Bispecific Antibody Pretargeting PET (ImmunoPET) with an ¹²⁴I-Labeled Hapten-Peptide

Bispecific Antibody

The humanized anti-CEA × anti-HSG recombinant bs-mAb, designated hBS14, which is divalent for CEA and monovalent to HSG, has been described previously (17). The bs-mAb was radioiodinated with Na¹²⁵I (Perkin Elmer) to a specific activity of 41.4–47.4 MBq/nmol using a chloramine-T method (18). Size-exclusion high-performance liquid chromatography (HPLC) showed the product as a single peak with <5% unbound radioiodine. The immunoreactive fraction to CEA was determined as described previously and was in excess of 90% (17).

Peptide Synthesis

IMP-288 (DOTA-d-Tyr-d-Lys(HSG)-d-Glu-d-Lys(HSG)-NH₂; MH⁺: 1,453) was synthesized on solid phase using the Fmoc strategy. The Sieber amide resin was obtained from Calbiochem-Novabiochem and the Fmoc amino acids were obtained from Advanced ChemTech. The DOTA-tris(t-butyl ester) was obtained from Macrocyclics. The crude peptide was purified by reverse-phase HPLC (RP-HPLC) on a Waters LC 4000 using a radial compression module containing three 40 × 100 mm Nova Pak column segments and a 40 × 10 mm precolumn, using a gradient (100% A, 0.1% ammonium acetate in water to 70:30 A/B, where B is 90% acetonitrile/10% water). IMP-288 was ≥99% pure by HPLC and showed a molecular-ion peak by electrospray mass spectrometry (ESMS) (HT Laboratories, Inc.) consistent with the intended structure.

For radioiodination, IMP-288 was first preloaded by adding a 4.5-fold mole excess of indium chloride in 6 mL of 0.01 mol/L citrate/sodium acetate buffer, pH 4.03, and heating at 50°C for 20 min. This material was purified on a Waters Sunfire Prep C₁₈ 5-μm, 19 × 150 mm column. ESMS confirmed a molecular-ion peak consistent with the desired peptide. The In⁺³-loaded IMP-288 used for radioiodination was given a new designation, IMP-325 (In-DOTA-d-Tyr-d-Lys(HSG)-d-Glu-d-Lys(HSG)-NH₂; MH⁺: 1,565).

Radiolabeling

IMP-288 was radiolabeled with ¹¹¹InCl₃ (IsoTex) using the previously reported procedure (16). A number of conditions for radioiodinating IMP-325 using Na¹³¹I were evaluated, but the following procedure provided the most consistent yields. Briefly, 4 μL of a 2 mmol/L solution of IMP-325 in 0.5 mol/L sodium phosphate, pH 4.0, were added to 100 μL of a buffer composed of 1.0 mol/L phosphate, 1.0 mol/L acetic acid, and 0.75 mol/L ammonium hydroxide, pH 2.0. The entire contents were transferred to the manufacturer's vial containing 370 MBq of Na¹³¹I (MDS Nordion or Perkin Elmer). Initially, the peptide and radioiodine solution were transferred to an IODO-GEN–coated vial (Pierce); however, we found that the iodination reaction proceeded more reliably when IODO-GEN was added as a microparticulate. For this procedure, 100 μg of IODO-GEN dissolved in 20 μL of dimethylformamide (DMF) were added to the vial containing the peptide-iodide solution and mixed briefly on a vortex mixer. After 15–20 min, the contents of the reaction vial, with a 150-μL wash, was added to 400 μL of a freshly prepared quenching buffer (0.1 mol/L cysteine in 0.5 mol/L sodium phosphate, pH 7.4) that was suspended on the top of a Waters Oasis HLB column (1 mL, 30 mg). Washing the column with 2.0 mL of the cysteine buffer followed by 2.0 mL of water removed the unbound radioiodide. The labeled peptide was eluted with 400 μL of 1:1 ethanol/water. Radiolabeling yields averaged between 70% and 80%, with specific activities ranging from 25.9 to 44.4 GBq (0.7–1.2 Ci)/μmol (all specific activity estimates assume that 100% of the peptide was recovered after purification). By RP-HPLC, the labeled peptide showed 2 peaks, likely representing mono- and diiodinated species. The purified products typically had <4% unbound radioiodine.

