Libman-Sacks endocarditis, characterized by Libman-Sacks vegetations, is common in patients with systemic lupus erythematosus and is commonly complicated with embolic cerebrovascular disease. Thus, accurate detection of Libman-Sacks vegetations may lead to early therapy and prevention of their associated complications. Although two-dimensional (2D) transesophageal echocardiography (TEE) has high diagnostic value for detection of Libman-Sacks vegetations, three-dimensional (3D) TEE may allow improved detection, characterization, and clinical correlations of Libman-Sacks vegetations.
Twenty-nine patients with systemic lupus erythematosus (27 women; mean age, 34 ± 12 years) prospectively underwent 40 paired 3D and 2D transesophageal echocardiographic studies and assessment of cerebrovascular disease manifested as acute clinical neurologic syndromes, neurocognitive dysfunction, or focal brain injury on magnetic resonance imaging. Initial and repeat studies in patients were intermixed in a blinded manner with paired studies from healthy controls, deidentified, coded, and independently interpreted by experienced observers unaware of patients’ clinical and imaging data.
The results of 3D TEE compared with 2D TEE were more often positive for mitral or aortic valve vegetations, and 3D TEE detected more vegetations per study and determined larger sizes of vegetations ( P ≤ .03 for all). Also, 3D TEE detected more vegetations on the anterior mitral leaflet, anterolateral and posteromedial scallops, and ventricular side or both atrial and ventricular sides of the leaflets ( P < .05 for all). In addition, 3D TEE detected more vegetations on the aortic valve left and noncoronary cusps, coronary cusps’ tips and margins, and aortic side or both aortic and ventricular sides of the cusps ( P ≤ .01 for all). Furthermore, 3D TEE more often detected associated mitral or aortic valve commissural fusion ( P = .002). Finally, 3D TEE detected more vegetations in patients with cerebrovascular disease ( P = .01).
Three-dimensional TEE provides clinically relevant additive information that complements 2D TEE for the detection, characterization, and association with cerebrovascular disease of Libman-Sacks endocarditis.
Libman-Sacks endocarditis, best characterized by Libman-Sacks vegetations, is common in systemic lupus erythematosus (SLE). Libman-Sacks vegetations are sterile abnormal growths of tissue around the heart valves with an autoimmune-mediated inflammatory and thrombotic pathogenesis. Libman-Sacks vegetations can be complicated with embolic cerebrovascular disease, peripheral arterial embolism, severe valve regurgitation, superimposed infective endocarditis, need for high-risk valve surgery, and increased mortality. Therefore, accurate detection of Libman-Sacks endocarditis may lead to early therapy and prevention of the development or progression of its associated complications. Multiplane two-dimensional (2-D) transesophageal echocardiography (TEE) has high diagnostic value for the detection of Libman-Sacks endocarditis. Three-dimensional (3D) TEE has emerged as an important complementary technique to 2D TEE for the detection and characterization of infective vegetations and valve tumors. However, no previous study has been reported on the diagnostic value and clinical relevance of 3D TEE in Libman-Sacks endocarditis. Therefore, this 3-year study was prospectively designed and conducted to determine the value of 3D TEE compared with 2D TEE for the detection, characterization, and association with cerebrovascular disease of Libman-Sacks vegetations.
This study was part of a 6-year (December 2006 to December 2012) research protocol approved by the National Institutes of Health and the institutional review board for the study of cardiovascular and cerebrovascular disease in patients with SLE. The results of that study in 102 subjects (76 patients with SLE and 26 healthy controls) using 2D TEE have been published. All participants provided informed consent.
From February 2010 (when 3D TEE became available at our institution) to December 2012 (the end of study funding and authorized enrollment), 29 patients with SLE (27 women; mean age, 34 ± 12 years; range, 18–55 years) prospectively underwent clinical and laboratory evaluations, brain magnetic resonance imaging (MRI), neurocognitive assessment, and 40 paired 3D and 2D transesophageal echocardiography studies (11 studies were repeated to reassess cardiac and brain disease after anti-inflammatory and/or antithrombotic therapy after patients’ first or during recurrent neurologic events). The 40 paired transesophageal echocardiographic studies constitute the basis for this study.
Patients with suspected heart and brain disease unrelated to SLE, including any febrile illness and infective endocarditis, were excluded from participation in the study.
Clinical, laboratory, brain MRI, and initial and repeat paired transesophageal echocardiographic studies in patients were intermixed in a blinded manner with paired studies from healthy controls and were deidentified, coded, and independently interpreted by experienced observers unaware of patients’ clinical and imaging data.
