Limitations of magnetic resonance imaging and ultrasound-directed (duplex) scanning in. the diagnosis of subclavian-axillary vein thrombosis.

Transcription

1 Limitations of magnetic resonance imaging and ultrasound-directed (duplex) scanning in the diagnosis of subclavian vein thrombosis William D. Haire, MD, Thomas G. Lynch, MD, Gmmar B. Lurid, MD, Robert P. Lieberman, MD, and James A. Edney, MD, Omaha, Neb. To investigate the potential role of magnetic resonance imaging and duplex scanning in the diagnosis of catheter-induced subclavian vein thrombosis, we correlated the results of 43 arm phlebograms with duplex scans; 28 of these phlebograms were also correlated with magnetic resonance imaging scans of the thoracic veins. Eighteen of the 43 phlebograms were normal, and all had normal magnetic resonance imaging and duplex studies. Eleven subclavian veins were totally occluded on phlebography; all had duplex scans, and five were also scanned with magnetic resonance imaging. Duplex scans detected 6 of 11 occlusions, whereas magnetic resonance imaging detected 4 of the 5 occlusions scanned. The five occlusions that were not detected by either magnetic resonance imaging or duplex scans were short segmental occlusions of the medial one third of the left subclavian vein. Of 14 nonocclusive thrombi seen on phlebography, duplex scans correctly identified 8. Magnetic resonance imaging was done on eight nonocclusive thrombi but identified only two. All abnormal findings on duplex scanning and magnetic resonance imaging were confirmed by phlebography. Short occlusions of the proximal portion of the left subclavian vein were often undetected by duplex scanning but occasionally seen with magnetic resonance imaging. Neither modality was sensitive to the presence of nonocclusive mural thrombi. Magnetic resonance imaging is highly reliable in ruling out the presence of a thrombotic process in the subclavian vein, but it may on occasion fail to detect the presence of subclavian thrombi. For this reason, in cases with suspected subclavian vein thrombosis magnetic resonance imaging cannot be used as the only diagnostic modality. Contrast phlebography has to be used for positive exclusion of the possibility of vein thrombosis. (J VAse SURG 1991;13:391-7.) Because functional vascular laboratory tests are not considered definitive, authorities have suggested :hat radiographic contrast medium or isotope phlebography is currently required to objectively establish the diagnosis of subclavian-axillary vein thrombosis. ~ However, in many clinical circumstances a noninvasive method of diagnosing this disorder would be desirable. We have prospectively applied duplex ultrasonography and magnetic resonance imaging (MRI) to a group of patients at risk for subclavian vein thrombosis and compared the results with standard contrast phlebography in an attempt to establish the accuracy of these noninvasive techniques. From the Departments of Internal Medicine (Dr. Hake), Surge D, (Drs. Lynch and Edney), and Radiology (Drs. Lund and Lieberman), University of Nebraska Medical Center, Omaha. Presented at the Second Annual American Venous Fortmq Meeting, Coronado, Calif., Feb. 2i to 23, Reprint requests: William D. Hake, MD, Department of Internal Medicine, University of Nebraska Medical Center, 42nd and Dewey, Omaha, NE /1/25130 MATERIAL AND METHODS Patients. The risk factor for subclavian vein thrombosis in all patients was the presence of a silicone rubber catheter placed in the vein,~ M1 but four examinations were performed as part of a prospective evaluation of thrombotic complications of central venous catheters. The experimental protocol was approved by the Institutional Review Board of the University of Nebraska Medical Center, and written informed consent was obtained from all participants. Participants in this study underwent diagnostic imaging examinations before catheter removal or when signs or symptoms of subclavian vein thrombosis developed. The four examinations performed separate from this prospective study were performed to evaluate symptoms of acute subclavian vein obstruction. Informed consent for performance of the examinations was obtained in these individuals. Phlebography. A 21-gauge needle was inserted into a vein of the forearm. Thirty milliliters of contrast material (Conray, iothalamate meglumine) containing 282 mg I/ml was injected by hand with the 391