A similar microparticulate IODO-GEN procedure was used for ¹²⁴I radiolabeling of the peptide; however, 6 nmol of a sodium iodide solution were added to the radiolabeling mixture to improve yields. Na¹²⁴I was purchased from Eastern Isotopes. After 30 min at room temperature, the reaction was quenched and separated in an identical manner as stated above. Radiolabeling yields with ¹²⁴I were commonly 50%–60%, with an estimated specific activity of ∼29.6 GBq/μmol. By RP-HPLC, the diiodinated peptide represented ∼40% of the product with ≤1.0% unbound ¹²⁴I.

The purified ¹²⁴I-IMP-325 used for animal studies was diluted in phosphate-buffered saline containing 1 mg/mL ascorbic acid to 74 MBq/mL. The product was stored overnight at 4°C and then shipped the next morning for injection in animals. RP-HPLC analysis of the ¹²⁴I-peptide stored in this manner showed an increase from 0.8% to 5.4% unbound ¹²⁴I after 17 h, which increased to ∼10% after 30 h. However, a 1:100 diluted sample used to analyze the initial postlabeling run was reexamined after it was stored at 25°C overnight in the HPLC autoinjector, and it showed the amount of unbound ¹²⁴I was unchanged from the previous day. Thus, the higher concentration of the product that was administered to the animals resulted in some radiolysis of the ¹²⁴I-peptide.

¹²⁴I-Labeled humanized anti-CEA (hMN-14, labetuzumab; Immunomedics (16)) Fab′ was prepared using a similar procedure as described for radioiodinating the bs-mAb but, as with the ¹²⁴I-peptide radiolabeling, 6 nmol of a sodium iodide solution were added to increase incorporation. The final product was isolated in phosphate-buffered saline, pH 7.4, containing 1% human serum albumin, and adjusted to a concentration of 74 MBq/mL. It was stored at ∼4°C overnight and then shipped by same-day courier to Memorial Sloan-Kettering Cancer Center (MSKCC) and Center for Molecular Medicine and Immunology (CMMI) for animal injections. Size-exclusion HPLC showed a single peak with <1% unbound radioiodine.

In Vivo Pretargeting

All animal studies were performed in compliance with the Guide for the Care and Use of Laboratory Animals in accordance with National Institutes of Health policies. Female nude mice of 8–10 wk of age (athymic NCr-nu/nu, National Cancer Institute colony; Taconic Laboratories) were implanted subcutaneously with 1 × 10⁷ LS174T human colon cancer cells (American Tissue Culture Collection) that were freshly harvested from culture and washed in phosphate-buffered saline, pH 7.4, before implantation (0.5 mL). Seven to 11 d after implantation, studies were initiated.

After an initial study that confirmed optimal pretargeting with the ¹³¹I-IMP-325 could be achieved using the same conditions as reported previously for other labeled di-HSG peptides (3,17) (not shown), 13 animals were given 40 μg (0.5 nmol) of the hBS14 bs-mAb, containing a trace amount of ¹²⁵I-hBS14 (74 kBq), followed 24 h later with ∼1.07 MBq (0.05 nmol) of the radiolabeled peptide. External scintigraphy and tissue counting were used to determine uptake and distribution at 1, 4, and 24 h. Appropriate windows were set in the γ-scintillation counter to discriminate each of the radioactive elements, correcting the ¹²⁵I channel for downscatter of ¹³¹I. Two animals receiving the ¹³¹I-peptide alone were imaged at 2 and 4 h and then necropsied. In addition, 8 animals were given 40 μg of the hBS14 bs-mAb followed 24 h later with 0.05 nmol (1.07 MBq) of ¹¹¹In-IMP-288. Five animals were imaged at 1 h and then necropsied, with the other 3 animals imaged and necropsied at 24 h. None of the animals was given a radioiodine-blocking regimen deliberately, so that dehalogenation could be assessed in the images (e.g., thyroid and stomach uptake). Imaging was performed for 20 min using an ADAC Solus imaging system with a high-energy collimator for ¹³¹I and a medium-energy collimator for ¹¹¹In, with animals placed in prone position on the face of the collimator. No background correction was applied, but the intensity was varied by adjusting the color scale so that the pixels in the region with the most activity first became saturated, with the exception of the 1-h images, where the high activity in the urinary bladder was ignored. The adjusted nuclear images were then flipped to match photographs of the animals lying on the collimator.

