Value of lower extremity venous examination in the diagnosis of pulmonary embolism duplex Lois A. Killewich, MD, PhD, Janice D. Nunnelee, RNC, BSN, and Arthur I. Auer, MD, Baltimore, Md,, and St. Louis, Mo. Purpose: This study tests the hypothesis that the absence of deep venous thrombosis (DVT) on lower extremity color-flow venous duplex examination (LECFD) combined with a non-high-probability ventilation/perfusion (V/Q) scanning result rules out pulmonary embolus (PE). The use of LECFD as a diagnostic aid for PE is based on data that show that 90% of PE originate from lower extremity DVT, and therefore on the assumption that PE cannot be present if DVT is not present as the source. Methods: Over a 3-year period 51 patients with clinically suspected PE underwent LECFD and pulmonary angiography (PA) within 72 hours of each other. Forty-one patients also underwent V/Q scanning during the same time period. The results of LECFD and V/Q scans were compared with the results of PA, the gold standard for the diagnosis of pulmonary emboli. Results: Results of LECFD were positive for DVT in seven of 16 cases of angiographically documented PE. Thus the "sensitivity" of LECFD as a diagnostic aid in cases of suspected PE is 44%. The results of LECFD and V/Q scans were combined and compared with the results of PA. A high-probability V/Q scanning result was considered positive for PE. Intermediate- and low-probability scanning results were considered negative, because in the literature the decision to begin heparin therapy is not made on the basis of this result. The combination of test results was considered positive if either test result was positive and negative only if results of both were negative. With these criteria the combination of test results was positive in only 62% of cases of angiographically documented PE. Conclusions: If treatment of suspected PE were based on LECFD alone or on duplex combined with V/Q scanning, 40% to 50% of patients with PE would remain untreated. In cases of suspected PE where these noninvasive tests do not confirm its presence, PA should be performed. (J VASC SURG 1993;17:934-9.) The diagnosis of pulmonary embolism remains problematic. Clinical assessment and simple laboratory tests such as arterial blood gas measurements are notoriously inaccurate) Pulmonary angiography is accurate in more than 90% of cases, but it is invasive and costly. To overcome these limitations ventilation/perfusion scanning was introduced as a noninvasive alternative. Although initially it was believed From the University of Maryland (Dr. Killewich), Baltimore, and St. John's Mercy Medical Center (Dr. Auer and Ms. Ntmnelee), St. Louis. Presented at the Sixteenth Annual Meeting of the Midwestern Vascular Surgical Society, Cleveland, Ohio, September 11-12, 1992. Reprint requests: Arthur I. Auer, MD, Vascular and General Surgery, Ltd., 621 S. New Ballas Rd., St. Louis, MO 63141. Copyright 1993 by The Society for Vascular Surgery and International Society for Cardiovascular Surgery, North American Chapter. 0741-5214/93/$1.00 +.10 24/6/44138 934 to be highly accurate, recent studies have demonstrated inadequacies. 2-5 The Prospective Investigation of Pulmonary Embolism Diagnosis (PIOPED) Study 2 showed that most patients with angiographically documented pulmonary emboli had abnormal ventilation/perfusion scan results. However, only a minority of the abnormal scans were high probabil-: ity, which correlated well with the presence of pulmonary emboli on pulmonary angiography. Thirty-three percent of patients with intermediateprobability scan results and 16% with lowprobability scan results also had angiographically documented pulmonary emboli. Although a patient with a high-probability scan result could reasonably be assumed to have a pulmonary embolus, an intermediate- or low-probability scan could not be relied on to Me out pulmonary embolus, particularly if clinical suspicion was high.
