Our study represents the largest investigation to date on the association of CT-derived EAT volume and density with postprocedural outcomes after TAVR. Moreover, it is the first of its kind in a US-based cohort comprising diverse race/ethnic backgrounds and with simultaneous cardiac damage assessment. The main findings of our study can be summarized as follows: (1) Low EAT volume (< 49 cm3) was independently associated with higher 4-year all-cause mortality, even when adjusted for CD, and contrary to what has been demonstrated in previous literature; (2) Higher EAT density (≥ -86 HU) was also associated with increased all-cause mortality, independent of clinical variables but not when AS cardiac damage stage was accounted; (3) EAT volume progressively decreased, and EAT density increased with worsening AS cardiac damage stage; and (4) AS cardiac damage stage was the strongest independent predictor of mortality in this diverse cohort.

In the widespread TAVR era, CT has become an indispensable tool for adequate patient selection and accurate preprocedural TAVR planning based on its ability to predict intraprocedural and early postprocedural complications [27,28,29,30,31]. In addition, CT could provide a more comprehensive phenotype involving not only the valve and myocardium but also more systemic disease markers such as the psoas muscle, subcutaneous fat, and bone density, all of which have shown a direct impact on longer-term outcomes [6, 32,33,34]. In this context, EAT could also serve as an opportunistic imaging biomarker for further refining TAVR therapy candidates or guiding patient postprocedural care.

The ultimate mechanistic explanation for EAT and its interplay with different CVD diseases remains a matter of debate [35, 36]. It is accepted that the nature of this relation is dynamic and varies with the age and severity of certain pathological conditions [9].

Our observation that lower (in contrast to higher) EAT volumes were associated with higher mortality represents a unique perspective to the existing literature [10, 20]. We believe this could be explained by the fact that we evaluated a population with more severe disease progression, as depicted by cardiac damage stage ≥ 2 in the vast majority of patients. In this specific setting, EAT may become dysfunctional and incur fibrotic and apoptotic changes, leading to an overall lower volume. Previous observational studies have found a significant reduction in both EAT volume and thickness among patients with HF with reduced LVEF (HFrEF) compared with both healthy individuals and those with preserved ejection fraction [35]. Furthermore, reduced EAT in HFrEF patients has been linked to worse LV function, adverse myocardial remodeling, and myocardial damage by troponins [37]. This observation supports the hypothesis that the reduction in EAT volume can be attributed to cardiac cachexia and catabolic-related adverse effects [35]. Large studies from North America, Europe, and Asia have shown that underweight patients undergoing TAVR exhibit higher rates of short and mid-term mortality than normal or overweight individuals, thereby supporting the hypothesis of a cardiovascular obesity paradox in this population [38,39,40]. However, our findings were independent of patient’s BMI and remained significant after adjusting for EAT volume by height.

Additionally, in patients with more advanced or end-stage organ disease and elderly individuals, the thermogenic function of EAT can be decreased, with reciprocal increases in the expression of genes encoding profibrotic and pro-apoptotic factors [9, 41]. The present study extended these observations to patients with severe AS undergoing TAVR.

Interestingly, higher EAT density, likely reflecting increased inflammatory activity, was strongly associated with mortality even after adjusting for age, sex, and BMI. This observation aligns with recent work by Salam et al. [19] in 1,197 patients undergoing TAVR. The authors found that a higher EAT density (>-81 HU) was strongly associated with increased 2-year mortality independent of clinical characteristics and surgical risk scores. In another recent study by Sato et al. [21] among 125 consecutive TAVR patients, authors showed higher EAT density (>-74.3 HU) predicted the occurrence of major adverse cerebral and cardiovascular events (MACCE) (AUC = 0.685) and the EAT density and EuroSCORE were independently associated with MACCE. Recently, EAT density was found to be related to COVID-19 severity, independent of the presence of coronary artery disease [42]. These findings are supported by the concept that aging, environmental factors, and genetics can lead to EAT adipocyte dysfunction, producing pro-inflammatory adipokines implicated in CVD pathogenesis [9]. However, the relationship between EAT density and CVD, especially for coronary atherosclerosis, has been considered controversial, with some studies showing direct and other inverse correlation [43].

