Nuclear James L. Lear, MD Paul A. Bunn, MD Duane C. Bloedow,

Improved Monoclonal Unbound

K. Kasliwal, PhD #{149} Angela J. Feyenabend, MD #{149} Jonathan G. Dienhart, MD #{149} Rene Gonzalez, MD a Timothy K. Johnson, Stephen W. Maddock, PhD #{149} Stephan D. Glenn, PhD

#{149} Ravindra #{149} David

PhD

a

Tumor Imaging Antibodies Antibody with

Imaging of tumors by using radiolabeled monoclonal antibodies (MoAs) is hindered by the presence of background activity. To reduce this problem, the authors investigated the process of removing labeled MoAs from plasma at selected times by means of extracorporeal immunoadsorption. In seven patients with either lung or breast carcinoma, an indium-ill-labeled munine antibody was intravenously administered. Six to 24 hours later, immunoadsorption was performed by passing the patients’ plasma through a goat anti-mouse antibody column connected to a plasma separator. Whole-body computer images were obtained before and after the treatment. Blood pool activity in the images was reduced by an average of 59%, while tumor activity dropped by only 10%. Tumor-toblood activity ratios therefore more than doubled, improving by an avenage of 121% between the pre- and posttreatment image sets. Eight of 12 areas of known disease and three areas of unknown but later documented disease were detected after the immunoadsorption process, while the three areas of unknown disease and three of the areas of known disease were not detected in the preclearance images. Thus, the feasibility of using extracorponeal immunoadsonption to improve MoA imaging of tumors was demonstrated.

T

terms:

mental

a

Radionuclide

Radionuclide

imaging, imaging,

neoplasms

Radiology

1991; 179:509-5 12

experi-

in diagnosis

of

P. Pratt, PhD

PhD

a

a

with Radiolabeled by Plasma Clearance of Anti-antibody Column’

imaging of tumors by using radiolabeled monocbonal antibodies (MoAs) is often hampered by the presence of high levels of background activity in the blood pool. Various methods used to reduce this background activity in attempts to improve imaging have recently been described. These include the use of Fab fragments, which clear from the circulation more rapidly than intact antibodies (1,2), the injection of a second antibody to clear the blood of the imaging antibody (3,4), the use of metabobizable chelates, which reduce background activity through release of the radiolabel from the circulating antibody (5), the use of reversibly labeled antibodies from which the Iabel can be removed by injecting a hapten chase (6), and pretargeting of tumors with nonradiobabeled antibody followed by in vivo labeling (7). Also, antibody images have been improved by mathematically subtracting blood pool images obtained in the same patient with a separately administered intravascular tracer (8). While all of these approaches offer potentially reduced background activity, problems remain in their practical use. These problems include deHE

creased tumor uptake with techniques in which the labeled compound clears too rapidly from the plasma, uncertainties in extrapobating mouse toxicity results to those from human studies with techniques that require second injections of antibodies or haptens, and difficulty in aligning and subtracting dual-radionuclide images. We therefore explored a new approach to the selective reduction of background activity: removal of unbound radiolabeled antibody from the plasma by means of extmacorporeal immunoadsorption. This technique offers the advantage of the ability to control the time at which the circulating activity is removed without the need for injection of additional, possibly immunoreactive, compounds. This control of antibody removal time is important because background activity can normally take several days to drop to acceptable levels for imaging, while tumor uptake usually peaks much earlier.

MATERIALS Radiolabeled

AND

METHODS

Antibodies

Munine antibody leah, Fla) was used

KC-4G3 (Coulter, Hiain our study. KC-4G3

is an IgG3 antibody that recognizes a high molecular weight (438,000 d) human milk-fat globule membrane antigen (9). From

(J.L.L., (R.K.K., D.C.D.,

the

Divisions

of Nuclear

This antigen is expressed on many epithelial tumors, including human breast and non-small cell lung carcinomas (9). While some antigen is shed into serum in

Medicine

A.J.F., J.P.P.), Radiological Sciences T.K.J.), Medical Oncology (P.A.B., R.G.), and Pharmacy (D.C.B.), Universi-

ty of Colorado

Index

Medicine

Health

A034, COBE Coulter

4200 E Ninth Laboratories, Immunology,

From

the

1990

Sciences

RSNA the

Center,

Box

Ave. Denver, CO 80262; Denver (S.W.M.); and Hialeah, Fla (S.D.G.). scientific

supported

by

RSNA

tion Fund in part by

as a Scholar. This National Institutes

assembly.

Research

and

the

J.L.L. Educa-

work is supported of Health grant

NS-26657, National Cancer Institute 42767, Clinical Research Center grant

grant CARR-5i,

and

Received

the

Cancer

November 3, 1991;

20, revision

January

i8.

