Abstract
15O-Gas PET is useful for evaluating hemodynamic status in patients with ischemic cerebrovascular disease. To reduce examination time and exposure to radioactive gas, we assessed a count-based method with shorter continuous 15O2 gas inhalation. Methods: Twenty-five patients (66 ± 13 [mean ± SD] y old) with unilateral cerebrovascular stenoocclusive disease were examined by use of measurements of asymmetric oxygen extraction fraction (OEF) elevation. Dynamic PET scans of 1 min per frame were obtained starting 2 min after the beginning of 15O2 inhalation at a constant flow rate (740 MBq/min). Each subject also underwent C15O and H215O PET with the bolus administration method. To evaluate the effects of different scan start times and durations during 15O2 inhalation, we extracted and summed individual 15O2 PET data from the dynamic 15O2 dataset. Count-based OEF (cbOEF) images were calculated from 15O2 and H215O PET images. The asymmetric indices (AI) of cbOEF (cbOEF-AI) were obtained from regions of interest drawn on territories of the bilateral middle cerebral artery. These AI were compared with the AI of quantitative OEF (qOEF-AI). Results: The slopes of the regression lines and the coefficients of correlation between qOEF-AI and cbOEF-AI were close to 1.00 and greater than 0.79, respectively, regardless of different scan start times and durations. The cbOEF-AI obtained with a longer scan duration were closer to the qOEF-AI than those obtained with a shorter scan duration. Longer scan durations also provided better coefficients of correlation between cbOEF-AI and qOEF-AI regardless of scan start times. The coefficients of correlation between cbOEF-AI and qOEF-AI were greater than 0.90, except for cbOEF-AI obtained from 15O2 images at 2–3 min after 15O2 inhalation. Conclusion: The cbOEF obtained by 15O2 imaging from 4 min after 15O2 inhalation to 7 min or longer can correctly diagnose misery perfusion. The less invasive count-based PET method used in this study will be able to reduce examination time, exposure time, and stress for patients with ischemic cerebrovascular disease.
Oxygen-15-gas brain PET is the most reliable examination technique for evaluating misery perfusion, defined as impaired hemodynamics with a regional increase in the oxygen extraction fraction (OEF), in imaging modalities. Patients with this condition have a higher risk of stroke recurrence than patients with normal OEF (1–4). Several PET methods have been developed and used to calculate quantitative OEF (qOEF) for the diagnosis of misery perfusion (5–9). However, these methods require arterial blood sampling during the PET examination, which necessitates a long examination time and introduces the risk of bleeding from the arterial line.
Count-based OEF (cbOEF) methods can noninvasively evaluate asymmetric increases in OEF with a simple calculation as a substitution for qOEF measurements (3,10–13). In a previous study, we evaluated whether the asymmetry index (AI) of cbOEF (cbOEF-AI) could appropriately detect misery perfusion (13). The cbOEF-AI obtained with continuous 15O2 gas inhalation and bolus H215O injection could correctly estimate the AI of qOEF (qOEF-AI) without a C15O image for cerebral blood volume (CBV) correction, whereas the cbOEF-AI obtained with continuous inhalation of 15O2 and C15O2 required CBV correction (13). Although the former method can reduce the examination time for C15O2 scans, the cbOEF method with continuous 15O2 inhalation still requires a long examination time to achieve an equilibrium of cerebral radioactivity.
The purposes of this study were to reduce examination time and to assess whether a new method with shorter continuous 15O2 inhalation can appropriately evaluate misery perfusion. For these purposes, dynamic PET data acquisition was started before the steady state was reached during continuous 15O2 inhalation. The appropriate examination time for the use of the cbOEF method to evaluate side-by-side OEF differences in cerebrovascular diseases was estimated.
MATERIALS AND METHODS
Subjects
The subjects were 25 patients (15 men and 10 women; age [mean ± SD], 66 ± 13 y) with unilateral cerebrovascular stenoocclusive disease. The patients had occlusion (n = 9) or stenosis (n = 16; diameter reduction >70%) of the unilateral internal carotid artery (n = 20) or the middle cerebral artery (MCA) (n = 3). Because the remaining 2 patients had unilateral arterial occlusion and mild stenosis (<70%) of the contralateral side, that is, right MCA occlusion with mild left MCA stenosis and left internal carotid artery occlusion with mild right internal carotid artery stenosis, the side of the arterial occlusion was defined as the ipsilateral side. Seven patients had experienced transient ischemic attacks; 10 had experienced a nondisabling hemispheric stroke with minor cerebral infarctions, as shown on MRI; and 8 had no neurologic symptoms. The study was approved by the Ethical Committee of the University of Fukui Faculty of Medical Sciences. Written informed consent was obtained from each subject before the study.
PET Procedures
All scans were acquired in the 2-dimensional mode with a whole-body tomography scanner (Advance; GE Healthcare), which permits the simultaneous acquisition of 35 image slices with an interslice spacing of 4.25 mm (14). Performance tests showed the intrinsic resolutions of the scanner to be 4.6–5.7 mm in the transaxial direction and 4.0–5.3 mm in the axial direction. We obtained a blank scan before beginning the PET examination with the 68Ge/68Ga line source. Patients were positioned on the scanner bed with their heads immobilized with a head holder.
