Int. J. Cancer: 51,935-941 (1992) 0 1992 Wiley-Liss, Inc.
--I@@& -"&?F,
Publicatioi of the International Unlon Agans' Cancer Publicationde I Union InternationaleContre le Cancer
PHYSIOLOGICAL FACTORS INFLUENCING RADIOANTIBODY UPTAKE: A STUDY OF FOUR HUMAN COLONIC CARCINOMAS Rosalyn D. BLUMENTHAL, Robert M. SHARKEY, Rina KASHI.Ana M. NATAIF and David M. GOLOLNBERG~ Garden State Cancer Center and Center for Molecular Medicine and Immunology, 1 Brure Street, Newark NJ 07103, USA. We evaluated the accretion of '3'l-labeled NP-4 anticarcinoembryonic antigen (CEA) into 4 size-matched human colonic carcinomas grown S.C. in nude mice. Antibody uptake for LS I74T and GW-39 tumors was relatively high (I9 to 23% ID/g on day 3), whereas moderate uptake was seen in the Moser tumor (7.5% on day 3) and low uptake was detected in the GS-2 tumor (I.8% on day 3). Blood clearance of radioantibody was twice as fast in mice with GS-2 tumors than in mice with GW-39, LS174T or Moser tumors. Seven physiological parameters that might influence radioantibody accretion were evaluated in order t o better understand the differences in observed tumor targeting: vascular volume, blood flow rate, vascular permeability, tumor antigen content, serum antigen content and complexation of radioantibody, intratumoral antigen distribution, and intracellular antigen distribution. Although marked variability in vascular physiology, antigen content and antibody complexation of the 4 tumors grown in the same host and site existed, it was insufficient to explain the differences in antibody uptake. However, intra-tumoral distribution of antigen, and sub-cellular accessibility of antigen for radioantibody were important considerations. GS-2 tumors are well differentiated and have polarized cells. CEA in GS-2 is largely inaccessible t o radioantibody; most of the antigen is located in the lumen of the glands or on the apical surface of gland cells and most of the antibody distributes to the stromal region on the basolateral surface. The low antibody targeting in GS-2 could therefore be explained by restricted intra-tumor accessibility of antibody. Scatchard analysis of NP-4 binding t o Moser cells under non-internalizing and internalizing conditions revealed that 90% of the antigen is found within the cell, unavailable t o bind with the NP-4 antibody, which is slow to internalize. In contrast, CEA in LS I74T cells was almost entirely accessible. The reduced antibody targeting to Moser xenografts might therefore, be explained by restricted antibody accesribility at the cellular level. 1992 Wilty-I,iwInc. ,
Monoclonal antibodies (MAbs) developcd against tumorassociated antigcns and linked to drugs, toxins or radionuclides offers a unique approach for tumor-specific therapy. Somc initial success has been reported in experimental animal models (Aboud-Pirak et al., 1989; Marks et al., 1990) and in patients (Sears et al., 1985; Takahashi et al.>1988). The utility of new antibody conjugates is typically assessed in pre-clinical animal models. Although several animal models have been employed (eg., tumors transplanted into nude mice and rats and immunosuppressed micc, tumors grown in the hamster cheek pouch, and induced murinc tumors), the most common model is the nude mouse bcaring a S.C. human tumor (Gallagher, 1983). Since the site of implantation of a tumor is known to influence various properties of the tumor, including growth rate, invasiveness, metastatic behavior, vascularization, and degree of nccrosis (Kyriazis et al., 1978), we postulated that the same tumor implanted in various locations within a host would accrete radioantibody diffcrcntly. Upon testing this hypothcsis with the GW-39 human colonic carcinoma, we observed that thc magnitudc of tumor uptake of an anti-CEA MAb (NP-4) was in fact dependcnt on thc animal hosting the tumor (e.g., nude mouse or hamster) and on the site of implantation of thc tumor (s.c. liver, lung, muscle, cheek pouch; Blumenthal et al., 1989). Diffcrcnces in several tumor parameters that might influence radioantibody accretion were identified including vascular activity (blood flow rate, vascular volume, vascular permeability), total tumor antigen content,
and tumor size. The importancc of these 3 physiological parameters has bcen suggested by other investigators as well (Greiner et al., 1987; Sands et al., 1988). Since the animal host and the site of implantation can control tumor physiology and therefore also influence antibody accretion, we questioned how tumor physiology and antibody accretion would vary when different tumors were implantcd in the same site (s.c.) and in the same host (nude mouse). Using several S.C.nude-mouse xenografts, Buchsbaum et al. (1988) have recently demonstrated that tumor accretion and blood levcls o f an iodinated antibody could vary as much as 3-fold depending on the tumor. Thc rcason for this variability in antibody uptake was not analyzed; howcvcr, the tumor antigen content was suggested as an important factor. In this set of studies, we have demonstrated as much as a 7-fold difference in thc uptake of an anti-CEA MAb in 4 size-matched CEAproducing human colorectal tumors of varying histopathology (GW-39, LS174T, Moser, and GS-2). We have also evaluated several physiological parameters (vascular activity, serum antigen content and antibody complexation, tumor antigen content and intratumoral and intracellular antigen distribution) in thesc 4 xcnografts in an effort to determine which factor(s) is/are responsible for thc obscrved differences in MAb accretion. Our results highlight the importancc of antigen accessibility at thc intra-tumor and intra-cellular levels in tumors that have an adequate vascular supply and sufficient antigen in the total tumor for targeting. MATERIAL AND METHODS
Human tumorxenqrafi models Thc following serially propagated human colonic xenografts were used for these studics: GW-39, a signet-ring carcinoma; LS174T, a moderately differentiated adenocarcinoma; GS-2, a well-diffcrcntiatcd adenocarcinoma; and Moser, a poorly differentiated adenocarcinoma (from Dr. Chakeberty, Houston, TX). Tumors were passed through a 40-mesh screen, and rinsed with 0.9% sterilc NaCl to yield the desired cell suspension. Nude mouse S.C.tumors were initiatcd with 0.2 rnl of a 20% suspension into 6- to 8-week-old female nu/nu mice (Harlan Sprague Dawlcy, Indianapolis, IN). All studies were done on 0.4 to 0.7 g tumors to reduce thc possibility for any effect due to tumor size. CW-39 and TS174T tumors reach this size within 3 weeks of transplantation, whereas Moser and GS-2 tumors require 3 to 4 months. Radioantibody preparaiioti NP-4 anti-CEA MAb was purified from mousc ascites using Protein A and ion-exchange chromatography over S-Sepharose (Pharmacia, Piscataway, NJ). 131T-NP-4was prcparcd by the 'To whom correspondence and reprint requests should be addressed. Abbreviufions: BF, blood flow rate; CEA, carcinoembryonicantigen; ICH, immunocytochemistry; IDig, injected dose per gram; MAb. monoclonal antibody; MAR, microautoradiography; VP, vascular permeability; W,vascular volume.
Received: January-10,1992 and in revised form March 17,1992.
936
BLUMENTHAL ET AL.
chloramine-T mcthod as described by Blumenthal et al. (1989). Frce radioiodine was separated from antibody-bound iodine by passage over a PD-I0 column (Pharmacia) equilibrated with 0.04 M PBS (0.04 M phosphate, 0.15 M NaCI, 0.02% NaN3), p H 7.4, containing 1% human serum albumin. Routine quality assurance of radiolabcled antibody revealed no detectable aggregates, 2 to 4% free radioiodine by size exclusion HPLC using a Zorbax GF-250 (Dupont, Wilmington, D E ) column, and 75 to 95% of the radiolabeled anti-CEA antibody bound to the CEA-Affi-gel immunoadsorbent.
early as 15 to 30 minutes aftcr injcction of the antibody. Tumor-bearing micc were bled 1 hr or 24 hr after injection of 25 pCi of "'I-NP-4. Plasma was isolated and filtered, and lo5 CPM in 100 pl was injected. The running buffer was 0.1 M Na2HP04with 0.2% azidc, p H 6.5. Eighty tubes were collected and counted and the percentage of counts in the void region (Vo) and in the native region was dctermincd. Complexed antibody is expected to a have a shorter half-life than free antibody in blood.
