Leaflet Immobility and Thrombosis in Transcatheter Aortic Valve Replacement

Arnold C. T. Ng; David R. Holmes; Michael J. Mack; Victoria Delgado; Raj Makkar; Philipp Blanke; Jonathon A. Leipsic; Martin B. Leon; Jeroen J. Bax


Eur Heart J. 2020;41(33):3184-3197. 

In This Article


Thrombosis plays a central role in some patients with early THV leaflet dysfunction. Resolution of the echocardiographic and MDCT THV abnormalities [e.g. hypoattenuated leaflet thickening (HALT)] with anticoagulation strongly suggests a thrombosis pathology, but specific insights into the pathophysiology of initial thrombus formation on the leaflets remain unclear. Whether unrecognized or untreated early THV thrombosis represents a long-term risk factor for subsequent SVD from tissue degeneration and/or fibrosis is unclear. However, determining its impact on THV longevity is of great clinical importance and can only be evaluated by long-term outcome studies. To develop effective risk stratification tools to identify at-risk patients, formulate effective treatment strategies, and determine appropriate surveillance protocols, it is important to identify the mechanisms leading to THV thrombosis.

Based on the principles of Virchow's triad, factors contributing to THV thrombosis can be categorized into alterations in the constitution of blood, endothelial dysfunction, or alterations in blood flow. For example, elderly TAVR patients more likely have underlying prothrombotic comorbidities such as cancer. Another potential prothrombotic aetiology concerns the recovery of large von Willebrand factor multimers following the treatment for aortic stenosis. Usually, the loss of these macromolecules due to shear stress is associated with bleeding diathesis in severe aortic stenosis. Recently, Yamashita et al.[10] observed rapid correction of these high-molecular weight von Willebrand factor multimers post-aortic valve replacement, and patients were characterized as being in a 'von Willebrand factor predominant state' between post-operative Days 8 and 22, predisposing them to thrombosis rather than bleeding even in the early stage after surgery. These factors may potentially contribute to the pathogenesis of HALT early after TAVR.

Regarding endothelial dysfunction, the previous study demonstrated that bioprosthetic tissues undergo four phases of healing after implantation, starting from initial platelet and fibrin deposition to inflammation, granulation tissue, and eventual fibrous encapsulation.[11] Factors resulting in delayed re-endothelialization would theoretically increase thrombotic risks. Jilaihawi et al.[12] have suggested a systematic and quantitative 4D MDCT analysis protocol evaluating stent frame-related factors that could potentially contribute to THV thrombosis, including native commissural/bioprosthetic leaflet orientation, stent frame expansion, stent frame fracture, depth, and symmetry of implantation. Theoretically, technical complications such as stent frame fracture could result in delayed re-endothelialization and increased risk of THV thrombosis. Midha et al.[13] reported that over-expansion was associated with the higher incidence of THV thrombosis. The authors suggested that over-expansion of the THV stent may increase endothelial injury and provide a nidus for thrombus formation.

Notably, however, majority of studies on leaflet re-endothelialization is related to surgical valves with limited analyses on TAVR valves. However, endothelial-like cells noted on the leaflet surfaces of explanted TAVR often have an abnormal morphology suggestive of endothelial dysfunction.[14] Studies on endothelial dysfunction of native valve leaflets have demonstrated impaired nitric oxide generation and subsequent activation and proliferation of interstitial valve cells, increased reactive oxygen species generation with promotion of osteogenic differentiation of the interstitial valve cells, and inflammation with recruitment of immune cells within the leaflets.[15–17] Valvular endothelial dysfunction may also be related to valve toxicity related to fixatives such as glutaraldehyde used in manufacturing process.[18]

However, most studies have predominantly focused on alterations in blood flow leading to THV thrombosis.[8,13,19–26] Transcatheter heart valve thrombosis is typically localized to the aortic side of the leaflets. Although all leaflets can be involved, previous studies suggested that the leaflet corresponding to the native right coronary cusp may be more commonly affected.[27,28] Analysis of published literature revealed several common flow-related predictors of THV thrombosis including height of annular deployment and neo-sinuses, valve-in-valve procedures/patient–prosthesis mismatch, and reduced cardiac output.

