Introduction	Choose Top of pageABSTRACTIntroduction <<Materials and MethodsResultsDiscussionReferencesCITING ARTICLES

Pulmonary embolism and deep venous thrombosis should be considered part of the same pathologic process because more than 90% of pulmonary emboli arise from lower extremity deep venous thrombosis [1]. Combined CT venography and helical pulmonary angiography is a new diagnostic examination by which radiologists can check both the pulmonary arteries for pulmonary embolism as well as the deep veins of the abdomen, pelvis, and legs for thrombosis in a single study [2, 3]. The purpose of our study was to determine for a large population of patients the degree of venous enhancement routinely obtained on combined CT venography and pulmonary angiography using a 3-min delay.

Materials and Methods	Choose Top of pageABSTRACTIntroductionMaterials and Methods <<ResultsDiscussionReferencesCITING ARTICLES

We retrospectively identified all adult patients who had clinical indications of pulmonary embolism and who had been referred for combined CT venography and helical pulmonary angiography at a single institution over a 29-month period. Only those patients whose CT examinations did not show evidence of deep venous thrombosis (on the basis of the official CT reports) were retained for our study. Patients underwent CT pulmonary angiography after antecubital IV injection of 150 mL of iohexol (240 mg I/mL) at a rate of 3-5 mL/sec. The patients' informed consent was obtained for the administration of IV contrast material. After helical CT of the pulmonary arteries was performed on the patients, 10-mm-thick axial images taken 50 mm apart were obtained from an area ranging from the diaphragm to the ankles beginning 3 min after the start of the IV contrast injection. No additional contrast medium was administered for the venous images. The timing for the venous images was performed by the CT technologist using a clock or stopwatch. Approximately 20 venous images were acquired for each patient, and the time required for CT venographic examination was 30-40 sec at most.

The CT venographic images for all patients were retrospectively reviewed by one of two radiologists. The CT venographic images were reloaded from optical disks onto a CT monitor or workstation. Hounsfield unit measurements were obtained at preselected levels for each patient; these included the inferior vena cava at or just below the level of the kidneys, the right internal or external iliac veins at the level of the sacroiliac joints, the right and left common femoral veins at or just below the level of the pubic symphysis, the right and left superficial femoral veins at the level of the mid thighs, and the right and left popliteal veins at or just above the knee (Fig. 1A,1B,1C,1D,1E). A single Hounsfield unit measurement was obtained from the center of each vessel using a region of interest that was approximately half the diameter of the vessel being examined.

View larger version (145K)

Fig. 1A. —Representative images from single CT venographic study for a 50-year-old woman shows levels used for measuring venous enhancement. Venographic image shows region-of-interest cursor placed in center of inferior vena cava (1). Hounsfield unit (H) measurement was 106.21 H.

View larger version (148K)

Fig. 1B. —Representative images from single CT venographic study for a 50-year-old woman shows levels used for measuring venous enhancement. Venographic image shows pelvis with internal iliac veins labeled 1 (113.88 H) and 2 (117.69 H).

View larger version (125K)

Fig. 1C. —Representative images from single CT venographic study for a 50-year-old woman shows levels used for measuring venous enhancement. Venographic image shows lower pelvis with common femoral veins labeled 1 (127.33 H) and 2 (119.21 H).

View larger version (72K)

Fig. 1D. —Representative images from single CT venographic study for a 50-year-old woman shows levels used for measuring venous enhancement. Venographic image shows thighs with superficial femoral veins labeled 1 (127.20 H) and 2 (127.82 H).

View larger version (57K)

Fig. 1E. —Representative images from single CT venographic study for a 50-year-old woman shows levels used for measuring venous enhancement. Venographic image obtained at level of distal femurs shows proximal popliteal veins labeled as 1 (127.56 H) and 2 (112.90 H).

