Incorporating coronary artery calcium scoring in the prediction of obstructive coronary artery disease with myocardial ischemia: a study with sequential use of coronary computed tomography angiography and positron emission tomography imaging

Background Additional strategies are needed to refine the referral for diagnostic testing of symptomatic patients with suspected coronary artery disease (CAD). We aimed to compare various models to predict hemodynamically obstructive CAD. Methods and results Symptomatic patients with suspected CAD who underwent coronary artery calcium scoring (CACS) and sequential coronary computed tomography angiography (CCTA) and [15O]H2O positron emission tomography (PET) myocardial perfusion imaging were analyzed. Obstructive CAD was defined as a suspected coronary artery stenosis on CCTA with myocardial ischemia on PET (absolute stress myocardial perfusion ≤ 2.4 mL/g/min in ≥ 1 segment). Three models were developed to predict obstructive CAD-induced myocardial ischemia using logistic regression analysis: (1) basic model: including age, sex and cardiac symptoms, (2) risk factor model: adding number of risk factors to the basic model, and (3) CACS model: adding CACS to the risk factor model. Model performance was evaluated using discriminatory ability with area under the receiver-operating characteristic curves (AUC). A total of 647 patients (mean age 62 ± 9 years, 45% men) underwent CACS and sequential CCTA and PET myocardial perfusion imaging. Obstructive CAD with myocardial ischemia on PET was present in 151 (23%) patients. CACS was independently associated with myocardial ischemia (P < .001). AUC for the discrimination of ischemia for the CACS model was superior over the basic model and risk factor model (P < .001). Conclusions Adding CACS to the model including age, sex, cardiac symptoms and number of risk factors increases the accuracy to predict obstructive CAD with myocardial ischemia on PET in symptomatic patients with suspected CAD. Supplementary Information The online version contains supplementary material available at 10.1007/s12350-022-03132-z.


INTRODUCTION
Traditionally, myocardial ischemia has been the gatekeeper for invasive coronary angiography and subsequent revascularization. 1 However, many symptomatic patients with suspected coronary artery disease (CAD) do not have myocardial ischemia. [2][3][4][5] Hence, alternative strategies are warranted in order to improve the referral for ischemia testing of this specific group of patients. Currently, European guidelines recommend physicians to estimate the pre-test probability of obstructive CAD-as a surrogate of myocardial ischemia-using the Diamond-Forrester approach by integrating age, sex and cardiac symptoms. 6,7 Additional information on the clinical profile of patients, such as the presence and extent of risk factors for cardiovascular disease and coronary artery calcium (CAC), holds potential to further refine these often overestimating pre-test probabilities of myocardial ischemia. 7,8 Coronary artery calcium scoring (CACS) seems particularly desirable since it is easily performed using non-contrast computed tomography (CT), requiring no intravenous contrast, low radiation exposure and lower costs (as compared to contrast-enhanced CT). 9 Also, the extent of CACS has been described to correlate well with ischemia. 10,11 Nevertheless, the optimal use of CACS in improving the pre-test probability assessment of ischemia has yet to be established in a large contemporary patient cohort. 7 Therefore, the present study aimed to compare three models to predict obstructive CAD with myocardial ischemia on positron emission tomography (PET) in symptomatic patients with suspected CAD: (1) a basic model: including age, sex and cardiac symptoms, (2) a risk factor model: adding number of risk factors to the basic model, and (3) a CACS model: adding CACS to the risk factor model.

Study design and patients
The study population included consecutive symptomatic patients with suspected CAD, who were referred for a PET/CT evaluation at the Turku University Hospital, Turku, Finland between 2007 and 2011. A detailed study design has been previously published. 12 Of those enrolled, 717 patients underwent (1) CACS and (2) sequential coronary computed tomography angiography (CCTA) and [ 15 O]H 2 O PET myocardial perfusion imaging to detect potential myocardial ischemia. The ethics committee of the Hospital District of South-West Finland approved the study protocol and waived the need for patients' written informed consent. The study complied with the principles of the Declaration of Helsinki. Patients with unavailable data on cardiac symptoms (n = 25) or who failed to follow the sequential protocol (n = 45) were excluded. Hence, the present study consisted of 647 patients ( Figure 1).

