Author + information
- Received October 19, 2015
- Revision received March 3, 2016
- Accepted March 10, 2016
- Published online October 1, 2016.
- Kevin P. Cohoon, DO, MSc,
- Matylda Mazur, MD,
- Robert D. McBane, MD,
- Naser Ammash, MD,
- Samuel J. Asirvatham, MD and
- Waldemar E. Wysokinski, MD, PhD∗ ()
- ↵∗Reprint requests and correspondence:
Dr. Waldemar E. Wysokinski, Division of Cardiovascular Medicine, Mayo Clinic and Foundation for Education and Research, 200 First Street SW, Rochester, Minnesota 55905.
Objectives The authors tested the hypothesis that the inflammatory response of nonvalvular atrial fibrillation (NVAF) is associated with elevated soluble CD40 ligand (sCD40L).
Background NVAF is generally believed to be an inflammatory disease process. sCD40L represents a sensitive in vivo indicator of platelet activation and may serve as an “inflammatory and thrombotic thermometer.”
Methods Plasma sCD40L was measured using enzyme-linked immunoadsorbent assay in 109 NVAF cases (60.9 ± 15 years of age; 26% women) and 48 normal sinus rhythm (NSR) controls (62.3 ± 15 years of age; 44% women). Patients were separated by arrhythmia duration of <1 month (n = 21), 1 to 12 months (n = 18), and >12 months (n = 70).
Results Median sCD40L level was significantly higher in NVAF cases than in NSR controls (321 pg/ml vs. 238 pg/ml, respectively; p = 0.029). This difference was driven by higher levels in patients with NVAF duration for <1 month (552 pg/ml) and 1 to 12 months (328 pg/ml). NVAF patients with arrhythmia duration for over 1 year had sCD40L levels not significantly different from those of NSR controls. An sCD40L concentration of 552 pg/ml distinguished NVAF patients with dysrhythmia duration of <1 month (area under the curve [AUC] of 0.72; p = 0.0010) or duration for ≤12 months (AUC: 0.69; p = 0.0003) from NSR controls. Circulating sCD40L levels were also significantly higher among patients with mild spontaneous echocardiogram contrast (SEC) (p = 0.0378) and those with moderate SEC (p = 0.007) compared with NSR controls.
Conclusions sCD40L levels are significantly higher in NVAF patients than in NSR controls but only for up to 1 year after development of dysrhythmia. An sCD40L concentration of 552 pg/ml can help to assess development or recurrence of asymptomatic NVAF.
Atrial fibrillation is a common cardiac dysrhythmia associated with an increased risk of cardioembolic stroke in the setting of left atrial appendage thrombosis (LAAT) (1–4). The timing of atrial fibrillation initiation or recurrence is relatively easy if associated with typical symptoms. However, asymptomatic atrial fibrillation is frequent and presents challenges to the clinician with regard to management. Two major clinical scenarios related to this problem are evaluation of cause of cryptogenic stroke and assessment of possible recurrence of atrial fibrillation after original induced or spontaneous reversal to sinus rhythm.
There is growing evidence that provocation, initiation, and progression of atrial fibrillation carries an inflammatory component related directly to the extent of arrhythmia burden (5,6). This is particularly evident with increasing left atrial blood stasis and LAAT formation (7). Whether this inflammatory “expansion” is exclusive to the initial phase of arrhythmogenesis or persists thereafter is unknown. It is plausible that the inflammatory component subsides during the scarification phase of endomyocardial and myocardial fibrosis. Conversely, it is likewise possible that variables governing inflammation persist throughout the lifetime of the individual once established.
CD40 ligand (CD40L), a member of the tumor necrosis factor transmembrane protein superfamily, is promptly expressed on the platelet cell surface following activation and thrombus participation (8). CD40L is subsequently cleaved generating soluble CD40 ligand (sCD40L). On the platelet cell surface, sCD40L and CD40L induce endothelial cells to recruit leukocytes and promote thrombin generation through overexpression of tissue factor (9,10). Although 95% of circulating sCD40L comes from platelets, this measurement may represent a sensitive indicator of platelet activation and may serve as an “inflammatory and thrombotic thermometer” for the individual (11,12).
