Diagnosis of Deep Venous Thrombosis

Transcription

1 810 Diagnosis of Deep Venous Thrombosis A Prospective Study Comparing Duplex Scanning to Contrast Venography Lois A. Killewich, MD, PhD, Geri R. Bedford, BS, Kirk W. Beach, MD, PhD, and D.E. Strandness Jr., MD Duplex scanning has been proposed as a safe alternative to contrast venography for diagnosing deep venous thrombosis, but its accuracy has not been proved. In this prospective, double-blind study of 47 patients, the sensitivity and specificity of duplex scan criteria were determined relative to contrast venography for lower extremity deep venous thrombosis. Criteria considered to show the presence of deep venous thrombosis included visualization of thrombus (T), absence of spontaneous flow by Doppler ultrasonography (F), absence of phasicity of flow with respiration (P), and incompressibility of the vein with probe pressure (VC). When analyzed individually, the variables T and F had low sensitivities (50%o and 76%) but high specificities (92% and 100%). VC had low values for both (79% and 67%, respectively). The best single variable was P (sensitivity and specificity=92%). The best combinations of variables were T+P (sensitivity=95%, specificity= 83%), T+ F+P (sensitivity= 95%, specificity=83%), F+P (sensitivity and specificity=92%), and F+T (sensitivity=92%, specificity=87%o). The low specificity of vein incompressibility was secondary to cases in which normal veins were difficult to compress in the thigh. All false-negative cases were from isolated calf vein thrombi. We conclude that isolated criteria from duplex scanning should not be used to diagnose deep venous thrombosis. In cases of suspected calf vein thrombosis, repeat duplex examination should be obtained in 3-4 days to determine the most appropriate therapy. In equivocal cases of proximal vein thrombosis, a contrast venogram should be obtained. (Circulation 1989;79: ) Acute deep venous thrombosis is the most common disease of the vascular system. As many as 800,000 new cases occur annually. Of these, up to one third of the patients experience a pulmonary embolism, many of which are fatal.' Moreover, up to 50% of the patients eventually develop the postthrombotic syndrome.2 The costs to society in terms of lost productivity and personal suffering are enormous. Both invasive and noninvasive methods have been used in diagnosing deep venous thrombosis. The contrast venogram is the standard to which the various noninvasive methods have been compared.3'4 Although highly accurate in diagnosing calf vein thrombi. the contrast venogram may miss more proximal iliac vein thrombi because of dilu- From the Department of Surgery and Section of Vascular Surgery, University of Washington School of Medicine, Seattle, Washington. Supported by National Institutes of Health Grant HL Address for correspondence: D.E. Strandness Jr., MD, Department of Surgery, RF-25, University of Washington School of Medicine, Seattle, WA Received May 17, 1988; revision accepted November 22, tion of the dye. In addition, this invasive technique can cause unsafe side effects, for example, local irritation at the site of cannulation of the vein, thrombosis of leg veins, and allergic reactions from the contrast material. Noninvasive methods include continuous wave Doppler ultrasonography,5 impedance plethysmography,6-8 fibrinogen labeled with 1251,8 and duplex scanning. Continuous wave Doppler and plethysmography are accurate in diagnosing proximal deep venous thrombosis but miss nonocclusive thrombi and those confined to the calf. Fibrinogen labeled with 125j is highly accurate in the calf but is insensitive for proximal deep venous thrombosis. In addition, it requires a waiting period of hours after injection before a diagnosis can be made, and it cannot be used in pregnant or lactating women. More recently, investigators have adapted realtime B-mode ultrasonography and duplex scanning, which combines B-mode ultrasonography with pulsed Doppler, to the diagnosis of deep venous thrombosis.9 '4 In our laboratory, we have conducted a prospective, double-blind study to compare duplex scanning with contrast venography. In particular, we have determined which of the many

