9-25: Pulmonary Venous Thromboembolism

Pulmonary venous thromboembolism, often referred to as pulmonary embolism (PE), is a common, serious, and potentially fatal complication of thrombus formation within the deep venous circulation. PE is the third leading cause of death among hospitalized patients. Despite this prevalence, most cases are not recognized antemortem, and less than 10% of patients with fatal emboli have received specific treatment for the condition. Management demands a vigilant systematic approach to diagnosis and an understanding of risk factors so that appropriate preventive therapy can be given.

Many substances can embolize to the pulmonary circulation, including air (during neurosurgery, from central venous catheters), amniotic fluid (during active labor), fat (long bone fractures), foreign bodies (talc in injection drug users), parasite eggs (schistosomiasis), septic emboli (acute infective endocarditis), and tumor cells (renal cell carcinoma). The most common embolus is thrombus, which may arise anywhere in the venous circulation or heart but most often originates in the deep veins of the lower extremities. Thrombi confined to the calf rarely embolize to the pulmonary circulation. However, about 20% of calf vein thrombi propagate proximally to the popliteal and iliofemoral veins, at which point they may break off and embolize to the pulmonary circulation. Pulmonary emboli will develop in 50–60% of patients with proximal deep venous thrombosis (DVT); half of these embolic events will be asymptomatic. Approximately 50–70% of patients who have symptomatic pulmonary emboli will have lower extremity DVT when evaluated.

PE and DVT are two manifestations of the same disease. The risk factors for PE are the risk factors for thrombus formation within the venous circulation: venous stasis, injury to the vessel wall, and hypercoagulability (Virchow triad). Venous stasis increases with immobility (obesity, stroke, bed rest—especially postoperative), hyperviscosity (polycythemia), and increased central venous pressures (low cardiac output states, pregnancy). Vessels may be damaged by prior episodes of thrombosis, orthopedic surgery, or trauma. Hypercoagulability can be caused by medications (oral contraceptives, hormonal replacement therapy) or disease (malignancy, surgery) or may be the result of inherited gene defects. The most common inherited cause in white populations is resistance to activated protein C, also known as factor V Leiden. The trait is present in approximately 3% of healthy American men and in 20–40% of patients with idiopathic venous thrombosis. Other major risks for hypercoagulability include the following: deficiencies or dysfunction of protein C, protein S, and antithrombin; prothrombin gene mutation; hyperhomocysteinemia and the presence of antiphospholipid antibodies (lupus anticoagulant and anticardiolipin antibody).

PE has multiple physiologic effects. Physical obstruction of the vascular bed and vasoconstriction from neurohumoral reflexes both increase pulmonary vascular resistance. Massive thrombus may cause right ventricular failure. Vascular obstruction increases physiologic dead space (wasted ventilation) and leads to hypoxemia through right-to-left shunting, decreased cardiac output, and surfactant depletion causing atelectasis. Reflex bronchoconstriction promotes wheezing and increased work of breathing.

A. Symptoms and Signs

The clinical diagnosis of PE is notoriously difficult for two reasons. First, the clinical findings depend on both the size of the embolus and the patient’s preexisting cardiopulmonary status. Second, common symptoms and signs of pulmonary emboli are not specific to this disorder (Table 9–18).

Indeed, no single symptom or sign or combination of clinical findings is specific to PE. Some findings are fairly sensitive: dyspnea and pain on inspiration occur in 75–85% and 65–75% of patients, respectively. Tachypnea is the only sign reliably found in more than half of patients. A common clinical strategy is to use combinations of clinical findings to identify patients’ risk for PE. For example, 97% of patients in the original Prospective Investigation of Pulmonary Embolism Diagnosis (PIOPED I) study with angiographically proved pulmonary emboli had one or more of three findings: dyspnea, chest pain with breathing, or tachypnea. Wells and colleagues have published and validated a simple clinical decision rule that quantifies and dichotomizes this clinical risk assessment, allowing diversion of patients deemed unlikely to have PE to a simpler diagnostic algorithm (see Integrated Approach to Diagnosis of Pulmonary Embolism, below).

B. Laboratory Findings

The ECG is abnormal in 70% of patients with PE. However, the most common abnormalities are sinus tachycardia and nonspecific ST and T wave changes, each seen in approximately 40% of patients. Five percent or less of patients in the PIOPED I study had P pulmonale, right ventricular hypertrophy, right axis deviation, and right bundle branch block.

