A new method of intrathecal PO2, PCO2, and pH measurements for continuous monitoring of spinal cord ischemia during thoracic aortic clamping in pigs

A new method of intrathecal PO2, PCO2, and pH measurements for continuous monitoring of spinal cord ischemia during thoracic aortic clamping in pigs Lennart Christiansson, MD, Anders Hellberg, MD, Itaru Koga, MD, Stefan Thelin, MD, PhD, David Bergqvist, MD, PhD, Lars Wiklund, MD, PhD, and Sadettin Karacagil, MD, PhD, Uppsala, Sweden

Background. Impaired spinal cord circulation during thoracic aortic clamping may result in paraplegia. Reliable and fast responding methods for intraoperative monitoring are needed to facilitate the evaluation of protective measures and efficiency of revascularization. Methods. In 11 pigs, a multiparameter PO2, PCO2, and pH sensor (Paratrend 7, Biomedical Sensors Ltd, United Kingdom) was introduced into the intrathecal space for continuous monitoring of cerebrospinal fluid (CSF) oxygenation during thoracic aortic cross-clamping (AXC) distal to the left subclavian artery. A laser-Doppler probe was inserted into the epidural space for simultaneous measurements of spinal cord flux. Registrations were made before and 30 minutes after clamping and 30 and 60 minutes after declamping. The same measuring points were used for systemic hemodynamic and metabolic data acquisition. Results. The mean CSF PO2 readings of 41 mm Hg (5.5 kPa) at baseline decreased within 3 minutes to 5 mm Hg (0.7 kPa) during AXC (P < .01). Spinal cord flux measurement responded immediately in the same way to AXC. Both methods indicated normalization of circulation during declamping. Significant (P < .01) changes were also observed in the CSF metabolic parameters PCO2 and pH. Conclusions. In this experimental model of spinal ischemia by AXC, online monitoring of intrathecal PO2, PCO2, and pH showed significant changes and correlated well with epidural laser-Doppler flowmetry (P < .01). (Surgery 2000;127:571-6.) From the Departments of Anesthesiology, Surgery, and Thoracic Surgery, University Hospital, Uppsala, Sweden

SPINAL CORD ISCHEMIA resulting in paraplegia remains the most devastating complication after thoracoabdominal aortic aneurysm repair, occurring in 5% to 10 % of patients.1,2 Monitoring methods for predicting the degree of spinal cord ischemia during interruption of spinal cord blood flow caused by aortic cross-clamping are important. They may guide us in deciding if it is necessary to reimplant intercostal arteries within the excluded aortic segment to reverse the ischemic insult on the spinal cord, which would be crucial for the clinical outcome after thoracoabdominal aortic replaceSupported by The Laerdal Foundation for Acute Medicine (1460/96), the Swedish Medical Research Council (00759), and the Departments of Anesthesiology and Surgery, Uppsala University Hospital. Accepted for publication December 7, 1999. Reprint requests: Lennart Christiansson, MD, Department of Anesthesiology, University Hospital, Uppsala SE - 751 85, Sweden. Copyright © 2000 by Mosby, Inc. 0039-6060/2000/$12.00 + 0 doi:10.1067/msy.2000.105036

11/56/105036

ment. Various neurophysiologic monitoring methods have been proposed for this purpose but have not yet reached universal acceptance.3-7 Recently, Ishizaki et al8 have reported the use of intrathecal oxygen tension monitoring with a rigid mass spectrometer probe during aortic occlusion in dogs to identify spinal cord feeding arteries. Perfusion of these arteries improved intrathecal oxygen tension and spinal cord-evoked potentials. The aim of the present study was to investigate the intrathecal use of a multiparameter flexible sensor, originally designed for intra-arterial online blood gas monitoring, for continuous registration of cerebrospinal fluid (CSF) oxygenation during thoracic aortic cross-clamping in pigs. The measurements were compared with laser-Doppler flowmetry and the changes in the intrathecally measured metabolic parameters are discussed. MATERIAL AND METHODS The experiments were performed in 9 male and 2 female native piglets with a mean weight of 25 ± 4 kg and a calculated mean body surface area of SURGERY 571

572 Christiansson et al

Surgery May 2000

Fig 1. The Paratrend 7 sensor. Schematic drawing showing a cross section of the distal part of the intrathecal sensor.

