Nucl. hfed. Biol. Vol. 19, No. 6. pp. 659-668, Int. .I. Radiat. Appl. Instrum. Part B Printed in Great Britain. All rights reserved

1992 Copyright


0883-2897/92 SO0 + 0.00 1992 Pergamon Press Ltd

Tumor Localization and Radioimaging with Mixtures of Radioiodinated Monoclonal Antibodies Directed to Different Colon Cancer Associated Antigens JAMES W. FLESHMAN, JUDITH M. CONNETT*, DAVID M. NEUFELD, TODD J. GARVIN and GORDON W. PHILPOTT Department of Surgery, Jewish Hospital, Washington University Medical School, 216 S. Kingshighway, St Louis, MO 63110, U.S.A. (Received 2 January 1992) After demonstrating enhanced tumor cell binding with a mixture of monoclonal antibodies (MAbs) in vitro, biodistribution and immunoscintigraphy studies with 3 radioiodinated anti-colon cancer MAbs and a non-specific control MAb (MOPC) were conducted in a human colon cancer (GW-39)-hamster model system. Each of the specific MAbs, but not MOPC, demonstrated extensive tumor binding and in scintigrams affected visualization of all large tumors (> 0.85 g) over background. Using single MAbs, few small tumors (0.19-0.50 g) were defined above background (O-29%). However, with combinations of these specific MAbs small tumors were more frequently defined in scintigrams (4367%). Radioimages using higher doses of MAbs and small, younger tumors more clearly demonstrated the superiority of a MAb mixture. These results confirmed that combinations of MAbs to different antigens can detect smaller tumors with better tumor localization when compared to component MAbs used singly. This study supports the concept that tumor targeting and detection may be enhanced with appropriate mixtures of MAbs.

Introduction MAbs directed to cancer-associated antigens have great potential for improving tumor detection and treatment (Buraggi et al., 1991; Epenetos and Kosmas, 1989; Goldenberg et al., 1989; Halpern and Dillman, 1987; Larson, 1990; Schlom and Weeks, 1985). A number of MAbs to colon cancer associated antigens have shown specific in vivo targeting to human colon cancer (Bishof-Delaloye et al., 1989; Chatal et al., 1984; Fenwick et al., 1989; Granowska et al., 1990). Although tumor localization and radioimaging with these reagents have been somewhat successful, significant improvement is needed before considering their routine use in the clinical setting. While there are still problems with radiolabeled antibodies such as dehalogenation, immunogenicity of xenoproteins, slow circulatory clearance and high background in liver or spleen (e.g. Halpern and Dillman, 1987), increasing problems related to tumor associated antigens (TAA) also have become evident. Many TAA are found in low concentrations on tumor cells and are heterogeneously distributed within the same tumor as well as among tumors of similar histologic patterns (Hart and Fidler, 1981; *Author for correspondence. NMB 19/fr-D

Olsson et al., 1984; Stramignoni et al., 1983). These latter problems have been approached in several ways, including unmasking TAA sites by removal of sialic acid residues, increasing concentrations of TAA with cytokines, synchronizing cells in the cell cycle and use of mixtures of MAbs directed to different TAA (Andrew et al., 1990; Durrant et al., 1989; Herlyn et al., 1985; Munz et al., 1986; Tagliabue et al., 1986). The approach of using mixtures of MAbs offers the potential of reaching greater numbers of cells within a given tumor as well as increasing the total amount of MAb that can be bound to the tumor. Although preliminary success has been reported in several in vitro and in vivo studies using mixtures of antibodies (Andrew et al., 1990; Chatal et al., 1984; Connett et al., 1990; Mujoo et al., 1991; Munz et al., 1986; Tagliabue et al., 1986), at least one study has failed to show mixture superiority (Matzku et ai., 1989). In colorectal cancer preliminary reports have suggested that tumor deposits might be detected at a higher rate with combinations of radiolabeled MAbs (Chatel et al., 1984), or F(ab’), fragments (Munz et al., 1986). The purpose of this study was to compare the tumor localization and clearance properties, and the radioimaging potential of three anti-colon cancer MAbs used alone and in combination. Two are 659




directed to the same epitope of human carcinoembryonic antigen (CEA) and the third, MAb 1A3, to a non-CEA colon cancer associated antigen. Previous immunohistological studies established that MAb lA3 binds well to most human colorectal cancers while showing little or no cross-reactivity to normal gastrointestinal tissue or to any of 19 other normal tissues tested (Connett et al., 1987). These three MAbs have a high affinity for the human colon carcinoma cell line (GW-39) carried in hamsters (> lOmaM), while non-specific MOPC does not. These MAbs in this tumor-animal model provide an opportunity to test whether mixtures of specific MAbs might improve tumor imaging over single MAbs with radioimmunoscintigraphy. Materials

and Methods

Monoclonal antibodies

The three specific anti-colorectal cancer MAbs were all developed in our laboratory following procedures previously described (Fenwick et al., 1989; Lockhart et al., 1980; Wahl et al., 1983a). The fourth monoclonal antibody, MOPC (American Type Culture Collection, Rockville, Md, U.S.A.), was used as a negative control. All antibodies were purified from mouse ascitic fluid, and each final antibody preparation was 2 85% pure. All 4 MAbs were of the immunoglobulin (Ig)G, , kappa subclass. Competitive binding experiments indicated that MAbs 2930 and 45-9 reacted with the same epitope and thus, these two anti-CEA MAbs were not tested in combination in vivo.

