Bispecific Antibody Pretargeting of Tumor Neovasculature for Improved Systemic Radiotherapy of Solid Tumors

Radioimmunotherapy, the systemic administration of radioisotope-labeled tumor-targeting monoclonal antibodies (MAb), has been successful as a treatment for certain types of non–Hodgkin's lymphoma but has had little success in the treatment of most solid tumors (1). This is due to the fact that, at maximally tolerated doses, radioimmunotherapy is not able to deliver a sufficiently high enough radiation dose to have a significant effect on the growth of solid tumors especially when applied as a monotherapy. In radioimmunotherapy, the hematologic toxicity caused by the high and long-lasting blood activities of MAbs is dose limiting. This is due to the continuous source of radiation to the highly radiosensitive red marrow. To overcome this limitation of directly radiolabeled IgG MAbs, enzymatic fragmentation or genetic engineering has been used to create constructs that clear rapidly from the blood and achieve maximum tumor accumulation more rapidly than a full-size MAb. Although the tumor-to-blood ratios can be improved significantly with these constructs because of their faster clearance from the blood, both the total amount of radioactivity delivered to the tumor as well as the tumor residence time are decreased (2). Thus, the advantage of being able to administer higher activities is offset by a decreased total tumor dose. This is especially true when MAbs, fragments thereof, are radiolabeled with radiometals or lanthanides because the bifunctional chelators needed for labeling become trapped in the proximal tubules of the kidney, leading to unwanted and high radiation doses to this organ that often are higher or equal to the tumor doses achieved (3).

One approach to overcome the limitations posed by direct radiolabeling of MAbs or their smaller fragments/constructs is the use of pretargeting (4, 5). The pretargeting systems solve the slow blood clearance problem by attaching the radionuclide to a very small compound that is characterized by rapid body elimination via renal clearance (∼80% of the injected activity excreted in the urine within 1-2 hours in mice). Moreover, these small molecules are able to quickly extravasate and diffuse to their tumor targets very efficiently. Because of their rapid diffusion and kidney elimination, these small molecules need to be very tumor selective and their specific targeting needs to occur with a high affinity, which can be achieved, for example, by making use of biotin/avidin binding, hapten/antibody binding, or the complementary oligonucleotide recognition systems (6). In this study, we used a bispecific Mab (bsMAb), consisting of one binding moiety specific for the angiogenesis-associated target, extra domain B (ED-B) fibronectin, and the other specific for a small hapten that can be radiolabeled either with ¹¹¹In or ⁹⁰Y after attachment of a bifunctional chelator. The unlabeled bsMAb is given first and allowed to accumulate in the tumor. Subsequently, the rapidly clearing radiolabeled hapten is injected. The interval between the bsMAb targeting and the subsequent injection of the radiolabeled compound is dependent on the blood clearance properties of the bsMAb and has to be optimized. If given too soon, a sizable fraction of the radiolabeled hapten-peptide binds to the bsMAb in the blood, slowing its clearance from the circulation. This type of pretargeting also benefits from the use of bivalent hapten-peptides, which greatly increase tumor uptake and retention, a mechanism known as the “affinity enhancement system” (7). High cure rates can be achieved by pretargeted radioimmunotherapy in mice bearing human colon carcinoma xenografts, significantly improving therapeutic responses if compared with directly radiolabeled IgG or F(ab′)₂ (8–13). In addition, Kraeber-Bodere et al. (14) have shown that repeated pretargeting applications increase the therapeutic efficacy without increasing its systemic toxicity in a human medullary thyroid cancer mouse model and that pretargeted radioimmunotherapy can be used with paclitaxel combination therapy to further improve efficacy in this animal model.

