Focal peripheral lucency beyond an occluded vessel, often accompanied by mild dilation of the central pulmonary vessels, known as Westermark’s sign, (Fig. 27-1), is a very nonspecific finding seen in only 7% to 14% of cases of documented PE in the PIOPED I study. Westermark’s sign is thought to be caused by embolic obstruction of the pulmonary artery or hypoxic vasoconstriction secondary to ventilation of poorly perfused lung. This sign often is a subtle finding, in many cases not recognized prospectively, and can be mimicked by other common lung diseases, such as emphysema.

Enlargement of the central pulmonary vasculature also may occur with PE and also is frequently subtle and easily overlooked. This finding may be the result of distention of the vessel by thrombus or by acute rise in pulmonary arterial pressure secondary to the presence of distal emboli. Enlargement of the right descending pulmonary artery, occasionally with a “sausage-like” configuration, may be seen in a number of patients with acute PE, and the size and shape of the artery may normalize following resolution of the embolic event. Enlargement of the right descending pulmonary artery is not specific to PE and may result from pulmonary hypertension of any cause.

Pulmonary edema rarely may occur in association with PE. This finding most often is seen in patients with underlying cardiopulmonary disease and may be caused by left ventricular failure precipitated by PE.

Focal parenchymal abnormalities, particularly atelectasis, were the most common chest radiographic abnormalities in patients with PE in the PIOPED I series, occurring in just over two thirds
of the patients. Linear opacities often occur near the lung bases and are thought to represent areas of subsegmental atelectasis related to mucous plugging, hypoventilation, or, perhaps, to distal airway closure or focal depletion of surfactant. Such opacities commonly are transient; if they persist, they may represent areas of scarring secondary to prior infarction.

Focal air-space consolidation may occur in patients with PE and may represent pulmonary hemorrhage without infarction or true pulmonary infarction with ischemic necrosis of lung tissue. Estimates of the frequency of pulmonary infarction in patients with PE vary from 10% to 60%. Infarction is most likely to occur when diminished cardiopulmonary reserve is present because both the pulmonary and bronchial arterial systems are impaired. Infarcts often are multiple and occur most frequently in the subpleural regions of the lower lobes, usually within 12 to 24 hours of the onset of symptoms. Infarcts are variable in size and often do not show an air bronchogram, a finding that may favor the diagnosis of infarction over pneumonia. Infarcts typically are ill defined but may progress over several days to a discrete focal opacity. The classic description of a pulmonary infarct, the “Hampton hump,” is a circumscribed, subpleural opacity with a rounded or truncated convex medial border facing toward the pulmonary hilum (Fig. 27-2). This finding is neither common nor specific, however.

Cavitation within a bland (uninfected) infarct is uncommon. Bland infarct cavitation is more likely when the infarct is larger than 4 cm in diameter. When cavitation occurs, it usually is apparent within 2 weeks of the appearance of the air-space opacity.

When air-space opacities are secondary to pulmonary hemorrhage without infarction, resolution of the abnormality usually is rapid; however, true infarction with ischemic tissue necrosis usually takes weeks or months to resolve, leaving linear scars or occasionally associated with pleural thickening. Resolution of pulmonary infarction has been likened to the “melting of an ice cube,” implying that an infarct clears by peripheral dissolution, whereas pneumonia will gradually clear in an irregular, patchy fashion.

Pleural effusion is detected on chest radiography in about half of patients with PE and usually is unilateral and small. When pulmonary infarction occurs, pleural effusions may be larger, hemorrhagic, and may take longer to resolve. Diaphragmatic elevation is common in patients with PE, but this finding is nonspecific.

Chest radiographic abnormalities in patients with PE usually are nonspecific and neither establish nor exclude the diagnosis of PE. The sensitivity and specificity of chest radiography for the diagnosis of PE are only 33% and 59%, respectively. The main value of chest radiographs is for the detection of diagnoses that may clinically simulate PE, such as pneumothorax, pulmonary edema, pneumonia, or rib fractures. In addition, a recent chest radiograph is required for the interpretation of ventilation/perfusion () scintigraphy.

