O R I G I N A l , ARTICLES

Perioperative monitoring of blood flow in femoroinffagenicular vein grafts with Doppler ultrasonography: A preliminary report John A. Brennan, MB, ChB, Abigail J. Thrush, MSc, David H. Evans, P h D , and Peter R. F. Bell, M D , Leicester, United Kingdom A system for monitoring blood flow in femorodistal vein grafts with Doppler ultrasonography in the immediate postoperative period has been developed. Twenty-three grafts have been monitored for periods of up to 72 hours. Seventeen grafts have remained patent at minimum follow-up of 6 months, and six grafts occluded in the immediate postoperative period. Successfifl Doppler recordings were obtained in 16 successful and all failed grafts. Fast Fourier transform analysis of the Doppler signals was performed, and pulsatility index and time-averaged mean velocity were derived from the spectral information. Successful grafts displayed hyperemic flow with pulsatility index < 2 and time-averaged mean velocity > 10 cm/sec. Failed grafts could be classified in two groups: those that occluded less than 24 hours after operation and those that occluded after 24 hours after operation. Short-term failure was categorized by highly pulsatile flow, with pulsatility index rising rapidly and time-averaged mean velocity falling correspondingly before actual cessation of flow. Delayed failure was less well defined but was suggested by failure to develop, or early deviation from, the hyperemic flow se6n in successful grafts. Occlusion was heralded by development of the pulsatile pattern seen in the short-term failure group. (I V~sc SuxG 1991;13:468-74.)

The technique of femorodistal bypass grafting by use of in situ saphenous vein is now well established in thc management of patients with critical lower limb ischemia. ~ Considerable experience has been gained in recent years with regard to patient selection and improvements in operative technique, leading to increasingly successful results? '~ In spite of these advances, however, a group of patients remain in w h o m graft failure occurs in thc immediate postoperative period? In the absence of technical errors the cause of failure is likely to be inadequate distal runoff, rcsulring in poor flow within the graft, and subsequent thrombosis. Some grafts, however, particularly From the Department of Surgery and the Department of Medical Physics (Ms. Thrush, Mr. Evans), Leicester Royal Infirmary, Leicester, United Kingdom. Reprint requests: J. A. Brennan, MB, ChB, Department of Surgery, Clinical SciencesBuilding, Leicester Royal Infirmary, Leicester LE2 7LX~United Kingdom. 24/1/26790 468-

vein, could be salvaged before they become thromboric provided objective evidence of impending failure can be obtained. Previous studies have shown that in femoropopliteal vein grafts performed for claudicarion, volume blood flow increases in the first 24 hours compared to the value measured at the end of operation, high flow being maintained for 4 to 5 days before declining to normal levels. 4-6 This response was found to be most pronounced in those patients with the most severe ischemia, and is thought to be due to dilation of the runoffvessels with a consequent fall in peripheral vascular resistance. It is conceivable that in the presence of critical ischemia this hyperemic response may be delayed or defective resulting in poor graft flow, which may fall below the thrombotic threshold, 7 in the critical early postoperative period. To improve our understanding of the events that take place, particularly in grafts that occlude early, a system capable of closely monitoring graft hemody-

Volume 13 Number 4 April 1991

namics is required. Previous work has been carried out with implantable electromagnetic flow probes, 4'5 but these are impractical for regular clinical use and only provide a figure for volume blood flow, which is of limited use in isolation. In contrast, the study of blood flow by use of Doppler ultrasonography has developed enormously in recent years, and directly relevant physiologic information can now be obtained from spectral analysis of Doppler-derived velocity waveforms. 8 This technology is applied in duplex scanners, which provide a combination of a B-mode ultrasound image of blood vessels with a facility for blood flow measurement. These machines are cumbersome to use and are not ideally suited for intensive monitoring in the early postoperative period where there is a requirement for frequent measurements. In an attempt to overcome the ~,roblems of early data collection we have developed a practical Doppler system designed to record blood flow from fizmorodistal vein grafts. This article describes the system, a n d we report our early experience of it in clinical use. MATERIAL A N D M E T H O D S Description o f monitoring system

