Background
Incidental extracardiac findings (ECFs) have been described and studied in myocardial perfusion imaging, cardiac computed tomography, and cardiac magnetic resonance scanning. The literature is surprisingly limited with regard to ECFs in echocardiography. The aim of this study was to evaluate the prevalence and the clinical significance of ECFs in routine echocardiographic studies.
Methods
The literature in other cardiovascular modalities was searched to identify and classify ECFs. ECFs in reports of transthoracic and transesophageal echocardiographic studies performed at Temple Health Network between 2009 and 2011 were sought. A sensitivity analysis was performed by reviewing the actual echocardiographic images for a subset of studies ( n = 350) to determine the sensitivity and specificity of the results. The electronic medical records of patients with ECFs on echocardiography were then retrospectively reviewed, except for those with pleural effusions and descending aortic atheroma.
Results
A total of 41,067 echocardiographic studies performed between September 2008 and September 2011 (39,269 transthoracic and 1,798 transesophageal studies) were screened. Of these studies, 66.5% were performed in the inpatient setting and 33.5% in the outpatient setting. The prevalence of ECFs was 4.4% (1,797 findings) and was constant during the study years. Pleural effusion was the most common ECF on transthoracic echocardiography, while descending aortic atheroma was the most common ECF on transesophageal echocardiography. Detailed chart reviews were performed in all patients with ECFs, except those with pleural effusion and descending aortic atheroma (351 cases). ECFs on echocardiography led to new diagnoses and altered management in the majority of patients with vascular or liver findings.
Conclusions
In this large consecutive series, ECFs on echocardiography were relatively uncommon and had variable clinical implications. The majority or ECFs are likely low-risk findings (pleural effusion, ascites, and hiatal hernia) and can be managed conservatively. “Higher risk” findings such as liver abnormalities, inferior vena cava filling defects, mediastinal masses, and descending aortic dilatation frequently lead to significant changes in clinical management. There is a need for uniform reporting, appropriate training, and the establishment of national guidelines for ECFs on echocardiography.
Extracardiac findings (ECFs) are not uncommon on cardiac imaging examinations. With the substantial increase in the use of noninvasive cardiac imaging, the recognition and management of ECFs have become important. Several investigations have suggested that the prevalence of ECFs is 1% to 4% on myocardial perfusion imaging and ranges between 10% and 60% on cardiac computed tomography (CT) and cardiac magnetic resonance scanning. Although echocardiography has always been the most widely used cardiac imaging modality, there is a paucity of data with regard to ECFs with the routine use of this modality. In a recent study of 1,008 patients, the authors found a prevalence of 7.5% on transthoracic echocardiography of noncardiac pathology. However, the study was limited by including echocardiographic images obtained only from subcostal views and interpreted by an experienced radiologist trained in cardiac imaging rather than by cardiologists. We sought to determine the prevalence and clinical significance of ECFs in a large, “real-world,” consecutive series of clinically indicated transthoracic and transesophageal echocardiograms interpreted by experienced cardiologists.
Methods
All adult transthoracic and transesophageal echocardiographic studies performed within the Temple Health Network (Temple University Hospital, Episcopal Hospital, and Jeanes Hospital) from September 2008 to September 2011 were studied. Routine, urgent, and technically limited studies were included. Stress echocardiographic examinations were excluded. Transesophageal and transthoracic echocardiographic reports were downloaded separately to a central database in Portable Document Format on password-protected computers. The study was approved by the institutional review board at Temple University.
Because there is no consensus definition of ECFs on echocardiography, we surveyed the scientific literature across other cardiac imaging modalities (cardiac CT, cardiac magnetic resonance, and myocardial perfusion imaging) to define and categorize ECFs. We considered findings in the liver, abdomen, lung, pleura, mediastinum, vena cava, pulmonary artery, and descending aorta extracardiac and included them in the study. Pericardial and ascending aortic findings were excluded, as we considered these to be cardiac findings.