Microazctoradiograph~~ (MAR) and immunocytochemistry (ICH) Tumors were removed 24 hr aftcr injcction of 100 pCi of lz51-NP-4 and h c d in cold acidiethanol. Serial 5-p sections were cut from paraffin-embedded sections and rehydrated. Alternate sections were used lor MAR and ICH. Slides for M A R were dipped into an aqueous (1:l) dilution of Kodak NTB-2 emulsion. Following 7-day exposure in light-tight boxes at 4"C, the slides were proccssed using Kodak Dektol developer and GBX h e r at 15 to 19°C. Sections for ICH were stained by sequential treatment with the primary antibody, biotinylated IgG horsc anti-mouse (1:200), avidin peroxidase Vascular measurements and 5 mg diaminobenzidine with 0.02% H202.All sections Vascular volume (VV) and vascular permeability (VP) were were counterstained with Harris' hematoxylin (Goldenbcrg ef quantitated using an in vivo labeling mcthod of rcd blood cells al., 1978). Regions rich with autoradiographic grains from 3 of (44). Animals were injccted i.v. with 2.5 ng stannous chloride thc tumors (GW-39, GS-2, and Moser) were photographed for (DupontiNEN Products, N. Billerica, MA), followed 30 min comparison of intra-tumor distribution of antibody, but d o not later by intravenous injection of 25 to SO pCi 99mTc04 reflect the amount of grains ( k ,the magnitude of antibody (Mallinckrodt, St. Louis, MO) and I-125-labeled irrelevant uptake) across whole-tumor sections. Similarly, parallel secisotype matched antibody (AFP-7-31 obtained from Immuno- tions for ICH were photographed to compare antigen location medics, Newark, NJ). Onc hour later the animals were with antibody distribution. but the amount of antigen in this anesthetized and blood was collected by intracardiac puncture. particular region may not reflect the amount in the section. The animals werc killed by cervical dislocation and the tissues were removed, weighed and counted with a gamma scintilla- Quantitation of membrane-associated and intra-cellular antigen Antigen content was measured by a direct antibody binding tion counter. VV and VP were calculated using the following assay under non-intcrnalizing and internalizing conditions. formulas: For the assay, cells were removed from culture flasks by gentle VV = ml bloodig tissue = (99mTc/gtissue)l(99mTc/gblood) scraping, rather than by trypsinization that might damage the cell mcmbranc; 5 x los cells were suspended in 100 p1 of PBS, VP = total plasma in tissue - intravascular plasma and the binding assay was run for 3 hr at 4°C (non-internalizing = [(I-l25/g tumor) - (I-125ig plasma)] conditions) or at 37°C (internalizing conditions). NP-4 does not internalize at 4°C and internalizes slowly over -5 hr at - [W *( 1-HCTI100)1. 37°C (data not shown). '.j11-NP-4 ( - 5 pCi/pg) was added in Blood flow (BF) was calculatcd by measuring 86Rbuptake 2 125 p1 (2.5 ng to 1 pg protein dosc range) ? 100 pg of min after injection. The tissues were collected, weighed and unlabeled NP-4 to the cell suspension. The cells were washed counted. The following formula was used to calculatc blood 2~ with PBS + 1% horse serum + 0.02% azide to remove flow: pl/g/min = pcrcent injected doseigram x total blood excess unbound antibody. Thc cell pellet was then counted in a volume. For all studies involving animals, care was provided gamma counter. The amount of specific binding was calculated and analyzed by the Scatchard method. Each point was done in according to institutional animal welfare guidelines. duplicate. The total number o l antigen molecules (N) was CEA measurements dctcrmined from the x-intercept and the affinity of antigen for C E A was cxtracted from tumors by homogenization in water NP-4 (K,) was calculated from the slope of the line relating (5:l v/w) with a Brinkman Polytron (Westbury, NY) for 15 scc bound antibody to bound/frec. Thcsc studies wcrc done with followed by sonicatictn for 15 scc on ice. Membrane material the IS174T and the Moser colonic lines only because neither was pelleted with centrifugation at 40$00 g for 30 min? and GW-39 nor GS-2 grow well in vitro. CEA content was mcasured in the supernatant. Heparinized blood samples were centrifuged to remove cells and CEA in Statistical analysis Results for vascular activity and antigen content are rethe plasma was measured. C E A quantitation was performed using a "sandwich"-type enzyme immunoassay involving 2 ported as the mean ? S.E. for a sample number of 17 to 44. MAbs against CEA. The niicrotiter plate was sensitized with Comparison of vascular function bctwccn thc 4 tumor xethc first MAb, sample o r standard was added and incubated nografts was perforrncd by a one-way analysis of variance for 90 rnin at 3TC, and thcn peroxidase-conjugated anti- (F-test). mouse antibody was added, followed by substrate prepared from o-phcnyldiamine. After 30 min, the reaction was stopped RESULTS with H2S04and the absorbance rcad at 490 nm. Each plate Antibody accretion contained 2 blank wells (buffer only) and 2 background wells Tumor accretion of 1311-labeledNP-4 anti-CEA IgG into 4 (all additions except antigen). size-matched colonic carcinomas is depicted in Figure 1 (left Complexation measurernrnts panel). Maximal accretion, as measurcd by thc %TD/g occurs The amount of MAb complexed by serum CEA was deter- 72 hr post injection in GW-39 (19.6%), Moser (7.5%) and mined by size-exclusion HPLC using a Zorbax GF-250 column. LS174T (22.9%) tumors, and 24 hr post injection for the GS-2 With this animal-antibody system, complexation occurs as (4.4%) tumor. Maximum blood concentrations were in the
Biodistnbution studie.7 When tumors were approximately 0.4 g in size, nudc micc were injected i.p. with 25 pCi l3II-NP-4 IgG (1.5 pg). Groups of 5 animals were killcd 1 , 3 , 7 and 14 days after radioantibody injection. The radioactivity in tumor and normal tissues was quantitated by gamma scintillation counting and recorded as the pcrcent injccted dose per gram (%ID/g). All data were corrected for physical decay of the isotope. Blood half-life was determined from this data by a pharmacokinetics program.
937
PHYSIOLOGICAL FACTORS INFLUENCING RADIOANTIBODY UPTAKE 30
~~
~
40
~
,
Blood
Tumor
GW-39
a
20
Q
::
0
o LS174T
pz
W GS-2
0 MOSER
u 8 c
0 W
'Z -
8
10
30 100
200
300
4
1
C
z? 0
I
Hours Post Radioantibody Injection
FIGURE1 - Differences in the magnitude and kinetics of radiolabeled NP-4 uptake in size-matched GW-39, LS174T, Moser and GS-2 tumors grown S.C. in the nude mouse (left panel) and differences in blood clearance of radioantibody in mice bearing each of 4 colorectal tumors (right panel). Results are expressed as the mean 2 S.D. for 5 to 10 tumors and for 5 blood samples.
range of 14 to 20% ID/g for mice bearing GW-39, LS174T, or Moser tumors, and the biological half-life of the radioantibody in blood ranged between 112 and 118 hr for these 3 tumors. The maximal blood levels of antibody in mice-bearing GS-2 tumors was 12% ID/g. The antibody half-life in blood when mice carried this tumor was only 57 hr. In an effort to explain the differences in tumor uptake of radioantibody seen in the 4 size-matched colonic carcinomas, we evaluated 7 physiological parameters: 3 tumor vascular parameters (VV, BF and VP); tumor and serum antigen content; serum complexation of antibody; antigen accessibility within the intact tumor; and accessibility of antigen within the tumor cells. Vascular measurements Figure 2a demonstrates that the B F to GW-39 tumors (32 pl/g/min) was about 2-fold higher than to LS174T, GS-2 and Moser tumors (12.7-16.8 pl/g/min; p < 0.001). In contrast, the tumor VV (Fig. 2b) was more variable between the 4 tumor types. GW-39 (17.8 pl/g) had the lowest W, followed by LS174T (27.3 pl/g), Moser (35.7 pl/g), and GS-2 (51.0 pl/g). These results indicate an absence of a relationship between VV and BF, suggesting the presence of ultrastructural differences in the vessels formed in each of the colonic tumors. VP to an IgG (Fig. 2c) was similar for GW-39, LS174T and Moser (15.1, 15.7, and 12.2 pl/g/hr respectively). The VP for GS-2 was about 50% less (6.7 pl/g/hr; p < 0.002 vs. LS174T; p < 0.001 vs. GW-39; p < 0.05 vs. Moser). Although there exists marked variability in the vascular physiology of different types of tumors grown in the same host and site, the vascular properties of a particular tumor do not correlate well with the observed accretion of antibody and therefore are insufficient to explain the observed difference in tumor accretion of MAb, and in particular the poor uptake seen in GS-2 and only moderate uptake seen in Moser tumors.
Tumor and serum antigen content We evaluated CEA content in size-matched tumors of all 4 cell lines and in the serum of animals bearing these tumors.
GW-39
LS174T
GS-2 MOSER
FIGURE2 - Radiotracer results of (a) BF, (b) W and (c) VP to an IgG in each of 4 colorectal xenografts. Results are expressed as the mean 2 S.E. for 6 to 24 tumors. TABLE I CEA CONTENT IN TUMOR AND SERUM OF NUDE MICE ~
BEARlNG S.C. COLONIC TUMORS Tumor
Tumor CEA (I.dg)'
GW-39 LS174T MOSER GS-2
30.91 rfr 2.65 24.29 2 3.46 59.01 2 6.59 140.80 t 20.63
(N)
]#I
SerumCEA (ngiml)'
(N)
10.76 t 1.10 3.87 t 0.94 156.34 rfr 29.09 143.68 t 25.57
i?:]
'Mean t S.E. The results are presented in Table I. GS-2, which accretes the least antibody, has the highest antigen content (mean ? S.E. of 140.80 ? 20.63 pg/g; N = 18) and Moser, which accretes only a moderate amount of radioantibody, has the second highest tumor antigen content (mean of 59.01 ? 6.59 pg/g; N = 23). GW-39 and LS174T tumors have 30.91 & 2.65 (N = 44) and 24.29 f 3.46 pg/g (N = 18) of CEA respectively. The 2 tumors with the highest antigen content also released the largest amount of CEA into the circulation. Moser- and GS-2-bearing mice had 156.34 ? 29.09 (N = 20) and 143.68 ? 25.57 (N = 18) ng/ml CEA respectively, while GW-39- and LS174T-bearing mice had 93 to 97% less antigen in their serum. The high serum antigen content in GS-2- and Moserbearing mice suggested that low tumor targeting with antibody may be due to complexation of antibody with serum antigen, resulting in a rapid clearance of the MAb. Complexation studies done either 1 to 2 hr or 24 hr after an injection of 12jI-NP-4 into tumor-bearing mice revealed that less than 3% of the total injected antibody was complexed in GW-39- and LS174T-bearing mice, and only 7 to 9% of the in GS-2- and Moser-bearing mice (Table 11). The low complexation obselved in all 4 lines is consistent with the 24-hr blood values presented in Figure 1. It is important to note that even though complexation is 9% for Moser-bearing mice, the blood clearance pattern of antibody is similar to mice bearing GW-39 or LS174T with lower serum CEA and lower Dercentages of
938
BLUMENTHAL ETAL.
