Reagents and Methods for PET Using Bispecific Antibody Pretargeting and ⁶⁸Ga-Radiolabeled Bivalent Hapten-Peptide-Chelate Conjugates

MATERIALS AND METHODS

⁶⁷GaCl₃ in 0.1N HCl was obtained from Nordion, ¹¹¹InCl₃ was purchased from IsoTex Diagnostics, and sodium (¹²⁵I) iodide was bought from Perkin-Elmer Life Sciences Inc. An ionic ⁶⁸Ge/⁶⁸Ga generator was purchased from DuPont Radiopharmaceuticals and eluted with 0.5N–1.0N solutions of ultrapure HCl prepared using OPTIMA grade concentrated HCl (Fisher Scientific) and Millipore 18-MΩ water. All radiolabelings were performed in 3N HCl acid-washed and Millipore water-rinsed CZ resin vials (West Pharmaceutical Services). Radiolabeled products were analyzed on a size-exclusion high-performance liquid chromatography (SE-HPLC) GF-250 column (Bio-Rad) attached to a Millennium system (Waters Corp.) equipped with a Packard FLO-ONEβ-in-line radiomatic detector. The SE-HPLC columns were equilibrated in 0.2 mol/L sodium phosphate buffer, pH 6.8, and samples were run at 1 mL/min.

The peptide designated IMP 241 [DOTA-Phe-Lys(HSG)-d-Tyr-Lys(HSG)-NH₂] (35) was synthesized using an Fmoc-based solid-phase strategy on an Advanced ChemTech automated peptide synthesizer. Briefly, IMP 241 was synthesized on a Sieber amide resin (2.1 g, 0.62 mmol/g) with the following amino acids (6 equivalents per coupling) added in the order shown: Fmoc-Lys(Aloc)-OH, Fmoc-d-Tyr(But)-OH, Fmoc-Lys(Aloc)-OH, Fmoc-Phe-OH, and DOTA-tris(t-butyl) ester (Macrocyclics, Inc.). A single diisopropylcarbodiimide (DIC) coupling overnight was used with the DOTA. Each amino acid was double-coupled for 1 h using DIC as the activating agent in the presence of N-hydroxybenzotriazole (HOBt), followed by a second 1-h coupling using HOBt, O-benzotriazole-N,N,N′,N′-tetramethyl-uronium-hexafluoro-phosphate and diisopropylethylamine. The Aloc side chains were removed with the Pd catalyst in the usual way (36) and the trityl-HSG-OH (6 equivalents) was double-coupled to the lysine side chains, as described above. The peptide was cleaved from the resin with trifluoroacetic acid, precipitated in ether, and purified by reverse-phase HPLC to obtain the desired peptide (ESMS MH+ 1471).

