Tumor targeting properties of monoclonal antibodies with different affinity for target antigen CD44V6 in nude mice bearing head-and-neck cancer xenografts

Authors


Abstract

The CD44 protein family consists of isoforms with tissue-specific expression, which are encoded by standard exons and up to 9 alternatively spliced variant exons (v2–v10) of the same gene. The murine MAbs U36 and BIWA-1, directed against overlapping epitopes within the v6 region of CD44, have previously been shown to efficiently target HNSCC. We herein report on the construction of 1 chimeric (BIWA-2) and 2 humanized (BIWA-4 and BIWA-8) derivatives of BIWA-1. Together with U36 and BIWA-1, these new antibodies were evaluated for affinity to the antigen in vitro as well as for biodistribution and efficacy in RIT using nude mice bearing the HNSCC xenograft line HNX-OE. As determined by surface plasmon resonance, the MAbs bound to CD44v6 with an up to 46-fold difference in affinity (Kd ranging from 1.1 × 10−8 to 2.4 × 10−10 M) with the following ranking: mMAb U36 < hMAb BIWA-4 < hMAb BIWA-8 < mMAb BIWA-1 ∼ cMAb BIWA-2. To evaluate their in vivo tumor-targeting properties, 2 MAbs with identical murine or human isotype were labeled with either 131I or 125I and administered simultaneously (50 μg/10 μCi each) as pairs showing a stepwise decrease in the difference in affinity: U36 vs. BIWA-1 (35.0-fold difference), BIWA-4 vs. BIWA-2 (14.0-fold) and BIWA-4 vs. BIWA-8 (4.0-fold). Biodistribution was assessed at 1, 2, 3 or 4 and 7 days after injection. Remarkably, for all 3 MAb pairs tested, the lower-affinity MAb showed a higher degree and specificity of tumor localization. The difference in tumor localization was more pronounced when the difference in affinity was larger. For example, 3 days after injection, the lower-affinity mMAb U36 showed a 50% higher tumor uptake than the higher-affinity mMAb BIWA-1, while blood levels and uptake in organs were similar. After labeling with 186Re (300 or 400 μCi), the same MAb pairs showed RIT efficacy consistent with the biodistribution data: 186Re-U36 was more effective than 186Re-BIWA-1, 186Re-BIWA-4 was slightly more effective than 186Re-BIWA-2 and 186Re-BIWA-4 and 186Re-BIWA-8 demonstrated similar efficacy. Based on these data, we conclude that antibodies with markedly lower affinity to a given target antigen (e.g., U36, BIWA-4) may show superior tumor targeting in comparison with higher-affinity versions of these antibodies. © 2002 Wiley-Liss, Inc.

The use of radiolabeled MAbs has been recognized as a realistic option for improvement of diagnosis and treatment of cancer. While in the last decade MAbs have been administered to thousands of patients with various types of tumor for both diagnostic and therapeutic purposes, the application of MAbs in head-and-neck oncology has not kept pace. One of the main reasons for this slow progress has been the lack of MAbs with a high specificity for head-and-neck cancer, in particular for HNSCC, which accounts for approximately 90% of head-and-neck tumors.

In recent years, a panel of MAbs has been developed that is capable of selective HNSCC targeting, as demonstrated in clinical radioimmunoscintigraphy/biodistribution studies with HNSCC patients. Among the best-qualified antibodies are the mMAbs U36 and BIWA-1 (formerly VFF18).1, 2 U36 and BIWA-1 bind to overlapping epitopes in the variable domain v6 of the cell-surface antigen CD44 and have been characterized extensively.3, 4 CD44 isoforms, which arise from differential splicing of up to 9 variant exons, show a tissue-specific expression pattern. Homogenous expression of v6-containing CD44 isoforms has been observed in squamous cell carcinoma of the head and neck, lung, skin, esophagus and cervix, while heterogeneous expression was found in several other tumor types, including adenocarcinomas of the breast, lung, colon, pancreas and stomach.4 Among normal tissues, CD44v6 isoform expression has been observed only in a subset of epithelial tissues, such as breast and prostate myoepithelium and skin and bronchial epithelium.4 Soluble v6-containing CD44 fragments have been detected in the blood of cancer patients as well as of healthy individuals.5

The physiologic role of CD44 variants has been the subject of many investigations. Evidence has been found for involvement in adhesion,6 signal transduction,7 cell migration,8 growth factor binding9 and tumor metastasis formation.10 With respect to the latter, especially v6-containing CD44 isoforms have attracted much attention since they appear to be capable of conferring metastatic potential to tumor cells originally not able to metastasize,10 while outgrowth of metastases was inhibited by MAbs directed against the v6 domain.11 Overexpression of CD44v6 in tumors has been correlated with reduced survival of patients with breast and colon cancer and with non-Hodgkin's lymphoma.12, 13, 14

