Pulsed Doppler velocity patterns produced by arterial anastomoses

Ulrruvound in Med. & Biol. Vol. 9, No. I. pp. 79-87, 1983 Printed in the U.S.A.

PULSED

Departments

0301~5629/83/01007949$03.00/0 Pergamon Press Lid.

DOPPLER VELOCITY PATTERNS BY ARTERIAL ANASTOMOSES

PRODUCED

D. F. BANDYK,R. E. ZIERLER,G. A. BERNIand B. L. THIELE of Surgery, University of Washington School of Medicine, and the Veterans Administration 4435 Beacon Avenue So., Seattle, WA 98108, U.S.A.

Medical Center,?

Abstract-Centerstream velocity waveforms produced by end-to-end and end-to-side anastomoses constructed in the dog illeofemoral arterial system were studied with a 20MHz pulsed Doppler velocimeter combined with spectral analysis. Flow disturbance was identified by changes in spectral width during the systolic phase of the cardiac cycle. Measurement of the maximum frequency and the spectral width at peak systole was used to quantify the magnitude of flow disturbance at varying locations proximal and distal to the anastomoses. Disruption of the normal laminar flow pattern observed in the unoperated dog artery was evident distal to both anastomotic configurations. An increase in spectral width reflecting disturbed flow was maximal during the deceleration phase of systole. Flow disturbance was localized to a zone within one diameter distal to the anastomosis and dissipated rapidly downstream. The velocity spectrum changes observed downstream of an anastomosis resemble the flow disturbances produced by low grade, nonpressure reducing arterial stenoses. This study suggests that spectral analysis of pulsed Doppler waveforms is a potentially useful method of anastomosis assessment both to rule out major flow disruption produced by technical error, and to provide insight into the role of turbulence in the development of anastomotic intimal hyperplasia. Key Words: Ultrasound,

Pulsed Doppler, Arterial anastomosis,

The purpose of this study was to determine if ultrasound, a noninvasive and widely used clinical diagnostic tool, is capable of identifying flow disturbance produced by an arterial anastomosis. Quantification of the flow disturbance was attempted by comparison of pulsed Doppler waveform parameters obtained in normal, unoperated arteries to those obtained downstream of an arterial bypass graft anastomosis. The effect of anastomosis configuration and the type of graft material was also assessed.

INTRODUCTION Turbulence at an arterial anastomosis can affect volume flow, produce local hemodynamic effects on the adjacent vessel wall and influence the development of intimal hyperplasia and late anastomotic stenosis. Traditional methods of determining the hemodynamic changes produced by an anastomosis have relied on documenting changes in the pressure waveform (Santiago et al. 1981) or flow rate (Szilagyi et al. 1960). While these parameters are important, reduction in flow rate and mean pressure occurs only with major flow disruption as would result from a technical error in construction; such as a significant lumen stenosis, intimal flap, or thrombus formation. In experimental hemodynamic models, hot film has anemometry been the standard for identification of flow disturbances in the arterial system. This method is sensitive and has demonstrated distuibed flow downstream of arterial stenoses. (Giddens et al. 1976) and end-to-side anastomoses (Rittgers et al. 1978) which would not be expected to produce abnormal pressure or flow gradients. Aside from its many technical problems, this method is not suitable for use in the clinical setting because of its invasive nature; requiring a needle to be placed through the vessel wall. tsupported by the Veterans

Administration

Spectral analysis, Flow disturbance.

METHODS Centerstream pulsed Doppler velocity waveforms recorded proximal and distal to arterial anastomoses were compared to normal waveforms in the ileofemoral arterial systems of 24 mongrel dogs. Thirty arterial bypass grafts were interposed between the external iliac and femoral artery using both end-to-end (E-E) and end-to-side (E-S) anastomoses. The bypass graft material was autologous external jugular vein in 20 grafts and 4 mm I.D. polytetrafluoroethylene (PTFE) in 10 grafts. Surgical preparation

Dogs, weighing between 2S-3QKg, were anesthetized with intravenous pentabarbital, intubated and maintained under general anesthesia using methoxyflurane by inhalation. The hemodynamic status of the animal was monitored with a con-

Research

Service. 79

80

D.

