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


Antithrombotic therapy post-bioprosthetic valve replacement is a balance between the risks of thromboembolism vs. bleeding. Following TAVR, the American Heart Association/American College of Cardiology (AHA/ACC) guidelines, the ESC/EACTS guidelines, and the latest TAVR expert consensus statement all recommend dual antiplatelet therapy (DAPT) for 3–6 months post-procedure (Table 3).[49–51] For patients with high bleeding risks, the ESC/EACTS guidelines also suggested single antiplatelet therapy (SAPT) may be more appropriate (Class IIb, level of evidence C). This was based on three small randomized controlled trials including 421 TAVR patients comparing DAPT vs. SAPT. These trials suggested that TAVR could be safely performed using SAPT without increased procedural morbidity and mortality, and at 6 months follow-up.[52–54] Subsequent meta-analyses of these three trials with other observational studies suggested that DAPT after TAVR was associated with higher rates of bleeding, but there were no differences in the incidences of death, ischaemic events, myocardial infarction, or strokes.[55,56]

It is possible that TAVR patients may be at higher risk for valve thrombosis compared with SAVR patients, and the routine clinical/research use of MDCT following TAVR has demonstrated a significant proportion of patients develop subclinical THV thrombosis despite DAPT. Often, treating THV thrombosis is based on the clinical judgement of variables such as acuity of patient presentation and severity of symptoms secondary to valve leaflet dysfunction. Clearly, patients presenting with heart failure and mean transvalvular gradient of 60 mmHg should be treated more urgently than those with asymptomatic HALT but normal mean gradients. As open thoracotomy for SAVR was initially deemed inappropriate for these high-risk TAVR patients in the first place and therefore redo-SAVR for THV thrombosis is unlikely to happen, there are two potential treatment strategies for THV thrombosis: conservative surveillance vs. anticoagulation.

Conservative Surveillance

One of the advantages of bioprosthetic valve prosthesis is avoiding anticoagulation associated with mechanical prosthesis. Currently, many TAVR patients have an increased risk of bleeding due to advanced age or multiple comorbidities. Therefore, conservative surveillance of THV thrombosis may be more appropriate for TAVR patients with absent clinical symptoms or haemodynamically significant valve leaflet dysfunction. Sondergaard et al.[24] was first to provide some insights into the natural history of THV thrombosis where 84 patients (61 TAVR and 23 SAVR) from the SAVORY registry underwent two protocol-driven 4D MDCT examinations with unchanged antithrombotic medication between the scans. Baseline and follow-up 4D MDCT were performed at a mean of 140 ± 152 and 298 ± 141 days post-valve implantation, respectively, and all scans were evaluated at core laboratories blinded to baseline variables, antithrombotic medication, clinical events, and outcomes of previous CT scans. At baseline, 38.1% of patients had HALT and 20.2% had HAM. At follow-up, 15.5% of patients had progression of the abnormality, 10.7% showed regression, and 73.8% showed no change. All patients with THV thrombosis were asymptomatic throughout the study duration. On multiple logistic regression, anticoagulation with either vitamin K antagonist (VKA) or non-vitamin K oral anticoagulants (NOAC) was significantly associated with non-progression. No patients on NOAC had progression of THV thrombosis. In the PARTNER 3 CT substudy, HALT spontaneously resolved in 56% of all patients (i.e. both TAVR and SAVR) at 1 year. Vice versa, 21% of all TAVR and SAVR patients developed new HALT at 1 year.[41] These observations suggest that the time course of THV thrombosis is highly variable.


Currently, oral anticoagulation following TAVR is only recommended when there are other indications for anticoagulation such as atrial fibrillation.[49] In patients with surgical bioprosthetic valve thrombosis, the ESC/EACTS guidelines recommend anticoagulation with VKA and/or unfractionated heparin (UFH) as first-line therapy (Class I, level of evidence C).[49] However, there are no guideline treatment recommendations for subclinical THV thrombosis in TAVR patients. It is increasingly recognized that anticoagulation with either VKA or NOAC, not DAPT, reduces the incidence and promotes the regression of THV thrombosis.[21–23,27,33,34,39,48] For example, of the 58 patients who were diagnosed with THV thrombosis from the RESOLVE and SAVORY registries, all 36 patients who were anticoagulated for 3 months (24 with VKA and 12 with NOAC) had restoration of normal leaflet motion on follow-up MDCT.[21] In the remaining 22 patients who were not anticoagulated, 20 patients had either persistent or progressive leaflet abnormality.[21] Other smaller series also reported high proportion of patients with the resolution of THV thrombosis with anticoagulation.[23,27,33,39]

