A prospective randomized study of cerebrospinal fluid drainage to prevent paraplegia after high-risk surgery on the thoracoabdominal aorta

Transcription

1 A prospective randomized study of cerebrospinal fluid drainage to prevent paraplegia after high-risk surgery on the thoracoabdominal aorta E. Stanley Crawford, MD, Lars G. Svensson, M~B, Phi), Kenneth R. Hess, MS, Salwa S. Shenaq, M_D, Joseph S. CoseUi, MD, Hazim J. Sail, MD, Prita K. Mohindra, MD, and Victor Rivera, MD, Houston, Texas This article is concerned with the study of the effect of several variables, principally that of cerebrospina/fluid drainage, on the incidence of neurologic deficit in a prospective randomized series of patients with extensive aneurysms of the descending thoracic and abdominal aorta (thoracoabdominal type I and II). Forty-six patients had cerebrospinal fluid drainage, and 52 were controls, with a total of 98 available for study. Cerebrospinal fluid pressure was continuously monitored in the former group and pressure maintained -<10 mm Hg in 20, -<15 mm Hg in 20, and >15 mm Hg in 6 patients during period of aortic damping. The method of treatment including reattachment of intercostal and lumbar arteries (p = 0.2), temporary atriofemoral bypass during aortic occlusion (p = 0.3), and spinal fluid drainage (p = 0.8) were not statistically significant in reducing the incidence of neurologic deficits. Thus cerebrospinal fluid drainage as we used it, was not beneficial in preventing paraplegia. On appropriate statistical analysis we found that the only significant predictor of delayed deficits was postoperative hypotension (p = 0.006). (J VAsc SURG 1990;13:36-46.) We have used various methods to prevent paraplegia from occurring during operations for descending and thoracoabdominal aortic aneurysm. These include surface hypothermia, cardiopulmonary bypass with and without profound hypothermia, temporary atriofemoral pump bypass with and without somatosensory evoked potential monitoring, reattachment of intercostal and lumbar arteries, and use of various drugs such as steroids, oxygen radical scavengers, mannitol, barbiturates, and calcium channel blockers. We have not been able to demonstrate that any of these methods are superior to simple crossclamping and proximal blood pressure control in prevention of this complication in treatment of 584 patients with aneurysm of the descending thoracic aorta and 1062 patients with thoracoabdominal aortic aneurysms between June 1960 and September s From the Departments of Anesthesia, Neurology, and Surgery, Baylor College of Medicine and The Methodist Hospital, Houston. Presented at the Forty-fourth Annual Meeting of the Society for Vascular Surgery, Los Angeles, Calif., June 4-6, Reprint requests: E. Stanley Crawford, MD, Department of Surgery, Baylor College of Medicine, One Baylor Plaza, Houston, TX /6/ Cerebrospinal fluid drainage (CSFD) (cisterna magna) has been shown repeatedly to reduce the incidence of paraplegia during long periods of simple aortic cross-clamping in the dog? s However, CSFD in the pig during aortic cross-clamping has been shown not to reduce the incidence of paraplegia. 9 Lumbar CSFD alone during aortic cross-clamping experiments in the baboon has not been accompanied by a reduction in the incidence of paraplegia even though the circulation of the lumbar cord was better maintained.i Similar experiments in the baboon adding intrathecal papaverine or temporary aortic bypass did reduce the incidence of paraplegia, l Cerebrospinal fluid drainage was performed through laminectomy exposure of the dura in these pig and baboon experiments. Arterial spasm associated with this method of exposure may account for the difference in the results, dog versus pig and baboon. Thus the differences between species in these cases should be interpreted with caution. No experiment has been reported either in the dog, the baboon, pig, or other animal that duplicates the human situation consisting of excision and graft replacement with permanent interruption of some or all of the intercostal and lumbar arteries or blind reattachment of these vessels during a long (30 to