Clinical, Brain MRI, and Laboratory Evaluations
On enrollment, patients were prospectively assessed for the presence of (1) acute focal and nonfocal neurologic syndromes of stroke, transient ischemic attack (TIA), confusional state, cognitive dysfunction, or seizures ; (2) focal brain injury on MRI, defined as old or recent cerebral infarcts or periventricular or deep white matter abnormalities ; and (3) global neurocognitive dysfunction, defined as ≥1.5 SDs below the mean of healthy controls. Patients were also characterized with regard to SLE duration, activity, damage, standard serology, autoantibodies including antiphospholipid antibodies, and anti-inflammatory and anti-thrombotic therapy ( Table 1 ).
|Duration of SLE (y)||6.2 ± 5.6|
|Age at diagnosis of SLE (y)||27.5 ± 11.1|
|Total SLEDAI (U)||12.6 ± 11.8|
|Total SLICC (U)||2.8 ± 1.8|
|White blood cell count (cells/cm 3 )||6.9 ± 2.9|
|Hemoglobin (mg/dL)||12.9 ± 2.0|
|Platelets (per cm 3 )||243.6 ± 76.7|
|Creatinine (mg/dL)||1.33 ± 1.81|
|Serum albumin (mg/dL)||3.6 ± 0.6|
|dsDNA titer||247.3 ± 540.1|
|ANA titer||895.8 ± 818.0|
|C3 (mg/dL)||93.1 ± 39.8|
|C4 (mg/dL)||14.8 ± 8.8|
|CH50 (mg/dL)||69.1 ± 44.1|
|Quantitative d -dimer (mg/dL)||0.95 ± 1.2|
|C-reactive protein (mg/dL)||1.3 ± 1.2|
|Erythrosedimentation rate (mm/h)||40.5 ± 23.8|
|Smith antibody positive||14 (48%)|
|SSA antibody positive||12 (41%)|
|SSB antibody positive||9 (31%)|
|Any antiphospholipid antibody positive||16 (58%)|
|β 2 glycoprotein I antibody positive||8 (28%)|
|Lupus-like inhibitor positive||10 (34%)|
|IgG, IgM, or IgA anticardiolipin positive||14 (48%)|
|Prednisone therapy||16 (55%)|
|Prednisone current dose (mg/d)||21.0 ± 44|
|Prednisone average dose (mg/d)||5.4 ± 4.5|
|Years of prednisone||4.8 ± 4.6|
|Cyclophosphamide therapy||10 (34%)|
|Any immunosuppressive therapy (beyond prednisone)||12 (41%)|
|Hydroxychloroquine or chloroquine therapy||20 (69%)|
|Aspirin, clopidogrel, or warfarin||11 (38%)|
Participants underwent concurrent, complete 3D and 2D TEE with Philips iE33 systems (Philips Medical Systems, Andover, MA) using a 7-MHz 3D matrix-array transducer. Images were digitally acquired for offline reconstruction (for 3D transesophageal echocardiographic images) and interpretation.
Using 2D TEE and from the basal to the midesophageal level, the mitral and aortic valves were imaged in multiple planes at a depth of 4 to 8 cm with a narrow sector scan to improve image resolution.
Using 3D TEE and from the basal to the midesophageal level, 0° to 120° views of the mitral valve and 30° to 60° short-axis and 110° to 130° long-axis views of the aortic valve were acquired. To achieve the highest spatial and temporal resolution, electrocardiographically triggered multiple-beat (four beats) high-density full-volume images with the narrowest possible sector scan were obtained during breath holding (expiration). Three-dimensional transesophageal echocardiographic full-volume data sets were acquired with gain and compression settings of ≥50 units and then cropped and multiplane transected to obtain en face atrial and left ventricular views of the mitral valve and aortic root and left ventricular outflow tract views of the aortic valve.
Two-dimensional and 3D transesophageal echocardiographic images were acquired during the same examination, and both types of images were collected by the same examiner using the same probe.
Also, 2D transesophageal echocardiographic images were acquired in a systematic order using single-plane or, infrequently, x-plane tomographic planes. In addition, volumetric images were obtained in a systematic manner independently of identifying or not valve pathology on single tomographic planes. Furthermore, for best quality matching of 2D and 3D transesophageal echocardiographic images, 3D transesophageal echocardiographic volumetric images were obtained immediately after the best corresponding 2D images.