The preparation of a bsAb with specificities against carcinoembryonic antigen (humanized anti-CEA) and histaminyl-succinyl glycine (murine anti-HSG), prepared as Fab′ × Fab′ dimaleimido- chemically cross-linked F(ab′)₂ fragment, has been described (37). The bispecific diabody referred to as BS1.5H was constructed from the variable domains of hMN-14 and humanized variable domains of m679 using established methods (38). Briefly, V_H and V_K sequences of hMN-14 were amplified by polymerase chain reaction (PCR) from a vector that was constructed for expressing hMN-14 Fab′ (39). The V_H and V_K domains of m679 were amplified by PCR from a plasmid containing the complementary DNA sequences of m679 heavy and light chains (Z. Qu, unpublished data, 2000). Humanized versions of m679V_H and m679V_K were generated by PCR-based mutagenesis. The strategy used for humanization was to retain all CDR amino acid sequences and any framework residues known to interact with CDR regions. Amino acid residues of mouse framework regions that are not represented in the database of human frameworks were replaced with the most common amino acid found in human sequences at that position. Thirteen amino acid substitutions were made to humanize V_H and V_K of m679. The dicistronic expression cassettes for BS1.5H are shown in Figure 1. Chemically competent Escherichia coli BL21-pLysS cells (Novagen) were transformed with the expression plasmid for BS1.5H following the manufacturer’s recommendations. Transformants were plated overnight at 37°C on LB agar plates supplemented with kanamycin sulfate (100 μg/mL) and chloramphenicol (34 μg/mL). Transformed colonies were used to inoculate shaker flask cultures of Difco 2× YT broth (Becton Dickinson) supplemented with kanamycin sulfate (100 μg/mL) and chloramphenicol (34 μg/mL). Cultures were shaken at 37°C to an OD₆₀₀ of 1.6–1.8. An equal volume of room temperature 2× YT media supplemented with antibiotics and 0.8 mol/L sucrose was added to the cultures, which were then transferred to 20°C. After 30 min of shaking at 20°C, expression was induced by the addition of isopropyl thiogalactose to a final concentration of 40 μmol/L and the incubation was continued at 20°C for 15–18 h. Bacteria were pelleted by centrifugation at 10,000g. The cell pellets were frozen and thawed and then resuspended in lysis buffer (2% Triton X-100, 300 mmol/L NaCl, 10 mmol/L imidazole, 5 mmol/L MgSO₄, 25 units/mL benzonase, 50 mmol/L NaH₂PO₄, pH 8.0) using an amount equal to 1% of the culture volume. The suspension was homogenized by sonication and clarified by centrifugation at 40,000g. The soluble extracts were loaded onto Ni-NTA agarose (Qiagen, Inc.) columns using about 1.0 mL Ni-NTA resin per liter of culture. Columns were washed with a buffer containing 20 mmol/L imidazole and eluted with 250 mmol/L imidazole. The Ni-NTA eluate was loaded directly onto a Q-Sepharose (Amersham Pharmacia Biotech) anion-exchange column, and the flow-through fraction containing the products was collected. The yield of BS1.5H was about 0.5 mg/L of culture.

For a typical radiolabeling with ⁶⁸Ga, IMP 241 (11.4 μL of a 2.2 × 10⁻³ mol/L stock solution in 0.5 mol/L ammonium acetate, pH 4.0; 2.5 × 10⁻⁸ mol) was mixed with 1.0 mL of 2.5 mol/L ammonium acetate, pH 5.5, in an acid-washed 10-mL CZ-Resin vial. The vial was sealed with an acid-washed Flurotec stopper, crimped closed, and swirled to mix. The ⁶⁸Ge/⁶⁸Ga generator was then eluted directly into the IMP 241 vial using a nonmetallic catheter (VWR International), and vent needle, with 2 mL 1N HCl, after running the first milliliter of generator eluent to waste. The final pH of the labeling mixture was 4.0. The vial was heated at 95°C for 15 min. After cooling for 5 min, an aliquot was withdrawn and diluted to 1.85 MBq/mL with sterile saline for radioanalysis. Ten to 12 μL of the diluted radiolabeled peptide containing 18–22 kBq ⁶⁸Ga were mixed with a 20-fold molar excess m679 IgG and applied to the SE-HPLC column. Column recovery of radioactivity was estimated by collecting the entire eluate from the 20-min run and counting 3 × 1-mL aliquots against 3 × 1-mL aliquots of a standard prepared by diluting the same volume of labeled peptide in 20 mL elution buffer. Averages of the triplicate counts were taken and compared with estimate column recovery of applied radioactivity.

For a typical radiolabeling with ⁶⁷Ga, 200 μL 2N HCl (OPTIMA) were drawn up into the barrel of a 1-mL syringe and, with a vent needle in place, the acid was added to a crimp-cap vial of 104 MBq (2.8 mCi) ⁶⁷Ga chloride (Nordion) using a nonmetallic catheter. The vial was mixed by inversion and vortex and then spun in a clinical centrifuge at ∼3,000 rpm for ∼2 min, after which it was a left at room temperature for 1 h. In an acid-washed 1.8-mL Eppendorf vial, 500 μL 2.5 mol/L NH₄OAc, pH 6.9, was mixed with 11.4 μL of a stock solution of IMP 241 peptide (2.2 × 10⁻³ mol/L, 2.5 × 10⁻⁸mol). This diluted solution of IMP 241 in the acetate buffer was drawn into the barrel of a 1-mL syringe and added to the vial of ⁶⁷Ga through the nonmetallic catheter. The vial and contents were then heated at 95°C for 15 min to effect radiolabeling. Radioanalyses were performed as described above for the ⁶⁸Ga analog. ¹¹¹In-IMP 241 was prepared as described previously (35).

