Novel Antibody Vectors for Imaging

Abstract

Non-invasive molecular imaging approaches include nuclear, optical, MRI, CT, ultrasound and photoacoustic imaging, which require accumulation of a signal delivered by a probe at the target site. Monoclonal antibodies (mAbs) are high affinity molecules that can be used for specific, high signal delivery to cell surface molecules. However, their long circulation time in blood makes them unsuitable as imaging probes. Efforts to improve antibodies pharmacokinetics without compromising affinity and specificity have been made through protein engineering. Antibody variants that differ in antigen binding sites and size have been generated and evaluated as imaging probes to target tissues of interest. Fast clearing fragments such as single-chain Fv (scFv; 25 kDa) with one antigen binding site (monovalent) demonstrated low accumulation in tumors due the low exposure time to the target. Using scFv as building block to produce larger, bivalent fragments such as scFv dimers (diabodies, 50 kDa) and scFv-fusion proteins (80 kDa minibodies and 105 kDa scFv-Fc) resulted in higher tumor accumulation due to their longer residence time in blood. Imaging studies with these fragments following radiolabeling have demonstrated excellent, high contrast images in gamma cameras and PET scanners. Several studies have also investigated antibody fragments conjugated to fluorescence (near infrared dyes), bioluminescence (luciferases) and quantum dots for optical imaging and iron oxides nanoparticles for MRI. However, these studies indicate that there are several factors that influence successful targeting and imaging. These include stability of the antibody fragment, the labeling chemistry (direct or indirect), whether critical residues are modified, the number of antigen expressed on the cell, and whether the target has a rapid recycling rate or internalizes upon binding. The preclinical data presented are compelling and it is evident that antibody-based molecular imaging tracers will play an important future role in the diagnosis and management of cancer and other diseases.

INTRODUCTION

Monoclonal antibodies (mAbs) have long been considered attractive candidates for targeted therapy and diagnostics due to their highly specific targeting ability. However, despite considerable efforts in developing mAb-based therapeutics for more than 30 years, initial progress was slow because poor performance of rodent mAbs in humans [i.e. human anti-mouse antibody (HAMA) responses, short half-lives and inability to trigger human effector functions]. With the advances in protein engineering techniques, genetics and proteomics, the pharmaceutical industry has embraced mAbs as a new group of targeted drugs which has led to numerous FDA approved therapeutic mAbs.

For diagnostic imaging, only a handful of mAbs have been approved in the United States for single photon emission computed tomography (SPECT) imaging (Table 1). Most of these are no longer marketed in the US; only ProstaScint, LeukoScan, Bexxar and Zevalin are currently used in the clinic. Since these imaging agents are derived from murine mAbs, repeated use in humans limited except in patients with low grade B-cell lymphomas; a disease characterized by reduced host-immune recognition. In addition to straight diagnositic applications, antibody imaging can provide targeting and dosimetry information that can guide therapy. For example, both ¹³¹In-labeled Bexxar and ¹¹¹In-labeled Zevalin can be used in combination with ⁹⁰Y and ¹³¹I radioimmunotherapy, respectively, in patients with B-cell lymphomas. Image-guided therapies would be beneficial to other malignancies as well, and antibodies, due to their biological specificity, continue to be a promising avenue for developing new imaging probes for targeted treatment planning and monitoring.

TABLE 1.

Antibody-based imaging agents in the clinic

Generic name (trade name)	Company	Date of FDA approval	Antibody, format	Target	Radiolabel	Approved indications
Satumomab pendetide* (OncoScint)	Cytogen	December, 1992	B72.3 mouse IgG1	TAG-72	111In	Colorectal and ovarian carcinoma
Arcitumomab (CEA-Scan)	Immunomedics	June, 1996	IMMU-4, mouse IgG1 Fab’	CEA	99mTc	Colorectal, breast, and small cell lung carcinoma
Imciromab pentetate* (Myoscint)	Centocor	July, 1996	R11D10, mouse IgG2a Fab’	Myosin	111In	Myocardial necrosis
Nofetumomab merpentan (Verluma)	Boehringer Ingelheim	August, 1996	NR-LU-10, mouse IgG2b Fab	40-kDa glycoprotein	99mTc	Small cell and non-small cell lung carcinoma
Capromab pendetide (ProstaScint)	Cytogen	October, 1996	7E11-C5.3, mouse IgG1	100-kDa glycoprotein	111In	Prostate carcinoma
Ibritumomab tiuxetan (Zevalin)	Spectrum Pharms	February, 2002	2B8, mouse IgG1	CD20	111In	Non-Hodgkin lymphoma
Tositumomab (Bexxar)	SmithKline Beecham	June, 2003	B1, mouse IgG2a	CD20	131I	Non-Hodgkin lymphoma
Fanolesomab* (NeutroSpec former LeuTech)	Palatin Technologies	July, 2004	RB5, mouse IgM	CD15	99mTc	Appendicitis
Bectumomab (LymphoScan)	Immunomedics	Not in US	LL2, mouse IgG2a Fab’	CD22	99mTc	Non-Hodgkin lymphoma
Votumumab (HumaSPECT)	Intracel	Not in US	88BV59, human IgG3	Altered cytokeratins	99mTc	Colorectal, ovarian and breast carcinoma
Igovomab* (Indimacis-125)	CIS Bio International	Not in US	OC125, mouse IgG1 F(ab’)2	CA-125	111In	Ovarian cancer
Sulesomab (LeukoScan)	Immunomedics	Not in US	IMMU-MN3 mouse IgG1 Fab’	NCA-90	99mTc	Osteomyelitis, appendicitis, inflammatory bowel disease

^*no longer marketed

The discipline of molecular imaging is one of the most rapidly growing areas of science. It offers the ability to visualize, characterize and measure processes on molecular and cellular levels non-invasively in living systems. The key players for obtaining this information are the molecular imaging agent (probe or tracer) and the target which can be intracellular or cell surface proteins. Radiolabeled probes for SPECT or positron emission tomography (PET), offer visualization of physiological and biochemical changes. Conventional imaging modalities on the other hand [radiography, ultrasonography (US), computed tomography (CT) and magnetic resonance imaging (MRI)], offer visualization of non-specific changes related to morphology. PET is a highly sensitive radiotracer imaging modality with only 10⁻¹¹ – 10⁻¹² mol/L levels of probe required for detection ¹. Although numerous tracers for imaging cancers by PET have been developed, only the glucose analogue [¹⁸F]fluoro-2-D-deoxyglucose (FDG) which is a substrate for hexokinase in glucose metabolism, and [¹⁸F]fluoride ions such as ¹⁸F-NaF₂ for bone imaging (incorporated in the hydroxyapatite crystals in bone) are approved by FDA as PET probes. The major issues associated with FDG are non-specific and elevated uptake in inflammatory or infectious lesions, and variable physiological uptakes in normal tissues/organs that can be confused with malignant neoplasm ². In addition, lack of uptake in metabolically inactive malignant tissues will fail to be detected by FDG-PET. One approach to overcome the limitations of FDG-PET is to use targeting molecules/ligands that can specifically localize to cell-surface markers such as growth factor receptors, adhesion molecules, and differentiation and activation markers. Targets that have been subjected to intense research for oncological molecular imaging include angiogenesis ³, apoptosis ⁴, signal transduction and protein interaction networks ⁵, receptor- or enzyme-based ^{6, 7} and metabolic imaging ⁸ as well as inquiries about multidrug-resistance ^{9, 10} and gene delivery and expression ¹¹. Identification and validation of novel molecular markers have introduced several new targeted drugs into the clinic. Additionally, technical advances in instrumentation have enabled the combination of anatomy and molecular events into a single image through fused imaging modalities such as PET/CT, PET/MRI and SPECT/CT.

