| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Pediatric Imaging |
1 From the Departments of Radiology (D.W.W.S., E.W.) and Anesthesiology (C.F., D.S.J.), University of Washington School of Medicine, Seattle, Wash; and Department of Epidemiology, University of Washington School of Public Health and Community Medicine, Seattle, Wash (S.R.H.). Received August 27, 2003; revision requested November 10; revision received January 14, 2004; accepted February 4. Supported by academic funding from the Department of Anesthesiology at the University of Washington and the use of equipment from the Thermo Respiratory Group. Address correspondence to C.F., Department of Anesthesiology, Montreal Children’s Hospital, 2300 rue Tupper, Montreal, QC, Canada H3H 1P3 (e-mail: chantal.frigon@muhc.mcgill.ca).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
MATERIALS AND METHODS: At brain MR imaging during general anesthesia with propofol, two FLAIR sequences were performed in 20 children with American Society of Anesthesiologists physical status classification system grades of 3 or lower. The first FLAIR sequence was performed with the child breathing 100% oxygen; the second was performed with the child breathing 30% oxygen. CSF signal intensity was quantified on a three-point ordinal scale (0 = hypointense to brain parenchyma, 1 = isointense to brain parenchyma, 2 = hyperintense to brain parenchyma) by a pediatric neuroradiologist who was blinded to the FIO2 level. The Wilcoxon signed rank test was used to determine if CSF hyperintensity was correlated with FIO2.
RESULTS: CSF hyperintensity was present in all 20 children (age range, 1.9–16.7 years; 12 children were boys) when the FIO2 was 100%. The hyperintensity partially or completely disappeared in the basilar cisterns (P < .001) and cerebral sulcal subarachnoid space (P < .001) after FIO2 was reduced from 100% to 30%.
CONCLUSION: These findings are consistent with the hypothesis that increased arterial oxygen tension and consequently increased CSF PO2 resulting from administration of high FIO2 during general anesthesia are responsible for the increased CSF signal intensity noted on brain FLAIR MR images.
© RSNA, 2004
Index terms: Anesthesia • Brain, MR, 10.121413, 10.121415 • Cerebrospinal fluid, MR, 167.121413, 167.121415 • Magnetic resonance (MR), in infants and children, 10.121413, 10.121415, 167.121413, 167.121415 • Oxygen
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
In other reports, however, propofol or other anesthetic agents have been suggested as the potential cause of the CSF hyperintensity. Recent in vitro observations (5) support the hypothesis that increased oxygen tension is responsible for CSF hyperintensity during FLAIR MR imaging. To address these differing conclusions, we conducted a prospective study designed to test the hypothesis that high levels of the fraction of inspired oxygen (FIO2) during general anesthesia cause subarachnoid CSF hyperintensity during FLAIR MR imaging.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
Children were excluded if they were suspected of having conditions (such as meningitis or subarachnoid tumor) that are known to produce hyperintense signal in the CSF at FLAIR MR imaging. To ensure that high FIO2 led to a substantial elevation of the PaO2 (partial pressure of arterial oxygen), children were excluded if they had cyanotic heart disease, obesity (>140% of ideal weight), or pulmonary disease other than mild asthma. Children who were allergic to propofol were also excluded.
After anesthesia was induced intravenously or with a mask, each child’s airway was secured with an endotracheal tube or a laryngeal mask airway. Anesthesia was maintained with an infusion of 125–250 µg of propofol (Diprivan; AstraZeneca) per kilogram of body weight per minute, and children breathed spontaneously through a Mapleson circuit with 100% oxygen during transport from the induction room to the MR imaging unit.
MR Imaging
MR imaging was performed with a 1.5-T Symphony MR imaging unit (Siemens, Erlangen, Germany). Two FLAIR sequences (repetition time msec/echo time msec/inversion time msec, 10 000/117/2500) were performed in each child. The first FLAIR sequence was performed with the child breathing 100% oxygen because the induction of anesthesia and transportation were accomplished with administration of 100% oxygen. Immediately after induction of anesthesia, a transcutaneous oxygen monitor (MicroGas 7650 Sensor; SensorMedics, Yorba Linda, Calif) was placed on the skin of the forearm. Because the sensor was not MR imaging compatible, the transcutaneous oxygen measurement was obtained and the sensor was removed just before the child entered the MR imaging unit.
Immediately after the completion of the first FLAIR sequence, the FIO2 was turned down to a 30% oxygen-air mixture and kept at that level for the remainder of the examination. A second FLAIR sequence was performed at the end of the examination while the FIO2 was 30%. To allow equilibration to occur at the lower FIO2, the second FLAIR sequence was performed at least 9 minutes after the oxygen had been turned down. The propofol infusion remained on throughout the examination, including the performance of both FLAIR sequences. Both FLAIR sequences were performed before contrast material enhancement if contrast material was needed for the elective brain MR imaging examination.
