Relationship of spinal cord blood flow to vascular anatomy during thoracic aortic cross-clamping and shunting

J THORAC CARDIOVASC SURG 91:71-78, 1986

Relationship of spinal cord blood flow to vascular anatomy during thoracic aortic cross-clamping and shunting No satisfactory explanation exists as to why paraplegia occurs despite distal aortic perf~ion during thoracic aortic operatiolti. We studied the hemodynamics, paraplegia rate, and spinal cord blood flow with radioactive microspheres in 17 male adult baboo~ with particular reference to the arteria radicularis magna. The groups consisted of control animals, animals subjected to cross-clamping for 60 minutes, and animals with aorto-aortic shunts operational for 60 minutes. There were no significant left ventricular hemodynamic advantages with shunting. Shunting significantly increased lumbar spinal cord blood flow (P = 0.0009), which correlated with the distal aortic mean pressure (r = 0.59, p = 0.008). However, lower thoracic spinal cord blood flow did not increase during shunting (p = 0.2) and did not correlate with the distal aortic pressure (r = 0.11, p = 0.64~ This is due to the vascular anatomy of the anterior spinal artery, which was, as in man, smaller above (0.278 mm) than below (0.744 mm) the entry of the arteria radicularis magna. Resistance to flow, as calculated by Poiseuille's equation, was 51.7 times greater up the anterior spinal artery as compared with downthis artery. The vascular anatomy explains the absence of paraplegia in one baboon in the cross-clamp group and paraplegia in one baboon in the shunt group. Thus, distal aortic perf~ion protects the spinal cord below the arteria radicularis magna but not above it

Lars G. Svensson, M.B., B.Ch., M.Sc., F.R.C.S., F.C.S.(SA), Elizabeth Rickards, B.Sc.(Hon), Anne Coull, B.Sc., Geoffrey Rogers, M.Sc., Claus J. Fimmel, M.D., and Ronald A. Hinder, M.B., B.Ch., Ph.D., F.R.C.S., Johannesburg, South Africa

RraPlegia caused by distal hypotension is a disastrous complication following cross-clamping of the thoracic aorta. The incidence varies from 0.1% after repair of coarctation of the aorta I to 24%2 for repair of traumatic rupture of the aorta or repair of a thoracoabdominal aneurysm. Various methods of spinal cord protection have been suggested, including cardiopulmonary bypass,' hypothermia,' and ventricular or proximal aorta-to-distal aorta heparin-impregnated shunts.' Spinal blood flow with thoracic aortic cross-clamping or shunting has previously not been adequately studied. In a reviewof 596 traumatic ruptures of the aorta, the incidences of paraplegia following cardiopulmonary From the MRC University Circulation Research Unit, Departments of Surgery and Physiology, University of the Witwatersrand, Johannesburg, South Africa. Received for publication Dec. 27, 1984. Accepted for publication March 7, 1984. Address for reprints: Lars G Svensson, Department 'of Surgery, Medical School, York Road, Parktown, 2193, Johannesburg, South Africa.

bypass, shunts, and simple aortic cross-clamping without distal perfusion were 2.2%, 2.3%, and 5.8%, respectively," The mortality rates were 16.7%, 11.4%, and 5.8%, respectively, caused primarily by the hemorrhagic complications of heparinization during cardiopulmonary bypass or from the sites of insertion of the shunt. 6 Because of the higher mortality and sometimes greater morbidity, Crawford,' Delsakey," Najafi," and their associates have advocated that a distal perfusion technique should not be used. No satisfactory explanation exists as to why paraplegia occurs with distal perfusion techniques.' 10 There has been a discrepancy between the clinical studies and experimental studies in the reported incidence of paraplegia. The first reason for this is that patients who undergo repair of coarctation of the aorta or atherosclerotic aneurysms may have extensive collateral channels of blood supply to the distal aorta. The second reason is that species differences may be important. For example, in the dog the anterior spinal artery (ASA) is not a continuous vessel from the base of the brain to the distal spinal cord as it is in man. II 71

The Journal of Thoracic and Cardiovascular Surgery

7 2 Svensson et al.