0.8 ± 0.1.9 All animals were treated in compliance with the “Guiding Principles in the Care and Use of Animals” approved by the Council of The American Physiological Society.33 The experiments conformed to the national guidelines for animal care and were approved by the Institutional Animal Care and Ethics Committee. After intramuscular anesthesia induction with 3 mg/kg of tiletamin, 3 mg/kg of zolazepam (Zoletil 100, Reading, France), 2.2 mg/kg of xylazine (Rompun, Bayer, Germany), and 0.04 mg/kg of atropine sulfate, the animals were placed in a dorsal recumbent position. An intravenous line was inserted and 20 mg of morphine administered before a tracheotomy was performed. A Siemens 900 C ventilator (Siemens-Elema, Sweden) was connected, and the pigs were ventilated with a tidal volume of approximately 10 mL/kg at an appropriate respiratory rate of 25/minute to produce normocapnia (PaCO2 40 to 43 mm Hg) at baseline conditions. FIO2 was set at 0.4 to assure good oxygenation (PaO2 100 to 200 mm Hg) throughout the experiment. A positive end-expiratory pressure of 4 cm H2O was used. The anesthesia was supplemented with intravenous pancuronium bromide (8 mg) followed by a continuous, triple drug, anesthetic infusion of 25 mg/mL of glucose at an hourly rate of 4 mL/kg (mean total 432 mL ± SD 74 mL). The admixtures were calculated for the following dosages: ketamine, 20 mg × kg-1 × h-1; morphine, 0.48 mg × kg-1 × h-1; and pancuronium bromide, 0.24 mg × kg-1 × h-1. Isotonic saline (10 mL × kg-1 × h-1; mean total, 1382 mL ± SD 473 mL) was given for basal fluid requirements and replacement of perioperative losses. Normothermic temperatures (38 ± 0.5°C) were maintained with the use of prewarmed intravenous fluids and a thermostatically controlled heating pad. The mean blood loss during the whole experiment was 322 mL ± SD 166 mL, (ie, less than 20% of the circulating blood volume). For substitution, the animals received 450 to 750 mL of Dextran 70 (Macrodex, Medisan, Sweden). During the first minutes after declamping, all pigs

were given 200 mL of tromethamine/bicarbonate buffer (Tribonat, Pharmacia-Upjohn, Sweden). Mean urine production throughout the experiments was 363 mL ± SD 156 mL. Pressure monitoring catheters were inserted into the right carotid artery and the right femoral artery without ligation, for recording of arterial pressure proximal and distal to the aortic crossclamp. The observed reduction of mean distal pressure to 14 ± 2 mm Hg indicated complete clamping. A flow-directed Swan-Ganz catheter and a separate central venous catheter were inserted through the right jugular vein for central hemodynamic measurements. The Intellicath/Vigilance monitor system (Baxter Healthcare Corp, USA) was used for continuous cardiac output10 and mixed venous saturation monitoring. All pressures were recorded with Ohmeda transducers (Medical Devices Division, USA) connected to a Siemens Monitor 1281 (Siemens Medical Electronics, USA). A urinary catheter was inserted transvesically through a small incision for urine output monitoring. After these preparations, a left thoracotomy was performed through the fourth intercostal space and the thoracic aorta was isolated distal to the left subclavian artery. A laminectomy was performed at the lower thoracic level and through an over-needle introducer, a multiparameter PO2, PCO2, and pH-sensor was inserted into the intrathecal space. The intrathecal sensor used in this study has been designed for intra-arterial on-line blood gas monitoring11,12 and contains 2 modified optical fibers: for the measurements of PCO2 and pH, a miniaturized polarographic Clark oxygen electrode, and a thermocouple for temperature measurement (Paratrend 7, Biomedical Sensors Ltd, High Wycombe, United Kingdom). The void between each sensor is filled with acrylamide gel containing phenol red (Fig 1). The cylindrical construction of the sensor allows measurement over the entire surface of the probe. The diameter of the probe is 0.5 mm and the 4 sensing components are located at different intervals