These data indicate that CD44v6 is an attractive target for MAb-based therapy, especially for the treatment of HNSCC in which the level of expression is very favorable. Since HNSCCs are intrinsically radiosensitive, RIT might be particularly suitable. For the development of optimal RIT, several parameters must be considered. An important one is the radionuclide to be used. Since rhenium-186 (186Re) is a promising candidate radionuclide for RIT, we have put effort into the production of optimal 186Re–MAb conjugates.15 Other parameters that can be considered for optimizing RIT are the MAb form (e.g., murine vs. chimeric vs. humanized constructs) and the affinity of the MAb. Administration of mMAbs to patients usually induces HAMA responses, as observed for U361 and BIWA-1.2 HAMA responses can lead to rapid clearance of the injected MAb from the body, thus reducing MAb uptake by the tumor,16 and eventually to anaphylactic reactions. One possibility to reduce HAMA responses is the use of human–mouse cMAbs, which are composed of the variable regions of the murine MAb fused to the heavy and light chain constant region of the human immunoglobulin or of hMAbs that retain only a small fraction of the original murine protein sequence required for antigen binding. These modifications, however, may result in MAbs with lower affinity. Decrease of affinity is generally considered to be undesirable, based on animal studies in which it was shown that tumor uptake of low-affinity MAbs is less than that of high-affinity MAbs.17, 18, 19, 20, 21 Whether these results are generally applicable remains to be determined, especially because the MAbs used in many of these comparative studies were unrelated and directed against different epitopes on the same antigen.17, 18, 19, 20 Since the accessibility of such epitopes might be different, data on the relationship between MAb affinity and tumor uptake should be interpreted with caution.

In the present study, we describe the construction of 1 chimeric and 2 humanized derivatives of BIWA-1. Together with mMAbs U36 and BIWA-1, these antibodies were evaluated for their affinity to the target antigen in vitro, while biodistribution and therapeutic efficacy were studied in nude mice bearing human HNSCC xenografts.

Abbreviations:

HAMA, human antimouse antibody; HNSCC, head-and-neck squamous cell carcinoma; HPLC, high-performance liquid chromatography; ka, association rate; kd, dissociation rate; Kd, dissociation constant; MAb, monoclonal antibody; cMAb, chimeric MAb; hMAb, humanized MAb; mMAb, murine MAb; MAG3, S-benzoylmercaptoacyltriglycine; RIT, radioimmunotherapy; SDS-PAGE, sodium dodecyl sulfate-polyacrylamide gel electrophoresis; TFP, 2,3,5,6-tetrafluorophenol; TLC, thin-layer chromatography; VH, variable heavy chain; VL, variable light chain.

MATERIAL AND METHODS

MAbs

mMAb BIWA-1 (formerly VFF18, IgG1) was generated by immunizing BALB/c mice with glutathione S-transferase fusion protein containing the human CD44 domains v3–v10.4 The epitope recognized by BIWA-1 has been mapped to amino acids 360–370 in domain v6 of CD44 (numbering according to Kugelman et al.22). The batch used for the present studies was obtained after purification on protein G-sepharose and dialysis against PBS.

mMAb U36 (IgG1) was derived after immunization of mice with the HNSCC cell line UM-SCC-22B. The epitope recognized by U36 was mapped to amino acids 365–376 of CD44v6, indicating that U36 and BIWA-1 recognize overlapping epitopes.3 The batch used for the current studies was supplied by Centocor (Leiden, the Netherlands). U36 was purified from a concentrated tissue culture supernatant by affinity chromatography on protein A-sepharose and further purified on Q-sepharose.

Generation of cMAbs and hMAbs

mRNA was isolated from the BIWA-1 hybridoma cell line using the QuickPrep mRNA Purification Kit (Pharmacia, Uppsala, Sweden). cDNA from VH and VL was generated by RT-PCR.

