Comparison between Internalizing Anti-HER2 mAbs and Non-Internalizing Anti-CEA mAbs in Alpha-Radioimmunotherapy of Small Volume Peritoneal Carcinomatosis Using 2 1 2Pb

Background and Purpose We assessed the contribution of antibody internalization in the efficacy and toxicity of intraperitoneal α-radioimmunotherapy (RIT) of small volume carcinomatosis using 212Pb-labeled monoclonal antibodies (mAbs) that target HER2 (internalizing) or CEA (non-internalizing) receptors. Materials and Methods Athymic nude mice bearing 2–3 mm intraperitoneal tumor xenografts were intraperitoneally injected with similar activities (370, 740 and 1480 kBq; 37 MBq/mg) of 212Pb-labeled 35A7 (anti-CEA), trastuzumab (anti-HER2) or PX (non-specific) mAbs, or with equivalent amounts of unlabeled mAbs, or with NaCl. Tumor volume was monitored by bioluminescence and survival was reported. Hematologic toxicity and body weight were assessed. Biodistribution of 212Pb-labeled mAbs and absorbed dose-effect relationships using MIRD formalism were established. Results Transient hematological toxicity, as revealed by white blood cells and platelets numbering, was reported in mice treated with the highest activities of 212Pb-labeled mAbs. The median survival (MS) was significantly higher in mice injected with 1.48 MBq of 212Pb-35A7 (non-internalizing mAbs) (MS = 94 days) than in animals treated with the same activity of 212Pb-PX mAbs or with NaCl (MS = 18 days). MS was even not reached after 130 days when follow-up was discontinued in mice treated with 1.48 MBq of 212Pb-trastuzumab. The later efficacy was unexpected since final absorbed dose resulting from injection of 1.48 MBq, was higher for 212Pb-35A7 (35.5 Gy) than for 212Pb-trastuzumab (27.6 Gy). These results also highlight the lack of absorbed dose-effect relationship when mean absorbed dose was calculated using MIRD formalism and the requirement to perform small-scale dosimetry. Conclusions These data indicate that it might be an advantage of using internalizing anti-HER2 compared with non-internalizing anti-CEA 212Pb-labeled mAbs in the therapy of small volume xenograft tumors. They support clinical investigations of 212Pb-mAbs RIT as an adjuvant treatment after cytoreductive surgery in patients with peritoneal carcinomatosis.


Introduction
Peritoneal carcinomatosis is relatively common in gynecological or digestive cancers or primary peritoneal malignancies, such as mesothelioma or peritoneal serous carcinoma [1]. The combination of cytoreductive surgery to treat the visible disease, and hyperthermic intraperitoneal chemotherapy (HIPEC) can improve the patients' median survival [2,3,4,5,6]. Nevertheless, this approach is associated with high post-operative morbidity (30-50%) and mortality (4%) due to surgery complications and/or chemotherapy side effects [7,8].
Several studies in rats have demonstrated that radioimmunotherapy (RIT) could be an alternative approach to HIPEC [9,10,11,12]. However, the results of the only phase III clinical trial on intraperitoneal RIT for ovarian cancer by injection of 90 Y-HMFG1 monoclonal antibodies (mAbs) were rather unsatisfactory [13], possibly due to the low absorbed doses to the tumors and high incidence of extraperitoneal disease recurrence [14]. The choice of 90 Y may be questionable for RIT of small volume tumors because the emitted b particles have a long range in matter (0.05-12 mm) and thus they may cause bone marrow toxicity due to non-specific cross fire irradiation. Moreover, as they have very low linear energy transfer (LET = 0.2 keV/mm) they are poorly cytotoxic per unit dose.
Conversely, alpha particles constitute attractive candidates for RIT of single cells or small volume tumors (for reviews [15,16] ) because they have shorter path length (40-100 mm) and higher LET (50-270 keV/mm) than compared to beta particles and thus they are highly deleterious locally. Moreover, new in vivo nanogenerators of alpha radionuclides, such as 225 Ac/ 213 Bi or 212 Pb/ 212 Bi that generate 213 Bi and 212 Bi respectively, and new chelating agents that improve the radionuclide-mAb complex stability have improved the availability of alpha particle emitting isotopes for clinical RIT (for review [17]).
Alpha particle emitters, such as 213 Bi, 211 At and 212 Bi (generated from 212 Pb), have been coupled to monoclonal antibodies, peptides or liposomes for treating leukemia [18,19], breast [20,21], prostate [22,23,24], ovarian [25,26,27,28,29], colorectal [30,31,32] and bladder [33] cancers in mice. Most of the preclinical studies on RIT with 212 Pb [30,31,34,35,36] and the ongoing clinical phase I study in the USA have targeted the human epidermal growth factor receptor 2 (HER2). As anti-HER2 mAbs are internalized in the cytoplasm after receptor binding (for review [37]), 212 Pb-mAb internalization could contribute to RIT efficacy and toxicity. Indeed, internalization may be associated with high radioactivity uptake via cell surface receptor recycling and it may also help retaining radioisotope daughters (including, for 212 Pb, the two alpha emitters 212 Bi and 210 Po, and the beta emitter 208 Tl) within the cytoplasm of targeted cells. Some have suggested that, 212 Pb-mAb internalization and the subsequent acidic catalysis within lysosomes may lead to the dissociation of the radio-metal from the chelator and to the release of isotopes from the targeted cells that may produce toxic effects, such as bone toxicity [38].
Therefore, the aim of our work was to compare the efficacy and toxicity of non-internalizing 212 Pb-35A7 (anti-carcinoembryonic antigen, CEA) mAbs, which mostly remain at the cell surface, and of internalizing 212 Pb-trastuzumab (anti-HER2) mAbs in RIT of small volume peritoneal tumors that express CEA (high level) and HER2 (lower level) receptors.

