Effects of prosthesis-patient mismatch on cardiac function and clinical outcome after transcatheter aortic valve implantation

Patients with severe symptomatic AS scheduled to undergo transfemoral TAVI in our center were prospectively enrolled between September 2018 and May 2020. Selection criteria included: age>40, severe AS (aortic valve area, AVA) <1.0 cm², indexed AVA<0.6 cm²/m², peak aortic jet velocity ≥4 m/s, or mean gradient ≥40 mmHg. Exclusion criteria included: hypertrophic cardiomyopathy, prosthetic aortic valve, non-transfemoral TAVI, and poor acoustic window. All patients underwent Heart Team evaluation and were deemed eligible for TAVI based on current guideline recommendations. All procedures were performed in a hybrid operating room with the participation of both an interventional cardiologist and a cardiovascular surgeon.

All patients underwent percutaneous transfemoral TAVI with balloon-expandable valves. Transesophageal echocardiography was used during the procedure for additional guidance and assessment. Clinical, biological, and procedural data were collected. The primary outcome was all-cause mortality (obtained through a query of the National Register of population records), registered three years after TAVI. Informed consent was obtained from each patient. The study protocol, which conforms to the ethical guidelines of the 1975 Declaration of Helsinki, was reviewed and approved by the institutional Ethics Committee.

All patients underwent a comprehensive echocardiogram performed by experienced echocardiographers both before and 30 days after TAVI using a Vivid E95 ultrasound system (General Electric Healthcare, Horten, Norway), as previously described [13]. Data were digitally stored for offline analysis using commercially available software (EchoPac version 203; GE Medical Systems, Horten, Norway), and images were analyzed by a single trained cardiologist according to current guidelines [14]. The evaluation included standard parameters used to assess AS severity: peak aortic jet velocity, peak and mean pressure gradients across the aortic valve (using modified Bernoulli equation), and AVA (using continuity equation). In the parasternal long-axis view, LV dimensions were assessed, and LV mass was calculated using Devereux’s formula and indexed to body surface area [14]. LV end-diastolic and end-systolic volumes were measured in the apical 4-chamber and 2-chamber views and indexed to body surface area [14]. LV ejection fraction (EF) was calculated according to the Simpson’s biplane method [14]. Left atrial volumes were measured by the biplane method of disks and indexed for body surface area [14]. Transmitral flow was assessed by PW Doppler to measure the peak early (E) and late (A) diastolic velocities, and tissue Doppler imaging of the mitral annulus on the apical 4-chamber view was used to measure the e’ velocities at both the lateral and septal sites to calculate the E/e’ ratio.

Myocardial deformation analysis using speckle tracking echocardiography (STE) was performed to assess LV and LA function. The evaluation included STE analysis for LV function: LV global longitudinal strain (GLS); LA function: LA global longitudinal strain (LAε, reservoir function), LA systolic strain rate (SSr, reservoir function), LA early diastolic strain rate (ESr, conduit function), LA late diastolic strain rate (ASr, contractile function). Complete myocardial deformation analysis was possible in 132 out of 160 patients. Negative values of strain parameters are used as moduli (positive numbers) for ease of analysis. RV function was assessed by measuring TAPSE, the peak systolic myocardial velocity at the lateral site of the tricuspid annulus (S’RV), RV fractional area change (FAC) and RV longitudinal strain parameters by STE: peak values of global RV strain (RV-GLS), RV free wall longitudinal strain (RV-FWLS) and the interventricular septum longitudinal strain (RV-IVS). The right ventricular systolic pressure was calculated from the peak velocity of the tricuspid regurgitant jet using the Bernoulli equation, and the right atrial pressure (determined by the diameter and inspiratory reduction of the inferior vena cava) was added [15].