Figure 1 shows the timing of PET examinations in the present study. Before tracer administration, a small cannula was placed in the right brachial artery for blood sampling, and a transmission scan was obtained for 10 min in each subject for attenuation correction. For 15O-gas PET scans, a nasal tube was placed in the nose of patients, and an oxygen mask was used to cover the nose and the mouth. 15O-Gas was administered from the tube, and radioactive gas in the expired air was vacuumed from the mask. Subjects inhaled C15O2 as a single dose of 1,000 MBq for CBV measurements (14). PET was started at least 30 s after the tracer count had peaked in the brain and was continued for 3 min. Continuous 15O2 inhalation was started at a constant flow rate of 740 MBq/min. Dynamic PET scans of 1 min per frame were obtained for 11 min starting 2 min after the beginning of continuous 15O2 inhalation. Next, bolus H215O scans were obtained with a 3-min acquisition starting at the time of bolus injection of the tracer at 740 MBq. Arterial blood was sampled, and radioactivity in the blood samples was immediately measured with a scintillation counter to determine arterial blood activity (13,14). Reconstructions of PET data were performed by use of filtered backprojection with a Hanning filter at a cutoff of 6.0 mm full width at half maximum in the transaxial direction.
Scan protocol and data preparation for 15O-gas PET. For cbOEF calculation, subsets of 15O2 dynamic data were extracted at 2, 3, 4, 5, 6, and 7 min (bars below diagram, from top to bottom, respectively) after beginning of 15O2 inhalation up to 13 min. cbOEF-AI was calculated from 15O2 images and H215O PET scans for 3 min. qOEF was calculated from 15O2 images obtained from 8 min to 13 min after 15O2 inhalation and H215O images with arterial blood data. C15O data were used for CBV correction of qOEF.
Parametric images of cerebral blood flow, CBV, OEF, and cerebral metabolic rate for oxygen were calculated from PET images and arterial blood data (5,7,14). In the count-based method, individual 15O2 data were extracted from the original dynamic 15O2 dataset and summed to simulate different acquisition start times and scan durations during 15O2 inhalation (Fig. 1). cbOEF images were calculated from ratios of 15O2 and H215O images by use of 15O2 images variously extracted from the dynamic PET data (Table 2).
Data Analysis
Regional values were obtained from regions of interest (ROIs) drawn bilaterally on 3 slices of the cerebral cortex as reported previously (13). In brief, elliptic ROIs of 15 × 50 mm were placed bilaterally on cortical territories of the MCA at the 3 slice levels of the centrum semiovale. Before the ROIs were placed, each image of hemodynamic parameters was normalized anatomically with SPM2 (Wellcome Department of Cognitive Neurology, Institute of Neurology, London, U.K.). The ROIs placed in the ipsilateral hemisphere of normalized MR images were copied symmetrically at corresponding regions in the contralateral hemisphere. The same ROIs were applied to all parametric images for each subject. The values obtained from the ROIs in each hemisphere were averaged.
The cbOEF-AI was compared with the qOEF-AI, which was obtained from 15O2 images summed for 8–13 min after 15O2 inhalation. The slopes of the regression lines and the coefficients of correlation (r) between cbOEF-AI and qOEF-AI were calculated, and the mean distance of all plots from the line of identity (bias) was determined. The cbOEF-AI was applied to the diagnosis of misery perfusion on the basis of absolute OEF values (>52.0%) obtained from healthy volunteers at the University of Fukui (13,14). The threshold of 1.15 for cbOEF-AI, which was equivalent to that of 1.17 for qOEF-AI in our previous study (13), was applied to the cbOEF method in the present study. Differences between the 2 hemispheres for each hemodynamic parameter were compared statistically by use of repeated-measures analysis of variance and a paired t test. P values of less than 0.05 were considered to indicate a significant difference.
RESULTS
Regional parametric values for each hemisphere in all patients are presented in Table 1. All parameters except for CBV were affected by stenoocclusive lesions in the ipsilateral hemisphere. Table 2 shows the slope, r, and bias obtained from the relationship between qOEF-AI and cbOEF-AI. The slopes of the regression lines were close to approximately 1.00, and the correlations between qOEF-AI and cbOEF-AI were greater than 0.79, although both acquisition start times and durations were different. Longer acquisition times showed a better correlation and smaller bias in the relationship between cbOEF-AI and qOEF-AI, irrespective of the acquisition start times for 15O2 images. Figure 2 shows changes in cbOEF-AI at each 15O2 acquisition start time and duration in 3 patients with misery perfusion. Continuous 15O2 inhalation required at least 7 min to detect misery perfusion at a threshold of 1.15 for cbOEF-AI, even though the acquisition start times were varied.
