Vol. 4, 1833-1 841, August 1998 Clinical Cancer Research 1833

Advances in Brief

Positron Emission Tomography-based Boron Neutron Capture

Therapy Using Boronophenylalanine for High-Grade

Gliomas: Part 111

Yoshio Imahori,2 Satoshi Ueda, Yoshio Ohmori,

Kenji Sakae, Tsukasa Kusuki, Tooru Kobayashi,

Masao Takagaki, Koji Ono, Tatsuo Ido, and

Ryou Fujii

Department of Neurosurgery, Kyoto Prefectural University ofMedicine, Kyoto 602 [Y. I., S. U., Y. 0., K. S., T. Ku.]; KyotoUniversity Research Reactor Institute, Osaka 590-04 [T. Ko., M. T.,

K. 0.]; Cyclotron and RI Center, Tohoku University, Sendai [T. I.];

and Cyclotron Unit, Nishijin Hospital, Kyoto 602 [R. F.], Japan

Abstract

Based on pharmacokinetic findings of fluorine-18-

labeled L-fluoroboronophenylalanine by positron emis-

sion tomography (PET), methods for estimating tumor

iOB concentration were devised. In clinical practice of

boron neutron capture therapy (BNCT) for high-grade

gliomas, a large amount of L-boronophenylalanine (L�iOB�

BPA)-fructose solution is used. Under these conditions, a

slow i.v. infusion of L-10B-BPA-fructose solution should be

performed for BNCT; therefore, the changes over time iniOB concentration in the target tissue were estimated by

convoluting the actual time course of changes in plasma‘#{176}Bconcentration with a PET-based weight function in-

cluding the proper rate constants [K1 (ml/g/min), k2

(min1), k3 (mint), and k4 (min’)]. With this method,

the estimated values of ‘#{176}Bconcentration in gliomas werevery close to the ioB levels in surgical specimens. Thisdemonstrated the similarity in pharmacokinetics between

fluorine-18-labeled L-fluoroboronophenylalanine and

L-’#{176}B-BPA. This method, using the appropriate rate con-

stant, permits the determination of tumor ioB concentra-

tion and is widely suitable for clinical BNCT, because the

averaged PET data are enough to use in future patientswithout individual PET study.

Received 12/18/97; revised 5/22/98; accepted 7/1/98.The costs of publication of this article were defrayed in part by thepayment of page charges. This article must therefore be hereby markedadvertisement in accordance with 18 U.S.C. Section 1734 solely to

indicate this fact.

I Supported in part by Grants-in-Aid for Science Research 06282254,06671411, 07274259, 8671602, and 09671439 from the Ministry of

Education, Science, Sports and Culture of Japan.2 To whom requests for reprints should be addressed, at Department of

Neurosurgery, Kyoto Prefectural University of Medicine, Kawarama-chi-Hirokoji, Kamigyo 602-0841, Kyoto, Japan. Phone: 075-251-5543;

Fax: 075-251-5544; E-mail: imahon@koto.kpu-m.ac.jp.

Introduction

BNCT3 requires selective delivery of a boron-containing

drug to the tumor, followed by irradiation with neutrons (1-5).

For estimation of the radiation dose to the tumor, it is essential

to know the concentration of boron in the tumor at the time of

BNCT. However, direct measurement at the time of BNCT is

impossible; therefore, new approaches to the estimation of the

boron content of tumors during BNCT are required. When

actually performing BNCT, we must reconfirm the tumor

level by measuring the L-’#{176}B-BPA in the arterial blood. We

administered a large amount of L-’#{176}B-BPA-fructose solution i.v.

before BNCT, admitted the patient to the irradiation room, and

began neutron irradiation, but the collection of arterial blood

was possible for at least 1-2 h, until the beginning of neutron

irradiation. Thus, the input function of L-’#{176}B-BPA can be de-

termined by blood sampling during this 1-2 h, and, finally, the

neutron dose must be determined based on the tumor lO� level.

Under these conditions, estimation of ‘#{176}Blevel based on the

incorporation constant (ic*) and the utilization ratio (Ur*; Refs.

6-8) is often inappropriate. In this study, our main purpose is to

solve these actual problems. We evaluated the tumor pharma-

cokinetics ofL-’8F-’#{176}B-FBPA based on comparison with L-’#{176}B-

BPA using PET and assessed the similarity of L-’8F-’#{176}B-FBPA

accumulation to that of L-’#{176}B-BPA. If the ‘#{176}Bconcentration

calculated using the rate constant obtained by PET using L-

‘#{176}B-FBPA and the actually measured ‘#{176}Binput function is close

to that in the surgical specimens, similarity in pharmacokinetics

between L-’8F-1#{176}B-FBPA and L-’#{176}B-BPA will be confirmed.

We also designed a basic method for measuring the tumor

concentration of L-’#{176}B-BPA using L-’8F-’#{176}B-FBPA for clinical

BNCT.

Patients and Methods

Clinical Use of L-18F-’#{176}B-FBPA. The performance ofL-’8F-’#{176}B-FBPA PET studies on humans and the quality control

of L-’8F-’#{176}B-FBPA followed the guidelines established by the

PET Committee of Nishijin Hospital (Kyoto, Japan) in January

1991. The synthesis method and purification of L-’8F-’#{176}B-FBPA

are detailed in our previous report (9).

