Author + information
- Received June 21, 2016
- Revision received September 12, 2016
- Accepted October 20, 2016
- Published online January 16, 2017.
- Sonny Palmer, MBBS, DMedScia,b,∗ (, )
- Nicholas Child, MBBS, PhDa,
- Mark A. de Belder, MA, MDa,
- Douglas F. Muir, MDa and
- Paul Williams, MDa
- aThe James Cook University Hospital, Middlesbrough, United Kingdom
- bSt Vincent’s Hospital, Melbourne, Australia
- ↵∗Reprint requests and correspondence:
Dr. Sonny Palmer, St Vincent’s Hospital Melbourne, Department of Cardiology, Melbourne, Victoria, Australia.
Objectives The aim of this study was to describe the incidence and clinical impact of left atrial appendage thrombus (LAAT) in a population referred for transcatheter aortic valve replacement (TAVR) and to examine the role of cardiac computed tomography (CCT) in the diagnosis of LAAT.
Background Atrial fibrillation is common in patients undergoing TAVR. Embolization of LAAT is a potential mechanism of periprocedural stroke. The incidence and clinical impact of LAAT in a TAVR cohort have not been reported, and the optimal method for diagnosing LAAT remains unclear.
Methods Dual-phase cardiac computed tomographic scans were examined for the presence of LAAT in 198 consecutive patients referred for consideration of TAVR. Findings on CCT were compared with those on transesophageal echocardiography (TEE) when both modalities were available.
Results The incidence of LAAT on CCT was 11% in the overall cohort and 32% in patients with atrial fibrillation. Two patients (1.6%) had LAAT on CCT but were not known to have histories of atrial fibrillation. Ninety-eight patients also underwent TEE. Compared with TEE, CCT had sensitivity and specificity of 100% and 98%, respectively, and a negative predictive value of 100%. In the 124 patients who underwent TAVR, the in-hospital stroke rate was 4.8%. The risk for stroke appeared higher in patients with LAAT (20% [2 of 10]) compared with patients without LAAT (3.8% [4 of 105]).
Conclusions The incidence of LAAT in patients considered for TAVR is high, and LAAT embolization may represent a clinically relevant cause of periprocedural stroke. Dual-phase CCT is an accurate modality for the diagnosis of LAAT. It may obviate the need for pre-procedural TEE. The presence of LAAT should be examined in all patients undergoing TAVR and strategies developed for those patients in whom LAAT is identified.
- cardiac computed tomography
- left atrial appendage thrombus
- transcatheter aortic valve replacement
- transesophageal echocardiography
Atrial fibrillation (AF) is the most common chronic arrhythmia and is an important modifiable risk factor for stroke (1–4). It is associated not only with increased mortality in stroke (1–5) but also ischemic heart disease (1–7), heart failure (5,8), and valvular heart disease (9). Formation and embolization of left atrial appendage thrombus (LAAT) is the dominant mechanism for the increased stroke risk in patients with AF, and the prevalence of LAAT in these patients has been reported to be between 5% and 15% (10). As the incidence of AF increases with age (11), it is not surprising that AF is common in patients currently referred for transcatheter aortic valve replacement (TAVR) with a reported prevalence of between 30% and 40% (12–15).
Stroke remains a feared complication of TAVR, with a reported incidence of up to 5% at 30 days (12,13). The etiology of periprocedural cerebrovascular accident is likely multifactorial, and mechanisms include embolization of calcified material, dislodgement of atheromatous debris during wire and device manipulation, thrombus formation, and cerebral hypoperfusion during rapid ventricular pacing. However, embolization of LAAT has recently been described during TAVR and may represent an additional mechanism of periprocedural stroke (16) (Figure 1).
