Author + information
- Received January 26, 2015
- Revision received March 11, 2015
- Accepted March 26, 2015
- Published online June 9, 2015.
- Syahidah S. Tamin, MBBS∗,
- Joseph J. Maleszewski, MD∗,†,
- Christopher G. Scott, MS‡,
- S.K. Khan, MBBS§,
- William D. Edwards, MD∗,†,
- Charles J. Bruce, MD∗,
- Jae K. Oh, MD∗,
- Patricia A. Pellikka, MD∗ and
- Kyle W. Klarich, MD∗∗ ()
- ∗Division of Cardiovascular Diseases, Mayo Clinic, Rochester, Minnesota
- †Division of Anatomic Pathology, Mayo Clinic, Rochester, Minnesota
- ‡Division of Biomedical Statistics and Informatics, Mayo Clinic, Rochester, Minnesota
- §Department of Radiology, Mayo Clinic, Rochester, Minnesota
- ↵∗Reprint requests and correspondence
: Dr. Kyle W. Klarich, Division of Cardiovascular Diseases, Mayo Clinic, 200 First Street SW, Rochester, Minnesota 55905.
Background Papillary fibroelastomas (PFE) are benign neoplasms with little available outcome data.
Objectives This study sought to describe the frequency and clinical course of patients with surgically removed PFE and echocardiographically suspected, but unoperated, PFE.
Methods Mayo Clinic pathology and echocardiography databases (January 1, 1995, to December 31, 2010) were queried, resulting in 511 patients: group 1 (n = 185), including patients with surgically removed, histopathologically confirmed PFE; group 1a (n = 94; 51%) with PFE removed at primary surgery; and group 1b (n = 91; 49%) with PFE removal at time of another cardiac surgery. Group 2 (n = 326) patients had echocardiographic evidence of PFE but no cardiac surgery to remove PFE.
Results Group 1 had mean age of 63 ± 14 years (116 women [63%]). During the study period, we identified 112 cardiac myxomas in the pathology database and 142 in the echocardiographic database. Mean age in group 2 was 67 ± 14 years (162 women [50%]). PFE occurred most commonly on cardiac valves (n = 400 [78%]). In group 1, transient ischemic attack or stroke was the presenting symptom in 58 patients (32%). With surgical removal of valvular PFE, the valve was preserved in 92 (98%). Recurrence was documented in 3 patients (1.6%). Follow-up stroke risk in groups 1, 1a, and 1b at 1 year was 2%, 0%, and 4%; at 5 years, 8%, 5%, and 11%, respectively. Cerebrovascular accident risk in group 2 at 1 and 5 years was 6% and 13%.
Conclusions In patients with echocardiographically suspected PFE who do not undergo surgical removal, rates of cerebrovascular accident and mortality are increased.
Cardiac papillary fibroelastomas (PFE) are small, benign endocardial lesions that are clinically important because of their documented embolic potential. Although generally considered a common benign cardiac tumor, they are not the most common. In 1 autopsy series (1), PFE were the third most frequently occurring cardiac tumor, behind cardiac myxoma and lipoma. Histologically, PFE are avascular structures composed of fibroelastic tissue surrounded by endocardium, imparting a characteristic echocardiographic appearance (2–4).
Historically, PFE were diagnosed incidentally at autopsy, thus reflecting probable underdiagnosis. With increasing use of echocardiography, improved resolution, and transesophageal techniques, more PFE are now being detected antemortem. We sought to better elucidate the true frequency of PFE and to define clinical characteristics and outcomes of surgically treated, histopathologically confirmed cases in a single, large clinical practice. Additionally, we sought to better understand the clinical course of PFE suspected on echocardiography where the patient did not undergo excision.
This study was approved by the Mayo Clinic Institutional Review Board. The pathology and echocardiography databases of Mayo Clinic (Rochester, Minnesota) were queried for cases of PFE and cardiac myxoma occurring between January 1, 1995, and December 31, 2010, for retrospective review. Inclusion and exclusion criteria were applied (Figure 1). Patients were excluded from the study if any of the following was present:
• Patient declined access to their records for research.
• Inflammatory or infectious disease was present at time of incident PFE (n = 224).
