Deep Venous Thrombosis with Suspected Pulmonary Embolism

Thoracic Imaging

Peter A. Loud, MD Deep Venous Thrombosis Douglas S. Katz, MD Dennis A. Bruce, MD with Suspected Pulmonary Donald L. Klippenstein, MD Zachary D. Grossman, MD Embolism: Detection with Combined CT Venography Index terms: Computed tomography (CT), 1 angiography, 9*.129142, and Pulmonary Angiography 9*.12915, 9*.12916 Embolism, pulmonary, 60.72 Pulmonary angiography, 944.12914, 944.12915, 944.12916 PURPOSE: To determine the frequency and location of deep venous thrombosis at Veins, thrombosis, 9*.751, 9*.12914 computed tomographic (CT) venography after CT pulmonary angiography in a large series of patients clinically suspected of having pulmonary embolism and to Radiology 2001; 219:498–502 compare the accuracy of CT venography with lower-extremity venous sonography. Abbreviation: MATERIALS AND METHODS: Venous phase images were acquired from the DVT ϭ deep venous thrombosis diaphragm to the upper calves after completion of CT pulmonary angiography in 650 patients (373 women, 277 men; age range, 18–99 years; mean age, 63 years) 1 From the Department of Radiology, to determine the presence and location of deep venous thrombosis. Results of CT Roswell Park Cancer Institute, Elm and Carlton Sts, Buffalo, NY 14263 (P.A.L., venography were compared with those of bilateral lower-extremity venous sonog- D.L.K., Z.D.G.), and the Department raphy in 308 patients. of Radiology, Winthrop University Hospital, Mineola, NY (D.S.K., D.A.B.). RESULTS: A total of 116 patients had pulmonary embolism and/or deep venous From the 1999 RSNA scientific assem- thrombosis, including 27 patients with pulmonary embolism alone, 31 patients with bly. Received July 7, 2000; revision re- quested August 19; revision received deep venous thrombosis alone, and 58 patients with both. Among 89 patients with October 4; accepted October 11. Ad- deep venous thrombosis, thrombosis was bilateral in 26, involved the abdominal or dress correspondence to P.A.L. (e- pelvic veins in 11, and was isolated to the abdominal or pelvic veins in four. In mail: [email protected]). patients in whom sonographic correlation was available, CT venography had a 2 9*. Vascular system, location unspec- sensitivity of 97% and a specificity of 100% for femoropopliteal deep venous ified. thrombosis. © RSNA, 2001 CONCLUSION: Combined CT venography and pulmonary angiography can accurately depict the femoropopliteal deep veins, permitting concurrent testing for venous thrombosis and pulmonary embolism. CT venography also defines pelvic or abdominal thrombus, which was seen in 17% of patients with deep venous thrombosis.

Pulmonary thromboembolism is a feared complication of deep venous thrombosis (DVT). The mortality rate in untreated cases is 25%–30%, whereas the mortality rate in treated cases decreases to 5%–8% (1). More than 90% of pulmonary emboli arise from deep veins of the legs and pelvis, and the primary risk factor for recurrent pulmonary embolism is the Author contributions: Guarantors of integrity of entire study, presence of residual proximal venous thrombosis (2,3). P.A.L., D.S.K.; study concepts, P.A.L.; Unfortunately, both pulmonary embolism and DVT are conditions that are notoriously study design, P.A.L., D.S.K.; literature difficult to diagnose clinically. Diagnostic algorithms for the evaluation of suspected research, P.A.L.; clinical studies, P.A.L., thromboembolism have traditionally included ventilation-perfusion lung scanning and D.S.K.; data acquisition and analysis/ interpretation, P.A.L., D.S.K.; statistical conventional pulmonary angiography to evaluate the lungs and lower-extremity sonog- analysis, P.A.L.; manuscript prepara- raphy to evaluate the leg veins, but they have recently evolved to include computed tion and definition of intellectual con- tomography (CT) (4). CT pulmonary angiography is increasingly being used to evaluate tent, P.A.L.; manuscript editing and suspected pulmonary embolism because it accurately defines emboli to the level of seg- revision/review, all authors; manuscript final version approval, P.A.L., mental pulmonary arteries and reveals other nonembolic causes of thoracic symptoms D.S.K. (5–7). Because pulmonary embolism and venous thrombosis are different aspects of the same

