INTRODUCTION
The left atrial appendage (LAA) is a small, irregular shape structure, outpouching from the left atrium (LA) of the heart [23]. LAA is a remnant of the primitive atrium developed in early gestation. It is proven that LAA takes part in the regulation of volume homeostasis through its mechanical function and neurohormonal secretion [12]. Moreover, the LAA is a well-established source of cardiac thrombogenesis and arrhythmogenesis [20]. For more than half of a century, the LAA is considered a primary source of cardioembolic events in patients with atrial tachycardia (atrial fibrillation [AF]) [28]. Numerous morphological features of the LAA, which include multilobular and trabeculated structure and small orifice with a narrow neck, predispose to turbulent blood flow and local blood stasis, thereby increasing the risk for thrombosis [20, 21]. Nevertheless, no consensus was reached as to which specific LAA morphometric features increased the risk of thrombogenicity.
Anticoagulation therapy is a standard of care to prevent cardioembolic stroke in patients with AF [14]. In patients with a high risk of bleeding, contraindications to anticoagulants, intolerance, or low patient compliance the LAA closure was proposed as an alternative approach to eliminate cardioembolic stroke risk [11]. This led to the development of several surgical and percutaneous approaches to close/exclude the LAA based on three concepts: plug, pacifier principle, and ligation [4]. Among the ligation techniques, the LARIAT (SentreHEART, Inc., Redwood City, CA, USA) procedure is a distinguished percutaneous adaptation of the surgical exclusion of LAA with a suture, used as an alternative to open-heart surgery [2]. Although the LARIAT procedure can be performed effectively with acceptably low complications, some short- and long-term complications of the intervention may occur [16]. The frequency of peri- and post-procedural complications can be lowered by adequate patients selection thanks to computed tomography (CT) evaluation of the LAA shape and size and detailed pre-procedural planning [26]. Morphology of the LAA can significantly affect pathophysiological properties of the LAA, choice of closure technique, the course of the procedure, its results, and complications [4, 23]. The present study aims to compare the LAA morphometrical features between patients with AF subjected to LAA percutaneous closure using LARIAT device and sinus rhythm patients using cardiac CT.
MATERIALS AND METHODS
The current study was conducted at the Department of Anatomy and the Department of Cardiovascular Surgery and Transplantology at the Jagiellonian University Medical College, Krakow, Poland. The study was conducted according to the principles expressed in the 1975 Declaration of Helsinki and was approved by the Bioethical Committee of the Jagiellonian University in Krakow, Poland (No. 1072.6120.4.2020).
Study population
The study evaluated 101 contrast-enhanced electrocardiograms (ECG) gated CT scans of hearts of which 51 were of consecutive adult patients with AF and who underwent percutaneous closure of the LAA with the LARIAT device (41.2% females, mean age of 61.0 ± 10.35 years) (study group). The CT scans were performed prior to the LARIAT procedure. The remaining 50 scans (control group) were of randomly selected adult patients subjected to non-invasive coronary arteries evaluation (50% females, mean age of 57.8 ± 14.45 years) of which inclusion criteria were: no history of ischaemic stroke, no AF, and absence of any significant structural heart defects. All sample patients gave written informed consent. The medical history of all included patients was reviewed, and the following data were extracted: sex, age, body mass index, existing comorbidities (i.e. hypertension, hyperlipidaemia, chronic heart failure, coronary artery disease, chronic obstructive pulmonary disease, diabetes mellitus, history of stroke, and history of myocardial infarction), heart rhythm status, and the medication’s usage. The following scores were calculated: Canadian Cardiovascular Society (CCS) grading of angina pectoris and New York Heart Association (NYHA) Functional Classification and CHA2DS2-VASc Score for Atrial Fibrillation Stroke Risk.
