Lot-to-Lot Variation for Commercial High-Sensitivity Cardiac Troponin: Can We Realistically Report Down to the Assay’s Limit of Detection?

The release of high-sensitivity cardiac troponin T (hs-cTnT) and troponin I (hs-cTnI) assays in the US and worldwide has resulted in changes to the triage strategy for patients who present to the emergency department with ischemic symptoms suggestive of acute coronary syndrome including chest pain. For rule-out of acute myocardial infarction (MI), many emergency departments have been able to reduce the frequency of blood collection from the time of presentation plus 3–6 hours down to presentation plus 1–3 hours while maintaining a high negative predictive value ≥99.5%) (1,). Investigators have also shown that if the time from onset of symptoms to acute MI is 1–2 hours, very low cardiac troponin (cTn) concentrations can be used to rule out acute MI with a single admission sample (2–4,). If the patient presents earlier than 1–2 hours, this strategy may not be optimal (5). To fully implement a single-sample rule-out protocol, clinical laboratories must be able to report low cTn concentrations. Although laboratories in countries outside the US are able to report values down to the assay limit of detection (LoD), the US Food and Drug Administration (FDA) allows clinical laboratories to report cTn values down to only the assay’s limit of quantification (LoQ), which is conventionally defined as the lowest concentration with a 20% CV and is usually higher than the LoD. One of the FDA’s major concerns is assay variability. Inadequate performance between reagent lots has been documented for other widely used immunoassays. For cTn specifically, recall notices have been posted by the FDA for the Biosite Triage Assay, Beckman Access, and Roche cTnI (see Supplemental Data for additional references). An editorial from the FDA implies that variability of results between lots at low hs-cTn concentrations may give a false impression of low patient risk (6,). Data have been reported that this could lead to inappropriate baseline rule-out of MI, followed by emergency department discharge (7). In fact, the FDA already has policies in place to protect against the use of inappropriately low LoDs when assigning values. The FDA’s standard submission protocol for assigning the LoD requires manufacturers to determine this parameter over multiple lot numbers. The assay’s LoD concentration is defined by the highest value of the evaluated lots, usually rounded up. When LoD values are determined independently, they tend to be many folds lower than the package insert claims. For example, the LoD for the Beckman hs-cTnI assays is listed as 2.1 ng/L and was determined in 2 separate studies as 0.32 and 0.37 ng/L (8,). Similarly the Pathfast manufacturer’s published LoD was 7.0 ng/L (whole blood) and 5.0 ng/L (plasma) in its FDA submission, but an independent assessment yielded a value of 2.3 ng/L (9). This conservative assignment of LoD strongly supports the utility of package insert–listed LoD values as consistent, safe, and effective. The clinical experience worldwide of using hs-cTn assays provides valuable assistance regarding how clinical laboratories in the US can improve test performance at cTn concentrations near and below the LoQ. A key is the use of QC materials that challenge the low end of the reportable assay range. This recommendation was made by the AACC Academy, and the IFCC Committee on Clinical Applications of Cardiac Bio-Markers recommended the use of a QC material between the LoD and the lowest of the 99th percentile upper reference limits (10). Nevertheless, most commercial manufacturers of cTn reagents and QC materials do not yet offer QC materials that have low cTn concentrations. Consequently, preparation of preassayed frozen pooled serum or plasma pools is appropriate. The use of fresh and appropriately selected patient samples could be used to construct pools near the LoD that would be tested in addition to recommended QC materials for hs-cTn assays. The international experience with these low-troponin-concentration pools is documented. In Canada, after implementing a low control, Kavsak et al. (11,) assessed EDTA plasma samples (fresh, not frozen, to mimic real-time testing) and determined lot-to-lot variability for an hs-cTnI assay (Abbott Diagnostics US package insert: LoD, 1.7 ng/L; LoQ, 3.5 ng/L) with dozens of calibrators and reagent lots over 2 years. The overall mean difference between lots for 118 samples with concentrations measurable only by hs-cTn assays (i.e., ≤10 ng/L) was −0.07 ng/L (SD, 0.6 ng/L; 95% CI limits of agreement by Bland-Altman method: −1.3 to 1.1 ng/L). In Italy, Aloisio et al. (12,) used a lower QC material (near the LoD) for the hs-cTnT assay (Roche Diagnostics: LoD, 5 ng/L; LoQ, 6 ng/L). These investigators evaluated accuracy of this procedure by using external proficiency testing samples on a monthly basis. Before the implementation of this QC material, the results of 10 (40%) of 25 clinical laboratory–reported monthly challenges did not meet the maximum total error of ±22.5% recommended by the Australasian Association of Clinical Biochemists. After implementation, failure was reduced to only 1 of 26 monthly challenges (3.8%). During the intervening 51 months, the laboratory had multiple reagent and calibrator lot changes. Long-term data obtained from Norway (2013–2019 for hs-cTnT) (13) and from the Kavsak group in Canada (2014–2019 for hs-cTnI) also provide compelling evidence of assay stability and consistent lot-to-lot performance for both hs-cTnT (7–9 different frozen serum pools with hs-cTnT concentrations from 3.4 to 20.6 ng/L, as measured with the initial lot) and hs-cTnI (fresh EDTA plasma with hs-cTnI concentrations from 0.3 to 5.4 ng/L, as measured with the initial lot) (Fig. 1). The median differences for hs-cTnI (n = 103 paired samples) and hs-cTnT (n = 162 pairs) were −0.1 and −0.4 ng/L, respectively. The observed lot-to-lot bias was similar across concentrations well below the 99th percentile: −1.7 to 2.3 for hs-cTnI and −2.4 to 1.6 for hs-cTnT. Importantly, as shown in Fig. 1, the plot of differences in hs-cTnI and hs-cTnT concentrations between new and initial lots of reagents was flat—that is, the lot-to-lot bias was not greater at the lower cTn concentrations. Plot of cardiac troponin concentration differences between the new lot versus the current hs-cTnI and hs-cTnT. ○, Abbott EDTA plasma (data from Canada); •, Roche serum (data from Norway). Plot of cardiac troponin concentration differences between the new lot versus the current hs-cTnI and hs-cTnT. ○, Abbott EDTA plasma (data from Canada); •, Roche serum (data from Norway). These data provide promising evidence that the variability of different manufacturer lots can be mitigated for laboratories that use appropriate QC materials and methods and other quality assurance practices that challenge the low end of the calibration curve. These minimal observed biases between lots indicate that the published total error criteria can be used at concentrations below and above the LoD. This approach can be used to mitigate analytical error, thereby enabling laboratories to clinically report accurate concentrations at the LoD and clinicians to use values near the LoD. Analytical considerations aside, many published clinical studies support reporting down to the LoD with minimal (<1%) risk of 30-day adverse outcomes. Even studies using assays with relatively high imprecision (20%–30% CV) between the LoD and LoQ have demonstrated very high negative predictive value (>99%) for MI and low rates (<1%) of 30-day events using LoD thresholds. US centers are already validating and adopting accelerated diagnostic pathways that utilize a single-draw rule-out at the LoQ. However, use of the LoQ would be expected to result in lower negative predictive value (and the possibility of missed MIs) than if the LoD were used (14,). The FDA’s decision to censor concentrations between LoD and LoQ because of concerns regarding false-negative readings may actually lead to more missed MIs because of implementation of single-draw rule-out protocols at higher thresholds. Lowering the reporting limit to the LoD will also be important in the emerging role of hs-cTn for risk stratification in primary care. In a meta-analysis conducted by Parikh et al. (15,), individuals who had hs-cTnT values between the limit of blank and LoD had worse outcomes than those with values below the LoD. Note that at this time, we are not advocating that clinical laboratories should report values below the LoD (10). We are asking the FDA to provide guidance to manufacturers regarding what studies to perform to demonstrate lot-to-lot stability and imprecision, thereby enabling the reporting of assay results down to the LoD. In the interim, we advocate that the FDA permit clinical laboratories to report hs-cTn assay concentrations down to the LoD if they can demonstrate stability over many lots, as part of the Laboratory Developed Test exception. To facilitate such a use, the FDA would have to grant permission for manufacturers to enable laboratory reporting of hs-cTn assay values between the LoD and LoQ—something that global vendors are currently disabling only for US customers. From the Biomarkers of Acute Cardiovascular Disease Division of the AACC and the IFCC Committee on Clinical Applications of Cardiac Bio-Markers. hs-cTnT, high-sensitivity cardiac troponin T; hs-cTnI, high-sensitivity cardiac troponin I; MI, myocardial infarction; cTn, cardiac troponin; LoD, limit of detection; FDA, US Food and Drug Administration; LoQ, limit of quantification. 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. A.H.B. Wu, administrative support, provision of study material or patients; P.A. Kavsak, statistical analysis, provision of study material or patients; R.H. Christenson, provision of study material or patients; F.S. Apple, statistical analysis, administrative support, provision of study material or patients; A.S. Jaffe, statistical analysis, administrative support. Upon manuscript submission, all authors completed the author disclosure form. Disclosures and/or potential conflicts of interest. R.H. Christenson, The Journal of Applied Laboratory Medicine, AACC; D.N. Greene, The Journal of Applied Laboratory Medicine, AACC; F.S. Apple, Clinical Chemistry, AACC. P.A. Kavsak, Beckman, Roche, Siemens; R.H. Christenson, Beckman Coulter, Siemens Healthineers, Roche Diagnostics, Quidel Diagnostics, Becton Dickinson, Sphingotec; F.S. Apple, HyTest, Instrumentation Laboratory, Qurvo, Siemens Healthcare, Brava Dx; W.F. Peacock, Abbott, Beckman, Instrument Labs, Ortho Clinical Diagnostics, Osler, Roche, Quidel, Siemens; J.E. Hollander, Roche; J. de Lemos, Quidel Cardiovascular, Ortho Clinical Diagnostics; D. Morrow, Bayer Pharma, InCarda (DSMB), Merck, Novartis, Roche Diagnostics; J. Januzzi, Roche, Abbott, Siemens, Quidel; A.S. Jaffe, Abbott, Siemens, Roche, Beckman-Coulter, Quidel, ET Healthcare, Sphingotec, Brava, Blade, Novartis; A.H.B.Wu, Siemens, Brava, ET Healthcare, Quidel. None declared. P.A. Kavsak, Beckman, Roche, Siemens; R.H. Christenson, Beckman Coulter, Siemens Healthineers, Becton Dickinson, Quidel Diagnostics, Roche Diagnostics, Sphingotec; F.S. Apple, national/international meetings sponsored by Abbott and Siemens; J. Januzzi, Roche, Abbott, Quidel. F.S. Apple, Abbott, Abbott POC, Siemens, Beckman Coulter, Ortho-Clinical Diagnostics, Quidel; W.F. Peacock, Abbott, CSL Behring, Ortho Clinical Diagnostics, Roche, Siemens; J. Januzzi, Abbott; A.H.B. Wu, Siemens, Abbott. None declared. McMaster University has filed patents in the acute cardiac biomarker area with P. Kavsak listed as an inventor.