Twenty-five patients were excluded from further analysis for one of three reasons—venous enhancement could not be measured at all present levels, the images were missing from the reloaded examination, or streak artifacts from a prosthesis or other orthopedic hardware were present on the images. For the 429 CT examinations that we successfully measured at all present levels, we calculated the mean Hounsfield unit measurement and the corresponding standard deviations for each level.

Results	Choose Top of pageABSTRACTIntroductionMaterials and MethodsResults <<DiscussionReferencesCITING ARTICLES

The mean Hounsfield unit measurements for the 429 examinations are summarized in Table 1. The standard deviations were all between 20 and 22 H (Table 1). The range and distribution of all measurements are shown in Figure 2A,2B,2C,2D,2E for the inferior vena cava (Fig. 2A), the iliac veins (Fig. 2B), the common femoral veins (Fig. 2C), the superficial femoral veins (Fig. 2D), and the popliteal veins (Fig. 2E).

View Larger Version

TABLE 1 Hounsfield Unit Measurements for 429 CT Venographic Examinations

View larger version (24K)

Fig. 2A. —Range and distribution of all Hounsfield unit measurements at each venous level for all 429 CT venographic examinations (number of examinations shown on y-axes, Hounsfield measurement levels on x-axes; measurements for right shown in white, for left and for vena cava in black). Note that measurements are 61-140 H for most CT examinations at all venous stations. Range and distribution of Hounsfield unit measurements for inferior vena cava (A), for right and left iliac veins (B), for right and left common femoral veins (C), for right and left superficial femoral veins (D), and for right and left popliteal veins (E).

View larger version (28K)

Fig. 2B. —Range and distribution of all Hounsfield unit measurements at each venous level for all 429 CT venographic examinations (number of examinations shown on y-axes, Hounsfield measurement levels on x-axes; measurements for right shown in white, for left and for vena cava in black). Note that measurements are 61-140 H for most CT examinations at all venous stations. Range and distribution of Hounsfield unit measurements for inferior vena cava (A), for right and left iliac veins (B), for right and left common femoral veins (C), for right and left superficial femoral veins (D), and for right and left popliteal veins (E).

View larger version (28K)

Fig. 2C. —Range and distribution of all Hounsfield unit measurements at each venous level for all 429 CT venographic examinations (number of examinations shown on y-axes, Hounsfield measurement levels on x-axes; measurements for right shown in white, for left and for vena cava in black). Note that measurements are 61-140 H for most CT examinations at all venous stations. Range and distribution of Hounsfield unit measurements for inferior vena cava (A), for right and left iliac veins (B), for right and left common femoral veins (C), for right and left superficial femoral veins (D), and for right and left popliteal veins (E).

View larger version (29K)

Fig. 2D. —Range and distribution of all Hounsfield unit measurements at each venous level for all 429 CT venographic examinations (number of examinations shown on y-axes, Hounsfield measurement levels on x-axes; measurements for right shown in white, for left and for vena cava in black). Note that measurements are 61-140 H for most CT examinations at all venous stations. Range and distribution of Hounsfield unit measurements for inferior vena cava (A), for right and left iliac veins (B), for right and left common femoral veins (C), for right and left superficial femoral veins (D), and for right and left popliteal veins (E).

View larger version (31K)

Fig. 2E. —Range and distribution of all Hounsfield unit measurements at each venous level for all 429 CT venographic examinations (number of examinations shown on y-axes, Hounsfield measurement levels on x-axes; measurements for right shown in white, for left and for vena cava in black). Note that measurements are 61-140 H for most CT examinations at all venous stations. Range and distribution of Hounsfield unit measurements for inferior vena cava (A), for right and left iliac veins (B), for right and left common femoral veins (C), for right and left superficial femoral veins (D), and for right and left popliteal veins (E).