498 disease, a single study that accurately de- with a section thickness of 3–5 mm and a Sensitivity, specificity, and positive fines both processes would be a valuable pitch of 1.8 or 2.0 from the diaphragm to and negative predictive values from CT addition to the diagnostic regimen. Com- the aortic arch during a single breath venography, compared with lower-ex- bined CT venography and pulmonary hold, beginning 20–25 seconds after the tremity venous sonography, were calcu- angiography was reported in 1998 (8). start of contrast medium infusion. At lated. Cases with CT-depicted DVT iso- This test, which consists of helical CT both hospitals, transverse venous images, lated to the calf veins, iliac veins, and pulmonary angiography followed by ve- 5–10 mm thick, were acquired at 5-cm vena cava were excluded from these cal- nous phase CT performed from the dia- intervals during approximately 40 sec- culations, as sonography of these areas phragm to the calves, allows concurrent onds from the diaphragm to the upper was not routinely performed. evaluation of pulmonary embolism and calves. At one hospital, the venous study DVT. This technique uses the venous en- began 31⁄2 minutes after the start of con- hancement that follows rapid peripheral trast material infusion and included im- RESULTS venous infusion of iodinated contrast ages acquired from the upper calves up to medium for helical CT pulmonary an- the diaphragm. At the other, the venous Among 650 patients examined, 534 giography and therefore requires no ad- study began 3 minutes after contrast ma- (82%) showed no evidence of pulmonary ditional contrast medium to image the terial administration and included im- embolism or DVT. Among 116 (18%) deep veins. Findings of several subse- ages acquired from the diaphragm to the with positive CT scans—that is, those quent studies (9–11) in which CT venog- upper calves. Both protocols, therefore, with evidence of pulmonary embolism raphy was compared with lower-extrem- were used to image the femoropopliteal and/or DVT—58 had both pulmonary ity sonography have indicated that it is veins between 3 and 4 minutes after the embolism and DVT (Fig 1), 27 had pul- accurate for the evaluation of femoro- administration of contrast material. A to- monary embolism without DVT, and 31 popliteal DVT. tal of 18–20 venous images were typi- had DVT without pulmonary embolism The purpose of this study was to deter- cally obtained. (Fig 2). mine the frequency and location of DVT All CT scans were evaluated for DVT by In 26 of the 89 patients with DVT, with combined CT venography and pul- one of three radiologists (P.A.L., D.S.K., thrombus in the lower extremities was monary angiography in 650 consecutive D.L.K.) blinded to the results of any pre- bilateral. The most proximal location of patients referred for evaluation of sus- vious venous imaging. All radiologists thrombus defined at CT venography was pected pulmonary embolism. In a subset were fellowship trained in body imaging the calf veins or popliteal vein in 25, the of 308 patients, we compared the results and had an additional 4–8 years in prac- superficial femoral vein in 18, the com- with those of lower-extremity venous son- tice. Criteria for a diagnosis of DVT were mon femoral vein in 31, the iliac veins in ography. an intraluminal filling defect or localized five, and the inferior vena cava in 10 (Fig nonopacification of a venous segment. 3). DVT was found only in the superficial The location of DVT was recorded in all femoral vein in one patient and only in MATERIALS AND METHODS patients. In 85 patients, pulmonary em- the common femoral vein in two. Four bolism was diagnosed at helical CT pul- patients had thrombus only in the iliac Between September 1997 and April 2000, monary angiography. veins or inferior vena cava. 650 consecutive patients suspected of hav- The results of all patients undergoing A total of 308 patients underwent bilat- ing pulmonary embolism (373 women, bilateral lower-extremity venous sonog- eral lower-extremity sonography within 24 277 men; age range, 18–99 years; mean raphy within 24 hours before or after CT hours of CT. In this group, femoropopli- age, 63 years) underwent CT venography examination were reviewed. Sonography teal DVT was found at CT in 63 patients, and pulmonary angiography. The two was performed as part of the patient’s and DVT was found at sonography in 65. participating institutions were Winthrop clinical evaluation and involved the use There were two false-negative CT scans. University Hospital, Mineola, NY, and of a standard compression and Doppler Undetected thrombus in the superficial Roswell Park Cancer Institute, Buffalo, technique from the popliteal trifurcation and common femoral veins in these pa- NY, which contributed 509 and 141 cases, to the inguinal level, which included the tients represented short areas of clot, pre- respectively. The patient population in- popliteal vein and the superficial, deep, sumably missed due to the 5-cm section cluded 71 patients from an earlier prelim- and common femoral veins (12). Sono- interval used. In both patients, pulmonary inary study (10). Other patients were graphic findings were considered positive embolism was seen at helical CT pulmo- excluded during that period due to contra- if thrombus prevented complete collapse nary angiography. The sonograms of four indications to contrast medium adminis- of the vein during manual compression patients were initially interpreted as neg- tration, such as allergy or renal insuffi- and caused a lack of flow at Doppler ex- ative for femoropopliteal DVT. The cor- ciency, and unavailable venous access. amination. The reported results of sonog- rect diagnosis of DVT, initially revealed Slightly different protocols were used raphy were reviewed independently by at CT venography, was confirmed in these at the two participating hospitals. Either one of two authors (P.A.L., D.S.K.), and cases with repeat sonography. These gen- 120 mL of iohexol 350 (Omnipaque; Ny- findings were compared with CT veno- erally represented sonographically chal- comed Amersham, Princeton, NJ) was ad- graphic results. In four cases in which CT lenging cases, including one in a patient ministered at a rate of 3 mL/sec or 150 venography revealed femoropopliteal with isolated thrombus in the Hunter ca- mL of iohexol 240 (Omnipaque; Ny- DVT and initial sonographic findings nal (adductor canal), and one in a patient comed Amersham) was administered at a were negative, a repeat sonographic ex- with thrombus in one limb of a dupli- rate of 3–5 mL/sec through an intrave- amination was performed at the clini- cated popliteal vein. The other two were nous catheter in the arm. To evaluate the cian’s request, with particular attention seen in a superficial femoral vein and a pulmonary arteries, a scanner (HiSpeed to the area of concern at CT. In all other distal popliteal vein. Among cases with Advantage; GE Medical Systems, Milwau- cases, sonographers were blinded to CT sonographic findings for comparison, kee, Wis) was used to generate images venographic results. there were 243 true-negative findings, 63

