Discrimination of Types of Venous Emboli Using Doppler Ultrasound
Article Outline
- Abstract
- Introduction
- Materials and Methods
- Results
- Discussion
- Conclusions
- Acknowledgment
- References
- Copyright
Background
Many studies have identified features of the emboli traveling in the arterial system by Doppler ultrasound, and estimated their composition and size for anticipation of cerebral infarct. Another concern is features of the emboli in the venous system for anticipation of pulmonary embolism (PE). The objective is to prove that the emboli in the venous system can be discriminated by Doppler ultrasound in animal study and to assess whether PE can be predicted by using this technique in clinical case.
Methods
Animal study—lard oil, thrombus, or bone marrow was injected to the femoral veins in pigs. High intensity transient signals were transhepatically identified at the inferior vena cava using Doppler ultrasound. Intensity volume (dB) and frequency shift (Hz) of each signal were obtained. The cutoff values at which thrombi, fat emboli, and marrow emboli could be discriminated were calculated by receiver-operating characteristic curves analysis.
Human study—Subjects were 47 patients who underwent total knee arthroplasty. On postoperative day 0 and 1, High intensity transient signals were identified at the affected common femoral vein using Doppler ultrasound. Contrast computed tomography was done, and the patients were divided into two groups according to the presence of PE. The two groups were distinguished by intensity volume (dB) and frequency shift (Hz) of each signal and calculated the sensitivity and specificity. The statistics analysis was done as in animal study.
Results
Animal study—Thrombus and bone marrow could be discriminated from lard oil well at frequency shift of 208 Hz and 196 Hz (with sensitivity of 86.9, 85.7%, and specificity of 82.3%, 91.8%, respectively).
Human study—The sensitivity and specificity were 76.7 and 75.0% to discriminate patients with and without PE at frequency shift of 208 Hz.
Conclusions
Our animal study results indicated that emboli could be identified and discriminated in the venous system. Clinical study informed that the frequency shift (208 Hz) of the Doppler signal was the best parameter to discriminate PE with and without PE in patients with total knee arthroplasty.
Introduction
Continuous wave Doppler monitoring is a noninvasive examination generally useful for estimation of the cerebral circulation and detection of microemboli,1 and is mainly used in cardiac surgery, carotid endarterectomy, and carotid artery stenting.2, 3
It is clinically very important to discriminate the emboli traveling in the vascular channel, and each embolus sends out various ultrasonic signals due to their ultrasonic characteristics. Relatively many studies have identified features of the emboli traveling in the arterial system by continuous wave Doppler monitoring, and estimated their composition and size.4 However, few studies on Doppler-detected emboli in the venous system have been reported. Only one study reports that the emboli can be identified in the venous system by Doppler ultrasound, and their size can be estimated using the intensity volume (dB) to some extent5; however to our knowledge, no other studies have analyzed and compared the features of emboli in the venous system using Doppler ultrasound. It is well known that deep vein thrombosis (DVT) in the lower leg can be associated with and be the cause of pulmonary embolism (PE) after total knee arthroplasty (TKA),6 and reports suggest that as many as two-third of patients undergoing TKA have PE, many of which are clinically silent.7 The composition of PE could be thrombus, marrow emboli, or fat emboli, which usually occur after TKA.8 Thrombus usually results in PE, whereas fat and marrow emboli can present as cerebral infarction (CI) and/or postoperative delirium.9 The examination method to predict PE or CI has not yet been established, and we can only evaluate the risks based on the clinical findings obtained using Wells score and Geneva score.10, 11 In this study, we confirm that it is possible to detect micro emboli in the venous system. Using an animal model, we characterize ultrasound characteristics of Doppler-detected High intensity transient signals (HITS) in the venous system from different sources (fat, marrow, and clot) and develop ultrasound parameters for HITS in the venous system derived from a fat source, a marrow source, and a clot source, and determine which ultrasound parameter is best able to distinguish between HITS derived from marrow, fat, and clots, and apply this information to a population of patients after TKA to diagnose PE.