Patient-prosthesis mismatch (PPM) was classified based on EOAi measured at 30 days post-procedure by transthoracic echocardiography using the continuity equation in line with the current guidelines [15]. Additionally, PPM BMI-adjusted was reclassified according to lower EOAi values for obese patients (BMI≥30 kg/m²) as stated in the guidelines [15]. Patient-prosthesis mismatch (PPM) was classified as follows: No-PPM – EOAi>0.85cm²/m², PPM - EOAi≤0.85cm²/m², further divided in moderate PPM – EOAi>0.65cm²/m² and ≤0.85cm²/m², and severe PPM – EOAi≤0.65cm²/m².

Additionally, PPM BMI-adjusted was reclassified according to lower EOAi values for obese patients (BMI≥30kg/m²) as follows: No-PPM BMI-adjusted - EOAi>0.70cm²/m², PPM BMI-adjusted – EOAi≤0.70cm²/m², further divided in moderate PPM BMI-adjusted – EOAi>0.55cm²/m² and ≤0.70cm²/m², and severe PPM BMI-adjusted - EOAi≤0.55cm²/m².

Patients were grouped according to measured PPM (PPM_M) into PPM_M group: EOAi≤0.85 cm²/m² (n=64) and No-PPM_M group: EOAi>0.85 cm²/m² (n=96). The PPM_M group was further divided into the predicted PPM_M group (n=39) and unpredicted PPM_M group (n=25), according to valve-size predicted PPM, using EOA based on normal reference values of size and model for published data [16]. Predictors and predictive value of PPM_M, predicted PPM_M, and unpredicted PPM_M were analyzed based on the primary endpoint, 3-year all-cause mortality. Analysis of PPM_M adjusted for BMI was also performed revealing the same significant results as in the presented analysis.

Continuous variables are given as mean ± standard deviation and compared using the Student t-test. Discrete variables were expressed as counts and percentages, and comparisons between groups were done with the χ² or Fisher’s exact test, when appropriate. For comparisons between subgroups, the Kruskal-Wallis test, Wilcoxon rank sum tests using pairwise comparisons, and the Chi-square test for comparing proportions (of categorical variables), between >2 groups have been used. Bonferroni method was used to adjust p-values for multiple comparisons. Long-term clinical outcomes were estimated using the Kaplan-Meier method, with comparisons made using the log-rank test (overall or pair wise as appropriate). Univariable analysis (linear and binary logistic) was used to identify potential predictors of patient prosthesis-mismatch. After careful selection of variables based on clinical judgment, univariable assessment (p<0.05), exclusion of variables showing collinearity (Pearson’s coefficient >0.6), and multiple testing to ensure stability, a multivariable model has been fitted (by stepwise multivariable regression analysis, linear and binary logistic). Univariable predictors of all-cause mortality were determined using Cox proportional hazards (Enter). Multivariable analysis was also performed in a similar fashion (Forward Wald). A two-sided P-value of 0.05 was considered statistically significant for all tests. All analyses were conducted using SPSS 21.0 (SPSS, Inc., Chicago, IL, USA).

According to measured EOAi of 160 patients during the 30-day follow-up post-TAVI, 94 patients did not develop PPM_M (60%), while 47 (29.4%) and 17 (10.6%) patients had moderate and severe PPM_M, respectively, yielding a total of 64 patients with PPM_M (40%). After adjustment for BMI thresholds, only 50 patients had PPM_M (31.3%), of which 34 (21.3%) with moderate and 16 (10.1%) with severe PPM_M.

Except for higher incidence of females (57.8% vs. 39.6%, p=0.024) and higher BMI (28.5±5.0 vs. 26.9±4.5, p=0.033) in the PPM_M group, no significant differences in age (76.3±7.8 vs. 76.6±7.1 years, p=0.8), body surface area (p=0.1), prevalence of obesity (p=0.2), nor in the presence of cardiovascular risk factors (p>0.2 for all) were observed between groups (Table 1).

