The human TNBC cell line MDA-MB-468 was obtained from LGC Promochem (Molsheim, France). The MDA-MB-468 cell line was cultured in adherent-cell monolayers in RPMI medium (Gibco BRL, Cergy-Pontoise, France) supplemented with 10% bovine calf serum (Gibco BRL), 1% glutamine (L-glutamine 200 mM; Gibco BRL), and 1% antibiotic (penicillin 100 U/ml, streptomycin 100 U/ml; Gibco BRL). The human myeloma cell line U266 used for the radiolabeled B-B4 immunoreactivity assay was obtained from the American Type Culture collection (Rockville, MD, USA).

NMRI-nu (nu/nu) mice over 8 to 10 weeks of age were grafted subcutaneously in the right flank with 5 × 10⁶ MDA-MB-468 cells in 0.1 mL of PBS. The animals were housed in aseptic conditions. Lugol's 1% solution was added to drinking water (0.1 mL/L) 2 days before RIT and then 2 weeks after injecting the radioiodinated reagent. The mice were injected with radiolabeled mAb 24 days after MDA-MB-468 engraftment once the tumor volume had reached a mean volume of 108 ± 55 mm³. The mice were housed in our animal core facility according to ongoing national regulations issued by INSERM and the French Department of Agriculture. The experiments performed in this study were approved by the local veterinary services (license number B44.565).

The reference antibody was the B-B4 anti-CD138 mAb (IgG1). This mouse IgG1 was kindly provided by Diaclone (Besançon, France). The 7D4 mAb was used as the isotype-matched control in the biodistribution studies 21 . This mAb recognizes the CMH-peptide complex formed by peptide MAGE3 271-279 and HLA-A2. This epitope is not expressed on MDA-MB-468 cells. The antibodies were labeled with ¹²⁵I (PerkinElmer 16, Avenue du Québec, Courtaboeuf Cedex 1, France) using the iodogen method 22 . The ¹²⁵I-labeled mAb was purified on a PD10 column. The B-B4 mAb was also labeled with ¹³¹I (MDS Nordion, Zoning Industriel, Avenue de L'Espérance, Fleurus, Belgium) using the same iodogen method. The specific activity ranged from 185 to 200 MBq/mg. Radiolabeling efficiency estimated by instant thin-layer chromatography (ITLC) was above 95%. The ¹³¹I-labeled antibody was not purified further. Iodine-¹²⁴ (IBA molecular, Chemin du cyclotron, 3, Louvain-la-Neuve, Belgium)

labeling of the B-B4 mAb for immuno-PET imaging was performed according to a previously described method 22 . Briefly, 1.7 mg (350 μL) B-B4 in 0.1 M phosphate buffer pH 7 and 20 μl (0.1 mg) of iodogen in DMF were added to a glass vial containing 230 μL (222 MBq) of Na¹²⁴I. After 15 min of incubation under gentle mixing at room temperature, the radiolabeled antibody was purified by gel filtration on NAPTM-5 columns (GE Healthcare UK, Little Chalfont Buckinghamshire, UK). The ¹²⁴I-B-B4 was recovered in 1.15 ml. The specific activity was 90 mBq/mg, and the radiochemical purity of the purified radiolabeled antibody was greater than 98% as determined by ITLC-SG chromatography using trichloroacetic acid at 10% w/v in distilled water as a mobile phase.

The immunoreactivity of radiolabeled B-B4 was determined according to the "Lindmo" cell-binding assay using CD138-positive U266 cells 23 . The number of antigen-binding sites per MDA-MB-468 cells was determined by Scatchard analysis as previously described 21 . The binding data were subjected to nonlinear regression analysis using a one-site equilibrium binding equation with Prism software.