2 392 Haire et al. Journal of VASCULAR SURGERY Table I. Comparison of the venogram and duplex scan results in imaging subclavian veins Duplex scan Results of phlebography Partial Complete Normal occlusion occlusion Normal ~ Partial occlusion Complete occlusion ~ -- 6 Total ~Occlusion of proximal portion of left subclavian vein. Table II. Comparison of the venogram and MRI scan results in imaging subclavian veins MRI scan Results of phlebography Partial Complete Normal occlusion ocdusion Normal Partial occlusion Complete occlusion Total arm in a dependent position. As the arm was raised, 10 serial x-ray films of the upper chest and shoulder were obtained at the rate of one per second. Complete venous occlusion was diagnosed when both of the following features were present: (1) A complete cessation of flow of contrast at any point along the length of the subclavian vein, and (2) filling of collateral vessels from the vein lateral to the point of cessation of flow. Partial obstruction was diagnosed when a vessel demonstrated both of the following criteria: (1) A narrowing of the lumen of the vessel at any point along its length, but with flow through the entire length of the vein and entry of the contrast medium into the superior vena cava, and (2) no evidence of collateral circulation around the point of narrowing... Duplex ultrasound. Real-time B-mode imaging of,the: :: sui~ciavian Vein with simultaneous pulsed DOppler flow velocity measurement was conducted with the padent in the supine position. The neck was extended and the head turned away from the side of the examination. The subclavian vein was evaluated in the longitudinal and transverse planes from the sternum to the axilla. The vein was identified by its relationship to the subclavian artery, and windows superior and inferior to the clavicle were used to permit visualization of the entire length of the vessel. Imaging was accomplished by use of a Hoffrel model 518 SD (Hoffrel Instruments Inc., Norwalk, Conn.) sector scanner and a 7.5 MHz probe to visualize the subclavian vein. A 5 MHz pulsed Doppler was used to assess the venous flow velocity characteristics. A study was defined as normal if the vein demonstrated both of the following characteristics: (1) The vein collapsed with deep inspiration and probe pressure, and (2) the flow velocity was phasic and varied with respiration and/or atrial contraction. Complete occlusion was defined as a vein that could not be compressed with probe pressure and had no blood flow detected by Doppler. A partial occlusion was defined by either incomplete compressibility or complete compressibility with continuous, nonphasic flow on Doppler. Magnetic resonance imaging. Magnetic resonance imaging was done with a 1.5 T (Tesla) syste~ (General Electric, Milwaukee, Wis.). At the time 6~ this study cine techniques were not available for this system. Five millimeters thick contiguous coronal spin echo images were obtained by use of a 24 cm field of view and a 128 x 256 acquisition matrix. Respiratory gating was used with a repetition time TR = 600 msec and an echo time TE = 20 msec. Additional 10 mm thick axial images through the superior vena cava were obtained by use of a low flip angle (30 degrees) and gradient recall echos with a TR = 100 msec. With the spin echo images a vein was considered occluded if no structure demonstrating flow void signal was seen in the expected location of the subclavian or innominate veins. Ifa venous structure with a flow void was seen, the veins were considered to be patent. Partial occlusion was diagnosed when flow void was seen but with an area of significant narrowing along the course of the vein. On the gradient echo axial images loss of signal in the superior vena cava was seen in the location of the catheters. Additional intraluminal signal loss was considered evidence of thrombosis. Data analysis. The phlebograms were read without knowledge of the results of the MRI or the duplex scan. The duplex ultrasound was evaluated without knowledge of the results of either the MRI or the arm ph!ebogram. The MRI scans were read without knowledge ofth e results of either the phlebogram or the duplex ultrasound. The scans of patients with discordant results were reviewed in conference by all authors to ensure agreement on scan interpretation before data analysis. During this review no changes were made in the original interpretations of the scans. RESULTS Forty-three arm phlebograms were obtained with duplex ultrasound examinations shortly thereafter.

3 Volume 13 Number 3 March 1991 Magnetic resonance imaging and duplex scanning in subclavian vein thrombosis 393 Fig. 1. A, Venogram of a normal subclavian and axillary vein. B, MR1 scan oft_he same patient shows a patent subclavian vein. Twenty-eight of these phlebograms had MRI scans for correlation as well. The results are summarized in Tables I and II. Eighteen of these phlebograms were normal; all had normal duplex and MRI (15 venograms) scan results (Fig. 1). Eleven subclavian veins were found to be totally occluded on phlebography. All 11 had duplex ultrasonography and 5 also had MRI. Duplex ultrasonography detected 6 of the 11 occlusions, whereas MR[ detected 4 of 5 obstructions (Fig. 2). The five total occlusions that were not

4 394 Haire et al. Journal of VASCULAR SURGERY Fig. 2. A, Venogram of a left subclavian vein totally obstructed by thrombus (arrow). B, MRI scan of the same vein shows lack of flow signal in the innominate vein and the medial portion of the subclavian vein (arrow). detected by either MRI or duplex scanning were all short segmental occlusions of the medial one third of the left subclavian vein (Fig. 3). All of these obstructed veins had good flow in the distal two thirds of their length and had large collaterals around the obstruction. The venous anatomy of the one occluded left subclavian vein that was correctly identiffed by duplex scan (Fig. 2, A) was not appreciably

