Applied Radiation and Isotopes 61 (2004) 887–891

ARTICLE IN PRESS

*Correspond

2820-1095.

E-mail addr

0969-8043/$ - se

doi:10.1016/j.ap

MicroPET-based pharmacokinetic analysis of the radiolabeledboron compound [18F]FBPA-F in rats with F98 glioma

J.C. Chena,*, S.M. Changa, F.Y. Hsub, H.E. Wanga, R.S. Liuc

aDepartment of Medical Radiation Technology, Institute of Radiological Sciences, National Yang-Ming University,

155 Li-Nong Street, Sec 2, Taipei, 112, TaiwanbDepartment of Radiological Technology, Yuanpei University of Science and Technology, Taiwan

cSchool of Medicine, National Yang-Ming University, Taipei, Taiwan

Abstract

Boron neutron capture therapy (BNCT) is one of the effective methods of radiation therapy for the treatment of

tumors such as malignant glioma. Boronophenylalanine (10B-BPA) solution has been used as a potential boron carrier

for such a treatment. The aim of this study is to investigate 4-borono-2-[18F]-fluoro-l-phenylalanine-fructose

([18F]FBPA-F) in rats injected in the brain with glioma using in vivo small animal positron emission tomography (PET)

imaging (microPET). Male Fischer 344 rats with F98 glioma in the left brain were used for these studies. Dynamic PET

imaging of [18F]FBPA-F was performed on the 13th day after tumor inoculation. Arterial blood sampling was

performed to obtain an input function for tracer kinetic modeling. The accumulation ratios of [18F]FBPA-F for the

glioma-to-normal brain approached 3. The uptake characteristics of BPA-F and [18F]FBPA-F were similar. The results

indicate that 4 h after BPA-F injection would be the optimal irradiation time for BNCT. Rate constants were estimated

using a three-compartment model. This study provides useful information for the clinical application of BNCT in

patients with brain tumors.

r 2004 Elsevier Ltd. All rights reserved.

Keywords: 4-borono-2-[18F]fluoro-l-phenylalanine-fructose; Boron neutron capture therapy; MicroPET; Boronophenylalanine-

fructose (BPA-F); F98 glioma

1. Introduction

Boron neutron capture therapy (BNCT) uses a

thermal, hyperthermal or epithermal neutron source to

bombard 10B atoms inside a patient’s brain and this

produces short-range alpha particles via a nuclear

reaction that are able to kill tumor cells effectively

(Slatkin, 1991). BNCT has been proposed for the

selective destruction of infiltrating cells in brain tumors

since 1936 (Locher, 1936). 10B-BPA has been used to

treat animals with brain tumors in preclinical trials

(Coderre, 1993). Clinical trials have been conducted in

the USA, Europe and Japan using 10B-containing

ing author. Tel.: +886-2826-7282; fax: +886-2-

ess: jcchen@ym.edu.tw (J.C. Chen).

e front matter r 2004 Elsevier Ltd. All rights reserve

radiso.2004.05.056

compounds such as BPA or sulfhydryl borane

(Na2B12H11SH, or BSH) (Hatanaka and Nakagawa,

1994; Diaz, 2003; Hideghety et al., 1999).

In Taiwan, the reactor at National Tsing-Hua

University (THOR) is under remodeling to become a

dedicated facility for BNCT and is ready for preclinical

trials for the treatment of gliomas and hepatomas

in May, 2004. Its success will mainly depend on a

differential uptake of 10B-BPA between tumor and

normal tissues. According to Soloway et al. (1998), a

tumor-to-normal tissue uptake ratio (T/N) of 3:1 is

desirable. However, it is difficult to directly measure 10B

levels at the time of BNCT. Imahori et al. (1998b) used

radioactive analogs of 10B ([18F]-FBPA) as a probe to

analyze its kinetics in vivo using PET. Yoshino et al.

(1989) used BPA conjugated with fructose to increase its

solubility, so that drug uptake by the tumor is enhanced.

d.

ARTICLE IN PRESSJ.C. Chen et al. / Applied Radiation and Isotopes 61 (2004) 887–891888

Ishiwata et al. (1992) studied 10B concentrations in

melanoma-bearing mice in vivo using 18F as a probe in

clinical PET imaging. The aim of our study was to use a

dedicated high-resolution small animal PET scanner

(microPET) to image the brain of glioma-bearing

Fischer 344 rats and to calculate accurate T/N ratios

by dynamic PET imaging as well as to carry out a

quantitative study on the pharmacokinetics of the tracer

([18F]FBPA-F) in tumors and normal tissues.

Fig. 1. Three-compartment model for [18F]FBPA-F uptake

with four rate constants.