Materials and Methods
Cell Line and mAbs HER2-positive vulvar squamous carcinoma A-431 cells obtained from ATCC were transfected with constructs encoding CEA and luciferase [12]. Cells were grown in Dulbecco's Modified Eagle Medium supplemented with 10% fetal calf serum, 1% penicillin/streptomycin and 1% geneticin at 37uC in a humidified atmosphere containing 5% CO 2 . The IgG1k 35A7 mAb against the CEA Gold 2 epitope was obtained from hybridoma kindly provided by Dr J-P Mach, Lausanne, Switzerland [39] and the anti-HER2 mAb trastuzumab (HerceptinH, Genentech, San Francisco, CA) was also used. The non-specific PX IgG1 mAb was obtained from the ATCC mouse hybridoma P3X63Ag8 [40] and was used for control experiments. The 35A7 and PX mAbs were purified from mouse hybridoma ascitic fluids by ammonium sulfate precipitation followed by ion exchange chromatography on DE52 cellulose (Whatman, Balston, UK CEA and HER2 Expression Levels 500610 4 A-431 cells were incubated with 20 mg/ml 35A7 or trastuzumab for 1.5 h, then washed twice with PBS before incubation with 5.4 mg/ml anti-mouse IgG-FITC antibody produced in goat (F2653, Sigma-Aldrich, St. Louis, MO, USA) or 7.6 mg/ml anti-human IgG-FITC antibody produced in goat (F9512 Sigma-Aldrich, St. Louis, MO, USA). We determined the number of fluorophores per antibody by measuring optical density at both 280 nm (mAb absorption) and 495 nm (FITC absorption) for given concentration of mAb and found 7.5 FITC per anti-35A7 and 13.1 FITC per anti-trastuzumab mAbs. Samples were analyzed with a Cytomics FC 500-MCL Flow Cytometer (Beckman Coulter, Roissy, France) by recording 5,000 events to analyze CEA/HER2 expression using WinMDI software. Control groups consisted of cells incubated with the secondary antibody only.
In addition, the number of HER2 and CEA receptors at the cell surface has also been determined using two in vitro kit assays (QifikitH, Dako, France; CellQuant CalibratorH, Biocytex, France).