To reduce interpretation bias toward 3D TEE, the following steps were taken: (1) No live 3D images were obtained or recorded during the performance of TEE. (2) All deidentified 2D transesophageal echocardiographic studies were first interpreted by one experienced observer (C.A.R.). In 22 randomly selected 2D transesophageal echocardiography studies, interobserver agreement (C.A.R. and G.C.) for detection of vegetations of the mitral, aortic, and either valve were 82% (κ = 0.61), 96% (κ = 0.91), and 91% (κ = 0.80), respectively. (3) At a later time and to preserve the highest image resolution, 3D transesophageal echocardiographic studies of the mitral and aortic valves were identified by analyzing the volumetric data sets using multiplanar reconstruction in the ultrasound imaging systems by an experienced observer (D.M.) who was not an interpreter of transesophageal echocardiographic studies. These images were usually added at the end of the study. (4) Finally, 3D transesophageal echocardiographic studies were independently interpreted by two experienced observers (C.A.R. and K.T.) unaware of earlier 2D transesophageal echocardiographic interpretations. For all 3D transesophageal echocardiographic studies, interobserver agreement (C.A.R. and K.T.) for the detection of vegetations of the mitral, aortic, and either valve were 89% (κ = 0.78), 97% (κ = 0.92), and 96% (κ = 0.92), respectively.
Criteria for 3D and 2D TEE Interpretation
Libman-Sacks vegetations by either technique were defined as abnormal localized, protruding, and sessile echodensities >3 mm in diameter with well-defined borders either as part of or adjacent to valve leaflets, annulus, subvalvular apparatus, or endocardial surfaces. The cutoff of >3 mm in diameter was adopted to prevent misinterpretation of Lambl’s excrescences as vegetations. Lambl’s excrescences are thin (usually ≤2 mm in width, rarely up to 3 mm), elongated (usually >5 mm in length), independently hypermobile, and homogeneously echoreflectant structures located at the coaptation point and atrial side of the mitral valve and ventricular side of the aortic valve. By both techniques, location of the mitral valve vegetations with regard to anterior or posterior leaflets; basal, middle, or distal portions of the leaflets; corresponding scallops (anterolateral [A1, P1], middle [A2, P2], or posteromedial [A3, P3]); and atrial or ventricular side of the leaflets was determined. Location of the aortic valve vegetations with regard to left, right, or noncoronary cusps; tip, body, or margin of the cusps; and ventricular or aortic side of the cusps was also determined.
Using multiplane 2D TEE, the maximum diameter and area of vegetations were measured. With 3D TEE, the maximum diameter and area of vegetations were measured from the anteroposterior, superoinferior, and mediolateral dimensions. These 3D transesophageal echocardiographic measurements were performed offline using QLAB (Philips Medical Systems) and multiplanar reconstruction mode. For both techniques and using electronic calipers, three measurements during three different cardiac cycles were performed, and the maximum diameters and areas were analyzed.
The presence and severity of associated mitral valve anterolateral or posteromedial commissural fusion was assessed from the atrial view and defined as mild, moderate, or severe when involving less than or equal to one-third, up to two-thirds, or the entire corresponding commissural scallops, respectively. Aortic valve commissural fusion was assessed from the aortic root view and defined as mild, moderate, or severe when the corresponding cusps were fused less than or equal to one-third, up to two-thirds, or extended to the cusps’ central coaptation point, respectively.
Finally, the detection of valve vegetations by 3D and 2D TEE in relation to acute clinical neurologic syndromes, focal brain injury on MRI, neurocognitive dysfunction, or the combination of these three outcomes was determined.
Descriptive statistics are presented as frequency (percentage) and as mean ± SD. Paired comparison of the number of 3D versus 2D transesophageal echocardiographic studies with vegetations and the frequency of vegetations by either technique in relation to cerebrovascular disease were performed using the McNemar test. The individual mean counts of vegetations per study are reported as Poisson means. These mean counts with regard to locations (distal vs middle or proximal leaflet and middle [A2 or P2] vs anterolateral or posteromedial [A1, P1 or A3, P3] scallops for the mitral valve leaflets and left coronary cusp, noncoronary cusp, or right coronary cusp and cusp tip, body, or margin for the aortic valve cusps) as repeated factors were done by Poisson regression (PROC GENMOD in SAS; SAS Institute, Inc, Cary, NC). The comparisons of diameters and areas of vegetations between 3D and 2D TEE were done using paired t tests. Two tailed P values ≤ .05 were considered to indicate statistical significance. All statistical analyses were performed using SAS version 9.3.
Clinical and Laboratory Evaluations
Among the 40 patients with SLE who underwent TEE, a total of 34 (85%) had one or more manifestations of cerebrovascular involvement: 23 (58%) patients had acute neurologic syndromes, 25 (63%) had focal brain injuries on MRI, and 27 (68%) had neurocognitive dysfunction. As shown in Table 1 , patients with SLE were young, had a mean disease duration of 6.2 years, and had high levels of inflammatory markers; nearly 60% had positive antiphospholipid antibodies, the majority (55%) had been receiving prednisone for nearly 5 years, 41% were on nonsteroidal immunosuppressive therapy, and 38% were on antiplatelet or anticoagulant therapy.