Tumor sections (4 μm) were cut from the tissue microarray blocks and placed on superfrost slides. The immunochemical technique was performed with an automated immuno-stainer (Labvision, Fremont, CA, USA) using the strepatavidin-biotin amplification technique (ChemMate kit, Dako, Glostrup, Denmark) after appropriate antigen retrieval in EDTA buffer (pH 8) at 95°C. This involved the application of a specific primary antibody to syndecan-1/CD138 (clone MI15, 1:100, Dako). A secondary anti-mouse antibody conjugated to peroxidase was used for revelation. Peroxidase activity was revealed using 3,3'-diamino-benzidine for 5 min. Sections were counterstained with Harris hematoxylin for 3 min. Negative controls were obtained by omitting primary antibodies.

MDA-MB-468 tumor-bearing mice were given 4 MBq of ¹²⁵I-labeled B-B4 (5 μg) via the tail vein. The mice were killed 4, 24, 48, 72, and 96 h after injection (three mice per time point), the tumor and organs were removed, weighed, and the radioactivity counted using a gamma counter. The results were expressed as percent ID per gram. The same protocol was applied for biodistribution of the control isotype-matched 7D4 mAb.

Immuno-PET imaging with ¹²⁴I-labeled B-B4 was performed on six NMRI-nu (nu/nu) mice bearing MDA-MB-468 xenografts. An activity of 3.3 MBq of ¹²⁴I-labeled B-B4 (90 μg) was injected intravenously, and PET images were acquired 1, 2, 3, 4, and 8 days after injection, with an Inveon PET scanner (Siemens Medical Solutions, Knoxville, TN, USA) under anesthesia (isoflurane-O2). A CT scan was performed using the docked CT module (Siemens Medical Solutions, Knoxville, TN, USA).

PET imaging with ¹⁸FDG was performed on three NMRI-nu (nu/nu) mice bearing MDA-MB-468 xenografts receiving an intravenous injection of 6.2 MBq of ¹⁸FDG. PET images were acquired 1 h after injection. PET data were collected over a period of 20 min (one bed position), and the 3D sinograms were reconstructed using a 3D ordered subset expectation maximization followed by a maximum a posteriori algorithm (3D OSEM-MAP).

The mice were injected i.v. in the lateral tail vein with escalating activity of ¹³¹I-labeled B-B4 mAb. The amount of B-B4 mAb was adjusted to 120 μg by adding unlabeled B-B4 antibody and the injected doses were adjusted to a constant volume of 0.2 ml with sterile PBS. These groups received 0.0 MBq (n = 4), 11.1 MBq (n = 4), 14.8 MBq (n = 4), 18.5 MBq (n = 4), 22.2 MBq (n = 4), or 25.9 MBq (n = 4) of ¹³¹I-labeled B-B4 mAb. This assay was repeated under the same experimental conditions with 25.9 (n = 3) and 37 MBq (n = 3). The maximum tolerated dose (MTD) was determined using the data obtained in the two assays as the dose just below the one at which at least one mouse died or lost more than 10% of its weight before treatment. The mice were weighed weekly for 96 days, and blood samples were taken from the inner border of the eye at 0, 14, 28, 40, 55, and 90 days after ¹³¹I-labeled B-B4 mAb injection. Leukocyte and platelet numbers were determined using a cell counter (Melet-Schloesing Laboratories, Cergy-Pontoise France).

An additional RIT experiment was performed under the same conditions as those described for the dose escalation at the MTD (22.2 MBq) (n = 4) and at a lower dose of 14.8 MBq (n = 4) compared to the control group (n = 5). Tumor volumes were measured with a sliding caliper twice a week for 96 days measuring tumor length (L), width (w) and thickness (t). Tumor volume (V) was calculated according to the formula: V = π/6 × L × w × t. In order to increase statistical power, the results of tumor growth assessed by this assay were pooled with those of the corresponding mice groups receiving the same doses in the dose escalation assay. The effect of unlabeled B-B4 mAb (120 μg) and the ¹³¹I-labeled isotype-matched mAb control was tested in comparison with a PBS control group in an independent assay.

The parameters used to evaluate the efficacy of each type of treatment were the minimal relative volume (the ratio of the smallest measured tumor volume to the initial tumor volume before treatment) and the event defined as the growth delay (the time required for the tumor to double in size after measurement on the day of treatment). The mean tumor volume was calculated for each group on each day of measurement. Tumor responses were categorized as follows: cure (tumor disappeared with no recurrence at the end of the study 96 days later), complete response (CR) (tumor disappeared for at least 7 days but later re-grew), partial response (PR) (tumor volume decreased by 50% or more for at least 7 days but then re-grew).