The system consists of a Sonicaid BV380 bidirectional continuous-wave Doppler unit, and a custom built 4 M H z probe. This is connected via a custom built timer and control unit to a cassette tape recorder that records the forward and reverse audio Doppler signals onto separate channels. Initial recordings were made with a Uher CR160 recorder with standard tapes, but this has been replaced with a Sony (Sony Sound Tec Corp., Tokyo, Japan) 2CD-D10 digital audio tape unit. The latter is advantageous as it automatically records the date and time of any' recordings and has a wider dynamic range. This avoids problems with recording levels, allowing more data to be extracted from each recording. The timer unit controls a relay that switches power to the Doppler unit and simultaneously activates the tape recorder. There is a facility for making automatic recordings at 15- or 30-minute intervals, each providing 1 minute of continuous data. The system can also be activated manually at any time allowing additional recordings of any length to be made. When the unit is in record mode a volume control makes it possible to listen to the signal from the graft, a feature that provides the surgeon or nurse fai~liar with l~e Doppler audio signal with helpful information regarding early graft function. The custom built probe (Fig. 1) was made with

Perioperative bloodflow monitoring i~ vein grafts 469

Fig. 1. Diagram of the Doppler transducer. The molded resh~ probe (a) houses the piezoelectric disks (b) adjacent to a hollow (c) that is filled with gel before being sited over the graft. two piezoelectric ceramic disks, one transmitting and one receiving, set in a hollow in a dome-shaped mold of cold curing epoxy resin. The discs have been arranged so that the resultant ultrasound beam subtends an angle of 45 degrees to the flat surface, which is attached to the patient, and their centerlines cross at 2 cm from the probe surface. The probe has been specifically designed to study grafts lying in a relatively superficial position. Beyond a depth of 3 to 4 cm the probe picks up poor signaJs that are not analyzable, so it is not possible to monitor deeply tunnded grafts. To obtain a satisfactory signal the incision at the elected site of probe placement (generally the lower thigh) was closed with a subcuticular suture and covered with a clear Opraflex (Neuwied, Federal Republic of Germany) dressing. Recordings were commenced in the immediate postoperative period, in the operating theater recovery area. The probe was filled with ultrasound gel, placed over the graft, and its position was adjusted until the best signal was obtained. It was then fastened in place by use of a double-sided adhesive ring (Fig. 2). An initial 1-minute recording was made before setting the system to automatic mode. Once correctly positioned the probe provided satisfactory recordings for periods of up to 7 days. Although the probe remained in place constantly throughout the monitoring period, it was designed with a long flexible lead that could be disconnected from the unit at any time, causing minimal hindrance to patient care. On completion of recording the tapes were replayed through a real-time fast Fourier transform analyzer, 1° and the maximum frequency and intensity-weighted mean frequency (IWMF) envelopes were extracted for each period of recording. The IWMF is the mean frequency of the Doppler signal taking into account the fraction of blood cell reflectors moving at each velocity over the pulse cycle. From these data the time-averaged mean velocity

Journal of VASCULAR SURGERY

470 B r e n n a n et al.

if flow is hyperemic with forward flow occurring throughout the cardiac cycle, and rises in the presence of increasing peripheral vascular resistance. In the normal popliteal artery the PI is > 6.

Patients

Fig. 2. Photograph of the probe in position over a graft. The clear dressing facilitates accurate locating of the probe and allows ultrasound transmission. The lead from the probe is attached to the Doppler unit, on top of which is the recording unit. (TAMV) and the mean pulsatility index (PI) were calculated and plotted against time. In the former case the frequency was converted to velocity (V) by assuming the graft to be parallel to the skin and hence at 45 degrees to the ultrasound beam, because in the Doppler equation the following occurs: cFd 2Ft cos0 where c = speed of sound in blood; Fd = Doppler shift frequency; Ft = transmitted frequency; 0 = angle of insonation (45 degrees). By use of the IWMF in the Doppler equation the velocity calculated is the TAMV, and since volume flow-T A M V x cross-sectional area, it can be seen that a proportional change in TAMV represents an equivalent change in volume flow, provided that the cross-sectional area remains constant. The PI is defined as the peak to peak excursion of the maximum frequency envelope divided by the mean value over the cardiac cycle. ~ This ratio is low V