Aortic Root and Neo-sinuses

Insights arising from cardiac magnetic resonance on aortic valve sparing surgery for aortic regurgitation showed that the sinuses of Valsalva generate vortices that form during early systole and persist into early diastole and may thus play a role in reducing the risk of thrombus formation on the aortic side of the leaflets (Figure 2).[29,30] Using aortic root phantoms, Jahren et al.[31] showed that the aortic root morphology can affect blood flow behind the TAVR prosthesis. Specifically, there was absent vortex formation within the sinuses and resultant relative blood stasis behind the TAVR leaflets. Unlike SAVR whereby the native leaflets are removed, TAVR results in a new small 'neo-sinus' located between the displaced diseased native valve leaflets and the TAVR leaflets where thrombus usually forms (Figure 3). Detailed in vitro modelling showed increased blood stasis within these neo-sinuses as quantified by blood residence time in TAVR compared with SAVR (Figure 4, top panel).[19] The volume of these neo-sinuses also varies according to the TAVR type, position in relation to the aortic annulus, and the degree of apposition to the native valve (which is further influenced by local native valve characteristics such as calcification, as well as the implanted TAVR size). This was demonstrated by Midha et al.[13] who showed that supra-annular TAVR deployment resulted in nearly a seven-fold reduction in the size of the stagnation zone within the neo-sinus and a shorter blood residence time. However, a supra-annular TAVR deployment has to be balanced against the risk of coronary artery occlusion.

Figure 2.

Vortices generated within the sinuses of Valsalva using time-resolved three-dimensional magnetic resonance phase contrast imaging. Top row: particle paths at peak systole delineate haemodynamics in the thoracic aorta of 60-year-old patient with sinus prosthesis (A), of a healthy 53-year-old age-matched volunteer (B), and a 30-year-old young volunteer (C). Bottom row: sinus vortices in the right and left coronary sinus as visualized by instantaneous streamlines. Dashed lines demonstrating sinus borders and dotted lines delineating vortex direction. These vortices may play a role in reducing the risk of thrombus formation on the aortic side of the leaflets in native and surgical aortic valve replacements. Vmax, peak velocity. Reproduced with permission.30

Figure 3.

Transcatheter aortic valves divide the native aortic sinus of Valsalva into two new spaces: a smaller native sinus between the aortic wall and the native aortic valve leaflet (#) and a new 'neo-sinus' bounded by the native aortic valve leaflet and the transcatheter aortic valve leaflet (*) where thrombus is usually observed. TAVR, transcatheter aortic valve replacement.

Figure 4.

Contours of blood residence time (TR ) in seconds on the aortic valve leaflets and sinuses at the end of diastole. Top panel: comparison between surgical vs. transcatheter aortic valve replacements. Bottom panel: comparison between supra-annular vs. intra-annular valve-in-valve deployment. Modified and reproduced with permission.19,32.

Valve-in-valve Procedure and Patient–Prosthesis Mismatch

With valve-in-valve procedures, these neo-sinuses are confined by the degenerated surgical bioprosthesis frame that circumferentially surrounds the TAVR leaflets. In vitro modelling also showed increased blood stasis on TAVR leaflets following valve-in-valve procedures.[32] Similar to previous studies evaluating THV thrombosis for TAVR implanted in native aortic valves, a supra-annular TAVR deployment in valve-in-valve procedures had significantly shorter blood residence time within the neo-sinuses compared with an intra-annular position (Figure 4, bottom panel). Clinical studies have since corroborated the suggestion that valve-in-valve procedures may potentially increase THV thrombotic risks.[22,33] In the multicentre registry by Del Trigo et al.[22] including 1521 TAVR patients, there were 68 cases of THV thrombosis defined by echocardiography. The authors found that a higher body mass index (BMI), smaller TAVR size, and valve-in-valve procedures were independently associated with valve thrombosis on multivariable analysis. Similarly, Jose et al.[33] also identified valve-in-valve procedures (in addition to increased BMI, absence of anticoagulation, and use of balloon-expandable valve) as independent determinants of THV thrombosis.

Valve-in-valve procedures are also more likely to result in patient–prosthesis mismatch with a smaller indexed effective orifice area. Moreover, Abdel-Wahab et al.[20] also demonstrated that a smaller TAVR size in relation to body surface area was associated with higher incidence of THV thrombosis. Similarly, the recently published FRANCE TAVR registry suggested that a higher BMI, previous TAVR (i.e. valve-in-valve), smaller prosthesis size (≤23 mm), moderate-to-severe chronic renal failure, and absence of anticoagulation were independently associated with SVD defined as an increase in mean gradient ≥10 mmHg or new mean gradient ≥20 mmHg (a surrogate for possible THV thrombosis).[34]

It is interesting to note that a larger body size with a relatively smaller prosthetic aortic valve size are conditions with increased severe patient–prosthesis mismatch.[35] Although mild or even moderate patient–prosthesis mismatch may not impact on clinical outcomes, severe mismatch may result in haemodynamic flow perturbation, which could accelerate either thrombus or pannus formation. The multicentre registry by Yanagisawa et al.[26] including 485 TAVR patients also concluded that severe patient–prosthesis mismatch was associated with higher likelihood of early THV thrombosis (6.7% vs. 0.9%, P = 0.02). Clinically, severe patient–prosthesis mismatch is associated with adverse outcomes and patients more likely to experience SVD early (2–3 years) after implantation.[36] Flameng et al.[36] showed that patients with severe patient–prosthesis mismatch who develop SVD were more likely to present clinically as valve stenosis on follow-up. However, the proportion of these patients with SVD due to valve thrombosis is unknown.