Today’s antibody engineering technologies enable routine production of novel fully human or humanized mAbs from display libraries or from immunized transgenic mice carrying human immunoglobulin genes. Despite the development of non-immunogenic antibodies, only HumaSPECT (not approved in US) is human amongst the diagnostic mAbs/Fab listed in Table 1. A major disadvantage of using intact antibodies as imaging probes is that they circulate in the blood for several days. Early studies of antibody-based imaging agents showed that blood clearance was inversely related to the size of the protein i.e. clearance rate of Fab or Fab’ > F(ab’)₂ > IgG ^{12, 13}. For this reason Fab fragments, produced by proteolytic digested mAbs were developed into imaging agents (Table 1). However, the disadvantage of this fragment is that it has only one antigen binding site (monovalent), which reduces the overall functional affinity (avidity) of the antibody. The production of genetically engineered antibody fragments of different sizes and valencies started with the introduction of a single-chain variable fragment (scFv, 25 kDa) (Figure 1) ^{14, 15}. Using scFv as building block, larger antibody fragments such as diabodies (dimers of scFv, 50 kDa), minibodies (dimers of scFv-C_H3, 80 kDa) and scFv-Fc dimers (105 kDa) have been generated (Figure 1) ¹⁶. These fragments have exhibited better tumor penetration, faster clearance kinetics and excellent tumor to blood ratios, which are desirable properties for an imaging probe. The blood clearance and tumor uptake curves of radioiodinated anti-carcinoembyonic antigen (CEA) scFv, diabody, minibody and a rapid clearing scFv-Fc fragment are shown in Figure 2. The rapid elimination has been attributed to their size and lack of interaction with the neonatal Fc receptor (FcRn).

Fig. 1.

Schematic presentation of an intact antibody and engineered antibody fragments derived from it including a single variable domain fragment (Fv), single chain Fv (scFv), diabody, minibody and scFv-Fc. Molecular weights are indicated below each fragment. V_L = variable light (light green); V_H = variable heavy (dark green), C_L = constant light (pink); C_H = constant heavy (blue).

Fig. 2.

Blood clearance (A) and tumor uptake (B) curves of radioiodine-labeled anti-CEA T84.66 antibody fragments in mice carrying subcutaneous LS174T human colon carcinoma xenografts. Standard errors are shown. Uptake is expressed as percentage of injected dose per gram (% ID/g). The scFv-Fc fragment, modified to clear quickly by the introduction of two mutations in the Fc region that interfere the FcRn interaction, is referred to as double mutant (DM). The longer persistence of radiolabeled fragments in the circulation leads to higher and more persistent tumor uptakes, but longer times are required to obtain high contrast imaging. Data are from ²⁶,³⁹,¹⁰⁵.

Once the optimal antibody-based vector has been generated they can be tagged with radionuclides (PET, SPECT), magnetic nanoparticles (MRI) or fluorescence/bioluminescence (optical) probes that enable a variety of imaging modalities. Although PET is more sensitive than SPECT, the latter offer advantages such as broad availability and lower costs of radioisotopes and γ-scanning instruments. In addition, since SPECT radionuclides emit photons with different energies, several biomarkers can potentially be distinguished at the same time, as opposed to PET radionuclides which all emit the same energy photons (511 keV). MRI is a powerful imaging modality with regards to high spatial resolution (≤ 100 microns) and tomographic capabilities, but the low signal sensitivity has been a major limitation for molecular imaging. However, the recent development and the increased sensitivity of magnetism-engineered iron oxide (MEIO) nanoprobes obtained when conjugated to Herceptin may make antibody fragments attractive MRI imaging agents ¹⁷. The extremely high sensitivity of optical imaging approaches (10⁻¹⁵ – 10⁻¹⁷ mol/L for bioluminescence; 10⁻⁹ – 10⁻¹² mol/L for fluorescence ¹⁸) for real-time imaging of small animal models provide low cost alternatives to nuclear medicine imaging approaches. The major disadvantage of optical imaging using visible light is poor tissue penetration (<1 cm) due to scatter and absorption of light. Bioluminescence has some advantages over fluorescence, since signal is produced in the presence of substrate (no excitiation light source is needed) and there is no background autofluorescence. However, for optical imaging, tomographic imaging remains a challenge. This review focuses on the use of engineered antibody fragments as probes or vectors for non-invasive imaging of tumors, with an emphasis on nuclear medicine applications.

RECOMBINANT ANTIBODY FRAGMENTS FOR IMAGING

1. Monovalent single-chain Fv fragments

The scFv fragment consists of variable light (V_L) and heavy (V_H) domains joined by a flexible peptide linker that can be 12–25 residues long (Figure 1). The resulting protein being 1/6 of the size of an intact antibody (150 kDa) retains the specificity and affinity of the parental antibody. Although the V domains can be in either orientation i.e. V_L-linker-V_H or V_H-linker-V_L, the orientation can affect expression ^{19, 20}, stability and antigen binding activity ²¹. Early biodistribution and imaging studies in planar gamma cameras revealed that although scFv molecules cleared very rapidly from blood, they were able to localize efficiently to their target ^22–27. However, issues with direct iodination such as loss of immunoreactivity and deiodination reduce the signal in the target tissue. For this reason scFv molecules have been radiolabeled with radiometals and other radionuclides. Several groups have used technetium-99m for two reasons: 1) its short physical half-life (6 h) is compatible with the short biological half-life of scFv, and 2) carrier-free ^99mTc is generator produced and readily available in hospital nuclear medicine departments. Genetic engineering approaches for site-specific ^99mTc-labeling have been created by C-terminal addition of a cystinyl peptide (G)₄C ²⁸ or a cys-his₆ peptide ²⁹ for chelation chemistry, metallothionein-myc peptide for direct radiolabeling ²⁷ and his₆ tag for using Tc(I)-carbonyl complex ³⁰. Radiometal labeling of scFv with ¹⁷⁷Lu and ¹¹¹In following conjugation to DOTA or DTPA, respectively ^{31, 32} resulted in elevated kidney retention and only a few %ID/g in the tumor in xenograft-bearing mice. When a disulfide stabilized anti-Tac scFv (dsFv) radiolabeled with ¹⁸F was evaluated in ATAC4 tumor bearing mice, maximum retention in the tumor was 4.2% ID/g at 45 min p.i. and maximum retention in the kidneys was 303% ID/g at 15 min p.i. ³³. Thus, kidneys are the dose-limiting organ of this fragment. Fluorescence imaging using fluorescent-conjugated scFv molecules for non-radioactive immunedetection of tumors has also been investigated ^{34, 35}. Using cyanine fluorochromes Cy7 (infrared) and Cy5.5.18 (far infrared) rapid secretion via the renal route was observed early whereas strong accumulation of fluorescence in tumor was evident at 24 h p.i.

