This cross-sectional, retrospective, observational and descriptive study included patients with severe isolated AS referred to Hospital Universitario de Navarra for aortic valve (AV) replacement from June 2013 to February 2015. AS was diagnosed following the European Society of Cardiology recommendations.20 Serum were collected 24hours before the surgery and kept at −80° C until batch analyses. Concomitant coronary artery disease (CAD) was evaluated. Informed consent was obtained. The study protocol agreed with the ethics guidelines of the 1975 Declaration of Helsinki and was approved by the human research committee of our institution (Comité Ético de Experimentación Clínica. Gobierno de Navarra, Departamento de Salud; Ethics number: 2015/26).

Transthoracic echocardiography was performed with a VIVID7 3.5MHz ultrasound scanner (GE Ultrasound) or an IE-33 (Phillips Healthcare), according to current guidelines. Transaortic maximum and mean pressure gradients and AV area were assessed by continuity equation.

CMR imaging studies were performed using a 1.5 T scanner (Avanto). Steady-state free precession sequences were used to assess LV dimensions, mass and function, and AV area. The LGE images were acquired 10 to 15minutes after injection of intravenous gadolinium contrast agent (Dotarem, 0.1 mmol/kg body weight). Inversion recovery-prepared spoiled gradient echo images were acquired to detect areas of LGE. Inversion delay times were optimized to null normal myocardium. Cine images (steady-state free precession) were used to assess LV volumes, mass, and function. Endocardial and epicardial LV borders were manually contoured using ARGUS software (Siemens Medical Solutions). AS severity was assessed using CMR-derived planimetry of the AV area as well as phase-contrast sequences to determine flow velocities.

LGE mass was semiautomatically calculated using QMass MR7 software. Intramyocardial (diffuse or focal and including the interventricular junction) and subendocardial fibrosis, as well as widespread diffuse fibrosis from the mitral and aortic rings, were considered pathological.

Circulating sST2 levels were quantified using a commercially available enzyme-linked immunosorbent assay (Presage, Critical Diagnostics) following the manufacturer's protocol.

Categorical variables are expressed as percentages and were compared using the chi-square test, or Fisher exact test, as appropriate. Continuous variables were assessed using the unpaired Student t- test and 1-way analysis of variance if normally distributed. Otherwise, the Mann-Whitney U test or Kruskal-Wallis test were used. Post hoc tests were performed when appropriate. The Pearson or Spearman correlation coefficients were calculated. Linear regression modeling was used to show the clinical and biological markers able to stratify/identify patients with RMF. GraphPad Prism 9.0 software, SPSS version 28.0.1.0 or the R statistical package, v. 3.6 were used.

The data supporting the findings of this study are available from the corresponding author, upon reasonable request.

This cross-sectional, retrospective, observational and descriptive study included 79 all-coming patients (60.8% men, 73.0 [68.0-78.0] years) with severe AS. The baseline and demographic characteristics of our cohort are displayed in table 1. Based on previously published yields of serum ST2 in AS patients undergoing valve replacement,21 we calculated the magnitude of effect size to be 0.625.

To validate the sample size and power of our cohort, Cohen's F effect size (f2=0.39 and α=0.05) were calculated. The power of our sample size was calculated to be> 90%.

Replacement myocardial fibrosis is associated with greater left ventricular remodeling and dysfunction

Patients were classified as LGE-positive or -negative in table 1. All variables recorded in table 1 were compared among the LGE groups using chi-square tests for categorical variables, and unpaired univariate Student t or Mann-Whitney U tests for continuous variables. Importantly, LV RMF was detected in 53.2% patients and was more likely in men (71.4% vs. 48.6%, P=.039). CAD prevailed among positive LGE patients (50% vs. 27%, P=.037). Positive LGE patients had higher LV dilation, enhanced LV mass and lowered ejection fraction as evidenced by the CMR parameters LV end-diastolic diameter (P=.001), interventricular septum thickness (P=.017), indexed end-diastolic volume (EDV) (P=.010), indexed end-systolic volume (ESV) (P<.001), indexed LV mass (P=.001) and left ventricular ejection fraction (LVEF) (P <.001), all regardless of the degree of stenosis. Only positive LGE patients (23.8%, P=.001) had LVEF <50%.

Levels of sST2 were higher in patients with RMF than in those without (37.2 (31.2 to 39.8) vs. 18.9 [16.0 to 22.7], P <.001). On receiver operating characteristic (ROC) curve analysis and Youden's statistic, the sST2 cutoff for the identification of patients with RMF was ≥ 28.2 ng/mL (88% sensitivity, 100% specificity; 0.942 AUC, P <.001) (figure 1). Patients with sST2≥ 28.2 had enhanced LV hypertrophy compared with those with sST2 <28.2 (interventricular septum thickness, P=.022; indexed LV mass, P=.010) (table 1 of the supplementary data). Of note, sST2≥ 28.2 was found only in LGE-positive patients.