The immunoreactivity (IR) of the MAbs was routinely tested using radioimmunoassays (RIAs) under conditions of antigen excess (Lindmo et al., 1984). Target antigens included purified CEA, GW-39 and SW1 116 cells. After radiolabeling, MAb IA3 showed immunoreactivity values of 80-90%, while values of Z&65% for 45-9 and 2930, and 2% for control MOPC were routinely obtained. The lower IR values for the anti-CEA MAbs were due to the presence of contaminating irrelevant light chain synthesized by both anti-CEA hybridoma lines. Using lactoperoxidase techniques (Marchalonis et al., 1971), specific activity of labeled antibody ranged from 2.5 to 15.7 pCi/pg. In vitro MAb binding studies used human colon cancer LS 174T cells (American Type Culture Collection) as target cells in a direct binding RIA. Cells maintained in McCoy’s medium with 15% donor calf serum were removed from monolayers by trypsinization (0.25% trypsin with 0.02% EDTA in saline; 37”C, 30min), washed 3 times, and adjusted to 5 x lo5 cells/ml/tube in phosphate buffered saline (PBS) containing 1% bovine serum albumin (BSA). 5 or 1Opg of [“‘I]MAb (lA3, 2930, MOPC or IA3 + 2930) was added in 5OkL PBS. Incubation was at 4°C for 4 h with constant shaking. Cells were pelleted by centrifugation and washed 3 times in cold

et al.

PBS containing 1% BSA. Cell pellets were counted in a Beckman y 300 counter. Tumor model

A transplantable human signet ring colon cancer, GW-39, was carried in the right thigh musculature of male Golden Syrian hamsters weighing 80-100 g (Fenwick et al., 1989). Intramuscular injection of 0.054.5 mL of a 50% ceil suspension (v/v) yielded 0.14-l .Og tumors in 5-7 days. Hamsters were injected with radioiodinated MAbs at different times after tumor transplantation so that localization in different size tumors could be evaluated. In most experiments (Tables 24, Figs l-4), tumors were older than 5 days when MAbs were injected. In one radioimaging experiment (Table 5, Fig. 5) 1.5 day old tumors (0.3-0.4 g at the time of injection) were used. All animals received Lugol’s solution in their drinking water 2-3 days prior to MAb injection. Using a computer program developed in our laboratory and GW-39 growth curve data derived from 177 untreated hamsters whose tumors were harvested at daily intervals for a month after transplantation, we calculated tumor weight at the time of MAb injection. The predicted weight was dependent upon the interval between tumor transplantation and MAb injection, the size of the tumor inoculum, as well as the measured weight of the tumor at the time of harvest. Experimental design

A double-label experiment was performed with each anti-tumor antibody (2930, 45-9, lA3) and the non-specific MOPC, to confirm tumor localization of the specific MAbs. A mixture of ‘3’I-labeled specific MAbs (5 pg) and ‘251-labeled non-specific antibody (5 pg) in 0.1-0.4 mL was injected intracardially (ic.) into hamsters carrying approx. 1 g GW-39 tumors. Each animal received l-10 PCi of r3’I and ‘2SI-labeled antibodies. Animals were sacrificed 2, 4, 7 or 11 days after antibody injection (n = 3-5 for each time point). Tumors and tissues (blood, skin, muscle, urinary bladder, heart, lungs, liver, spleen, kidneys, stomach and small intestine) were harvested, rinsed in PBS and blotted dry. After weighing, tissues were counted in a dual chamber y-counter. Counts per minute were corrected for channel spill-over and physical decay. Several parameters were determined by computer program: (1) the percent of injected specific antibody and MOPC that were present per g of tumor or tissue (%ID/g); (2) tumor or tissue to blood ratios; (3) tumor to non-tumor tissue ratios (T/NT); (4) the specific localization ratios (antibody T/NT + MOPC T/NT); (5) tumor weight at time of MAb injection. In another series of experiments, radioimages were obtained by y-scanning three groups of hamsters to identify the smallest tumors which could be defined by ‘3’I-labeled MAbs using simple scanning techniques without background subtraction or computer smoothing. Tumor images were visually compared,



MAb mixtures

first, to the opposite thigh which had no tumor, and secondly, to the liver, heart, kidney and stomach where background was highest. Tumors were classified as “defined” when at least 3 observers independently agreed that the scans showed more radioactivity (higher density) in the tumor than in any other area. A few tumors were visualized without being defined. These tumors were distinguishable compared to the opposite thigh, but the density of radioactivity in the tumor was less than the background levels seen over the liver. Full body images were obtained every 24 h after antibody injection using a “pho gamma” camera through a 4 mm pinhole collimator. Isotope range was 360 keV with a 20% window and 7000-10,000 counts were collected on Polaroid film using one of two lenses. Although these lenses differed slightly in their image magnification properties and lens aperture, no differences in results were found when a single animal was imaged with both lenses. All scintiscans in these experiments were interpreted without further processing. Hamsters were sacrificed after the tumor was clearly defined, or alternately, 8-11 days post-MAb injection when undefined tumors were incapable of being defined by these techniques. Tumors and tissues were harvested, weighed and the 13’1counted. The first group in these radioimaging experiments was injected i.c. with 20 or 40 pg of ‘3’I-labeled specific or non-specific antibody (2930, 45-9, lA3 or MOPC) in 0.14.3 mL PBS. The second group was injected i.c. with a mixture of 2Opg [‘3’I]anti-CEA MAb (2930 or 45-9) and 2Opg [13’I]1A3 MAb in 0.3-0.5 mL PBS. Tumors ranged from 0.25 to 3.0 g for both groups at harvest. In a second radioimaging experiment, the tumor model was further optimized to enhance uptake of MAbs and background subtraction techniques were used to define tumor scans at earlier times after MAb injection. Four groups of hamsters (injected with 0.5 mL of 50% tumor cell suspension in their right thighs) were injected i.c. with [‘-“I]MAbs 1.5 days after tumor inoculation. Three days later hamsters were scanned with a LFOV y-camera with a 6mm pinhole collimator collecting 50,000 counts that were stored in a computer. The four groups were: (1) MOPC (3OOpg, 65 p(i); (2) 45-9 (3OOpg, 88 PIi); (3) lA3 (300 pg, 90 PCi); and (4) lA3 plus 45-9 (150 pg of each, 92 PCi total). The images were digitized and the radioactive density in fields of interest (FOI) were determined in: (1) tumor, (2) a comparable area in the left control leg, and (3) the area over the liver, kidney and heart region (upper abdominal background). These FOI were compared by the formula: (tumor FOI - left leg FOI) + upper abdominal background FOI. Also, a tumor index was calculated [(tumor counts per pixel-left leg counts per pixel) + tumor weight at injection] to normalize any possible effects of variation in tumor weights. Scintiscans were evaluated in two ways: first, without any background subtraction; and second with sub-