Cardiac Output

In their in vitro study on fluid mechanics and neo-sinuses in TAVR, Midha et al.[13] found that reduced cardiac output resulted in larger stagnation zone and increased blood residence time, theoretically increasing the risk of THV thrombosis. This hypothesis was supported by two clinical studies showing reduced cardiac output as an independent predictor for THV thrombosis.[21,26] Chakravarty et al.[21] included patients from the Assessment of Transcatheter and Surgical Aortic Bioprosthetic Valve Thrombosis and its Treatment with Anticoagulation (RESOLVE) and Subclinical Aortic Valve Bioprosthesis Thrombosis Assessed with Four-Dimensional Computed Tomography (SAVORY) registries and showed that reduced left ventricular ejection fraction (LVEF) was independently associated with THV thrombosis. Similarly, Yanagisawa et al.[26] reported that patients with low-flow, low-gradient severe aortic stenosis had higher incidence of early leaflet thrombosis on multivariable analysis (odds ratio 2.71, 95% confidence interval 1.11–6.62; P = 0.03). However, other studies failed to identify cardiac output as a predictor of THV thrombosis.[23,37,38]

Balloon-expandable vs. Self-expanding Valves

Conflicting results exist on the thrombotic risks of various balloon-expandable vs. self-expanding TAVR in head-to-head comparisons.[26,33,37,39,40] In the Repositionable Percutaneous Replacement of Stenotic Aortic Valve Through Implantation of Lotus Valve System–Randomized Clinical Evaluation (REPRISE III) trial that randomized 912 participants into the self-expanding CoreValve™ vs. the balloon-expandable Lotus™ valve, the self-expanding valve had better forward flow dynamics, effective orifice area, and mean gradient due to its supra-annular design and positioning.[39] Transcatheter heart valve thrombosis was identified in 16 cases during routine echocardiographic follow-up, all of which occurred with the balloon-expandable valve (3.0% vs. 0%, P < 0.01). However, there was no difference in all stroke rates (8.4% in balloon-expandable vs. 11.4% in self-expanding valves, P = 0.75). Similarly, Jose et al.[33] reported a higher incidence of THV in balloon-expandable valves compared with self-expanding valves. Conversely, other studies did not observe any differences in THV thrombosis rates between the balloon-expandable vs. self-expanding valves.[26,37,40] Interestingly, Yanagisawa et al.[26] found that the Edwards Sapien 3™ had a significantly higher incidence of early leaflet thrombosis compared with the Sapien XT™ (17.9% vs. 4.1%, P < 0.001). On further inspection, the patients with Sapien 3™ valve thrombosis had their TAVR implanted in a lower position compared with those without thrombosis. This was consistent with the aforementioned in vitro study showing that a lower TAVR implantation (i.e. intra-annular deployment) results in larger neo-sinuses with increased stagnation zone and longer blood residence time independent of valve type.[13]

Transcatheter Aortic Valve Replacement vs. Surgical Aortic Valve Replacement

The Placement of Aortic Transcatheter Valves (PARTNER) 3 trial randomized severe aortic stenosis patients of low surgical risk to undergo either TAVR with Sapien 3 or SAVR.[2] The primary objective of the PARTNER 3 computed tomography (CT) substudy was to evaluate HALT and reduced leaflet motion (RLM) in a subset of patients from the larger randomized trial.[41] Four-dimensional MDCT was performed at 30 days and 1 year, and examinations were all interpreted in a CT core laboratory. Of the 408 patients, 346 and 312 patients had evaluable CT examinations at 30 days and 1 year, respectively. At 30 days, the incidence of HALT was significantly higher for TAVR compared with SAVR [22 of 165 patients (13.3%) vs. 6 of 119 patients (5%), P = 0.03; relative risk ratio 2.64, 95% confidence interval 1.11–6.32]. However, the incidence of HALT was no longer different at 1 year [42 of 153 patients (27.5%) vs. 22 of 109 patients (20.2%), P = 0.19; relative risk ratio 1.38, 95% confidence interval 0.87–2.18].

In the Evolut Low-Risk substudy that utilized the self-expanding TAVR, Blanke et al.[42] reported similar rates of THV thrombosis between TAVR and SAVR at both 30 days and 1 year. At 30 days, the frequency of HALT was 17.3% in TAVR patients vs. 16.5% in SAVR patients and the frequency of RLM was 14.6% in TAVR patients vs. 14.3% in SAVR patients. At 1 year, the frequency of HALT was 30.9% in TAVR patients and 28.4% in SAVR patients (28.4%) and the frequency of RLM was 31% in TAVR patients vs. 27% in SAVR patients.