Between 2008 and 2020 all consecutive patients aged ≥18 years with symptomatic severe AS treated with TAVI at a single university center were prospectively included in a dedicated TAVI database. Patients with incomplete echocardiographic data, normal flow low gradient AS or aortic regurgitation predominance were then excluded from analysis.

The decision to perform TAVI was discussed by a heart team, considering both surgical AVR and TAVI risk. Patient evaluation included medical history, physical examination, 12-lead electrocardiography, blood analysis, transthoracic echocardiography (TTE), coronary angiography and cardiac, thoracic, abdominal and pelvic computed tomography (CT) with contrast.

Echocardiographic measurements were obtained pre-TAVI and at hospital discharge. All images and measurements were acquired from the standard views and digitally stored for offline analysis. Left ventricle ejection fraction was calculated using the Simpson's Rule. Aortic valve (AV) velocity and gradients were assessed with Doppler TTE using the Bernoulli's principle. Stroke volume (SV) was measured using the velocity time integral in the LV outflow tract with pulsed-wave Doppler; then aortic valve area (AVA) was calculated by means of the continuity equation.

The diagnosis of severe AS was made according to the guidelines if:6,20

• -

  AVA <1 cm2 or <0.6 cm2/m2 and a mean AV gradient ≥40 mmHg or peak velocity jet ≥4m/s (HG AS)

• -

  AVA <1 cm2 or <0.6 cm2/m2 and a mean AV gradient <40 mmHg (LF-LG AS) if the following criteria were met:

  • -

    Classic LF-LG (cLF-LG) when LVEF <50% and:

    • -

      Dobutamine stress echocardiography with contractile reserve (increase ≥20% in SV) and a mean AV gradient increase for ≥40 mmHg maintaining an AVA <1 cm2 or

    • -

      AV score calcium >2000 Agatston Units (AU) in men or >1300 AU in women on CT.21

  • -

    Paradoxical LF-LG (pLF-LG) when LVEF ≥50% and SV <35 mL/m2, if high AV score calcium as defined for cLF-LG.

The anesthetic technique was individualized, and the procedure was performed under general anesthesia, continuous sedation or local anesthesia with no sedation. Transesophageal echocardiography guidance was not routinely used. The access route (transfemoral, trans-subclavian, transaortic or transcava) was selected according to the results of the CT angiography.

After the procedure all patients were admitted to an intensive care unit for at least 24 hours. Data on procedural success and periprocedural complications were collected for each patient according to the Valve Academic Research Consortium 2 (VARC-2) criteria.22

Patients were treated with antithrombotic therapy, consisting of: dual antiplatelet therapy (acetylsalicylic acid plus clopidogrel) for one to three months, followed by simple antiplatelet therapy; or simple antiplatelet therapy if high bleeding risk; or oral anticoagulation if there was another clinical indication.

Clinical follow-up and post discharge events were analyzed at clinical visits, through phone contact and assessing medical records. Patients were followed up at one, six and 12 months at a TAVI outpatient clinic at our center, and at least yearly thereafter by an attending cardiologist. All clinical events were defined according to the VARC-2 criteria.22

Continuous variables were reported as the mean (standard deviation) and were compared using the two-sample t-test. Categorical variables were expressed as frequencies and percentages and were compared using the chi-squared or Fisher's exact test, as appropriate.

Univariate Cox proportional hazard regression model with corresponding 95% confidence intervals was performed to assess the effect of LF-LG AS on multiple outcomes and a multivariate Cox regression model was used to assess the same effect with adjustment for differences in baseline characteristics.

A propensity score matching analysis was performed to control for the possible bias induced by the heterogenicity among the groups’ baseline characteristics, which may influence the occurrence of LF-LG AS. The propensity scores were estimated for each patient using a multivariable logistic regression and patients were then matched 2:1 using the nearest neighbor method and similarity between baseline characteristics in this matched population confirmed. Then the effect of LF-LG AS on the outcomes was assessed using the Cox regression model.

A sub-analysis of the LF-LG AS group to compare outcomes in reduced ejection fraction (rEF), LVEF<40% vs. preserved or mildly reduced ejection fraction (pEF) LVEF≥40%, considering the cut-off for LVEF used in heart failure (HF) guidelines,23 was employed with the same statistical methods.

A two-sided p value <0.05 was considered significant. Statistical analysis was performed with SPSS for Windows version 26 (IBM Corp, Armonk, New York).

