Do Indomethacin and Propofol Cause Cerebral Ischemic Damage?: Diffusion-weighted Magnetic Resonance Imaging in Patients Undergoing Craniotomy for Brain Tumors

The Research Ethics Committee of the County of Aarhus, Denmark, approved the study. The investigation was conducted in accordance with the “Note for Guidance on Good Clinical Practice.” Monitoring of the study was performed by the Good Clinical Practice Unit, Århus University Hospital, Århus, Denmark.

After written informed consent was obtained, nine patients (American Society of Anesthesiologists physical status I or II) undergoing supine-positioned elective craniotomy for supratentorial brain tumors of 3 cm or larger (measured as largest diameter in any plane on magnetic resonance [MR] images) were included in the study. Exclusion criteria were age younger than 18 yr or older than 70 yr, pregnancy or nursing, history of allergic reactions to prostaglandin inhibitors, arterial hypertension (diastolic pressure > 110 mmHg), cardiac failure (New York Heart Association class III or IV), moderate to severe chronic pulmonary insufficiency, renal or hepatic dysfunction/disease, peptic ulcer, and treatment with indomethacin or other nonsteroidal antiinflammatory drugs.

The first MRI examination was performed in the awake patient the day before surgery. Magnetic resonance imaging sequences consisted of a three-plane localizer followed by three axial, anatomical scans: a three-dimensional T1-weighted spoiled gradient recall sequence, a T2-weighted sequence, and a fluid-attenuated inversion recovery (FLAIR) sequence, to obtain images for outlining the extent of tumor, edema, and older (> 12 h) ischemic lesions. All three sequences were recorded in a 24 × 24-cm field of view, with image acquisition parameters (time of repetition − TR/time of echo − TE/time of inversion − TI) being TR/TE = 30/2.8 ms (ms), flip angle 20°, TR/TE = 4,000/102 ms, and TR/TE/TI = 9,000/120/2,200 ms. The slice thickness was 1.2 mm (120 slices) for the three-dimensional spoiled gradient recall sequence and 5 mm with a 1.5-mm gap (24 slices in 256 × 256 resolution) for the T2 and FLAIR sequences. To detect acute ischemia, an axial DWI sequence was acquired (spin-echo echo planar imaging, TR/TE = 5,000/81.5 ms, slice number and dimensions as for the FLAIR/T2 scans in a 96 × 96 resolution). The diffusion-weighted scan consisted of one unweighted scan followed by three scans with diffusion-weighting gradients applied in three orthogonal directions (diffusion weighting factor b = 1,000). Based on this, maps of the apparent diffusion coefficient (ADC) were calculated. In cases where lesions are observed on DWI, ADC allows acute lesions (lesion ADC is lower than that of normal tissue up to 5 days after an infarct) to be distinguished from older lesions or edema (ADC is higher than that of normal tissue in vasogenic edema and necrosis for detection of previous cerebral injuries or acute cerebral ischemia).16,17 

Perfusion imaging was performed by dynamic (spin-echo echo planar imaging) tracking of a bolus of 0.2 mmol/kg gadolinium-DTPA (Magnevist®; Schering AG, Berlin/Bergkamen, Germany), injected at a rate of 5 ml/s, with an 8-s delay, using an MR-compatible power injector (Medrad, Pittsburgh, PA). This bolus was immediately followed by injection of an equal volume of physiologic saline, also at a rate of 5 ml/s. Eleven or twelve slices were obtained, covering the tumor area. The acquisition parameters were TR/TE = 1,500/75 ms, flip angle 20°, 96 × 96 resolution, 24 cm field of view, 5-mm slice thickness, and 1.5-mm interslice gap with a 55-s acquisition time.

A postcontrast T1-weighted spoiled gradient recall sequence was acquired for determination of tumor-contrast enhancement and to assess tumor grade. The duration of the examination was 30–35 min.

On the day of surgery, the patient was anesthetized and transported to the MR scanner. The second MRI examination was performed before administration of indomethacin to evaluate possible ischemic effects of the initial propofol–fentanyl anesthesia (MRI was performed approximately 70 min after induction of anesthesia). Indomethacin was then administered as an intravenous bolus dose of 0.2 mg/kg followed by infusion at 0.2 mg · kg⁻¹· h⁻¹. The third MRI examination was initiated 5 min after administration of the indomethacin bolus dose, with the DWI sequence performed 20 min after indomethacin administration. Both MRI examinations consisted of the same sequences as the initial scan (see above), except for the FLAIR sequence, which would be insensitive to acute ischemic lesions. The indomethacin infusion was terminated after completion of the third MRI examination. Data regarding mean arterial blood pressure, Sjvo², arteriovenous oxygen difference (AVDo²), arterial oxygen tension (Pao²), and arterial carbon dioxide tension (Paco²) was collected immediately before commencement of the second MRI examination and again immediately after termination of the third MRI examination. The duration of the two examinations, including indomethacin administration, was approximately 60 min. After completion of the MRI examinations, the patient was transported to the operating room for surgery. The fourth MRI examination was performed in the awake patient 2 days after surgery to detect possible “late” ischemic lesions, using the same MRI protocol as the first MRI examination.

