Invasive Hemodynamic Assessment Is Mandatory to Confirm Bioprosthetic Valve Dysfunction and Failure Following TAVR

Abstract

Echocardiography is commonly used for clinical evaluation of bioprosthetic valves. However, the clinical significance of an isolated high gradient detected by echocardiography is unclear. This uncertainty arises from discrepancies between echocardiographic and invasive gradient measurements in normally functioning prosthetic valves, that is attributed to limitations of the Bernoulli principle among other factors. Furthermore, the relationship between echocardiographic gradient and clinical outcomes is complex and nonlinear. To avoid unnecessary interventions, it is crucial to confirm an unfavorable gradient detected by echocardiography through invasive measurements, particularly when valve thrombosis is absent.

Introduction

Assessment of valvular hemodynamics, by echocardiography, has evolved as the primary method for evaluation of native and prosthetic valvular stenosis, replacing invasive methods in most cases. ¹ Non-invasive estimation of transvalvular pressure gradients is achieved using the simplified Bernoulli equation, where the peak instantaneous pressure gradient = 4 ${\overline{v}}^2$ ( $v$  = peak transvalvular velocity) from which the mean gradient (MG) is derived. Additionally, the aortic valve area is calculated using the continuity equation.

Nevertheless, a discrepancy in gradients between echocardiography and invasive measurements has been observed in normally functioning valves after both surgical aortic valve replacement (SAVR) and transcatheter aortic valve replacement (TAVR), which is referred to as discordance. ² This review will discuss the role of invasive hemodynamics in bioprosthetic valve dysfunction (BVD) following TAVR.

Bernoulli Principles and Limitations in Aortic Stenosis

Hatle et al first described the use of Bernoulli equation to estimate pressure drop across valvular stenosis initially across the mitral valve and eventually expanded across the aortic valve (AV) in aortic stenosis (AS).^3,4

Bernoulli equation describes the relationship between pressure and kinetic energy. ⁵ It is predicated on the first law of thermodynamics and the principle of energy conservation for an incompressible, inviscid, laminar fluid, with steady velocity, and a single level of obstruction, where a rise in kinetic energy (velocity) is associated with a reduction in pressure (convective forces). ⁴

To account for non-convective forces, the Bernoulli equation accounts for changes in flow velocity over time with ventricular contraction (flow acceleration) and incorporates a viscous constant for frictional losses (viscous losses). When ignoring potential energy, the Bernoulli equation describes blood flow, from the left ventricle (LV) across the AV as follows ⁴ :

$$\mathrm{\Delta }P(P_{LV}-P_{AV})={\displaystyle \frac{1}{2}\rho ({\overline{v}}_{AV}^2-{\overline{v}}_{LV}^2)+\rho {\int }_{LV}^{AV}\left({\displaystyle \frac{\mathrm{\partial }v}{\mathrm{\partial }t}}\right)ds+R(\mu )}$$

$$\mathrm{\Delta }P=Convectiveacceleration+Flowacceleration+ViscousLosses$$

where $\mathrm{\Delta }P(P_{LV}-P_{AV})$  = transaortic pressure gradient, $({\overline{v}}_{AV}^2-{\overline{v}}_{LV}^2)$ difference in the squares of average velocity between LV and AV, ρ = mass density of blood, R = viscous resistance, μ = viscosity.

In AS, the non-convective forces are considered negligible and ignored. Furthermore, the proximal flow velocity ( ${\overline{v}}_{LV}$ ) across the left ventricular outflow tract (LVOT) is considered negligible when the distal AV velocity ( ${\overline{v}}_{AV}$ ) is > 3 m/s and is ignored in the Simplified Bernoulli ⁶ :

Modified Bernoulli – $\mathrm{\Delta }P(P_{LV}-P_{AV})=\frac{1}{2}\rho ({\overline{v}}_{AV}^2-{\overline{v}}_{LV}^2)$ $=4V_{AV}^2-4V_{LV}^2$