Correlations between dose and degree of toxicity (body weight loss, platelets, leukocytes) or efficacy (tumor growth) were made using the nonparametric Spearman's test. Groups of interest (radioimmunotherapy doses) were also compared using the nonparametric ANOVA with Bonferroni correction. Curves for event-free survival (the time required for tumors to reach at least twice the initial volume) were calculated according to the Kaplan-Meier method and compared using the Log-rank test. All analyses were two-sided. P values < 0.05 were considered significant. Analyses were performed using SAS 9.1 (SAS Institute, Cary, NC, USA) and Stata 10.0 SE (StataCorp, College Station, TX, USA). The comparison of the mean tumor volume curves was performed using a nonparametric test with an on-line software from UCL Institut de Statistique (Louvain, Belgium) http://www.stat.ucl.ac.be/ISpersonnel/lecoutre/Tgca/french/help/help.htm.

CD138 expression was studied upon equilibrium binding of the specific B-B4 antibody to the human breast cancer MDA-MB-468 cell line. The expression of CD138 on this cell line was rather low; only 1.19 × 10⁴ ± 9.27 × 10² CD138 sites per cell were detected with saturating amounts of B-B4 mAb (Figure 1A). This result is consistent with previous work indicating that MDA-MB-468 is among the lowest syndecan-expressing breast cancer cell lines 24 . The affinity (dissociation constant) of B-B4 mAb was 4.39 ± 1.10 nM. The immunoreactivity assessed using the method described by Lindmo 23 (Figure 1B) was close to 80%, showing that the labeling did not modify the binding of B-B4 mAb to the CD138 antigen. For convenience, the immunoreactivity assay was performed on the multiple myeloma cell line U266 that expresses at least ten times more CD138 antigen than MDA-MB-468 cells.

Syndecan-1 expression was further investigated by immunohistochemistry on xenograft tumors in order to validate the preclinical model for CD138 targeting. The lack of estrogen or progesterone receptor and Her2/neu expression on the MDA-MB-468 cell line was consistent with the capacity of this cell line to form tumors in nude mice without additional hormone treatment. Figure 2 presents micrographs of an MDA-MB-468 xenograft with hematoxylin-eosin staining (panel A) and CD138 immunostaining (panel B). The trabeculae of carcinoma cells were observed with high nuclear pleomorphism, high mitotic activity, and a central necrosis. Despite the relatively low expression of CD138 on cultured MDA-MB-468 cells, intense membrane immunostaining of carcinoma cells from tumor xenografts was visualized by CD138 immunostaining with the B-B4 mAb.

In mice injected with ¹²⁵I-labeled B-B4 mAb, maximum tumor uptake was reached at 24 h (13.8% ± 1.8% ID per gram) and was maintained over time with a tumor uptake of 8.4% ± 1.9% ID per gram at 96 h (Figure 3A). In normal tissues, the maximum accumulation was observed at 4 h but rapidly decreased with kinetics similar to those for blood (Figure 3C). A higher level of ¹²⁵I-labeled B-B4 mAb was observed in the more vascularized organs such as the heart and lungs, in keeping with their higher blood content. The level of ¹²⁵I-labeled B-B4 mAb in the blood reflected the serum stability of radiolabeled B-B4 mAb.