in uivo

traction from the entire scan of both background (cts/pixel in the left control leg FOI) and non-specific tumor uptake of MOPC (9.4 cts/pixel) from the entire scan. The cts/pixel subtracted from the scans ranged from 13.4 to 16.1. Statistical comparisons were determined by the one tailed Student’s t-test, linear regression model and univariate analyses. Ranking results before performing a general linear regression analysis verified the applicability of the above tests. Results In vitro binding studies In vitro MAb binding studies, conducted under conditions of functional antigen saturation, showed significantly increased MAb binding to target cells with the combination of MAbs lA3 and 2930 (Table 1). Specific MAb binding (total bound minus nonspecific MOPC bound) approached additivity with the mixture, with 263 ng calculated (73 + 190) and 222 ng measured. Doubling the amount of single MAbs injected did not result in significant improvements in the amount of MAb specifically bound to tumor (P > 0.2). In contrast the mixture increased specific MAb binding by at least 21 ng which equated to an increase of 1.7 x 10’ molecules of MAb bound per cell. Double-label in vivo studies

The three specific MAbs studied (2930, 45-9, lA3) behaved in a similar fashion in this GW-39 tumor model of human colon cancer. Double-label experiments with 13’I-specific MAb and [‘25I]MOPC showed tumor accretion of all three anti-colon carcinoma MAbs, reaching about 2% of injected dose per g of tumor 2 days post-injection, followed by a slow steady decline of all three MAbs after 2-4 days (Fig. 1). By 11 days the concentrations within the Table

I, Increased tumor cell binding in uitro with a combination of MAbs Added kg)

Specific MAb bound (ng)


IO 20

73.3 f 3.0 74.2 f 4.3


IO 20

190. I f 6.0 200.6 f 5.7

lo+ IO



IA3 + 2930

*Significantly different (P < 0.02~.0001) than means of IA3 or 2930 alone. Each value is the mean f SEM for 5 samples. [12’I]MAb (lA3; 2930; IA3 + 2930; MOPC) was added to 5 x IO’ LSl74T cells at 4°C for 4 h. The cells were washed x 3 in cold PBS + I % BSA and counted on a y-counter. Total cpm bound were measured and mean nanograms bound calculated from specific activity (IA3 and 2930= 11.6 and MOPC IS.1 aCi/rg). MAb specifically bound is the total bound minus the mean of non-specific MOPC bound (16.2 and 31.3 ng) at the same added concentration. No significant differences in specific MAb binding to tumor cells were found when 2 doses of the same MAb were compared (IO vs 2Opg).





0.5 . 1

















DAYS AFTER INJECTION Fig. 1. Tumor uptake of MAbs in the double-label experiment. Each point represents the mean k SEM of 3-5 animals with GW-39 tumors (total 43) injected with ‘2sI-labeled MOPC (5 L(g)and one of 3 specific MAbs (5 pg) labeled with “‘I: l ,2930; n , 45-9, A, 1A3, 0, MOPC with 2930; 0, MOPC with 45-9; A, MOPC with IA3.

Fig. 2. Tumor or tissue to blood ratios for MAbs in the double-label experiment as described in the legend to Table 3. Each point represents the mean k SEM of 3-5 animals. Lung and liver are presented as examples. All other normal tissues had similar ratios. In tumor: 0, 2930; n , 45-9; A, lA3; +, MOPC. In lung: a, 2930; CR,45-9; A, IA3. In liver: 0, 2930; 0, 45-9; A, lA3.

tumor had dropped to about 1% per g (0.9-l .4%). This decline was primarily due to increasing tumor weight since the mean per cent of injected dose (for the three specific MAbs) in the entire tumor was 3.2 f 0.8% at 4 days versus 3.0 + 0.5% injected dose at 11 days. The control MOPC also had some tumor accretion; however, this was 2-3-fold less than that of any of the three specific MAbs (Fig. 1). Increased injection doses of 20 and 40 pg (immunoscintigraphy studies) gave comparable results with 0.9-1.3% of the injected dose bound per g of tumor after 9-10 days. Blood levels of all three of the specific MAbs were similar with 34%ID/g remaining in the circulation 2 days after injection, decreasing to about 2% by day 4 which was similar to concentrations in tumors (Table 2). By 11 days, all three specific MAbs were significantly lower in blood than in tumors (Fig. 2). Specific binding was also demonstrated by comparison of MAb uptake in tumors versus normal tissues. The tumor to non-tumor ratios of the three specific MAbs for the ten normal tissues examined ranged

from 1.6 + 0.3 (SEM) (tumor/lung for 45-9) to 41.3 f 5.2 (tumor/muscle for 45-9). By 4 days after MAb injection, the binding of 2930, 45-9 and lA3 in tumor was 47 times that in the liver and kidney (tumor/liver 4.8 f 0.2 for 2930; 7.3 f 0.4 for 45-9; 5.7 f 0.3 for 1A3 and tumor/kidney 4.1 f 0.8 for 2930; 5.9 f 0.7 for 45-9; 4.4 & 0.2 for lA3), and these values increased further at later time points. The specific localization ratios (Table 3) demonstrated the ability of each MAb to localize in GW-39