The fibronectin splice variant B (ED-B) is considered a pan-species angiogenesis-associated tumor matrix marker that is expressed in close proximity to the tumor neovasculature and in the stroma of a large variety of hyperplastic and neoplastic tissues. It is also expressed in healthy tissues during angiogenic processes (15). The completely human anti-ED-B single-chain Fv (scFv) MAb fragment L19 with a subnanomolar affinity has been generated (16). Derivatives of scFv L19 have been applied successfully in different strategies for tumor imaging and therapy in both animal models (17–19) as well as in patients for immunoscintigraphic detection of primary tumors and their metastases (20). ED-B is considered to be a promising target for radioimmunotherapy of solid tumors because, being a matrix antigen expressed in close vicinity of new tumor vessels, it is rapidly accessible via the circulation, and its relationship to the angiogenic processes in the tumor might facilitate selective destruction of tumor vasculature (21).

Berndorff et al. (21) recently examined three derivatives of the L19 anti-ED-B MAb (i.e., a 50,000-Da scFv, the 75,000-Da small immunoprotein (SIP), and the 150,000-Da IgG) radiolabeled with ⁹⁰Y or ¹³¹I and determined that ¹³¹I-L19-SIP had the most favorable biodistribution properties and superior antitumor effects compared with the other radiolabeled derivatives investigated. L19-SIP is a divalent scFv fusion protein encoding an IgE C_H4 domain for dimerization (21). To investigate the potential of the pretargeting system in combination with the ED-B tumor matrix marker, scFv AP39 (derived from scFv L19) was used to form a bsMAb with the murine anti–histamine-succinyl-glycine (HSG) 679 Fab′ (AP39xm679; Fig. 1). As a hapten, ¹¹¹In-labeled HSG was used. For subsequent in vivo characterization, the human glioblastoma xenograft model U251 was chosen. In this tumor model, ED-B expression can be detected in the perivascular space as well as in the stroma.⁶

Antibody derivatives. scFv AP39 was generated by insertion of the peptide sequence (Gly)₃-Cys-Ala at the COOH-terminal end of the V_L chain of the scFv L19. Expression in Pichia pastoris and product purification were done according to Cunha et al. (22). Briefly, the AP39-encoding DNA sequence was cloned into the P. pastoris expression vector pPIC9K (Invitrogen, Carlsbad, CA). This construct contains a methanol-inducible promoter (AOX1) and a signal sequence (from yeast α factor) for expression and secretion of the recombinant product. Stably transfected yeast clones (based on strain GS115) were established by electroporation and subsequent antibiotic selection. The clones were cultivated at 30°C in buffered minimal glycerol-complex medium or basal mineral medium, and methanol was added for promoter induction during the expression phase. The recombinant product had a correctly processed terminus and high antigen-binding activity. AP39 was purified from culture supernatant by protein A affinity chromatography and size exclusion chromatography followed by Tris (2-carboxy-ethyl)-phosphin-HCl treatment (for reduction of covalent dimers).

The purified AP39 forms a noncovalent dimeric scFv molecule (M_r ∼50,000 Da). L19-SIP is a 75,000-Da SIP that was expressed in mammalian cells and purified by immunoaffinity chromatography as described earlier (23). Different L19-based MAb formats have been shown to have similar binding affinities when measured in solution to avoid avidity effects, but when the antigen is bound to a solid phase, these bivalent L19-based formats show better performance due to slow disassociation rates (24).

The murine anti-HSG IgG (m679) was purified from tissue culture medium by protein A and ion exchange chromatography. The F(ab′)₂ was prepared by pepsin digestion. The AP39xm679 bsMAb was made by cross-linking reduced sulfhydryl groups with N,N′-1,2-phenylenedimaleimide (PDM; Aldrich, Milwaukee, WI). Briefly, AP39 and 679 F(ab′)₂ were both reduced with 2 mmol/L Tris (2-carboxy-ethyl)-phosphin-HCl (Pierce, Rockford, IL) in 0.1 mol/L phosphate buffer (pH 7.4) for 30 minutes at 37°C. The reduced proteins (AP39-SH and 679 Fab′-SH) were dialyzed into coupling buffer [50 mmol/L sodium acetate, 0.5 mmol/L EDTA (pH 5.0)]. AP39-SH was then derivatized with a 500-fold molar excess of PDM by adding an aliquot of 20 mmol/L PDM (prepared in 90% N,N′-dimethylforamide) to a final concentration of 4 mmol/L. After 30 minutes at room temperature, the reaction mixture was dialyzed into coupling buffer to remove unreacted PDM. AP39-PDM was then combined with 679 Fab′-SH at a molar ratio of 1.5:1 and, after 90 minutes at room temperature, was quenched with 10 mmol/L cysteine. The solution was adjusted to pH 7.0 with Na₂HPO₄.