journal OF VASCULAR SURGERY Volume 17, Number 5 Killewich, Nunnelee, and Auer 935 Given the concerns about pulmonary angiography, some investigators have explored the possibility of using noninvasive testing for venous thrombosis in the lower extremities in cases of suspected pulmonary embolus, 68 In part this is based on the assumption that up to 90% of pulmonary emboli originate from lower extremity deep venous thrombosis. 9 In addition, venous duplex scanning as a diagnostic test for deep venous thrombosis is now widely available and highly accurate. 1 Others have combined ventilation/perfusion scanning with noninvasive testing for " :ep venous thrombosis.3-s In either case the assumption is that because most pulmonary emboli originate from lower extremity deep venous thrombosis, a negative lower extremity noninvasive test result, especially when combined with an intermediate- or low-probability ventilation/perfusion scan result, should rule out pulmonary embolus. This study demonstrates the fallacy of this approach. PATIENTS AND METHODS From January 1989 to February 1992, 51 patients at St. John's Mercy Medical Center in St. Louis underwent pulmonary angiography and lower extremity color-flow venous duplex examination within 72 hours of each other. Forty-one patients also underwent ventilation/perfusion scans in the same time period. In all cases the venous duplex examinar;q,.n and ventilation/perfusion scanning were performed before pulmonary angiography and were performed for the indication of suspected pulmonary embolism, either hypoxemia, shortness of breath, or deterioration of ventilator settings. In all cases the changes that led to the suspicion of pulmonary embolism occurred suddenly. In no cases were tests performed for signs or symptoms of deep venous thrombosis or swollen leg. A review of the charts suggested that in all cases the clinician had a relatively high suspicion of pulmonary embolus. The results of these tests were reviewed and constitute the basis for this report. There were 24 women and 27 men in the study. The average age was 69 years, range 39 to 87 years. Pulmonary angiography was performed by cannulating the femoral vein with the Seldinger technique and then advancing the catheter through the heart into the pulmonary arteries. Iodinated contrast material was injected while films were taken in the anteroposterior projection. Angiography results were considered positive if an embolus was identified as obstructing a vessel or if the outline of an embolus (filling defect) within a vessel was identified. Lower extremity color-flow venous duplex exam- ination was performed with an Acuson 128 colorflow scanner (Acuson, Inc., Mountain View, Calif.). The common femoral, superficial femoral, proximal deep femoral, popliteal, posterior tibial, and peroneal veins of both legs were imaged. All veins were examined in longitudinal direction to determine the presence of thrombi; pulsed-wave Doppler ultrasonography was used to determine the presence of spontaneous, phasic, and augmentable flow. Veins were also examined in transverse direction for compressibility with probe pressure. A scan result was considered positive for deep venous thrombosis in a venous segment if thrombus was visualized, if the vessel was not fully compressible in transverse view, or if spontaneous, phasic, and augmentable flow were absent. In a previous article n the sensitivity and specificity of conventional duplex scanning compared with contrast venography for the diagnosis of deep venous thrombosis in our noninvasive vascular laboratory were shown to be 100% and 78%, respectively. More recent unpublished data comparing color-flow duplex scanning with venography demonstrated both sensitivity and specificity rates greater than 90% (Auer AI. Unpublished data, June 1992). Ventilation/perfusion scanning was performed according to the technique described in the HOPED Study. 2 Ventilation scanning was performed with xenon 133. Films were taken in the posterior view with the patient erect if possible. Perfusion scanning was performed with technetium 99m macroaggregated albumin. Images were taken in the anterior, posterior, lateral, and both posterior oblique views. Criteria for low-, intermediate-, and high-probability scans were also the same as those used in the HOPED Study and are listed in Table I. RESULTS In Table II the results of lower extremity colorflow venous duplex examination performed in cases of suspected pulmonary embolism are compared with the results of pulmonary angiography, which is the gold standard for the diagnosis of pulmonary embolism. Sixteen patients had pulmonary emboli diagnosed by pulmonary angiography, and seven of these patients had deep venous thrombosis diagnosed by duplex scanning. The sensitivity of venous duplex scanning as a diagnostic tool for pulmonary embolus is therefore 44%. Thirty of the 35 patients with negative pulmonary angiography results had negative venous duplex scan results, whereas five patients had deep venous thrombi without associated pulmonary emboli. Forty-one of the 51 patients in this study also