Surgery Volume 127, Number 5

along the distal 4 cm of the polyethylene probe permeable for O2 and CO2. The outer surface of the probe is coated with covalently bonded heparin to prevent fibrin deposition. The flexible wire part of the device has a length of 600 mm. The monitor continuously displays pH, PCO2, PO2 and temperature. The readings are given at the actual temperature of the animal, or, if chosen, the thermocouple corrects all measurements to 37°C. Just above the level of the multiparameter sensor insertion, a laser-Doppler probe was placed into the epidural space and connected to a laser-Doppler flowmeter (PF 2B, Perimed, Sweden). Blood gas samples were examined in the automated blood gas analyzer (ABL 300; Radiometer, Copenhagen, Denmark) and with the oximeter calibrated for pig hemoglobin. After baseline hemodynamic, laser Doppler, and CSF (PCO2, PO2, and pH) measurements, the thoracic aorta distal to the origin of the subclavian artery was cross-clamped for 30 minutes. To control proximal hypertension (mean arterial pressure < 150 mm Hg) after cross-clamping, a sodium-nitroprusside infusion (mean total dosage 45 mg) was started for cardiac unloading during clamping. Shortly before declamping, this infusion was replaced by a dopamine infusion (mean total dosage 68 mg) for vasoconstriction and inotropic support. Complete recordings were made at the following intervals: (1) baseline after stabilization (which was reached 30 to 90 minutes after surgical preparation), (2) at the end of the 30-minute crossclamp period, (3) 30 minutes after declamping, and (4) 60 minutes after declamping. At the end of the experiments, the still anesthetized pigs were given a lethal injection of potassium chloride. Data are presented as mean ± standard deviation. The statistical evaluations were based on oneway analyses of variance for repeated measures. Statistical testing was performed for the spinal measurement series with post-hoc comparisons of the different measuring points (StatView 5.0 1998; SAS Institute Inc, Cary, NC). To correlate intrathecal PO2 to laser-Doppler flux, a cross-tabulation of changes during clamping of more than 50% of baseline was performed. Chi-square statistics for small expected frequencies was applied (two-tailed Fisher’s exact test with Yates’ continuity correction). A P level of < .01 was considered significant. RESULTS Intrathecal PO2, PCO2, and pH measurements at various stages of aortic cross-clamping are shown in the Table. These changes are illustrated by a 2-hour trend of the Paratrend registration from one of the experimental animals (Fig 2). Highly significant

Christiansson et al 573

Fig 2. Paratrend registration; online print-out. Trend registration of 2 hours showing changes in intrathecal pH, PCO2 and PO2 in response to clamping and declamping of the aorta. The figure shows the fast reaction of PO2 and gradual changes of pH and PCO2 in response to aortic clamping. After declamping, pH and PCO2 show gradual but, compared with clamping, inverse changes; and PO2 shows an initially fast increase with transition to hyperemia before stabilizing. Conversion factor kPa → mm Hg = 7.5.

changes upon clamping (P < .01) were seen in intrathecal oxygenation and the metabolic parameters as registered by Paratrend. Mean PO2 decreased from baseline 41 to 5 mm Hg (5.5 to 0.7 kPa) within 3 minutes. PCO2 and pH changed gradually during the whole clamping period. Mean PCO2 of baseline 61 mm Hg (8.1 kPa) increased to 119 mm Hg (15.9 kPa) and pH (baseline mean, 7.14) reached 6.75 just before declamping. The dynamics of these changes are shown in the trend registration of Fig 2 where gas tensions are given in kPa with conversion factor 7.5 for mm Hg. The recordings of systemic hemodynamics, blood gas analyses, and spinal measurements are summarized in the Table. There was a sudden decrease in spinal cord flux (SCF) immediately after the aortic cross-clamping, presented as percent of baseline measurements after correction for the biological zero (Table). Before declamping, the mean SCF was 26% of the baseline (P < .01). All animals demonstrated a prompt increase in SCF after removal of the crossclamp. In 10 out of 11 animals there was a hyperemic response for about 10 minutes followed by stabilization at values slightly lower (mean 87% of baseline), but not significantly different, from baseline. The laser Doppler flux and the Paratrend registrations during clamping differed significantly (P < .01) from both the baseline and reperfusion phase.