Figure 1.

Receiver operating characteristic (ROC) curve and cutoff score for circulating replace by sST2 for the diagnosis of replace by RMF in patients with severe AS. Area under the curve (AUC) for replace by sST2: 0.942 (88% sensitivity, 100% specificity; 0.942 AUC, P <.001).

(0.07MB).

LGE LV mass positively correlated with CMR parameters suggesting a higher LV dilation (LV end-diastolic diameter r=0.3733, P=.001; ESV r=0.5370, P=.002; EDV r=0.3631, P <.001) (figure 2A-C) and hypertrophy (interventricular septum thickness r=0.3514, P=.002; LV mass r=0.4133, P=.001) (figure 2D,E), and with a significantly lowered LVEF (r=−5390, P <.001) (figure 2F). Circulating sST2 levels were associated with higher LV mass (r=0.3977, P=.002) (figure 2G). Importantly, LGE LV mass was strongly correlated with circulating sST2 levels (r=0.7497, P <.001) (figure 2H).

Figure 2.

Late gadolinium enhancement mass correlation with cardiac magnetic resonance parameters and circulating levels of sST2. LGE mass was correlated with LV dilation markers (A-C), LV hypertrophy (D-E), LV dysfunction (F). Circulating sST2 was correlated with LV mass (G), LGE mass (H) and LGE mass only in men (I) when studying each sex group. ESV, end-systolic volume; IVST, interventricular septum thickness; LGE, late gadolinium enhancement; LV, left ventricle; LVEDD, left ventricle end-diastolic diameter; sST2, soluble ST2.

(0.36MB).

We investigated different profiles of RMF, namely midwall fibrosis and subendocardial fibrosis. Multivariate comparisons among the 3 groups are shown in table 2, and the corresponding post hoc analyses can be found in table 2 of the supplementary data. As shown in table 1, 42 (53.2%) patients had RMF, of which 23 (54.8%) patients had midwall fibrosis and 19 (42.4%) had subendocardial fibrosis (table 2). Concomitant CAD and prior anterior myocardial infarction were more frequent in patients with subendocardial fibrosis than in those with midwall fibrosis (78.9% vs. 27.0%, P p=.001; 21.1% vs. 4.3%, P=.008, respectively). Alignment of coronary arterial segments and myocardial perfusion territories was reported in 14 out of 15 patients with CAD and positive LGE subendocardial fibrosis (73.7% of all patients with subendocardial fibrosis). In 1 patient with CAD (5.26%), subendocardial fibrosis was not associated with the ischemic coronary artery, and 4 patients with subendocardial fibrosis did not have previous CAD.

Overall, patients with LGE-positive subendocardial fibrosis had significantly more dilated and worsened LVEF (even <50%) (LV end-diastolic diameter: 50.0 [47.0-55.5] vs. 47.0 [44.0-50.0], P=.019; indexed EDV: 69.0 [56.5-102.0] vs. 67.0 [54.8-71.2], P=.037; indexed ESV: 32.0 [20.0-48.0] vs. 19.0 [10.8-23.0], P=.001; LVEF: 55.6±16.9 vs. 70.4±9.7, P=.001), while patients with midwall fibrosis were more prone to develop LV hypertrophy, as shown by the indexed LV mass (g/m2): 97.0 [87.0-128.0] vs. 84.0 [68.0-93.5], P=.004.

Levels of sST2 were higher in patients with midwall fibrosis than in those with subendocardial fibrosis (38.5 [37.0-40.2] vs. 31.5 [28.2-35.9], P=.006). Nevertherless, no LGE mass differences were found between midwall and subendocardial fibrosis.

LVEF and sST2 levels are indepently associated with RMF in AS patients with preserved LVEF (> 50%)

LV dysfunction was only reported in patients with RMF (table 2). Two groups of severe AS patients with preserved LVEF (> 50%) were distinguished: a) those with no fibrosis (-LGE) and b) with fibrosis (+LGE) (table 3). A tendentiously increased LV hypertrophy (interventricular septum thickness, P=.052; indexed LV mass, P=.052; LGE mass, P=.001; LGE LV mass, P=.001) and significant dilation (LV end-diastolic diameter, P=.022; and indexed ESV, P=.044) were reported in +LGE. That was in parallel with worsened LV function (LVEF % in −LGE vs. +LGE was 68.5 [64.8-77.2] vs. 65.0 [56.0-76.0], P=.001, respectively). Of note, brain natriuretic peptide (BNP) levels were significantly higher in patients with fibrosis than in those without (188 [79.5-283.5] vs. 78.6 [46-219], respectively, P=.025). Strikingly, sST2 levels were significantly higher in patients with LVEF> 50% and concomitant fibrosis than in those without (38.0 [31.8-40.0] vs. 18.9 [16.0-22.7], P<.001), in line with our findings in table 1.

