Introduction	Choose Top of pageABSTRACTIntroduction <<Subjects and MethodsResultsDiscussionReferencesCITING ARTICLES

Venous thromboembolism is a common and important cause of morbidity and mortality in the critical care setting [1]. Among patients who died while in the ICU, pulmonary embolus has been reported in 7–27% (mean, 13%) of postmortem examinations, and pulmonary embolus was thought to have contributed to death in 0–12% (mean, 3%) [2–6]. Deep venous thrombosis (DVT) is considered to be the most important causative factor of pulmonary embolism [7]. The overall incidence of DVT in the ICU population ranges from 0% to 33%, with variability based on DVT prophylaxis [2, 8, 9]. The critically ill patient often has at least one risk factor predisposing to thromboembolism, such as recent surgery, trauma, burn, sepsis, prolonged bed rest, malignancy, cardiac failure, or acute myocardial infarction [2]. Because of the critical care patient's condition, body wall edema, or surgical dressings, the signs and symptoms of DVT are often masked, making a clinical diagnosis problematic at best. Thus, the role of imaging becomes more critical, even when clinical suspicion is low.

Because venous thromboembolism represents a spectrum of disease from DVT to pulmonary embolism, a combined approach to the diagnosis is desirable. That is, even in the absence of DVT, pulmonary embolism may still be present and vice versa, despite the relationship. Furthermore, DVT may not be suspected in an ICU patient until the patient manifests signs or symptoms suggestive of pulmonary embolism. CT pulmonary angiography has gained acceptance as a first-line imaging study in suspected acute pulmonary embolism, replacing traditional ventilation–perfusion scintigraphy at many institutions [10]. Although the negative predictive value of CT pulmonary angiography approaches 96–98%, many still recommend evaluating the deep venous system, because the legs may harbor an unsuspected clot that will ultimately result in pulmonary embolism [11]. Venous sonography is the most widely used technique for the diagnosis of DVT [11].

Relatively recently, CT venography (CTV) has been documented as offering a rapid and available alternative to venous sonography of the lower extremities in the evaluation for DVT, with sensitivity and specificity reported to be in the range of 89–100% and 94–100%, respectively [12–15]. Indirect CTV coupled with CT pulmonary angiography provides a single examination capable of evaluating both the pulmonary arterial system and the pelvic and lower extremity venous system.

To our knowledge, the literature reviewed comparing the accuracy of indirect CTV with that of lower extremity duplex sonography in DVT evaluation has not specifically addressed the ICU population. The primary aim of this study was to determine if CTV offers an accurate alternative to venous sonography in the first-line evaluation for DVT in the high-risk critical care population being evaluated for pulmonary embolism.

Subjects and Methods	Choose Top of pageABSTRACTIntroductionSubjects and Methods <<ResultsDiscussionReferencesCITING ARTICLES

A waiver of consent was granted to this study by our institutional review board because the study was considered to be within the standard of care for the diagnosis of thromboembolic disease. Between June 2002 and April 2003, patients referred clinically for the exclusion of pulmonary embolus and who were currently admitted to either a surgical or a medical ICU were considered for study. Any patient who could not undergo sonography within 24 hr was excluded. Sixty-one patients who were 20–85 years old formed the study group.

CT was performed on MDCT scanners (MX8000D or MX8000, Philips Medical Systems). CT pulmonary angiography was performed using our standard pulmonary embolism protocol, with 3.2-mm collimation (2.5-mm nominal slice thickness) from the lung bases to the lung apices in a single breath-hold. Images were reconstructed at 1.6-mm intervals (1.25-mm nominal slice width). Approximately 140 mL of nonionic contrast material was injected by power injection at a rate of 3–4 mL/sec, depending on the IV catheter, with a standard prescanning delay of 20 sec. A bolus chaser was not used.

View larger version (163K)

Fig. 1A —66-year-old woman with deep venous thrombosis on CT and sonography. Axial CT image reveals filling defect in left superficial femoral vein (arrow).