574 Christiansson et al

Surgery May 2000

Table. Hemodynamic, metabolic, and spinal data (standard deviation) Parameter Cardiac output L/min Heart rate beats/min Mean arterial pressure mm Hg Central venous pressure mm Hg Mean pulmonary arterial pressure mm Hg Pulmonary capillary wedge pressure mm Hg Hemoglobin g/L a-Lactate mmol/L a-Base excess a-pH a-PCO2 kPa* a-PO2 kPa* Arterial oxygen saturation % Mixed venous PO2 kPa* Mixed venous oxygen saturation % Paratrend pH Paratrend PCO2 kPa* Paratrend PO2 kPa* Laser-Doppler % of baseline

Clamping 30 min

Baseline 4.2 (1.3) 112 (19) 109 (15) 9 (2) 24 (4) 13 (4) 78 (10) 1.4 (0.4) 5.2 (1.3) 7.46 (0.03) 5.5 (0.2) 22.3 (3.0) 99 (0) 5.3 (0.7) 63 (10) 7.14 (0.18) 8.1 (0.6) 5.5 (1.6) 100 (0)

4.6 (0.8) 212 (22) 139 (18) 10 (2) 26 (4) 14 (6) 81 (14) 6.7 (0.8) -3.1 (1.8) 7.44 (0.03) 4.0 (0.3) 25.3 (3.2) 99 (0) 8 (0.7) 86 (7) 6.75 (0.16) 15.9 (3.3) 0.7 (1.1) 26 (21)

Declamping 30 min 6.1 (1.0) 186 (33) 85 (12) 10 (2) 30 (7) 16 (7) 71 (9) 8.5 (1.2) -4 (2.4) 7.27 (0.05) 6.7 (0.6) 20.4 (4.0) 98 (1) 6.5 (0.7) 66 (8) 7.01 (0.19) 8.6 (0.8) 6.0 (2.3) 84 (43)

Declamping 60 min 6.2 (1.3) 180 (30) 82 (15) 10 (2) 27 (5) 14 (5) 68 (8) 6.9 (1.2) -1.4 (2.3) 7.33 (0.05) 6.3 (0.5) 20.4 (4.4) 98 (1) 5.9 (0.8) 64 (9) 7.10 (0.12) 8.4 (0.8) 5.4 (2.3) 87 (51)

*Conversion factor kPa → mm Hg = 7.5.

Measurements at 30 and 60 minutes after declamping showed no significant difference compared with the baseline. All measurements were performed after stabilization, but we noted changes in flux immediately after aortic clamping and declamping, and Paratrend showed changes in oxygen tension within its response time of 3 minutes. Good correlation (P < .01) was shown between the 2 methods regarding the changes in SCF and intrathecal PO2 during aortic cross-clamping. DISCUSSION Reliable methods for monitoring spinal cord perfusion and, ideally, function during descending thoracic or thoracoabdominal aortic replacement have important implications on the technical strategies for the prevention of paraplegia. Somatosensory-evoked potentials have been successfully used in some series to detect deterioration of spinal cord function during aortic replacement, but they have been criticized by others because they may fail to directly monitor the function of anterior spinal cord tracts during aortic occlusion.1,3-7 Some authors have advocated the use of myogenic motor responses evoked either by transcranial or spinal electrical stimulation in an attempt to overcome some of the limitations of somatosensory-evoked potentials. They have been shown to be effective in assessing spinal cord ischemia during thoracoabdominal aortic repair.4,5,7 However, hemodynamic changes associated with aor-

tic cross-clamping might interfere with the neurophysiologic monitoring methods; and, furthermore, changes in these evoked potentials may only be detectable after a significant ischemic spinal cord injury.1 The laser-Doppler flowmetry technique has been validated for experimental spinal cord blood flow measurements.13 Continuous registration of spinal cord microcirculation correlated well with the microsphere technique in the discrimination of rapid dynamic changes but also in the reflection of relative changes during longer standing dysregulation of spinal cord blood flow.14 Epidural laserDoppler recordings have been shown to predict histopathologic15 and clinical outcome in a spinal cord ischemia model induced by thoracic aortic occlusion in cats.16 Disadvantages of the laserDoppler technique are the difficulties of obtaining absolute values and the sensitivity to artifacts. Because the application of the technique requires laminectomy, it is not appropriate for routine use in human beings. The reason for epidural placement of the laser-Doppler probe in our experiments was to prevent spinal fluid leakage that would jeopardize the intrathecal Paratrend recordings. Wadouh et al17 studied hypoxia as the primary reason for spinal cord injury by directly measuring decreased oxygen tension on the spinal cord surface in pigs during aortic cross-clamping. Several other methods have been used for the direct or indirect monitoring of spinal cord oxygenation,