Fragments were cloned into the TA cloning vector pCR II (Invitrogen, Groningen, the Netherlands) and sequenced. Two expression vectors derived from the plasmid pAD CMV123 were constructed, carrying the constant regions of human γ-1 and of the human κ light chain, respectively. Subsequently, the VH and VL fragments of BIWA-1 were cloned into the corresponding expression vectors in front of the constant regions. The chimeric antibody was named cMAb BIWA-2. Humanized versions of the BIWA-1 heavy and light chain variable regions (generated by CDR grafting) were cloned in front of the immunoglobulin constant regions of the above-mentioned expression vectors. The resulting MAb was named hMAb BIWA-4. In addition, a mutated version of BIWA-4, called hMAb BIWA-8, was constructed with the aim of improving binding affinity. This was achieved by back-mutating 2 amino acids of the light chain framework 2 and keeping the humanized heavy chain unchanged.

Recombinant MAbs were stably expressed in dihydrofolate reductase-deficient Chinese hamster ovary cells by electroporation with heavy and light chain expression plasmids. Cells were seeded into 96-well microtiter plates at densities of 500 and 100 cells/well in selection medium (α-MEM with 10% dialyzed FCS). When colonies became visible (after approx. 14 days), culture supernatants were tested for their IgG content by ELISA and the best producers expanded. Gene amplification was performed by culturing in the presence of increasing concentrations of methotrexate (20–500 nM).

Laboratory-scale production of chimeric and humanized MAbs was performed in a proprietary medium (Boehringer Ingelheim, Biberach, Germany) containing 1% FCS. IgG fractions were purified from tissue culture supernatants by affinity chromatography on protein A-sepharose. Purity was tested by SDS-PAGE and high-performance size-exclusion chromatography.

Evaluation of antibody affinity

Kinetic and affinity constants were measured using recombinant antigen on a BIAcore (Uppsala, Sweden) 2000 system. A glutathione-S-transferase fusion protein containing domains v3–v10 of human CD44 (GST/CD44v3–v10, 20 μg/ml) was immobilized on a CM5 sensor chip by the amine-coupling method according to the manufacturer's instructions, using 10 mM sodium acetate (pH 5.0) as coupling buffer. MAb (35 μl) at various concentrations (8–67 nM) in HBS [10 mM HEPES (pH 7.4), 150 mM NaCl, 3.4 mM EDTA, 0.05% BIAcore surfactant P20] were injected over the antigen-coated surface at a flow rate of 5 μl/min. Dissociation of the MAb was assessed for 5 min in buffer flow (HBS). Between 2 analyses, the surface of the chip was regenerated with a single pulse of 15 μl of 30 mM HCl. Analysis of the data and calculation of kinetic constants were performed with the BIAcore BIAevaluation software, version 2.1. ka, kd and Kd were assessed for all antibodies.

Relative binding affinities were also evaluated by competitive cell ELISA. Human A431 cells, originating from an epidermoid carcinoma of the vulva and known to express high levels of CD44v6, were seeded in 96-well tissue culture plates in 200 μl/well RPMI-1640 with 10% FCS at a density of 2.5 to 5 × 105 cells/ml. Plates were incubated overnight at 37°C in a humidified incubator with 5% CO2 in air. After removal of medium, cells were washed once with PBS, fixed with 96% ethanol for 1 min and washed again with PBS. cMAb BIWA-2, hMAb BIWA-4 and hMAb BIWA-8 (prediluted to 10 μg/ml) were applied in 1:2 serial dilutions (8 steps) in 100 μl/well in PBS/0.5%BSA/0.05% Tween-20 (assay buffer) and incubated for 30 min at room temperature. Prediluted mMAb BIWA-1 (100 μl, 20 ng/ml) was added and plates were incubated for 2 hr at room temperature on an orbital shaker. Control samples contained prediluted samples only, without BIWA-1 (0% control) or BIWA-1 only without any competing antibodies (100% control). After washing 3 times with PBS/0.05% Tween 20 (washing buffer), 100 μl of the secondary antibody (peroxidase-conjugated goat antimouse Fc, diluted 1:15,000 in assay buffer; Dako Copenhagen, Denmark) were added for detection of mMAb BIWA-1 and the plates incubated for 1 hr at room temperature on an orbital shaker. After washing 3 times with washing buffer, plates were developed with 100 μl/well tetramethylbenzidine substrate solution (Kierkegaard and Perry, Gaithersburg, MD). The reaction was stopped after 15 min with 50 μl/well 1 M phosphoric acid. Absorbance was measured in an ELISA plate reader at 450 nm (reference 610–690 nm).