Cardiac computed tomography
Every patient underwent a per-protocol contrast-enhanced ECG gated CT scan. All scans were performed in sinus rhythm. If the patient’s heart rate was over 70 bpm before the cardiac CT examination, the patient was administered 10 or 40 mg of propranolol or 40 mg of verapamil, depending on the physician’s recommendation. The examination was performed using a 64-row dual-source scanner (Siemens, Erlangen, Germany). The image acquisitions were collected during a deep inspiration breath-hold. The imaging parameters for dual-source CT were: a tube voltage of 100–120 kV and an effective tube current of 350–400 mA. The collimation and temporal resolution were 2 × 32 × 0.6 mm and 165 ms, respectively. The time of arrival of the contrast agent to the ascending aorta was determined at the level of the tracheal bifurcation and achieved with the test bolus method. Then, the 15 mL of contrast agent was infused, which was followed by 20 mL of saline. The contrast agent was injected at a dose of 1.0 mL/kg at a rate of 5.5 mL/s followed by a 40 mL saline chaser with the same infusion rate. In the test bolus, the acquisitions delay was the time of maximum density of the ascending aorta with an additional 6 s of delay. Images were reconstructed with a B26f and B46f kernel and an image matrix of 512 × 512 pixels. A multiphase reconstruction (from 10% to 100%) was done. Moreover, 30% (left ventricle end-systole) and 70% (left ventricle end-diastole) image reconstructions were evaluated. Then, the initial quality check was performed using multiplanar reconstructions.
Image postprocessing and analysis
Postprocessing analyses of CT scans were performed on dedicated workstations (Dell, USA). Three-dimensional (3D) reconstructions of the heart subcomponents (LAA, LA, and left ventricle) were created using volume-rendering and segmentation techniques using Innovation Suite 23 (Materialise, Belgium) software in both end-diastolic and end-systolic cardiac phases. Manual and semi-automatic methods of segmentation were used. Next, raw data and visualizations were subjected to multidirectional morphological and morphometrical analyses.
First, two independent, blinded to patients history investigators classified LAA based on its shape into one of the three groups: cauliflower, chicken wing, or arrowhead as discussed previously (Fig. 1) [23]. Additionally, some individual lobes have been established. The LAA orifice and body were measured on the 3D reconstructions in the end-diastolic phase of the heart using virtual callipers (Fig. 2). Orifice shape, dimensions, area, and perimeter were measured. In addition, the maximal LAA body length (from orifice to the most far point) and width (in the widest point at the LAA base) were obtained. Depth of the LAA chamber was measured as a distance from the orifice plane to the last perpendicular and the orifice plane. For the chicken wing type, the angle of the bend and lengths from orifice to bend and from the bend to the LAA apex were also evaluated. The main dimensions of the left ventricle and atrium were estimated. The volume of the LAA left atrium and the left ventricle was measured in end-diastolic and end-systolic cardiac phases. Then, ejection fractions of the above-mentioned chambers were calculated.
To reduce human bias, all measurements were recorded by two independent, blinded to patients history investigators. If results between the two researchers varied by more than 10%, both measurements were repeated. The mean of the two new values was calculated and reported as the final value.
Statistical analyses
Data are shown as mean values with corresponding standard deviations (± SD) or in percentages. The Shapiro–Wilk test was used to determine if the quantitative data were normally distributed, and Levene’s test was performed to verify a relative homogeneity of variance. The Student t-tests, Wilcoxon signed-rank tests, and the Mann–Whitney U tests with Bonferroni corrections were used to conduct statistical comparisons. The non-parametric Kruskal–Walli’s test was used to compare values between groups. Bonferroni correction was applied because of multiple comparisons. Correlation coefficients were calculated to assess whether there was a statistical dependence between the analysed parameters. A p-value lower than 0.05 was considered statistically significant. Statistical analyses were performed using StatSoft STATISTICA 13.3 software for Windows (StatSoft Inc., Tulsa, OK, USA).
