ORIGINAL ARTICLE Annals of Nuclear Medicine Vol. 7, No. 4, 245-250, 1993 Quantilication of human splenic blood flow (Quantitative measurement of splenic blood flow with H,*=0 and a dynalnic state method : 1) Atsushi OGURO,* Hiroki TANIGUCHI,* Hiroshi KOYAMA,* Hiroki TANAKA,* Keigo MIYATA,* Kazumi TAKEUCHI,* Tadashi INABA,** Hisamitsu NAKAHASHI** and Toshio TAKAHASHI* *First Department of Surgery, Kyoto Prefectural University of Medicine * *Nishljin Hospita/, Kyoto, Japan Positron emission tomography (PET) by means of a dynamic state method and H2150 was performed to quantify splenic blood flow in 20 patients who had no hepatic functional dis-orders. Non-1inear regression analysis was applied to determine splenic blood fiow. In calculating arterial input function for the spleen, our original exponential method was used to facilitate computerization. Mean splenic blood flow per 100 g of spleen (SBF) was 168.0 ml/min/lOO g with a standard error (SE) of 12.4 ml/min. The mean spleen-blood partition coefficient for water (p) was 0.767 with a SE of 0.020. Significant correlations were noted between the values for SBF obtained by the exponential method and linear method in which individual increasing values for arterial 150 concentration were used rectilinearly (r=0.96, p<0.005) and also between the values for p obtained by the two methods (r=0.95, p<0.005). In order to validate the application of a one compartment model to an organ with a large blood volume such as the spleen, a further experiment was performed with a water fiow model simulating splenic circulation. We succeeded in quantifying regional splenic blood fiow by PET. It was thought that the quantification of splenic blood fiow by our method would be beneficial in the study of splenic circulation, which is expected to be altered under conditions of portal hypertension, liver dysfunction and shock, etc. Key words: splemc blood fiow PET H 150 INTRODUCTION SEVERAL 'rECHNIQUES are currently used to determine human splenic blood flow both semi-quantitatively and noninvasively. Examples including the inhalation of inert gasses such as 133Xe and the subsequent monitoring of the clearance of the gas with an external scintillation counter,1 measurement of the splenic uptake of 111ln-1abeled platelets following their intravenous injection2 and measurement of blood velocity in the splenic vein by Doppler Ultra-sonography3 have made it possible to determine splenic blood flow safely. However, because of the potency of the isotopes used, it is difficult to perform quantification or serial examinations. In addition, Doppler Ultrasonography sometimes fails in locating the splenic vein and measuring its diameter. There-fore, these methods of determining splenic blood flow are not considered totally satisfactory in spite of their non-invasive merit. In the present study we applied positron emission tomography (PET) to quantifying human splenic blood flow with H2150 in which it was possible to safely determine not only the splenic blood fiow, but also the blood flow to all organs adjacent to the spleen. Generally organic blood flows measured by PET are blood flows in regional tissues which contain little blood volume. We ventured to measure splenic blood fiow where there is a big blood volume and examined the validity of the measurement. SUBJECTS AND METHODS Mathematical model and theory A one compartment model was adopted for the analysis of splenic blood flow. However, this model is suitable for organs in which blood volumes are negligibly small and the spleen is far from suitable. We therefore performed a further series of ex-periments with a blood flow model, details of which are shown in the section on Model analysis. The input function (Ca(t)) may be expressed as the time activity curve (TAC) for the aorta, and the flow of H2150 into the splenic artery is determined from total splenic blood flow (F) with activity Ca(t) (Fig. l). The radioactivity in the spleen (Cs(t)) is then washed out by the blood flow. The differential equation is : where to represents the time at which the inpl function is maximized. The left sides of both equ~ tions (d) and (e) are derived from the input functiol Following determination of Ca(t), equation (b) ca be solved. Since the values in the expression show in (c) are measurable by PET, F/ V and p are dete] mined in the same manner. Model analysis Simulating the human splenic blood circulation, a water flow model was designed. As shown in Fig. 2, two pellucid acryl boxes (an imitation of lungs with a volume of 500 m/; pulmonary box and of spleen with a volume of 200 m/; splenic box) were connected by a silicon tube 52 cm long and with an internal diameter of 0.8 cm. Before the splenic box a flow meter was inserted (MODEL RK400, KOFLOC Co., Tokyo, Japan) and valves for sampling water were set up just before and behind the splenic box. A valve for infusion of indigocarmine (DAIICHI Pharmaceut., Tokyo, Japan) was set before the pul-monary box. Under the condition of regular water flow through the flow meter, 2 ml of indigocarmine was infused and I m/ water samples were obtained from the both valves at O, 5, lO, 15, 20, 25, 30, 35, 40, 45, 60, 120, 180, and 240 sec. after the infusion. With a spectrophotometer (HITACHI 220A), the indigocarmine concentration in each water sample was measured. Then Simplex method analysis deter-mined the simulated Ca(t) and Cs(t). The water used was stored and the volume was counted for 240 sec. in order to estimate the error of the fiow meter. Patients Twenty patients who were free of diseases causing chronic hepatic damage such as liver cirrhosis or chronic hepatitis were investigated. The patients included I I males and 9 females whose ages ranged from 32 to 77 years (mean : 55.4 years). Eight of the patients were diagnosed as having metastatic liver tumors. The conditions of 8 other patients were as follows: 3 hemangiomas of the liver, 2 hepatic hilar bile duct carcinomas, I gallbladder carcinoma, 1 hepatocellular carcinoma, and I angiomyolipoma of the liver. Liver tumors could not be identified in the other 4 patients. Methods The PET system (HEADTOME Ill SET-120W, Shimadzu Co., Kyoto, Japan) was used with a whole body collimator and a cyclotron with a 150 gas production system (BC-1710, Japan Steel Works. Muroran, Japan). The performance characteristics of the PET system in this study were set as follows : an image resolution of 8.2 m. m in full width at half-maximum (FWHM), and a slice thickness of 1 1 mm (direct) and 13 mm (cross plane) FWHM. The matrix size of the image was 128 x 128 with a 2 mm pixel size. The slice interval of the planes was 15 mm. The scan position for each patient was determined by computed tomography. About 20-30 mCi ot~ H2150 was administered by intravenous bolus injection. Emission data for 3 slices in every 1.5 cm interval were collected simultaneously. The patients had measurements taken every 5 sec. for one minute, followed by measurements at 30 sec. intervals for 4 min. The count data obtained from the brachial arterial blood samples were substituted for Ca(t). Arterial blood samples were obtained from the left brachial artery at 10, 15, 20, 25, 30, 35, 60, 120, 180 and 240 sec. after the beginning of the emission scan. The values for A1, A2, B1, B2, C1, C2, D1, and D2 were then detenrrined by the Simplex method6 based on non-1inear regression analysis. Regions of interest (ROls) were placed on the spleen, and time-activity curves (TAC) were then obtained. Corrections for the physical decay of 150 (half-life 123 seconds) were made every 2.5 seconds. In the present study, an average F/V for each slice derived from one scanning process was adopted as the splenic blood flow for the patient. Initially, the linear method was utilized, where individual values for the arterial 150 concentration are used rectilin-early.7 The exponential method was then added and the values for F/V and p calculated by the two methods were compared by paired t-testing. The value for V was calculated by the analysis of serial splenic imagings by computed tomography (_CT).8 RESULTS In the model analysis, when the water flow was set at 100, 200, 300, and 400 m//min, the water flow in the splenic box was calculated as 121, 198, 370, and 398 m//min, respectively, actual water flow being 1 18, 231, 330, and 440 m//min, respectively (Fig. 3). We took the specific gravity of the spleen to be l because the mean specific gravity of 13 surgically removed spleens was 1.004 (the mean specific gravity~ of spleens from 4 patients with liver cirrhosis was 0.966 and the mean specific gravity of spleens from 9_ patients who had no hepatic functional disorders was 1.021). Blood flow data for the 20 patients pro duced by the exponential method are as follows. Splenic blood flw per 100 g of splenic tissue (SBF= F/V) ranged from 48.0 to 258.9 ml/min/lOO g (mean : 168.0, standard error (SE): 12.4); F ranged from 71.9 to 333.7 ml/min (mean: 196.8, SE: 20.1); p ranged from 0.680 to 0.981 (mean: 0.767, SE: 0.020); and V ranged from 46.0 to 243.1 ml (mean: 125.2, SE: 13.5). SBF and p calculated by the linear method ranged from 47.5 to 236.0 (mean : 160.7, SE: 12.0) and from 0.681 to 0.979 (mean: 0.765, SE : 0.019). An actual image of the spleen is shown in Fig. 4 and an example of the determination of Ca(t) and Cs(t) by the exponential method and the linear method are presented in the Fig. 5A and 5B. Significant correlations were noted between the values for SBF derived by the two methods (r=0.96, p<0.05) (Fig. 6A) and also between the values of p determined by the two methods (r=0.95, p< 0.005) (Fig. 6B). DISCUSSION In this report, we have examined a technique for quantifying splenic blood flow by PET. The merits of scanning by PET include very little tissue invasion, with the exception of two intraluminal insertions. one into a vein for the infusion of H2150 and the other into the brachial artery in order to monitor and sample arterial blood. The method of scanning the spleen is technically simple and failure is rare since the areas scanned enable the easy location of target organs. In addition, repeated measurements by means of this method can be performed as 150 has a very short half-life (123 seconds). Local blood flow measurements by PET with 150 has been previ-ously employed in the examination of cerebral blood flow.9 However, we modified this technique by adopt-ing the input function derived from the fitting by the exponential method during the time period 0