Simplified Bernoulli – $\mathrm{\Delta }P(P_{LV}-P_{AV})=\frac{1}{2}\rho ({\overline{v}}_{AV}^2)=4V_{AV}^2$

Pitfalls of Simplification of Bernoulli in Prosthetic Valves

Bernoulli mainly accounts for the pressure drop due to convective acceleration (area reduction or stenosis), in a quadratic relation to area (inversely) and velocity (directly), while considering non-convective forces—flow acceleration and friction losses—as insignificant. On the other hand, flow acceleration exhibits a linear relationship with the diameter and length of the conduit, hence, is highly dependent on valve design and size through which blood flows.^5,7 In prosthetic valves with substantial long frames, such as self-expanding valves (SEV), flow acceleration likely plays a more pronounced role in contributing to pressure drop resulting in an underestimation of the gradient by echocardiography.

Simplified Bernoulli equation presumes the proximal velocity to be minimal compared to the ${\overline{v}}_{AV}$ . Although reasonable with AS when ${\overline{v}}_{AV}$  > 3 m/s, in normal prosthetic valves with a lower ${\overline{v}}_{AV}$ or with ${\overline{v}}_{LV}$  > 1.5 m/s, this leads to an overestimation of the gradient.

Pitfalls in Implementation of Bernoulli in Prosthetic Valve Evaluation

Flow Profile: Laminar Versus Turbulent, Uniform Versus non-Uniform

In standard prosthetic valve designs that produce turbulent blood flow, the velocity profile is typically flat without a distinct peak velocity, and the pressure drop/flow rate relationship is quadratic. Conversely, in valve designs that produce laminar flow, the velocity profile is parabolic with a clear peak velocity, and the pressure drop/flow rate relationship transitions from quadratic to linear. ⁸ Therefore, in prosthetic valves of comparable geometric areas, laminar flow results in a higher peak velocity, a reduced pressure drop, and an increased Bernoulli-derived gradient compared to turbulent flow.

Bernoulli principle presupposes a uniform velocity profile across the flow column—wherein the $\overline{v}$ term representing the average blood velocity across the flow column is taken to be the maximal measured velocity. ⁹ In laminar flow, the maximum velocity may not reflect the average velocity. ¹⁰ Consequently, using the maximal velocity in place of the average velocity would result in an overestimation of the pressure gradient. ¹¹

Flow Convergence: Multiple Versus Single Level

Bernoulli assumes a single-level stenotic orifice within a “short tube”, wherein a single flow convergence occurs. ¹² Nevertheless, in normally functioning bioprostheses, there are multiple levels of flow convergence; beneath the valve frame, beneath the leaflets, and at the exit from the valve frame.^13,14 With multiple levels of flow convergences, or stenoses in series (AS and coexisting LVOT obstruction), Bernoulli-derived gradients are inaccurate.^6,15 This inaccuracy results in an overestimation of the gradient due to the failure of several previously discussed assumptions.

Pressure Recovery

Pressure recovery (PR) occurs above the AV, where a decrease in blood flow velocity allows for partial recovery of dissipated pressure, resulting in a lower pressure gradient across the AV. ⁵ Echocardiography, which measures the highest velocity, ignores PR, leading to an overestimation of the MG compared to the one measured invasively. ¹⁶ In native AS, the degree of PR is influenced by the size of the proximal ascending aorta, the aortic valve area, and the presence of eccentric jets. Although non-invasive methods like the Energy Loss Index account for PR in AS, they are not applicable to eccentric jets or prosthetic valves. ¹⁷ Post-TAVR, while PR does occur, the discrepancy between echocardiographic and invasive MG measurements exceeds the effects of PR. ¹⁸

Discordance in Echocardiographic and Invasive Derived Gradients in TAVR

Invasive gradient directly measures the difference between the LV and the AO pressures, without the assumptions and simplification of the Bernoulli equation, and accounts for PR. In contrast, the echocardiogram-derived gradient overlooks PR, non-convective forces, proximal velocity, and is constrained by the Bernoulli equation's inherent assumptions.

