A retrospective study of patients seen over two years at the emergency department of Sweden's Karolinska University Hospital has demonstrated that using a very low cutoff point when measuring the protein biomarker cardiac troponin T could accurately rule out heart attack in a large proportion of patients complaining of chest pain.
The study — released during the annual meeting of the American College of Cardiology in Washington, DC last week and simultaneously published in the ACC's journal — analyzed two years worth of emergency room chest pain patients who had at least one high-sensitivity measurement of the biomarker. Looking at these patients' outcomes, the study authors calculated the ability of a very low, or "undetectable," reading — lower than five ng/l — to predict both future heart attack and death up to a year later.
High-sensitivity troponin T assays detect troponin levels significantly below the threshold of earlier, non-high sensitivity assays. In that sense, the high-sensitivity troponin detected by these methods is treated as a distinct biomarker with potentially unique diagnostic and prognostic implications from the more abundant, lower-sensitivity troponin.
In the context of myocardial infarction, or heart attack, high-sensitivity measurement of troponin T has been identified as a promising tool to rapidly and simply rule out patients with chest pain who are unlikely to be suffering an infarction and can be discharged from the hospital without additional testing, potentially reducing prolonged observation of patients, hospital admissions, ER overcrowding, and associated costs.
In the Karolinska study, only 15 patients out of several thousand with below-threshold hs-cTnT and clear ECGs suffered heart attacks within a month of their initial presentation in the emergency room. Thus, the absolute risk for a patient with troponin below the "undetectable threshold" and no other indication of heart issues in their ECG measurements was 0.17 percent in the study cohort.
Based on this, the team calculated that the negative predictive value, or the ability of their simple cutoff strategy to rule out patients unlikely to have a heart attack, was nearly 100 percent for heart attacks within one month, and remained similar for events up to a year later.
Martin Holzmann, one of the study's senior authors, said at the ACC meeting that based on these retrospective results, he and his colleague believe that sending patients home who have undetectable hs-cTnT and no abnormal ECG results could prevent 20 to 25 percent of all admissions at their hospital due to chest pain — between 600 and 700 unnecessary hospitalizations every year.
According to the study authors, chest pain is responsible for an estimated 15 to 20 million emergency visits per year in Europe and the US. Currently, serial testing of troponin levels using non-high sensitivity troponin assays is one of several clinical measures used to diagnose myocardial infarction, but according to the study authors, there is significant room for improvement since only 10 to 20 percent of patients admitted to the hospital with chest pain are actually diagnosed with myocardial infarction during their stay.
The Karolinska study is not the first to suggest that chest pain patients with an undetectable level of troponin using high-sensitivity assays could be ruled out in the emergency setting as unlikely to have a myocardial infarction. Previous prospective studies — including one British-led effort published in the Journal of the American College of Cardiology in 2011 and another Swiss study published in 2013 in the International Journal of Cardiology — have also suggested that the predictive value of undetectable hs-cTnT for myocardial infarction might be near 100 percent.
In the Karolinska team's retrospective study, Holzmann and his colleagues analyzed the outcomes of 14,636 patients seen at Karolinska whose hs-cTnT levels were tested at some point during their workup, approximately 61 percent of whom — about 9,000 — had troponin levels below the five ng/l "undetectable" threshold.
Among this subset, only 39 patients — less than 0.5 percent — had a myocardial infarction within 30 days, and only 15 — less than 0.2 percent — were also without ischemic changes in their electorcardiograms that might otherwise indicate elevated risk. The researchers hypothesized that because the troponin measurements in these patients were taken relatively early — less than two hours after their symptoms started — subsequent measurements may have revealed rising troponin levels and indicated risk for an MI.
Based on these data, the researchers calculated that the negative predictive value of having a heart attack for patients with an "undetectable" first hs-cTnT level and no ischemic ECG changes was 99.8 percent. The negative predictive value for death during that time was 100 percent.