Discussion	Choose Top of pageABSTRACTIntroductionMaterials and MethodsResultsDiscussion <<ReferencesCITING ARTICLES

The combination of CT venography and helical pulmonary angiography is a promising diagnostic test, first introduced in 1998, that allows radiologists to evaluate both the pulmonary arteries for embolism and the deep veins of the abdomen, pelvis, and legs for thrombosis in a single examination [2,3,4,5]. Because more than 90% of pulmonary emboli arise from lower extremity deep venous thrombosis [1], pulmonary embolism and deep venous thrombosis should be considered part of the same pathologic process. CT pulmonary angiography is being increasingly used for examining patients who are thought to have pulmonary embolism. The addition of CT venography permits radiologists to examine a patient's entire subdiaphragmatic deep venous system, an important feature considering the primary prognostic factor for recurrent pulmonary embolism is residual deep venous thrombosis [6]. CT venography can also reveal deep venous thrombosis in regions of the deep venous system not routinely examined at sonography [7], including the inferior vena cava, pelvic veins, and calf veins.

A preliminary study of combined CT venography and pulmonary angiography in 71 patients revealed deep venous thrombosis in 19 patients, 12 of whom had pulmonary embolism [3]. Deep venous thrombosis was detected as a filling defect surrounded by opacified blood or as a nonopacified lumen with the enhancement of the venous wall [3]. In this group of 71 patients, there was exact correlation between CT venography and venous sonography in the femoropopliteal deep venous system. In addition, CT venography revealed pelvic extension of deep venous thrombosis in six patients and isolated vena caval thrombosis in one patient. Similarly, Censullo et al. (Censullo ML et al., presented at the Radiological Society of North America meeting, December 1999) recently reported that the addition of CT venography to CT pulmonary angiography increased the accuracy of detection of thromboembolic disease from 69% to more than 90%. Very recently, Cham et al. [4] performed a prospective multicenter study using combined CT venography and pulmonary angiography on 541 patients whom the researchers believed to have pulmonary embolism. Deep venous thrombosis was found in 45 patients (8%), including 16 in whom no pulmonary embolism was identified on CT. The number of patients with the diagnosis of thromboembolic disease therefore increased by 18% with the addition of CT venography [4].

To exclude or diagnose deep venous thrombosis on CT venographic images, we chose a delay of 3 min to allow time for uniform venous opacification. We hypothesized that at 3 min from the start of IV contrast medium injection for CT pulmonary angiography, a high level of opacification would be present in the deep venous system. In a preliminary study by Loud et al. [3] using 120 mL of nonionic contrast medium (350 mg I/mL) and a delay of 3.5 min, the mean Hounsfield unit measurements of nonthrombosed veins were 99 H in the mid inferior vena cava, 94 H in the common femoral vein, and 96 H in the popliteal vein; the corresponding standard deviations for these veins were 19, 19, and 21 H. Sixty-five of the 71 patients in that study received 120 mL of nonionic contrast material (350 mg I/mL). The current study, using 150 mL of nonionic contrast material (240 mg I/mL), confirms our hypothesis: Mean Hounsfield unit measurements at all venous stations were 91-97 H, and standard deviations were 20-22 H. Similarly, in the study of 541 patients by Cham et al. [4] (using a protocol of 140 mL of nonionic contrast material, 300 mg I/mL, and a 120-sec delay for CT venography after the completion of CT pulmonary angiography), the mean Hounsfield unit measurement of the common femoral vein was 101 H.

As shown in Figure 2A,2B,2C,2D,2E, most Hounsfield unit measurements at all venous stations ranged from 61 to 140 H. Of all venous stations with the exception of the superficial femoral veins, approximately only 20 or fewer of the 429 CT venographic examinations (≤ 5%) reviewed for our study had Hounsfield unit measurements below 61 H, and virtually no patients had measurements below 41 H. In the study by Loud et al. [3], the mean density of venous thrombi in 19 patients (sonographically confirmed in the femoropopliteal regions) was 31 ± 10 H. In the study by Cham et al. [4], the mean density of 43 thrombosed common femoral veins was 51 H, higher than the 31 H obtained in the study by Loud et al., but again well below the 60 H above which almost all venous stations in our study measured. These data strongly suggest that there should be little overlap in density between patent and thrombosed veins in almost all CT venographic examinations and that the high levels of venous enhancement routinely obtained on CT venography using our or similar protocols should maximize the opportunity to detect deep venous thrombosis.