Radioiodination of antibodies

Iodination of MAbs was performed essentially as described by Haisma et al.,24 using either 125I (100 mCi/ml) or 131I (200 mCi/ml), both purchased from Amersham (Aylesbury, UK). One milligram of MAb IgG dissolved in 500 μl PBS (pH = 7.4) and 1 mCi 125I or 131I were mixed in a vial coated with 75 μg Iodogen (Pierce, Oud Bijerland, The Netherlands). After 5 min of incubation at room temperature, free iodine was removed by gel filtration on a PD10 column (Pharmacia-LKB, Woerden, the Netherlands). After removal of unbound 125I or 131I, the radiochemical purity always exceeded 97%, as determined by TLC and HPLC procedures described previously.15 No aggregates or fragments were formed, as assessed by HPLC analysis.

Preparation of 186Re-labeled MAbs

186Re-labeled MAbs were prepared according to a multistep procedure using the chelate MAG3, as previously described.15 In this procedure, a solid-state synthesis for the preparation of 186Re-MAG3 is followed by esterification with TFP and conjugation of the reactive 186Re-MAG3-TFP ester to the MAb. After conjugation, the 186Re-labeled MAb was purified on a PD10 column. After removal of unbound 186Re, the radiochemical purity always exceeded 98%.

Binding assay for radiolabeled antibodies

In vitro binding characteristics of the labeled MAbs used in the biodistribution and therapy studies were determined in an immunoreactivity assay essentially as described previously.15 To test the binding of iodinated or 186Re-labeled MAbs, UM-SCC-11B cells (kindly provided by Dr. T.E. Carey, University of Michigan, Ann Arbor, MI) fixed in 0.1% glutaraldehyde were used. Five serial dilutions (ranging from 5 × 106 cells/tube to 3.1 × 105 cells/tube) were prepared with 1% BSA in PBS. Excess of unlabeled MAb IgG was added to a second tube with the lowest concentration of cells to determine nonspecific binding. IgG labeled with 10,000 cpm of 125I, 131I or 186Re was added to each tube; and samples were incubated overnight at 4°C. Cells were spun down, radioactivity in the pellet and supernatant was measured in a γ-counter (LKB-Wallace 1282 CompuGamma; Kabi Pharmacia, Woerden, the Netherlands) and the percentage of bound and free radioactivity was calculated. Data were graphically analyzed in a modified Lineweaver-Burk plot and immunoreactivity was determined by linear extrapolation to conditions representing infinite antigen excess.

Biodistribution studies in HNSCC-bearing nude mice

For biodistribution experiments, nude mice bearing s.c.-implanted human HNSCC xenografts (HNX-OE) were used as described previously.15 Female mice (Hsd athymic nu/nu, 25–32 g; Harlan, Zeist, the Netherlands) were 8–10 weeks old at the time of the experiments. Three biodistribution experiments were conducted with mice bearing 1 or 2 tumors ranging 30–470 mm3. In the first experiment, 10 μCi (50 μg) 131I-labeled mMAb U36 were injected simultaneously with 10 μCi (50 μg) 125I-labeled mMAb BIWA-1 in mice bearing tumors of 133 ± 28 mm3 (n = 20 mice, 37 tumors). In the second experiment, 10 μCi (50 μg) 131I-labeled hMAb BIWA-4 and 10 μCi (50 μg) 125I-labeled cMAb BIWA-2 were coinjected in mice bearing tumors of 167 ± 31 mm3 (n = 21 mice, 32 tumors). In the third experiment, 10 μCi (50 μg) 131I-labeled hMAb BIWA-4 and 10 μCi (50 μg) 125I-labeled hMAb BIWA-8 were coinjected in mice with tumors of 130 ± 21 mm3 (n = 23 mice, 40 tumors). Conjugates were i.v.-injected in a volume of 100 μl after dilution in 0.9% NaCl. To obtain a comparable blood/body clearance of the coinjected MAbs, only MAbs with an identical murine or human isotype were combined. The antibody dose (total dose 100 μg/mouse) was chosen to be high enough to prevent rapid isotype-related elimination of the MAb from the blood25, 26 and low enough to prevent antigen saturation in the tumor.

At indicated time points after injection, mice were anesthetized, bled, killed and dissected. Besides the tumors, the following organs were removed: liver, spleen, kidney, heart, stomach, ileum, colon, bladder, sternum, muscle, lung, skin and tongue. After weighing, radioactivity in tumors, blood and organs was counted in the dual-isotope γ-counter, with automatic correction for the 131I comptons in the 125I window setting. Radioactivity uptake in these tissues was calculated as the percentage of the injected dose per gram of tissue (%ID/g).

All animal experiments were performed according to the principles of laboratory animal care and the Dutch national law Wet op de Dierproeven (Stb 1985, 336). Until the day of MAb administration, mice were routinely housed under specific pathogen-free conditions in sterile cages in a humidity- and temperature-controlled clean room (classification 2000 according to Federal Standard 209d). On the day of injection, mice were transported to the Radio Nuclide Center, Vrije Universiteit and sterile radioimmunoconjugates were administered under aseptic conditions in a laminar flow hood.

