Reply: Prognostic Role of Pericardial Fat on the Incidence of Heart Failure

We thank Dr Bonapace and colleagues for their kind comments on our article (1). In the MESA (Multi-Ethnic Study of Atherosclerosis), hepatic and splenic densities, expressed in Hounsfield units (HU), were available on nonenhanced computed tomographic (CT) scans at baseline on 6,615 and 4,396 participants, respectively (2). In prior analyses, hepatic fat per se was not associated with cardiovascular disease outcomes (3). Nonetheless, as suggested, we have examined its influence on the association between pericardial fat volume (PFV) and newly diagnosed heart failure (HF). For this purpose, we defined hepatic attenuation as an average of mean HU from 2 regions of interest, one anterior and the other posterior, each >100 mm² in area, in the right hepatic lobe. We measured splenic attenuation on one region of interest >100 mm² in area. Intra/inter-reader reproducibility was high for measurements of both hepatic attenuation (r = 0.96/0.99) and splenic attenuation (r = 0.99/0.99) (2). We considered mean hepatic attenuation <40 HU (that correlates with >30% hepatic fat) or mean hepatic-to-splenic attenuation ratio <1.0 (ie, reversal of hepatic-to-splenic attenuation ratio) as indicative of fatty liver disease (2).

To examine the impact of nonalcoholic fatty liver disease (NAFLD) on the association between PFV and newly diagnosed HF, we excluded 606 participants due to missing information on PFV (n = 29), hepatic attenuation (n = 195), or alcohol use (n = 48), and history of heavy alcohol use (>7 drinks per week in women; >14 drinks per week in men) (n = 334); thus, 6,208 (91.1%) of 6,814 participants were eligible. An additional 2,091 with missing information on splenic attenuation were excluded in analyses requiring indexation of hepatic-to-splenic attenuation.

PFV was negatively/inversely correlated, albeit poorly-to-weakly, with mean hepatic attenuation (r = −0.27; P < 0.001) and mean hepatic-to-splenic attenuation ratio (r = −0.17; P < 0.001). High PFV (≥70 cm³ in women; ≥120 cm³ in men) compared with normal PFV had lower mean hepatic attenuation (56.1 ± 13.8 vs 61.6 ± 10.0; P < 0.001) and mean hepatic-to-splenic attenuation ratio (1.14 ± 0.32 vs 1.21 ± 0.23; P < 0.001) suggestive of higher hepatic fat content. High PFV compared with normal PFV had a greater proportion of mean hepatic attenuation <40 HU (10.9% vs 3.8%; P < 0.001) and mean hepatic-to-splenic attenuation ratio <1.0 (26.5% vs 12.4%; P < 0.001) suggestive of higher prevalence of NAFLD.

In sex-stratified and age-adjusted, partially adjusted, and fully adjusted models (1), the association between PFV and newly diagnosed HF was essentially unchanged after additional adjustment for hepatic fat estimates (Figure 1). Specifically, in sex-stratified and fully adjusted models, high PFV was associated with hazard ratios of 1.78 (95% confidence interval [CI]: 1.40-2.25; P < 0.001) with mean hepatic attenuation value in the model, 1.73 (95% CI: 1.37-2.18; P < 0.001) with mean hepatic attenuation <40 HU, 1.87 (95% CI: 1.42-2.47; P < 0.001) with mean hepatic-to-splenic attenuation ratio, and 1.86 (95% CI: 1.41-2.46; P < 0.001) with mean hepatic-to-splenic attenuation ratio <1.0. Of note, none of the hepatic fat measures were associated with newly diagnosed HF (all P ≥ 0.33).

Figure 1.

PFV and the Risk of HF Adjusted for Hepatic Fat Estimates

Elevated pericardial fat volume (PFV) was associated with a greater risk of heart failure (HF) after adjustment for computed tomographic estimates of hepatic fat. *High PFV was defined as ≥70 cm³ in women and ≥120 cm³ in men. †Hazard ratios are shown on a logarithmic scale. Vertical bars = 95% confidence intervals. ‡HU = Hounsfield Units. §r = correlation coefficient, P = probability value, n = number of participants. ‖Values are mean ± SD (n) or n (%). P values were calculated using Student’s t-test for continuous and chi-square test for categorical variables. #Excluded additional 26 participants in analyses involving hepatic attenuation and 21 in analyses pertaining to hepatic-to-splenic attenuation ratio due to missing information on HF events. **Py = person-years of follow-up. ††Race, cigarette smoking, alcohol consumption, and vigorous physical activity at baseline. ‡‡Hypertension, diabetes mellitus, and dyslipidemia at baseline and interim myocardial infarction during follow-up.

In summary, CT-based estimate of hepatic fat or NAFLD was cross-sectionally associated with pericardial fat; however, it did not substantially influence the prospective association between pericardial fat and incident HF. This finding further reinforces the notion that pericardial fat predisposes to the development of HF irrespective of fat depots elsewhere in the body.