Subclinical leaflet thrombosis of bioprosthetic aortic valves is characterized by hypoattenuated leaflet thickening (HALT) and reduced leaflet motion of the bioprosthetic valve leaflets noted on 4-dimensional computed tomography (CT). The early reports on subclinical leaflet thrombosis noted this finding to be present in up to 10% to 15% of patients after transcatheter aortic valve replacement (TAVR), more frequently in transcatheter compared with surgical bioprosthetic valves.[2–4] This finding was routinely missed on transthoracic echocardiograms because of the relatively normal aortic valve gradients.[1,2] A few studies also noted an association between subclinical leaflet thrombosis and neurological events (strokes or transient ischemic attacks).[1,2] Anticoagulation was noted to be effective in the prevention of subclinical leaflet thrombosis in both randomized trials and registries1,2,6 and in the treatment of subclinical leaflet thrombosis in registries.[1,2,4] After these initial reports, the US Food and Drug Administration mandated incorporation of CT substudies into the low-risk TAVR trials to understand the natural history of subclinical leaflet thrombosis. At the same time, several industry-sponsored randomized trials were initiated to evaluate the impact of routine anticoagulation after TAVR on clinical outcomes and subclinical leaflet thrombosis. Although many questions concerning subclinical leaflet thrombosis have been answered in randomized clinical trials, several questions remain unanswered (Figure).
Current state of post-TAVR subclinical leaflet thrombosis.
CT indicates computed tomography; DAPT, dual antiplatelet therapy; DOAC, direct oral anticoagulant; and TAVR, transcatheter aortic valve replacement.
In this issue of Circulation, 2 simultaneously published articles (by Fukui et al and Park et al) further our understanding of the impact of transcatheter valve frame geometry on subclinical leaflet thrombosis and the impact of routine anticoagulation with edoxaban after TAVR on subclinical leaflet thrombosis, new cerebral ischemic lesions, and neurocognitive function.
Fukui et al evaluated the association between transcatheter aortic valve prosthesis frame geometry and the presence of HALT and clinical outcomes. In this prospective nonrandomized single-center study of 565 patients undergoing cardiac CT screening for HALT at 30 days after TAVR, the authors analyzed the TAVR valves for deformation of the prosthetic valve, asymmetric prosthesis leaflet expansion, prosthesis sinus volumes, and commissural alignment. The authors propose the novel prosthesis deformation index for balloon-expandable valves, defined as follows: average of the internal stent frame dimension at prosthesis leaflet outflow and leaflet inflow divided by internal stent frame dimension at prosthesis waist. A relatively larger prosthesis deformation index represents a conformation more comparable to an hourglass shape, whereas a smaller index represents a cylindrical orientation. The prosthesis deformation index was not calculated for the self-expanding valves because of the inherent hourglass shape of the self-expanding frame. The authors also evaluate eccentricity of the transcatheter heart valves (both balloon-expandable and self-expanding valves). A relatively larger eccentricity represents a configuration more comparable to an oval shape, and a smaller index that is close to 0 more closely represents a circular expansion. The ideal shape of the balloon-expandable valves is circular expansion in cross section and cylindrical expansion in long axis. For self-expanding valves, the ideal valve expansion is circular expansion in cross section; the longitudinal shape remains hourglass for the self-expanding frame. The prosthesis deformation index was predictive of subclinical leaflet thrombosis for balloon-expandable valves, whereas greater eccentricity of the transcatheter valve frame was predictive of subclinical leaflet thrombosis for self-expanding valves. Asymmetric prosthesis leaflet expansion and smaller prosthesis neo-sinus volumes were predictive of subclinical leaflet thrombosis for both balloon-expandable and self-expanding valves. Commissural alignment, implantation depth, canting, and CT-derived left ventricular stroke volume index were not predictive of subclinical leaflet thrombosis. The presence of HALT in the study was associated with increased risk of mortality, cardiac death, or the composite of all-cause mortality and heart failure hospitalization at 1 year, with adjustment for age, sex, and comorbidities. In both the studies, HALT was not associated with bioprosthetic valve hemodynamics.