-

-

-

To ascertain its potential the system described above was used to monitor 23 femorodistal vein grafts for periods of between 8 and 72 hours after operation. Twenty grafts were performed for critical ischemia and three for severe ( < 50 m) clandication. In twenty-two patients the long saphenous vein (19 in sit-u, 3 reversed) was used but was unsuitable in one, and a reversed arm vein was used. The distal anastomosis was to the below-knee popliteal artery in seven, the tibioperoneal trunk in three, the peroneal in seven, the anterior tibial in four, and the dorsal pedal in two. Completion angiography was pez formed routinely to identify technical errors. Seventeen grafts have remained patent at a minimum follow-up of 6 months, and six grafts, all performed for critical ischemia, failed at intervals ranging from 12 hours to 10 days. Five of these were reexplored at which time no obvious cause for failure could be found. In four, thrombectomy alone was performed but all rcoccluded between 2 hours and 9 days. In one patient an in situ graft to the peroneal artery was relocated to a closed popliteal segment, and although the graft remained patent, there was no clinical improvement. In all six cases the end result of early occlusion was a major amputation. RESULTS Satisfactory recordings were taken from 22 of the 23 grafts. The unsuccessful case was a result 6[ placing the probe too proximally in the thigh. Data were therefore obtained from 16/17 successful grafts and all 6 failed grafts. Analysis of the early cases revealed a high level of background noise generated by electrocardiographic monitoring equipment in the operating theater environment. This interference resulted in the first few hours of recording being unsuitable for computer analysis, and attempts to reduce it with a mains filter were unsuccessful. To capture data in the immediate postoperative period it has been necessary to temporarily switch off the patient's monitor during recordings. N o such problems have been encountered on the ward.

Successful grafts Fifteen of the 16 successful grafts demonstrated stable hemodynamics during the recording period (Fig. 3). This was reflected by an initial fall in PI,

Volume 13 Number 4 April 1991

which remained 10 em/sec in all 15 grafts and was > 20 cm/sec in 9, indicating high flow (e.g., for a 5 mm vein a TAMV of 20 cm/sec is equivalent to volume flow of 235 ml/min). In six grafts therefore the T A M V was maintained between 10 to 20 cm/sec, and although this indicates lower flow, all have gone on to long-term patency. The common feature seen in these grafts was the maintenance of both the measured parameters, with fluctuations around the steady state, once hyperemic flow was established. The remaining graft in this group showed a progressive departure from the steady state over 72 hours, with the PI rising slowly from 2 to 3 and the T A M V falling from 25 to 15 cm/sec. This early ~hange toward a more normal pulsatile pattern may reflect the fhct that the graft was performed for claudication rather than tissue necrosis where one would expect: any initial hyperemia to persist longer. That the changes seen in this case were not detrimental is borne out by the fact that the graft remains patent at 18 months. Failed grafts Nine sets of recordings are available for the six failed grafts and five revisions. Three categories emerged, (1) grafts failing within 24 hours, (2) grafts failing after a number of days, and (3) one graft remaining patent with no clinical improvement. Failure within 24 hours (four grafts). These grafts produced a characteristically pulsatile signal, with reversed flow in diastole. This was illustrated graphically by a rapidly rising PI and a corresponding fall in T A M V (Fig. 4). Although the audible signal was clearly pulsatile in two patients, analysis showed that there was no net forward flow. In this situation the audible signal presumably represents an oscillating column of blood in a graft with thrombosed runoff.