The next series of studies compared the biodistribution and targeting of ¹²⁴I-IMP-325 to ¹⁸F-FDG- or ¹²⁴I-labeled hMN-14 anti-CEA Fab′. Two sets of animals were examined: one at MSKCC (microPET and necropsy at conclusion of imaging) and one at CMMI (necropsied at 3 planned intervals). Imaging studies were performed at MSKCC using a Focus 120 microPET unit (Concorde Microsystems, Inc.). A minimum of ∼20 million events was acquired over ∼5 min using an energy window of 350–750 keV, a coincidence timing window τ of 6 ns, and a span of 3. These list-mode data were sorted into 2-dimensional histograms by Fourier rebinning, and transverse images were reconstructed by filtered backprojection and by ordered-subsets expectation maximization (OSEM) iterative reconstruction into 128 × 128 × 94 (0.86 × 0.86 × 0.76 mm) for the Focus 120. The microPET imaging studies were quantitative—that is, the voxel values in the reconstructed images were in terms of percentage injected dose per gram (%ID/g). Data were corrected for the duration of the image acquisition, dead time, randoms, branching ratio, and physical decay of the radionuclide using hardware and software (ASIPro) provided by the manufacturer (Concorde Microsystems). The fully corrected counting rate densities (in cps/voxel) were then converted to activity concentrations (MBq/mL) by applying a measured system calibration factor (MBq/mL/cps/voxel) applicable to mice.

At MSKCC, 3 LS174T-bearing nude mice were given an intravenous injection of the hBS14 bs-mAb (160 μg; 2 nmol). Twenty-four hours later, an intravenous injection of ¹²⁴I-IMP-325 (0.2 nmol; 7.4 MBq in 0.1 mL) was given (n = 3). No radioiodine-blocking regimen was administered. All 3 animals were anesthetized and imaged at ∼1–2, 4–6, and 24–27 h after the radiolabeled peptide injection and necropsied after the final imaging session. Three other groups of animals were examined. One group (n = 3) was fasted overnight and given an intravenous injection of ¹⁸F-FDG (30 MBq per mouse; Eastern Isotopes) and, after imaging at ∼1.5 h, were necropsied for tissue distribution determinations. Other animals received ∼7.4 MBq (0.2 nmol) of ¹²⁴I-IMP-325 alone (no pretargeting; n = 3) or ¹²⁴I-hMN-14 Fab′ (7.4 MBq, 12 μg; n = 3), followed by sequential imaging at the same intervals as the pretargeted animals, and then necropsied at the conclusion of the last imaging study.

All images were adjusted without background correction. The maximum intensity for the images of the pretargeted animals and the animals given ¹⁸F-FDG was adjusted so that the highest-activity voxels of the tumor corresponded to the highest-intensity gray scale. Whenever possible, the maximum intensity of the sequential images of the individual animals was adjusted to the identical intensity used for the animal's 1-h image, so that quantitative comparisons could be made of the tumor and tissue distribution over the specified time interval for that animal.

At CMMI, groups of 5 LS174T-bearing animals were necropsied at 1, 3, and 24 h after the radioactivity injection. The animals received identical doses for the pretargeted peptide, peptide alone, and ¹²⁴I-hMN-14 Fab′. An additional group was given 7.4 MBq of ¹²⁴I-hMN-14 Fab′ that was supplemented with unlabeled hMN-14 Fab′ to increase the total protein dose to 100 μg.