In the recently published GALILEO trial that was specifically designed to evaluate the clinical implications of THV thrombosis, TAVR patients were randomized to experimental low-dose rivaroxaban at 10 mg daily plus aspirin 75–100 mg daily for 3 months, followed by rivaroxaban 10 mg daily monotherapy vs. control aspirin 75–100 mg daily plus clopidogrel 75 mg daily for 3 months followed by aspirin 75–100 mg daily monotherapy.[57] No patient had an underlying baseline indication for chronic anticoagulation. The trial was terminated prematurely in November 2018 after interim analysis due to safety concerns with anticoagulation. At the time of trial termination, only 42% of the total planned 440 primary efficacy events (defined as combined higher death or first thromboembolic event) had occurred. Based on the intention-to-treat analysis, patients randomized to low-dose rivaroxaban had significantly higher risk of death or first thromboembolic event (hazard ratio 1.35, 95% confidence interval 1.01–1.81; P = 0.04) and higher all-cause mortality (hazard ratio 1.69, 95% confidence interval 1.13–2.53). However, the higher all-cause mortality was primarily driven by non-cardiovascular causes. Patients were also more likely to have bleeding complications, and there was a trend towards a significant difference in the primary safety outcome (defined as the composite of life-threatening, disabling, or major bleeding according to the Valve Academic Research Consortium) (hazard ratio with rivaroxaban 1.50, 95% confidence interval 0.95–2.37; P = 0.08).[57] As the study was terminated early with less than half of the projected primary efficacy events, it is difficult to assess the overall risks vs. benefits ratio especially in the context of the GALILEO-4D substudy.

In the GALILEO-4D substudy, 231 TAVR patients underwent 4D MDCT and echocardiograms at 3 months with the primary endpoint of ≥grade 3 (i.e. >50%) RLM.[48] Anticoagulation with rivaroxaban with aspirin was associated with a lower incidence of RLM (between-group difference −8.8%, 95% confidence interval −16.5% to −1.9%; P = 0.01) and HALT (between-group difference −20.0%, 95% confidence interval −30.9% to −8.5%) compared with DAPT. Cross-sectionally, there were no differences in mean transvalvular gradients between both treatment arms at 3 months, between patients with or without ≥grade 2 RLM, between patients with or without ≥grade 3 RLM, or between patients with or without HALT.[48] However, longitudinally (when comparing baseline and follow-up echocardiographic examinations) patients with RLM or HALT were more likely to have ≥5 or ≥10 mmHg increase in mean gradient at follow-up. Overall, the numbers of patients with moderate haemodynamic SVD or clinical thromboembolic events were too small for meaningful interpretation.

Although anticoagulation will result in the resolution of THV thrombosis, the optimal duration of anticoagulation is unknown. When anticoagulation was ceased in patients from the RESOLVE and SAVORY registries, THV thrombosis recurred in 50% of patients after a mean time of 164 ± 109 days.[21] Yanagisawa et al.[26] also reported that the incidences of late THV thrombosis at 6 months, 1 year, 2 years, and 3 years were 7.1%, 11.3%, 12.7%, and 16.9%, respectively. Finally, the median time to THV thrombosis ranged from 5 to 379 days (Table 1 and Table 2).[37,43,46] These data suggest that a longer period of antithrombotic/anticoagulation therapy post-TAVR may be warranted, and it is unclear if the eventual 'protective' re-endothelialization of the bioprosthetic valve is enough to overcome the permanent 'prothrombotic' effects of altered flow (i.e. neo-sinuses, reduced cardiac output) across the valve. On the other hand, routine indiscriminate use of anticoagulation seems inappropriate in these TAVR patients with high bleeding risks as demonstrated in the GALILEO trial.[57] Therefore, future studies are clearly required to determine the optimal antithrombotic regimen after TAVR in terms of specific agent, dose, and duration of therapy.

Finally, the Antiplatelet Therapy for Patients Undergoing Transcatheter Aortic Valve Implantation (POPular-TAVI) trial recently published the results for their cohort B TAVR patients who have an established indication for long-term anticoagulation such as atrial fibrillation.[58] Patients receiving oral anticoagulation alone had a significant lower risk of bleeding compared with the combination of oral anticoagulation plus clopidogrel (21.7% vs. 34.6%, P = 0.01; relative risk ratio 0.63, 95% confidence interval 0.43–0.90). In the trial, the secondary composite endpoint #1 combined the risks of cardiovascular death and thromboembolism (i.e. CVA and myocardial infarction) with bleeding, and the secondary composite endpoint #2 combined the risks of cardiovascular death and thromboembolism without bleeding. For the secondary composite endpoint #1, oral anticoagulation alone was not inferior to combination therapy (31.2% vs. 45.5%, difference −14.3 percentage points, 95% confidence interval for non-inferiority −25.0 to −3.6 percentage points) and was superior to combination therapy (relative risk ratio 0.69, 95% confidence for superiority 0.51–0.92). For the secondary composite endpoint #2, oral anticoagulation alone was not inferior to combination therapy (13.4% vs. 17.3%, difference −3.9 percentage points, 95% confidence interval for non-inferiority) but was not superior to combination therapy (relative risk ratio 0.77, 95% confidence interval for superiority 0.46–1.31). Therefore, compared with the GALILEO trial, the POPular-TAVI trial suggested that selective single agent anticoagulation without antiplatelet therapy can be safely used in TAVR patients with pre-existing indications for long-term anticoagulation and is associated with a lower risk of bleeding.