This was an observational study using a prospectively maintained database of TAVI procedures performed at a single institution from 2012 to 2023. Definitions and terminology were consistent with the Society of Thoracic Surgeons (STS)–American College of Cardiology (ACC) Transcatheter Valve Therapy (TVT) Registry. All patients undergoing TAVI with both invasive and echocardiographic post-procedural mean aortic valve pressure gradients were included in the study. Mean valvular gradient measurements were collected in the immediate post-procedural setting after deployment of the transcatheter heart valve. Echocardiographic gradients were measured according to current recommendations from the American Society of Echocardiography [9]. Invasive gradients were obtained from simultaneous left ventricular (LV) and aortic pressures measured via pigtail catheters. This study was approved by the Institutional Review Board of the University of Pittsburgh on 4/17/2019 (STUDY18120143), with written consent waived.

This study aimed to assess the correlation between post-procedural TTE and invasive pigtail catheter mean aortic valve pressure gradients in patients undergoing TAVI. The correlation between the invasive and the TTE pressure gradient before TAVI was not assessed in this study. In severe AS, TTE is the gold standard for quantifying stenosis severity because of the simplified Bernoulli equation, which assumes laminar flow at a single level of stenosis and treats flow acceleration and viscous forces as negligible. However, these assumptions may not be true for non-restrictive transcatheter heart valves. For this reason, invasive pressure gradients are unlikely to significantly diverge from TTE pressure gradients when the valve is severely stenotic, while they are more likely to diverge when the valve is non-restrictive [10]. By implication, correlating TTE and invasive pressure gradients before TAVI is unlikely to be related to correlations after TAVI.

The primary outcome was the correlation between TTE and invasive mean pressure gradient measurements after TAVI. Secondarily, this study sought to determine risk factors for discordance between TTE and invasive pigtail gradients $>$10 mmHg, measured as the raw difference between each modality (i.e., TTE pressure gradient minus invasive pressure gradient). This is consistent with prior studies, where $>$10 mmHg has defined clinically significant discordance [7, 8]. Moreover, short-term and long-term clinical outcomes were determined for patients with a TTE-invasive discordance $>$10 mmHg. A Cox proportional-hazards model was built for survival to assess the potential relationship between discordance and the long-term hazard of death. The proportional-hazards assumption was tested by Schoenfeld Residuals. All follow-up data were obtained from a clinical warehouse that contains long-term survival data for patients undergoing cardiac operations at the University of Pittsburgh Medical Center. All patients had longitudinal follow-up at 1-month, 6-months, 1-year, and yearly thereafter. Survival status was obtained via chart review of these follow-up appointments as well as patient phone calls, all of which were verified with the Social Security Death Index at the time of this study. All time-to-event data were right-censored. For all analyses, missing data were handled via single imputation by group median if less than 15% of the total cohort was missing.

Immediately after valve deployment, LV and aortic pressures were recorded simultaneously. LV pressure was obtained retrograde with a pigtail positioned as apically as feasible in the LV, while aortic pressure was obtained in the proximal ascending aorta via a second pigtail through contralateral access. All transducers were flushed and zeroed to atmospheric pressure. Waveforms were checked for dampening prior to acquisition. The mean transvalvular gradient was calculated from the difference between LV and aortic pressures integrated over systole, which were averaged across three consecutive beats at end-expiration.

Continuous variables were presented as mean $\pm$ standard deviation for normally distributed data or median and interquartile range for nonnormally distributed data. Categorical data were reported by frequency and percentage. Normally distributed continuous variables were analysed by the Student t-test, whereas nonnormally distributed continuous variables were analysed with the nonparametric Mann-Whitney U test. A $\chi$² or Fisher exact test was used to compare categorical variables between groups.

Spearman’s coefficient was used to assess the correlation between TTE and invasive gradients (where coefficient values from 0.90 to 1.0 indicate very strong correlation; 0.70 to 0.89 strong correlation; 0.40 to 0.69 moderate correlation; and 0.40 weak correlation). Logistic regression was used to determine risk factors for TTE-invasive discordance $>$10 mmHg, while multivariable Cox proportional-hazards regression was performed for long-term survival. All statistical analyses were performed with SAS/STAT version 15.2 software (SAS Institute, Cary, NC, USA). All tests were 2-sided with an $\alpha$ level of 0.05 designated to indicate statistical significance.

A total of 1589 patients underwent TAVI with available TTE and invasive pressure gradients. Baseline characteristics are reported in Table 1. In this cohort, 782 (49.2%) of the implanted valves were self-expanding valves (SEV) and 807 (50.8%) were balloon-expanding valves (BEV). Additionally, 1474 (92.8%) procedures involved TAVI in native valves, while 115 (7.2%) were valve-in-valve (ViV) procedures. Transcatheter heart valves were additionally stratified by size, with small valves defined as Evolut (Medtronic, Minneapolis, MN, USA) valves $\le$26 mm, Sapien (Edwards Lifesciences, Irvine, CA, USA) valves $\le$23 mm, and Navitor/Portico (Abbott Vascular, Santa Clara, CA, USA) valves $\le$25 mm. While 1309 (82.4%) patients had a large transcatheter heart valve implanted, 280 (17.6%) patients had a small valve implanted. Based upon consensus, recent randomized controlled trials have defined the small aortic annulus as a mean annular diameter 23 mm or an annular area $\le$430 mm² [11, 12]. Evolut $\le$26 mm, Sapien $\le$23 mm, and Navitor/Portico $\le$25 mm valves fit into small aortic annuli according to these definitions, while larger aortic annuli can only accommodate larger sizes of the commercially available valves. As such, this study’s definition of small and large valve implants is based upon previous scientific literature, rather than industry designation.

