Impending Paradoxical Embolism FREE

In utero, the foramen ovale provides a means of bypassing the nonfunctional lung circuit by ensuring unidirectional blood flow from the right atrium (RA) to the left atrium (LA). After birth, the pressure in the LA rises in response to increases in systemic vascular resistance that results in the closure of the patent foramen ovale (PFO), and thereby prevents left-to-right (L-R) shunting of the circulation. During the first year of life, the septum primum is permanently sealed to the septum secundum by the formation of fibrous adhesions. In some instances, the foramen ovale does not seal completely, yielding a persistent PFO.^1- 2 Moreover, valvular competence is no guarantee of protection from a paradoxical embolism (PDE). Pathologic conditions that reverse the normal pressure gradient can cause transient valvular regurgitation, allowing a venous thrombus access to the systemic circulation.^3- 4

Paradoxical embolism refers to the embolic entry of a venous thrombus into the systemic circulation through a right-to-left (R-L) shunt.⁵ Prior to the advent of echocardiography, impending paradoxical embolism (IPDE) was usually discovered only on postmortem examination. Ingham³ described 6 patients with PDE preceded by pulmonary embolism (PE) whose autopsy findings revealed a venous thrombus in transit through a PFO. Johnson⁶ reviewed 39 autopsy reports in which the embolus lay in situ through a PFO, and 43 cases in which the diagnosis of PDE appeared certain.

The first case of PDE diagnosed before death was reported in the international literature in 1930.^7- 8 Cheng⁹ reported in 1976 that only 20 of 150 documented diagnoses of PDE had been made before death. We conducted a search of the English-language literature and found 29 reported antemortem cases of IPDE in which a venous thrombus extends from the RA through a PFO into the LA.^5,10- 32

A 55-year-old black woman with polycystic liver and kidney disease complicated by chronic ascites presented with a chief complaint of orthostatic syncope. Also, she reported generalized weakness but denied chest pain, palpitations, or prior syncope. The remainder of the review of her systems was unremarkable and she denied alcohol or tobacco use.

Physical examination revealed the following: body temperature, 36.7°C; regular tachycardia; blood pressure, 110/80 mm Hg supine; and tachypnea. She did not exhibit orthostatic changes in blood pressure or pulse, and there was no paradoxical pulse. On examination, the patient appeared cachectic, but there was no cyanosis. Examination of the neck revealed marked jugular venous distention but no Kussmaul venous sign. Her chest was clear on auscultation, and there was no cardiac murmur or rub. Abdominal examination findings revealed massive hepatomegaly and ascites. Although there was bilateral edema of her lower extremities, findings from the remainder of her physical examination were unremarkable.

Laboratory tests revealed the following values: white blood cells, 9.7×10⁹/L (85% neutrophils); hematocrit, 0.30; platelets, 198×10⁹/L; serum urea nitrogen, 6.4 mmol/L (18 mg/dL); creatinine, 132.6 µmol/L (1.5 mg/dL); and albumin, 33 g/L. Results of liver function tests were normal. The international normalized ratio was 1.35 seconds and the partial thromboplastin time was 23.8 seconds. Arterial blood gas while receiving 4 L of oxygen revealed a pH of 7.46; PCO₂, 23 mm Hg; and Pao₂, 64 mm Hg. Urinalysis demonstrated protein levels of 20 g/L. The chest radiograph revealed mild elevation of the right hemidiaphragm and bibasilar atelectasis, and the electrocardiogram revealed sinus tachycardia with low voltage but no electrical alternans, ischemia, P pulmonale, S₁Q₃T₃ pattern, or right ventricular hypertrophy.

A transthoracic echocardiogram (TTE) was obtained to evaluate for the presence of pericardial effusion and cardiac tamponade and revealed the presence of a large multilobulated mass in all 4 chambers consistent with thrombus, mild mitral and tricuspid regurgitation, and a mildly enlarged right ventricle and atrium (Figure 1). Additionally, a transesophageal echocardiogram (TEE) further revealed the large pedunculated echodense mass lodged in a PFO. The mass lesion prolapsed intermittently throughthe tricuspid and mitral valves (Figure 2).

