BACKGROUND: Myocardial infarction is diagnosed when biomarkers of cardiac necrosis exceed the 99th centile, although guidelines advocate even lower concentrations for early rule-out. We examined how many myocytes and how much myocardium these concentrations represent. We also examined if dietary troponin can confound the rule-out algorithm.
METHODS: Individual rat cardiac myocytes, rat myocardium, ovine myocardium, or human myocardium were spiked into 400-μL aliquots of human serum. Blood was drawn from a volunteer after ingestion of ovine myocardium. High-sensitivity assays were used to measure cardiac troponin T (cTnT; Roche, Elecsys), cTnI (Abbott, Architect), and cardiac myosin-binding protein C (cMyC; EMD Millipore, Erenna®).
RESULTS: The cMyC assay could only detect the human protein. For each rat cardiac myocyte added to 400 μL of human serum, cTnT and cTnI increased by 19.0 ng/L (95% CI, 16.8–21.2) and 18.9 ng/L (95% CI, 14.7–23.1), respectively. Under identical conditions cTnT, cTnI, and cMyC increased by 3.9 ng/L (95% CI, 3.6–4.3), 4.3 ng/L (95% CI, 3.8–4.7), and 41.0 ng/L (95% CI, 38.0–44.0) per μg of human myocardium. There was no detectable change in cTnI or cTnT concentration after ingestion of sufficient ovine myocardium to increase cTnT and cTnI to approximately 1 × 108 times their lower limits of quantification.
CONCLUSIONS: Based on pragmatic assumptions regarding cTn and cMyC release efficiency, circulating species, and volume of distribution, 99th centile concentrations may be exceeded by necrosis of 40 mg of myocardium. This volume is much too small to detect by noninvasive imaging.
Cardiac troponin (cTn),5 released in the setting of myocardial necrosis, is engrained in the universal definition of myocardial infarction (1). This definition incorporates a diagnostic threshold concentration at the 99th centile for the general population. More recent guidelines suggest rule-in and rule-out cutoffs widely spaced around the 99th centile and utilize small changes in concentration between first and second blood draw (the δ values) to identify acute myocardial injury (2). The guideline published by the European Society of Cardiology lists δ values derived from observational studies (3–5) and advocates changes in the concentration of high-sensitivity cardiac troponin T (hs-cTnT; Roche) <3 ng/L and hs-cTnI (Abbott) <2 ng/L, to identify patients for early discharge (rule-out). The low magnitude of the absolute and δ concentrations used to rule out myocardial infarction (MI) have stimulated debate (6–9). To our knowledge, no previous study has established how many myocytes, or how much myocardium, needs to undergo necrosis to exceed the 99th centile, rule out cutoff, or rule out δ values. We set out to address this question by simulating myocardial injury using defined numbers of cardiac myocytes and quantities of myocardium spiked into human serum.
Adult rat ventricular myocytes were isolated from the hearts of male Wistar rats weighing approximately 200–250 g (B&K Universal Ltd.) by a collagenase-based enzymatic method as described in (10). In brief, hearts were excised from terminally anesthetized and heparinized (60 mg/kg sodium pentobarbitone and 100 U sodium heparin, intraperitoneally) rats. Excised hearts were immediately cannulated and initially perfused for 5 min with HEPES-Tyrode solution containing the following (mmol/L): 130 NaCl, 4.5 MgCl2, 0.4 NaH2PO4, 0.75 CaCl2, 4.2 HEPES, 20 taurine, 10 creatine, and 10 glucose. Hearts were then consecutively perfused with Ca2+-free HEPES-Tyrode solution containing 100 μmol/L EGTA (10 min) and HEPES-Tyrode solution containing 100 μmol/L CaCl2 and 1 g/L type II collagenase (Worthington Biochemical Corp., 8 min). All solutions were gassed with 100% O2 and maintained at 37 °C. Hearts were then removed from the perfusion apparatus, the ventricles were cut into small pieces, and agitated for a further 7 min at 37 °C. Isolated myocytes were separated from the undigested ventricular tissue by filtering through 200-μm nylon gauze, and the cells were allowed to settle by gravity (8 min). The supernatant was removed and replaced with HEPES-Tyrode solution containing 1% BSA and 500 μmol/L CaCl2. Myocytes were again allowed to settle, the supernatant was removed, and the cells were finally pooled and resuspended in 30 mL of HEPES-Tyrode solution containing 1 mmol/L CaCl2. The pooled isolated myocytes were pelleted by brief centrifugation at 50g and washed at room temperature with modified M199 culture medium (Invitrogen) containing 2 mmol/L Creatine, 2 mmol/L carnitine and 5 mmol/L taurine supplemented with 100 IU/mL penicillin/streptomycin. Following further centrifugation at 50g, myocytes were finally resuspended in modified M199 medium. Myocytes were then plated onto 6-well culture plates precoated with laminin and allowed to attach for 90 min in an incubator (37 °C, 5% CO2). Unattached cells were removed after preplating for 2 h and the culture medium was replaced with fresh modified M199 medium, and the cells were maintained overnight. Cultured rat cardiomyocytes were subsequently resuspended in Tyrode solution. Trypan blue staining revealed a viability of 45%. Cells were allowed to settle in Tyrode's solution. The solution was subsequently removed and replaced with 10 mL Tris (20 mmol, pH 7.5), then centrifuged at 94g for 3 min. The wash supernatant was discarded, and the cell pellet resuspended in fresh Tris solution. Cell count was calculated using an automated cell counter (Bio-Rad TC20™). The results of the automated cell counter were calibrated to manual cell counts attained by visual inspection with a hemocytometer. The 10 mL of resuspended pellet was then ultrasonicated (6 × 10-s bursts on ice, with 10-s intervals on ice). Dilutions of this solution were then spiked into 400 μL of banked human serum.