Radioiodination of IMP-325

A number of iodination reactions of the peptide IMP-325 were performed with ¹²⁵I, ¹³¹I, and ¹²⁴I. The initial reactions were performed using IODO-GEN–coated vials under conditions typically used for the radioiodination of peptides (19). Most notably, this procedure uses a pH of 7.5 for the radioiodination reaction. However, using this method, the radiolabeled peptide contained several different products by HPLC, most likely representing the incorporation of radioiodine on the imidazole ring of the HSG hapten, as well as on the tyrosine (20). By lowering the pH of the reaction mixture, the imidazoles would be protonated, making them less reactive with electrophilic iodide. Additional test labels performed at progressively lower pH confirmed that the products had fewer iodinated species, until at a pH of 4.0 or lower, only 2 primary peaks were seen by RP-HPLC. These 2 peaks were assumed to be mono- and diiodotyrosine peptides based on previously published experience (21). The formation of the second (longer retention time) peak could also be minimized as the pH was lowered and even eliminated if a high molar ratio of peptide to iodide was used.

We initially performed the radioiodination in IODO-GEN–coated vials but found the yield to be highly variable, especially with high-specific activity iodide. The addition of cold iodide improved the yield for the ¹²⁴I-labeled peptide and increased the proportion of the second peak (diiodotyrosine) (Figs. 1A and 1B). The reaction consistency was also improved by dissolving the IODO-GEN in DMF and then precipitating it in the iodine vial in the presence of the peptide (Figs. 1C and 1D (22)). The preparation of ¹³¹I-IMP 325 was also sensitive to the ratio of iodide to moles of peptide, but when the reaction was performed as described in the Materials and Methods, the radiolabeling yields (70%–80%) and specific activity (25.9–44.4 GBq/μmol) were consistent. The ratio of monoiodopeptide to diiodopeptide varied from batch to batch of ¹³¹I, ranging from 70:30 (monoiodopeptide/diiodopeptide) to 10:90, but within the same batch, the ratios were consistent from run to run.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 1. 

RP-HPLC illustrates final radiolabeled products obtained under different radioiodination conditions. (A and C) ¹²⁴I-Labeled products obtained using Na¹²⁴I with (C) or without (A) addition of carrier-added iodide. Addition of carrier improved yields, with formation of both mono- and diiododinated forms. (B and D) ¹³¹I-Labeled products prepared using IODO-GEN–coated vials vs. microparticulate IODO-GEN. UnB = unbound radioiodine.

In Vivo Pretargeting

Initial studies confirmed that optimal pretargeting with ¹³¹I-IMP-325 could be achieved by first administering the hBS14 bs-mAb, followed 24 h later with the radiolabeled peptide (not shown). A 10-fold mole excess of the hBS14 was given in relation to the moles of radiolabeled peptide injected. The targeting properties were first examined using varying doses of the hBS14 bs-mAb, with a constant amount of ¹³¹I-peptide given 24 h later.

Whole-body imaging studies were performed with ¹³¹I-IMP-325 in pretargeted animals at 1, 4, and 24 h after the peptide injection (Fig. 2). At 1 h, most of the ¹³¹I radioactivity was already filtered from the kidneys and in the urinary bladder. Tumor uptake was evident, but image intensity could not be increased further without significant septal penetration emanating from the high ¹³¹I activity levels in the urinary bladder. Images taken at 2 h from animals that had only received ¹³¹I-IMP-325 (without pretargeting) indicated no selective ¹³¹I-peptide uptake in the tumor. The 1-h images of the animals pretargeted with ¹¹¹In-IMP-288 could be adjusted to highlight tumor uptake without artifacts.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 2. 

Whole-body γ-images of ¹³¹I-IMP-325 in animals given hBS14 bs-mAb 24 h earlier (+ bs-mAb) or animals given only peptide alone (− bs-mAb). Images on the right show pretargeting under the same conditions using ¹¹¹In-IMP-288. T = tumor; UB = urinary bladder. Arrows show tumor location on photographs of animals.