A ventilation-perfusion lung scan was obtained and revealed multiple perfusion defects in both lung fields consistent with bilateral pulmonary emboli. A Doppler ultrasonogram of the patient's lower extremities revealed deep venous thrombosis (DVT). A hypercoagulability profile was positive for anticardiolipin IgG antibody.

The patient was heparinized but did not receive thrombolytic therapy because of concerns over possible arterial embolization of the thrombus. Three days later she experienced cardiac decompensation manifested as shock that was refractory to intravenous fluids, inotropic agents, and vasopressors, and underwent emergent cardiotomy for removal of the biatrial mass. Intraoperative findings revealed an enlarged and hypokinetic right ventricle and a 10×1-cm tubular blood clot extending from the tricuspid valve and crossing a PFO and septum into the LA, with prolapse through the mitral valve. The PFO, the mass, and the surrounding septal wall were excised, and the resulting septal defect was surgically repaired.

Subsequent hospital treatment included the placement of a Greenfield filter in the patient's inferior vena cava to prevent recurrent embolic events originating from her lower extremities. Her hospital course was complicated by the development of cardiogenic shock resulting from right ventricular failure, acute renal failure, respiratory failure requiring intubation and subsequent tracheostomy, bacterial peritonitis, and extensive thrombosis of the left upper extremity. Heparin therapy was begun for the thrombosis of her upper extremity on postoperative day 9, but on postoperative day 13, the patient developed hemorrhagic pericarditis, which was complicated by pericardial tamponade and cardiac arrest. The patient was resuscitated by bedside thoracotomy, open pericardiocentesis, and subsequent evacuation of a pericardial hematoma.

Following these complications, the patient made steady clinical improvement and was discharged from the hospital on day 44. Her renal, respiratory, and cardiac function returned to baseline, and she suffered no neurological sequelae. At 15 months of follow-up, she was still doing well.

A MEDLINE search of the literature was conducted for 1983 to 1996 using the key words paradoxical embolism, impending paradoxical embolism, thrombus, patent foramen ovale and thrombus, thrombus and right side of the heart, and intracardiac thrombus or mass. Additionally, selected references cited in articles were evaluated for pertinence and applicability.

Thirty cases of IPDE were reviewed (Table 1). Detailed analysis of summarized data proved difficult because of insufficient information. Many cases were reported as incidental findings among a series of patients being investigated for other purposes,^{14,17,26,31,33} as studies about PDE,⁵ as a PDE,^16,18 or in letter²³ or picture^13,22 form.

The clinical findings of IPDE were nonspecific and usually resulted from PE or systemic arterial embolism (Table 1). Those suggestive of PE included chest pain, dyspnea, syncope, presyncope, hypotension, a change in the results of cardiac auscultatory examination (increased S₂, new murmur, or additional heart sounds), hypoxia, tachycardia, electrocardiographic changes (S₁Q₃T₃), hemoptysis, and jugular venous distention. Findings suggesting arterial embolism included neurological changes consistent with a stroke or transient ischemic attack, a pulseless or cold extremity, flank pain with hematuria, and abdominal pain with bloody diarrhea. Noncategorized findings included coma, decreased breath sounds, bilateral pleural effusions, edema, acute myocardial infarction, and arrhythmias.

Of the 25 patients with known presentation, 14 patients (56%) presented with at least 1 of the findings suggestive of PE, and 2 (8%) also had symptoms suggestive of DVT. Seven patients (28%) presented with arterial embolism. Two patients (8%) presented with both PE and arterial embolism, and 2 patients (8%) had presentations that could not be categorized.

Table 2 shows a breakdown of the individual presenting signs and symptoms ranked in order of frequency.

No specific preexisting illness was consistently noted, and no comorbid conditions were listed for 13 patients. Risk factors for the development of venous thromboembolism were common and included cancer, recent surgery, immobilization, medical history of DVT and/or PE, obesity, old age, and the presence of anticardiolipin antibody. Of those with known comorbid conditions, 13 (76%) had at least 1 risk factor. Three patients (18%) may have had prior undiagnosed PDE events, including stroke (n=2) and both a stroke and a peripheral arterial embolism (n=1).

Seventeen patients had documentation of DVT (n=1), PE (n=7), or both (n=9). Of the 16 patients with documented PE, 12 (75%) had evidence of multiple emboli to the lung.