Experiments with cultured rat myocytes were repeated using 4 different human serum samples to account for donor-dependent interaction between human serum and rat protein. Experiments were repeated once for 2 of the serum donors using a different stock solution of cultured myocytes, so as to account for variation between culture preparations. Cells were spiked into serum in increments of 10 cells, ranging from 1 cell to 90 or 100 cells (limited by the availability of cells and serum). These repeat experiments resulted in a total of 62 samples for assessment of linear correlation.
Human myocardium was obtained from an explanted failing heart under Ethical Approval from the Royal Brompton and Harefield Trust BRU Biobank, which complies with the Helsinki declaration of 1975. The tissue was transported in cardioplegia, and frozen at −80 °C. Frozen myocardium was weighed (the exact weight was recorded), and the tissue was crushed in a percussion mortar for 10 s. Buffer solution [50 mL Tris, pH 7.5, with 1 tablet protease inhibitor (complete EDTA-free, Roche)] was added to the pulverized tissue (1 mL of buffer per 100 mg of tissue). The subsequent solution was ultrasonicated on ice (6 × 10-s bursts on ice, with 10-s intervals on ice). Following ultrasonication, the solution was centrifuged at 21.130g for 30 mins at 4 °C. The supernatant was frozen in liquid nitrogen and then stored at −80 °C. Dilutions of this solution were then spiked into 400 μL of banked human serum.
Experiments using human myocardium were limited in number by availability of myocardial tissue, and were repeated using serum obtained from 3 different donors. Myocardium was spiked into serum at a concentration of 1 μg, 10 μg, and increments of 10 μg up to 100 μg. With addition of blank controls (serum + buffer), this resulted in a total of 36 samples for assessment of linear correlation.
DIETARY TROPONIN CONSUMPTION
Ovine left ventricular myocardium was boiled in water for 3 h and then mechanically homogenized using a glass handheld homogenizer. Buffer solution [50 mL Tris, pH 7.5 with 1 tablet protease inhibitor (complete EDTA-free, Roche)], was added to the homogenized tissue (1 mL of buffer per 100 mg of tissue). The subsequent solution was ultrasonicated on ice (6 × 10-s bursts on ice, with 10-s intervals on ice). Following ultrasonication, the solution was centrifuged at 21.130g for 30 mins at 4 °C. The supernatant was frozen in liquid nitrogen and then stored at −80 °C. Dilutions of this solution were then spiked into 400 μL of banked human serum to generate a calibration curve. A healthy human volunteer with a baseline serum troponin (T and I) below the limit of detection had a 200-g dietary load of ovine left ventricular myocardium boiled for 3 h. Serial venipuncture was performed from an antecubital fossa vein at 15, 60, 120, 180, 240, 1320, and 1640 min after ingestion.
The concentrations of cTnI and cTnT were measured using contemporary high-sensitivity assays [Abbott Architect, limit of detection (LoD) 1.9 ng/L; and Roche Elecsys, LoD 5 ng/L, respectively], (11–13). Cardiac myosin-binding protein C (cMyC) was measured by EMD Millipore on the Erenna® platform using proprietary reagents as recently described (LoD of 0.4 ng/L) (14). A triplicate standard curve was run and used to interpolate the data.