• Transesophageal echocardiography (TEE) did not confirm findings of PFE suspected on transthoracic echocardiography (TTE) (n = 30) or, on review of images, PFE was not confirmed (n = 18).
• Surgeon could not confirm PFE or echocardiography-identified mass at surgery (n = 10) or histopathologic evaluation did not support the diagnosis (n = 33) (Table 1).
Patients were separated into 2 groups. Group 1 met histopathologic criteria for diagnosis of PFE. This group was divided into group 1a, including patients for whom PFE was the primary reason for surgical intervention, and group 1b, including patients for whom PFE excision was not the primary indication for surgery. Group 2 met echocardiographic criteria for PFE.
Histopathologic diagnosis was rendered by a cardiovascular pathologist (J.J.M. or W.D.E.) based on presence of branching endocardium-lined fronds containing an avascular fibroelastic core. To assess the relative frequency of PFE, we queried institutional pathology databases for the incidence of cardiac myxoma during the study period.
To identify patients with echocardiographic evidence of PFE, we queried the Echocardiography and Hemodynamic Laboratory database to identify all patients with the echocardiography diagnosis PFE. Echocardiograms were reviewed using previously published echocardiographic characteristics of PFE (3,4): a small mobile mass attached to an endocardial surface, with independent motion and a stippled or shimmering border (especially notable under high resolution or by TEE). These were specifically distinguished from Lambl excrescences, which are linear echocardiographic densities.
Follow-up data for stroke, peripheral embolism, and death were obtained from the medical records.
Continuous variables were summarized as mean ± SD or median (interquartile range [IQR]), as appropriate. These variables were compared between groups using a 2-sample test or nonparametric Wilcoxon rank sum test. Categorical variables were summarized as counts and percentages and compared between groups using the Pearson chi-square test.
Time-to-event outcomes were defined as time from index date to event for those with events and last known follow-up for those without events. The Kaplan-Meier method was used to describe event rates at time points of interest, and event rates were compared between groups using the log-rank test. Observed events in these patients were compared with published rates (5), age- and sex-matched, and compared with the 1-sample log-rank test.
To examine the association of baseline variables with outcomes, Cox proportional hazards regression models were used. The results of these models were summarized with hazard ratio (HR) and 95% confidence intervals (CIs). As a secondary analysis to compare mortality between patients with and without surgery, a propensity-based approach was used. The propensity of having surgery was estimated on the basis of variables in Table 2 in addition to New York Heart Association class and size of PFE. Patients were then grouped by quartiles of propensity score (area under the curve: 0.77), and Cox regression was used to estimate the effect of surgery within each stratum. These estimates were combined for presentation using a weighted average approach (6). Statistical analysis was performed using JMP (SAS Institute, Inc., Cary, North Carolina) and SAS (version 9.3, SAS Institute). Statistical significance was set at p ≤ 0.05, and 2-sided tests were used for all analyses.
During the study period, 575,772 unique patients underwent echocardiography at Mayo Clinic. After exclusions, the study group contained 511 patients (mean age: 66 ± 14 years; 60% women), representing 0.089% of all patients undergoing echocardiography during the 16 years. Patient characteristics in group 1 (n = 185) and group 2 (n = 326) are shown in Table 2. In both groups, PFE most commonly occurred on valves — aortic (n = 304 [59%]), mitral (n = 64 [13%]), tricuspid (n = 23 [4%]), and pulmonary (n = 9 [2%]) — with the remainder (n = 111 [22%]) on nonvalvular endocardial surfaces. During this period, 112 histopathologically confirmed cardiac myxomas were identified in the pathology database and 142 in the echo database.
Patients in group 1 had a mean age at diagnosis of 63 ± 14 years; 116 were women (63%). Most were initially identified with echocardiography (n = 152 [82%]). PFE diagnosis was made by preoperative echocardiography in 119 patients (64%), intraoperative TEE in 33 (18%), surgical inspection in 12 (7%), and gross or histopathologic evaluation alone in 19 (10%). In the 98 patients with both TTE and TEE, PFE could be seen only by TEE in 32 (33%) and on both TTE and TEE in 50 (51%).
Indications for echocardiography are shown in Table 3. Group 1 had more structural heart disease (p < 0.001); group 2 had more atrial fibrillation (p < 0.001) and neurologic events (NE) (including transient ischemic attack [TIA], stroke, or both) (p = 0.045).