The optimal timing for CT venography has been debated; however, to our knowledge, only a relatively small number of patients have been examined in studies on this issue. Yankelevitz et al. [8] obtained time-density curves of the common femoral veins in 20 patients after CT pulmonary angiography. Measurements were obtained every 30 sec for 5 min after the administration of 140 mL of 300 mg I/mL. The mean measurement at peak contrast was 95 H. The time required to obtain peak venous enhancement varied but increased slowly and gradually compared with the density of the adjacent common femoral artery. Eighty-five percent of patients were within 90% of the peak Hounsfield unit value at 3 min from the start of IV contrast injection, and near-peak enhancement was achieved in most patients after 2 min from the start of the injection [8].

Matar et al. (Matar LD et al., presented at the Radiological Society of North America meeting, December 1999) measured venous density in the common femoral vein in 11 patients after the administration of 150 mL of 300 mg I/mL of IV contrast material, every 15 to 30 sec beginning 50 sec from the start of IV contrast medium injection. The peak enhancement delay time was 94 sec. Mean Hounsfield unit measurement was 112 H (range, 73-187 H; SD, ± 26). In seven patients, there was an early peak and a slow decrease in enhancement. In four patients, enhancement peaked after 125 sec. These investigators therefore suggested that a delay of less than 3 min may be optimal.

Patel et al. (Patel S et al., presented at the Society of Thoracic Radiology meeting, March 2000) performed combined CT venography and pulmonary angiography in 70 patients after administrating 150 mL of IV contrast medium. A delay of 2.5 min was used for the venous imaging for the first 35 patients, and a 3-min delay was used for the remaining 75 patients. The mean Hounsfield unit measurement at the 2.5-min delay versus the mean measurement at the 3-min delay was 107 versus 98 H for the inferior vena cava, 110 versus 100 H for the common iliac vein, 106 versus 91 H for the common femoral vein, and 108 versus 96 H for the popliteal vein. The venous attenuation differences were all significantly greater statistically for the 2.5-min delay compared with such differences for the 3-min delay with the exception of the findings for the inferior vena cava. Although some of these studies suggest that the optimal delay is less than 3 min from the start of IV contrast medium injection, the mean level of enhancement routinely achieved in patent deep veins below the diaphragm using any of these CT protocols is so much higher than the mean density of venous thrombosis that exact timing using one specific protocol may not be crucial for most patients.

There are some limitations to our study. We did not perform a formal qualitative analysis on the presence or absence of mixing artifacts on the venous images, but we never observed such artifacts. However, Garg et al. [5] did note mixing artifacts in two of 70 CT venographic studies using a 3-min delay, which resulted in false-positive examinations for these two patients. These authors now recommend use of a 4-min delay for patients with suspected slow flow or abnormal hemodynamic status but agree with us regarding the effectiveness of a 3-min delay for examining most patients. In addition, a small number of patients in our study had to be excluded because of incompletely stored examinations or artifacts from orthopedic hardware, but we do not believe that these exclusions biased the overall results.

In conclusion, we found that beginning CT venography of the abdomen, pelvis, and lower extremities 3 min after the start of injection of 150 mL of nonionic contrast medium (240 mg I/mL) for helical CT pulmonary angiography routinely produced high mean levels of enhancement—between 91 and 97 H—at all venous stations in our group of 429 examinations. Several studies to date using this and similar protocols for CT venography have shown high accuracy compared with the results obtained from sonography [3, 4]. Further investigations would be useful to examine how to strike a balance between decreasing the dose of contrast medium required for multidetector CT pulmonary angiography and maintaining good levels of enhancement on CT venography, as well as to further refine the optimal slice thickness and slice interval while simultaneously minimizing radiation doses [5, 9].