View larger version (82K)

Fig. 1B —66-year-old woman with deep venous thrombosis on CT and sonography. Color Doppler sonogram reveals absence of flow. LSFV = left superficial femoral vein.

Indirect CTV was begun 180 sec after the initiation of the contrast bolus, and 10-mm axial images were acquired at 20-mm intervals from the renal veins through the popliteal fossa as previously described by Garg et al. [16].

CT pulmonary angiography with indirect CTV studies were prospectively reviewed on soft copy by a fellowship-trained thoracic radiologist who was unaware of the sonography results. For interpretation, multiplanar reformatted images of the pulmonary circulation were used as necessary. The observer completed a standardized grading sheet that assessed CT pulmonary angiography quality (Table 1), the presence (central, segmental, or subsegmental) or absence of pulmonary embolus, the Hounsfield unit measurements at each segment of the venous system from the inferior vena cava through the popliteal veins, and the presence or absence of DVT in each segment.

View Larger Version

TABLE 1: Image Quality Grades of CT Pulmonary Angiography

The criterion used to diagnose DVT was the presence of a definite intraluminal filling defect. Opacification of the veins was considered poor (< 60 H) or adequate (≥ 60 H) on the basis of results previously reported by Baldt et al. [17].

Bilateral extremity venous sonography was performed within 24 hr of the CT examination by certified sonography technologists. Bilateral lower extremity duplex Doppler venous sonography was performed on several sonographic machines (models 700 and 900 [GE Healthcare] or Sequoia 512 [Acuson]) based on availability and technologist preference. The technologists were free to use various transducers at their discretion as they would in normal clinical practice. The examination consisted of images obtained with compression, augmentation followed by color flow, and Doppler waveforms from the common femoral vein to the popliteal trifurcation, which included gray-scale and color Doppler waveform evaluations. The study was evaluated by an experienced sonographer who was blinded to the indirect CTV results. The criterion used to diagnose DVT included a combination of lack of vessel compression, absent flow, and lack of augmentation.

The results of both indirect CTV and bilateral lower extremity Doppler venous sonography were compared with a clinical standard. This standard was chosen to reflect the reality of clinical practice and to take into account the limited sensitivity of venous sonography in the patient population. The clinical standard defined the presence of DVT as diagnosis of DVT either on venous sonography or on indirect CTV in the presence of definitive pulmonary embolus on CT pulmonary angiography. Both indirect CTV and venous sonography were compared with this standard.

View larger version (140K)

Fig. 2A —49-year-old woman with pulmonary embolus and deep venous thrombosis. Axial CT image through pulmonary arteries reveals filling defect in right lower lobe artery (arrow). Note also peripheral pulmonary infarct (I).

View larger version (170K)

Fig. 2B —49-year-old woman with pulmonary embolus and deep venous thrombosis. Axial CT image through right lower extremity reveals thrombus in right superficial femoral vein (arrow).

View larger version (88K)

Fig. 2C —49-year-old woman with pulmonary embolus and deep venous thrombosis. Color Doppler sonogram, considered false-negative, reveals thrombus in right superficial femoral vein.

Results	Choose Top of pageABSTRACTIntroductionSubjects and MethodsResults <<DiscussionReferencesCITING ARTICLES

Ten patients (16.4%) of the 61 evaluated were determined to have DVT according to the clinical standard, seven by venous sonography and three by combined positive CT pulmonary angiography and indirect CTV (Figs. 1A, 1B, 2A, 2B, 2C, 3A, 3B). Three patients with DVT on venous sonography but negative indirect CTV results had DVT in the right popliteal, right superficial femoral, and left common femoral veins. Three patients with positive results for DVT on both CT pulmonary angiography and CTV but with negative results on venous sonography had DVT in the left popliteal vein and right superficial femoral vein (n = 2). The sensitivity, specificity, negative predictive value, and positive predictive value were 70% (95% confidence interval, 41.6–98.4%), 96% (90.8–101.4%), 94% (50.6–104.9%), and 77% (87.9–100.6%), respectively, for indirect CTV. The sensitivity and negative predictive value of sonography against the clinical standard were 70% (41.6–98.4%) and 94% (88.3–100.6%).