The major limitations of scFv molecules are low functional affinity and short in vivo half-life (Figure 2) due to their valency and small size, respectively. Consequently, rapid dissociation from the target antigen due to monovalent binding will result in modest retention time in the target and potentially poor image quality ³⁶. Still they remain attractive candidates as they can easily and cost effectively be expressed in bacteria. To extend the in vivo half-life, scFv molecules have been conjugated to polyethylene glycol (PEG) polymers and human serum albumin (HSA). Following site-specific PEGylation of 5, 20 and 40 kDa maleimide–PEG polymers to a scFv, prolonged biological half-lives in vivo correlating to the molecular mass of the polymer was observed ³⁷. HSA has a serum half-life of 19 days and Yazaki et al ³⁸ fused HSA to anti-CEA T84.66 scFv molecule and evaluated the fusion molecule by both SPECT and PET using different radionuclides. SPECT images with ¹²⁵I and ¹¹¹In-DOTA-labeled scFv-HSA demonstrated rapid clearance and excellent tumor uptake. PET imaging was carried out with one tumor bearing mouse injected with ⁶⁴Cu-DOTA-scFv-HSA. In this animal, tumor localization was evident at 4 h which reached highest intensity at 24 h p.i. The anti-CEA scFv-HSA fusion protein primary clearance route was through the liver (MW = 90 kDa). It exhibited significant slower blood clearance and a higher tumor uptake than the minibody ³⁹. These results are encouraging and be utilized as an alternative way to overcome the very rapid clearance of scFv fragments.

2. Di- and multivalent single-chain Fv fragments

The monovalency of scFv is a significant limitation to tumor retention and numerous approaches to genetically engineer monovalent scFv into multivalent fragments with greater avidity have been pursued. Joining two anti-TAG-72 CC49 scFv molecules in tandem to produce divalent sc(Fv)₂ have resulted in spontaneous dimerization to a tetravalent [sc(Fv)₂]₂ with affinity and tumor retention similar to the intact parental antibody, but with a more rapid blood clearance ⁴⁰. In vivo studies with ^99mTc-CC49 sc(Fv)₂ and [sc(Fv)₂]₂ demonstrated longer blood clearance and higher tumor retention with the tetravalent [sc(Fv)₂]₂ ⁴¹. Chemical cross-linking of anti-TAG B72.3 scFv molecules using bis-maleimide cross-linkers have been employed to produce dimers or trimers ⁴². Both exhibited rapid clearance and similar tumor uptakes (~10% ID/g at 24 h p.i) in vivo, but high kidney retention was observed only with the radiometal labeled dimer. In another study, SPECT imaging of tumor bearing mice injected with a covalent linked ¹²⁵I-sc(Fv)₂ (divalent) against placental alkaline phosphatase (PLAP), showed improved tumor uptake and longer retention time in the tumor compared to the monomeric scFv ⁴³. Similarly, improved tumor uptake over the 741F8 scFv monomer was achieved when it was made into a dimer, bridged via a C-terminal gly₄cys₂ peptide ⁴⁴.

Use of multimerization domains or sequences to produce high avidity scFv molecules has also been explored. Streptavidin-scFv fusion for production of tetravalent and bifunctional scFv molecules ^{45, 46} has largely been exploited for in vivo pretargeting approaches for delivery of therapeutics conjugated to biotin ^47–51. However, streptavidin-based pretargeting approaches have two potential limitations: renal toxicity ⁵⁰ and highly immunogenic streptavidin ⁵². PET imaging using ⁶⁸Ga-DOTA-biotin-streptavidin revealed high blood-pool and renal activities at 90 min p.i. ⁵⁰. SPECT images of ⁶⁷Ga-DOTA-biotin after pretargeting with succinylated anti-TAG-72 CC49 scFv-streptavidin showed targeting to the tumor (3.8% ID/g at 24 h). Biodistribution data showed that activities in blood and organs were below 1% ID/g and that there was 30% reduction of kidney uptake relatively to the control. Recently two research groups have used collagen derived sequences to produce scFv-trimers ^{53, 54}. Fluorescent imaging was carried out with anti-CEA MFE-23 and anti-laminin L36 scFv molecules fused to the N-terminal collagen XVIII NC1 subdomain that forms non-covalent bound trimers of collagen alpha chains ⁵⁴. The so-called “trimerbodies” were labeled with the near-infrared fluorochrome cyanine-5 (Cy5) and administered to tumor bearing mice. Strong signal was observed in the tumors as early as 3 h p.i. that persisted to 48 h, whereas mice injected with control trimerbody showed no signal in the tumors. Monomeric scFv showed localization to tumor at 3 h only, but at a much lower level. Another domain that has been used used for multimerization of scFv is the p53 tetramerization domain ^55–58. Various versions of leucine zippers or four-helix bundles have been fused to scFv C-terminus via IgG3 hinge region for production of dimers ^{59, 60}. In vivo studies with these dimers, demonstrated intermediate blood clearance activity relatively to the monomer and the intact antibody ⁶⁰. In a later study, tumor targeting of mono-, di-, and tetrameric variants of anti-Her2 4D5 scFv was compared ⁵⁸. C-terminal helix-turn-helix domain and human p53 protein was used to make dimers and tetramers, respectively. In vivo, the ^99mTc-labeled tetramer demonstrated the longest serum persistence and highest tumor accumulation (4.3% ID/g at 24 h p.i.). Other multimerization domains that have been used are C4 binding protein to produce scFv octamers ⁶¹, and recently dimeric and tetrameric scFv molecules fused to uteroglobin was described ⁶². However, these have not been evaluated in vivo. The above in vivo studies provide compelling evidence that the avidity effect is important for tumor retention. In addition, since blood clearance is inversely related to the size, the larger sized antibody fragments all demonstrate slower clearance and longer exposure to the target, thus improving tumor targeting and retention.