For all samples, the biomarker concentration in the blank control was subtracted from the total biomarker concentration of each sample, to control for variation in troponin concentrations within the human serum or the background signal generated by buffer alone. Samples of Tris buffer added to banked human serum (serving as controls for each experiment) returned cTnT values between LoD and 6.98 ng/L, cTnI values between LoD and 5.00 ng/L, and cMyC values between 11.37 and 26.78 ng/L.
Likely compatibility of the high sensitivity troponin assays with rat cTn was assessed by basic local alignment search tool (BLAST) comparison of the amino acid sequence of rat troponin and the detection/capture epitopes for the high-sensitivity assay antibodies (see Fig. S1 in the Data Supplement that accompanies the online version of this article at http://www.clinchem.org/content/vol63/issue5). The amino acid sequence for these epitopes was largely conserved between the human and rat proteins. This molecular suggestion of compatibility was borne out in the strong signal detected by the assays in rat and ovine cardiac tissues.
Linear regression analysis was used to assess correlation, and standardized residuals greater than ±3 SDs were excluded as outliers (n = 1). Statistical analysis was conducted using SPSS version 22 (IBM Corp.) and R version 3.3.0 (GUI 1.68, The R Foundation for Statistical Computing).
The cMyC assay did not detect rat cMyC since the capture and detection antibodies are directed at human-specific sequences [see Fig. 3 in (15)]. There is a strong linear correlation between rat cardiomyocyte number and cTn concentration (cTnI; R2 = 0.58, P <0.001, n = 61; cTnT; R2 = 0.83, P <0.001, n = 62; Fig. 1). We were able to detect, in 400 μL of serum, a cTn increase resulting from a single cardiomyocyte. The slope coefficients for both cTnI and cTnT were similar [cTnI slope = 18.9 ng.L−1/cell (95% CI, 14.7–23.1); cTnT slope = 19 ng.L−1/cell (95% CI, 16.8–21.2) and the lines of regression did not deviate significantly from the origin [cTnI y-intercept = −44.6 ng.L−1 (95% CI, −128.8 to 39.5) and cTnT y-intercept = 24.4 ng.L−1 (95% CI, −18.9 to 67.7]. One outlier was identified during linear regression with a standardized residual >3 SDs. No other outliers were identified, and in the context of strong linear correlations either side of this point, this outlier was assumed to represent a user-operated pipetting error.
In experiments using human myocardium, both the cTns and cMyC were strongly linearly correlated with mass of myocardium [cTnI R2 = 0.92, P <0.001, n = 36, slope = 4.3 ng.L−1/μg (95% CI, 3.8–4.7), y-intercept = 4.4 ng.L−1 (95% CI, −20.1 to 28.8)], [cTnT R2 = 0.93, P <0.001, n = 36, slope = 3.9 ng.L−1/μg (95% CI, 3.6–4.3), y-intercept = 19.0 ng.L−1 (95% CI, −1.6 to 39.5)], [cMyC R2=0.96, P < 0.001, n = 36, slope = 41.0 ng.L−1/μg (95% CI, 38.0–44.0), y-intercept = 91.1 ng.L−1 (95% CI, −79.3 to 261.4)] (Fig. 2).
Cooked ovine myocardium had a much greater troponin content than human myocardium, and a robust linear correlation was established between mass of myocardium and troponin release [hs-cTnI R2 = 0.992, P <0.0001, n = 12, slope = 4928 ng.L−1/μg (95% CI, 4616–5241)], [hs-cTnT R2=0.998, P < 0.0001, n = 12, slope = 11512 ng.L−1/μg (95% CI, 11225–11798)]. Despite this extreme sensitivity, at all measured time points following an oral load of similarly processed ovine myocardium, the concentration of both cTnI and cTnT remained below the LoD for their respective assays when measured in human peripheral venous circulation (Fig. 3).
This study documents the extreme sensitivity of the high-sensitivity cTn assays, which are capable of detecting release from a single cardiomyocyte in a 400-μL blood sample. All investigated biomarkers correlate strongly with the mass of human myocardium. The observation that each microgram of human myocardium releases less cTn than a single cardiomyocyte most likely results from the heterogeneous cellular makeup of the human heart samples and the difficulty in efficiently liberating sarcomeric protein.
Despite the extreme sensitivity of cTn assays, we were unable to detect exogenous cTn in the peripheral blood stream after an oral load.