F.

BANDYK et ul.

tinuous EKG recording and a catheter placed in the carotid artery connected to a pressure recording apparatus. Hemodynamic support was provided by intravenous infusion of lactated Ringer’s solution. After a low midline laparotomy, the retroperitoneum was entered at the aortic bifurcation. The external iliac artery was mobilized as was the femoral artery distal to its deep femoral branch at the inguinal ligament. A 10 cm length of external jugular vein was dissected from a separate neck incision and the branches ligated with 4-O silk suture. After removal, the vein was flushed with heparinized saline at zero resistance. The animal was then systemically heparinized with a dosage of 1 mg/Kg. Atraumatic vascular clamps were applied to the proximal and distal ileofemoral arterial segment and a bypass graft of either reversed external jugular vein or 4 mm I.D. PTFE was placed between the proximal iliac artery and the femoral artery distal to the deep femoral branch. Both E-E and E-S anastomoses were constructed using a continuous suture with 6-O polypropylene monofilament and 2.5X optical magnification. The anastomosis length was 2-2.5 times the host vessel diameter. Following completion of the bypass graft anastomoses, the intervening arterial segment was doubly ligated with 2-O silk and divided. Arterial pressure catheters were placed in the distal aorta and via a sidebranch of the superficial fermoral artery in the thigh to obtain simultaneous pressure recordings across the graft.

0.125 mm’. Two Doppler quadrature signals were recorded on two channels of a Hewlett-Packard instrumentation tape recorder (model No. 3964A) at a tape speed of 15 i.p.s. During recordings, the Doppler transducer, which was mounted in a 16 gauge needle, was applied directly to the external surface of the artery and acoustically coupled with saline. An angle of 60” was maintained between the Doppler probe and the logitudinal axis of the artery by use of an acrylic guide. The sample volume was adjusted to an appropriate range, calculated from caliper diameter measurements of the vessel and the Doppler angle, to ensure that centerstream flow was being recorded. Baseline recordings were obtained from the unoperated iliac and femoral artery prior to bypass graft placement. Thirty minutes after restoration of blood flow through the bypass graft, centerstream velocity recordings were sequentially obtained one host vessel diameter proximal to the anastomosis, and l-3 dia. downstream of the anastomosis. Figure 1 depicts the pulsed Doppler sample sites for a distal graft E-S anastomosis. Only the distal graft anastomosis constructed in an E-S configuration was evaluated, while both proximal and distal graft anastomoses in an E-E configuration were included in the study. Systemic pressure of the dogs was maintained at 100 ? 10 mm Hg throughout the period of velocity recordings. A real time fast Fourier transform (FFT) spectrum analyzer (Angioscan, Unigon Industries, Inc., Mt. Vernon, New York) was used to visually monitor the velocity recordings. This permitted selection of representative velocity spectra at each sample site for off-line recording and subsequent of noise analysis, and avoided the recording and artifact. At the conclusion of the velocity data recording, simultaneous pressure recordings across the

Hemodynamic measurements The velocity sensing apparatus consisted of a direction sensitive 20 MHz pulsed Doppler velocimeter (C. J. Hartley, Ph.D., Methodist Hospital, Houston, TX) with a measured sample volume of

DOPPLER PROBE (20 Megahedd

SAMPLE

VOLUME-O.125mm‘

I

MID

I

1Dia.

X = Sample site Y

=Range

/

Zbia.

I 3Dia.

of Doppler

Fig. 1. A schematic representation of a distal by pass graft end-to-side anastomosis. Sample sites of a pulsed Doppler obtained from the centerstream of the vessel are shown.

20

MHz single gated

Pulsed Doppler velocity patterns produced by arterial anastomoses

Fig. 2. Velocity spectrum obtained with an FFT spectrum analyzer. This allows measurements of the maximum frequency in systole (F,) and the spectral width (SW,) occurring at peak systole. The ratio SWJF, termed systolic spectral broadening, was used as an index of turbulence in the comparison of waveform patterns.