Despite the high rate of AF in the TAVR population, the incidence and clinical outcomes of LAAT have not been previously reported, and the presence of LAAT was an exclusion criterion in the PARTNER (Placement of Aortic Transcatheter Valves) I study (12). Furthermore, the optimal method for diagnosing LAAT remains unclear. Cardiac computed tomography (CCT) has emerged as a reliable alternative to transesophageal echocardiography (TEE) in patients with AF undergoing cardioversion, AF ablation, or evaluation for a potential cardioembolic focus (10), but it has not been examined in a TAVR population.
Thus, we aimed to describe the incidence and clinical impact of LAAT in a population referred for TAVR and to examine the role of CCT in the diagnosis of LAAT.
CCT became a routine investigation for patients referred for consideration of TAVR at our institution in July 2013. All cardiac computed tomographic scans performed on consecutive patients between July 2013 and October 2015 were examined for the presence of LAAT. The images of all patients who also underwent TEE were reviewed.
Baseline clinical characteristics and procedural details were collected, including patient demographics, burden of AF and risk scores, the prescription of anticoagulant therapy, technical aspects of the procedure, and clinical outcomes. TAVR was performed with the Edwards SAPIEN XT and SAPIEN 3 devices (Edward Lifesciences, Irvine, California).
Clinical outcomes, such as mortality, vascular complications, bleeding, and stroke or transient ischemic attack, were diagnosed and defined as per the Valve Academic Research Consortium–2 guidelines (17). Periprocedural stroke was any stroke that occurred during the index admission. All patients underwent clinical review at 6 weeks post-TAVR and then underwent follow-up echocardiography at 6 months. Mortality was tracked independently by the Office of National Statistics. Mortality and clinical outcomes were also collected as per the British Cardiovascular Intervention Society and UK TAVI (United Kingdom Transcatheter Aortic Valve Implantation) database.
CCT was performed with a second-generation dual-source computed tomographic scanner (Somatom Definition Flash, Siemens Medical Solutions, Erlangen, Germany) in the craniocaudal direction during a single breath-hold.
Beta-blockers were used at the discretion of the imaging cardiologist. A test bolus was used before image acquisition. Twenty milliliters of a noniodinated contrast agent, Omnipaque (350 iodine mg/ml; GE Healthcare, Cork, Ireland), was administered using a power injector (Ulrich Medical, CT Emotion, Ulm, Germany) at a rate of 6 ml/s through an 18-gauge cannula inserted into a forearm vein. After the administration of the contrast agent, 30 ml normal saline was administered at a rate of 6 ml/s through the same forearm vein. A region of interest was plotted against the aorta and a geometric curve was acquired. The curves were analyzed and the time taken to maximal enhancement was measured to determine the optimal scan delay. Subsequently, using electrocardiographic gating, a bolus of 60 ml of Omnipaque, followed by 30 ml normal saline solution, was administered intravenously at a rate of 6 ml/s using the power injector. Our institution’s TAVR protocol consisted of 2 acquisitions following the single bolus. The exact timing of the 2 acquisitions depended on patient factors determined at the time of the test bolus. The first scan was a retrospective acquisition at approximately 20 seconds post-bolus. The second flash acquisition occurred approximately 40 seconds post-bolus. All patients underwent computed tomographic studies consisting of both these cardiac and aortic acquisitions.
The computed tomographic parameters were as follows: detector collimation, 128 × 0.6 mm; slice acquisition, 2 × 128 × 0.6 mm by means of a z-flying focal spot; gantry rotation time, 280 ms; tube voltage, 100 to 120 kV; tube current–time product, 280 to 380 mAs; and pitch, 0.28 to 3.4, adapted according to heart rate. All retrospective studies were reconstructed at approximately 70% of the R-R interval (best end-diastolic phase) with a slice thickness of 0.6 mm and a reconstruction increment of 0.4 mm using a cardiac kernel (I30f).
Two experienced cardiologists independently reviewed the cardiac computed tomographic images of all patients. Each cardiologist was blinded to the clinical history. A consensus opinion was reached when interpretation differed (1 case).
