ORIGINAL ARTICLE 

Annals of Nuclear Medicine Vol. 14, No. 6, 433-440, 2000 

A new method to estimate rCBF using IMP and SPECT without 
ny blood sampling 

Sadahiko NISHIZAWA,* Toshiki SHIOZAKI,* Makoto UENO,* Hiroshi TOYODA,* Taro SHIMONO,* Yoko KAMOTO,* Toru FUJITA,* Yoshiharu YONEKURA,** and Junji KONISHI* 

*Department of Nuclear Medicine and Diagnostic Imaging, Graduate School of Medicine, Kyoto University 
**Biomedical Imaging Research Center,  Fukui Medical University 

We developed and evaluated a method to measure rCBF without any blood sampling by using iodine-123IMP and SPECT. An integral of arterial input function, int Ca(t)dt, can be expressed as TC(T)/CO, where TC(T) is radioactivity delivered to the body in T minutes and CO is cardiac output. If T is acceptably small, rCBF can be determined by means of a microsphere model analysis with IMP as Cb(T)/(TC(T)/CO), where Cb(T) is cerebral radioactivity at T minutes. We derived TC(T) and CO from a chest dynamic scan. The method was applied to 45 patients who underwent rCBF studies (58 studies) with arterial blood sampling (ABS). Data from the chest scan were analyzed in comparison with ABS data in the first 28 studies, and equations for correction yielding an accurate TC(T)/CO were derived. The validity of the proposed method was evaluated in the subsequent 30 studies. The method yielded rCBF (rCBF-test) which agreed well with rCBF obtained by a two-compartment model analysis of dynamic SPECT and ABS data (rCBF-ref) with the mean and SD of differences between rCBF-test and rCBF-ref being 1.0 and 2.7 ml/100 g/min, respectively. In eleven subjects who underwent more than two studies, a percentage change in rCBF-test between the studies also closely approximated that of rCBF-ref (y=1.11x+2.63, r=0.92). The method can be used with acceptable reliability to measure rCBF without any blood sampling. 

Key words: iodine-123 IMP, regional cerebral blood flow, SPECT 


INTRODUCTION 

Iodine-123-N-isopropyl-p-iodoamphetamine (IMP) and single photon emission computed tomography (SPECT) have been used for the quantitative measurement of regional cerebral blood flow (rCBF) according to the microsphere (MS) model1-3 or two-compartment model4-7 analysis. Because of a linear relationship between IMP uptake and blood flow in regional brain tissues and the favorable chemical properties of IMP in SPECT acquisition and 

 Received February 23, 2000, revision accepted September 21, 2000. 
 For reprint contact: Sadahiko Nishizawa, M.D., Ph.D., Biomedical Imaging Research Center, Fukui Medical University, 23 Shimoaizuki, Matsuoka-cho, Yoshida-gun, Fukui 910-1193, JAPAN. 
 E-mail: sadahiko@fmsrsa.fukui-med.ac.jp 

determining the concentration of IMP in blood, these methods proved to provide reliable rCBF values. Recently several simplified methods have been reported,8-10 but these methods still require arterial blood sampling (ABS) which is invasive and demanding in a daily clinical setting. 
 A recent report suggested that rCBF could be calculated with IMP and SPECT without any blood sampling.11 The method uses a planar dynamic scan of the chest for 3 minutes after an injection of IMP to estimate cardiac output and the integral of the input function (see Theory). The method is theoretically promising but a validation study is needed before routine clinical use. In this study, we tested the method in a group of patients with a variety of neurological disorders by comparing the input function and rCBF estimated by the method with those obtained by using a two-compartment model analysis of dynamic SPECT and ABS data. 