5 Volume 13 Number 3 March 1991 Magnetic resonance imaging and duplex scanning in subclavian vein thrombosis 395 different from that of the five not detected by duplex scan (Fig. 3, A). MI of the five total occlusions of the right subclavian vein were detected by duplex scanning. Duplex scanning revealed only 8 of the 14 nonocclusive thrombi seen on phlebography. Magnetic resonance imaging was done on eight veins with nonocclusive thrombi but identified only two. Thirty-two phlebograms were obtained on individuals with no symptoms, and 11 were obtained to evaluate symptoms of acute thrombosis. Those obtained on individuals with no symptoms were done a mean of 41.6 days (range 24 to 69 days) after catheter placement, suggesting that any thrombi seen were less than a few months old. Five total occlusions were seen in these patients, only two of which were visualized by duplex scanning. SIX complete occlu- ~.ons and 3 partial occlusions were found in the 11 phlebograms obtained on patients with acute symptoms suggesting subclavian vein thrombosis. Only four of the six symptomatic complete occlusions were seen by duplex scanning. MI three symptomatic partial occlusions were detected by duplex scanning. DISCUSSION Real-time ultrasonography has been found to be useful in the diagnosis of venous thrombosis of the legs. 2,3 When applied to the subclavian veins this technique has been found helpful in a limited number of patients. 4-7 These four reports describe the ultrasound findings of 24 subclavian veins proven phlebography to be totally obstructed by thrombus. Ultrasound findings of only six normal veins on phlebography and no veins only partially occluded by :hrombi have been reported. 7 Consequently statements of the sensitivity and specificity of this technique cannot be made. Magnetic resonance imaging can visualize venous thrombi in the legs and abdomen, 81 but experience with the thoracic vasculature is limited to the arterial system or the superior vena cava.lm2 This study evaluates the role of duplex ultrasonography and MR/in the noninvasive diagnosis of subclavian vein thrombosis by use of the reporting standards previously published. ~ As required by these standards, we have (1) studied a population homogenous with respect to the cause of subclavian thrombosis--central venous catheterization, (2) used contrast medium phlebography to define the venous anatomy--both in cases of normal and abnormal veins, (3) described the technique of performance of the noninvasive tests and phlebography, and (4) included the results of these tests in a negative control population (patients with central venous catheters with normal venous anatomy proved on venography. In this study of 25 subclavian vein thrombi, we found both MRI and duplcx ultrasonography to be very specific in their ability to image thrombotic abnormalities of the subclavian vein. All abnormalities identified by these modalities wcre confirmed by arm vein phlebography. This occurred with veins only partially obstructed by a mural thrombus as well as with veins whose lumens were totally occluded. False-positive duplex or MRI examination outcomes were not seen. This study suggests that if an abnormality is detected by either MRI or duplex scanning it is unlikcly to be an artifact of technique. The sensitivity of these studies is, however, not as good as their specificity. This is particularly true in cases ofnonobstructive mural thrombi, where only 8 of 14 were imaged by duplex scanning, and 2 of 8 were visualized by MRI. Similar results have been obtained when ultrasound is used to image nonoc~ clusive thrombi in the deep veins of the legs. 13 Thc insensitivity of duplex ultrasonography to totally occlusive thrornbi in the subclavian vein was, surprisingly, limited to the left subclavian vein. At this site duplex detected only one of five totally occlusivc thrombi. On the right side duplex detected all five totally occlusive thrombi found by phlcbography. All of the left-sided thrombi not visualized by duplcx scanning were short segmental occlusions of the proximal one third of the subclavian vein. All of these veins had good flow through the distal two thirds of their length. The reason for nonvisualizafion of these occlusions can only be postulated. In the patients with asymptomatic occlusions thc thrombi could have been several wccks old at thc time of examination. This may have allowed time for the development of collateral flow around the occlusive lesion. However, since our criteria for normality on duplex scanning required both detection of flow and compressibility, detection of collateral flow in these patients with no symptoms would not have been sufficient to allow them to be classified as normal. Additionally, since duplex scanning did not visualize three of five asymptomatic ocdusions and two of the six symptomatic occlusions, neither the age of the thrombus nor the absence of symptoms of venous occlusion appear to be major determinants of the sensitivity of this diagnostic modality. Acoustic shadowing from the davicle results in nonvisualization of a short segment of the subclavian vein34 This may have contributed to our inability to visualize short segmental occlusions, such as the one in Fig. 3. However, all five occlusions of the right subclavian vein were short

