Feedback-Controlled Bolus plus Infusion (FC-B/I) Method for Quantitative Drug Assessment in Living Brain with PET

Overview of Feedback-Controlled Bolus Plus Infusion Method

The block diagram of FC-B/I method was shown in Figure 1A. Feedback-controlled bolus plus infusion system consisted of an animal PET scanner (Hamamatsu SHR-7700, Hamamatsu Photonics, Hamamatsu, Japan), a realtime reconstruction computer, a feedback control computer and a syringe pump (CFV-3200, Nihon Koden, Tokyo, Japan) connected to a venous vessel of a monkey leg.

OPEN IN VIEWER

Figure 1.

The semantic diagram of feedback-controlled bolus plus infusion (FC-B/I) method. (A) Infusion rate of the labeled compound was controlled using feedback information of time-activity curves (TAC) derived from the target region of interest (ROI) placed on the striatum and a real-time reconstruction system. After the reaching the equilibrium state, the feedback loop was quitted, followed by the loading of a testing drug. (B). Time sequence of positron emission tomography (PET) operation by FC-B/I method. Positron emission tomography scan was started by bolus injection of the labeled compound using bolus mode, followed by the determination of an aimed activity level during the bolus mode. Once an aimed activity level was determined, the system was changed to feedback control mode to attain the first equilibrium state. After attained to the equilibrium, an infusion rate was fixed to constant speed to maintain the equilibrium (fixed speed infusion mode). When the first equilibrium was maintained enough long, a drug was then administered and the second equilibrium was observed.

Before a PET scan, the position of striatal ROI was determined from MRI (magnetic resonance imaging) image of each monkey, which was previously coregistered to individual PET image coordinate. The ROI was used to derive a time history of radioactivity concentrations from the PET real-time reconstructed images. And 12 mL of [ ¹¹ C]raclopride solution in a syringe was set on the infusion pump.

A multiframe PET scan was started simultaneously with the start of bolus injection of [ ¹¹ C]raclopride. One-fifth of the total volume was injected within the first 30 seconds in a bolus to make the radioactivity concentration in the target ROI near equilibrium state. The real-time image reconstruction computer made a real-time image to calculate time-activity curves (TAC) in the ROI. The real-time images were reconstructed by filtered back projection with Hanning filter of 8.0 mm FWHM every 30 seconds.

When the TAC reached the peak, an aimed equilibrium activity level for feedback control was determined as an averaged value of the TAC in the last 3 minutes around the peak (Figure 1B, equilibrium level estimation mode). Once the aimed equilibrium activity level was attained, the system operation mode was switched to feedback-controlled infusion mode shown in Figure 1B. In this mode, infusion speed was controlled to keep the current TAC value at the aimed activity level using a proportional-integral-derivative (PID) control theory ¹⁴ (Supplementary Text; Supplementary Figure S1).

When a percent difference of the TAC against the equilibrium activity level was within ± 1% in the latest 10 minutes, the feedback-controlled infusion operation was switched to a fixed speed infusion mode. The infusion speed was set to the value of total infused volume divided by a total time of feedback control duration.

Chemicals

[ ¹¹ C]raclopride was synthesized by the standard method at Hamamatsu Photonics PET Center. ¹⁵ In this study, radioactive purity of [ ¹¹ C]raclopride was 498% and specific activity was 64.9 ± 18.3 GBq/μmol (mean ± s.d.).

Animals

Three male rhesus monkeys (Macaca mulatta) weighting from 3.9 to 6.2 kg, 5 to 8 years old were used in this study. They were maintained and handled in accordance with the recommendations of the US National Institutes of Health and the guidelines of the Central Research Laboratory, Hamamatsu Photonics.

Magnetic resonance imaging of monkeys were taken at National Institute for Physiological Sciences with a 3.0-T MR imager (Allegra; Siemens, Erlangen, Germany) under anesthesia with pentobarbital. The stereotactic coordinates of PET and MRI were adjusted based on the orbitomeatal line with a specially designed head holder.

After the overnight fast, the monkey was sat to the monkey chair and its head was fixed with stereotactic coordinates paralleled to the orbitomeatal line. A cannula was implanted into the saphena of a monkey leg for administration of [ ¹¹ C]raclopride and another cannula was put into the posterior tibial artery of the other leg to get arterial blood samples.

Positron Emission Tomography Measurement Protocol

Beside the FC-B/I method, other two types of administration protocols were conducted. One was a bolus injection method, which was a standard procedure and the labeled ligand was bolusly injected, and another was a conventional bolus plus infusion (B/I) method, in which a partial volume of labeled ligand was bolusly injected and the remainder was infused at a constant rate.

A high-resolution animal PET scanner (SHR-7700, Hamamatsu Photonics) ¹⁶ was employed, and all data were acquired in 2D mode. To correct the sensitivity of the scanner detectors, a blank scan of 120 minutes duration was performed using ⁶⁸Ge–⁶⁸Ga rotation rod source before emission study within a day. Transmission scan for 30 minutes duration was conducted. Positron emission tomography data were reconstructed by filtered back projection method with Hanning filter of 4.5 mm FWHM to produce a 256 × 256 × 31 voxel image, and each voxel size was 1.2 × 1.2 × 3.6 mm³.