Preparation of L�i�B�BPA�Fructose Complex. L-’#{176}B-

BPA is relatively insoluble in water. For i.v. administration, we

prepared a L-’#{176}B-BPA fructose-complex solution as reported by

Yoshino et a!. (10). This complex dissociates and reaches an

3 The abbreviations used are: BNCT, boron neutron capture therapy;PET, positron emission tomography; L-”F-1#{176}B-FBPA, fluorine-l8-la-

beled L-fluoroboronophenylalanine; L-’#{176}B-BPA, L-boronophenylala-nine; GBM, glioblastoma multiforme; ROI, region of interest; All,anaplastic astrocytoma WHO grade II; Am, anaplastic astrocytoma

WHO grade III.

on April 9, 2021. © 1998 American Association for Cancer Research.clincancerres.aacrjournals.org Downloaded from

k3 �

k4!uI�

i81?.jO�fl3pA

in tumor

A Ki

‘8F-’#{176}B-FBPAin plasma _______

. k2

metabolitesIn plasma

B

r

r��i�BPAI in plasma

1(4

complex

Ki

k2

....x

k3 �

k4.I�

lSFJOB4�pA

in tumor

1834 PET-based BNCT Using L-’#{176}B-BPA: Part II

the mean value. Using this as a criterion, we limited macro-

Fig. 1 Modified three-compartment model of

L-’8F-’#{176}B-FBPA and L-’#{176}B-BPA. A, this model is

adapted to a four-parameter model by adding anadditional serial tissue compartment with anabolic

and reverse process rate constants k3 and k4, re-spectively. K1 and k2 refer to forward and reversetransport of L-”F-’#{176}B-FBPA across the blood-

brain barrier, respectively. The pharmacokinetics

of L�ISF�tt�B�FBPA were analyzed using the three-

compartment model by K, (mllg/min), k, (min’),

k3 (min’), and k4 (min ‘).B,this model was also

adopted to analyze the pharmacokinetics of L-’#{176}B-

BPA described here. The L-’#{176}B-BPA-fructosecomplex is dissociative and reaches an equilibriumbetween free molecule and the complex in the

diluted condition in plasma. Kd and Ka representthe dissociation and association process of thecomplex, respectively.

Table 1 Rate constants of L-”F-’#{176}B-FBPA in patients with gliomas

Rate constants of L-’8F-’#{176}B-FBPA in patients with gliomas, including 17 cases of high-grade gliomas and 4 cases of Al!. The values of the rate

constants (K,, k2, k3, and k4) are given as the mean ± SD. Control was established at the corresponding region in contralateral brain in each case.

Grade n K, (mug/mm) k2 (min’) k3 (mm’) k4 (mm’)

GBM 1 1 0.040 ± 0.007 0.034 ± 0.009 0.018 ± 0.007 0.01 1 ± 0.005

Alil 6 0.039 ± 0.025 0.030 ± 0.013 0.025 ± 0.014 0.01 1 ± 0.007All 4 0.021 ± 0.006 0.030 ± 0.005 0.025 ± 0.005 0.009 ± 0.009Control 21 0.01 1 ± 0.003 0.025 ± 0.009 0.033 ± 0.015 0.009 ± 0.0 10

equilibrium between free molecules and the complex in the

diluted condition present in the plasma (Fig. 1B; Ref. I 1 ). The

solution of L- ‘#{176}B-BPA-fructose complex consisted of L-

BPA (1 g):D-(-)-fructose (2.22 g):H20 (30 ml):lN NaOH (5.55

ml). Thereafter, the solution was slowly added to 0.8 ml of iN

HC1, and the pH value was adjusted to 7.5. To buffer the

solution, 7% NaHCO3 was added. The molar concentration of

the final solution was 1 15 m�vi, and the ‘#{176}Bconcentration was

5.52 mrsi. The molar ratio of L-’#{176}B-BPA:fructose was 1:2.57.

Finally, the fructose complex was prepared by filtration using a

Millipore filter (MILLEX-GS; 0.22 p.m; Waters). The L-’#{176}B-

BPA solution was then tested for a pH value of 8.5. The i.v.

infusion rate was 10 mI/mm.

Patients. The subjects were the same group of patients

described in part I (6). Pharmacokinetic analyses could be

performed for 17 patients with high-grade gliomas and 4 pa-

tients with All in the group, and the rate constants were calcu-

lated (Table 1). Seven of the patients (the same set of patients

described in Table 2 of part I; Ref. 6) underwent resection of the

tumor after PET studies using L�ISF�i�B�FBPA, and the ‘#{176}B

concentration estimated using PET data was compared with that

in the tissue obtained at surgery (Table 2). Surgery was per-

formed for all patients, and in each case, the diagnosis was

confirmed histologically. The degree of malignancy was classi-

fled histologically using the WHO grading criteria (All, Alli,

and GBM).

Dynamic PET Method. The spatial resolution in PET

imaging was 8.2 mm at full-width half-maximum with plane

resolution, whereas the average axial resolution was 12.8 mm.

X-ray computed tomography and magnetic resonance imaging

were also performed for all patients. Dynamic images of L-’8F-

‘#{176}B-FBPA were obtained using PET in all cases. PET scans

were conducted with a tomograph using a HEADTOME III

(Shimadzu Co., Kyoto, Japan). The planes of the PET scans

were the same as those in the parallel computed tomography

studies. Local cerebral blood volume was measured after bolus

inhalation of ‘5O-labeled carbon monoxide gas (12). A dose of

1-1 .5 mCi/lO kg body weight of L-’8F-’#{176}B-FBPA was injected

i.v. over 40 s. Dynamic PET study was begun when whole brain

activity reached a value greater than that of the background

activity. PET data were collected continuously for nine 2-mm

periods and six 4-mm periods, making a total of 15 periods over

a total of 42 mm (Fig. 2, A and B). The four initial arterial blood

samples were obtained at 5-s intervals; subsequent samples were

obtained at gradually longer intervals (0.5-10 mm), making a

total of 21 samples over a period of 42 mm (Fig. 3A). The time

course changes in plasma L-’5F-’#{176}B-FBPA levels were fitted

using Equation A (see “Appendix”).

