According to currently published reports only roughly one third of all heart catheterizations performed are leading to intervention . To improve diagnostic yield the current guidelines of international cardiology societies suggest non-invasive tests for the detection of CAD in symptomatic patients, such as stress echocardiography with reported pooled sensitivity of 80-85%, and specificity of 85-86%, before invasive procedure . Calcium scoring is a non-invasive modality which detects coronary calcifications being a surrogate marker for cardiac atherosclerosis. According to current literature the amount of calcified tissue correlates with the risk of developing adverse cardiac events in the future, such as myocardial infarction or sudden cardiac death. In addition it seems that CS is a better risk predictor for prospective cardiac events than currently available risk stratification score systems, such as Framingham or ATPIII risk score, in the way that CS reflects individual atherosclerotic disease burden. Especially the exclusion of coronary calcifications has been shown to exclude significant CAD in symptomatic patients [11–13]. However, the amount of calcium does not correlate well with stenosis severity . Thus, aim of our study was to evaluate the use of an approach combining CS determination and SPECT-MPI. Especially we intended to identify the settings in which one can refrain from additional imaging or incremental information is won by the combination of both modalities.
In our patient population of 351 consecutive patients admitted for invasive coronary angiography, disease prevalence of significant CAD needing intervention was 38% reflecting a typical population undergoing ICA. As expected, specificity and positive predictive value of coronary calcium scoring were rather low. Thus, detection of hemodynamically relevant stenosis of epicardial vessels utilizing this modality is not practicable. However, a high sensitivity of 99.2%, and a high NPV of 98.5% suggest CS being a useful tool for ruling out CAD by exclusion of calcified plaque tissue. In these patients an additional SPECT-MPI did not further increase diagnostic accuracy as the negative predictive value of SPECT-MPI is limited to 92.1%. Following these results we would have correctly refrained from ICA and SPECT-MPI in 66 patients (18.8%) of our study population. Although there are some publications questioning the safety of a zero to low coronary calcium scores for the purpose of CAD exclusion in symptomatic patients, several studies in large patient cohorts consistently have shown a high NPV of >95% utilizing a threshold of zero calcium similar to our study. But also in these studies the safety of ruling out CAD is reduced by using higher thresholds like a CS of 10 or a CS above the 25th age and sex adjusted percentile instead of the exclusion of calcified tissue [11–13]. The same was true for our study: increasing CS thresholds would lead to an increase in specificity and PPV. Concomitantly sensitivity and NPV decreased to a value of 40.6% and 72.5%, respectively. Thus CS would no longer be suitable as a filter for invasive angiography and still not sufficient for the detection of relevant stenosis with a PPV of 84.4%. Still, diagnostic accuracy of CS at a threshold of 100 is comparable to the diagnostic performance of SPECT-MPI.
Due to low specificities of positive coronary calcium scores, a functional test like SPECT-MPI following initial CS in case of calcified tissue detection is reasonable. In our 284 patients with detectable coronary calcifications 51.1% (n=145) showed ischemia in SPECT-MPI followed by intervention in 80.0% (n=116) of those patients. Specificity of this combined approach increased to 81% with a disease prevalence of now 46.5%. Additionally a positive ischemic burden and its relation to a specific coronary vessel territory, provides helpful information for guiding interventional procedure.
In remaining 139 patients with positive calcium scores, SPECT-MPI excluded significant CAD in 123 patients (88.5%). Although CS did not help further with respect to CAD detection in these particular patients, it nonetheless provides individual prognostic information and estimates risk of developing prospective major coronary events. In that respect CS might outperform SPECT-MPI in symptomatic patients in the long-time follow-up . An overall of 16 patients (11.5%) were misclassified by SPECT-MPI. Mean CS of these patients exceeded 300, with below 100 in only three cases, so that the majority of these patients presumably would have been put on secondary prevention therapy due to the CS results . Furthermore it remains questionable whether these patients in particular would benefit from additional revascularization according to the nuclear medicine sub-study of the Clinical Outcomes Utilizing Revascularization and Aggressive Drug Evaluation (COURAGE) trial . In this study Shaw et al. compared the effectiveness of adding coronary revascularization in order to minimize stress-induced ischemia with that of optimal medical therapy, utilizing SPECT-MPI to find the end point. They found that the outcome in patients with stable CAD was most strongly associated with the amount of residual ischemia after medical, or revascularization therapy, or both.
The utilization of different scanner technologies (MDCT and DSCT scanners) poses the risk of inconsistent calcium score values. According to a currently published head-to-head comparison of these two scanner types by Ghadri et al. revealed an excellent inter-scanner agreement for Agatston scores with a correlation coefficient of 0.976 despite relatively wide limits of agreement and a coefficient of variation 15.1% even in case of different vendors . Calcium mass score as well as calcium volume score were also evaluated showing also excellent correlation, but wider limits of agreement in Bland-Altman Analysis. Interestingly the authors found that the use of different workstations had a greater influence on the comparability than the scanner technology. To account for those possible sources of error a comparable acquisition protocol for both scanner technologies routinely is used at our institution and the same workstation with the same software solution was utilized to calculate the Agatston score. Indication for revascularization was solely based on detection of significant stenosis (≥75%) in QCA and was independently made by the examiner without necessary verification of existing ischemia in order to guarantee for a blinded study design. This approach to revascularization procedure represents clinical practice at the time of study initiation. For comparison of SPECT-MPI a functional invasive measurement such as fractional flow reserve (FFR) would have been more appropriate, but FFR usually is not a widely available diagnostic modality. The use of the ≥75%-stenosis-degree criterion in order to define hemodynamically relevant CAD theoretically results in a higher number of false-positive findings with initially lower specificity and PPV. The likelihood of a ≥50%-stenosis causing ischemia, which serves as CAD defining endpoint in the majority of studies is relatively low and does not warrant revascularization procedure in the majority of cases unless prove of ischemic burden. Additionally selection bias must be considered while interpreting the data presented. First our evaluation was based on patients with clinical indication for ICA as referred by primary-care providers that do not necessarily sent unselected chest-pain patients for cardiac evaluation. Second our study was conducted at a single university site including only regional patients. Thus our study population may not reflect general symptomatic CAD patients.