Chacma baboons, Papio ursinus, were chosen as the experimental species because we had shown by angiography and anatomic dissections in these animals that the ASA is continuous." In addition, the distal spinal cord is supplied as it is in human beings by a large radicular artery, the arteria radicularis magna (ARM), also known as the artery of Adamkiewicz. We wished to determine, in a species with a spinal cord anatomy similar to that of man, whether an aorto-aortic shunt increases distal spinal cord blood flow and to relate the ARM and ASA gross anatomy to spinal cord blood flow and the hemodynamics. Methods Seventeen male adult Chacma baboons weighing 13.6 to 26.8 kg (mean 20.2 kg) were immobilized by ketamine hydrochloride (10 rug/kg) and maintained under general anesthesia by a continuous infusion of sodium thiopental* (Intraval sodium) by a Harvard pumpt (Model 944A) at a rate of 5 mg/min. An inflatable, cuffed, endotracheal tube was inserted and the animal was ventilated with a respirator (Harvard Model 607) at a tidal volume of 10 ml/kg. The minute volume was increased up to 150 ml/kg while the left lung was collapsed. Arterial and venous blood gases were analyzed every half hour (blood gas analyzer G 13, Micro Autocal pH:!:) Inhaled oxygen concentrations were controlled between 60 and 120 mm Hg by an Afrox M-70-00 gauge. Expired carbon dioxide concentrations, maintained between 30 and 40 mm Hg, were measured by the Anarad gas analyzer§ (Model PM-20) and the respiratory rates were altered accordingly. Periodic hyperventilation with an Ambu bag II was performed to prevent atelectasis and to expand the left lung. Blood hematocrit value, hemoglobin concentration, and oxygen saturation were checked every half hour (Haemofuge, Heraeus, Haematocrit reader, Hawksley, and Co-Oximeter ILl82 analyzerf). Ringer's lactate or normal saline solution warmed to 38 C was infused at a rate of 10 ml/kg/hr by an IVAC 630 volume pump (Model 631-EE) through a 16 G Jelco intravenous Teflon catheter~ inserted into the antecubital fossa. The animals were given 3,000 to 4,000 IV of heparin. Blood that accumulated in the left pleural 0

cavity during the operation was collected, heparinized, and reinfused. In addition, unrecoverable blood losses were replaced intravenously by an equivalent volume of Ringer's lactate solution. Serum acid-base and pH changes did not require correction except for 50 mmol of sodium bicarbonate, which was administered routinely after unclamping of the aorta. The rectal temperature was monitored by a YSI-tele-thermometer* (Model 43TI) and kept at 36° to 37 C by warming blankets. An electrocardiographic tracing (standard Lead on a Beckman oscilloscopef (Type EO-I 18) and recorded on an eight-channel Beckman Dynograph (Type R411, with a 9806A AC coupler and Beckman 5411 chart drive) was used to monitor cardiac electrical activity during the experiment. Polyethylene catheters were placed in the arch of the aorta via the right brachial artery, in the distal abdominal aorta via the right femoral artery, and in the right atrium via the left femoral vein. A Swan-Ganz thermodilution cardiac output catheterf (1.5 mI, 93A-131-7F) was inserted through the right femoral vein and guided under radiographic control (Siremobil 2S§) into a distal pulmonary artery. The catheters were connected to strain-gauge transducers (Gould P231D, P23Gb, P23AA, P23BBID. Pressures were monitored and recorded on the Beckman recorder with Type 9803 strain-gauge couplers. In the control animals, a pigtailed catheter for the injection of the radioactive microspheres was advanced into the left ventricle via the left femoral artery. In animals undergoing operations on the thoracic aorta, two left lateral thoracotomies were performed through the fourth and eighth intercostal spaces; these incisions provided access to the proximal and distal descending thoracic aorta and to the left atrium. A Millar Mikro-Tip pressure transducer catheter~ was then guided under radiographic control into the left ventricle from the left axillary artery by way of the arch of the aorta through the aortic valve. Once the radiograph and pressure tracing confirmed that the left ventricle had been entered, the catheter was gently withdrawn until ventricular ectopic activity ceased to be seen on the oscilloscope-monitored electrocardiogram. The rate of pressure change over the first time derivative 0

m

*Maybaker, Dagenham, Essex, United Kingdom.

*Yellow Springs Instrument Co., Yellow Springs, Ohio.

tHarvard Apparatus Co. Inc., Natick, Mass.

tBeckman Instruments, Inc., Fullerton, Calif.

:j:lnstrumentation Laboratories, Inc., Lexington, Mass.

:j:American Edwards Laboratories, Irvine, Calif.

§Anarad, Inc., Santa Barbara, Calif.

§Siemens Corp., Iselin, N. J.

II Ambu

II Gould

International, Glostrup, Denmark.

'l[Jelco Laboratories, Raritan, N. J.