TABLE I1 - PERCENT OF I-131-NP-4COMPLEXING WITH SERUM CEA IN NUDE MICE REARING COLONlC TUMORS
DISCUSSION
In order to improve the efficacy of antibody-conjugate Percent NP-4 complexed therapy, it is necessary to identify which tumors are most likely Tumor 1 hr 24 hr to be responsive to this modality of treatment and to identify the factors that regulate antibody targeting. Several parame1.7 t 0.3 0.5 t 0.1 GW-39 ters that may regulate antibody accretion have been identified, 2.7 t 0.2 1.1 t 0.0 LS174T including: factors associated with antibody availability, e.g., 10.8 t 2.8 7.1 t 2.7 GS-2 9.0 ? 3.4 7.8 2 2.9 MOSER antibody form, antibody affinity, protein dose, complexation with circulating antigen, and complexation with a human Results represent the average of 2 to 6 mice. anti-murine antibody; factors associated with antigen availability, eg., antigen density, antigen heterogeneity, intra-tumor complexed antibody. In general, unless serum antigen levels and intra-cellular location of antigen; and factors associated become very high ( > 150 ng/ml), NP-4 does not complex with with the tumor target, eg., size and site of the tumor, and circulating antigen (Sharkey et al., 1990) because the amount tumor vascular parameters (Blumenthal et al., 1990). Studies of antibody ( 1.5 Kg) far exceeds the amount of circulating in tumor-bearing animals have attempted to elucidate the antigen. It seems unlikely, therefore, that complexation of relative importance of each of these immunological and radioantibody with serum antigen can explain the differences physiological factors and to devise methods to overcome the in tumor targetability. limitations imposed by any one of these factors. Although vascular activity was an important parameter Antibody distribution and intra-tumoral antigen location when we evaluated how well a radioantibody targeted a single A quantitative assessment of antigen content does not provide an indication of how much of the antigen is accessible tumor cell line when the host and/or site of tumor growth was to bind antibody. The intra-tumoral location of CEA was varied (Blumenthal et al., 1989), it was not sufficient to explain evaluated by direct ICH and the distribution of 1251-labeled differences in targeting of several tumor cell lines grown in the NP-4 was assessed in a parallel section of the same tumor by same host and site. Since the animal host and the site of tumor MAR in GW-39, GS-2, and Moser xenografts 1-day after growth were held constant, the source of variability in BF, VV injection of NP-4 (Fig. 3). CEA is distributed throughout and VP between the 4 tumor lines used in these studies must GW-39 and Moser tumors, and the NP-4 antibody is predomi- come from the tumor itself and the interaction of tumor and nantly restricted to the perimeter of the GW-39 tumor and host as the tumor establishes itself. The process of neovascularalong the perimeter and in several distinct foci in Moser ization at the site of the tumor is controlled in part by factors tumors. Regions of overlapping antigen and antibody can be secreted by the tumor (Blood and Zetter, 1990). The most visualized in both of these tumors. In GS-2 tumors, CEA is specific of these mitogens includes vascular endothelial growth predominantly localized in the lumen of glands and intra- factor, vascular permeability factor, and angiogenin. Other cellularly along the apical surface of gland cells. Most of the host factors such as platelet-derived endothelial cell growth antibody is seen in the stromal region along the basolateral factor and angiotropin also exist. In addition, the host can surface of the glands. Therefore, although GS-2 has large produce substances that inhibit the migration and proliferaamounts of CEA, most of the antigen is inaccessible in this tion of endothelial cells, such as thrombospondin and platelet factor IV. The molecular mechanisms responsible for the tumor. production of these tumor factors and for the complex interacPhysiological barriers affecting antigen accessibility may occur not only at the level of the whole tumor, but also at the tion between tumor and host to control release of non-tumor level of the cell, i.e., intracellular vs. membrane-associated and factors remain unknown. The fact that vascular activity differs extracellular antigen. We evaluated binding of a specific from tumor to tumor (Fig. 2; Sands et al., 1988), suggests that antibody (NP-4) to Moser and LS174T cells in non-permeable the amount of these factors may also differ. Serum antigen levels and antibody complexation are imporand permeable conditions in vitro. Table I11 summarizes the results of these antibody-binding studies. The amount of tant in regulating the distribution of administered antibody bound CEA in LS174T is the same in non-permeable (8.0 X lo4 (Beatty et al., 1990). Complexation of circulating antibody with sites/cell) and permeable conditions (8.9 x lo7 sites/cell), i.e., shed antigen can affect the kinetics of antibody clearance from all antigen is accessible. The affinity of the MAb for the the blood and the availability of free antibody for the tumor. antigen was 1.3 to 2.0 x lo8M/1. The amount of CEA in Moser Serum taken from a patient with a high antigen level inhibited under non-permeable conditions is 6.0 x lo4 sites/cell (20% the in vitro binding of antibody to tumor cells expressing the less than LS174T) and is 10-fold higher under permeable antigen (Nadler et al., 1980). Increased levels of circulating conditions (5.6 x 105 sites/cell). The affinity constant in non- antigen also increased the amount of antibody needed to permeable and permeable conditions alike was 9.0 X lo7 to achieve tumor penetration (Meeker et al., 1985). Experiments 1.5 x lo8 M/l. Additional evidence suggesting that the large with tumor-bearing mice have demonstrated that the greater amount of antigen made by Moser cells is not accessible comes the amount of CEA in the serum, the lower the uptake of from in vitro immunoassay results of the media taken from iodinated antibody in tumor and liver (Hagan et al., 1986). cultures containing equal numbers of Moser or LS174T cells. However, the anti-CEA MAb used in these studies (NP-4) was Media from Moser cultures contain 1372 ng/ml of CEA, shown in clinical trials not to complex readily with circulating whereas media from LS174T cultures contain 6720 ng/ml CEA, unless serum levels are very high (Sharkey et al., 1990). (results not shown). This experimental design assumes that In these studies, complexation was not observed in plasma sub-cellular antigen distribution in cultured cells reflects taken 1 hr after NP-4 injection, however, we did not sample antigen distribution in cells grown in an intact tumor, which earlier time points (5 to 15 min). Since antibody clearance in may not be true. We chose to perform this analysis in vitro Moser-bearing mice is similar to that observed in GW-39- or because we felt that physical and chemical disruption of a LS174T-bearing mice with lower percentages of antibody tumor xenograft might alter the antigen distribution of the complexation, it appears that antibody complexation may not cells derived from the tumor. Overall, these results demon- be a significant influence on Moser tumor accretion of antistrate the importance of identifying both the intra-tumoral body. However, there is still a possibility that an additive effect distribution and the sub-cellular location of antigen in addition of the low level of MAb complexation with continuously to providing an overall quantitation of total tumor antigen released antigen into the circulation does influence MAb availability. content.
-
PHYSIOLOGICAL FACTORS INFLUENCING RADIOANTIBODY UPTAKE
NP-4 MICROAUTORADIOGRAPHY
939
CEA IMMUNOCYTOCHEMISTRY
RCURE 3 - MAR of a 100-pCidose of Iz5I-NP-4IgG 24 hr after injection (left panel) and ICH of CEA (right panel) in serial sections of GW-39, GS-2 and Moser tumors. Arrows point to areas rich in deposited antibody (left) or antigen (right). Bar in upper left panel, 60 p.
ICH is a well-established technique for identifying the location of antigen within various tissues (Goldenberg et al., 1978). Coupled with MAR, it becomes a powerful tool for determining the accessibility of antigen for administered antibody. Poor antigen accessibility, as seen in the GS-2 tumor, has also been observed in another antigen-antibody model (Pervez et aZ., 1989). Pervez’s study demonstrated that the tumor cells in HRA-19, a well-differentiated adenocarcinoma of human large-bowel origin, form polarized acini and the luminal antigenic sites are inaccessible. This observation raises
the question of whether it is possible to alter the polarity that separates the apical located antigen and basolateral distributed antibody. Several approaches have been used successfully in vitro to alter epithelial cell polarity and to cause redistribution of surface macromolecules (Ziomek et aZ., 1980; Pisam and Ripoche, 1976). However, the application of these methods (e.g., chelating agents, protamine, hypertonic solutions) have not been investigated and may not be practical for an in vivo tumor model unless the polarity of tumor cells could be specifically altered.