ECFs were reported using Centricity software (GE Healthcare, Little Chalfont, United Kingdom) using primarily free-text typing. Nine echocardiography faculty members were surveyed for all possible phrases commonly used by them when reporting ECFs. A comprehensive list of keyword and phrases was then developed and used to search all Portable Document Format files in the database. A sensitivity test was performed to evaluate the accuracy of our keyword list by manually searching 1,000 echocardiographic reports for ECFs. Two different operators performed the search independently using Adobe Reader 9.5.0 (Adobe, Inc, San Jose, CA). The two operators cross-verified their findings in 20% of the studies. We also used Crystal Reports software (SAP AG, Walldorf, Germany) to search for certain characteristics of the study (inpatient vs outpatient, limited vs complete studies).
A detailed review of the actual images was also performed for a subset of transthoracic studies (sensitivity analysis). Two experienced readers (level II and III trained, both diplomates of the American Board of Echocardiography) were asked to interpret the images and note ECFs on a spreadsheet. The readers were blinded to the original reports of the studies.
Finally, we conducted a chart review using our electronic medical records system (Alpha Image, EPIC, PACS, and Centricity) for all patients with ECFs, with the exception of those with pleural effusion. Patients were considered to have adequate clinical follow-up if they had >6 months of contact with health care providers (inpatient or outpatient visits) or if they died during the hospitalization or during the follow-up period. The chart review aimed at finding whether the ECFs were new, resulted in dedicated follow-up testing, or had any significant clinical impact.
Results
Incidence of ECFs
A total of 41,067 echocardiographic studies performed between October 2008 and September 2011 (39,269 transthoracic and 1,798 transesophageal echocardiographic studies) were screened. Of these studies, 27,310 (66.5%) were performed in the inpatient setting and 13,757 (33.5%) in the outpatient setting. There prevalence of ECFs was 4.4% (1,797 findings) overall and was constant during the 3 years studied ( Figure 1 ). Pleural effusion followed by ascites and liver abnormalities were the most common ECFs on transthoracic echocardiography ( Figure 2 ). Descending aortic atheroma followed by pleural effusion were the most common ECFs on transesophageal echocardiography. Table 1 illustrates the incidence of each ECF stratified by transthoracic versus transesophageal echocardiography.
| ECF | TTE ( n = 39,269) | TEE ( n = 1,798) |
|---|---|---|
| Pleural effusion | 1,032 (2.6%) | 45 (2.5%) |
| Lung or mediastinal mass | 4 (<0.1%) | 1 (<0.1%) |
| Ascites | 129 (0.3%) | 3 (0.2%) |
| Liver abnormalities | 41 (0.1%) | 0 (0%) |
| Hernia | 13 (<0.1%) | 1 (<0.1%) |
| DA dilatation | 141 (0.4%) | 0 (0%) |
| DA thrombus or ulcer | 3 (<0.1%) | 34 (2%) |
| DA severe atheroma | 4 (<0.1%) | 107 (6%) |
| Mild to moderate DA atheroma | 19 (<0.1%) | 206 (11.5%) |
| IVC thrombosis | 10 (<0.1%) | 2 (0.1%) |
| Pulmonary embolism | 2 (0.1%) | 0 (0%) |
Sensitivity Analysis
It is possible that searching the reports only might have led to underestimation of ECFs if the initial echocardiographic reader recognized the finding but failed to comment on it in his or her report. To examine the sensitivity of the original “routine” read to detect ECFs in our database, a dedicated secondary review of the actual echocardiographic images was performed by two experienced readers on a selected subset of 350 transthoracic echocardiographic images (55 with ECFs and 295 without ECFs). The agreement between the original echocardiographic report and the secondary review was 89% overall (75% for studies with ECFs and 95% for studies without ECFs) and was similar for the two readers (89% and 90%). The secondary review disagreed with the echocardiographic report in 14 of 55 patients with ECFs (five with pleural effusion, four with descending aortic atheroma, two with ascites, two with liver abnormalities, and one with a hernia) and in 15 patients without ECFs (five with liver abnormalities, five with descending aortic atheroma, two with pleural effusion, two with ascites, and one with a hernia). In three cases, there was a disagreement on the actual ECF (ascites vs right pleural effusion). The original reports detected ECFs with 73% sensitivity, 95% specificity, 75% positive predictive value, and 95% negative predictive value compared with the dedicated secondary review ( Table 2 ).