Choice of ROIs. All regions of macroscopically necrotic

tumor were excluded when ROIs were designated. The variation

in radioactivity associated with each pixel was less than 1 8% of

on April 9, 2021. © 1998 American Association for Cancer Research.clincancerres.aacrjournals.org Downloaded from

A B

p.9-6 Tumor area

/

70’

I Fitting result K(1) .0306075

c� K(2) .0206767I K(3) .020022I K(4) .0084036

Normal area,t5

9,I

p/ Fitting result K(l) .0093911

1 K(2) .0257794c� K(3) .024827

K(4).01l0718

Clinical Cancer Research 1835

Table 2 Segmental convolution method for estimation of the tumor ‘#{176}Blevel

The values shown under “Estimated ‘#{176}B�’were obtained by the segmental convolution method using rate constants obtained by PET (EquationsC-H, see “Appendix”). These ‘#{176}Blevels were estimated according to the resection time on operation (see Fig. 6). L-’#{176}B-BPA solution (3.38-I 13.6

mg/kg body weight) was infused for 20-45 mm. “Tumor ‘#{176}B��shows the ‘#{176}Blevels in tumor tissue resected after an infusion of L-’#{176}B-BPA, as

measured using the prompt -y method.

Grade, total injection Rate constants

dose of L-’#{176}BPA

Case (mg/kg body weight) K, (ml/g/min) k2 (mm ‘) k1 (min ‘) k4 (mm ‘) Estimated ‘#{176}B(p.g/ml) Tumor “B (pg/ml)

1 All!, 51.0 0.031 0.034 0.025 0.009 4.66 (71 mm) 2.1

2 GBM, 33.8 0.043 0.022 0.004 0.015 3.72 (50 mm) 2.3

3 All!, I 13.6 0.027 0.038 0.025 0.023 7.15 (60 mm) 5.4

4 GBM, 71.4 0.039 0.045 0.018 0.013 7.48 (77 mm) 6.0-8.0

5 GBM, 83.3 0.027 0.039 0.01 1 0.013 5.13 (76 mm) 3.5-4.06 GBM, 76.9 0.035 0.034 0.018 0.010 5.77 (1 12 mm) 5.4

7 GBM, 76.9 0.041 0.037 0.019 0.01 1 10.63 (82 mm) 9.5

Fig. 2 Time-activity data ofL-’8F-’#{176}B-FBPA and the fitting results. A, time-activity data in the tumor region. B, time-activity data in the normal brain

region. Fine dotted lines represent the fitting results obtained by the nonlinear least squares best regression method. The pharmacokinetics of

IS�IOi�p�p� were analyzed using a three-compartment model. as shown in Fig. IA. The equation (Equation B in “Appendix”) proposed by Huang

et al. (20) is a generalized model of the theorem on which the three-compartment model is based. The nonlinear least squares best regression method

was used to obtain the rate constants (K,, k,, k3, and k4). For the nonlinear least squares best regression method. an appropriate initial value is required.The method for obtaining this initial value is described in our previous report (9).

scopic heterogeneity to a minimum when designating ROIs.

ROl images consisted of 49-171 pixels on the tumor center and

a nontumor control area; hence, they had a minimum voxel

volume of 2.94 cm3. Patients with a hot area (active area)

smaller than the above-mentioned area were excluded from the

evaluation. The tumor areas of all 2 1 cases presented in this

study satisfied this voxel volume requirement. No partial vol-

ume corrections were performed for any patient. We designated

several ROIs from tumor-affected areas and used the region

with the highest values as a representative ROI. In pharmaco-

kinetic analyses, we corrected the time-activity curve for over-

estimation of ‘8F-’#{176}B-FBPA incorporation by cerebral blood

volume to obtain accurate values of the rate constants [K,

(mL/g/min), k2 (min’), k3 (min’), and k4 (min’)] in the

three-compartment model, as reported in our previous study (9).

Correction of Blood Concentration Data. Generally,plasma:whole blood ratios of ‘8F radioactivity and ‘0B level are

1.3 for both L-’8F-’#{176}B-FBPA and L-’#{176}B-BPA. We determined

the time course changes in both ratios. The plasma:whole blood

ratios of ‘8F radioactivity and ‘#{176}Blevels changed little over

time. Based on these results, we judged it reasonable to treat the

ratio as a constant. When ‘#{176}Blevels of whole blood are meas-

ured by the prompt �l’ method, they should be corrected by

multiplication using a factor of 1 .3, because the input function is

based only on plasma concentrations. Ishiwata et a!. ( I 3) re-

ported that the amounts of labeled metabolites of L-’8F-’#{176}B-

FBPA in arterial plasma were small during the 1-h animal

experiments. In our human PET studies, the metabolic fractions

were negligible during the 42-mm period after injection, be-

cause we observed that more than 95% of the total radioactivity

in plasma was due to the free ‘8F-’#{176}B-FBPA fraction during the

42-mm arterial sampling. These findings suggest that plasma

metabolite correction is not necessary, and that the metabolism

of L- ‘#{176}B-BPAcan be treated like that of ‘ 8F- ‘#{176}B-FBPA.

Three-Compartment Model of L�iSF�i�B�FBPA and

L�iOB�BPA. Our previous data obtained by dynamic PET

studies using ‘8F-’#{176}B-FBPA suggested that the time-activity

curves reveal a characteristic accumulation of ‘8F activity. After

a rapid increase to a quasi-steady state, accumulation remained

stable (Fig. 2, A and B). When the dynamic PET data were

on April 9, 2021. © 1998 American Association for Cancer Research.clincancerres.aacrjournals.org Downloaded from

A B

40000

30000

20000

1 0000

Iplasma

whole blood

40 60 80

mm

0 tj.i tj

1836 PET-based BNCT Using L-’#{176}B-BPA: Part II

, I;

� . �0

Fig. 3 Input functions of L-’8F-’#{176}B-FBPAon two differential iv. administrations. A, all pharmacokinetic analyses were performed by using the input

function administered by a bolus injection for 40 s and arterial blood sampling for 42 mm after the injection. X axis, time course; Y axis, ‘8Fradioactivity of plasma or whole blood. Radioactivity in the blood increased rapidly until I mm after the bolus injection and subsequently decreased

rapidly, permitting three-exponential fittings, as shown in Equation A (see “Appendix”). Plasma:whole blood ratios of ‘8F radioactivity and the ‘#{176}Blevel are 1 .3. We determined the time course changes in ratios. The plasma:whole blood ratios of ‘8F radioactivity changed little over time. Basedon these results, we judged it reasonable to treat the ratio as a constant (6). B, with slow infusion of L-’8F-’#{176}B-FBPA over 20 mm, radioactivity in

the blood increased gradually and then decreased rapidly after discontinuation of the infusion. In this case, the input function permits two-exponentialfunction, but we adopted the segmental convolution method to obtain the accurate C�’t).

