Clinical Relevance of the 99th Percentile Upper Reference Limit for High-Sensitivity Cardiac Troponin Assays

Contemporary clinical use of cardiac troponins (cTns) represents a quintessential bench-to-bedside research success story, spanning more than half a century from the first description of the troponin protein complex (1,). Following the development in the 1990s of the first scalable cTn assay that could be implemented for clinical use (2), large multicenter clinical studies around the turn of the century brought increasing recognition of cTns as the gold standard biomarkers of myocardial injury. The more recent regulatory approval of cTn assays with greater sensitivity and overall improved analytical performance (i.e., high-sensitivity cardiac troponin [hs-cTn] assays) denotes the latest chapter in this continuously evolving story, presenting both new opportunities and new challenges. According to the 2018 Fourth Universal Definition of Myocardial Infarction (UDMI), acute myocardial injury is represented by an acute rise and/or fall in cTn values, with at least 1 cTn value above the assay’s 99th percentile upper reference limit (URL) (3,). In turn, when considered in the appropriate clinical context suggestive of myocardial ischemia, biomarker evidence of acute myocardial injury is essential for the diagnosis of myocardial infarction (MI). Although this definition is based on sound clinical evidence and has undeniable pathophysiologic plausibility, it also has important limitations. First, despite recommendations from the UDMI committee to classify injury as acute if cTn concentrations are above the 99th percentile and if they change by >20% between serial measurements (3,), little evidence supports this criterion for magnitude of change. Second, the 99th percentile URLs for cTn assays are, by convention, derived from normal reference populations (4,), but there is no consensus regarding the most appropriate inclusion and exclusion criteria to define the normal reference population, nor is there consensus regarding statistical methodology. These limitations are particularly relevant in the era of hs-cTn assays, which can reliably detect cTn in a much larger proportion of presumably healthy populations than previous-generation assays (5). The greater analytical sensitivity of hs-cTn assays has some downsides for clinicians in the busy emergency room setting, where these tests are most commonly used. Simply replacing older assays with newer hs-cTn assays using manufacturer-recommended 99th percentile URLs may be problematic because many additional patients will be identified with cardiac injury, with most not meeting criteria for acute MI (6,). In the absence of specific recommendations for the interpretation of low to mildly increased cTn values, this deluge of abnormal cTn values may create challenges for clinicians, leading to potential misdiagnosis and unnecessary further testing (7,). However, even mild cTn increases below the 99th percentile URL are associated with structural heart disease and increased risk of mortality in the general population (8,) and are associated with notably increased risk among patients seen in the emergency department and ruled out for MI (9). Additional research is needed to define optimal evaluation and management strategies for patients with cardiac injury in the absence of acute MI. Although the 99th percentile URL remains front and center in the UDMI, another inherent limitation that has been accentuated with the transition to hs-cTn assays is dependence of the 99th percentile URL on sex, age, and possibly race. When a contemporary high-sensitivity cardiac troponin T (hs-cTnT) assay was studied in healthy individuals with no subclinical cardiovascular disease from 3 independent population-based cohorts, 99th percentile URLs were markedly higher in men compared with women, increased with age in both sexes, and trended higher in black participants (10). Notably, hs-cTnT values in that study were above the manufacturer-reported 99th percentile URL in >10% of men aged 65–74 years. The dependence of 99th percentile values on sex, age, and possibly race increases the risk of MI misdiagnosis when uniform thresholds are used for all patients. In this context, the study by Apple et al. (11) in this issue of Clinical Chemistry is particularly important because it represents a major collaborative initiative that provides important novel insight into the uncertainties and limitations of using manufacturer-recommended 99th percentile URLs for hs-cTn assays. The investigators determined the sex-specific 99th percentile URLs for different high-sensitivity cardiac troponin I (hs-cTnI) and hs-cTnT assays using the Universal Sample Bank initiated at the 2015 annual meeting of the AACC. The study population consisted of almost 700 women and men selected using a health questionnaire, with exclusion of statin users and additional exclusions based on hemoglobin A1c, N-terminal pro-B-type natriuretic peptide, and estimated glomerular filtration rate cutoffs. In addition to the use of a standard biomarker-based set of criteria to better define a healthy reference population, this work is notable for its scope, comparing 12 different hs-cTn assays (9 hs-cTnI and 3 hs-cTnT) in the same blood samples, and using 3 distinct statistical approaches to determine 99th percentile URLs for each assay. The study’s key findings are that 99th percentile URLs determined from the Universal Sample Bank vary between assays, both overall and by sex; differ substantially from the manufacturer-recommended values; and are affected by the statistical method used. These findings further underscore the urgent need for the development and implementation of better algorithms for hs-cTn assay interpretation in clinical practice, including reconsideration of currently used 99th percentile URLs based on manufacturer’s individual reference populations. However, in our opinion, a major limitation of this study requires consideration before recommending specific revisions to existing 99th percentile URLs: the Universal Sample Bank was derived