Elevated sCD40L has been found in various diseases and cardiovascular conditions that is associated with atrial fibrillation and has been shown to predict ischemic stroke and other adverse cardiovascular events (13–16). Increased sCD40L is an independent predictor of atrial fibrillation development after coronary artery bypass grafting (17). In those patients with pre-existing atrial fibrillation, sCD40L concentrations exceeding 476 pg/ml had a nearly 5-fold higher likelihood of suffering a vascular event (18). However, the relationship between elevated sCD40L and atrial fibrillation was not consistently demonstrated in case-control studies (19–21). It has been concluded that this association is weak at best and is likely explained by common comorbidities seen in this patient population (22). However, in a cohort of paroxysmal atrial fibrillation patients undergoing pulmonary vein isolation, atrial fibrillation induced by catheter stimulation in the electrophysiology laboratory resulted in rapid and significant rise of sCD40L levels (23). This contradiction may reflect a transitory nature of sCD40L elevation, which may be very time-dependent relative to the trigger or onset of this dysrhythmia.
We therefore sought to test the hypothesis that sCD40L is augmented only during the initial onset of nonvalvular atrial fibrillation (NVAF) and is governed by the degree of left atrial blood stagnation. To test this hypothesis, we compared plasma sCD40L concentrations in patients with different atrial fibrillation durations and in normal sinus rhythm (NSR) controls.
Materials and Methods
Study design, patient selection, recruitment, and clinical and echocardiographic data collection and assessment have been described previously (7). Briefly, all patients with NVAF (cases) who had transesophageal echocardiogram (TEE) ordered by their primary physician or cardiologist (from October 4, 2007, to April 27, 2009) were approached for study participation, unless they had: 1) acute illness, stroke, myocardial infarction, or surgery within 30 days; 2) more than moderate heart valve disease; 3) artificial heart valves; 4) previously unprovoked venous or arterial thrombosis; 5) prior major bleeding unrelated to warfarin therapy; 6) liver disease; 7) active malignancy; or 8) hormonal stimulation (estrogen/progesterone therapy or pregnancy). Control subjects in NSR with no history of atrial fibrillation were recruited from the Primary Care Internal Medicine clinic during their annual medical examination. From our original cohort (7), a subgroup of subjects who were not on antiplatelet therapy was randomly sampled for analysis of this study. The details of the screening process and patient recruitment are provided in the Online Figure.
TEE was performed as previously described (24,25). LAAT was defined as an echogenic mass in the appendage or body of the atrium, distinct from the underlying endocardium and pectinate muscles and detected in more than 1 imaging plane (24,25). Spontaneous echocardiogram contrast (SEC) was defined as a pattern of dynamic “smoke-like,” slowly swirling, intracavitary echo densities imaged with gain settings adjusted to eliminate background noise. SEC was graded as “absent” (Fatkin criterion: 0), “mild” (Fatkin criteria: 1 and 2), “moderate” (Fatkin criterion: 3), or “severe” (Fatkin criterion: 4), according to the published echocardiographic criteria by Fatkin et al. (26) with modification. Left atrial appendage emptying velocity profiles were measured over 5 consecutive cardiac cycles, with the sample volume positioned 1 cm within the orifice of the left atrial appendage (LAA) (27). The left ventricular ejection fraction (LVEF) was visually estimated. Aortic atherosclerosis severity was defined as “simple” when atheroma thickness was <4 mm and immobile when “severe” when atheroma exceeded 4 mm or contained mobile components (27). Left atrium volume index was assessed by TEE performed within 1 month of the TEE study and calculated by using the biplane area-length method (27). All echocardiographic images were analyzed by the study cardiologist (N.A.), who was blinded to clinical and laboratory data. Control subjects in NSR did not have evaluations completed with TEE.
Study definitions and event adjudication
NVAF was defined as atrial fibrillation occurring in the absence of rheumatic mitral stenosis or heart valve prosthesis (biological or mechanical, respectively). Atrial fibrillation was classified as paroxysmal, persistent, or permanent according to current guidelines (28). Paroxysmal atrial fibrillation was defined as recurrent arrhythmia of variable duration but lasting <7 days. Persistent atrial fibrillation was defined as arrhythmia lasting longer than 7 days but <12 months. Permanent atrial fibrillation was defined as arrhythmia lasting longer than 12 months. These distinctions were made by history and confirmed by either electrocardiography or Holter monitoring, when available. The duration of atrial fibrillation for each individual patient was defined as the interval between the date of initial diagnosis of this dysrhythmia, confirmed by electrocardiography, and the time of blood collection for the study. The CHADS2 (congestive heart failure, hypertension, age ≥75 years, diabetes mellitus, stroke/transient ischemic attack) and CHA2DS2-VASc (congestive heart failure, hypertension, age ≥75 years, diabetes mellitus, stroke/transient ischemic attack, vascular disease, sex) scores were assigned for each case and control (29,30).
