Chapter 11. Pulmonary Embolism

Approximately 600,000 patients per year are diagnosed with pulmonary embolism (PE).1 By far, most pulmonary emboli are thromboembolic in nature, although other causes, such as air, fat, amniotic fluid, tumor, and septic emboli, do occur.

The risk factors for thromboembolic PE mirror those for deep venous thrombosis (DVT) and include recent surgery, malignancy, confinement to a hospital or nursing home, confinement to a bed (“bed rest”), immobility, oral contraceptives or hormone replacement therapy, and paresis of an extremity.2 The mortality rate is high. In a large observational study of 2,388 hospitalized patients that evaluated cause of death, it was found that approximately 10% of all deaths were due to PE.3 With prompt diagnosis and treatment, mortality can be significantly impacted and the incidence of long-term complications such as chronic thromboembolic pulmonary hypertension and cor pulmonale can be reduced.

The clinical diagnosis of PE can be very challenging. The classic findings—or triad—of dyspnea, pleuritic chest pain, and tachycardia can be found in up to 95% of patients with confirmed PE. However, they are very nonspecific and could be indicative of any number of disorders. Furthermore, the first two, being symptoms, are difficult or impossible to evaluate in patients who are intubated and/or sedated, such as many ICU patients. Still, PE should be considered in any patient with any of these signs or symptoms.

Dyspnea is the most common symptom related to PE and is caused by a ventilation–perfusion (V/Q) mismatch. Chest pain is the second most common symptom, although up to one third of patients with PE deny chest pain or complain only of a vague chest discomfort. Hemoptysis secondary to PE is due to pulmonary infarction and is an uncommon and late finding. When present, it should significantly raise clinical suspicion. Fever is also rare but, if present, is usually “low grade” (<102°F), and appears more commonly when hemoptysis is present. Tachycardia is due to the cardiopulmonary stress created by the PE, but lacks sensitivity. Approximately 50% of all patients ultimately diagnosed with PE never demonstrated a persistent heart rate above 100 beats/min.3 Furthermore, patients on chronic β-blocker therapy will likely not develop a tachycardic response. Other clinical findings include diaphoresis, anxiety, cough, rales, murmur, syncope, cyanosis, and altered mental status. Cardiac arrest occurs in approximately 2% of patients suffering an acute PE, with pulseless electrical activity the initial rhythm in 60% of patients and asystole in 33%.4

Patients presenting with any of the above clinical findings along with unilateral arm or leg swelling have a significant risk for both DVT and PE; yet the finding of a single swollen limb is often overlooked. In the previously mentioned study, of the 2,388 autopsies performed, 83% of the patients with confirmed PE had evidence of DVT in their lower extremities but only 19% of them had reported symptoms of DVT prior to death.2

The clinical presentation of PE can be very nonspecific and the consequences of misdiagnosis can be dire. Therefore, clinicians should have a low threshold for ordering diagnostic testing. Because the diagnosis of PE is often elusive, a large body of current research focuses on diagnostic strategies to assist in accurately identifying patients with an embolism.

Any patient complaining of chest pain should have an electrocardiogram (EKG) and chest x-ray (CXR). These tests are nonspecific, however, and are more useful for ruling out other diagnoses than for diagnosing PE, per se.

The most commonly seen EKG abnormalities are sinus tachycardia and nonspecific ST-segment and T-wave changes.4 The classic EKG pattern of S1Q3T3, often quoted as “pathognomonic” for PE, is infrequently found. Furthermore, S1Q3T3 is not a specific finding for PE, so even when the pattern does appear, the diagnosis may still be other than PE. When it is found in the setting of PE, it is most commonly seen with massive acute PE and cor pulmonale. Overall, the EKG is best used to rule out other pathology, such as myocardial infarction and pericarditis.

Similarly, CXR is not sensitive for PE. Up to 25% of CXRs in patients with confirmed PE are normal. Nonspecific findings seen in patients with PE include atelectasis and pleural effusion, but these findings are equally common in patients without embolism.5 The classic CXR findings of Hampton's Hump (a wedge-shaped consolidation in a peripheral lung field) and Westermark's sign (dilation of proximal pulmonary veins with decreased vascular markings in the periphery) are also very uncommon; however, when present, they should raise clinical suspicion significantly. Like the EKG, the CXR is best done to rule out other pathology such as pneumonia or pneumothorax, or to diagnose or lend weight to an alternate diagnosis such as aortic dissection or esophageal rupture that would alter the diagnostic workup.