The resulting AP39xm679 Fab conjugate was purified using two affinity chromatography steps. Affinity chromatography with IMP-291-Affigel (25), which binds 679 MAb, removed unreacted AP39-PDM. The IMP-291-purified material was further purified using protein A agarose, which specifically binds AP39 but not 679 Fab′ and removes unreacted 679 Fab′. Purity of the material was determined by size exclusion high-pressure liquid chromatography (HPLC) and SDS-PAGE, whereas the bispecific binding of the AP39xm679 bsMAb was determined by BIAcore analysis.

Radiolabeling of antibodies and HSG-hapten-peptide. Radioiodination was done following a slightly modified version of the original Iodo-Gen method (26) using Iodo-Gen precoated iodination tubes (Pierce). Briefly, 200 μg (L19-SIP) or 280 μg (AP39xm679 bsMAb) of protein in 500 μL PBS were labeled with 0.5 mCi of Na ¹²⁵I (Amersham Biosciences, Little Chalfont, United Kingdom). After incubation at room temperature for 10 minutes, the radiolabeled protein was separated from free radioiodine by gel filtration using NAP5 columns (Amersham Biosciences, Upssala, Sweden) that were pretreated with 0.5% bovine serum albumin in PBS and equilibrated and eluted with PBS.

Each radiolabeled protein was tested for purity and immunoreactivity before its use. Size exclusion HPLC was done using a TSK-gel column SWxL, 300 × 7.8 mm (Tosoh Haas, Montgomeryville, PA), with TSK buffer (0.1 mol/L Na₂HPO₄, 0.1 mol/L Na₂SO₄, 0.05 mol/L NaN₃) as the mobile phase and a flow rate of 0.5 mL/min.

The immunoreactivity of the radioimmunoconjugates was determined using a test column composed of a Pasteur pipette filled with 250 μL ED-B Sepharose resin (27) and saturated with 0.5% bovine serum albumin in PBS. The test column was loaded with ∼0.2 MBq radiolabeled protein in 500 μL of 0.5% bovine serum albumin in PBS and rinsed with 2 × 500 μL of 0.5% bovine serum albumin in PBS to remove the nonimmunoreactive fraction. The immunoreactive fraction was then eluted with 2 × 750 μL of aqueous triethylamine solution (0.1 mol/L). The radioactivity of the fractions and that remaining on the column was measured with a dose calibrator. The immunoreactive fraction was expressed as the percentage radioactivity bound to the column. After purification, the concentration of AP39xm679 bsMAb was determined by measurement of the absorbance of the sample with an UV spectrometer, and the specific activity needed for the biodistribution studies was adjusted by addition of nonlabeled protein.

The hapten-peptide used in these studies was the divalent HSG-1,4,7,10-tetra-azacylododecane N,N′,N′,N′-tetraacetic peptide IMP-241 (28) provided by Immunomedics, Inc. (Morris Plains, NJ). The hapten-peptide was dissolved in 0.5 mol/L ammonium acetate (pH 5.5) to a concentration of 2.0 × 10⁻³ mol/L and radiolabeled with ¹¹¹InCl₃ (Amersham Health, Little Chalfont, United Kingdom) according to Sharkey et al. (28). Unbound ¹¹¹In was measured by instant TLC and done using a water-ethanol-ammonia mixture (5:2:1) as a mobile phase. Additionally, reversed-phase HPLC was done using a Zorbax Bonus-RP column, 250 × 6 mm (Tosoh Haas), with an acetonitrile plus 0.1% trifluoroacetic acid/water plus 0.1% trifluoroacetic acid gradient (0-45% organic phase in 15 minutes) and a flow rate of 1.5 mL/min. The ¹¹¹In-labeled hapten-peptide solution was used for the biodistribution studies without further purification.