Data Collection
Arterial oxygen saturation, end-tidal carbon dioxide (ETCO2), infusion rate of propofol, ventilation method, transcutaneous oxygen level, time it took to perform each FLAIR sequence, and demographic characteristics (age, sex, and weight) were noted (C.F.). The anesthesiologist and the radiologist performing the examination were aware of the FIO2 level. However, the study pediatric neuroradiologist (D.S.), who had 15 years of experience in interpreting pediatric brain MR images, evaluated the FLAIR MR images for CSF hyperintensity while blinded to the FIO2 level.
To ensure that the study radiologist was unaware of the FIO2, because the FLAIR sequence with the higher FIO2 had always been performed first, the order in which the study radiologist read the examination results was determined by using a random number table. Film annotations were covered with a mask at the time of the reading to hide the time of the examination and any patient-identifying information. To quantify CSF hyperintensity in the basilar cisterns and cerebral sulcal subarachnoid space, the radiologist used an ordinal scale from 0 to 2 in which a score of 0 indicated that the CSF was hypointense to brain parenchyma (normal CSF signal intensity on FLAIR MR images); a score of 1, that the CSF was isointense to brain parenchyma (mild abnormality in the CSF signal intensity); and a score of 2, that the CSF was hyperintense to brain parenchyma (marked abnormality in the CSF signal intensity).
Statistical Analysis
Data were analyzed by using the STATA 7 statistics package (STATA Corporation, College Station, Tex). The Wilcoxon signed rank test was used to determine if CSF hyperintensity was observed more frequently in the presence of high or low FIO2. Arterial oxygen saturation, ETCO2, and propofol infusion rate were compared for the two FIO2 levels by using the Wilcoxon signed rank test because these data were not normally distributed. P < .05 was considered to indicate a statistically significant difference.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
|
|
|
|
|
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
CSF hyperintensity is usually caused by diseases such as blood conditions or other conditions that result in increased protein concentration in the CSF (1). The hyperintensity is a result of shortened T1 relaxation time associated with the increased protein in the CSF.
The mechanism that can explain CSF hyperintensity in patients without CSF abnormalities who are receiving supplemental oxygen relates to the fact that oxygen, being slightly paramagnetic, also produces a mild shortening in T1 relaxation time, and the increased arterial oxygen will result in increased CSF oxygen tension.
Although measurement of CSF PO2 in our patients was not possible, there is evidence that breathing 100% oxygen will elevate the CSF oxygen tension. In dogs, an arterial PO2 of 585 mm Hg after the administration of 100% oxygen for 10 minutes resulted in a cisternal CSF PO2 of about 155 mm Hg (6). Dunkin and Bondurant (7) also demonstrated that patients who hypoventilate and become hypercarbic increase their CSF PO2 by augmenting their cerebral blood flow. Hypercarbia was not a factor for the abnormal FLAIR MR images observed in the present study because the ETCO2 of the children remained constant between the two FLAIR sequences.
Prior investigators have speculated that anesthetic agents might be responsible for the CSF hyperintensity seen on FLAIR MR images obtained in anesthetized patients, while others have proposed that supplemental oxygen is responsible. After demonstrating that the T1 value of propofol approaches that of CSF, Filippi et al (3) suggested that propofol might have been responsible for the CSF hyperintensity that they retrospectively documented in anesthetized patients. Although it is possible that propofol may have caused mild CSF hyperintensity in a minority of our cases, this would not fully explain the association of CSF hyperintensity with administration of 100% oxygen that we observed. If propofol was the primary cause of CSF hyperintensity on FLAIR MR images, the proportion of abnormal FLAIR MR images should have been the same at low and at high concentrations of oxygen because the patients were receiving essentially the same propofol infusions in both circumstances. Instead, we found that CSF hyperintensity was observed with much greater frequency and degree among patients breathing 100% oxygen.
In vitro evidence for oxygen’s potential effect on the CSF signal intensity has been reported by Deliganis et al (5), who evaluated the MR imaging properties of tubes containing saline and various anesthetic agents (propofol, sevoflurane, or isoflurane) in room air and at 100% oxygen. They found minimal differences in T1 between the various anesthetic agents and saline but a significant T1 shortening for all the solutions at 100% oxygen. In the same study, the authors also conducted a retrospective evaluation of MR images obtained in nine patients and a prospective evaluation in a single volunteer. The authors found that CSF hyperintensity on FLAIR MR images was associated with high FIO2.