Inc., Medical Products Div., Oxnard, Calif.

'l[Millar Instruments, Inc., Houston, Texas.

Volume 91

(dP/ dT derivative) was determined by a Beckman Type 9841 nystagmus velocity coupler and recorded on the Beckman Dynograph. A soft polyetheylene catheter for the injection of the microspheres was placed into the left atrium through the left auricle and ligated in place with a purse-string suture. The following pressures were recorded continuously: left ventricular end-diastolic and systolic, left and right atrial, aortic arch systolic, distal aortic systolic, and pulmonary artery. In addition, the left ventricular mean pressure and pulmonary capillary wedge pressures were regularly recorded at the intervals listed below. Baselinetracings were checked before each recording and standardized if necessary with a mercury Baumanometer.* Cardiac output was determined by an American Edwards cardiac output computer (Model 9510A). Injected boluses of normal saline were below a temperature of 3° C as measured by a finelycalibrated alcohol thermometer. The median of five cardiac output measurements was used. The following three experimental groups were formed. Control group. This group consisted of four control animals in which no thoracic operation was performed. Hemodynamic monitoring was performed every quarter hour. Spinal cord blood flow was determined by radioactive microspheres injected at the start of the experiment and 120 minutes later. Aortic cross-clamp group (AX). Five animals underwent thoracotomy and aortic cross-clamping. The aorta was cross-clamped for 60 minutes immediately beyond the left subclavian artery and immediately above the diaphragm. The occluded segment of aorta was not opened. The distal aorta was not perfused. Blood flow studies with radioactive microspheres were done after the thoracotomy and before aortic cross-clamping, 30 minutes after cross-clamping, after 60 minutes of crossclamping, and after the aorta had been unclamped for 30 minutes. Shunt group (SH). Seven baboons underwent crossclamping immediately distal to the left subclavian artery and above the diaphragm as described in the AX group; but in addition, a 7 rom internal diameter shunt was connected between a T piece inserted into the aorta at the left subclavian artery and immediately above the diaphragm. The T piece took up most of the lumen and was ligated in place by encircling the aorta with silk ties. Flow through the shunt was measured by a 16 mm Narcomatic electromagnetic flowmeter probe'[ situated *W. A. Baum Co., Copiague, N. Y. [Narco Scientific Div., Ft. Washington, Pa.

73

Thoracic aortic operations

Number 1 January, 1986

• Control proximal • 0------ ----0 Conrrot distal • Ax proximal ... t r - - - - - - 6 Ax distal • SH proximal • 0,-,-,- ,-,-0 SH distal SH group

,.;s.C.:_

Ax group /-

Shunt

XC

.: .:

180 160 140 0'

I

E E ~ :l en en

120 100

5- 80

Q>

c

o

Q>

~

60 40 20 0 B

A S A L

p x R e E 5

x SX

e He 253

0

0

xu

e or s H 6

N X e 5

U N X 3

0

0

Fig. 1. Mean aortic pressures in control animals, during cross-clamping alone (Ax group), and shunting (SH group). PRE, Pre-cross-clamping. XC. Cross-clamped time elapsed. UNXC. Uncross-clamped time elapsed. SH. Shunted time elapsed.

on a separate piece of homogenous ascending aorta in series with the shunt. The shunt was opened for 60 minutes, at which time it was clamped and the aorta was unclamped. Blood flow studies were carried out before cross-elamping, within 15 minutes of cross-clamping, after 60 minutes of shunting, and 30 minutes after unclamping the aorta. Measur ements . Mean pressures were recorded before thoracotomy, after thoracotomy, 5, 10, 20, 30, 45, and 60 minutes after aortic cross-clamping, and 5, 10, 20, and 30 minutes after unclamping of the aorta. In the SH group, mean pressure recordings were taken at the same intervals during the period of cross-clamping and shunting. Cardiac output measurements were performed prior to thoracotomy, after thoracotomy, 5, 30, and 60

The Journal of Thoracic and Cardiovascular Surgery

7 4 Svensson et al.