MATERIALS AND METHODS 

Theory 
According to the microsphere model, rCBF (ml/g/min) is calculated as follows, 
where Cb(T) (uCi/g) is regional radioactivity of the brain tissue at T minutes after an injection of the tracer, and Ca(t) (uCi/ml) is the radioactivity-time course of the tracer in arterial blood. If cardiac output (CO: ml/min) and the total amount of tracer delivered to the whole body in T minutes (TC(T): uCi) are known, int Ca(t)dt (min*uCi/ml) can be replaced by TC(T)/CO (min*uCi/ml). 
 The cerebral perfusion tracer, IMP, is known to accumulate in the lungs after an intravenous injection and is gradually released from the lungs into the systemic circulation. As a chemical microspheric substance, a lipophilic portion of IMP goes into the brain and is retained without significant loss from the brain, at least for several minutes after an injection. Assuming that peak counts of the lung represent the injected dose and re-circulation of the tracer is negligible in T minutes. TC(T) is replaced by an injected dose (D: uCi) multiplied by the ratio of clearance from the lung in T minutes (R(T)) and a mean lipophilic fraction of IMP (Fr) as follows, 
From the peak activity of the lung (Lpeak) and activity of the lung at T minutes (L(T)), R(T) is calculated as follows, 
 Cardiac output can be estimated from first pass data for the right heart or pulmonary artery trunk of a chest dynamic scan after an injection of IMP.11,12 Therefore, rCBF can be calculated non-invasively by using the microsphere model with IMP as follows, 

Subjects 
Forty-five patients (age: 59.1;-15.4, male/female: 36/9) with a variety of neurological disorders were included in this study (Tables 1a and 1b). As nine patients were studied twice and two were studied three times before and after therapeutic procedures or during the course of clinical follow-up, fifty-eight studies were included in this study. During the study, frequent ABS was performed in all patients. We divided the 58 studies into 2 groups. The first 28 studies in 23 patients (age: 56.7+-17.7, male/female: 17/6) were analyzed in comparison with ABS data to derive equations for correction yielding an accurate input function from the chest dynamic scan data (first group, Table 1a). The subsequent 30 studies in 22 patients (age: 61.6+-12.4, male/female: 19/3) served to confirm the validity of the proposed method with the equations for correction (second group, Table 1b). 
 The study was performed according to the guidelines of the ethical committee of Kyoto University Hospital for human studies, and all subjects gave informed consent. 

General procedure 
We used a three-head SPECT scanner (PRISM3000, Picker Int. Ohio) equipped with a low energy general purpose collimator. Patients underwent a dynamic planar scan of the chest with one of the detectors for 3 minutes (0.5 sec/frame, 7.1x7.1 mm/pixel, 64x64 matrix) followed by a dynamic SPECT scan of the brain for 20 minutes (1 min/frame) after a bolus injection of IMP (Fig. 1). Twenty 1-minute raw data were reconstructed to ten 2-minute SPECT frames (3.4x3.4 mm/pixel with 14.2 mm in thickness, 64x64 matrix) by a filtered back-projection method with a Butterworth filter. One 10-minute SPECT frame from about 10 to 20 minutes after an injection was also reconstructed for subjects in the second group. Details of performance of the scanner and image reconstruction procedures were reported previously.6 
 We loaded IMP (111-167 MBq in 1-1.5 ml) in an extension tube (volume 3.8 ml) connected to a 10-ml syringe in advance and injected it as a bolus into the antecubital vein in 5 seconds with 10 ml of saline. Radioactivity in a syringe and extension tube was measured before and after an injection by means of the scanner in all studies and by means of a dose calibrator In 39 studies to confirm the dose which entered the body. For measurement with the scanner, the distance between the surface of the collimator and the syringe containing the tracer was set at 10 cm in all studies. Because radioactivity measured with the scanner agreed well with that measured with a dose calibrator, with the mean and SD of differences between two measurements being 1.0% and 3.9%, respectively, we measured an injected dose (D in Eq. 2) by means of the scanner only in the rest of the studies. 
 Data acquisition for the brain was started about 6 minutes' after an injection of IMP (mean +- SD: 6.06+-0.93 minutes in the first group and 5.66+-0.61 minutes in the second group, no significant difference between the two groups). During scans, nineteen arterial blood samples were taken through a small catheter inserted into the radial artery (every 5 seconds from 10 to 50 seconds, at 60, 75, 90, 120 and 150 seconds, and at 3, 5, 10, 20 and 30 minutes after an injection). 