For purification of ⁶⁷Ga-IMP 241, a Waters Oasis HLB 1-mL extraction cartridge was fitted with a 1–1/2 in. (3.8 cm) 19-gauge needle and placed in a syringe shield fixed to a ring stand. The needle end was inserted through the septum of an empty crimp-cap 2-mL glass collection vial. Also inserted through the septum of the vial was a 1–1/2 in. 18-gauge needle attached to one end of a 24-in. (61.0 cm) double male monitoring line. The other end of the line was attached to a 3-way stopcock fitted to a 5-mL syringe. In this manner, liquid was drawn through the cartridge into the collection vial by use of vacuum. The cartridge was prepared for use by drawing through 500 μL methanol, followed by 500 μL water, and, finally, 500 μL 2.5 mol/L NH₄OAc, pH 6.9, buffer. The ⁶⁷Ga-IMP 241 labeling solution was drawn into a 3-mL syringe and applied to the top of the prepared cartridge. The labeling liquid was drawn onto and through the column and into the collection vial. This first vial contained nonpeptide-bound ⁶⁷Ga waste. A second empty collection vial was then attached and a wash volume of 1.0 mL 2.5 mol/L NH₄OAc, pH 6.9, was drawn through the column. The ⁶⁷Ga-labeled peptide was eluted into a third empty collection vial with 500 μL methanol and water, 1:1.

In vitro stability on gallium-labeled IMP 241 was assessed by dilution of an original ⁶⁷Ga-IMP 241 sample 10-fold in human serum and incubation at 37°C. Aliquots were withdrawn at time points from 30 min through 4 h, mixed with a 50-fold molar excess of m679 IgG, and applied to SE-HPLC for analysis. Column recoveries of applied radioactivity were determined.

All animal experiments were conducted in accordance with National Institutes of Health guidelines on the humane use of animals and with prior approval of the institutional animal care and use committee. In vivo targeting experiments were performed with the ⁶⁷Ga-IMP 241 analog, due to the 68-min half-life of the ⁶⁸Ga. Athymic nude mice bearing GW-39 human colon tumor xenografts in groups of 4 or 5 per time point were used. For an experiment with the chemically produced construct, animals were given an injection of 15 μg (1.5 × 10⁻¹⁰ mol) hMN-14 × m679 F(ab′)₂ bsAb, radioiodinated with ¹²⁵I (Perkin-Elmer) using the standard IODO-GEN method (Pierce Chemical Co.), 24 h before administration of the ⁶⁷Ga-IMP 241 peptide. Animals were then given 1.5 × 10⁻¹¹ mol of peptide, a 10-fold molar deficit of peptide (25,35), equivalent to about 92.5 kBq (∼2.5 μCi) ⁶⁷Ga per animal. Five animals were killed at each of 1, 2, 3, 4, and 6 h after injection, with tissues (tumor, blood, liver, kidney, spleen, lungs, stomach, small and large intestine, and bone) quantified for both ¹²⁵I and ⁶⁷Ga uptake, using dual-energy counting with cross-over correction. A separate series of animals was given the ⁶⁷Ga-IMP 241 alone, without bsAb pretargeting.

