Cumulative survival curves for patients with and without residual impairment of functional capacity were constructed using the Kaplan–Meier method and compared with the log-rank test and Cox-proportional hazards model.

The baseline, periprocedural, and follow-up characteristics in the patients stratified according to the functional status at follow-up are summarized in a Supplementary Table. All relevant baseline and periprocedural factors were tested for association with residual impairment of functional capacity after TAVI. Table 1 summarizes the univariable and multivariable predictors.

On multivariable logistic regression analysis, atrial fibrillation/flutter [odds ratio-OR, 2.08 (1.21–3.58), p = 0.008], low-flow–low-gradient AS [OR, 1.97 (1.09–3.57), p = 0.026], chronic obstructive pulmonary disease [OR, 1.92 (1.19–3.12), p = 0.008], and lower hemoglobin [OR, 1.11 (1.01–1.21) for each g% decrement, p = 0.036] were independently associated with residual impairment of functional capacity after TAVI. Although device failure (mainly driven by a higher trans-prosthetic valve pressure gradient-Supplementary Table) was associated with residual impairment of functional capacity in univariable analysis [OR, 1.73 (1.04–2.88), p = 0.034], it was not an independent predictor in the multivariable analysis.

Echocardiographic follow-up was available in 532 patients and was performed at a median interval of 366 (161–736) days after TAVI. As shown in Fig. 1, apart from left ventricular mass index (LVMi) which improved significantly in both groups with no between-group difference at follow-up, reverse cardiac remodeling was less efficient in patients with residual impairment of functional capacity.

LV diastolic diameter (LVDD), although similar at baseline, was significantly larger in those with residual impairment of functional capacity at follow-up (52.8 ± 9.7 vs. 50.4 ± 8.8, p = 0.008). Impaired LV ejection fraction (LVEF <50%) was significantly less at follow-up as compared to baseline only in those with NYHA I symptoms. Accordingly, although not significantly different at baseline, the prevalence of impaired LVEF at latest follow-up was significantly higher in those with residual impairment of functional capacity (26.4 vs. 16.2%, p = 0.007). Mitral regurgitation (MR) severity was similar in both groups at baseline and significantly improved at latest follow-up in those with NYHA I symptoms (moderate–severe MR, 13% at follow-up vs. 18% at baseline, p = 0.005) but not in those with residual impairment of functional capacity (moderate–severe MR, 26% at follow-up vs. 23% at baseline, p = 0.44). Consequently, MR severity at follow-up was significantly higher in those with residual impairment of functional capacity than in those with NYHA I symptoms (moderate–severe MR, 26% vs. 13%, p < 0.001).

During the entire follow-up period, all-cause mortality was significantly higher in those with residual impairment of functional capacity than in those who recovered a normal functional status (32.6 vs. 12.7%, log-rank p < 0.001) (Fig. 2a). Similarly, cardiac mortality was significantly higher in those with impaired functional capacity (14.4 vs. 5.4%, p < 0.001) (Fig. 2b). After adjustment for the aforementioned LV remodeling markers (LVEF, LVDD, and MR at latest follow-up), the association between residual impairment of functional capacity and all-cause mortality [hazard ratio-HR: 2.37 (95% CI: 1.51–3.72), p < 0.001] and cardiac mortality [HR: 2.16 (95% CI: 1.08–4.35), p = 0.030] remained significant.

To explore whether residual mild impairment of functional capacity (NYHA II) after TAVI can be detrimental, survival analysis was repeated after dividing the patients into three groups; normal (NYHA I), mildly-impaired (NYHA II), and moderate–severely impaired functional capacity (NYHA III or IV). All-cause mortality was higher in those with mild impairment (26.4%) than in those who recovered a normal functional capacity [12.7%, log-rank p < 0.001, HR: 2.02 (95% CI: 1.10–3.72), p = 0.023]. Cardiac mortality was also higher in those with mild impairment (10.4%) than in those who recovered a normal functional capacity [5.4%, log-rank p = 0.021, HR: 2.08 (95% CI: 1.42–3.07), p < 0.001]. In patients who had a residual moderate–severe functional impairment, mortality was very high (53.7%, with 27.8% being cardiac) and was significantly higher than those with mild impairment [all-cause mortality: log-rank p < 0.001, HR: 2.57 (95% CI: 1.60–4.11), p < 0.001; cardiac death: log-rank p < 0.001, HR: 3.31 (95% CI: 1.68–6.53), p = 0.001]. Survival curves for the three groups are displayed in Fig. 3.

The dismal prognosis of symptomatic severe AS if managed conservatively drew the interest to developing prognosis-modifying strategies. TAVI emerged as a prognosis-modifying intervention with an un-equivocal mortality benefit compared to conservative management in patients who cannot undergo surgery [17] and compared to surgical management in high-risk [8] and intermediate-risk [6, 18] patients. However, physicians’ and patients’ appraisal of risks and benefits may differ [19], and symptomatic relief is, for some patients, a priority. In a study by Hussain et al., the majority of patients undergoing SAVR for severe AS were willing to accept considerably higher risk of perioperative death than what is considered by physicians/guidelines as “acceptable” [20]. Patients who had more severe symptoms and lower quality of life as well as those with pulmonary disease, impaired LVEF, or lower transaortic pressure gradient were more likely to accept a high/prohibitive risk of perioperative death if a normal health is likely to be restored after valve replacement [20]. These results emphasize the importance of symptomatic improvement (vs. mere survival) among the priorities of AS patients especially those with more severe symptoms.