2. Materials and methods

2.1. Materials

We used a microPET R4 (Concorde, USA) supported

by the PET gene probe core established for the National

Research Program for Genomic Medicine (NRPGM) to

carry out high-resolution rat imaging. The microPET

R4 has a spatial resolution of around 1.6mm FWHM at

center of the field of view (FOV) using the ordered-

subsets expectation maximization (OSEM) reconstruc-

tion method. 18F was produced from the cyclotron

(MC17F, Scanditronix) at National PET/Cyclotron

Center. The preparation of [18F]FBPA-F has been

described previously by Imahori et al. (1998a,b) and

was carried out as modified by Wang et al. (2004). The

F98 glioma cells were a gift from Dr. Barth. We used

three male Fischer 344 rats (12–14 week old, 250–280 g)

from the animal center of National Yang-Ming Uni-

versity. Each tumor-bearing rat was anesthetized with

Urethane (0.9ml). At this point, F98 rat glioma cells

were injected into the left brain region. The animal

experiments were approved by the Laboratory Animal

Care Panel of National Yang-Ming University.

2.2. Methods

Dynamic PET images of the F98 glioma-bearing rats

were obtained using the microPET R4 system and this

produced 63 image slices over a 7.89 cm axial field of

view (FOV), with a slice thickness of about 1.25mm.

The in-plane spatial resolution at the center of the FOV

is 1.6mm using the OSEM reconstruction method. Each

microPET scanning was performed on the 13th day after

tumor implanation. Each tumor-bearing rat was an-

esthetized and injected with about 20MBq [18F]FBPA-F

(B4.3� 10�7mg BPA/kg body weight) through the

lateral tail vein. All injections were bolus injections.

Each rat was imaged in the prone position. Each

dynamic data acquisition was acquired using ten 3-min

frames, followed by seven 30-min frames up to 4 h after

injection (non-uniform time sampling). All images were

reconstructed using ordered-subsets expectation max-

imization, with a 256� 256 pixel image matrix, 16

subsets, 4 iterations, and use of a Gaussian filter.

Arterial blood samples were obtained at the same time

intervals as the dynamic scans during the microPET

procedures, and a total of 17 samples (0.1ml each) over

a period of 4 h period were collected.

Time-activity curves (TACs) were plotted for the

tumor and the normal tissue located in the non-tumoral

control area. Regions of interest (ROIs) of each tumor

were drawn from each image plane in which tumor was

visible on the final time frame. The mean of radioactivity

concentration of the tumor ROI at different time frames

was determined. Reference (normal) ROIs were drawn

from the final frame at the end of scanning in the

contralateral normal brain and applied to all images in

the dynamic sequence.

Quantitative knowledge of the tissue kinetic para-

meters in the regions of the brain can offer information

such as the metabolic rate or 10B level and is useful in

clinical applications. Dynamic microPET imaging with

injection of radioactive tracer can be used for these

measurements. We used a modified three-compartment

physiological model of [18F]FBPA-F by K1 (ml/gmin),

k2 (min�1), k3 (min

�1), and k4 (min�1) as shown in Fig. 1

(Imahori et al., 1998a), where K1 and k2 refer to forward

and reverse transport rate of [18F]FBPA-F across the

blood–brain barrier to the selected ROIs, respectively,

and k3 and k4 represent anabolic (specific binding to the

target) and reverse process rate constants, respectively.

This model has been validated in a previous study by

Imahori et al. (1998a). The kinetic parameters of the

tissue can be estimated from the dynamically acquired

microPET images with blood sampling as the input

function. The nonlinear least squares regression method

was used to estimate the rate constants (K1, k2, k3, and

k4). We used Matlab 6.5 for image processing tool

development running under Windows XP professional

OS. ROIs and TACs and the rate constant estimations

were all calculated using in-house designed Matlab

codes.

3. Results

3.1. Micropet scanning of 18F-FBPA-Fr in F98 glioma-

bearing fischer 344 rats

Transaxial views of microPET images of F98 glioma-

bearing Fischer 344 rats at the 15th time frame were

ARTICLE IN PRESS

Fig. 2. The 27th tomographic slice of microPET images of a

F98 glioma-bearing Fischer 344 rat at the 15th time frame after

intravenous injection of 20MBq [18F]FBPA-F (transaxial

view).

Fig. 3. Time-activity curve of tumor and normal brain tissue in

F98 glioma-bearing Fischer 344 rats calculated from the

microPET images after intravenous injection of 20MBq

[18F]FBPA-F.

Fig. 4. Time course of the tumor-to-normal ratios of

[18F]FBPA-F.

Fig. 5. A plasma time-activity curve obtained by arterial blood

sampling.