For the entire cohort, TTE and invasive mean gradients correlated moderately well (Spearman $\rho$ = 0.401, p 0.001, Fig. 1, Table 2A), with a median absolute difference of 2.2 [1.0–4.0] mmHg between measurement modalities. Summary statistics of post-implantation gradients in TAVI subgroups are listed in Tables 2A,2B,2C,2D. Invasive and TTE gradients were more strongly correlated in SEVs than BEVs (Spearman $\rho$ = 0.447 vs. 0.345, respectively, Fig. 2, Table 2B). Similarly, TTE and invasive measurements were better associated in small valves compared to large valves (Spearman $\rho$ = 0.455 vs. 0.375, respectively, Fig. 3, Table 2C), while ViV TAVI pressure gradients were better correlated than native TAVI gradients (Spearman $\rho$ = 0.575 vs. 0.357, respectively, Fig. 4, Table 2D). Figs. 1,2,3,4 are scatter plots of TTE versus invasive pressure gradients, and the curves in each figure depict a linear regression that models the relationship between TTE and invasive measurements.

Fig. 1.

Correlation between invasive and echocardiographic mean transvalvular aortic gradients immediately post-TAVI.

Fig. 2.

Correlation between post-TAVI invasive and echocardiographic mean gradients in balloon- vs. self-expanding valve designs.

Fig. 3.

Correlation between post-TAVI invasive and echocardiographic mean gradients in large vs. small valve implantation.

Fig. 4.

Correlation between post-TAVI invasive and echocardiographic mean gradients in native vs. valve-in-valve implantations.

To assess agreement between TTE and invasive pressure gradients, we constructed a Bland-Altman plot. The bias (i.e., mean difference between TTE and invasive pressure gradients) was –0.31 mmHg (SD 3.88), with 95% limits of agreement of –7.92 to 7.30 mmHg (Supplementary Fig. 1). This near-zero bias indicated no systematic difference between each modality of measuring pressure gradients. 95% of pairs differed by less than 8 mmHg, and visual inspection of the Bland-Altman plot showed no proportional bias across the measurement range.

The difference between the TTE and invasive pressure gradient was calculated and called the “TTE-invasive discordance”. The TTE-invasive discordance was compared to the TTE gradient to determine if the TTE-invasive discordance changed as the TTE gradient increased. Correlation between TTE-invasive discordance and TTE mean gradients was weak (Spearman $\rho$ = 0.107, p 0.001).

Twenty-five patients (1.6%) had a TTE-invasive discordance that exceeded 10 mmHg. Patients with a TTE-invasive discordance $>$10 mmHg had significantly higher TTE gradients compared to those with a TTE-invasive discordance $\le$10 mmHg (12.0 mmHg vs. 6.0 mmHg, p 0.001). There were no differences in most immediate post-TAVI outcomes in patients with a TTE-invasive discordance $>$10 mmHg versus patients with a TTE-invasive discordance $\le$10 mmHg, including in-hospital mortality, permanent pacemakers, renal failure, and stroke (p $>$ 0.05, Table 3). However, patients with TTE-invasive discordance $>$10 mmHg had more blood product transfusions (12.0% vs 3.8%, p = 0.038) and more major vascular complications (4.0% vs 0.2%, p 0.001). On univariable logistic regression, risk factors for a TTE-invasive discordance $>$10 mmHg included age, ViV TAVI (p 0.001), and a small valve TAVI (p = 0.005), but not SEVs vs. BEVs (p = 0.281) (Table 4).

Median [IQR] follow-up for the entire cohort was 3.3 [2.1–4.9] years and there were a total 505 (31.8%) deaths. On multivariable Cox regression, a TTE-invasive discordance $>$10 mmHg was not significantly associated with an increased hazard of death after TAVI (HR: 1.37; 95% CI: 0.73–2.54; p = 0.326, Table 5). Variables that were found to be significantly associated with death included diabetes, atrial fibrillation, and pre-procedure creatinine.

To determine if the raw pressure gradients measured by either modality were separately more predictive of survival, another Cox model was built, with the post-procedural TTE and invasive pressure gradients being treated as independent variables for predicting death. For this model, the same covariates listed in Table 5 were utilized. Neither the TTE pressure gradient (HR: 1.02 [per 1 unit increase in mmHg], 95% CI: 0.90–1.16, p = 0.750) nor the invasive pressure gradient (HR: 1.10 [per 1 unit increase in mmHg], 95% CI: 0.99–1.22, p = 0.080) were independently associated with mortality.

This study has several important limitations. First, the retrospective observational design of the study may introduce considerable selection bias. Gradient data were measured and collected by a single operator, thereby introducing a potential source of measurement bias. Moreover, TTE gradients were acquired immediately following TAVI, with patients in a supine position. This may have resulted in improper capturing of peak aortic velocities, potentially leading to underestimation of TTE mean gradients. Finally, this study’s data are from a single centre, which may limit the generalizability of the findings to other institutions with different experiences with TAVI.

Invasive and TTE pressure gradients were obtained in close temporal succession after valve deployment, but not simultaneously. Thus, interval changes in several parameters — such as blood pressure, heart rate/rhythm, preload/afterload, respiration, and level of sedation — may have introduced variability between modalities. Additionally, this is a single-center retrospective study, presenting inherent limitations for generalizability and leaving room for residual or unmeasured confounding despite multivariable adjustment.