ACCF 2012 Expert Consensus Document on Practical Clinical Considerations in the Interpretation of Troponin Elevations: A Report of the American College of Cardiology Foundation Task Force on Clinical Expert Consensus Documents

Robert A. Harrington, MD, FACC, Chair⁎⁎Former Task Force member during this writing effort.

Deepak L. Bhatt, MD, MPH, FACC, Co-Chair

Jeffrey L. Anderson, MD, FACC

Eric R. Bates, MD, FACC⁎⁎

Charles R. Bridges, MD, MPH, FACC⁎⁎

Mark J. Eisenberg, MD, MPH, FACC⁎⁎

Victor A. Ferrari, MD, FACC

John D. Fisher, MD, FACC

Mario J. Garcia, MD, FACC

Timothy J. Gardner, MD, FACC

Federico Gentile, MD, FACC

Michael F. Gilson, MD, FACC

Adrian F. Hernandez, MD

Mark A. Hlatky, MD, FACC⁎⁎

Alice K. Jacobs, MD, FACC⁎⁎

Sanjay Kaul, MBBS, FACC

Jane A. Linderbaum, MS, CNP-BC, AACC

David J. Moliterno, MD, FACC

Debabrata Mukherjee, MD, FACC⁎⁎

Robert S. Rosenson, MD, FACC⁎⁎

James H. Stein, MD, FACC⁎⁎

Howard H. Weitz, MD, FACC

Deborah J. Wesley, RN, BSN⁎⁎

• Preamble…...2429

• 1Introduction…...2429
  • 1.1Document Development Process…...2429
    • 1.1.1Writing Committee Organization…...2429
    • 1.1.2Document Development Approval…...2429
  • 1.2Conceptual Model…...2430
• 2Interpretation…...2430
  • 2.1Analytical Issues…...2432
  • 2.2Statistical Issues…...2433
• 3Troponins in Acute Coronary Syndromes…...2435
  • 3.1Impact of Improved Sensitivity Troponin Assays…...2436
• 4Non-ACS Ischemic Troponin Elevations…...2437
• 5Troponins in PCI and CABG…...2438
  • 5.1Biomarkers With PCI Procedures…...2438
  • 5.2Biomarkers With CABG Procedures…...2439
• 6Troponins in Nonischemic Clinical Conditions…...2439
  • 6.1Nonischemic Conditions With a Current or Potential Clinical Role for Troponin Measurement…...2440
    • 6.1.1Heart Failure…...2440
    • 6.1.2Pulmonary Embolism…...2441
    • 6.1.3Chronic Kidney Disease…...2441
    • 6.1.4Sepsis…...2442
    • 6.1.5Chemotherapy-Associated Cardiac Toxicity…...2442
    • 6.1.6Assessing Cardiotoxicity in Drug Development…...2443
• 7Other Nonischemic Conditions in Which Interpretation Creates Clinical Uncertainty…...2443
  • 7.1Infection and Myocarditis…...2443
  • 7.2Myopericarditis…...2444
• 8Summary and Overview of Recommendations…...2444

  Appendix 1. Relevant Author Relationships With Industry and Other Entities…...2453

  Appendix 2. Relevant Peer Reviewer Relationships With Industry and Other Entities…...2454

  Appendix 3. Additional Nonischemic Syndromes in Which Troponin Testing May Have Potential Clinical Application…...2457

  • 1.1Amyloidosis…...2457
  • 1.2Cardiac Transplant Monitoring…...2457
  • 1.3Blunt Cardiac Injury…...2457
  • 1.4Noncardiac Surgery…...2458
  • 1.5Thermal Injury…...2458

  Appendix 4. Other Nonischemic Conditions in Which Interpretation of Troponin Testing Has Created Clinical Uncertainty…...2458

  • 1.1Subarachnoid Hemorrhage…...2458
  • 1.2Stroke…...2459
  • 1.3Endocarditis…...2459
  • 1.4Cardiac Tumors and Systemic Malignancies…...2459
  • 1.5Hematologic Conditions…...2459
  • 1.6Neuromuscular and Myopathic Conditions…...2459
  • 1.7Autoimmune and Connective Tissue Diseases…...2460
  • 1.8Arrhythmia Treatments and Resuscitation…...2460
  • 1.9Metabolic Disorders…...2461
  • 1.10Chronic Obstructive Lung Disease…...2461
  • 1.11Autonomically Mediated Disorders…...2461
  • 1.12Pregnancy and Related Conditions…...2461
  • 1.13Strenuous Exercise…...2462
  • 1.14Rhabdomyolysis…...2462
  • 1.15Aortic Dissection…...2462

  Appendix 5. Abbreviation List…...2463

  ACCF Staff List…...2445

  Key Words…...2453

  References…...2445

This document has been developed as an Expert Consensus Document (ECD) by the American College of Cardiology Foundation (ACCF), American Association for Clinical Chemistry (AACC), American College of Chest Physicians (ACCP), American College of Emergency Physicians (ACEP), American College of Physicians (ACP), American Heart Association (AHA), and Society for Cardiovascular Angiography and Interventions (SCAI). Expert Consensus Documents are intended to inform practitioners, payers, and other interested parties of the opinion of ACCF and document cosponsors concerning the evolving areas of clinical practice and/or technologies that are widely available or new to the practice community. Topics chosen for coverage by ECDs are so designed because the evidence base, the experience with technology, and/or clinical practice are not considered sufficiently well developed to be evaluated by the formal ACCF/AHA Practice Guidelines process. Often the topic is the subject of considerable ongoing investigation. Thus, the reader should view the ECD as the best attempt of the ACCF and document cosponsors to inform and guide clinical practice in areas where rigorous evidence may not yet be available or evidence to date is not widely applied to clinical practice. When feasible, ECDs include indications or contraindications. Some topics covered by ECDs will be addressed subsequently by the ACCF/AHA Practice Guidelines Committee.

The ACCF Task Force on Clinical Expert Consensus Documents (TF CECD) makes every effort to avoid any actual or potential conflicts of interest that might arise as a result of an outside relationship or personal interest of a member of the writing panel. Specifically, all members of the writing panel are asked to provide disclosure statements of all such relationships that may be perceived as relevant to the writing effort. This information is documented in a table, reviewed by the parent task force before final writing committee selections are made, reviewed by the writing committee in conjunction with each conference call and/or meeting of the group, updated as changes occur throughout the document development process, and ultimately published as an appendix to the document. External peer reviewers of the document are asked to provide this information as well. The disclosure tables for writing committee members and peer reviewers are listed in (12) and (13), respectively, of this document. Additionally, in the spirit of complete transparency, writing committee members' comprehensive disclosure information—including relationships with industry and other entities that do not pertain to this document—are available online. Disclosure information for members of the ACCF Task Force on Clinical Expert Consensus Documents—as the oversight group for this document development process—is also available at www.cardiosource.org/ACC/About-ACC/Leadership/Guidelines-and-Documents-Task-Forces.aspx.

The work of the writing committee was supported exclusively by the ACCF without commercial support. Writing committee members volunteered their time to this effort. Meetings and/or conference calls of the writing committee were confidential and attended only by committee members.

Robert A. Harrington, MD, FACCChair, ACCF Task Force on Clinical Expert Consensus Documents

Deepak L. Bhatt, MD, FACCVice Chair, ACCF Task Force on Clinical Expert Consensus Documents

The writing committee was commissioned by the ACCF TF CECD and consisted of members representing 7 societies: ACCF, AACC, ACCP, ACEP, ACP, AHA, and SCAI. Prior to the commencement of the writing process, authors reported all relevant relationships with industry within the previous 24 months. Authorship reflects 1 chair and 5 additional members with no relevant RWI. Relationships were managed in accordance with the disclosure policy in place as of September 2009, as noted in the Preamble. The ACCF disclosure policy was subsequently revised, but it did not apply to this writing effort, which was already in progress. Coordination and staff support were provided by the ACCF.

The writing committee convened by conference call and e-mail to finalize the document outline, develop the initial draft, revise the draft per committee feedback, and ultimately sign off on the document for external peer review. All participating organizations participated in peer review, resulting in 22 reviewers representing 170 comments. Comments were reviewed and addressed by the writing committee. A member of the ACCF TF CECD served as lead reviewer to ensure that all comments were addressed adequately. Both the writing committee and TF CECD approved the final document to be sent for board review. The ACCF Board of Trustees reviewed the document, including all peer review comments and writing committee responses, and approved the document in July 2012. This document is considered current until the TF CECD revises or withdraws it from publication.

Since the introduction of troponin testing in the early 1990s, there have been questions about the relationship between the physiological finding of elevated troponin as a marker of myocardial necrosis and the clinical significance of the finding and the nomenclature that should be attributed to it. This early experience with testing clearly demonstrated that an elevated troponin level identified patients at increased risk for adverse outcomes whether the clinical diagnosis was unstable angina, myocardial infarction (MI), or a noncoronary etiology. This experience taught us much about both analytical and clinical sensitivity and specificity of cardiac markers, especially in the absence of a putative operational standard (clinical, imaging, or laboratory), against which the clinical states of unstable angina and MI could be defined.