Estimation of cardiac output 
First pass data for the pulmonary artery trunk on a chest dynamic scan were used to estimate CO.11,12 This procedure was a modified version of the established procedure for measuring CO with a Tc-99m-labeled blood pool tracer.13,14 Briefly, applying the concept of the Stewart-Hamilton principle, the index of CO (COI) is calculated as follows,
where Dc (counts/min) is an injected dose counted by means of a gamma camera for one minute, A (cm2) is the area of a ROI placed on the pulmonary artery trunk and B (counts) is the area under a time-activity curve from the ROI. A rectangular ROI 14.2x14.2 mm in size was used as A. A least squares exponential fit to a downward slope of the curve was performed to calculate B13 (Fig. 2). COI (unit: cm2/min) was converted to CO (liter/min) according to an equation which was derived from a relationship between COI and CO measured by the cardiac echo Doppler method.12 Accuracy and reproducibility of CO measurement by the cardiac echo Doppler method have been well validated.15,16 The equation used for conversion from COI to CO in this study was as follows, 

Estimation of an input function from a lung curve
To estimate an accurate TC(T) from a lung washout curve for 3 minutes, we analyzed the washout curve in comparison with ABS data in the first group of 28 studies. Two large ROIs covering both lung fields were drawn to obtain a curve for washout of the tracer from the lungs (Fig 3). The washout ratio in 3 minutes (R(3 )) was defined as (Lpeak-L(3))/Lpeak (L(3): radioactivity in the lungs at 3 minutes after an injection). As a first step, to test assumptions described in the theory that peak counts for the lungs represented an injected dose and re-circulation of the tracer was negligible, we compared R(3) with the corresponding value calculated with CO, D and ABS data. Through this comparison, we derived an equation for correction yielding more accurate R'(3) from R(3). In the second step, we compared two procedures to estimate the washout ratio in T minutes (R'(T)) from R'(3). In the first procedure we used an extrapolated lung washout curve (R(t)) obtained by fitting a downward slope of the curve with a two-exponential curve (therefore, R'(T)=R'(3)*(R(T)/R(3)). In the second procedure, we estimated R'(T) from R'(3) multiplied by a population mean of int.Ca(T)dt/ int.Ca(T)dt (standard ratio: SR(T)), which was calculated by using the data from the first 28 subjects (therefore, R'(T)=SR(T)*R'(3)). Ratios of R'(T) to R'(3) obtained through these procedures were compared to individual int.Ca(T)dt/ int.Ca(T)dt obtained by using ABS data. 

Determination of cerebral blood flow 
Four large regions of interest (ROIs) were manually drawn for each patient. One ROI covered the whole brain and two ROIs included large cortical areas of each cerebral hemisphere in a slice of the corona radiata and lateral ventricles. The forth ROI was set on the cerebellum. The ROI covering the whole brain was used to calculate mean CBF (mCBF-test) by using data from the chest dynamic scan and the first frame of the dynamic SPECT scan as follows, 
where Cwb(T) is total radioactivity in the brain in one image slice. Then, to reduce statistical noise, regional values for CBF (rCBF-test) were calculated on 10-minute SPECT images by redistributing the value for mCBF-test according to ratios of regional to total radioactivity of the brain. 
 A two-compartment model analysis of the dynamic SPECT and ABS data was used to determine CBF in the four ROIs used as references (mCBF-ref for the whole brain CBF and rCBF-ref for regional CBF). Because the dynamic SPECT in this study lacked data for first 5-6 minutes, we confirmed that rCBF determined with SPECT data lacking for the first 6 minutes agreed well with rCBF determined with a full data set in another group of 35 patients (Fig. 4). We used 0.75 for a lipophilic fraction of IMP in the calculation of mCBF-test, whereas a standard time-course of a lipophilic fraction of IMP previously determined in 20 subjects was used for ABS data in the two-compartment model analysis. 

