Highly sensitive troponin assays — a two-edged sword?

The advent of cardiac troponin (cTn) assays has redefined acute myocardial infarction (AMI) and revolutionised the care of patients with suspected AMI presenting to emergency departments (EDs).1 So central has cTn measurement become to the diagnosis of AMI that, since 2000, the formal criteria start with detection of rise and/or fall in serum troponin levels (with at least one value above the 99th percentile of the value distribution of a reference population for an assay with optimal precision at this level, defined as a coefficient of variation ≤ 10%), to which clinical evidence of myocardial ischaemia is added in regard to symptoms, electrocardiogram (ECG) changes or findings on cardiac imaging.2 This revised definition of AMI, with cTn assays using 99th percentile cut-off values, has altered the epidemiology of the disease. Data from Western Australia suggest that in the two decades before the advent of cTn testing in 1998, age-specific hospitalisation rates for AMI had decreased by an average of 30%, but this downward trend was abolished between 1998 and 2004.3

Using the logic that detection of even lower levels of cTn may assist in earlier AMI diagnosis and improved risk stratification, each new generation of cTn assays has been developed with the aim of greater sensitivity. As the limit of detection has progressively decreased, so has the 99th percentile threshold for labelling a given cTn value as abnormal and potentially diagnostic of AMI. This has occurred in tandem with improved ability to detect small changes in cTn levels at these lower concentrations. While first-generation assays identified 99th percentile concentrations at 0.5 μg/L, new fifth-generation highly sensitive (hs) cardiac troponin T (cTnT) immunoassays can identify concentrations as low as 14 ng/L with reasonable precision.4 Indeed, cTnT levels as low as 3 ng/L can now be detected in up to 66% of seemingly healthy individuals, compared with detection of cTnT in 10% to 20% of the same individuals using assays that are currently available for clinical use.5

It is assumed that increasingly sensitive cTn assays will result in higher diagnostic rates for AMI and that, as a consequence of providing appropriate therapies to patients with newly diagnosed AMI, risk of death and recurrent cardiac events will fall. A recent before–after study involving 2092 patients with suspected acute coronary syndrome (ACS) has provided early support for this assumption — lowering the 99th percentile cut-off value from 200 ng/L to 50 ng/L for a sensitive assay led to an 8 percentage point increase in the number of patients diagnosed with AMI, while incidence of death and recurrent AMI at 12 months among patients with cTn cut-off values between 51 ng/L and 199 ng/L significantly decreased from 39% to 21%.6 In other studies, the use of hs-cTn assays compared with previous-generation assays has resulted in between 9% and 27% of patients with chest pain being recategorised to AMI on the basis of hs-cTnT results alone.7-9

The advent of hs-cTn assays also holds the promise of detecting or ruling out AMI earlier as a result of improved sensitivity. In a study of 718 patients that compared three sensitive cardiac troponin I (cTnI) assays and one hs-cTnT assay with the standard fourth-generation cTnT assay,10 the sensitivity and negative predictive value of the hs-cTnT assay were superior to those of the standard assay among patients who presented within 3 hours of symptom onset (Box 1). In another study, use of hs-cTnT assays reduced the average time to confirm or exclude AMI after presentation from 4 hours to 71 minutes.11 However, a more recent study comparing a hs-cTnI assay with a contemporary cTnI assay, using a diagnostic cut-off value at the 99th percentile of 30 and 32 ng/L respectively, revealed identical negative predictive values on admission (94.7% v 94.0%) and at 3 hours after admission (99.4% for both).12

Where hs-cTn may be superior as a rule-out criterion relates to whether troponin can be detected at all on presentation to the ED. In a prospective evaluation of two patient cohorts totalling 1618 patients presenting with chest pain, only one of 355 patients (0.3%) who had undetectable cTnT according to a hs-cTnT assay at presentation to the ED (< 3 ng/L) subsequently had elevated levels consistent with AMI.13 This equated to sensitivity ranging from 99.8% to 100% and negative predictive value ranging from 99.4% to 100%.

However, despite their potential to facilitate decision making and their early adoption by many laboratories, hs-cTn assays have amplified some longstanding questions regarding all cTn assays. To determine the utility of hs-cTn assays in routine clinical practice, these questions need to be answered.

While the only hs-cTn assay which is currently commercially available in Australia has already been adopted by many laboratories, its use is by no means universal, although the shift to hs-cTn assays is likely to accelerate with the introduction of one or more hs-cTnI assays in 2012. However, hs-cTn assays have the potential to be a two-edged sword if they are not used carefully — for example, without monitoring for potentially deleterious effects. There is little doubt that positivity rates for hs-cTn assays will be considerably greater than for current assays. For example, an Italian study reported a 2.4-fold increase in the proportion of ED patients with positive test results — from 19% to 45% — following introduction of hs-cTnT assays, which led to an 85% increase in the number of patients hospitalised (based on a comparison of the first 3 months after assay introduction and the same period in the previous year).28 Positivity rates in a New Zealand study more than doubled — from 22% to 50% of patients (based on assessment of patient blood samples using both hs-cTnT and fourth-generation assays).9

This increased positivity rate includes many false positives for AMI as a result of the lower specificity and lower positive predictive value of hs-cTn assays compared with fourth-generation assays (Box 1). Among 1000 patients with an AMI probability of 10%, it has been estimated that hs-cTn assays (compared with standard assays) would produce 12 additional positive results that meet the definition of AMI and 117 that do not meet the definition; with an AMI probability of 17%, hs-cTn assays would produce 21 additional positive results that meet the definition and 108 that do not (Box 2).29 In low-risk patients (AMI probability, ≤ 5%), additional numbers of positive results meeting the AMI definition would be eight or fewer, and those not meeting the definition would be up to 126.29

These estimates pose a considerable logistical challenge for cardiologists, general physicians and ED physicians as the vastly increased numbers of patients who test positive for cTn using hs-cTn assays could invoke more serial cTn testing and further cardiac investigations, resulting in longer hospital stays, ED overcrowding, and more admissions to acute medical assessment and chest pain units. This increased resource utilisation may not be offset by more rapid discharge and avoidance of testing in the no more than 25% of patients with chest pain who test negative for cTn using hs-cTn assays at presentation.

More before–after studies involving hospitals that have introduced hs-cTn assays would provide clarification about real-world effects. The opportunity to conduct prospective studies that assess clinical outcomes and cost-effectiveness of hs-cTn assays in comparison to existing cTn assays should be seized as these new assays are introduced into sites where they have not previously been used. At the very least, we advise clinicians to interpret hs-cTn assay results with caution (Box 3) while awaiting results of future studies.