Normally functioning prosthetic valves with different designs, exhibit different flow profiles, multiple levels of flow convergence, and pressure recovery.^7,13,14 Consequently, echocardiography may yield “valve-specific” gradients that are discordant to invasive measurements even when the geometrical areas and invasive mean gradients are identical. Accounting for proximal LVOT velocity and PR does not completely explain the discordance observed post-TAVR^5,18 confirming the limitations of Bernoulli equation in this population beyond merely its simplification.

Indeed, multiple early reports demonstrated significant discrepancies between Doppler-derived and catheter-derived gradients across various SAVR valve designs.^9,19–22 These differences were noted even in the absence of prosthetic valve stenosis, were more marked in smaller valves, varied between valve designs, and were independent of sinotubular junction and the impact of PR.

Significant discrepancies in concomitantly obtained echocardiographic and invasive MG also exist in TAVR valves.^23–25 A study of > 5000 patients showed good correlation and agreement between these two measurements in AS, a weak correlation and agreement in native TAVR, and only moderate correlation with weak agreement in valve-in-valve (ViV)-TAVR. ²³

Absolute discordance increases with higher echocardiographic gradient and discordance is more pronounced (>10 mm Hg) in smaller valves, balloon-expandable valves (BEV), higher stroke volume index (SVi), younger age, and with ViV-TAVR. Predictors of echocardiographic MG > 20 mm Hg included small valve size, BEV, ViV-TAVR, SVi, and younger age, while predictors of invasive MG > 20 mm Hg included SEV, ViV-TAVR, pre-TAVR invasive MG, and body mass index. Similarly, a randomized study involving patients with failed small surgical valves undergoing TAVR demonstrated that echocardiographic MG were higher in BEV compared to SAV, despite similar invasive gradient measurements. ²⁶

Another study described the relationship between flow and invasive MG pre- and post-TAVR using high fidelity pressure catheters and dobutamine infusion. ²⁷ It demonstrated that in AS, the flow-gradient relationship is represented by a combined quadratic orifice and linear resistive model. Post-TAVR, however, this relationship tends to conform more to a purely linear resistive model. Interestingly, a 2.2-fold increase in post-TAVR valve resistance (the ratio of gradient change to flow change) was observed in SEVs compared to BEVs, further demonstrating the distinct flow patterns and physiological responses associated with different valve designs.

Transvalvular Gradient and Clinical Outcomes Following TAVR

Impact of Elevated Echocardiographic Gradient on Clinical Outcomes

Multiple studies have failed to demonstrate an association between elevated echocardiographic MG and mortality. In the PARTNER 3 (Placement of Aortic Transcatheter Valves) trial, there was no association between the 30-day echocardiogram severe prosthesis-patient mismatch (PPM) or MG ≥ 20 mm Hg and 1 or 5-year clinical outcomes for all annular sizes.^28,29 Similarly, a propensity matched analysis of STS-ACC TVT (Society of Thoracic Surgeons–American College of Cardiology Transcatheter Valve Therapy) registry patients undergoing TAVR with BEV did not find an association between echocardiographic MG > 20 mm Hg and 1 and 3-year clinical outcomes^30,31 or any valve type following ViV-TAVR and 1 year outcomes. ³²

A study of patients undergoing ViV-TAVR in the PARTNER 2 ViV registry, did not find a higher incidence of adverse clinical outcomes at 3-years in those with elevated echocardiographic gradient, and similar findings were observed at 8-years in the ViV International Data registry.^33,34

Non-Linear Relationship Between Echocardiogram Gradient and Mortality

A multicenter retrospective registry including both native and ViV-TAVR, described a non-linear relationship between echocardiographic MG and 2-year all-cause mortality (ACM). ³² A low MG (<10 mm Hg) was associated with a higher adjusted 2-year ACM compared to MG 10–20 mm Hg, while association of a high MG (>20 mm Hg) with ACM did not reach statistical significance. Patients exhibiting low echocardiographic MG were older, with reduced left ventricular ejection fraction (LVEF) and SVi.