#{163}�iCij(t)

Fig. 4 Segmental convolution method for estimation of the tumor ‘#{176}B

level. The input function [C�(t)] was obtained by arterial blood sam-pling.j, segment number; t, time (minutes); C(t), the estimated ‘#{176}Blevelat time t; G�, slope of the first order function, Cpj(t); K�, Y axis interceptof Cpj(t).

represented as Gjedde-Patlak plots (14, 15), a positive slope was

obtained, suggesting the involvement of the putative metabolic

pool of this tracer in tumor cells (9). We used a compartment

model that is adapted to a four-parameter model by adding an

additional serial tissue compartment with anabolic and reverse

process rate constants k3 and k4, respectively; the incorporation

into proteins is then described by rate constant k5, but it should

be ignored because of less incorporation into proteins (13). This

method has already been established as an analytical model of

amino acids by Wienhard et a!. (16); a similarity in the model

analysis between the characteristics of ‘ 8F)fluorotyrosine

in their report and the characteristics of ‘8F-’#{176}B-FBPA was

found (9, 17-19). Therefore, the pharmacokinetics of ‘8F-’#{176}B-

FBPA were analyzed using a modified three-compartment

model by K, (mug/mm), k, (min ‘), k3 (min ‘), and k4 (min ‘)

as shown in Fig. 1A. The equation (Equation B in “Appendix”)

proposed by Huang et al. (20) is a generalized model of the

theorem on which the three-compartment model is based. The

nonlinear least squares best regression method was used to

obtain the rate constants (K, , k2, k3, and k4). Using this method

of analysis, the rate constants were determined in each glioma

patient. Details of the relationship between each rate constant

and malignancy were presented in our previous study (9).

Segmental Convolution Method for Estimation of theTumor ioB LeveL In clinical practice of BNCT, 600-800 ml

of L-’#{176}B-BPA-fructose solution are used, and infusion requires

about 60-80 mm. Under these conditions, convolution of the

time course changes in plasma ‘#{176}Bconcentration in the form of

L-’#{176}B-BPA determined by Equation D (in “Appendix”), as

shown in Fig. 4, can be used in Equation C (in “Appendix”). In

the segmental convolution method, the estimated values of

tissue ‘#{176}Bconcentration of L-’#{176}B-BPA can be calculated from

the values of K,, k2, k3, and k4 obtained by PET using L-’8F-

‘#{176}B-FBPA for each patient or the mean values shown in Table

1 . If the ‘ #{176}Bconcentration calculated using the rate constant

obtained by L-’8F-’#{176}B-FBPA PET and the actually measured

‘#{176}Binput function is close to that in surgical specimens, simi-

larity in pharmacokinetics between L-’8F-’#{176}B-FBPA and L-’#{176}B-

BPA will be confirmed. Using this method, we also preopera-

tively compared the estimated values with the ‘#{176}Blevels in

seven patients with high-grade gliomas (anaplastic astrocytoma

on April 9, 2021. © 1998 American Association for Cancer Research.clincancerres.aacrjournals.org Downloaded from

A

I

B

E

0 40 80 120 160 200

mm

Plasma 18F-level

Estimated tumor 18F-level

0 20 40 60 80

mm

brain 18F-level

and GBM). The estimated ‘#{176}Blevels were obtained using Equa-

tions B-H (in “Appendix”).

Results

Input Function of L�i$F�i�B�FBPA Obtained by Arte-rial Sampling To facilitate the evaluation of pharmacokinet-

ics, a bolus injection of L-’8F-’#{176}B-FBPA over 40 s was used for

dynamic PET studies. Radioactivity in the blood increased rap-

idly until 1 mm after the bolus injection and subsequently

decreased rapidly, permitting three-exponential fittings, as

shown in Equation A (Fig. 3A). With the slow infusion of

L-’8F-’#{176}B-FBPA, radioactivity in the blood increased gradually

and then decreased rapidly after discontinuation of the infusion

(Fig. 3B). These findings suggest that L-’8F-’#{176}B-FBPA is

cleared immediately from the blood. Fig. 3B shows the changes

in blood radioactivity after a slow infusion of L-’8F-’#{176}B-FBPA

Clinical Cancer Research 1837

Fig. 5 Assessment of the suitabil-

ity of the segmental convolution

method. Representative images (A

and B) are of patients with glioblas-toma. As shown in both A and B,the estimated values by the segmen-tat convolution corresponded wellwith the actual measurements by

PET in each case. The estimatedcurves were plotted by the convo-lution of the input function [C�*(t)]of ‘8F radioactivity obtained by

continual arterial blood specimen

collection at the same time, with themean values of K,-k4 shown in Ta-

ble 1 substituted into the weight

function. #{149},C.*(t) of tumor lesionbased on actual PET study (tumor

‘8F level on PET); �, the estimatedci *(t) (the estimated tumor ‘ 8F 1ev-ci). #{149},C*(t) of the normal brainbased on actual PET study (normal

brain ‘8F level on PET); 0, the es-

timated C. *(t) (estimated normal

brain ‘8F level).

on April 9, 2021. © 1998 American Association for Cancer Research.clincancerres.aacrjournals.org Downloaded from

Plasma 10B-level

Estimated 10B-level

20

16’

12

Plasma 10B-level

Estimated 10B-level

0

8

4

50 100 150 200

mm25 50

mm

Fig. 6 Segmental convolution method for the estimation of the tumor ‘#{176}Blevel. A comparison between the estimated values and the ‘#{176}Blevels of

surgical specimens in case 6 (A) and case 7 (B) in Table 2 are shown. We used the proper rate constants as a weight function obtained by PET in

each patient. The estimated curves were calculated by the input function of the ‘#{176}Blevel obtained by continuous arterial blood sampling.