C

RSNA,

League

of Colorado.

most

patients,

levels

are

generally

low

enough so that if antibody doses greater than 1 mg are administered, the tumor uptake is not affected by the presence of serum antigen (10). Purified KC-4G3 was conjugated i-(para-isothiocyanatobenzyl)-diethyl-

with

enetniaminepentaacetic conjugate was then

acid (1 1) and the radiolabeled with in-

Abbreviation:

=

1990; revision requested January received January 10; accepted

Address 1991

reprint

requests

to J.L.L.

MoA

monoclonal

antibody.

509

dium-iii mCi in

by In-ill

incubating approximately with approximately

10 2 mg

of antibody-diethylenetriaminepentaacetic acid

conjugate.

purified

Fine

by

preparation

a Supenose

(Phanmacia NJ)

The

with using

was

12 HR

Chemicals,

column

Piscataway,

low-pressure

liquid

chroma-

tography. The final sterile, pyrogen-free, product was more than 98% pure In-illdiethylenetniaminepentaacetic acid-KC-

4G3 with mCi/mg ity

a specific activity of 3.8 (± 1.0) and an average immunoreactiv-

of 74%.

Immunoadsorption The device

Device

extracorporeal immunoadsonption consisted of a polycarbonate

umn

connected

to a Centry

TPE

separator (Cobe Laboratories, Cob) (Fig 1). The column pohycanbonate lently

wafers

bound

against minimal proteins).

of mouse

was Total

and

Seven

IgG

to human antibody

Pre

Column

Sample

Poil

(with

plasma (approx-

present to bind extraconporeal

10-12 vol-

COBE TPE Plasma Separator

Procedures

patients,

six with

carcinoma

and

nocarcinoma, their clinical and metastatic means

non-small

one

with

ade-

Figure

were studied. As pant of wonkup, sites of primary disease were determined and

the radioimaging consent and the

1.

Extracorponeal

computed

were to each

was

was antigen

Duplicate

i-mL

obtained 2,

then patient

performed,

obtained reactive

from each fraction

12,

24,

48,

tional samples ly before and tion

Table 1 Change in Antibody

was with

pie

radiolabeled

was

first

nected

column

Activities

Due to Extracorporeal

and Immunoadsorption (h)

Tumor/

Whole

Liver

Ventricle

Tumor

Liver

Body

1 2

20 7

-73 -69

-26 -41

-14 -15

-55 -55

-8

+104

+24

3

-10

+100

+32

-9 -ii -10

+146 +134 +121

+75 +48 +45

22

-71

-32

-20

-58

4

13

-55

-32

-i8

-55

5

20

-72

-40

-17

-66

10 6

-72 -66 -68

-48 -40 -37

-32 -19 -19

-63 -62 -59

6 7 Mean

plas-

after

procedure was administration A blood

from

each

and placed

perfusion

sam-

patient of the antipatient

a femoral and con-

pump.

Im-

phoresis autoradiography. Whole-body anterior ages were obtained immunoadsorption

and before with

posterior

and after the a gamma camera

(Gemini 700; Technicare [GE tems, Milwaukee]) interfaced

tosh

Ilci

computer

im-

(Apple

Medical Systo a Macin-

Computer,

Cu-

high-resolution

Radiology

Left

Body

analyzed

a

(%)

Plasma

pertino, Calif) (13). Each image was acquired for 30 minutes with a matrix of 600 X 150 pixels with 16-bit depth and

510

Immunoadsorption

Tumor/

Whole

munoadsonption was performed until two to three total patient-plasma volumes (estimated from standard height-weight charts) had been treated. After use, the column was eluted and the contents were with

immunoad-

connection.

anticoagulation, catheter was

to the

=

MoA

Patient

the

and used to verify the effectiveness column at clearing the radiolabeled body from the plasma. Then, each underwent vein-access

IA

Administration

by gamma

antibody.

obtained

study.

were

in

determined

The immunoadsonption performed 6-24 hours of the

in this

Between

30 minutes, and 1, and 96 hours. Addiobtained immediatethe immunoadsonp-

decay

used

Time

admin-

samples

Radioactivity

ma samples counting

72,

were after

procedure.

intravenously over 1 hour.

plasma

at 5 minutes,

4, 8,

device

to-

of the In-i 1 l-diethylenetniaminepentaacetic acid-KC-4G3 was determined for each patient dose (12). Infusions of approximately 4 mCi of the radiolabeled antibody istered

immunoadsonption

sorption.

endoscopy,

radiography, (CT).