The presence of LAAT was graded as definite, absent, or possible. Definite thrombus was defined as a filling defect in both phases of imaging (retrospective and flash acquisitions) (Figures 2A and 2B). Absent thrombus was defined as an absence of a filling defect in both phases of imaging or when the retrospective phase was suspicious for LAAT, but the subsequent flash phase did not demonstrate a filling defect (Figures 2C and 2D). Possible thrombus was diagnosed on any scan in which LAAT could not be excluded. This may have been due to several factors, including the presence of pectinate muscles, movement artefact, and reduced opacification of the entire left atrial appendage.
Standard transesophageal echocardiographic views (Epiq 7, Phillips Healthcare, Eindhoven, the Netherlands) were used to confirm the severity of aortic stenosis, determine annular size when required, evaluate for concomitant valve disease, and visualize the left atrial appendage and identify LAAT and/or spontaneous echocardiographic contrast. All images were transferred to Prosolv (FUJIFILM Medical Systems, Chicago, Illinois) and analyzed offline.
Two experienced cardiologists independently reviewed the transesophageal echocardiographic images. Each cardiologist was blinded to the clinical history and the results of CCT. On TEE, thrombus was defined as a well-circumscribed, homogeneous, echo-reflective mass (Figure 3). A consensus opinion was reached when interpretation differed (2 cases).
Data were analyzed using SPSS version 23 (SPSS, Chicago, Illinois). Continuous variables are expressed as mean ± SD. Categorical variables are reported as frequencies and percentages. Comparisons between categorical variables were evaluated using the Fisher exact test or the Pearson chi-square test as appropriate. Using TEE as the reference standard, the sensitivity, specificity, positive predictive value, negative predictive value, and accuracy of CCT for detecting LAAT were calculated. A p value < 0.05 was considered to indicate statistical significance.
One hundred ninety-eight patients with severe symptomatic aortic stenosis were referred for consideration for TAVR and underwent CCT during the study period. The baseline characteristics and echocardiographic parameters of the patients are depicted in Table 1. The overall study cohort represented a high-risk population with a mean age of 83 years and a mean logistic European System for Cardiac Operative Risk Evaluation score of 25. Sixty-nine patients (35%) were in AF, with a mean CHA2DS2-VASc score of 4.0.
Definite LAAT was identified on CCT in 22 patients (11%), and there was no evidence of thrombus in 166 patients (84%). Thrombus could not be excluded on CCT in the remaining 10 patients (5.1%). Thus, 188 patients (95%) had diagnostic CCT. AF was a significant risk factor for definite LAAT identified on CCT, which was present in 20 of 63 patients (32%) with AF compared with 2 of 125 (1.6%) without histories of AF (odds ratio: 19.8; 95% confidence interval: 4.5 to 88; p < 0.0001). The incidence of previous stroke in the overall cohort was 21.7%. There was no significant difference in any other baseline characteristic between those with LAAT and those without.
Comparison with TEE
Ninety-eight patients (49%) from the overall cohort underwent TEE (Table 2). There was agreement between the imaging modalities in 96 patients (98%). In 2 patients, definite LAAT was diagnosed on CCT but was not seen on TEE. Both patients had histories of AF and received anticoagulation. In 1 of these patients, spontaneous echocardiographic contrast was present on TEE. There were no cases of LAAT identified on TEE and a negative finding on CCT.
Using TEE as the reference standard, the overall sensitivity and specificity of CCT for the detection of definite LAAT were 100% and 98%, respectively. The positive and negative predictive values were 75% and 100%, respectively, and the diagnostic accuracy was 98%.
One hundred twenty-four of the patients (63%) referred to the heart team for consideration for aortic valve intervention subsequently underwent TAVR. During the study period, the procedural success rate was 100%. Thirty-day mortality was 2.4% (3 of 124), the rate of major vascular complications was 2.4% (3 of 124), the rate of life-threatening bleeding was 1.6% (2 of 124), and the rate of major bleeding was also 1.6% (2 of 124). The periprocedural pacemaker insertion rate was 4.0% (5 of 124). In this cohort, histories of AF were present in 42 patients (34%), and the mean CHA2DS2-VASc score was 4.2. Twenty patients received anticoagulation with warfarin and 9 with novel oral anticoagulant agents; the remainder did not undergo anticoagulation.