RESULTS
Clinical characteristics of studied patients are presented in Table 1. Among the study group, 29.4% of patients had a history of ischaemic stroke. Arterial hypertension was more common in the study group than in control subjects (90.2 vs. 38.0%, p < 0.001). The clinical manifestation of chronic heart failure symptoms was significantly more severe in study group patients (NYHA scale, Table 1). Moreover, patients subjected to the LARIAT procedure have significantly higher thromboembolic risk than control patients (CHA2DS2-VASc, Table 1). No other significant differences in patients’ demographic and clinical data were found.
|
Variable |
Study group (n = 51) |
Control group (n = 50) |
P |
|
|
Females (%) |
41.2% |
50.0% |
0.09 |
|
|
Age [years] |
61.04 (10.35) |
57.80 (14.45) |
0.20 |
|
|
Body mass index [kg/m2] |
28.8 (5.4) |
28.6 (4.9) |
0.83 |
|
|
Non-sinus tachycardia |
51/100.0% |
0/0.0% |
< 0.001 |
|
|
Arterial hypertension |
46/90.2% |
19/38.0% |
< 0.001 |
|
|
Hyperlipidaemia |
3/5.9% |
2/4.0% |
0.67 |
|
|
Chronic heart failure |
10/19.6% |
6/12.0% |
0.30 |
|
|
Coronary artery disease |
11/21.6% |
6/12.0% |
0.20 |
|
|
Chronic obstructive pulmonary disease |
2/3.9% |
4/8.0% |
0.39 |
|
|
Diabetes mellitus type II |
6/11.8% |
5/10.0% |
0.78 |
|
|
History of myocardial infarction |
1/2.0% |
2/4.0% |
0.55 |
|
|
History of ischaemic stroke |
15/29.4% |
0/0.0% |
< 0.001 |
|
|
CCS: |
||||
|
I |
5/9.8% |
3/6.0% |
0.48 |
|
|
II |
8/15.7% |
4/8.0% |
0.24 |
|
|
III |
0/0.0% |
1/2.0% |
0.31 |
|
|
IV |
0/0.0% |
0/0.0% |
– |
|
|
NYHA: |
||||
|
I |
14/27.5% |
6/12.0% |
0.05 |
|
|
II |
20/39.2% |
8/16.0% |
0.01 |
|
|
III |
3/5.9% |
0/0.0% |
0.08 |
|
|
IV |
0/0.0% |
0/0.0% |
– |
|
|
CHA2DS2-VASc: |
||||
|
0 (males) or 1 (females) — “low risk” |
2/3.92% |
22/44.0% |
< 0.001 |
|
|
1 (males) — “low-moderate risk” |
11/21.6% |
5/10.0% |
0.11 |
|
|
≥ 2 — “moderate-high risk” |
38/74.5% |
23/46.0% |
0.01 |
|
The chicken wing shape was the most common type of LAA both in the study and control group (45.1 vs. 46.0%, p = 0.93), followed by arrowhead (29.4 vs. 20.0%, p = 0.27) and cauliflower (25.5 vs. 34.0%, p = 0.35) (Table 2). Although no statistically significant differences in LAA type distributions were found between groups, the study group was characterized by LAAs with a lower number of lobes (Table 2).
