Reply to: Quantification of perivascular inflammation does not provide incremental prognostic value over myocardial perfusion imaging and calcium scoring

Abstract

We read with interest the paper by Bengs et al. [1] on the prognostic value of vascular inflammation measured by analysing perivascular fat attenuation on coronary computed tomography angiography (CCTA) for major adverse cardiovascular events (MACEs). In a total of 314 patients who were followed up for a median of 2.7 years, a cut-off in pericoronary fat CT attenuation ≥ − 70.1 HU around the right coronary artery (RCA) did not predict MACEs (defined as the composite of cardiac death, non-fatal myocardial infarction, hospitalization for any cardiac reason, and late revascularization during follow-up) [1]. We would like to endorse the authors for their decision to explore the prognostic value of pericoronary fat attenuation, a recently introduced CT-basedmetric of vascular inflammation [2], in a population of patients undergoing combined CCTA and SPECT/MPI. However, we believe that there are certain limitations in the presented study that hinder the extrapolation of meaningful conclusions on the prognostic value of pericoronary fat attenuation. The authors report measurement of RCA pericoronary fat attenuation, but it is unclear whether scan acquisition settings and biological and anatomical factors were taken into consideration in the calculations of reported fat attenuation measurements. Therefore, although the anatomical coronary segments used for analysis were similar to previous studies [2, 3], it is questionable whether the measured pericoronary fat attenuation corresponds to the perivascular Fat Attenuation Index (FAI) as it has been originally developed and validated [2, 3]. We have recently highlighted the difference between non-adjusted metrics of perivascular adipose tissue attenuation compared to the fully adjusted FAI metric [4]. Pericoronary fat attenuation measurements have to take into account various technical, anatomical, and biological factors, for which they need to be corrected for to accurately provide information on vascular inflammation. This is the reason why FAI has superior prognostic value to the unadjusted pericoronary fat attenuation measurements, as we have previously discussed [4]. Another concern relates to the small sample size. The study of Bengs et al. [1] was clearly underpowered to study hard clinical endpoints; hence, a composite definition of MACEs was used that included softer endpoints such as hospitalization for any cardiac reason, and late revascularization during follow-up. Also, from the original cohort of 492 patients, only a final of 314 patients were included in the analysis, which suggests high risk of selection and/or attrition bias. There are also certain additional issues in the study design that should be weighed against. A cut-off of − 70.1 HU previously suggested by our group [3] referred to the fully adjusted fat attenuation (i.e. FAI), and not the average pericoronary fat attenuation. Also, the population of the study by Bengs et al. [1] is strikingly different to the typical populations undergoing clinically indicated diagnostic CCTA,which comprises lower risk cohorts,with lower prevalence of coronary stenoses and known coronary artery disease. Therefore, previously used cut-offs in FAI [3] may not be directly transferable to all populations. A cohort-specific cut-off should be explored by including FAI as a continuous variable in the regression models used. Besides, experts strongly advise against the dichotomization of continuous variables in clinical models which may give rise to spurious effects [5]. This article is part of the Topical Collection on Cardiology.