Detection and Characterization of Libman-Sacks Vegetations by 3D and 2D TEE
As shown in Table 2 , the results of 3D TEE compared with 2D TEE were more often positive for mitral valve, aortic, and either valve vegetations ( P < .05 for all); 3D TEE detected a larger number of aortic valve vegetations and either valve vegetations (Poisson regression, P ≤ .001 for both), with a trend toward mitral valve vegetations ( P = .09); and 3D TEE determined larger diameter and area of mitral valve vegetations ( P < .05 for both) and larger maximum diameter of aortic valve vegetations ( P = .005).
|Valve||3D TEE ( n = 40)||2D TEE ( n = 40)||P|
|Studies with vegetations|
|Mitral valve||18 (45%)||14 (35%)||.046 ∗|
|Aortic valve||19 (48%)||12 (30%)||.008 ∗|
|Either valve||26 (65%)||20 (50%)||.01 ∗|
|Number of vegetations|
|Mitral valve||59 (1.48)||42 (1.05)||.09 †|
|Aortic valve||31 (0.78)||15 (0.38)||<.001 †|
|Either valve||90 (2.25)||57 (1.43)||.001 †|
|Maximum diameter and area of vegetations|
|Mitral valve vegetations (diameter, mm)||9.16 ± 5.76||5.3 ± 4.15||.03 ‡|
|Mitral valve vegetations (area, cm 2 )||0.58 ± 0.46||0.26 ± 0.34||.049 ‡|
|Aortic valve vegetations (diameter, mm)||5.59 ± 1.61||3.9 ± 1.26||.005 ‡|
|Aortic valve vegetations (area, cm 2 )||0.15 ± 0.07||0.12 ± 0.08||.21 ‡|
Also, as shown in Table 3 , 3D TEE compared with 2D TEE improved the localization of vegetations. Three-dimensional TEE detected more vegetations on the anterior mitral leaflet, anterolateral (A1, P1) and posteromedial (A3, P3) mitral scallops, and the ventricular side or both ventricular and atrial sides (vegetations probably extending into both sides) of the leaflets ( P < .05 for all) ( Figures 1–3 , Videos 1A–1F, 2A–2D, 3A, and 3B ; available at www.onlinejase.com ). In addition, 3D TEE detected a larger number of vegetations on the anterior or posterior mitral leaflets with an oval, tubular or roll-like, coalescent, or clustered shape affecting two or three contiguous scallops ( P = .03) ( Figures 1–3 ). The locations of mitral valve vegetations on the ventricular side or on both sides of the leaflets and vegetations affecting two or more contiguous scallops were originally described by Libman and Sacks. Furthermore, 3D TEE detected more aortic valve vegetations on the noncoronary cusps, on the coronary cusps’ tips and margins, and on both the aortic and ventricular sides (vegetations extending into both sides) of the cusps ( P ≤ .02 for all) ( Figures 4 and 5 , Videos 4A, 4B, 5A, and 5B ; available at www.onlinejase.com ).
|3D TEE ( n = 40)||2D TEE ( n = 40)||P ∗|
|Anterior leaflet||35 (0.88)||19 (0.48)||.02|
|Posterior leaflet||21 (0.53)||19 (0.48)||.68|
|Distal portion of anterior or posterior leaflets||41 (1.03)||34 (0.85)||.26|
|Middle or proximal portion of mitral leaflets||15 (0.38)||4 (0.10)||.07|
|Middle (A2 or P2) scallops||30 (0.75)||26 (0.65)||.46|
|Anterolateral (A1, P1) or posteromedial (A3, P3) scallops||26 (0.65)||12 (0.30)||.046|
|Ventricular side||9 (0.23)||6 (0.15)||.07|
|Atrial to ventricular side (protruding through leaflet)||4 (0.10)||2 (0.05)||.14|
|Left ventricular or both atrial and ventricular side||11 (0.28)||6 (0.15)||.04|
|Vegetations involving two or three contiguous scallops||15 (0.38)||7 (0.18)||.03|
|Right coronary cusp||8 (0.20)||7 (0.18)||.55|
|Left coronary cusp||9 (0.23)||3 (0.08)||.05|
|Noncoronary cusp||14 (0.35)||5 (0.13)||.002|
|Coronary cusps tip||18 (0.45)||8 (0.20)||.009|
|Coronary cusps body||3 (0.08)||5 (0.13)||.46|
|Coronary cusps margin||10 (0.25)||2 (0.05)||.004|
|Ventricular side||6 (0.15)||6 (0.15)||1.00|
|Aortic side||6 (15%)||2 (5%)||.14|
|Ventricular to aortic side (protruding trough)||11 (28%)||1 (3%)||.02|
|Both aortic or aortic and ventricular side||16 (40%)||3 (8%)||.01|