|
Parameter |
Study group (n = 51) |
Control group (n = 50) |
P |
|
|
LAA shape (%): |
||||
|
Cauliflower |
25.5 |
34.0 |
0.35* |
|
|
Chicken wing |
45.1 |
46.0 |
0.93* |
|
|
Arrowhead |
29.4 |
20.0 |
0.27* |
|
|
Number of LAA lobes (%): |
||||
|
1 |
39.2 |
16.0 |
0.01** |
|
|
2 |
33.3 |
20.0 |
0.13** |
|
|
3 |
19.6 |
34.0 |
0.10** |
|
|
≥ 4 |
7.8 |
30.0 |
0.01** |
|
|
LAA orifice shape round/oval [%] |
6.0/94.0 |
29.6/70.4 |
0.001 |
|
|
LAA orifice antero-posterior diameter [mm] |
19.3 (4.12) |
17.2 (4.0) |
0.01 |
|
|
LAA orifice transverse diameter [mm] |
25.1 (5.1) |
20.5 (4.4) |
0.001 |
|
|
LAA orifice area [mm2] |
387.2 (133.9) |
327.1 (128.3) |
0.02 |
|
|
LAA orifice perimeter [mm] |
70.2 (12.5) |
61.2 (11.6) |
0.04 |
|
|
LAA body length [mm] |
47.4 (15.4) |
43.7 (10.9) |
0.17 |
|
|
LAA width at the base [mm] |
24.7 (5.6) |
24.4 (5.8) |
0.81 |
|
|
LAA chamber depth [mm] |
17.7 (3.5) |
16.5 (3.8) |
0.11 |
|
|
LAA end-diastolic volume [mL] |
9.0 (3.4) |
5.4 (3.8) |
0.001 |
|
|
LAA end-systolic volume [mL] |
10.9 (3.6) |
10.6 (4.3) |
0.69 |
|
|
LAA ejection fraction [%] |
16.4 (14.9) |
48.2 (12.9) |
0.001 |
|
|
Left atrial end-diastolic antero-posterior diameter [mm] |
40.1 (6.9) |
36.0 (7.1) |
0.001 |
|
|
Left atrial end-diastolic volume [mL] |
109.5 (38.9) |
93.8 (30.8) |
0.02 |
|
|
Left atrial end-systolic volume [mL] |
129.5 (41.7) |
112.5 (34.3) |
0.04 |
|
|
Left atrial ejection fraction [%] |
13.9 (10.8) |
17.7 (6.5) |
0.04 |
|
|
Left ventricular end-diastolic inner diameter [mm] |
45.5 (6.8) |
46.4 (6.3) |
0.47 |
|
|
Left ventricular end-diastolic volume [mL] |
124.5 (48.1) |
133.5 (39.1) |
0.15 |
|
|
Left ventricular end-systolic volume [mL] |
68.7 (40.5) |
65.7 (25.0) |
0.56 |
|
|
Left ventricular ejection fraction [%] |
46.7 (14.3) |
50.7 (12.2) |
0.14 |
|
In both analysed groups, the LAA orifice was predominantly oval (study group: 94.0% vs. control group: 70.4%), and significantly fewer round orifices were found in the study group (p = 0.001, Table 2). Moreover, the size of the orifice was found to be significantly larger in LARIAT patients compared to the control group (higher diameters, orifice area, and orifice perimeter, Table 2).
Several major LAA morphometrical features were analysed (body length, body width, and chamber depth). However, no statistically significant differences were found in those measured parameters between the two groups (all p > 0.05, Table 2). No differences between groups were noted in LAA end-systolic volumes, but a significant difference was observed in the end-diastolic cardiac phase (Table 2). The calculated LAA ejection fraction was significantly lower in the study group compared to healthy patients (16.4 ± 14.9 vs. 48.2 ± 12.9%, p = 0.001). For the chicken wing type, the angle of the bend was significantly smaller in LARIAT patients than in the control group (62.1 ± 10.4 vs. 68.5 ± 13.9°, p = 0.011). The LAA body lengths measured from orifice to bend and from bend to LAA apex showed no differences between analysed groups (to bend: 22.7 ± 3.8 vs. 18.4 ± 5.5 mm, p = 0.103; from bend: 28.1 ± 4.6 vs. 27.3 ± 5.8 mm, p = 0.443).
A significantly larger LA was observed in the study group compared to controls (larger anteroposterior diameter and volume) with reduced LA ejection fraction in LARIAT patients (13.9 ± 10.8 vs. 17.7 ± 6.5%, p = 0.04) (Table 2). No differences were found in measured left ventricle dimensions, volumes, and ejection fraction between investigated groups (all p > 0.05, Table 2). No other significant correlations and differences were found between all measured morphometric parameters.
DISCUSSION
The LAA morphology plays a significant role in the pathogenesis of ischaemic stroke. It was proven that complex LAA (non-chicken wing shape) is an independent risk factor for cryptogenic ischaemic stroke [6, 27]. However, little is known about the LAA anatomy that influences the course and results of the LAA closure procedures. Without a doubt, a significant morphologic remodelling of the LA occurs in patients with non-sinus rhythm [3, 25]. Nevertheless, a lack of consensus is observed regarding the LAA structural alterations related to rhythm and hemodynamical disturbances [3, 13]. Although it is hard to believe that the remodelling may concern the type (shape) of the LAA body, the changes in the LAA orifice shape and size, LAA body size, and LAA function may be substantial.