ABGs usually reveal acute respiratory alkalosis due to hyperventilation. The arterial PO₂ and the alveolar-arterial oxygen difference (A–a–DO₂) are usually abnormal in patients with PE compared with healthy, age-matched controls. However, ABGs are not diagnostic: among patients who were evaluated in the PIOPED I study, neither the PO₂ nor the A–a–DO₂ differentiated between those with and those without pulmonary emboli. Profound hypoxia with a normal chest radiograph in the absence of preexisting lung disease is highly suspicious for PE.

Plasma levels of D-dimer, a degradation product of cross-linked fibrin, are elevated in the presence of thrombus. Its value is as a diagnostic tool to exclude PE. Using a D-dimer threshold between 300 and 500 ng/mL (300 and 500 mcg/L), a rapid quantitative enzyme-linked immunosorbent assay (ELISA) has shown a sensitivity for venous thromboembolism of 95–97% and a specificity of 45%. Therefore, a D-dimer less than 500 ng/mL (less than 500 mcg/L) using a rapid quantitative ELISA provides strong evidence against venous thromboembolism, with a likelihood ratio of 0.11–0.13. Due to much higher false-positive rates, D-dimer is not useful in the hospital setting.

Serum troponin I, troponin T, and plasma B-type natriuretic peptide (BNP) levels are typically higher in patients with PE compared with those without embolism; the presence and magnitude of the elevation are not useful in diagnosis, but correlate with adverse outcomes, including mechanical ventilation, prolonged hospitalization, and death.

C. Imaging and Special Examinations

1. Chest radiography

The chest radiograph is necessary to exclude other common lung diseases and to permit interpretation of the ventilation-perfusion (V̇/Q̇) scan, but it does not establish the diagnosis by itself. The chest radiograph was normal in only 12% of patients with confirmed PE in the PIOPED I study. The most frequent findings were atelectasis, parenchymal infiltrates, and pleural effusions. However, the prevalence of these findings was the same in hospitalized patients without PE. A prominent central pulmonary artery with local oligemia (Westermark sign) or pleural-based areas of increased opacity that represent intraparenchymal hemorrhage (Hampton hump) are uncommon (eFigure 9–21). Paradoxically, the chest radiograph may be most suggestive of PE when normal in the setting of hypoxemia.

2. CT-pulmonary angiography (PA)

Helical CT-PA is used as the initial diagnostic study in North America for suspected PE (eFigure 9–22). CT-PA requires administration of intravenous radiocontrast dye but is otherwise noninvasive. A high-quality study is very sensitive for the detection of thrombus in the proximal pulmonary arteries. Comparing CT-PA to the V̇/Q̇ scan as the initial test for PE, detection of thrombi is roughly comparable, although more alternative pulmonary diagnoses are made with CT-PA scanning.

Test characteristics of CT-PA vary widely by study and facility. Factors influencing results include patient size and cooperation, the type and quality of the scanner, the imaging protocol, and the experience of the interpreting radiologist. The 2006 PIOPED II study, using multi-detector (four-row) helical CT and excluding the 6% of patients whose studies were “inconclusive,” reported sensitivity of 83% and specificity of 96%.

A 15–20% false-negative rate is high for a screening test and raises the practical question whether it is safe to withhold anticoagulation in patients with a negative CT-PA. Research data provide two complementary answers. The insight of PIOPED I, that the clinical assessment of pretest probability improves the performance of the V̇/Q̇ scan, was confirmed with CT-PA in PIOPED II, where positive and negative predictive values were highest in patients with concordant clinical assessments but poor with conflicting assessments. The negative predictive value of a normal CT-PA in patients with a high pretest probability was only 60%. Therefore, a normal CT-PA alone does not exclude PE in high-risk patients, and either empiric therapy or further testing is indicated.

A large, prospective trial, the Christopher Study, incorporated objective, validated pretest clinical assessment into diagnostic algorithms using D-dimer measurement. In this study, patients with a high pretest probability and a negative CT-PA who were not receiving anticoagulation had a low (less than 2%) 3-month incidence of subsequent PE. This low rate of complications supports the contention that many false-negative studies represent clinically insignificant, small distal thrombi and provides support for monitoring most patients with a high-quality negative CT-PA off therapy (see Integrated Approach to Diagnosis of Pulmonary Embolism below). The rate of false-positive CT-PA and overtreatment of PE has not been as well studied to date.