In a second pretargeting experiment in the same xenograft model system, the construct BS1.5H was administered first, followed by IMP 241 labeled with either ⁶⁷Ga or ¹¹¹In. We had previously established the preferred targeting and blood clearance time of the ¹²⁵I-BS1.5H before radiolabeled peptide administration at 15 h (data not shown). The same molar dose of the BS 1.5H (1.5 × 10⁻¹⁰ mol) as the chemical construct was given 15 h before the same 10-fold molar deficit of ∼ 92.5 kBq ⁶⁷Ga-IMP 241 peptide (1.5 × 10⁻¹¹ mol). Animals were killed at 1, 3, and 24 h in the ⁶⁷Ga-IMP 241 groups and at 3 and 24 h in the ¹¹¹In-IMP 241 groups, with the same range of tissues (except bone) taken and counted in appropriate energy windows for each nuclide.

RESULTS

The synthesis of the IMP 241 peptide on solid phase was routine. Interestingly, the presence of the protected DOTA did not interfere with the aloc group cleavage, which used a palladium catalyst. Also, there was no palladium left in the final purified product, so there was no inhibition of metal-DOTA binding due to nonremoved palladium metal.

The dicistronic expression cassette shown in Figure 1 encodes 2 heterologous polypeptides designed to pair with each other and form 1 functional binding site each for CEA and HSG via domain swapping (39). BS1.5H was obtained from the soluble fraction and purified by immobilized metal affinity chromatography using Ni-NTA followed by Q-Sepharose anion-exchange chromatography. The characterization of BS1.5H included the determination of purity by SE-HPLC and sodium dodecyl sulfate polyacrylamide gel electrophoresis, molecular size by SE-HPLC, binding affinity for CEA by competitive enzyme-linked immunosorbent assay, binding affinity for HSG by BIAcore, and binding valency for CEA by BIAcore.

At receipt, the nominal 740-MBq (20 mCi) ⁶⁸Ge/⁶⁸Ga generator contained 1.23 GBq (33.2 mCi) ⁶⁸Ge and eluted at 70.6% efficiency according to the manufacturer. That should have yielded 868 MBq (23.44 mCi) at first elution, but only 696 MBq (18.82 mCi) were obtained (yield ∼ 57%). With suitable in-growth time allowed, recoveries remained between 55% and 65% during >1 y of constant use and hundreds of elutions. Although 20 mL 1N HCl was recommended, equivalent yields were recovered using 10 mL, and little loss was evident even when using only 5 mL of eluent, since most ⁶⁸Ga eluted in the 2- to 4-mL volume fraction, as measured from the beginning of the elution time. In addition, the generator could usefully be eluted with 0.5N rather than 1.0N HCl, with only about 10% loss in recovered ⁶⁸Ga over what one would expect with 1.0N HCl (2).

A SE-HPLC radiomatic analysis of the ⁶⁸Ga-IMP 241 after mixing with the hMN-14 × 679 F(ab′)₂ bsAb is shown in Figure 2. Also shown, for in vitro stability, is a typical radiomatic SE-HPLC analysis of a trace taken after 4 h at 37°C. This trace is unchanged from the original SE-HPLC analysis of the ⁶⁷Ga-IMP 241-m679 IgG complex or from analyses taken at the earlier times from 30 min through 3 h. Essentially 100% of the detected activity elutes at the retention time of ⁶⁷Ga-IMP 241-m679 IgG, with column recoveries of applied radioactivity between 74% and 82% at each time interval tested.

In a targeting experiment with the chemically cross-linked bsAb trace-radiolabeled with ¹²⁵I, ¹²⁵I-hMN-14 × 679-F(ab′)₂ was given 24 h before ⁶⁷Ga-IMP 241 to nude mice bearing GW-39 tumors at specified times after injection (5 animals per group). Data for the bispecific protein are given in Table 1, and data for the ⁶⁷Ga-IMP 241 are given in Table 2. The biodistribution of ⁶⁷Ga-IMP 241 administered alone, without bsAb pretargeting, is in Table 3, with all data in Tables 1–3 given in terms of percentage injected dose per gram (%ID/g) of tissue (±SD). Biodistribution data for ⁶⁷Ga-IMP 241 from this experiment are reported in terms of tumor-to nontumor ratio in Table 4. Tumor uptake was between 8.3 and 16.0 %ID/g over the 6 h of the experiment and, at each time point and for every tissue, exceeded uptake in all normal tissues. Blood and kidney activities averaging 8.2 and 6.3 %ID/g at 1 h dropped to 2.1 and 3.1 %ID/g, respectively, at 2 h after injection. Residual activity at the later time points averaged around 3.5 %ID/g in the kidney and about 1.5 %ID/g in the blood out to 6 h after injection. Tumor uptake remained high throughout the duration of the experiment, indicating that targeting was specific and due to the prelocalized bsAb, evidenced by the lack of tumor uptake when injecting the ⁶⁷Ga-IMP 241 alone (Table 3). Tumor-to-normal tissue ratios are given in Table 4, and only tumor-to-kidney ratios are <5:1 from 2 h after injection onward.