Evaluation of the method 
The validity of the proposed method was evaluated in the second group of 30 studies. The reliability and reproducibility were assessed by using the mean and SD of differences in obtained values between two methods or between two studies.17 Linear regression analyses were also per-formed to correlate obtained values between two methods or between two studies. To exemplify the reliability of the method in a wide range of CBF, we made a comparison of rCBF-test and rCBF-ref, which were obtained from regions with high CBF (normal cortex and cerebellum) and low CBF (ischemic cortex). The reproducibility of the method was assessed by seeing the stability and/or alteration of CO, R'(3) and rCBF-test in thirteen pairs of two consecutive studies. Because CO and R'(3) did not seem to be affected by therapeutic procedures for the brain, we compared CO and R'(3) in two studies. Regarding rCBF, there might bc a difference between two studies because nine out of 11 patients underwent cerebral revascularization surgery. We, therefore, compared changes in the rCBF-test from the first to the second study with those of rCBF-ref. Thirteen consecutive studies were available in 11 patients, and two cortical ROIs in each patient were used for the analysis. 


RESULTS 

In the 28 studies of the first group, CO was in a physiological range between 3.78 and 5.74 liter/min (mean+-SD: 4.62+-0.49), R(3) between 0.12 and 0.50 (0.32+-0.10) and mCBF-ref between 16.8 and 40.5 ml/100g/min (27.5+-6.7). Because R(3) was linearly correlated with, but constantly smaller than, a corresponding value, CO*int.Ca(T)dt/D (Fig. 5), we defined R'(3) as follows, 
 The standard ratio of int.Ca(T)dt to int.Ca(T)dt (SR(T)) obtained from ABS data in the first group was expressed as follows (Fig. 6), 
Variations in SR(T) were small, especially within 10 minutes after an injection (coefficient of variation at 7 minutes: 2.9%), and SR(T) represented the individual int.Ca(T)dt/int.Ca(T)dt better than that obtained from the extrapolated lung curve (Fig. 7). Therefore, the final equation to calculate mCBF-test was determined as follows, 
In the 30 studies of the second group, CO ranged between 3.63 and 7.55 liter/min (mean+-SD: 5.24+-0.81), R(3) between 0.14 and 0.49 (0.34+-0.10) and mCBF-ref between 18.1 and 37.8 ml/100g/min (28.2+-5.3). There was a significant difference in CO (p<0.01), but not in R(3) and mCBF-ref, between the first and second groups. When Eq. 10 was applied to this group, the rCBF-test correlated and agreed well with rCBF-ref, showing a correlation coefficient of 0.92 (Fig. 8a) and the mean and SD of differences between rCBF-test and rCBF-ref of 1.0 and 2.7 ml/100g/min, respectively (Fig. 8b). 
 In patients who underwent more than two studies on separate days, percentage changes in rCBF-test between the two consecutive studies showed a linear relationship with those in rCBF-ref (Fig. 9). The mean and SD of differences between changes in rCBF-test and rCBF-ref were 3.6 and 8.8%, respectively. The mean and SD of differences between the two consecutive studies were 0.33(6.3%) and 0.86 (16.5%) liter in CO, and 0.001 (0.1%) and 0.031 (7.1%) in R'(3), respectively. 