J.C. Chen et al. / Applied Radiation and Isotopes 61 (2004) 887–891 889

displayed on the monitor. The 15th time frame

corresponds to 4 h postinjection. From the 63 spacial

slices available, we chose the 27th slice, which showed

clear uptake of [18F]FBPA-F in tumor region, for

further ROI analysis. From the 27th slice, we drew the

same ROI crossing from 25th to 29th slices and from

these ROIs, we calculated the mean and standard

deviation for all time frames. Fig. 2 shows such a 27th

slice image with ROIs indicated by the arrows. The

glioma in the left brain is indicated by the hot spot with

high radioactivity levels in the ROI. The tumor time-

activity curve of mean [18F]FBPA-F concentration in the

ROI at different time frames is shown in Fig. 3. Note

that all the images shown here were decay corrected. We

also drew an ROI for the normal brain area in the same

image as shown in Fig. 2. The normal ROI is indicated

by the right arrow. The normal tissue time-activity curve

is also shown in Fig. 3, which shows the mean activity of

the normal ROIs at different time frames. The

accumulation of radioactivity in normal brain tissue

increased at first and then gradually reached a stable

state from 2 up to 4 h after drug injection. Fig. 4 shows

the ratios of the mean activity values in the tumor ROIs

to the mean values in the normal ROIs at different time

frames. As shown in Fig. 4, 1 h after injection, time no

ARTICLE IN PRESS

Table 1

A pilot study of tracer kinetic parameters in F98 glioma-bearing Fisher 344 rats�

Grade K1 k2 k3 k4

Glioma 0.1609� 0.0009 0.4118� 0.001 0.0856� 0.0012 0.0077� 0.0012

Normal 0.1078� 0.0018 0.4026� 0.0092 0.1031� 0.0062 0.0176� 0.0011

�Values of the rate constants (K1 (ml/g min), k2 (min�1), k3 (min

�1), k4 (min�1) are given as mean7s.d., where n=3 in each glioma

and normal group.

J.C. Chen et al. / Applied Radiation and Isotopes 61 (2004) 887–891890

longer determined the T/N ratios, and they became

steady (approaching 3:1). Fig. 5 shows the plasma time-

activity curve, which was used as an input function for

estimating the rate constants in the tracer kinetic

modeling. Each point on the curve corresponded to an

arterial blood sampling at that time frame and the

activity concentration of the plasma was assayed using a

g-scintillation counter (Cobra II Autogamma; Packard).

The uptake of [18F]FBPA-F in the plasma was expressed

in counts per minute (cpm) corrected for decay and

normalized as the activity concentration (Bq/cm3).

3.2. Tracer kinetic modeling of[18F]FBPA-F in glioma-

bearing rats

The pharmacokinetics of [18F]FBPA-F was examined

using this dynamic microPET study. A three-compart-

ment model using rate constants (K1, k2, k3, and k4), as

shown in Fig. 1, was used for the kinetic analysis (see

Imahori et al., 1998a,b for a solution of the three-

compartment model). We used blood sampling and its

timing information from rats together with the tracer’s

half-life and then applied the nonlinear least-squares

regression method to estimate the rate constants.

Table 1 shows the values of K1, k2, k3, and k4 obtained

from the glioma-bearing rats in tumor ROIs and normal

ROIs. In both the tumor and normal control cases, k3 is

greater than k4.

4. Discussion

The K1 value in glioma group (three rats) differed

from that in normal group (three rats) as shown in Table

1. These findings indicate that tracer uptake capacity,

associated with tumor malignancy, depends on K1,

which is an indicator of the transport process. This

finding can be compared to a recently published paper

by Takahashi et al. (2003) that examined the correlation

between metabolic values for radiolabeled BPA-F and

tumor malignancy, survival, etc.

One of the important factors for BNCT is the T/N

ratio. This ratio not only indicates tumor malignancy

but also can be used to suggest the optimal neutron

irradiation time for BNCT to minimize dose irradiation

to normal tissues. In this study, the ratio approached a

stable value of 3 after 1 h postinjection. The imaging

findings have been correlated with previous studies by

Wang et al. (2004) in which the animals were sacrificed

at various time points. However, considering the

significant differences in pharmacological parameters

of the rat as compared to human, the optimal irradiation

time determined by this study cannot be directly applied

for clinical BNCT.

Another important factor for BNCT is the 10B level or

the uptake of boron compounds. Ishiwata et al. (1992)

had shown that the uptake of 10B-BPA was similar to

that of [18F]-FBPA in terms of the pharmacokinetics.