PFE location and size are shown in Figure 2. In group 1, mean PFE size was 9.76 ± 5.94 mm (IQR: 6 to 12 mm) by echocardiography, which was larger than in group 2 (7.57 ± 3.30 mm; IQR: 5 to 10 mm) (p < 0.001). In groups 1 and 2 combined, 183 patients had a previous NE (Table 4). Group 2 had more patients presenting with a history of NE than group 1 did (58 [32%] vs. 125 [38%]; p = 0.045). Group 2 had more NE as indication for echocardiography (107 [33%] vs. 45 [24%]; p = 0.04). In both groups, patients with an NE as a presenting symptom had significantly smaller PFE at diagnosis (7.13 ± 4.21 mm; IQR: 4 to 8.5 mm) than those who presented with other indications (8.95 ± 4.43 mm; IQR: 6 to 10 mm) (p < 0.001).
In group 1, the PFE was valvular in 153 patients. The predominant location was the aortic valve in 96 (63%), followed by left atrium or ventricle in 24 (16%), mitral valve in 14 (9%), tricuspid valve in 9 (6%), right atrium or ventricle in 9 (6%), and pulmonary valve in 1 (<1%). In 126 patients (82%), there was no functional valvular abnormality, but 23 (15%) had at least moderate dysfunction likely due to underlying valve disease (not to the PFE), including rheumatic disease (n = 6), myxomatous mitral valve disease (n = 5), previous heart surgery (n = 4), and post-inflammatory valve disease (n = 4), and 1 patient each with degenerative valve disease, congenital heart disease, pulmonary hypertension, and aortic root dilation. Only 1 patient had mild-moderate regurgitation with no obvious valvular abnormality other than PFE. Thirty-eight patients (21%) had multiple PFE (range 2 to 40).
The valvular PFE in group 2 had only mild valvular stenosis or regurgitation in 300 patients (92%). Forty-four patients (14%) had more than 1 PFE. An attempt was made to determine predictive features of PFE. Echocardiographic characteristics of the PFE were not significantly associated with a cerebrovascular accident (CVA): size (RR: 1.04; 95% CI: 0.94 to 1.16; p = 0.41), mobility (RR: 0.96; 95% CI: 0.52 to 1.78; p = 0.89), and aortic valvular location (RR: 1.26; 95% CI: 0.58 to 2.77; p = 0.56).
In 18 patients of group 1, PFE was diagnosed on pathologic inspection alone: 6 (33%) on excised aortic valves; 4 (22%) on left ventricular outflow tract endocardium; 3 (17%) on excised mitral valves; 2 (11%) on excised tricuspid valves; and 1 each (6%) on the right ventricular outflow tract endocardium, the left atrium, and a pacemaker lead. In this group, 8 (44%) had congenital heart disease, 6 (33%) rheumatic heart disease, 2 (11%) degenerative fibrocalcific aortic valve disease and hypertrophic cardiomyopathy, 1 (6%) myxomatous mitral valve disease, and 1 (6%) PFE was on the pacemaker lead. These patients were all in group 1b.
Compared with group 1b, group 1a patients had more NE, more chest pain, and significantly more were in New York Heart Association class I (Table 5) at diagnosis, but ages were similar (p = 0.16).
Group 1 follow-up
Median post-operative clinical follow-up was 1.6 years (IQR: 0.3 to 4.2 years), including echocardiography in 122 patients (66%) at a median of 2.9 years (IQR: 0.9 to 5.3 years). Recurrent PFE was documented in 3 patients (1.6%). In group 1a (n = 94), the native valve was preserved in 92 patients (98%). One patient required an aortic valve replacement at the time of surgery because the valve could not be repaired after the PFE excision and another patient had an attempted aortic valve repair at the time of PFE removal that later failed, necessitating valve replacement.
In group 1, overall risk of CVA was 2% at 1 year and 8% at 5 years (p = 0.003) (Table 6). Ten CVA were recorded during follow-up compared with 4.1 expected CVA based on age- and sex-matched rates. In group 1a, risk of CVA was 0% at 1 year and 5% at 5 years (p =0.81), with 2 CVA recorded in follow-up (1.7 expected CVA). In group 1b, risk of CVA was 4% at 1 year, 11% at 5 years (p < 0.001); 8 CVA were recorded in follow-up versus 2.4 expected.