Materials and Methods
Animal Study
Three neutered pigs were used (32, 31, and 35 kg). General anesthesia with 35% oxygen and 1.0% isoflurane was administered, and electrocardiogram and transcutaneous oxygen saturation were monitored. After median laparotomy and intravenous injection of 10,000 units of novo heparin, the left femoral vein were exposed and a 16F nasogastric tube (for injection of clot) and an 18F catheter (for injection of marrow or lard oil) were inserted to each vein. Human thrombin of 100 units was mixed with 10 mL blood collected from pigs, and was transferred and kept for 2 hours in a tube (5 mm in diameter and 10 mm in length, or 7 mm in diameter and 20 mm in length) to make thrombi. The thrombotic size was determined to be equal to the diameter of vein in the lower extremities. The prepared thrombi were injected into the femoral vein through a nasogastric tube. Lard pig oil was used to simulate fat emboli. The oil was fused at the temperature same as that in the living body (37°C) and injected in 0.1 mL aliquots. Bone marrow was collected using a 16F puncture needle after the ilium was exposed. Normal saline was slowly drip-infused so that the oil or bone marrow did not remain in the tube. A probe was transhepatically fixed using Doppler ultrasound so that an incidence angle of the Doppler became 60° at the retrohepatic inferior vena cava (IVC) of the liver, and HITS was monitored. The blood flow was considered to be detected when HITS were detected at the IVC by flushing 10 mL normal saline in which air was dissolved, and when respiratory variation of venous blood flow was observed. The identified HITS were analyzed and investigated.
Ethics on Animal Study
The animal study was reviewed by the Committee on Ethics on Animal Experiments in the Factory of Tokyo Medical and Dental University and The law (No. 105) and Notification (No. 6) of the Government. “Principles of Laboratory animal Care” formulated by the National Society for Medical Research and the “Guide and Use of Laboratory Animals” prepared by the Institute of Laboratory Animal Resources, National Research Council, and published by the National Academy Press, revised in 1996 was followed.
Human Study
The subjects were 65 patients (94 extremities) who underwent TKA due to knee osteoarthritis at Tokyo Medical and Dental University Hospital between September 2008 and May 2009. The deep vein was evaluated by Duplex ultrasound with compressive technique or maneuver before operation. Postoperative PE was confirmed by contrast computed tomography (CT). Seven patients (11 extremities) could not receive contrast CT due to renal dysfunction or a history of allergy. One patient (one extremity) was complicated by perforation of the digestive tract after operation. Seven patients (eight extremities) did not agree to receive postoperative CT. DVT was observed before operation in three patients (five extremities). One of them was given an IVC filter preoperatively. All the above patients were excluded from analysis. Consequently, 47 patients (69 extremities; nine males, and 38 females) with mean age of 75 years (±7.8) were analyzed in this study. The mean height and weight were 151 cm (±9.1) and 59 kg (±12.0), respectively. The patients wore support hoses and foot pump, A-V Impulse™ (Orthofix, McKinney, TX) on returning to their recovery rooms till rising from the bed in the following morning. Enoxaparin sodium (Sanofi-Aventis, Paris, France) (4,000 U/d) was subcutaneously injected from 12 hours after operation to postoperative day 11 (POD 11). HITS were measured at affected common femoral veins on POD 0 (within 1 hour after operation) and POD 1 (24 hours after operation) (Fig. 1). HITS were recorded for consecutive 7.5 minutes after adjusting the gain to make the venous flow signal have a pale blue color, corresponding to the background intensity 3–9 dB. Timing to observation was determined by the clinical simplicity at the bed side. On POD 4, the presence of DVT was assessed again using Duplex ultrasound. Contrast CT was done on POD 4 regardless of clinical indication. All clinical protocols were approved by the ethical committee of Tokyo Medical and Dental University. Informed consent was obtained before the operation.
Fig. 1
High-intensity transient signals were measured at the ipsilateral common femoral vein to total knee arthroplasty.
Continuous Wave Doppler Monitoring
A 2-MHz probe was used for Doppler ultrasound monitoring (EME Pioneer, Nicolet Biomedical, Companion, Madison, WI). The same type of machine and probe was used in the animal study and in the human study. The insonation depth was 46-58 mm, the sample volume 10 mm which was set to be equal to IVC diameter, the threshold 3 dB, and the scale width 25-50 cm/s. High-pass filter was set to be 150 Hz. Time overlap window was 75%, and amplitude was decreased to 31%. Gain was controlled so that the color tone of flow pattern became pale blue. The confidence level was set to 60%.5
Data Analysis
Signals of the same nature appeared with a sudden increase of intensity, transiently and irregularly, regardless of respiratory variations, and produced characteristic sounds that were identified as HITS. The following criteria for HITS were used for offline analysis: intensity increase of at least 3 dB above background, short duration (300 milliseconds or less), unidirectional, random appearance, and characteristic sound on the audible output.