Table 2. Blood levels of monoclonal Day 2






3.1 + 0.3

1.9 f 0.2


3.6 f 0.3

2.4 f 0.2

I.2 f 0.01 1.6 f 0.1

0.8 + 0.1 1.0*0.1

45-9 MOPC

3.1 kO.1 4.4 * 0.2

2.1 io.2 3.6 + 0.3

1.2*0.1 2.0 * 0. I

0.5 kO.1 l.O&O.l


3.3 * 0. I 4.8 f 0.2

1.850.1 2.1 f 0.2

0.9kO.l 1.7 f 0.1

0.4+0.1 0.9 f 0.1

*Mean percent injected dose per g of blood f SEM (n = 3-5). Each value was calculated from cpm per g of blood at time of sacrifice divided by the total corrected cpm of “’ I or “I injected into each .^^ ammal x IUU.

Table 3. Antibody

specific localization


Day 2




LiIW 2930 45-9 IA3

1.9 * 0.2 1.9 + 0.2 2.0 k 0.3

2.8 & 0.3 6.9 + 2.8 3.2 F 0.6

3.5 f 0.5 4.8 f 0.7 3.8 i 0.5

3.4 + 0.2 7.9 f 1.0 4.2 k 0.6

Kidney 2930 45-9 IA3

1.9*0.1 I.8 fO.l 2.2 * 0.3

2.7 f 0.3 6.8 + 2.7 3.2 + 0.4

3.4 f 0.6 4.8 * 0.7 4.0 rt 0.5

3.4 + 0.2 7.9* I.1 4.4 + 0.6

Lung 2930 45-9 IA3

1.6+0.2 2.0 + 0.2 2.2 * 0.3

2.9 f 0.3 1.2 k 2.7 3.3 f 0.5

3.6 f 0.6 4.9 + 0.8 4.2 + 0.5

3.6 f 0.2 8.2 f 0.9 4.6 f 0.6

MUSCk 2930 45-9 IA3

2.0 5 0.2 2.OkO.l 2.4 f 0.3

2.8 k 0.3 7.3 + 2.9 3.2 f 0.4

3.3 +_0.6 5.0 f 0.7 4.1 f 0.5

3.1 * 0.2 8.1 k I.3 3.7 t 0.4

Blood 2930 45-9 IA3

2.0 + 0.2 1.7 f 0.1 2.1 kO.2

2.8 + 0.3 7.0 + 2.3 3.1 fU.5

3.4 k 0.6 4.8 + 0.7 3.8 f 0.6

3.4 f 0.2 8.1 + 1.0 4.3 f 0.7

*Each value is the mean * SEM of specific localization ratios determined from 3 to 5 animals. A total of 43 hamsters carrying _ I g GW-39 tumors in the right thigh were injected with ‘3’I-labeled MAb (5 pg; l.4-15.OpCi/pg) and ‘Z’I-labeled MOPC (5 pg; 2.5-10.3 pCi/pg). Ratios were calculated from the formula: (T/NT for MAC) + (T/NT for MOPC).

Fig. 3. Representative scintiscans of hamsters with GW-39 tumors in the right thigh injected 11 days earlier with a single radiolabeled MAb. All scintiphotos were taken of hamsters in supine position as panel (a) demonstrates and were without background subtraction. The hamster in (a) was carrying a 4-5 g tumor (arrow); (b) a tumor weight of 0.84 g when injected with 13’I-labeled MOPC (15 pg, 56 PCi); (c) a tumor weight of 0.88 g when injected with “‘I-labeled IA3 (20 fig, 92 PCi); and (d) a tumor weight of 0.73 g when injected with “‘I-labeled 45-9 (14pg, 51 PCi); 10,000 counts were collected for all images. Scintiphotos in (b) and (d), were photographed using one lens (f-stop 9.3) while another lens with slightly higher magnification and wider lens aperture (f-stop 8.0) was used for (c). For this figure, the images in (b) and (d) were enlarged approx. 2-fold to make comparisons easier.


Fig. 5. Representative scintiphotos from four groups of hamsters 3 days after injection of: (a,e) ‘r’ I-labeled MOPC (0.3 mg, 65 nCi); (b,f) 13’I-labeled lA3 (0.3 mg, 90 PCi); (c,g) “‘I-labeled 45-9 (0.3 mg, 85 PCi); and (d,h) ‘3’I-labeled IA3 and ‘3’I-labeled 45-9 (total 0.3 mg, 92 PCi). The tumor weights at the time of injection of MAbs were (a) 0.36 g, (b) 0.29 g, (c) 0.30 g and (d) 0.29 g. The upper row (ad) depicts the original, non-subtracted scans and the bottom row (e-h) shows computer subtracted images. Background subtraction was done by subtracting first the cts/pixel in the left leg field of interest and then by subtracting the average remaining cts/pixel in the MOPC tumor (9.4 cts/pixel) from the entire scan. The actual cts/pixel subtracted from each scan were (a) 14.7, (b) 13.4, (c) 14.7 and (d) 16.1.