As troponin assays become more sensitive, the issues plaguing clinicians will become more frequent and more complex. Although there are substantial discussions related to assay characteristics (e.g., sensitivity, precision, and reference limits), the distinction between diagnosis and prediction of risk in populations, and the consequences of false-positive and false-negative results, the common focus is on improving patient care and outcomes.

What has become extremely clear is that much of the interpretation of the test results must consider the clinical context in which the measurement was made. For example, the interpretation of a positive troponin in a patient presenting with ischemic chest pain will (and must) be different from that in the patient undergoing a procedure or presenting with acute onset dyspnea, fever and hypotension, or renal failure. Furthermore, there is an increasing appreciation for the nonischemic versus ischemic etiologies of troponin release, and for the latter, an appreciation for differentiating the nuances of acute coronary syndromes (ACS) versus non-ACS etiologies. The most important nuance to understand is that an elevated troponin is a finding that represents the likely occurrence of myocardial necrosis and does not in and of itself provide any indication of the etiology.

(Figure 69_gr1) shows a conceptual model for clinical distribution of elevated troponin. The key to understanding this concept is to appreciate that not all elevated troponin results represent an MI and that not all myonecrosis results from an ACS event, even when ischemic in etiology. Although the finding of an elevated troponin carries an increased risk for subsequent adverse clinical outcomes in many circumstances, inappropriate treatments driven by troponin elevation alone could impart an even higher risk.

More than 30 years ago, a landmark Joint Report from the International Society and Federation of Cardiology and World Health Organization defined the criteria for the diagnosis of ischemic heart disease (1). In this report, the diagnosis of acute MI was based on a consensus of 2 of the following 3 features: 1) clinical history; 2) electrocardiographic findings; and 3) temporal changes in serum enzymes. A diagnosis based on a consensus was necessary due to the heterogeneity of clinical symptoms at presentation, the fact that the ECG is frequently equivocal, and because enzyme biomarkers available at the time were not specific for myocardial injury. However, in the early 1990s, the situation changed with the development of cardiac troponin T and I assays. Initial studies showed that with the exception of rare analytical false positives (2), the presence of cardiac troponin in blood indicated that cardiac injury had occurred. Therefore, clinicians rapidly came to consider cardiac troponin biomarkers to have virtually 100% predictive accuracy for MI.

Although early cardiac troponin assays were considered as a replacement test for creatine kinase (CK)-MB measurement, equivalence between the markers could not be demonstrated. In 12% to 39% of patients who were negative for CK-MB, cardiac troponin results were positive (3). These data raised the question as to whether discordant troponin and CK-MB results were falsely positive or indicative of a more sensitive test that classified patients more accurately. Subsequent meta-analyses answered this question by showing that patients with positive troponin results indeed had a higher risk for adverse outcomes ((4),5) even in the absence of recurrent ischemic injury.

The question then became, what cutoff should be used for the diagnostic and prognostic interpretation of cardiac troponin? Several studies indicated that even minor elevations in cardiac troponin were associated with an increased risk in patients within the continuum of ACS ((6),(7),(8),9). The notion that any amount of myocardial necrosis caused by ischemia should be labeled as MI, and the evolution of sensitive and specific technologies, including cardiac troponin assays, necessitated the re-evaluation of established definitions of MI (10). On the heels of a declaration made by the National Academy of Clinical Biochemistry (NACB) in reference to the need for incorporation of troponin into the diagnosis on MI (11), a joint committee of the European Society of Cardiology (ESC) and the American College of Cardiology Foundation was convened in 1999 to re-examine the MI definition. The result was a consensus that the preferred biochemical marker for detecting myocardial necrosis was cardiac troponin and that a maximal concentration of troponin T or I that exceeded the operative threshold on at least 1 occasion during the first 24 h after an index clinical ischemic event indicated MI (12). The operative threshold was defined as the 99th percentile of the values for a reference control group and was based on the consensus that an acceptable false-positive rate would be approximately 1%.

The emerging role of cardiac troponin as a powerful tool for MI diagnosis and risk stratification led professional organizations to issue guidance statements on its usage. From the laboratory medicine perspective, the NACB recommended cardiac troponin as the preferred marker for risk stratification of suspected ACS patients and for establishing the diagnosis of MI (11). A low cutpoint at the 99th percentile of a reference control population was championed by the NACB guidelines, in agreement with the earlier statement by the ACC/ESC/AHA on redefinition of MI (10).

In 2007, a second global task force comprised jointly of representatives from the ESC, ACCF, AHA, and WHF was convened to update the 2000 consensus document on redefinition of MI (12). This task force concluded that the term MI should be used when there is evidence of myocardial necrosis in a clinical setting consistent with myocardial ischemia and in association with the following criteria for diagnosis of MI: 1) A rise and/or fall of biomarkers (preferably troponin); 2) sudden cardiac death; 3) elevations in biomarkers after percutaneous coronary intervention (PCI) in patients having normal pre-intervention troponin levels; 4) elevations in biomarkers in patients following coronary artery bypass grafting (CABG) and with normal baseline troponin levels; or 5) pathological findings of an acute MI. This document defined MI according to 5 classifications, as shown in (Figure 69_gr2): Type 1 is termed spontaneous MI, which is related to ischemia due to a primary coronary event such as plaque rupture, erosion/fissuring or dissection; Type 2 is ischemia related to either increased oxygen demand or decreased supply; Type 3 is related to sudden unexpected cardiac death; Type 4a is associated with PCI, and 4b is associated with documented stent thrombosis; and Type 5 is associated with CABG. A major refinement in the 2007 document was the stipulation that a rise and/or fall of cardiac biomarkers (preferably troponin) is necessary. Not included in the earlier global task force document, but consistent with the NACB guidelines, this stipulated rise and/or fall mandates serial sampling of troponin in all patients suspected of having an acute spontaneous (Type 1) MI. Although the global task force document did not specify the degree of rise or fall that would be diagnostic of Type 1 MI, the earlier NACB document recommended a change of 20% at 3 to 6 h from a previous sample. Both documents recommended sampling at the baseline, approximately 6 to 9 h later, and again between 12 and 24 h from the baseline. To consider 2 troponin measures to be different requires them to vary by a difference of >3 standard deviations of the variance of the measures. For most robust assays, the variance is approximately 5% to 7%. Therefore, although empirical data supporting this degree of change are limited, a 20% change between successive values should be statistically different and also produce a value >99th percentile. However, other factors, including interindividual variability, may affect this parameter and become more important as increasingly precise assays are available. Point-of-care testing may be useful as a screening tool, but most point-of-care assays are only semiquantitative. To confirm a rise and/or fall from an initially positive assay would require serial quantitative testing, and in general, high-quality quantitative assays are preferred.

Most recently, the “Third Universal Definition of Myocardial Infarction” document was published in 2012 (13). Although refinements have been made to the thresholds and supporting information needed for the use of troponin to define MI in the setting of PCI and CABG, the general classification framework created by the 2007 Universal Definition of MI (12) document was carried forward. (Figure 69_gr2) aligns the model presented in (Figure 69_gr1) with the framework for the universal definition of MI. It is incumbent on all practitioners to fully understand the implications of an elevated troponin level in a given patient in order to initiate the appropriate treatment and to optimize outcomes. This is extremely important, not only in distinguishing Type 1 from Type 2 MI, but also in distinguishing ischemic from nonischemic causes and in understanding the non-MI cohort of patients denoted on the right side of (Figure 69_gr1).