RIT studies in nude mice

Animal RIT studies were performed to compare the therapeutic efficacy of the different MAbs labeled with 186Re. The immunoreactive fractions of the conjugates always exceeded 75%. Three therapy experiments were conducted with mice bearing 1 or 2 HNX-OE tumors ranging 45–195 mm.3186Re doses were chosen to be the maximum tolerated dose (i.e., 400 μCi) or lower (300 μCi). The maximum tolerated dose level is defined as the dose resulting in 5–15% body weight loss. In previous studies, it has been shown that the biodistribution of 186Re-labeled MAbs and iodinated MAbs is similar in this animal model15, 27, 28 and that neither 186Re nor iodine is retained in the tumor cell upon MAb internalization.27 In the first experiment, mice were given a single i.v. injection of either 300 μCi (100 μg) 186Re-labeled mMAb U36 or 300 μCi (100 μg) 186Re-labeled mMAb BIWA-1. In the second experiment, either 300 μCi (100 μg) 186Re-labeled hMAb BIWA-4 or 300 μCi (100 μg) 186Re-labeled cMAb BIWA-2 were administered. In the third experiment, either 400 μCi (100 μg) 186Re-labeled hMAb BIWA-4 or 400 μCi (100 μg) 186Re-labeled hMAb BIWA-8 were administered. Average tumor volumes were similar for all experimental groups: experiment 1, 95 ± 34 mm3 (n = 7 mice, 12 tumors) for the 186Re-mMAb U36–treated group, 91 ± 15 mm3 (n = 7 mice, 12 tumors) for the 186Re-mMAb BIWA-1–treated group and 99 ± 54 mm3 (n = 6 mice, 11 tumors) for the control group; experiment 2, 101 ± 35 mm3 (n = 7 mice, 12 tumors) for the 186Re-hMAb BIWA-4–treated group, 92 ± 43 mm3 (n = 7 mice, 12 tumors) for the 186Re-cMAb BIWA-2–treated group and the control group was the same as in experiment 1; experiment 3, 105 ± 43 mm3 (n = 8 mice, 13 tumors) for the 186Re-hMAb BIWA-4–treated group, 100 ± 42 mm3 (n = 8 mice, 13 tumors) for the 186Re-hMAb BIWA-8–treated group and 110 ± 46 mm3 (n = 7 mice, 11 tumors) for the control group. During treatment, tumors were measured twice weekly and tumor volumes relative to the volume at the start of treatment were calculated. Toxicity was monitored by measurement of body weight twice weekly. Mice were killed when one of the tumors exceeded 1,000 mm3.

Statistics

Differences in tissue uptake and tumor-to-blood uptake ratios between coinjected MAbs were statistically analyzed for each time point with Student's t-test for paired data. Differences in average tumor volume between the various RIT treatment groups were statistically analyzed for each time point with Student's t-test for independent samples.

RESULTS

In vitro binding characteristics of the CD44v6-specific MAbs

The binding affinities of the 5 MAbs were analyzed using recombinant antigen as well as human tumor cell lines. Kinetic and affinity constants were evaluated by surface plasmon resonance using GST/CD44v3-v10 as immobilized antigen. Table I shows the ka,kd and Kd values. mMAb BIWA-1 and cMAb BIWA-2, containing identical variable regions, have similar ka,kd and Kd values and show the highest affinity. In contrast, mMAb U36 and hMAb BIWA-4 have lower ka and higher kd values, resulting in markedly lower Kd values (factors 35.0 and 10.5, respectively). hMAb BIWA-8, containing back-mutations in the light chain framework region 2, showed a marked decrease of kd, resulting in increased affinity compared to hMAb BIWA-4.

Table I. Kinetics and Affinity Constants of MAbs Directed Against CD44v6
Antibodyka (M−1 · sec−1)kd (sec−1)Kd (M)Kd relative to murine BIWA-1
Murine BIWA-11.3 × 1054.2 × 10−53.2 × 10−101.0
Murine U361.5 × 1041.7 × 10−41.1 × 10−835.0
Chimeric BIWA-21.7 × 1054.1 × 10−52.4 × 10−100.7
Humanized BIWA-46.5 × 1042.2 × 10−43.4 × 10−910.5
Humanized BIWA-87.5 × 1046.3 × 10−58.4 × 10−102.6

The relative binding affinities of the cMAb and the hMAbs were also evaluated in a competitive cell ELISA using human A431 tumor cells (Fig. 1). In accordance with the affinity measurements on recombinant antigen, cMAb BIWA-2 was the most effective competitor, followed by hMAb BIWA-8 and hMAb BIWA-4. Similar results (not shown) were obtained with 2 other human HNSCC cell lines (FaDu and LICR-LON-HN5).