2 Volume 13 Number 1 Januaq, 1991 Cerebrospinal fluid drainage and paraplegia after high-risk aortic surgery 37 III IV Fig. 1. Drawings illustrate the classification of thoracoabdominal aortic aneurysms used in this study. 60 minutes) period of aortic cross-clamping. Regardless, CSFD with and without intrathecal papaverine injection is being used with increasing frequency during aortic operations in patients for treatment of aneurysms of varying extents and with variable risks of paraplegia (3% to 40%). The authors of these studies report "encouraging results," or 95% to 100% successful results in prevention of paraplegia. 67.H-~s These observations warrant a more extensive controlled evaluation in the human situation. Our institutional review board would not give unanimous approval for the use of intrathecal papaverine in 1987 and again in 1988 but did approve lumbar CSFD Aug. 16, 1988, provided that drainage was limited to 50 ml in excess of that spilled at time of insertion of the cerebrospinal fluid (CSF) catheter, in a prospective randomized study with sufficient statistical power in a series of patients who had the highest risk of paraplegia. The purpose of this article is to present the results of this study performed during the period Sept. 12, 1988, to Oct. 31, METHODS A review of historical data up to Sept. 12, 1988, in 1062 patients treated for thoracoabdominal aortic aneurysm revealed the expected incidence of lower limb neurologic events that vary according to extent Table I. Treatment of thoracoabdominal aortic aneurysms, June 20, 1960, to Sept. 11, 1988 No. of Extent etiology patients Paraparesis Paraplegia TAA I Nondissection (6%) 12 (6%) Dissection 41 4 (10%) 4 (10%) TAA II Nondissection (12%) 30 (15%) Dissection (16%) 18 (20%) TAA III Nondissection (2%) 6 (3%) Dissection 25 2 (8%) 2 (8%) TAA IV Nondissection (1%) 5 (2%) Dissection 11 0 (0%) 1 (9%) Total (6%) 78 (7%) Twenty-five patients excluded, 11 operative deaths, 14 preoperative neuromuscular deficits. and cause of aneurysm (Table I, Fig. 1). Patients with involvement of most or all of the descending thoracic and abdominal aorta down to renal arteries were classified as group I; most or all of the descending thoracic and most or all of the abdominal aorta, group II; half or less of the descending thoracic aorta and most or all of the abdominal aorta, group III; and upper half or all the abdominal aorta, group IV.

3 38 Crawford et al. Journal of VASCULAR SURGERY Incldon~) In ContrOl Group 5% 50O0 J- 5O0O 1000 o Z I I I I I 0 0% 5% 10% 15% 20% 25% 30% Incidence In Treatment Group Fig. 2. Nomogram for calculating minimal sample size for a statistically valid study with a dichotomous endpoint (alpha = 0.05, beta = 0.2, power = 80%, based on Pearson chi-square test). Moreover, nearly 20% of patients in group I and II had previous or concurrent aneurysmal disease of the ascending and transverse aortic arch indicating that these two groups had truly extensive disease. The neurologic deficits in the historical series were either transient or permanent and were present at the time of recovery from anesthesia or appeared at later intervals up to 4 years after operation. The incidence of neurologic deficits after operation for both dissection and nondissection according to extent of disease is shown in Table I. To select a group of patients that would provide results with adequate statistical power and that would be available in a reasonable period of time, certain assumptions were required. The first was an event rate of 25% in controls and a 5% rate in patients with CSFD. These numbers required 98 group I and II patients for adequate statistical power (alpha = 0.05, beta = 0.2, power = 80%, based on Pearson chi-square test) (Fig. 2). Such a series of patients could be obtained within I year if the study included patients with both nondissection and dissection aneurysms (Table II). Limiting the study to groups III and IV or a mixture of these as in the cited previous reports would require up to 568 patients by the same calculations assuming 5% neurologic deficits in controls and 1% in the treated group (Fig. 2). CLINICAL MATERIAL Graft replacement was used in treatment of 210 patients with aneurysms of varying extent during the study period (Table II). Of these, 148 (70%) were patients with aneurysms of group I and II extent and were potential candidates for CSFD study. Fortyeight patients were excluded because of rupture, emergency operation for symptoms, previous brain or cord disease, bleeding dyscrasias (including chronic warfarin [Coumadin] therapy), previous low back operation or lumbosacral disk disease, fever, infection or mycotic aneurysms, refusal to give informed consent, heparinization for total cardiopulmonary bypass, profound hypothermia, and brachiocephalic arrest of circulation. Therefore 100 patients, 13 with aneurysms of the ascending and transverse aortic arch and aortic valve insufficiency to be submitted to elective operation, gave written fully informed consent on the day before operation for the operation with CSFD. Thirty-three of these patients had aortic dissection, and 67 had nondissection disease. These patients were subsequently block randomized prospectively to either the control group or the CSFD group by computer-generated sequence in opaque, sealed envelopes immediately before surgery. Randomization was stratified according to extent expected to be resected (group I or II) and cause