Conjugation of scFv to nanoparticles has also been described ^63–65. One study used anti-Her2 quantum dot-conjugated immunoliposomes-based nanoparticles (QD-ILs) for optical imaging of xenografts ⁶⁴. These nanoparticles consisted of a liposomal core with incorporated PEG-linked anti-HER2 scFv and QDs. In vivo studies revealed that the QD-ILs exhibited intermediate clearance (T_1/2 = ~2.9 h) relatively to free QDs (T_1/2 <10 min) and free liposomes (T_1/2 = ~7.7 h). However, optical imaging showed that both targeted and non-targeted nanoparticles exhibited equal tumor accumulation (~14% of total body fluorescence) at 24 hours p.i. The authors explained this was consistent with the enhanced permeability and retention effect previously observed with long circulating liposomes. Thus, antibody mediated in vivo targeting was not observed in this study. In another study, anti-MUC-1 scFv fragments were conjugated to superparamagnetic iron oxide (SPIO) nanoparticles ⁶³. At 24 hours after injection, 5% ID/g was in the tumor as shown by whole-body autoradiography. In a recent study, an anti-epidermal growth factor receptor (EGFR) scFv was conjugated to surface functionalized QDs and magnetic iron oxide (IO) nanoparticles ⁶⁵. In vivo targeting of both nanoparticles following systemic delivery was demonstrated in an orthotopic pancreatic cancer model. Frozen tissue sections of mice injected with anti-EGFR-scFv-QD were examined by fluorescence microscopy, whereas mice injected with anti-EGFR-scFv-IO were evaluated by MRI imaging. Higher tumor uptake than background activity was observed with both nanoparticles. However, uptakes were also observed in the liver and spleen, although lower than those seen with the non-targeting nanoparticles.

3. Diabody (scFv dimer)

The easiest approach to engineer scFv dimers and higher orders of multimers is shortening the linker to <10 residues between the variable domains. The shorter linker opens up the scFv and forces it to cross-pair with another scFv molecule to form a non-covalent bound dimer that is called diabody (Figure 1) ⁶⁶. By reducing the linker length further (≤ 3 aa) non-covalently bound scFv trimers and tetramers can be produced ⁶⁷. However, the pattern of oligomerization is not straight forward as it is influenced by the variable domain sequence and their orientation in the scFv fragment ⁶⁸. The stability of trimers and tetramers is an issue as demonstrated in vivo with an ¹¹¹In-labeled anti-Lewis Y trimeric/tetrameric “multimer” ⁶⁹. Although tumor retention at 6 hours p.i. was similar to that with the F(ab’)₂ at 24 hours p.i, high renal accumulation was observed which was probably due dissociation into monomers. The authors conclude that instability will prevent development of these fragments into imaging agents.

Diabodies on the other hand form stable non-covalent dimers that have been shown to have a slightly longer blood clearance than scFv molecules (Figure 2). The larger size (50 kDa) and divalency have been attributed to their improved tumor uptake ⁷⁰. Early SPECT imaging studies with that ¹²³I-labeled anti-CEA T84.66 diabody demonstrated that tumors could be readily imaged at 6 h p.i ⁷¹. This fragment reached a maximum tumor uptake at 2 h of 13.68(1.49)% ID with a tumor to blood ratio of 2.9. At 6 h the activity in the tumor was 9.23(0.44)% ID/g and the tumor to blood ratio was 9.4. The uptake in the tumor at 24 h was similar to that of another anti-CEA, CB/ior-CEA.1, diabody (5 aa linker) that showed accumulation of 3.1% ID/g at 24 hours ⁷². The anti-CEA T84.66 diabody has been radiolabeled with several different positron emitting radionuclides (¹⁸F, ⁶⁴Cu, ¹²⁴I) and evaluated in tumor-bearing mice by small animal PET (Figure 3A). ¹²⁴I –labeled diabody exhibited excellent, high contrast images with target to background ratios of 3.95 at 4 h and 10.93 at 18 h ⁷³. Less contrast was achieved with ⁶⁴Cu-DOTA diabody (Figure 3A). The anti-CEA diabody was also radiolabeled with ¹⁸F using N-succinimidyl-4-¹⁸F-fluorobenzoate (¹⁸F-SFB) and evaluated by PET imaging in xenografted mice ⁷⁴. Clear delineation of the positive tumor was obtained as early as 1 hour after injection of the ¹⁸F-FB-T84.66 diabody with clearance predominantly via the kidneys. This study suggests that the diabody is compatible with a short lived radionuclide for same day imaging.

Fig. 3.

Small animal PET (A) and optical imaging (B) of anti-CEA T84.66 diabody in LS174T xenograft-bearing mice ⁷³,⁷⁴,⁷⁶,⁷⁷. A. Coronal PET images of mice injected with diabody following radiolabeling with different positron emitting radionuclides. Specific localization to the positive tumor (arrow) and no localization to the negative C6 rat glioma tumors (arrowhead) are seen in all the images. Renal clearance is evident at 4 hours p.i. with ⁶⁴Cu- and ¹⁸F-labeled diabodies which is expected for a fragment of < 60 kDa in size. High contrast imaging is achieved with ¹²⁴I-labeled diabody at 20 hour p.i. due to almost complete elimination of any background activity, and a PET/CT overlaid image is shown for anatomical reference. B. Bioluminescence imaging of mice injected with diabody-renilla (Db-RLuc8) and –gaussia (Db-GLΔ15) luciferase fusion proteins at 4 hours p.i. The control mouse injected with Rluc8 only is shown. Specific targeting to the positive tumor is seen with both fusion proteins.

Enzymes that can catalyze a light-producing reaction (a process called bioluminescence) are known as luciferases. The light is produced following injection of a substrate for the enzyme. Luciferases are mainly used as reporter genes with firefly luciferase being the most commonly used for imaging ⁷⁵. Other luciferases such as renilla luciferase (RLuc) and gaussia luciferase (GLuc) from the sea pansy Renilla reniformis marine copepod Gaussia princeps, respectively, have been fused to the anti-CEA diabody and evaluated by optical bioluminescence imaging (Figure 3B) ^{76, 77}. Signal in the tumor was observed from 2 to 24 h with the diabody-RLuc8 variant with the highest intensity at 4 to 8 hours p.i. ⁷⁷. The tumors in mice injected with diabody-GLucΔ15, also exhibited highest intensity at 4 hours p.i ⁷⁶. However, although the diabody-GLucΔ15 was brighter than diabody-RLuc8 in vitro, the overall in vivo performance of diabody-GLucΔ15 was poorer than that of the diabody-RLuc8 with regards to stability, tumor uptake and sustained targeting out to 24 hours p.i. Still, these studies show that diabodies can be used as targeted vehicles to carry functional units for optical imaging of cell surface targets.