Clinicians must be aware that all troponin assays are not created equal, and they must understand the characteristics and potential limitations of the specific assay used in their practice. This is because the susceptibility of troponin assays to potential interfering substances, such as heterophile antibodies and rheumatoid factor, can vary widely. Cardiac troponin is a complex analyte, and the regions of the troponin molecule targeted by the antibodies comprising these immunoassays are an important consideration for assay performance. Furthermore, assays have become, over time, increasingly sensitive, with improved analytical precision. This has resulted in a wide spectrum of assay quality in practice. Ultimately, this variability in quality has led to confusion in clinical practice and in the literature because varying cutoffs and decisions limits have been used. These decision limits have not always been the same for a given assay or between users of the same assay, and some have changed between earlier and later generations of the same assay. Thus, one study may not be comparable to the next in a similar population, and a test in one hospital may not have the same meaning in another. Assays are heterogeneous in their ability to accurately and reliably measure in the range of the 99th percentile of troponin values (i.e., the 95% confidence interval [CI] can be rather narrow for some assays but much wider for others) (14). Different interpretations of a “reference control population” upon which the 99th percentile of cardiac troponin values is based further complicates interpretation. Finally, measurement of cardiac troponin is not currently standardized. Therefore, unlike glucose, total cholesterol, and many other common measurements, troponin values vary from assay to assay, and assays have very different values for 99th percentile of normal. The NACB has developed analytical recommendations for troponin assays (15), and 1 publication has proposed a system of “grading” assays (16). With a centrally maintained, continuously updated database of assays, their functional characteristics and overall “grade” on these parameters would facilitate assay selection through competitive pressure to promote assay quality. Importantly, for the multiple troponin I assays in existence, regulations to ensure standardization of assays to the National Institute of Standards and Technology reference material (NIST #2921) would make it more feasible for clinicians to readily compare troponin levels measured in different laboratories or hospitals with different assays or generations of assays. This may be particularly important as patients are transferred from one facility to another. It is recommended that reference interpretive thresholds should be established for each cardiac biomarker, based on a population of normal, healthy individuals without a known history of heart disease. Creation of a healthy subject sample bank—such that all assay manufacturers could establish the 99th percentile of their assay against a common standard population of uniform size and clinical characteristics—would eliminate variability related to the population selected and obviate the need for each hospital or clinic to independently carry out this task. The NACB analytical document also recommends 1 threshold for optimal use of the cardiac biomarkers, troponin I and troponin T. Importantly, assays for cardiac biomarkers should improve towards a total imprecision (% coefficient of variation) of <10% at the 99th percentile reference limit. Even now, “high-sensitivity” troponin assays are being developed and are in use in some areas of the world. These assays have substantially lower limits of detection (in the picogram per milliliter range versus the current fourth-generation assays in the nanogram per milliliter range) as well as improved assay precision. Clinicians, laboratorians, clinical pathologists, and other users must communicate to assure that their troponin assays are in sufficient compliance with these recommendations and that all groups understand the characteristics of the assay in clinical use at a given facility.

Determining whether a troponin elevation represents a Type 1 MI is dependent upon the pre-test probability of ACS due to atherothrombosis (i.e., atherosclerotic plaque rupture, fissuring, and erosion). The concept of pre-test probability has been well understood for several decades. Pre-test probability for obstructive coronary artery disease (CAD) has been formally quantified in reference to coronary angiography, based on observable clinical characteristics such as age, sex, risk factors, and the quality of presenting symptoms (17). Similarly, factors that suggest a high pre-test probability of ACS include typical symptoms (rest or crescendo angina), ischemic ECG changes (ST-segment depression of >1.0 mm or T-wave inversion) or wall-motion abnormalities on echocardiography (or other imaging tests), and the presence of CAD risk factors or history of CAD (18). Because troponin elevation may be due to myocardial necrosis from causes other than atherothrombosis, there is no single putative standard for defining the presence or absence of MI as reliably as that of coronary angiography for the diagnosis of obstructive CAD. Furthermore, many of the same demographic factors that suggest a higher probability of ACS are also implicated in patients with non-ACS causes of troponin elevation, including heart failure, which often coexists. As a result, interpretation of the results of troponin testing for diagnosis of MI must be considered in the context of the pre-test probability of ACS, which is less formally quantitative than the pre-test probability of CAD.

Despite this limitation, Bayes' theorem is equally operative, if less precise. Thus, assuming a sensitivity of 100% among patients with a high pre-test probability of atherothrombotic ACS, in the range of 90% (typical chest pain with clinical and electrocardiographic evidence of ischemia), the post-test probability (predictive accuracy) for positive troponin is over 95%, even if the false-positive rate is as much as 40% (point B in Figure 69_gr3). On the other hand, if pre-test probability is low, in the range of only 10% (as for patients with atypical symptoms and nonspecific ECG changes), the post-test probability is around 50%, even if the false-positive rate is only 10% (point A in Figure 69_gr3). The difference in post-test probabilities becomes more pronounced at low pre-test probabilities, highlighting the impact of specificity of troponin in patients presenting with low pre-test probability of ACS. Thus, although myocardial necrosis is present, even high values of troponin do not establish a diagnosis of ACS with confidence if the pre-test probability is low. Conversely, low values do not reliably exclude the diagnosis of ACS if pre-test probability is high.

Therefore, from a Bayesian perspective, troponins are no different from any other imperfect diagnostic test, and even putative “high-sensitivity” troponin assays must obey the mathematical laws of probability. Just as a tool is only as good as its operator, a diagnostic test can be only as good as its interpretation. Expecting troponin testing to provide all the answers without including the proper clinical context can lead to erroneous diagnoses (19).

A semiquantitative summary of positive and negative predictive accuracies of troponin measurement is displayed in (Table 1), recognizing that these parameters are also influenced by the prevalence of the disease in the population studied. If troponin (or any other laboratory test) is applied indiscriminately in broad populations with a low pre-test probability of atherothrombotic disease, given its high sensitivity but low specificity for ACS among these patients, the positive predictive value (PPV) for non–ST-segment elevation MI is greatly diminished. Thus, from a diagnostic standpoint, even when the troponin is “positive”—especially a weak positive—the post-test probability for atherothrombotic ACS is still low in a patient with low pre-test probability of atherothrombotic ACS (e.g., a young woman with atypical symptoms or an elderly patient with nonspecific symptoms admitted with pneumonia). Although looking for a characteristic rise and/or fall in troponin is essential for the diagnosis of MI as proposed by both the NACB and the universal definition of MI as discussed earlier, ignoring pre-test probability often results in a high rate of misdiagnosis, especially when clinical symptoms are less typical.

Although some have advocated floating cutpoints in different clinical contexts (lowered cutpoints trading specificity for increased sensitivity, and higher cutpoints trading sensitivity for increased specificity) (20), the superior strategy would be to determine the patient-specific post-test probability of infarction given the patient-specific estimate of pre-test probability and the patient-specific observed troponin level. An algorithm that takes into consideration the pre-test probability based on clinical presentation and ECG changes, age, renal function, and a higher troponin T cutpoint was claimed to allow for more accurate diagnosis of ACS (20). However, the formal integration of pre-test probabilities with clinical predictors and troponin levels awaits validation in prospective studies.

Another feature that might help discriminate ischemia-induced cardiac injury from nonspecific myocardial damage is the kinetics of the marker. An elevated troponin level that is relatively constant over an appropriate sampling interval (e.g., baseline at 6 to 9 h and again at 12 to 24 h; a so-called “smoldering” pattern) is more likely to be caused by chronic diseases, such as renal failure, heart failure, myocarditis, or amyloidosis. However, episodic and lower-level changes, even below the reference limit, could represent ischemia-induced injury. By contrast, although data are limited in clinical practice, a dynamic change from the baseline value may be more suggestive of an acute MI. Primarily on the basis of assay characteristics (which are in flux) and statistical considerations, the NACB has recommended a 20% change at 3 to 6 h from the baseline value to be suggestive of an MI that is either evolving (a troponin increase) or resolving (a troponin decrease) (11). However, although using such a change may discriminate acute myocardial injury, it does not discriminate acute injury as a result of ACS from other causes of acute myocardial injury (e.g., pulmonary embolus or myocarditis). Further, the degree of elevation above the reference limit may also provide clues as to the etiology of the infarction (21), and further study will be needed to understand whether the same degree of change is relevant at low levels of baseline troponin elevation as at higher levels.

With the high analytical and clinical specificity of cardiac troponin assays and the pragmatic need for early management decisions, clinicians commonly diagnose acute MI on the basis of a single abnormal troponin value, especially those derived from highly sensitive troponin assays. Although in certain circumstances (e.g., a patient presents with their last chest pain >24 h prior, or for patients who present with moderate to high pre-test probability of MI), a single troponin may be sufficient. This is especially the case when the timing of symptoms is uncertain or in the setting of low pre-test probability for ACS. Relying on a single troponin value should be avoided in favor of serial testing as recommended by the 2012 ESC/ACCF/AHA/WHF consensus document (13). Of course, treatment for clear ST-segment elevation MI based on clinical and ECG criteria should not be delayed for troponin testing or the results of even a single test.

In summary, the major challenge of troponin testing in clinical practice, as for any other laboratory test, is often an inappropriate request and improper interpretation of the results, not the marker itself. Troponin evaluation should be performed only if clinically indicated, and elevated troponin must always be interpreted in the context of the clinical presentation. Only by doing so will troponin evaluation allow for optimal interpretation, diagnosis, risk stratification, and patient management.