For group 1, 30-day survival was 99% (100% for group 1a and 96% for group 1b). In group 1a, median survival was 9.4 years. At 1 year and 5 years, 95% and 86% of patients, respectively, were alive, compared with 94% and 86% in group 1b who had surgery for other indications (p = 0.32, log-rank test) and also had a PFE excised (Figure 3).
Group 2 follow-up
Follow-up was available in 317 patients (97%), median length 1.7 years (IQR: 0.2 to 4.8 years). At 1 and 5 years, the risk of CVA was 6% and 13%, respectively, and 99% and 98% were without clinical features of peripheral embolism. There were 29 CVA during follow-up versus 8.4 CVA expected (p < 0.001). The risk of CVA compared with the age- and sex-matched rates is outlined in Figure 4. Fifty-four patients (17%) were referred for, but did not ultimately undergo, surgery. Observed survival compared with age- and sex-matched rates (Figure 5) was significantly decreased (p < 0.001).
Only 8 patients in group 2 did not also have other risk factors for embolic events (e.g., atherosclerosis, atrial fibrillation, systemic hypertension).
Of 121 patients in group 2 with a history of NE, anticoagulant medications were taken by 87 (72%) at the time of PFE diagnosis. These therapies included warfarin in 27 (22%), aspirin in 57 (47%), clopidogrel in 1 (1%), and dual antiplatelet therapy in 2 (2%). There was no significant difference in the occurrence of subsequent stroke between patients treated with warfarin or antiplatelet drugs versus those not taking medication. The 5-year rates of freedom from CVA in patients taking warfarin, aspirin, or clopidogrel (alone or with aspirin) were 86%, 87%, and 91%, respectively, which was not significantly different from those not taking medication (p = 0.39).
Comparison of groups 1 and 2
One- and 5-year survival rates were 98% and 84% for group 1 and 87% and 67% for group 2 (p = 0.02, log-rank test). HRs were estimated using Cox proportional hazards regression to compare groups, and with age adjustment, the HR for group 1 versus group 2 was 0.67 (95% CI: 0.44 to 0.97; p = 0.03). A similar HR was estimated in the propensity stratification approach (HR: 0.68; p = 0.08). In group 1a at 1 year and 5 years, 95% and 86% of patients were alive compared with 94% and 84% in group 1b (p = 0.32, log-rank test). These groups also had similar survival after age adjustment (p = 0.57).
A difference in mortality was observed between group 1a and group 2 (HR: 0.51; 95% CI: 0.29 to 0.92; p = 0.02) but not between group 1b and group 2 (HR: 0.69; 95% CI: 0.43 to 1.10; p = 0.12) (Figure 3). However, with age adjustment, similar results were observed for comparison among groups: Group 1a versus group 2 (HR: 0.54; 95% CI: 0.30 to 0.99; p = 0.05) and group 1b versus group 2 (HR: 0.71; 95% CI: 0.44 to 1.15; p = 0.16). The estimated difference in mortality between group 1a and group 2 remained similar in the propensity-based analyses, but it was no longer significant (HR: 0.61; p = 0.10). Observed rates for survival and CVA in group 2 were compared with published rates matched for age and sex; the observed rates were higher than expected for both endpoints (p < 0.001) (Figures 4 and 5).
This study was undertaken to describe PFE frequency, clinical importance, and characteristics. To our knowledge, this is the largest series of patients with PFE, with follow-up in both operated and unoperated patients. We reached 5 important conclusions from this work.
First, PFE is more common than cardiac myxoma, at a rate of approximately 2:1, for a rate of 1 PFE per every 1,100 echocardiograms in our referral base population. Assuming these lesions are indeed neoplastic, this rate makes them the most common benign primary cardiac neoplasms of adulthood.
Additionally, a significant association exists between clinically diagnosed PFE and NE. Both mortality and risk of subsequent CVA are greater in patients with PFE identified with echocardiography and not removed than with age- and sex-matched rates. The risk of CVA is 6% at 1 year and 13% at 5 years.
PFE cannot be stratified into high- and low-risk types based on echocardiographic characteristics.
PFE can be safely excised with preservation of the native valve in experienced surgical centers. Long-term results are excellent in patients undergoing primary excision.