Cardiac computed tomographic analysis identified definite LAAT in 10 patients (8.1%) and no evidence of thrombus in 105 patients (85%). Possible LAAT was identified in the remaining 9 patients (7.3%). All 10 patients with definite LAAT on CCT had histories of AF. Eight of these patients were receiving oral anticoagulant agents. Of the 9 patients with possible thrombus on CCT, 6 had histories of AF and 3 had no histories of AF.
The overall stroke rate in this TAVR cohort was 4.8% (n = 6). One case occurred with transapical SAPIEN XT valves, 3 cases with transfemoral SAPIEN XT devices, and the 2 remaining cases with transfemoral SAPIEN 3 valves. Two of these patients had definite LAAT on CCT. Both of these patients had clinical diagnoses of AF and were therapeutically anticoagulated with warfarin at the time of the CCT. They both had stopped warfarin 5 days prior to the TAVR procedure. The first patient had a CHA2DS2VASc score of 6 and underwent a transapical procedure (SAPIEN XT) and was managed conservatively. The second patient had a CHA2DS2VASc score of 5 and underwent a transfemoral procedure (SAPIEN 3). This patient underwent periprocedural thrombolysis without complication.
The remaining 4 patients who experienced strokes had no LAAT on CCT; none of these patients had AF diagnoses or were receiving anticoagulation. Therefore, the stroke rate was 20% (2 of 10) in patients with LAAT compared with 3.8% (4 of 105) in those without LAAT. Stroke management for each patient was delivered as per the national stroke guidelines. Overall, there was 1 disabling stroke among the 6 patients.
Eight patients (80%) with LAAT on CCT went on to undergo TAVR successfully without clinical evidence of a procedural stroke. Six of these patients received anticoagulation (2 with warfarin and 4 with novel oral anticoagulant agents), and 2 were not receiving anticoagulation.
The key finding of this study is that LAAT is common in TAVR patients and was found in 11% of all patients referred for TAVR and in 32% of those with prior diagnoses of AF. Second, the presence of LAAT was associated with a higher risk for procedural stroke. Finally, in line with other studies, we show that CCT is a reliable imaging modality for the diagnosis of LAAT in a TAVR population.
The high incidence of AF (35%) seen in our cohort is in line with previous TAVR studies and reflects the older valve disease population that is currently considered suitable for treatment (12,13,15). The incidence of LAAT in TAVR patients has not previously been reported. In this study, LAAT was present in one-third of patients with AF, which is substantially higher than that reported in other series of patients with AF. For instance, in a recent meta-analysis of 19 studies, the mean incidence of LAAT in patients with AF was 8.9% (10). The incidence of thrombus depends on the patient cohort studied: in the Romero et al. (10) study, the incidence was typically <5% in a patients referred for pulmonary vein isolation but higher in those being investigated for a potential cardioembolic focus for stroke.
The reasons for the high incidence of LAAT in TAVR patients with AF are likely multifactorial. These patients by definition have valvular disease, and many will have histories of left ventricular systolic dysfunction. They are also usually older, and there are high rates of hypertension, vascular disease, and previous stroke, all of which are strong predictors of stroke risk in patients with AF. This is supported by the high CHA2DS2-VASc scores seen in our study.
We have previously demonstrated that LAAT embolization can occur during a TAVR procedure (16). There are several potential mechanisms for this phenomenon. First, rapid ventricular pacing for balloon aortic valvuloplasty or TAVR deployment may predispose to thrombus expulsion. Second, inadvertent wire passage from the left ventricle through the mitral valve apparatus into the left atrium may directly embolize thrombus from the appendage. Finally, LAAT may embolize if the patient requires cardioversion during the procedure, for instance because of the development of a ventricular arrhythmia. In the cohort of patients who underwent TAVR in our study, the rate of stroke was higher in patients with LAAT (20%) than in those without LAAT (3.8%). Although the absolute number of events is small and definitive conclusions cannot be drawn, this finding does raise the possibility that LAAT embolization may represent a clinically important cause of procedural stroke.