Diabodies to several other targets have also been evaluated in vivo. The human anti-Her2 C6.5 diabody is another well-characterized antibody fragment that has demonstrated significant tumor uptakes (~6.5 %ID/g and a tumor-to-blood ratio of about 9 at 24 hours p.i.) in xenograft-bearing animals ^{70, 78}. This diabody was radiolabeled with ¹²⁴I using both direct and indirect labeling methods ^{79, 80}. PET images of mice bearing Her2-positive xenografts following injection of directly radioiodinated diabody, showed excellent localization to tumors as well as significant uptake in the thyroid and stomach which are the natural sites of iodine metabolism. Indirectly labeled diabody with the water-soluble form of Bolton-Hunter reagent, SHPP, on lysine residues attenuated the uptake of free, metabolized radioiodine in thyroid and stomach but did not enhance the uptake in the tumors ⁸⁰. Animals injected with SHPP-labeled diabody had a higher retention of radioactivity in non-targeted tissues than animals injected with directly labeled diabody, resulting in a higher background signal in the images. However, tumors were clearly delineated at 48 hours p.i. More recently, a humanized anti-PSCA diabody was evaluated by PET imaging in mice bearing prostate xenografts following radioiodination with ¹²⁴I ⁸¹. In this study, activity in the tumor was observed at 4 hours with most of the activity gone at 20 hours p.i., but the highest tumor-to-background-ratio was obtained at 12 hours p.i. Better tumor accumulation in human prostate xenografts was recently achieved with ⁸⁶Y-CHX-A”-DTPA-antimindin/RG-1 19G9 diabody ⁸². At 48 h post injection, significant accumulation of activity was seen in the tumors by PET, whereas no activity in the tumor was detected with the scFv monomer. Diabodies against PSCA and Her2 have also been radiolabeled with ¹⁸F-SFB and evaluated by PET in xenograft bearing mice ^{83, 84}. Localization to tumors and high contrast images were obtained by 4 to 6 hours p.i. suggesting that that the half lives of the nuclide and the antibody fragment were compatible for early imaging.

One issue with radiolabeling smaller antibody fragments is that reactive residues may be present in critical regions of the antibody. For this reason approaches for site specific modification have been developed. The anti-ED-B domain of fibronectin L19 diabody and two variants with different ^99mTc binding motifs (i.e. Gly₃CysAla or His₆ peptide for direct site-specific radiolabeling) C-terminally were evaluated in murine F9 teratocarcinoma tumor-bearing mice ⁸⁵. Although the in vitro binding to ED-B fibronection was the same, their biodistribution differed. All three showed similar tumor uptakes of 7.7–9.4% ID/g at 3 hours that was reduced to 2.8–5.7% ID/g at 24 hours p.i. The L19 Db that was randomly chemically modified by a MAG2-type chelator and indirectly radiolabeled demonstrated slower blood clearance than the L19 Db-Gly₃CysAla which resulted in a higher background activity in the scintigraphic images. The L19 Db-His₆ showed elevated accumulation of radioactivity in the kidneys (145% ID/g at 3 hours) in comparison to the other two Db variants (8.6 and 15.2% ID/g at 3 hours). At 24 hours the activity of L19 Db-His₆ in the kidney was reduced to 72.4% ID/g versus 2.6 and 3.0% ID/g of the other two diabodies. The authors conclude that L19 Db-Gly₃CysAla showed the most favorable biodistribution and imaging properties which could potentially be useful to image angiogenesis-associated ED-B fibronectin-expressing tumors.

Another approach has been to add a C-terminal cysteine, preceded by two glycines, to diabodies for thiol-chemistry ⁸⁶. Cys-diabodies against CEA, Her2, CD20 and PSCA have been generated and shown that this fragment will exists exclusively as disulfide-bonded dimer that can be reduced and chemically modified using a thiol-specific bifunctional chelating agent, for attaching radionuclides ⁸⁶, fluorophores ⁸⁷ and for targeting nanoscale particles such as quantum dots (Qdots) ⁸⁸. The anti-CEA cys-diabody was evaluated in vivo following site specific radiololabeling with ⁶⁴Cu to a macrocyclic chelate DOTA peptide–hexanevinyl sulfone (DOTA-GLGK-HVS) conjugated to the C-terminal cysteine ⁸⁶. The ⁶⁴Cu-DOTA-GLGK-HVS Cys-diabody behaved similarly to its parental diabody with specific targeting to tumors observed by PET imaging at 4 and 18 h post-injection. More recently an anti-CD20 cys-diabody was compared with its parental diabody by PET imaging following random radiolabeling with ¹²⁴I ⁸⁹. Again, very similar biodistribution was observed for the two fragments suggesting that the tethering of the C-termini of the diabody subunits did not affect the in vivo tumor targeting and blood clearance properties. Thus, this format appears to be a general platform for site-specific conjugation of antibody fragments with functional payloads.

4. Minibody (scFv-C_H3 dimer)

To overcome the accumulation in the kidneys seen with the diabodies, larger fragments that the exceeded the renal clearnce threshold (<60 kDa) have been designed (Figure 1). For example, human immunoglobulin constant domains have been used to produce intermediate-sized antibody fragments with longer serum-half-lives (6–11 hours) (Figure 2). The first scFv molecule that was joined to human IgG1 C_H3 domain was derived from the anti-CEA T84.66 mAb and was referred to as a minibody ⁹⁰. The fusion was either by a two amino acid spacer (LD minibody) that formed non-covalent dimer or via the human IgG1 hinge and a flexible linker peptide (FLEX minibody) that formed covalent dimers ⁹⁰. The ¹²³I-FLEX minibody exhibited superior tumor targeting (32.9% ID/g) at 6 hours after injection, relatively to that of the ¹²³I-LD minibody (16.4% ID/g) which had substantially faster blood clearance kinetics. At 6 hours, the tumor-to-blood ratio of FLEX-minibody was only 2 which increased to 64.9 at 48 hours p.i. Initial SPECT images with the anti-CEA ¹²³I-minibody showed localization to the tumors at 4 hours, and that the background was almost completely cleared at 19 hours allowing distinct delineation of the tumors. As a result of these initial studies, the FLEX minibody format was used in all subsequent studies.

ImmunoPET with the anti-CEA minibody is shown in Figure 3. This minibody demonstrated its suitability as an imaging agent with rapid high-level accumulation of activity in xenografts and rapid disappearance from the circulation ^{73, 91}. Serial small animal PET imaging with ⁶⁴Cu-DOTA-minibody demonstrated rapid and specific localization to the positive tumor with increasing accumulation of activity over time (from 2 to 24 hours p.i.) that was >20% ID/g at 24 hours with a tumor-to-soft-tissue ratio of 6 ⁹¹. As expected, high persistent activity in the liver was also seen with activity moving into the GI-tract at the late time-point. PET images with ¹²⁴I-minibody produced excellent images at 18 hours p.i. with tumor uptakes of 21% ID/g and tumor-to-background ratio of 26.2 ⁷³. The high ratio is due to the rapid clearance from the blood combined with fast metabolism and clearance of radioiodine from non-targeted tissues. In comparison to FDG scan, both radiometal labeled and radioiodinated minibody produced a much cleaner and higher contrast image.