Anti-colon cancer MAb mixtures in viuo


above background levels of radioactivity in the liver, kidney and heart. A significant proportion of tumors . weighing between 0.50 and 0.85 g were defined by 0.9 ._...... ~ ..........._.... ~ ....__._.._.......___............_................................... each MAb (67% for lA3, 75% for 45-9 and 83% for D 0.8 . 2930). In contrast, no tumors weighing less than cz . t 0.50 g were defined using MAb lA3, and only 3 of 25 g 0.7 0 tumors (12%) weighing less than 0.50 g were defined when MAbs 45-9 or 2930 were used singly (Fig. 4). g . 1 : .: l.O-T##-yp



5 g








w i 8



0.3 0.2




.0 L 0 __--o__-_-o---__-_---_p--__-__ -




ii .

i .0



. . 0


---;__---?__--__B_____B_____&___ i 0

0. I

Q 8

&a 8 0








Fig. 4. Comparison of weights of GW-39 tumors at the time of injection of different i3’I-lab&d MAbs correlated with ability to define tumor over background in scintiscans. Closed symbols represent defined tumors; open symbols, non-defined tumors. Tumors 2 1.Og are all noted at the I .Og level. Animals were injected with 20 pg (circles) or 40 jig (squares) of single MAb (2930,45-9 or 1A3) or 20 pg each (circles) of 45-9 + IA3 or 2930 + lA3. Tumors 20.85 g (dashed line) all defined. The dotted line defines a “window” of tumor sizes (0.19-0.50 g) within which combinations of MAbs showed improved tumor definition over single MAbs.

tumor compared to normal tissues and to nonspecific MOPC. This ratio (calculated for each MAb for all tissues) ranged from 3 to 8 by day 11 postinjection. MAb lA3 localized in the tumor as well as 2930; 45-9 had slightly higher specific localization ratios than 2930 and lA3 in this experiment. However, the differences in tumor binding among lA3, 2930 and 45-9 were minimal compared to the differences between all of the three specific MAbs and control MOPC. Immunoscintigraphy-single


Radioimaging experiments confirmed the ability of the three MAbs studied (lA3, 2930, 45-9) to localize and image (SW-39 tumors by external scintiscanning. Tumors of sufficient size (20.85 g) at the time of injection concentrated more radioactivity than any other area of the body so that tumors could be defined above background for each of these specific MAbs (Fig. 3). MOPC, on the other hand, did not define any tumors in these experiments. It is evident from the first radioimaging experiment, devised to compare tumor weights with tumor definition, that each of the three MAbs used alone was able to define all tumors greater than 0.85 g but no tumor smaller than 0.19 g (Fig. 4). Some tumors smaller than 0.19 g were visualized compared to the opposite control thigh, but they were not defined

combination studies

Improved imaging results were seen in the tumorbearing hamsters treated with MAb mixtures (Fig. 4). The combinations of lA3 + 2930 and lA3 + 45-9 resulted in 58% of the smaller tumors (0.19-0.50 g) being defined. This represented an increase from O-29% for the single MAbs to 43367% for the mixtures (O/6 for lA3; l/13 for 2930; 2/7 for 45-9; 8/12 for 45-9 + lA3; 3/7 for 2930 + lA3). This increase was statistically significant (P < 0.03, twotailed test using Pearson’s approximation to the normal). Furthermore, this improvement in scanning definition with mixtures of MAbs was not due simply to a larger injected dose. Increasing single MAb injections from 20 to 4Opg did not provide better tumor definition because background radioactivity increased at the same rate as tumor uptake. Further evidence that the MAb combinations provided better tumor localization than single MAbs is demonstrated in the %ID/g of tumor and the T/NT ratios from these imaging experiments (Table 4). Comparing the animals with tumors in the injection weight range of 0.3-1.3 g that were followed for a similar period after injection (9-10 days) showed that the two groups receiving MAb combinations had significantly (P < 0.01) higher %ID/g of tumor and higher T/NT ratios for most tissues. Since these two parameters (%ID/g tumor and T/NT ratios) were, at the doses used, independent of injected dose (Fenwick et al., 1989), the observed increases confirmed the better tumor localization of MAb combinations. Tumor size did not influence MAb uptake in these experiments as we have shown that MAb I A3 uptake (%ID/g tumor) was not influenced by tumor weight in this range (0.3-1.3 g) when tumors had been established for 5 days or more (Connett et al., 1989). The quality of tumor images, however, is influenced by tumor size since tumor visualization is dependent upon the total amount of [13’I]MAb localized in the tumor. Larger tumors are easier to visualize than smaller ones even with the same %ID/g. In the second scanning experiment (Fig. 5, Table 5) smaller tumors were used with higher MAb doses so that the experimental conditions approached functional antigen saturation (Fenwick et al., 1989). This also demonstrated the superiority of the lA3 + 45-9 mixture for tumor imaging [Fig. 5(d) and (h)] compared to MAbs used alone. Although the two specific MAbs produced better scans than MOPC, the best tumor images were with the mixture. This was clearly true for the unaltered scans [Fig. 5(a-d)], and even


JAMES W. FLESHMAN ct al. Table 4. Comparison of tumor localizatton with MAba used smelv and m combrnattont

MAb IA3 45-9 2930 IA3 + 45-9 IA3 + 2930

No. of animals

%lD$ g tumor

II 12 13 I9 20

0.93 + 0.01 *** 1.18 k O.l** 0.96 + 0. I *** 1.41 io.1 1.49+0.1




T:NT ratios: ~___. _~_~ ._ LlVcr

Blood I.1 *0.1*** 1.5 * 0.2 1.2 2 0. I l ** 1.7 +0.1 l.6kO.l

4.9 4.9 4.5 6.3 6.4

_+0.7’1 : 0.7’ _t 0.4*** & 0.3 + 0.3

Lung 2.3 * 2.6 i 2.4 + 3.5 i 4.0 +

0.3” 0.4’ 0. I *** 0.2 0.4

Kidney 4.1 4.6 4. I 5.5 5.8

+ i * + ?