81

D. F.

82

bypass graft were performed. This measurement of peak systolic and mean gradients. The frequency response of the transducer system with 20 gauge attached was flat to 60 Hz.

BANDYK et al.

allowed pressure pressure catheters

Signal analysis

The Doppler signal was processed by playing back the tape at 15 i.p.s. with the two quadrature channels connected to the input of a real time Radionics 8000 FFT spectrum analyzer (Radionics Medical, Scarborough, Ontario, Canada). Quantification of the Doppler spectra was based on peak systolic frequency and spectral width. Peak systole was selected since it was a clearly definable point in each pulse cycle and was also the time interval during which velocity changes were most apparent. Measurement of maximum systolic frequency (F,), the maximum velocity in the sample volume, was performed using the spectrum analyzer frequency cursor and visual inspection of the waveform (Fig. 2). At the maximum systolic frequency, spectral width (SW,), representative of the range of velocities in the sample volume, was determined from the amplitude versus frequency display by including the high and low frequencies within -6 dB of the mode frequency (frequency of maximum amplitude). At each sample site, the waveform parameters, F,, SE,, and the ratio SWJF,, termed systolic spectral broadening, was measured for three consecutive heart cycles and the values averaged. The systolic spectral broadening ratio was used as a normalized index of the degree of turbulence since spectral width increases as peak frequency increases even under laminar ‘flow conditions. Differences in the waveform parameters were assessed using Student’s t test.

RESULTS

The pulsed Doppler velocity spectra from the normal, unoperated dog iliac and femoral arteries are shown in Fig. 3. Maximum centerstream velocity can be calculated from the hard copy output of the spectrum analyzer by measurement of the peak frequency and utilizing the Doppler equation. The unoperated dog artery contains laminar plug flow with the red blood cells in the sample volume moving at similar speeds and therefore producing a similar shift in the backscattered Doppler signal. When spectral analysis of the Doppler signal is performed, the spectral width is narrow and constant throughout the cardiac cycle. Visual inspection of the velocity waveforms distal to both E-E and E-S anastomosis (Fig. 4) show a disruption of the normal flow pattern. The flow disturbance is evident by an increase in spectral width indicating a greater variation on red cell velocities in the sample volume. The increase in spectral width is maximal in the deceleration phase of systole but becomes narrow and constant during diastole. Spectral broadening of the velocity waveform was localized in the centerstream flow to within 2 diameters distal to the anastomoses. The flow disturbance appears to dissipate at distances greater than 2 vessel diameters where a laminar flow pattern reappears. Quantification of the flow disturbance observed in systole was attempted by measurement of SW,, and the ratio SWJF,. This ratio allowed the comparison of Doppler waveforms at sample sites in normal arteries to those obtained proximal and distal to anastomoses. The measured waveform parameters of flow in normal arteries and 1 vessel diameter distal to E-E and E-S anastomoses are shown in Table 1 for both autologous vein and PTFE bypass grafts. A significant increase in

Table 1. A comparison of centerstream velocity waveform parameters (F,-maximum systolic frequency; SW,-spectral width; SW$F,-systolic spectral broadening) measured at peak systole in normal arteries to values obtained one vessel diameter dtstal to end-to-end and end-to-side anastomoses utilizing autologous vein and PTFE graft material. Values express as f 1 S.D. Normal iliac

(n=l6)

I

Artery Femoral

(n=l6)

Diameter Autologous

Distal

to Anastomosis PTFE

Vein

E-E(n=IO)

E-S(n=lO)

E-S

(n=lO)

SWPMZ)

1170+540

17392392

2035+llYO

4034+1520"

6400+2250"

SW /F P P

0.18~

0.20+0.04

0.38+0.09""

0.50+0.ll" -

0.65+0.20"

*

p < 0.01

** p c

0.05

0.05

Pulsed Doppler velocity patterns produced by arterial anastomoses

NORMAL

ARTERY

Fig. 3. Normal velocity spectrum in unoperated dog iliac and femoral artery.