940
BLUMENTHAL ETAL. TABLE 111-BINDING PARAMETERS OF NP-4
IrG FOR CEA IN MOSER AND LS174T CELLS LS 174T
Moser
Number of sitedcell Binding affinity Ka (M/1)
Internalizing
Noninternalizing
Internalizing
6.0 x 104
5.6 x 105
7.9 x 104
8.95 x 104
1.5 x 10*
8.9 x 107
2.04
1.33 x lo8
Noninternalizing
Our results highlight the importance of analyzing intratumoral and sub-cellular distribution of antigen, rather than simply presenting results of total extractable antigen. This finding has ramifications for 2 current lines of investigation: relating antigen content to antibody accretion, and assessing the effect of therapy on antigen expression. Modification of antigen expression in tumors has been an area of considerable interest, due to the potential that such an approach might have at increasing antibody localization. Several methods to regulate CEA levels have been reported, including the use of interferon (Guadagni et aZ., 1990), the use of a site-selective cyclic AMP analogue (Guadagni et al., 1991), and exposure of tumor cells to radiation (Hareyama et al., 1991). How these treatments affect the intra-tumoral and sub-cellular distribution of antigen remains to be elucidated. Others have been interested in the expression of antigen following therapy to determine whether tumor escape was the result of an absence of sufficient antigen and to establish the potential advantage of multiple doses of radioantibody. Esteban et al. (1991) have demonstrated a decrease in CEA content in resurgent LS174T xenografts 6 weeks after treatment with Y-90-anti-CEA MAb. Irrelevant antibody therapy also resulted in a partial reduction in CEA. After 3 passages, the CEA levels in reappearing tumors had increased but had not returned to control levels. In contrast, immunohistologic studies by Schlom et al. (1990)
X
lo8
found the same antigenic phenotype in tumors that escaped antibody therapy as in untreated tumors. We have observed no change or small increases in total antigen expression following radioimmunotherapy of GW-39 tumors, although the distribution of a mucinous antigen (CSAp) shifted from being partially extra-cellular to being almost exclusively extracellular (Blumenthal et al., 1991). It is important to determine the distribution of antigen in addition to assessing the total quantity of antigen, since this could affect the ability to retarget tumor with a second dose of antibody. Although antibodies need to be evaluated clinically to be certain of their targeting ability, animal models can be a useful pre-clinical screen for assessing tumor targeting in vivo and for providing additional information for in vitro selection criteria. However, the results presented here, along with information from our previous studies (Blurnenthal et al., 1989), highlight the importance of evaluating various animal models before establishing the potential usefulness of a particular antibody conjugate. ACKNOWLEDGEMENTS This work was supported in part by USPHS grant CA39841 from the National Institutes of Health.
REFERENCES
ABOUD-PIRAK, E., HURWITZ, E., BELLOT,F., SCHLESSINGER, J. and SELA, M., Inhibition of human tumor growth in nude mice by a conjugate of doxorubicin with monoclonal antibodies to epidermal growth factor receptor. Proc. nat Acad. Sci. (Wash.), 86, 3778-3781
GALLAGHER, B.M., Monoclonal antibodies: the design of appropriate carrier and evaluation systems. In: R.M. Lambrecht and W.C. Eckelman (eds.), Animal models and radiotracer design, 61-105, Springer
(1989).
GOLDENBERG, D.M., SHARKEY, R.M. and PRIMUS, F.J., Immunocytochemical detection of carcinoembryonic antigen in conventional histopathologyspecimens. Cancer, 42,1546-1553 (1978). GREINER, J.W., GUADAGNI, F., NOGUCHI, P., PESTKA, S., COLCHER, D., FISHER,P.B. and SCHLOM, J., Recombinant interferon enhances monoclonal-antibody targeting of carcinoma lesions in vivo. Science,
BEATTY, J.D., BEATTY, B.G., O'CONNER-TRESSEL, M., Do, T. and PAXTON,R.J., Mechanisms of tissue uptake and metabolism of radiolabeled antibody. Role of antigen-antibody complex formation. CancerRes., 50,840s-845s (1990).
BLOOD, C.H. and ZETTER,B.R., Tumor interactions with the vasculature: angiogenesisand tumor metastasis. Biochem. biophys. Acfa, 1032, 89-118 (1990).
BLUMENTHAL, R.D., SHARKEY, R.M. and GOLDENBERG, D.M., Current perspectives and challenges in the use of monoclonal antibodies as imaging and therapeutic agents. Advanc. Drug Deliv. Rev., 4, 279-318 (1990).
BLUMENTHAL, R., SHARKEY, R.M., KASHI,R. and GOLDENBERG, D.M., Changes in GW-39 tumor physiology following a sub-optimal dose of radioimmunotherapy. Z'roc. Amer. Cancer Assoc., 32, 260 (abstract 1548) (1991).
(1983).
235,895-898 (1987).
GUADAGNI, F., TOKTORA, G., ROSELLI, M., CLAIR,T., CHO-CHUNG, Y.S., SCHLOM, J. and GREINER, J.W., Carcinoembryonic antigen regulation in human colorectal tumor cells by a site-selective cyclic AMP analogue: a comparison with interferon-gamma.Znt. J. Cancer, 48,413-422 (1991).
GUADAGNI, F., Win, P.L., ROBBINS, P.F., SCHLOM, J. and GREINER, J.W., Regulation of carcinoembryonic antigen in different human colorectal tumor cells by interferon. Cancer Res., 50,6248-6255 (1990). HAGAN,P.L., HALPERN, S.E., DILLMAN, R.O., SHAWLER, D.L., JOHNSON, D.E., CHEN,A., KRISHNAN, L., FRINCKE, J., BARTHOLOMEW, R.M., DAVID,G.S. and CARLO, D., Tumor size: effect on monoclonal antibody uptake in tumor mode1s.J. nucl. Med., 27,422-427 (1986). HAREYAMA, M., IMAI,K., KUBO,K., SHIDOU, M., OOUCHI, A., YACHI, A. and MORITA, K., Effect of radiation on the expressionof carcinoembryonic antigen of human gastric adenocarcinoma cells. Cancer, 67,
BLUMENTHAL, R.D., SHARKEY, R.M., KASHI,R., NATALE, A.M. and GOLDENBERG, D.M., Influence of animal host and tumor implantation site on radioantibody uptake in the GW-39 human colonic cancer xenograft. Znt. J. Cancer, 44,1041-1047 (1989). BUCHSBAUM, D., LLOYD,R., JUNI,J., WOLLNER, I., BRUBAKER, P., J., BURNS,F., STEPLEWSKI, Z., COLCHER, D., HANNA,D., SPICKER, SCHLOM, J., BUCHEGGER, F. and MACH,J.P., Localization and imaging 2269-2274 (1991). of radiolabeled monoclonal antibodies against colorectal carcinoma in KYRIAZIS, A.P., DIPERSIO, L., MICHAEL, G.J., PESCE,A.J. and STINtumor-bearing nude mice. Cancer Res., 48,4324-4333 (1988). NETT, J.D., Growth patterns and metastatic behavior of human tumors ESTEBAN, J.M., KUHN,J.A., FELDER,B., WONG,J.Y.C., BATTIFORA,growing in athymic mice. CancerRes., 38,3186-3190 (1978). H., BEATTY, J.D., WANEK, P.M. and SHIVELY, J.E., Carcinoembryonic D., BJORN,M.J., LEI, M. and BAUMAL, R., antigen expression of resurgent human colon carcinoma after treat- MARKS,A,, ETTENSON, ment with therapeutic doses of gOY-carcinoembryonicantigen mono- Inhibition of human tumor growth by intraperitoneal immunotoxins in nude mice. Cancer Res., 50,288-292 (1990). clonal antibody. Cancer Rex, 51,3802-3806 (1991).
PHYSIOLOGICAL FACTORS INFLUENCING RADIOANTIBODY UPTAKE
MEEKER,T.C., LOWDER,J., MALONEY, D.G., MILLER,R.A., THIELMAN,K., WARNKE,R. and LEVY,R., A clinical trial of anti-idiotype therapy for B-cell malignancy. Blood, 62,988-995 (1985). NADLER,L.M., STAHENKO, P., HARDY,R., KAPLAN,W.D., BUTTON, L.N., KUFE,D.W., ANTMAN, K.H. and SCHLOSSMAN, S.F., Serotherapy of a patient with a monoclonal antibody directed against a human lymphoma-associated antigen. Cancer Res., 40,41474154 (1980). PERVEZ,S., KIRKLAND, S.C., EPENETOS, A.A., MOOI,W.J., EVANS, D.J. and KRAUSZ,T., Effect of polarity and differentiation on antibody localization in multicellular tumor spheroid and xenograft models and its potential importance for in vivo imrnunotargeting. Int. J. Cancer, 44, 940-947 (1989). PISAM,M. and RIPOCHE, P., Redistribution of surface macromolecules in dissociated epithelial cells. J. Cell Biol., 71,907-920 (1976). SANDS, H., JONES,P.L. and SHAH,L., Correlation of vascular permeability and blood flow with monoclonal antibody uptake by human Clousner and renal-cell xenografts. Cancer Res., 48,188-193 (1988). SCHLOM, J., MOLINOLA, A,, SIMPSON, J.F., SILER,K., ROSELLI, M., HINKLE,G., HOUCHENS, D.P. and COLCHER,D., Advantage of dose
941
fractionation in monoclonal antibody-targeted radioimmunotherapy. J. nut. Cancer Inst., 82,763-771 (1990). SEARS,H.F., HERLYN,D., STEPLEWSKI, Z . and KOPROWSKI, H., Phase-I1 clinical trial of a murine monoclonal antibody cytotoxic for gastrointestinal carcinoma. Cancer Res., 45,5910-5913 (1985). SHARKEY, R.M., GOLIIENBERG, D.M., GOLDENBERG, H., LEE, R.E., BALLANCE, C., PAWLYK, D., VARGA,D. and HANSEN,H.J., Murine monoclonal antibodies against carcinoembryonic antigen: immunological, pharmacokinetic and targeting properties in humans. Cancer Res., 50,2823-2831 (1990). TAKAHASHI, T., YAMAGUCHI, T., KITAMURA,K., SUZUYAMA, H., HONDA,M., YOKOTA,T., KOTANAGI, H., TAKAHASHI, M. and HASHIMOTO,Y., Clinical application of monoclonal-antibody-drug conjugates for immunotargeting chemotherapy of colorectal carcinoma. Cancer, 61,881-888 (1988). ZIOMEK,C.A., SCHULMAN, S. and EDIDIN,M., Redistribution of membrane proteins in isolated mouse intestinal epithelial cells. J. Cell B i d , 86,849-857 (1980).