Twenty-two of the 30 patients had data reported about arterial embolic events:

Eight (73%) of the 11 patients who had acute arterial embolism had involvement of 1 site (36%); however, 3 (27%) had involvement at multiple sites, and included 1 patient with cerebral and peripheral involvement; 1 with splenic, mesenteric, and peripheral involvement; and 1 with carotid, renal, splenic, mesenteric, and peripheral involvement.

All patients underwent diagnosis with either TTE or TEE. Transthoracic echocardiography was not performed in 4 patients (13%); TEE was not performed in 15 patients (50%). Eleven patients (36%) underwent both studies. Eight patients (31%) who had negative or indeterminate TTE results underwent a TEE that established the diagnosis. Three cases (27%) of IPDE diagnosed by TTE were confirmed with TEE. Only 2 patients (7%) underwent contrast echocardiography. No comparison of sensitivity and specificity of these modalities can be inferred from these data.

Treatment was described for 26 patients (Table 1 and Table 3). Six (23%) were treated medically, 12 (46%) were treated surgically, and 8 (31%) received both medical and surgical treatment. Outcome was described for 25 patients. Nineteen (76%) recovered and 6 (24%) died.

Table 3 compares treatment modalities and outcomes in 23 patients for whom complete treatment and outcome data were available. Of the 6 who received medical treatment only, 4 (67%) recovered and 2 (33%) died. Heparin appeared as efficacious as urokinase and tissue-type plasminogen activator, but not streptokinase. However, conclusions about comparable efficacies cannot be drawn from these data because of the small numbers of patients. Of 10 patients who were treated with surgery, 8 (80%) recovered and 2 (20%) died. Seven patients were treated with a combination of heparin and/or thrombolysis and surgery, and all survived.

The clinical triad suggesting PDE was first described by Johnson⁶ and included (1) DVT and/or PE, (2) an intracardiac communication that permits an R-L shunt, and (3) arterial embolism not arising from the left side of the heart. However, a PDE could only be classified as proven if a venous thrombus was found within an intracardiac defect (usually a PFO) at autopsy.³⁴

The most important clue in the diagnosis of PDE is the presence of PE followed by systemic embolization.³⁵ Arterial embolism is the sine qua non of PDE.³⁴ Currently, a diagnosis of PDE requires that the following 4 criteria be met: (1) presence of DVT or PE, (2) abnormal communication between the right (venous) and left (systemic) circulations, (3) clinical, angiographic, or pathologic evidence for systemic embolism, and (4) presence of a favorable pressure gradient promoting R-L shunting.⁵

It is well established that DVT may be clinically occult. Even in patients with documented PE, DVT may escape clinical detection in more than 50% of cases.³⁶ Haeger³⁷ reports that of 512 patients treated for DVT, results of phlebographic studies were normal in 46%. Cranley et al³⁸ appropriately noted that the findings from physical examination are neither sensitive enough to exclude DVT, nor specific enough to be a basis for treatment.

Stöllberger and coworkers³⁹ describe 264 patients with clinically suspected arterial embolic events. All underwent contrast TEE and 49 (19%) were found to have a PFO. Of 42 patients who underwent bilateral ascending venography, DVT was documented by venographic study in 24. The authors conclude that when a PFO is detected in a patient with embolism, occult leg DVT is frequently present. Additionally, more patients with DVT than without had a history of a thromboembolism, and nearly one third of DVTs are recurrent events.⁴⁰ This suggests that in patients with embolic events, a history of DVT and the onset of symptoms associated with a Valsalva maneuver should alert the clinician to the consideration of a PDE.³⁹ Similar findings are reported by Hanna et al,⁴¹ who retrospectively identified 79 patients with PFO from 615 consecutive TEE studies. Sixteen patients (20%) had a prior symptomatic stroke thought to be secondary to the PFO and 5 (31%) had a documented concurrent DVT.