Finally, recurrence of a PFE after removal develops in 1.6% of patients and holds potential implications for follow-up.
Frequency of PFE
Initially, there appears to be a temporal increase in PFE diagnosis from 0.019% with the original 1997 report from our institution (3) to 0.089% seen here. However, multiple reasons may account for this rise, including increasing use of echocardiography, better technology with improved resolution, enhanced awareness of PFE, and an aging population. A less likely reason: an actual increase in PFE incidence.
Cardiac myxomas are widely recognized as the most common primary benign cardiac neoplasms of adulthood, but most of the supporting data were derived from an autopsy series (1,2). During the study time frame, we identified 112 pathologically confirmed cardiac myxomas at our institution, compared with 185 PFE. This suggests that PFE, rather than cardiac myxoma, is the most common benign cardiac neoplasm of adulthood, assuming PFE is indeed a neoplastic (and not reactive or hamartomatous) lesion.
Two previously published studies have implications in comparison with our study. We confirmed several observations and were concordant with the findings of Sun et al. (4): PFE most commonly arose on valves, though not infrequently on nonvalvular endocardial surfaces (22%), and relatively small lesions (<1 cm) did not cause valvular dysfunction. Compared with a meta-analysis by Gowda et al. (7), our study had significantly more women (p = 0.001). All studies showed a direct correlation of PFE incidence with age: mean age was in the sixth decade in the study by Sun et al. (4), the seventh decade in the present study, and a peak incidence in the eighth decade in the meta-analysis (7). These data suggest that although PFE can be seen at any age, the risk increases with age.
All 3 studies suggest an association of PFE with thromboembolic events. The incidence of a presenting NE reported in the previous studies was 17% (7) and 13% (4). In our study, the indication for echocardiography was NE in 32% of group 1 (CVA: 13%; TIA: 19%) and 36% in group 2 (CVA: 24%; TIA: 13%). Follow-up in group 2 revealed an 8.9% incidence of new CVA. Unlike previous landmark studies, ours is the first to show the incremental risk of PFE on subsequent CVA above baseline risk, which in unoperated PFE was 6% at 1 year and 13% at 5 years, which are higher than age- and sex-matched rates.
Sun et al. (4) found that the incidence of CVA or TIA in 45 patients was 6.6% when monitored for 11 ± 22 months after the echocardiography-based diagnosis of PFE. The present study’s findings are similar: group 2 had a CVA incidence of 6% at 1 year (p = 0.78). Of note, 33% of group 2’s patients had already had a TIA or CVA or both, and they were older than the group in Sun et al. (4) (67 ± 15 years vs. 58 ± 16 years).
Indeed, even in group 1 after PFE excision, a small risk of CVA persisted, likely reflecting the background risk of CVA in this patient population, per age group (5,8). In group 1a, where the main reason for surgery was PFE removal, post-operative CVA risk was not greater than expected (p = 0.81), whereas in group 1b, where the patient population was sicker at the time of surgical intervention, it is not surprising that CVA risk was higher than expected (p < 0.001). Some of this risk may be due to recurrent PFE, and this notion requires further investigation. In the past, our group has suggested that endocardial injury (e.g., surgery) may predispose to PFE (3,9).
In these published studies, treatment was determined by the managing clinician and the patient. Thus, the decision to have the PFE surgically excised was influenced by perceived risk versus benefit. The analysis by Gowda et al. (7) determined that PFE mobility was the only characteristic predictive of embolism. In our model, we identified no single echocardiographic characteristic predictive of such risk.
A previous surgical series by Ngaage et al. (10) compared patients in whom PFE was the primary indication for surgery to those in whom PFE was an incidental surgical finding. This led us to look at group 1 in the distinct subgroups of group 1a and group 1b. Both the present study and the one by Ngaage et al. (10) showed the high success rate of shave resection, sparing the valve.