DISCUSSION 

In this report we demonstrated that rCBF could be calculated by using IMP and SPECT without any blood sampling by adding a chest dynamic scan for 3 minutes after an injection of IMP. But assumptions referred to in the theory, that peak counts of the lung represented the injected dose and that re-circulation of the tracer was negligible for several minutes, did not seem to be valid in our procedures. The input function obtained from dynamic chest scan data was systematically smaller than the input function obtained from ABS data. It is well known that smokers have slow washout of IMP from the lung.18 Peak counts of the lung would be affected by how rapidly the tracer was given as a bolus. Photon attenuation by the chest wall and/or a partial volume effect may result in errors in the estimation of CO. The equation for conversion from COI to CO was derived from the comparison with the cardiac echo Doppler method.12 Therefore, errors in the cardiac echo Doppler method may propagate, eventhough the accuracy and reproducibility of the method have been validated.15,16 There are many factors that cause errors in estimating the input function and it seems difficult to analyze each of them. Therefore, we found a practical solution. We derived two equations from the first group of patients to correct an input function obtained from the lung washout curve by analyzing with ABS data. These equations worked well in the second group of patients that was independent of the first group, indicating the validity of the methods with these equations for correction in the general population. 
 The mean and SD of differences between rCBF-test and rCBF-ref were 1.0 and 2.7 ml/100g/min (3.2 and 8.4%. respectively, when expressed as a percentage), indicating that the reliability was in an acceptable range as a method without any blood sampling. Okubo et al.19 evaluated the accuracy and reliability of several simplified methods for quantifying rCBF by using IMP and SPECT. They compared rCBF obtained with those simplified methods with that obtained by means of a non-linear least squares fitting (NLLSF) method based on a two-compartment model analysis, that was the same as that used in this study. Their results showed a mean error of 6.8% when the microsphere model with continuous ABS was used and 10.4% when the microsphere model with a standard input function calibrated with an arterial blood sample at a single time point. Although they did not show a definition of the mean error, it was probably an average of absolute differences expressed as a percentage between the rCBF tested and rCBF obtained by means of the NLLSF method divided by rCBF obtained by means of the NLLSF method. We obtained a mean error of 7.0% in this study when calculated in this way, and value was comparable with theirs. Therefore, our method seems to be sufficiently reliable as a method without any blood sampling and acceptable for clinical application. 
 We could not evaluate the reproducibility of the method in the test-retest conditions in our clinical study but we compared the CO and R'(3) for the two studies, because CO and R'(3) did not seem to be affected by therapeutic procedures for the brain. Although CO was not well reproduced in some patients, it may have varied due to physical and mental stress during the study, especially at the time of the injection. Compared with CO, R'(3) was well reproduced with an acceptably small SD of differences (7.1%). The reproducibility of the rCBF-test may be difficult to assess because nine of 11 patients received a revascularization procedure between the two studies but we found a linear relationship between changes in the rCBF-test and rCBF-ref between the first and second studies. An increase or decrease in the rCBF-ref after therapeutic procedures was well represented by the rCBF-test and the change in the rCBF-test was small in patients without a significant change in the rCBF-ref. 
rCBF values calculated by the proposed method were considerably lower than generally accepted rCBF values,20 but were comparable to the reported rCBF values obtained with IMP and SPECT6-8,19,21 and those obtained by means of a two-compartment model analysis of dynamic SPECT and ABS data in this study. This underestimation of rCBF can be mainly attributable to the limited spatial resolution of the SPECT scanner and the limited extraction fraction of IMP as suggested previously.7-9,20 As shown by the reliability and reproducibllity in this study, the method seems to be sufficiently acceptable for clinical use. Although the measurement of rCBF may not be necessary for many patients, it is informative for management of patients with cerebrovascular diseases, especially those with an occlusive disease of the cerebral major artery, by revealing the magnitude of ischemia, monitoring response to therapy and predicting the outcome.21 This noninvasive method that can be performed in usual clinical settings is widely applicable and beneficial for patients. 