A multivariate linear regression model using backward selection revealed that LVEF and sST2 levels were independedently associated with RMF levels (table 4). The lower the LVEF, the higher the burden of RMF and viceversa. For each 1% increase in LVEF, the RMF was decreased by 0.22 units. Interestingly, regardless of the LVEF, sST2 levels remained associated with RMF: 1 ng/mL sST2 was associated with an increase of 0.17 units of fibrosis. Distribution of residuals of the model were approximately normal, but with increassing variance. Thus, to further confirm the significance of the observed associations, the direction and the relative magnitude, Yeo-Johnson transformations were applied to the dependent variable (RMF) after calculating the lambda value (−0.75) by the maximum likelihood method.22 Modeling the relationship among our transformed dependent variable (RMF) and the independent variables (sST2 and LVEF) revealed no differences to the original multivariate linear regression model (table 3 of the supplementary data). Finally, linear regression assumptions were checked with gvlma (Global Validation of Linear Models Assumptions) package in R.23 Since myocardial fibrosis could be secondary to AS or CAD, we further analyzed our data by multivariate linear regression excluding those patients with myocardial infarction and with CAD (table 4 of the supplementary data). These findings are consistent with our original model (table 4). We finally analyzed the internal validity of our results by bootstrapping, as specified in Methods. The results are displayed in table 5 of the supplementary data and confirm the statistical robustness of our analyses.

Pathological sex-related differences have been proposed to underlie the phenoytpe of AS. Accordingly, our cohort was additionally grouped by sex (table 5). Differences between the sexes for the recorded quantitative parameters were studied using univariate analyses. Smoking and history of atrial fibrillation differed significanlty between the sexes (P <.001 and P=.043, respectively). As expected, AV area was higher in men than in women (P=.009). Nevertheless, gradients (maximum gradient P=0.213; mean gradient P=0.164 and LVEF (P=.377) did not differ between the sexes.

CMR parameteres revealed higher LV dilation and hypertrophy in men than in women, in association with an enhanced prevalence of the overall RMF (62.5% vs. 38.7%, P=.039). Of note, circulating levels of profibrotic sST2 were significantly higher in men than in women (31.8 [22.7-39.3] vs. 21.6 [16.3-30.2], P=.007).

Sex-related differences were observed on comparison of the impact of RMF in LGE-positive and LGE-negative patients within each sex group (table 6). In men, RMF was associated with dilation (LV end-diastolic diameter: 53.0 [47.5-56.8] vs. 47.5 [44.5-51.0], P=.009; indexed ESV: 31.5 [23.0-46.5] vs. 19.0 [11.0-23.0], P=.002) and LV dysfunction (LVEF: 57.0 [48.5-67.0] vs. 71.0 [66.0-82.0], P=.006), with these findings not being reported in LGE-positive vs. LGE-negative women.

RMF was associated with high levels of sST2 regardless of sex (table 6). However, sST2 was correlated with LGE mass (r=0.6604, P <.001) (figure 2I) only in me.

This study has some limitations. First, we cannot gather enough evidence on causal associations due to the inherent limitations of the cross-sectional experimental design of our study. Second, based on previous publications, we confirm that the apparently small sample size of our study is adequate and has a power over 90%. However, it is indeed a limitation to studying the sex-dependent expression of sST2 and its association with CMR parameters and RMF. Third, our cohort represents a real-life population of consecutive patients with severe isolated AS. Nevertheless, a selection bias is highly likely due to our exclusion criteria. For example, excluding AV regurgitation or moderate-to-severe concomitant mitral valve disease, both of which are highly prevalent in degenerative AS, might have increased the percentage of patients with bicuspid AV. Other exclusion criteria such as CMR contraindication might not significantly interfere with the external validity of our results, as in the clinic these patients would not undergo RMF quantification by CMR. In contrast, patients with isolated AS and a contraindication to CMR could clearly benefit from the assessment of circulating sST2. Finally, although we statistically confirmed the adequate sample size and the internal validity of our model, we cannot exclude the possibility that some exclusion criteria may have influenced the unusual prevalence of bicuspid AV in our cohort and may pose a threat to the external validity of our results. Replication studies in larger series will be required to assess sex differences, to validate the causality of the associations reported here, and to circunvent the effect of the high bicuspid AV numbers in our cohort.