In addition to providing diagnostic information, troponin elevation in the setting of a clinical presentation with ACS is independently associated with worse clinical outcomes ((6),9), irrespective of the result of CK-MB testing ((22),23). The yield of serial testing, both diagnostically and prognostically, in this setting is small beyond 8 h (24). In ACS, cardiac troponins also offer clinicians a valuable tool for therapeutic decision making. The underlying rationale, or so-called “troponin hypothesis,” is predicated on observations that ACS patients who are troponin-positive are more likely than troponin-negative patients to have more complex lesions with greater thrombus burden, a greater propensity for platelet embolization and distal microvascular obstruction that will lead to impaired epicardial coronary and myocardial tissue perfusion, as well as depressed left ventricular (LV) function ((25),(26),27). Treatment strategies such as intravenous glycoprotein (GP) IIb/IIIa inhibitors (abciximab, tirofiban, and lamifiban) ((8),(26),(28),29), the low-molecular-weight heparins (enoxaparin and dalteparin) ((30),31), and an early invasive strategy ((27),32) appeared to be more beneficial in troponin-positive patients than in troponin-negative patients. However, when the troponin hypothesis was examined prospectively in the GUSTO IV (Global Use of Strategies to Open Occluded Coronary Arteries) trial—assessing whether benefit from upfront initiation and sustained treatment with abciximab was limited to troponin-positive patients—treatment with abciximab offered no benefit in patients with elevated levels of troponin undergoing primarily conservative medical management (33). Furthermore, in contrast to low-molecular-weight heparins and GP IIb/IIIa inhibitors, the benefit of treatment with clopidogrel was not shown to differ among patients with and without elevated troponin in the CURE (Clopidogrel in Unstable Angina to Prevent Recurrent Events) trial (34). Thus, the troponin hypothesis may not be applicable in the setting of all therapeutic interventions for ACS.

Similarly, randomized trials evaluating early versus selective invasive strategy in ACS have yielded inconsistent results. The results from the ICTUS (Invasive versus Conservative Treatment in Unstable Coronary Syndromes) trial, which only enrolled patients with elevated troponins, showed no apparent treatment advantage for the early invasive strategy, but also no harm, and may have been limited by selecting only on troponin status and entering lower global risk patients (35). Disparate responses to an invasive strategy in patients with elevated troponin levels across trials (harm in Vanquish [Veterans Affairs Non–Q-Wave Infarction Strategies In-Hospital], neutral effect in ICTUS, and benefit in FRISC II [Fragmin and Fast Revascularisation during Instability in Coronary artery disease], TACTICS–TIMI 18 [Treat Angina with aggrastat and determine Cost of Therapy with Invasive or Conservative Strategy–Thrombolysis In Myocardial Infarction 18], and VINO [Value of First Day Angiography/Angioplasty in Evolving Non–ST-Segment Elevation Myocardial Infarction] trials) highlight the limitations of cardiac biomarkers used as a dichotomous variable as a single risk index. However, in aggregate, meta-analyses have demonstrated benefit from an early invasive strategy among troponin-positive patients ((36),37). Still, additional risk stratification beyond troponin alone may help to refine populations that benefit. In retrospective analysis of the FRISC II data (38), there was a 40% relative risk reduction in death or MI only in patients with both troponin T >0.03 ng/ml and ST-segment depression on admission ECG, whereas participants with only 1 of these variables had no benefit. In the RITA-3 [Third Randomised Intervention Treatment of Angina] trial, 9 factors besides the treatment group emerged as multivariate predictors of outcomes at 5 years (39). Patients derived the greatest benefit from an early invasive strategy in the highest quartile of risk score based on these predictor variables. These observations underscore the importance of global risk assessment rather than using any single risk marker for therapeutic triage. The 2011 focused update on Unstable Angina/Non–ST-Elevation Myocardial Infarction guidelines reflected these findings by recommending the early invasive strategy as Class I, Level of Evidence: A in patients identified to be at high risk, based on a “combination” (not “any 1”) of several risk variables, including elevated troponin (18).

Ideally, refinement of the use the troponin testing should ensure that it is an important element of global risk assessment in a clinically driven management algorithm. One such algorithm is shown in (Figure 69_gr4). Patients with elevated troponin and a high pre-test probability of ACS (based on presenting symptoms, risk factors, history of CAD, ECG, or wall motion changes) are most likely to derive benefit from a treatment strategy aimed at coronary thrombosis (e.g., aggressive antiplatelet therapy, coronary angiography, and revascularization). Patients, then, could be further stratified based on the risk characteristics into those likely to benefit from an early invasive strategy (for high-risk characteristics) or early conservative strategy (for low-risk characteristics). Patients with elevated troponin and a low pre-test probability of ACS are unlikely to derive a large incremental benefit from aggressive treatment strategy. In such patients, the main goal would be to identify the underlying cause of the troponin elevation—conditions such as myocarditis, pericarditis, cardiac contusion, sepsis, pulmonary embolism (PE), and heart failure. Therapy in these circumstances should target the underlying cause. Treatment in patients without troponin elevation, but with a high pre-test probability of ACS, should be directed by identification of other markers of risk. Those with high-risk features should be considered for early invasive management, whereas those with low-risk features could be managed with either an invasive strategy or conservative strategy that is dictated by clinical course and functional testing and depends on clinical judgment of the individual patient circumstances.

In summary, the troponin test should be integrated with assessment of other clinical factors that influence diagnosis and prognosis to provide a foundation for selection of the most clinically appropriate treatment strategy in patients with ACS without ST-segment elevation. Global risk assessment rather than any single risk marker should be the main driver of therapeutic decision making.

Previously, the commercially available troponin assays used in clinical practice lacked the stringent precision (10% coefficient of variation at the 99th percentile cutoff) advocated by the universal definition of MI (11). However, so-called high-sensitivity (or ultrasensitive) assays have been developed that meet these requirements. Several studies have reported enhanced diagnostic and prognostic accuracy of these high-sensitivity troponin assays across a spectrum of patients with cardiovascular disease, including ACS ((40),41), heart failure (42), and chronic stable CAD without LV systolic dysfunction (43). In comparison with standard assays, the high-sensitivity assays used in the studies by Reichlin and Keller (44) showed remarkably increased sensitivity and increased early detection of myocardial necrosis, but this was associated with decreased specificity. The discrimination from the reference population was favorable, and overall diagnostic accuracy was driven by the increased sensitivity. If, as in previous studies, the prognostic value of small elevations of troponin remains, we may soon use them in new ways. Rapidly repeated determinations may accelerate triage, especially in the emergency department. Also, the ability to detect incremental changes at levels below those previously detectable may lead to identification of patients with ischemic events who previously would have gone unrecognized and for whom additional testing for diagnostic clarification, risk stratification, or treatment may be indicated. Extending the range of detection may also lead to the use of continuous elevations instead of cutpoints in risk models.

However, the utility of high-sensitivity troponin testing for rapid triage or incremental identification of ACS patients with previously subclinical ischemic events may be limited by delays in patient awareness and travel to acute care facilities (45) as well as influenced by the pre-test likelihood of ACS (19). In addition, it is unclear what the prognostic or therapeutic implications of these increasingly sensitive and precise assays will be in general clinical use. Cohort studies show that as the definition of MI includes lower-risk patients, the number of outcome events decreases (46), and the case fatality rate decreases for the same diagnosis, thus creating an “era” effect (i.e., confounding by year of testing) (47). Still, for patients with detectable troponin levels and a clinical presentation consistent with moderate- to high-risk ACS, even if early intervention with antithrombotic therapy or an invasive assessment is not clearly indicated, it may identify a referral group for stress testing or other noninvasive means for further risk assessment.

Routine detection of troponin levels using high-sensitivity assays that yield a continuous gradient in apparently normal subjects may make it difficult to differentiate myocardial necrosis related to plaque rupture in ACS patients (those who might benefit from aggressive treatment strategies) from necrosis in non-ACS patients. As an example, Venge et al. (48) detected troponin I in 95% of a normal healthy reference population with values that overlapped those in a random subsample of ACS patients in the GUSTO IV trial. Although cardiac events were significantly more frequent in the GUSTO IV patients, the discrimination was affected by case mix. Other studies suggest that low-level troponin elevations will be commonly detected by high-sensitivity assays, particularly in populations with stable coronary disease and heart failure; therefore, this finding reflects a shift, as assays become more sensitive, from detecting acute illness to identifying underlying chronic illness. In the Val-Heft (Valsartan Heart Failure) trial on heart failure and the PEACE (Prevention of Events with Angiotensin Converting Enzyme Inhibition) trial on chronic coronary disease, nearly all patients had detectable troponin by high-sensitivity assays, which were associated with subsequent risk of mortality and heart failure ((43),49). Such widespread low levels of detectable troponin in populations with these assays will make it challenging to interpret low-level troponin elevations. However, these same features may open an era of population screening for subclinical disease and monitoring for disease tempo. In 1 population random sample, troponin T was detectable by high-sensitivity troponin T assay in 25% of the cohort, including 16% of the cohort after restricting to those without known chronic illnesses (such as diabetes, chronic kidney disease [CKD], hypertension, or coronary disease) (50). These levels correlated with the measures of subclinical cardiovascular disease, including high coronary calcium scores and greater LV mass adjusted for body surface area.