from a self-selected volunteer population comprising younger individuals (mean age, 41 years), not representative of the target population typically undergoing evaluation for possible MI, who are on average 2 decades older. This is especially relevant because in the population-based cohorts discussed earlier, similar differences in age range were associated with 50%–100% higher 99th percentile URLs in older versus younger individuals, even after careful exclusion of participants with subclinical heart disease (10). We applaud the concept of creating a universal sample bank by this team of experts but believe the actual 99th percentile URL values reported here are of uncertain applicability, given their derivation from a reference population that does not reflect the demographics of the target population in which the test is used. This issue notwithstanding, the study by Apple and colleagues adds to mounting evidence that reliance on 99th percentile URLs for hs-cTn assays may be inherently flawed. This threshold was always arbitrary, even with older generation assays. With hs-cTn assays, however, the influence of age and sex (and possibly race), as well as common comorbidities such as chronic kidney disease, is magnified, to the point where it is questionable whether it is even possible to have a uniform 99th percentile threshold. One could speculate that future advances in precision medicine, including incorporation of machine learning and artificial intelligence-based algorithms, may eventually result in the implementation of MI diagnostic criteria that are not only sex- and age-specific but also tailored to each patient based on a panel of relevant individual characteristics. Preliminary evidence supports, for example, a machine learning approach using age, sex, and paired hs-cTn values as superior to a standard 0/3-hour hs-cTn protocol using the 99th percentile URL (12). Until such strategies are validated and operationalized, the most sensible solution to the conundrum surrounding the 99th percentile URL for hs-cTn assays may be to de-emphasize the importance of rigid hs-cTn cutoffs in clinical decision-making while placing more emphasis on appropriate strategies for MI rule-out and on serial cTn changes for acute MI diagnosis. Innovative protocols incorporating hs-cTn assays have been developed that capitalize on the augmented sensitivity and enhanced precision of these assays to allow more efficient rule-out of MI in the emergency department. Importantly, several of these protocols do not use the 99th percentile value at all, bypassing the issues raised above (13, 14,). The high precision of the hs-cTn assays allows exclusion of very small changes over serial measurements in 1–2 h (13,), and the high sensitivity allows ruling out MI using a single blood draw in some patients (15,). These protocols have been demonstrated to be safe, with a very high negative predictive value for MI, and result in MI rule-out in more patients and in shorter time compared with standard strategies focused on the 99th percentile URL (16). These algorithms do not solve the inherent uncertainties related to the optimal cutoff values for acute myocardial injury and MI rule-in. However, by focusing on change values and de-emphasizing the arbitrary 99th percentile, they are less influenced by age, sex, and comorbidity effects. Additional research should result in further refinement of algorithms to reliably confirm or rule out acute MI using hs-cTn, without necessarily depending on 99th percentile URL cutoffs. We commend Apple and colleagues for an important study that attempts to harmonize methods for determination of 99th percentile URLs for hs-cTn assays. The authors clearly demonstrate that we should not simply use manufacturer-recommended cutoffs for these assays. In fact, we believe that the implications of their study go even further, raising questions as to whether even a carefully determined 99th percentile URL can be considered reliable. Beyond this, we question whether the 99th percentile value is necessary, or even helpful, in contemporary clinical practice. Although we recognize the continued importance of diagnostic standards, particularly for endpoint adjudication and assessing epidemiologic trends, we believe that emerging strategies focusing on change values will prove more accurate, reliable, and clinically actionable than those using the 99th percentile threshold values. cTns, cardiac troponins; hs-cTn, high-sensitivity cardiac troponin; UDMI, Universal Definition of Myocardial Infarction; URL, upper reference limit; MI, myocardial infarction; hs-cTnT, high-sensitivity cardiac troponin T; hs-cTnI, high-sensitivity cardiac troponin I. All authors confirmed they have contributed to the intellectual content of this paper and have met the following 4 requirements: (a) significant contributions to the conception and design, acquisition of data, or analysis and interpretation of data; (b) drafting or revising the article for intellectual content; (c) final approval of the published article; and (d) agreement to be accountable for all aspects of the article thus ensuring that questions related to the accuracy or integrity of any part of the article are appropriately investigated and resolved. Upon manuscript submission, all authors completed the author disclosure form. Disclosures and/or potential conflicts of interest: Employment or Leadership: None declared. Consultant or Advisory Role: J. de Lemos, Roche Diagnostics, Abbott Diagnostics, Ortho Clinical Diagnostics, and Quidel Cardiovascular. Stock Ownership: None declared. Honoraria: None declared. Research Funding: M.O. Gore, a career development grant (K23-HL131939) from the National Heart, Lung, and Blood Institute (NHLBI), a University of Colorado School of Medicine Fund to Retain Clinical Scientists (FRCS) research award from the Doris Duke Charitable Foundation, and a Pilot Study Research Grant Program for Junior Investigators from Denver Health and Hospital Authority; J. de Lemos, Abbott Diagnostics and Roche Diagnostics. Expert Testimony: None declared. Patents: None declared.