CAC Testing Improves Pretest Probability of Obstructive CAD

Adding clinical risk factors and a coronary artery calcium (CAC) score to the conventional diagnostic algorithm for chest pain improves the prediction and discrimination of patients with suspected coronary artery disease, according to a new study. The new risk-factor model, when used along with CAC, categorizes more patients with a low clinical likelihood of having CAD and can help avoid further testing, say researchers.

“Currently, if you have a patient with chronic chest pain, you use age, sex, and the type of symptoms to calculate the pretest probability of disease and from there you decide whether you shouldn’t test any further, whether you should test noninvasively with CT angiography or functional testing, or whether you should send them for invasive angiography,” lead investigator Simon Winther, MD, PhD (Gødstrup Hospital, Herning, Denmark), told TCTMD. “It’s actually quite simple. It’s only based on three variables.”

The investigators’ aim, he continued, was to first incorporate clinical risk factors into the pretest probability (PTP) assessment of obstructive CAD, and then to include the CAC score to determine if it provided any further value for estimating the likelihood of disease. Winther said that when physicians calculate the pretest probability of obstructive CAD, they typically will keep in mind the patient’s risk factors, but the new study allows them to put some hard numbers behind their clinical intuition. 

The study, published in the November 24, 2020, issue of the Journal of the American College of Cardiology, includes two new PTP models. With the risk factor clinical likelihood (RF-CL) model, which takes into account family history, smoking status, dyslipidemia, hypertension, and diabetes, patients were simply characterized as having 0-1, 2-3, or 4-5 risk factors, and this number was then included with age, sex, and symptoms to calculate the PTP of CAD. The CAC clinical likelihood (CAC-CL) algorithm incorporated the calcium score into the risk-factor model.

These two models were then compared with the PTP model currently recommended by the European Society of Cardiology, which is based on the Diamond-Forrester approach and includes sex, age, and type of symptoms.

The study included four large cohorts of patients with symptoms suggestive of obstructive CAD who were referred for noninvasive testing using CT angiography. With the training cohort, which was used to develop the models, there were 41,177 patients who underwent CT angiography between 2008 and 2017 at 13 Danish hospitals. The validation cohorts included 9,383 patients in the Western Denmark Heart Registry, 1,675 patients from the Danish Study of Noninvasive Testing in Coronary Artery Disease (Dan-NICAD), and 4,403 patients from the Prospective Multicenter Imaging Study for Evaluation of Chest Pain (PROMISE).

Overall, the area under the receiver-operating characteristic curve (AUC) of the RF-CL model in the validation cohort was 74.9 (95% CI 73.7-76.1), which was higher than the 72.3 (95% CI 71.0-73.6) observed in the standard PTP model. In total, more patients in the validation cohort had a less than 5% chance of having obstructive CAD when assessed using the RF-CL algorithm as compared with the conventional PTP model (5,919 vs 1,708 patients).

Overall, 38% and 11% of patients had a low clinical likelihood of CAD using the RF-CL and PTP models, respectively.

In their paper, the researchers present their data in a table that incorporates symptoms (nonanginal pain, atypical angina or dyspnea, or typical angina), age, and sex, along with the number of risk factors. Taken together, these variables provide a more-accurate estimate of the clinical likelihood of obstructive CAD, they say.  

“Based on that simple table, you can improve the risk classification of the patient,” said Winther.

When the researchers turned to calcium, they found the prevalence of obstructive CAD increased with greater CAC burden in the basic PTP and RF-CL models. In terms of diagnostic performance, the AUC of the CAC-TL model in the validation cohort was 84.9 (95% CI 84.0-85.9) and this was higher than the AUC in the PTP and RF-CL models. In total, 54.1% of patients (n = 7,801) in the validation cohort had less than a 5% likelihood of obstructive CAD using the CAC-CL model, which again was higher than what was observed in the RF-CL and PTP models.