--I@@& -"&?F,
Publicatioi of the International Unlon Agans' Cancer Publicationde I Union InternationaleContre le Cancer
PHYSIOLOGICAL FACTORS INFLUENCING RADIOANTIBODY UPTAKE: A STUDY OF FOUR HUMAN COLONIC CARCINOMAS Rosalyn D. BLUMENTHAL, Robert M. SHARKEY, Rina KASHI.Ana M. NATAIF and David M. GOLOLNBERG~ Garden State Cancer Center and Center for Molecular Medicine and Immunology, 1 Brure Street, Newark NJ 07103, USA. We evaluated the accretion of '3'l-labeled NP-4 anticarcinoembryonic antigen (CEA) into 4 size-matched human colonic carcinomas grown S.C. in nude mice. Antibody uptake for LS I74T and GW-39 tumors was relatively high (I9 to 23% ID/g on day 3), whereas moderate uptake was seen in the Moser tumor (7.5% on day 3) and low uptake was detected in the GS-2 tumor (I.8% on day 3). Blood clearance of radioantibody was twice as fast in mice with GS-2 tumors than in mice with GW-39, LS174T or Moser tumors. Seven physiological parameters that might influence radioantibody accretion were evaluated in order t o better understand the differences in observed tumor targeting: vascular volume, blood flow rate, vascular permeability, tumor antigen content, serum antigen content and complexation of radioantibody, intratumoral antigen distribution, and intracellular antigen distribution. Although marked variability in vascular physiology, antigen content and antibody complexation of the 4 tumors grown in the same host and site existed, it was insufficient to explain the differences in antibody uptake. However, intra-tumoral distribution of antigen, and sub-cellular accessibility of antigen for radioantibody were important considerations. GS-2 tumors are well differentiated and have polarized cells. CEA in GS-2 is largely inaccessible t o radioantibody; most of the antigen is located in the lumen of the glands or on the apical surface of gland cells and most of the antibody distributes to the stromal region on the basolateral surface. The low antibody targeting in GS-2 could therefore be explained by restricted intra-tumor accessibility of antibody. Scatchard analysis of NP-4 binding t o Moser cells under non-internalizing and internalizing conditions revealed that 90% of the antigen is found within the cell, unavailable t o bind with the NP-4 antibody, which is slow to internalize. In contrast, CEA in LS I74T cells was almost entirely accessible. The reduced antibody targeting to Moser xenografts might therefore, be explained by restricted antibody accesribility at the cellular level. 1992 Wilty-I,iwInc. ,
Monoclonal antibodies (MAbs) developcd against tumorassociated antigcns and linked to drugs, toxins or radionuclides offers a unique approach for tumor-specific therapy. Somc initial success has been reported in experimental animal models (Aboud-Pirak et al., 1989; Marks et al., 1990) and in patients (Sears et al., 1985; Takahashi et al.>1988). The utility of new antibody conjugates is typically assessed in pre-clinical animal models. Although several animal models have been employed (eg., tumors transplanted into nude mice and rats and immunosuppressed micc, tumors grown in the hamster cheek pouch, and induced murinc tumors), the most common model is the nude mouse bcaring a S.C. human tumor (Gallagher, 1983). Since the site of implantation of a tumor is known to influence various properties of the tumor, including growth rate, invasiveness, metastatic behavior, vascularization, and degree of nccrosis (Kyriazis et al., 1978), we postulated that the same tumor implanted in various locations within a host would accrete radioantibody diffcrcntly. Upon testing this hypothcsis with the GW-39 human colonic carcinoma, we observed that thc magnitudc of tumor uptake of an anti-CEA MAb (NP-4) was in fact dependcnt on thc animal hosting the tumor (e.g., nude mouse or hamster) and on the site of implantation of thc tumor (s.c. liver, lung, muscle, cheek pouch; Blumenthal et al., 1989). Diffcrcnces in several tumor parameters that might influence radioantibody accretion were identified including vascular activity (blood flow rate, vascular volume, vascular permeability), total tumor antigen content,
and tumor size. The importancc of these 3 physiological parameters has bcen suggested by other investigators as well (Greiner et al., 1987; Sands et al., 1988). Since the animal host and the site of implantation can control tumor physiology and therefore also influence antibody accretion, we questioned how tumor physiology and antibody accretion would vary when different tumors were implantcd in the same site (s.c.) and in the same host (nude mouse). Using several S.C.nude-mouse xenografts, Buchsbaum et al. (1988) have recently demonstrated that tumor accretion and blood levcls o f an iodinated antibody could vary as much as 3-fold depending on the tumor. Thc rcason for this variability in antibody uptake was not analyzed; howcvcr, the tumor antigen content was suggested as an important factor. In this set of studies, we have demonstrated as much as a 7-fold difference in thc uptake of an anti-CEA MAb in 4 size-matched CEAproducing human colorectal tumors of varying histopathology (GW-39, LS174T, Moser, and GS-2). We have also evaluated several physiological parameters (vascular activity, serum antigen content and antibody complexation, tumor antigen content and intratumoral and intracellular antigen distribution) in thesc 4 xcnografts in an effort to determine which factor(s) is/are responsible for thc obscrved differences in MAb accretion. Our results highlight the importancc of antigen accessibility at thc intra-tumor and intra-cellular levels in tumors that have an adequate vascular supply and sufficient antigen in the total tumor for targeting. MATERIAL AND METHODS
Human tumorxenqrafi models Thc following serially propagated human colonic xenografts were used for these studics: GW-39, a signet-ring carcinoma; LS174T, a moderately differentiated adenocarcinoma; GS-2, a well-diffcrcntiatcd adenocarcinoma; and Moser, a poorly differentiated adenocarcinoma (from Dr. Chakeberty, Houston, TX). Tumors were passed through a 40-mesh screen, and rinsed with 0.9% sterilc NaCl to yield the desired cell suspension. Nude mouse S.C.tumors were initiatcd with 0.2 rnl of a 20% suspension into 6- to 8-week-old female nu/nu mice (Harlan Sprague Dawlcy, Indianapolis, IN). All studies were done on 0.4 to 0.7 g tumors to reduce thc possibility for any effect due to tumor size. CW-39 and TS174T tumors reach this size within 3 weeks of transplantation, whereas Moser and GS-2 tumors require 3 to 4 months. Radioantibody preparaiioti NP-4 anti-CEA MAb was purified from mousc ascites using Protein A and ion-exchange chromatography over S-Sepharose (Pharmacia, Piscataway, NJ). 131T-NP-4was prcparcd by the 'To whom correspondence and reprint requests should be addressed. Abbreviufions: BF, blood flow rate; CEA, carcinoembryonicantigen; ICH, immunocytochemistry; IDig, injected dose per gram; MAb. monoclonal antibody; MAR, microautoradiography; VP, vascular permeability; W,vascular volume.
Received: January-10,1992 and in revised form March 17,1992.
936
BLUMENTHAL ET AL.
chloramine-T mcthod as described by Blumenthal et al. (1989). Frce radioiodine was separated from antibody-bound iodine by passage over a PD-I0 column (Pharmacia) equilibrated with 0.04 M PBS (0.04 M phosphate, 0.15 M NaCI, 0.02% NaN3), p H 7.4, containing 1% human serum albumin. Routine quality assurance of radiolabcled antibody revealed no detectable aggregates, 2 to 4% free radioiodine by size exclusion HPLC using a Zorbax GF-250 (Dupont, Wilmington, D E ) column, and 75 to 95% of the radiolabeled anti-CEA antibody bound to the CEA-Affi-gel immunoadsorbent.
early as 15 to 30 minutes aftcr injcction of the antibody. Tumor-bearing micc were bled 1 hr or 24 hr after injection of 25 pCi of "'I-NP-4. Plasma was isolated and filtered, and lo5 CPM in 100 pl was injected. The running buffer was 0.1 M Na2HP04with 0.2% azidc, p H 6.5. Eighty tubes were collected and counted and the percentage of counts in the void region (Vo) and in the native region was dctermincd. Complexed antibody is expected to a have a shorter half-life than free antibody in blood.