936 Killewich, Nunnelee, and Auer JOURNAL OF VASCULAR SURGER~ May 1993 Table I. Criteria for classification of ventilation/perfusion scan results* High probabifity Two or more large segmental perfusion defects without corresponding ventilation or roentgenographic abnormalities or substantially larger than either matching ventilation or chest roentgenogram abnormalities Two or more moderate segmental perfusion defects without matching ventilation or chest roentgenogranl abnormalities and one large mismatched segmental defect Four or more moderate segmental perfusion defects without ventilation or chest roentgenogram abnormalities Intermediate probability Not falling into other categories Borderline high or borderline low Low probability Nonsegmental perfusion defects Single moderate mismatched segmental perfusion defect with normal chest roentgenogram Any perfusion defect with a substantially larger chest roentgenograrn abnormality Large or moderate segmental perfusion defects involving no more than 4 segments in 1 lung and no more than 3 segments in 1 lung region, with matching ventilation defects either equal to or larger in size and chest roentgenogram either normal or with abnormalities substantially smaller than perfusion defects More than 3 small segmental perfusion defects with a normal chest roentgenogram *Adapted from the PIOPED Study 2. Table II. Sensitivity and specificity of lower extremity venous duplex ultrasonography Lower extremity venous duplex (n) Pulmonary angiography Positive Negative Total Positive (n) 7 9 16 Negative (n) 5 30 35 Total (n) 12 39 51 Sensitivity is 44% (7 of 16), specificity is 86% (30 of 35), positive predictive value is 58% (7 of 12), and negative predictive value is 77% (30 of 39). underwent ventilation/perfusion scanning within the 72-hour time period. Five scans were categorized as high probability, 25 as intermediate probability, and 11 as low probability. Two (40%) of the highprobability scans, nine (36%) of the intermediateprobability scans, and two (18%) of the lowprobability scans showed angiographically documented pulmonary emboli. Table III shows the correlation between ventilation/perfusion scans and pulmonary angiography for the diagnosis of pulmonary embolus. For this analysis intermediate- and low-probability scan results were considered negative for pulmonary emboli. The categorization of intermediate-probability scan results as negative was chosen, because they were categorized this way in previously published reports in which decision-making regarding suspected pulmonary embolism is based on noninvasive testing. 3s High-probability scan results were considered positive. The sensitivity (percentage of high-probability scan results) was 15%, and the specificity was 89%. In Table IV the results of lower extremity venous duplex and ventilation/perfusion scans were combined, and the combination of the results was compared with pulmonary angiography for the diagnosis of pulmonary emboli. For the purposes of this analysis the combination of results was cons i~: ered positive for pulmonary emboli if either test result was positive, and negative only if both results were negative. Intermediate- and low-probability scan results were considered negative. The sensitivity of this combination is 62%, the specificity 78%, positive predictive value 57%, and negative predictive value 81%. DISCUSSION In this study the correlation between a positive result for deep venous thrombosis on lower extremity color-flow venous duplex scanning and a positive result for pulmonary embolism on pulmonary angiography was low. Only 44% of patients wig" angiographically documented pulmonary emboli had a positive venous duplex scan result. If treatment of suspected pulmonary embolus were based on this test alone, more than 50% of patients would have remained untreated. Most patients without angiographically documented pulmonary emboli also had negative venous duplex examination results. Because this is a retrospective study, data regarding clinical suspicion for pulmonary embolus were not uniformly recorded at the time the original event occurred. We therefore do not believe we can