In another population study of community-dwelling individuals over age 65 years who had no prior documented heart failure, 66% of subjects had detectable troponin levels by high-sensitivity troponin T assay, which were strongly associated with subsequent death or heart failure events (51). Furthermore, changes in troponin levels correlated with changes in risk, such that those with initially detectable troponin whose levels increased by >50% on subsequent testing had increased risk. However, those whose levels fell >50% on serial testing had a reduction in risk in comparison with those patients with >50% change. These results suggest a potential role for high sensitivity in monitoring treatment response and will potentially usher in a new era of directed therapy. In addition, high-sensitivity troponin may have a role as part of a biomarker score for population screening as suggested by the MORGAM (Multinational Monitoring of Trends and Determinants in Cardiovascular Disease [MONICA], Risk, Genetics, Archiving, and Monograph) biomarker project—where a combination of high-sensitivity troponin I, N-terminal pro-brain natriuretic peptide (NT-proBNP), and high-sensitivity C-reactive protein contributed significantly to clinical features in stratifying risk for long-term cardiovascular events (52).

Prospective studies are needed to determine the clinical effect of using new high-sensitivity assays, both in respect to their relationships with outcomes in population screening and for patients with suspected ACS and chronic disease populations, as well as in regard to risk–benefit tradeoffs for treatment or additional testing in such populations.

The 2007 Joint ESC/ACCF/AHA/WHF consensus document on “Universal Definition of Myocardial Infarction” defined Type 2 MI as myocardial necrosis secondary to ischemia from either “increased oxygen demand or decreased supply” (12). This designation has been maintained in the 2012 version of the universal definition of MI document (13). How frequently Type 2 MI occurs is unclear, in part because of reporting variability. There are only a few cohort studies (all retrospective) that have attempted to quantify the incidence of non-ACS causes of ischemia-mediated troponin elevations. In the largest series to date, 1.6% of 1,093 patients had Type 2 MI when the 2007 Joint ESC/ACCF/AHA/WHF criteria were applied (53). The reported causes for non-ACS ischemic troponin elevations vary by study and include paroxysmal atrial fibrillation, supraventricular tachycardia, ventricular tachycardia, hypoxia, severe anemia, or gastrointestinal hemorrhage ((20),(53),54). All of these studies suffered from inadequate characterization for the rate of underlying CAD in the population studied, without which it is difficult to estimate the true frequency of Type 2 MI.

Many non-ACS, demand-mediated ischemic conditions may cause increased troponin levels (see Figure 69_gr1 for examples). Case reports and case series tend to focus on patients with elevated troponin and normal coronary arteries, but it is important to understand that these non-ACS ischemia-mediated conditions may also unmask underlying CAD. For example, though cocaine is thought to elevate troponin acutely from a combination of coronary spasm and sympathomimetic effects, it also accelerates the development of epicardial CAD. Coronary arteriograms in these patients have revealed that approximately 80% of such patients have significant underlying CAD ((55),56). Two studies have investigated causes of troponin elevations in patients with normal coronary arteriograms. In a study of 144 patients, non-ACS ischemic causes of elevated troponins included 35 patients (24%) with tachyarrhythmia, 2 (1.4%) with LV hypertrophy, 1 (0.7%) with malignant hypertension, 2 (1.4%) with coronary vasospasm, and 9 (7%) with gastrointestinal bleeding (57). In another study of 21 patients, causes of elevated troponin were tachycardia in 6 patients (28.5%) and extreme physical exertion in 2 patients (9.5%) (58).

In summary, there is insufficient evidence to provide strict guidelines as to how to differentiate between ACS and non-ACS ischemia-induced troponin elevations without taking into account the clinical presentation. Though disruption of epicardial blood supply from emboli or coronary spasm can result in ischemic ECG changes and serial troponin changes indistinguishable from an MI caused by plaque rupture, the other causes of non-ACS ischemic troponin elevations may result in a more subtle increase with less change evident on serial determinations. When deciding whether or not to further investigate the possibility of CED in these patients, it is necessary to make an assessment of pre-test probability that the troponin elevation is due to underlying CAD versus one of the many non-ACS causes of troponin elevations.

In this section, we review and define the current status of troponin assays for detection of periprocedural myonecrosis and clinical MI. In this context, “procedures” refers to both PCI and cardiac surgery procedures, primarily CABG. The ESC/ACCF/AHA/WHF Task Force for Redefinition of MI have reviewed and revised the universal definition of MI (13). Our discussion of post-revascularization troponin adheres to these newly established and widely adopted definitions and is confined to Type 4a and Type 5 MI, which are relevant to PCI and CABG, respectively.

The occurrence and adverse consequences of periprocedural MI for both PCI and CABG are well known. Early assessments depended predominantly upon a combination of clinical observations, ECG changes, and cardiac biomarkers such as total CK. With development of the relatively myocardium-specific marker CK-MB and troponins, detection became more accurate and precise. However, what level of biomarker elevation reflects “clinically significant injury” and whether this laboratory threshold is related to immediate adverse outcomes or delayed adverse events remains uncertain.

Several studies have linked post-PCI CK or CK-MB elevation in the 3× to 8× upper limit number (ULN) range to increased mortality ((59),(60),61). The 2011 update of the PCI guidelines noted that more frequent requirements for revascularization procedures and a higher risk of death or subsequent MI were associated with elevated cardiac biomarkers (62). These guidelines recommended that for patients in whom a clinically driven CK-MB determination was made—a CK-MB increase of >3× ULN should be treated as signifying an MI—and provided a Class IIb recommendation that post-PCI enzymes levels be measured in all patients. It is recognized that the threshold specified in the PCI guidelines may change in accordance with the recently published 2012 update to the Universal Definition of MI that now specifies a >5× ULN troponin elevation and clinical evidence of MI to define a PCI-related MI (13). There were no recommendations for further workup of smaller elevations in an otherwise asymptomatic patient.

Several studies have examined the relevance of PCI-related troponin elevations ((63),(64),(65),(66),(67),68). The results and conclusions of these studies, like the CK-MB data that preceded them, have been inconsistent secondary to small sample sizes, different elevation thresholds, and varying analytic techniques. Two meta-analyses, one using older and less sensitive troponin assays (68) and the other using a newer generation with more sensitive assays (based on the 99th percentile criteria) (65), concluded that post-procedure troponin elevations were associated with adverse outcomes, including long-term death or MI. Troponin elevations might have particularly important prognostic implications in circumstances where intraprocedural complications have occurred and result in angiographic evidence of flow impairment (e.g., side branch closure, transient decreased TIMI flow grade, or embolization).

A report from the ACUITY (Acute Catheterization and Urgent Intervention Triage Strategy) trial suggested that spontaneous MI (i.e., clinical ischemic events unrelated to a procedure, not including periprocedural MI) was significantly associated with subsequent mortality (66). Consistent with these observations, a consecutive cohort study at a single institution of 2,352 patients with pre- and post-PCI biomarkers for elective or urgent PCI found that long-term prognosis was most closely related to the baseline, pre-procedure troponin value rather than the post-PCI value (69). Post-procedure cardiac troponin T values did not contribute to the prediction of death or MI when added to the pre-procedure risk estimates. Of particular importance was that this report used the fourth-generation troponin T assay, an important difference from earlier reports in which earlier generation, less sensitive assays were used. Similar findings were also reported in a single-center cohort study of 5,847 consecutive patients treated with nonemergency PCI and who were assayed with a third-generation troponin T assay (70). However, in another study of ACS patients, elevated troponin I levels post-PCI remained prognostically significant even after adjusting for pre-procedure troponin elevation (63). Thus, although some studies suggest that periprocedural MI may be more related to baseline risk, atherosclerotic burden, and procedural complexity, without being an independent mortality prognosticator, it is premature to exclude the prognostic importance of detecting post-PCI troponin elevations.

A particular challenge with these studies is that they examine only relative increases above the ULN. This is particularly problematic with the very low ULNs of newer generation troponin assays with which even a 5× elevation may still be a very low absolute troponin level. Considering this and conflicting results from prior studies, the examination of the relationships for PCI-related troponin elevation with outcomes may be best assessed using absolute elevations, and further study is needed to understand whether there is a threshold effect. Although there is general consensus that large troponin elevations are associated with negative prognostic implications, it is less clear what constitutes a large elevation and whether the negative prognosis is limited to in-hospital, is short term, or persists long term ((67),(69),71). As evidence of these uncertainties, the current ESC guidelines for PCI do not recommend the use of troponins after elective PCI (72). Further studies are required to resolve these uncertainties.

Recent data from the National Cardiovascular Data Registry (NCDR) show that most institutions do not routinely check troponins post-procedure. Whether this is a response to the lack of scientific consensus or to other factors is unknown. Regardless, NCDR data suggest that institutions that routinely perform and report troponin values appear to have higher post-procedure MI rates than those that do not. This disparity may be due to reporting imbalances as opposed to an actual difference in outcomes. In the absence of clear data indicating the proper clinical threshold to diagnose a MI, most troponin elevations are coded as MI. Thus, institutions that voluntarily check and report their values may be placing themselves in a disadvantaged position relative to those that do not. Although it is premature to use such data as a PCI performance measure, systematic collection of post-procedure troponin data within the NCDR may, nevertheless, be considered appropriate to determine true rates and outcomes of periprocedural troponin elevations and to create a robust dataset that will fill an important knowledge gap and facilitate the establishment of evidence-based periprocedural management.