Failure within a few days (four grafts). This group was much less well defined. Two grafts, failing after 5 and 10 days, displayed the hyperemic flow pattern seen in successful grafts, with PI < 2 and T A M V > 1 0 cm/sec, throughout the recording period. In neither case did analysis reveal changes in flow to suggest that subsequent occlusion was likely. We assume that recording for longer periods would have documented the changes that took place before occlusion. One graft that failed after 9 days demonstrated initially hyperemic flow with progressive deviation

Perioperative bloodflow monitoring in vein graf~s 471

8

a

a z H

H

m J

i

2

.

b

+

.~ SO-

m=°unD°Dum°="Domnmnmn°~mo"m°Dn %Bm%a~mu%un~a~n%mmff% 20 > H

~0

I

I

I

I

I

I

I

I

I

I

1tl211

TIME

I

I

AFTER

I

I

',

~ I

I

:

:

24

OPERATION

', [

i

:

I

i

i

I

I

!

i

36

(HOURS)

Fig. 3. Graph of (a) pulsatility index, and (b) IWMV (intensity-weighted mean velocity) plotted against hours after operation, showing the hyperemic pattern seen in successfial grafts. toward pulsatile flow (Fig. 5), similar to the isolated case in the successful group. This graft, however, was performed for critical ischemia rather than clandication, suggesting that in the presence of severe ischemia an early departure from the expected hyperemic flow is of more concern. The fourth graft (Fig. 6) had a persistently low TAMV < 10 cm/sec associated with a PI > 2. This unfavorable flow pattern persisted for 3 days before assuming the pulsatile pattern seen in the early failure group, the graft occluded shortly thereafter. Hemodynarnic failure (one graft). This graft had failed within 24 hours and was relocated to a closed popliteal segment. Although it was still patent at the time of amputation 13 days later, analysis revealed a poor flow pattern with a gradually rising PI from 2 to 2.5 and a TAMV < 10 cm/sec. Poor flow demonstrated at this proximal level is in keeping with the clinical picture of poor runoff and lack of clinical improvement. Considering then the group as a whole, the following four distinct hemodynamic patterns were observed: (1) Stable hyperemia (PI < 2 and TAMV > l0 cm/sec) was seen in 17 grafts, 15 of which remained patent. (2) Progressive deviation from

Journal of VASCULAR SURGERY

472 B r e n n a n et al.

Fig. 4. Graph of (a) pulsatility index, and (b) IWMV (intensity-weighted mean velocity) plotted against hours after operation, showing rapid development of highly pulsatile flow in a graft failing within 24 hours. Adjacent to this is an example of the pulsatile waveform generated by computer analysis of the signal, showing reverse flow in diastole. With hyperemia there is forward flow throughout the pulse cycle.

I I m~ iiRRllmimmlgimmmwRiINll WmlllllRmmmllilll~l

initial hyperemia was seen in two grafts, one of which remained patent. (3) Rapid development ofpulsatile flow was seen in four grafts all of which occluded within 12 hours. (4) Persistently poor flow (PI > 2 and T A M V < 10 cm/sec) was seen in two grafts, one occluded after 3 days and the other remained patent without clinical improvement. Since it would be expected that all successful grafts should display stable hyperemia in the perioperative period, and that any deviation from such a pattern is likely to indicate a potentially failing graft, it can be seen that the recordings provided by this new system reliably predicted graft outcome in 22 out of 25 (88%) cases.

I w mmMm~

o o 3o-

°°~ ooOoOO°ooOOoo~°o°Oo.

o o°

o

°°"°°°%°

o

o

Oo o

°° o o °

D

D D

D

2o.

H

lo.

J

]

I

J

I

I

I

I

I

Et [ 2 1

[

I

J

I

J

I

I

J

I

I

~l

T~ME

J

J l l L I

AFTER

J]l

,,

' ', : : : : : ' , ; ; :

OPERATION

48

tHOURB1

Fig. 5. Graph of (a) pulsatility index, and (b) 1WMV (intensity-weighted mean velocity) plotted against hours after operation, showing progressive departure from an initially hyperemic pattern.

DISCUSSION This preliminary study has shown that by use of the monitoring system we have developed, it is possible to closely follow the hemodynamic course of femorodistal vein grafts in the immediate postoperative period, and that the data obtained correlate well with early graft outcome.