The data were stored in a hard disk with sound track software (Win TCD EXE Version: 3.3.4.11 EME Pioneer, Nicolet Biomedical, Companion, Madison, WI). One save time was 4.1 seconds. The embolic signals were discriminated from artifacts based on the tunability of waveforms of reflected waves using the soundtrack function in off-line mode.12 Among waveforms of HITS, those that were constant for not less than 5 milliseconds were used to calculate the frequency (Hz), and it was obtained by subtracting wave number from persistence time (millisecond) of the concerned signals (Fig. 2).
Fig. 2
Venous emboli could be detected as high-intensity transient signals by Doppler ultrasound. White arrow: high-intensity transient signals (above). Using the soundtrack function in off-line mode, the frequency shift was calculated. Wave number into persistence time of the concerned signals (white arrow of broken line) gives frequency shift (below).
Statistical Analyses
Mann–Whitney test for independent samples were used to compare the differences in the Doppler characteristics among embolic materials. All p values of less than 0.05 were considered to be statistically significant. The maximum-likelihood estimation was employed with a bivariate binomial model using ROCKIT 0.9.1 BETA Version (IBM compatible version The University of Chicago, Chicago, USA) to calculate cutoff values.
Results
Animal Study
After excluding artifacts, a total of 147 single Doppler embolic signal was detected. Those with less than 60% probability were excluded, and 11 thrombi (7 × 20 mm), six thrombi (5 × 10 mm), 28 fat emboli, and 15 marrow emboli were compared. (Table I) The cutoff values were obtained from the receiver-operating characteristic (ROC) curves prepared based on the differences in the intensity volume and frequency shift. According to the ROC curves prepared to discriminate thrombi (7 × 20 mm and 5 × 10 mm) and fat emboli, the sensitivity, specificity, and accuracy at the cutoff value were 73.7, 77.5, and 77.0%, respectively, when using the intensity volume (10.9 dB) for discrimination; and 86.9, 82.3, and 88.9%, respectively, when using the frequency shift (208 Hz) for discrimination. According to the ROC curves for fat emboli and marrow emboli, the sensitivity, specificity, and accuracy at the cutoff value were 72.0, 80.0, and 74.4 %, respectively, using the intensity volume (10.9 dB); and 85.7, 91.8, and 90.7 %, respectively, using the frequency shift (196 Hz). According to the ROC curves for marrow emboli and thrombi (7 × 20 mm and 5 × 10 mm), the sensitivity, specificity, and accuracy at the cutoff value were 52.5, 66.6 and 65.6 %, respectively, using the intensity volume (8.9 dB); and 72.7, 74.3 and 71.9%, respectively, using the frequency shift (161 Hz). Figure 3 shows a scatter graph of the intensity volume and frequency shift of HITS identified among thrombotic emboli. With any types of embolus, both the sensitivity and specificity were higher when classified using the frequency shift than using the intensity volume. Fat emboli could be discriminated from thrombi and marrow emboli, but the sensitivity and specificity decreased when comparing thrombi and marrow emboli.
Table I. Intensity volume and frequency shift of emboli used in the animal study
| Study | Composition | Size | Intensity volume (dB) (mean ± SD) | Frequency shift (Hz) (mean ± SD) |
|---|---|---|---|---|
| Animal | Lard oil | Unknown | 14.5 ± 4.8 | 242 ± 37 |
| Thrombi | 5 × 10 mm 7 × 20 mm | 9.7 ± 1.8a | 178 ± 20b | |
| Marrow | Unknown | 9.3 ± 3.1a, c | 161 ± 33b, d |
ap < 0.01 relative to the intensity volume of fat emboli. |
bp < 0.01 relative to the frequency shift of fat emboli. |
cp = 0.41 relative to the intensity volume of thrombi. |
dp = 0.01 relative to the frequency shift of thrombi. |
Fig. 3
Scatter graph against animal study (lard oil, thrombus, marrow). Horizontal line: The cutoff value calculated based on the receiver-operating characteristic (ROC) curves using the frequency shift. Vertical line: The cutoff value calculated based on the receiver-operating curves using the intensity volume. Solid line, fat versus thrombi; broken line, fat versus marrow; dotted line, marrow versus thrombi. Scatter graph of thrombi (5 × 10 mm and 7 × 20 mm) and marrow emboli had a lot more overlaps than that of fat emboli and thrombi, and that of fat emboli and marrow emboli.