View larger version (74K)

Fig. 3A —84-year-old man with deep venous thrombosis. Color Doppler sonogram reveals absent flow in right popliteal vein (R POP V).

View larger version (141K)

Fig. 3B —84-year-old man with deep venous thrombosis. Axial CT image, considered false-negative, at same level as A fails to show evidence of filling defect. Segment was considered technically inadequate because of poor enhancement.

Four (6.6%) of the 61 patients had pulmonary embolism, three of whom had negative venous sonography studies and one of whom had indirect CTV and venous sonography with positive results for DVT. In two cases, indirect CTV was considered to be false-positive because of the lack of pulmonary embolism and the lack of confirmatory sonographic findings. Table 2 lists the location of DVT according to our study. In only one instance was there venous thrombosis in the pelvic veins; this patient also had DVT in the lower extremities on both CT and venous sonography.

View Larger Version

TABLE 2: Deep Venous Thrombosis Locations on CT Venography and Lower Extremity Sonography

The image quality grade for CT pulmonary angiography was noted as excellent in 20, good in 26, adequate in eight, and poor in seven patients. The quality of CT pulmonary angiography did not appear to affect the quality of CTV. Although indirect CTV examinations could technically be performed in all patients, 41 (5.2%) of 793 venous segments were not evaluable because of beam-hardening artifact, excess noise, poor venous opacification, or the presence of a catheter, as listed in Table 3. Of these, 16 segments were in regions traditionally evaluable with venous sonography. No venous sonography studies were indeterminate for DVT.

View Larger Version

TABLE 3: Image Quality of CT Venography (CTV) at Individual Segments

Discussion	Choose Top of pageABSTRACTIntroductionSubjects and MethodsResultsDiscussion <<ReferencesCITING ARTICLES

In the critical care setting, venous thromboembolic disease contributes significantly to morbidity and mortality. Most critically ill patients have at least one risk factor predisposing them to thromboembolism [2]. In the ICU, the incidence of asymptomatic DVT has been documented to be as high as 33% [2, 8, 9] and even higher in postoperative orthopedic patients if DVT prophylaxis is not used [18]. Currently, color-flow venous duplex scanning of the proximal and distal veins is the standard for routine clinical assessment of possible lower extremity DVT [11]. The underlying rationale for ordering lower extremity venous sonography in symptomatic patients is that a diagnosis of DVT may indirectly suggest the diagnosis of pulmonary embolism [11], based on the assumption that most pulmonary emboli originate from the lower extremity deep venous system. The sensitivity and specificity of venous sonography in DVT evaluation in the femoral and popliteal veins is 95% [17, 19, 20]. However, these results were obtained under ideal circumstances in symptomatic patients. In the asymptomatic patient and in the critical care setting, the sensitivity of venous sonography decreases significantly when compared with that of venography [7, 19, 21, 22]. Venous sonography is also less accurate for the diagnosis of DVT in the calf veins and pelvic veins [17, 19, 20]. Lower extremity venography is the gold standard in DVT evaluation, but it is often limited by logistics and is technically difficult in the ICU population [23, 24].

Although approximately 90% of pulmonary emboli are thought to originate in the lower extremity deep venous system [25], the sensitivity of sonography as a diagnostic tool for pulmonary embolism has reportedly been low (44%) [26], with two other studies noting between 19% and 77% of patients with angiographically proven pulmonary embolism to have normal Doppler venous flow analysis [23, 26]. Because of the low sensitivity of lower extremity venous sonography in predicting the presence of pulmonary embolism, it is not an appropriate initial test when pulmonary embolism is clinically suspected. Also, despite the documented accuracy of sonography in DVT evaluation of the femoral and popliteal veins, its use can be compromised in the ICU, where factors such as immobilization devices, bandages, lower extremity edema, open wounds, and tenderness can render the examination technically inadequate [11, 19]. The sensitivity and positive predictive value of compression sonography screening for proximal DVT in certain postoperative populations have been as low as 62% and 25%, respectively [7, 21]. Based on the significant impact of thromboembolic disease on the ICU population, and the just-mentioned pitfalls of lower extremity venous sonography, a single accurate examination evaluating both the pulmonary arterial system and lower extremity venous system would offer a significant advantage. We believe that CT pulmonary angiography with CTV fills this role.

