Cerebral Swelling after Normothermic Cardiopulmonary Bypass 

After we received approval from the hospital ethics committee and informed consent, seven patients (aged 54–72 yr; mean, 63 yr) suitable for MR imaging who were having routine CABG were studied. Anesthesia was induced with 10 micro gram/kg fentanyl, 0.05 mg/kg midazolam, and 0.1 mg/kg pancuronium and maintained with a 0.5–3 mg [center dot] kg sup -1 [center dot] h sup -1 propofol infusion with an oxygen and air mixture for intermittent positive pressure ventilation (fractional inspired concentration of oxygen, 0.5) to a fractional expired concentration of carbon dioxide in expired gas of 3.5 to 4 kPa. Lactated Ringer's solution was given before bypass to maintain central venous pressure > 1 cm H²O. No aprotinin was used in these patients.

Bypass was carried out with pulsatile perfusion at 2.4 l/m sup 2 through a membrane oxygenator with 40-micro meter arterial line filtration; blood cardioplegia was used. The bypass prime (2 l crystalloid) was prewarmed and the patient was actively warmed to 37 [degree sign] Celsius on starting bypass from the mild hypothermia (35 [degree sign] Celsius) that resulted during preparation for bypass. Aortic manipulation was minimized with a single cross-clamp to reduce embolization. Mean arterial pressure was maintained between 40–70 mmHg with 1-mg boluses of methoxamine or glyceryl trinitrate infusion. After separation from bypass, glyceryl trinitrate was infused at 3–20 mg/h to facilitate transfusion of the pump reservoir blood; the rate of glyceryl trinitrate and transfusion were adjusted to keep the central venous pressure at < 7 cm H²O.

Propofol anesthesia was continued throughout transfer to and during the scanning in the MR unit. Electrocardiogram, pulse oximetry, arterial and central venous pressures, fractional inspired concentration of oxygen, and the fractional expired concentration of carbon dioxide in expired gas were monitored throughout the operation, transfers, and MR imaging. Chest drain losses and urine volume were recorded every 15 min in the MR suite.

To confirm the absence of postbypass hyperemia transcranial Doppler measurement of middle cerebral artery velocity was made through the right temporal window (LogiDop 3; Scimed Ltd., Bristol, UK) at a depth of 4–5 cm as an index of change in cerebral blood flow (rather than absolute level). Jugular bulb saturation was measured from samples drawn through a 7.5-French fiberoptic thermodilution catheter (Oximetrics-Abbott Laboratories, North Chicago, IL) to assess potential cerebral oxygen desaturation. Samples were taken 5 min before the start of bypass; 10', 30', and 40' into bypass; and 10' after discontinuing bypass. After preinsertion calibration and flushing with saline, the catheter was inserted retrograde through an 8-French introducer with a hemostatic valve into the right internal jugular vein until resistance at the base of the skull was felt. The catheter was withdrawn slowly until the saturation showed a step increase reflecting facial vein contribution; the catheter was then advanced 3 mm and secured. In vivo calibration was done with a sample from the catheter distal port, which was checked on the co-oximeter (IL 701). Jugular bulb saturation samples were drawn and measured on a co-oximeter. Temperature was measured from the jugular bulb catheter thermistor and from a nasopharyngeal probe to confirm the lack of temperature gradient between cerebral venous blood and brain.

Baseline MR images were obtained the evening before surgery (pre), followed by immediate images at the end of surgery before transfer to the intensive care unit (immediate). Late images (late) were obtained 7 days after surgery to determine if cerebral tissue had returned to normal. Two sequences were used on a 1.0T Picker HPQ system, a 4-min T1-weighted spin-echo sequence 720/20 ms TR/TE, 20 contiguous 7-mm slices on a 192 x 256 matrix with 2 NEX, followed by a 14-min fluid attenuated inversion recovery (FLAIR) sequence 6500/2100/160 msec TR/TI/TE, 10 7-mm slices with a 7-mm gap between each slice on a 128 x 256 matrix with 1 NEX. The T1 sequence was used to determine positioning to match the previous sequences; great care was taken to match the orientation and depth of postoperative scans to the preoperative positions. By design, FLAIR sequences enhance the contrast between cerebrospinal fluid and gray matter as the inversion recovery sequence is set around the null point for cerebrospinal fluid.

There were six sets of images for each patient (pre, immediate, or late: T1 and FLAIR). To minimize any bias, each set of images was presented in random order for visual assessment of cerebral swelling based on width and clarity of cerebral sulci on our standard five-point score (0, none; 1, slight; 2, mild; 3, moderate; 4, severe) and for ventricular size (enlarged, normal, or compressed). Each set was scored independently by two radiologists. For swelling, their individual scores were recorded and then an average score was derived for each set of images: Change from pre to immediate or late images was then calculated for T1 and FLAIR for each patient. These difference scores were tested for significance using a two-tailed sign test.

Bypass lasted 66 +/- 15 min, and aortic cross-clamp time lasted 35 +/- 7 min. The hematocrit concentration decreased from 38 +/- 5% to 22 +/- 3%. Central venous pressure increased from 1.6 +/- 2 mmHg before bypass to 4.2 +/- 3 mmHg 10 min after bypass. Patients were extubated without complication 2.6 to 6.4 h after return from the MR unit (our routine average). No patient had visible conjunctival edema. Nasopharyngeal temperature was 35.1 [degree sign] Celsius (range, 34.8–35.5 [degree sign] Celsius) before bypass and constant at 37 [degree sign] Celsius (range, 36.2–37.5 [degree sign] Celsius) throughout bypass; pulse oximetry was 98–100% throughout the study. Table 1shows results of jugular bulb saturation, middle cerebral artery velocity, systolic and diastolic blood pressure, and partial pressure of carbon dioxide, which were unchanged. All patients had uncomplicated operative and immediate postoperative courses and showed no gross neurologic defects.