Human Study
A total of 135 single embolic signal from 47 patients (69 extremities) was used after excluding artifacts based on the HITS diagnostic criteria as in the animal study. Postoperative Duplex ultrasound revealed DVT in 24 patients (52.1%). Among 11 patients in whom HITS was identified after operation, four patients had PE. PE was found in two patients without HITS after operation. The cutoff values for patients with and without PE in whom HITS was identified were obtained from the ROC curves. The sensitivity and specificity at the cutoff value were 48.4 and 50.0%, respectively, when using the intensity volume (10.5 dB) for discrimination; and 76.7 and 75.0%, respectively, when using the frequency shift (208 Hz) for discrimination. Figure 4 shows a scatter graph of the intensity volume and frequency shift of HITS identified after operation. With using the intensity volume, either the sensitivity or specificity was not thought to be sufficient for discrimination between patients with and without PE. However, the cutoff value obtained from the ROC curve using the frequency shift (208 Hz) corresponded to the cutoff value to discriminate fat emboli and thrombi in our animal study.
Fig. 4
Scatter graph against human study. Horizontal line: The cutoff value calculated based on the ROC curve using the frequency shift. Vertical line: The cutoff value calculated based on the ROC curve using the intensity volume. Solid circle, pulmonary embolism (+); asterisk, pulmonary embolism (–).
Discussion
Discrimination of Emboli Using Doppler Ultrasound
Our study suggested that fat emboli could be discriminated from thrombi and marrow emboli by frequency analysis of venous emboli. Detection of arterial emboli is first reported at carotid surgery, and Doppler ultrasound has been a noninvasive examination that is useful for predicting CI in patients who have stenosis of carotid artery,3 and who undergo heart bypass and valve replacement.2, 13 Recently, several investigators have reported that discrimination of emboli can be done for some extent by using continuous wave Doppler monitoring. However, to our knowledge, there is no report concerning the emboli traveling in the veins. Regarding arterial emboli, the intensity volume, duration, sample-volume length (duration × velocity), and measured embolic power are reported to be useful for discriminating between gaseous emboli and particulate microemboli.4, 14 As equipment used for phantom studies, multifrequency transcranial Doppler is suggested to be useful,15 but whether it can perform clear discrimination has been still unknown. It is reported that microbubbles, fat emboli, and thrombi can be discriminated using the intensity volume in arterial circulation in rabbits.16 In this study, using the intensity volume, fat emboli could be discriminated from thrombi and marrow emboli even in the venous system to some extent (with sensitivity of 73.7 and 72.0%, and specificity of 77.5 and 80.0%, respectively). Regarding frequency analysis, as the ultrasound reflects diffusely in HITS of microbubbles and fatty oil, scattered waves are observed with discontinuous frequencies. In contrast, as reflected ultrasound wave form appears linear form in HITS of solid emboli, continuous frequency distribution can be observed.17 Therefore, discrimination between microbubbles and fatty oil is thought to be very difficult. However, discrimination between solid microemboli and fatty oil using the frequencies is thought to be theoretically possible with regard to waveform and frequency distribution even in the venous system, and we could discriminate them with high sensitivity and specificity in our study. Hanzawa and Rodriguez et al. have suggested that microbubbles and solid microemboli can be discriminated using frequencies of 400 and 895 Hz as cutoff values, respectively.4, 17 According to the principle of Doppler effect, the difference of frequencies basically correlates with the velocity. Discrimination is performed in the aorta in pigs with the artificial heart lung and middle cerebral artery, respectively in their study, and the cutoff values are thought to be influenced by the difference of the arterial velocity in which HITS is identified. The velocity of emboli is slower in the venous system than in the arterial system. Therefore, the cutoff value of frequency shift tended to be low (208 Hz) in our study. It is reported that the marrow cells flow in blood after TKA,8 and we used bone marrow aspirated from the ilium of pigs in this study. However, both the sensitivity and specificity were lower with discrimination between marrow emboli and thrombi, than with discrimination between marrow emboli and fat emboli. This might be because the characteristics of marrow emboli are similar to those of thrombi, as it is reported that the bone marrow flowing in blood induces clotting and platelet aggregation around the clot.18 In addition, the path of emboli causing HITS did not necessarily parallel to the vessels. The venous wall was easily deformed by external pressure, the velocity changed due to respiration, and the emboli seemed to travel in a more spiral manner (or with more often whirling) depending on their mass and form in the venous system than in the arterial system. Therefore, the path of marrow emboli might be differed from that of thrombi, which made it difficult to discriminate them using the intensity volume or frequency shift in the venous system.