Animals. Female nude mice (NMRI-Foxn1^nu) were obtained from Taconic Europe (Bomholtgard, Denmark), and tumors were induced by inoculation of 2.2 × 10⁶ human glioblastoma (U251) cells s.c. into the right hind limb. Biodistribution experiments were carried out after 20 to 28 days when tumors reached a mass of 50 to 300 mg. All animal experiments were done in compliance with the German law on the protection of animals.

Biodistribution studies. Biodistributions of nonpretargeted ¹²⁵I-labeled bsMAb (∼1.1 μg, 61 kBq) and ¹²⁵I-L19-SIP (∼0.45 μg, 48.5 kBq) were obtained at the indicated time points after i.v. injection of the radiolabeled MAb in a total volume of 0.1 mL. At the designated times after injection, animals were anesthetized and then sacrificed by decapitation. The dissected organs and collected excretions were counted for radioactivity in a gamma counter (Compugamma, LKB Wallac, Turku, Finland), and values of the percentage of injected dose per gram (% ID/g) and % ID were calculated.

For pretargeting studies, mice were injected i.v. with 1.5 × 10⁻¹⁰ mol (12 μg, 222-444 kBq ¹²⁵I) of the bsMAb (molecular weight assumed to be M_r 77,000 Da) in 0.1 mL followed 25 or 41 hours later with an i.v. injection (0.1 mL) of ¹¹¹In-labeled IMP-241 (1.5 × 10⁻¹¹ mol, 296 kBq). After the animals were anesthetized and sacrificed by decapitation (at the indicated time points), the dissected organs and collected excretions were counted for radioactivity as described above for the nonpretargeted studies.

Mouse dosimetry. Self-to-self doses for mouse organs were estimated based on medical internal radiation dose (29) using the biodistribution results of the labeled MAb and peptide. Using ¹²⁵I and ¹¹¹In as surrogates, dosimetric calculations were made for the corresponding therapeutic isotopes ¹³¹I and ⁹⁰Y as described earlier (21). About the dose-limiting organ, which was either the red marrow (directly targeted L19-SIP) or the kidneys (pretargeted peptide-hapten), maximum activities to be injected per mouse were determined based on published human tolerance doses of TD_5/5 = 2.5 Gy for the red marrow and TD_50/5 = 20 Gy for the kidneys (30). These maximum tolerated activities per mouse were used to calculate absolute tumor doses (in Gy) for a 100 mg normalized tumor mass in relation to the respective dose-limiting organ (therapeutic index).

Characterization of the AP39xm679 bsMAb. The affinity-purified preparation of AP39xm679 bsMAb was shown by size exclusion HPLC (Fig. 2A) to consist of a major peak at 8.89 minutes corresponding to the expected molecular size of ∼75,000 Da, an ascending shoulder at 8.11 minutes, and a minor peak at 10.36 minutes. A buffer peak at 11.44 minutes was also observed. Surface plasmon analysis (BIAcore) indicated that the purified AP39xm679 contained binding specificities for both HSG and ED-B (Fig. 2B and C).

Radiolabeling of antibodies and HSG-hapten-peptide. The radiochemical yield of the radioiodinations ranged between 60% and 90%, and the radiochemical purity, as determined by size exclusion HPLC, was always >80% with <5% of free iodide. For all iodinated proteins, the immunoreactivity was >80%. The radiochemical purity of the ¹¹¹In-labeled hapten-peptide, as determined by instant TLC and reversed-phase HPLC, was always >90% with no free ¹¹¹In in the samples.

Biodistribution of ¹²⁵I-labeled AP39xm679 bsMAb and L19-SIP in tumor-bearing mice. The biodistribution and pharmacokinetic data for the ¹²⁵I-labeled AP39xm679 bsMAb compared with the L19-SIP in tumor-bearing mice are shown in Tables 1 and 2 Both products cleared rapidly from the blood, with ∼1.3% ID/g in the blood 1 day after their injection and with maximum tumor accumulation occurring at 4 hours after injection and with similar uptake (averaging between 17% and 20% ID/g) and retention. These data suggest that the bsMAb retained its ability to target ED-B as well as L19-SIP did, with similar pharmacokinetic properties.