Table I. Cardiac index (CI), total peripheral resistance (TRP), and rate pressure product (RPP) (±SEM) in control, aortic cross-clamp (AX), and shunt (SH) groups Measurement Group

Time Basal

CI (L/min/nr) Control 1.8 (0.06) 2.40 (0.22) AX SH 2.40 (0.24) TPR (Jif dyne. sec . cmr'} 7.04 (0.75) Control AX 5.14 (0.61) SH 4.58 (0.39) RPP (10<) 2.57 (0.26) Control AX 2.31 (0.09) 1.80 (0.08) SH

Pre-XC

XC-30

1.89 (0.20) 2.22 (0.23)

1.39 (0.80) 1.70 (0.25)

4.81 (0.38) 4.51 (0.37)

6.22 (1.0) 8.13 (l.l)

1.64 (0.07) 1.72 (0.12)

1.68 (0.14) 1.90 (0.27)

SH-60

Un-XC5

Un-XC30

2.19 (0.30)

2.35 (0.46) 1.90 (0.29)

2.00 (0.13) 1.66 (0.20)

3.45 (0.47)

2.13 (0.64) 3.71 (0.48)

3.20 (0.45) 4.20 (0.45)

1.40(0.17)

1.07 (0.18) 1.27 (0.15)

1.24 (0.18) 1.24 (0.19)

TJ20 1.60 (0.07)

7.50 (0.70)

2.09 (0.17)

Legend: Pre-XC, Before cross-clamping. XC,Time aftercross-clamp. SH, Time aftershunt insertion. Un-XC, Time after unclamping. T120, Control value after 120 minutes.

minutes after cross-clamping of the aorta, and 5 and 30 minutes after unclamping of the aorta. In addition, animals with aorto-aortic shunts in situ had their cardiac outputs measured at 5, 30, 45, and 60 minutes after opening the shunt. Tissue blood flow was determined by means of 15 ~m radioactive microspheres labeled with 51Cr, 141Ce, 46Sc, and 103Ru.· Approximately 1 million spheres added to 5 ml of 10% mannitol and normal saline solution was mixed with an MSE 5-78 sonicator and injected into the left atrium at body temperature over 10 seconds. Arterial reference samples were withdrawn with a Harvard infusion-withdrawal pump (Model 944A) from the proximal aorta at a constant rate of 12 ml/min for a period of 80 seconds starting 5 seconds prior to the microsphere injections. The withdrawn blood was weighed and the volume calculated (weight X 1.04). Thirty minutes after unclamping of the aorta, a craniotomy was performed. The precentral motor cortex was stimulated with a platinum electrode connected to a dual stimulator type 3060A t and peripheral motor responses were noted. The motor responses were confirmed by electrical stimulation of the midbrain long tracts with a Model 755 blended coagulator (Bircteler). The baboons were killed with a 50 ml intravenous bolus of saturated potassium chloride. The descending thoracic aorta was occluded between vascular clamps and a 16 G Jelco catheter inserted into the occluded segment. Fifty milliliters of a mixture of liquid detergent, methylene blue, water, and microfined barium

sulfate with carbon dioxide" was then injected into the occluded segment over 5 minutes by gentle hand pressure on a 60 ml syringe. The spinal cord was then carefully removed using a Stryker saw.] The diameter of the distended spinal cord blood vessels was measured using a vernier scale. The diameter of the ARM and the ASA above and below the site of entry of the ARM was measured and the level noted. The spinal cord was sectioned into four cervical, eight thoracic, and four lumbar segments. The level of division between lower thoracic spinal cord and lumbar spinal cord was immediately above the site of entry of the ARM. Blood and tissue samples were placed in vials, dissolved with hot nitric acid, and counted in a well-scintillationgamma counter. The total activity and tissue blood flows were calculated on a Hewlett-Packard 85 computerj by the method described by Heyman and associates." The study was approved by the University of the Witwatersrand Senate Animal Ethics Committee. The data were stored on an Apple computer§ and analyzed on an IBM 4331 computed using the SAS package.1[ Results are expressed as means ± standard deviation/standard error of the mean. As the number of experiments in each group are small, p values are expressed with a prefix indicating which statistical test was used. T indicates Student's t test and W indicates

*New England Nuclear, North Billerica, Mass.

IIIBM Corp., Armonk, N. Y. ~SAS Institute, Inc., Cary, N. C.

tLectromed, Jersey Island, United Kingdom.

*Baritop, Noristan Group, Pretoria, South Africa. tStryker Corp., Kalamazoo, Mich. :j:Hewlett-Packard Co., Palo Alto, Calif. §Apple Computers, Cupertino, Calif.