Christiansson et al 575

Surgery Volume 127, Number 5

such as platinum electrodes and other polarographic probes,17-19 platinum electrodes in combination with hydrogen injection,20-22 and mass spectrometer or near-infrared spectroscopy.23-25 Different locations for these measurements have been used: spinal cord surface,17 brain tissue,19 epidural space,23,24 and intrathecally.18,24 Ishizaki et al24 measured intrathecal, epidural, and spinal cord oxygen tension with a mass spectrometer in a dog model of graded spinal cord ischemia and showed a significant correlation between intrathecal and spinal cord oxygen tension. The same authors have examined the value of intrathecal oxygen tension monitoring in another experimental model in dogs for identification of spinal cord feeding arteries during aortic occlusion.8 The selective perfusion of the feeding arteries to the spinal cord not only reversed spinal cord ischemia but also improved spinal cord function indicated by spinal-evoked potentials. However the lowest safe limit of intrathecal oxygen tension that will prevent ischemic damage was not clear. The major concern with the above-mentioned method was the PO2 probe because it was relatively rigid and likely to damage the spinal cord during insertion. The authors concluded that it would be necessary to develop a softer and more flexible probe for possible clinical use. The intrathecal multiparameter sensor used in our study has been in clinical practice for some years for continuous intra-arterial blood gas monitoring and has been validated against conventional blood gas analysis.11,12 It has also been tested by others in locations such as brain tissue,25-27 gastrointestinal lumen,28 and subcutis.29 The sensor can be introduced percutaneously. Few reports are found in the literature on normal values of PO2, PCO2, and pH in CSF at different locations. Fleckenstein et al18 reported PO2 of 61 to 64 mm Hg in the ventricle and about 45 mm Hg in the subarachnoid space of normoxic (PaO2, 100 mm Hg) cats, dogs, and pigs. In the current study of pigs, we report a baseline intrathecal PO2 of 41 ± 12 mm Hg, a PCO2 of 61 ± 5 mm Hg, and a pH of 7.14 ± 0.18. For human brain tissue, Hoffman et al26 demonstrated a PO2 of 37 ± 12 mm Hg, a PCO2 of 49 ± 5 mm Hg, and a pH of 7.16 ± 0.08 in noncompromised anesthetized patients and a PO2 of < 20 mm Hg, a PCO2 of > 60 mm Hg, and a pH of < 7 as critical levels for the development of ischemia. The same group reported later27 concordant normal data with critical values of PO2 < 10 mm Hg, PCO2 > 60 mm Hg, and pH < 6.8. Our group30 has reported percutaneous, intrathecal measurements of PO2 64 mm Hg, PCO2 49 mm Hg, and pH 7.27 in an anesthetized patient before tho-