4 Volume 13 Number 1 Jmuary 1991 Cerebrospinal fluid drainage and para_plegh~ after high-risk aortic surgery 39 Table II. Thoracoabdominal aortic aneurysms according to treatment, Sept. 12, 1988 to Oct. 31, 1989 (210 patients) Randomized (CSFD) (No CSFD) Not randomized No. No. No. Etiology extent patients PAR PLG patients PAR PLG patients PAR PLG TAA I Nondissection 15 2 (13%) 1 (7%) 18 1 (6%) 1 (6%) 9 1 (11%) 0 (0%) Dissection 4 1 (25%) 0 (0%) 7 2 (29%) 0 (0%) 6 2 (33%) 0 (0%) TAA II Nondissection 16 5 (31%) 1 (6%) 18 4 (24%) 4 (19%) 23 3 (13%) 1 (4%) Dissection 11 3 (27%) 1 (9%) 9 4 (29%) 1 (14%) 8 0 (0%) 0 (0%) TAA III Nondissection (6%) 0 (0%) Dissection (0%) 0 (0%) TAA IV Nondisscction (7%) 0 (0%) Dissection Total 46 *~ 11 (24%) 3 (7%) (21%) 6 (12%) 112 ~*~ 10 (9%) 1 (1%) PAR, Paraparesis; PLG, paraplegia. **One patient who died in the operating room is omitted. ***Three patients with preoperative neuromuscular deficits are excluded. (nondissection or dissection) predicted from CT scanning or aortography or both. Thus the resulting treatment intended by randomization was CSFD in 51 and no drainage in 49 and aortic reconstructive techniques, including extensive intercostal and lumbar arterial reattachment when possible and atriofemoral pump bypass when indicated as previously reported.~3'~ After randomizing on the day of operation, four patients to receive CSFD were disqualified for this form of treatment; two withdrew consent for CSFD--fever and urinary tract infection developed in one overnight, and the intrathecal catheter could not be inserted in the other patient. Operation was performed in these four patients without CSFD, and thus they were included in the control group. One patient in the control group with concurrent aneurysm of the ascending aorta and transverse aortic arch and aortic valve insufficiency had acute pulmonary edema from heart failure during induction of anesthesia. Operation was aborted at this time, and 4 days later he had successful aortic valve, ascending aorta, and transverse aortic arch replacement with composite valve graft. He is currently waiting for distal aneurysm replacement. Thus 47 patients had CSFD, and 52 were controls. One patient receiving CSFD died of myocardial infarction at the end of operation, and thus 46 patients with CSFD and 52 control patients survived long enough to evaluate neurologic complications, 100% for 15 days, 97% at 30 days, 93% at 90 days, and 89% at 180 days. The time and causes of death are shown in Table III. CEREBROSPINAL FLUID DRAINAGE A catheter was introduced by lumbar puncture by use of local anesthesia at L4-5 or L3-4. It was advanced 5 to 10 cm within the intrathecal space by means of a guidewire as previously described? 3 The cerebrospinal fluid pressure (CSFP) was continuously monitored throughout the operation, but only initial levels and those observed every 15 minutes were recorded in the protocol to be transferred to computer. After complete anesthesia and when the left lung was collapsed, 20 ml of CSF was withdrawn. After aortic clamping, an attempt was made to maintain CSFP between 10 and 15 mm Hg (as in the Mayo Clinic protocol) by withdrawing additional fluid. 17 Our protocol differed by maintaining normothermia and being limited to a withdrawal of a maximum of 50 ml in excess of loss during insertion of the catheter, which was exceeded in only one patient (120 ml). ~3 The volumes removed in our patients, including an estimate of loss during insertion of catheter, varied from 24 to 120 ml (median volume, 52.5 ml) (Fig. 3). The CSFP was consistently maintained at 10 mm Hg or less in 20, 15 mm Hg or less but higher than 10 mm Hg in 20, and consistently greater than 15 mm Hg in 6 surviving patients. The catheter was removed at the end of operation before transfer to the intensive care unit. All pertinent preoperative, operative, technical,

5 40 Crawford et al. Journal of VASCULAR SURGERY Table III. Randomized patients (N = 99), Cause of death (N = 11) from operation to last follow-up (17 mo.) Method Protocol of Days No. Cause of death treatment Deficit postoperative 11 Myocardial infarction CSFD Unknown Control No Respiratory. and renal failure Control Yes Multiple organ failure Control Yes Multiple organ failure Control Yes 41 6 Multiple organ failure CSFD Yes Multiple organ failure CSFD No 62 7 Rupture proximal aneurysm Control Yes Multiple organ failure Control Yes Multiple organ failure Control Yes Cardiac, stroke CSFD No 152 Table IV. Neuromuscular deficit according to treatment received (98 patients) Drainage* Control** p value*** Immediate 10/46 (22%) 11/52 (21%) 0.9 Delayed~ 4/36 (11%) 6/41 (15%) Total 14/46 (30%) 17/52 (33%) 0.8 *One patient who died in the operating room is omitted. **One patient whose operation was cancelled is omitted. ***Pearson chi-square test. TTwenw-one patients with immediate deficits excluded. and postoperative hemodynamic data were