2 Killewich et al Diagnosis of Deep Venous Thrombosis 811 criteria obtainable by duplex scanning correlate best with the findings of contrast venography. We report our results with a group of 47 patients. Methods The patients referred for study came from three sources in the University of Washington School of Medicine system-the University Hospital (186 patients), the Veterans Hospital (six patients), and Harborview Medical Center (nine patients). The patients from the Veterans Hospital and Harborview Medical Center were brought to our attention when scheduled for venography. For the total population studied at the University Hospital, the patients were classified by nature of the findingsnormal (129 patients), positive (52 patients), and equivocal (five patients). An equivocal scan was defined as indeterminate for either normalcy or the presence of disease. The decision for venography was made by the referring service and not by us, although we requested the study to confirm our results. A total of 153 patients did not undergo venography, leaving 47 patients who had a total of 50 venograms performed. There were 33 men and 14 women. The youngest patient was 21 years of age, and the oldest was 83 years of age, the mean age was 51 years. All patients gave informed consent for participation in the study. One of the patients had bilateral deep venous obstruction and thus had both legs studied by venography. One patient was treated with urokinase. After treatment, another duplex study and venogram were done to estimate the degree of lysis. No lysis had occurred, and this was confirmed by the duplex and venographic studies. The second patient, a quadriplegic with pulmonary emboli, had a negative duplex scan and venogram. Because of repeated pulmonary emboli, repeat duplex scans were done, which confirmed the previous negative findings. A second venogram revealed the presence of calf vein thrombi. Risk factors for the development of acute deep venous thrombosis included the following: previous deep venous thrombosis (17 patients), bed rest for greater than 1 week (17 patients), leg trauma in the previous month (18 patients), surgery in the previous month (15 patients), the presence of neoplasm (11 patients), a family history of superficial varicosities (eight patients), a history of superficial varicosities in the patient (seven patients), a previous pulmonary embolus (six patients), a family history of deep venous thrombosis (six patients), a history of congestive heart failure (four patients), a history of prolonged travel with sitting for greater than six hours in the previous month (four patients), previous stasis ulcers (two patients), pregnancy (one patient), and oral contraceptive use (one patient). Venous duplex examination was performed with the Advanced Technical Laboratories Ultramark 8 (Bothell, Washington), with the patient at in the reverse Trendelenburg position. The iliac veins were imaged with a 3.0- or 5.0-MHz transducer. The common femoral, superficial femoral, and popliteal veins were imaged with a 7.5 -MHz transducer. The superficial femoral vein was examined at three locations in the thigh, distal to the bifurcation of the common femoral vein, in the midthigh region, and at the entrance to the adductor canal. Suspected calf vein thrombosis was evaluated by imaging the posterior tibial veins in the lower leg with the 10.0-MHz transducer. Pulsed Doppler venous signals were obtained with a 5.0-MHz transducer. No attempts were made to image either the anterior tibial veins or the peroneal veins routinely. All veins were examined in the longitudinal section for the presence of thrombi and with the pulsed Doppler for the presence of spontaneous flow, phasicity of flow with respiration, augmentation of flow with distal compression, augmentation of flow with release of proximal compression or release of Valsalva, and reversal of flow (or reflux) with proximal compression or Valsalva. The veins were then examined in the transverse section for compressibility with probe pressure. The duplex scan was considered positive for deep venous thrombosis in a particular venous segment when thrombus was visualized, when the vein was not completely compressible, when absence of flow by Doppler was present, or when absence of phasicity of flow with respiration