Oxygen saturation (SaO2) is an important tool when assessing for the possibility of a PE, and any degree of hypoxemia must be investigated. Conversely, an SaO2 of 100% while breathing room air tends to favor the absence of a PE. It may also be useful to obtain an ABG to further assess the Pao2 and Paco2, and calculate the alveolar–arterial (A–a) gradient, as the sensitivity of an abnormal Pao2 and/or A–a gradient is 90%.7 Furthermore, for those patients with PE, an ABG may help risk stratify and guide treatment as the Pao2 appears to correlate well with the extent of pulmonary vascular occlusion. However, a normal SaO2 and/or ABG alone cannot exclude PE, as up to 18% of patients with this diagnosis will still have an SaO2 of 100% and a Pao2 between 85 and 105 mm Hg.6

Compared with the above testing modalities, the concept of testing for D-dimers is relatively new. As a thrombus is forming, natural anticoagulation systems within the body begin to break it down. A degradation product of cross-linked fibrin, D-dimers are released as a thrombus is broken down. Many studies have evaluated the utility of D-dimer levels for predicting the likelihood of PE. It was found that among patients with a low pretest probability, a normal D-dimer was associated with at least a 99% likelihood of not having a PE. Conversely, approximately 95% of patients with confirmed PE had an abnormal D-dimer. Current evidence indicates that D-dimer analysis using enzyme-linked immunosorbent assays (ELISA) yields the best results with >95% sensitivity. A negative quantitative ELISA result (<500 ng/mL) is generally felt to have a sufficient negative predictive value for PE when the pretest probability of a patient is low. Further complicating the assessment, however, is the multitude of D-dimer assays available, each with its own sensitivity and specificity. It is therefore important to be aware of the parameters of the assay available at a particular institution, as well as the threshold used for diagnosing a positive result. In general, however, D-dimer assays are only useful when negative and then only in the setting of a low-risk patient.7–10

The traditional test for evaluating for PE was a V/Q scan. A V/Q scan can be useful for the evaluation of a PE, and the PIOPED study published in 1990 had much to do with its widespread use. The study, which enrolled almost 1,000 patients across six centers, included inpatients and outpatients, although neither the number of each nor, for inpatients, floor versus ICU status was reported.

In the study, clinicians determined the probability of PE prior to a V/Q scan being performed. It was found that patients with a high clinical probability and high-probability V/Q scan had a 95% likelihood of having a PE. Patients with a low clinical probability and low-probability V/Q scan had only a 4% likelihood of having a PE. Perhaps most useful, a V/Q scan read as “normal” virtually excluded PE11 (Note: “normal,” not “low probability”). Unfortunately, in the ED, only approximately 33% of patients will have a normal V/Q scan and only another 10% will have a high-probability V/Q scan, leaving a large percentage of patients for whom this test cannot be considered diagnostic. These patients require further diagnostic studies to rule out PE. Therefore, many institutions have not made this their initial study of choice.

Pulmonary angiography has long been considered the “gold standard,” and in some institutions it is the next study of choice in patients with nondiagnostic V/Q scans. However, due to the invasiveness of this procedure and the exposure to a significant contrast load, as well as the need for interventionalists to perform this study, it is reserved for patients in whom noninvasive studies are either equivocal or discordant with strong clinical suspicion. It may also be considered in patients for whom treatment of a PE is highly risky (e.g., metastatic disease of the brain and CNS) and the practitioner desires to be “as sure as possible” in weighing the risks and benefits of treatment.

CT pulmonary angiography has become the diagnostic modality of choice in most institutions across the United States. It is easily available and it has the benefit of detecting alternative pulmonary abnormalities.2–14 In one study, approximately 7% of CTs negative for PE found other abnormalities that required immediate attention, and another 10% required follow-up attention. In a large study of 824 patients, 83% of patients with a PE had a positive study (Figure 11-1), and 96% of patients without a PE had a negative study. CT angiography is also the study of choice in patients with an infiltrate on CXR or a history of emphysema, situations in which a V/Q scan is unlikely to be normal. There is great variation in sensitivity from institution to institution; this variability is dependent on the experience of the person reading the study, study quality, and resolution of the CT scanner. Furthermore, a pulmonary CT angiogram subjects the patient to radiation, high-pressure injection of nephrotoxic contrast medium, and transport out of the controlled environment of the ED or ICU. The clinician should take these concerns into consideration when ruling out PE as a diagnosis.

Echocardiography may be useful in unstable patients who cannot be transported, or for further confirmatory data in patients with suspected PE and equivocal results. It may reveal abnormalities such as increased right ventricular size, decreased right ventricular function, or tricuspid regurgitation.