In a second targeting experiment, using the construct BS1.5H for pretargeting, we were able to reduce levels of circulating bsAb still further than that achievable with the chemical construct, to 0.36 ± 0.07 %ID/g at 15 h after injection, which was chosen as the optimum time point for injection of the ⁶⁷Ga-IMP 241. In this experiment, ⁶⁷Ga-IMP 241 blood retention was 1.6 and 0.1 %ID/g at 1 and 3 h after administration, respectively, of the peptide, and at 16 and 19 h after administration of the BS1.5H. Tumor uptake was 7.9 and 13.2 %ID/g at 1 and 3 h, respectively (Table 5), leading to tumor-to-blood ratios of 41:1 and 137:1, at 1 and 3 h after administration of ⁶⁷Ga-IMP 241, respectively (Table 6). ⁶⁷Ga uptake in other normal tissues was close to zero and, in any event, ≤0.5 %ID/g, with the exception of the kidney, which had uptakes of 5.8 and 5.7 %ID/g at 1 and 3 h after injection, respectively.

DISCUSSION

Targeting with bsAbs followed by radiolabeled bivalent haptens offers a unique opportunity to produce an entirely new generation of radioimaging agents, advantageously coupling disease specificity with exceptional target-to-nontarget ratios of disease-localized radionuclide (26,35). A particular advantage is to be expected in the delivery of PET nuclides using this technology, since the short half-lives of ¹⁸F and ⁶⁸Ga make them incompatible with longer circulating primary targeting species such as intact antibodies. The major disadvantage is the added complexity of the approach, which involves preparation and dosing of 2 separate reagents and consequential timing and ratio issues. These are in addition to other aspects, such as making a judicious choice of reagents to take into development as well as solving more practical radiolabeling matters.

Regarding the bsAb component, the poor yields obtained when preparing Fab′ × Fab′ bsAbs with chemical combination techniques (often in the region of 20%–30%) have been known for >20 y, and a satisfactory process that overcomes this problem has not been developed to date. Also, this approach tends to result in <100% pure, fully structurally defined protein products, which is problematic for clinical development. Therefore, constructs that can be prepared in high yield in bacteria, yeast, or mammalian cells, in a fully defined structural manner at ≥95% purity, are preferred (26).

In an animal pretargeting study performed with the chemically cross-linked bsAb, the agent performed well in targeting the ⁶⁷Ga-IMP 241 specifically and maintaining it in the xenograft throughout the course of the experiment. The tumor-to-tissue ratios in Table 4 suggest that imaging with this agent from 2 h through 6 h after injection would be feasible. However, this F(ab′)₂ behaves in circulation and clearance as one would expect, and the residual bsAb remaining in blood (Table 1) serves to capture a relatively large amount (1.3–2.1 %ID/g between 2 and 6 h after injection; Table 2) of later injected ⁶⁷Ga-IMP 241, which itself shows no blood retention when given alone (Table 3). In turn, this translates into useful, but less than optimal, tumor-to-blood ratios of 6:1 to 12:1 between 2 and 6 h after injection of the ⁶⁷Ga-IMP 241.