Conjugation and Radiolabeling
Trastuzumab, 37A7 or PX were conjugated with TCMC (Macrocyclics, Dallas, TX, USA) using a 12-fold molar excess of ligand to mAb as described in [43]. TCMC was chosen based on its stability at low pH, because DOTA undergoes acidic catalysis within lysosomes after internalization and this can dissociate the radio-metal from the chelator, leading to release of isotopes and toxicity [38,44,45]. The mAb final concentration was quantified using the BCA Protein Assay Reagent (Pierce, Netherlands). The number of TCMC molecules linked to the mAbs was determined using a spectrophotometry assay based on the titration of the lead-Arsenazo(III) complex [46] and was about 8-10 TCMC/mAb. Trastuzumab-TCMC was from AREVA Med LLC (Bethesda, MD, USA).
The 224 Ra/ 212 Pb generators were provided by AREVA Med SAS (Bessines-sur-Gartempes, Haute-Vienne, France) and radiolabeling with 212 Pb was performed as described by Dong [47]. Then, 1 mg mAb-TCMC was incubated with 37 MBq 212 Pb at 37uC for 1 hour and the reaction quenched with 4 mL 0.1 M EDTA. Specific activities were generally around 37 MBq/mg for the three mAbs. The labeling yield (ratio 212 Pb/ 212 Pb-mAbs) was assayed using SG-ITLC 10-cm strips (Gelman Sciences, Ann Arbor, MI USA) developed in 0.15 M NH4OAc buffer, pH 4.0. 212 Pb-mAbs were retained at the origin whereas 212 Pb acetate migrated with the solvent front. The strips were dried, cut into 1 cm segments, and counted in a gamma-counter (Hewlett Packard, Palo Alto Instrument, Ca, USA). It was generally ,2%. After conjugation, the ITLC analysis demonstrates the absence of remaining unconjugated TCMC post diafiltration, therefore the arsenazo assay in combination with the protein quantification is sufficient to estimate the mole ratio of chelate to antibody. Additionally, since the chelation is performed at a predefined ratio of 1 mg per 37 MBq, the ITLC post chelation with 212 Pb demonstrates that we consistently have a high labeling yield as well as a consistent recovery yield on the desalting column.
Immunoreactivity of 212 Pb-mAbs against CEA or HER2 was assessed in vitro by direct binding assays using sepharose activated beads (GE Healthcare) coated with human recombinant CEA and HER2. The binding percentage was determined by measuring the antigen-bound radioactivity after overnight incubation followed by 2 washes with phosphate-buffered saline. It was shown to range from 70% to 80%.

Animal Model
Swiss nude mice (7 week/old females) from Charles River were acclimated for 1 week before experimental use. They were housed at 22uC and 55% humidity with a light-dark cycle of 12 h. Food and water were available ad libitum. Body weight was determined weekly and the mice were clinically examined throughout the study.

Tumor Growth Follow-up by Bioluminescence Imaging
Tumor growth was followed weekly by in vivo bioluminescence imaging after ip injection of 200 mL luciferin (0.1 mg luciferin/g) as described above [12]. For this purpose, we previously calibrated the bioluminescence signal (photons/s) as a function of tumor weight (g) as described in [12] and reported in Figure 1A. Typically, mice bearing intraperitoneal A-431 tumors xenografts were imaged and next sacrificed for collection and measurement of tumor nodules. The sum of nodules masses per mice was calculated and correlated to the bioluminescence signal. We found a good linearity between the bioluminescence signal and the tumor weight for tumor weight between 0.01 g and 0.08 g. For larger tumors, the dose-response relationship was next saturated, and tumor size was, therefore, underestimated. Mice were sacrificed by CO 2 asphyxiation, when the bioluminescence signal reached 2.0610 9 photons/s for two consecutive times (weekly measurements). Dissection revealed that the real tumor weight was then 2-3610 21 g.
Hematologic toxicity was evaluated using the scil Vet abc system (SCIL Animal Care Co.) and animal weight was determined weekly.

Biodistribution Experiments and SPECT-CT Imaging
To assess the biodistribution of 212 Pb-mAbs, mice were xenografted with A-431 cells as described above. Four days later, mice were ip injected with 0.37 MBq (37 MBq/mg) 212 Pb-35A7 or 212 Pb-trastuzumab and 30 mg of the relevant unlabeled mAb. At each time point (1, 6, 11, 22, 33 and 44 h after injection), 3-4 animals/group were anesthetized, bled and dissected. The uptake of radioactivity (UOR Biodis ) of tumor nodules and organs was measured using a c-counter. The percentage of injected activity per gram of tissue (%IA/g) was then calculated as described in Santoro et al. [12].
Due to lack of detection sensitivity, whole-body SPECT/CT images using 212 Pb-mAbs could not be acquired even for the highest (1.48 MBq) injected activities. Tumors were thus imaged at various times (24 h, 48 h) following ip injection of 18.5 MBq 125 I-mAbs using a 4-head multiplexing multipinhole NanoSPECT camera (Bioscan Inc.).

