Preparing for High-Sensitivity Cardiac Troponin Assays
Preparing for High-Sensitivity Cardiac Troponin Assays
With the ability to precisely measure small concentrations of cTn, clinicians will be faced with the challenge of distinguishing patients who have acute problems from those with chronic elevations from other causes. Using the fourth-generation cTnT assay, approximately 0.7% of patients in the general population have modest elevations >99th percentile URL. In the same population, this number was 2% with the hs-cTnT assay. Of that number, only half had documentation (even with imaging) of cardiac abnormalities. If the prevalence of a positive cTnT is 2% in the general population, it will likely be 10% or 20% in the emergency department (ED) and even higher in hospitalized patients, as these patients often have cardiac comorbidities.
Measurement of changes in hs-cTn over time (δ hs-cTn) improves the specificity of hs-cTn for the diagnosis of acute cardiac injury. However, it does so at the cost of sensitivity. With contemporary assays, differences in analytical variation have been used to define an increasing pattern. At elevated values, CV for most assays is in the range of 5% to 7%, so a change of 20% ensures that a given change is not caused by analytical variation alone. At values near the 99th percentile URL, higher change values are necessary. The situation with hs-cTn assays is much more complex, as the following outline shows:
These issues pose a major challenge even for defining the ideal delta change value and provide the reasons why the use of this approach will reduce sensitivity (Figure 2).
(Enlarge Image)
Figure 2.
Defining the Optimal Delta: Tension Between Sensitivity and Specificity
There is a reciprocal relationship between sensitivity and specificity. With marked percentage changes, specificity is improved at the expense of sensitivity, and at lower values, the opposite occurs.
In addition, there is controversy in regard to the metrics that should be used with high-sensitivity assays. The Australian-New Zealand group proposed a 50% change for hs-cTnT for values below 53 ng/l and a 20% change above that value. The 20% change is much less than conjoint biological and analytical variation. A number of publications have suggested the superiority of absolute δ cTn compared to relative δ cTn in discriminating between AMI and non-AMI causes of elevated cTn. However, the utility of the absolute or relative δ cTn appears to depend on the initial cTn concentration, and the major benefit may be at higher values. A recent publication by Apple et al. calculates deltas in several different ways with a contemporary assay and provides a template for how to do such studies optimally. If all studies were carried out in a similar fashion, it would help immensely. In the long run, institutions will need to define the approach they wish to take. We believe this discussion is a critical one and should include laboratory, ED, and cardiology professionals.
Discriminating Between Acute and Nonacute Causes of hs-cTn Elevations
With the ability to precisely measure small concentrations of cTn, clinicians will be faced with the challenge of distinguishing patients who have acute problems from those with chronic elevations from other causes. Using the fourth-generation cTnT assay, approximately 0.7% of patients in the general population have modest elevations >99th percentile URL. In the same population, this number was 2% with the hs-cTnT assay. Of that number, only half had documentation (even with imaging) of cardiac abnormalities. If the prevalence of a positive cTnT is 2% in the general population, it will likely be 10% or 20% in the emergency department (ED) and even higher in hospitalized patients, as these patients often have cardiac comorbidities.
Measurement of changes in hs-cTn over time (δ hs-cTn) improves the specificity of hs-cTn for the diagnosis of acute cardiac injury. However, it does so at the cost of sensitivity. With contemporary assays, differences in analytical variation have been used to define an increasing pattern. At elevated values, CV for most assays is in the range of 5% to 7%, so a change of 20% ensures that a given change is not caused by analytical variation alone. At values near the 99th percentile URL, higher change values are necessary. The situation with hs-cTn assays is much more complex, as the following outline shows:
Change criteria are unique for each assay.
It will be easy to misclassify patients with coronary artery disease who may present with a noncardiac cause of chest pain but have elevated values. They could be having unstable ischemia or elevations caused by structural cardiac abnormalities and noncardiac discomfort. If hs-cTn is rising significantly, the issue is easy but if the values are not rising, a diagnosis of AMI still might be made. If so, some patients may be included as having AMI without a changing pattern. This occurred in 14% patients studied by Hammarsten et al.. If patients with elevated hs-cTn without a changing pattern are not called AMI, should they be called patients with "unstable angina and cardiac injury" or patients with structural heart disease and noncardiac chest pain? Perhaps both exist?
The release of biomarkers is flow-dependent. Thus, there may not always be rapid access to the circulation. An area of injury distal to a totally occluded vessel (when collateral channels close) may be different in terms of the dynAMIcs of hs-cTn change than an intermittently occluded coronary artery.
Conjoint biological and analytical variation can be measured. They are assay-dependent, and the reference change values range from 35% to 85%. The use of criteria less than that (which may be what is needed clinically) will thus likely include individuals with changes caused by conjoint biological and analytical variation alone. This has been shown to be the case in many patients with nonacute cardiovascular diagnoses.
Most evaluations have attempted to define the optimal delta, often with receiver operator curve analysis. Such an approach is based on the concept that sensitivity and specificity deserve equivalent weight. But higher deltas improve specificity more and lower ones improve sensitivity and it is not clear that all physicians want the same tradeoffs in this regard. ED physicians often prefer high-sensitivity so that their miss rate is low (<1%), whereas hospital clinicians want increased specificity. This tension will need to be addressed in defining the optimal delta.
The delta associated with AMI may be different from that associated with other cardiac injury. In addition, women have less marked elevations of cTn in response to coronary artery disease and in earlier studies were less apt to have elevated values. Given their pathology is at times different, it may be that different metrics may be necessary based on gender.
Some groups have assumed that if a change is of a given magnitude over 6 hours, it can be divided by 6 and the 1-h values can be used. This approach is not data driven, and biomarker release is more likely to be discontinuous rather than continuous. In addition, the values obtained with this approach are too small to be distinguished from a lack of change with most assays.
These issues pose a major challenge even for defining the ideal delta change value and provide the reasons why the use of this approach will reduce sensitivity (Figure 2).
(Enlarge Image)
Figure 2.
Defining the Optimal Delta: Tension Between Sensitivity and Specificity
There is a reciprocal relationship between sensitivity and specificity. With marked percentage changes, specificity is improved at the expense of sensitivity, and at lower values, the opposite occurs.
In addition, there is controversy in regard to the metrics that should be used with high-sensitivity assays. The Australian-New Zealand group proposed a 50% change for hs-cTnT for values below 53 ng/l and a 20% change above that value. The 20% change is much less than conjoint biological and analytical variation. A number of publications have suggested the superiority of absolute δ cTn compared to relative δ cTn in discriminating between AMI and non-AMI causes of elevated cTn. However, the utility of the absolute or relative δ cTn appears to depend on the initial cTn concentration, and the major benefit may be at higher values. A recent publication by Apple et al. calculates deltas in several different ways with a contemporary assay and provides a template for how to do such studies optimally. If all studies were carried out in a similar fashion, it would help immensely. In the long run, institutions will need to define the approach they wish to take. We believe this discussion is a critical one and should include laboratory, ED, and cardiology professionals.
Source...