We have the view that a bsAb used for 2-step imaging needs to use a species with an almost complete serum clearance in a reasonable time period after administration. We chose 24 and 15 h as time intervals (before injection of ⁶⁸Ga) for the bispecific chemical conjugate and the BS1.5H, respectively, based on previously determined optimum clearance times for each species. Rapid and complete serum clearance obviates the need for bsAb clearing agents or blocking agents, which would engender considerably more complexity regarding doses and timings (35). Without a clearing agent, and with less than optimal bsAb blood clearance, one would have a radiolabeled peptide displaying lower than ideal target-to-tissue ratios due to radiolabel capture by circulating bsAb. In this work, the BS1.5H construct has been shown to be more effective than the corresponding Fab′ × Fab′ chemically manufactured bsAb in targeting ⁶⁷Ga, when tested in the same xenograft model, primarily due to its more complete clearance, with <1 %ID/g in circulation at the time of ⁶⁷Ga-IMP 241 injection. Full details of BS1.5H properties will be published elsewhere. Despite the near-complete blood clearance, the BS1.5H retained strong tumor uptake, leading to outstanding tumor-to-blood ratios of 41:1 and 137:1 at 1 and 3 h after injection, respectively. As shown in Table 6, all tumor-to-normal tissue ratios were >40:1 by 3 h after injection, with the exception of the kidney, which contained 5.7–5.8 %ID/g at 1 and 3 h, respectively. Due to the much lower ⁶⁷Ga-IMP 241 uptake when given alone, this suggests that some renal capture by residual bsAb may be taking place, although some loss of ⁶⁷Ga from the complex may also be involved, since ¹¹¹In-IMP 241 renal uptake is lower in the same targeting system. In any event, these relatively low levels may not present much of a problem for imaging purposes.

As a primary requirement, the radiolabeled entity that is administered after the bsAb should be designed from the outset to have desirable biophysical properties. Ideally, it should not accumulate in any normal tissues and should be rapidly flushed via the urine, if not captured by pretargeted bsAb. ⁶⁷Ga-IMP 241, administered alone, deposited between 1.7 and 2.6 %ID/g in the kidney over the 6-h course of the experiment, with all other tissues having very low uptakes, which suggests at least adequate stability (Table 3).

For this bsAb targeting system, the most valid comparison for the data we obtained with that previously reported probably involves the work of Schuhmacher et al. (40). These investigators injected xenografted mice with an anti-MUC-1 × anti-HBED-CC-gallium F(ab′)₂ × Fab′ bsAb, followed by a dose of a Ga-HBED-CC-apotransferrin blocker, and then followed by the ⁶⁷Ga-HBED-CC complex in a carefully determined set of timings and dosages. They obtained tumor-to-blood, tumor-to-kidney, tumor-to-lung, and tumor-to-liver ratios of ⁶⁷Ga of around 2.5:1, 2.5:1, 3:1, and 6:1 at 3 h after injection, which they state are comparable with work by others for early time points after injection. Although these are good results, they do not approach the ratios reported herein, with tumor-to-blood, tumor-to-kidney, tumor-to-lung, and tumor-to-liver ratios of ⁶⁷Ga of around 137:1, 2.3:1, 41:1, and 106:1 at the same 3 h after injection time point when using BS1.5H pretargeting followed by ⁶⁷Ga-IMP 241. Reasons for the lower ratios observed previously in this general pretargeting area are multifactorial, with choice of antibody, antibody affinities, antigen-shedding tendencies, inadequate blood clearances, the added complexity inherent in 3-step systems, less than ideal radiolabels, use of radiolabeled agents that are too hydrophobic for excellent tissue (particularly renal) clearance, and use of monovalent rather than bivalent radiolabeled haptens as some of the factors involved.