75 100 125

1838 PET-based BNCT Using L-’#{176}B-BPA: Part II

A B

Case 6 Case 710�

8

� 4.

2

over 20 mm. It shows a time-activity curve for L-’8F-’#{176}B-FBPA

obtained under conditions similar to those used for BNCT. In

BNCT, a large amount of L-’#{176}B-BPA-fructOse solution is in-

fused over 60 mm. Segmental convolution using Equations C-H

therefore seems to be appropriate for slow iv. infusion.

Pharmacokinetics of L�lSF�iOB�FBPA Determined byPET. Based on the results of dynamic studies, we calculated

the rate constants (K, , k2, k3, and k4) for the three-compartment

model in Fig. iA. Radioactivity in tumor tissue increased rapidly

for 20 mm after bolus injection and increased gradually there-

after (Fig. 2, A and B). The rate constants obtained are shown in

Table 1. When BNCT is performed clinically, L-’#{176}B-BPA is

infused at 170-280 mg/kg body weight, and the tumor ‘#{176}Blevel

during BNCT is 20-35 p�g/ml (in ppm). In this case, estimation

of the ‘#{176}Blevel based on Ur* and lc* is often inappropriate, but

segmental convolution of the serially determined plasma ioB

concentrations in Equation D yields accurate estimations (Fig.

4). To use Equation E, K,, k2, k3, and k4 should be obtained

beforehand by PET using L-’8F-’#{176}B-FBPA for each patient, or

the mean values shown in Table 1 can be used. The input

function of L-’#{176}B-BPA in plasma, C�(t), was obtained by the

prompt “�‘ measurement connected with the reactor.

Fig. 5, A and B, shows examples of a PET study to assess

the suitability of the segmental convolution method. Both im-

ages are of patients with glioblastoma. These two patients un-

derwent slow infusions of L-’8F-’#{176}B-FBPA, and we performed

a continual measurement by PET and plotted C1*(t). In this PET

measurement, slow infusions of L-’8F-’#{176}B-FBPA were per-

formed under conditions of administration quite similar to those

of clinical BNCT. The closed squares show values of C� *(t) for

the tumor obtained from the PET study, and the open squares

show the estimated curves, which were plotted by convolution

of the input function [C�*(t)] of ‘8F radioactivity obtained by

continual arterial blood specimen collection at the same time,

with the mean values of K,-k4 shown in Table 1 substituted into

the weight function. As shown, the values estimated by the

segmental convolution method corresponded well with the PET

measurements in each case, confirming the suitability of the

segmental convolution method.

Therefore, as shown in Table 2, we administered subthera-

peutic doses of L-’#{176}B-BPA to seven patients and plotted the

estimated curves of the ‘#{176}Blevels using the method of calcula-

tion described above. The estimated curves were calculated by

the input function of the ‘#{176}Blevel obtained by continuous

arterial blood sampling. Typical examples are shown in Fig. 6,

A and B. In the seven cases shown in Table 2, we were able to

substitute the proper rate constants determined by PET in Equa-

tion C. We compared the estimated values with the ‘#{176}Blevels in

surgical specimens for these seven cases. As shown in Table 2,

the estimated values tended to be higher than those for surgical

specimens but were similar to them, except in cases 1 and 2. On

PET, the highest radioactivity area can be set as the concerned

area. As a result, the highest value will be the representative

value. On the other hand, in the case of the surgical sample,

selection of only a high area of ‘#{176}Bis impossible. For this

simple reason, a discrepancy develops between them. However,

in patients from whom a homogenous tumor sample was ob-

tamed, the surgical sample is close to the estimated value, and

the data are considered as validating the adequacy of the present

method.

Discussion

In Vivo L-’#{176}B-BPAPharmacokinetics Determined withL-’8F-10B-FBPA PET. In the treatment of gliomas, L-’8F-

‘#{176}B-FBPA uptake permits clear visualization of the main part

of the tumor and helps to determine the extent of tumor

resection required. Infiltrating cells are present around the

main part of the tumor, and treatment of this area with BNCT

is considered to be the most reasonable approach. We as-

sumed that infiltrating cells have the same biochemical prop-

erties as cells in the main part of the tumor, which could be

on April 9, 2021. © 1998 American Association for Cancer Research.clincancerres.aacrjournals.org Downloaded from

E

3

Plasma iSFIevel

Tumor 18F4evel on PET

20 40 60 80 100 120

mm

140

Clinical Cancer Research 1839

Fig. 7 Differences in the courses of time-activity

curves between tumor lesion and normal brain

tissue. The time-activity curves of ‘�F (in cps/ml)

were real data obtained in a patient with GBM.

When PET scans were performed 1-2 h after theinjection, we found a decrease in radioactivity inthe tumor region after a while. This seemed to bedue to the catabolic process, as suggested by theaction of k4 in the tumors. X axis, time (minutes):

Y axis, ‘8F activity.

macroscopically observed by PET, and planned the use of

thermal neutron irradiation in BNCT. With this concept for

PET-based BNCT, clinical application ofBNCT using L-’#{176}B-

BPA was successful. L-’#{176}B-BPA is cleared rapidly from the

blood but is retained in tumor tissue and actively taken up by

proliferating cells. Its accumulation in tumors was demon-

strated by dynamic PET using L-’8F-’#{176}B-FBPA (6, 9). An

important factor in BNCT is the determination of the ‘#{176}B

concentration in both the tumor tissue and normal brain. An

increase in the tumor:normal ratio increases the efficacy of

BNCT, because it allows a greater tumor dose to be delivered

per dose administered to normal tissue, and the tumor dose is

limited by normal tissue tolerance (4). Appropriate indica-

tions for BNCT include a tumor:normal ratio above 2.5 (21).

