New chelating agent for attaching indium-111 to monoclonal antibodies

248

Bioconjugate Chem. 1992, 3,248-255

New Chelating Agent for Attaching Indium-111 to Monoclonal Antibodies: In Vitro and in Vivo Evaluation R. Subramanian,* J. Colony, S. Shaban, H. Sidrak, M. V. Haspel, N. Pomato, M. G. Hanna, Jr., and R. P. McCabe Organon Teknika, Biotechnology Research Institute, 1330A Piccard Drive, Rockville, Maryland 20850. Received January 13, 1992

11lIn possesses excellent radiophysical properties suitable for use in immunoscintigraphy of cancerous tissues when attached to an antitumor antibody. However, lllIn has a tendency to accumulate in normal tissues such as liver. Instability of the linkage between "'In and antibody may contribute to this problem. To avoid this, we developed a new bifunctional chelating agent, 1,3-bis[N-[N-(2-aminoethyl)-2-aminoethyl]-2-aminoacetamidol-2- (4-i~o""cyanatobenzyl)propane-N,N,N',N~',N'~',N"",Nffrff,N""'-octaaceticacid (LiLo),that forms a kinetically stable chelate with metal ions such as indium. Using LiLo, indium-111 was conjugated to a human monoclonal antibody, 16.88. Competitive binding analysis revealed that the 16.88-LiLo conjugate is as immunoreactive as the unconjugated native antibody. This conjugate was compared with l"In-16.88, where diethylenetriaminepentaaceticacid dianhydride (DTPAa) was used as the chelating agent. In vitro stability studies showed that "'In was more stably bound to 16.88-LiLo than to 16.88-DTPA. Biodistribution studies in athymic mice bearing colorectal tumor xenografts indicated less liver retention with 16.88-LiLo than with 16.88-DTPA. These results demonstrate that LiLo is superior to DTPAa for attachment of "'In to the monoclonal antibodies.

A recently completed imaging study in colon cancer patients using l3lI-1abeled human monoclonal antibody (16.88) indicated a sensitivity in tumor detection of 79% for lesions >2 cm that express the relevant antigen (I). Encouraged by this result, we are developing improved methods of tumor imaging with human monoclonal antibodies to increase the sensitivity of detection of lesions 5000).At intervals an aliquot of the solution was removed and analyzed by thin-layer chromatography. lllIn not bound to 16.88 migrated as In-DTPA (Rf 0.7; See Table I), whereas indium bound to 16.88 remained at the origin. Measuremdnts were performed in triplicate. TLC plates were cut into equal portions and counted in a y counter as described by Meares et al. (15). Serum stability measurements were done as described by Deshpande et al. (9). Tissue Distribution Studies. Tumor retention studies were conducted in athymic mice bearing 0.8-1.5 cm diameter xenografts of the THO colon carcinoma prepared from enzymatically dissociated cells as described elsewhere (10,17).Six-to-eight-week old athymic male mice (Balb/ c, nu/nu) were injected in the tail vein with 50 pg (50 KCi) quantities of 16.88-DTPA-ll’In or 16.88-LiLo-lllIn. Animals were sacrificed at intervals from 1 h to 8 days after administration. Blood was collected and sera counted for lllIn at periodic intervals. Liver, kidney, spleen, intestine, thigh muscle, and femur were excised, weighed, and counted for ll1In. Results were expressed as a percentage of the injected dose per organ for dosimetry measurements and as a percentage of the injected dose per gram of tissue or per milliliter of serum for distribution studies. The counts were decay corrected. Whole-body clearance was monitored with a dose calibrator. Absorbed dose estimates were calculated using the MIRD formalism of the Society of Nuclear Medicine (18). Pharmacokinetic data were calculated using least squares regression analysis and a microcomputer program (RStrip, Micromath Scientific Software, Salt Lake City, UT). RESULTS

Synthesis. A diagram of the LiLo synthesis procedure is provided in Scheme I. The intermediates diester (I), diamide (11),diamine (111), and LiLo esters (IV-VI) were identified by spectral means. The structure of LiLo (VII) was confirmed by spectral methods such as FABMS, IR, and UV/vis. The chromatographic behavior of radiolabeled LiLo in methanol/ammonium acetate solvent system resembled the behavior of ethyl ester of LiLo in methanol/ ammonia solvent system. Silica gel TLC analysis showed that the greater the number of carboxyl groups per chelate, the less the chromatographic mobility of the chelate on silica gel (Table I). DTPA chelates with five carboxyl

252

Subramanian et ai.