Similarly, two recent studies from the STS-ACC TVT registry, one involving patients undergoing native TAVR with BEV and 3-year ACM, and the other involving patients undergoing ViV-TAVR with either BEV or SEV and 1-year ACM, also demonstrated a non-linear relationship between MG and ACM. Both studies found increased ACM at very low (<10 mm Hg) and extremely high (>30 mm Hg) echocardiographic MG when compared to MG of 10–30 mm Hg (Figure 1).^31,35

Figure 1.

(top panel) The discordance between echocardiographic and invasive MG in TAVR valves can be attributed, in part, to the simplifications and assumptions inherent in the Bernoulli equation. The relationship between echocardiographic MG and mortality is complex and nonlinear. Invasive MG is directly measured, less influenced by flow variations, and more closely associated with mortality. (Bottom panel) Representative images of echocardiographic MG (13 mm Hg) and invasive MG (2 mm Hg) for the same patient post-TAVR, demonstrating discordance. The middle figure is a scatterplot of invasive and echocardiographic MG, demonstrating a weak correlation following TAVR. Abbreviations: MG, mean gradient; TAVR, transcatheter aortic valve replacement. Adapted from. ²³

A frequently cited but incorrectly extrapolated study suggesting an association between elevated MG and clinical outcomes is an analysis of the National Echo Database Australia, which showed an increased 5-year ACM in patients with moderate or severe MG elevation (20-40 or >40 mm Hg) after aortic valve replacement (AVR). ³⁶ However, applying the findings of this study to all bioprosthetic valves in clinical practice comes with several caveats. Primarily, the study focused on surgical AVR (81% of participants), and nearly half of the subjects had mechanical valves, which have different flow profiles compared to bioprosthetic valves. Additionally, the gradients were taken from the last available echocardiogram, omitting previous gradients, changes in gradient over time, and the duration of valve implantation. This is particularly relevant as patients with the markedly abnormal hemodynamics had gradients as high as 40 mm Hg, suggesting potential prosthetic valve stenosis or thrombosis over time, indicative of actual valve dysfunction rather than normal function. Similar to previously discussed studies, a J-shaped risk pattern emerged when survival was stratified by gradient quantile distribution, revealing increased ACM at the lowest and highest quantiles of gradient (<7 and >20 mm Hg, respectively). Moreover, patients with abnormal hemodynamics had lower EF and lower SVi.

Finally, an analysis of pooled PARTNER2A and S3i studies showed that TAVR and SAVR patients with low flow and severe measured PPM experienced worse 5-year clinical outcomes (cardiac death or rehospitalization), despite exhibiting lower MG, compared to normal flow severe PPM. ³⁷ Patients with severe AS in a low flow state (SVi < 35 ml/m²), whether they have preserved or reduced ejection fraction, generally experience worse clinical outcomes after TAVR compared to those with normal flow, despite having lower gradients post-TAVR. ³⁸

In contrast to echocardiographic MG, low invasive MG in normal bioprostheses seems to better reflect valve hemodynamics and is less affected by myocardial function and low flow, as evidenced by LVEF and echocardiographic SVi.^23,32 As invasive measurement provides a directly measured gradient that is reproducible, rather than an estimate, it offers greater reliability compared to echocardiographic MG. Furthermore, a study comparing echocardiographic and invasive gradients post-TAVR found that while different echocardiographic gradients (<10 mm Hg, 10-20 mm Hg, and > 20 mm Hg) were associated with higher EF and SVI, these indices were similar across increasing invasive gradients.