We studied the factors contributing to accumulation. An

increase in K, is the principal factor affecting the accumula-

tion of L-’8F-’#{176}B-FBPA (6, 9). As described above, L-’5F-

‘#{176}B-FBPA accumulation is accomplished within 42 mm after

bolus i.v. injection in dynamic PET studies. The rapid

achievement of equilibrium (within 42 mm) was attributed to

K,. When PET scans were performed in patients with high-

grade gliomas during 1-2 h after injection, we found a

decrease in radioactivity in the tumor region after a while

(Fig. 7). This seemed to be due to catabolism, as suggested by

the action of k4 in tumors. Kabalka et a!. (22) recently

showed that the four-compartment model can be used for the

evaluation of the tracer kinetics of L-’8F-’#{176}B-FBPA. They

obtained the arterial blood concentration input function from

a ROl representing vascular activity in the internal carotid

artery. Although their method is different from our compart-

ment model, the results of the kinetics were essentially sim-

ilar to ours. Assessment based on these characteristics of

� suggests that accurate ‘#{176}Blevels can be

estimated within 1-2 h after iv. infusion of L-’#{176}B-BPA.

Furthermore, the � concentration in normal tissue must be

considered. Briefly, a rapid increase in the tumor ‘#{176}Blevel is

promoted by K, and suppressed by k, until blood ‘#{176}Blevels

decrease rapidly after the end of the iv. infusion; then the ‘#{176}B

level is maintained by k3 and decreased by k4 in tumor tissue

after a while. Although a similar process occurs in normal

tissue, normal tissue has a higher k3 (0.033 ± 0.015) than

does tumor tissue (GBM, 0.018 ± 0.007; AIlI, 0.025 ±

0.014). Fig. 7 shows that differences between tumor and

normal tissue in time course changes decreased with time in

the L-’5F-’#{176}B-FBPA study. This suggests that ‘#{176}Blevels in

normal tissue gradually approach the tumor ‘#{176}Blevel as a

result of the higher k3 value in normal tissue. It may be

appropriate to start BNCT when the difference in ‘#{176}Bcon-

centration between normal tissue and tumor is marked.

Pathological Features and Pharmacokinetics. There

are some important problems regarding pathological features

and pharmacokinetics in gliomas. One of them is whether the

cellularity of the tumor sample relates to the boron content. In

general, the cellularity of tumors is the standard of malignancy.

The detailed comparison of malignancy and L-’8F-’#{176}B-FBPA

uptake was described in our previous report (9). Elowitz et a!.

(23) also referred the correlation. However, because micro-

scopic heterogeneity is observed in tumor tissues, a numerical

determination of the tumor cellularity using a part of the tumor

tissues is generally difficult, although the cellularity can be

classified into various stages based on the evaluation of pathol-

ogists. Therefore, we did not compare the cellularity with the

boron contents. We also recognize that classifying the cellularity

by certain numerical indices is an important subject to under-

stand the detailed mechanism of ‘#{176}B-BPAuptake and the pro-

tection of the brain. Regarding these subjects, future experi-

ments should be performed to solve this problem, including the

administration of a compound, a cell population marker as a

positron tracer, and a simultaneous scan by L-’8F-’#{176}B-FBPA.

Then, discussion comparing the results of positron tracer ad-

ministration with scanning data is thought to be necessary.

Another problem is that if only a relatively small percent-

age of the tumor cells are undergoing mitosis at any one time,

then the uptake of L-’8F-’#{176}B-FBPA would significantly under-

estimate the true size of the tumor. When we investigated the

relationship between the malignancy of the tumor and L- ‘ 8F-

on April 9, 2021. © 1998 American Association for Cancer Research.clincancerres.aacrjournals.org Downloaded from

1840 PET-based BNCT Using L-’#{176}B-BPA: Part II

�Ci3(t) = Ci�(t) - C�_,(t) (G)

‘#{176}B-FBPA uptake, it was clinically found that the malignancy is

greatly dependent on the capacity of amino acid transport by

tumors. However, we have encountered cases of low-grade

tumors with marked amino acid uptake. It was considered that

these low-grade tumors might correspond to metabolically ac-

tive nonproliferating cells. Furthermore, although these tumors

were pathohistologically classified as low grade, the prognoses

of the patients were unsatisfactory (9). In general, however,

macroscopic analysis is the objective level for the methodology

described here, and it is difficult to differentiate the small

percentage of metabolically active cells included in the entire

tumor mass. Because there is the possibility of underestimating

regions with the proliferating potential, treatment of this area

with BNCT should be considered as a concept of the target area.

Conditions for Quantitative Measurement of Tissue ‘#{176}BLevels in Clinical BNCT. We found that the segmental con-

volution method is appropriate for clinical BNCT. This method

uses two different modalities. Equation E involves data from

PET and the prompt ‘y method. The rate constants obtained by

PET are used in the weight function in Equation C. The input

function describing changes over time in i% levels, C�(t), is

also involved in Equation C. In the present study, when we

compared the ‘#{176}Bvalues estimated by PET with those of sur-

gical specimens, the estimated values were found to be very

close to the iO� levels in the surgical specimens, as shown in

Table 2. The similarity between L�i8F�i�B�FBPA and L-’#{176}B-

BPA in their pharmacokinetics was thus verified. Therefore, the

proper rate constants for L-’#{176}B-BPA in each subject were re-

garded as similar to those for L-’8F-’#{176}B-FBPA obtained by PET

(Fig. 1, A and B).

When actually performing BNCT, we must reconfirm the

tumor 10B level by measurement of L-’#{176}B-BPA in arterial blood.