The different specific activities of the radiolabels of the ¹²⁵I-AP39xm679 used for the initial investigation of biodistribution and for monitoring of the bsMAb in the pretargeting experiments did not significantly affect biodistribution and tumor uptake (data not shown).

AP39xm679-pretargeted ¹¹¹In-hapten-peptide. Previous experience with bsMAb pretargeting systems, including bsMAbs based on the 679 anti-HSG MAb, suggested that optimal pretargeting results can be obtained when (a) 10-fold more moles of the bsMAb are given compared with the moles of radiolabeled hapten-peptide to be injected, (b) the hapten-peptide should be radiolabeled at its highest specific activity, and (c) the blood concentration of the bsMAb should be ≤1.0% ID/g before the radiolabeled hapten-peptide is given (31). The blood clearance data for the ¹²⁵I-AP39xm679 bsMAb indicated that the ¹¹¹In-hapten-peptide should be given between 24 and 48 hours after the bsMAb injection (Table 1). Therefore, two different groups of mice were examined, one where the ¹¹¹In-hapten-peptide was given at 25 hours and the other at 41 hours. In both cases, the molar ratio of bsMAb to peptide was 10:1. Biodistributions in tumor-bearing mice were carried out up to 168 hours after injection of the radiolabeled hapten-peptide.

The tumor and tissue uptake and clearance for the pretargeted ¹¹¹In-hapten-peptide using the two intervals tested, along with the data for the ¹²⁵I-L19-SIP, are shown in Tables 2 and 3A and B as well as in Fig. 3. The variation of the clearance time did not affect biodistribution and tumor targeting of ¹¹¹In-hapten-peptide. Moreover, tumor uptake of the ¹¹¹In-hapten-peptide remained in the tumors at a more sustained level than that seen with the directly radiolabeled ¹²⁵I-L19-SIP (Tables 2 and 3A and B; Fig. 3A). The ¹¹¹In-hapten-peptide was efficiently cleared from the body through the urine, whereas only a small fraction was found in the feces (Tables 3A and B and 4).

In addition to the uptake seen in tumors, the ovaries and the uterus showed an increased uptake of pretargeted ¹¹¹In-hapten-peptide (Table 3A and B). Similarly, both the bsMAb and the L19-SIP showed an enhanced uptake in these tissues (Tables 1 and 2).

In contrast to pretargeting settings, the uptake of ¹¹¹In-hapten-peptide alone in tumor and normal tissue (including ovaries and uterus) was rather low (Table 4).

Dosimetry calculations. Based on the results from the biodistribution experiments with AP39xm679-pretargeted ¹¹¹In-hapten-peptide and ¹²⁵I-L19-SIP, the self-to-self organ doses in mice were estimated for the respective therapeutic radioisotopes ⁹⁰Y and ¹³¹I, respectively. The highest calculated dose levels can be observed in tumor as well as in ovaries and uterus. Pretargeting the tumors with AP39xm679 using either a 25-hour or a 41-hour clearance time yielded a much higher radiation dose to the tumors if compared with the directly targeted L19-SIP (2,568 and 2,278 mGy/MBq for the 25-hour and 41-hour clearance times, respectively, versus 939 mGy/MBq for L19-SIP). The increased dose levels in both uterus and ovaries are consistent with known ED-B expression in these organs (32). The radiation doses for the other investigated tissues and organs were comparable in the pretargeting and the directly radiolabeled L19-SIP groups (data not shown).

