ESC: DeFACTO misses mark but shows promise for diagnosing CAD

(A) Coronary CT angiography indicates obstructive stenosis (white arrow) in the proximal portion of the left anterior descending (LAD) artery. (B) Angiography confirms the LAD stenosis (red arrow) (C) Noninvasive computation of FFR from FFRCT of the first diagonal branch (0.79) and distal LAD (0.57). Source: J Am Coll Cardiol 2011;58[19]:1989-1997.

DeFACTO, a study designed to assess the performance of a noninvasive method for gauging the physiologic significance of coronary artery disease (CAD), failed to meet its prespecified primary endpoint, researchers reported Aug. 26 at the European Society of Cardiology Congress in Munich. But that may be the result of the study’s high standards, according to the author of an editorial that accompanied the study’s simultaneous publication online in the Journal of the American Medical Association, who agreed with the researchers that the approach holds promise.

In the latest study on the novel technique, James K. Min, MD, of the division of cardiology at Cedars-Sinai Heart Institute in Los Angeles, and colleagues assessed the diagnostic performance of noninvasive fractional flow reserve (FFR) computed from coronary CT angiography (CCTA). In the DISCOVER-FLOW study, their FFR_CT technique was shown to be more accurate than CCTA for detecting ischemia-causing coronary lesions, with accuracy, sensitivity and specificity values of 84.3 percent, 87.9 percent and 82.2 percent, respectively, compared with 58.5 percent, 91.4 percent and 39.6 percent, respectively, for CCTA (J Am Coll Cardiol 2011;58[19]:1989-1997).

DeFACTO (Determination of Fractional Flow Reserve by Anatomic Computed Tomographic Angiography) enrolled 252 stable patients with suspected or known CAD from 17 centers in five countries who underwent CT, invasive coronary angiography (ICA), FFR, and FFR_CT between October 2010 and October 2011.

CT, ICA, FFR and FFR_CT were interpreted in blinded fashion by independent core laboratories. Researchers compared the accuracy of FFR_CT plus CT for diagnosis of ischemia with an invasive FFR reference standard. Ischemia was defined by an FFR or FFR_CT of 0.80 or less, while anatomically obstructive CAD was defined by a stenosis of 50 percent or larger on CT and ICA.

The primary endpoint was the diagnostic accuracy of FFR_CT plus CT for diagnosis of per-patient ischemia compared with an FFR reference standard. Their prespecified primary outcome goal was for a level of per-patient diagnostic accuracy of greater than 70 percent of the lower bound of a 95 percent confidence interval.

They found that 54.4 percent of the patients had an abnormal FFR. Diagnostic accuracy, sensitivity, specificity, positive predictive value and negative predictive value of FFR_CT plus CT were 73 percent, 90 percent, 54 percent, 67 percent and 84 percent, respectively, on a per-patient basis. By comparison, the diagnostic accuracy of CT alone was 64 percent. FFR_CT showed superior discrimination compared with CT alone. But the lower bound for diagnostic accuracy was 67 percent—three percentage points below the prespecified threshold.

Min and colleagues wrote that the findings should be considered a proof of concept of the feasibility of the technique. “At the patient level, FFR_CT, when added to CT, improved diagnostic accuracy vs. CT alone, driven by improvements in sensitivity as well as specificity,” they wrote. “These results suggest that FFR_CT can impart considerable discriminatory power to identify and exclude ischemia in patients with suspected CAD.”

The authors explained that they selected the prespecified endpoint of 70 percent accuracy based on other studies that showed 70 percent to be at the midpoint of reported diagnostic accuracies for stress imaging. “In this study, FFR_CT demonstrated a per-patient diagnostic accuracy of 73 percent, with confidence intervals that suggest a diagnostic accuracy as low as 67 percent and as high as 78 percent,” they wrote. “These findings establish a performance of FFR_CT that is within the range of conventional stress imaging testing.”

In an accompanying editorial, Manesh R. Patel, MD, of the Duke Clinical Research Institute in Durham, N.C., agreed with Min et al that FFR_CT plus CT was associated with improved diagnostic accuracy and discrimination when compared with CT alone. He praised the researchers for setting the bar high by comparing their approach “with a reference standard of both invasive angiography and invasive FFR. This change in reference standard may in part explain some of the accuracy findings.”

Patel wrote that FFR_CT might be seen as being limited, but argued that would be “a naïve conclusion, likely based on the published diagnostic performance of noninvasive tests compared only with invasive angiography. If the existing noninvasive imaging technologies were compared with invasive angiography plus FFR, it is highly likely that the published diagnostic performance would be reduced.”

He noted that in real-world practice current noninvasive technologies used for diagnosis of stable, intermediate risk CAD patients prior to invasive angiography do not perform at published diagnostic levels. “Hence, the current report describes an important noninvasive technology that may improve existing care and has the potential to outperform established noninvasive technologies,” he argued.  

Patel recommended that future FFR_CT studies include a wider array of patients with pretest risks, comparisons of technologies beyond invasive angiography, endpoints such as resource utilization and clinical outcomes and local site rather than core lab participation in the imaging analyses. “It is with these types of continued rigorous studies that noninvasive technologies such as FFR_CT plus CT may move the clinical community closer to the holy grail of a high-quality combined anatomic and functional test for detection of CAD that improves efficiency and patient outcomes,” Patel concluded.    

DeFACTO was funded by HeartFlow.