At present, CAC testing isn’t typically performed as part of the chest-pain diagnostic algorithm. While the CAC-CL model resulted in superior clinical performance, there are logistical and operational hurdles that prevent its use, said Winther. It would require an outpatient chest-pain clinic with a cardiac CT program, along with conventional testing such as echocardiography and ECG testing, but if such a clinic was in place, “you could stop half of the patients there because you know the calcium score is very, very low and the pretest probability would be less than 5%,” he said. “Then you wouldn’t have to do anything else. And a calcium score is very simple. It takes 5 to 10 minutes.”

Winther stressed that CAC is not a diagnostic test, but rather one to stratify patient risk or the probability of stenosis. In this setting, though, “it can move your pretest probability a lot,” he said.

A Couple of Options for CAC in Obstructive CAD

Ron Blankstein, MD (Brigham and Women’s Hospital, Boston, MA), who wasn’t involved in the study, said CAC testing has two potentially useful roles in the evaluation of patients without known CAD. In low-risk patients, a score of 0 can identify those extremely unlikely to have obstructive CAD, who have an excellent prognosis, and in whom further testing can be deferred.

“However, the current study now also nicely shows how information on the severity of CAC can significantly improve our ability to identify patients who have a higher pretest probability of having obstructive CAD,” he told TCTMD. “This is important because when we use prior algorithms which rely on age, sex, and type of symptoms, the ability to accurately predict the pretest probability of obstructive CAD is limited. It is often overestimated and has reduced specificity. Moreover, contemporary [PTP] tools show that almost all patients have a low or intermediate pretest probability of obstructive CAD, and thus these models have not been able to reliably identify high-risk patients.”

Based on that simple table, you can improve the risk classification of the patient. Simon Winther 

When significant CAC is present, which is CAC > 100, and especially > 400 as in the present study, this information accurately identifies patients at high risk of obstructive CAD who warrant further testing, said Blankstein. It might also suggest the type of test needed, he added. For example, in high-risk patients with significant CAC and a PTP > 70%, functional testing may be preferred, while in low- or intermediate-risk patients, coronary CT angiography might be more useful. 

In an editorial, Khurram Nasir, MD, MPH (Houston Methodist DeBakey Heart and Vascular Center, TX), Jagat Narula, MD, PhD (Icahn School of Medicine at Mount Sinai, New York, NY), and Martin Bødtker Mortensen, MD, PhD (Aarhus University Hospital, Denmark), point out that even updated and recalibrated PTP scores overestimate risk of obstructive CAD, leading to a low prevalence of either ischemia or obstructive CAD on noninvasive imaging.

The CAC-CL algorithm in the present study had high sensitivity and a negative predictive value (NPV) of 98%. While it didn’t show that a CAC score of 0—the so-called power of zero—offered a favorable prognosis among stable symptomatic patients, both PROMISE and SCOT-HEART showed that the NPV was extremely high among those without any coronary calcification, suggesting it might be an optimal diagnostic strategy for identifying a very low-risk subgroup. 

For Nasir, Narula, and Mortensen, the results of the present study support the use of CAC as part of the early diagnostic workup of patients with suspected CAD. “With extensive evidence, further strengthened by the current study, it is hard to contemplate how the upcoming US chest-pain-management guidelines could not adapt the central role of CAC testing as a gatekeeper for both advanced anatomical and functional testing,” they write.

While this approach appears “prudent,” they also acknowledge the massive operational and financial challenges of adopting this strategy. “However, with the evolving healthcare landscape and greater emphasis on value-based reimbursements (eg, bundled payments), we believe it would not be long before this stepwise hybrid protocol to optimize patient selection for downstream testing would be endorsed as a Class IIa recommendation.”

The American Heart Association/American College of Cardiology, in cooperation with several other groups, last updated the guidelines for the management of patients with NSTE ACS in 2014, and the new guidelines are expected in 2021.