Potential of Immuno–Positron Emission Tomography for Tumor Imaging and Immunotherapy Planning

In this issue of Clinical Cancer Research, Börjesson et al. (1) report the results of the first clinical trial using the long-lived positron emitter ⁸⁹Zr for immuno–positron emission tomography (PET). A series of 20 patients with squamous cell carcinoma of the head and neck were imaged up to 6 days after receiving ⁸⁹Zr-labeled chimeric (mouse/human) U36 monoclonal antibody (mAb), which binds to the v6 domain of CD44 (2). By selecting patients scheduled to undergo neck dissection and at high risk for having neck lymph node metastatic spread, histopathologic results provided a gold standard for evaluation of the sensitivity and accuracy of immuno-PET for the detection of primary and metastatic disease. Börjesson et al. (1) showed that the sensitivity and accuracy of ⁸⁹Zr-labeled chimeric U36 mAb imaging was at least as good as computed tomography/magnetic resonance imaging with tumor delineation optimum at later imaging times.

Molecular imaging in its more practical manifestations offers the exciting prospect of providing noninvasive evaluation of tumor metabolic status, location, and response to therapy. PET has been a key technology for these applications because its sensitivity allows the use of labeled probes at concentrations that do not perturb tumor metabolism or saturate molecular signatures that are up-regulated on tumor cells as a result of oncogenic transformation. The prototypical PET tracer, 2-[¹⁸F]fluoro-2-deoxy-d-glucose, is used routinely to detect malignancies that are characterized by an increased rate of glycolysis. Most other PET imaging agents currently undergo-ing clinical evaluation [e.g., [¹⁸F]fluorothymidine and O-(2-[¹⁸F]fluoroethyl)-l-tyrosine] share two properties with 2-[¹⁸F]fluoro-2-deoxy-d-glucose: use of the widely available 2-hour half-life (t_1/2) ¹⁸F as the radiolabel and a relatively low molecular weight to facilitate rapid distribution and tumor accumulation (3). However, these PET tracers also share a potential weakness: reliance on an uptake mechanism that is operant in normal tissues and, in some cases, other pathologic conditions, such as inflammation.

An alternative approach to tumor imaging is immuno-PET, which attempts to combine the advantages of PET technology with the potential tumor specificity of mAbs. The most common strategy has been to label a variety of mAb fragments instead of intact immunoglobulins. The rationale is that these smaller proteins clear more rapidly from normal tissues, thereby permitting tumor delineation at imaging times compatible with shorter t_1/2 and more conveniently available positron emitters, such as ¹⁸F and ⁶⁴Cu. A disadvantage of mAb fragments, however, is that expedited normal organ and blood pool clearance is achieved at the expense of tumor localization in terms of both peak retention and residence time. This is certainly problematic for mAb-based therapeutics and also may compromise some of the most useful potential applications of immuno-PET as well.

One of the most important principles of radiopharmaceutical design is to select a radionuclide with a physical t_1/2 compatible with the pharmacokinetics of the intended molecular carrier. Intact mAbs, particularly chimeric constructs with human IgG constant regions, clear from the vascular compartment with half-times of the order of days. When used for radioimmunotherapy, whole immunoglobulins are most frequently paired with β-emitting radionuclides with physical t_1/2s of similar duration as IgG biological clearance half-times. Notable examples are the Food and Drug Administration–approved radioimmunotherapy agents, Zevalin (Biogen Idec, Inc., Cambridge, MA), labeled with 2.7-day ⁹⁰Y, and Bexxar (GlaxoSmith Kline, Research Triangle Park, NC), labeled with 8.1-day ¹³¹I. A second reason for selecting these radionuclides is that they are both readily available. Unfortunately, one of the major impediments to the development of immuno-PET has been the limited availability of positron emitters with t_1/2s of similar duration.

The approach taken by Börjesson et al. (1) is a departure from the previous immuno-PET studies with intact mAbs that have predominantly used ¹²⁴I as the radiolabel. Although the 4.2-day t_1/2 of this radiohalogen is well suited to intact mAb pharmacokinetics, ¹²⁴I possesses other characteristics that can complicate its use for PET imaging. The positrons emitted by this radionuclide are of relatively high energy, which causes the intrinsic spatial resolution loss to be greater than experienced with the benchmark positron emitter ¹⁸F (4). In addition, 600 to 1,700 keV γ-rays are also emitted during ¹²⁴I decay. These emissions, along with the long range of its positrons in tissue, are features that can compromise image quality and hinder accurate quantification (5). Labeling of mAbs via direct substitution of radioiodine on constituent tyrosine residues can be readily accomplished; however, mAbs labeled in this fashion are subject to dehalogenation in vivo by endogenous deiodinases. Furthermore, if the mAb undergoes internalization after binding to its molecular target, lysosomal proteolysis of the mAb generally results in rapid egress of radioiodine from the tumor cell (6). Although multiple ¹²⁴I-labeled mAbs have been studied in animal models, only a few have progressed to the stage of clinical evaluation (7–9), and these have generally involved only a limited number of patients. It is anticipated that progress in immuno-PET with ¹²⁴I-labeled mAbs should be aided by the recent commercial availability of this radionuclide and the development of improved methods for protein radioiodination (10).

Zirconium-89 is an intriguing alternative radionuclide for immuno-PET, having a 3.3-day physical t_1/2 and emitting positrons with an energy that results in an intrinsic spatial resolution loss of only 1 mm compared with a loss of 2.3 mm for ¹²⁴I (5). Translation of the conceptual appeal of ⁸⁹Zr for immuno-PET into the clinically evaluable tracer described by Börjesson et al. (1) required several important radiochemical advances (11, 12). The first task was to develop a convenient method for producing ⁸⁹Zr, which was done by irradiation of a natural yttrium target with 14-MeV protons in a medical cyclotron. Isolation of ⁸⁹Zr from the target material was achieved using a column of hydroxamate, providing the desired radionuclide in high yield and radiochemical purity. Levels of ⁸⁹Zr sufficient to provide >50 clinical doses of ⁸⁹Zr-labeled mAbs could be produced from a single irradiation and at a reasonable cost, making more widespread investigations of the imaging potential of ⁸⁹Zr-labeled mAbs feasible.

An additional obstacle that had hindered clinical evaluation of ⁸⁹Zr-labeled mAbs was the lack of appropriate methodology for attaching the radiometal to the mAb in a stable fashion. Most of the metallic radionuclides used for labeling mAbs are trivalent species and form stable complexes with 1,4,7,10-tetra-azacylododecane-based or diethylenetriaminepentaacetic acid–based bifunctional chelates. The chemistry of zirconium is different, however, because this element is only stable in the IV oxidation state. Desferrioxamine B (Df), a drug used clinically for the removal of excess iron, was selected as the chelate for ⁸⁹Zr because Df contains three hydroxamate groups, which form a stable complex with zirconium (13). The next task was optimizing the bond between the ⁸⁹Zr-Df complex and the mAb. In the past, a thioether bond, formed by the reaction of N-(S-acetyl)mercaptoacetyl-Df and maleimide functionalized mAbs, had been investigated for this purpose (14); however, the integrity of the mAb-chelate conjugates was not maintained under physiologic conditions. To circumvent this problem, a novel bifunctional chelator was synthesized consisting of a Df moiety for metal binding and a 2,3,5,6-tetrafluorophenol N-succinyl ester to permit coupling to lysine residues on the mAb (11). Unlike mAb-chelate conjugates linked by a thioether bond, the amide bond generated with the 2,3,5,6-tetrafluorophenol N-succinyl ester reagent was stable in vivo. Because the mAb is premodified by reaction with 2,3,5,6-tetrafluorophenol N-succinyl ester, preparation of ⁸⁹Zr-labeled mAbs for clinical use can be accomplished in a single step. Taken together, the radiochemical methodologies developed by these investigators are important because they are adaptable to other centers and other mAbs and should allow researchers to evaluate the clinical potential of immuno-PET in different patient populations with a variety of mAbs.