Bloconjugate Chem., Voi. 3, No. 3, 1992

Table I. Chromatographic Behavior of Radiometal Chelates chelate l1IIn-EDTA l1IIn-DTPA 111In-LiLo "'In-16.88 I

I

~ " " " ' 1

t

t*

Rf

110

0.9 0.7

loo

0.6 0.0 '

'"""'/

"""'~

'

'""".I

I

"""'1

0,48

70t

0.4

-

60 0.32 0.24

0 ." " 08

:

10-1

100

101

102

.I

1

-

80

2

X 116 H l L l L o I n l l l l l R.OID.LlI"I1,

t 101.1

I

n m

a

0

I

I

I

I

4

6

8

10

Days

" C

m

, 2

Figure 3. Stability ~tudies-16.88-LiLo-~l~In( 0 ) vs 16.88DTPA-lllIn (+): stability studies were carried out in PBS buffer (pH 7.2,37 "C) with excess DTPA. Error bars represent the SD values.

[16.88] ug/ml Figure 1. Competitive binding assay: the immunoactivity of the antibody (16.88-LiLo, 0 )was estimated by comparison with native antibody (16.88, +), in its ability to compete with Iz5I16.88 for binding to the cognate antigen for 16.88 (CTA-1).

-m

1

0

3

6

9

12

15

I

Fraction # Figure 2. Radiochromatography: the radiolabeled immunoconjugate ( 0 )was purified by gel filtration chromatography. The labeled antibody came off the column in the first peak. (Radiochemical purity as determined by TLC analysis is >95%.)

groups moved slower (Rj 0.7) than EDTA chelates with four carboxyl groups (Rj 0.9). LiLo with eight carboxyl groups had an Rf value of 0.6 in this system. The radiolabeled antibody chelate conjugate (16.88-LiL0-~llIn) under our experimental conditions stayed at the origin (Rj 0-0.2). Conjugation and Labeling. The bifunctional chelating agent LiLo was easily coupled to HSA and monoclonal antibody 16.88. The immunoreactivity of the antibody was not affected by the presence of LiLo or the conjugation procedures used. Competitive binding studies demonstrated that 16.88-LiLo behaves similarly to 16.88 in its ability to bind to antigen CTA-1 (Figure 1). The immunoconjugate was labeled by incubating with lllIn in a buffer solution containing acetate and citrate. The radioimmunoconjugate was purified by gel filtration chromatography (Figure 2). Stability Studies. The stability of DTPA and LiLo conjugateswas determined in a chalienge experiment with a 5000molar excess of DTPA to conjugate. Under identical experimental conditions, 16.88-LiLo retained lllIn much