Image acquisition and analysis
Patients were scanned using a hybrid 64-detector row PET/CT scanner (GE Discovery VCT or GE D690, General Electric Medical Systems, Waukesha, Wisconsin). Protocols regarding image acquisition and analysis have been reported in detail. 12,13 CACS CACS was calculated from non-contrast CT scans according to the Agatston algorithm. 14 Scores were categorized into 0, 1-99, 100-399 and C 400.
Sequential CCTA and PET myocardial perfusion imaging CCTA was performed using intravenous low-osmolar iodine (48-155 mL; 320-See related editorial, pp. 189-192 400 mg/mL) as a contrast agent. 12,13 Prior to acquisition, intravenous metoprolol (0-30 mg) was administered to achieve heart rates \ 60/min. Sublingual nitroglycerin (800 lg) or isosorbide dinitrate (1.25 mg) was administered to achieve maximal coronary vasodilatation. Subsequently, according to study design, all patients with a suspected obstructive stenosis C 50% on CCTA by visual inspection of the attending physician underwent PET myocardial perfusion imaging to detect potential myocardial ischemia. PET myocardial perfusion imaging was performed using dynamic acquisition with [ 13 For stress, adenosine (rate: 140 lg/kg/min) was infused 2 min before the stress scan to induce maximal vasodilation. Patients received instructions to avoid caffeine 24 h prior to the scan, considering its interaction with adenosine. Stress scans were quantitatively analyzed according to the 17segment American Heart Association model using dedicated software (Carimas version 1.1.0, Turku, Finland) by an experienced physician, blinded to clinical or other data. 15,16 Absolute stress myocardial perfusion was generated in mL/g/min for the segments and left ventricle as a whole (not for all).

Obstructive CAD-induced myocardial ischemia
The reference standard for myocardial ischemia was defined as an absolute stress myocardial perfusion B 2.4 mL/g/min in C 1 segment on PET. 12 PET myocardial perfusion imaging was not performed in patients without a suspected obstructive stenosis on CCTA by study design. This specific group was considered to not have obstructive CAD-induced myocardial ischemia.

Statistical analysis
Normally and non-normally distributed continuous data are presented as means ± standard deviations (SD) and medians with interquartile ranges (IQR), respectively. Categorical data are presented as frequencies with percentages. First, comparisons of continuous data were performed with the Independent-Samples T test, Mann-Whitney U test, one-way analysis of variance or Kruskal-Wallis test, as appropriate. Comparisons of categorical data were performed using the v 2 test. Also, the Diamond-Forrest approach was applied to visualize the distribution of obstructive CAD with myocardial ischemia among patients according to age, sex and cardiac symptoms. 6,7 Additionally, negative predictive values (NPV) and positive predictive values (PPV) were calculated with different cut-points of CACS. Second, models were developed for the prediction of obstructive CAD-induced myocardial ischemia using logistic regression analysis. Uni-and multivariate logistic regression analysis was performed to assess the association between selected variables versus myocardial ischemia. In a stepwise manner, three prediction models were defined: (1) basic model: including age, sex and cardiac symptoms, (2) risk factor model: adding number of risk factors to the basic model, and (3) CACS model: adding CACS to the risk factor model. Measures of association were expressed as odds ratios (OR) with 95% confidence intervals (CI). Goodness of model fit was compared with the likelihood ratio test. Third, performance of the models was evaluated using discriminatory ability. Discriminatory ability was assessed using area under the receiver-operating characteristic curves (AUC), integrated discrimination improvement (IDI) and net reclassification improvement (NRI). AUCs were compared with the DeLong's test. 17,18 A two-sided P-value of \ .05 was considered statistically significant, and all statistical analyses were performed with R (version 3.0.3, R Development Core Team, Vienna, Austria), SPSS software (version 26, SPSS IBM Corp., Armonk, New York) and MedCalc software (version 19.2.0, Ostend, Belgium).