.gurnal OF VASCULAR SURGERY Volume 17, Number 5 Killewich, Nunnelee, and Auer 937 Table III. Sensitivity and specificity of ventilation/perfusion scanning results Ventilationperfusion scan (n) Pulmonary angiography Positive Negative Total Positive (n) 2 11 13 Negative (n) 3 25 28 Total (n) 3-'6 4--[ Sensitivity is 15% (2 of 13), specificity is 89% (25 of 28), positive predictive value is 40% (2 of 5), and negative predictive value is 69% (25 of 36). Table IV. Sensitivity and specificity of ventilation/perfusion scanning + lower extremity,~lous duplex Ventilationperfusion + duplex (n) Pulmonary angiography Positive Negative Total Positive (n) 8 5 13 Negative (n) 6 22 28 Total (n) 14 27 41 Sensitivity is 62% (8 of 13), specificity is 78% (22 of 28), positive predictive value is 57% (8 of 14)m and negative predictive value is 81% (22 of 27). categorize our findings on the basis of the clinician's original level of suspicion. What we can say, however, is that our review demonstrated that in all cases the suspicion for pulmonary embolism was based on an acute change in the patient's condition, either hypoxemia, shormess of breath, or deterioration of yentilator settings. No tests were done for suspected aeep venous thrombosis or swollen leg. We believe therefore that the clinical suspicion in all cases was at least relatively high. Although this is the first study in which pulmonary angiography has been compared with lower extremity color-flow venous duplex examination, previous studies have demonstrated that the correlation between pulmonary emboli and noninvasive testing for lower extremity deep venous thrombosis is low. 6-8 Schiff et al. 7 showed that results of continuous-wave Doppler scanning were positive for deep venous thrombosis in 38% of patients with angiographically documented pulmonary emboli, :vhereas Barnes et al.6 found that only 28% of patients with pulmonary emboli demonstrated by ventilation/perfusion scanning had positive Doppler examination results. Smith and Iber 8 showed that only 25% of patients with intermediate-probability or indeterminate ventilation/perfusion scan results had positive venous duplex examination results. Although the actual percentages in these studies and ours vary, it seems clear that a large proportion of venous duplex examination results are negative for deep venous thrombosis when performed for symptoms of pulmonary embolus rather than for symp- toms of deep venous thrombosis. We believe the reason for the large number of negative study results is that thrombus that causes the pulmonary embolus originates in a location not assessed by lower extremity noninvasive testing, such as the upper extremity, iliac, or abdominal/retroperitoneal veins, or that the entire thrombus embolizes before scanning is begun. The results of ventilation/perfusion scanning compared with pulmonary angiography in this study are similar to previously published results. 2-s As in these studies, only a minority of test results indicated high probability. A significant proportion of abnormal, non-high-probability scanning results occurred in patients with angiographically documented pulmonary emboli. Clearly, pulmonary embolus should not be excluded in patients with non-highprobability scanning results, particularly if clinical suspicion is high. Sixty percent of patients (three of five) with positive ventilation/perfusion scanning results had negative pulmonary angiography results. Two of the three false-positive test results occurred in patients with previous pulmonary emboli. It has been shown that ventilation/perfusion scanning cannot distinguish acute from chronic pulmonary emboli. 2 If these results are discarded the sensitivity of ventilation/perfusion scanning is two of three cases, or 67%. Although this number is still lower than previously published figures, 2 the small number of cases in our study makes it somewhat unreliable. In general we categorize high-probability scanning results as posi-
JOURNAL OF VASCULAR SURGER~ 938 Killewich, Nunnelee, and Auer May 1993 five unless the patient has had a previous pulmonary embolus. In cases of abnormal, non-high-probability ventilation scanning results, the addition of a lower extremity color-flow venous duplex examination did not rule out all pulmonary emboli. The sensitivity of this combination was 62%. If treatment of suspected pulmonary embolus were based on the strategy of a ventilation/perfusion scan and a lower extremity venous duplex scan, approximately 40% of patients with angiographically documented pulmonary emboli would remain untreated. Hull et al. ~ have advocated treatment of pulmonary emboli based on a ventilation/perfusion scan followed by serial noninvasive tests for lower extremity venous thrombosis. In their study, treatment was withheld safely if the ventilation/perfusion scan was not high probability and if impedance plethysmography results remained normal for 2 weeks. Because our study included the results of only a single lower extremity noninvasive examination, the results are not fully comparable. Nevertheless, treatment should clearly not be based on the results of one ventilation/perfusion scan and one venous duplex scan. Moreover, given the low complication rate for pulmonary angiography, it is our