Minibodies have also been produced against Her2 and prostate stem cell antigen (PSCA) expressed in breast and prostate cancers ^92–94. Although both anti-Her2 and anti PSCA minibodies demonstrated similar pharmacokinetics to the anti-CEA minibody and localized specifically to tumors, they accumulation in tumors was considerable lower. Activity in the tumor was about 5% ID/g at 12 to 24 h with anti-Her2 ¹³¹I-10H8 minibody with tumor-to-blood ratios of 1.1 at 12 h that increased to 3 at 48 hours ⁹⁴. This lower activity was explained by internalization of the minibody upon binding, followed by metabolism and rapid excretion of radiolabel from the cells. Biodistribution with anti-Her2 ¹¹¹In-DOTA-10H8 minibody demonstrated that the radioactivity remained in the tumor, increasing the tumor-to-blood ratio to 13.3 at 48 hours p.i. ⁹³. Unexpectedly, the kidneys contained higher activity than the liver which was attributed to the presence of a cross-reactive antigen. A different minibody derived from Herceptin (Herc. Mb) was evaluated by PET (Figure 4) ⁹³. Although PET imaging with anti-Her2 ⁶⁴Cu-DOTA-4D5v8 minibody showed targeting to positive tumor, only about 4% ID/g was in the tumor from 4 to 48 hours p.i. Since the activities in the liver and kidneys were 3 to 4-fold higher, high background and less contrast were achieved with this fragment. When this fragment was radioiodinated with ¹²⁴I, less activity was seen in the tumor, whereas clear delineation of the tumor was achieved with the anti-Her2 ¹²⁴I-C6.5 minibody (Figure 4). This difference is probably due to different affinities of the fragments and that C6.5 internalization rate varies in different cell lines ⁹⁵. Biodistribution with the anti-PSCA ¹³¹I-2B3 minibody resulted in accumulation of 9% ID/g in the tumor, and the tumor-blood-ratio was 2.8 at 21 hour p.i. ⁹². Figure 4 shows PET imaging with the anti-PSCA ¹²⁴I-2B3 minibody. Specific localization to PSCA expressing LAPC-9 xenografts, with uptake of 5.2% ID/g and tumor-to-blood ratio of 1.1 at 21 hours was achieved ⁹². This was close to the uptake (6.6% ID/g) obtained with the intact parental antibody at 168 hours p.i. ⁹⁶. Thus, in this tumor model early, good contrast images of PSCA expressing xenografts were obtained with the minibody. Recently, a minibody against CD20 antigen expressed on B-cell lymphoma was generated from rituximab (Rituxan, Genentech) ⁹⁷. This fragment differed from the above minibodies in that the 218 linker ⁹⁸ was used between the variable domains, and the scFv was fused to the complete human IgG4 hinge and C_H3 domain. PET images with anti-CD20 ⁶⁴Cu- and ¹²⁴I-rituximab minibody (Rx. Mb) are shown in Figure 4. The ¹²⁴I-rituximab minibody showed excellent targeting (12.9% ID/g) to tumor with very little activity in negative tumor and other tissues and a tumor to blood ratio of 7.5 at 21 hours, whereas less contrast was obtained with the ⁶⁴Cu-DOTA-Rx. Mb ⁹⁷. These studies show that although the pharmacokinetics is similar for this antibody fragment format in the different animal models, tumor uptake varies depending on factors such as antigen expression, whether the target internalizes or not upon antibody binding, radiolabel and specificity/affinity.

Fig. 4.

Small animal PET imaging of xenograft-bearing mice injected with ¹²⁴I-labeled (top) and ⁶⁴Cu-DOTA minibodies (bottom) against different cell surface antigens. Minibodies against CEA (T84.66 Mb) ⁷³, CD20 (Rx. Mb) ⁹⁷, Her2 (Herc. Mb) ⁹³ were radiolabeled and imaged with both radionuclides. The anti-Her2 C6.5 Mb and anti-PSCA 2B3 Mb ⁹² were radiolabled with ¹²⁴I only. All ¹²⁴I-labeled minibodies, except the ¹²⁴I-Herc. Mb, produce high contrast images at 18–21 hours p.i. Less contrast is achieved with ⁶⁴Cu-DOTA minibodies due to high background activity. Hepatic clearance is evident with the ⁶⁴Cu-DOTA minibodies as expected for a fragment of > 60 kDa in size.

Another similar sized antibody fragment referred to as “small immunoproteins” (SIP) consisting of scFv joined to the C_H4 domain of the secretory isoform S2 of human IgE for dimerization and further stabilized by a C-terminal disulfide bond has been generated ⁹⁹. Tumor targeting of anti-ED-B domain of fibronectin L19-SIP was 2–5 times higher than that of (scFv)₂, reaching a maximum uptake at 4–6 hours p.i. Accumulation of radioactivity in the mouse embryonal teratocarcinoma F9 tumor model was 3–4 times higher (12–18% ID/g) than in SKMEL-28 tumor model due to their difference in angiogenic activity. Although PET imaging of ⁷⁶Br-L19-SIP showed accumulation in F9 tumors from 5 to 48 h, persistent activity in the blood and stomach due to debromination resulted a tumor to background ratio of <2 at 48 hours p.i ¹⁰⁰. Recently, this fragment was radiolabeled with ¹²⁴I and evaluated in mice bearing human head and neck squamous cell carcinoma (HNSCC) FaDu xenografts ¹⁰¹. In this model, tumor accumulation reached a maximum of approximately 11% ID/g at 6 hours p.i., and at 48 hours the tumor uptake was about 5% ID/g and the tumor to blood ratio was 13.5 which increased to 45.9 at 72 hours. This was in concordance with ¹³¹I-labeled L19-SIP biodistribution, thus authors conclude that imaging with iodine-124 can be used to predict ¹³¹I-L19-SIP biodistribution. Another SIP has been made from affinity matured human anti-tenascin-C G11 mAb ¹⁰². However, only 0.85% of ID/g of ¹²⁵I-G11-SIP accumulated in U87 human glioblastoma zenografts at 24 hours p.i. which was similar to the tumor uptake (0.71% ID/g) of ¹²⁵I-labeled G11 scFv alone at 24 hours. The authors do not address this low uptake and one can only speculate that such low uptake was due to instability of the protein, low antigen expression in target and/or internalization of the G11-SIP with rapid dehalogenation and secretion of the label in this tumor model.

Recently, a so-called “miniantibody” against mindin/RG-1 composed of IgE-C_H4 dimerization domain and IgG1 hinge and modified first β-strand peptide residues of IgG1 C_H2 was evaluated by PET following conjugation of CHX-A”-DTPA and radiolabeling with ⁸⁶Y ⁸². At 24 hours p.i. about 4% ID/g was in the blood and about 13% ID/g was localized to LNCaP (prostate cancer) xenografts. At 72 hours p.i. the tumors were still visible and the background activity was greatly reduced. When the same fragment was labeled with ¹¹¹In, biodistribution studies revealed that the activities in the tumors and blood at 48 hours were approximately 14% ID/g and 1% ID/g, respectively. This resulted in a tumor to blood ratio that was better than that of the intact antibody (2.2 to 1), but worse than that of the diabody (80 to 1) at 48 hours p.i. Thus, the authors conclude that the diabody may be more suitable for radiodiagnostic applications.