Tumor size (calculated from the modified spheric volume equation:

), tumor localization, and the degree of midline shift were determined by an experienced neuroradiologist (C. G.) from the structural MR images obtained from the first MRI examination. The histopathologic diagnosis was obtained from the neuropathology report.

The patients were premedicated with diazepam (5–15 mg) orally 1 h before anesthesia. For induction, propofol (1.2–2.5 mg/kg) supplemented with fentanyl (2–4 μg · kg⁻¹· h⁻¹) was used. Cisatracurium (0.2 mg/kg) was administered to facilitate tracheal intubation. Anesthesia was maintained with propofol (6–12 mg · kg⁻¹· h⁻¹) and fentanyl (1.5–3 μg · kg⁻¹· h⁻¹). Neuromuscular blockade was maintained with cisatracurium and monitored by train-of-four stimulation. Controlled ventilation (fraction of inspired oxygen [Fio²] 50–60% by oxygen–air) was applied, and the patients were ventilated with Paco²and Pao²levels that we attempted to keep between 30–40 mmHg and greater than 100 mmHg, respectively. A decrease in systolic pressure exceeding 20 mmHg compared with the preoperative level was treated with 5–10 mg intravenous ephedrine. Normal saline, 15 ml/kg, was infused for the first hour after induction, followed by normal saline at 2–4 ml · kg⁻¹· h⁻¹. Moreover, to counteract the blood pressure decrease observed after induction of anesthesia, 6% hydroxyethyl starch, 500 ml, was infused over a period of 30 min. Monitoring consisted of continuous electrocardiography, pulse oximetry (Datex® AS3; Datex, Helsinki, Finland), and rectal temperature monitoring. After induction of anesthesia, a radial artery catheter was inserted for continuous mean arterial pressure monitoring and blood sampling. A jugular bulb catheter was inserted retrogradely into the internal jugular vein for jugular bulb pressure monitoring and jugular venous blood sampling. The tip of the catheter was placed high in the jugular bulb, and the position was confirmed by x-ray control. After stable physiologic conditions were ensured, the patient was transferred to the MR scanner. During the MRI examinations, hemodynamic parameters were continuously monitored with an MR-compatible monitor, and controlled ventilation was applied with an MR-compatible ventilator.

Observation of clinical signs of cerebral ischemic damage associated with the anesthetic regimen and administration of indomethacin was based on a thorough neurologic examination performed by a neurosurgeon 1 or 2 days after the operation.

Diffusion-weighted images (together with ADC maps) and FLAIR images were reviewed systematically, supplemented by the use of T1- and T2-weighted images by a board-certified neuroradiologist with extensive experience in interpretation of DWI images (C. G.). The neuroradiologist evaluated the DWI and ADC maps from the four MRI examinations obtained in each patient in random order and was blinded to the sequence in which they appeared. In all cases, DWI and ADC maps were analyzed first, followed by FLAIR images. A DWI scan was considered positive if the scan revealed an area of hyperintensity on DWI and hypointensity on ADC maps not present in the first MRI examination, thus signifying a newly developed acute ischemic lesion.

Physiologic parameters are presented as “raw” data and median values and ranges. Paired analysis of physiologic parameters obtained before and after indomethacin administration was performed with the Wilcoxon signed rank test. Statistical significance was considered at the 5% level.

The purpose of the perfusion-weighted information was to determine the effects of indomethacin and propofol on cerebral blood flow physiology. These results will be presented elsewhere.

Patient characteristics, neuroradiologic data (tumor size, tumor localization, and midline shift), and histopathologic diagnoses are given in table 1. In patient 5, it was not possible for the neuropathologist to determine the tumor type.

Induction doses of fentanyl and propofol were 2.5 (1.8–3.2) μg/kg and 1.2 (0.7–1.80) mg/kg, respectively, and maintenance doses were 1.8 (1.4–2.20) μg · kg⁻¹· h⁻¹and 7.0 (4.5–10) mg · kg⁻¹· h⁻¹, respectively. The interval from induction of anesthesia to DWI was 86 (71–91) min.