As mentioned above, the input function of L-’#{176}B-BPA can be

determined by blood sampling during this 1-2 h. The tumor iOi�

level can be more precisely determined by convoluting the input

function of C�(t) to Equations B-E using K1, k2, k3, and k4. In

evaluating indications for BNCT, Ic* and Ur* are useful for

estimating the dose of L-10B-BPA required (6), but the tumor

10B level calculated by the segmental convolution method based

on the prompt -y measurement is more reliable for the perform-

ance of BNCT.

In the former (6) and the present studies, we evaluated two

different methods ofestimating i% concentration. The first method

permits estimation by PET alone and is useful for determining the

indications for BNCI’. In the second method, the proper rate

constant for brain tumors is used, and the i0J3 concentration in

tumor tissue is obtained using the input function for plasma ‘#{176}B

concentration. This method is useful for the performance of clinical

BNCT, and with it, tissue concentration can be measured even after

slow infusion. Systematic PET-based BNCT is possible using one

or the other of those two methods according to the purpose. With

the first method, the administered dose of L-’#{176}B-BPA can be

estimated for individual patients by PET, and patients can be

selected for future BNCT. With the second method, the concentra-

tion in the tumor can be accurately determined at the time of

neutron irradiation using L-’#{176}B-BPA, and the method is widely

suitable for clinical BNCT, because the averaged PET data as

shown in Table 1 are enough to use in future patients without

individual PET study. The development of this PET-based BNCT

system has made it possible for us to perform BNCT accurately

using reliable objective data. Using this system, we have performed

BNCT with thermal neutrons on patients with high-grade gliomas

(24, 25).

In conclusion, tumor � #{176}Blevels can be more accurately

determined by convoluting the input function to the weight

function using rate constants. The similarity in pharmacokinet-

ics between L�i8F�iOB�FBPA and L�i�B�BPA was also con-

firmed. Using the PET-based BNCT system described here,

clinical performance of BNCT using L-’#{176}B-BPA is possible.

Acknowledgments

We acknowledge the technical support and effort of Kazuo Wakita

and Hitoshi Horii (Cyclotron Unit, Nishijin Hospital, Kyoto, Japan). We

gratefully thank Dr. Kiichi Ishiwata (PET Center, Tokyo MetropolitanInstitute of Gerontology, Tokyo, Japan) for helpful advice regarding‘8F’o��PA synthesis and its biological properties.

Appendix

Cp*(t) = �

i= I

K,Ci(t) = [(k3 + k4 -

( a2 �I)

with

+ (a2 k3 - k4)e_a2h] #{174}Cp(t)

a, [(k2 + k3 + k4) - �(k2 + k3 + k4)2 - 4k2k4]/2

(A)

a2 [(k2 + k3 + k4) + �(k2 + k3 + k4)2 - 4k2k4]12 (B)

Ci(t) = (Ae�’ + Be�2) #{174}Cp(t)

with

K,A= (k3+k4-a,)

(a2 - a,)

K,B=

(a2 - a,)

tJ � 1 < t <

Cp�(t) G�t + K�

C�(t) = f’(Ae�_� + Be�2(T)) X (G�T +

AGCi�(t) - -y�(e’” + a,t - 1) + � + a2t - 1)

a, a2

(C)

(D)

(E)

AK BK.- -�(e” - 1) - � 1) (F)

on April 9, 2021. © 1998 American Association for Cancer Research.clincancerres.aacrjournals.org Downloaded from

Clinical Cancer Research 1841

References

Ci(t) = � �Ci�(t)

I . Sweet, W. H. Early history of development of boron neutron capturetherapy of tumors. J. Neuro-Oncol., 33: 19-26. 1997.

2. Asbury, A. K., Ojemann, R. G., Nielsen, S. L., and Sweet, W. H.

Neuropathologic study of fourteen cases of malignant brain tumor

treated by boron-lO slow neutron capture radiation. J. Neuropathol. Exp.Neurol., 31: 278-303. 1972.

3. Hatanaka, H. Experience of boron neutron capture therapy for ma-

lignant brain tumors with special reference to the problems of postop-

erative CT follow-ups. Acta Neurochir. Suppl., 42: 187-192, 1988.

4. Barth, R. F., Soloway, A. H., and Fairchild, R. G. Boron neutron

capture therapy of cancer. Cancer Res., 50: 1061-1070, 1990.

5. Soloway, A. H., Barth, R. F., Gahbauer, R. A., Blue, T. E., and

Goodman, J. H. The rationale and requirements for the development ofboron neutron capture therapy of brain tumors. J. Neuro-Oncol., 33:

9-18, 1997.

6. lmahori, Y., Ueda, S., Ohmori, Y., Sakae, K., Kusuki, T., Kobayashi.T., Takagaki, M., Ono, K., Ido, T., and Fujii, R. Positron emissiontomography-based boron neutron capture therapy using boronophenyl-alanine for high-grade gliomas: part I. Clin. Cancer Res., 4: 1825-1832,1998.

7. Mishima, Y., Imahori, Y., Honda, C., Hiratsuka, J., Ueda, S., and

ldo, T. in vivo diagnosis of human malignant melanoma with positron

emission tomography using specific melanoma-seeking ‘8F-DOpA an-

alogue. J. Neuro-Oncol., 33: 163-169, 1997.

8. Imahori, Y., Ueda, S., Ohmori, Y., Yoshino, E., Ono, K., Kobayashi,

T., Ido, T., and Mishima, Y. A basic concept for PET-BNCT system. In:

Y. Mishima (ed), Cancer Neutron Capture Therapy, pp. 691-696. New

York: Plenum Press, 1996.

9. Imahori, Y., Ueda, S., Ohmori, Y., Kusuki, T., Ono, K., Fujii, R., andIdo, T. Fluorine-l8-labeled fluoroboronophenylalanine and PET in pa-tients with glioma. J. Nucl. Med., 39: 325-333, 1998.

10. Yoshino, K., Suzuki, A., Mon. Y., Kakihana, H., Honda, C..Mishima, Y., Kobayashi, T., and Kanda, K. Improvement of solubility

ofp-boronophenylalanine by complex formation with monosaccharides.