Although a PFO is the most frequent conduit for R-L shunts (>70%),^27,29 other abnormal communications reported include atrial septal defect, pulmonary arteriovenous malformation, ventricular septal defect, Ebstein anomaly, and patent truncus arteriosus.^{5- 6,42- 43}

In his review of 30 patients, Loscalzo⁵ reports the following arterial sites of PDE: 37% cerebral, 49% peripheral, 9% coronary, 1% renal, and 1% splenic. Other possible outcomes of this thromboembolic event likely include spontaneous lysis and fragmentation.²⁸

Autopsy data suggest that a PFO may be present in 11% to 35% of the normal population.^7,44- 45 In studying 965 normal hearts, Hagen and coworkers⁴⁵ found the incidence of a PFO for all age groups to be 27.3%. They found both sexes equally represented, with each exhibiting a decreasing incidence of PFO with aging. The age-related incidence was 34.3% for ages 0 to 29 years; 25.4% for ages 30 to 79 years; and 20.2% for ages 80 to 99 years.

Thompson and Evans⁷ examined the findings from 1100 consecutive autopsy reports and found a "probe patent" foramen ovale (measuring 0.2-0.5 cm) in 29%, and a "pencil patent" foramen ovale (measuring 0.6-1.0 cm) in 6%. Bridges et al⁴⁶ found a mean PFO size of 12.3 mm, using cineangiogram balloon sizing in their study of 36 patients who had suspected PDE.The significant difference in mean size in the study by Bridges et al⁴⁶ vs that in other studies may be explained by selection bias and the fact that formalin-fixed autopsy specimens shrink in size.

Patent foramen ovale in older patients is usually large (mean, 3.4 mm in the 1st decade vs 5.8 mm in the 10th decade), suggesting that a smaller PFO may seal with advancing age.^44- 45 Elderly patients may be at increased risk for passage of thrombus through a larger PFO.⁴⁴ Moreover, the prevalence of DVT, one of the main sources of PDE, is increased in the elderly.³⁶ Patent formen ovale has allowed passage of thrombi large enough to occlude the aortic bifurcation in 3 cases.⁴⁷ The thrombus in one case of IPDE, which was surgically extracted, measured 25 cm in length.²⁰ Conversely, cases of PDE have been reported in patients with PFOs smaller than 0.6 cm.⁶

In patients with normal hemodynamics and a PFO, there are no detectable abnormal findings in their medical history or on their physical examinations, chest radiographs, or electrocardiograms.^34,43 However, in pathologic conditions in which right atrial pressure (RAP) becomes elevated above left atrial pressure (LAP), a R-L shunt may occur. Shunting through a PFO has been demonstrated by TEE microbubble contrast studies (contrast echocardiography) in 3% to 10% of the population.⁴⁴ In another study,⁴⁸ TEE demonstrated R-L shunting in 14 (18%) of 76 healthy subjects with no clinical evidence of atrial septal defect. The incidence of PFO detected by contrast echocardiography is lower than that reported on autopsy studies and may be explained by the fact that even with provocative maneuvers (eg, Valsalva or cough), LAP may remain higher than RAP,⁴⁹ thereby preventing the demonstration of a functional PFO before death.

Reported causes of R-L shunting include (1) RA hypertension resulting from structural or mechanical disorders of the tricuspid valve^4,9,50; (2) RV hypertension; (3) RV failure with increased end-diastolic pressure⁹; (4) positive pressure ventilation; (5) positive end-expiratory pressure; (6) hypoxemia; (7) myocardial infarction of the right side of the heart⁵¹; (8) cardiopulmonary bypass; (9) platypnea and orthodeoxia associated with pneumonectomy or chronic liver disease; (10) neurosurgical procedures complicated by air embolism⁴⁵; and (11) Valsalvalike maneuvers in normal patients,^9,45,48,51 including defecation,⁵ micturition, and sneezing.²⁹

Among the numerous causes of R-L shunting, the most common is acute PE.^42,51- 52 An estimated 630000 cases of symptomatic PE occur annually in the United States,¹⁶ and it may be the third most common cause of death.³⁶ Pulmonary embolism has been demonstrated in as many as 85% of diagnosed cases of PDE.^5,21,27 Acute PE is generally considered a prerequisite for PDE.^4,6,43