Baseline demographic, echocardiographic, tomographic, laboratorial and procedural characteristics are shown in Table 1. Of 81 patients (16.9%) with LF-LG AS, 39 had pLF-LG and 42 had cLF-LG. A dobutamine stress echocardiography was performed in five of them, all with contractile reserve and mean gradient rising to ≥40 mmHg). Patients with LF-LG AS had worse baseline characteristics, a higher Society of Thoracic Surgeons score and New Euroscore II risk, higher NT pro natriuretic peptide B, more frequent LVEF <40% and tricuspid annular plane systolic excursion <17 mm, coronary artery disease (including previous myocardial infarction and coronary artery bypass graft), poor vascular accesses and periprocedural angioplasty. Periprocedural complications were similar between groups (14.8% in LF-LG and 16.3% in HG, p=0.741). After the procedure, 56.4% of patients with initial rEF saw an improvement to ≥40%, but a significant difference was still found in mean LVEF pre-hospital discharge (47% in LF-LG vs. 55% in HG patients, p<0.0001) between groups.

Patients were followed for a mean period of 21±21 months (minimum 0 and maximum 112 months). In univariable analysis (Table 2), LF-LG AS was associated with worse one year mortality, one year and long-term functional class, and one year and long-term heart HF hospitalizations. When adjusted to the differences in baseline characteristics, in a Cox regression multivariable analysis (Table 3), LF-LG AS was still associated with worse functional class at one year (Figure 1) and in the long-term, and with one year (Figure 2) and long-term HF hospitalizations.

Figure 1.

Multivariate Cox regression analysis for functional class at one year in low flow low gradient versus high gradient aortic stenosis.

(0.1MB).

Figure 2.

Multivariate Cox regression analysis for heart failure hospitalizations at 1-year in low flow low gradient versus high gradient aortic stenosis.

(0.09MB).

After propensity score matching (121 HG and 79 LF-LG AS patients), the outcomes remained very similar, with worse functional class at one year (p=0.033) and long-term (p=0.018), and with one year (p=0.032) and long-term (p=0.005) HF hospitalizations.

In a first sub-analysis considering only patients with LF-LG, despite significant differences in baseline characteristics in cLF-LG and pLF-LG, there were no differences when outcomes (mortality, HF hospitalizations and functional class) were compared between these two sub-groups.

However, when patients were sub-divided considering LVEF ≥ or <40% (pEF or rEF) significant differences were found. Patients with rEF were younger and had worse baseline characteristics, with higher risk scores and natriuretic peptides. They were more commonly smokers and had peripheral artery disease more frequently (Table 4). There were no differences in periprocedural complication rates (22.2% in rEF versus 11.1% in pEF, p=0.185). In the sub-group of patients with rEF, there was an improvement in mean LVEF from 28% to 34%, but it remained inferior to pEF patients (53%, p<0.0001). About one third of rEF patients recovered LVEF≥40%. Despite similar procedural success and younger age, after multivariable Cox regression analysis (Tables 5 and 6), adjusting for the different baseline characteristics, there were significant differences in global (Figure 3) and cardiovascular mortality.

Figure 3.

Multivariate Cox regression analysis for mortality in preserved ejection fraction versus reduced ejection fraction low flow low gradient aortic stenosis.

(0.08MB).

When outcomes were compared among the three groups of patients (HG as control, pEF LF-LG and rEF LF-LG), in a multivariable Cox regression analysis (Table 7), there were significant differences in the composite end-point of death and HF hospitalizations (Figure 4), mostly due to HF hospitalizations which were significantly different among the three groups (Figure 5), while global and cardiovascular mortality were only significantly different in the rEF LF-LG sub-group (Figure 6).

Figure 4.

Multivariate Cox regression analysis for death+heart failure hospitalizations in high gradient versus preserved ejection fraction low flow low gradient and reduced ejection fraction low flow low gradient aortic stenosis. pEF: preserved ejection fraction; rEF: reduced ejection fraction; LF-LG: low flow low gradient; HG: high gradient; HR: hazard ratio.

(0.13MB).

Figure 5.

Multivariate Cox regression analysis for heart failure hospitalizations in high gradient versus preserved ejection fraction low flow low gradient and reduced ejection fraction low flow low gradient aortic stenosis.

(0.12MB).

Figure 6.

Multivariate Cox regression analysis for cardiovascular mortality in high gradient versus preserved ejection fraction low flow low gradient and reduced ejection fraction low flow low gradient aortic stenosis.

(0.11MB).