In our study, indirect CTV, along with CT pulmonary angiography, was performed in 61 symptomatic ICU patients, with lower extremity duplex venous sonography performed within 24 hr. CTV and sonography were compared with an outcome standard. In none of the literature reviewed was the ICU population addressed specifically, and in none of the literature was an outcome standard used. The sensitivity and specificity of CTV were 70% and 96%, respectively, with a positive predictive value of 77% and negative predictive value of 94%. Duwe et al. [14] performed CTV with CT pulmonary angiography in 74 patients and used lower extremity venous sonography as a standard. In their study, CTV had a sensitivity of 89%, a specificity of 94%, a positive predictive value of 64%, and a negative predictive value of 98%. Loud et al. [13] performed indirect CTV after CT pulmonary angiography and also used lower extremity venous sonography as the standard. In that study, the sensitivity and specificity of CTV were 97% and 100%, respectively, for evaluation of femoropopliteal DVT. Loud et al. also found 17% of DVT in the pelvic or abdominal veins could not be imaged by sonography. In our study, in only one patient was DVT noted in a vein that could not be imaged by lower extremity venous sonography.

In a study performed by Baldt et al. [17], CTV was compared with conventional venography as the standard. In that study, Baldt et al. noted a sensitivity of 100%, specificity of 96%, positive predictive value of 91%, and negative predictive value of 100% for CTV in DVT evaluation. In our study, as noted previously, the sensitivity of CTV is notably lower than in the prior studies comparing CTV with sonography or venography. However, the sensitivity of CTV and the sensitivity of sonography in our study are similar to those of prior studies evaluating the accuracy of sonographic diagnosis of DVT in similar high-risk asymptomatic populations. We believe that the lower sensitivity of CTV in DVT diagnosis in our study is most likely the result of our critical care population, in which similar obstacles, such as orthopedic hardware, venous catheters, and so forth, that often limit sonographic evaluation can also adversely affect indirect CTV. Because sonography is assumed to have a positive predictive value of 100% but remains an imperfect standard, it is possible that any or all of the CTV findings that were considered false-negative may in fact be true-negative. Similarly, CTV findings considered false-positive on the basis of lack of pulmonary embolism and negative sonography may in fact be true-positive. A retrospective unblinded review of these two cases suggests that in at least one case sonography should have been considered false-negative.

Our study has a number of limitations. First, our outcome standard has not been previously validated. Although no prior validation exists, we think that this standard more accurately reflects the pathophysiology of venous thromboembolic disease than does lower extremity venous sonography alone, given the limitations of sonography in our population. In the ICU population we believe venography would be an impossible standard to use because of logistics, the clinical status of our population, and the technical difficulty of performing venography in critically ill patients.

Second, the perfect specificity of lower extremity venous sonography may not reflect the test's true characteristics but may be a result of how the clinical diagnosis of DVT is defined. Previous published studies have determined the specificity of sonography in the evaluation of DVT of the femoral and popliteal veins to be approximately 95% [19, 20].

Another potential limitation of our study was that only a single expert observer, who was involved in the study design, performed the interpretation of CT pulmonary angiography and CTV. This design introduces bias into the interpretation of the images and may not depict the normal situation of daily radiology practice.

In summary, CT pulmonary angiography with CTV offers a single first-line diagnostic test in the ICU for evaluation of suspected DVT that offers high accuracy in diagnosing pulmonary embolism and results comparable to those of lower extremity venous sonography. Therefore, good-quality CTV obviates sonography. In equivocal cases or with poor venous opacification, confirmation with sonography is still recommended.