Necropsy data in Table 1 are given as a %ID per organ to reflect the amount of activity in each tissue as seen by external scintigraphy, but the tissue weights (normal tissue weights taken from 1-h necropsy data) are provided for conversion to %ID/g. The tumors had substantially higher amounts of the pretargeted ¹³¹I- or ¹¹¹In-peptide than any of the normal tissues at 1 h, with >30 times the amount of activity in the tumor than in the liver or spleen and ≥15 times more activity in the tumor than in the blood and lungs. As expected, the kidney and stomach had the highest amount of ¹³¹I-activity of the normal tissues, as the peptide is cleared from the body by urinary excretion, with uptake in the stomach indicative of the metabolic processing of radioiodine (no premedication to reduce gastric or thyroid uptake), albeit a portion of this uptake could be attributed to the small amount (∼4%) of unbound radioiodine coinjected with the ¹³¹I-IMP-325. Neither of these organs was visualized on the 1-h image, because when expressed as %ID per organ, tumor uptake was ≥15 times higher than in kidneys or stomach. Except for an ∼10-fold increase in the amount of activity in the stomach at 1 h for the ¹³¹I-pretargeted peptide compared with the pretargeted ¹¹¹In-peptide, there were no appreciable differences in the tumor and organ uptake. By 4 h, the radioactivity in the urinary bladder had been voided and, therefore, the ¹³¹I-images could be adjusted to illustrate the enhanced tumor uptake, whereas uptake in all normal tissues had decreased. At 4 h, the pretargeted peptide was 70-fold higher in the tumor than with the peptide alone. At 24 h, only the activity in the tumor was visualized, with tumor-to-blood ratios of >200:1.

With evidence that the peptide could be radioiodinated and perform well in a pretargeting setting, an evaluation of the ¹²⁴I-labeled peptide for PET was undertaken. Figures 3–⇓5 compare the images obtained in animals receiving the various agents. Thyroid and stomach activity was more prominent in the pretargeted ¹²⁴I-peptide images (e.g., Fig. 2 vs. Fig. 3), presumably because of the higher level of unbound ¹²⁴I but also the tumors in these animals were smaller (∼0.2–0.5 g). The smaller tumors were all clearly seen on the 1-h images without substantial radioactivity observed in any other organ (except for the nonvoided activity in the urinary bladder). A transverse slice through the region of kidneys and tumor is shown for animal 1 and illustrates the relative ease with which the tumor can be visualized against the major organ involved in the excretion of the radiolabeled peptide. Within each of the pretargeted ¹²⁴I-peptide animals, the image intensity was normalized to the level used at 1 h and, therefore, the coronal slices shown in Figures 3A, 3C, and 3D reflect the quantitative changes occurring from ∼2 to 24 h. Tumor intensity appears unchanged over 24 h, whereas the background in the normal tissues continued to decrease. The region of interest (ROI) for the tumor and left kidney taken from transverse slices over a 24-h period confirmed this and indicated that tumor-to-kidney ratios were initially about 2:1, increasing by nearly 2-fold at 24 h (Table 2). Necropsy data from a separate group of animals indicated the same trend of relatively constant tumor activity with decreasing activities in normal tissues (Table 3). The necropsy results indicated that tumor-to-blood ratios were 9.1 ± 2.6 at 1 h, which doubled at 3 h, and then increased to >200:1 by 24 h.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 3. 

Focus 120 microPET images of ¹²⁴I-IMP-325 pretargeted with hBS14 (A, C, and D) or alone (not pretargeted; B). (A) From left to right, photograph of AN1, 2 transverse slices (0.82 mm) at level of tumor (arrows) and kidneys (K), taken ∼2 h after injection, and coronal slices obtained at ∼2, 6, and 24 h after injection. (B) From left to right, sagittal and coronal views obtained 1.9 h after injection. (C) From left to right, photograph of AN3 and coronal slices obtained at ∼2, 6, and 24 h after injection. (D) From left to right, coronal slices obtained at ∼2, 6, and 24 h after injection. All pretargeted images are adjusted to same intensity as 2-h image for each animal to illustrate retention and clearance of ¹²⁴I-IMP-325 over time. At 24-h necropsy, tumors were 0.448, 0.285, and 0.483 g in pretargeted animals and 0.472 g in animal given ¹²⁴I-IMP-325 alone. H = heart; S = stomach; Thy = thyroid; UB = urinary bladder.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 4. 

Focus 120 microPET images of 3 animals injected with ¹²⁴I-hMN-14 Fab′. (A) Transverse slices (0.82-mm thick) in same plane to highlight tumor (arrows) localization at ∼2, 6, and 24 h, from top to bottom, respectively, for each animal. Six-hour images are taken at same intensity as 2-h images, but intensity was increased substantially for 24-h images. Coronal sections taken from 2-, 6-, and 24-h intervals (B, C, and D, respectively) illustrate elevated uptake of radiolabeled Fab′ initially in kidneys (K) and stomach (S). Uptake in these tissues was cleared over time, but there was persistent thyroid (Thy) uptake. Tumor weights at necropsy (24 h) were 0.596, 0.173, and 0.280 g, respectively.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 5. 