3. Ventilation-perfusion lung scanning

A perfusion scan is performed by injecting radiolabeled microaggregated albumin into the venous system, allowing the particles to embolize to the pulmonary capillary bed. To perform a ventilation scan, the patient breathes a radioactive gas or aerosol while the distribution of radioactivity in the lungs is recorded. A defect on perfusion scanning represents diminished blood flow to that region of the lung. This finding is not specific for PE. Defects in the perfusion scan are interpreted in conjunction with the ventilation scan to give a high, low, or intermediate (indeterminate) probability that PE is the cause of the abnormalities. Criteria for the combined interpretation of ventilation and perfusion scans (commonly referred to as a single test, the V̇/Q̇ scan) are complex, confusing, and not completely standardized. A normal perfusion scan excludes the diagnosis of clinically significant PE (negative predictive value of 91% in the PIOPED I study). A high-probability V̇/Q̇ scan is most often defined as having two or more segmental perfusion defects in the presence of normal ventilation and is sufficient to make the diagnosis of PE in most instances (positive predictive value of 88% among PIOPED I patients). V̇/Q̇ scans are most helpful when they are either normal (eFigure 9–23) (eFigure 9–24) or indicate a high probability of PE. Such readings are reliable—interobserver agreement is best for normal and high-probability scans—and they carry predictive power. The likelihood ratios associated with normal and high-probability scans are 0.10 and 18, respectively, indicating significant and frequently conclusive changes from pretest to posttest probability.

However, 75% of PIOPED I V̇/Q̇ scans were nondiagnostic, ie, of low or intermediate probability. At angiography, these patients had an overall incidence of PE of 14% and 30%, respectively.

One of the most important findings of PIOPED I was that the clinical assessment of pretest probability could be used to aid the interpretation of the V̇/Q̇ scan. For patients with low-probability V̇/Q̇ scans and a low (20% or less) clinical pretest probability of PE, the diagnosis was confirmed in only 4%. Such patients may reasonably be observed off therapy without angiography. All other patients with nondiagnostic V̇/Q̇ scans require further testing to determine the presence of venous thromboembolism.

4. Venous thrombosis studies

Seventy percent of patients with PE will have DVT on evaluation, and approximately half of patients with DVT will have PE on angiography. Since the history and physical examination are neither sensitive nor specific for PE and since the results of V̇/Q̇ scanning are frequently equivocal, documentation of DVT in a patient with suspected PE establishes the need for treatment and may preclude further testing.

Commonly available diagnostic techniques include venous ultrasonography, impedance plethysmography, and contrast venography. In most centers, venous ultrasonography is the test of choice to detect proximal DVT. Inability to compress the common femoral or popliteal veins in symptomatic patients is diagnostic of first-episode DVT (positive predictive value of 97%); full compressibility of both sites excludes proximal DVT (negative predictive value of 98%). The test is less accurate in distal thrombi, recurrent thrombi, or in asymptomatic patients. Impedance plethysmography relies on changes in electrical impedance between patent and obstructed veins to determine the presence of thrombus. Accuracy is comparable though not quite as high as ultrasonography. Both ultrasonography and impedance plethysmography are useful in the serial examination of patients with high clinical suspicion of venous thromboembolism but negative leg studies. In patients with suspected first-episode DVT and a negative ultrasound or impedance plethysmography examination, multiple studies have confirmed the safety of withholding anticoagulation while conducting two sequential studies on days 1–3 and 7–10. Similarly, patients with nondiagnostic V̇/Q̇ scans and an initial negative venous ultrasound or impedance plethysmography examination may be monitored off therapy with serial leg studies over 2 weeks. When serial examinations are negative for proximal DVT, the risk of subsequent venous thromboembolism over the following 6 months is less than 2%.

Contrast venography remains the reference standard for the diagnosis of DVT (eFigure 9–25). An intraluminal filling defect is diagnostic of venous thrombosis. However, venography has significant shortcomings and has been replaced by venous ultrasound as the diagnostic procedure of choice. Venography may be useful in complex situations where there is discrepancy between clinical suspicion and noninvasive testing.