Volume 219 ⅐ Number 2 Deep Venous Thrombosis: CT Venography and Pulmonary Angiography ⅐ 499 Figure 1. Transverse contrast-enhanced CT venogram and pulmonary angiogram in a 71-year-old man with pulmonary embolism and DVT. (a) CT scan obtained at the level of the lower chest shows an embolus in a right lower-lobe pulmonary artery (arrow). (b) Venous phase CT scan obtained at the level of the knee shows DVT as a nonenhancing ﬁlling defect that expands the left popliteal vein (arrow). (c) CT scan obtained at the inguinal level shows thrombus in the left common femoral vein (arrow).

true-positive findings, two false-negative findings, and no false-positive findings. The sensitivity and specificity of CT venography for femoropopliteal DVT, compared with sonography, were 97% and 100%, respectively. Positive and negative predictive values were 100% and 99%, respectively. Figure 2. Transverse CT scan obtained at the midthigh level in a 68-year-old woman with DVT shows a nonenhancing thrombus in DISCUSSION the duplicated left superficial femoral veins (arrows). CT pulmonary angiography failed to show pulmonary embolism. One of the criticisms of helical CT pulmonary angiography as a screening test for pulmonary thromboembolism is that use could potentially eliminate the need 31 (35%), and the iliac veins or vena cava emboli in small subsegmental pulmo- for separate venous imaging in many pa- in 15 (17%). While most DVT propagate nary arterial branches may be missed, tients, and it could expedite the diagnos- craniad from the calf, we found a number leading to inadequate treatment in pa- tic work-up of thromboembolism. of isolated thrombi in the femoral veins, tients at risk for larger, potentially fatal The frequency and location of DVT re- the pelvis, and the abdomen. These may recurrent emboli, namely, patients with vealed with combined CT venography represent fragments of previously larger DVT. Therefore, lower-extremity venous and pulmonary angiography in this study clots, as most of these patients had pul- sonography has been advocated for use are similar to those previously reported monary embolism. in patients with leg symptoms suggestive (14) with the use of conventional venog- In 31 (5%) of 650 patients examined, of DVT or in those with a negative or raphy in patients with pulmonary embo- DVT was detected with no evidence of equivocal finding at helical CT pulmo- lism. Analysis of data in 89 patients with pulmonary embolism. This increased the nary angiography (13). If the accuracy of DVT showed the upper location of throm- number of positive scans by 36% (from combined CT venography and pulmo- bosis to reach the calf or popliteal vein in 85 to 116). It is possible that subsegmen- nary angiography for the detection of 25 (28%), the superficial femoral vein in tal pulmonary emboli were undetected in DVT is similar to that of sonography, its 18 (20%), the common femoral vein in some of these patients at CT. However,

500 ⅐ Radiology ⅐ May 2001 Loud et al Figure 3. Transverse CT scans in a 62-year-old woman with DVT extending to the inferior vena cava. CT pulmonary angiography failed to show pulmonary embolism. (a) Upper pelvic image shows thrombus in the right common iliac vein (arrow). (b) Lower abdominal image shows thrombus in the inferior vena cava (arrow).