Uptake of Radioactivity per Organ and Tumor, Dosimetry
The uptake of radioactivity (UOR) per tissue (kBq) in RIT experiments (UOR RIT ) was extrapolated from biodistribution experiments by multiplying UOR Biodis by the ratio between the highest activity used in RIT and biodistribution experiments, namely 4.

UOR RIT (kBq)~UOR Biodis (kBq)|4
We thus considered that the weight of healthy tissues did not change during the study and that it did not differ between RIT and biodistribution experimental conditions. However, during the two days following the injection of radiolabeled mAbs, tumor weight was measured in biodistribution experiments and found to be slightly higher than compared to RIT experiments at 1.48 MBq of 212 Pb-mAbs because of the lower activity injected. Tumor weight in RIT experiments was determined from the bioluminescence signal using calibration curves and was 9610 23 g in both 212 Pb-35A7 and 212 Pb-Trastuzumab groups. Then, for tumors, UOR RIT was calculated by multiplying UOR Biodis per gram of tumor by the measured tumor weight in RIT conditions and next by the ratio between the highest activity used in RIT and biodistribution experiments, namely 4: This approach was supported by the finding that UOR Biodis increased linearly with tumor weight and was validated in [12]. The cumulated activity per tissue (Ã rs ), was next calculated by measuring the area under the UOR RIT curves.
To estimate the mean absorbed dose delivered to the different organs and tumors, we assumed that the energy emitted during the decay of 212 Pb and its daughters was deposited locally and totally absorbed. This is a relevant assumption for alpha particles and for most of the beta particles emitted within this decay chain. Thus, the mean absorbed dose to organs/tumors was determined by multiplying Ã rs by 8.7 MeV, which corresponds to the overall energy released by the alpha (7.8 MeV) and beta (0.9 MeV) particles emitted during the decay of 212 Pb, 212 Po, 208 Tl [48]. For the mean absorbed dose delivered to tumors, the contribution from free (i.e., unbound) radiolabeled antibodies in the peritoneal cavity was not taken into account as short-ranged alpha particles should irradiate tumors only superficially and since we finally did not observe any therapeutic effect of non-specific 212 Pb-PX mAb on survival (Figure 2).
In this dosimetric analysis, two cell S-values (S NrCy and S NrCs ) were also determined for A431 cells and for 212 Pb and its daughters. Only the alpha particle contribution was considered in these calculations and since the tumors were much larger than the path length of the alpha particles, we calculated the mean absorbed dose in the same manner for both mAbs. Calculations were carried out with a validated small-scale dosimetry code [49] and results were compared with the values of the MIRD tables [25].

Statistical Analysis
Kaplan-Meier survival estimates were calculated from the xenograft date to the date of the event of interest (i.e., bioluminescence of 2610 9 photons/s) and compared with the log-rank test. Statistical analyses were performed using STATA 10.0.

Tumor Growth and Survival
Using flow cytometry analysis, we determined a higher signal of fluorescence for CEA than for HER2 receptors in A-431 cells ( Figure 3A) though the number of FITC per anti-35A7 mAb was lower than the number of FITC per anti-trastuzumab mAb and that the concentration of FITC-mAb was higher for anti-Trastuzumab. However, if affinity of 35A7 for CEA was greater than affinity of trastuzumab for HER2, we couldn't accede to affinity value of both types of FITC-mAbs for their epitopes. Therefore, we confirmed these data by using in vitro kit assays (QifikitH, Dako, France; CellQuant CalibratorH, Biocytex, France), which similarly indicated that 1665610 3 and 200635610 3 HER2 and CEA receptors, respectively, were expressed at the surface of A-431 cells.
We also confirmed that anti-HER2 (trastuzumab) mAbs, are internalized after receptor binding, while anti-CEA (35A7) 212 Pb-mAbs, remain at the cell surface ( Figure 3B). Tumor growth was evaluated by bioluminescence imaging (Figure 1A, left panel) using the calibration curve described in Materials and Methods. At day 4 post-graft, the mean tumor weight (about 0.01 g) was distributed among 7.264.8 nodules/mouse ( Figure 1A, middle panel). Only the biggest tumor nodules were detected by SPECT-CT imaging at day 1 or 2 after ip injection of 125 I-mAbs ( Figure 1A