Volume 91 Number 1 January. 1986

Table

Thoracic aorticoperations 7 5

n. Aorto-aortic shunt hemodynamics (±SEM)

~I------,-Time (minutes)

Flow (Lrrnin) Flow (nil/kg/min) Cardiac output

5

30

60

1.30 (.19) 56 2.19 (.25)

1.49 (.27) 64 2.16 (.34)

1.76 (.37) 75 2.28 (.36)

59

68

77

112 (II)

100 (7)

90 (6)

94 (II)

82 (5)

72 (5)

18

18

18

(Lrmin) Flow/ cardiac output (%) Proximal mean pressure Distal mean pressure Gradient (mm Hg)

the Wilcoxon test. Fisher's exact test was used to compare the rates of paraplegia. Results Hemodynamics.

Control group. The animals remained hemodynamically stable over the 120 minute period (Fig. 1). There was no change in the cardiac index (Tp = 0.06), total peripheral resistance (Tp = 0.07), or the rate pressure product (Table I).

Aorticcross-elamp and shunt groups. After thoracotomy, both the mean proximal and distal aortic blood pressures fell (Fig. 1). With the aorta cross-clamped, proximal aortic hypertension developed in both groups while the distal aortic pressure fell precipitously. After 30 minutes of aortic cross-clamping in the AX group and after 5 minutes of blood being shunted to the distal aorta in the SH group, there was a significant difference in the distal aortic pressures (Tp = 0.0002, Wp = 0.01) (Fig. 1). Immediately after the aorta was unclamped after 60 minutes of cross-clamping or shunting, the mean proximal pressure (Tp = 0.04, Wp = 0.03), distal aortic pressure (Tp = 0.01, Wp = 0.01), and total peripheral resistance (Tp = 0.02, Wp = 0.03) were lower in the AX group than in the SH group. Nevertheless, 5 minutes after the aorta was unclamped there was no significant difference in any of the hemodynamic parameters (mean proximal and distal pressures, cardiac index, total peripheral resistance, or rate pressure product) between the nonshunted (AX) and shunted (SH) groups at the end of the experiments (Table I). There was no significantdifference in the left ventricular end-diastolic pressure between the AX and SH groups (4 mm Hg after 60 minutes of cross-clampingin the AX group and 2.5 mm Hg after 60 minutes of shunting in the SH group). The mean time taken to insert the shunt was 21.4

~ T-12 A.R.M. •

~ ~I I~ Basal

Lower Thorocic

, Tp=0'2

[

Lumbar

'Tp=o-OOOg'

--.. X X CC

30 15

min min

X S

C H

6060

min min

UNXC 30min

Fig. 2. Segmental spinal cord blood flow in control, crossclamped (Ax), and shunted (SH) animals. VA, Vertebral artery. ASA, Anterior spinal artery. ARM, Arteria radicularis magna. XC, Cross-clamped time elapsed. SH60, 60 minutes of shunting. UNXC, Uncross-clamped for 30 minutes. Note that second blood flow study was performed 30 minutes after cross-clamping in the AX group and 15 minutes after cross-clamping in SH group, prior to opening shunt.

minutes (SEM ± 2.16). During this time the distal aorta was poorly perfused. Table II shows the hemodynamics of the shunt in the SH group. The proportion of the cardiac output shunted to the distal aorta increased from 59% at the start to 77% after 60 minutes of shunting. Similarly, the shunt mean blood flow increased from 56 ml/kg/min at the start to 75 ml/kg/min after 60 minutes of shunting. The mean gradient between the proximal and distal aorta was 18 mm Hg both at the beginning and at the end of 60 minutes of shunting. There was a positive correlation between the mean proximal pressure and the cervical spinal cord blood flow (r = 0.4, Tp = 0.1) in the AX group. Mean proximal arterial pressure correlated inversely with the lumbar spinal cord blood flow (r = -0.4, Tp = 0.05). In the AX group, the distal aortic pressure correlated with the upper thoracic (r = 0.54, Tp = 0.03), lower thoracic (r = 0.45, Tp = 0.07), and lumbar spinal cord (r = 0.58, Tp = 0.003) segmental blood flows. In the SH group the distal aortic pressure correlated with lumbar spinal cord blood flow (r = 0.58, Tp = 0.008) but not with the lower thoracic spinal cord blood flow (r = 0.11, Tp = 0.64).

The Journal of Thoracic and Cardiovascular Surgery

7 6 Svensson et al.