Before

cell

breast

of bronchoscopy,

standard mography informed patient,

cova-

antibody

180 mL.

Patients

by

was

goat

cross-reactivity Enough goat

was

lung

to which

Fc portion

imately 30 mg) mg of KC-4G3.

ume

Lakewood, housed three

a nonspecific

the

cob-

plasma

polyacrylamide

gel

electno-

was

displayed

on a 1,280 gray

scale

X

1,024

monitor.

pixel

experienced nuclear (J.L.L., A.J.F.) drew oven the visualized left

ventricle,

copied

of images. gions were

Total activities determined,

connected. corresponding

The

subsequent

were

set

whole

posttreatment were then

gions

Two

and

medicine regions tumors,

body

on

of images. The to the pretreatment

geometric anterior

computed

physicians of interest liver, spleen, the regions set

within the reand decay was means of the and posterior neand used for the

analyses.

May

1991

(decrease in plasma concentration times estimated total plasma volume). Polyacmylamide gel electrophoresis autoradiogmaphy of the column contents indicated that the madioactivity that was removed consisted mostly of intact, labeled KC-4G3, with a small and variable amount madiolabeled immune complexes. Left ventricular region of interest counts from the computer images creased an average of 59%. On the other hand, tumor counts dropped much less after immunoadsorption-

Figure

2.

Computer

visibility.

images

Anterior

(a)

immunoadsorption sites

of

has

night

disease

been

pulmonary

treatment

image.

as a cold

than

the

left

A right in

Note

rather

(p)

the

also hot

of extraconponeal

images

were

approximately in

reduced. artery

the effect

posterior

procedure,

metastatic

activity

show

and

lung

are

hihan

hours

more

lesion

pretreatment

the

lesion

20

was

of

the

the

before MoA

and

by blood is well

metastasis

nonspecific

the

the

activity in

antibody

blood

Detected

with

Staging

Radioantibody

Patient

Staging Studies

the

pool

which

appears

by

normal

the

Imaging,

Imaging

Before Immunoadsorption

Follow-up Studies

After Immunoadsorption

1

2

2

4

4

0 2

0 0

0 0

0 2

4

2

0

2

3

5

0

0

0

0

6 7

2 4

1 2

1 4

2 4

12

5

11

15

Note-Staging

of increased

studies

included

activity,

with

plain

radiography

the exception

Immunoadsorption times to process the two to three patient-plasma volumes ranged from 3 to 4 hours. All patients tolerated the procedure well, and no adverse reactions attnibuted to the procedure occurred. However, one patient died the foblowing day from another cause.

179

a

Number

of one

and

liver

CT.

lesion

Based

RESULTS

Volume

Radioantibody

2

Abnormal

MoA

that

had decreased

on

pre-

and

activity

DISCUSSION

post-

2 3

Total

areas

Studies,

only 10%. Thus, the tumor-to-blood activity ratio increased significantly after the immunoadsomption, rising by an average of 121% (Fig 2). Before immunoadsorption, five of 12 known disease sites were visualized on the images (Table 2). After immunoadsomption, three additional sites of known disease became visible on the antibody images. Also, three areas not known to have disease (on the basis of staging studies) demonstrated abnormal antibody uptake. All three of these areas were shown to represent metastatic disease at foblow-up radiobogic examinations.

in the

liver.

Table 2 Number of Disease Sites and Follow-Up Studies

de-

Two

pool

proved), of

after

after visualized

(biopsy uptake

on tumor just

administration.

demonstrated

hidden

image

of a liver

because

just

after

clearly

that anterior

presence

immunoadsonption

obtained

of

consisted

of

activity.

posttreatment

plasma activities, an average of 68% of the antibody in the blood was memoved over the 2-4 hours of the immunoadsorption procedure (Table 1). The total amount of radioactivity in the column at the end of each immunoadsomption procedure was almost equal to that estimated to have been lost from the patient’s plasma

The immunoadsonption procedure dramatically changed the time course of radiolabeled antibody activity in the blood. Almost 70% of the activity in the plasma was lost during the treatment, a rate more than 10-fold that which occurs through normal redistribution of KC-4G3 activity (10). In other words, a decrease in plasma activity of less than 7% would have occurred during the same time penod without use of extraconporeab immunoadsomption. The removal of circulating antibody by means of extracorponeal immunoadsomption dramatically improved tumor-to-blood pool activity ratios in the MoA images. The increase averaged 121% and was caused by the large decrease in circulating antibody activity with little loss of radiolabeled antibody from the tumoms. Tumors were therefore better visualized after plasma clearance, and the sensitivity of the examination for enabling detection of areas of disease increased. In two of seven patients, areas of disease were detected with radioimaging after plasma antibody clearance that were not detected with staging radiography and CT. The procedure was tolerated well