Toxicity of 212 Pb Radioimmunotherapy
No weight loss was observed in mice treated with 212 Pbtrastuzumab or 212 Pb-35A7 ( Figure 1C). Hematological toxicity of unlabeled and 212 Pb-labeled mAbs was assessed by measuring white blood cells ( Figure 4) and platelets number ( Figure 5) over 30 days following treatment. Results were normalized to those of the NaCl group for PX and to those obtained with unlabeled 35A7 or trastuzumab for 212 Pb-labeled specific mAbs. All mice treated with 212 Pb-mAbs showed a transient and activity-dependent hematological toxicity. Reduced number of white blood cells and platelets was also observed following treatment with 0.74 and 1.48 MBq of the two specific 212 Pb-mAbs.

Biodistribution of 212 Pb-labeled mAbs
212 Pb-mAb biodistribution was determined over 44 h in tumor nodules and healthy tissues and uptake of radioactivity was measured (UOR Biodis ). Maximal concentrations for tumors were 20.568.9 and 16.2566.9 of the injected activity per gram of tissue (%IA/g) for 212 Pb-35A7 and 212 Pb-trastuzumab, respectively. Maximal uptake was measured 1 h after injection for trastuzumab while it was measured at 11 h for 35A7. However, the latter measurement is associated to large error bars and we previously showed that maximal uptake of 125 I-35A7 in peritoneal tumor was observed 1 h after ip injection [11]. Then, maximal uptake measured 11 h after injection is likely to be overestimated or attached to large uncertainty. Maximal uptake was next measured in blood with 10.463.9 and 9.2264.0%IA/g for 212 Pb-35A7 ( Figure 6A) and 212 Pb-trastuzumab ( Figure 6B), respectively. Next, liver, kidneys and lungs showed the highest values with 7.762.7, 8.262.5, and 6.962.0%IA/g ( 212 Pb-35A7, Figure 6A) and 7.763.2, 7.363.7, and 3.261.4%IA/g ( 212 Pb-trastuzumab, Figure 6B), respectively.

Uptake of Radioactivity and Dosimetry
As a preliminary step towards the assessment of the absorbed dose, the UOR Biodis values expressed in kBq were multiplied by 4, as described in Materials and Methods, which corresponds to the ratio between the highest activity used in RIT experiments (1.48 MBq) and the activity injected for biodistribution analysis  (Figure 7). The uptake of radioactivity during RIT UOR RIT was then obtained and was maximal at 1 h after injection of 212 Pb-35A7 ranged between 1.46 kBq (bone) and 349.2 kBq (carcass, data not shown), with 3.77 kBq for tumor nodules, 7.29 kBq for spleen, 35  by the total amount of energy emitted by 212 Pb and its daughters (that were assumed to be deposited locally and totally absorbed) and by dividing the result by the organ mass. The mean absorbed doses were at the highest activity, 35.5 Gy for 212 Pb-35A7 and 27.6 Gy for 212 Pb-Trastuzumab in tumors and 10.9 and 7.2 Gy, respectively, in blood ( Figure 7C). 212 Pb-35A7 mAbs were less efficient than 212 Pb-trastuzumab, although the dose absorbed by the tumor was higher for 212 Pb-35A7.