better than 16.88-DTPA (Figure 3). After 8 days of incubation 95% of "'In was bound to 16.88-LiLo, whereas only 80% of lllIn was bound to 16.88-DTPA. Biodistribution Studies. In vivo stability of the 16.88LiLo-ll'In and 16.88-DTPA-lllIn conjugates was studied in athymic mice bearing human colon tumor xenografts. Retention of "'In in various normal tissues as well as in the tumor xenograft was determined as a percentage of the injected dose per gram of tissue (Table 11). Stability of 16.88-LiL0-~llIn compared with the DTPA conjugate was most apparent in the liver, where significant differences between the two conjugates were seen from 4 h (p < 0.005) to 24 h after administration (p < 0.020). During the distribution phase up to 4 h following administration apparent differences were not significant (Figure 4A). A typical biphasic serum clearance curve was obtained with both l6.88-LiLo-ll1In and 16.88-DTPA-lllIn. Least squares regression analysis gave an initial distribution phase half-life of 39 min with an elimination phase halflife of 10.5 h for 16.88-LiL0-~~~In. The values for 16.88DTPA-lllIn (2.2 and 16.3h, respectively) suggested longer retention of 16.88-DTPA-lllIn in the circulation, but these differences were not significant and area under the curve determinations actually indicated longer retention of 16.88-LiLo-l111n (Figure 4B). Maximum blood concentration with a 50-pg dose of the LiLo-chelate conjugate was 13.3 wg/mL and was 12.3 pgImL for the DTPA conjugate. At 4 and 24 h following administration, when the liver values showed the greatest difference, no significant differences in lllIn retention in other normal tissues was apparent. Similarly, whole-body clearance of lllIn occurred at the same rate with both chelates (Figure 4C). These results were similar to those obtained with lZ5I-16.88(27). Somewhat more activity was seen in tumor xenografts of animals receiving 16.88-LiL0-~llIn ( m u = 3.4%) than in those from the 16.88-DTPA-lllIn group (max 2.9%) (Table 111, but the differences between the two groups were not significant. For both groups tumor-to-blood ratios were greater than unity from 2 days after administration, ranging from 2.4:l to 29:l for the LiLo group and 2.6:l to 16:l for the DTPA group. Tumor-to-liver ratios over the 8-day evaluating period averaged 0.30 f 0.06 for 16.88-LiL0-~~~Inand 0.14 f 0.05 for 16.88-DTPAll1In. Dosimetry calculations based on the murine data indicated that patients receiving 16.88-LiL0-~~~In would receive less radiation exposure to normal tissues than

Bioconjugate Chem., Vol. 3, No. 3, 1992 253

"'In-mAb for Cancer Imaging Using a New Chelator

x cl

A 15

\

w

u)

0 0

12

0

w

c V

W '3

6

z

U

w

I

I 0.16

I 1

I 2

1

1

4

0.i6

1

2

4

I 90W

-

v)

8 3 k

0

Y

70

50-

n

n 30-

-

10

I

I

I

I

0.16

1

2

4

I I I

0.16

1

2

1

I

4

6

DAYS POST ADMINISTRATION DAYS POST ADWINIS"RAT1ffl Figure 4. Tissue uptakes of 16.88-DTPA-11*In and 16.88-LiLo-1111n in nude mice: (A) liver, (B) serum, (C) whole body, (D) tumor. Human colon xenografts were developed in 6-8-week-old athymic Balb/c mice from enzymatically dissociated human tumor cells. Fifty micrograms of the labeled antibodies was injected into tail veins of the mice for biodistribution studies (n= 6-12). 0 ,DTPA; 0,LiLo. Table 11. Tissue Distribution of Xenograft (n = 6-12) ~~

lllIn

from 16.88-LiL0-~~~In or 16.88-DTPA-1111n in Nude Mice Bearing Human Colon Tumor

~

% injected dose per gram of tissue (*SD)

tissue serum

lh 34.7 (8.1) 12.9 (7.6) 3.5 (0.9) 7.4 (2.3) 4.1 (2.6) 2.7 (1.0)

2h 30.0 (9.2) 8.6 (3.6) 3.7 (0.9) 7.3 (3.6) 2.0 (0.83) 3.2

skeletal muscle

0.8

tumor

(0.67) 1.9 (0.5)

0.6 (0.36) 1.7 (0.27)

liver spleen kidney sm. intestine bone (femur)

(1.21)

16.88-LiL0-~~lIn 4h Id 2d 6.2 1.9 25.2 (11.5) (1.6) (0.4) 9.7 11.7 9.0 (3.1) (4.5) (3.6) 5.9 6.6 8.7 (3.7) (2.4) (2.1) 8.9 7.0 5.6 (3.3) (2.6) (1.7) 1.3 1.1 1.6 (0.51) (0.56) (0.41) 4.3 3.4 4.4 (1.10) (1.73) (1.71) 0.5 0.4 0.3 (0.17) (0.23) (0.21) 2.6 3.4 2.7 (0.29) (1.14) (0.43)