83

84

D. F.

BANDYK et al.

Fig. 4. Velocity spectra obtained 1, 2 and 3 vessel diameters distal to an end-to-end anastomosis (vein graft) and an end-to-side anastomosis (vein graft). Spectral broadening evident during systole and maximal during the deceleration phase of systole. Maximum flow disturbances are seen at the one-diameter sample site with the three diameter site showing almost a normal flow pattern.

Pulsed Doppler velocity patterns produced by arterial anastomoses

spectral width (p < 0.01) was measured distal to E-S anastomoses constructed between a host artery and a graft of either vein or PTFE material. Distal to E-E anastomoses constructed with autologous vein, spectral width changes were not statistically different (p = 0.2). Use of the systolic spectral broadening ratio allowed normalization on the absolute spectral width relative to the peak spectral frequency. In Fig. 5, the systolic broadening ratio is plotted to compare this index of disturbed flow in normal arteries to that calculated 1 diameter proximal and distal to both anastomotic configurations. The ratio was significantly increased distal to both E-E (p < 0.05) and E-S anastomoses (p < 0.01) when compared to values of this parameter in normal arteries and proximal to the anastomoses. Flow disturbance were greatest distal to anastomoses constructed with PTFE relative to autologous vein, but the difference was not significant (p < 0.10). Assessment of the pulsed Doppler waveform at the midpoint of an anastomosis (Fig. 1) revealed a narrow spectral width throughout the cardiac cycle with a systolic spectral broadening ratio similar to normal arterial flow. This finding suggests that the origin of the flow disturbance is at the inlet to the host vessel. No mean pressure (+-2mm Hg) or systolic pressure (+-5 mm Hg) gradient was measured across the bypass grafts as determined by simultaneous proximal and distal high fidelity pressure recordings.

T

E-S

MEANtS.

1.0

i

IPTFE)

0.75

Normal

1

E-E Anastomosas

Artery

IVEiN

p co.01

SWP/Fp 0.5

pco.01

(VEIN)

11 0.65f0.20 0.5t0.11 vein

0.25 I 20f0.06

I

0.05 Iliac h=16) Normal

Femoral h=16) Artery

Proxina, End-to-End ln=10)

I 1-D Distal *nartolnosis

Proximal I-D Dstal End-to-Side Anastomosis (n=16)

Fig. 5. A comparison of systolic spectral broadening ratio (SW,/F,) measured in unoperated iliac and femoral artery to that obtained proximal and one diameter downstream of end-to-end (E-E) anastomosis. Vertical bars indicate mean value t I standard deviation.

85

DISCUSSION

Theoretically, the conversion of a tapered, elastic arterial conduit into a complex geometric configuration of an anastomosis with varying wall compliance should alter the pattern of blood flow. Flow pattern changes produced by anastomosis have been demonstrated in rigid model systems (Logerfo et al. 1979) and in animals using hot-film anemometry (Rittgers, et al. 1978). While these investigatiions have demonstrated areas of boundary layer separation and skewed velocity profiles downstream of an anastomosis, the results are difficult to extrapolate to the clinical setting. In the case of hot film anemometry, the results cannot be verified in humans because of the invasive nature of the technique. The detection of blood flow abnormalities with ultrasound has been demonstrated using both pulsed and continuous wave Doppler methods. Quantification of disturbed flow, particularly due to minor lesions, requires the use of a pulsed Doppler system which allows study of the velocity pattern at specific points in the vessel lumen. Interpretation of Doppler velocity spectra is based primarily on changes in peak systolic frequency and spectral width (Fell et al. 1981). In vessels with normal laminar flow, the velocities of red blood cells within a pulse Doppler sample volume are similar, and FFT spectral analysis will show a narrow range of frequencies. As laminar flow decays, movement of red blood cells in the sample volume becomes more random and results in a greater variation in the Doppler shift of the backscattered Doppler signal; and is termed spectral Recognizing pattern changes in broadening. spectral and temporal characteristics of a pulsed Doppler waveform is currently being used clinically to determine the extent of carotid occlusive disease (Greene et al. 1982). This study was designed to determine if a high frequency pulsed Doppler device was capable of identifying flow disturbances produced by an arterial anastomosis and in addition to quantify what these changes are. Velocity waveform changes in centerstream flow were identified downstream from both E-E and E-S anastomoses using a 20 mHz pulsed Doppler velocimeter. Fast Fourier Transform spectral analysis of the Doppler signal allowed both qualitative visual assessment of the velocity waveform and quantitative comparison of flow patterns based on velocity parameters measured at peak systole. An increase in spectral width, reflecting disturbed flow, was apparent during systole distal to both types of anastomoses compared to flow in un-