Congestive heart failure was defined as the presence of clinical symptoms and signs of heart failure within the previous 3 months with or without evidence of LV systolic dysfunction by echocardiography (29). Diabetes mellitus was diagnosed on the basis of criteria recommended by the American Diabetes Association (31). Stroke, transient ischemic attack, and systemic embolization were defined by criteria proposed by the American Heart Association (32).
Plasma sCD40L analysis
For NVAF patient (cases) scheduled for elective cardioversion or radiofrequency ablation, phlebotomy was uniformly performed prior to the procedure. The time interval between phlebotomy and TEE was <24 h. For each subject, 20 ml of citrated blood was collected by antecubital venipuncture using a 19-gauge thin-walled “butterfly” needle with a short plastic extension tube. Plasma samples, obtained by centrifugation (2,000 × g for 10 min), were snap frozen and archived at −70°C until used.
sCD40L was measured in plasma by a 4.5-h solid-phase enzyme-linked immunoadsorbent assay technique using the Quantikine human CD40 ligand immunoassay (R&D Systems, Inc., Minneapolis, Minnesota). Each assay was performed in duplicate. Intra-assay precision coefficient of variation (CV) was 4.5% to 5.4%; interassay precision CV was 6.0% to 6.4%.
Demographic, clinical, and echocardiographic characteristics and sCD40L concentrations were compared between cases and controls. Continuous variables (mean ± SD) were compared between groups by using a 2-sample Student t test. Categorical variables (%) were compared using the Pearson chi-square test for independence. Ordinal variables (percent; median with quartiles) were compared using the Wilcoxon rank-sum test. Covariates incorporated into the univariate model included age, sex, congestive heart failure, hypertension, coronary artery disease, diabetes mellitus, previous stroke or transient ischemic attack, previous vascular disease, warfarin therapy, and statin therapy. We adjusted for age, sex, congestive heart failure, hypertension, diabetes mellitus, and previous stroke/transient ischemic attack. These parameters were included in the multivariate analyses to adjust for risk factors for LAAT. The C-index, or area under the curve, was used to evaluate the efficiency of sCD40L levels for detecting the atrial fibrillation duration time periods (<1 month, 1 to 12 months, and >12 months) by a receiver operating characteristic (ROC) curve. Using the cutpoint determined by the ROC curve, we dichotomized sCD40L concentrations according to whether the circulating levels were under or over the calculated threshold. This dichotomy for sCD40L levels was assessed in the logistic models to explore the overall predictive value of sCD40L (odds ratio [OR]; 95% confidence interval [CI]; and p value) in duration of NVAF. This was also done without the dichotomy for sCD40L levels. Statistical testing used the 2-tailed alpha level of 0.05 for significance. Data analyses were conducted using JMP version 9.0.1 software (SAS Institute Inc., Cary, North Carolina).
Demographic and clinical characteristics are presented for 109 NVAF cases and 48 NSR controls and separately for NVAF patients according to arrhythmia duration of <1 month, 1 to 12 months, and >12 months (Table 1). There was a higher proportion of women in NSR controls group than in NVAF cases. NVAF patients had a higher prevalence of hypertension and were more frequently treated with warfarin than NSR controls. All other demographic and clinical features, including mean CHADS2 and CHA2DS2-VASc scores, were similar between groups. There were no significant differences when demographic variables were stratified by atrial fibrillation duration.
Echocardiographic findings stratified by atrial fibrillation duration for cases are shown in Table 2. NVAF patients with arrhythmia duration >12 months had the highest EF. There were only 3 cases in this group with LVEF ≤40%. The left atrial volume index did not vary by atrial fibrillation duration. Although differences in aortic atheromatous disease differed by duration stratification, there were only small numbers of patients within each group with complex atheroma. By definition, permanent atrial fibrillation percentages were higher in those patients with longer atrial fibrillation duration. All other echocardiographic findings in the NVAF cases were distributed evenly among all periods of atrial fibrillation duration.