The cardiovascular response to PE is dependent on the amount of embolic occlusion and the presence of preexisting cardiopulmonary disease.³⁶ Sudden occlusion of the left pulmonary artery causes a 29% rise in mean pulmonary artery pressures with a simultaneous fall in systemic arterial pressure.³ It has been estimated that occlusion of 33% or greater of the pulmonary arterial tree must be present before RAP exceeds LAP.⁷ Sharma and coworkers⁵³ found that pulmonary artery pressure rises proportionally to the extent of pulmonary vascular obstruction; however, mean pulmonary artery pressure did not exceed 40 mm Hg even with occlusion of 50% or greater of the pulmonary vascular bed. Significant RAP elevation occurred only when mean pulmonary artery pressure was 30 mm Hg or greater, suggesting that acute PE may lead to PDE only if it precipitates a rise in mean pulmonary artery pressure greater than 30 mm Hg. Furthermore, chronic pulmonary hypertension due to a single episode of PE probably does not occur.³⁶

Kasper et al²⁶ describe 85 patients with hemodynamically significant PE who underwent echocardiographic evaluation for PFO. Thirty-three patients (39%) were found to have had a PFO. Patients with a PFO and a large, hemodynamically significant PE were more likely to have a hospital course complicated by PDE and require resuscitation, intubation, or the use of vasopressor agents.²⁶ Chronically elevated pressures of the right side of the heart have been associated with 5% of PDE cases.²¹ These findings suggest that patients with PE might benefit from routine echocardiographic screening for PFO, but this warrants further investigation.

Echocardiography plays an indispensable role in the diagnosis and management of PFO and PDE^54- 55 and should be considered in patients with PE of undefined origin.⁵⁶ More recently, the diagnosis and management of these entities have been greatly improved with the advent of TEE, which is superior to TTE in detecting intracardiac masses^57- 58 and PFO.^{12,33,59- 65}

The echocardiographic characteristics of an intracardiac thrombus are those of a mobile mass of irregular shape, serpentine or lobulated, whose configuration changes during the cardiac cycle.^17,66 The major limitation of TTE is the inability to differentiate between an intracardiac thrombus and a myxoma.¹⁷ These studies suggest that the sensitivity for detecting intracardiac shunt or thrombus may be up to 3 times higher with TEE than TTE.⁵⁹

Manning and coworkers¹⁴ describe 231 patients who underwent TEE prior to elective mitral valve repair or left atrial tumor excision. Six percent of patients were found with TEE to have an LA thrombus and surgical confirmation was made in 86%, yielding a sensitivity of 100% and specificity of 99% for diagnosis of intracardiac thrombus. However, echocardiography still has limitations for the detection of thrombi, since LV thrombi measuring 1.5 cm are usually visualized but those that are 4 to 8 mm are not.⁶⁷

Finding a PFO can be as important as identifying an intracardiac mass because of the inherent risk of PDE. Chen et al⁶⁴ used contrast TTE and TEE to diagnose PFO in 32 patients undergoing cardiac catheterization for cardiac abnormalities. Transesophageal echocardiography detected 20 PFOs with Valsalva maneuvers and 14 with normal breathing, while TTE detected only 12 and 8, respectively. All but 1 PFO detected with TEE was confirmed by surgery or catheterization (with 1 false-positive finding from a patient with RA myxoma). When compared with TTE for the diagnosis of PFO, TEE had a sensitivity of 100%, and a specificity of 97%, while the results for TTE were 63% and 78%, respectively. However, an IPDE may yield a false-negative result on contrast echocardiography because of thrombotic occlusion of the PFO.³²

Contrast echocardiography is a safe (0.062% risk for transient adverse effects) and useful modality for the diagnosis of PDE,^68- 69 and may be superior to conventional echocardiography in detecting a PFO. Cheng⁷⁰ states that it is essential that physicians request bubble contrast studies in evaluating PDE, since these maneuvers may more effectively exclude the diagnosis.

A positive contrast study result⁶⁴ for a PFO occurs when 2 to 5 microbubbles measuring 3 to 5 µm cross the interatrial septum within 3 cycles of complete opacification of the RA.^24,41,60,71 However, contrast echocardiography provides only a qualitative result,⁴⁹ since no correlation exists between the amount of contrast passing the septum and the size of the defect.⁷²

The sensitivity of the contrast study may be increased with a Valsalva maneuver.^48,60,71,73 Lynch et al⁴⁸ show that contrast echocardiographic detection of interatrial R-L shunt increases almost 4-fold when performed with a Valsalva maneuver.