Based on preoperative echocardiographic evaluation, the patients in the PPM group had smaller AVA and AVAi, with no difference in mean gradient or LVOT diameter between groups. Impaired LA function (LA-GLS, SSr, ASr) was more frequent in the PPM group, despite similar LAVI between groups (Appendix 1). According to preoperative angio-CT evaluation, the patients in the PPM group had smaller aortic roots (smaller aortic annulus, smaller LVOT, lower LCA and RCA origin with smaller sinuses), but no differences in calcium load or calcium disposition, nor in the presence of bicuspid valve (Appendix 2). There were no differences between groups regarding intraprocedural features, except for a higher rate use of 20/23 valves in the PPM group (Appendix 3).

All echocardiographic parameters describing AS severity, LA reservoir, contractile function, LV systolic and diastolic function improved significantly after the procedure as previously presented[13], but also in PPM_M and No-PPM_M groups, respectively (Table 2).

At 1 month after TAVI, there were no significant differences in terms of indexed LV mass, LV volumes, LVEF, and LV global longitudinal strain (p=NS) between patients with or without PPM (Table 3). Post-TAVI prevalence of paravalvular aortic regurgitation or mitral regurgitation was also not different between the two groups (p=0.96, p=0.12, respectively). Although LA indexed area and volume were similar between groups, in the PPM_M group we found a significantly worse systolic global LA strain (13.7±6.1 vs. 16.9±7.4%, p=0.005) and LA contractile function (ASr: -1.0±0.6 vs. -1.3±0.6 in the No-PPM_M group, p=0.013).

Subgroup analysis of patients with PPM according to predicted vs. unpredicted PPM_M revealed larger LV volumes, significant mitral regurgitation, and larger aortic root in the unpredicted PPM_M group (Appendix 4). Complete data regarding preoperative CT evaluation and procedural characteristics are available in the appendix (Appendix 5 and 6). After TAVI, unpredicted PPM_M was associated with worse subclinical LV dysfunction and larger LV volumes (Appendix 7).

Two separate analyses were performed: one based on preoperative characteristics to determine factors associated with the occurrence of PPM_M, and one based on postoperative characteristics, to analyze the relationship between PPM_M and cardiac function after TAVI. In the multivariable regression model (Table 4), higher BMI (per unit increase OR 1.01, p=0.012) and smaller valve number (OR 0.79, p=0.001) were independent predictors for PPM_M before the procedure, while LA-GLS (OR 0.92, p =0.022) was independently associated with PPM_M after the procedure. There was no significant association between PPM_M and age, BSA, obesity, aortic valve calcium score, LV mass index, oversizing, or post-dilatation. After TAVI, LA-GLS (OR 0.93, p=0.014) was independently associated with unpredicted PPM_M, while LV-GLS (OR 0.81, p=0.015) was independently associated with predicted PPM_M.

3-year mortality was higher in the PPM_M group (23.4% vs. 10.4% in the No-PPM_M group, p=0.026), even after BMI-adjustment (25.5% vs. 11.2%, p=0.019). Reclassification of PPM_M group into No-PPM_M, moderate PPM_M, and severe PPM_M showed a trend for higher mortality in both moderate and severe PPM_M as compared to No-PPM_M but no overall significant difference (23.4% and 23.5% vs. 10.4%, p=0.104). Valve-predicted PPM underestimates the incidence of measured PPM_M (27.5% vs. 40.0% PPM_M, p=0.011) and is not predictive for mortality (HR=0.67, p=0.42; AUC=477, p=0.71). Reclassification of PPM_M group into predicted PPM_M and unpredicted PPM_M, according to valve-size predicted PPM, demonstrated a progressively unfavorable prognosis from No-PPM_M, to predicted and unpredicted PPM_M in terms of all-cause mortality at 3 years follow-up (10.4%, 16.0% and 28.2%, respectively; p=0.036). Kaplan-Meier survival analysis depicts event-free survival during follow-up, with differences between groups becoming apparent after 1-year follow-up and achieving statistical significance at 3-year follow-up (Figure 1).