Microazctoradiograph~~ (MAR) and immunocytochemistry (ICH) Tumors were removed 24 hr aftcr injcction of 100 pCi of lz51-NP-4 and h c d in cold acidiethanol. Serial 5-p sections were cut from paraffin-embedded sections and rehydrated. Alternate sections were used lor MAR and ICH. Slides for M A R were dipped into an aqueous (1:l) dilution of Kodak NTB-2 emulsion. Following 7-day exposure in light-tight boxes at 4"C, the slides were proccssed using Kodak Dektol developer and GBX h e r at 15 to 19°C. Sections for ICH were stained by sequential treatment with the primary antibody, biotinylated IgG horsc anti-mouse (1:200), avidin peroxidase Vascular measurements and 5 mg diaminobenzidine with 0.02% H202.All sections Vascular volume (VV) and vascular permeability (VP) were were counterstained with Harris' hematoxylin (Goldenbcrg ef quantitated using an in vivo labeling mcthod of rcd blood cells al., 1978). Regions rich with autoradiographic grains from 3 of (44). Animals were injccted i.v. with 2.5 ng stannous chloride thc tumors (GW-39, GS-2, and Moser) were photographed for (DupontiNEN Products, N. Billerica, MA), followed 30 min comparison of intra-tumor distribution of antibody, but d o not later by intravenous injection of 25 to SO pCi 99mTc04 reflect the amount of grains ( k ,the magnitude of antibody (Mallinckrodt, St. Louis, MO) and I-125-labeled irrelevant uptake) across whole-tumor sections. Similarly, parallel secisotype matched antibody (AFP-7-31 obtained from Immuno- tions for ICH were photographed to compare antigen location medics, Newark, NJ). Onc hour later the animals were with antibody distribution. but the amount of antigen in this anesthetized and blood was collected by intracardiac puncture. particular region may not reflect the amount in the section. The animals werc killed by cervical dislocation and the tissues were removed, weighed and counted with a gamma scintilla- Quantitation of membrane-associated and intra-cellular antigen Antigen content was measured by a direct antibody binding tion counter. VV and VP were calculated using the following assay under non-intcrnalizing and internalizing conditions. formulas: For the assay, cells were removed from culture flasks by gentle VV = ml bloodig tissue = (99mTc/gtissue)l(99mTc/gblood) scraping, rather than by trypsinization that might damage the cell mcmbranc; 5 x los cells were suspended in 100 p1 of PBS, VP = total plasma in tissue - intravascular plasma and the binding assay was run for 3 hr at 4°C (non-internalizing = [(I-l25/g tumor) - (I-125ig plasma)] conditions) or at 37°C (internalizing conditions). NP-4 does not internalize at 4°C and internalizes slowly over -5 hr at - [W *( 1-HCTI100)1. 37°C (data not shown). '.j11-NP-4 ( - 5 pCi/pg) was added in Blood flow (BF) was calculatcd by measuring 86Rbuptake 2 125 p1 (2.5 ng to 1 pg protein dosc range) ? 100 pg of min after injection. The tissues were collected, weighed and unlabeled NP-4 to the cell suspension. The cells were washed counted. The following formula was used to calculatc blood 2~ with PBS + 1% horse serum + 0.02% azide to remove flow: pl/g/min = pcrcent injected doseigram x total blood excess unbound antibody. Thc cell pellet was then counted in a volume. For all studies involving animals, care was provided gamma counter. The amount of specific binding was calculated and analyzed by the Scatchard method. Each point was done in according to institutional animal welfare guidelines. duplicate. The total number o l antigen molecules (N) was CEA measurements dctcrmined from the x-intercept and the affinity of antigen for C E A was cxtracted from tumors by homogenization in water NP-4 (K,) was calculated from the slope of the line relating (5:l v/w) with a Brinkman Polytron (Westbury, NY) for 15 scc bound antibody to bound/frec. Thcsc studies wcrc done with followed by sonicatictn for 15 scc on ice. Membrane material the IS174T and the Moser colonic lines only because neither was pelleted with centrifugation at 40$00 g for 30 min? and GW-39 nor GS-2 grow well in vitro. CEA content was mcasured in the supernatant. Heparinized blood samples were centrifuged to remove cells and CEA in Statistical analysis Results for vascular activity and antigen content are rethe plasma was measured. C E A quantitation was performed using a "sandwich"-type enzyme immunoassay involving 2 ported as the mean ? S.E. for a sample number of 17 to 44. MAbs against CEA. The niicrotiter plate was sensitized with Comparison of vascular function bctwccn thc 4 tumor xethc first MAb, sample o r standard was added and incubated nografts was perforrncd by a one-way analysis of variance for 90 rnin at 3TC, and thcn peroxidase-conjugated anti- (F-test). mouse antibody was added, followed by substrate prepared from o-phcnyldiamine. After 30 min, the reaction was stopped RESULTS with H2S04and the absorbance rcad at 490 nm. Each plate Antibody accretion contained 2 blank wells (buffer only) and 2 background wells Tumor accretion of 1311-labeledNP-4 anti-CEA IgG into 4 (all additions except antigen). size-matched colonic carcinomas is depicted in Figure 1 (left Complexation measurernrnts panel). Maximal accretion, as measurcd by thc %TD/g occurs The amount of MAb complexed by serum CEA was deter- 72 hr post injection in GW-39 (19.6%), Moser (7.5%) and mined by size-exclusion HPLC using a Zorbax GF-250 column. LS174T (22.9%) tumors, and 24 hr post injection for the GS-2 With this animal-antibody system, complexation occurs as (4.4%) tumor. Maximum blood concentrations were in the
Biodistnbution studie.7 When tumors were approximately 0.4 g in size, nudc micc were injected i.p. with 25 pCi l3II-NP-4 IgG (1.5 pg). Groups of 5 animals were killcd 1 , 3 , 7 and 14 days after radioantibody injection. The radioactivity in tumor and normal tissues was quantitated by gamma scintillation counting and recorded as the pcrcent injccted dose per gram (%ID/g). All data were corrected for physical decay of the isotope. Blood half-life was determined from this data by a pharmacokinetics program.
937
PHYSIOLOGICAL FACTORS INFLUENCING RADIOANTIBODY UPTAKE 30
~~
~
40
~
,
Blood
Tumor
GW-39
a
20
Q
::
0
o LS174T
pz
W GS-2
0 MOSER
u 8 c
0 W
'Z -
8
10
30 100
200
300
4
1
C
z? 0
I
Hours Post Radioantibody Injection
FIGURE1 - Differences in the magnitude and kinetics of radiolabeled NP-4 uptake in size-matched GW-39, LS174T, Moser and GS-2 tumors grown S.C. in the nude mouse (left panel) and differences in blood clearance of radioantibody in mice bearing each of 4 colorectal tumors (right panel). Results are expressed as the mean 2 S.D. for 5 to 10 tumors and for 5 blood samples.
range of 14 to 20% ID/g for mice bearing GW-39, LS174T, or Moser tumors, and the biological half-life of the radioantibody in blood ranged between 112 and 118 hr for these 3 tumors. The maximal blood levels of antibody in mice-bearing GS-2 tumors was 12% ID/g. The antibody half-life in blood when mice carried this tumor was only 57 hr. In an effort to explain the differences in tumor uptake of radioantibody seen in the 4 size-matched colonic carcinomas, we evaluated 7 physiological parameters: 3 tumor vascular parameters (VV, BF and VP); tumor and serum antigen content; serum complexation of antibody; antigen accessibility within the intact tumor; and accessibility of antigen within the tumor cells. Vascular measurements Figure 2a demonstrates that the B F to GW-39 tumors (32 pl/g/min) was about 2-fold higher than to LS174T, GS-2 and Moser tumors (12.7-16.8 pl/g/min; p < 0.001). In contrast, the tumor VV (Fig. 2b) was more variable between the 4 tumor types. GW-39 (17.8 pl/g) had the lowest W, followed by LS174T (27.3 pl/g), Moser (35.7 pl/g), and GS-2 (51.0 pl/g). These results indicate an absence of a relationship between VV and BF, suggesting the presence of ultrastructural differences in the vessels formed in each of the colonic tumors. VP to an IgG (Fig. 2c) was similar for GW-39, LS174T and Moser (15.1, 15.7, and 12.2 pl/g/hr respectively). The VP for GS-2 was about 50% less (6.7 pl/g/hr; p < 0.002 vs. LS174T; p < 0.001 vs. GW-39; p < 0.05 vs. Moser). Although there exists marked variability in the vascular physiology of different types of tumors grown in the same host and site, the vascular properties of a particular tumor do not correlate well with the observed accretion of antibody and therefore are insufficient to explain the observed difference in tumor accretion of MAb, and in particular the poor uptake seen in GS-2 and only moderate uptake seen in Moser tumors.
Tumor and serum antigen content We evaluated CEA content in size-matched tumors of all 4 cell lines and in the serum of animals bearing these tumors.
GW-39
LS174T
GS-2 MOSER
FIGURE2 - Radiotracer results of (a) BF, (b) W and (c) VP to an IgG in each of 4 colorectal xenografts. Results are expressed as the mean 2 S.E. for 6 to 24 tumors. TABLE I CEA CONTENT IN TUMOR AND SERUM OF NUDE MICE ~
BEARlNG S.C. COLONIC TUMORS Tumor
Tumor CEA (I.dg)'
GW-39 LS174T MOSER GS-2
30.91 rfr 2.65 24.29 2 3.46 59.01 2 6.59 140.80 t 20.63
(N)
]#I
SerumCEA (ngiml)'
(N)
10.76 t 1.10 3.87 t 0.94 156.34 rfr 29.09 143.68 t 25.57
i?:]
'Mean t S.E. The results are presented in Table I. GS-2, which accretes the least antibody, has the highest antigen content (mean ? S.E. of 140.80 ? 20.63 pg/g; N = 18) and Moser, which accretes only a moderate amount of radioantibody, has the second highest tumor antigen content (mean of 59.01 ? 6.59 pg/g; N = 23). GW-39 and LS174T tumors have 30.91 & 2.65 (N = 44) and 24.29 f 3.46 pg/g (N = 18) of CEA respectively. The 2 tumors with the highest antigen content also released the largest amount of CEA into the circulation. Moser- and GS-2-bearing mice had 156.34 ? 29.09 (N = 20) and 143.68 ? 25.57 (N = 18) ng/ml CEA respectively, while GW-39- and LS174T-bearing mice had 93 to 97% less antigen in their serum. The high serum antigen content in GS-2- and Moserbearing mice suggested that low tumor targeting with antibody may be due to complexation of antibody with serum antigen, resulting in a rapid clearance of the MAb. Complexation studies done either 1 to 2 hr or 24 hr after an injection of 12jI-NP-4 into tumor-bearing mice revealed that less than 3% of the total injected antibody was complexed in GW-39- and LS174T-bearing mice, and only 7 to 9% of the in GS-2- and Moser-bearing mice (Table 11). The low complexation obselved in all 4 lines is consistent with the 24-hr blood values presented in Figure 1. It is important to note that even though complexation is 9% for Moser-bearing mice, the blood clearance pattern of antibody is similar to mice bearing GW-39 or LS174T with lower serum CEA and lower Dercentages of
938
BLUMENTHAL ETAL.