In the present study, two cardiac (atrial fibrillation and low-flow–low-gradient AS) and two non-cardiac (COPD and anemia) baseline clinical conditions were identified as independent predictors of impaired functional status after TAVI. The benefit of identifying these markers during the decision-making process prior to TAVI is twofold: (1) to predict the functional outcome and counsel the patient in light of the lower probability of restoring a normal functional capacity and (2) to stimulate correction of these conditions when possible knowing that failure to control these conditions will impair the functional gains from TAVI. Although not among the independent predictors, two markers of cardiac hemodynamics (transaortic valve PG and brain natriuretic peptide) seem to have an added value to the LVEF in predicting functional status after TAVI (Supplementary Table). This finding is in line with previous studies that concluded that the indices of LV mechanics other than the volumetric LVEF (e.g., longitudinal strain [21]) as well as markers of elevated LV pressure (e.g., brain natriuretic peptide [22]) are crucial in predicting the functional status in patients with severe AS. In fact, a “low-flow” status leading to a low-gradient severe AS reflects the combination of a small LV cavity, a severe diastolic dysfunction, and an impaired longitudinal contractility. The combination of both low transvalvular gradient and low ejection fraction portends significantly worse outcomes [23, 24]. These data together might explain why the mere reduction of LVEF at baseline was not an independent predictor of the functional outcomes after TAVI, while the combination of reduced LVEF and relatively low transaortic valve PG was.

Many attempts have been made to improve the predictability of TAVI outcomes, including the development of specific TAVI outcome-prediction scores. Although the inclusion of frailty and functional parameters into the predictive models has improved their performance as compared to surgical risk models, the accuracy of those models remain modest [16]. The complexity of the cardiovascular morbidity in patients with severe degenerative AS probably plays an important role in this suboptimal performance of predictive models.

Reduced arterial compliance is an important contributor to the increased afterload and to the adverse cardiac remodeling in AS patients [25]. This arterial component of the AS disease complex is likely even more pronounced in TAVI candidates, who are typically older with multiple risk factors for atherosclerosis, than SAVR candidates. In AS patients referred for valve replacement, a higher arterial stiffness correlates with less LV mass regression and with more adverse cardiac events after SAVR [26] and TAVI [27]. Yotti et al. [4] studied arterial function before and after TAVI and reported an increase in arterial load after the procedure resulting in a residual elevation of LV pressure in 70% of patients. Moreover, myocardial response to AS involves variable degrees of myocardial fibrosis [28], the extent of which correlates with NYHA functional class at baseline [29], and predicts the improvement in NYHA class after valve replacement [29]. Arterial stiffness and myocardial fibrosis are two examples of important contributors to the impaired cardiac performance in AS patients that might attenuate the benefit from TAVI. Therefore, the classic screening of patient’s symptoms, comorbidities and valvular/myocardial function might not reflect the complete spectrum of the actual patient morbidity.

This study has a number of limitations. Echocardiographic data were reported by the treating centers without independent core lab adjudication and follow-up echocardiographic data were missing in some cases.

The list of predictors of functional recovery after TAVI that has been investigated in the present analysis included valvular and cardiac function, as well as major comorbidities. However, markers of frailty and surrogates for arterial function and myocardial fibrosis were not included in our analysis. In addition, the socioeconomic and educational status of the patient as well as the involvement in regular exercise or rehabilitation programs might also play a role in determining the functional outcome of these patients. For future studies, we suggest to study the relation of those factors to the functional recovery after TAVI.

Clinicians assign a given patient to an NYHA class on the basis of their subjective interpretation of reported symptoms, and accordingly, interobserver variability of the functional assessment is a potential downside of this classification. In spite of this limitation, a higher NYHA class was shown in the present study to be a marker of objective adverse cardiac remodeling and, more importantly, of a higher mortality risk. It turns out that, in spite of its limitations, this simple tool that is still used to decide the timing of intervention (as recommended by clinical practice guidelines) can still be crucial in post-TAVI clinical assessment. Other more objective, more quantitative, and purely patient-reported multidimensional assessment tools have been suggested as better indices of the quality of life. These multidimensional tools (e.g., EuroQol-5L and SF-36 questionnaires) involve dimensions that are more determined by the extra-cardiac morbidities and general frailty (e.g., pain/discomfort, anxiety/depression, and independent self-care). TAVI that effectively relieves AS and its relevant symptoms and improves survival cannot reverse non-cardiac pathologies that profoundly impacts on the patient’s overall quality of life. Previous studies revealed that the general health status visual analog scale improves after TAVI by only 2.7–7.0% (mainly driven by improvements in mobility and usual activity dimensions, while the other dimensions showed only very modest change) [30] and that the EQ-5D index also shows a modest improvement (+7% at 1 year) [31].