was present. There were two exceptions to these criteria. First, we did not consider the isolated finding of incompressibility of the inferior vena cava, common iliac, or superficial femoral vein as it enters Hunter's canal as evidence of deep venous thrombosis. In some patients, the course of these veins are too deep to be fully compressible under normal conditions. Second, absence of spontaneous flow in the posterior tibial veins at the ankle was not considered as indicative of deep venous thrombosis because certain environmental conditions may limit spontaneous flow here. Rather, in these segmnents we defined positive Doppler findings as the absence of flow after distal augmentation. When the examination was completed, the referring service was informed of the results along with a recommendation for a confirming venogram. The referring service did not always follow our recommendations based upon the major medical problem for which the patient was being treated, the potential risks associated with the venogram, or the discomfort and problems associated with the dye load. We would recommend a follow-up duplex examination to reassess the deep veins. Contrast venography was performed by the method of Rabinov and Paulin.4 All venograms and duplex examinations were performed within 1 week of each other, and the results of one were not known when the other was performed. Results The four variables we evaluated as diagnosing the presence of acute deep venous thrombosis were

3 812 Circulation Vol 79, No 4, April 1989 TABLE 1. Sensitivity and Specificity ofvisualization of Thrombus Visualization of thrombus (n) Positive Negative Total Sensitivity is 50% (19 of 38) (95% confidence limits are 34% and 66%); specificity is 92% (11 of 12) (95% confidence limits are 62% and 98%); positive predictive value is 95% (19 of 20) (95% confidence limits are 69 and 100); negative predictive value is 37% (11 of 30) (95% confidence limits are 14 and 59). visualization of thrombi, absence of spontaneous flow by Doppler, absence of phasicity of flow with respiration by Doppler, and incompressibility of the vein in the transverse section with probe pressure. Tables 1 through 4 show the sensitivity, specificity, and 95% confidence limits obtained when each variable is compared with the results of contrast venography. Visualization of thrombus had a sensitivity of 53%, a positive predictive value of 95%, and a specificity of 92% with a negative predictive value of 37%. In the 19 cases where we were unable to visualize a thrombus, no flow was detected by Doppler, and venography confirmed the presence of the thrombus. Absence of spontaneous flow by Doppler had a sensitivity of 76%, a positive predictive value of 100%, and a specificity of 100% with a negative predictive value of 57%. The lower sensitivity and negative predictive value represented cases of nonocclusive thrombi in which flow was still detectable. Absence of phasicity of flow with respiration had the best sensitivity and specificity for the individual variables, 92% for each. The positive and negative predictive values were 95%. Incompressibility of the vein with probe pressure was poor in both areas, with a sensitivity of only 79% and specificity of 67%. The positive predictive value was 88%; the negative predictive value was 50%. Table 5 shows the sensitivities and specificities of variable combinations. In this analysis, a duplex scan was considered positive for deep venous thrombosis when any of the variables were positive, and it was considered negative only when all variables were negative. T represents visualization of throm- TABLE 2. Flow Sensitivity and Specificity of Absence of Spontaneous Absence of spontaneous flow (n) Positive Negative Total Sensitivity is 76% (29 of 38) (95% confidence limits are 63% and 90%); specificity is 100% (12 of 12) (95% confidence limits are 88% and 100%); positive predictive value is 100% (29 of 29) (95% confidence limits are 85 and 100); negative predictive value is 57% (12 of 21) (95% confidence limits are 29 and 85). TABLE 3. Sensitivity and Specificity of Absence of Phasicity of Flow With Respiration Absence of phasicity of flow with respiration (n) Positive Negative Total Sensitivity is 92% (35 of 38) (95% confidence