0.4** 0.3** 0.3*** 0.2 0.3

*Significantly different than the mean for appropriate MAb mixtures using one-tailed Student’s l-test. P < 0.05; **p < 0.01; ***p < 0.001. tData derived from animals in imaging experiments (Fig. 4) whose tumors were harvested 9-10 days after Injection of MAb(s). Table includes data from 2 additmnal hamsters not scanned for technical reasons but counted IO days after treatment with “‘I-labeled 45-9 or “‘I-labeled 2930 in the same fashion as animals in Fig. 4. Calculated mea” tumor weights at injection included all tumors I” the range for 0.3 k 0.07 to I .3 & 0.2 g. Data for single MAbs injected with 20 or 4Opg were combined since T/NT ratms and % bound per g tumor were similar at both doses. iMeans 2 SEti.

more pronounced in the background subtracted images [Fig. 5(e-h)]. In agreement with the biodistribution data, the MAb mixture showed increased tumor accretion in the scans [Fig. S(d) and (h)] without the concomitant high background in the upper abdominal area demonstrated by single MAbs [Fig. SC-c), ([email protected] Lastly, the quantitative analyses of FOI in radioimages (Table 5) is further evidence of the superiority of MAb mixtures. The tumor FOI for IA3 + 45-9 had 45% more cts/pixel than either IA3 or 45-9 used alone at similar doses (300 pg of total IgG) and only in the mixture were the cts/pixel in the tumor higher than in the upper abdominal background FOI. When either background subtraction techniques (tumor FOI minus left leg FOI) or tumor weights (tumor index) were factored into the comparison, the mixtures showed significantly higher tumor binding indices than any of the MAbs used alone, confirming the biodistribution and scanning results. Discussion This study demonstrates that combinations of MAbs in vitro can increase specific binding to tumor cells, and, when administered in an in uivo model system, can result in better radioimaging of smaller

tumors and in better tumor localization when compared to the use of single component MAbs. The significant increases in total specific MAb bound in vitro (Table 1) and in u&o (Tables 4 and 5) to tumor cells with mixtures of MAbs suggest that antibody accretion within tumors can be increased, or at least maintained at higher levels for longer periods by using MAb mixtures. Increasing the amount of injected MAb within the range of 5-4Opg did not change the percent of injected dose bound per g of tumor. This result is consistent with other reports (Duewell et al., 1986; Fenwick et al., 1989; Sharkey et al., 1987; Wahl et al., 1983a) and suggests that these doses are below antigen saturation levels. The superior scanning qualities of the MAb mixtures were best demonstrated with small tumors, high doses of MAbs (300~0 and computer background subtraction of images. At these near optimal conditions, which approached functional antigen saturation in this model (Fenwick et al., 1989) it was not surprising that targeting two different populations of tumor antigen sites resulted in more tumor binding than targeting only one. Thus, shortly after injecting MAbs IA3 and 45-9, the images produced by the mixture were significantly better than those from either MAb used singly at twice the dose: that is, raising the dose of single MAbs from 150 to 300 pg

Table 5. Radioactive density (cts/pixel) in scintlscans of young GW-39 tumors treated with high doses of “’ I-labeled MAbst Fields of interest (FOI) Tumor MAb MOPC 45-9 IA3 IA3 + 45-9

IS.1 i: 17.6 + 16.7 & 25.2 +

I.6 4.8 2.9 1.8

Left leg Upper abdomen (cts/pixel) 3.8 4.3 2.6 5.2

_+ 1.1 + 0.4 _+ I .O + 1.0

21.7 + 18.9 + 21.0 rt 17.9F

0.4 0.2 2.1 1.8

(Tumor FOI - Left leg FOI) fupper abdomen FOI 0.53 + 0.70 * 0.67 k 1.12 f

0. I * 0.2** 0. I * 0.1

Tumor index (ctsipixellg) 27.8 42.8 41.7 65.8

f f f k

X3’** 4.7 5.9 3.6””

*Significantly different (P < 0.01) from IA3 + 45-9 mixture; **significantly different (P = 0.05) from IA3 + 45-9 mixture; ***significantly different (P < 0.05) from the other 3 groups: and ****significantly different (P < 0.005) from the other 3 groups. t1 I hamsters with 1.5 day old GW-39 tumors in the right thigh were injected ix. with “’ I-labeled MAb as follows: MOPC (3OOpg. 65 PCi; tumor wt at injection = 0.40 i 0.06g; n = 2); 45-9 (300 Pg, 85 FCi; tumor WI = 0.31 -+ 0.07 g; n = 3); IA3 (3OOflg. 9OpCi; tumor wt=0.34+0.05g; n =3); IA3 (ISOpg) plus 45-9 (150/(g) (total 92pCi; tumor wt = 0.31 f 0.06g; n = 3). 3 days later animals were scanned, the images digitized (64 x 64) and the radioactive density determined in fields of interest (FOI) over the tumor, a comparable area in the left control leg, and the upper abdominal background area. The weight of the tumor at injection was calculated from the harvest weight and tumor age by a computer program. Tumor index was calculated from [tumor (cts/pix) - left leg (cts/pix)] + tumor wt (g). AI1 values are means k SD.