6 396 Haire et al. Journal of VASCULAR SURGERY Fig. 3. A, Venogram of a left subclavian vein thrombosis not visualized by either MRI or duplex scanning. Note the short segmental occlusion of the subclavian vein (arrow) and the patency of the innominate vein. B, MRI scan of the same vein with flow signal suggests patency along its entire length. proximal segmental occlusions, and they were readily seen with duplex scanning. Also, the clavicle would not be expected to interfere with subclavian veto imaging by MR/. Magnetic resonance imaging scan- ning identified four of the five totally occlusive thrombi scanned. It is interesting to note that the one totally occlusive thrombus missed by MRI was a short segmental occlusion of the left subclavian

7 Volume 13 Number 3 March 199i A4agnetic resonance imaging and duplex scanning in subclavian vein thrombosis 397 vein. The reasons for the difficulty in imaging the occlusions in the proximal portion of the left subclavian vein are not clear and deserve further study with larger numbers of patients and with potentially more sensitive techniques such as newer duplex scanners with better depth penetration and resolution, color-flow Doppler, and low flip angle cine MRI techniques. However, in a recent report of the use of color-flow Doppler scanning in this setting, the only occlusions not visualized wcre in the proximal portions of the veins, 14 suggesting that this may be an anatomic location that is not reliably visualized by current ultrasound technology. In summary, both duplex ultrasonography and MRI are very specific in their ability to image catheter-induced subclavian vein thrombosis. Neither modality gave false-positive rcsults. In clinical ~,ractice this would allow patients with suspected subclavian vein thrombosis to be screened with one of these modalities. If the result of either duplex scanning or MRI suggested the presence of thrombosis, therapy could proceed without need for contrast medium phlebography. However~ since these noninvasive studies have failed to detect a significant number of thrombi, a normal study does not reliably exclude the possibility of thrombosis. In cases where the probability of subclavian vein thrombosis is high but both MRI and duplex scans are normal, phlebography is necessary to rule out a thrombosis. Further study in this area is warranted in an effort to enhance the sensitivity of the noninvasive diagnosis of subclavian vein thrombosis. REFERENCES 1. Porter JM, Rutherford RB, Clagett GP~ et al. Reporting standards in venous disease. J VAsc SURG 1988;8: Rollins DL, Semrow CM, FreideH ML, Catligaro KD, Buchbinder D. Progress in the diagnosis of deep venous thrombosis: the efficacy of real-time B-mode ultrasonic imaging. J VASC SURG 1988;7: Lensing AWA, Prandoni P, Brandies D, et al. Detection of deep vein thrombosis by real-time B-mode ultrasonography. N Engl J Med 1989;320: Sullivan ED, Peter DJ, Cranley li. Real-time B-mode venous ultrasound. J VASC SUF, G 1984;1:465-7i. 5. Giatini D, Kaftori JK, Pery M, Engel A. High-resolution realtime ultrasonography: diagnosis and follow-up of jugular and subclavian vein thrombosis. J Ultrasound Med I988;7: 62I Hubsch PJS, Stiglbauer RL, Schwaighofer BWAM, Kainberger FM, Barton PEA. Internal jugular and subclavian vein thrombosis caused by central venous catheters: evaluation using Doppler blood flow imaging. J Ultrasound Med 1988;7: Falk RL, Smith DF. Thrombosis of upper extremity thoracic inlet veins: Diagnosis with duplex doppler sonography. AJR i987;149: Spritzer CE, Sussman SK, Blinder RA, Saeed M, Herfkens RI. Deep venous thrombosis evaluation with limited-flipangle, gradient-refocused MR imaging: preliminary experience. Radiology 1988;166: Le W HM, Newhouse IH. MR imaging of portal vein thrombosis. AJR;151: Mintz MC, Lex T DW, Axel L, et al. Puerperal ovarian vein thrombosis: MR diagnosis. AJR 1987;149: li. White RD, Higgins CB. Magnetic resonance imaging of thoracic vascular disease. J Thorac Imag 1989;4: McMurdo KK, de Geer G, Webb WR, Gamsu G. Normal and occluded mediastinal veins: MR imaging. Radiology 1986;159: Borris LC, Christiansen HiM, Lassen MR, Olsen AD, Schott P. Comparison of real-time B-mode ultrasonography and bilateral ascending venography for detection of postoperative deep vein thrombosis following elective hip surgery. Thromb Hemostas 1989;6i:363-5.!4. Krmdson GJ, Weidmeyer DA, Erickson SJ, et al. Color Doppler s0nograp_hic imaging in the assessment of upper-extremity deep venous thrombosis. AIR 1990;154: