Heterogeneity of left ventricular remodelling patterns across severe aortic stenosis phenotypes

Background

Severe aortic stenosis (AS) imposes chronic pressure overload on the left ventricle (LV), triggering adaptive hypertrophy to preserve cardiac output. Over time, these compensatory changes may progress to maladaptive remodelling, contributing to ventricular dysfunction and adverse outcomes. LV structural adaptation to pressure overload is highly variable, and the mechanisms underlying this heterogeneity remain incompletely understood.

Purpose

To evaluate LV geometry across severe AS phenotypes and examined its association with prognosis.

Methods

Patients with severe AS diagnosed between 2012 and 2020 were retrospectively identified and classified into 3 phenotypes: high-gradient (HG) (mean gradient [MG] ≥40mmHg, aortic valve area [AVA] ≤1.0 cm²), classical low-flow, low-gradient (c-LFLG) (AVA≤1.0, MG<40, stroke volume index [SVi] ≤35 mL/m², LVEF <50%), and paradoxical LFLG (p-LFLG) (AVA≤1.0, MG<40, SVi≤35, LVEF≥50).

LV geometric patterns were defined according to standard cut-offs for relative wall thickness (RWT) and LV mass index (LVMI): normal geometry (normal RWT/normal LVMI), concentric remodelling (increased RWT/normal LVMI), concentric hypertrophy (CH) (increased RWT/increased LVMI), and eccentric hypertrophy (EH) (normal RWT/increased LVMI).

The association between LV geometry and mortality was assessed using a multivariate logistic regression.

Results

A total of 569 echocardiograms from 380 patients (53.2% females, mean age 77.8 ±8.4 years) were analysed: 341 (59.9%) had HG, 118 (20,7%) c-LFLG, and 110 (19,3%) p-LFLG AS.

p-LFLG patients were more often females (66% vs 54% in HG and 42% in c-LFLG; p=0.012); c-LFLG patients had higher prevalence of coronary artery disease (43% vs 22% in HG and 25% in p-LFLG; p<0.001).

As expected, SVi and MG were lower in both LFLG phenotypes compared with HG AS, while LVEF was lower in c-LFLG. AVA was similar across groups. c-LFLG patients had larger indexed LV end-diastolic volumes (72±24.5mL/m² vs 57±15.6 in HG and 49±14.3 in p-LFLG) and a higher prevalence of >grade II mitral regurgitation (62% vs 34% in HG and 41% in p-LFLG).

LV geometric patterns differed significantly across AS phenotypes (χ², p<0.001). CH predominated in both HG and p-LFLG (76.8% and 67.3%, respectively), whereas c-LFLG showed a more heterogeneous distribution, with higher prevalence of EH (44.6%)

In multivariate analysis, LV geometry was not independently associated with mortality (p=0.192).

Conclusion

LV geometry varied across AS phenotypes, with CH predominating in HG and p-LFLG, and a higher prevalence of EH in c-LFLG AS. In this cohort, LV geometric pattern was not associated with mortality.