Focus 120 microPET images of animals bearing LS174T human colorectal cancer xenografts 1.7 h after ¹⁸F-FDG. All images were taken at same maximum intensity and background set to 0.0. (A) Photographs of 3 tumor-bearing animals. (B) Coronal slice (0.82-mm thick) taken to highlight uptake in tumor (arrows), as well as in bone marrow (BM) of shoulder and in hind quarters. Heart (H) and brain can also be seen in some images. Bars represent planes used for transverse and sagittal sections. (C) Transverse sections that highlight tumor or heart wall uptake. Sagittal sections illustrate the side view of tumor (D) and a plane that highlights heart, bone marrow, and urinary bladder (UB) activity (E). Tumor weights were 0.425, 0.321, and 0.153 g, respectively.

The images of the animals given the ¹²⁴I-peptide alone required considerable intensity enhancement to assess the peptide's distribution even at ∼2 h, as the peptide alone cleared so rapidly from the tissues. Even a tumor of ∼0.5 g could barely be identified on the 1.9-h microPET images and, except for prevoided activity in the bladder and kidneys, there was relatively little activity remaining in the body (Fig. 3B). Over time, the activity levels decreased so dramatically that it was difficult to visually distinguish focal organ or tumor uptake from general body background activity (not shown).

In contrast to the excellent tumor localization in the pretargeted ¹²⁴I-IMP-325 animals, tumors were less distinct at 2 and 6 h in the animals given ¹²⁴I-anti-CEA Fab′ because of an appreciably higher level of activity in the kidneys, stomach, and thyroid (Fig. 4). The images and the ROI data (Table 2) show a rapid clearance of radioiodine from the kidneys, but according to the ROI and necropsy data (Table 3), tumor-to-kidney ratios were <1.0 through 6 h. The imaging and necropsy data indicated that the activity in stomach of animals given ¹²⁴I-Fab′ was 5–8 times higher than in the pretargeted animals at 1 and 3 h, but by 24 h this activity had cleared to just 0.5 ± 0.2 %ID/g, providing a favorable tumor-to-stomach ratio. The 2- and 6-h images were adjusted to the same intensity to illustrate the quantitative changes occurring in the tissue distribution of the radioactivity at these time intervals. Even though the intensity setting of the 24-h images of the animals given ¹²⁴I-Fab′ had to be increased considerably and, therefore, cannot be quantitatively compared with the earlier images, the tumors (even one that was 0.173 g) were clearly visible without appreciable activity in the normal tissues. Necropsy data indicated that tumor uptake had decreased by as much as 8-fold from the level measured at 1 and 3 h. Importantly, tumor-to-kidney ratios were higher for the pretargeted peptide than with ¹²⁴I-Fab′ even through 24 h (e.g., P = 0.037 at 24 h using a 2-tailed t test assuming equal variance; Excel 2003). A separate group of animals given a higher protein dose of the Fab′ (100 μg) had similar tumor and tissue uptake values as the lower protein dose (12 μg), indicating that the higher protein dose did not affect tumor localization.

Tumors were relatively well visualized with ¹⁸F-FDG at ∼1.7 h after its injection (Fig. 5), but there was substantially more activity in the bone marrow, heart wall, and brain than in the pretargeted animals. The elevated uptake in these tissues is consistent with the known distribution of ¹⁸F-FDG, but because these tissues were not included in the necropsy panel, the relative uptake in the bone marrow and heart was assessed by ROI analysis. As shown in Table 4, uptake in the heart wall was about the same or nearly 2-fold higher than the tumor. Bone marrow uptake was 1.2- to 1.7-fold higher than that of the tumor. When compared with the necropsy data from the pretargeted animals at 1 h, the results suggest a higher tumor uptake for the pretargeted animals, but with higher tumor-to-blood ratios for ¹⁸F-FDG. The pretargeted animals had improved tumor-to-nontumor ratios for all other tissues, except the kidney.