Patients
Baseline characteristics of the patients are shown in Table 1. In total, 647 patients (mean age 62 ± 9 years, 45% men) underwent CACS and sequential CCTA and [ 15 O]H 2 O PET myocardial perfusion imaging for ischemia assessment. CCTA ruled out an obstructive stenosis in 338 patients; they were considered to not have obstructive CAD-induced myocardial ischemia (and did not undergo PET myocardial perfusion imaging by the sequential study design) ( Figure 1). CCTA revealed a suspected obstructive stenosis in 309 patients. Obstructive CAD with myocardial ischemia on PET was present in 151 (23% out of 647) patients. Patients with myocardial ischemia were older (63 ± 8 years vs. 61 ± 10 years, P = .002), more often male (72% vs. 37%, P \ .001) and presented more frequently with typical angina (37% vs. 22%, P \ .001) as compared to patients without ischemia. In addition, patients with myocardial ischemia had more risk factors for cardiovascular disease (P \ .001) and used more medications (P B .007). The distribution of ischemia among patients based on the Diamond-Forrester approach according to age, sex and cardiac symptoms was demonstrated in Supplemental Table 1.
Sequential CCTA and PET myocardial perfusion imaging Details regarding sequential CCTA and PET myocardial perfusion imaging are shown in Figure 1. In patients with obstructive CAD-induced myocardial ischemia, a median of 10 segments (IQR 5-15 segments) was affected. Patients with myocardial ischemia had a reduced global stress myocardial perfusion as compared to patients without ischemia on PET (2.3 ± .7 mL/g/min vs. 3.9 ± .9 mL/g/min, P \ .001) ( Table 2).

Prediction of obstructive CAD with myocardial ischemia
Model development using logistic regression analysis In the univariable analysis, age, male sex, typical angina, all individual cardiac risk factors (except for family history of CAD) and the number of risk factors per-patient were each associated with obstructive CAD-induced myocardial ischemia (P B .005). Furthermore, CACS was a significant univariable predictor of myocardial ischemia, both as a continuous (P \ .001) and categorized score (P \ .001) ( Table 3). In the multivariable analysis, prediction models of ischemia were defined using a stepwise approach: (1) basic model: including age, sex and cardiac symptoms, (2) risk factor model: adding number of risk factors to the basic model, and (3) CACS model: adding CACS to the risk factor model (  Model performance using discriminatory ability AUC for the discrimination of obstructive CAD with myocardial ischemia was .746 (95% CI .701-.791) for the basic model, .790 (95% CI .751-.830) for the risk factor model and .849 (95% CI .813-.884) for the CACS model ( Figure 4). The CACS model had a significantly better discriminatory ability than the basic model (P \ .001) and risk factor model (P \ .001). Also, the CACS model provided incremental predictive information over the basic model (IDI = .176, P \ .001 and NRI = .633, P \ .001) and risk factor model (IDI =  Table 2).

DISCUSSION
The present study evaluated 647 symptomatic patients with suspected CAD from a large contemporary patient cohort, who underwent CACS and sequential CCTA and [ 15   Bold values are statistically significant (P \ .05) Values are presented as mean ± SD, median (IQR) or n (%) CACS, coronary artery calcium score; CAD, coronary artery disease; PET, positron emission tomography. Definitions: *Values only available for patients who underwent PET myocardial perfusion imaging, as depicted in Figure 1 imaging for ischemia assessment. We compared three models to predict obstructive CAD with myocardial ischemia on PET: (1) a basic model, (2) a risk factor model and (3) a CACS model. CACS was strongly and independently associated with myocardial ischemia. Moreover, by incorporating CACS into the pre-test  probability assessment, the discrimination of ischemia significantly improved compared to the basic model and risk factor model. These findings suggest a possible role for routine CACS detection in symptomatic patients in order to refine referral for ischemia testing, by either triaging them away from (in case of low CACS) or towards (in case of high CACS) this test. Particularly, the NPV of CACS = 0 was excellent (97.8%) irrespective of the cardiac symptoms at presentation (96.0-98.5%). Our approach is an example of the stepwise application of non-invasive imaging tests, which in turn could lead to more cost-effective care.