Volume 13 Number 4 April 1991

Perioperative bloodflow monitoring in vein grafts 473

30

w J

~

>

H

~c

°%°°°°° Oo

ll~fl

o

o

oo~

o o o °o o° ~°°°°o°O°.°o°°oooOoooo°oOO~Oo°°oo • m ......

,

. . . . . . . . . . .

o~ o°

~4 . . . . . . . . . . . TIME

AFTER

o Oo°o°O°oOO

3'6 . . . . . . . . . . .

OPERATION

o~

° ~ ~o° o ~ o~

o

~o °oO°~o°O°°°°°"%°o

4e ....

llllLlido

o o o°

o

I

o °% %° %%° llllL~°$°

(HOURS)

Fig. 6. Graph of (a) pulsatility index, and (b) IWMV (intensity-weighted mean velocity) plotted against hours after operation, showing persistently poor flow with eventual development of abnormal pulsatility before actual failure after 67 hours. After a :successful femorodistal reconstruction there is a period of hyperemic flow, the duration of which is related to the severity of the preoperative ischemia. 3-s ]In the presence of tissue necrosis, hyperemia can be expected to persist for a period of weeks. 8 Intraoperative measurement of blood flow on graft completion can predict subsequent outcome, ~2 but as flow is dependent on a number of interacting factors, such as systemic blood pressure, vein diameter, and outflow disease, isolated flow measurements may be potentially misleading. Analysis of Doppler waveforms obtained from grafts allow other aspects of flow to be studied, and provides a more complete picture of graft hemodynamics. Bandyk et al.8 characterized hyperemic flow by intraoperative assessment of in situ vein grafts using a handheld Doppler probe. In their study all successful grafts displayed antegrade flow throughout the pulse cycle and a peak systolic velocity > 40 cm/sec. In our study we chose to measure PI since it is easily derived from the Doppler specmun and provides a measure of peripheral resistance. In the presence of forward diastolic flow the peak-to-peak excursion of the Doppler waveform is reduced, and this is reflected in a low PI. We looked at T A M V rather than peak systolic velocity, since it is directly related to voltu-ne flow. If the vein diameter at the site of probe placement is known, volume flow is easily calculated. For the purposes of this study, however, we assumed that the graft diameter remained constant, and that any change in T A M V was accompa-

nied by an equivalent change in volume flow. All successful grafts in our study had a PI < 2 and T A M V > 10 cm/sec, and we can conclude that this combination is indicative ofhyperemic flow, since we know that a graft successfully performed for severe ischemia will exhibit hyperemia after operation. Early deviation toward more pulsatile flow was seen in only one graft in this group. This can be explained by the fact that the graft was performed for claudication, where one would expect a quicker return to normal flow. None of the occluded grafts in our study had any identifiable cause for failure other than poor runoff, and it is interesting that this resulted in two distinct categories of failure, that is, failure within, or after, 24 hours. In the early failing group, occlusion was clearly identified by a dramatic rise in PI associated with a corresponding fall in TAMV. Once this pattern developed it was followed by occlusion within 6 hours. It was observed in one of the late failure group after a prolonged period of stable, but poor flow. It is reasonable to assume that the preocclusion pattern would have been seen in the other three late failures if monitoring had been continued for longer periods. It was disappointing, however, that in two cases there was no indication during the early recording that subsequent occlusion was likely. The remaining graft in this group showed a gradual, early departure from hyperemia that was not expected, since the indication for surgery was rest pain~

474

J'ournalof VASCULAR SURGERY

B r e n n a n et al.