The present study had some limitations. CSF hyperintensity remained in some patients (but to a lesser degree) when they were breathing 30% oxygen. Mild CSF hyperintensity was seen in the basilar cisterns in eight patients (40%) and in the cerebral sulcal subarachnoid space in seven patients (35%) after the FIO2 was lowered to 30%. Because all the patients initially received 100% oxygen, which was reduced to 30% for the second FLAIR sequence, it is possible that the oxygen tension in the CSF did not have time to equilibrate with the blood and remained high when the second set of FLAIR images was obtained. It is alternatively possible that breathing 30% oxygen may elevate the CSF oxygen tension sufficiently in some patients to cause mild CSF hyperintensity. According to the alveolar gas equation, the administration of 30% oxygen could result in a maximum PaO2 of 166 mm Hg in an intubated patient (8).
The transcutaneous oxygen monitor used in this study was not MR imaging compatible; hence, we could not measure the PaO2 while the patient was in the imaging unit and had an FIO2 of 30%. Transcutaneous PaO2 measurement was also limited by clinical time constraints. Optimum measurement accuracy required up to 30 minutes of equilibration to allow tissue warming under the sensor probe. However, so that the clinical procedure would not be delayed, the measurements were obtained in less time. This would tend to decrease the PaO2 measurement obtained and may account for the lower PaO2 measured with the patients at an FIO2 of 100% than would be predicted by using the alveolar gas equation (according to which an FIO2 of 100% could yield a PaO2 of up to 671 mm Hg). It is also possible that the lower measured PaO2 accurately reflected a lower value that resulted from reduction of lung volume and pulmonary atelectasis associated with the induction of anesthesia.
Our ability to establish the proposed causality was limited in this study by the lack of direct CSF PO2 measures to correlate with CSF hyperintensity; instead we relied on indirect measures of FIO2 and transcutaneous PO2 and extrapolation from results of animal models (as discussed above). We did take steps to reduce the likelihood that this investigation would be biased. This study had a crossover design, which permitted each child to serve as his or her own control subject. To reduce the possibility of observer bias, the radiologist interpreted the images obtained with the FLAIR sequences without knowledge of the FIO2 level or the temporal order in which the sequences had been performed.
Finally, the hyperintensity was observed in the cisterns and sulci but not in the ventricles. Our speculation is that the relative exposure per unit volume of CSF to oxygen diffusing from blood is much greater in the CSF along the pial-lined surfaces of the brain in the sulci and cisterns than in the large volume of CSF in the ventricles, which, although they are exposed to the vascular choroid plexus, are primarily lined by less vascular ependyma.
In conclusion, the results of this prospective study demonstrate that increased arterial oxygen tension, resulting from administration of high FIO2 during general anesthesia, correlates with increased CSF signal intensity on brain FLAIR MR images and that increased arterial oxygen resulting in increased CSF PO2 is probably the cause of the CSF hyperintensity. Patients who are breathing gas with a high FIO2 may show hyperintense CSF on FLAIR MR images in the absence of CSF disease.
|
FOOTNOTES |
|---|
Authors stated no financial relationship to disclose.
Author contributions: Guarantors of integrity of entire study, C.F., D.W.W.S.; study concepts, C.F., D.W.W.S., D.S.J.; study design, C.F., D.W.W.S., D.S.J., S.R.H.; literature research, C.F., D.W.W.S.; clinical studies, C.F., D.W.W.S., E.W.; data acquisition, C.F.; data analysis/interpretation, C.F., D.W.W.S.; statistical analysis, C.F., S.R.H.; manuscript preparation, C.F.; manuscript definition of intellectual content and final version approval, C.F., D.W.W.S., E.W., D.S.J.; manuscript editing, C.F., D.W.W.S.; manuscript revision/review, all authors
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  B.E. Hamilton and G.M. Nesbit Delayed CSF Enhancement in Posterior Reversible Encephalopathy Syndrome AJNR Am. J. Neuroradiol., March 1, 2008; 29(3): 456 - 457. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. L. Stuckey, T. D. Goh, T. Heffernan, and D. Rowan Hyperintensity in the Subarachnoid Space on FLAIR MRI Am. J. Roentgenol., October 1, 2007; 189(4): 913 - 921. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 F. T. Braga, A. J. da Rocha, R. B. Fonseca, C. Frigon, D. W. W. Shaw, S. Heckbert, and E. Weinberger Supplemental Oxygen Concentration and Increased Signal Intensity of Cerebrospinal Fluid on FLAIR MR Images * Dr Frigon and colleagues respond: Radiology, June 1, 2005; 235(3): 1085 - 1086. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