Ultrasound techniques used in the evaluation of DVT variously include real-time gray scale imaging (with and without compression), continuous wave and pulsed Doppler, color Doppler imaging, and ancillary techniques, such as the Valsalva maneuver and manual blood flow augmentation. A high-frequency linear array transducer is preferred to provide optimal spatial resolution. For larger patients, a lower-frequency transducer may be required to provide adequate tissue penetration to visualize the deep venous system of the lower extremity successfully. The lower extremity veins are imaged in both longitudinal and transverse planes from the level of the inguinal ligament to the popliteal trifurcation, including the common femoral vein, the superficial femoral vein, the popliteal vein, and the saphenous vein at its junction with the common femoral vein. Normal veins appear as tubular, anechoic structures. Although thrombosis occasionally may be seen with gray-scale sonography (Fig. 27-3), real-time imaging alone is not sufficient to exclude DVT, because thrombus may have variable echogenicity and often is anechoic, especially when acute; therefore, compression ultrasonography has become the most reliable maneuver to assess for DVT. With compression ultrasonography, the venous system is visualized in the transverse plane and serially compressed from the inguinal ligament to the popliteal fossa in 1- to 2-cm intervals by exerting gentle pressure with the transducer. The diagnosis of DVT is established by demonstrating lack of venous compression due to intraluminal thrombus (Fig. 27-4).

Additional methods employed during lower extremity venous ultrasound examinations for suspected DVT include the Valsalva maneuver, spectral Doppler analysis, and color Doppler analysis. In response to a Valsalva maneuver, a normal vein dilates to more than 50% of its original diameter as a result of impaired venous drainage upstream from the area sampled, whereas veins with acute thrombus have pathologic changes in their walls that prevent such dilation. Lack of appropriate response to a Valsalva maneuver also may indicate thrombosis or obstruction of more central veins outside the field of view, such as the inferior vena cava (IVC). The Valsalva maneuver requires adequate patient cooperation and generally is limited to assessment of the common femoral vein, which is sufficiently large to demonstrate the caliber changes induced by the altered blood volume caused by the maneuver. Although an abnormal venous response to the Valsalva maneuver supports the diagnosis of DVT when there is a lack of venous compressibility, and a normal response may corroborate the findings of a normal compression ultrasound examination, normal or abnormal Valsalva maneuvers alone are not sufficiently sensitive or specific to establish or exclude the diagnosis of DVT.

Spectral Doppler analysis is particularly useful for vessels that cannot be visualized directly (such as the medial portions of the subclavian veins, the central IVC, the superior vena cava [SVC], and brachiocephalic veins). The spectral Doppler waveform of patent central vessels normally shows respiratory phasicity. A monophasic waveform suggests venous obstruction remote from the point of venous interrogation. This abnormal waveform can indicate central DVT, although stenosis or extrinsic compression of the central veins may result in a similar waveform.

Color Doppler imaging is a useful addition to lower extremity compression ultrasonography. Color Doppler is valuable for identifying deep venous structures and interrogating deep vessels where the application of direct venous compression is difficult, such as the superficial femoral vein in the adductor hiatus and the iliac veins (Fig. 27-5). In patients who may be difficult to image, such as obese or postoperative patients, or those with swollen extremities, color Doppler imaging is a useful tool for identifying and interrogating venous anatomy. Venous thrombosis is shown on color Doppler imaging as absence of color flow within the vessel lumen (Fig. 27-6) at baseline and with augmentation. In patients with clinically suspected DVT and a technically adequate examination, color Doppler imaging demonstrates high sensitivity (95%) and specificity (98%) for the diagnosis of femoropopliteal DVT. However, for asymptomatic, high-risk patients (e.g., patients who have just undergone orthopedic surgery or trauma patients), the sensitivity of color Doppler imaging for the detection of femoropopliteal DVT is much lower, perhaps due to the presence of short-segment or nonocclusive thrombus, a relatively high proportion of thrombi limited to the calf veins, or an overall decrease in prevalence of lower extremity DVT in settings where routine DVT prophylaxis is used.

Flow augmentation using spectral Doppler is performed by placing the Doppler gate on the examined vein in the longitudinal plane and manually compressing the calf. A normal response is a rapid rise and fall in blood flow velocity in the interrogated vessel. Such a response implies that the venous system is patent between the point of interrogation and the area of manual compression. Lack of a normal response to flow augmentation may indicate nonocclusive thrombus, although this observation is not specific for this diagnosis.