Fig. 3. Lower spinal cord vascular anatomy shows mean diameters (in millimeters). 0.688 = Mean diameter of ARM; 0.278 = mean diameter of ASA above ARM; 0.744 = mean diameter of the ASA below ARM. Spinal cord blood ftow. Fig. 2 shows the spinal cord segmental blood flows. Basal blood flows were similar in all groups. No significant fall in segmental spinal cord blood flow occurred in the control group. Increased blood flow occurred in the AX group at the end of the experiments. This was most marked in the lumbar spinal cord segment in the paralyzed baboons (Tp = 0.02). The lower thoracic spinal cord blood flow in the SH group did not increase once the shunt was established (Tp = 0.2); however, a highly significant increase in lumbar spinal cord blood flow occurred (Tp = 0.00(9). Paraplegia. Paraplegia occurred in none of the baboons in the control group, in four of five in the AX group, and in one of seven in the SH group (p = 0.044). In the AX baboon without paraplegia, the mean lumbar blood flow was 7.4 rnl/IOO gm/min compared with 1.2 rnl/l00 gm/min in the paraplegic baboons. In the SH baboon with paraplegia, the mean basal thoracic spinal cord blood flow was 21.5 rnl/l00 gm/rnin and dropped to 4.9 rnl/l00 gm/rnin with distal aortic perfusion compared with 8.6 SD ± 2.4 rnl/IOO gm/min in nonparaplegic baboons. The mean lumbar spinal cord blood flow during shunting in the paraplegic baboon was 26.8 rnl/l00 gm/min, Anatomy. The ARM entered the ASA in equal frequencies from the right and left sides. The most common level of entry was the twelfth thoracic vertebra (N = 7) followed by the eleventh thoracic vertebra (N = 5), with a range from the tenth thoracic to the third lumbar vertebra. No ARM arose from the excluded segment of aorta; in all SH animals the ARM was perfused by the shunt. The mean diameter of the ARM

was 0.688 mm (SEM 0.066) (Fig. 3). The mean diameter of the ASA above the entry of the ARM was only 0.278 mm (SEM 0.035), whereas it was 0.744 mm (SEM 0.074) below the entry (Tp = 0.0001, Wp = 0.00(3). Discussion Segmental spinal cord blood flow with aortic crossclamping and shunts and its relationship to the vascular anatomy of the spinal cord have not previously been studied. The only studies of spinal cord blood flow are those reported by Gelman and colleagues," who studied the upper and lower spinal cord blood flow in crossclamped dogs and the effect of sodium nitroprusside, and by Laschinger and associates," who studied spinal cord blood flow in relation to distal aortic blood pressures during atriofemoral bypass. We determined how an aorto-aortic shunt influenced spinal cord blood flow and related the findings to the vascular anatomy of the spinal cord. The most striking finding in our study is that a shunt does not increase the lower thoracic spinal cord blood flow immediately above the ARM, despite the ARM being perfused by the shunt. Distal perfusion of the aorta with a shunt markedly increases lumbar spinal cord blood flow below the level of entry of the ARM. Furthermore, distal aortic pressure correlates well with lumbar spinal cord blood flow but does not correlate with lower thoracic spinal cord blood flow. This can be explained by the interesting vascular anatomy of the ASA with a mean diameter above the site of entry of the ARM of 0.278 mm compared with 0.744 mm below the entry. By substitution in Poiseuille's equation

Volume 91 Number 1 January. 1986

Thoracic aortic operations 7 7

Q = 1lT" (PI - P2)/8L17

resistance is found to be inversely proportional to the fourth power of the radius. Thus, the resistance to blood flow up the ASA is 51.7 times greater than blood flow down the ASA. Distal perfusion of the aorta thus providesadequate lumbar spinal cord protection, but the arrangement of the vascular anatomy leaves the lower thoracic spinal cord at risk from ischemia. That paraplegia occurred in only one of the seven SH baboons indicates that distal perfusion of the aorta is not entirely protective. This is best explained by the vascular anatomy, which allows little of the shunted blood access to the lower thoracic spinalcord. The paraplegic SH baboon had an inadequate thoracic spinal cord blood flow despite an adequate lumbar spinal cord blood flow during distal perfusion. The lower thoracic spinal cord therefore seems to be mainly dependent for its oxygen supply on blood flowing down the ASA from the cervical and upper thoracic spinal segments. This is supported by the work of Fried, Di Chiro, and Doppmann," who found that ligation of the ASA above the ARM usually did not cause paraplegia in monkeys, but ligation below the ARM usually produced paraplegia. Analysisof diagrams of the blood supply to the spinal cord shows that the same discrepency in diameter of the ASA above and below the ARM exists in man." The mean diameter above and below were 0.283 mm and 0.508 mm, respectively. These measurements were not taken in the immediate vicinity of the ARM and were from undistended postmortem specimens. Thus, the resistance to upward flow is at least 11.3 times greater in