CACS in asymptomatic patients: preventative care
Anatomical imaging with CACS has been initially introduced as a screening tool for CAD in asymptomatic patients with the aim of improving cardiovascular risk assessment and guiding primary preventative care. [19][20][21] Regarding cardiovascular risk assessment, various large long-term population-based studies have uniformly reported on the association between CACS and major adverse cardiac events in asymptomatic patients without known CAD. [22][23][24][25] Especially, a CACS = 0 has been linked to a very low risk of adverse events (power of zero). 23,26,27 Regarding preventative care strategies, it has been clearly demonstrated that a CACS = 0 can reclassify a large subset of asymptomatic patients (44%) in whom statins would have been otherwise considered or recommended (atherosclerotic cardiovascular disease risk score C 5%) according to existing guidelines. 28

CACS in symptomatic patients: ischemia
On the other hand, anatomical imaging in symptomatic patients with suspected CAD has the aim to identify hemodynamically obstructive CAD (coronary artery stenosis C 50%) that causes ischemia. 29 Few studies have reported on the association between CACS and myocardial ischemia on PET in symptomatic patients with suspected CAD. [30][31][32] Schenker et al. analyzed 695 symptomatic patients with suspected CAD, who underwent CACS and PET myocardial perfusing imaging using a hybrid PET/CT scanner. 30 In line with our results, a stepwise increase was demonstrated in the frequency of myocardial ischemia with increasing CACS (16% for CACS = 0 to 49% for CACS C 1000). Furthermore, adding CACS to a model including age, sex, cardiac symptoms and risk factors improved the  31 Applying this strategy, a strong association was shown between CACS = 0 and the absence of myocardial ischemia, yielding a negative predictive value of 100%. Again, these results were overall highly consistent with the findings in the current study, showing only a 2% prevalence of myocardial ischemia in patients with CACS = 0. Similar findings were derived from studies using single photon emission computed tomography as the reference standard for myocardial ischemia. 11,29 However, it should be noted that PET has enhanced diagnostic performance over single photon emission computed tomography, in particular when myocardial perfusion is quantitatively analyzed. 33 Additionally, all latest generation PET scanners are combined with a CT scanner into a hybrid system, of which the low-dose non-gated CT transmission scan can be used to not only perform attenuation correction of the PET images but also to perform visual assessment of CAC. 34,35 With the rapid development of artificial intelligence with sophisticated algorithms, this approach holds potential for the automated assessment of CAC from non-gated CT scans. 36,37 Limitations Some limitations of the present study need to be addressed. First, our study had a retrospective observational design with limitations such as (unmeasured) confounding factors and selection bias. For instance, of those enrolled in the registry, CACS was performed per protocol in all patients for risk stratification purposes, but not analyzed in some patients due to logistical or technical reasons. 12 Second, PET myocardial perfusion imaging was not performed in patients without suspected obstructive stenosis on CCTA according to study design. Absence of myocardial ischemia in this specific group of patients was therefore assumption-based, but in line with published literature. 31 Nevertheless, we acknowledge that diffuse, heterogenous CAD or microvascular dysfunction could have contributed to downstream myocardial perfusion abnormalities. [38][39][40][41] Unfortunately, we were not able to analyze this in detail due to the sequential design of the study. Third, PET myocardial perfusion findings were solely interpreted on a per-patient basis, since CACS was not available on a per-vessel basis. Lastly, utilizing CACS as a gatekeeper to ischemia testing still needs prospective and randomized data. However, our study adds to the wealth of data suggesting that patients with CACS = 0 are at low risk (but not risk free). To this end it should be emphasized that clinical decisions should always be individualized.

NEW KNOWLEDGE GAINED
The stepwise application of non-invasive imaging tests, including an initial CACS, can potentially refine the referral for ischemia testing of symptomatic patients with CAD.

CONCLUSION
In symptomatic patients with suspected CAD, a CACS model including age, sex, cardiac symptoms, number of risk factors and CACS allows for accurate and superior prediction of obstructive CAD with myocardial ischemia on PET.

Disclosures
Dr. Wang is supported by a research grant from the University of Turku. Dr. Saraste received speaker or consultancy fees from Amgen, Abbott, Astra Zeneca, Bayer, Boehringer Ingelheim and Pfizer. Dr. Knuuti received consultancy fees from GE Healthcare and AstraZeneca and speaker fees from GE Healthcare, Bayer, Lundbeck, Boehringer Ingelheim, Pfizer and Merck, outside of the submitted work. Dr. Bax received speaker fees from Abbot Vascular and Edwards Lifesciences. The Department of Cardiology, Leiden University Medical Center, Leiden, the Netherlands has received unrestricted research grants from Bayer, Abbott Vascular, Medtronic, Biotronik, Boston Scientific, GE Healthcare and Edwards Lifesciences. The remaining authors have no relevant disclosures.

Open Access
This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.