Echocardiographic characteristics and diagnosis of PFE
The overall incidence of PFE in patients referred for echocardiography was 0.089%. Of note, 47% of the total study group had both TTE and TEE, 27% had TEE only, and 17% had TTE only; it is possible that if TEE were done for all patients, PFE incidence might be higher. Our PFE size range, from 2 to 40 mm, is smaller than reported in the meta-analysis by Gowda et al. (7) but similar to that of Sun et al. (4). Although the location and distribution of the PFE in our study are concordant with those reported in previous studies, ours showed more aortic valve PFE (64% vs. 45% (4) and 31% (7)). The next most likely location is the mitral valve, followed in decreasing frequency by cardiac chambers, tricuspid valve, and pulmonary valve. Similar to the study by Sun et al. (4), we found PFE in left-sided chambers to be significantly larger than those on left-sided valves (p = 0.005). In accord with previous studies, we found that the PFE were not typically thought to cause valvular dysfunction.
The indications for diagnostic echocardiography were significantly different between groups 1 and 2. More patients in group 1 had structural heart disease, whereas more patients in group 2 had an NE, atrial fibrillation, or heart failure. Nonvalvular PFE occurred more frequently in the left-sided chambers. Right-sided PFE were significantly larger than left-sided, possibly because right-sided lesions are less clinically apparent and manifest later. An almost equal proportion of PFE arose from the aortic and ventricular surfaces of the aortic valve; this varies from 2 previous studies where the former (4) and latter (10) predominated.
Approximately one-quarter of all patients in both groups had PFE detected on TEE but not TTE, reinforcing TEE’s importance when investigating embolic phenomena. There were 34 patients who had PFE suggested from echocardiography, but the lesion proved to represent something else histologically (Table 1). This underscores that echocardiography can suggest, but cannot ultimately make, the diagnosis. This finding also highlights the need for echocardiographers to be familiar with PFE characteristics. Almost one-fifth of group 1 had their PFE unexpectedly diagnosed during surgery with intraoperative TEE surveillance, which is routinely used in most cardiovascular surgical cases in our institution, highlighting the utility of such evaluation. The majority of PFE missed on TEE were detected in the setting of a serious operation for underlying cardiac disease. A previous meta-analysis (7) reported that most PFE were discovered incidentally in patients being evaluated for CVA, which is similar to what we found. The most likely reason PFE might not have been detected on TTE but was seen on TEE is lesion size.
The incidence of 1.6% for echocardiographically detected, post-operative recurrence of PFE has not been reported previously (4,7,10). These recurrences were diagnosed at 1, 5, and 6 years after the original excision. PFE are thought to be either neoplastic, hamartomatous, or reactive lesions. New PFE may arise at or near the site of previous resection due to endocardial injury, but also they may arise from lesional cells left behind during incomplete resection. The surgeon may be able to tell patients that the PFE was grossly removed, but cannot tell them that another PFE will not develop in the future near that site or elsewhere in the heart. We had 1 patient who had undergone a subaortic septal myectomy with a single PFE excised and then had more than 40 PFE removed from the same site many years later (9). Her surgeon described the ventricle as looking like a shag carpet.
In group 1a, which had PFE excision as a solitary procedure, post-operative survival and 1-year survival were 100%, and 1-year freedom from CVA was 100%. At 5 years, there were 2 CVA, no evidence of recurrence, and no residual valvular regurgitation. Two patients needed aortic valve replacement. Group 1b contained patients with serious cardiac disease requiring surgical intervention. Thus, it was a sicker group than group 1a was and, not surprisingly, had a different post-operative outcome. In group 2, 8.9% of patients had a subsequent CVA. No identifiable echocardiography characteristic of the PFE was significantly associated with CVA. We therefore could not define high- or low-risk characteristics of PFE, unlike in the study by Gowda et al. (7), wherein characteristics of mobility and aortic location were associated with increased risk of embolism. Their observation may be due to greater sample size.
This is the first report of event-free and overall survival in unoperated PFE patients. In group 2, patients had higher incidences of CVA and death than expected based on age- and sex-matched rates, and CVA risk was 6% at 1 year and 13% at 5 years. Patients in group 2 were older and had more traditional risk factors for CVA than did group 1, yet the risk of CVA and death held up in age-adjusted modeling and the incidence of risk factors was similar to the general population when adjusted for age. This information may be useful when counseling patients who have an incidental PFE on echocardiography but no other abnormality necessitating surgery.
In group 2, 55 patients (17%) did not undergo PFE excision despite the managing physician’s recommendation. The risk-to-benefit ratio of cardiac surgery was likely influenced by older age, comorbid conditions, and perhaps the uncertainty of embolic risk. In some referral cases, the patient declined surgery even though the surgeon was prepared to operate. This study adds to the current literature by quantifying the risk of PFE and its association with CVA in follow-up compared with background risk.