REFERENCES 

1. Kuhl DE, Barrio JR, Huang SC, Selin C, Ackermann RF, Lear JL, et al. Quantifying local cerebral blood now by Nisopropyl-p-[123I]iodoamphetamine (IMP) tomography. J Nucl Med 23: 196-203, 1982. 
2. Podreka I, Baumgartner C, Suess E, Muller C, Brucke T, Lang W, et al. Quantification of regional cerebral blood flow with IMP-SPECT. Reproducibility and clinical relevance of flow values. Stroke 20: 183-191, 1989. 
3. Murase K, Tanada S, Mogami H, Kawamura M, Miyagawa M, Yamada M, et al. Validity of microsphere model in cerebral blood flow measurement using N-isopropyl-(iodine-123) iodoamphetamine. Med Phys 17: 79-83, 1990. 
4. Greenberg JH, Kushner M, Rango M, Alavi A, Reivich M. Validation studies of iodine-123-iodoamphetamine as a cerebral blood flow tracer using emission tomography. J Nucl Med 31: 1364-1369, 1990. 
5. Yonekura Y, Nishizawa S, Mukai T, Iwasaki Y, Fukuyama H, Ishikawa M, et al. Functional mapping of flow and back-diffusion rate of N-isopropyl-p-iodoamphetamine in human brain J Nucl Med 34: 839-844, 1993. 
6. Nishizawa S, Yonekura Y, Tanaka F, Fujita T, Tsuchimochi S, Ishizu K, et al. Evaluation of a double-injection method for sequential measurement of cerebral blood flow with iodine-123-iodoamphetamine. J Nucl Med 36: 1339-1345, 1995. 
7. Yokoi T, Iida H, Itoh H, Kanno I. A new graphic plot analysis for cerebral blood flow and partition coefficient with iodine-123-iodoamphetamine and dynamic SPECT validation studies using oxygen-15-water and PET. J Nucl Med 34: 498-505, 1993. 
8. Iida H, Itoh H, Nakazawa M, Hatazawa J, Nishimura H, Onishi Y, et al. Quantitative mapping of regional cerebral blood flow using iodine-123-IMP and SPECT. J Nucl Med 35: 2019-2030, 1994. 
9. Iida H, Itoh H, Bloomfield PM, Munaka M, Higano S, Murakami M, et al. A method to quantitate cerebral blood flow using a rotating gamma camera and iodine-123 iodoamphetamine with one blood sampling. Eur J Nucl Med 21: 1072-l084, 1994. 
10. Odano I, Okubo M, Takahashi M. Quantification of cerebral blood flow and partition coefficient using iodine-123-iodoamphetamine. J Nucl Med 38: 1248-1253, 1997. 
11. Yonekura Y, Sugihara H, Taniguchi Y, Aoki E, Furuichi K, Miyazaki Y. Quantification of brain perfusion SPECT with N-isopropyl-p-iodoamphetamine using noninvasive microsphere method: estimation of arterial input by dynamic imaging KAKUIGAKU (Jpn J Nucl Med) 34: 901-908, 1997. 
12. Sugihara H, Yonekura Y, Miyazaki Y, Taniguchi Y. Estimation of cardiac output by first-pass trans it of radiotracers. Ann Nucl Med 13: 299-302, 1999. 
13. Glass EC, Rahimian J, Hines HH. Effect of region of interest selection of first-pass radionuclide cardiac output determination. J Nucl Med 27: 1282-1292, 1986. 
14. Kelback H, Hartling OJ, Skagen K, Munck O, Henriksen O, Godtfredsen J. First-pass radionuclide determination of cardiac output: an improved gamma camera method. J Nucl Med 28: 1330-1334, 1987. 
15. Dubin J, Wallerson DC, Cody RJ, Devereux RB. Comparative accuracy of Doppler echocardiographic methods for clinical stroke volume determination. Am Heart J 120: 116-123, 1990. 
16. Sun JP, Pu M, Fouad FM, Christian R, Stewart WJ, Thomas JD. Automated cardiac output measurement by spatiotemporal integration of color Doppler data. In vitro and clinical validation. Circulation 95: 932-939, 1997. 
17. Bland JM, Altman DG, Statistical methods for assessing agreement between two methods of clinical measurement. Lancet i: 307-311, 1986. 
18. Yonekura Y, Fujita T, Nishizawa S, Iwasaki Y, Mukai T, Konishi J. Temporal changes in accumulation of N-isopropyl-p-iodoamphetamine in human brain: relation to lung clearance J Nucl Med 30: 1977-1981, 1989. 
19. Ohkubo M, Odano I, Takahashi M. A comparative study of simple methods to measure regional cerebral blood flow using iodine-123-IMP SPECT. J Nucl Med 38: 597-601, 1997. 
20. Lassen NA. Normal average value of cerebral blood flow in younger adults is 50 ml/100g/min. J Cereb Blood Flow Metab 5: 347-349, 1985. 
21. Hatazawa J, Iida H, Shimosegawa E, Sato T, Murakami M, Miura Y. Regional cerebral blood now measurement with iodine-123-IMP autoradiography: normal values, reproducibility and sensitivity to hypoperfusion. J Nucl Med 38: 1102-1108, 1997.