We further investigated the specificity of B-B4 mAb tumor uptake by comparing it with the biodistribution of an isotype-matched control antibody. To this end, we used the 7D4 IgG1 antibody developed in our laboratory, which does not bind to MDA-MB-468 cells. Tumor uptake measured with the nonspecific 7D4 mAb was subtracted from that detected with the B-B4 mAb in the same assay. The results presented in Figure 3B clearly show that all organs except the tumor displayed an identical B-B4 and 7D4 mAb uptake, with differences that did not exceed 1% ID per gram. Conversely, a clear and specific binding of ¹²⁵I-labeled B-B4 mAb was observed in the tumor. This specific ¹²⁵I-labeled B-B4 mAb uptake increased rapidly on the first day post injection and at a lower rate up to 72 h thereafter (Figure 3D). The tumor uptake was long lasting since 7 days after injection, it reached at least 80% of the maximum measured at 72 h (data not shown). Nevertheless, B-B4 tumor uptake was relatively low as compared to other tumor-specific antibodies. Consequently, in this MDA-MB-468 tumor model, the tumor/blood ratio is almost 1:1, as compared to 2:1 in other models. These observations can be explained by the low CD138 expression by this breast cancer cell line, together with the central necrosis in the tumors at the time of the biodistribution study.

The availability of isotopes of the same element, such as ¹²⁴I and ¹³¹I or ⁸⁶Y and ⁹⁰Y, offers the possibility to image tumors by immuno-PET with the antibody used for therapy with identical labeling technology and, therefore, the same pharmacokinetics. This should afford more accuracy in dosimetry calculations before therapy and in providing information on tumor antigen expression complementary to metabolic ¹⁸FDG-PET imaging. We thus used the B-B4 mAb labeled with ¹²⁴I for immuno-PET imaging in mice engrafted with MDA-MB-468 cells. Imaging was performed on days 1, 3, and 8 for mice injected with 3.5 MBq ¹²⁴I-labeled B-B4 mAb or 1 h after ¹⁸FDG injection. Tumor imaging with ¹²⁴I-labeled B-B4 was as informative as ¹⁸FDG for tumor visualization, even 1 day after antibody injection (Figure 4). Immuno-PET images recapitulated the results of biodistribution at all times, with a clear tumor imaging at day 1 together with an intense activity in the heart and blood vessels. At day 3, when antibody tumor uptake had reached its maximum, the staining was intense and comparable with that of the heart, which is in accordance with the equal uptake of the antibody in the blood and tumor observed at 72 h in the biodistribution study. Finally, at day 8, the persistence of the labeled antibody within the tumor, together with the decline in blood content, led to a well-contrasted image of the tumor with little signal remaining in normal tissues. This imaging assay validates the feasibility of syndecan-1 targeting for immuno-PET imaging with the B-B4 mAb.

We performed a dose escalation study from 11.1 to 37 MBq with a 3.7-MBq increment of ¹³¹I-labeled B-B4 mAb. Hematological toxicity was assessed by platelet and leukocyte numeration (Figure 5). For platelets, the nadir was observed at day 14. A dose-dependent drop in platelet count was observed from 11.1 to 25.9 MBq reaching respectively 67.4% and 38.6% of the platelet number before treatment (0.56 × 10⁶/mm³ vs. 1.45 × 10⁶/mm³ and 0.99 × 10⁶/mm³ vs. 1.47 × 10⁶/mm³, respectively). A more dramatic drop was observed at 37 MBq where the platelet count at day14 decreased to 15% of the pretreatment value, arguing for an acute hematological toxicity at this injected activity. The platelet count reverted to normal by day 40 at injected activities ranging from 11.1 to 25.9 MBq. The evolution of leukocyte count over time was similar to that of platelets, with a nadir at 14 days. The leukocyte count at injected activities ranging from 14.8 to 25.9 MBq decreased respectively to 19.7% and 7.9% of the pretreatment value (1,000 vs. 5,075/mm³ and 460 vs. 5,847/mm³, respectively). An injected activity of 11.1 MBq was less toxic with a leukocyte count of 47.0% at day 14 and a rapid recovery of normal counts as early as 26 days after treatment. With the exception of the 37 MBq group in which all mice died, a progressive recovery in leukocyte number occurred after the nadir and a normal count was reached 70 days after injection of ¹³¹I-labeled B-B4 mAb. In summary, activities ranging from 11.1 to 25.9 MBq induced a dose-dependent hematological toxicity that spontaneously resolved in 40 days for platelets and 70 days for leukocytes. An injected activity of 37 MBq was clearly toxic.