Overall, sCD40L concentrations were higher in NVAF cases than in NSR controls (p = 0.029) (Table 3). Among NVAF patients, sCD40L levels were higher in those with heart failure and in those with hypertension. Levels did not vary by any of the other collected demographic or clinical variables among cases or controls, including treatment with warfarin or statins. sCD40L levels were highest among cases with a short duration of atrial fibrillation. Beyond the first month, levels varied inversely by atrial fibrillation duration (Table 3, Figure 1). This relationship between sCD40L and atrial fibrillation duration was most striking in younger cases (<65 years of age) and in males and those with heart failure. Type of atrial fibrillation was not associated with different level of sCD40L.
The relationship between echocardiographic measurements and sCD40L stratified by atrial fibrillation duration is shown in Table 4. Levels of sCD40L were significantly higher in those cases with LVEF <40. sCD40L levels otherwise did not differ by other echocardiographic measurements. Specifically, levels did not vary by left atrial volume index or function, aortic atheroma, SEC, or presence of left atrial appendage thrombus. For individuals with long-standing atrial fibrillation (>12 months), sCD40L levels were significantly lower and varied by low ejection fraction, severely enlarged left atrium, and presence of simple aortic atheroma (Table 4). The differences in sCD40L levels between patients with left atrial thrombus who had dysrhythmia duration of 1 to 12 months and those with duration longer than 12 months were particularly high (1,168 pg/ml [887 to 1,449 pg/ml] vs. 278 pg/ml [178 to 591 pg/ml], respectively; p = 0.0094).
Discrimination of rhythm status by sCD40L
To determine whether specific sCD40L concentrations could be used to discriminate between patients with NVAF and those with sinus rhythm, the area under the ROC curve for sCD40L was analyzed by atrial fibrillation duration (Figure 2). For those patients with NVAF duration of <1 month, the best sCD40L cutpoint was 552 pg/ml (AUC: 0.72; p = 0.0010), giving a sensitivity of 0.52 and specificity of 0.99 (Figure 2A). The positive predictive value is 73.3%, and the negative predictive value is 81.5%.
For NVAF lasting ≤12 months, the best cutpoint was also 552 pg/ml (AUC: 0.69; p = 0.0003). This cutoff concentration point showed a sensitivity of 0.46 and a specificity of 0.99 (Figure 2B). Positive predictive value was 81.8%, and negative predictive value was 67.7%. For those with NVAF >12 months, no potential cutpoint for sCD40L levels could be identified (Figure 2C).
Logistic regression analysis was used to further evaluate the significance of arrhythmia duration and sCD40L levels stratified at the concentration of 552 pg/ml. High levels of sCD40L (>552 pg/ml) were associated with NVAF of <1 month by univariate logistic regression (OR: 12.1; 95% CI: 3.40 to 51.5; p < 0.0001). After we adjusted for age, sex, and CHADS2 score, sCD40 levels exceeding the cutpoint of 552 pg/ml remained an independent predictor for NVAF duration of <1 month (OR: 30.35; 95% Cl: 5.72 to 265); p = <0.0001).
For assessment of patients with NVAF duration ≤12 months, sCD40L was also analyzed using the same cutpoint. High levels of sCD40L (>552 pg/ml) were associated with NVAF ≤12 months by univariate logistic regression (OR: 9.43; 95% CI: 3.07 to 35.85; p <0.0001). After we adjusted for age, sex, and CHADS2 score, sCD40 concentrations >552 pg/ml remained an independent predictor for NVAF duration of ≤12 months (OR: 10.28; 95% Cl: 3.14 to 42.54); p < 0.0001). When sCD40L levels were nondichotomized, multivariate regression demonstrated sCD40L levels remained an independent predictor of NVAF <1 month (OR: 1.003; 95% Cl: 1.002 to 1.007 per unit change; p < 0.0001) and NVAF ≤12 months (OR: 1.002; 95% Cl: 1.001 to 1.004 per unit change; p = 0.0002) after adjustment for age, sex, and CHADS2 score.