belief that proceeding directly to this test is more appropriate and certainly cheaper than repeating noninvasive tests over a 2-week period. In conclusion, our strategy in cases of suspected pulmonary embolism is to obtain a ventilation/perfusion scan first. If the result is high probability, the patient has not had previous emboii, and clinical probability is high, then we conclude that an acute embolus is present and treat appropriately. If the result of the scan is normal, we accept this as conclusive evidence of the absence ofemboli. In cases of abnormal, non-high-probability scans, previous emboli, or low clinical suspicion, we obtain pulmonary angiograms and base treatment on the results. We obtain lower extremity color-flow venous duplex scans on all patients to determine whether a deep venous thrombus is present, and we base treatment of thrombi on these test results, We do not use the results as a basis of treatment for suspected pulmo- nary embolism. Although heparin is used in treatment of both disorders, we believe it is important to know whether the patient has had a pulmonary embolic event, because this may assist in directing other aspects of treatment, such as placement of a caval filter. We strongly encourage clinicians to be liberal in their use of pulmonary angiography, because this remains the definitive test for the diagnosis of pulmonary embolism. REFERENCES 1. Stein PD, Alavi A, Gottschalk A, et al. Usefulness of noninvasive diagnostic tools for diagnosis of acute pttlmon~i~ embolism in patients with a normal chest radiograph. Am J Cardiol 1991;67:1117-20. 2. The PIOPED Investigators. Value of the ventilation/perfusion scan in acute pulmonary embolism: results of the Prospective Investigation of Pulmonary Embolism Diagnosis (PIOPED) Study. JAMA 1990;263:2753-9. 3. Hull R_D, Hirsh J, Carter CJ, et al. Pulmonary angiography, ventilation lung scanning, and venography for clinically suspected pulmonary embolism with abnormal perfusion lung scan. Ann Intern Med 1983;98:891-9. 4. Hull RD, Hirsh J, Carter CJ, et al Diagnostic value of ventitation/perfusion lung scanning in patients with suspected pulmonary embolism. Chest 1985;88:819-28. 5. Hull RD, Raskob GE, Coates G, Panju AA, Gill GJ. A new non-invasive management strategy for patients with suspected pulmonary embolism. Arch Intern Med 1989;149:2549-55. 6. Barnes RW, Kinkead LR, Wu KK, Hoak JC. Venous thrombosis in suspected pulmonary embolism: incidence detectable by Doppler ultrasound. Thromb Haemost 1976; 36:150-6, 7. Schiff MJ, Feinberg AW, Naidich ]'B. Noninvasive ven~s examinations as a screening test for pulmonary embolism. Arch Intern Med 1987;147:505-7. 8. Smith LL, Iber C. Venous duplex ultrasound in the diagnosis of pulmonary embolism. J Vasc Technol 1989;2:227-30. 9. McLachlin J, Paterson JC. Some basic observations on venous thrombosis and pulmonary embolism. Surg Gynecol Obstet 1951;93:1-9. 10. Killewich LA, Bedford GR, Beach KW, Strandness DE Jr. Diagnosis of deep venous thrombosis: a prospective study comparing duplex scanning to contrast venography. Circulation 1989;79:810-4. 11. Langsfeld M, Hershey FB, Thorpe L, et al. Duplex B-mode imaging for the diagnosis of deep venous thrombosis. Arch Surg 1987;122:587-91. Submitted Sept. 14, 1992; accepted Nov. 6, 1992. DISCUSSION Dr, Fred N. Littooy (Maywood, Ill.). This study examines both color-flow duplex lower extremity venous examinations and ventilation perfusion scans and analyzes their role singly and in combination in making the difficult diagnosis of pulmonary embolism that we as clinicians often must tackle. I should point out that the authors used a vigorous venous color-flow duplex examination and analyzed their ventilation/perfusion lung scans according to
:/OURNAL OF VASCULAR SURGERY Volume 17, Number 5 IGllewich, Nunnelee, and Auer 939 the same technique described in the PIOPED Study, which stands as the landmark prospective study, to date, comparing ventilation/perfusion scans with pulmonary angiography in the diagnosis of pulmonary embolism. This allows for comparisons but also adds the information obtained from lower extremity noninvasive examination of these patients in whom, at least historically, 90% would have a pulmonary embolus that arose from the deep veins of the pelvis or legs. From the PIOPED Study an abnormal ventilation/perfusion scan result had a sensitivity of 98% but only a 10% specificity. However, the positive predictive value of a high probability scan was 88% in that study. ~rom other studies, only about 10% of patients who die of pulmonary embolism have clinical manifestations of deep venous thrombosis. On the other hand, venography or venous duplex examinations of the lower extremities have positive results in 50% to 90% of patients in whom the diagnosis of pulmonary embolism is made. A major problem arises