Figure 1

A. Kaplan-Meier survival plot illustrating cumulative survival in patients with PPM_M vs. No-PPM_M; B. Kaplan-Meier survival plot illustrating cumulative survival in patients with moderate PPM_M vs. severe PPM_M vs. No-PPM_M; C. Kaplan-Meier survival plot illustrating cumulative survival in patients with predicted PPM_M vs. unpredicted PPM_M vs. No-PPM_M.

None of the baseline characteristics were predictive for mortality in the univariate analysis, therefore were not included. Procedural and post-TAVI echocardiographic parameters were further analyzed, together with various definitions of PPM. While increased transvalvular aortic mean gradient, impaired LA-GLS, and LV diastolic dysfunction were associated with mortality in the univariate analysis, only the presence of PPM_M was independently associated with mortality (Table 5).

In the present study, we found an incidence of PPM_M of 40%, of which 47 (29.4%) and 17 (10.6%) patients had moderate and severe PPM_M, respectively. Adjustment for BMI thresholds decreased the incidence of PPM_M to 31.3%, with a similar proportion of severe PPM_M (10.1%). The reported incidence of PPM after TAVI varies between 6 - 46% for moderate PPM and 1-15% for severe PPM [12,17–19], according to the definition used [8,20,21], the type of valve [22–24] and positioning [25,26]. The relatively high incidence of PPM_M in our study may be related to the exclusive use of balloonexpandable valves, which have higher rates of PPM compared to self-expandable valves, especially in small valves [25,27]. Moreover, use of newer balloon-expandable valves has been associated with lower paraprosthetic leak rate, but higher PPM rate compared to prior generation, due to flared inflow stent-frame morphology [28,29]. Concerns about accurate EOA measurement have led to comprehensive research using multimodality imaging and different parametric approaches to efficiently establish PPM [17,30,31]. Studies have shown a lower rate of predicted PPM based on implanted valve size or CT-measured LVOT area, which was consistent with our results and led to separate analyses of predicted and unpredicted PPM_M [8,20]. A potential cause of PPM_M overestimation may be the pressure recovery phenomenon, which can be responsible for this EOA discrepancy in relation to the patient’s anatomy, especially in a small aortic root [21,30,32]. A recent study suggests that this effect may be more important in TAVI than in SAVR and with balloon-expandable valves than with self-expandable valves [31].

Previous studies have shown that preoperative prediction of PPM in TAVI patients based on valve type, annular size, and BSA varies from measured PPM_M, rendering EOAi charts utility to predict actual PPM_M rather limited[8]. Our results are consistent with previously published data and add information regarding patients with unpredicted PPM_M.

Based on perioperative characteristics, PPM_M was predicted by BMI (higher risk per unit increase in BMI) and valve size (larger valve size protective for PPM_M). While predicted PPM_M was independently associated with BMI and valve size in a similar fashion, unpredicted PPM_M was independently associated with post-dilation (protective effect). Previous studies have shown reduced rates of moderate or severe prosthesis-patient mismatch after post-dilation, with no evidence for short-term structural deterioration of the balloonexpandable transcatheter valve [33]. Although post-dilation was associated with higher rates of early stroke, there was no significant association between post-dilation and mortality [33]. Oversizing also had a protective effect for unpredicted PPM_M but failed to be significant in the multivariate analysis. This may suggest a certain degree of under expansion, despite larger aortic root and valve size in the unpredicted PPM_M group vs. predicted PPM_M group.