Valsalva maneuvers have been associated with 15% of PDE cases.^5,21 Valsalva maneuvers cause RAP to increase transiently above LAP, thereby reversing the normal L-R gradient that keeps the flap closed. Thus, provocative maneuvers, such as Valsalva or cough, could potentially dislodge an entrapped thrombus³²; however, no cases of this phenomenon have been reported.

Stoddard et al⁷⁴ report that the cough test is superior to the Valsalva maneuver in the diagnosis of PFO during contrast TEE. During this procedure, the patient coughs 3 to 5 times in rapid succession immediately after opacification of the RA with contrast medium. Among the 73 patients in their study, a PFO was detected with the cough test in 32 (43.8%), the Valsalva maneuver in 24 (32.9%), and quiet breathing in 18 (24.7%).

Intracardiac shunts may be demonstrated by cardiac catheterization. Characteristics of an intracardiac shunt include (1) typical oxygen saturation changes (oxygen step-up), (2) atrial pressure gradients, or (3) visual evidence of contrast medium traversing an abnormal communication.^1,43 To increase detection, it may be necessary to have the patient perform a Valsalva maneuver or cough during the cardiac catheterization to demonstrate the shunt.^43,75 Contrast TTE is at least as sensitive as cardiac catheterization.⁷⁵ Contrast TTE is also more sensitive than oximetry and dye-dilution curves; however, its sensitivity for detecting R-L shunts was found to be relatively low (64%).⁷² Presumably, contrast TEE would be more sensitive than TTE in the detection of R-L shunts, since it has been demonstrated to be more sensitive than TTE for the detection of intracardiac masses and PFO.

Nemec et al⁶⁵ found that shunting may be demonstrated using microbubble contrast medium study techniques while monitoring right middle cerebral artery blood flow with transcranial Doppler ultrasonography (TCD). They studied 32 patients (21 had unexplained arterial emboli) and compared the results of TTE, TEE, and TCD. Using TEE as the criterion standard, contrast TCD had a sensitivity and specificity of 100%. This suggests that contrast TCD may be a useful alternative modality in the detection of R-L shunting if TEE is unavailable or if the primary indication for TEE is to exclude an interatrial R-L shunt.

When an arterial embolism occurs without the presence of documented risk factors, the probability that the embolism originated in the left side of the heart is small; thus, PDE should be considered, and a search for DVT or PE should be made.⁴³ The finding of acute PE is prima facie evidence of DVT, and it also demonstrates that embolization has occurred. Pulmonary embolism is present in most reported cases of PDE.³⁴

It has been estimated that the yearly incidence of stroke in the United States is 400000.⁴⁶ Approximately one sixth of all ischemic strokes are due to cerebral embolism of a cardiac origin.⁷⁶ Several reports indicate that PDE may be more frequently involved.^{34,49,71,73,77} Prior to the widespread availability of TTE, Caplan et al⁷⁸ demonstrated that 35% of patients meeting criteria for embolic stroke had no identifiable arterial or intracardiac source. Additionally, the National Institute of Neurological Disorders and Stroke data bank⁷⁹ showed that roughly 40% of 1273 strokes reported were without an identifiable cause.

Harvey and coworkers⁸⁰ describe 11 patients younger than 50 years with unexplained stroke. Using contrast TTE, 8 patients were found to have a R-L shunt and 4 were subsequently found to have a PFO demonstrated at surgery. Hanna and coworkers⁴¹ reviewed 615 consecutive TEE reports and identified 79 patients with a PFO. Thirty-nine (53%) of these patients had symptomatic strokes and in 16 (41%) the stroke was thought to be due to PDE.