Discussion
Here, we investigated the efficacy and toxicity of 212 Pb-labeled anti-CEA (non-internalizing; 35A7) and anti-HER2 (internalizing; trastuzumab) mAbs in RIT of small tumors. To this aim, we xenografted nude mice with carcinoma A-431 cells that express both CEA (high level) and HER2 (lower level) receptors.
We then injected them with 212 Pb-35A7 and 212 Pb-trastuzumab mAbs that were previously coupled to TCMC to provide resistance to intralysosomal acid catalysis [44], seen by others, with internalized 212 Pb-DOTA-mAb complexes leading to release and accumulation of 212 Pb in bone and causing bone marrow toxicity [45]. One hour after injection, the bone uptakes of 212 Pbtrastuzumab (1.560.6%) and 212 Pb-35A7 (2.260.2%) were comparable. However, the bone absorbed dose calculated over 72 h was slightly higher for 212 Pb-trastuzumab (3.8 Gy) than for 212 Pb-35A7 (2.5 Gy). The higher bone absorbed dose of 212 Pbtrastuzumab was not associated with higher hematological toxicity as the WBC and platelet count nadir were more marked in mice treated with 212 Pb-35A7.
Although the expression level of HER2 was lower than that of CEA receptors, the cumulative UOR RIT by tumors was in the same range for both targeting models (2.8610 8 for 212 Pb-35A7 and 1.9610 8 Bq.s for 212 Pb-trastuzumab). This suggests that the lower HER2 expression was partly compensated by HER2 recycling at the cell surface, where receptors could be targeted again by 212 Pb-mAbs. Nevertheless, the final tumor mean absorbed dose was still higher for 212 Pb-35A7 (35.5 Gy) than for 212 Pb-trastuzumab (27.6 Gy).
Calculation of the median survival showed a dose-dependent MS increase in mice treated with increasing activities of the two mAbs. However, if our study showed that the blood absorbed dose, higher for 212 Pb-35A7 (10.9 Gy) than for 212 Pb-trastuzumab (7.5 Gy) mAbs, was in agreement with the higher hematological toxicity of 212 Pb-35A7 mAb (Figure 4), it also highlighted the lack of absorbed dose-effect relationship between tumor absorbed dose and survival because 212 Pb-trastuzumab mAbs were more efficient than 212 Pb-35A7 mAbs.
Though the range of absorbed doses that we found was quite in agreement with data from literature [50], a possible explanation for this lack of correlation relies on the approaches and assumptions done to calculate the mean absorbed dose. Alphaparticle dosimetry has been reviewed extensively in MIRD22 pamphlet abridged in [51]. First, we took into account the mean energy (8.7 MeV) released by alpha and beta particles that are emitted during the decay of 212 Pb and its daughters, but not the emitter's subcellular localization (cell surface or cytoplasm). Nevertheless, from the calculations of the S-values for A-431 cells (cell and nucleus diameters: about 1464.5 and 9.261.9 mm [52]), we derived that, when considering alpha-emission only, every decay occurring in the cytoplasm delivered a 1.5 times higher dose to cells than a decay occurring at the cell surface (for the whole chain of decay S NrCy = 3.06610 22 Gy/Bq.s and S NrCs = 1.98610 22 Gy/Bq.s). If subcellular localization is an important parameter when evaluating the efficacy in isolated tumor cells, or very small micro-metastases, its influence is negligible in our work since tumors of a few millimeters in diameter are studied. Specifically, by using a validated small-scale dosimetry code [49], we found that the mean absorbed dose delivered to one A-431 cell surrounded by similar cells (a cell packing of 0.74 was assumed in this model and one 212 Pb atom attached to each cell) was the same (0.64 Gy) with internalizing and non-internalizing radiolabeled vectors. This result could be explained by the fact that, in 1-2 mm (or larger tumors) tumors, 95% of the received dose is due to cross-fire. This observation also invalidates the hypothesis that the release from TCMC in the extracellular space of the short lived 212 Po (T 1/2phys = 2.9 10 27 s) might reduce the mean tumour absorbed dose when noninternalizing mAbs are used.
Secondly, we assumed that for both 212 Pb-mAbs S-values, 212 Pb was in equilibrium with its daughters. This is true for analysis times longer than 5 hours following elution of the radiolabeled mAbs through the gel size exclusion column after radiolabeling. However, as the same hypothesis was used for both targeting models, this overestimation of the mean absorbed doses cannot interfere with our conclusions.
Thirdly, the distribution of internalizing and non-internalizing 212 Pb-mAbs within tumors and subsequent absorbed dose distribution at the organ scale would deserve to be further investigated since mean absorbed dose does not allow taking into account heterogeneity in dose distribution that could be associated to higher therapeutic effects per Gy of internalizing 212 Pb-mAbs.