limits are 79% and 97%); specificity is 92% (11 of 12) (95% confidence limits are 62% and 98%); positive predictive value is 97% (35 of 36) (95% confidence limits are 81 and 99); negative predictive value is 79% (11 of 14) (95% confidence limits are 41 and 92). bus, F represents absence of spontaneous flow, P represents absence of phasicity of flow with respiration, and VC represents vein The highest sensitivities were obtained with T + P, T+F+P, and T + F + P + VC. The combination of four variables, however, had a low specificity. The highest specificities were obtained with F + T and F+ P, which also had the best overall combination of sensitivity and specificity. Any combination including vein incompressibility had a low specificity. These data suggest that the Doppler variables, absence of spontaneous flow and absence of phasicity of flow with respiration, are the most reliable and accurate in diagnosing deep venous thrombosis. Many thrombi will not be visualized by duplex scanning, and vein incompressibility will overdiagnose a significant number of cases. (Negative predictive value is 50%.) Table 6 shows the locations of the false-negative cases by duplex scanning. Regardless of the variable tested, most of these were in the calf, extending into the knee and distal portion of the adductor canal. Interrogation of only the distal posterior tibial veins clearly misses a significant number of isolated calf vein thrombi. Table 7 shows the locations of the false-positive cases by duplex scanning. Eighty-three percent (five of six) of the cases were in the thigh, and 80% of those were due to overdiagnosis by the vein incompressibility variable. This variable seems unreliable in evaluating veins in this region of the lower extremity. TABLE 4. Sensitivity and Specificity of Vein Incompressibility Vein incompressibility (n) Positive Negative Total Sensitivity is 79% (30 of 34) (95% confidence limits are 66% and 92%); specificity is 67% (8 of 12) (95% confidence limits are 40% and 93%); positive predictive value is 88% (30 of 34) (95% confidence limits are 67 and 95); negative predictive value is 50% (8 of 16) (95% confidence limits are 18 and 82).

4 Killewich et al Diagnosis of Deep Venous Thrombosis 813 TABLE 5. Sensitivity and Specificity of Variable Combinations Variables* Sensitivity (%) Specificity (%) T+P 95 (82,98) 83 (52,95) T+F+P 95 (82,98) 83 (52,95) T+F+P+VC 95 (82,98) 58 (30,86) F+P 92 (79,97) 92 (62,98) F+T 87 (76,98) 92 (62,98) T+F+VC 87 (76,98) 67 (40,93) F+VC 84 (73,96) 67 (40,93) T+VC 82 (69,94) 67 (40,93) Values in parentheses are 95% confidence limits. Discussion As duplex scanning has gained in precision and popularity, it has been adapted for the purpose of diagnosing deep venous thrombosis. The technique is safer than invasive techniques, and it provides a diagnosis in a more timely and efficient manner than most noninvasive techniques. In our study, we compared duplex scanning and contrast venography in a prospective, double-blind fashion in diagnosing deep venous thrombosis. In general, we found a sensitivity and specificity for duplex scanning relative to contrast venography of 85-95%. We believe that duplex scanning can be used to diagnose deep venous thrombosis in most cases. However, we caution against the nonselective application of duplex scanning to all cases of suspected deep venous thrombosis without a confirmatory venogram. First, we believe that the use of isolated criteria (visualization of thrombus and vein incompressibility), as advocated by some groups,9'12 will lead to a significant number of false-positive and false-negative tests. Second, isolated calf vein thrombi may be missed by duplex scanning, although the clinical significance of this can be questioned. Sullivan et a19 claim that deep venous thrombosis can be diagnosed solely on the finding of thrombus on the B-mode ultrasonogram and that acute deep venous thrombosis can be differentiated from chronic TABLE 6. Location of False-Negative Cases by Duplex Scanning Calf-distal Adductor Cases adductor canal canal-pelvis Total leg Variables* (n) (n) (n) (n) T (79) 2 (11) 2 (11) F 9 6 (67) 2 (22) 1 (11) VC 8 8 (100) 0 (0) 0 (0) P 3 2 (67) 1 (33) 0 (0) T+F+P 2 2 (100) 0 (0) 0 (0) T+F+P+VC 2 2 (100) 0 (0) 0 (0) Values in parentheses are percentages of number of cases. TABLE 7. Location of False-Positive Cases by Duplex Scanning Calf-distal Adductor Cases adductor canal canal-pelvis Variables* (n) (n) (n) T 1 0 (0) 1 (100) F 0 0 (0) 0 (0) P 1 1 (100) 0 (0) VC 4 0 (0) 4 (100) Values in parentheses are percentages of cases. changes. In our study, only 50% of the cases of venogram-documented deep venous thrombosis had visualizable thrombus. Furthermore, we have found that if we follow patients with deep venous thrombosis over time, the thrombi become anechoic in 90% of cases by 1-2 weeks after the initial presentation. In these cases, we are confident that the vessel remains occluded, because we are not able to detect flow with Doppler. Other studies have shown that the echogenicity of in vitro clot changes with time.15,16 These studies have shown that clot is echogenic for the first hours after formation but then becomes progressively anechoic owing to the breakdown of red blood cells and their replacement in the clot by fibrin. Although we cannot fully explain the discrepancy between our findings and those of Sullivan et al,9 it seems apparent that, under the conditions we used to perform venous duplex examination, thrombus is not uniformly echogenic. We caution against its use as an isolated criterion. Investigators9-11,14 have placed heavy emphasis on the use of incompressibility of the vein in transverse section with probe pressure to diagnose deep venous thrombosis. At least one group12 has advocated this as the sole criterion necessary for diagnosis. Others have suggested that imaging without the concomitant use of Doppler is sufficient for diagnosis. We caution against this as well. The sensitivity of vein incompressibility is known to be low in the abdomen and pelvis, where it is often not possible to compress veins because of overlying structures or bowel gas. In addition, because of the depth of the vein's course, it is often not possible to compress the superficial femoral vein where it enters the adductor canal in normal subjects. In our study, we also found a very low specificity of vein incompressibility in the main portion of the thigh. Our inability to compress normal veins did not correlate with previous events in this region, such as a deep venous thrombosis or surgical procedures, which may render veins incompressible in the absence of acute deep venous thrombosis. We therefore conclude that vein incompressibility should be used with caution. In our study, the criteria obtainable by Doppler, the absence of spontaneous flow, and the absence

5 814 Circulation Vol 79, No 4, April 1989 of phasicity of flow with respiration remained the most reliable criteria for diagnosing deep venous thrombosis. The most effective use of the duplex scanner may be to use ultrasonography to locate the vein and then the Doppler to determine the presence of thrombosis. There were a significant number of cases of deep venous thrombosis of the calf in our study that were not detected. Our technique for assessing calf vein thrombosis involved interrogating the posterior tibial veins at the ankle. Interrogation of the six posterior tibial, anterior tibial, and peroneal veins in the calf is a difficult and time-consuming task. The peroneal veins in particular may be very difficult to identify.'1 Moreover, thrombi in other areas of the calf such as the soleal sinuses are invariably missed. Although duplex scanning clearly misses some isolated calf vein thrombi, whether isolated calf vein thrombi are clinically significant is not clear. Most investigators agree that these do not result in fatal pulmonary emboli.17 They become significant in the 20% of cases in which propagation into the popliteal vein occurs because these thrombi are associated with an increased risk of pulmonary emboli.18 The ultimate utility of any diagnostic method depends upon its impact on therapy and outcome. At worst, these translate into the penalty the patient pays for false-positive and false-negative tests. A false-positive test would in most instances result in the patient being treated by anticoagulation. The duration of treatment would depend upon the location of the possible thrombus. If the error were for the calf veins, opinion varies