Biller et al⁷⁷ found that of 20 younger patients who had stroke, transient ischemic attack, or retinal artery embolus, 18 (90%) had evidence of a R-L shunt. Furthermore, stroke had occurred in 5 (25%) of these patients after events associated with a Valsalva maneuver, such as coitus, lifting, push-ups, or defecation. Similar findings were reported in the studies of Webster et al,⁷³ Lechat et al,⁴⁹ and Pearson and colleagues.⁶²

The Second Report of the Cerebral Embolism Task Force in 1989⁷⁶ stated that the likelihood of a PFO being unrelated and coincidental in young adults with unexplained stroke is about 1 in 3. It recommended that contrast echocardiography with Valsalva maneuver be performed in all young adults in whom unexplained stroke occurred.

Several studies suggest that a PFO may be an important risk factor in all age groups.^44,63,71 Di Tullio et al⁷¹ and De Belder et al⁴⁴ demonstrate that independent of age, patients with cryptogenic stroke had a higher prevalence of PFO than patients with stroke secondary to a known cause. Di Tullio and colleagues found a PFO in 18% of 146 patients in whom stroke occurred, and that the cryptogenic strokes were more frequently associated with a PFO in both younger (<55 years; 48% vs 4%) and older (≥55 years; 38% vs 8%) patients. De Belder et al⁴⁴ found a 26% prevalence of PFO in patients with cryptogenic stroke.

Based on these reports, PDE should always be included in the differential diagnosis of potential cardiac sources of stroke.⁷⁶ One should specifically mention PDE when ordering an echocardiogram to rule out the source of embolus so that appropriate Valsalva, cough, and contrast maneuvers may be performed.⁸¹ Di Tullio and colleagues⁷¹ suggest that because of the higher incidence of both stroke and DVT with advancing age, PFO detection in the elderly might be of greater relevance than its detection in younger patients. Furthermore, this study recommends that when a diagnostic workup fails to reveal a cause for stroke, all patients, regardless of age, should undergo contrast echocardiography.

Clinically, the diagnosis of PDE or IPDE is presumptive, relying on circumstantial evidence and a high index of suspicion. Obviously, simultaneous PE and systemic emboli strongly suggest the diagnosis of PDE or IPDE, although it is not a diagnostic prerequisite.

Patients present with nonspecific symptoms that are related to either PE or arterial embolism. A helpful historical clue would be the onset of the patient's symptoms with a Valsalva-type maneuver. In patients with cyanotic congenital heart disease and arterial embolism, PDE is likely; however, PDE should be suspected in any patient with unexplained arterial embolism.⁴²

In their review of right heart entrapment of PE in transit, Farfel et al¹⁷ note that reduction in cardiac output produced light-headedness, dizziness, fainting, and/or hypotension in 50% of patients and jugular venous distention in 30%. Dyspnea was present in 66% and chest pain in 33%. Additional diastolic sounds were present in nearly 50% and parasternal systolic murmurs were present in 33%. Cyanosis, cough, and hemoptysis were rare, and DVT was present in 18% of cases.

Because of the importance of DVT as an initial source of PDE, rigorous exclusion of DVT is especially important in patients with unexplained arterial embolic events. Patients with stroke with underlying chronic obstructive pulmonary disease or pulmonary hypertension should also undergo assessment for a PFO with contrast echocardiography.⁴²

As outlined in Figure 3, patients with arterial embolism (without evidence of PE) should have an evaluation beginning with TTE. If no intracardiac or proximal arterial source is located, then contrast TTE with cough test and Valsalva maneuver should be performed. If contrast TTE results are negative, contrast TEE should be performed. If TEE is unavailable, contrast TCD should be performed. Contrast TCD can be considered first as an alternative to contrast TEE.

At this time, no controlled clinical trials for the treatment of IPDE have been performed. Farfel et al¹⁷ describe 49 patients with RA thromboemboli. Patients who received surgical embolectomy had a 15% mortality rate. Those treated medically with thrombolytic agents, anticoagulants, or best supportive therapy had a 50% mortality rate. Four patients in this study had IPDEs. Of these, 3 were treated surgically and survived, and 1 was treated unsuccessfully with streptokinase. These authors thought that IPDE was best managed surgically. Similar data were reported by Armstrong et al.⁸²

In their meta-analysis of 71 studies, Kinney and Wright⁸³ report an overall mortality rate of 31% for patients with intracardiac clot. The presence of PE in conjunction with atrial clot significantly increased mortality. They found no difference in survival whether the patient was treated medically or surgically, and conclude that drug therapy should be chosen.