This study provides evidence that circulating levels of sCD40L are elevated in NVAF patients compared to those in NSR controls but only in those patients with duration of arrhythmia ≤12 months. Patients with atrial fibrillation duration over 1 year had sCD40L levels no different from controls. This is in keeping with prior study demonstrating a prompt rise in sCD40L upon induction of atrial fibrillation (23). The mechanism responsible for this rise is not precisely clear. Possibilities could include a response to either the chaotic electrical activity of this dysrhythmia or the mechanical dysfunction of the atria. Regardless of the inciting mechanism, the ultimate response leads to platelet activation with sCD40L release. This response is prompt, occurring within minutes of atrial fibrillation initiation. The current study shows that this process does not appear to be persistent, however. In our cohort, sCD40L levels were highest in patients with atrial fibrillation duration of <1 month. For those patients with atrial fibrillation duration <12 months, these levels continued to be elevated regardless of persistence of atrial fibrillation episodes. Within the 4 weeks of atrial fibrillation, CD40L levels trended higher in those with paroxysmal arrhythmia (594 pg/ml; range: 309 to 923 pg/ml) compared with those with persistent arrhythmia (512 pg/ml; range: 204 to 660 pg/ml) (Table 4). These findings together with the observations from the acute provocation of atrial fibrillation during ablation procedure (23) indicate that, although this dysrhythmia leads to rapid and significant rise of sCD40L levels, this elevated level is not significantly affected and remains high regardless of whether the episodes of atrial fibrillation are persistent or only intermittent. In summary, these results suggest that early in the natural course of atrial fibrillation, there is a strong stimulus for platelet activation; however, this platelet stimulation does not persist beyond the period of 12 months. Beyond 1 year, levels are similar to those of NSR control patients. At first glance, these findings appear to contrast with the prevailing hypothesis that thrombotic propensity is directly related to atrial fibrillation burden or traditional clinical risk factors of thromboembolism in this dysrhythmia. Rather, it is possible that the pathogenesis of thrombosis evolves over time. This is not to imply that platelets do not participate in thrombi occurring beyond 1 year of atrial fibrillation duration. These findings may suggest that the tumor necrosis factor transmembrane protein superfamily is not principally involved in the platelet participation in the more chronic setting of permanent atrial fibrillation. Indeed, histologic assessment of aged thrombi retrieved from the left atrial appendage confirm substantive differences of platelet content compared with acute embolized thrombi (3). This temporal relationship between sCD40L liberation and atrial fibrillation duration might also explain the lack of correlation between these variables and another marker of platelet activation, endothelial perturbation, and atrial fibrillation burden noted in previous studies (19–22). Without accounting for atrial fibrillation duration, these associations may be missed. Moreover, our finding can shed a new light on prior report that elevated sCD40L was an independent predictor of atrial fibrillation development after coronary artery bypass grafting (16); these patients with elevated sCD40L might in fact have already had undiagnosed asymptomatic paroxysmal atrial fibrillation before heart surgery. The behavior of sCD40L in the late course of atrial fibrillation beyond 12 months might reflect an inflammatory evolution dominated by scarification of endothelial and myocardial tissues. The introduction of a “package of care” including anticoagulation, antiplatelet therapy, and vascular risk factor modification might also affect sCD40L levels, as postulated by Choudhury et al. (19).
In the present study, we provide evidence that sCD40L levels may help differentiate the timing of atrial dysrhythmia. We used a cutpoint of 552 pg/ml to distinguish between patients who had had atrial fibrillation for up to 1 year and patients with sinus rhythm. One could envision a number of clinical scenarios where this measurement could be useful. For example, in the evaluation of patients with cryptogenic stroke and normal sinus rhythm on electrocardiography, elevated levels could prompt the clinician to more aggressively evaluate the patient for subclinical paroxysmal atrial fibrillation as a mechanism. Another example could include the periodic evaluation of asymptomatic patients following atrial fibrillation ablation or cardioversion (electrical or chemical) to determine whether recurrent subclinical atrial fibrillation is present. More than academic in nature, this finding would be anticipated to alter therapy.
NVAF patients with mild to moderate SEC have significantly higher sCD40L concentrations than atrial fibrillation patients without SEC or control subjects in NSR. This association was not seen in patients with severe SEC or LAAT. In general, left atrial size correlates directly with duration of dysrhythmia. The degree of blood stagnation also reflects left atrial size. Whereas 13 of 20 of our patients (65%) with severe SEC had atrial fibrillation for longer than 12 months, these findings may simply reflect longer duration of arrhythmia. Interestingly, 2 NVAF patients with LAAT and duration of dysrhythmia for <12 months had one of the highest concentrations of sCD40L (887 and 1,449 pg/ml).