Based on our results, if we assume a circulating plasma volume and biomarker distribution of 2.75 L without clearance, the 99th centile concentrations for high-sensitivity assays for cTnT (Roche, 14 ng/L), cTnI (Abbott, 26 ng/L), and cMyC (87 ng/L) can be exceeded by necrosis of 0.025 g, 0.042 g, and 0.015 g of myocardium, respectively (for calculation, see the online Data Supplement). The recent guideline by the European Society of Cardiology defines “rule-out” and δ values for cTn, below which cardiac injury is unlikely (2). These values are close to the LoD concentrations of the high-sensitivity assays for cTnT and cTnI (5 ng/L and 1.9 ng/L, respectively). Our experiments suggest that 9 mg and 3 mg of human myocardial necrosis are required to increase cTnT and cTnI above their LoDs as measured by high-sensitivity assays, respectively. The corresponding value for cMyC (LoD 0.4 ng/L) is 0.07 mg. Our experiments simulate a scenario of complete myocardial necrosis, with subsequent rapid reperfusion and distribution of the coronary effluent into the systemic circulation. We have also ignored the circulating species of cTnI, cTnT, and cMyC. Collectively these, and other unknown factors, make it likely that diagnostic thresholds would in reality require myocardial necrosis more substantial than we have predicted.
The models of myocardial necrosis we have adopted are reductionist and convenient but differ markedly from necrosis of blood-perfused myocardium in vivo. Firstly, the process of cardiomyocyte necrosis in vivo is more complex than tissue homogenization. The vast majority of cTnI, cTnT, and cMyC resides in the crystalline sarcomere and release from this compartment is slow. The cause for this slow and incomplete release is uncertain but is likely related to the quality of reperfusion since the temporal profile of cTnT differs markedly between alcohol septal ablation (low microvascular reflow) and cardioplegia (high microvasculature reflow) (15). In addition, myocardial cTnT can be readily released from myocardium by serum alone, without the need of specialist extraction buffers (16). Although, we didn't measure the cTnI, cTnT, or cMyC remaining in the insoluble fraction (pellet after centrifugation) following homogenization, it is likely extraction was inefficient since the concentration of cTnT we observed in human myocardium is lower than those published previously (16, 17). Secondly, the protracted release in vivo provides ample opportunity for posttranslational modifications, that will be absent in our models of rapid myocardial homogenization in calcium-free lysis buffers containing protease inhibitors. For example, cTnI, cMyC, and cTnT appear in the circulation as peptides, as well as intact proteins (15, 18–20). In the case of cMyC, calpain-mediated cleavage is regulated by phosphorylation events within the M domain that impede the formation of an N-terminal peptide (21) that is both immunogenic (22) and negatively inotropic (23). Immunoassays may not have the same sensitivity for these modified forms of cTnI, cTnT, and cMyC as they do for the parental unmodified protein (18, 24). If the cleavage event occurs between the epitopes recognized by the capture and detection antibodies, the immunoassay will become insensitive. Conversely, other posttranslational modifications are known to enhance the sensitivity of immunoassays. For example, the sensitivity of assays for cTnI can be enhanced by oxidation leading to intramolecular disulfide formation and also when cTnI is in a binary or ternary complex with cTnC and cTnT (18). Furthermore, the abundance of circulating cTnI, cTnT, and cMyC peptides changes over time in an individual patient as well as between patients (15, 18, 20). There can also be marked differences between individual patients in the proportions of cTnT and cTnI that appear in apo vs binary and ternary forms (18). These complexities were absent in our experiments where heart tissue was rapidly homogenized, protease inhibitors prevented protein cleavage, opportunities for oxidation were limited and there was no added calcium to maintain ternary complexes of cTnI/cTnC/cTnT (18). These details illustrate the simplistic assumptions made in our in vivo extrapolation of volumes of myocardial necrosis needed to cross diagnostic thresholds. Nonetheless, they do not invalidate the extreme analytic sensitivity of the cTnT, cTnT, and cMyC assays and the microscopic nature of the myocardial necrosis events that drive clinical decision-making.
Several publications have previously challenged the thresholds for AMI (acute MI) rule-out as well as the application of very small δ values in clinical practice (6–8, 25). In their analysis, Chenevier-Gobeaux et al. questioned the cutoff for rule-out at the limit of detection (5 ng/L for cTnT), prompting Peter Kavsak to comment that there might “be other analytical factors at play that affected the performance of hs-cTnT” (6, 7). Hickman et al. (8) further highlighted the challenges associated with such narrow decision limits and the use of deltas when the cTn release might have plateaued—as is the case with late presentations to the emergency department and cTn release caused by events other than atherosclerotic plaque instability. Turer et al. have added to the diagnostic conundrum by describing low-level cTnT release in the context of ischemia without frank infarction (25). Although new myocardial scarring on cardiac magnetic resonance imaging (cMRI) has been correlated to cTn measured with a second-generation assay (26), this is limited by the inability of cMRI to detect infarct size smaller than a gram of myocardium. To our knowledge, no previous study has directly correlated cardiac damage at a tissue or cellular level to biomarker concentrations measured using contemporary high-sensitivity assays.
