LAUSR

www.thelancet.com Published online March 19, 2017 http://dx.doi.org/10.1016/S0140-6736(17)30764-X 1 Excellent outcomes of transcatheter aortic valve replacement (TAVR) have been experienced by patients with aortic stenosis at high and intermediate risk of surgery. Findings from large randomised trials have shown survival with TAVR that is similar to or improved compared with bioprosthetic surgical aortic valve replacement (SAVR), and very low stroke rates have been observed with new-generation devices. Investigators of echocardiographic follow-up studies have consistently reported low transvalvular gradients up to 5 years after TAVR and SAVR, with slightly greater aortic valve areas after TAVR than after SAVR. Against this background, the occurrence of subclinical valve leaflet thrombosis in patients, detected with CT after TAVR or SAVR, has been described. In The Lancet, Tarun Chakravarty and colleagues report data from two large registries (SAVORY and RESOLVE) of 890 patients undergoing TAVR or SAVR with follow-up CT (626 [70%] in the RESOLVE registry and 264 [30%] in the SAVORY registry). Masked analyses of all CT scans, echocardiograms, and neurological events were done. Subclinical leaflet thrombosis, defined as moderate or severe restriction of leaflet motion with corresponding CT-derived hypoattenuating lesions, was detected in 106 (12%) patients, including five (4%) of 138 who had SAVR and 101 (13%) of 752 who had TAVR (p=0·001). A greater proportion of patients with subclinical leaflet thrombosis had aortic valve gradients of more than 20 mm Hg and increases in aortic valve gradients of more than 10 mm Hg (12 [14%] of 88) than did those with normal leaflet motion (seven [1%] of 632; p<0·0001). Leaflet thrombosis was less frequently observed in patients using warfarin or novel oral anticoagulants (NOACs; eight [4%] of 224) than in those using dual antiplatelet or monoantiplatelet therapy (98 [15%] of 666; p<0·0001). Subclinical leaflet thrombosis was associated with development of non-procedural stroke or transient ischaemic attack during follow-up (5·71 vs 1·75 events per 100 person-years; p=0·004). Several issues warrant discussion to put these results into perspective. Dynamic four-dimensional CT imaging was used for detection of subclinical thrombosis. Consensus definitions and quantification of leaflet thrombosis with CT need to be established before prospective study and clinical use. A second issue relates to the discrepancy between CT and echocardiographic findings. Investigators of previous small studies reported a 10–15% prevalence of subclinical thrombosis with CT, whereas elevated gradients (a mean gradient of >20 mm Hg) with echocardiography were infrequent. Similar findings were present in the large RESOLVE and SAVORY registries. These observations imply that CT detects early subclinical thrombosis, whereas echocardiography detects the late consequences of thrombosis—ie, valvular stenosis. These results also indicate that not all thrombosis results in valve degeneration and stenosis—ie, early thrombosis might resolve without permanent clinical sequelae. The optimal CT timing after valve implantation to detect meaningful leaflet thrombosis is thus unknown. The timing of imaging might also affect the proportions of leaflet thrombosis with different valve types (ascertainment bias). CT scans in this study were obtained earlier after TAVR (median 58 days [IQR 32–236]) than after SAVR (163 [79–417]). Although time from implantation to CT was not an independent correlate of leaflet thrombosis, given residual confounding, this difference in timing might partly underlie the lower proportion of leaflet thrombosis detected with SAVR than with TAVR. Review of the distributions of time to CT versus leaflet thrombosis might provide additional insight. Other unmeasured confounders (eg, frailty or immunological factors) might also have predisposed the TAVR group to a higher proportion of leaflet thrombosis. However, the authors describe intrinsic structural, manufacturing, and functional differences between surgical and transcatheter valves, which might differently affect valve predisposition to thrombosis. The proportions of subclinical leaflet thrombosis varied from 0% to 30% with different transcatheter valves. Given the small sample size of each valve type studied (including surgical valves), as well as differences in patient characteristics, anticoagulation regimens, and timing of imaging, this study cannot be used to draw conclusions that different valves cause different proportions of leaflet thrombosis. Bioprosthetic surgical and transcatheter heart valve thrombosis