collected prospectively on a protocol form and then collated for entry onto computer for statistical analysis. Neuromuscular function was graded every morning after surgery, and all neuromuscular deficits were documented. When such occurred, a neurologist examined the patient and graded the deficit without knowledge of the intraoperative treatment. Neurologic deficits noted at the time of recovery from anesthesia were classified as immediate and those occurring after being normal for 3, 4, 5, 7, and 22 days were classified as delayed (Tables IV to VI). The degree of paralysis was graded and reported separately for the immediate and delayed groups at the time of occurrence, discharge or death, follow-up or death (90 days), survivors at last follow-up and finally, the cumulative survival and neurologic status both immediate and late at last follow-up (Tables V, VI). Neurologic deficits either unilaterally or bilaterally were scored as follows: minimal or no motion, 1; motion but not against resistance or gravity, 2; motion against resistance and gravity but unable to stand or walk, 3; able to stand and walk with assis- tance, 4. Those with scores of 1 to 2 were considered paraplegic and those with 3 to 4 to have paraparesis. STATISTICAL ANALYSIS The data entered on the computer were analyzed by use of the BMDP statistical package. Preoperative, etiologic, technical, and postoperative variables were compared by Pearson's univariate chi-square test for balance between the control and treatment groups (Table VII), and for their effect on the occurrence of neuromuscular deficits occurring within 30 days after surgery (Table VIII). Multiple logistic regression was used to determine the independent determinants of neuromuscular deficits occurring within 30 days after surgery. The distributions of neuromuscular scores were compared between the treatment and control group by use of the Wilcoxon ranksum test. The sample size calculations were made with standard formulas, 18 Kaplan-Meier estimation was used to calculate survival probabilities during the follow-up period. Simple linear regression and correlation were used to analyze the relationships between CSFP, central venous pressure, and mean arterial pressure. RESULTS OF SPINAL FLUID DRAINAGE No difference was found in the incidence or degree of neurologic deficits that occurred in these patients irrespective of whether data were analyzed according to treatment assigned or treatment received (Tables IV to VI) and regardless of CSFP levels maintained during operation. Neurologic deficits occurred in the patient with the highest CSFP as well as in one of the patients who had the lowest CSFP from whom the largest volume (120 ml) of CSF was

6 Volume 13 Number 1 January 1991 Cerebrospinal fluid drainage and paraplegia after high-risk aortic surgery Number of m lowest volume: 24ml median volume: 52.5ml largest volume: 120ml Patients 6 s I ' 0 i I I I I I III Total CSF volume drained (ml) Fig. 3. Histogram of total CSF volume drained in 47 patients. Table V. Immediate neurologic deficits (drainage deficits = 10, control deficits = 11) Total patients p value* Score at onset Score CSFD Control Score at discharge/death Score CSFD Control Cumulative score at follow-up/death** Score CSFD Control Score of survivors on last follow-up** Score CSFD Control *Wilcoxon rank-sum test comparing the distributions of scores between treatment groups (CSFD vs Control). **Follow-up ranged from 86 days to 17 months. removed. Of the six patients whose spinal cord function was intact who required reoperation 9 hours to 7 days after operation for bleeding, to evacuate clotted hemothorax, or other organ injury, neurologic deficits occurred in none (0/4) of the CSFD group and in all (2/2) of the controls (p = 0.07, Fisher exact test). In addition, hypotension (less than 100 mm Hg systolic) occurred in 31 patients who did not have immediate deficits. Of these 31 patients 2/16 (12%) CSFD and 6/15 (40%) controls (p = 0.08) developed neurologic deficits after periods of hypotension that occurred 3 to 22 days after operation. Those events causing hypotension in patients who developed neurologic deficits were due to cardiac arrhythmias in four, temporary respiratory failure and cardiac arrest in one, low cardiac output from chronic coronary artery disease in one, dehydration and hypovolemia from excessive diuresis in one, myo-