PE and Fat Emboli after TKA
After TKA and long-bone fracture, CI due to fat emboli and PE due to thrombi may occur.9, 19, 20, 21 It is reported that HITS is observed in the affected common femoral vein immediately after TKA in most cases, which disappears within 168 hours after operation, and is not correlated with DVT, PE, or D-dimer level.21, 22 In our study, PE was diagnosed by contrast-enhanced CT in patients with DVT using postoperative Duplex ultrasound, and who had a lot of HITS judged as thrombi based on the frequency shift in the affected common femoral vein. In our study, two patients showed no HITS after operation and developed PE. HITS were measured in 1 hour after operation. It is reported that HITS are observed in the right atrium/ventricle 10-15 seconds after tourniquet release, and reached a peak 24-45 seconds after the release, and that 78.9-89.5% of patients show HITS 1-2 hours after TKA.8, 21 Therefore, the reason why PE did not show HITS in the two patients could be because HITS had already disappeared before the examination. The method to discriminate emboli in the venous system could also be applied to the patients undergoing orthopedic surgery, those with long-bone fracture, and those with DVT for predicting the occurrence of PE. In this study, subjects received anticoagulant therapy and foot pump with wearing support hoses after operation, and the number of PE cases was small. Therefore, we should perform another study with larger number of subjects in human study.
Conclusions
Various emboli could be discriminated in the venous system in pigs based on the intensity volume and frequency shift using Doppler equipment for embolus detection. Clinical study indicated the possibility to discriminate between patients with and without PE at frequency shift of 208 Hz in patients undergoing TKA.
The authors thank Dr. Kurihara N and Dr. Notani H for their helpful comments.
References
- . Computerized detection of cerebral emboli and discrimination from artifact using doppler ultrasound. Stroke. 1993;24:1667–1672
- . The impact of microemboli during cardiopulmonary bypass on neuropsychological functioning. Stroke. 1994;25:1393–1399
- . The neurological morbidity of carotid revascularisation: using markers of cellular brain injury to compare cea and cas. Eur J Vasc Endovasc Surg. 2007;34:552–560
- . Transcranial doppler characteristics of different embolic materials during in vivo testing. J Neuroimaging. 2002;12:259–266
- . Detection of venous emboli using doppler ultrasound. Eur J Vasc Endovasc Surg. 2008;35:96–101
- . Occurrence of pulmonary thromboembolism immediately after arthroplasty. Nucl Med Commn. 2001;22:1237–1242
- . Establishing a universal protocol for deep vein thrombosis following orthopedic surgery: total knee arthroplasty. Orthopedics. 1996;19(Suppl.):6–8
- Emboli observed with use of transesophageal echocardiography immediately after tourniquet release during total knee arthroplasty with cement. J Bone Joint Surg Am. 1998;80:389–396
- . The fat embolism syndrome. J Bone Joint Surg. 1974;56b:408–416
- . Assessment of deep vein thrombosis or pulmonary embolism by the combined use of clinical model and noninvasive diagnostic tests. Semin Thromb Hemost. 2000;26:643–656
- Prediction of pulmonary embolism in the emergency department: the revised geneva score. Ann Intern Med. 2006;144:165–171
- Consensus on microembolus detection by TCD. International consensus group on microembolus detection. Stroke. 1998;29:725–729
- Reduction of gaseous micro embolism during aortic valve replacement using a dynamic bubble trap. Gen Physiol Biophys. 2006;25:207–214
- . A comparison of four methods for distinguishing doppler signals from gaseous and particulate emboli. Stroke. 1998;29:1133–1138
- . Online automatic discrimination between solid and gaseous cerebral microemboli with the first multifrequency transcranial doppler. Stroke. 2002;33:1975–1980
- . Detection of arterial emboli using Doppler ultrasound in rabbits. Stroke. 1991;22:253–258
- . Clinical significance of frequency analysis of high-intensity transient signals (HITS) by TCD in cardioplumonary bypass. Neurosonology. 1999;12:64–70
- . Alterations in pulmonary function, coagulation and fat metabolism in patients with fractures of the lower limbs. Clin Orthop Relat Res. 1976;115:248–267
- . Transcardiac echocardiography during invasive intramedullary procedures. J Bone Joint Surg Br. 1995;77:450–455
- . Outcome of calf deep-vein thrombosis after total knee arthroplasty. J Bone Joint Surg Br. 2003;85:841–844
- . Prolonged asymptomatic micro-embolism after hip or knee arthroplasty. J Bone Joint Surg Br. 1997;79:269–272
- . Does venous microemboli detection add to the interpretation of d-dimer values following orthopedic surgery?. Ultrasound Med Biol. 1999;25:637–640
PII: S0890-5096(10)00135-4
doi:10.1016/j.avsg.2010.03.004
© 2010 Annals of Vascular Surgery Inc. Published by Elsevier Inc All rights reserved.