The 10B-BPA per total injection dose estimated by

inductively coupled plasma-atomic emission spectro-

scopy (ICP-AES) corresponded almost 1:1 relative to

the 18F-FBPA as estimated by specific activity (Chandra

et al., 2002). Thus from the tracer concentration we can

estimate the concentration of the boron compounds

in vivo solely from microPET imaging. This shows the

advantages of PET imaging as a non-invasive in vivo

tool in monitoring the drug distribution before BNCT.

5. Conclusions

We used a high-resolution microPET to image [18F]-

FBPA-F biodistribution in F98 glioma-bearing Fischer

rats in vivo. The images show a high tumor-to-normal

uptake ratio (B3:1). This microPET imaging can be

used with [18F]-FBPA-F as a probe for 10B-BPA-F in

BNCT. This preclinical study provides useful informa-

tion for future clinical application of BNCT in patients

with brain tumors such as gliomas.

Acknowledgements

This research was supported by Grants NSC89-2745-

P-368-001, NSC91-2745-P-010-001 and NSC92-2745-P-

010-001 from National Science Council, Taipei, Taiwan.

We thank the staff of the National PET and Cyclotron

Center in Taipei Veterans General Hospital, who kindly

provided the radiopharmaceuticals. The PET Gene

Probe Core for providing the microPET scanning is

also gratefully acknowledged.

ARTICLE IN PRESSJ.C. Chen et al. / Applied Radiation and Isotopes 61 (2004) 887–891 891

References

Chandra, S., Kabalka, G.W., Lorey 2nd, D.R., Smith, D.R.,

Coderre, J.A., 2002. Imaging of fluorine and boron from

fluorinated boronophenylalanine in the same cell at

organelle resolution by correlative ion microscopy and

confocal laser scanning microscopy. Clin. Cancer Res. 8 (8),

2675–2683.

Coderre, J.A., 1993. A phase I biodistribution study of p-

boronophenylalanine. In: Gabel, D., Moss, R. (Eds.), Boron

Neutron Capture Therapy: Toward Clinical Trials of Glioma

Treatment. Plenum Press, New York, pp. 111–121.

Diaz, A.Z., 2003. Assessment of the results from the phase I/II

boron neutron capture therapy trials at the brookhaven

national laboratory from a clinician’s point of view. J. Neuro-

oncol. 62, 101–109.

Hatanaka, H., Nakagawa, Y., 1994. Clinical results of long

surviving brain tumor patients who underwent boron

neutron capture therapy. Inst. J. Radiat. Oncol. Biol. Phys.

28, 1061–1066.

Hideghety, K., Sauerwein, W., Haselsberger, K., et al., 1999.

Postoperative treatment of glioblastoma with BNCT at the

Petten irradiation facility. Strahlenther Onkol. 175 (suppl

2), 111–114.

Imahori, Y., Ueda, S., Ohmori, Y., et al., 1998a. Fluorine-18-

labeled fluroboronophenylalanine pet in patients with

glioma. J. Nucl. Med. 39, 325–333.

Imahori, Y., Ueda, S., Ohmori, Y., et al., 1998b. Positron

emission tomography-based boron neutron capture therapy

using Boronophenylalanine for high-grade gliomas: Part I.

Clin. Cancer Res. 4, 1825–1832.

Ishiwata, K., Shiono, M., Kubota, K., et al., 1992. A unique

in vivo assessment of 4-[10B]borono-l-phenylalanine in

tumour tissues for boron neutron capture therapy of

malignant melanomas using positron emission tomography

and 4-borono-2[18F]fluoro-l-phenylalanine. Melanoma

Res. 2, 171–179.

Locher, G.L., 1936. Biological effects and therapeutic possibi-

lities of neutrons. Am. J. Roentgenol Radium Ther. 36,

1–13.

Slatkin, D.N., 1991. A history of boron neutron capture

therapy of brain tumors: postulation of a brain radiation

dose tolerance limit. Brain 114, 1609–1629.

Soloway, A.H., Tjarks, W., Barnum, B.A., et al., 1998. The

chemistry of neutron capture therapy. Chem. Rev. 98,

1515–1562.

Takahashi, Y., Imahori, Y., Mineura, K., 2003. Prognostic and

therapeutic indicator of fluoroboronophenylalanine posi-

tron emission tomography in patients with gliomas. Clin.

Cancer Res. 9 (16 Pt 1), 5888–5895.

Wang, H.E., Liao, A.H., Deng, W.P., et al., 2004. Evaluation of

4-borono-2-18F-fluoro-l-phenylalanine-fructose as a probe

for boron neutron capture therapy in glioma-bearing rat

model. J. Nucl. Med. 45, 302–308.

Yoshino, K., Susuki, A., Mori, Y., et al., 1989. Improvement of

solubility of p-boronophenylalanine by complex forma-

tion with monosaccharides. Strahlenther Onkol. 165,

127–129.