects of biotin on troponin levels used Roche assays, which are affected variably by biotin [63]. The frequency of biotin elevations and their impact on high-sensitivity troponin assays is being explored in ongoing studies. The clinical impact of these observations is discussed separately. (See "Troponin testing: Clinical use", section on 'Unanticipated results'.)

Falsely low values can also occur with hemolysis with hs-cTnT and with some non-high-sensitivity cTnI assays [64], and is a major problem if one does not have good sample quality. In addition, there are anti-troponin antibodies that were initially thought to be specific for cTnI. It is now clear that they bind to a complex composed of troponin T, troponin I, and troponin C, which is one of the multiple fragments of troponin that is released [60]. Therefore, it is likely that they modestly reduce cTnI as well as cTnT. It is usually the case that they do not change diagnostic signals because the complex of troponin T, troponin I, and troponin C is only one of a series of proteins that are released in response to cardiac injury.

Each commercially available troponin assay is variably sensitive to these sources of error, and this variability may cause some of the disagreement between troponin assays [65].

SUMMARY AND RECOMMENDATIONS

●Troponin values within the normal range likely come from a mixture of truly normal individuals (with detectable values) and individuals with comorbidities reflected by low but detectable values. (See 'Normal range' above.)

●Troponin elevations (above the 99^th percentile) can be due to structural heart disease in the absence of any acute process. (See "Troponin testing: Clinical use", section on 'Diagnosis of acute MI'.)

●For most clinical purposes, it appears that cardiac troponin T and I have equal utility. (See 'What is troponin' above.)

●Troponin assays cannot be substituted one for another; the values of the various cardiac troponin assays are not calibrated to agree with each other. Thus, it is important to know the established 99^th percentile upper reference limit for the relevant assay(s) used in your institution. Evidence-based, assay-specific values and change ("delta") criteria should be used when evaluating patients with chest pain. (See 'Normal range' above and "Troponin testing: Clinical use".)

●The sensitivity of present-day assays vary markedly from insensitive assays (many of the point-of-care assays) to highly sensitivity assays (hs-cTn). (See 'Defining a high-sensitivity assay' above.)

●With the use of hs-cTn assays, most healthy ("normal") individuals have small but detectable levels of troponin in their blood. (See 'Normal range' above.)