5. scFv-Fc (scFv-C_H2-C_H3 dimers)

Larger antibody fragments that include the entire Fc region (Figure 1) have also been evaluated by small animal PET imaging (Figure 5). However, the neonatal Fc receptor (FcRn) which binds to the IgG Fc region is responsible for the prolonged serum half-life of intact antibodies and imaging with intact antibodies requires days to clear from the blood in order to produce high contrast images. However, clearance can be tailored by modifying the FcRn interaction in the Fc region. This was elegantly shown in a study with anti-CEA T84.66 scFv-Fc fragments ¹⁰³. Site-specific mutations of the IgG1 Fc residues involved in this interaction resulted in five variants (I253A, H310A, H435Q, H435R, and H310A/H435Q) that each exhibited distinct blood clearances in mice that ranged from 83.4 to 7.96 hours, which was much faster than that of the wild-type (~12 days). SPR analyses using human FcRn, also demonstrated binding of scFv-Fc WT whereas no interaction was obtained with the scFv-Fc H310A/H435Q fragment¹⁰⁴. These fragments were also radioiodinated with ¹²⁴I and evaluated by PET in xenografted mice. Although the wild-type had the highest tumor uptake (42.5% ID/g at 48 h p.i.) it also had the lowest contrast relatively to the intermediate clearing (I253A and H310A) and the fast clearing (H310A/H435Q) scFv-Fc fragments. The blood clearance of the scFv-Fc DM shown in Figure 2A, demonstrates that this fragment persists longer in the circulation than the minibody fragment that results in high tumor uptake (Figure 2B). The PET images of ¹²⁴I- and ⁶⁴Cu-DOTA- scFv-Fc H310A/H435Q also referred to as scFv-Fc double mutant (scFv-Fc DM) show high, specific tumor uptakes in xenografts (Figure 5A). The pharmacokinetics of radioiodinated versus radiometal-labeled I253A, H310A and H310A/H435Q scFv-Fc variants were further evaluated in xenografted mice in order to predict their therapeutic potential ¹⁰⁵. Tumor uptakes were inversely related to blood clearance and hepatic radiometal activity correlated with the blood clearance rate of the fragment i.e. faster clearance resulted in higher activity. Based on the biodistribution data with ¹²⁵I and ¹¹¹In, it was predicted that the fast clearing scFv-Fc DM would be able to deliver >7,000 cGy to the tumor with favorable tumor to liver and kidney ratios when radiolabeled with ¹³¹I, whereas as for ⁹⁰Y therapy a slow clearing antibody would be the protein of choice as the liver/kidney activities would be lower.

Fig. 5.

Small animal PET imaging of xenograft-bearing mice injected with ⁶⁴Cu and/or ¹²⁴I-labeled scFv-Fc double mutant (DM) fragments targeting CEA (A) and CD20 (D) ⁹⁷,¹⁰³. The background activity of ¹²⁴I-anti-CD20 scFv-Fc with IgG4 Fc is less (1.9% ID/g at 21 hours p.i.) ⁹⁷ than that seen with ¹²⁴I-anti-CEA scFv-Fc with IgG1 Fc region (6.9% ID/g at 12 hours and 2.8% ID/g at 24 hours p.i.) ¹⁰⁵, suggesting a more rapid clearance rate of this anti-CD20 fragment. This is consistent with published observation of the different blood clearances of human IgG1 and IgG4 isotypes. Again, ⁶⁴Cu-labeled fragments produce less contrast than ¹²⁴I-labeled fragments and hepatic clearance is evident.

The scFv-Fc DM fragment has also been evaluated in other tumor models. An anti-Her2 scFv-Fc DM fragment was evaluated by PET in xenografted mice following conjugation with DOTA and radiolabeling with ⁶⁴Cu ⁹³. The scFv-Fc DM fragment exhibited over 2-fold improved tumor targeting and reduced kidney activity relatively to the anti-Her2 minibody fragment which was probably due to its slightly longer residence time in the blood. When this fragment was labeled with ¹²⁴I, very low activity was observed in the tumor. In another recent study, an anti-CD20 scFv-Fc DM fragment with IgG4 Fc was evaluated in tumor bearing mice following radioiodination with ¹²⁴I (Figure 5B) ⁹⁷. Although rapid tumor localization was observed, the activities in the tumor and the blood were 2-fold lower than of the minibody. Such difference was not observed with the anti-CEA (Figure 2A) and anti-Her2 fragments. Thus it appears to be related to the change of IgG Fc isotype. Intact human IgG4 has been reported to clear faster than human IgG1 in both humans and mice ^{106, 107}. Consistent with this, a lower background activity of the radioiodinated anti-CD20 scFv-Fc DM fragment than that of the anti-CEA scFv-Fc DM at 18–20 hours p.i. was observed. Thus, the introduced mutations in IgG4 Fc may accelerate the scfv-Fc fragment to clear faster than the minibody. Figure 5B also shows the images obtained with ⁶⁴Cu-DOTA scFv-Fc DM fragment.

A rapid method to generate scFv-Fc fragments targeting endothelial cell from binders isolated from phage display libraries and in vivo evaluation by SPECT have been described ¹⁰⁸. Using this approach, three antibodies were identified that, when converted to scFv-Fc using mouse IgG1 Fc, targeted the lungs in rats with uptakes of 30–40% ID/g at 1 and 2 hour p.i. Since the blood activities were less than 2% ID/g, high contrast images were achieved early.

IMPACT OF ANTIBODY MODIFICATIONS ON IMAGING

From the numerous studies performed with the different engineered antibody fragments there are several factors such as the molecular weight, Fc domains, valency and specificity that will influence targeting and pharmacokinetics. Larger fragments above 60 kDa will clear via the liver, whereas those below this size will clear via the kidneys. Presence of Fc domains will increase the serum residence time, and increase the exposure time to the target. However, for imaging purposes a more rapid clearance from the blood is warranted in order to increase contrast and sensitivity. Difficulties in localizing and accumulating activity in target tissue are related to specificity and avidity (i.e. number of antigen binding sites). In addition, the number of target antigen per cell and whether it internalizes or not will affect accumulation in target tissue. Antibody mediated internalizing of targets followed by catabolism via lysozomal targeting and protolytic degradation will reduce tissue uptake. The rate of antigen internalization varies among different antigens. For example CEA internalizes slowly at a rate of approximately 0.001 minutes^{−1 109}, whereas Her2 has up to 100 times faster internalization rate when associated with EGFR ¹¹⁰. This may explain the lower accumulation of activity in Her2 tumors observed with ¹²⁴I-labeled Herceptin minibody and scFv-Fc DM in Figures 4 and 5B. In contrast, tumor was clearly delineated with ¹²⁴I-labeled C6.5 minibody (Fig. 5B). The C6.5 diabody has been reported to differ in internalization rate in different cell lines [i.e. relatively slowly by SK-OV-3 cells (10% in 3 hours) and rapidly by MDA-361/DYT2 cells] ⁹⁵. Thus, it is possible that the C6.5 minibody internalizes more slowly than the Herceptin minibody in this tumor model. Since it appears that the mAb and scFv internalization rates are determined by antigen recycling, rather than antigen endocytosis ^{111, 112}, antibodies with greater dissociation rates than antigen internalization rates can minimize catabolism.