The study may not have the power to evaluate the benefit of anticoagulation, antiplatelet therapy, or dual antiplatelet therapy. Nevertheless, we did not observe a significant difference in the occurrence of subsequent CVA among patients treated with such agents.
The increased occurrence of CVA in patients with PFE, despite comorbid conditions, is also a novel finding. We have found previous observations in studies by Sun et al. (4) and Gowda et al. (7) to be complementary to our present observations and expand current understanding through the follow-up comparison with age- and sex-matched rates.
At our tertiary referral center, we recommend that patients who are good surgical candidates (Society of Thoracic Surgeons score <1%) with left-sided PFE (regardless of size, mobility, or location) consider surgical excision. We base this recommendation on our practice’s excellent outcomes, the data suggesting that excision substantially decreases CVA risk from PFE, and the fact that the valve can usually (98%) be spared when excision is the primary surgical indication (Central Illustration). Clearly, institutional surgical expertise factors importantly in treatment decision making. Advances such as robotic approaches may further improve outcomes (11).
When a patient is not a surgical candidate or refuses surgery, we have been recommending long-term treatment with antiplatelet agents. Because supporting data are limited, we base this on the histologic observation that surface thrombi may occur.
Importantly, this is an echocardiography-based study. Although a PFE has classic echocardiography features, these characteristics cannot provide a diagnosis but instead provide only a differential diagnosis of the mass (3,4,7). We advise that patients have blood cultures, antiphospholipid antibodies, and be screened for lupus.
The current study is limited primarily by its retrospective nature and the inherent issues with follow-up in this type of study. It is also important to note the referral bias of our tertiary care center. The propensity-based approach, though an effective way to control for potential confounders, results in loss of power because of stratification into quartiles.
This large, single-center study confirms that PFE most commonly occur on cardiac valves, although they may be found on any endocardial surface. Although PFE were usually small, patients not undergoing surgery face increased risk of stroke and mortality. Surgical excision is effective, with a high likelihood of valve preservation and low recurrence rate when performed at a high-volume tertiary care center. However, recommendations for an aggressive surgical approach need to be placed in context of the local surgical expertise. Although this is the current approach at our institution, management may best be sorted out ultimately with a randomized trial.
COMPETENCY IN MEDICAL KNOWLEDGE: PFE should be considered in the differential diagnosis of source of embolism work-up. This study is the first large single-center review of the management of a relatively rare benign cardiac neoplasm. It suggests that PFE should be considered for removal in patients with low surgical risk.
TRANSLATIONAL OUTLOOK: Further clinical trials are needed to verify whether removal prevents central neurological events (CVA and TIA) and mortality, to define the optimal patients for surgical removal, and to determine whether medical management with antiplatelet or anticoagulation will improve outcomes, especially in patients who are not candidates for surgical intervention or who decline surgical intervention.
The authors have reported that they have no relationships relevant to the contents of this paper to disclose.
Presented as a poster and Rapid Fire Session at the annual meeting of the American Society of Echocardiography, on July 2, 2012, in Washington, DC.
- Abbreviations and Acronyms
- confidence interval
- cerebrovascular accident
- hazard ratio
- interquartile range
- neurologic event(s)
- papillary fibroelastoma(s)
- relative risk
- transesophageal echocardiography
- transient ischemic attack
- transthoracic echocardiography
- Received January 26, 2015.
- Revision received March 11, 2015.
- Accepted March 26, 2015.
- American College of Cardiology Foundation
- McAllister H.A. Jr..
- Klarich K.W.,
- Enriquez-Sarano M.,
- Gura G.M.,
- Edwards W.D.,
- Tajik A.J.,
- Seward J.B.
- Sun J.P.,
- Asher C.R.,
- Yang X.S.,
- et al.
- Go A.S.,
- Mozaffarian D.,
- Roger V.L.,
- et al.,
- for the American Heart Association Statistics Committee and Stroke Statistics Subcommittee
- National Institutes of Health
- Bonnichsen C.R.,
- Burkhart H.M.,
- Klarich K.W.,
- Suri R.M.