Further improvement in the system as described here may be sought by making a bsAb that retains 2 binding sites for the target cell antigen and 1 for the radiolabeled moiety (40), but which retains the complete serum clearance properties, at some defined interval between 15 and 48 h after injection. This might be expected to increase the absolute amount of bsAb targeted and retained at the tumor. In reference to this point, it must be borne in mind that an ideal agent would also be capable of delivering a therapeutic radionuclide. In this respect, use of a bivalent CEA-binding bsAb may be preferred, and use of a minimally residualizing radiolabeled agent in the kidney may be mandatory. With regard to the chelating moiety itself, though a purist might claim that NOTA would be a better choice than DOTA as the chelate to ensure maximal ⁶⁸Ga stability, the stability of the ⁶⁷Ga-IMP 241 complex appears adequate for early time-point imaging, while use of the DOTA chelate enables an easier development path for IMP 241 analogs of indium, yttrium, lutetium, bismuth, and actinium and other similar metals of interest.

Successfully making a clinically useful set of reagents for radioimaging generally requires that radiolabeling processes remain relatively simple. This has not been true in the past for procedures involving ⁶⁸Ga, with generator-eluent drying, concentration, or ethereal extraction of highly dilute ⁶⁸Ga being developed to overcome radiolabeling problems. However, given the short-lived, high-energy emissions, these procedures are not practical for widespread adoption in a routine clinical setting. The ⁶⁸Ge/⁶⁸Ga generator used in this work was of the type first described by Loc’h et al. in 1980 (2), and our review of the original article suggested that handling and elution of the generator could be changed to obtain the ⁶⁸Ga in a smaller volume of less concentrated HCl. In turn, this might enable a facile labeling to be developed, with the only drawback being a small, and acceptable, reduction in elutable ⁶⁸Ga. It proved possible to obtain the ⁶⁸Ga in 2- to 4-mL volumes, allowing use of a kit containing the peptide and sufficient buffer to neutralize the 0.5N–1.0N HCl and thereby allow ⁶⁸Ga labeling of IMP 241 by direct elution of the generator.

Radiolabeling of the IMP 241 peptide with ⁶⁷Ga or ⁶⁸Ga could be accomplished in about 15 min at an elevated temperature with apparent yields of at least 80%, as analyzed by SE-HPLC of bsAb-bound radiogallium. Given the inconsistencies in measuring column recoveries in this complex system, the real incorporation may be >80%. For the animal studies, we purified ⁶⁷Ga-IMP 241 from possible contamination by other ⁶⁷Ga species, using an extraction cartridge process that took 5–10 min and gave the desired agent in an aqueous alcohol solution. Using ethanol in lieu of methanol, this certainly would not be incompatible with the 68-min half-life of ⁶⁸Ga in a clinic or radiopharmacy, although further process work designed to guarantee >95% incorporation and dispensing with the purification would be desirable. As with the SE-HPLC, recoveries from the cartridge as the desired radiolabeled peptide were again generally in the 80% range, with losses spread throughout the system (i.e., in tubing, other fractions, syringes, and so forth), most probably due to handling losses.

Handling and use of the ⁶⁸Ge/⁶⁸Ga generator was simple, with its reliability established by hundreds of elutions over the course of this work. The finding that the generator could be usefully eluted in much smaller volumes of HCl enabled us to devise reagents that might be adapted to a kit formulation for ready and easy ⁶⁸Ga labeling in the future. Three or 4 elutions of the generator are readily made in any 6- to 8-h period, and the generator has ingrowth of ⁶⁸Ga to 50% of its maximum after just only over 1 h since the previous elution. The major cold metal contamination detected by atomic absorption was tin (data not shown), probably in the stannate form, as described earlier (12), but which did not significantly affect ⁶⁸Ga-labeling procedures. This stannic oxide generator has been available for 23 y, although a new type using ⁶⁸Ge adsorbed onto a titanium dioxide support has recently become commercially available and may replace the Loc’h generator in time. However, the newer generator’s performance has not been adequately described in the literature as yet.

CONCLUSION

These studies demonstrate that the interweaved issues involved in bringing a bsAb targeting system for PET to clinical fruition can be addressed successfully. The outstanding target-to-tissue ratios obtained in these experiments encourage us to continue to refine and develop our agents still further, in the hope that they may soon realize their promise clinically, not only for improved cancer imaging but also for other diseases that have appropriate molecular targets for bsAbs.