man. The findings of paraplegia in four of five AX baboons and a lowmean lumbar spinal cord blood flow at the end of cross-clampingsuggest that the lumbar spinal cord is the most sensitive area to ischemia in the AX baboons. The concept that the lumbar spinal cord is the most sensitive area to ischemia is further supported by the finding that the AX baboon that did not develop paraplegia had a greater lumbar spinal cord blood flow, probably because of the larger ASA present in this baboon. The increase in proximal blood pressure correlated with an increase in the cervical cord blood flow. Thus the AX baboons, with higher proximal pressures than the SH baboons, probably had better protection to the lower thoracic spinal cord by virtue of our hypothesis that most of the lower thoracic spinal cord blood flow comes from the upper spinal cord radicular arteries. Furthermore, Gelman and colleagues" showed that when proximal hypertension is decreased by sodium nitroprusside, distal aortic perfusion of tissues also

decreases. The hyperemic response demonstrated at the termination of the experiments in the AX group may have been caused by vasodilation after severe tissue ischaemia during cross-clamping. It may be asked if distal perfusion rates by way of the shunts were adequate because the flow rate required for spinal cord protection is unknown." We achieved mean shunt flow rates of 56 ml/kg/min at the start, which increased to 75 ml/kg/min by the end of the period of shunting with a mean distal pressure of 72 mm hg. This accounted for 77% of the cardiac output, and the mean gradient across the shunt was 18 mm Hg at the end of shunting; therefore, we believe that the flow rates through our shunts were adequate. In the clinical situation such high flow rates per kilogram may not be achieved, because a relatively smaller shunt, usually of a 7.5 or 9 mm internal diameter, must be used. This may account for the discrepancy between SH animals and the clinical situation, where there is usually a significant gradient between the proximal and distal aorta. As shown by Laschinger and colleagues," a distal perfusion pressure greater than 40 mm Hg appears to be important for spinal cord protection. We found a shunt to be most effective in preventing paraplegia during cross-clamping. Laschinger and coworkers" showed in six dogs that varying the distal aortic pressure between 40 and 100 mm Hg results in no significant difference in spinal cord blood flow. Only at a distal perfusion pressure of 40 mm Hg did somatosensory-evoked potentials disappear. They also reported on patients who had a shunt or femorofemoral bypass pump method of distal perfusion of the distal aorta." The distal clamp was situated at the level of the sixth to eleventh thoracic vertebrae in two of the patients whose somatosensory-evoked potentials disappeared. Because the ARM enters at about this level,the ARM may have been occluded in the one patient who developed paraplegia despite distal perfusion. Neither of the two patients with distal aortic pressures greater than 60 mm Hg without distally assisted perfusion developed paraplegia. Others have reported that a distal perfusion technique does not always protect against paraplegia and may even increase the rate of paraplegia.": 18, 19 Distal perfusion techniques with a flow rate greater than 60 ml/kg/rnin have been suggested by Hug and Taber" to unload the left ventricle. They showed that 20 to 30 ml/kg/min was adequate for renal protection. Our shunt mean flow rates of 56 to 75 ml/kg/min adequately decompressed the left ventricle. Roberts and colleagues" showed that distal perfusion at a rate of 33 to 40 ml/kg/min has no significant advantage for cardiac and renal preservation during aortic cross-clamping. We

7 8 Svensson et al.

confirmed the findings of Roberts and colleagues" that there was an insignificant difference in the left ventricular end-diastolic pressure between the AX and the SH groups of animals both at the beginning and at the end of the cross-clamp and shunt periods. Despite the increase in the total peripheral resistance during crossclamping in the AX group, there was not a significantly smaller cardiac index or greater work performed (rate pressure product) than in the SH group. More sensitive tests of left ventricular function, such as integration of the area under the left ventricular ejection curve, might have detected a difference. Our hemodynamic studies were unable to detect any significant hemodynamic advantage of distal shunting. We conclude from these results that, during thoracic aortic cross-clamping without distal perfusion, the lumbar spinal cord is the most sensitive area to ischemia. A distal shunt increases lumbar spinal cord blood flow but does not alter thoracic spinal cord blood flow protection during prolonged thoracic aortic cross-clamping. This is because of the vascular anatomy at this level. There were no hemodynamic advantages in the SH group of animals. We believe that for a shunt to be effective, the ARM must be adequately perfused, but even this does not guarantee a good blood flow to the lower thoracic spinal cord, and paraplegia may therefore still ensue. We are grateful to Professor H. T. Groenefeldt, Institute for Biostatistics, South African Medical Research Council, for assistance with the statistical analysis.