4d 0.3 (0.1)

6.6 (2.7) 8.3 (3.9) 3.5 (1.0) 0.5 (0.15) 2.1

(1.02) 0.2 (0.13) 2.3 (0.46)

patients receiving an equal dose of 16.88-DTPA-lllIn (Table 111). The difference was most apparent in the estimate of the liver dose, where the LiLo patients would receive slightly more than half the dose of the patients receiving 16.88-DTPA-l"In. DISCUSSION

DTPA dianhydride has been widely used to link radiometals such as lllIn to proteins such as human serum albumin and monoclonal antibodies. However, these coupling reactions are difficult to control because hydrol-

8d 0.08

(0.02) 5.1 (2.8) 6.6 (2.4) 2.0 (0.6) 0.2 (0.0) 1.2

(0.57) 0.1 (0.0)

1.4 (0.5)

lh 32.7 (2.1) 15.3 (0.8)

4.6 (1.9) 5.8 (0.56) 3.5 (0.42) 2.1 (0.21) 6.5 (0.14) 1.4 (0.28)

16.88-DTPA-ll'In 2h 4h Id 2d 28.0 18.6 4.6 1.3 (0.2) (8.6) (1.9) (0.7) 16.3 10.1 14.7 12.6 (3.9) (4.9) (5.3) (5.5) 4.2 4.6 5.7 6.1 (2.2) (0.7) (0.9) (1.5) 6.8 6.2 7.4 4.9 (0.57) (0.67) (1.89) (2.04) 2.7 1.9 1.3 1.0 (0.35) (0.85) (0.3) (0.35) 2.1 2.3 2.5 2.8 (0.42) (0.10) (0.5) (0.70) 0.4 1.4 0.2 0.2 (0.0) (0.06) (0.06) (0.06) 2.3 2.1 2.9 2.0 (0.3) (1.07) (0.2) (0.2)

4d 0.2 (0.06)

8d 0.07 (0.02) 8.0 5.4 (2.5) (3.0) 4.2 3.1 (0.8) (0.7) 4.1 2.6 (1.04) (0.78) 0.4 0.2 (0.10) (0.06) 1.1 1.0 (0.31) (0.15) 0.1 0.1 (0.06) (0.06) 1.7 0.7 (0.44) (0.22)

ysis of cyclic anhydride competes with the protein coupling reaction. In addition, because of the presence of two anhydride moieties, aggregate formation frequently occurs. Although DTPA dianhydride can be coupled to human IgM class antibodies, preparation and purification by gel filtration chromatography can result in considerable loss of radioimmunoconjugate. Halpern et al. reported that only 30 5% of ll'In-labeled human IgM was recovered when DTPA was used as the chelating agent (19). To avoid these losses and to obtain kinetically inert radioimmunoconjugates with a variety of radiometals, we

254

Subramanian et al.

Bloconjugate Chem., Vol. 3, No. 3, 1992

Table 111. Estimated Radiation Dose to Patients per or Millicurie of Administered 16.88-LiL0-~~~In 16.88-DTPA-1111n rads/mCi 16.88-LiL0J~~In 16.88-DTPA-lllIn tissue preclinical preclinical whole body blood spleen kidneys liver bone bone marrow