Strahlenther. Onkol., 165: 127-129, 1989.

I 1. Kakihana, H., Yoshino, K., and Mod, Y. Relation between struc-tures of p-boronophenylalanine related compounds and their usefulnessfor boron neutron capture therapy. In: A. H. Soloway et al. (eds.),Advances in Neutron Capture Therapy, pp. 257-260. New York: Plc-num Press, 1993.

12. Phelps, M. E., Huang, S-C., Hoffman, E. J., and Kuhl, D. E.

Validation of tomographic measurement of cerebral blood volume withC-l 1-labeled carboxyhemoglobin. J. Nucl. Med., 20: 328-334, 1979.

13. Ishiwata, K., Ido, T., Kawamura, M., Kubota, K., Ichihashi, M.,

and Mishima, Y. 4-Borono-2-[’8F]fluoro-D,L-phenylalafline as a tar-

get compound for boron neutron capture therapy: tumor imaging

(H) potential with positron emission tomography. Nucl. Med. Biol., 18:745-751, 1991.

14. Gjedde, A. High- and low-affinity transport of o-glucose from

blood to brain. J. Neurochem., 36: 1463-1471, 1981.

15. Patlak, C. S., Blasberg, R. G., and Fenstmacher. J. D. Graphical

evaluation of blood-to-brain transfer constants from multiple-time up-take data. J. Cereb. Blood Flow Metab., 3: 1-7, 1983.

16. Wienhard, K., Herholz, K., Coenen, H. H., Rudolf, J., Kling, P.,Stocklin, G., and Heiss, W. D. Increased amino acid transport into brain

tumors measured by PET of L-(2-’8F)fluorotyrosine. J. Nucl. Med.. 32:

1338-1346, 1991.

17. Coenen, H. H., Kling, P., and Stocklin, G. Cerebral metabolism ofL-[2-l8F]fluorotyrosine, a new PET tracer of protein synthesis. J. Nucl.Med., 30: 1367-1372, 1989.

18. Willemsen, A. 1. M., van Waarde, A., Paans, A. M. J., Pruim, J.,

Luurtsema, G., Go, K. G., and Vaalburg, W. in vivo protein synthesisrate determination in primary or recurrent brain tumors using L-[l-’ ‘C]-

tyrosine and PET. J. Nucl. Med., 36: 41 1-419, 1995.

19. Keen, R. E., Barrio, J. R., Huang, S-C., Hawkins, R. A., and Phelps,M. E. In vivo cerebral protein synthesis rates with leucyl-transfer RNAused as a precursor pool: determination of biochemical parameters tostructure tracer kinetics models for positron emission tomography.J. Cereb. Blood Flow Metab., 9: 429-445, 1989.

20. Huang, S-C., Phelps, M. E., Hoffmann, E. J., Sideris, K., Selin,

C. J., and Kuhl, D. E. Non-invasive determination of local cerebral

metabolic rate of glucose in man. Am. J. Physiol., 238: E69-E82, 1980.

2 1 . Laramore, G. E., and Spence, A. M. Boron neutron capture therapy

(BNCT) for high-grade gliomas of the brain: a cautionary note. Int. J.Radiat. Oncol. Biol. Phys., 36: 241-246, 1996.

22. Kabalka, G. W., Smith, G. T., Dyke, J. P., Reid, W. S., Longford,

C. P. D., Roberts, T. 0., Reddy, N. K., and HUbner, K. F. Evaluation offluorine- 18-BPA-fructose for boron neutron capture treatment planning.J. Nucl. Med., 38: 1762-1767, 1997.

23. Elowitz, E. H., Bergland, R. M., Coderre, J. A., Joel, D. D., Chadha,

M., and Chanana, A. D. Biodistribution of p-boronophenylalanine inpatients with glioblastoma multiforme for use in boron neutron capture

therapy. Neurosurgery (Baltimore), 42: 463-469, 1998.

24. Ueda, S., Imahori, Y., Ohmori, Y., Yoshino, E., Ono, K.,

Kobayashi, T., Takagaki, M., Oda, Y., Ido, T., and Mishima, Y. Positronemission tomography and boron neutron capture therapy system to the

patient with malignant brain tumor: the first clinical trial using ‘#{176}B-

BPA. in: Y. Mishima (ed), Cancer Neutron Capture Therapy, pp.

823-827. New York: Plenum Press, 1996.

25. Ueda, S., Imahori, Y., Ohmori, Y., Kusuki. T., Yoshino, E.,Takagaki, M., Kobayashi, T., Ono, K., Fujii, R., and Ido, 1. Positronemission tomography boron neutron capture therapy system and clinical

application to patients with malignant glioma. In: B. Larsson, J.Crawford, and R. Weinreich (eds.), Advances in Neutron Capture Ther-

apy, Vol. II, Chemistry and Biology, pp. 280-285. Amsterdam: ElsevierScience Publishers, B. V., 1997.

on April 9, 2021. © 1998 American Association for Cancer Research.clincancerres.aacrjournals.org Downloaded from

1998;4:1833-1841. Clin Cancer Res   Y Imahori, S Ueda, Y Ohmori, et al.   II.

parttherapy using boronophenylalanine for high-grade gliomas: Positron emission tomography-based boron neutron capture

  Updated version

  http://clincancerres.aacrjournals.org/content/4/8/1833

Access the most recent version of this article at:

   

   

   

  E-mail alerts related to this article or journal.Sign up to receive free email-alerts

  Subscriptions

Reprints and

  .pubs@aacr.orgDepartment at

To order reprints of this article or to subscribe to the journal, contact the AACR Publications

  Permissions

  Rightslink site. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC)

.http://clincancerres.aacrjournals.org/content/4/8/1833To request permission to re-use all or part of this article, use this link

on April 9, 2021. © 1998 American Association for Cancer Research.clincancerres.aacrjournals.org Downloaded from