NEW YORK (GenomeWeb) – Proteomic diagnostics firm Prevencio and collaborating researchers are presenting data at this week's American College of Cardiology Scientific Session in Washington, D.C. on a biomarker panel that appears useful for predicting major adverse cardiac events (MACE) including stroke and myocardial infarction.
Rhonda Rhyne, Prevencio's president and CEO, said the company is pursuing commercialization plans for the panel, which it has named HART CVE (cardiovascular events) in parallel with its commercialization efforts for its HART CAD proteomic test for identifying patients with coronary artery disease.
The company plans to have kits for both tests completed in 2018, after which it will begin studies in support of US Food and Drug Administration clearance, with launches targeted for 2019," she said. She added that Prevencio was also considering a "parallel lab-developed test path" in which they would launch the tests next year.
In the study being presented this week, researchers led by James Januzzi, a cardiologist at Massachusetts General Hospital who has collaborated with Prevencio on several biomarker development efforts, looked at 927 patients undergoing coronary angiography and identified a set of four protein markers that outperformed conventional clinical measures for identifying patients likely to suffer a MACE within one year of testing.
The different in performance was modest, with the study showing an area under the curve of .79 for the four-marker panel compared to .75 for clinical measures. But, Januzzi said, the study's main significance lay in its demonstration that protein markers were competitive with, if not better than, existing clinical variables.
"I don't think that we're suggesting that this .04 difference in [AUC] is going to make clinicians hang up their stethoscopes for good," he said. "But it really establishes a proof of concept that [molecular] markers can stand quite equally with clinical variables."
"Ultimately, it's reasonable to expect that it's going to be a combination of clinical variable and biomarkers that is optimal for evaluating patients," he added.
The study looked at patients from the Catheter Sampled Blood Archive in Cardiovascular Diseases (CASABLANCA) study, which Prevencio also used to develop protein markers for the HART CAD test.
The researchers used a 649-subject training set and a 278-patient validation set to test the usefulness of more than 50 clinical variables and 109 molecular markers for predicting patients who would go on to suffer cardiovascular death, myocardial infarction, and/or stroke over the course of the next year.
"What we found was that while clinical variables alone were quite useful for predicting prognosis, in an unbiased model that allowed for anything in the model that had strong enough statistical meaning, clinical variables were actually knocked out of the model," Januzzi said. "The final four [best] predictors of outcome in these patients were concentrations of four [protein] biomarkers."
The four markers were NT-proBNP, kidney injury molecule 1, osteopontin, and tissue inhibitor of metalloproteinase-1. The presence of ProBNP was unsurprising, Januzzi said, given that it is a well-established marker of heart muscle stress. The other three markers have been identified as prognostic for cardiac events in previous studies, he said, "but this is the first time we've put them all together."
At the cutoff chosen for their analysis, the four-marker panel showed 64 percent sensitivity, 76 percent specificity, 28 percent positive predictive value, and 93 percent negative predictive value.
The high negative predictive value suggests the markers could be used as a rule-out test to identify patients who are at low risk of a MACE and so don't require adjustments to their treatment or more intensive monitoring.
In general, Januzzi noted, "biomarker studies focused on prognosis show pretty consistently that the negative predictive value of the biomarkers is better than the positive predictive value."
"In other words, if you're healthy, you're healthy. If the marker is low, you're going to have a good prognosis," he said. "On the other hand, if the marker is elevated, the positive predictive value tends to be lower."
In part, he suggested, this could be due to the limited follow up in such studies. "Just because an event doesn't happen [during a study], doesn't mean an event won't happen if the patient is followed long enough," he said.
"For instance, he added, "most people in my field would agree that it's not normal to have an elevated biomarker of disease like NT-proBNP. It's just a question of how long you can wait before an event may occur."
Despite these limitations, Januzzi said the markers could have some value as positive predictors. He said that adjusting the cutoffs to maximize positive predictive value, the researchers could bring this measure into the mid-30 percent range, which he described as "a very respectable level."
Used in this way, the test could help identify patients who should be referred for more intensive evaluation or help physicians consider adjustments in their medication, he said.
It could also prove useful for enriching patient cohorts for clinical trials, he said, allowing researchers to select patients more likely to undergo a cardiac event and thereby reduce the number of patients they need to enroll in a trial for, for instance, a new drug aimed at lowering stroke or heart attack risk.
Januzzi said he and his colleagues are now hoping to validate their findings in additional cohorts and then develop antibodies to the four proteins better suited to use in a targeted, high-throughput assay.
"Presently, these four markers are measured in the context of a microarray, but there are 105 other markers in that array," he noted.
He added that he also hopes to expand the profile of patients the assay is tested in. The CASABLANCA cohort consisted of patients who underwent coronary and peripheral angiography at Mass General between 2008 and 2011 and were referred for those procedures for reasons including myocardial infarction, unstable angina pectoris, and heart failure, along with nonacute indications like evaluation of stable chest pain and failed stress testing or pre-operative evaluation before heart valve surgery.
"I think it's always a good idea to examine any new biomarker technology in a broad range of patient types," Januzzi said. "So, we'd like to look at patients with stable coronary disease, patients with unstable disease, with heart failure, patients at risk for stroke. I think that it's important to always diversify the portfolio whenever possible."