While there is a significant body of evidence to suggest that di- and tripeptides, derived from dietary protein can cross the gastrointestinal tract into the portal circulation, the absorption of larger intact polypeptides is controversial (27). Based on the closely spaced capture/detection epitopes used by the cTnI and cTnT assays, 20-mer polypeptides should generate a signal (see Appendix in the online Data Supplement). Given the exceptional sensitivity of the high-sensitivity troponin assays, we hypothesized that we would detect longer cTnI and cTnT polypeptides in the systemic circulation. This hypothesis was supported by the fact that both of the high-sensitivity assays for cTnT and cTnI were substantially more sensitive to cTn in cooked ovine than in human myocardium. Following a meal of 200 g of ovine myocardium, venous blood was sampled from a healthy human volunteer at hourly time intervals. Assuming that (a) the human digestive tract is able to liberate all the cTnT in ovine myocardium in a manner similar to ultrasonication, and (b) the entire quantity of liberated cTnT is able to cross into the systemic circulation, extrapolation of the calibration curve indicates that serum cTnT should be increased by 335 mg/L. The hs-cTnT assay has sensitivity to detect 5 ng/L increments in serum troponin. As such, only 1.5 × 10−6% of the dietary cTnT needed to be released from the myocardium and absorbed into systemic circulation to produce a detectable increase in serum cTnT. Nonetheless, measured cTnT remained consistently below the LoD at all time points after the dietary load. The same observation was made with cTnI. Even with the unrealistic assumption of complete liberation of troponin from the full mass of ingested myocardium, the orders of magnitude involved suggest the gastrointestinal tract is virtually impervious to absorption of intact polypeptides of troponin.
In conclusion, we have, for the first time, correlated the 99th centile thresholds of cTn to the approximate mass of myocardium undergoing complete necrosis. Our experiments have revealed the exquisite sensitivity of the contemporary biomarker assays, with necrosis of just 40 mg of myocardium, equivalent to 0.015% of the heart, sufficient to increase serum concentrations above the 99th centile.
↵† Jack Marjot and Thomas E. Kaier contributed equally to the work, and both should be considered as first authors.
↵5 Nonstandard abbreviations:
- cardiac troponin;
- high-sensitivity–cardiac troponin T;
- myocardial infarction;
- limit of detection;
- cardiac myosin-binding protein C;
- cardiac MRI.
Author Contributions: All authors confirmed they have contributed to the intellectual content of this paper and have met the following 3 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; and (c) final approval of the published article.
Authors' Disclosures or Potential Conflicts of Interest: Upon manuscript submission, all authors completed the author disclosure form. Disclosures and/or potential conflicts of interest:
Employment or Leadership: M. Iqbal, Viapath pathology; S. Harding, Imperial College London.
Consultant or Advisory Role: S. Harding, BHF, MRC, Rosetrees, and British Government Science and Technology Committee.
Stock Ownership: None declared.
Honoraria: None declared.
Research Funding: The UK Department of Health through the National Institute for Health Biomedical Research Centre Award to Guy's & St Thomas' National Health Service Foundation Trust. T.E. Kaier, British Heart Foundation (TG/15/1/31518 and FS/15/13/31320); S. Harding, BHF; M. Marber, The Medical Research Council (UK; G1000737), Guy's & St Thomas' Charity (R060701, R100404), and British Heart Foundation (TG/15/1/31518 and FS/15/13/31320).
Expert Testimony: None declared.
Patents: M. Marber, patent no. US 8,546,089 B2, a patent held by King's College London for the detection of cMyC as a biomarker of myocardial injury.
Role of Sponsor: The funding organizations played no role in the design of study, choice of enrolled patients, review and interpretation of data, and final approval of manuscript.
Millipore Sigma (Hayward, CA) was contracted to undertake the analyses of cMyC on a fee-for-service basis and holds no commercial interest.
- Received for publication July 26, 2016.
- Accepted for publication January 6, 2017.
- © 2017 American Association for Clinical Chemistry