by the patients, tions poreal

and

no adverse

occurred. However, immunoadsorption

extracoris not

Radiology

reaca

a

511

trivial process. Compared with most imaging procedures, extracompomeal immunoadsomption is more invasive, since it requires either very good pemipherab venous access or the placement of a central venous line for 2-4 hours and concomitant anticoagulation. Therefore, despite the improved imaging of tumors that it provides, it is unlikely to be used for routine staging and follow-up of malignancies. Extracorporeal immunoadsomption should, thus, be further examined with attention to specific indications. The potentially improved sensitivity in detection of disease sites compared with that of conventional procedures could be important in specific instances in which accurate staging is necessary. Also, the substantial meduction

in

plasma

and

high

doses

is considered and, further investigation.

of labeled

tories ing

1.

Radiology

with

COBE

the

Immunology used

in this

8.

Labora-

anti-antibody

3.

4.

6.

7.

WahI

for supply-

DM,

HS,

tion

study.

Pnimus

9.

of cancer

Hofheinz

CW,

Philpott

GW.

Goodwin

CF.

DA,

Meares

10.

Dienhart

DC,

C, et

al. Use of specific antibody for rapid clearance of circulating blood background from radiolabeled tumor imaging proteins. Eur J Nuci Med 1984; 9:209-215. Goodwin DA, Meares CF, McCall MJ, et al. Cheiate conjugates of monoclonal antibodies for imaging lymphoid structures in the rat. J Nucl Med 1985; 26:493-502. MK,

Goodwin

DA,

Meares

CF.

al. Metabolizable In-i 1 1 chelate conjugated anti-idiotype monoclonal antibody for radioimmunodetection of lymphoma in mice. Eur J NucI Med 1986; 12:455-460. Goodwin DA, Meares CF, David GF, et al. Monoclonal antibodies as reversible equihibrium carriers of radiopharmaceuticals. Int J Nucl Med Biol 1986; 13:383-391. Goodwin DA, Meares CF. McCall MJ, McTigue M, Chaovapong W. Pre-targeted immunoscintigraphy of murine tumors indium-i

J NucI

1 1-labeled

Med

Cancer

DC,

Miller

Schmelter

a monoclonal

7076. Brechbiel

et al. benzyl)

MW,

12.

Gansow

I,

Bunn

JL,

et al.

cancers

KC-4G3,

OA,

and

Atcher

tumor

Chem

Boven

PA.

Lear

lung

of l-(p-isothiocyanatoof DTPA and

Inorganic

Lindmo

on huAssoc

a human milk fat globuRes 1990; 50:7068-

labeling

studies. 2781.

et al.

recognizes

antibody,

Synthesis derivatives

antibody

RF,

cell

Res C,

KC-4G3,

of non-small

which recognizes lin antigen. Cancer

Ben-

antibodies

antigen.

antibody,

Imaging

ii. Diamonti

FH,

a novel widely expressed antigen man epithelial cancers. Proc Am Cancer Res 1987; 28:391.

Im-

and tumor locahizaF(ab’)2. J NucI Med

Deland

radioactive

D, Dienhart

with Parker

EE,

Radioimmunodetec-

with

Monoclonal

RL,

Kim RFJ.

to carcinoembryonic 1980; 40:2984-2992.

proved radioimaging tion with monoclonal 1983; 24:316-325.

Haseman

Goldenberg ne

Wilbanks AB, Peterson JA, Miller S. et al. Localization of mammary tumors in vivo with 1-131 labeled Fab fragments of antibodies against mouse mammary epithelial (MME) antigens. Cancer 1981; 48:1768-

haptens.

a

Coulter

antibody

with

512

us

1775. 2.

merits U

and

the

We thank

providing

References

antibody

therefore,

for

column

5.

whole-body

activities compared with tumor activity increases relative tumor radiation dose. This feature could be useful in cases in which radioimmunothemapy

with

Acknowledgments:

EDTA:

imaging

1986;

E, Cuttitta

Determination

RW,

25:2772-

J,

F, Fedorko of the

immuno-

reactive fraction of radiolabeled monoclonal antibodies by linear extrapolation to binding at infinite antigen excess. J Im-

et

munol 13.

Lear

Methods JL,

Feyerabend quisition, the

personal

Pratt

1984; JP,

AJ. display,

72:77-89.

Roberts

DR.

Gamma

camera

and

microcomputer.

processing

Johnson

I,

image

ac-

with Radiology

1990; 175:241-245.

bifunctional

1988;

29:226-234.

May

1991