Regardless of the technique used, residual leaks after the LAA are commonly observed [9]. Our study shows that in AF patients, the biggest morphological changes are found in the LAA orifice, crucial for all LAA closure procedures. Left atria remodelling associated with enlargement in heart failure and arrhythmia patients changes the shape and size of the LAA opening. LAA orifices in diseased patients are significantly larger and more oval, which affects the risk of leakage [18]. An incomplete sealing may potentially occur in cases where an appropriately large device is used to close the LAA when implanted not in the round and when in oval or in irregular shape orifice [18]. Thus, operators should be aware of this change in LAA orifice morphology and choose the appropriate tool for the LAA closure to avoid danger to the health and life of patients’ peri-device leaks. The observed reduced LAA ejection fraction in research group is another important observation emerging from the current work. The LAA ejection fraction was decreased even despite performing tomography scans during sinus rhythm. This reduced contractile function of the LAA may be the indirect result of remodelling of the LA and altered atrial blood pressure, which influence structure of the LAA itself leading to impaired contractility of LAA. In paroxysmal AF remodelling of LA is most often negligible and reversable. The duration time of LA rhythm disturbances negatively affects the structure of the atrium. Persistent AF strongly promotes electrical remodelling and pressure/volume overload of LA promotes interstitial fibrosis leading to structural remodelling [23]. It follows that irreversible changes in LA morphology and consequently in LAA structure may start to develop in patients with AF lasting more than one week [1, 5]. On the other hand, the increased left atrial pressure may create to high resistance for the LAA to generate the effective force of its contraction. Previous studies showed that LAA ejection fraction may be an independent predictor of cardiac thrombo-embolization [8,22], which alters LAA emptying and predisposes to local blood stasis and its increased thrombogenicity [19].
Cardiac CT seems to be the preferable tool for pre-procedural imaging of the LAA as it provides the most faithful, 3D image of the LAA [24]. Moreover, it helps visualize other important heart cavities and main vessels (together with coronary vasculature), in which knowledge of morphology is crucial to avoid serious peri-procedural complications [7, 23]. Our study also indicates that LAA 3D visualization may provide useful information for the operator on the LAA morphology, thereby influencing the choice of the technique of its closure. The creation of a fully personalized virtual 3D model of the patient’s LAA may be completed in less than an hour. This model can easily be subjected to multi-directional measurement or used to simulate procedures [10]. Moreover, the created model may be visualized using virtual or augmented reality techniques or using the 3D printed technique to better assist the clinicians [17]. Unfortunately, one of the obstacles to the use of this wonderful tool in all medical centres is the relatively low availability and high cost of software enabling the segmentation and visualization of the CT data, as well as the lack of experience of the staff in manufacturing the 3D models [15].
Some limitations to this study should be considered. First, this is a single-centre study with a relatively small sample size. A limited number of patients in both groups does not allow a comprehensive statistical analysis to determine which LAA morphometrical factors may contribute to the occurrence of a stroke or are associated with success rate or complications during the LARIAT procedure. Moreover, only Caucasian subjects were studied, and no inter-racial differences are studied. Despite these limitations, we believe that the current study has provided insight into the morphological characteristics of the LAA in different groups of patients.
CONCLUSIONS
Presented morphometrical data should be considered during planning and when performing procedures targeted to LAA closure, as significant differences in the anatomy of the LAA may be encountered in patients with AF. Substantial morphometrical differences in LAA orifice are found, which is significantly larger and more oval in patients with AF compared to healthy controls. Although no difference in LAA body type and size is observed, the LAA ejection fraction is significantly lower in AF patients.
Funding
This study was supported by the National Centre for Research and Development under the LIDER programme (LIDER/7/0027/L-10/18/NCBR/2019).