The compression ultrasound examination has been proved to be accurate in multiple studies that compared compression ultrasonography with CV and impedance plethysmography in symptomatic patients. When results of various series are combined, compression ultrasonography demonstrates a sensitivity of at least 93% and a specificity of 98% for femoropopliteal DVT, and these numbers would probably be enhanced some with the routine addition of spectral and color Doppler techniques. The overall diagnostic accuracy of compression ultrasonography in the symptomatic patient is lower, however, when the analysis includes potential calf vein thrombosis.

The reliability of a negative compression ultrasound examination is high. Outcome studies have shown no deleterious effects resulting from withholding anticoagulation therapy in patients with clinically suspected DVT who have negative compression ultrasonography examinations at presentation and at follow-up testing over the next week.

Data on the accuracy of compression ultrasonography in asymptomatic, high-risk patient populations (e.g., following orthopedic surgery) are conflicting, with some series showing poor sensitivity for lower extremity ultrasound (including spectral and color Doppler techniques), but other studies showing better results. Conflicting data may be the result of differences in study design and study populations.

The importance of diagnosing isolated calf vein thrombosis is controversial. Although many investigators postulate that clinically significant PE does not originate from calf veins, not all sources agree that calf vein thrombosis is self-limited. Venographic studies have shown that 40% of untreated calf vein thrombi will remain below the knee, 40% will lyse, and 20% may extend into the femoropopliteal system. Once calf vein thrombi extend above the knee, 50% may be associated with abnormal scintigraphy, indicating possible PE. Many physicians do not treat calf vein thrombosis aggressively, but anticoagulation certainly is warranted for patients with proximal thrombi, so it may be clinically useful to detect calf vein thrombi before proximal migration has occurred. Calf vein thrombosis can be demonstrated with serial imaging with compression ultrasonography or by direct examination of the veins of the calf. Outcome studies employing compression ultrasonography in symptomatic patients have shown that it is safe to withhold anticoagulation in patients with a negative initial compression ultrasound examination, provided that symptoms prompting the initial compression ultrasound examination do not persist. If lower extremity symptoms persist following the initial examination, a repeat sonogram should be obtained; a small number of patients with an initially negative examination but persistent symptoms will be found to have femoropopliteal DVT on the subsequent compression ultrasound exam. This observation presumably reflects proximal migration of previously undetected calf vein thrombosis.

Another method of addressing suspected calf vein thrombosis is direct examination of the calf veins with ultrasound. The technique of calf vein sonography is not uniformly standardized, and rates of technically adequate examinations vary widely because calf vein sonography is difficult to perform. Several studies have demonstrated encouraging results for calf vein thrombosis detection when examinations are technically satisfactory, though results are variable. Again, color Doppler imaging is a useful adjunct to routine compression ultrasound in the evaluation of symptomatic patients with potential calf vein thrombosis. However, compression ultrasonography, with or without color Doppler, is an insensitive test for calf vein thrombosis for asymptomatic high-risk patients. As with compression ultrasonography in the calf, the utility of color Doppler imaging may be limited by the difficulty in obtaining a technically adequate study.

Ultrasound is an excellent screening tool for the assessment of potential upper extremity thrombosis. Routine compression ultrasonography can be utilized in the jugular and brachial venous systems, but the medial aspects of the subclavian veins, the brachiocephalic veins, and the SVC are inaccessible to direct compression. Therefore, interrogation of these vessels requires spectral and color Doppler techniques. Thrombosis may identified by the absence of color Doppler flow within the vessel lumen, occasionally accompanied by echogenic material completely or partially filling the vessel lumen. Spectral Doppler analysis is particularly valuable in the evaluation of potential upper extremity thrombosis. The patency of central venous structures that cannot be visualized directly is inferred by the presence of normal respiratory phasicity. Monophasic flow in the jugular or subclavian veins suggests a central venous abnormality such as DVT, stenosis, or extrinsic compression. Familiarity with these methods is important because the incidence of PE in the setting of upper extremity DVT may be as high as 12%. Furthermore, the incidence of upper extremity DVT has risen in recent years, largely due to the increased use of long-term indwelling catheters for total parenteral nutrition, chemotherapy, and other indications.