Changes in cbOEF-AI for 3 patients with misery perfusion, as determined from 15O2 images with various scan durations started at 2 min (A), 3 min (B), 4 min (C), and 5 min (D) after beginning of 15O2 inhalation. Misery perfusion was defined in our previous study as increase in qOEF of 52.0% or greater, which is equivalent to threshold of 1.17 for qOEF-AI (qOEF-AI = 1.17 for • and ▴ and 1.36 for ▪). Threshold of 1.15 for cbOEF-AI in present study (dashed line) was equivalent to threshold of 1.17 for qOEF-AI in our previous study. Regardless of different acquisition start times, continuous 15O2 inhalation for 7 min or longer would be required to correctly detect misery perfusion.
Hemispheric Differences in Cerebrovascular Diseases (n = 25)
Correlations Between qOEF-AI and cbOEF-AI for Different Acquisition Start Times and Scan Durations (n = 25)
DISCUSSION
The cbOEF method could identify patients with misery perfusion, even when 15O2 images were acquired before an equilibrium of blood and brain radioactivities was reached. The major advantage of the cbOEF method is that it is a simpler and quicker method of PET examination than the conventional qOEF method, which requires arterial blood sampling and complicated data analysis. Patients with cerebrovascular diseases in Japan have usually been examined with quantitative methods, such as the steady-state method for 15O-gas tracers (15). Although cbOEF methods that make use of bolus tracer administration to estimate the regional AI of OEF have been developed (10–12), we applied continuous 15O2 inhalation and a bolus H215O injection to calculate the cbOEF in the present study (13). This cbOEF-AI method does not need a C15O image for CBV correction, and the results obtained with this method showed good correlations between qOEF-AI and cbOEF-AI. However, a long waiting time for achieving equilibrium during continuous 15O2 inhalation was required. A shorter examination time for this method was evaluated in the present study by use of dynamic 15O2 data acquisition started before the steady state was reached. The 15O2 images were acquired from 2 min after the start of continuous 15O2 inhalation because the level of radioactivity in the brain was not high enough at an inhalation time of less than 2 min.
As shown in Table 2, even when brief 15O2 inhalation was applied to the cbOEF method, cbOEF-AI showed good linearity with qOEF-AI. The correlations between cbOEF-AI and qOEF-AI were greater than 0.90, except for 15O2 imaging at 2–3 min after 15O2 inhalation. This very early and short scan timing resulted in high levels of noise in the 15O2 images, and quality was poor. Longer scan durations also yielded better correlations and smaller biases between qOEF-AI and cbOEF-AI, despite different acquisition start times.
Three of 25 patients had significantly higher OEF in the ipsilateral hemisphere. 15O2 inhalation times of 7 min or longer should be used to diagnose misery perfusion with the cbOEF method, although the sample population of patients with misery perfusion in the present study may have been small (Fig. 2). To minimize the PET examination time, 15O2 images summed from 4 to 7 min would be appropriate, because they provided the maximal r value (0.97) (Table 2). Therefore, this cbOEF method could reduce 15O2 inhalation time to about half that in the steady-state method. The total examination time should be 30 min or less with this method when CBV correction of OEF is not used.
Cerebral vascular volume is usually increased in the impaired circulation and is considered to affect cbOEF values. However, in our previous study, CBV correction did not improve the accuracy of estimation of the AI of OEF with the bolus 15O-water injection method, whereas estimation with continuous C15O2 inhalation required CBV correction for a better correlation (13). This is why we used the former method without CBV correction. A simple 15O-gas PET method would be promising for the evaluation of cerebral hemodynamics, especially with a PET/CT scanner, which can lead to difficulties in arterial sampling because of its deep gantry. The use of new scanners, such as PET/CT and PET/MRI scanners, would be useful for motion correction during and between scans because precise attenuation correction would be beneficial for the cbOEF method, which compares the left-to-right asymmetric changes. On the other hand, in the application of 15O-gas methods for 3-dimensional-mode scanners (as opposed to 2-dimensional-mode scanners), random and scattered coincidences from high levels of radioactivity in the nasal cavity and the body should be carefully corrected.
CONCLUSION
For cbOEF-AI with continuous 15O2 inhalation, PET acquisition from 4 to 7 min after the start of 15O2 inhalation provided an appropriate estimation of qOEF-AI even before an equilibrium of radioactivity was reached and could properly diagnose patients with misery perfusion. The cbOEF method described here can considerably reduce examination time, exposure time, and stress for patients compared with conventional methods.
Acknowledgments
The authors thank Shingo Kasamatsu, Katsuya Sugimoto, Tetsuya Mori, and other staff of the Biomedical Imaging Research Center, University of Fukui, for their technical and clinical support. This study was partly funded by Grants-in-Aid for Scientific Research from the Japan Society for the Promotion of Science (17209040, 18591334, and 19790861) and by the 21st Century COE Program (Medical Science).
Footnotes
-
COPYRIGHT © 2008 by the Society of Nuclear Medicine, Inc.
References
- Received for publication September 9, 2007.
- Accepted for publication October 29, 2007.
Jump to section
Related Articles
Cited By...
- No citing articles found.