TABLE I1 - PERCENT OF I-131-NP-4COMPLEXING WITH SERUM CEA IN NUDE MICE REARING COLONlC TUMORS
DISCUSSION
In order to improve the efficacy of antibody-conjugate Percent NP-4 complexed therapy, it is necessary to identify which tumors are most likely Tumor 1 hr 24 hr to be responsive to this modality of treatment and to identify the factors that regulate antibody targeting. Several parame1.7 t 0.3 0.5 t 0.1 GW-39 ters that may regulate antibody accretion have been identified, 2.7 t 0.2 1.1 t 0.0 LS174T including: factors associated with antibody availability, e.g., 10.8 t 2.8 7.1 t 2.7 GS-2 9.0 ? 3.4 7.8 2 2.9 MOSER antibody form, antibody affinity, protein dose, complexation with circulating antigen, and complexation with a human Results represent the average of 2 to 6 mice. anti-murine antibody; factors associated with antigen availability, eg., antigen density, antigen heterogeneity, intra-tumor complexed antibody. In general, unless serum antigen levels and intra-cellular location of antigen; and factors associated become very high ( > 150 ng/ml), NP-4 does not complex with with the tumor target, eg., size and site of the tumor, and circulating antigen (Sharkey et al., 1990) because the amount tumor vascular parameters (Blumenthal et al., 1990). Studies of antibody ( 1.5 Kg) far exceeds the amount of circulating in tumor-bearing animals have attempted to elucidate the antigen. It seems unlikely, therefore, that complexation of relative importance of each of these immunological and radioantibody with serum antigen can explain the differences physiological factors and to devise methods to overcome the in tumor targetability. limitations imposed by any one of these factors. Although vascular activity was an important parameter Antibody distribution and intra-tumoral antigen location when we evaluated how well a radioantibody targeted a single A quantitative assessment of antigen content does not provide an indication of how much of the antigen is accessible tumor cell line when the host and/or site of tumor growth was to bind antibody. The intra-tumoral location of CEA was varied (Blumenthal et al., 1989), it was not sufficient to explain evaluated by direct ICH and the distribution of 1251-labeled differences in targeting of several tumor cell lines grown in the NP-4 was assessed in a parallel section of the same tumor by same host and site. Since the animal host and the site of tumor MAR in GW-39, GS-2, and Moser xenografts 1-day after growth were held constant, the source of variability in BF, VV injection of NP-4 (Fig. 3). CEA is distributed throughout and VP between the 4 tumor lines used in these studies must GW-39 and Moser tumors, and the NP-4 antibody is predomi- come from the tumor itself and the interaction of tumor and nantly restricted to the perimeter of the GW-39 tumor and host as the tumor establishes itself. The process of neovascularalong the perimeter and in several distinct foci in Moser ization at the site of the tumor is controlled in part by factors tumors. Regions of overlapping antigen and antibody can be secreted by the tumor (Blood and Zetter, 1990). The most visualized in both of these tumors. In GS-2 tumors, CEA is specific of these mitogens includes vascular endothelial growth predominantly localized in the lumen of glands and intra- factor, vascular permeability factor, and angiogenin. Other cellularly along the apical surface of gland cells. Most of the host factors such as platelet-derived endothelial cell growth antibody is seen in the stromal region along the basolateral factor and angiotropin also exist. In addition, the host can surface of the glands. Therefore, although GS-2 has large produce substances that inhibit the migration and proliferaamounts of CEA, most of the antigen is inaccessible in this tion of endothelial cells, such as thrombospondin and platelet factor IV. The molecular mechanisms responsible for the tumor. production of these tumor factors and for the complex interacPhysiological barriers affecting antigen accessibility may occur not only at the level of the whole tumor, but also at the tion between tumor and host to control release of non-tumor level of the cell, i.e., intracellular vs. membrane-associated and factors remain unknown. The fact that vascular activity differs extracellular antigen. We evaluated binding of a specific from tumor to tumor (Fig. 2; Sands et al., 1988), suggests that antibody (NP-4) to Moser and LS174T cells in non-permeable the amount of these factors may also differ. Serum antigen levels and antibody complexation are imporand permeable conditions in vitro. Table I11 summarizes the results of these antibody-binding studies. The amount of tant in regulating the distribution of administered antibody bound CEA in LS174T is the same in non-permeable (8.0 X lo4 (Beatty et al., 1990). Complexation of circulating antibody with sites/cell) and permeable conditions (8.9 x lo7 sites/cell), i.e., shed antigen can affect the kinetics of antibody clearance from all antigen is accessible. The affinity of the MAb for the the blood and the availability of free antibody for the tumor. antigen was 1.3 to 2.0 x lo8M/1. The amount of CEA in Moser Serum taken from a patient with a high antigen level inhibited under non-permeable conditions is 6.0 x lo4 sites/cell (20% the in vitro binding of antibody to tumor cells expressing the less than LS174T) and is 10-fold higher under permeable antigen (Nadler et al., 1980). Increased levels of circulating conditions (5.6 x 105 sites/cell). The affinity constant in non- antigen also increased the amount of antibody needed to permeable and permeable conditions alike was 9.0 X lo7 to achieve tumor penetration (Meeker et al., 1985). Experiments 1.5 x lo8 M/l. Additional evidence suggesting that the large with tumor-bearing mice have demonstrated that the greater amount of antigen made by Moser cells is not accessible comes the amount of CEA in the serum, the lower the uptake of from in vitro immunoassay results of the media taken from iodinated antibody in tumor and liver (Hagan et al., 1986). cultures containing equal numbers of Moser or LS174T cells. However, the anti-CEA MAb used in these studies (NP-4) was Media from Moser cultures contain 1372 ng/ml of CEA, shown in clinical trials not to complex readily with circulating whereas media from LS174T cultures contain 6720 ng/ml CEA, unless serum levels are very high (Sharkey et al., 1990). (results not shown). This experimental design assumes that In these studies, complexation was not observed in plasma sub-cellular antigen distribution in cultured cells reflects taken 1 hr after NP-4 injection, however, we did not sample antigen distribution in cells grown in an intact tumor, which earlier time points (5 to 15 min). Since antibody clearance in may not be true. We chose to perform this analysis in vitro Moser-bearing mice is similar to that observed in GW-39- or because we felt that physical and chemical disruption of a LS174T-bearing mice with lower percentages of antibody tumor xenograft might alter the antigen distribution of the complexation, it appears that antibody complexation may not cells derived from the tumor. Overall, these results demon- be a significant influence on Moser tumor accretion of antistrate the importance of identifying both the intra-tumoral body. However, there is still a possibility that an additive effect distribution and the sub-cellular location of antigen in addition of the low level of MAb complexation with continuously to providing an overall quantitation of total tumor antigen released antigen into the circulation does influence MAb availability. content.
-
PHYSIOLOGICAL FACTORS INFLUENCING RADIOANTIBODY UPTAKE
NP-4 MICROAUTORADIOGRAPHY
939
CEA IMMUNOCYTOCHEMISTRY
RCURE 3 - MAR of a 100-pCidose of Iz5I-NP-4IgG 24 hr after injection (left panel) and ICH of CEA (right panel) in serial sections of GW-39, GS-2 and Moser tumors. Arrows point to areas rich in deposited antibody (left) or antigen (right). Bar in upper left panel, 60 p.
ICH is a well-established technique for identifying the location of antigen within various tissues (Goldenberg et al., 1978). Coupled with MAR, it becomes a powerful tool for determining the accessibility of antigen for administered antibody. Poor antigen accessibility, as seen in the GS-2 tumor, has also been observed in another antigen-antibody model (Pervez et aZ., 1989). Pervez’s study demonstrated that the tumor cells in HRA-19, a well-differentiated adenocarcinoma of human large-bowel origin, form polarized acini and the luminal antigenic sites are inaccessible. This observation raises
the question of whether it is possible to alter the polarity that separates the apical located antigen and basolateral distributed antibody. Several approaches have been used successfully in vitro to alter epithelial cell polarity and to cause redistribution of surface macromolecules (Ziomek et aZ., 1980; Pisam and Ripoche, 1976). However, the application of these methods (e.g., chelating agents, protamine, hypertonic solutions) have not been investigated and may not be practical for an in vivo tumor model unless the polarity of tumor cells could be specifically altered.