And followed over a year, this predictive power held up, dropping only to 99.4 percent. At one year of follow-up there were 38 deaths in the cohort, of which 32 were caused by cancer and only two were cardiovascular, Holzmann reported.
Holzmann told PGx Reporter after the presentation of the study that he and his colleagues would now like to follow up their retrospective analysis with a study actually using hs-cTnT measurements to rule out myocardial infarction and evaluating the impact of that strategy.
"What we need to do now to answer the question of whether we can use this or not is to do an intervention [study]," he said. "At our site we have 4,000 patients per year, so we could do that only at our site and see if it actually reduces the numbers of admissions. But first we have to get funding."
The Roche hs-cTnT assay the Karolinska hospital used during the retrospective study period has been available in Europe for several years. In the US, however, neither that assay nor any other high-sensitivity troponin tests have been approved for clinical use.
Serving as a panelist during the ACC presentation of the study, the Mayo Clinic's Alan Jaffe raised the possibility that very high-quality traditional troponin assays that are currently approved in the US might be able to achieve similar results to the Swedish study. Jaffe reasoned that the Roche test the group used is not as "high-sensitivity" as some of the newer hs-cTnT tests, and may actually be a lot closer to a regular-sensitivity troponin assay.
"Testing this strategy with some of the better US assays might be worthwhile," Jaffe said. "But using this, or other high-sensitivity assays in the US is not going to happen at least for a year or so," he added. Studies of high-sensitivity troponin assays are underway in the US, he said, but no tests have been cleared yet for clinical use by the US Food and Drug Administration.
Jaffe also shared some skepticism that the results the Swedish team shared have closed the case on high-sensitivity troponin's rule-out ability. Specifically, Jaffe suggested that it may not be clear that all the patients being ruled out by undetectable hs-cTnT levels really should be ruled out.
"In many previous studies, people have used very high cutoffs and so are probably missing some small [myocardial infarctions]. Some other studies haven’t done good follow-up ascertainment of events," he said. "Maybe it was done perfectly in this trial." But in terms of what was presented, Jaffe said he wasn't yet convinced.
"I think the strategy will be proven" in the long term, he said. "But I have concerns about whether we have dotted all the i's and crossed all the t's."
Holzmann stressed, however, that there were numerous indications in the Karolinska data to suggest that the biomarker was truly distinguishing those with little risk of myocardial infarction with high sensitivity and specificity.
For example, he said, there were "1,900 patients who had a second troponin measurement with undetectable levels" after a first undetectable result. Furthermore, in 90 percent of patients, a third test yielded undetectable levels, as well. "Nothing happens to these patients," Holzmann said. "They really are low risk." He added that it was also encouraging that the undetectable cutoff method was performed with such a high negative predictive value regardless of other risk factors like age and sex.
The group hopes lingering questions will be answered if they are able to begin a follow-up study to track whether discharging patients with undetectable troponin can actually reduce admissions and associated costs.
Additionally, Holzmann said, the group has also considered the issue of what to do with patients with intermediate troponin levels. In the Karolinska study, about 20 percent of the total cohort had hs-cTnT levels between five and 14 ng/l.
Among these patients, the absolute risk for myocardial infarction within 30 days was three percent, and the negative predictive value for myocardial infarction was 97 percent, compared to the 0.17 and 99.8 percent respectively for those with troponin below the lower threshold.
According to Holzmann, though troponin alone might not be a sufficient tool to rule out myocardial infarction in these patients, using a combination of other cardiac biomarkers could possibly allow more accurate risk assessments.
Ruling out cardiovascular events and disease in chest pain patients is an attractive diagnostic target. Prevencio, a proteomics company working with technology licensed from the University of Pittsburgh, said last year that it is developing a multiplex protein test to rule out obstructive coronary artery disease in patients presenting with chest pain in the emergency room. Meanwhile, CardioDx currently markets a gene expression test to rule out coronary artery disease in patients with chest pain in non-emergency settings.