regarding the type of therapy and, if given, its duration. However, if the error involved the more proximal veins, the patient would be treated for 3-6 months. The false-negative test is more serious, because the patient would not be treated for a potentially lethal disorder. This is particularly true for thrombi in the proximal leg veins. For undetected calf vein thrombi, the outcome would not be as serious, except in those cases where propagation to the popliteal vein occurred. Because of these concerns, we have adopted the following policy regarding the use of duplex scanning in the diagnosis of deep venous thrombosis: 1) Equivocal tests for involvement of the proximal deep veins must, in most cases, be followed by a contrast venogram. An equivocal test in our lab is defined as one that is indeterminate in arriving at the diagnosis. 2) In cases of suspected calf vein thrombi, repeat studies in 2-3 days should be obtained before a final decision is made regarding therapy. We realize this study may be faulted for the lack of attention to the anterior tibial and peroneal veins and to the more proximal deeper portions of the posterior tibial veins. We believe that assessment of these veins with conventional duplex scanning is difficult and time consuming. Moreover, the clinical significance of isolated thrombi in them is questionable; also, large and clinically significant calf vein thrombosis is detectable by changes in the posterior tibial veins. In our initial study, therefore, we have turned our attention only to the posterior tibial veins. We are currently evaluating other duplex scanning methods, including the use of color flow, to give a more complete assessment of calf vein thrombosis. References 1. Coon WW, Willis PW: Deep venous thrombosis and pulmonary embolism: Prediction, prevention and treatment. Am J Cardiol 1959;4: Strandness DE Jr, Langlois Y, Cramer M, Randlett A, Thiele BL: Long-term sequelae of acute venous thrombosis. JAAL 1983;250: Lea Thomas M: Phlebography. Arch Surg 1972;104: Rabinov K, Paulin S: Roentgen diagnosis of venous thrombosis in the leg. Arch Surg 1972;104: Strandness DE Jr, Sumner DS: Ultrasonic velocity detector in the diagnosis of thrombophlebitis. Arch Surg 1972; 104: Cranley JJ, Gay AY, Grass AM, Simeone FA: A plethysmographic technique for the diagnosis of deep vein thrombosis of the lower extremities. Surg Gynecol Obstet 1973; 136: Hull R, Van Aken WG, Hirsch J, Gallus AS, Hoicka G, Turpie AGG, Walker I, Gent M: Impedance plethysmography using the occlusive cuff technique in the diagnosis of venous thrombosis. Circulation 1976;53: Hull R, Hirsch J, Sackett DL, Taylor DW, Carter C, Turpie AGG, Zielinsky A, Powers P, Gent M: Replacement of venography in suspected venous thrombosis by impedance plethysmography and 125-I-fibrinogen leg scanning: A less invasive approach. Ann Int Med 1981;94: Sullivan ED, Peter DJ, Cranley JJ: Real-time B-mode venous ultrasound. J Vasc Surg 1984;1: Dauzat MM, Laroche JP, Charras C, Blin B, Domingo-Faye MM, Sainte-Luce P, Domergue A, Lopez FM, Janbon C: Real-time B-mode ultrasonography for better specificity in the noninvasive diagnosis of deep venous thrombosis. J Ultrasound Med 1986;5: Langsfeld M, Hershey FB, Thorpe L, Auer AI, Binnington HB, Hurley JJ, Woods JJ: Duplex B-mode imaging for the diagnosis of deep venous thrombosis. Arch Surg 1987; 122: Cronan JJ, Dorfman GS, Scola FH, Schepps B, Alexander J: Deep venous thrombosis: Ultrasound assessment using vein compression. Radiology 1987;162: Raghavendra BN, Horii SC, Hilton S, Subramanyam BR, Rosen RJ, Lam S: Deep venous thrombosis: Detection by probe compression ofveins. J Ultrasound Med 1986;5: Oliver MA: Duplex scanning in venous disease. Bruit 1985; 9: Coelho JCU, Sigel B, Ryva RC, Machi J, Renigers SA: B-mode ultrasonography of blood clots. J Clin Ultrasound 1982;10: Alanen A, Kormano M: Correlation of the echogenicity and structure of clotted blood. J Ultrasound Med 1985;4: Browse NL, Clemenson C, Croft DN: Fibrinogen detectable thrombosis in the legs and pulmonary embolism. Br Med J 1974;1: Kakkar VV, Howe CT, Flang C, Clarke MB: Natural history of post-operative deep vein thrombosis. Lancet 1969; 2: KEY WORDS * venous thrombosis * diagnosis * ultrasound * duplex scanning