940
BLUMENTHAL ETAL. TABLE 111-BINDING PARAMETERS OF NP-4
IrG FOR CEA IN MOSER AND LS174T CELLS LS 174T
Moser
Number of sitedcell Binding affinity Ka (M/1)
Internalizing
Noninternalizing
Internalizing
6.0 x 104
5.6 x 105
7.9 x 104
8.95 x 104
1.5 x 10*
8.9 x 107
2.04
1.33 x lo8
Noninternalizing
Our results highlight the importance of analyzing intratumoral and sub-cellular distribution of antigen, rather than simply presenting results of total extractable antigen. This finding has ramifications for 2 current lines of investigation: relating antigen content to antibody accretion, and assessing the effect of therapy on antigen expression. Modification of antigen expression in tumors has been an area of considerable interest, due to the potential that such an approach might have at increasing antibody localization. Several methods to regulate CEA levels have been reported, including the use of interferon (Guadagni et aZ., 1990), the use of a site-selective cyclic AMP analogue (Guadagni et al., 1991), and exposure of tumor cells to radiation (Hareyama et al., 1991). How these treatments affect the intra-tumoral and sub-cellular distribution of antigen remains to be elucidated. Others have been interested in the expression of antigen following therapy to determine whether tumor escape was the result of an absence of sufficient antigen and to establish the potential advantage of multiple doses of radioantibody. Esteban et al. (1991) have demonstrated a decrease in CEA content in resurgent LS174T xenografts 6 weeks after treatment with Y-90-anti-CEA MAb. Irrelevant antibody therapy also resulted in a partial reduction in CEA. After 3 passages, the CEA levels in reappearing tumors had increased but had not returned to control levels. In contrast, immunohistologic studies by Schlom et al. (1990)
X
lo8
found the same antigenic phenotype in tumors that escaped antibody therapy as in untreated tumors. We have observed no change or small increases in total antigen expression following radioimmunotherapy of GW-39 tumors, although the distribution of a mucinous antigen (CSAp) shifted from being partially extra-cellular to being almost exclusively extracellular (Blumenthal et al., 1991). It is important to determine the distribution of antigen in addition to assessing the total quantity of antigen, since this could affect the ability to retarget tumor with a second dose of antibody. Although antibodies need to be evaluated clinically to be certain of their targeting ability, animal models can be a useful pre-clinical screen for assessing tumor targeting in vivo and for providing additional information for in vitro selection criteria. However, the results presented here, along with information from our previous studies (Blurnenthal et al., 1989), highlight the importance of evaluating various animal models before establishing the potential usefulness of a particular antibody conjugate. ACKNOWLEDGEMENTS This work was supported in part by USPHS grant CA39841 from the National Institutes of Health.
REFERENCES
ABOUD-PIRAK, E., HURWITZ, E., BELLOT,F., SCHLESSINGER, J. and SELA, M., Inhibition of human tumor growth in nude mice by a conjugate of doxorubicin with monoclonal antibodies to epidermal growth factor receptor. Proc. nat Acad. Sci. (Wash.), 86, 3778-3781
GALLAGHER, B.M., Monoclonal antibodies: the design of appropriate carrier and evaluation systems. In: R.M. Lambrecht and W.C. Eckelman (eds.), Animal models and radiotracer design, 61-105, Springer
(1989).
GOLDENBERG, D.M., SHARKEY, R.M. and PRIMUS, F.J., Immunocytochemical detection of carcinoembryonic antigen in conventional histopathologyspecimens. Cancer, 42,1546-1553 (1978). GREINER, J.W., GUADAGNI, F., NOGUCHI, P., PESTKA, S., COLCHER, D., FISHER,P.B. and SCHLOM, J., Recombinant interferon enhances monoclonal-antibody targeting of carcinoma lesions in vivo. Science,
BEATTY, J.D., BEATTY, B.G., O'CONNER-TRESSEL, M., Do, T. and PAXTON,R.J., Mechanisms of tissue uptake and metabolism of radiolabeled antibody. Role of antigen-antibody complex formation. CancerRes., 50,840s-845s (1990).
BLOOD, C.H. and ZETTER,B.R., Tumor interactions with the vasculature: angiogenesisand tumor metastasis. Biochem. biophys. Acfa, 1032, 89-118 (1990).
BLUMENTHAL, R.D., SHARKEY, R.M. and GOLDENBERG, D.M., Current perspectives and challenges in the use of monoclonal antibodies as imaging and therapeutic agents. Advanc. Drug Deliv. Rev., 4, 279-318 (1990).
BLUMENTHAL, R., SHARKEY, R.M., KASHI,R. and GOLDENBERG, D.M., Changes in GW-39 tumor physiology following a sub-optimal dose of radioimmunotherapy. Z'roc. Amer. Cancer Assoc., 32, 260 (abstract 1548) (1991).
(1983).
235,895-898 (1987).
GUADAGNI, F., TOKTORA, G., ROSELLI, M., CLAIR,T., CHO-CHUNG, Y.S., SCHLOM, J. and GREINER, J.W., Carcinoembryonic antigen regulation in human colorectal tumor cells by a site-selective cyclic AMP analogue: a comparison with interferon-gamma.Znt. J. Cancer, 48,413-422 (1991).
GUADAGNI, F., Win, P.L., ROBBINS, P.F., SCHLOM, J. and GREINER, J.W., Regulation of carcinoembryonic antigen in different human colorectal tumor cells by interferon. Cancer Res., 50,6248-6255 (1990). HAGAN,P.L., HALPERN, S.E., DILLMAN, R.O., SHAWLER, D.L., JOHNSON, D.E., CHEN,A., KRISHNAN, L., FRINCKE, J., BARTHOLOMEW, R.M., DAVID,G.S. and CARLO, D., Tumor size: effect on monoclonal antibody uptake in tumor mode1s.J. nucl. Med., 27,422-427 (1986). HAREYAMA, M., IMAI,K., KUBO,K., SHIDOU, M., OOUCHI, A., YACHI, A. and MORITA, K., Effect of radiation on the expressionof carcinoembryonic antigen of human gastric adenocarcinoma cells. Cancer, 67,
BLUMENTHAL, R.D., SHARKEY, R.M., KASHI,R., NATALE, A.M. and GOLDENBERG, D.M., Influence of animal host and tumor implantation site on radioantibody uptake in the GW-39 human colonic cancer xenograft. Znt. J. Cancer, 44,1041-1047 (1989). BUCHSBAUM, D., LLOYD,R., JUNI,J., WOLLNER, I., BRUBAKER, P., J., BURNS,F., STEPLEWSKI, Z., COLCHER, D., HANNA,D., SPICKER, SCHLOM, J., BUCHEGGER, F. and MACH,J.P., Localization and imaging 2269-2274 (1991). of radiolabeled monoclonal antibodies against colorectal carcinoma in KYRIAZIS, A.P., DIPERSIO, L., MICHAEL, G.J., PESCE,A.J. and STINtumor-bearing nude mice. Cancer Res., 48,4324-4333 (1988). NETT, J.D., Growth patterns and metastatic behavior of human tumors ESTEBAN, J.M., KUHN,J.A., FELDER,B., WONG,J.Y.C., BATTIFORA,growing in athymic mice. CancerRes., 38,3186-3190 (1978). H., BEATTY, J.D., WANEK, P.M. and SHIVELY, J.E., Carcinoembryonic D., BJORN,M.J., LEI, M. and BAUMAL, R., antigen expression of resurgent human colon carcinoma after treat- MARKS,A,, ETTENSON, ment with therapeutic doses of gOY-carcinoembryonicantigen mono- Inhibition of human tumor growth by intraperitoneal immunotoxins in nude mice. Cancer Res., 50,288-292 (1990). clonal antibody. Cancer Rex, 51,3802-3806 (1991).
PHYSIOLOGICAL FACTORS INFLUENCING RADIOANTIBODY UPTAKE
MEEKER,T.C., LOWDER,J., MALONEY, D.G., MILLER,R.A., THIELMAN,K., WARNKE,R. and LEVY,R., A clinical trial of anti-idiotype therapy for B-cell malignancy. Blood, 62,988-995 (1985). NADLER,L.M., STAHENKO, P., HARDY,R., KAPLAN,W.D., BUTTON, L.N., KUFE,D.W., ANTMAN, K.H. and SCHLOSSMAN, S.F., Serotherapy of a patient with a monoclonal antibody directed against a human lymphoma-associated antigen. Cancer Res., 40,41474154 (1980). PERVEZ,S., KIRKLAND, S.C., EPENETOS, A.A., MOOI,W.J., EVANS, D.J. and KRAUSZ,T., Effect of polarity and differentiation on antibody localization in multicellular tumor spheroid and xenograft models and its potential importance for in vivo imrnunotargeting. Int. J. Cancer, 44, 940-947 (1989). PISAM,M. and RIPOCHE, P., Redistribution of surface macromolecules in dissociated epithelial cells. J. Cell Biol., 71,907-920 (1976). SANDS, H., JONES,P.L. and SHAH,L., Correlation of vascular permeability and blood flow with monoclonal antibody uptake by human Clousner and renal-cell xenografts. Cancer Res., 48,188-193 (1988). SCHLOM, J., MOLINOLA, A,, SIMPSON, J.F., SILER,K., ROSELLI, M., HINKLE,G., HOUCHENS, D.P. and COLCHER,D., Advantage of dose
941
fractionation in monoclonal antibody-targeted radioimmunotherapy. J. nut. Cancer Inst., 82,763-771 (1990). SEARS,H.F., HERLYN,D., STEPLEWSKI, Z . and KOPROWSKI, H., Phase-I1 clinical trial of a murine monoclonal antibody cytotoxic for gastrointestinal carcinoma. Cancer Res., 45,5910-5913 (1985). SHARKEY, R.M., GOLIIENBERG, D.M., GOLDENBERG, H., LEE, R.E., BALLANCE, C., PAWLYK, D., VARGA,D. and HANSEN,H.J., Murine monoclonal antibodies against carcinoembryonic antigen: immunological, pharmacokinetic and targeting properties in humans. Cancer Res., 50,2823-2831 (1990). TAKAHASHI, T., YAMAGUCHI, T., KITAMURA,K., SUZUYAMA, H., HONDA,M., YOKOTA,T., KOTANAGI, H., TAKAHASHI, M. and HASHIMOTO,Y., Clinical application of monoclonal-antibody-drug conjugates for immunotargeting chemotherapy of colorectal carcinoma. Cancer, 61,881-888 (1988). ZIOMEK,C.A., SCHULMAN, S. and EDIDIN,M., Redistribution of membrane proteins in isolated mouse intestinal epithelial cells. J. Cell B i d , 86,849-857 (1980).