By showing the ability to delineate sites of primary and metastatic disease via qualitative PET imaging, Börjesson et al. (1) have provided impetus for extending immuno-PET beyond lesion detection to applications, in which the quantitative capabilities of PET imaging can be fully exploited. PET is the nuclear imaging method of choice for tracer quantification because more accurate attenuation and scatter correction can be done compared with single-photon emission computed tomography and planar imaging. It should be noted that scan interpretation in the Börjesson et al. (1) study was done on images that were not attenuation corrected. The protocol involved administration of relatively low activity levels of ⁸⁹Zr-labeled mAbs (75 MBq) to minimize patient radiation dose, which is primarily due to the high-energy, high-abundance γ-ray (909 keV, 99.9%) emitted during the decay of ⁸⁹Zr. This is a potential limitation of ⁸⁹Zr for PET imaging because the low count rates resulted in noisy images when an attenuation correction was applied. The development of imaging protocols and procedures that balance clinical practicality with the ability to acquire the data required for PET quantification of longer-lived radionuclides, such as ⁸⁹Zr, is needed and should be feasible. An additional refinement that would greatly facilitate image interpretation and quantification is to overlay the immuno-PET images with those obtained with a higher resolution anatomic modality, such as computed tomography or magnetic resonance imaging. Hybrid PET/computed tomography scanners are now widely available and should be ideal for applications, such as immuno-PET that are critically dependent on an accurate determination not only of tracer activity concentration but also of tumor volume.

mAb-based therapeutics have been under development for >30 years, and progress has not been as rapid as initially anticipated. Nonetheless, at least nine antibodies, immunotoxins, and radionuclide conjugates have received Food and Drug Administration approval, and many more are undergoing clinical evaluation in a wide variety of patient populations (15). In principle, immuno-PET could be an invaluable tool in expediting the development of immunotherapeutics and tailoring these treatment strategies for the characteristics of individual patients (16). Serial quantitative PET offers the possibility of noninvasively providing accurate cumulative activity concentrations of the labeled mAb in normal tissues and tumors. Knowledge of the level of tumor accumulation of the labeled mAb could facilitate the selection of patients with sufficient targeting to likely benefit from a specific form of immunotherapy without the need for a biopsy. Moreover, tumor concentrations of mAb based solely on biopsies may not be predictive because of heterogeneities in target molecule expression and mAb delivery. Furthermore, it is difficult to obtain an appreciation of the dynamics of tumor localization and the area under the tumor delivery curve from invasive tumor sampling techniques.

Immuno-PET-acquired cumulative tumor activity concentration data could be used in several ways, provided that concordance between PET radionuclide distribution and that of the therapeutic radionuclide (or immunotoxin or mAb alone) can be shown. For example, the level of administered therapeutic could be adjusted for each patient to achieve a desired radiation-absorbed dose to tumor. Correlation of treatment efficacy with tumor radiation dose could then be investigated not only among patients but also for different lesions in the same patient. Finally, immuno-PET might be a valuable tool for optimization of immunotherapeutic strategies, potentially allowing comparison of constructs and dosing regimens as well as the evaluation of modifiers of mAb tumor delivery, such as hyperthermia (17).

Because of the large field of view of modern PET scanners, data for the quantification of cumulative mAb concentrations in normal organs can generally be obtained during the same imaging session. This information could greatly facilitate the implementation of phase I dose escalation protocols, which should optimally be designed based on the radiation-absorbed dose to critical normal organs rather than on the level of radioactivity administered to the patient (18). In addition, immuno-PET could provide valuable information of the relationship between measured radiation dose in normal tissues and toxicity to better define the maximum tolerated dose.

In conclusion, targeting malignant cell populations via antibody-based constructs retains its initial conceptual appeal; however, the effect of these reagents on cancer diagnosis and treatment has been limited. The report by Börjesson et al. (1) in this issue of Clinical Cancer Research has shown the feasibility of enhancing patient imaging with labeled mAbs by exploiting the excellent spatial resolution of PET. Research directed at harnessing the quantitative potential of immuno-PET is under way and, if successful, could provide invaluable tools for expediting the development of immunotherapeutic strategies and tailoring their implementation to the characteristics of individual patients.