5. Pulmonary angiography

Pulmonary angiography is the historical reference standard for the diagnosis of PE. An intraluminal filling defect in more than one projection establishes a definitive diagnosis. Secondary findings highly suggestive of PE include abrupt arterial cutoff, asymmetry of blood flow—especially segmental oligemia—or a prolonged arterial phase with slow filling. Pulmonary angiography was performed in 755 patients in the PIOPED I study. A definitive diagnosis was established in 97%; in 3% the studies were nondiagnostic. Four patients (0.8%) with negative angiograms subsequently had pulmonary thromboemboli at autopsy. Serial angiography has demonstrated minimal resolution of thrombus prior to day 7 following presentation. Thus, negative angiography within 7 days of presentation excludes the diagnosis.

Pulmonary angiography it is a safe but invasive procedure with well-defined morbidity and mortality data. Minor complications occur in approximately 5% of patients. Most are allergic contrast reactions, transient kidney injury, or percutaneous catheter–related injuries; cardiac perforation and arrhythmias are reported but rare. Among the PIOPED I patients who underwent angiography, there were five deaths (0.7%) directly related to the procedure.

Pulmonary angiography is indicated in any patient in whom the diagnosis is in doubt when there is a high clinical pretest probability of PE or when the diagnosis of PE must be established with certainty, as when anticoagulation is contraindicated or placement of an inferior vena cava filter is contemplated.

6. MRI

MRI has sensitivity and specificity equivalent to contrast venography in the diagnosis of DVT. It has improved sensitivity when compared with venous ultrasound in the diagnosis of DVT, without loss of specificity. The test is noninvasive and avoids the use of potentially nephrotoxic radiocontrast dye. However, artifacts introduced by respiratory and cardiac motion have limited the use of MRI in the diagnosis of PE. New techniques have improved sensitivity and specificity to levels comparable with helical CT, but MRI remains primarily a research tool for PE.

An integrated approach to the diagnosis of PE uses the clinical likelihood of venous thromboembolism derived from a clinical prediction rule (Table 9–19) along with the results of diagnostic tests to come to one of three decision points: to establish venous thromboembolism (PE or DVT) as the diagnosis, to exclude venous thromboembolism with sufficient confidence to discontinue anticoagulation therapy and monitor the patient, or to refer the patient for additional testing. An ideal diagnostic algorithm would proceed in a cost-effective, stepwise fashion to come to these decision points at minimal risk to the patient. Most North American centers use Wells’ clinical prediction rule (Table 9–19) or something similar to guide a rapid D-dimer and CT-PA–based diagnostic algorithm (Figure 9–7). This approach is highly effective. The incidence of PE following a negative integrated clinical, D-dimer, and CT-PA evaluation that excludes PE is comparable to that seen following the traditional gold standard of a negative pulmonary angiography. The algorithm used in a recent, high-quality study by the PEGeD investigators (Figure 9–7) was associated with only one VTE event in 3 months of follow-up of 1863 patients who did not receive an initial diagnosis of pulmonary embolism.

Concern over false-positive D-dimer measurements in low-risk patients that prompt unnecessary CT-PA has led researchers to develop the Pulmonary Embolism Rule-Out Criteria (PERC, Table 9–20). In a prospective analysis of 8138 emergency department patients at risk for PE, researchers identified 20% (1666) who had both a low (less than 20%) pretest probability of PE and met all eight PERC criteria. This group had a 1.0% incidence of PE, with one fatality. Patients with a Modified Wells Score of 4 or less who meet all eight PERC criteria do not need to undergo D-dimer testing and may be monitored clinically.

The standard V̇/Q̇ scan-based algorithm (Table 9–21) remains useful in many patients, especially those who are not able to undergo CT-PA (eg, those with advanced chronic kidney disease).

Clinical prediction scores, D-dimers, and other diagnostic tests have not been well validated during pregnancy. An 8-year European study of 395 pregnant women using a diagnostic algorithm, a sensitive D-dimer, venous ultrasonography, and CT-PA excluded PE if the diagnostic workup was negative; follow-up at 3 months resulted in a 0.0% rate of venous thromboembolism events among untreated participants. The overall incidence of PE in this population was 7.1%.

Venous thromboembolism is often clinically silent until it presents with significant morbidity or mortality. It is a prevalent disease, clearly associated with identifiable risk factors. For example, the incidence of proximal DVT, PE, and fatal PE in untreated patients undergoing hip fracture surgery is reported to be 10–20%, 4–10%, and 0.2–5%, respectively. There is unambiguous evidence of the efficacy of prophylactic therapy in this and other clinical situations, yet it remains underused. Only about 50% of surgical deaths from PE had received any form of preventive therapy. Discussion of strategies for the prevention of venous thromboembolism can be found in Chapter 14.

A. Anticoagulation

Anticoagulation is not definitive therapy but a form of secondary prevention. Heparin binds to and accelerates the ability of antithrombin to inactivate thrombin, factor Xa, and factor IXa. It thus retards additional thrombus formation, allowing endogenous fibrinolytic mechanisms to lyse existing clot. The standard regimen of heparin followed by 6 months of oral warfarin results in an 80–90% reduction in the risk of both recurrent venous thrombosis and death from PE (see Tables 14–16, 14–19, 14–20). Low-molecular-weight heparins (LMWHs) are as effective as unfractionated heparin in the treatment of venous thromboembolism. The 2016 CHEST Guideline and Expert Panel Report recommends direct oral anticoagulants over vitamin K antagonist (warfarin) and LMWH in all patients with venous thromboembolism without a cancer diagnosis and recommends LMWH for patients with cancer.

The optimal duration of anticoagulation therapy for venous thromboembolism is unknown. There appears to be a protective benefit to continued anticoagulation in first-episode venous thromboembolism (twice the rate of recurrence in 6 weeks compared with 6 months of therapy) and recurrent disease (eightfold risk of recurrence in 6 months compared with 4 years of therapy). These studies did not distinguish patients with reversible risk factors, such as surgery or transient immobility, from patients who have a nonreversible hypercoagulable state, such as factor V Leiden, antithrombin deficiency, antiphospholipid syndrome, or malignancy. An RCT of low-dose warfarin (INR 1.5–2.0) versus no therapy following 6 months of standard therapy in patients with idiopathic DVT was stopped early. The protective benefits of continued anticoagulation include fewer DVTs in addition to a trend toward lower mortality despite more hemorrhage in the warfarin group. Risk reductions were consistent across groups with and without inherited thrombophilia.

For many patients, venous thrombosis is a recurrent disease, and continued therapy results in a lower rate of recurrence at the cost of an increased risk of hemorrhage. Therefore, the appropriate duration of therapy needs to take into consideration the patient’s age, potentially reversible risk factors, likelihood and potential consequences of hemorrhage, and preferences for continued therapy. The 2016 CHEST Guideline and Expert Panel Report recommends 3 months of anticoagulation after a first episode provoked by a surgery or a transient nonsurgical risk factor. Extended therapy (no scheduled stop date) is recommended for an unprovoked episode with a low to moderate risk of bleeding. Patient sex and D-dimer level measured a month after stopping anticoagulant therapy may influence this treatment decision. Patients who continue to receive anticoagulation long term should be reassessed for venous thrombosis periodically (at least annually). For patients with cancer, extended therapy with LMWH is recommended regardless of bleeding risk. D-dimer testing has been suggested to identify those who may benefit from continued anticoagulation after 3 months of therapy, but clinical data have not supported its utility in this regard.

The major complication of anticoagulation is hemorrhage. Risk factors for hemorrhage include the intensity of the anticoagulation; duration of therapy; concomitant administration of medications, such as aspirin, that interfere with platelet function; and patient characteristics, particularly increased age, previous gastrointestinal hemorrhage, and coexistent chronic kidney disease.

The reported incidence of major hemorrhage following intravenous administration of unfractionated heparin is nil to 7%; that of fatal hemorrhage is nil to 2%. The incidence with LMWHs is not statistically different. There is no information comparing hemorrhage rates at different doses of heparin. The risk of death from another pulmonary embolism during subtherapeutic heparin administration in the first 24–48 hours after diagnosis is significant; it appears to outweigh the risk of short-term supratherapeutic heparin levels. The incidence of hemorrhage during therapy with warfarin is reported to be between 3% and 4% per patient year. The frequency varies with the target INR and is consistently higher when the INR exceeds 4.0. There is no apparent additional antithrombotic benefit in venous thromboembolism with a target INR above 2.0–3.0 (see Chapter 14).

