Considered a gold standard, cardiac troponin is unequaled in its specificity and sensitivity among biochemical markers for diagnosing myocardial infarction (MI). However, questions remain about how new high-sensitivity assays should be employed by clinicians, as these assays’ high analytical sensitivity tends to pick up conditions other than MI, such as heart failure and tachyarrhythmias. Researchers recently compared absolute versus relative change in three novel high-sensitivity assays. Their findings are the subject of this issue of Strategies.
A tradeoff of lower specificity for higher sensitivity is a familiar conundrum for laboratory medicine. But for troponin, the stakes are especially high: the older, less sensitive tests mean waiting for 6–12 hours of serial sampling, while a high-sensitivity assay improperly used can lead to overdiagnosis and unnecessary follow-up testing such as cardiac catheterization. As a result, the field of laboratory medicine has focused on how to interpret the rise and fall of troponin to make sure physicians can act rapidly and appropriately, based on reliable results.
In a recent study, Christian Mueller, MD, and an international group of colleagues compared absolute versus relative change using a high-sensitivity cardiac troponin T (hs-cTnT) assay and two high-sensitivity cardiac troponin I (hs-cTnI) assays among a cohort of 830 patients in the Advantageous Predictors of Acute Coronary Syndrome Evaluation (APACE) study (Am J Med 2013 126:781–8). APACE is a prospective, international, multicenter study of patients presenting to the emergency department with acute chest pain. Mueller and his colleagues found that absolute changes performed better in all three assays compared to relative changes, and that combining both metrics added some additional value for hs-cTnT.
For the hs-cTnT assay, the receiver operating characteristic (ROC) curves for absolute, relative, and combined change at 2 hours after presentation were 0.95, 0.75, and 0.98, respectively. In contrast, ROC curves for the first hs-cTnI assay were 0.97, 0.75, and 0.97, respectively; and 0.96, 0.73, and 0.96, respectively, for the second. Reclassification analysis confirmed the superiority of absolute versus relative changes.
“The data described in this paper highlight absolute change as the preferred metric to assess levels of cardiac troponin in the early diagnosis of acute myocardial infarction, extending and complementing previous studies on this topic,” Mueller said. “Hopefully this analysis will help to advance the discussion regarding these changes, as there is not a single study that has shown superiority of relative changes, while we now have analyses from five different datasets showing superiority of absolute changes.” Mueller is a professor in the department of internal medicine at University Hospital Basel in Switzerland.
With the results bolstering the case for using absolute change, clinicians and laboratorians could start to apply the findings of this research now, according to Mueller. “High-sensitivity cardiac troponin assays have been in use in Europe, Australia, and other countries for several years and this analysis may help them to better interpret changes in troponin,” Mueller said. However, several questions need further study, he noted, including timing of the second measurement, assay-specific and possibly patient subgroup-specific cutoff levels, as well as adding the marker copeptin.
Mueller and his colleagues’ research is significant because it not only looks at three different hs-cTn assays, but also examines combining relative and absolute change, noted Peter Kavsak, PhD, an associate professor in the department of pathology and molecular medicine at McMaster University in Hamilton, Ontario, Canada. Kavsak, who was not associated with the study, also has authored papers on hs-cTn, advocating more study of combining relative and absolute metrics.
“In our own studies, we’ve found that at lower concentrations an absolute change is more important, and then at very elevated concentrations, a percent change or relative change may be more important, so I think it’s notable that in this paper they examined combining the two,” Kavsak said. “It will be important to see, as others continue down this road, whether the data confirm that we should just stick with absolute change, or whether there might still be added utility to incorporating relative change.” Further research will also be needed to explain why combining the two metrics worked for hs-cTnT, but not hs-cTnI, he added.
The study typifies the “remarkably consistent” previous findings on the superiority of absolute change for hs-cTn, the authors note. “It is important to note that different findings may be obtained when one does not try to optimize accuracy but specificity,” they wrote. They remarked that in a frequently-cited study a high relative change—greater than 250%—appeared to offer the most pronounced increase in specificity (JAMA 2011;306:2684–93). “Although relative changes provided some added value on top of the baseline value in [area under the curve] analysis in our study, this was not evident once net reclassification improvement was used to quantify the difference,” according to the authors.
Mueller emphasized that he believes this oft-cited study has been misunderstood. “That study has been misinterpreted as evidence of the superiority of relative changes for achieving high specificity. This is incorrect, as that research team never compared absolute and relative changes in that paper. In contrast, in some of their recent analyses, they were able to confirm all other previous studies regarding the superiority of absolute changes.”
On the matter of 1-hour versus 2-hour periods of sampling after initial presentation, Kavsak advised caution, warning that although a 1-hour timeframe is appealing, it might not be workable. “In this study they found the optimal absolute delta at one hour to be five ng/L, but I would be cautious on that approach,” he said. “A change of five ng/L may well be within the analytical imprecision of the assay at concentrations near the 99th percentile.”
The authors emphasized that optimal cutoff values have to be defined for each assay. For example, the optimal 1-hour absolute changes were the same for all three assays, but this did not hold true for 2-hour changes, where ROC-derived optimal cutoffs were ng/L for the hs-cTnT assay and 10 ng/L for both hs-cTnI assays.
For physicians to properly utilize the new hs-cTn assays, laboratories will need to be clear about the unique delta cutoff for each assay and for the length of time between measurements, according to Kavsak. “I think an important message in this paper is that change is dependent on the timeframe in which you measure it—and I think this paper clearly indicates that with earlier timeframes, the values will be different. It is critical that laboratorians and physicians appreciate this fact when they incorporate these assays and change criteria into clinical algorithms. Obviously they can’t be applying change criteria derived from a high-sensitivity troponin T assay to a troponin I assay. But even more importantly, are they actually drawing samples at the correct timeframe to calculate and apply the delta? If not, then any diagnostic gains may well be lost by using inappropriate change criteria.”
Dr. Mueller has received research support from the Swiss National Science Foundation, the
Swiss Heart Foundation, the Novartis Foundation, the Krokus Foundation, Abbott, AstraZeneca, Biosite, Brahms, Nanosphere, Roche, Siemens, and the Department of Internal Medicine, University Hospital Basel, as well as speaker honoraria from Abbott, Biosite, Brahms, Roche, and Siemens.
Dr. Kavsak has received grants/consultant/honoraria from Abbott Diagnostics, Beckman Coulter, Randox Laboratories, and Roche Diagnostics. He is listed as an inventor on patents filed by McMaster University related to laboratory testing in acute cardiac care.