The presence of shed antigen in the blood will deplete the pool of free antibodies available to reach the target. Once bound, the abnormal physiology in solid tumors limits inward diffusion and extravasation across capillary walls into the surrounding tissue by the antibody due to presence of a high interstitial fluid pressure gradient. Although smaller antibody fragments can penetrate tumors better than larger fragments, the affinity of the scFv fragments has also shown to affect tumor uptake i.e. as affinity increases, penetration decreases which is referred to as the “binding-site barrier” ^113–115. Anti-Her2 scFv with different affinities demonstrated that a minimum binding affinity of 10⁻⁸ to 10⁻⁹ M was required for retention in tumors, as a scFv with 10⁻⁷ M affinity failed to accumulate more than the negative control scFv ¹¹⁶. Increasing the affinity to 10⁻¹⁰ to 10⁻¹¹ failed to increase the tumor accumulation ^{114, 117}. Thus, the higher tumor accumulation of C6.5 minibody in Figure 5B may be due to the higher affinity of C6.5 IgG (5.4 × 10⁻¹¹ M ¹¹⁸) versus that of Herceptin (5 × 10⁻⁹ M).

Other factors that may affect the targeting are the nature of label used and the labeling chemistry employed. Several positron emitters for immunoPET are currently under investigation. These can be grouped according to their physical half-lives (t_1/2) that ideally should be paired with the biological half-lives of the antibody fragments. Short-lived positron emitters are ⁶⁸Ga (t_1/2 = 1.13 h) and ¹⁸F (t_1/2 = 1.83 h) which are suitable for rapid clearing fragments such as scFv and diabodies. Intermediate-lived positron emitters are ⁶⁴Cu (t_1/2 = 12.7 h), ⁸⁶Y (t_1/2 = 14.7 h) and ⁷⁶Br (t_1/2 = 16.2 h) which are suitable for antibody fragments with intermediate clearance properties such as minibodies and modified scFv-Fc fragments. Long-lived positron emitters are ⁸⁹Zr (t_1/2 = 78.4 h) and ¹²⁴I (t_1/2 = 100.3 h) which are suitable for intact antibodies and scFv-Fc WT. Since long lasting radionuclides have been successfully used with rapid clearing antibody fragments and produced excellent, high contrast images ^{73, 97}, it does not seem to be crucial to pair the physical and biological half-lives of the radionuclide and antibody, respectively. One may consider the target and use bifunctional chelates for stable, indirect radiolabeling chemistry when targeting rapid internalizing targets, as the label will be trapped in the lysosomes leading to increased accumulation of activity over time. Since directly iodinated proteins will be rapidly catabolized with the radioactivity excreted and eliminated via the kidneys, only targets with slow internalization rates are suitable to be imaged with this label. However, there are exceptions as demonstrated with C6.5 Mb in Figure 4. Another issue is that critical residues may be modified and impair the function of the antibody, especially as the fragment become smaller. If this is the case, site-specific conjugation and radiolabeling chemistry can be employed in order to retain function.

OTHER PROMISING NEW DIRECTIONS

Over the past 10–15 years, the antibody engineering field has moved towards even smaller scaffolds, encompassing a range of protein domain-based frameworks that are neither conventional antibodies nor peptides. These are characterized by small molecular size (single polypeptide chain format), high stability (thermodynamic and chemical), high solubility, and architecture suitable for modifications. Other attractive features may include high bacterial expression for inexpensive production, absence of disulfide bonds and human origin. These protein based reagents represent a new generation of highly specific and sensitive binding molecules that are immunoglobulin or non-immunoglobulin based ^{119, 120}. Immunoglobulin based scaffolds include immunoglobulin variable fragments (Fv), single domain antibodies from camelids and sharks, T cell receptor (TCR), cell adhesion molecules and fibronectin domains. Among non-immunoglobulin based scaffolds are affibodies, ankyrins, adnectins, avimers and anticalins.

Nanobodies® are derived from naturally occurring antibodies that lack the light chain. An anti-EGFR llama single-domain antibody fragment (8B6 Nanobody; 12 kDa) was recently evaluated for in vivo radioimmunodetection ¹²¹. The 8B6 Nanobody was radiolabeled with ^99mTc via its C-terminal histidine tag and evaluated in mice bearing high (A431) or moderate (DU145) EGFR expressing xenografts by SPECT. Although rapid clearance (t_1/2β = 1.5 h) from the blood was observed, a significantly higher radioactive uptake was seen in the high EGFR expressing A431 xenografts, demonstrating specificity at 3 hours post injection. Another camelid Nanobody targeting CEA was also radiolabeled with ^99mTc on a hexahistidine tag and evaluated for tumor targeting by SPECT ¹²². At 3 hours p.i. 3.2% ID/g was in the positive tumor, which was 3 and 6 times higher than the activities in the negative tumor and blood, respectively. Images showed specific localization to the positive tumor, and background activities in the liver and kidneys.

Affibodies (7 kDa), derived from staphylococcal surface Protein A, are one of the most characterized non-immunoglobulin based scaffolds (reviewed in ¹²³). Affibodies to Her2 have been radiolabeled with ¹²⁵I, ¹¹¹In, ^99m Tc, for SPECT imaging and ¹⁸F for PET imaging. Clear, high contrast images of the tumors were seen early in all the imaging studies performed. However, high activity in the kidneys was observed with the radiometal labeled protein. The anti-Her2 affibody has also been labeled with Alexa Fluor for near-infra red imaging. A second affibody molecule against EGFR has also shown successful targeting of tumors in gamma-camera imaging. A clinical pilot study demonstrated that anti-Her2 affibody could delineate tumors as early as 2 hours after administration. Thus it appears that affibody may be favorable molecules as alternative imaging agents.

CONCLUSIONS

It is clear that engineering antibodies for tumor targeting is not a trivial task as several factors regarding the antibody and the target need to be taken into account. For imaging, homogeneous tumor penetration is less important since heterogeneous distribution will still identify the tumor provided that expression of surface molecules is sufficient to discriminate the target tissue from background activity. However, the imaging time frame and radiation exposure to other tissues are governed by specificity and blood clearance of the immunoprobe. Several intact antibodies and to a lesser extent antibody fragments have been evaluated in patients (reviewed in ^{124, 125}). Although preliminary clinical studies with antibody fragment with gamma-camera imaging show encouraging results, to date only intact antibodies have been evaluated by PET. As a result, there is great interest in clinical translation of antibody fragments and alternative affinity scaffolds for tumor targeting with improved properties (i.e. rapid blood clearance), ideal for molecular imaging of surface markers. In addition to enabling early, same day imaging of malignant lesions that will accelerate evaluation of patient care and treatment, antibody-based molecular imaging tracers will play an important future role in the diagnosis and management of cancer and other diseases.