Cardiometabolic risk assessment by imaging: current status and future perspectives

This editorial refers to ‘Incremental prognostic value of non-alcoholic fatty liver disease over coronary computed tomography angiography findings in patients with suspected coronary artery disease’, by Ichikawa et al., pp. 2059–2066. The close relationship between adipose tissue dysfunction and cardiovascular disease is well acknowledged.1 Dysregulated adipose tissue function can lead to vascular dysfunction via the release of pro-inflammatory adipokines that can affect the cardiovascular system and promote atherosclerosis.2 Since the in vivo assessment of adipose tissue function for clinical decision making is challenging, in everyday clinical practice adiposity status has been traditionally assessed by surrogate markers, such as anthropometric measurements of obesity (e.g., body mass index) or of ectopic fat deposition (e.g., waist-to-hip ratio). More recently, there has been a growing interest in the use of imaging for the quantitation of ectopic fat mass/volume or even the qualitative assessment of adipose tissue. In this issue of European Journal of Preventive Cardiology, Ichikawa et al.3 explored the prognostic value of hepatic steatosis (HS) for cardiovascular events over and above coronary anatomy in a Japanese cohort of 1,148 patients undergoing a coronary computed tomography angiography (CCTA) scan. HS, defined as a hepatic-to-spleen attenuation ratio of <1.0, was present in 22% of the study population, which was an overall low-risk population for cardiovascular events (incidence of 3.5%), typical of those undergoing diagnostic CCTA. In this cohort, HS was predictive of the composite endpoint of cardiac death, acute coronary syndrome, and late revascularization; moreover, HS enhanced risk discrimination and reclassification for cardiovascular events over and above Framingham risk score and adverse CCTA findings i.e., obstructive coronary artery disease (CAD) and/or presence of high-risk plaque (HRP) features. Importantly, HS remained an independent predictor of cardiovascular events even among high-risk patients, such as those with metabolic syndrome or diabetes.3 These are observational findings derived from a large clinical cohort which are potentially important since they (i) highlight the value of metabolic risk assessment as an integral part of risk stratification of stable patients, (ii) support the use of cardiometabolic imaging biomarkers in clinical practice, and (iii) demonstrate that a single CCTA scan can provide a wealth of diagnostic/prognostic information on top of coronary anatomy assessment. The use of CCTA has now a Class I recommendation for the diagnostic assessment of CAD in stable patients. Further to its diagnostic uses, CCTA is a highly attractive tool for screening purposes. Coronary calcium score (CCS) stands out as one of the simplest and most powerful biomarkers for the risk stratification of asymptomatic patients, providing surrogate information on coronary plaque burden by non-contrast CT. Administration of iodine contrast allows the luminal assessment of coronary arteries, and the extraction of valuable information on the presence of obstructive or non-obstructive CAD, as well as plaque phenotyping and the detection of HRP features, which all provide independent prognostic information for cardiovascular events.4 On top of the CCTA-provided insight into coronary anatomy and plaque-related metrics, other valuable information on patient’s risk profile can be extracted from the analysis of adjacent tissues to the heart. For example, adipose tissue can be easily identified on CT by its radiodensity or CT attenuation, typically defined as all voxels within the range of −190 to −30 Hounsfield Units. Quantitative information on epicardial or pericardial adipose tissue is independently associated with the risk of cardiovascular events on top of traditional risk factors in large clinical cohorts.5 But why are cardiac adiposity indices (i.e., adipose tissue mass) positively correlated with cardiovascular disease risk? Pericardial or epicardial adipose tissue volume is a good surrogate index of systemic obesity status, and more specifically of ectopic fat deposition.6 In clinical cohorts, thoracic adipose tissue volume by CT is strongly correlated with visceral adiposity, and both are positively associated with systemic insulin resistance6 and overall cardiometabolic risk profile.7 The findings of Ichikawa et al.3 are in the same line: they demonstrate the value of CT biomarkers of ectopic fat deposition for cardiovascular risk assessment. Therefore the provided evidence is not entirely novel; for example, CT biomarkers of ectopic adiposity, such as epicardial adipose tissue volume and non-alcoholic fatty liver disease, were independently associated with cardiovascular outcomes in a cohort of 2,068 patients undergoing a non-contrast CT scan for CCS and clinical follow-up of 14 years.8 However, the findings of Ichikawa et al.3 demonstrate also the incremental value of HS on top of obstructive CAD or HRP. Rest assured, the quest for clinically useful imaging biomarkers of metabolic risk has just begun. Unsupervised machine learning of whole-body CT scans can identify adverse metabolic phenotypes—of human fat and/or skeletal muscles—that are independently associated with adverse health outcomes.9 In addition to assessing for the presence of non-alcoholic fatty liver disease or to quantitate the CT volume of distinct adipose tissue depots, fat quality per se (e.g., assessed by fat CT attenuation) may be a useful surrogate marker of cardiometabolic risk.7 Also, along with the other fat depots, perivascular adipose tissue (PVAT) has now emerged as the next big imaging target in the field of cardiovascular medicine. PVAT imaging, so far feasible only by CCTA, can be used to extract information on coronary inflammation.10 Such information may be useful both for cardiac risk stratification purposes as well as in the clinical diagnosis of conditions such as vasospastic angina.10 Other non-invasive imaging modalities, such as Cardiac Magnetic Resonance, could also provide potentially useful information on ectopic fat deposition, and the detection of intra-myocardial fat.11 Last but not least, the application of machine learning and artificial intelligence in medical imaging is expected to lead to the development of novel biomarkers for more precise cardiometabolic risk assessment. CT radiomics have been used to extract information from adipose tissue about its structure and biology to further refine risk prognostication of stable patients.4 Convolutional neural networks and the deep learning of CT scans of human liver, adipose tissue or the heart have also been successfully applied to explore the wealth of data stored in medical images that remain hidden to the naked eye and are expected to generate new phenomaps of human health and disease. Technological advancements in non-invasive and cross-modality imaging in conjunction with the analysis of big data are expected to path the way for the development of novel imaging biomarkers of cardiometabolic risk. A pipeline for the development and validation of new imaging biomarkers of cardiometabolic risk is provided in Figure 1. Clinical researchers should function as the gatekeepers of this process, ensuring adequate technical and clinical validation, confirmation of the added prognostic value, and the net clinical benefit of any new imaging biomarker before introduction in clinical practice. A pipeline for the development and validation of new imaging biomarkers of cardiometabolic risk. The figure summarizes the four main pipelines (A–D) for the development and discovery of new imaging biomarkers of cardiometabolic risk. These approaches can then lead to translational imaging biomarkers (i.e. the translation of experimental or clinical findings by imaging) or even the generation of imaging phenomaps of cardiometabolic disease. A rigorous validation approach would be ultimately required before the introduction of these new biomarkers in clinical practice. ML, machine learning. Conflict of interest: none declared. The opinions expressed in this article are not necessarily those of the Editors of the European Journal of Preventive Cardiology or of the European Society of Cardiology.