B. Thrombolytic Therapy

Streptokinase, urokinase, and recombinant tissue plasminogen activator (rt-PA; alteplase) increase plasmin levels and thereby directly lyse intravascular thrombi. In patients with established PE, thrombolytic therapy accelerates resolution of emboli within the first 24 hours compared with standard heparin therapy. This is a consistent finding using angiography, V̇/Q̇ scanning, echocardiography, and direct measurement of pulmonary artery pressures. However, at 1 week and 1 month after diagnosis, these agents show no difference in outcome compared with heparin and warfarin. Hemodynamically stable patients with echocardiographic evidence of right heart strain from acute PE show no long-term improvement in function or mortality when given thrombolytic therapy. Indeed, clinical research shows no improvements in mortality in any patient population. Subtle improvements in pulmonary function, including improved single-breath diffusing capacity and a lower incidence of exercise-induced pulmonary hypertension, have been observed. The reliability and clinical importance of these findings is unclear. The major disadvantages of thrombolytic therapy compared with heparin are its greater cost and significant increase in major hemorrhagic complications. The incidence of intracranial hemorrhage in patients with PE treated with alteplase is 2.1% compared with 0.2% in patients treated with heparin. Current practice supports thrombolytic therapy for PE in patients at high risk for death from refractory hypotension or hypoxemia despite heparin therapy, patients in whom the more rapid resolution of thrombus may be lifesaving.

Absolute contraindications to thrombolytic therapy include active internal bleeding and stroke within the past 2 months. Major contraindications include uncontrolled hypertension and surgery or trauma within the past 6 weeks.

C. Additional Measures

Interruption of the inferior vena cava may be indicated in patients with a major contraindication to anticoagulation who have or are at high risk for development of proximal DVT or PE. Placement of an inferior vena cava filter is also recommended for recurrent thromboembolism despite adequate anticoagulation, for chronic recurrent embolism with a compromised pulmonary vascular bed (eg, in pulmonary hypertension), and with the concurrent performance of surgical pulmonary embolectomy or pulmonary thromboendarterectomy. Percutaneous transjugular placement of a mechanical filter is the preferred mode of inferior vena cava interruption. These devices reduce the short-term incidence of PE in patients presenting with proximal lower extremity DVT. However, they are associated with a twofold increased risk of recurrent DVT in the first 2 years following placement so plans must be usually made for their subsequent removal.

In rare critically ill patients for whom thrombolytic therapy is contraindicated or unsuccessful, mechanical or surgical extraction of thrombus may be indicated. Pulmonary embolectomy is an emergency procedure of last resort with a very high mortality rate. It is performed only in a few specialized centers. Several catheter devices to fragment and extract thrombus through a transvenous approach, or to allow for the directed administration of rt-PA, have been reported in small numbers of patients. Comparative outcomes with surgery, thrombolytic therapy, or heparin have not been studied.

PE is estimated to cause more than 50,000 deaths annually. In the majority of deaths, PE is not recognized antemortem or death occurs before specific treatment can be initiated. These statistics highlight the importance of preventive therapy in high-risk patients (Chapter 14). The outlook for patients with diagnosed and appropriately treated PE is generally good. Overall prognosis depends on the underlying disease rather than the PE itself. Death from recurrent thromboemboli is uncommon, occurring in less than 3% of cases. Perfusion defects resolve in most survivors. Chronic thromboembolic pulmonary hypertension develops in approximately 1% of patients. Selected patients may benefit from pulmonary endarterectomy.

Most patients with acute PE require hospitalization. The decision to admit patients with acute PE requires assessment of factors placing them at high risk, including their severity of illness (eg, severe hypoxemia), comorbidities (eg, DVT, cardiac dysfunction), educational needs (eg, lack of knowledge about PE and its management), and/or problematic social situations (eg, prior noncompliance with follow-up care). Some institutions and healthcare systems have demonstrated that carefully selected low-risk patients with acute PE can be safely and effectively managed as outpatients with the aid of integrated clinical decision support systems.