In summary, in the era of less sensitive biomarkers, 20% of patients who had an angiographically uncomplicated PCI experienced periprocedural enzyme elevations (73). The development of newer and more sensitive troponin assays has increased the ability to detect myonecrosis so that as many as 33% of patients undergoing elective PCI have elevated troponin post-procedure, but it has created confusion as to the clinical relevance of such findings following coronary procedures. At this juncture, there are data to support the concept that detection of troponin elevations post-procedure may be associated with an increase in long-term adverse events, particularly if the pre-procedure troponin is normal or falling. The updated PCI guidelines provide a Class I, Level of Evidence: C recommendation for performing post-procedure biomarker assays (including troponin testing) when an intraprocedural angiographic complication is identified or a patient has signs or symptoms suggestive of MI during or after PCI, and give a Class IIb recommendation for post-PCI testing in all patients (62). The 2012 revision of the universal definition of MI now recommends that the threshold for defining PCI-related MI (Type 4a) in patients with a normal pre-procedure troponin level should be a troponin elevation within 48 h post-procedure of >5× ULN with either symptoms of myocardial ischemia, new ischemic ECG changes, or documented complications during the procedure (13). If elevated pre-procedure levels are stable or falling, a >20% rise above the pre-procedure level with concurrent clinical criteria is necessary to define Type 4a MI. The NACB recommended testing with the caveat that insufficient data exist to propose a specific cutoff (71). This writing group supports these recommendations for testing and defining PCI-related MI. Further, it supports the NCDR effort to obtain routine periprocedure troponin measurements in all PCI patients to establish a database from which critical questions about the utility of troponin testing peri-PCI can be evaluated. We also encourage simultaneous collection of analytical parameters of the assays used to establish a robust database from which information about true rates of elevation, relevant diagnostic thresholds, and prognostic implications—in the context of data on other clinical factors that influence prognosis—can be established. Finally, all PCI patients should receive guidelines-recommended secondary prevention, and immediate clinical care should consider the patient's overall clinical status in addition to any biomarker testing results.

Because of sensitivity and lack of etiological specificity for the cause of myonecrosis, use of troponin to define perioperative MI in CABG patients is challenging. Based on data from studies using older generation assays, the NACB guidelines recommend that troponin elevation must exceed at least 5× ULN to define clinically relevant post-operative MI, with higher values associated with worse outcomes (71). However, given the variability in biomarker responses to surgery, the NACB advises that additional criteria over and above marker results are needed to define a CABG-related coronary vascular event (71). Most recently, recognizing that the threshold selection is arbitrary, the 2012 universal definition of MI document recommends that CABG-related MI be defined as a troponin elevation of >10× ULN when there is corresponding ECG (new Q waves), angiographic (occluded graft or newly occluded native vessel) or imaging (new loss of viable myocardium) evidence of MI (13). In the evolving era of highly sensitive troponin assays, no cutoff for clinically relevant post-operative MI has been defined or prospectively studied.

Nonischemic conditions often present with chest pain or other symptoms that create diagnostic uncertainty for the treating physician. Therefore, troponin may be ordered early in the assessment of the patient as part of the diagnostic evaluation for these conditions. Serum concentrations of cardiac troponins have been detected in many disease entities, aside from coronary and other primary cardiac conditions ((71),(74),(75),(76),77). In some cases, the mechanism of cardiac involvement is obvious (e.g., hypoxemia and secondary subendocardial ischemia from right ventricular (RV) pressure overload following a PE. In others, however, troponin release appears to represent a nonspecific “vital organ” response to systemic illness. As with any other test, troponin levels are only of value in nonischemic conditions when they contribute to accurate diagnosis or inform prognosis—and when such will affect treatment or clinical outcome.

Using troponin levels to estimate survival, patient appropriateness for aggressive therapy, identification of patients at risk for therapy-induced disease, and/or determination of need for prolonged inpatient monitoring have been reasonably well supported for some conditions: heart failure, PE, CKD, sepsis syndrome, chemotherapy-induced cardiomyopathy, amyloid light chain (“primary”) amyloidosis, and post-cardiac transplantation monitoring, as well as following noncardiac surgery, thermal injury, and blunt cardiac trauma. Fewer data have been published on other conditions; however, enough are available to suggest that further investigation might identify value in troponin monitoring: toxins/envenomation, endocarditis, severe metabolic conditions, decompensated chronic lung disease, primary hematologic conditions, stress/catecholamine-associated myocardial dysfunction, and subarachnoid hemorrhage.

The remainder of this section will focus on 4 conditions in which a potential clinically useful role for troponin testing exists and/or for which there is substantial confusion in interpretation of clinical practice (heart failure, PE, CKD, and sepsis). Discussion of other conditions for which troponin has a potential clinical role is available in Appendix 3. In general, until further data are available on how troponin testing may clearly change patient management, unless specifically stated, it is not recommended to measure troponin levels specifically for diagnosis or prognosis in the conditions discussed in this section or Appendix 3. It is important to point out that this document, in this section and in Appendix 3, reflects prevalences and associations with outcomes that were determined with a variety of assays over several years. Many studies used troponin cutoff values other than the 99th percentile, which is currently recommended, and many used insensitive assays or older assays that were much less sensitive than current generation assays. Therefore, prevalences and strengths of association across studies may vary due to variation in these parameters and may not reflect results that would be obtained with modern assays or in the future with high-sensitivity assays.

In both inpatient and outpatient heart failure populations, elevated troponin levels are common and associated with worse outcomes (Table 2). Rates and strengths of association with outcome vary widely depending on the troponin assay, assay generation, and cutoff used. These differences across studies create challenges in interpretation and generalization and will become an increasing challenge in heart failure, as seen with MI, since troponin assays continue to evolve and until there is adequate standardization across assays. In the large, multicenter ADHERE (Acute Decompensated Heart Failure Registry) National database, 81% of patients admitted with heart failure had troponin testing, and nearly 6.2% of patients had abnormal troponin test results (troponin I ≥1.0 μg/l or troponin T ≥0.1 μg/l) after excluding patients with serum creatinine >2.0 mg/dl (78). Hospital mortality among troponin-positive patients was 8.0%, compared with 2.7% among troponin-negative patients (adjusted odds ratio [OR]: 2.55) and was independent of an etiology of heart failure (ischemic or nonischemic). However, when a lower troponin I threshold is used (troponin I ≥0.4 μg/l or troponin T ≥0.01 μg/l), 75% of patients have detectable levels of troponin.

The key questions regarding troponins and heart failure are as follows: 1) in the evaluation and management of heart failure patients, how should troponin testing be used, if at all; and 2) how should clinicians respond to elevated troponin levels found in such patients? First, elevated troponin values have prognostic value in estimating future risks of death or hospitalization for heart failure patients with or without preserved ejection fractions. However, although elevated troponins have prognostic value in heart failure, they are poor diagnostic markers for an ischemic versus nonischemic etiology of heart failure or heart failure progression. For these reasons, the NACB's practice guideline on cardiac biomarker testing in heart failure gives only a Class IIb recommendation to use of troponins for risk stratification “beyond the setting of acute coronary syndromes.” In addition, the guideline specifically recommends against “routine biomarker testing for the sole purpose of risk stratification in patients with heart failure” (89) (p. e103).

Nevertheless, it is sometimes difficult to know after initial evaluation whether patients presenting with heart failure, particularly acute decompensated heart failure with an elevated troponin test, do so as a result of unstable coronary ischemia. The 2012 universal definition of MI document cautions that troponin elevation alone is not sufficient to establish the diagnosis of MI or type of infarction or etiology of troponin elevation in heart failure (13). Thus, for patients with known coronary disease and heart failure who were previously stable or for patients with new-onset heart failure who present with an elevated troponin level, it may be useful to further evaluate for evidence of obstructive coronary disease or acute plaque rupture with functional testing or coronary angiography. For an excellent summary of the current state of troponin testing heart failure and future directions to be explored and solidified in incorporating troponin testing into heart failure management, the reader is referred to an overview published by Kociol et al. (90).