In contrast to ¹³¹I-L19-SIP, where the red marrow was the dose-limiting organ when related to the human tolerance dose (TD_5/5 = 2.5 Gy), for the pretargeted ⁹⁰Y-hapten-peptide the kidneys were the dose-limiting organ (TD_5/5 = 20 Gy). It was calculated that 59 MBq (25-hour clearance interval) and 57 MBq (41-hour clearance interval) of AP39xm679-pretargeted ⁹⁰Y-hapten-peptide could be injected per mouse to achieve this maximally tolerated kidney dose (Fig. 4). This maximum tolerable activity would result in a tumor dose of 146 and 134 Gy, respectively. For ¹³¹I-L19-SIP, the corresponding maximum tolerable activity per mouse was 48 MBq, resulting in a tumor dose of 45 Gy (Fig. 4). The ratio between the U251 tumor (100 mg) and the dose-limiting organ was defined as the therapeutic index and was higher for ¹³¹I-L19-SIP (index value, 18) if compared with the AP39xm679-pretargeted ⁹⁰Y-hapten-peptide (index value, ∼7 for both investigated clearance intervals).

In this report, the radiotherapeutic potential of two different techniques, (a) direct targeting of the radioiodinated, divalent L19-SIP and (b) a two-step pretargeting method, was compared in a human glioblastoma xenograft mouse model. Both MAb constructs target ED-B fibronectin, a stromal tumor marker strongly associated with neovascularization. A novel bsMAb pretargeting system that uses an anti-HSG-HSG-peptide secondary recognition system was used because this is highly flexible and has the potential to target a variety of compounds (28). Because bsMAb can be fully humanized, these pretargeting agents would be less immunogenic than those using streptavidin. In most instances reported in the literature, bsMAbs are commonly prepared from Fabs or IgGs, but our results clearly indicate that chemical combinations of a Fab′ and a modified scFv (using an additional unpaired cysteine residue) can lead to the efficient generation of a functional bsMAb. Importantly, this bsMAb showed high tumor retention over a 3-day period, while at the same time quickly clearing from the blood and nontumor tissues. Good tumor retention and the fast blood clearance are highly favorable properties for a bsMAb to be used as a pretargeting tool.

¹¹¹In-labeled HSG-peptide, when used for pretargeting, showed a rapid and high (∼17% ID/g) tumor uptake that rivaled that of the ¹²⁵I-labeled L19-SIP, but tumor retention was longer with the pretargeted peptide. The enhanced retention most likely reflects the use of a radiometal-labeled pretargeted peptide, whereas the directly radiolabeled MAb was radioiodinated. Radiometal-labeled MAbs frequently have a longer retention in tumors than radioiodinated forms because radioiodinated MAbs can be dehalogenated, or when internalized and catabolized, radioiodine will be quickly eliminated from the cell, whereas radiometals are retained (33). ¹³¹I-L19-SIP was used because earlier studies had indicated that renal uptake of ¹¹¹In-L19-SIP was ∼2-fold higher than tumor uptake for this agent, with dosimetry data indicating unfavorable tumor-to-kidney ratios for ⁹⁰Y-L19-SIP (21). Thus, ¹³¹I-L19-SIP was found to be the best therapeutic agent compared with ¹¹¹In-L19-SIP and two other L19 MAb formats radiolabeled with ¹³¹I or ¹¹¹In/⁹⁰Y. In contrast, the pretargeting procedure used in this study has the advantage of being receptive to a variety of radionuclides, having been studied already with ¹³¹I, ^99mTc, ¹¹¹In, ⁹⁰Y, and ¹⁷⁷Lu (8, 28, 34), and in each instance, excellent tumor-to-nontumor ratios, even tumor-to-kidney ratios, were reported.

Previous studies suggested that the concentration of bsMAb in the blood needed to be ≤1.0% ID/g to facilitate optimal pretargeting (31). Therefore, we investigated two different clearance intervals in the U251 tumor model, 25 and 41 hours, which, based on the clearance of the AP39xm679 bsMAb, suggested blood concentrations of approximately 1.4% and 0.4% ID/g (i.e., at 24 and 48 hours, respectively). Regardless of the interval at which the ¹¹¹In-hapten-peptide was given, tumor uptake and the total amount of radioactivity delivered to the tumor at an estimated maximum tolerated dose were similar. This most likely reflected the relative stability of this bsMAb in the tumor at these two intervals, allowing for the optimal capture of the bsMAb in the tumor, whereas the bsMAb concentration in the blood was sufficiently low, even at 25 hours, to allow the peptide to clear from the blood relatively unabated.