86

D. F.

operated iliac and femoral arteries with maximal changes occurring in the deceleration phase of systole. The flow pattern during diastole was unaffected. The ratio of peak systolic spectral width to peak frequency, termed systolic spectral broadening, was used as a normalized index of turbulence to allow comparison of velocity waveforms in vessels of varying diameter and flow rate. This index indicated that the degree of disturbed flow was increased downstream of E-S anastomoses compared to E-E anastomoses, particularly when PTFE was used as the graft material. In addition, the origin of disturbed flow appeared to be at the inlet of the anastomosis to the host vessel. Similar velocity changes have been noted in investigations of flow disturbances downstream of low grade nonpressure-reducing arterial stenoses using both hot film anemometry (Giddens et al. 1976) and a pulsed Doppler system (Hutchinson et al. 1981) similar to that used in this study. Disturbed flow first appeared in the deceleration phase of systole and was dependent on flow rate. As the degree of stenosis was increased, disturbances appeared earlier in systole and were accompanied by an increase in peak velocity. In the case of a pressure reducing stenosis (-> 50% diameter reduction), disturbed flow was noted throughout the cardiac cycle. The absence of a pressure gradient across the anastomoses constructed in this study makes technical error an unlikely cause for the observed disturbed flow. Based on the temporal and spectral characteristics of the flow an anastomosis appears to have disturbance, similar hemodynamic effects as a low grade arterial stenosis. The waveform changes of disturbed flow were most evident at the one and two diameter downstream recording sites and had dissipated at the three diameter recording site. This localization of disturbed flow in close proximity to an anastomosis and the absence of distal propagation may have implications in the etiology of intimal neointimal fibrous Anastomotic hyperfiasia. hyperplasia (ANFH) occurs at all anastomoses and is most evident when using prosthetic grafts, particularly PTFE (DeWeese and Green, 1980). Numerous investigators have suggested that mechanical and hemodynamic factors play important roles in the pathogenesis of atherosclerosis and intimal hyperplasia. Lateral wall pressure (Texon, 1%3), turbulence (Wesolowski et al. l%S), and wall shear stress (Fry, 1960) have all been considered as important hemodynamic factors.

BANDYK et al.

Rittgers (1978) demonstrated skewed velocity profiles downstream of an E-S anastomosis and calculated abnormal shear forces in regions where the greatest proliferation of intima was found. Characterization of the disturbed flow was not possible because of the lack of direction sensing capability of hot film anemometry. The centerstream flow disturbances demonstrated in this study cannot be extrapolated to characterize ,flow at the vessel wall. In a study of the radial distribution of disturbed flow distal to concentric stenoses in the dog thoracic aorta, Hutchinson et al. (1981) found decreasing turbulence as one moves from the centerstream toward the wall. In contrast Teijeira et al. (1981) found significant skewing of the velocity profiles distal to an E-S anastomosis with regions of reverse flow near the wall suggesting boundary layer separation. These studies support our findings of centerstream flow disturbance in the proximity of an anastomosis, but further investigation is required to assess flow patterns- adjacent to the vessel wall and their late effects on wall morphology. At present, operative confirmation of a technically satisfactory anastomosis relies primarily on visual inspection, pulse palpation, arteriography and qualitative audio Doppler flow analysis (Mozersky et al. 1973). Flow disruption which produces a pressure gradient is easily recognized using spectral analysis of Doppler waveform (Hutchinson et al. 1981). Distal to non-pressure reducing anastomoses, as documented in this study, flow disturbances are located primarily at peak systole and during the deceleration phase. An increase in peak frequency with spectral broadening throughout the cardiac cycle represents major flow disruption consistent with a technical error. Further studies are in progress to characterize the flow disturbances with various technical problems such as intimal flap, anastomotic stenosis and thrombus formation. The present study showed clearly that altered velocity patterns are produced in the region of arterial anastomoses considered technically adequate by traditional methods of evaluation. Evaluation of the hemodynamic effects of an anastomosis using pulsed Doppler spectral analysis is simple to perform, reproducible, and can be performed at multiple sample sites relative to the anastomosis to obtain physiologic velocity data. The clinical significance of the waveform changes was not specifically addressed in this experimental study, as serial postoperative recordings would be required to determine if the flow disturbances per-