A basic understanding of normal pulmonary physiology is required to appreciate the alterations in lung function that occur in patients with PE. Reflex pneumoconstriction may occur in alveoli that are ventilated but not perfused (i.e., abnormally “high” ); abnormally low carbon dioxide tension elicits this response. However, this reflex is not commonly observed and generally is transient, because patients inhale carbon dioxide from tracheal dead space into the poorly perfused alveoli, and bronchial circulation continues to deliver some carbon dioxide to ischemic alveoli, thus moderating this reflex bronchoconstriction. Abnormalities of ventilation may produce regional alveolar hypoxia, which, in turn, induces reflex pulmonary vasoconstriction. Thus, alveolar hypoxia (i.e., areas of abnormally low ) causes redistribution of pulmonary blood flow away from hypoventilated alveoli. These pulmonary responses to alterations in regional ventilation and perfusion provide the basis for scintigraphy.

A normal ventilation scan shows relatively homogeneous pulmonary tracer activity on the single-breath and equilibrium images. During the washout phase, tracer activity slowly clears, with the bases clearing slightly more slowly than the remainder of the lungs. Clearing usually is complete in 2 or 3 minutes. ^99mTc-DTPA aerosols also demonstrate homogeneous tracer activity from lung apex to base but, unlike xenon scans, the trachea and bronchi normally may be seen. Occasionally, swallowed activity may be detected in the esophagus and stomach.

Normal perfusion scans reveal homogeneous pulmonary tracer activity with predictable defects in the expected locations of the heart, pulmonary hila, and aortic arch, depending on the projection obtained.

Various schemes have been devised to interpret scintigraphy. These schemes rely on the principle that PE causes decreased or absent pulmonary blood flow in a portion of lung, producing a perfusion defect. Because the alveoli serving these occluded vessels remain ventilated, a “mismatch” is created. Unfortunately, many causes of perfusion defects unrelated to PE exist, and PE itself does not always result in a mismatch. Hence, PE is neither diagnosed nor excluded by lung scanning; rather, the probability of PE in any given patient is generated by interpretation of the lung scan. These probabilities are based on criteria that evaluate the shape, number, location, and size of perfusion defects on the perfusion scan in combination with the findings on the ventilation lung scan and chest radiograph. Perfusion
defects are classified as lobar, segmental, or subsegmental; defects that do not conform to segmental lung anatomy are considered nonsegmental defects. The shape of perfusion defects also is important. Perfusion defects resulting from PE usually are wedge-shaped and contact the pleural surface (Fig. 27-7). Defects that do not extend to the pleural surface may show a rim of activity immediately beneath the pleural surface but peripheral to the perfusion defect—a finding known as the “stripe” sign. Such perfusion abnormalities usually are not the result of PE. The number of perfusion defects also may be of some value in determining the likelihood of PE. Solitary perfusion defects usually are not related to PE, whereas multiple subsegmental perfusion defects are associated with PE in up to 50% of cases. The size of a perfusion defect also is significant. Sizes are graded as follows: a small perfusion defect occupies less than 25% of an anatomic lung segment; a moderate perfusion defect represents 25% to 75% of a lung segment; and a large perfusion defect constitutes greater than 75% of an anatomic lung segment. The perfusion scan is interpreted with reference to the chest radiograph, and the size of a perfusion defect is compared to any corresponding chest radiographic abnormality. The chief interpretive schemes (McNeil, Biello, PIOPED, and revised PIOPED classification schemes) have been divided into four diagnostic categories based on the probability of PE at pulmonary angiography: high-probability, intermediate/indeterminate-probability, low-probability, and normal perfusion scintigraphy. In addition to these categories, the original PIOPED criteria included a very low probability group in which the prevalence of angiographically documented PE was 9%.

The clinical diagnosis of PE is even more difficult in patients with obstructive lung disease because the clinical manifestations of PE may easily be misinterpreted as a COPD exacerbation, and laboratory values are not sufficiently sensitive or specific to distinguish acute PE from a COPD exacerbation. Unfortunately, scintigraphy often is abnormal in patients with moderate or severe COPD. Nevertheless, it has been shown that the diagnostic utility of scintigraphy for the evaluation of PE is not diminished in the setting of pre-existing cardiopulmonary disease, although intermediate-probability scan interpretations do occur more often in such patients. Furthermore, the diagnostic accuracy of lung scanning in patients with COPD is improved when integrated with the clinical assessment of the likelihood of PE.