Finally, the discrepancy between absorbed doses and biological effects could also be due to biological phenomena. For instance, while anti-CEA mAbs do not interact with any identified signaling pathways, unlabeled anti-HER2 mAbs are known to block the cells in G1 phase of the cell cycle and may down regulate HER2 receptors and disruption of receptor dimerization and signaling through the downstream PI3K cascade [37]. Although, the tested amount (40 mg) of unlabeled mAb did not have any direct impact on tumor growth, we cannot exclude any synergetic effect between trastuzumab and 212 Pb irradiation that may influence the final outcome of therapy. This and the contribution of bystander effects need to be assessed in further studies.
It must also be kept in mind that calculating accurately the uncertainty associated to mean absorbed dose values is a tedious task since it combines several sources of uncertainties: uptake of radioactivity at each time point, final cumulated uptake of radioactivity, tumor and organ masses, and S-values. Such uncertainties were not calculated in the present study and statistics about differences between calculated mean absorbed doses could therefore not be established. Our study confirms the strong efficacy of RIT with 212 Pb-mAbs in animal models of cancer. When 212 Pb-mAbs were tested in previous preclinical studies [30,31,34,35,53], TCMC was used as chelator only in three works [30,31,35] and tumors were targeted with trastuzumab. Thus, Milenic et al. reported that in mouse models of pancreatic and colorectal peritoneal cancer [30] MS increased from 19 days (sham) to 56.5 days after ip injection of 0.74 MBq 212 Pb-trastuzumab. However, no difference was observed between mice treated with 0.48 and 0.74 MBq suggesting than the lowest activity could then be used. Tan et al. demonstrated that one intravenous injection of 0.74 MBq 212 Pbtrastuzumab delayed tumor growth without significant toxicity in an orthotopic model of human prostate tumor. In our study, we observed, as mentioned above, an absorbed dose dependent effect for each radiolabeled mAbs considered alone and non-internalizing 212 Pb-35A7 were shown to be more effective after injection of 0.37 MBq than the 2 other radiolabeled mAbs used at the same activity. Indeed, 0.37 MBq 212 Pb-trastuzumab (MS = 24 days versus 11 in controls) was less effective than 0.37 MBq 212 Pb-35A7 (MS = 42 days). This lack of efficacy could be artefactual because the two Kaplan Meyer survival curves were rather similar ( Figure 2) with sacrifice of the last mice at day 75 post-graft in both groups.
Moreover, if it was shown in previous studies that high activity (1.48 MBq) of irrelevant 212 Pb-mAbs was accompanied by significant toxicity [30] and was associated with some therapeutic effects [31], our study indicated that the MS of mice treated with the irrelevant 212 Pb-PX mAb, unlabeled mAbs or NaCl were not statistically different, demonstrating the lack of effect of nonspecific irradiation in this tumor model.
The high efficacy of both anti-HER2 and anti-CEA 212 Pb-mAbs in our study may be explained by the nature of the tumor cells, or more likely by the tumor volume at the time of treatment. We treated 10 mm 3 tumors while volumes ranged from 15 mm 3 [30] to 100 mm 3 in [35] in previous studies. Our results support the idea that RIT of solid tumors should be dedicated to small volume tumors, such as peritoneal carcinomatosis that can originate from CEA-positive ovarian or digestive tumors. Although anti-HER2 212 Pb-mAbs were the most efficient, the MS of mice treated with 1.48 MBq anti-CEA 212 Pb-mAbs was also strongly improved (94 days versus 18 days for NaCl-treated controls), supporting the hypothesis that intraperitoneal RIT with anti-CEA 212 Pb-mAbs could be an alternative to HIPEC in digestive cancers expressing CEA.
To date, clinical alpha particle RIT has been investigated in patients with hematological malignancies [54,55], ovarian [56], melanoma [57] or brain tumors [58]. Our study gives preclinical rationale for the ongoing Phase I study (NCT01384253) as the toxicity was less than previously reported suggesting the potential for dose escalation with acceptable toxicity and a higher therapeutic efficacy.

Conclusion
We have shown that internalizing anti-HER2 212 Pb-mAbs targets antigen expressing xenografts and are more efficient per Gy than non-internalizing anti-CEA 212 Pb-mAbs in reducing and eradicating tumor growth. Treatment was associated with only transient and tolerable hematologic toxicity. This preclinical data gives support to proceeding with clinical trials with this RIT agent.