7 42 Crawford et al. 7ourn~ o r VASCU LAB SI.'RGER'~ Table VI. Delayed neurologic deficits (Scores: drainage deficits = 4, control deficits = 6) Score at onset Total patients p value ~ CSFD Control Score at discharge or death CSFD 36 1" Control 41 2" 0 1" Score of survivors on last follow-up CSFD Control Patients with immediate deficits excluded. *Died in the hospital **Wilcoxon rank-sum test comparing the distributions of scores between treatment groups (CSFD vs Control). Table VII. Balance of variables between treatment groups (CSFD study, N = 98) Drainage Control Variable (N = 50) (N = 48) p value* Extent II re- 29 (58%) 25 (52%) 0.6 placed Aortic dissection 17 (34%) 14 (29%) 0.6 Females 19 (38%) 19 (40%) 0.9 Age ->65 yr 25 (50%) 26 (54%) 0.7 Preoperative high 37 (74%) 36 (75%) 0.9 blood pressure Atriofemoral 21 (42%) 18 (38%) 0.6 bypass Mannitol 37 (74%) 40 (83%) 0.3 Total clamp 26 (52%) 22 (46%) 0.5 time ->45 min. Lumbar and in- 40 (80%) 38 (79%) 0.9 tercostal reattachment Postoperative re- 5 (10%) 4 (8%) 0.8 operation Postoperative by- 17 (34%) 14 (29%) 0.6 potension *Pearson chi-square test. cardial infarction in one, and reoperation in two patients. These patients in whom neurologic deficits occurred from 3 to 22 days after operation were considered to have delayed deficits: 4/36 (11%) CSFD, and 6/41 (15%) control (p = 0.6). These events were more pronounced in the controls and occurred late and presumably after operative CSFP had returned to normal. Operative CSFD almost certainly had no beneficial effects in these cases. The randomizing process led to an excellent balance or match of the variables considered possibly significant in the development of lower extremity neurologic deficits (Table VII). Univariate analysis indicated no protective effect regardless of variable tested (Table VIII). CUMULATIVE RESULTS The 30-day survival of all 210 patients treated by operation during the study period was (204/210) 97%, and the Kaplan-Meier survival was 93% at 3 months, 90% at 6 months, and 86% at 1 year. The incidence of immediate neurologic deficits in the 210 patients is shown in Table II. The survival rates in the randomized patients at 30 days (95/99), 60 days (93/99), and 90 days (91/99) in percentage were 96%, 94%, and 92%, respectively. Most of the deaths occurring 0 to 152 days after operation in the latter group of patients were due to cardiac complications and multiple organ failure, which were related to operation (Table III). NEUROLOGIC EVENTS IN RANDOMIZED PATIENTS Neurologic deficits in the lower extremities of varying intensity and duration occurred in 31 (32%) of the 98 patients in the randomized series. The deficit was observed immediately (within 24 hours) of the operation in 21 (68%) and from 3 to 22 days (delayed) in 10 (32%) of those affected. The deficit was severe (paraplegia) at onset in 21 and milder (paraparesis) in 10. By the time of death or hospital discharge, 9 patients had fully recovered; however, 22 (22%) patients continued to have deficits, 15 had paraplegia, and 7 had paraparesis. Of the 90 (93%) patients surviving 90 days 6 (7%) were paraplegic and 6 (7%) had paraparesis, 4 of whom were able to walk with assistance (cane, walker, brace). No significant difference was found between treated and control groups. Univariate predictors of neurologic

8 Volume 13 Number 1 January, 1991 Cerebrospinal fluid drainage and paraplegia after high-risk aortic surgery 43 Table VIII. 30-day neuromuscular deficits according to treatment received (Drainage and Control) in subgroups of patients Drainage Control Group (N = 46) (N = 52) p value*** All patients 14/46 (30%) 17/52 (33%) 0.8 Extent replaced I 4/19 (21%) 4/25 (16%) 0.7 Extent replaced II 10/27 (37%) 13/27 (48%) 0.4 Dissection 5/15 (33%) 7/16 (44%) 0.6 Nondissection 9/31 (29%) 10/36 (28%) 0.9 Age <65 yr 5/23 (22%) 6/28 (21%) 1.0 Age ~65 yr 9/23 (39%) 11/24 (46%) 0.6 Atriofemoral bypass 4/19 (21%) 6/20 (30%) 0.5 Clamp and sew 10/27 (37%) 11/32 (34%) 0.8 ACCT <45 min. 2/22 (9%) 3/25 (12%) 0.8 ACCT ~45 min. 12/24 (50%) 14/27 (52%) 0.9 Intercostal and/or lumbar 13/38 (34%) 14/40 (35%) 0.9 reattachment Postoperative reoperation** 0/4 (0%) 2/2 (100%) 0.07**** Postoperative hypotension** 2/16 (12%) 6/15 (40%) 0.08 Accr, Aortic cross-clamp time. **Among patients who awoke with no deficits (N = 77). ***Pearson chi-square test. ****Fisher's exact test. Table IX. Univariate analysis of neuromuscular deficit within 30 days of surgery Variable Group Patients (%) PAR/PLG OR (95% CI) p value* Treatment received Drainage 46 (47%) 14 (30%) 0.90 (0.4, 2.1) Control 52 (53%) 17 (33%) Dissection Yes 31 (32%) 12 (39%) 1.60 (0.6, 3.9) No 67 (68%) 19 (28%) Age (yr) (52%) 11 (22%) (48%) 20 (43%) 2.69 (1.1, 6.5) Extent replaced I 44 (45%) 8 (18%) 1 II 54 (55%) 23 (43%) 3.34 (1.3, 8.5) Atriofemoral Bypass Yes 39 (40%) 10 (26%) 0.62 (0.3, 1.5) No 59 (60%) 21 (36%) Aortic cross-clamp (14%) 1 (7%) 1 time (rain) (34%) 4 (12%) 1.79 (0.2, 18) (30%) 14 (48%) 12.1 (1.4, 105) (22%) 12 (55%) 15.6 (1.7, 141) Intercostal or lumbar Yes 78 (80%) 27 (35%) 2.12 (0.6, 7.0) reattachment No 20 (20%) 4 (20%) Postoperative reop- Yes 6 (8%) 2 (33%) 3.94 (0.6, 25) eration** No 71 (92%) 8 (11%) Postoperative hypo- Yes 31 (40%) 8 (26%) 7.65 (1.5, 39) tension** No 46 (60%) 2 (4%) *Pearson