Association of Invasive Gradient and Clinical Outcomes

Few studies have explored the relationship between invasive gradient and clinical outcomes. A multicenter retrospective registry revealed that an intermediate invasive MG of 5–10 mm Hg, assessed immediately after TAVR, was linked to a higher 2-year ACM than a low invasive MG < 5 mm Hg. ³² (Figure 1). The study lacked the statistical power to assess outcomes for patients with an invasive MG > 10 mm Hg. Patients with lower invasive MG were slightly older with no significant difference in LVEF or SVi between the groups. These findings were replicated by a study from the Erasmus center that is currently in-press. ³⁹

Role of Echocardiography and Invasive Gradients post-TAVR

The Valve Academic Research Consortium-3 has eliminated a single echocardiographic measure as an indication of BVD and has necessitated the presence of morphological changes in the valve and a change in echocardiographic MG > 10 mm Hg resulting in a MG > 20 mm Hg with a concomitant decrease in the effective orifice area > 0.3 cm². However, an echocardiographic MG > 20 mm Hg continues to be a criterion for 30-day device success. ⁴⁰

Evaluating Immediate Device Success: Echo Versus Invasive Hemodynamics

Obtaining invasive hemodynamic data could offer complementary insights, upon observing an increased echocardiographic MG immediately post-TAVR or ViV-TAVR potentially obviating the need for additional interventions such as post-dilation or balloon fracture. This approach is examined in the prospective randomized ECHOCATH study (Valve Hemodynamic Optimization Based on Doppler-Echocardiography vs Catheterization Measurements Following ViV-TAVR; ClinicalTrials.gov Identifier: NCT05459233).

Furthermore, an increase in echocardiographic MG on the day of discharge is often noted, particularly with smaller BEV. ²⁴ This finding may not be suggestive of BVD or discordance; rather, it could indicate improved left ventricular function and transvalvular flow, the resolution of anesthesia, and lower blood pressure at discharge.

Long Term TAVR Surveillance: Echo Versus Invasive Hemodynamics

Echocardiography offers valuable supplementary details, including myocardial function, flow status, and aortic regurgitation. Nonetheless, a solitary echocardiographic gradient after TAVR holds limited clinical significance. Serial monitoring of echocardiographic changes may shed light on BVD in TAVR, as per VARC-3 criteria recommendations. The definition of a significant change remains unclear, however, and necessitates further research.

A small prospective study of post-TAVR patients meeting the VARC-3 criteria for ≥ stage 2 (moderate) hemodynamic valve deterioration (HVD), or an MG ≥ 20 mm Hg on follow-up imaging > 1-month post-TAVR, compared invasive and echocardiographic MG. ⁴¹ Structural valve dysfunction was ruled out by computed tomography. A total of 69.2% of patients had a simultaneous invasive MG < 20 mm Hg, and all patients who met VARC-3 criteria for ≥ stage 2 HVD had an invasive MG < 20 mm Hg. Echocardiographic MG was significantly higher than invasive MG (mean difference 11.5 ± 8.4 mm Hg; P < .001). The study highlights the significance of refraining from premature re-interventions based solely on echocardiographic measurements. Invasive hemodynamics should be considered for patients evaluated for redo AVR, especially when there is no valve thrombus evident on computed tomography and when patients present with equivocal symptoms, particularly in the absence of abnormal valve morphology or valve regurgitation.

The prospective multicenter DISCORDANCE TAVR study (Standardized Invasive Hemodynamics for Monitoring Acute and Long-Term Valve Performance in Patients with Elevated Gradients Post-Transcatheter Aortic Valve Replacement; ClinicalTrials.gov Identifier: NCT05459233) is investigating the role of invasive hemodynamics in the long-term follow-up and management of post-TAVR patients who meet the VARC3 criteria for ≥ stage 2 HVD.

Conclusion

Echocardiography is essential for evaluation and long-term monitoring of patients with TAVR. Changes in echocardiographic hemodynamics hold more significance than a solitary gradient measurement due to notable discrepancies between echocardiographic and invasive gradient measurements, a phenomenon known as discordance. Discordance is not entirely attributable to pressure recovery and arises from the limitations of the Bernoulli equation in normal bioprostheses. The relationship between echocardiographic gradient and long-term clinical outcomes is non-linear, complex, and most likely predominantly related to myocardial function, ejection fraction, and flow. To avoid unnecessary valve re-intervention, whether in the acute phase or over the long term, confirmation of unfavorable echocardiographic hemodynamics by invasive measurement is essential, especially after excluding valve thrombosis.