The transcatheter heart valves are packaged by the valve manufacturers in the fully expanded configuration, with no valve deformation, no frame eccentricity, symmetric leaflet coaptation, and optimal neo-sinus volumes. These valves are crimped on a catheter from the fully expanded configuration to facilitate delivery by the transcatheter approach. The findings from the study by Fukui et al suggest that the risk of subclinical leaflet thrombosis is the lowest when the transcatheter heart valves achieve full expansion to the precrimping configuration after deployment. Optimal valve expansion can often be achieved by performing predilation and postdilation. However, the efforts to achieve optimal valve expansion to decrease the risk of subclinical leaflet thrombosis should be carefully balanced against the risk of aortic root injury, pacemaker implantation, or strokes associated with postdilation of the transcatheter heart valves. In certain situations, the procedural strategy may be altered to achieve optimal stent frame expansion without compromising the safety of the procedure. For instance, the balloon-expandable SAPIEN 3 valves are often undersized and overexpanded or oversized and underexpanded for a similar range of aortic valve annular area. In such situations, an undersized and overexpanded stent frame may offer a more ideal stent frame expansion compared with an oversized but underexpanded valve frame.
Park et al in a simultaneous publication report the results of a randomized trial of routine anticoagulation with edoxaban compared with dual antiplatelet therapy (DAPT) after TAVR and evaluate the causal relationship of subclinical leaflet thrombosis with cerebral thromboembolism and neurological or neurocognitive dysfunction. In this study, the incidence of subclinical leaflet thrombosis was lower, although not statistically significantly, in the edoxaban group compared with the DAPT group (9.8% versus 18.4%; P=0.076). There was no significant difference in the proportion of patients with new cerebral lesions on brain magnetic resonance imaging (edoxaban versus DAPT, 25.0% versus 20.2%), and median total new lesion number and volume were not different between the 2 groups. In addition, no significant association was observed between the presence or extent of leaflet thrombosis and new cerebral lesions or a change of neurological or neurocognitive function.
The ADAPT-TAVR study (Anticoagulant Versus Dual Antiplatelet Therapy for Preventing Leaflet Thrombosis and Cerebral Embolization After Transcatheter Aortic Valve Replacement) effectively rules out any favorable impact of routine anticoagulation after TAVR on new cerebral lesions or neurocognitive changes. However, this study does not definitively establish or exclude a causal association between subclinical leaflet thrombosis and new cerebral lesions or neurocognitive changes, given the dynamic nature of subclinical leaflet thrombosis. In the natural history CT substudy of TAVR with the balloon-expandable SAPIEN 3 valves (Edwards LifeSciences, Irvine, CA) in patients at low surgical risk, HALT was noted to be a dynamic finding, with spontaneous resolution from 30 days to 1 year in 54% of patients and spontaneous appearance of new HALT from 30 days to 1 year in 21% of patients. Similar findings were noted in the CT substudy of TAVR with the self-expanding Evolut valve (Medtronic, Minneapolis, MN) in patients at low surgical risk. Patients in the ADAPT-TAVR study underwent CTs at a single time point (6 months). It is quite likely that a proportion of patients who did not have subclinical leaflet thrombosis at 6 months in the ADAPT-TAVR study had subclinical leaflet thrombosis at 30 days, with spontaneous resolution by 6 months. A definitive assessment of the impact of subclinical leaflet thrombosis on new cerebral lesions can be performed only in a study performing serial cardiac CTs and brain magnetic resonance images at multiple time points after TAVR, comparing patients who have no HALT at any time point during the study with those who have HALT at any time point on serial CTs.