Figure 1.

Evaluation of relative binding affinities tested in a competitive cell ELISA. IC50, concentrations of cMAb and hMAbs at which binding of mMAb BIWA-1 to attached A431 cells is reduced by 50%. IC50 values relative to BIWA-2 are indicated.

Biodistribution in HNSCC-bearing nude mice

Biodistribution studies were performed in HNX-OE xenograft-bearing nude mice. Two MAbs with identical murine or human isotype were labeled with either 125I or 131I and injected simultaneously (50 μg, 10 μCi each). Each pair of MAbs was selected to provide a stepwise decrease in the difference in affinities: mMAb U36 had 35.0-fold lower affinity than mMAb BIWA-1 (experiment 1), hMAb BIWA-4 had 14.0-fold lower affinity than cMAb BIWA-2 (experiment 2) and hMAb BIWA-4 had 4.0-fold lower affinity than hMAb BIWA-8 (experiment 3). The fractions of MAb binding to 5 × 106 cells and at infinite antigen excess were, respectively, 59.7% and 87.4% for 131I-mMAb U36 and 91.1% and 91.1% for 125I-mMAb BIWA-1 in experiment 1; 77.4% and 82.3% for 131I-hMAb BIWA-4 and 80.5% and 79.9% for 125I-cMAb BIWA-2 in experiment 2; and 77.3% and 74.5% for 131I-hMAb BIWA-4 and 91.8% and 92.1% for 125I-hMAb BIWA-8 in experiment 3.

The biodistributions in experiment 1 were determined at days 1, 2, 3 and 7 after injection; biodistributions in experiments 2 and 3 were determined at days 1, 2, 4 and 7 after injection (Figs. 2, 3). The average %ID/g and SEM of tumor, blood and various organs after simultaneous injection of 131I-mMAb U36 and 125I-mMAb BIWA-1 were determined and tumor-to-blood ratios calculated. Tumor uptake of low-affinity U36 was significantly higher than uptake of high-affinity BIWA-1 (p ≤ 0.001) at all time points (Fig. 2a), while no significant differences were found between the uptake values of these MAbs in blood (Fig. 2a) and normal tissues (data not shown) at 1, 2 and 3 days after injection. At day 3, 50% higher tumor uptake of U36 compared to BIWA-1 was observed. At day 7, BIWA-1 levels in blood and most of the organs were significantly lower than U36 levels, indicating more rapid clearance of BIWA-1 from the blood/body. Tumor-to-blood ratios were significantly higher for U36 than for BIWA-1 at all time points (Fig. 3a).

Figure 2.

Tumor and blood levels of coinjected (a, experiment 1) 131I-labeled U36 (10 μCi, 50 μg; black bars) and 125I-labeled BIWA-1 (10 μCi, 50 μg; gray bars), (b, experiment 2) 131I-labeled BIWA-4 (10 μCi, 50 μg; black bars) and 125I-labeled BIWA-2 (10 μCi, 50 μg; gray bars) and (c, experiment 3) 131I-labeled BIWA-4 (10 μCi, 50 μg; black bars) and 125I-labeled BIWA-8 (10 μCi, 50 μg; gray bars) in HNX-OE xenograft-bearing mice (n = 5 or 6) at 1, 2, 3 or 4 and 7 days postinjection. At the indicated days after injection, mice were bled, killed and dissected and the radioactivity levels (%ID/g ± SEM) of tumor, blood and several organs (data not shown) were assessed. Significantly different uptake between coinjected 131I-labeled MAb and 125I-labeled MAb: *p ≤ 0.001, **p ≤ 0.01 and ***p ≤ 0.05.

Figure 3.

Tumor-to-blood ratios of coinjected (a, experiment 1) 131I-labeled U36 (black bars) and 125I-labeled BIWA-1 (gray bars), (b, experiment 2) 131I-labeled BIWA-4 (black bars) and 125I-labeled BIWA-2 (gray bars) and (c, experiment 3) 131I-labeled BIWA-4 (black bars) and 125I-labeled BIWA-8 (gray bars). For details, see legend to Figure 2. Ratios were obtained from paired samples. Significantly different uptake between coinjected 131I-labeled MAb and 125I-labeled MAb: *p ≤ 0.001; **p ≤ 0.01 and ***p ≤ 0.05.