Given the often fatal consequences of either fragmentation or complete embolization of a thrombus trapped in a PFO, it is apparent that a wedged thrombus must be emergently removed. After removal of the thrombus, treatment of the underlying cause of the elevated RAP should be a primary concern. Reducing RAP to below that of LAP will restore hemodynamic homeostasis by reversing the shunt.¹

In Figure 4, we propose that treatment of IPDE should consist of initial systemic heparinization to reduce the immediate threat of further embolic phenomenon while awaiting emergent intracardiac embolectomy with PFO closure. The available literature suggests that embolectomy is the treatment of choice. Loscalzo⁵ recommends intracardiac embolectomy and inferior vena cava interruption. In special situations, such as refusal of an operation by the patient, thrombolysis may be an alternative in the absence of contraindications.²⁸ In IPDE, thrombolysis has the additional advantage of treating the PE and reducing pulmonary artery pressure. Thrombolysis may decrease the incidence of recurrent PDE and increase the patient's chance of surviving recurrent PE.²⁸

Several authors^{5,21,24,47,84} recommend that anticoagulation, unless contraindicated, should begin immediately once the diagnosis of arterial embolism has been established. Thrombolytic therapy should be considered if significant acute cor pulmonale develops because of acute PE.²⁴ Some studies^5,42,47 recommend thrombolysis in selected patients with suspected PDE and significant RA hypertension due to PE to decrease RAP and reduce the likelihood of recurrent PDE. Thrombolysis or embolectomy should also be used to treat peripheral emboli as indicated.^21,42

Inferior vena cava interruption, anticoagulation or antiplatelet therapy, surgical closure of the defect, and placement of a Greenfield filter have less defined roles.^5,10- 11,59 There are reports of recurrent PDE after placement of a Greenfield filter, and it has been shown that caval filters are not protective for emboli smaller than 3 mm.⁵⁹ Although some studies recommend embolectomy, inferior vena cava interruption, and repair of the cardiac defect,⁵⁰ there is little information published regarding long-term treatment and outcomes. Future clinical studies should be designed to address these issues and to determine whether prophylaxis of persons with recognized predispositions for PDE is beneficial.⁴² Another question is whether patients with hypercoagulable states should routinely be screened using contrast echocardiography for PFO, given its relatively high prevalence in the normal population.

Leonard et al⁴⁷ suggest that PFO closure may not be needed if RA hypertension and R-L shunt are thought to be temporary. Cheng⁸⁵ proposes that the treatment of choice for minute atrial septal defect as a cause of PDE is nonsurgical transcatheter closure. However, Murphy and coworkers⁸⁶ state that surgical closure is well established and has excellent long-term results. Surgical closure has been shown to reverse RV hypertrophy, dilatation, and RV conduction abnormalities. Even in patients older than 60 years, the procedure has an acceptably low morbidity and mortality rate.⁸⁷ Although nonsurgical transcatheter closure may be the future treatment of choice for atrial septal defect, no long-term results are currently available.^{42,46,88- 90}

Ultimately, the most cost-effective and efficacious diagnostic and treatment strategy is yet to be determined, and individual approaches will vary based on available hospital resources and expertise at any given institution. We propose a potential management algorithm in Figure 5.

In summary, our patient represents one of the few cases of IPDE reported before death. With the increasing prevalence of routine TTE and TEE, this finding may become more common. Any patient with neurological changes complicating cardiovascular events, DVT or PE, or any unexplained arterial embolism should be regarded with a high level of clinical suspicion for PDE. Suggestions for the evaluation and treatment of patients with IPDE and PDE have been outlined. The benefit of these approaches has not been tested in controlled clinical trials and requires further investigation. It is hoped that the use of TEE coupled with increased awareness of the phenomenon of PDE will help prevent catastrophic consequences by permitting timely surgical intervention.

Accepted for publication October 9, 1997.

This project was supported by Baptist Memorial Health Care Foundation grant 9506 (Dr Headley).

We wish to thank David Armbruster, PhD, for editorial assistance and Katrina Kimble for secretarial assistance in the preparation of the manuscript.

Reprints: A. Stacey Headley, MD, Room H316, 956 Court Ave, Memphis, TN 38163.