Anti-colon cancer MAb mixtures in vivo

did not show the enhancement seen with the mixture at 300 gg total dose. The imaging experiments done with lower doses of MAbs also showed a superiority of mixtures although the effect was less pronounced. The mechanism(s) explaining the higher tumor levels of mixtures of MAbs compared to single MAbs at the lower doses of MAbs is more obscure than when near saturating doses were used. Possibly, at lower MAb doses, only the more accessible antigen binding sites in the GW-39 tumor become occupied. Mixtures of MAbs might show some additive effects in these subpopulations of tumor antigens when compared to MAbs used alone. Enhanced binding with mixtures in non-saturating conditions has been seen in other in uitro studies (Mujoo et al., 1991). Undoubtedly, the in uivo system is complex, with MAb clearance, tumor penetration, antibody avidity and tumor antigen and circulating antigen concentrations all potentially important factors. Based on previous studies (Primus et al., 1976), circulating CEA probably was not a factor in this system. The non-specific tumor accretion of MOPC found in this study, confirms our previous reports showing some non-specific uptake of foreign proteins by actively metabolizing tumor cells, but at significantly lower concentrations than with specific antibodies (Fenwick er al., 1989; Wahl et al., 1983b). Characterization of the antigen recognized by MAb IA3 is ongoing. In extensive immunofluorescence studies MAb lA3 has shown a high degree of selective colorectal cancer binding when compared to normal tissues, and further, MAb lA3 is much more specific for colon carcinoma than either of the antiCEA MAbs used in this study. The pharmacokinetics for MAb lA3 were very similar to the two anti-CEA MAbs with no significant differences in tumor accretion or clearance rates from normal tissues. The immunoscintigraphy experiments demonstrated the ability of MAb lA3 to localize in GW-39 tumors and provide good external imaging under conditions similar to the ones required by MAbs 2930 and 45-9 as shown here and previously (Wahl et al., 1983a,b). Additionally, each of the three specific MAbs demonstrated good tumor delineation in scintiscans when compared to non-specific MOPC. It is possible that MAbs 2930 and 45-9 may have been able to define slightly smaller tumors than MAb lA3, but this study was not conclusive in this regard. Considerably more animals would be needed to establish the significance of these small differences (Fig. 4). Immunofluorescence studies comparing binding of MAbs 2930, 45-9 and IA3 to GW-39 tumors showed slightly higher binding intensities for the two anti-CEA MAbs, reflecting the different levels of CEA and lA3 antigen expression on the GW-39 tumor cells (Timmcke et al., 1987). The lower expression of the lA3 antigen on GW-39 tumor cells when compared to CEA, is offset by the more specific cancer targeting properties of lA3. Combined with other selective MAbs, MAb lA3 could prove useful


in clinical detection (and perhaps therapy) of colon cancer. This model of a human colon tumor (GW-39) carried in non-immunosuppressed hamsters provides a convenient way to assay for potentially useful MAb mixtures. However the model also has drawbacks. The degree of antigenic heterogeneity in GW-39 is considerably less than that of many other human colon tumors (Connett et al., 1987). Thus, GW-39 cannot be used as the only target to predict optimum MAb mixtures. The considerable variability in histologic and cytologic patterns of TAA distribution that has been described strongly supports the need for testing MAb mixtures on individual tumor tissue to determine optimum “cocktails” for each patient. However, the encouraging results described in in viuo studies suggest that the use of MAbs in combination will enable the detection of smaller tumors, as well as offering the prospect of delivering more antibody to the target cells for therapeutic trials. Acknowledgements-Support from The Jewish Hospital of St Louis, Washington University Hybridoma Center, Mallinckrodt Inc. and NC1 Grant CA 44728 is appreciated. Presented in part at the meetings of the American Assoc. Cancer Research, Atlanta, 1987; San Francisco, 1989. The authors thank Dr D. M. Goldenberg for the GW-39 tumor, C. J. Mathias, Dr M. Welch and Dr B. Fischer for scanning assistance, Dr W. Connett for performing statistical analysis and computer programming, M. Baumann, S. Rapp and M. Ruiz for excellent technical assistance in producing and characterizing the MAbs and G. Redding-Dennison for diligence in manuscript preparation.

References Andrew S. A., Teh J. G., Johnstone R. W., Russell S. M., Whitehead R. H., McKenzie I. F. and Pietersz G. A. (1990) Tumor localization by combinations of monoclonal antibodies in a new human colon carcinoma cell line (LIM1899). Cancer Res. 50, 5225-5230. Bishof-Delaloye A., Delaloye B., Buchegger F., Gilgien W., Studer A., Churched S., Give1 J. C., Mosimann F., Pettavel J. and Mach J. P. (1989) Clinical value of immunoscintigraphy in colorectal carcinoma patients: a prospective study. J. Nucl. Med. 30, 16461656. Buraggi G. L., Gasparini M. and Seregni E. (1991) Immunoscintigraphy of colorectal carcinoma with an anti-CEA monoclonal antibody: a critical review. Nucl. Med. Biol. 18, 45-50. Chatal J. F., Saccavini J. C., Fumoleau P., Douillard J. Y., Curtet C., Kremer M., Mevel B. and Koprowski H. (1984) Immunoscintigraphy of colon carcinoma. J. Nucl. Med. 25, 307-314. Connett J. M., Fenwick J. R., Timmcke A. E. and Philpott G. W. (1987) Characterization of the binding properties of murine monoclonal antibody (MAb)lA3, a newly described antibody displaying anti-human colon cancer selectivity. Proc. Am. Assoc. Cancer Res. 28, 352. Connett J. M., Zhu X., G’Halloran L. R. and Philpott G. W. (1990) Improved tumor uptake with a mixture of I-125 monoclonal antibodies (MAbs) in a colon cancer model. J. Nucl. Med. 31, 852.‘ Connett J. M.. Mathias C. J.. Welch M. J.. Zhu X. and Philpott G. W. (1989) Increased uptake’ of a unique radiolabeled anticolon cancer monoclonal antibody (MAb lA3) in younger tumors in a human colon cancer