Imaging began 64 min (range, 52–71 min) after the end of bypass. Five of seven patients showed marked cerebral swelling (mean, 3; range, 2–4) in the immediate postoperative images with marked obliteration of the sulci on the FLAIR sequences; there was no change in ventricle size. One patient showed minor and one no swelling (Table 2). The mean difference for all seven patients between immediate and preoperative FLAIR images was 2.4 (P = 0.03). The two observers differed by one grade in two sequences and agreed in all others.

One patient died after 48 h (of myocardial failure), one was discharged early, and one refused the late scan. In the remaining four patients, swelling had returned to normal in the late scans (+7 days). T1-weighted images showed no significant change in immediate or late images. Figure 1, a sequence of T1-weighted images at the cortical level, shows the accurate repositioning and global reduction in sulcal width. Figure 2, the same slices with FLAIR imaging, shows the greater contrast in sulci with cerebrospinal fluid nulling of the FLAIR sequence.

The use of normothermic bypass (37 [degree sign] Celsius) is controversial, and it is not yet known whether it is associated with more or less cerebral impairment on neuropsychometric testing. [7–10]In the present study using normothermic bypass, five of seven patients showed cerebral swelling on the immediate postoperative scans, whereas one patient showed complete absence and one only one point of swelling (mean, 2.4; range, 0–4). In the previous study of hypothermic bypass, [6]all six patients showed more than two points of swelling (mean, 2.5; range, 2–4). Although the techniques, protocols, and patient population were the same for the two studies, and for both studies interobserver variability was less than one grade, comparisons must be limited by the small study sizes and historical nature of the controls. The randomization of the three sets of images (preoperative, immediate, and late postoperative) should have minimized any bias in scoring, but the lack of randomized controls makes interpretation of the swelling speculative.

When hyponatremia is produced in animals, astrocytic swelling is seen with a microscopic profile similar to that after ischemia. [11]During hypothermic bypass with a crystalloid prime, colloid osmotic pressure decreases from 20 to 10.6 mmHg due to hemodilution, and intraocular pressure increases 31 mmHg, [12]although osmolarity shows little change. [13]There are no data during normothermic bypass, but the effects of hemodilution are likely to be similar. It is possible that these physiologic changes alone could cause mild cerebral edema.

Patients were extubated within 6 h of the MR imaging and showed no clinical evidence of conjunctival or cerebral edema; it is not known what, if any increase, in CSF volume or pressure is represented by these MR images, which show reduction in sulcal width but normal ventricular size. Olson [14]has proposed a system for measuring brain water content with MR; this could provide useful data but was not possible with the sequences used in this study. Sellman et al. [15]found no increase in ventricle size after CABG, and Schmidt et al. [16]found increased preoperative ventricle size but no further increase unless encephalopathy developed.

Returning the pump prime to the patient after bypass can result in high right heart pressures, which could promote cerebral edema. In this study, the increase in central venous pressure after bypass was small, possibly due to slower transfusion or to higher doses of vasodilators used during transfusion.

Exposure of blood to the bypass circuit and damage to cells and platelets by the pump heads cause an inflammatory response and release free radicals [17]; this produces cerebral swelling, [18]especially if the blood-brain barrier has been damaged by microemboli. Reperfusion injury after release of the aortic cross-clamp also results in activation of leukocytes, and administration of protamine causes a transient increase in cerebrovascular permeability. [19]It is not known if these factors would be different between normo- and hypothermic bypass.

Although regional or microvascular ischemia has not been excluded, no patients had jugular bulb temperatures > 37.5 [degree sign] Celsius or jugular bulb-nasopharyngeal gradients > 1 [degree sign] Celsius, and jugular bulb saturation remained > 55% in all patients. Because cerebral swelling was observed both during our previous hypothermic bypass study, which includes rewarming reported to be associated with indices of global cerebral ischemia, [20]and in our present normothermic study without rewarming, it seems reasonable to hypothesize that rewarming is not necessary to produce postbypass cerebral swelling.

The pattern of MR change in both bypass studies is described as cytotoxic [21](cellular may be a better term [22]), characterized histologically by generalized astrocytic swelling in the absence of a measurable increase in blood-brain barrier permeability: No specific mechanism is implied and there is rarely a single cause of edema. A mild degree of swelling is also seen in volunteers breathing 100% oxygen, although not after 40% oxygen. [23]In the current study, the fractional inspired concentration of oxygen was controlled to 40%, but it is possible that this level could still have some effect on MR images in patients after cardiopulmonary bypass.

T1-weighted images were not enhanced after intravenous gadolinium (molecular weight, 547) in our previous MR study, suggesting that the blood-brain barrier was intact 1 h after the end of bypass. Astrocytic swelling peaks at 40 min after experimental head injury and is prolonged to 100 min by hypoxia. It is possible, therefore, that increased blood-brain barrier permeability after minor cerebral injury may have occurred toward the end of bypass but had recovered by the time of the scans 1 h later. Because the FLAIR images are acquired 60–90 min after bypass, it is not known whether the swelling seen reflects changes occurring at the time of imaging or previous intraoperative changes.

In conclusion, cerebral swelling is seen after normothermic bypass, as assessed by FLAIR MR. The degree of swelling is similar to that previously seen after hypothermic bypass, even though jugular bulb hemoglobin desaturation and postbypass hyperemia were avoided. It remains to be determined whether the cause is physiologic or pathologic.