Anticoagulation is not 100% effective in the prevention of subclinical leaflet thrombosis. A small proportion of patients even on anticoagulation develop subclinical leaflet thrombosis. For instance, 9.8% of patients treated with edoxaban in the ADAPT-TAVR study developed subclinical leaflet thrombosis. Despite the dynamic nature of subclinical leaflet thrombosis, a proportion of patients continue to experience persistence or progression of the severity of subclinical leaflet thrombosis, progress to clinical valve thrombosis, or develop recurrent subclinical leaflet thrombosis after discontinuation of anticoagulation. The deformation of the TAVR prosthesis, asymmetric prosthesis leaflet expansion, or small prosthesis sinus volumes may be responsible for the persistence, progression, or recurrence of subclinical leaflet thrombosis and need to be systematically studied in future studies.
Fukui et al noted an increased risk of death, cardiac death, and heart failure hospitalization in patients with HALT. Park et al did not notice an association between subclinical leaflet thrombosis and new cerebral lesions on magnetic resonance imaging or change in neurocognitive function. Despite these 2 studies, the clinical significance of subclinical leaflet thrombosis remains uncertain. Previous studies have not reported an association between subclinical leaflet thrombosis and death or heart failure hospitalizations. Data on the association between subclinical leaflet thrombosis and the risk of stroke, TIA, or thromboembolic complications have been conflicting.
The randomized clinical trials of routine anticoagulation with rivaroxaban 10 mg once a day (GALILEO trial [Global Study Comparing a Rivaroxaban-Based Antithrombotic Strategy to an Antiplatelet-Based Strategy After Transcatheter Aortic Valve Replacement to Optimize Clinical Outcomes]) or apixaban 5mg twice a day (ATLANTIS trial [Anti-Thrombotic Strategy After Trans-Aortic Valve Implantation for Aortic Stenosis]) compared with DAPT reported increased risk of death or thromboembolic complications with routine anticoagulation compared with antiplatelet therapy in the absence of an established indication for anticoagulation. Both GALILEO and ATLANTIS noted a significantly lower incidence of valve thrombosis in patients treated with routine anticoagulation after TAVR. The similar efficacy of edoxaban on subclinical leaflet thrombosis noted in the ADAPT-TAVR study further confirms that the effectiveness of direct oral anticoagulants for the prevention of subclinical leaflet thrombosis is likely a class effect, regardless of the type of direct oral anticoagulant used. Whether routine anticoagulation in patients with adverse stent frame geometry, as described in this study, would yield different results remains to be seen. There were not enough clinical events in the ADAPT-TAVR study to make any definitive conclusions on the impact of anticoagulation with edoxaban on clinical outcomes after TAVR.
In summary, these 2 studies address a few important missing aspects of the subclinical leaflet thrombosis puzzle. First, optimal TAVR stent expansion or stent geometry is important. This should fuel research for optimal valve deployment techniques, better stent designs, and clinical trials to test these variables. Second, yet another direct oral anticoagulant resulted in a reduction in subclinical leaflet thrombosis, suggesting that the role of anticoagulation in the prevention of subclinical leaflet thrombosis may be a class effect, regardless of the type of direct oral anticoagulant used. Third, routine anticoagulation (with edoxaban) was not associated with a reduction in new cerebral lesions, at least up to 6 months. The authors need to be commended on undertaking these important studies to further our understanding of subclinical leaflet thrombosis.
Among the questions that remain unanswered, the following are particularly relevant. First, is the story of routine anticoagulation after TAVR over? TAVR is now being performed in younger patients in their mid-60s or 70s, who may have better risk-benefit profile with anticoagulation. The mean age of patients in the 2 studies and in the anticoagulation trials, including GALILEO and ATLANTIS trials, was ≥80 years. The results of the anticoagulation trials in the elderly may not be applicable to younger patients with lower bleeding risk and potentially better tolerance for anticoagulation. Second, does subclinical leaflet thrombosis affect valve durability? We will have the opportunity to study this in the next few years as we treat younger patients who are expected to outlive the valve. More rigorous registries and randomized trials with long-term follow-up and serial CT scans are needed to evaluate the impact of subclinical leaflet thrombosis on clinical outcomes and valve durability.
Sources of Funding
Circulation. 2022;146(6):494-497. © 2022 American Heart Association, Inc.