chi-square test. **Patients with immediate deficits excluded. OR, Odds ratio; CI, confidence interval; PAR/PLG, paraparesis/paraplegia. complications (within 30 days) were extent replaced (p = ), age (p = ), aortic cross-clamp time (p = ) and postoperative hypotension (p = ) (Table IX). The independent predictors (multivariate logistic regression) of deficits occurring during the first 30 days were age (p = ) and aortic clamp time (p < ), and the only significant predictor of delayed deficits was postoperative hypotension (p = ). Of particular relevance to this study is that the method of treatment including reattachment of intercostal and/or lumbar arteries (p ), temporary atriofemoral bypass during aortic occlusion (p = 0.3), and CSFD (p ) was not statistically significant

9 44 Crawford et al. Journal of VASCULAR SURGERY in reducing the incidence of neurologic complications in the lower extremities (Table IX). Cerebrospinal fluid drainage was presumably safe in these patients. Headaches and infection did not occur. The incidence of neurologic complications was in the range of historical cases, and 80% of the patients had magnetic resonance imaging 10 days after the onset of the deficit. Ischemic lesions of the grey matter were noted in those cases that had the most profound deficits, and these findings were associated with a poor prognosis for significant recovci~. DISCUSSION Although numerous factors may be involved in the development of paraplegia during or after graft replacement of aortic segments, interruption of circulation to the spinal cord is the principal one. The interruption may be temporary or permanent, the latter caused by ligation of one or more aortic branches that supply blood to the spinal cord. Toleration of aortic clamping (cord ischemia) is variable. Simple aortic clamping without sacrifice of intercostal arteries in the repair of patent ductus arteriosus by Crafoord ~9 in the 1940s in 25 patients was associated with paraplegia in 4 (16%) with aortic crossclamp times varying from 10 to over 30 minutes. Clamp times in those with deficits were 18 minutes in 1, less than 30 minutes in 3, and over 30 minutes in 2. Aortic clamp times in this series of patients who had type I aneurysms varied from 17 to 79 minutes, median 36 minutes (n = 45), and type II aneurysms varied from 29 to 109 minutes, median 54 minutes (n = 54), with the incidence of paraplegia increasing from 7% to 55% with increase in clamp time. About one third of this time was spent with extensive blind reattachment of intercostal and lumbar arteries. Unfortunately, CSFD in these cases did not significantly extend the safe time nor did it prevent delayed neurologic deficits. Our hope is that rapid identification of the route or routes of cord blood supply by a method recently reported can be successfully applied on a routine basis so that reattachment can be performed more rapidly, thus significantly reducing the cord blood flow interruption time. 2 Regardless, some method of temporary cord protection will be necessary in these extensive cases during the period of reattachment. Cerebros Spinal fluid drainage alone, as we used it, was not enough in these cases. We are hopeful that intrathecal papaverine with or without CSFD, systemic prostaglandin El, or direct arterial perfusion will better serve this purpose. Grateful acknowledgement to Barbara Brooks and Yvette Crear for computer entry and patient follow-up; Chip Brown for manuscript preparation; and Mrs. Carol Pienta-Larson for illustration. REFERENCES 1. Crawford ES, Crawford JL, Sail HJ, et al. Thoracoabdominal aortic aneurysms: preoperative and intraoperative factors determining immediate and long-term results of operations in 605 patients. J VASC SURG 1986;3: Crawford ES, Coselli JS, Sail HJ. Partial cardiopulmonary bypass, hypothermic circulatory arrest, and posterolateral exposure for thoracic aortic aneurysm operation. J Thorac Cardiovasc Surg 1987;94: Crawford ES, Mizrahi EM, Hess KR, Coselli JS, Sail HJ, Patel VM. The impact of distal aortic perfusion and somatosensory evoked potential monitoring on prevention of paraplegia after aortic aneurysm operation. J Thorac Cardiovasc Surg 1988;95: Miyamota K, Keno A, Wada T, Kimoto S. A new and simple method of preventing spinal cord damage following temporary occlusion of the thoracic aorta draining the cerebrospinal fluid. J Cardiovasc Surg 1960;1: Blaisdell FW, Cooley DA. The mechanism of paraplegia after thoracid aortic occlusion and its relationship to spinal fluid pressure. Surgery 1962;51: McCullough JL, Hollier LH, Nugent M. Paraplegia after thoracic aortic occlusion: influence of cerebrospinal fluid drainage. J Vasc SURG 1988;7: Bower TC, Murray MJ, Glovicki P, et al. Effects of thoracic aortic occlusion and cerebrospinal fluid drainage on regional spinal cord blood flow in