(GW-39) xenograft model. Proc. Am. Assoc. Cancer Res. 30, 359. Duewell S., Horst W. and Westera G. (1986) Uptake of a monoclonal antibody against CEA (Tumak 431/31) in a human colon tumor (Co-l 12) xenografted in the nude mouse. Cancer Immunol. Immunother. 23, 101-106. Durrant L. G., Robins R. A., Ballantyne K. C., Marksman R. A., Hardcastle J. D. and Baldwin R. W. (1989) Enhanced recognition of human colorectal tumour cells using combinations of monoclonal antibodies. Br. J. Cancer 60, 855-860. Epenetos A. A. and Kosmas C. (1989) Monoclonal antibodies for imaging and therapy. Br. J. Cancer 59, 152-155. Fenwick J. R., Philpott G. W. and Connett J. M. (1989) Biodistribution and histological localization of antihuman colon cancer monoclonal antibody (MAb)IA3: the influence of administered MAb dose on tumor uptake. Int. J. Cancer 44, 1017-1027. Goldenberg D. M., Goldenberg I., Sharkey R. M., Lee R. E., Higgenbotham-Ford E., Horowitz J. A., Hall T. C., Pinsky C. M. and Hansen H. J. (1989) Imaging of colorectal carcinoma with radiolabeled antibodies. Semin. Nucl. Med. 4, 262-281. Granowska M., Mather S. J., Britton K. E., Bentley S., Richman P.. Phillios R. K. S. and Northover J. M. A. of colorectal (1990) ,“,Tc radioimmunoscintigraphy cancer. Br. J. Cancer. (Suppl.) 62, 30-33. Halpern S. E. and Dillman R. 0. (1987) Problems associated with radioimmunodetection and possibilities for future solutions. J. Biol. Resp. Mod. 6, 235-262. Hart I. R. and Fidler I. J. (1981) The implications of tumor heterogeneity for studies on the biology and therapy of cancer metastasis. Biochem. Biophys. Acta 651, 37-50. Herlyn D., Powe J., Ross A. H., Herlyn M. and Koprowski H. (1985) Inhibition of human tumor growth by IgG,, monoclonal antibodies correlates with antibody density on tumor cells. J. Immunol. 134, 130&-1304. Larson S. M. (1990) Clinical radioimmunodetection 197881988: overview and suggestions for standardization of clinical trials. Cancer Res. (Suppl.) 50, 892s-898s. Lindmo T., Boven E., Cuttetta F., Federko V. and Bunn P. A. (1984) Determination of the immunoreactive fraction of radiolabeled monoclonal antibodies by linear extrapolation to binding at infinite antigen excess. J. Immunol. Meth. 72, 77-89.

Lockhart C. G., Stinson R. S., Margraf H. W., Parker C. W. and Philpott G. W. (1980) Production of anticarcinoembryonic antigen (CEA) antibody by somatic cell hybridization. Fedn Proc. 39, 3476. Marchalonis J. J., Cone R. E. and Santer V. (1971) Enzymic

cl a/.

iodination: a probe for accessible surface proteins of normal and neoplastic lymphocytes. Biochem. J. 124, 921-927.

Matzku S., Kirchgessner H., Schmid U., Temponi M. and Ferrone S. (1989) Melanoma targeting with a cocktail of monoclonal antibodies to distinct determinants of the human HMW-MAA. J. Nucl. Med. 30, 390-397. Mujoo K., Rosenblum M. G. and Murray J. L. (1991) Augmented binding of radiolabeled monoclonal antibodies to melanoma cells using specific antibody combinations. Cancer Res. 51, 2768-2772. Munz D. L.. Alavi A.. Koorowski H. and Herlvn D. (19861 Improved radioimmundimaging of human tumor ‘xeno: grafts by a mixture of monoclonal antibody F(ab’)? fragments. J. Nucl. Med. 27, 1739-1745. Olsson L., Sorensen H. R. and Behnke 0. (1984) Intratumoral phenotypic diversity of cloned human lung tumor cell lines and consequences for analyses with monoclonal antibodies. Cancer 54, 1757-1765. Primus J. F., Wang R. H., Cohen E., Hansen H. J. and Goldenberg D. M. (1976) Antibody to carcinoembryonic antigen in hamsters bearing GW-39 human tumors, Cancer Res. 36, 2 17&2 18 1.

Schlom J. and Weeks M. 0. (1985) Potential clinical ability of monoclonal antibodies in the management of human carcinomas. In Important Advances in Oncology (Edited by De Vita V. T., Hellman S. and Rosenberg S. A.), pp. 37-50. Lippincott, Philadelphia, Pa. Sharkey R. M., Primus F. J. and Goldenberg D. M. (1987) Antibody protein dose and radioimmunodetection of GW-39 human colon tumor xenografts. Inr. J. Cancer. 39, 611417.

Stramignoni D., Bowen R., Atkinson B. F. and Schlom J. (1983) Differential reactivity of monoclonal antibodies with human colon adenocarcinomas and adenomas. Int. J. Cancer 31, 543-552. Tagliabue E., Porro G., Barbanti P., Torre G. D., Menard S., Rilke F., Cerasoli S. and Colnaghi M. I. (1986) Improvement of tumor cell detection using a pool of monoclonal antibodies. Hybridoma 5, 107-l 15. Timmcke A. E., Connett J. M., Fleshman J. R., Neufeld D. M., Shemesh E. I. and Philpott G. W. (1987) Improved specific binding to human colon carcinoma using combinations of monoclonal antibodies (MAbs). Proc. Am. Assoc. Cancer Res. 28, 362.

Wahl R. L., Parker C. W. and Philpott G. W. (1983a) Improved radioimaging and tumor localization with monoclonal F(ab’),. J. Nucl. Med. 24, 316-325. Wahl R. L., Philpott G. W. and Parker C. W. (1983b) Monoclonal antibody radioimmunodetection of humanderived colon cancer. Invest. Radiol. 18, 58-62.

Tumor localization and radioimaging with mixtures of radioiodinated monoclonal antibodies directed to different colon cancer associated antigens.

After demonstrating enhanced tumor cell binding with a mixture of monoclonal antibodies (MAbs) in vitro, biodistribution and immunoscintigraphy studie...
2MB Sizes 0 Downloads 0 Views