The European Society of Cardiology recommends the use of the updated Diamond-Forrester score (DF-score) for pre-test probability stratification in patients with symptoms suggestive of coronary artery disease (CAD). However, recently CAD Consortium models scores have demonstrated superior risk calibration compared to DF-score when compared with coronary CT or invasive angiography as reference.
DF-score includes age, sex, and type of chest pain. CAD Consortium have developed 3 models: 1) A basic model, which includes the same variables as DF-score, 2) A clinical model included the basic model and additionally diabetes, hypertension, dyslipidemia, and smoking, and 3) A clinical + coronary artery calcium score (CACS) model included the CACS in addition to the variables in the clinical model.
To dated, CAD Consortium score has not been validated against myocardial perfusion imaging (MPI).
To investigated the performance of CAD consortium models with MPI as reference standard.
We included 1675 consecutive patients (male 49%, mean age 57±8 years) referred to coronary computed tomography angiography (CTA) with symptoms suggestive of CAD. A medical interview was conducted prior to coronary CTA to obtain clinical risk factors. All patients with suspected obstructive CAD by coronary CTA (>50% diameter stenosis) were subsequently randomized 1:1 to either cardiac magnetic resonance or myocardial perfusion scintigraphy.
A total of 391 patients (23%) had suspected obstructive CAD by coronary CTA of which 299 (76%) completed a MPI. Myocardial perfusion defects were identified at perfusion imaging in 65/299 (22 %) patients
The Calibration in the Large and the calibration slope for DF-score was -3.1 and 0.68, CAD consortium basic -1.8 and 0.66, CAD consortium clinical -2.0 and 0.71, CAD consortium clinical + CACS -2.0 and 0.78 (Fig).
Discrimination estimates by area under the receiver operating characteristic curve were: DF-score 0.71 (CI: 0.64-0.77), CAD consortium basic 0.71 (CI: 0.68-0.78), CAD consortium clinical 0.74 (CI: 0.68-0.80), CAD consortium clinical + CACS 0.85 (CI: 0.81-0.89), respectively (Fig).
In a large population of patients with low-intermediate risk profile, the new CAD consortium score seems to perform with improved calibration compared to the Diamond Forester score. However, with MPI as reference standard the CAD Consortium models still overestimated the pre-test probability of CAD.
The CAD consortium basic or clinical model does not seem to improve discrimination substantially compared to DF-score. However, addition of coronary calcium to the score optimizes discrimination and should be considered for optimal pre-test probability classification.