| Reader 1 | Reader 2 | Overall | |
|---|---|---|---|
| Total studies reviewed | 175 | 175 | 395 |
| Agreed with positive ECF | 23 | 18 | 41/55 (75%) |
| Agreed with negative ECF | 141 | 139 | 280/295 (95%) |
| Disagreed with positive ECF | 5 | 9 | 14/55 (25%) |
| Disagreed with negative ECF | 6 | 8 | 15/295 (5%) |
Clinical Follow-Up for ECFs
A detailed chart review was performed in all patients with ECFs except those with pleural effusion and descending aortic atheroma (351 cases). Clinical follow-up meeting the prespecified inclusion criteria was available in 80% of patients. The mean duration of follow-up was 17.8 months. Echocardiography was the first imaging modality to identify the ECFs in 88% of the cases. Follow-up imaging studies such as CT or abdominal ultrasound were pursued in only 118 patients (34%) ( Table 3 ). Follow-up studies were more likely to be ordered when the ECF was a mass in the liver or the mediastinum and when echocardiography was done in the inpatient rather than the outpatient setting.
| ECF | Incidence | TTE | Inpatient | Follow-up available | Follow-up studies | Echocardiography led to new diagnosis requiring further management |
|---|---|---|---|---|---|---|
| DA dilatation | 141 | 100% | 117 (83%) | 105 (74%) | 71 | 22 (31%) |
| Ascites | 132 | 97% | 107 (81%) | 114 (86%) | 20 | 0 (0%) |
| Liver abnormality | 41 | 98% | 23 (56%) | 36 (88%) | 12 | 3 (25%) |
| Hernia | 13 | 93% | 12 (92%) | 8 (62%) | 2 | 2 (100%) |
| IVC mass/thrombus | 12 | 100% | 11 (92%) | 12 (100%) | 7 | 7 (100%) |
| Mediastinal mass | 6 | 83% | 6 (100%) | 6 (100%) | 6 | 3 (50%) |
| Pulmonary embolus | 2 | 100% | 2 (100%) | 2 (100%) | 2 | 2 (100%) |
Ascites
Ascites was a new clinical finding in 132 patients, but only 20 of these had follow-up studies (abdominal ultrasound and/or CT); all confirmed the ascites. Risk factors for developing ascites were present in 97 of the patients (heart failure, renal failure, liver failure, and pericardial disease). Diagnostic paracentesis was offered to 11 patients, of whom three declined and eight underwent the procedure. All eight patients were found to have transudative ascites (six due to hepatic cirrhosis and two to advanced heart failure).
Liver Abnormalities
Forty-one studies reported liver abnormalities. The liver findings were known before the echocardiographic studies in seven patients. Of the remaining 33 patients, only 12 had follow-up studies (abdominal ultrasound, CT, or both). Table 4 outlines the echocardiographic description and the final diagnosis of each liver finding for those 12 patients. One patient (subject 10) had a new diagnosis of liver adenocarcinoma that was first identified on echocardiography. Figure 3 illustrates a case of large right kidney cyst mimicking a liver cyst on echocardiography. Figure 4 illustrates the echocardiographic appearance of a gallstone.
| Subject | Echocardiographic description | Follow-up | Dedicated imaging diagnosis |
|---|---|---|---|
| 1 | Echo-dense | CT | Large simple right upper pole renal cyst invaginating into the right lobe of the liver |
| 2 | Echo-dense | CT | Multiple simple renal cysts and a peripheral wedge shaped hypodensity of the right kidney likely due to chronic infarction |
| 3 | Echo-dense | CT, ultrasound | Subcapsular pseudocysts, multiple gallstones with biliary sludge |
| 4 | Echo-dense | Ultrasound | Large liver cyst |
| 5 | Echo-lucent | CT | Multiple liver cysts |
| 6 | Echo-lucent | Ultrasound | Hepatomegaly with fatty infiltration |
| 7 | Liver cyst | Ultrasound | Multiple liver cysts |
| 8 | Liver cyst | CT | Areas of hypoattenuation in the kidneys consistent with ischemic insults; small liver and massive ascites |
| 9 | Liver cyst | Ultrasound | Multiple liver cysts |
| 10 | Liver mass | CT, ultrasound | Liver mass with metastatic lesions in the abdomen |
| 11 | Gallstone | Ultrasound | Gallstone |
| 12 | Gallstone | Ultrasound | Gallstone |