with the intermediate probability or low probability ventilation/perfusion scan in which the result of pulmonary angiography is negative 67% and 84% of the time, respectively. The authors studied the concept that venous duplex examinations of the lower extremities would aid in the diagnosis of pulmonary embolism. They therefore have retrospectively studied 51 patients who underwent pulmonary angiography and color-flow duplex venous examinations. Forty-one patients also underwent ventilation/perfusion scans before arteriography was performed. They then examined the diagnostic accuracy of each noninvasive examination singly and in combination when compared with pulmonary angiography in the iagnosis of pulmonary embolism. In their study the result of the lower extremity venous duplex examination was not very sensitive at 44% and had only a 58% positive predictive value. Ventilation/perfusion scans when used alone and when only considering high-probability scans had only a 15% sensitivity and a positive predictive value of 40%. Combining the two tests increased the sensitivity to 62% and the positive predictive value to 57%. These findings are in conflict with other studies. High-probability ventilation/perfusion scans have a much higher sensitivity and positive predictive value in the PIOPED Study. Venous duplex scanning also has a better positive predictive value in other studies. Can the authors comment on these inconsistencies with their study? Were the pulmonary angiograms obtained within 24 to 72 hours of the onset of symptoms? Do you have a better characterization of your patient population, that is, was there a reason to suspect subclavian, superior vena cava, right atrial, or renal vein origins of these dots? Were hidden areas such as the pelvic veins a likely source in any of your patients? On the basis of your study, what do you now recommend as the optimal approach to the diagnosis of pulmonary embolism? Dr. Lois A, Killewich. Thank you very much. First, concerning the PIOPED Study, their reported sensitivity of 98% was a sensitivity for all of the abnormal categories of scans combined, and in our study sensitivity was reported for only high-probability scans; this is one difference between the studies. I am not sure what studies Dr. Littooy is referring to, but of the studies that I reviewed in the literature, there were basically two categories of studies. There were studies where some type of nonlnvasive test for lower extremity deep venous thrombosis was compared with some type of test for pulmonary embolus. In those studies the sensitivity or the percentage of studies of positive lower extremity noninvasive testing for deep venous thrombosis was in the range of about 50%. Some of those tests compared hand-held continuous-wave Doppler scanning with ventilation/perfusion scanning, and some of them compared duplex scanning. The studies are variable, but I think that the numbers we have are not that far off numbers in those studies. Now there is another group of studies that has been done primarily by a group in Canada who combined ventilation/perfusion scanning with serial lower extremity noninvasive testing, and their preferred test is impedance plethysmography. When they perform serial testing they find that they can exclude pulmonary embolism if the results of serial impedance plethysmography, which is performed over a 2-week period, continue to remain negative and the result of the ventilation/perfusion scan is not high probability. But again, they are using a series of noninvasive tests instead of a single test. The accuracy of the laboratory at St. John's in terms of its diagnosis of deep venous thrombosis is very high, probably at least 95%, but I do not have the exact figure. In answer to Dr. Littooy's question about why I think these test results are negative, I think that a lot of these pulmonary emboli are originating from sources that are not examined with lower extremity venous duplex scanning. I think that some come from the upper extremities. I think some come from the iliac or the pelvic veins - places that we do not examine. I think that, if you examined those locations with noninvasive testing, you might pick up more information, and it would be valid then to use the noninvasive test. You would, however, need to examine those other locations. Pulmonary angiography was performed within 72 hours of the development of symptoms. Regarding my recommendations for what I would do, I would obtain the noninvasive tests first. I would perform the lower extremity venous duplex examination to determine whether the patient had a deep venous thrombosis, I would still want to perform nonlnvasive testing to determine whether the patient had a pulmonary embolus. I would want to know that, but if both those test results were negative, then I would perform angiography.