Bolus Injection Method

Emission data of total 91 minutes duration were dynamically collected. Positron emission tomography scan with arterial blood sampling was simultaneously started with the bolus injection of [ ¹¹ C]raclopride during 30 seconds with the total volume of 3 mL.

The baseline and loading scans were performed in a day separated over 2.5 hours, and the sets of PET scans of the same animal were separated >1 month. For loading scans, cold raclopride at doses of 10 and 100 μg/kg was intravenously administered via the saphenous vein 20 minutes before the emission start.

B/I Method

Emission data of total 90 minutes duration were collected by multiframe timing as 1 minute × 90 frames. Positron emission tomography scan with arterial blood sampling was simultaneously started with bolus injection followed by continuous infusion using a infusion pump (CFV-3200, Nihon Koden) with 12 mL in a syringe (SS-10ESZ, Terumo, Tokyo, Japan). The volume of the bolus injection was set to 2.6 mL, and the infusion rate was subjected to K_bol ¹¹ , which was calculated by PMOD (PMOD Technologies, Zurich, Switzerland) using the previous measured data bolus injection scan to achieve an equilibrium in the striatum within 20 minutes. The infusion speed of each animal was set at 0.088, 0.084, and 0.074 mL/min (K_bol at 28.3, 29.7, and 33.7 minutes), respectively. Cold raclopride was then administered into the saphenous vein 45 minutes after the scan start.

The BP_ND was obtained as a ratio of tissue radioactivity concentration between the target (striatum) and reference (cerebellum) regions at equilibrium.

Feedback-Controlled Bolus Plus Infusion Method

To check an influence on bolus volume changes to FC-B/I control, three different bolus volumes were examined. The bolus volume of 2.6 mL was used in 10 μg/kg cold raclopride loading scans and the bolus volumes of 3.4, 6.5, and 2.6 mL were used at 100 μg/kg cold raclopride loading scans. Dynamic PET emission scan with FC-B/I method and arterial blood sampling were performed by the same manner as conducted in B/I method. The feedback-controlled infusion was started after 15 minutes and regulated its speed every 1 minute. Cold raclopride was administered into the saphenous vein 45 minutes after scan start.

Blood Sampling and Metabolite Analysis

Bolus Injection Method. Arterial blood samples of about 1 mL during the PET scan were collected for 90 minutes using a auto blood sampling system (DR-2, Eicom, Kyoto, Japan). Aliquots of 50 μL whole blood and 100 μL plasma centrifuged at 9,510 g for 2 minutes under room temperature (FB-8000, Kurabo, Osaka, Japan) were weighted (HR-120, A&D, Tokyo, Japan) and the radioactivity counts were measured by γ counter (1480 WIZARD, Perkin-Elmer, Waltham, MA, USA). For metabolite analysis, the other aliquots of 100 μL plasma were mixed with 200 μL ethanol and the mixture was centrifuged at 9,510 g for 5 minutes at room temperature. The supernatants were developed on thin layer chromatography plates (AL SIL G/UV, Whatman, Kent, England) with a mobile phase (chloroform:methanol:triethylamine = 36:4:1). At each sampling time point for the metabolite analysis, the ratios of radioactivity in the unmetabolized fraction to that in total plasma (metabolite plus unmetabolite) were determined using a phosphoimaging plate (FLA-7000, Fuji Film, Tokyo, Japan).

Bolus Plus Infusion and Feedback-Controlled Bolus Plus Infusion Methods. Arterial blood samples of about 1 mL during the PET scan were collected for 90 minutes using a autoblood sampling system (DR-2, Eicom, Kyoto, Japan). Thin layer chromatography metabolite analysis was also performed as described above.

Data Analysis. In the bolus injection method, multilinear analysis method^17,18 (MA1), which was modified method of Logan graphical analysis ¹⁹ (LGA) to reduce the noise-induced bias, was used to calculate total distribution volume (V_T) in each region and binding potential (BP_ND) of striatum (STR) to cerebellum (CER) was obtained.

In the B/I and FC-B/I methods, the ratio (R = C_STR/C_CER) that corresponded to DVR ²⁰ was calculated ² before and after the loading of cold raclopride, where C_STR and C_CER were the averaged radioactivity concentrations in STR and CER at equilibrium, respectively. These DVR values were then translated to BP_ND by the calculation of DVR – 1. The first equilibrium state level before drug administration (R_pre) was obtained by an average between 30 and 40 minutes of TAC data and the second equilibrium state (R_post) were an averaged value between 70 and 80 minutes of TAC data, respectively. V_T was also calculated by C_STR/C_p, where C_p was an averaged radioactivity concentration of metabolite-corrected plasma during the first or second equilibrium.