because the treatment of both DVT and been reported (9,17,18) after the admin- tremity symptoms or increased risk fac- pulmonary embolism includes anticoag- istration of standard contrast medium tors for DVT. Another limitation of our ulation, one could reasonably argue that doses at CT pulmonary angiography. study is the lack of confirmation of most missed subsegmental emboli, in the pres- Studies of femoral vein attenuation after cases of DVT depicted at CT venography ence of anticoagulation therapy for DVT, helical CT pulmonary angiography show in the veins of the upper calf, pelvis, and are of little consequence. a gradual decline after peak enhance- abdomen with findings from another im- In this study, the sensitivity and spec- ment (17). Although peak venous en- aging test. Comparison of CT venogra- ificity of CT venography, compared with hancement may occur 2–3 minutes after phy with conventional venography with bilateral venous sonography, were 97% the administration of contrast material the use of bilateral pedal venous injec- and 100% for femoropopliteal DVT de- in most patients, waiting 3–4 minutes tion, the accepted standard for the eval- tection, respectively. Two false-negative allows for uniform enhancement in all uation of lower-extremity and iliac DVT, CT venograms were related to short areas patients, including those with slower cir- would provide more complete confirma- of superficial and common femoral vein culation times. The principle underlying tion of CT venographic results but would clot that were likely missed due to the our slightly longer delay, therefore, is to require additional venipunctures and 5-cm section interval used for DVT achieve a diagnostic image in virtually all contrast medium administration. As with screening. Both of the patients with these patients, rather than an aesthetically su- sonography, differentiation of acute and findings had pulmonary embolism. Such perior image in most patients and a non- chronic DVT can be difficult with CT small clot fragments could presumably diagnostic image in the remainder. venography. Further research is needed be detected by decreasing the section in- Unlike sonography, CT venography to determine if certain CT features of terval to 2–3 cm. consistently depicts the large pelvic and DVT can be used to reliably predict the Our use of a 5-cm interval between ve- abdominal veins, which harbored clots age of thrombus. nous images is based on the fact that in 15 (17%) of 89 patients in whom DVT In 85 patients with pulmonary embo- small isolated thrombi are unusual (15) was seen at CT. Moreover, in three (4%) lism, DVT was depicted at CT venogra- and that even limited sonographic sur- of 85 patients with pulmonary embo- phy in 56 (66%). In previously published veys that include only the popliteal and lism, DVT was confined to the iliac veins findings of patients with pulmonary em- common femoral veins depict the great or inferior vena cava, with no evidence of bolism, DVT was found in 71%–83% of majority of venous thrombi (16). Some distal DVT. The detection of thrombi in patients with conventional bilateral leg researchers (11) obtain contiguous CT ve- the large pelvic or abdominal veins is venography (14,20) and in 38%–49% of nous phase images without a section in- important for prognostication in terms of patients with bilateral leg sonography terval. Given the high sensitivity and future pulmonary embolism and severity (21,22). This suggests that the overall specificity that we have shown by using a of postphlebitic symptoms (19). CT venog- DVT detection rate with CT venography 5-cm interval between venous images, raphy also provides the radiologist with a in our patient population more closely the elimination of the section interval useful road map for planning interven- approximates that of conventional venog- entirely would lead to only a minimal tional procedures, such as vena caval fil- raphy than that of sonography, possibly increase in sensitivity for DVT despite a ter placement or thrombolysis. due to detection of additional DVT in the substantial increase in radiation dose, One limitation of this study is the lack veins of the calf, pelvis, or abdomen. number of images, and cost. of sonographic correlation in the major- In summary, while helical CT pulmo- We acquire CT venous phase images ity of patients. The patients that did un- nary angiography addresses pulmonary 3–4 minutes after the initiation of con- dergo sonography had a higher preva- embolism and other nonembolic disease trast medium infusion into an arm vein. lence of DVT. Although we did not in the chest, the addition of venous This delay allows venous blood to mix evaluate the symptoms for which the pa- phase imaging of the legs, pelvis, and uniformly with contrast medium and re- tients were referred or predisposition to abdomen allows concurrent, accurate turn from the lower leg. Mean femoral DVT, it is likely that a higher percentage evaluation for underlying femoropopli- venous attenuations of 94–112 HU have of patients in this group had lower-ex- teal venous thrombus—the major risk