Pregnancy and PE

It has long been acknowledged that pregnancy increases the risk of venous thrombosis and thromboembolism, and that PE is a leading cause of maternal death.5 This population presents a unique set of diagnostic challenges. D-dimer levels naturally increase during normal pregnancy, rendering the test relatively useless in ruling out thromboembolism if routine thresholds of 0.4–0.5 mg/L are used as a screening tool.6

Controversy still exists as to imaging modalities to utilize in pregnant patients with suspected PE. While lower extremity US may be useful if positive, there are many issues that may negatively impact sensitivity in pregnant females (increased diameter of lower extremity veins, decreased lower extremity venous flow velocity, nonpathologic leg swelling common in pregnancy).7 Therefore, this test is helpful only if positive. Ultrasound, however, does not expose the fetus to ionizing radiation and therefore is often recommended for initial screening in this population.

V/Q scan is less likely to be equivocal among pregnant patients than in the general populace, but the negative predictive value of CT angiography remains higher than that of the V/Q scan.8 Also, radiation dosages to the fetus are actually lower with CT angiography.9 In spite of this, many physicians still prefer to order V/Q scans in their pregnant patients due to a misconception that this test exposes the fetus to less radiation.20

Risk Stratification and Pretest Probability

As noted above, the majority of diagnostic testing available for the diagnosis of PE has significant limitations or risks. Therefore, it has been suggested that decision making and diagnostic testing should also incorporate a pretest clinical probability score. In 2004, the PIOPED II study was published evaluating the utility of CT for a suspected PE. Patients were classified as low, moderate, or high probability using the Well's criteria for PE (Table 11-1). In patient groups with an intermediate or high pretest probability, a CT positive for PE was found in 89% and 99% of patients, respectively. In patient groups with a low or intermediate pretest probability, a CT positive for PE was found in 0.5% and 7% of patients, respectively.21 One cost-effective approach in institutions where an ELISA D-dimer is rapidly available is to initially obtain a D-dimer on all patients deemed low risk based on their pretest probability. If negative, a CT does not need to be performed. If the patient has a moderate or high pretest probability for PE, D-dimer testing is not rapidly available, or clinical suspicion remains high despite low pretest probability, a CT pulmonary angiogram is recommended as the initial diagnostic modality. Patients with a CT negative for PE with a high clinical suspicion should undergo further diagnostic testing, likely with pulmonary angiography.

The above approach still yields a significant population that has a low score on the Well's criteria, a positive D-dimer, and, ultimately, a CT angiogram negative for PE. This has prompted further research in an attempt to determine if it is possible to establish criteria that place a patient at such low risk that even laboratory evaluation is not necessary. The pulmonary embolism rule-out criteria (“PERC criteria”), first published in 2004, is such an attempt (Table 11-2). In this study, patients felt to be at risk for PE who met none of the PERC criteria had a 1.6% chance of having an embolism. None of the patients clinically considered unlikely to have PE and who “ruled out” per PERC actually had a PE.22 This study has been prospectively validated with similar results.23 Many clinicians now utilize a combination of “gestalt” and the PERC to determine a threshold for diagnostic testing; if the PERC is negative and the clinician has a low index of suspicion of PE in a particular patient, the risk of further testing outweighs benefit to the patient and no further testing is required to eliminate the diagnosis of PE.4 It is important to remember that the PERC rule was not studied or validated for inpatients either on the floor or in the ICU.

Treatment

The main goal of treatment of PE is to maintain hemodynamic stability, reduce clot burden, and prevent extending or recurrent thrombosis. Initiating treatment early also helps decrease the incidence of the delayed complications of PE, such as chronic thromboembolic pulmonary hypertension and cor pulmonale. If treatment is not initiated early, the mortality rate reaches 30%, with the majority of these patients dying within the first few hours due to continued or recurrent thrombosis.5

As with all conditions, initial care of any patient presenting with suspected or diagnosed PE should focus on the status of the airway, breathing, and circulation. Hypoxic patients not needing ventilatory support should be placed on supplemental oxygen, but patients unable to maintain their airway, or who are in respiratory distress, should be intubated. Intravenous fluid (IVF) may be beneficial, but should be administered cautiously because large volumes may precipitate right ventricular failure. If a patient with hypotension does not respond to 500–1,000 cm3 of IVF, consider initiating vasopressor therapy.

The classic approach to treatment has been with unfractionated heparin (UFH) and should mirror dosages seen in the treatment of DVT. It may be prudent to initiate anticoagulation therapy prior to the diagnosis of PE in any patient with a high pretest probability, clinically high suspicion for massive PE, or unstable vital signs, especially if imaging is expected to be delayed.