In a 2007 meta-analysis of the prognostic value of cardiac troponins in 20 studies of acute PE (91), the rate of elevated troponin levels ranged from 10% to 77% (median 39%). Overall, elevated troponin levels were associated with short-term (up to 30 days) all-cause mortality (OR: 5.24 [95% confidence interval (CI): 3.28 to 8.38]), with similar associations for troponin I and troponin T. Of 8 studies reporting specifically on death due to PE, elevated troponin levels were highly associated with fatal PE, OR: 9.44 (95% CI: 4.14 to 21.49; p < 0.00001). Elevated troponin levels were also associated with nonfatal complications of PE during hospitalization, OR: 7.03 (95% CI: 2.42 to 20.43; p = 0.0003). A subsequent update of this meta-analysis with 5 additional studies revealed similar associations (92). Cardiac troponin elevation in PE is believed to result from pulmonary vascular obstruction and vasoconstriction that causes a sudden increase in pulmonary vascular resistance, pulmonary artery pressure, and RV afterload. RV dysfunction itself is associated with increased mortality in PE patients ((93),(94),(95),(96),(97),(98),(99),(100),101), and troponin elevation may be an early and reliable marker for RV dysfunction ((102),103). Indeed troponin is more likely to be elevated in those with echocardiographically identified RV dysfunction ((104),(105),(106),(107),(108),(109),110). However, an elevated troponin can occur in patients without RV dysfunction. Unlike echocardiography, troponin testing is readily, cheaply, and rapidly available 24 h a day. Therefore, it might be a useful tool to identify patients with PE at higher risk of mortality and who may benefit from more aggressive treatment.

However, despite the strong associations of troponin elevation in PE with important clinical outcomes, how to respond to this prognostic information is presently unclear, and routine testing in suspected or confirmed PE is not indicated. Patients with massive PE and hemodynamic instability benefit from thrombolytic therapy, if not contraindicated, regardless of troponin elevation. Those with a relatively small PE and no RV dysfunction typically have an uncomplicated course ((111),112). When RV dysfunction is present and the patient is normotensive, some advocate consideration of thrombolytic therapy ((113),(114),115), whereas others do not ((116),(117),(118),(119),(120),121). Additionally, some authors would advocate surgical pulmonary embolectomy if the thrombus were centrally located ((122),123). The answer to this important therapeutic question awaits the results of a large randomized trial. For now, it is clear that elevated troponin levels are not uncommon in patients with PE, and they are associated with RV dysfunction and adverse outcomes, including death. Troponin testing has no current role in the diagnosis of PE, but in the absence of hemodynamic instability, it appears to have excellent negative predictive value (NPV) for in-hospital deaths ranging from 82% to 100% ((92),(102),(103),(109),(112),124). Such patients are very unlikely to benefit from aggressive treatment such as thrombolysis.

Although still somewhat controversial, elevated troponin levels in patients with reduced renal function (those with end-stage renal disease [ESRD] and on dialysis or those with moderate to severe renal impairment and residual renal function) are most likely not caused solely by decreased renal clearance ((76),(125),126). Intact troponins are large molecules; therefore, it is improbable that the kidneys are primarily responsible for their clearance from the serum. Although there are some data to suggest that residual renal function can affect troponin levels ((127),(128),129), other studies have not found this association ((130),(131),132). Diris et al. (133) have demonstrated that troponin T molecules are degraded into smaller fragments that are detected by assays and are small enough to be filtered by the kidney. These fragments might partially account for elevations of troponin T so often seen in patients with ESRD. On the other hand, elimination and half-life of troponin I after MI appears to be similar in those with normal renal function and ESRD (134).

Despite absence of a single unifying pathophysiological explanation for troponin elevation in patients with impaired renal function, the relationship with clinical outcomes is clear. In a 2005 meta-analysis (135), the rate of troponin T positivity ranged from 12% to 66%, and 0.4% to 38% for troponin I. Elevated troponin T was significantly associated with all-cause mortality with a relative risk of 2.64 (95% CI: 2.17 to 3.20). A significant association was also demonstrated with cardiac death, with a relative risk of 2.55 (95% CI: 1.93 to 3.37). Eight subsequent studies corroborated the relationship of troponin T elevation in CKD patients with varying degrees of severity (moderate to severe impairment with residual function to ESRD on dialysis) with all-cause mortality ((127),(128),(136),(137),(138),(139),(140),141). Because of the different assays and cutoffs, pooling of the 12 studies using troponin I was problematic; however, there was an association between elevated troponin I and all-cause mortality, relative risk 1.74 (95% CI: 1.27 to 2.38).

The NACB Laboratory Medicine Practice Guidelines (71) recommend the use of troponin for diagnosis of MI in all CKD patients (regardless of the severity of renal impairment) who have symptoms or electrocardiographic evidence of myocardial ischemia. The guidelines also advise relying on dynamic changes in troponin values of ≥20% in the 6 to 9 h after presentation to define acute MI in ESRD patients, who more frequently have chronically elevated troponin levels. The guidelines also state that troponins can be aids to risk stratification in ESRD patients and provide baseline values for comparison when there is an acute clinical change. On the basis of the relationship between troponin T levels and mortality in patients with severe renal impairment and ESRD on dialysis, the Food and Drug Administration has approved the use of troponin T for identifying CKD patients at high mortality risk. The National Kidney Foundation Disease Outcomes Quality Initiative Work Group (142) also recommends that troponin T levels be considered for risk stratification of chronic dialysis patients, but how the information should be used is still unclear. The Work Group also reinforces that the presence of a time-dependent elevation in troponin T or troponin I in the setting of ACS portends increased cardiovascular morbidity and mortality.

Detection of troponin elevations in patients with sepsis, septic shock, and the systemic inflammatory response syndrome is relatively common. The mechanism of troponin release in the absence of flow-limiting epicardial coronary atherosclerosis is uncertain but is felt to be related to the known occurrence of myocardial dysfunction in sepsis, and many possible contributors have been hypothesized and partially investigated ((143),(144),145). In a summary of studies published between 1998 and 2008 ((146),(147),(148),(149),(150),(151),(152),(153),(154),155), elevated troponin I or troponin T occurred in a median 62% of patients, an interquartile range of 43% to 85%, and most elevations were modest. There was no obvious association between troponin elevation and a prior history of ischemic heart disease. Of the studies that reported outcomes, 5 found a significant association between troponin positivity and death ((146),(149),(152),(153),154). Only 2 studies considered the independent association of troponin with outcome ((153),154), and in only 1 was it significant (153). Larger, more homogeneous studies using standardized troponin analysis are needed to clarify the role of troponin in risk stratification of septic patients.

In addition to mortality, troponin elevation in sepsis may also be associated with impaired LV function, which commonly occurs in septic patients (∼50% of patients with severe sepsis and septic shock) ((143),144). In summation, elevated troponin levels appear to be a potential marker for poor LV function in septic patients. More studies corroborating the association and how this mediates the association with mortality and, specifically, how tailored treatments might modify sepsis-related LV dysfunction and mortality are needed. At present, however, routine troponin testing in septic patients is not recommended.

Expert panels have identified troponin as the preferred biomarker to detect drug-induced cardiac injury (156). The ability of high-dose chemotherapy (including anthracyclines, cyclophosphamide, and perhaps platinum-based agents) to induce both transient (early) and permanent LV systolic dysfunction, diastolic dysfunction, and arrhythmias is well established. Several studies of troponin testing in patients treated with these chemotherapeutic agents revealed findings worth noting: 1) troponin positivity at almost any level, and at almost any time during multicycle chemotherapy regimens, identifies patients with a significantly increased risk of permanent or more severe reduction in LV systolic function ((157),158) and/or premature death (159); 2) the magnitude and frequency of troponin elevations correlate with cumulative drug dose, including from earlier courses/regimens (160); 3) following an early chemotherapy-induced reduction in LV ejection fraction (LVEF), patients without elevated troponin tend to show significant or total recovery of LV function over time ((157),161); 4) lower-level troponin elevation may be associated primarily with changes in diastolic ventricular performance ((162),163); and 5) the NPV of a normal troponin level is very powerful when patients are sufficiently stratified (164). Finally, results from a randomized trial of 473 patients who had an elevated troponin level within 72 h after high-dose chemotherapy suggested that administration of enalapril (2.5 mg daily started 1 month after the last dose of chemotherapy and increased in 3 subsequent steps to 20 mg daily for 1 year of treatment) may dramatically reduce the risk of developing LV dysfunction at 1 year (hazard ratio: 0.015 vs. placebo; no patients in enalapril group vs. 25 patients in the control group developed an absolute decrease in ejection fraction of 10% or more from baseline). These results suggest a role for troponin testing in guiding adjuvant therapy (165). On the basis of these data, troponin appears to be a useful tool in detecting cardiac toxicity and stratifying risk for the severity of ventricular dysfunction. If confirmed in additional, well-done randomized clinical trials, use of angiotensin-converting enzyme inhibitors may prove useful in preventing cardiotoxic effects of some chemotherapy.

Observational studies limited to children were less likely to document troponin elevations during anthracycline chemotherapy (166) or, when troponin was elevated, found no relation to systolic function (167). Other studies did correlate LV dilation to troponin elevation, at least in the short term (168). A key question in this population, as to whether troponin can predict long-term cardiovascular mortality, remains unanswered.