Changes in physiologic parameters before and after indomethacin administration are demonstrated in table 2. In patient 3, it was not possible to insert a jugular bulb catheter; therefore, no Sjvo²or AVDo²measurements were performed. Except for patients 4 and 6, all patients experienced an increase in mean arterial blood pressure after indomethacin administration; however, there was no statistically significant difference in the paired analysis. There was a nonsignificant tendency toward lower Sjvo²values (P = 0.12) and higher AVDo²(P = 0.075) values after administration of indomethacin. The Pao²values were higher during the third MRI examination (P = 0.03).

Ischemic lesions were not detected in the DWI, ADC, and FLAIR images in any of the nine patients. An example of DWI and FLAIR images obtained in a patient with gliosarcoma is shown in figure 1.

Because of insufficient emergence from anesthesia, probably caused by 5 h of surgery and hence anesthesia, patient 2 was transported to the neurointensive ward for further artificial ventilation. Three hours later, the patient underwent extubation and was fully awake and conscious, without neurologic symptoms. The remaining patients had an uneventful emergence from anesthesia and underwent extubation in the operating room. The patients were fully awake and conscious before transfer to the postoperative ward. Based on the postoperative neurologic examination by the neurosurgeon, none of the patients experienced any aggravation of their preoperative neurologic deficits (table 3).

Experimental data18and clinical studies4–7suggest that indomethacin may be useful in the management of increased ICP. The vasoconstriction and associated reduction in CBF induced by indomethacin decrease elevated ICP and improve cerebral perfusion pressure. However, the use of indomethacin is controversial because the indomethacin-induced decrease in CBF may lead to critical hypoperfusion and ischemic damage. Nilsson et al.  18reported that indomethacin elicited progressive changes in AVDo²and pH and slowing of the electroencephalogram, suggesting development of cerebral ischemia in a porcine model of intracranial hypertension. In patients with severe head injury and isoflurane anesthetized patients with cerebral tumors, indomethacin administered as a bolus dose alone or followed by infusion caused an average decrease in CBF to 31–20 ml · 100 g⁻¹· min⁻¹accompanied by increases in arteriovenous difference of lactate and AVDo²and a decrease in Sjvo²values.4,6,7In the current study, Sjvo²averaged 43%, which is in accordance with a study in propofol-fentanyl–anesthetized tumor patients where a continuous indomethacin infusion was associated with an average Sjvo²value of 46%.8The reported low CBF and Sjvo²and high AVDo²values in the above-mentioned studies suggest that the level of CBF under these conditions is critically low and eventually surpasses ischemic threshold. To our knowledge, this is the first study that addresses the question whether indomethacin induces ischemic brain damage. Using DWI, we were not able to demonstrate any ischemic brain lesions after indomethacin administration. It is important to note that DWI is not sensitive to ischemia per se  but detects ischemic injury. In view of the low Sjvo²values, we cannot rule out that ischemia occurred in these patients. However, our results suggest that CBF was not reduced to such an extent that ischemic damage occurred on the DWI images. We observed a minor increase in AVDo², suggesting limited decrease in CBF after indomethacin. This observation is in contrast to previous studies in which indomethacin caused a pronounced decrease in CBF4,7but agree with the findings of a recent study in propofol-fentanyl–anesthetized patients undergoing craniotomy.8We suggest that the limited reduction in CBF after indomethacin may be explained by the propofol-induced cerebral vasoconstriction.8Propofol may hence have caused vasoconstriction of the cerebral vessels to the extent that they had a limited capacity to constrict further during the influence of indomethacin. As a note of caution, our sample size is relatively small, and patient 4 seems to be an outlier with an Sjvo²of 63%; the removal of this value would have resulted in a significant decrease in Sjvo²after indomethacin administration. Therefore, interpretation of the Sjvo²and AVDo²values after indomethacin must be cautious, and we cannot exclude a type II error in this study.

Diffusion-weighted imaging is an established MRI technique that is very sensitive to acute cerebral ischemic damage. In experimental stroke models, ischemic regions become hyperintense on DWI within minutes,15,19and this seems to be similar in humans, where DWI lesions have been detected as early as 39 min after stroke onset.20Early ischemic brain injury detected by DWI correlates with postmortem infarction,21acute and chronic clinical severity,22and clinical outcome23,24and can approach a 100% detection rate for ischemic lesions in the hyperacute period.25,26 

The DWI measurements were performed 20 min after indomethacin administration. We cannot exclude that a longer duration of indomethacin infusion or another (longer) time interval between indomethacin administration and DWI acquisition would have produced different acute DWI results. However, previous studies have demonstrated that intravenously administered indomethacin elicits a pronounced decrease in CBF and mean cerebral artery blood flow velocity within 30–60 s after administration.4,8Therefore, ischemic changes on the postindomethacin DWI are time and flow dependent, and this examination therefore rules out any immediate damage with uncertainty. Although we believe that the damage would coincide with maximum pharmacologic action, we cannot exclude that later damage would escape this examination. However, his would be detected on the follow-up scan (structural as well as DWI images) performed 2 days after the operation.