First, the study methodology included longitudinal follow-up of a single cohort of patients with newly diagnosed atrial fibrillation. Ideally, these patients would be serially sampled for changes in sCD40L over time without a specific intervention to change the natural course of dysrhythmia. In clinical practice, it would difficult to identify a large cohort of patients who would be satisfied without at least a trial of sinus rhythm restoration. Indeed, this would not be ethical. The next step to further assess utility of sCD40L to discriminate between patients with atrial fibrillation recurrence and those without would be a prospective study to measure sCD40L before, immediately after restoration of NSR (spontaneous, chemical, cardioversion, pulmonary vein isolation), and 1 month (and/or 12 months) later correlated with the results of continuous rhythm monitoring. Second, we cannot exclude referral bias for the NVAF cases enrolled, as only those who had a TEE requested by their care provider were approached for recruitment. Third, determination of degree of SEC is subjectively quantified, and to improve reproducibility of this limitation, assessment was limited to 1 cardiologist blinded to both clinical and laboratory data. Fourth, patients receiving antiplatelet therapy were excluded in order to remove any source of interference of sCD40L by this class of inhibitors (33). This may have introduced some degree of selection bias against those individuals receiving either aspirin or a thienopyridine for atherosclerotic disease. Fifth and finally, several variables were tested in multivariate modeling, and it is possible that some of the statistical tests were underpowered, and significant results could be spurious. Multivariate models could well be overfitted.
In summary, we now provide evidence supporting a direct relationship between sCD40L levels with duration of NVAF and measurements of left atrial blood stasis up to the moderate intensity of SEC. Elevated sCD40L levels can help to assess development or recurrence of asymptomatic NVAF.
COMPETENCY IN PATIENT CARE AND PROCEDURAL SKILLS: In the present study, we provide evidence that sCD40L, a sensitive marker of platelet activation, may serve as an indicator of atrial fibrillation acuity. An sCD40L concentration of 552 pg/ml distinguishes among patients with dysrhythmia duration of <1 month and ≤12 months and sinus rhythm controls.
TRANSLATIONAL OUTLOOK: One could envision a number of clinical scenarios in which this measurement could be useful, including an assessment of cryptogenic stroke mechanisms and periodic monitoring of asymptomatic patients following atrial fibrillation ablation or cardioversion (electrical or chemical) to determine whether recurrent subclinical atrial fibrillation is present.
For a supplemental figure, please see the online version of this article.
This work was supported by the Department of Internal Medicine, Mayo Clinic, grant CR 17896; statistical support was provided by a Mayo Clinic Division of Cardiology internal grant. The authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- area under the curve
- CD40 ligand
- congestive heart failure, hypertension, age ≥75years, diabetes mellitus, stroke/transient ischemic attack, vascular disease, sex
- congestive heart failure, hypertension, age ≥75 years, diabetes mellitus, stroke/transient ischemic attack
- confidence interval
- left atrial appendage
- left atrial appendage thrombus
- left ventricular ejection fraction
- normal sinus rhythm
- nonvalvular atrial fibrillation
- odds ratio
- soluble CD40 ligand
- spontaneous echocardiogram contrast
- transesophageal echocardiogram
- Received October 19, 2015.
- Revision received March 3, 2016.
- Accepted March 10, 2016.
- American College of Cardiology Foundation
- Chung M.K.,
- Martin D.O.,
- Sprecher D.,
- et al.
- Issac T.T.,
- Dokainish H.,
- Lakkis N.M.
- Ammash N.,
- Konik E.A.,
- McBane R.D.,
- et al.
- Bavendiek U.,
- Libby P.,
- Kilbride M.,
- et al.
- Andre P.,
- Nannizzi-Alaimo L.,
- Prasad S.K.,
- et al.
- Lim H.S.,
- Blann A.D.,
- Lip G.Y.
- Schonbeck U.,
- Varo N.,
- Libby P.,
- et al.
- Santilli F.,
- Davi G.,
- Consoli A.,
- et al.
- Antoniades C.,
- Van-Assche T.,
- Shirodaria C.,
- et al.
- Ferro D.,
- Loffredo L.,
- Polimeni L.,
- et al.
- Lip G.Y.,
- Patel J.V.,
- Hughes E.,
- et al.
- Hammwöhner M.,
- Ittenson A.,
- Dierkes J.,
- et al.
- Lim H.S.,
- Willoughby S.R.,
- Schultz C.,
- et al.
- Melduni R.M.,
- Chandrasekaran K.,
- Friedman P.A.,
- et al.
- Fatkin D.,
- Kelly R.P.,
- Feneley M.P.
- ↵Oh JK SJ, Tajik AJ. The Echo Manual. 3rd edition. Philadelphia, PA: Lippincott Williams & Wilkins; 1994:109–119.
- Fuster V.,
- Rydén L.E.,
- Cannom D.S.,
- et al.
- Genuth S.,
- Alberti K.G.,
- Bennett P.,
- et al.