racoabdominal aortic surgery. Valadka et al25 have performed intracerebral measurements with 2 different techniques. From their discussion and the editorial views31 regarding critical levels of brain tissue oxygen tension, it can be concluded that interstitial PO2 < 6 mm Hg (0.8 kPa) indicates a very high risk of ischemic injury. In the PO2 range of 6 to 15 mm Hg (0.8 to 2.0 kPa), the risk is intermediate and more dependent on duration. In a recent review by Tønnessen,32 the maximum aerobic tissue PCO2 was proposed to be 75 to 90 mm Hg (10 to 12 kPa) with a range of 68 to 143 mm Hg (9 to 19 kPa) for different organs. During no-flow conditions, metabolic waste products are not transported in ischemic tissue whereas the readily diffusible carbon dioxide will to some extent equilibrate with surrounding tissues and fluids such as CSF. In addition to oxygen measurements, the registration of pH and PCO2 as metabolic parameters might be helpful in the identification of critical levels of oxygenation. The results of this experimental study demonstrate that thoracic aortic occlusion produces prompt alterations in CSF oxygenation and gradual changes in the metabolic parameters that can be continuously monitored with the flexible intrathecal multiparameter sensor. Paratrendderived CSF parameters during and after thoracic aortic occlusion correlate with changes in spinal cord microcirculation measured by epidural laserDoppler flowmetry. Although the use of intrathecal oxygen tension measurements has been validated in another experimental model in dogs,8,24 the findings from the current study should be interpreted cautiously regarding its practical implications because it lacks histopathologic, neurophysiologic, and clinical outcome data. Validation of PO2, PCO2, and pH in CSF as possible indicators of spinal cord ischemia during aortic cross-clamping requires further studies to determine the critical ischemic levels of the parameters measured by Paratrend 7. This also includes the relation to duration of spinal cord ischemia at different temperatures. These are the aims of ongoing experimental studies at our institution. REFERENCES 1. Shenaq SA, Svensson LG. Paraplegia following aortic surgery. J Cardiothorac Vasc Anesth 1993;7:81-94. 2. Svensson LG, Crawford ES, Hess KR, Coselli JS, Safi H. Experience with 1509 patients undergoing thoracoabdominal aortic operations. J Vasc Surg 1993;17:357-70. 3. Cunningham J Jr, Laschinger JC, Spencer FC. Monitoring of somatosensory evoked potentials during surgical procedures on the thoracoabdominal aorta. IV. Clinical observations and results. J Thorac Cardiovasc Surg 1987;94:275-85. 4. Matsui Y, Goh K, Shiiya N, Murashita T, Miyama M, Ohba J, et al. Clinical application of evoked spinal cord potentials

576 Christiansson et al

5.

6.

7.

8.

9.

10.

11.

12.

13.

14.

15.

16.

17.

18.

elicited by direct stimulation of the cord during temporary occlusion of the thoracic aorta. J Thorac Cardiovasc Surg 1994;107:1519-27. Mongan PD, Peterson RE, Williams D. Spinal evoked potentials are predictive of neurologic function in a porcine model of aortic occlusion. Anesth Analg 1994;78:257-66. Galla JD, Ergin A, Sadeghi AM, Lansman SL, Danto J, Griepp RB. A new technique using somatosensory evoked potential guidance during descending and thoracoabdominal aortic repairs. J Cardiac Surg 1994;9:662-72. deHaan P, Kalkman CJ, deMol BA, Ubags LH, Veldman DJ, Jacobs M. Efficacy of transcranial motor-evoked potentials to detect spinal cord ischemia during operations for thoracoabdominal aneurysms. J Thorac Cardiovasc Surg 1997;113:87-101. Ishizaki M, Sugiyama S, Uchida H, Nawa S, Shimizu N. Identification and selective perfusion of the spinal cordfeeding arteries by intrathecal PO2 monitoring for spinal cord protection. Eur J Vasc Endovasc Surg 1999;18:17-24. Morrison PR, Meyer MP. Constants for use in surface area formula: mammals. In: Altman PL, Dittmer DS, editors. Biology Data Book. Federation of American Societies for Experimental Biology; 1964. p. 120-1. Yelderman M, Quinn MD, McKown RC, Eberhart RC, Dollar ML. Continuous thermodilution cardiac output measurement in sheep. J Thorac Cardiovasc Surg 1992; 104:315-20. Venkatesh B, Clutton-Brock TH, Hendry SP. A multi-parameter sensor for continuous intra-arterial blood gas monitoring: a prospective evaluation. Crit Care Med 1994;22:588-94. Clutton-Brock TH, Fink S, Luthra AJ, Hendry SP. The evaluation of a new intravascular blood gas monitoring system in the pig. J Clin Monit 1994;10:387-391. Lindsberg PJ, O’Neill JT, Paakkari IA, Hallenbeck JM, Feuerstein G. Validation of laser-Doppler flowmetry in measurement of spinal cord blood flow. Am J Physiol 1989;257 (Heart Circ Physiol 26):H674-80. Lindsberg PJ, Jacobs TP, Frerichs KI, Hallenbeck JM, Feuerstein GZ. Laser-Doppler flowmetry in monitoring regulation of rapid microcirculatory changes in spinal cord. Am J Physiol 1992;263(Heart Circ Physiol 32):H285-92. Marsala M, Sorkin LS, Yaksh TL. Transient spinal ischemia in rat; characterization of spinal cord blood flow, extracellular amino-acid release and concurrent histopathological damage. J Cereb Blood Flow Metab 1994;14:604-16. Yamada T, Morimoto T, Nakase H, Hiribayahi H, Hiramatsu K, Sasaki T. Spinal cord blood flow and pathophysiological changes after spinal cord ischemia in cats. Neurosurgery 1998;42:624-34. Wadouh F, Arndt CF, Metzger H, Hartmann M, Wadouh R, Borst HG. Direct measurements of oxygen tension on the spinal cord surface of pigs after occlusion of the descending aorta. J Thorac Cardiovasc Surg 1985;89:787-94. Fleckenstein W, Nowak G, Kehler U, Maas AIR, Dellbrugge HJ, Jong DA, et al. Oxygen pressure measurements in cerebrospinal fluid. mt-Medizintechnik 1990;110:44-53.