Given this potential risk, we recommend that investigation for LAAT should be performed in all patients planned for TAVR. Although LAAT was identified mostly in patients in whom diagnoses of AF had already been established, there were 2 patients (1.6%) in whom LAAT was identified in the absence of a diagnosis of AF. Neither of these patients received anticoagulation, and it may be that these patients had occult or paroxysmal AF. Nonetheless this reinforces the importance of examining all patients for LAAT, not just those with known AF diagnoses. Furthermore, although the yield is low, clinicians should consider a Holter monitor for the diagnosis of occult AF.
Although oral anticoagulation is clearly recommended as the first-line therapy for patients with AF, LAAT is often identified despite the use of anticoagulant agents. For instance, in our TAVR cohort, 80% of the patients with LAAT were already receiving anticoagulation, and 2 strokes occurred in patients with LAAT receiving anticoagulation. Anticoagulation may also be contraindicated in some patients. Strategies are therefore required to manage patients with LAAT. In these cases we postpone TAVR and review the patient’s anticoagulation regimen and medication compliance. Warfarin using a higher target international normalized ratio (e.g., 2.5 to 3.5) or the use of a novel oral anticoagulant agent may be considered. We then repeat imaging of the left atrial appendage between 3 and 4 weeks to see if the thrombus has resolved (18).
In patients with persistent thrombus despite this treatment, a decision must be made as to whether to proceed with TAVR. In our study, 8 patients (80%) with LAAT underwent TAVR successfully and did not experience clinically significant strokes. Given the impressive symptomatic and prognostic gains from TAVR, we believe that this therapy should not be withheld, but patients with LAAT should be informed that there may be an increased risk for stroke. Strategies that may reduce the procedural risk for embolization should be considered. Periprocedural bridging therapy with low–molecular weight heparin may be adopted. Cerebral embolic protection devices such as Sentinel (Claret Medical, Santa Rosa, California) (19) or TriGuard (Keystone Heart, Caesarea Business Park, Israel) (20) may be considered in patients with suitable anatomy. The duration of rapid pacing should also be minimized, for instance by avoiding pre-implantation balloon aortic valvuloplasty (21,22) and considering the use of a self-expanding TAVR device without the need for rapid pacing.
Use of CCT to diagnose LAAT
We examined the use of CCT to diagnose LAAT in a TAVR population. TEE has historically been considered the gold standard for LAAT.
CCT is emerging as a viable alternative to TEE for the diagnosis of LAAT. In the recent meta-analysis by Romero et al. (10) comparing CCT with TEE for the diagnosis of LAAT, the overall sensitivity and specificity were 96% and 92%, respectively, with a negative predictive value of 99%. Specificity has been limited because blood stasis (analogous to spontaneous echocardiographic contrast on TEE) in the left atrial appendage can appear as a filling defect and be mistaken for thrombus. With the development of delayed-phase imaging, the rate of false-positive results can be reduced. In a subanalysis of studies from the Romero et al. (10) meta-analysis in which delayed-phase acquisition was performed, diagnostic accuracy improved substantially to give sensitivity and specificity of 100% and 99%, respectively, and positive and negative predictive values of 92% and 100%, respectively.
In our study, we used a delayed-phase protocol and obtained diagnostic-quality images in 95% of patients. Figure 2 illustrates the utility of delayed-phase imaging in improving the specificity of the scan. Using TEE as the reference standard, the overall sensitivity and specificity were 100% and 98%, respectively, and the positive and negative predictive values were 75% and 100%, respectively. These values are very similar to those reported in other AF populations and suggest that CCT is also an accurate imaging modality for LAAT in a TAVR population, including in those with AF, in whom image quality can be influenced by heart rate variability.
