Leaflet thrombosis in transcatheter aortic valves implants: a new marker of risk?

This editorial refers to ‘Influence of prosthesis related factors on the occurrence of early leaflet thrombosis after transcatheter aortic valve implantation’, by P. Breitbart et al., pp. 1082--1089.

Leaflet thrombosis (LT) in transcatheter valve implants (TAVR) is of concern as it may cause thromboembolism,^1,2 symptomatic obstruction,³ and reduced durability.^2–5 Durability is a key concern as the use of TAVR extends to intermediate- and lower-risk patients who are expected to have a longer life-span than the original high-risk cohorts.

LT may present as ‘silent’ thickening of the valve on computed tomography (CT) with or without mild restriction of leaflet motion. This occurs in about 10–18% TAVR compared with 4–7% of surgical replacement valves (SAVR).^1,5,6 The incidence is similar for all types of TAVR, whether balloon- or self-inflating⁶ other than the Portico for which an incidence as high as 40% has been reported.¹ LT affects one leaflet in three-fourths of cases, two leaflets in one-fourth and occasionally all three.¹ There may be elevated transvalvular gradients⁵ and these may progress^1,3,4,7 if no anticoagulation is given. Structural valve deterioration (SVD), defined by a rise in mean gradient alone, occurs in 15–20%^3,5 with LT and is associated with exertional breathlessness in 10%.³

The higher incidence of LT in TAVR compared with SAVR is unlikely to be an artefact of the timing of routine CT or selection bias. It is likely to be explained by: (i) differences in leaflet morphology which are thinner, 0.25 mm in TAVR compared with 0.4 mm in SAVR,⁸ and which may be damaged from crimping⁹ or balloon expansion¹⁰; (ii) increased stresses on the leaflets caused by the non-expansile TAVR stent, and the potential for asymmetric deployment⁸; and (iii) increased stasis downstream because of the presence of native valve tissue.^11,12 Patient factors, for example left ventricular (LV) impairment, age, and high body mass index may also contribute but probably equally in SAVR and TAVR.^2,4

In this issue, Breitbart et al.¹² examine the effect of TAVR expansion on the incidence of LT. They compared 55 patients with LT and two groups without LT, 85 from a sponsored study and a separate group of 55 matched patients, all receiving SAPIEN 3 valves. They performed CT scans before and after implantation and fused the images to produce a 3D image of the TAVR within the native root. They measured the cross-sectional area of the TAVR in the LV outflow tract, at its minimum ‘waist’ and in the outflow region in the aorta. Expansion was calculated as the mean of these three values expressed as a percentage of the nominal area. Waisting was calculated as the difference between the central area and the average of the inflow and outflow areas. They showed that LT was associated with reduced expansion leading to lower cross-sectional areas at all three levels and therefore a smaller waist. The optimal cut-point waist was 6.8% (sensitivity 87% and specificity 67% for the presence of LT). There were no significant differences between the groups in the rates of re-inflation and re-ballooning, in calcification scores in the landing zone, or the implantation depth. An analysis of the SAPIEN 3 in the RESOLVE trial¹³ also showed reduced waist as a risk for LT although measured differently as the difference between nominal and actual diameter.

How is the effect of under-expansion explained? Under- or asymmetric expansion is known to increase the stresses normally found at the commissures¹⁴ and also to extend high stresses to the belly of the leaflet.^8,14 Under- or asymmetric expansion has been linked to LT on pathological examination.¹⁵ A TAVR is implanted within the native valve and does not replace it. There is, therefore, a native sinus and a neo-sinus and asymmetric expansion may interfere with communication between the native and neo sinuses¹³ to encourage thrombus formation. By comparison, SAVR have symmetrical expansile stents with no possibility of incomplete or asymmetric expansion. The native valve tissue is excised before replacement so there is no native sinus and much less stasis downstream compared with a TAVR.^8,11,13

Anticoagulation is another important determinant of LT. Breitbart et al.¹² showed that only 10.9% with LT were taking anticoagulants compared with 43.6% in the matched control group without LT. A lower incidence of LT on oral anticoagulants, either vitamin K antagonists or novel oral anticoagulants, is a universal finding.^1–4 LT regresses if oral anticoagulants are started^1,2 but progresses on antiplatelet therapy.^3,4,7 LT recurs after oral anticoagulation is stopped and may then affect previously normal leaflets.⁷ This underlines that there is an enduring substrate and raises the possibility that anticoagulation should be recommended routinely for all TAVR.

However, routine anticoagulation would introduce an unacceptable risk of bleeding. Instead, it might be reasonable to anticoagulate only those in whom LT is shown. Unfortunately, LT can only reliably be detected by CT and not by transthoracic echocardiography and, although most common in the first 3 months, may occur at any time in the first year.⁷ The timing of a routine CT is therefore hard. The importance of the study by Breitbart et al.¹² is to suggest that reduced waisting immediately after implantation may be a marker of future LT. A randomized trial could be done of oral anticoagulation in TAVR with reduced waists controlling for other known risks including age and low LV ejection fraction.

The study had limitations. It is a small, retrospective examination of prospectively gathered data. The patients with LT and comparators without LT were separate groups and not consecutive patients. We do not know the incidence of a reduced waist in the whole population of TAVR patients. Clearly, the results may not be generalizable to TAVR other than the SAPIEN 3.

LT is the ‘elephant in the room’ of TAVR. It may contribute to early SVD^2–5 and yet CT is not performed as a clinical routine. It is true that judging the timing of CT to show LT is hard. Breitbart et al.¹² suggest an alternative approach of using markers of expansion to identify higher-risk patients for closer follow-up both clinically and in trials.

Conflict of interest: none declared.

The opinions expressed in this article are not necessarily those of the Editors of EHJCI, the European Heart Rhythm Association or the European Society of Cardiology.

References