Data Agreement with Conventional Methods

Distribution volume was utilized in PET neuroreceptor quantification. Distribution volume was defined as a ratio of radioactivity concentration between in arterial plasma and in a target tissue at equilibrium. ²⁰ B/I method realized the equilibrium states exactly, and measured the DV directly. Thus, B/I method could monitor a change of the DV caused by intervention such as a loading of psychotropic agents. The original B/I method aimed at an equilibrium state using a combination of a bolus injection and a continuous infusion without control. It was not, however, guaranteed to achieve an equilibrium state. Feedback-controlled bolus plus infusion was actively controlled the injection speed based on an instantaneous activity concentration in a tissue in real-time fashion to realize rapidly and stably. Despite of the variation of the initial bolus volume (2.6 to 6.5 mL) in FC/B/I, the equilibrium was well obtained, so that the PID algorithm and its parameters were suitable for the experiments. I have confirmed that same PID parameters obtained from [ ¹¹ C]raclopride can be used for other PET ligands such as [ ¹⁸ F]β-CFT-FE, a novel dopamine transporter PET ligand.^13,21

As a result of the correlation analyses, the ratio of radioactivity concentration in striatum to cerebellum obtained by the FC-B/I method agreed with that of B/I method (Figures 4C and 5C). These results indicated that the control of the infusion speed worked well. We also evaluated DVR and V_T obtained by FC-B/I and B/I method, and these were apparently higher than those obtained by bolus injection method (Figures 4A, Figures 4B and 5A, 5B). We assumed that the differences between the equilibrium methods and the bolus injection method were caused by the underestimation in a graphical analysis, such as LGA, of bolus injection method due to the noise of TAC data. ²² Although we adopted V_T of noise-robust analysis of MA1, ¹⁷ we compared MA1 to LGA to realize the effect of the noise of TAC. Regression line of V_T suggested that noise-induced underestimation in both LGA and MA1 was not observed clearly (data not shown). As an overestimation of V_T in B/I compared with that in the LGA or four-parameter model were suggested, ² our result of the striatum V_T in both FC-B/I and B/I methods were reasonable.

K_bol comparison with B/I method

Intrinsically, FC-B/I did not require K_bol because the injection rate was controlled so that the equilibrium was achieved. However, the bolus injection was conducted at the beginning of measurements due to realize rapid equilibrium. The resultant K_bol in FC-B/I method ranged from 49.6 to 69.5 minutes (60 ± 7.0 minutes, mean ± s.d.), which were similar to the K_bol values reported previously. ² It should be noted that these were almost twice of the K_bol of the B/I method (K_bol of 28.3 to 33.7 minutes) calculated by PMOD with our own TAC data from PET and input function from metabolite-corrected plasma data after bolus injection. These differences appeared in the shapes of TAC shown in Figure 2. The shape of B/I method showed no overshoot before the first equilibrium state, while FC-B/I method showed an overshoot induced by the bolus injection. Although the reasons why such discrepancy was observed remain unclear, we experienced similar result with [ ¹⁸ F]β-CFT-FE, thus the K_bol calculated by PMOD software was almost half of that determined with the results from the FC-B/I method (data not shown). It might need reconsideration of parameter setting for simulation of K_bol estimation with PMOD. To achieve the faster equilibrium level, the large K_bol was needed and the equilibrium level should be achieved around peak of the overshoot of TAC owing to the controlled injection in FC-B/I.

Selection of Target Region of Interest

Feedback-controlled bolus plus infusion used an averaged TAC derived from a defined ROI onto a target tissue, the striatum with high-specific binding sites of D₂R for [ ¹¹ C]raclopride. However, if the region with lower-specific binding sites was selected as a target ROI, the radioactivity level in the ROI might not be enough stable to control the feedback calculation. Therefore, we choose the striatum with high-specific binding to D₂R. Although the high-specific binding sites were selected as a target ROI, there was a controversial point that feedback control of the infusion speed could not continue after the loading because of the change in the radioactivity concentration by the drug. To solve the problem, the infusion rate of labeled compound was fixed to the averaged rate obtained in the period before drug administration to keep constant concentration of the labeled compound in the blood. As shown in Figure 2A, TAC in the striatum was reached peak level within about 15 minutes from the start of PET scan and maintained equilibrium state until the loading of cold raclopride, and TAC in the cerebellum with less D₂R kept first equilibrium state until the end of scan. Radioactivity in plasma input in Figure 3 kept an equilibrium level from 30 to 40 and from 60 to 80 minutes, respectively, except the timing just after the loading. These demonstrated that ROI could be set even in the high-specific binding region such as the striatum in our FC-B/I method for the assessment of drug effects with PET.

Suitability of Our Technique

I considered that the PET ligands suitable for the FC-B/I method have rapid clearance rate from the brain and blood. Since the FC-B/I system regulates the infusion rate of syringe pump with feedback control using PET data obtained in real-time from PET system, the PET system combined should allow FC-B/I system to receive PET data in real-time during the scan.