Surgery May 2000 19. Maas AIR, Fleckenstein W, de Jong DA, van Santbrink H. Monitoring cerebral oxygenation: experimental studies and preliminary clinical results of continuous monitoring of cerebrospinal fluid and brain tissue oxygen tension. Acta Neurochir 1993[Suppl]59:50-7. 20. Svensson LG, Patel V, Coselli JS, Crawford ES. Preliminary report of localization of spinal cord blood supply by hydrogen during aortic operations. Ann Thorac Surg 1990; 49:528-36. 21. Svensson LG, Crawford ES, Patel V, McLean TR, Jones JJ, DeBakey ME. Spinal oxygenation, blood supply localization, and function with aortic clamping. Ann Thorac Surg 1992;54:74-9. 22. Svensson LG. Intraoperative identification of spinal cord blood supply during repairs of descending aorta and thoracoabdominal aorta. J Thorac Cardiovasc Surg 1996;112: 1455-61. 23. Sugiyama S, Ishizaki M, Uchida H. Spinal epidural oxygen pressure and evoked spinal cord potential in relation to the severity of spinal ischemia during cross-clamping of the thoracic aorta. Acta Med Okayama 1993;47:369-76. 24. Ishizaki M, Sugiyama S, Uchida H, Nawa S, Shimizu N. Intrathecal oxygen concentration as a new indicator of spinal cord ischemia. Acta Med Okayama 1997;51:71-7. 25. Valadka AB, Gopinath SP, Contant CF, Uzura M, Robertson CS. Relationship of brain tissue PO2 to outcome after severe head injury. Crit Care Med 1998;26:1576-81. 26. Hoffman W, Charbel F, Edelman G. Brain tissue oxygenation, carbon dioxide, and pH in neurosurgical patients at risk for ischemia. Anesth Analg 1996;82:582-6. 27. Charbel FT, Hoffman WE, Misra M, Hannigan K, Ausman J I. Cerebral interstitial tissue oxygen tension, pH, HCO3, CO2. Surg Neurol 1997;48:414-17. 28. Knichwitz G, Rötker J, Bruessel T, Kuhmann M, Mertes N, Möllhoff T. A new method for continuous intramucosal pCO2 measurement in the gastrointestinal tract. Anesth Analg 1996;83:6-11. 29. Jedlinska B, Mellström Å, Hartmann M, Jönsson K. Comparison of tissue oxygen-tension measurements by different devices. An experimental study in pigs. Scand J Clin Invest 1998;58:63-72. 30. Christiansson L, Karacagil S, Thelin S, Hellberg A, Tydén H, Wiklund L, et al. Continuous monitoring of intrathecal pO2, pCO2 and pH during surgical replacement of type II thoracoabdominal aortic aneurysm. Technical note. Eur J Vasc Endovasc Surg 1998;15:78-81. 31. Prough DS, DeWitt DS, Zornow MH. Monitoring cerebral oxygenation in the twilight years of the decade of the brain. Editorial views. Crit Care Med 1998;26:1482-4. 32. Tønnessen TI. Biological basis for PCO2 as a detector of ischemia. Acta Anaesthesiol Scand 1997;41:659-669. 33. The American Physiological Society. Guide for the Care and Use of Laboratory Animals. 7th ed. Washington, DC: National Academy Press; 1996.