0.22 0.40

0.93 1.02 1.28 0.24 0.15

0.38 0.56 0.96 1.46 2.27 0.32 0.21

have prepared a new poly(amino carboxylate) chelator suitable for attachment to human monoclonal antibodies and other proteins. The synthesis of the bifunctional chelate involves the condensation between p-nitrobenzyl bromide and diethyl malonate. The diethyl @-nitrobenzy1)malonateobtained is converted to a diamine by reacting the ester with ammonia to give a diamide and then by reducing the diamide to a diamine. The diamine was coupled to DTPA andesterified to obtain LiLo ethyl ester. This compound was purified by silica gel column chromatography, reduced to an aminobenzyl derivative, and converted to a isothiocyanate derivative. This was further hydrolyzed in the presence of hydrochloric acid to obtain LiLo. The final step in the synthesis of LiLo involves the reaction between the diamine and DTPA anhydride. This reaction leads to the formation of LiLo ester and DTPA ester. Upon silica gel column chromatography using chloroform/methanol as the eluant, DTPA ester elutes first followed by LiLo ester. NMR, infrared, and mass spectral analyses of the product confirmed the structure of LiLo. As seen from the structure, LiLo has six nitrogen atoms and eight carboxyl groups suitable for binding radiometals (Chart I). Each arm can bind a metal ion, or alternately, both arms together can bind one or more metal ions. The nature of such binding may depend on the reaction conditions and the type of metal ions involved. Metal ions with very high coordination numbers (e.g., actinides, lanthanides, etc.) may bind easily to this molecule. We focused our efforts on binding of indium, a radiometal used in cancer imaging with radiolabeled antibodies. The results of the in vitro stability measurements demonstrated that 16.88-LiLo-lllIn was more stable than 16.88-DTPA-lllIn. The stability of "'In chelator was determined using G-50 gel filtration chromatography in normal human serum at 37 "C. I t was found that over a period of 7 days, less than 1%of the "'In was released from the chelate. In vitro stability studies have shown that LiLo may be suitable for attaching yttrium to 16.88 (data not shown). Current experiments evaluating the pharmacokinetic behavior of 16.88-LiL0-~~Y in animal model systems will be reported separately. Although the thermodynamic stability of a metal chelate is an important factor in determining the ease with which the chelation occurs,kinetic stability data are critical for in vivo applications of radiometal chelates. Kinetic stability of a radioimmunoconjugate will be influenced by several factors including the nature of the metal ion, the nature of the chelating agent, and the nature of the linkage between the chelator and the antibody, and, to some extent, the characteristics of the antibody. Earlier stability measurements demonstrated that chelators such as isothiocyanatobenzyl-EDTA or isothiocyanatobenzyl-DTPA attached to proteins are kinetically more stable than DTPA

cyclicanhydride linked proteins. Recent experiments have shown that benzyl-EDTA chelates attached to proteins are more stable in human serum than benzyl-DTPA chelate linked proteins (9). Similar results were also obtained by Carney et al. (20). There are several reasons for the instability of DTPA conjugates prepared from cyclic dianhydride. During conjugation of DTPA anhydride to protein one of the five carboxyl groups of DTPA is involved in coupling to an amine group of the protein, leaving four carboxyl groups available for binding to radiometals. Recent crystal structure studies have shown that In-DTPA complexation involves coordination between indium and five oxygens and three nitrogens of DTPA (21),necessitating the presence of at least five carboxyl groups for optimal chelation with indium. The p-nitrobenzyl group in the backbone carbon chain may further contribute to kinetic stability of LiLo-antibody conjugates. Meares and co-workers have shown that bulky substituents in the chelator molecule will reduce loss of lllIn from the conjugate to other metal-binding proteins in vivo (3,9). To reduce the uptake of radioactivity in normal organs, several approaches have been taken. A recent report suggested that the liver uptake of lllIn-labeled antibodies may depend on the method of purification (22). Another method to reduce the normal organ uptake is the use of cleavable linkers between the antibody and chelator. In an attempt to reduce the background activity, we conjugated a DTPA analogue to 16.88 through a diester linkage. Biodistribution studies in nude mice bearing THO human colon tumor xenografts indicated that diesterlinked conjugate was retained to a lesser extent in tumor and normal organs such as liver and kidney as compared to the peptide conjugate (23). LiLo contains an isothiocyanato group for coupling to antibody, ensuring that aggregation through antibody cross-linking is prevented. HPLC analysis using a molecular-sieve chromatography showed that, under similar experimental conditions, aggregate formation with LiLo was almost negligible (