dogs: correlation with neurologic outcome. J VAsc SURG 1989;9: Woloszyn "IT, Marini CP, Coons MS, et al. Cerebrospinal fluid drainage and steroids provide better spinal cord protection during aortic cross-clamping than does either treatment alone. Ann Thorac Surg 1990;49: Wadouh F, Lindemann EM, Arndt CF, Hetzer R, Borst HG. The arteria radicularis magna anterior as a decisive factor influencing spinal cord damage during aortic occlusion. J Thorac Cardiovasc Surg 1984;88: Svensson LG, Von Ritter CM, Groeneveld HT, et al. Crossclamping of the thoracic aorta. Influence of aortic shunts, laminectomy, papaverine, calcium channel blocker, allopurinol, and superoxide dismutase on spinal cord blood flow and paraplegia in baboons. Ann Surg 1986;204: Hollicr LH. Protecting the brain and spinal cord. J VAsc SURG 1987;5: Hollier LH, Symmonds JF, Pairolero PC, et al. Thoracoabdominal aortic aneurysm repair: analysis of postoperative morbidi~. Arch Surg 1988;123: Svensson LG, Stewart RW, Cosgrove DM, Lytle BW, et al. Preliminary results and rationale for the use of intrathecal papaverine for the prevention of paraplegia after aortic surgery. S Mr J Surg 1988;26: Svensson LG, Stewart RW, Cosgrove DM, et al. Intrathecal papaverine for the prevention of paraplegia after operation on the thoracic or thoracoabdominal aorta. J Thorac Cardiovasc Surg 1988;96: Svensson LG, Grum DF, Cosgrove DM, et al. Appraisal of cerebrospinal fluid alterations during aortic surgery, with in-

10 Volume 13 Number 1 January 1991 Cerebrospinal fluid drainage and paraplegia after high-risk aortic surgery 45 trathecal papaverine administration and cerebrospinal fluid drainage. J VASC SURG 1990;11: Crawford ES, Crawford JL. Diseases of the aorta including an atlas of angiographic pathology and surgical technique. Baltimore: Williams & Wilkins, Murray MJ, Werner E, Oliver WC Jr, Bower TC, Gloviczki P. Anesthetic management of thoracoabdominal aortic aneurysm repair: effects of csf drainage and mild hypothermia. Anesthesiology 1989;71:A Fleiss JL. Statistical methods for rates and proportions, second edition. New York: John Wiley and Sons, 1981: Crafoord C. In: Ekstrom G. The surgical treatment of patent ductus arteriosus: a clinical study of 290 cases. Acta Chir Scand (Suppl): , Svensson LG, Patel V, Coselli JS, Crawford ES. Preliminarv report of localization of spinal cord blood supply by hydrogen during aortic operations. Ann Thorac Surg 1990;49: DISCUSSION Dr. Peter Pairolero (Rochester, Minn.). Dr. Crawford's past experience with more than 500 patients with group I and II thoracic aortic aneurysms demonstrated that 22.5% developed neurologic complications, and that 12% became paraplegic. His statisticians then calculated that to adequately answer the question of whether spinal fluid drainage was of any value in protecting the spinal cord, 98 patients prospectively randomized into a two-arm study would be necessary for sufficient statistical power to demonstrate an 80% improvement rate, no small number when one considers the rarity of these extensive aneurysms. Neurologic complications occurred in 30% of the decompressed group and in 33% of the nondecompressed group. Long-term paraplegia in both groups was 7%. Thus Dr. Crawford's conclusion that CSFD was not beneficial in reducing paraplegia in man seems very appropriate and reasonable. Undoubtedly, paraplegia is multifactorial, and clearly loss of blood flow to the spinal cord is the principal factor. Some method of cord protection then will always be necessary during repair, Spinal fluid drainage alone is not enough. Perhaps the ultimate solution will be found in either preoperative or intraoperative localization of the artery of Adamkiewicz combined with intraoperative techniques to both protect and perfuse the spinal cord. Dr. Crawford, my only question is, what should we do until we find this ultimate solution? Dr. Charles Acher (Madison, Wis.). I question whether you achieved decompression of the spinal compartment by limiting spinal fluid extraction to only an average of 35 ml. In our experience the average amount of spinal fluid withdrawn is 85 ml, and we have gone up to 150 ml to control CSFP without adverse effect. A large body of experimental evidence exists that spinal cord ischemia may be adversely affected and potentiated by spinal opiates. This year we reported our experience over 6 years looking at patients treated with simple crossclamping; subsequently, with the adjuncts of CSFD; then, with CSFD and 48-hour naloxone infusion started during surgery. We saw a marked improvement in the neurologic deficit rate. The deficit rate with simple cross-clamping was 29%. With CSFD alone, it was 22%. So far in 22 patients with naloxone and CSFD drainage, it has been zero percent. These were comparably matched patients. Our patient population is a little bit