2

3 4

5 6

REFERENCES Brewer LA, Fosburg RG, Mulder GA, Verska JJ: Spinal cord complications following surgery for coarctation of the aorta. 1 THORAC CARDIOVASC SURG 64:368-379, 1972 Katz NM, Blackstone EH, Kirklin rw, Karp RB: Incremental risk factors for spinal cord injury following operation for acute traumatic aortic transection. 1 THORAC CARDIOVASC SURG 81:669-674, 1981 Stallone Rl, Iverson LIG, Young IN: Descending thoracic aortic aneurysm. Am 1 Surg 142:106-108, 1981 Coles lG, Gregory lW, Sima AF, Klement P, Tait GA, Williams WG, Baird Rl: Intraoperative management of thoracic aortic aneurysm. 1 THORAC CARDIOVASC SURG 85:292-299, 1983 Gott VL: Heparinized shunts for thoracic vascular operations. 1 THORAC CARDIOVASC SURG 14:219-220, 1972 Svensson LG, Antunes Ml, Kinsley RH: Traumatic rupture of the thoracic aorta-A report of 14 cases and a review of the literature. S Afr Med 1 67:853-857, 1985

The Journal of Thoracic and Cardiovascular Surgery

7 Crawford ES, Rubio PA: Reappraisal of adjuncts to avoid ischemia in the treatment of aneurysms of descending thoracic aorta. 1 THORAC CARDIOVASC SURG 66:693-704, 1973 8 DeBakey ME, McCollum CH, Graham 1M: Surgical treatment of aneurysms of the descending thoracic aorta. 1 Cardiovasc Surg 19:571-576, 1978 9 Najafi H, Javid H, Hunter 1, Serry C, Monson D: Descending aortic aneurysmectomy without adjuncts to avoid ischemia. Ann Thorac Surg 30:326-335, 1980 10 Symbas PN: Review of Roberts et al" 11 Di Chiro G, Fried-LC, Doppman lL: Experimental spinal cord angiography. Br 1 Radiol 43:19-30, 1970 12 Svensson LG, Hinder RA: Spinal cord anatomy of the baboon. S Afr 1 Surg (in press) 13 Heyman MA, Payne BD, Hoffman lIE, Rudolph AM: Blood flow measurements with radionuclide-labeled particles. Prog Cardiovasc Dis 20:55-79, 1977 14 Gelman S, Reves lG, Fowler K, Samuelson PN, Lell WA, Smith LR: Regional blood flow during cross-clamping of the thoracic aorta and infusion of sodium nitroprusside. 1 THORAC CARDIOVASC SURG 85:287-291, 1983 15 Laschinger lC, Cunningham lN, Nathan 1M, Knopp EA, Cooper MM, Spencer FC: Experimental and clinical assessment of the adequacy of partial bypass in maintenance of spinal cord blood flow during operations on the thoracic aorta. Ann Thorac Surg 36:417-426, 1983 16 Fried LC, Di Chiro G, Doppman 1L: Ligation of major thoraco-Iumbar spinal cord arteries in monkeys. 1 Neurosurg 31:608-614, 1969 17 Domisse GF: The blood supply of the spinal cord. 1 Bone Joint Surg 56B:225-235, 1974 18 Young, in Laschinger lC, Cunningham IN, Nathan 1M, Knopp EA, Cooper MM, Spencer FC: Experimental and clinical assessment of the adequacy of partial bypass in maintenance of spinal cord blood flow during operations on the thoracic aorta. Ann Thorac Surg 36:417-426, 1983 19 Appelbaum A, Karp RB, Kirklin lW: Surgical treatment for closed thoracic aortic injuries. 1 THORAC CARDIOVxsc SURG 71:458-460, 1976 20 Hug HR, Taber RE: Bypass flow requirements during thoracic aneurysmectomy with particular attention to the prevention of left heart failure. 1 THORAC CARDIOVxsc SURG 57:203-213, 1969 21 Roberts Al, Nora 1D, Hughes WA, Quintanilla AP, Ganote CE, Sanders lH, Moran 1M, Michaelis LL: Cardiac and renal responses to cross-clamping of the descending thoracic aorta. 1 THORAC CARDIOVASC SURG 86:732-741,1983