We think that the chosen combination o f PI and T A M V is appropriate for monitoring early graft flow, and the frequent measurements that the system provided enabled any trends in these parameters to be well defined. Although duplex scanning has been used to assess early graft flow, g we have found it impractical for the frequent monitoring that we feel is necessary to follow the trends seen in early failing grafts. A drawback o f the present system is that the results are analyzed retrospectively, but further development will enable on-line analysis to be performed so that the trends in PI and T A M V can be displayed at the bedside. At present it is possible to assess graft flow by operating the system manually and listening to the Doppler signal. Where an aggressive approach toward femorodistal reconstruction in the presence o f critical ischemia is adopted, the problem o f early postoperative graft failure still remains. ~Intervention to salvage a falling graft before actual occlusion improves long-term patency, 13 and this new system provides a means whereby early postoperative failure can be reliably predicted. The potential value o f this was demonstrated in our series, since intervention once a graft had occluded was unsuccessful in all instances.

REFERENCES 1. Leather RP, Shah DM, Chang BB, KauftnanJL. Resurrection of the in situ saphenous vein bypass: 1000 cases later. Ann Surg 1988;208:435-42.

2. Buchbinder D, Rollins DL, Semrow CM, Schuler lJ, Meyer JP, Flanigan DP. In situ tibial reconstruction. State of the art. or passing fancy. Ann Surg 1988;207:184-8. 3. Shoenfield NA, O'Donnell TF, Bush I-IL, Mackey WC, Callow AD. The management of early in situ saphenous vein bypass occlusions. Arch Surg 1987;122:871-5. 4. Cronenstrand R, Ekestrom S. Blood flow after peripheral arterial reconstruction I. Scand J Thorac Cardiovasc Surg 1970;4:159-71. 5. Renwick S, Gabe JT, Shillingford JP, Martin P. Blood flow after reconstructive arterial surgery measured by implanted electromagnetic flow probes. Surgery 1968;65:197-206. 6. Eastcott HHG. Arterial grafting for the ischemic lower limb. Ann Roy Coil Surg Eng 1953;13:177-84. 7. Sanvage LR, Walker MW, Berger KG. Current arterial prostheses. Arch Surg 1979;114:687-91. 8. Bandyk DF, KaebnickHW, BergaminiTM, Moldenhaner P, Towne JB. Haemodynamicsof in situ saphenous vein arterial bypass. Arch Surg 1988;123:477-82. 9. Thrush AJ, Evans DH. Simple system for automatic intermittent recording of blood flow in femorodistal bypass grafts using Doppler ultrasound. Med Biol Eng Comput 1990; 28:193-5. 10. Schlindwein FS, Smith MJ, Evans DH. Spectral analysis of Doppler signals and computation of the normalised first moment in real time using a digital signalprocessor.Med Biol Eng Comput 1988;26:228-32. 11. Gosling RG, King DH. Continuous wave ultrasound as an alternative and complement to X-rayin vascularexamination. In: Reneman RS, ed. Cardiovascular applications of ultrasound. Amsterdam: North Holland, 1974:266-82. 12. Terry HJ, Allan JS, Taylor GW. The relationship between blood flow and failure of femoropopliteal reconstructive arterial surgery. Br J Surg 1972;59:549-51. 13. Whittemore AD, Clowes AW, Couch NP, Mannick JA. Secondary femoropopliteal reconstruction. Ann Surg 1981; 193:35-42.

B O U N D V O L U M E S AVAILABLE TO S U B S C R I B E R S Bound volumes of the JOURNAL OF V^SCULAV, SUV,G~RY for 1991 are available to subscribers only, They may be purchased from the publisher at a cost of $59.00 for domestic, $79.13 for Canadian, and $75.00 for international subscribers for Vol. 13 (January to June) and Vol. 14 (July to December). Price includes shipping charges. Each bound volume contains a subject and author index, and all advertising is removed. Copies are shipped within 60 days after publication of the last issue in the volume. The binding is durable buckram with the journal name, volume number, and year stamped in gold on the spine. Payment must accompany all orders. Contact Subscription Services, Mosby-Year Book, Inc., 11830 Wesfline Industrial Drive, St. Louis, MO 63146-3318, USA. In the United States call toll free: (800)325-4177, ext. 4351. In Missouri call collect: (314)872-8370, ext. 4351. Subscriptions must be in force to qualify. Bound volumes are not available in place of a regular JouR_~ subscription.