different from Dr. Crawford's in that 50% of our patients have acute rupture or dissection and two thirds of our patients are type! or type II aneurysms. From our data, we could not differentiate whether CSFD played some role in minimizing spinal cord ischemia during the time of aortic occlusion. I would like to ask Dr. Crawford if he saw any difference in rapidity of recovery in those patients who did recover their neurologic function in the CSFD group, and if so did he measure any substances within the CSF, such as ph, O2 saturation, lactates, or even beta endorphins, to see if there was any difference in the patients who had CSFD with and without paraplegia. Dr. John Connolly (Irvine, Calif.). A long-time interest in paraplegia has indicated to me that when it occurs after aortic surgery, it is a result of, as most of us know, multiple causes. These are related either to temporary inadequate flow to the anterior spinal artery during crossclamping above the artery of Adamkiewicz and/or permanent interruption of critical blood supply to the cord as supplied by the artery of Adamkiewicz between T5 and L2 or interruption of sacral arteries to the terminal cord via the hypogastric arteries. Doppman at the NIH showed first in primates and then in patients that the artery of Adamkiewicz could be safely identified, and I encouraged our angiographer to do the same successfully. Obviously, if the surgeon permanently divides critical blood supply to the anterior spinal artery, whether it is the artery of Adamkiewicz or the arterial branches going to the terminal cord by permanent occlusion of both hypogastric arteries, control of CSFP during aortic surgery is not going to lower the incidence of paraplegia. The possible place for CSFP control would seem to me to be a method to permit a longer safe period of cord ischemic time to permit reinclusion of an identified artery of Adamkiewicz into a distal thoracic or thoracoabdominal prosthesis, or to permit safer, longer periods of aortic crossclamping for repair of deceleration thoracic aortic tears. Finally, although we will probably never have a corn-

11 46 Crawford et al. Journal of VASCULAR SURGERY pletely reliable method to prevent paraplegia, particularly for thoracoabdominal resection, I do believe that continued investigation of the effects of CSFP control along with avoidance of hypotension, which is another very critical factor, are promising attempts to lower the incidence of this most dreaded of complications. Dr. Crawford: I thank those who have discussed this paper, and I regret to tell you that I do not know how to prevent paraplegia after performing more than 2000 operations to replace segments of the descending thoracic and thoracoabdominal aorta by use of the various methods that have been recommended for this purpose. In reference to Dr. Connolly's comments on the safe angiographic indentification of the artery of Adamldewicz, we have not routinely used this technique because of the risk of arteriography, the time required for us to perform the identification (up to 3 hours), and the possible need for general anesthesia. In response to Dr. Acher's question on spinal compartment decompression, we drained an average of 52.5 ml of CSF, which is the recommended safe level advocated by neurosurgeons. We were successful in reducing CSFP to an average of 9 to 10 mm Hg during aortic crossclamping (consistently -< I0 mm Hg in 20 [43%] patients and consistently - 15 mm Hg in 40 [87%]). We measured ph and po2 in the CSF before, during, and after aortic cross-clamping in patients undergoing CSFD and found no significant differences between patients with and without immediate neuromuscular deficits. As far as recover), of neuromuscular function is concerned, of the 10 CSFD patients with immediate deficits, 3 had recovered by discharge, and 4 had recovered by follow-up, whereas of the 11 control patients with immediate deficits, none had recovered by discharge, and 1 patient had recovered by follow-up (these differenccs were not statistically significant). We have not routinely used naloxone because of its failure in our own experience in the treatment of patients who developed neurologic deficits. I would also like to remind Dr. Acher of the necessity of a prospective randomized study to determine the validi~, of his method. He must be aware that statistical power requires 100 patients in groups I and II and 560 patients in groups III and IV or a mixture of all four groups. Our current research includes rapid identification of routes of cord blood supplying by use of hydrogen injection of intercostal and lumbar arteries and electrical recording via intrathecal platinum electrodes and reattachment of these identified vessels. Our prelimina U results with this technique in the pig and humans will be presented later in the meeting.