LMWH has been shown to have similar efficacy and safety profiles as UFH in patients with PE and may be used in the treatment of acute PE. Some studies suggest it may also have a greater inhibition of in vivo thrombin generation.6 Since all LMWHs are cleared by the kidneys, care must be taken when administering the drug to patients with renal insufficiency; do not use LMWH in patients with a creatinine greater than 2.0 mg/dL. The recommended dosing is 1 mg/kg subcutaneously every 12 hours or 1.5 mg/kg subcutaneously once per day. Although no large studies have shown any mortality benefit, thrombolytic therapy does improve other important parameters such as right ventricular function and pulmonary perfusion. Most studies advocate the use of thrombolytic therapy in any patient with persistent hypotension or hemodynamic instability and proven or clinical suspicion of a massive PE. Other clinical circumstances that may also benefit include severe hypoxemia, large perfusion defects, severe right ventricular dysfunction, free-floating right ventricular thrombus, and a patent foramen ovale.27 The contraindications for thrombolysis for PE are the same as those when used for acute myocardial infarction or stroke. The most commonly used agents are alteplase (tPA) and urokinase.

Patients with contraindications for thrombolytics or who fail thrombolytic therapy may benefit from embolectomy. The procedure can be performed surgically or under fluoroscopic guidance with catheter fragmentation. These modalities of treatment are obviously very invasive and are limited only to institutions that have specialists experienced in these treatment modalities. Intraoperative mortality remains high, and there are no large studies available to guide decision making for transfer to tertiary care centers. It is likely that the high mortality is due to the extensive clot burden and hemodynamic compromise in patients considered for such interventions, rather than the actual procedure.28

Although not crucial in the acute setting (ED or ICU), at some point a decision must be made as to the optimal outpatient therapy. For cost considerations or because of patient ability, compliance, or preference, few patients maintain their therapy with outpatient injections of LMWH. For those patients who will not continue LMWH outside of the hospital, warfarin therapy should begin soon after the initiation of either UFH or LMWH. Once the INR is within 2.0–3.0 for 2 consecutive days, UFH or LMWH is discontinued. The patient should remain on warfarin for 3–6 months or indefinitely depending on his or her risk factors. Pregnant patients must be maintained on LMWH since warfarin is contraindicated as teratogenic and potentially lethal to the fetus, having been implicated in fatal fetal hemorrhage.

Venous air emboli are usually iatrogenic in nature, usually occurring during central line insertion (although they have been reported after peripheral IV insertion). For patients who decompensate during or immediately after central line placement or manipulation, a venous air embolus should be strongly considered. For subclavian and internal carotid lines, immediate aspiration from the central line is warranted since, if the catheter is in the right atrium, remaining air may be aspirated. If the procedure was a femoral central line or peripheral IV where the catheter does not extend to the heart, or if the procedure was a subclavian or internal carotid central line but aspirating does not help, a pulmonary artery catheter (PAC) may be placed and an attempt made to aspirate once the PAC is known to be in the right atrium. The patient should also be placed in the left lateral decubitus position as this position may “trap” the air in the right atrium, preventing further embolization to the lungs or beyond.

For patients who present after hyperbaric oxygen exposure (underwater divers, patients receiving hyperbaric oxygen therapy), the patient should be placed in the left lateral decubitus position. A PAC may similarly be placed and an attempt made at aspiration.

For female patients, especially if pregnant, who present to an emergency department with a history, symptoms, and signs consistent with PE along with a recent history of recipient orogenital intercourse, vaginal insufflation resulting in air emboli should be strongly considered. Therapy is the same as above.

For all patients with suspicion of air emboli, along with the above maneuvers, hyperbaric oxygen therapy may be considered if available and the patient may be safely moved to a chamber. For all patients in cardiac arrest unresponsive to CPR and ACLS, thoracotomy and needle aspiration of intracardiac air are appropriate.

Fat, or bone marrow, PE is a risk after any bone fracture or surgery, but most often occurs following long bone fracture or surgery. Unfortunately, while corticosteroids before surgeries such as intramedullary nailing have been shown to decrease the incidence, no therapy has been shown to adequately treat fat emboli once they occur. Since it can be difficult to differentiate a thromboembolic from fat PE, and since thromboembolic PE is more common, it is reasonable to treat for thromboembolic PE as above, and provide other supportive therapy as necessary. For severe hemodynamic collapse, consideration may be given to placing the patient on cardiopulmonary bypass.

Although amniotic fluid embolism is a consideration in the peripuerperal period, given the hypercoagulable state of pregnancy and increased risk of being bed bound before, during, and after vaginal delivery or C-section, thromboembolic PE is still more common. Therefore, since there has been no proven therapy for amniotic fluid embolism, and since thromboembolic PE is more common, as with fat embolism a reasonable approach is to treat for thromboembolic PE and provide other supportive therapy as indicated. As with antepartum patients discussed above, warfarin is teratogenic and should not be started until after delivery.

For septic emboli, the reader is referred to Chapter 38.