In the context of aortic stenosis, EAT could contribute to the progression of valvular calcification and adverse ventricular remodeling through paracrine effects [44]. Parisi et al. [16]. enrolled 95 patients with severe calcific AS who underwent cardiac surgery for AVR. EAT thickness was assessed using echocardiography in these patients, and inflammatory profiles were analyzed using cytokines measurements. EAT thickness was significantly higher in patients with AS than in the control group. Notably, the EAT secretome of patients with increased EAT thickness showed higher levels of inflammatory mediators. Furthermore, the thickness of EAT significantly correlated with the levels of different pro-inflammatory and pro-atherogenic cytokines, such as IL-6, TNF-α, MCP-1, and IL-1β, therefore, the greater the thickness of EAT, the greater the secretion of these mediators. Interestingly, we found that patients with lower volume but more inflamed (higher density) EAT did worse than those with high volume and density, perhaps reflecting an advanced stage of the disease where there is persistent inflammation despite the decrease in the volume/thickness of the adipose tissue. Interestingly, a pro-inflammatory activation of EAT in patients with AS and EAT involvement in aortic valve calcific degeneration has been suggested [44].

While the correlation between greater extravalvular damage and increased mortality is not a novel finding, we believe that the specific composition of our cohort, including its diversity in terms of race/ethnicity and socioeconomic status, offers a unique perspective on interpreting this relationship. For instance, in the major TAVR trials, race/ethnicity was only reported in the PARTNER 3 trial, with non-White patients being only 7.7% in the TAVR and 9.9% in the surgical AVR group [45]. The reason for the underrepresentation of ethnic groups in trials is not clear, but we also know that in clinical practice, non-Hispanic Black and Hispanic patients are less likely to receive TAVR therapy, even within metropolitan areas [46].

The prognostic impact of EAT volume was independent of CD stage, but EAT density was not. This could be explained by the fact that cardiac damage remains one of the strongest independent prognostic markers in this population, establishing a high bar for any increased prognostic value [4, 47]. Additionally, EAT density quantification is less standardized than volume, as it is known to vary under the influence of CT acquisition parameters and the presence or absence of contrast [43]. Interestingly, the fact that EAT volume decreased and EAT density increased with worsening CD stage could be related to the role of EAT as a marker of more severe cardiac remodeling and the failure of the adaptation mechanisms to AS [48].

Since preprocedural CT imaging is routinely performed for TAVR planning, EAT quantification could provide incremental risk prediction without additional testing or radiation exposure. However, standardized EAT thresholds are still needed, given the overlap in values between patients with and without CVD events [10]. Additionally, the influence of contrast-enhanced images in EAT volume assessment should be considered. Variations of the lower threshold have been reported to have a negligible effect on the total EAT volume [49]. Bucher et al. reported significant differences in epicardial fat volume between non-contrast and contrast-enhanced (130.7 ± 49.5 ml vs. 87.2 ± 38.5 ml, p < 0.001) data sets at a − 30 HU upper threshold. Mean EAT volume for contrast-enhanced data sets at a -15 HU upper threshold (102.4 ± 43.6 ml) could be approximated most closely by non-contrast scans at a -45HU upper threshold (105.3 ± 40.8 ml) [25]. In our series, an upper threshold of -15 HU was used.

Our study is limited by the retrospective and observational nature of our findings, precluding causal inference and the presence of residual confounding that may have influenced our results despite the efforts made to account for those. The impact of medications, biomarkers, and/or frailty scale on clinical outcomes is unknown. In addition, since only patients with severe AS undergoing TAVR were studied, it remains unclear whether findings apply to those with surgical aortic valve replacement, conservative treatment, or those who died before receiving any treatment. Finally, we selected all-cause mortality as our primary outcome to improve statistical power and minimize potential biases in determining specific causes of death. This context is crucial for interpreting our findings accurately.

Prospective studies with larger cohorts and longer follow-up periods are needed to validate our findings and elucidate the underlying mechanisms driving the observed associations.