CCT is now recommended as a routine investigation prior to TAVR to assess annular dimensions, coronary location, and iliofemoral access (23,24). The presence of LAAT can be routinely examined in all patients. Given the excellent negative predictive value if delayed-phase CCT gives good annular imaging and there is no LAAT, then pre-procedural TEE may not be necessary, simplifying the work-up of TAVR patients. TAVR is increasingly performed under local anesthesia with conscious sedation, with which TEE cannot readily be performed, and therefore a large proportion of patients may not require TEE at all. For instance, in our study, 84% of our patients did not have LAAT on CCT and potentially could have avoided TEE altogether.
There were several limitations to this study. First, this was a single-center and retrospective analysis. Second, not all patients underwent both CCT and TEE, leading to potential selection bias. Third, the relatively small study size means that definitive conclusions on the impact of LAAT on clinical outcomes cannot be drawn.
The incidence of LAAT in the population currently referred for TAVR is very high, and the presence of LAAT may be associated with an increased risk for periprocedural stroke. CCT is an accurate alternative imaging modality to TEE for the diagnosis of LAAT, particularly if delayed-phase imaging is available, and may eliminate the need for TEE in many patients. Assessment for LAAT should be performed routinely in all patients referred for TAVR and strategies developed for those in whom LAAT is identified. Further larger prospective studies are required to confirm whether the presence of LAAT is an independent risk factor for procedural stroke and to examine the efficacy of potential avoidance strategies.
WHAT IS KNOWN? Stroke is a feared complication of TAVR. Embolization of LAAT is a potential mechanism of periprocedural stroke.
WHAT IS NEW? The incidence of LAAT in 198 patients considered for TAVR was high (11%). The in-hospital stroke rate was 4.8%. Stroke risk was higher in the presence of LAAT (20%) than without (3.8%). Dual-phase CCT is an accurate modality for the diagnosis of LAAT.
WHAT IS NEXT? The presence of LAAT should be examined in all patients undergoing TAVR and strategies developed for those patients in whom LAAT is identified.
The authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- atrial fibrillation
- cardiac computed tomography
- left atrial appendage thrombus
- transcatheter aortic valve replacement
- transesophageal echocardiography
- Received June 21, 2016.
- Revision received September 12, 2016.
- Accepted October 20, 2016.
- American College of Cardiology Foundation
- Passman R.,
- Bernstein R.A.
- Lloyd-Jones D.M.,
- Wang T.J.,
- Leip E.P.,
- et al.
- Pizzetti F.,
- Turazza F.M.,
- Franzosi M.G.,
- Barlera S.,
- Tognoni G.
- Grigioni F.,
- Avierinos J.-F.,
- Ling L.H.,
- et al.
- Romero J.,
- Husain S.A.,
- Kelesidis I.,
- Sanz J.,
- Medina H.M.,
- Garcia M.J.
- Chugh S.S.,
- Blackshear J.L.,
- Shen W.K.,
- Hammill S.C.,
- Gersh B.J.
- White J.M.,
- Kodali S.K.
- Williams P.D.,
- de Belder M.A.,
- Maredia N.,
- Muir D.F.
- Kappetein A.P.,
- Head S.J.,
- Généreux P.,
- et al.
- January C.T.,
- Wann L.S.,
- Alpert J.S.,
- et al.
- Haussig S.,
- Mangner N.,
- Dwyer M.G.,
- et al.
- Lansky A.J.,
- Schofer J.,
- Tchetche D.,
- et al.
- Bandali A.,
- Parry-Williams G.,
- Kassam A.,
- et al.
- Toutouzas K.,
- Latsios G.,
- Stathogiannis K.,
- et al.
- Achenbach S.,
- Delgado V.,
- Hausleiter J.,
- Schoenhagen P.,
- Min J.K.,
- Leipsic J.A.
- Bax J.J.,
- Delgado V.,
- Bapat V.,
- et al.