Volume 219 ⅐ Number 2 Deep Venous Thrombosis: CT Venography and Pulmonary Angiography ⅐ 501 factor for subsequent embolism. Further- pulmonary angiography: a new diagnos- with symptomatic deep vein thrombosis. more, CT venography depicts pelvic and tic technique for suspected thromboem- Arch Intern Med 1993; 153:2777–2780. bolic disease. AJR Am J Roentgenol 1998; 16. Pezullo JA, Perkins AB, Cronan JJ. Symp- abdominal thrombi and thus provides an 170:951–954. tomatic deep vein thrombosis: diagnosis important advantage over lower-extrem- 9. Garg K, Kemp JL, McLaughlin CM, et al. with limited compression US. Radiology ity sonographic screening for DVT. Thromboembolic disease: prospective 1996; 198:67–70. comparison of combined CT pulmonary 17. Yankelevitz DF, Gamsu G, Shah A, et al. angiography and venography with bilat- Optimization of combined CT pulmo- References eral leg sonography in 64 cases (abstr). nary angiography with lower extremity 1. Chan CK, Matthay RA. Pulmonary AJR Am J Roentgenol 2000; 174(suppl): CT venography. AJR Am J Roentgenol thromboembolism. In: Fauci AS, Braun- 83. 2000; 174:67–69. wald E, Isselbacher KJ, et al, eds. Princi- 10. Loud PA, Katz DS, Klippenstein DL, Shah 18. Matar LD, Ramirez JA, McAdams HP, Far- ples of internal medicine. 5th ed. St Louis, RD, Grossman ZD. Combined CT venog- rell MA, Herndon JE. Optimal timing of Mo: Mosby, 1998; 499–504. raphy and pulmonary angiography in CT venography following CT pulmonary 2. Byrne JJ, O’Neil EE. Fatal pulmonary em- suspected thromboembolic disease: diag- angiography using a multidetector row boli: a study of 130 autopsy-proven fatal nostic accuracy for deep venous evalua- helical scanner—work in progress (abstr). emboli. Am J Surg 1952; 83:47–54. tion. AJR Am J Roentgenol 2000; 174:61– Radiology 1999; 213(P):472. 3. Kakkar VV, Howe CT, Flang C, Clarke, 65. 19. O’Donnell TF, Browse NL, Burnand KG, MB. Natural history of postoperative deep- 11. Cham MD, Yankelovitz DF, Shaham D, et Thomas ML. The socioeconomic effects vein thrombosis. Lancet 1969; 2:230–233. al. Deep venous thrombosis: detection by of an iliofemoral venous thrombosis. 4. Goodman LR. CT diagnosis of pulmonary using indirect CT venography. Radiology J Surg Res 1977; 22:483–488. embolism and deep venous thrombosis. 2000; 216:744–751. 20. Hull RD, Hirsh J, Carter CJ, et al. Pulmo- RadioGraphics 2000; 20:1201–1205. 12. Lewis BD. The peripheral veins. In: Ru- nary angiography, ventilation lung scan- 5. Van Rossum AB, Pattynama PMT, Tjin A mack CM, Wilson RW, Charboneau JW, ning, and venography for clinically sus- Ton ER, et al. Pulmonary embolism: vali- eds. Diagnostic ultrasound. 2nd ed. St pected pulmonary embolism with abnormal dation of spiral CT angiography in 149 Louis, Mo: Mosby, 1998; 943–958. perfusion lung scan. Ann Intern Med 1983; patients. Radiology 1996; 201:467–470. 13. Goodman LR, Lipchik RJ. Diagnosis of 98:891–899. 6. Mayo JR, Remy-Jardin M, Muller NL, et al. acute pulmonary embolism: time for a 21. Barnes RW, Kinkead LR, Wu KK, Hoak JC. Pulmonary embolism: prospective com- new approach. Radiology 1996; 199:25– Venous thrombosis in suspected pulmo- parison of spiral CT with ventilation-per- 27. nary embolism: incidence detectable by fusion scintigraphy. Radiology 1997; 205: 14. Girard P, Musset D, Parent F, Maitre S, Doppler ultrasound. Thromb Haemost 447–452. Phlippoteau C, Simonneau G. High prev- 1976; 36:150–156. 7. Kim KI, Muller NL, Mayo JR. Clinically alence of detectable deep venous throm- 22. Smith LL, Iber CI, Sirr S. Pulmonary em- suspected pulmonary embolism: utility of bosis in patients with acute pulmonary bolism: conﬁrmation with venous duplex spiral CT. Radiology 1999; 210:693–697. embolism. Chest 1999; 116:903–908. US as adjunct to lung scanning. Radiol- 8. Loud PA, Grossman ZD, Klippenstein DL, 15. Cogo A, Lensing WA, Prandoni P, Hirsh J. ogy 1994; 191:143–147. Ray CE. Combined CT venography and Distribution of thrombosis in patients

502 ⅐ Radiology ⅐ May 2001 Loud et al