Similar relationships were found in the evaluation of the 2 other MAb pairs, as partly demonstrated by Figure 2b,c and 3b,c. hMAb BIWA-4, while having the lower affinity, showed significantly higher tumor uptake (p ≤ 0.01) than cMAb BIWA-2 (Fig. 2b) and hMAb BIWA-8 (Fig. 2c) at all time points, while MAb levels in blood (Fig. 2b,c, respectively) and normal tissues (data not shown) were similar for these pairs of MAbs at days 1, 2 and 4 after injection. A 45% higher tumor uptake of BIWA-4 compared to BIWA-2 is illustrated in Figure 2b, while a 20% higher tumor uptake of BIWA-4 compared to BIWA-8 is illustrated in Figure 2c, at 4 days postinjection. At 7 days postinjection, BIWA-2 and BIWA-8 levels in blood and organs were mostly significantly lower than BIWA-4 levels, indicating more rapid clearance of these MAbs from the blood/body. Tumor-to-blood ratios were significantly higher for BIWA-4 than for BIWA-2 (Fig. 3b) and BIWA-8 (Fig. 3c) at all time points.

Consistent results were obtained from an additional experiment (data not shown) in which the radiolabels were exchanged: 125I-BIWA-4 vs.131I-BIWA-8 instead of 131I-BIWA-4 vs.125I-BIWA-8. Data from this latter experiment rule out the possibility that the type of radiolabel had influenced the pharmacokinetic behavior of the labeled MAb.

RIT in HNSCC-bearing nude mice

From the 3 biodistribution experiments it appeared that the low-affinity MAbs showed higher and more selective tumor uptake than the high-affinity MAbs and, thus, might be better suited for RIT. To test this possibility, the following treatment groups were compared in RIT studies with HNX-OE xenograft-bearing mice: experiment 1, 300 μCi 186Re-U36 or 300 μCi 186Re-BIWA-1 or saline as control; experiment 2, 300 μCi 186Re-BIWA-4 or 300 μCi 186Re-BIWA-2 or saline as control; experiment 3, 400 μCi 186Re- BIWA-4 or 400 μCi 186Re-BIWA-8 or saline as control.

In Figure 4, mean relative tumor volume (as a percentage of tumor volume at day 0) for the control and treatment groups is plotted against time. Tumors of mice in the control group in all 3 experiments showed exponential growth, with a tumor volume doubling time of about 7 days. In the groups treated with the 186Re-labeled anti-CD44v6 MAbs, tumors stopped growing, in some cases accompanied by tumor regression, shortly after injection of conjugates. However, all tumors ultimately regrew. As a control, the nonbinding 186Re-labeled hMAb F19 (directed against a human epitope on fibroblast activation protein) was evaluated in the same animal model. This radioimmunoconjugate did not induce growth cessation (data not shown).

Figure 4.

Therapeutic efficacy of 186Re-labeled CD44v6-specific MAbs in HNX-OE xenograft-bearing nude mice. Mice received saline (open circles) as control or (a, experiment 1) 300 μCi 186Re-U36 (solid squares) and 300 μCi 186Re-BIWA-1 (open triangles) or (b, experiment 2) 300 μCi 186Re-BIWA-4 (solid triangles) and 300 μCi 186Re-BIWA-2 (open squares) or (c, experiment 3) 400 μCi 186Re-BIWA-4 (solid triangles) and 400 μCi 186Re-BIWA-8 (open diamonds). Control groups in (a,b) are the same. Tumor size is expressed as average tumor volume (±SEM) during treatment relative to average tumor volume at the start of therapy.

In experiment 1, administration of 300 μCi 186Re-BIWA-1 resulted in a decrease of the tumor growth rate but not in a reduction of the mean tumor size. Administration of 300 μCi 186Re-U36, however, caused a reduction of the mean tumor volume from 185 to 120 mm3 between days 7 and 17 postinjection, after which tumors started growing again. The mean relative tumor volume in the 186Re-U36–treated group was significantly smaller (p < 0.001) than that of the 186Re-BIWA-1–treated group from day 14 on.

In experiment 2, administration of either 300 μCi 186Re-BIWA-4 or 300 μCi 186Re-BIWA-2 resulted in tumor growth arrest at day 7, with start of regrowth at day 17 postinjection. BIWA-4 tended to be more effective in RIT than BIWA-2 from day 14 on, but a significant difference between the mean relative tumor volumes was found only at day 14 postinjection (p < 0.05).

