Study population

Patients aged ≥ 18 years, diagnosed at the echocardiography laboratories of two French tertiary centres (Amiens and Lille) between 2000 and 2012 with AS (aortic valve area [AVA] ≤ 1.3 cm ² ) and a left ventricular ejection fraction (LVEF) ≥ 50%, and managed medically for at least 3 months after diagnosis, were prospectively identified and included in an electronic database.

We excluded patients with > mild aortic and/or mitral regurgitation; patients with prosthetic valves, congenital heart disease (with the exception of bicuspid aortic valves), supravalvular or subvalvular AS, or dynamic LV outflow tract obstruction; patients with angina, syncope or heart failure symptoms; and patients who denied authorization for research participation.

The present analysis included 331 AS patients who were asymptomatic or minimally symptomatic at the time of diagnosis. Symptoms were ascertained by each patient’s personal cardiologist. We considered as minimally symptomatic patients presenting with atypical chest pain and elderly patients with minimal dyspnoea not clearly related to AS.

The Charlson comorbidity index, summating the patient’s individual comorbidities, was calculated . The diseases recorded for calculation of this index and their corresponding point values were myocardial infarction (1); congestive heart failure (1); cerebrovascular disease with mild or no residual deficit (1); chronic lung disease (1); peripheral vascular disease (1); peptic ulcer disease (1); diabetes mellitus without end-organ damage (1); dementia (1); connective tissue disease (1); mild liver disease (1); hemiplegia (2); diabetes mellitus with organ damage (2); moderate or severe chronic renal impairment (2); solid organ malignancy (2); leukaemia (2); lymphoma (2); moderate or severe chronic liver disease (3); metastatic solid organ malignancy (6); and acquired immunodeficiency syndrome (6). Known coronary artery disease was defined as the presence of documented history of acute coronary syndromes, coronary artery disease previously confirmed by coronary angiography (reduction of the normal diameter by ≥ 50% in the left main coronary artery and by ≥ 70% in the right coronary, left anterior descending and circumflex arteries) or history of coronary revascularization, as previously reported .

We obtained institutional review board authorizations prior to conducting the study. The study was conducted in accordance with institutional policies, national legal requirements and the revised Helsinki Declaration.

Echocardiography

Doppler echocardiographic measurements included LV end-diastolic and end-systolic diameters, LVEF determined by the modified Simpson’s biplane method, LV stroke volume measured in the LV outflow tract by pulsed wave Doppler, aortic peak velocity, mean transvalvular gradient determined by the simplified Bernoulli equation and AVA determined by the continuity equation. Valvuloarterial impedance (Z _va ) was calculated using the following formula, as previously reported : Z _va = [mean transaortic gradient + systolic blood pressure]/stroke volume indexed to body surface area.

LV mass was calculated using the corrected formula of the American Society of Echocardiography, and indexed for body surface area . LV wall thickness and dimensions were estimated, whenever possible, from M-mode imaging or by default from two-dimensional images obtained in the parasternal long-axis view using the leading edge methodology. LV hypertrophy (LVH) was defined as LV mass index > 115 g·m ⁻² in men and > 95 g·m ⁻² in women . Relative wall thickness (RWT) was calculated for assessment of LV geometry using the following formula: (septal + posterior diastolic wall thickness)/LV diastolic diameter . Patients were classified according to four patterns of LV geometry: concentric remodelling; concentric hypertrophy; eccentric hypertrophy; and normal. LVCR was classically defined by RWT > 0.42 in the absence of LVH. Concentric and eccentric hypertrophy were defined when LVH was associated with RWT ≤ and > 0.42, respectively. Normal geometry was considered in the absence of LVH and RWT ≤ 0.42.

Clinical decision and follow-up

After initial medical management, treatment was either conservative or surgical, as deemed appropriate by the patient’s attending physician. Information on follow-up was obtained retrospectively by direct patient interview and clinical examination and/or by repeated follow-up letters, questionnaires, and telephone calls to physicians, patients, and (if necessary) next of kin.

The study endpoint was overall survival after diagnosis starting at baseline echocardiography, and was analysed according to conservative management, and conservative and/or surgical management. Survival analysis with conservative management was assessed until last follow-up with medical management (censored at surgery). Survival with conservative and/or surgical management encompassed medical and surgical management.

Statistical analysis

Continuous variables are expressed as mean ± standard deviation or median (interquartile range); categorical variables are summarized as numbers and percentages. The relationship between baseline continuous variables and the four patterns of LV geometry was explored using one-way analysis of variance. The Shapiro–Wilk test was used to verify whether the residuals obtained on analysis of variance approximated a normal distribution. When this test failed, non-parametric analysis of variance (Kruskal–Wallis test) was used. Pearson’s χ ² statistic or Fisher’s exact test was used to examine the association between the four patterns of LV geometry and baseline categorical variables. The significance between normal LV geometry (referent subgroup) and the other subgroups was examined if there was a significant difference across categories. Post hoc comparisons were performed using either Tukey’s comparison or Mann-Whitney U tests with Bonferroni’s correction for multiple comparisons, as appropriate.

For analysis of outcomes under medical management, data were censored at the time of cardiac surgery (if performed). The entire follow-up was used to analyse outcomes under conservative and/or surgical treatments. The effect of AVR on outcome was analysed as a time-dependent covariate using the entire follow-up.

The Kaplan-Meier method was applied to estimate mortality in the subgroups of patients defined by the four patterns of LV geometry (concentric hypertrophy, concentric remodelling, eccentric hypertrophy and normal). Two-sided log-rank tests were applied to compare mortality in the normal geometry subgroup versus the three other subgroups. Cox proportional hazard models using univariate then multivariable analysis determined all-cause mortality. Risk of death in the three subgroups (concentric hypertrophy, concentric remodelling and eccentric hypertrophy) was estimated versus the normal geometry subgroup. Model-building techniques were not used. Covariates entered in the models were considered of potential prognostic impact on an epidemiological basis; these covariates were age, sex, comorbidity index, hypertension, asymptomatic status at baseline, known coronary artery disease, history of atrial fibrillation, AVA and LVEF. AVR as a time-dependant variable was also added into Cox multivariable models analysing outcomes under conservative and/or surgical treatments. As coronary angiography was not performed in every patient, and as distinction between asymptomatic and minimally symptomatic patients in an elderly population is challenging, multivariable models without these two variables were also performed. For continuous variables, the assumption of linearity was assessed by plotting martingale residuals against independent variables. The proportional hazards assumption was confirmed using statistics and graphs based on the Schoenfeld residuals. A P -value < 0.05 was considered statistically significant. Statistical analyses and figures were obtained using PASW 18.0 (IBM, Inc., Bois-Colombes, France), R-3.0.3 (R Foundation for Statistical Computing, Vienna, Austria) and GraphPad Prism (GraphPad Software, La Jolia, CA, USA).