In-depth analysis of postprocedural parameters revealed that significant PPM_M is associated with impaired LA function, specifically impaired left atrial reservoir function. Previous published studies suggest that LA strain was associated with a significant stepwise decrease in all stages of diastolic dysfunction, supporting LA dysfunction as a potential marker of diastolic dysfunction [34]. Furthermore, LA strain has been proven useful in the non-invasive assessment of left ventricular filling pressures [35]. In the subgroup analysis, unpredicted PPM_M was associated with impaired LA reservoir function (lower LA-GLS values in the presence of PPM_M), while predicted PPM_M was associated with impaired LV global longitudinal strain (lower LV-GLS values in the presence of PPM_M). In other words, unpredicted PPM_M correlates with LA function impairment, while predicted PPM_M correlates with subclinical LV dysfunction. LA and LV reverse remodeling were not correlated with either of the PPM_M groups, based on 30-days echocardiography.

Despite being a controversial parameter [8,36–38], measured PPM_M assessed by echocardiography 30 days after the procedure, crude or BMI-adjusted, appears to select a population with higher mortality risk in our study. The excess mortality associated with unpredicted PPM_M may be accounted to some extent to LA impairment, as LA reservoir dysfunction (LA-GLS) is known to be related to increased LV filling pressures [35], which are higher in the presence of PPM[39]. As previously shown in the literature, reverse LA remodeling and improvement in LA reservoir function occur post-TAVI and are independently associated with major adverse cardiac events [40]. Lack of correlation between LA and LV reverse remodeling with either of the PPM_M groups could be related to early post-procedural echocardiography, without additional echocardiographic follow-up. Furthermore, LA strain is a marker of LV diastolic dysfunction and has been proven to be a useful tool for diastolic dysfunction assessment[41], but also a measure for the non-invasive assessment of LV filling pressures[35].

While TAVI reduces afterload and improves LV remodeling recovery, myocardial fibrosis may not completely reverse after the procedure, resulting in persistent diastolic dysfunction[42]. Moderate to severe diastolic dysfunction can still be observed up to 10 years after AVR, despite significant reductions in LV mass [39]. Studies investigating the impact of mismatch on heart failure concluded that diastolic dysfunction was a predictor of mortality, independent of PPM[43]. These results are inconsistent with our findings, which prove PPM_M a predictor of adverse outcomes, irrespective of diastolic dysfunction[39,43].Another study analyzing the effect of PPM on longterm outcomes in a large cohort of patients undergoing SAVR found no significant impact[44]. However, the definition of PPM was based only on predicted PPM, as no echocardiographic was available. Our analysis is consistent with these results; as predicted, PPM was not associated with increased all-cause mortality at 3 years. However, our study provided evidence that PPM cannot simply be prevented in the clinical setting just by relying on determining valve predicted EOAi and that predicted PPM underestimated measured PPM_M. Technical adjustments such as avoidance of underexpansion, higher depth of implantation, balloon postdilation, or use of self-expandable valves could improve these results but associate additional risks and are not always anatomically feasible [45].

Nonetheless, this controversial parameter, PPM_M determined by echocardiography at 30 days post-TAVI identifies a population with higher all-cause mortality at 3 years follow-up. Although extensive echocardiographic data in our study indicate that significant PPM_M does not hinder the immediate effects of intervention, it appears to influence the preexisting cardiac damage and contribute to other comorbid conditions that may lead to increased medium-term and potentially long-term mortality. The relatively low incidence of mortality in our study group (25 events, 15.6% at 3-year follow-up) is encouraging but limits extensive analysis. The causal relationship between PPM, left atrial function, and mortality remains yet to be determined.

This is a prospective study conducted on consecutive AS patients who underwent TAVI, resulting in a heterogeneous population in terms of cardiac damage and associated comorbidities. Despite that, all patients underwent TAVI with balloon-expandable valves. One limitation of the study consisted in lack of repeat echocardiographic evaluation after the initial 1-month follow-up visit. While 3-year mortality, obtained through queries of the National Register of population records, is a robust endpoint, no information regarding cause of death was available. The findings outlined throughout our study must be interpreted in the context of a relatively small number of patients but with comprehensive advanced echocardiographic assessment and require confirmation in larger studies.