Clinical27,28and experimental studies29have demonstrated that propofol elicits a decrease in CBF and CMRo². Propofol is widely used for neurosurgical procedures, and a recent study demonstrated that ICP is lower and cerebral perfusion pressure is higher during propofol–fentanyl anesthesia as compared with isoflurane or sevoflurane.9However, a high incidence of Sjvo²values less than 50% have been reported in propofol-anesthetized brain tumor patients8–11and nonneurosurgical patients.30The reported low Sjvo²values reflect an imbalance between global cerebral oxygen supply and demand and may be indicative of cerebral hypoperfusion and ischemia. A number of studies have demonstrated that the percentage reduction of CBF after administration of propofol is larger than the reduction of CMRo².10,31,32These findings suggest that propofol may have a direct cerebral vasoconstricting effect, beyond the associated decrease in CMRo²leading to a decrease of the CBF:CMRo²ratio. The current study, however, demonstrates for the first time that low Sjvo²values are not associated with ischemic damage as evaluated with DWI. Furthermore, no aggravation of clinical neurologic signs of cerebral ischemia were disclosed in the current study or in clinical studies of patients with cerebral tumors.8,12These observations naturally lead to the question whether the established thresholds of Sjvo²for detection of cerebral ischemia apply to these patients. In acute head injury, Sjvo²values less than 50% suggest hypoperfusion and Sjvo²values less than 40% are supposed to be associated with the development of global cerebral ischemia.33In awake, nonmedicated subjects, the mean value of Sjvo²ranges between 54.6%, 55.0%, and 57.1%,34–36with the lowest individual value of 44.7%.34In propofol-fentanyl–anesthetized patients undergoing craniotomy for cerebral tumor, Sjvo²averaged 57%.9If normal values of CBF and Sjvo²are considered to be 50 ml · 100 g⁻¹· min⁻¹and 55%, respectively, a 40% decrease in CBF caused by cerebral vasoconstriction would decrease CBF to 30 ml · 100 g⁻¹· min⁻¹and Sjvo²to 42%. Under this circumstance, the vasoconstriction is accompanied by cerebral oligemia, but not ischemic changes detectable on DWI images. The value of 42% corresponds to the median value of 43% observed in the current study during indomethacin administration. Comparing absolute levels of Sjvo²or AVDo²between patients with severe head injury and patients subjected to elective tumor resection in propofol–fentanyl anesthesia may be misleading. An Sjvo²of approximately 40% may not have the same significance in patients with cerebral tumors compared with head-injured patients with different pathology. Therefore, the ischemic threshold of Sjvo²(i.e. , 50%) in patients with severe head injury should not be applied to propofol-anesthetized patients. We therefore suggest a redefinition of the ischemic threshold for Sjvo²during propofol anesthesia.

The dose of indomethacin administered in the current study is based on a recent study in which indomethacin (bolus of 0.2 mg/kg followed by 0.2 mg · kg⁻¹· h⁻¹) caused a significant decrease in cerebral blood flow velocity before induction of anesthesia.8This finding is in accord with a study in healthy volunteers where indomethacin (bolus of 0.2 mg/kg followed by 0.2 mg · kg⁻¹· h⁻¹) caused a significant decrease in CBF, ranging between 29% and 37%, equal to a clinical study in head-injured patients where indomethacin (bolus of 30 mg followed by 30 mg/h) reduced ICP and caused a reduction in CBF averaging 15–26%.1,7 

In conclusion, the current study could not demonstrate any evidence of ischemic brain tissue damage associated with indomethacin and propofol, as evaluated by DWI.

The authors thank language secretary Anita Pachai for spell checking the manuscript; research radiographers Dora Zeidler and Anne Dorthe Blankholm, Department of Neuroradiology, Århus University Hospital, Århus, Denmark; and staff at the Department of Neuroradiology, Århus University Hospital, for their professional assistance with the magnetic resonance imaging examinations. Staff anesthesiologist Alp Tankisi, M.D., Department of Neuroanesthesia, Århus University Hospital, is acknowledged for his help with analysis of the blood samples.