In experiment 3, mice were treated with either 400 μCi 186Re-BIWA-4 or BIWA-8, which resulted in a decrease of the relative tumor volume to a minimum of 80 ± 62% and 98 ± 81%, respectively, at day 19. Thereafter, tumors started regrowth. No significant differences in relative tumor volume between the 2 treatment groups were observed.

These data indicate that the lower-affinity MAbs are more effective in RIT than the higher-affinity MAbs, provided that the difference in affinity is large enough (as is the case for U36 compared to BIWA-1 and, to a lesser extent, BIWA-4 compared to BIWA-2).

DISCUSSION

Our results demonstrate that the high affinity of MAbs BIWA-1, BIWA-2 and BIWA-8 (Kd 3.2 × 10−10, 2.4 × 10−10 and 8.4 × 10−10 M, respectively) does not result in improved tumor delivery compared to the lower-affinity antibodies U36 and BIWA-4 (Kd 1.1 × 10−8 and 3.4 × 10−9 M, respectively). In contrast, lower affinity turned out to be advantageous in our study with respect to selective tumor uptake.

Our results deviate from those of several other studies in which radioimmunoconjugates with different affinities were compared for tumor uptake and/or therapeutic efficacy. The group of Schlom17, 18 demonstrated improved tumor delivery and therapeutic efficacy in xenograft-bearing nude mice for higher-affinity MAbs directed against the pancarcinoma antigen TAG-72 in a comparison of MAb B72.3 (Kd 4 × 10−10 M) with the higher-affinity MAbs CC49 (Kd 6 × 10−11 M) and CC83 (Kd 6 × 10−11 M), each was reactive with a different epitope on TAG-72.18 Remarkably, once tested in patients, the targeting capacity of the MAbs appeared to be similar.29 In a couple of studies, MAbs 17-1A and 323/A3, both reactive with a different epitope on the pancarcinoma antigen Ep-CAM, were compared. Velders et al.30 showed that the higher-affinity cMAb 323/A3 (Kd 5 × 10−10 M) had consistently better efficacy than cMAb 17-1A (Kd 2 × 10−8 M) when administered as “naked” MAb to xenografted nude mice. Kievit et al.28 demonstrated that the higher-affinity mMAb 323/A3 targeted better to ovarian cancer xenografts but was more heterogeneously distributed compared to cMAb 17-1A. At equivalent radiation doses, 131I-labeled mMAb 323/A3 induced better growth inhibition than 131I-labeled cMAb 17-1A in 2 of 3 xenograft lines tested. Adams et al.21 generated affinity mutants of the human anti-HER2/neu single-chain Fv antibody C6.5 by site-directed mutagenesis and tested them for tumor targeting in xenograft-bearing SCID mice. Their studies are of particular interest because all of these mutants targeted the same antigenic epitope and demonstrated a 320-fold difference in affinity (3.2 × 10−7 to 1.0 × 10−9 M). Biodistribution studies with these scFv antibodies revealed an increase in degree and specificity of localization with increasing affinity. During the time period our experiments were performed, the same researchers performed similar biodistribution studies but now using C6.5 mutation variants with affinities as high as 10−11 M.31 They showed that quantitative tumor retention did not significantly increase with enhancements in affinity beyond 10−9 M. At affinities of 1.2 × 10−10 to 1.5 × 10−11 M, tumor retention even decreased. These latter data are in line with our present observations with intact MAbs.

It is tempting to speculate that the improved therapeutic efficacy of 186Re-U36 compared to 186Re-BIWA-1 is due to the higher tumor uptake, as observed with the lower-affinity MAb U36. However, other factors also might play a role. On the basis of mathematical models of the distribution of MAbs throughout tumors, Fujimori et al.32 postulated that the use of lower-affinity MAbs in RIT might be advantageous for antitumor effects. While high-affinity MAbs will become firmly bound to antigens located at the periphery of the tumor and, therefore, will be retarded in their penetration deeply into tumors (binding site barrier theory), such a phenomenon will be met to a lesser extent with low-affinity MAbs. Indeed, it has been demonstrated that, besides tumor localization (this report), tumor penetration might be restricted by the high affinity of MAbs, thus resulting in a decrease of the efficacy of MAb-targeted therapies.20, 31

In conclusion, our results demonstrate that a significant loss of affinity incurred by humanization of BIWA-1 does not have a negative impact on the tumor-targeting properties and therapeutic efficacy of the derivatives. hMAb BIWA-4 is currently being used in tumor uptake/biodistribution studies in patients with HNSCC.

Ancillary