In addition, there was also high uptake in several normal tissues, such as the uterus and ovaries. This targeting of radioactivity to uterus and ovary is consistent with the described ED-B expression in these organs (32). However, polyovulation in mouse ovaries is not comparable with the human situation and is therefore not considered a dose-limiting organ.

The use of medical internal radiation dose software-based calculations of radiodosimetry from biodistribution experiments is widely used to estimate the radiation doses delivered to different organs of interest. However, dosimetry based on medical internal radiation dose is limited because it uses several simplified assumptions, such as homogenous organ and tumor distribution of labeled compounds (29). Nevertheless, Berndorff et al. (21) showed the predictive power and the applicability of medical internal radiation dose dosimetry calculations in a tumor model by targeting ED-B fibronectin with different radiolabeled immunoconjugates (L19 scFv, L19-SIP, and L19-IgG) and were able to confirm the dose-limiting organ and corresponding maximum tolerated dose predicted by medical internal radiation dose based on myelotoxicity studies. According to this procedure, we identified the kidney as the dose-limiting organ as well as the corresponding maximum tolerated dose for AP39xm679-pretargeted ⁹⁰Y-hapten-peptide. In accordance, Karacay et al. (13) recently showed that, in a similar pretargeting system using a ⁹⁰Y-labeled peptide, renal toxicity was dose limiting. In our pretargeting approach, the calculated tumor dose at TD_5/5 for AP39xm679-pretargeted ⁹⁰Y-hapten-peptide was 146 Gy, which is 3-fold greater when compared with the TD_5/5 tumor dose of the directly targeted ¹³¹I-L19-SIP (45 Gy). It should also be noted that ⁹⁰Y has a greater energy emission than ¹³¹I (β; E_max = 2.27 MeV versus β; E_max = 0.81 MeV, respectively) and a shorter half-life (t_1/2 = 2.67 days versus t_1/2 = 8.06 days, respectively). In addition, ⁹⁰Y will deposit its energy out to 1.81 cm versus only 0.55 cm for ¹³¹I (35), thus making it a more desirable radionuclide for the treatment of solid tumors. In fact, when directly compared at equal radiation doses delivered to a human pancreatic adenocarcinoma in mice, ⁹⁰Y was shown to be superior to ¹³¹I in terms of therapeutic efficacy of these solid tumors (36).

In the current investigation, the advantage of pretargeting becomes most obvious when compared with ¹¹¹In-labeled L19-SIP, which has been investigated in F9 teratocarcinoma-bearing mice (21). We have observed that ED-B-based targeting in the F9 tumor model is comparable with the U251 tumor model (data not shown). However, at the TD_5/5, ⁹⁰Y-labeled L19-SIP delivers only 7 Gy of ⁹⁰Y to the F9 tumor, whereas in the current investigation we showed that, by using AP39xm679 pretargeting, a 21-fold higher ⁹⁰Y radiation dose to the U251 tumor could be achieved (i.e., 146 Gy). It seems that pretargeted radioimmunotherapy would be superior for the treatment of ED-B-expressing solid tumors when compared with ¹³¹I-L19-SIP, which was previously identified as the best-suited, directly labeled, immunoconjugate for radiotherapy (21).

In conclusion, we have shown that an AP39xm679 bsMAb platform to pretarget a subsequently given radiolabeled hapten-peptide to ED-B-expressing tumors is substantially superior to using a directly radiolabeled anti-ED-B L19-SIP. These data clearly indicate that this pretargeting system with a ⁹⁰Y-labeled hapten-peptide, with its advantageous properties (e.g., long-range radiation and cross-fire effect), is a solid basis for developing future strategies for pretargeted radioimmunotherapy of solid tumors.

We thank Melanie Appel, Guel Caglar, Joerg Pioch, Volker Stickel, Ingolf Weber, and Simone Zolchow for excellent technical assistance with regard to the radiolabeling, cell culture, and biodistribution studies; John Kopinski and Li Zeng for help in the preparation and characterization of the bispecific conjugates; and Dr. Guido Malawski for purification of scFv AP39.