In 1993, the FDA approved the first radiolabeled monoclonal antibody for diagnostic use in patients with cancer, CYT-103 (OncoScint OV/CR). This indium-111-labeled whole murine antibody to a tumor-associated glycoprotein, TAG-72, is found on a variety of mucin-producing adenocarcinomas. Many other antibodies recognizing other tumor-associated antigens are being investigated currently for tumor imaging. An understanding of the underlying principles of this unique modality is necessary for its appropriate and optimal use in the management of oncology patients.
The purpose of this paper is twofold: (1) to explore some of the technical issues surrounding the selection and optimal use of radiolabeled antibodies for diagnostic use and (2) to present a sampling of clinical trial results using some of the more promising agents for radioimmunodetection.
In order to image tumors with antibodies, one must target those antigens on the tumor cell that are different, either qualitatively or quantitatively, from antigens on surrounding normal cells. Ideally, the targeted antigens would be unique to tumor cells (ie, not found in any normal tissue in any amount). However, in the real world, most "tumor-associated antigens" are also found in some normal tissues, although sometimes at lower density or in less accessible sites (eg, intracellular) than in tumors. The sensitivity and specificity of radiolabeled antibody imaging depends, in part, on the degree of expression of accessible antigen in the tumor site vs normal tissue.
Among the first antigens targeted for immunoscintigraphy were the oncofetal antigens (eg, carcinoembryonic antigen [CEA] and alpha-fetoprotein [AFP]). Later, monoclonal antibody development led to the discovery of other, previously unidentified tumor-associated antigens, (eg, tumor-associated glycoprotein 72 [TAG-72], novel CEA epitopes, and epithelial membrane antigen [EMA]), which has expanded the repertoire of available agents.
In selecting radioisotopes for imaging, one must consider a variety of factors, many of which are mutually exclusive (the "no free lunch" phenomenon). For example, one would like to have high count rates while limiting the radiation dose to the patient. Therefore, isotopes that have a high efficiency of interaction with the gamma camera crystal (eg, technetium-99m) are preferred over those with a low efficiency (eg, iodine(Drug information on iodine)-131). Technetium-99m also has a relatively short half-life (approximately 6 hours), which, coupled with its relatively low-energy gamma photon, means that a larger administered activity (which translates to more counts per second) can be given for the same patient radiation dose. Iodine-131, on the other hand, emits not only a higher-energy gamma radiation but also beta particulate radiation, which increases patient radiation dose without contributing to imaging. Both technetium-99m and iodine-131 are relatively inexpensive--an advantage in this era of cost consciousness.
For optimal lesion detection, however, a high target-background ratio is needed, which means that blood and normal tissue levels have decreased while tumor levels remain high. With whole antibodies, this frequently requires waiting 24 hours or more for clearance from normal tissues to occur. The "ideal" isotope from the standpoints of efficiency and radiation dose, technetium-99m, has a short, 6-hour half-life. This means that by 24 hours, the count rate will have dropped to a point that would require prolonged imaging time to acquire suitable images, which, in turn, increases the probability that patient movement will degrade those images. Iodine-131, with its 8-day half-life, presents no such problem. The chemistry of technetium-99m is also less favorable for antibody labeling than is that of iodine.
Iodine-123 is chemically identical to iodine-131 but has no beta emission. Its efficiency of interaction with the gamma camera crystal is almost as good as that of technetium-99m. The half-life of iodine-123 is 13 hours, which is acceptable for 24-hour imaging but marginal past that point. Unfortunately, iodine-123 is relatively expensive and is less readily available than iodine-131 or technetium-99m.
Indium-111 has a favorable half-life (approximately 3 days) for delayed imaging. However, it tends to accumulate in normal liver tissue, which decreases its usefulness for the detection of liver metastases. It is also expensive, but is currently the only isotope used with an FDA-approved monoclonal antibody imaging agent.
The first antibodies used for imaging were produced by immunizing an animal (usually a mouse) with human tumor cells or cell extracts and then harvesting and purifying antibody from the animal's serum or ascitic fluid. This method yielded a variety of antibodies against a wide spectrum of antigens, some of which were "tumor specific" and others of which were more ubiquitous in tissues. Since these antibodies were derived from many B-lymphocyte clones, the term "polyclonal" is used to describe them. As might be expected, substantial lot-to-lot variation occurred when this production method was used.
In 1975, Kohler and Milstein  developed a method for selecting specific clones of cells that produced pure antibody against a single antigen; hence these antibodies were termed "monoclonal." As initially described, animals were immunized with a target substance (eg, whole cells, membrane extract, or a purified antigen source), as is the case for the production of polyclonal antibodies. Subsequently, however, the splenic B lymphocytes were harvested and fused with mouse myeloma cells to form immortal "hybridomas," which could be grown in cell culture. Each hybridoma clone produced a single antibody, which could be screened for immunoreactivity with the desired antigen. Cells from desirable clones could then be grown in animals (ascitic fluid) or in artificial cell culture systems to produce large amounts of pure antibody. This technology permitted the mass production of antibodies for clinical use.
Modifications of Monoclonal and Polyclonal Antibodies
Various modifications of these antibodies have been explored. Whole antibody has a rather slow clearance half-time from blood and normal tissues, necessitating prolonged delays between antibody injection and imaging. Not only does this inconvenience the patient and prolong the diagnostic process, but longer delays necessitate the use of isotopes that are suboptimal with respect to imaging characteristics and patient radiation dose.
Fab' Fragments--The fact that smaller molecules tend to clear more rapidly has led to the development of a technique whereby whole antibodies are digested into antibody fragments in such a way as to maintain immunoreactivity. Since the specificity of antibody binding depends on the variable region of the IgG molecule, elimination of the constant region should produce faster clearance without compro- mising affinity. Fab' fragments consist of variable regions of one heavy chain (VH) and one light chain (VL), while F(ab')2 fragments consist of two Fab(v) fragments connected at the hinge region. They are approximately one third and two thirds, respectively, the molecular weight of whole antibody .
Even smaller fragments consisting of only the hypervariable regions of the heavy and light chains have been produced, and are sometimes referred to as Fv fragments. The smallest unit that would retain immunologic specificity is the hypervariable region peptide or molecular recognition unit, the sites that confer the specific antigen-binding properties to the antibody molecule . Not only are these fragments cleared from nontumor sites more quickly due to their smaller size, they should also be better able to migrate through the extravascular space into the interior of the tumor, thus expanding the available antigen-binding sites .
Chimeric Antibodies--Another method of limiting the amount of mouse protein in an antibody is to fuse the variable region of the murine monoclonal antibody of interest to the constant region of human antibody, either chemically or using genetic engineering techniques . The resulting antibodies are known as "chimeric" antibodies (named for the chimera of Greek mythology, creatures that were composed of parts of several different animals). The antigenicity of chimeric antibodies is similar to that of antibody fragments.
"Humanized" antibodies have been developed using genetic engineering techniques to graft the complementarity-determining regions of mouse antibodies into human molecules . These antibodies would be expected to have antigenicity (incidence of human antimouse antibody formation) similar to FV fragments while retaining other characteristics of whole antibody.
Human Antibodies--A few fully human antitumor antibodies have been studied. Antigen-stimulated human B lymphocytes from cancer patients are virus transformed and grown in nutrient media within a device that allows harvesting of high-purity antibody. As there is no foreign (nonhuman) protein content, antigenicity of these human antibodies would be expected to be quite low. This has been found to be the case in clinical trials, in which no antigenic responses were detected in patients receiving multiple injections of human antibodies 
Bifunctional Antibodies--An additional modification is the development of bifunctional antibodies that bind to both tumor-associated antigens and to a radiolabeled ligand. This allows for a two-step process of radiolabeling in which the unlabeled antibody is injected first, followed by injection of the labeled ligand after clearance of unbound antibody from the blood pool . This technique significantly decreases background activity, allowing detection of smaller lesions and/or enhanced detection of liver lesions using indium-111 for labeling.