Pulsed Doppler velocity patterns produced by arterial anastomoses

occurs. Such longitudinal studies will allow correlation on the presence, location, and degree of flow disturbance with histological changes in arterial wall morphology. sist

or diminish

as arterial

wall

healing

REFERENCES DeWeese J. A. and Green R. M. (1980) Anastomotic neointimal fibrous hyperplasia. In Complications in Vascular Surgery, Grune and Stratton, Inc., New York. Fell G., Phillips D. J., Chikos P. M., Harley J. D., Thiele B. L. and Strandness D. E. Jr. (1981) Ultrasonic duplex scanning for disease of the carotid artery. Circulation 64, 1191-1195. Fry D. L. (1968) Acute vascular endothelial changes associated with increased blood velocity gradients. Circ. Res. 22, 165172. Giddens D. P., Mabon R. F. and Cassanova R. A. (1976) Measurement of disordered flows distal to subtotal vascular stenoses in the thoracic aortas of dogs. Circ. Res. 39,112-119. Greene F. J. Jr., Beach K., Strandness D. E. Jr., Fell G. and Phillips D. J. (1982) Computer based pattern recognition of carotid arterial disease using pulsed Doppler ultrasound. Ultrasound in Med. & Biol. 8, 161-176. Hutchinson K. J., Thiele B. L., Greene F. M. and Hokanson D. E. (1981) Radial distribution of disturbed flow distal to

87

stenoses in the canine thoracic aorta. (Submitted to J. Appl. Phy.). Hutchison K. J., Thiele B. L., Greene F. M., Strandness D. E. Jr. (1981) Detection of disturbed flow by computer processing of pulsed Doppler spectra. (Submitted to Circ. Res.) LoGerfo F. W., Soncrant T., Tee1 T. and Dewey C. F. Jr. (1979) Boundary layer separation in models of side-toend arterial anastomoses. Arch. Surg. 114, 1369-1373. Mozersky D. S., Sumner D. S., Barnes R. W. and Strandness D. E. Jr. (1973) Intraoperative use of a sterile ultrasonic flow probe. Sum. Gynecol. Obslet. 136,239-282. Riitgers S. E?, I&yannacos P. E., Guy J. F., Nerem R. M., Shaw G. M.. Hostetler J. R. and Vasko J. S. (1978) Velocitv distribution gnd intimal proliferation in autologous Gein grafts in dogs. Circ. Res. 42,792-800. Santiago E. J., Chatamra K. and Taylor D. E. M. (1981) Hemodynamic consequences of bypass arterial grafting below 8mm i.d. Hemodynamics of the Limbs-2, Toulouse, France, pp. 435-442. Szilagyi D. E., Whitcomb J. E., Schember W. and Waibel P. (1960) The laws of fluid flow and arterial grafting. Surgery 47, 55-73. Texon M. (1963) The role of vascular dynamics in the development of atherosclerosis. In Arherosclerosis and Its Origin. Academic Press, New York. Wesolowski S. A., Fries C. C., Sabini A. M., Sayer P. N. (1%5) Turbulence, intimal injury, and atherosclerosis. In Biophysical Mechanisms in Vascular Homeostasis and Infravascular New York. Thrombosis. Appleton-Centrury-Crofts,