Just as troponin is a marker for anthracycline cardiotoxicity in clinical practice, another area of intense interest in which troponin testing may be of value is monitoring for cardiotoxicity during the early phases of new drug development. Similar advantages of sensitivity and cardiac specificity are of value in this arena just as in clinical care. Similar considerations for population variation, definition of normal levels and relevant incremental changes (particularly with high-sensitivity troponin assays), and the lack of etiological specificity are operative in using troponin as a biomarker of cardiotoxicity in early drug development. The Cardiac Safety Research Consortium produced a white paper on this topic (169).

Troponin elevation has been reported to occur in a number of other nonischemic clinical conditions (170). In some conditions, there is a clear association of troponin elevation with adverse outcomes; in others, the relationship is less clear. In none of these conditions is there yet a clear or potential clinical indication for the use of troponin testing for diagnosis, risk stratification, disease state monitoring, or to tailor treatment. Because myocarditis and myopericarditis may be major contributors to nonischemic elevation of troponin in patients who present for acute evaluation of chest pain, they are discussed in this section. Other nonischemic conditions that may confound interpretation in the clinical setting are discussed briefly in Appendix 4. The literature discussed in this section and in Appendix 4 reflects prevalences and associations with outcomes that were determined with a variety of assays over several years. Many studies used troponin cutoff values other than the 99th percentile, which is currently recommended, and many used less sensitive or older assays that were much less sensitive than current generation assays. Therefore, prevalences and strengths of association across studies may vary due to variation in these parameters and may not reflect results that would be obtained with modern assays or in the future with high-sensitivity assays.

Many infectious and toxic agents have been linked to myocardial inflammation and dysfunction ((171),172), but information on serum troponin levels is scant. In the Myocarditis Treatment Trial (173), 34% of patients with biopsy-proven active myocardial inflammation had elevated troponin (>3.1 ng/ml) compared with only 11% with systolic heart failure but negative endomyocardial biopsies (174). Although not linked to prognosis or response to therapy, it was noted that elevated troponin levels were much more common among patients with ≤1 month of heart failure symptoms, suggesting it could have a role in determining the chronicity of individual patients' myocarditis. Additionally, 4.4% of enrolled patients had demonstrable anti-hepatitis C virus antibodies (of those, 30% demonstrated elevated troponin I and 48%, elevated troponin T), which was also possibly consistent with troponin identifying active inflammation and myocyte necrosis (175). Other randomized trials studying treatment for myocarditis/acute cardiomyopathy have not published troponin data from their populations (176).

Human immunodeficiency virus (HIV) may involve the heart in a number of ways. Between 9% to 34% of HIV-positive individuals will have demonstrable cardiac abnormalities. Present estimates suggest that troponin elevation is nonspecific in this population, and it roughly parallels the increased incidence of coronary events associated with HIV complications that impair coronary blood flow (drug-related hypercoagulability, endothelial dysfunction, and vasculitis) (177).

Reviews on Lyme carditis did not identify troponin data linked to diagnosis, treatment, or outcome ((178),179). A series of 91 children with dengue hemorrhagic fever or dengue shock syndrome found no patients with troponin T elevation, despite demonstrating that 36% of victims had significant acute reductions in LVEF over the course of the illness (180). Troponin T levels were not statistically different between Plasmodium falciparum–infected patients with uncomplicated malaria and those with clinical cardiac involvement (181). Of those 540,824 military personnel who received smallpox vaccinations, 67 developed myopericarditis, and 81.6% of these personnel showed a significant troponin I elevation (mean 14.1 ng/ml), occurring an average of 10 days post-immunization; 96% had fully recovered after 32 weeks, with normalized troponin I levels. Because samples were drawn on only vaccine recipients manifesting symptoms, and with the total recovery rate so high, this study suggested no prognostic value of troponin levels in this setting (182). In 2 small series, snake bites resulted in troponin elevation in 2 of 24 patients (183), but none of another 7 victims showed troponin elevation (184). Reports of other envenomations (e.g., jellyfish and scorpions) have demonstrated troponin elevations in as many as 20% of victims (185). No long-term outcomes have thus far been linked to troponin status in envenomations.

Acute rheumatic fever—always assumed to be a pancarditis based on demonstration of intramyocardial Aschoff bodies—manifests minimal if any troponin abnormalities ((186),(187),(188),(189),190). Thus, to date, there appears to be no value in measuring or following troponin in patients with acute rheumatic fever, with or without obvious acute carditis.

Cardiac troponin I elevation has been documented in 22% to 71% of patients with clinical pericarditis; positive levels ranged from 0.5 ng/ml to >50 ng/ml ((191),192). One study suggested that troponin I elevation was much more common among patients with idiopathic pericarditis than among those with a demonstrable cause of peri/myocardial inflammation (193). Overall, with follow-up as long as 31 months, in multivariable analyses, troponin I positivity was not predictive of any clinically relevant outcomes (symptom recurrence, hospital readmission, tamponade, constriction, or ventricular dysfunction) (193). Deaths in these studies were too infrequent to determine any relationship with troponin I levels (194), which in turn do not appear to add power to existing risk models (192).

Most current assays for cardiac troponin are robust with respect to both sensitivity and analytic performance around the lower limits of detectability. With rare exception, these assays are able to selectively detect cardiac troponin to the exclusion of troponin from other tissues. Therefore, the premise moving forward is that the values obtained are in fact accurate and do reflect a release of troponin from myocytes into the systemic circulation. (Table 3) summarizes answers to some of the frequent questions regarding the use of troponin in the clinical setting.

The challenge is how to calculate the specificity of troponin elevation for ACS and apply this to clinical decision making. Specificity requires both a clinical definition and an existing “gold standard” as a basis to compare the results of a test. The point of this paper is to provide the framework for clinicians to interpret the results of troponin testing in a useful mechanism-based construct. The first distinction to be made is that elevated troponin in and of itself does not indicate MI; rather, it is a sensitive and specific determinant of myocardial necrosis that is nonspecific relative to the etiology of that necrosis. The current diagnosis of MI is limited to a specific clinical condition in which myocytes are compromised by ischemia (see Figure 69_gr1), whether that is related to acute plaque rupture (Type 1), other ischemic etiologies (Type 2), or regional or global insults related to revascularization procedures (Types 4a and 5) (see Figure 69_gr2). As a further point of distinction, MI is not synonymous with ACS (plaque disruption with thrombosis), since ischemia can occur via a number of other mechanisms, including the most common coronary insufficiency resulting from fixed (stable) lesions and increased demand.

As troponin assays become more sensitive, there will be an increasing number of conditions discovered that result in low-level troponin elevations. It has been shown in population-based studies that with high-sensitivity assays, even a proportion of apparently healthy, normal population distributions will have detectable troponin levels, and a small proportion of apparently healthy individuals will have elevations above the 99th percentile. In that context, there will need to be a renewed reliance on the diagnostic model that was put forward by the World Health Organization in 1979, which required both ECG as well as the clinical components and biomarkers of necrosis to make a diagnosis of MI. This concept is addressed in Section (4), where the argument is made that the diagnosis of MI follows a Bayesian model, and as such, an important consideration in using these tests should be a pre-test probability assessment that will influence the post-test interpretation of the result. That model is best developed for the acute coronary syndromes, including MI.

Even when troponin elevation is not thought to be diagnostic for MI, there remains an imperative to determine the true etiology because in most cases, the result will provide some prognostic information. Because for most etiologies, little is known about specifically what to do clinically to manage patients in these settings, routine testing is not indicated.

From a clinician's perspective, the first priority is to understand when (and why) to order (or not order) a troponin test. The best value of troponin testing remains in the diagnosis of MI. Therefore, in the setting of symptoms suggestive of ischemia and a nondiagnostic ECG, serial troponin testing is invaluable and has high sensitivity and specificity (see discussion of Bayes' Theorem in Section 4.2), especially when temporal changes in troponin level are considered. Even in the setting of MI, it is important to understand the clinical context as treatment may vary considerably (e.g., between Type 1 and Type 2 MI). Therefore, it is crucial that the correct assignment is made according to the 2012 Universal Definition of Myocardial Infarction (13) and that the patient is treated accordingly.

It must be remembered that the sensitivity and specificity of troponin are for myocardial necrosis and not for infarction. Therefore, it is incumbent upon the clinician to attempt to ascertain the reason for an elevation as in most cases it provides important prognostic information and, in some cases, will guide therapy. Myocardial necrosis is a laboratory diagnosis that does not imply an etiology, whereas MI is a clinical diagnosis. As troponin assays become increasingly sensitive, understanding the clinical scenario will become increasingly important in deciding who to test, and integration of clinical data along with laboratory data will become even more crucial to the diagnosis resulting from testing.

William A. Zoghbi, MD, FACC, President

Thomas E. Arend Jr., Esq, CAE, Interim Chief Staff Officer

William J. Oetgen, MD, MBA, FACC, Senior Vice President, Science and Quality

Charlene L. May, Senior Director, Science and Clinical Policy

Dawn R. Phoubandith, MSW, Director, ACCF Clinical Documents

Tanja Kharlamova, Associate Director, Clinical Policy and Documents