A comparison of 111In- or 64Cu-DOTA-trastuzumab Fab fragments for imaging subcutaneous HER2-positive tumor xenografts in athymic mice using microSPECT/CT or microPET/CT

Background Our objective was to compare 111In- or 64Cu-DOTA-trastuzumab Fab fragments for imaging small or large s.c. tumor xenografts in athymic mice that display a wide range of human epidermal growth factor receptor-2 (HER2) expression using microSPECT/CT or microPET/CT. Methods Trastuzumab Fab were labeled with 111In or 64Cu by conjugation to 1,4,7,10-tetraazacyclododecane N, N', N'', N'''-tetraacetic acid (DOTA). The purity of 111In- and 64Cu-DOTA-trastuzumab Fab was measured by SDS-PAGE and HPLC. HER2 binding affinity was determined in saturation radioligand binding assays using SKBR-3 cells (1.3 × 106 HER2/cell). MicroSPECT/CT and microPET/CT were performed in athymic mice bearing s.c. BT-20 and MDA-MB-231 xenografts with low (0.5 to 1.6 × 105 receptors/cell), MDA-MB-361 tumors with intermediate (5.1 × 105 receptors/cell) or SKOV-3 xenografts with high HER2 expression (1.2 × 106 receptors/cell) at 24 h p.i. of 70 MBq (10 μg) of 111In-DOTA-trastuzumab Fab or 22 MBq (10 μg) of 64Cu-DOTA-trastuzumab Fab or irrelevant 111In- or 64Cu-DOTA-rituximab Fab. Tumor and normal tissue uptake were quantified in biodistribution studies. Results 111In- and 64Cu-DOTA-trastuzumab were > 98% radiochemically pure and bound HER2 with high affinity (Kd = 20.4 ± 2.5 nM and 40.8 ± 3.5 nM, respectively). MDA-MB-361 and SKOV-3 tumors were most clearly imaged using 111In- and 64Cu-DOTA-trastuzumab Fab. Significantly higher tumor/blood (T/B) ratios were found for 111In-DOTA-trastuzumab Fab than 111In-DOTA-rituximab Fab for BT-20, MDA-MB-231 and MDA-MB-361 xenografts, and there was a direct association between T/B ratios and HER2 expression. In contrast, tumor uptake of 64Cu-DOTA-trastuzumab Fab was significantly higher than 64Cu-DOTA-rituximab Fab in MDA-MB-361 tumors but no direct association with HER2 expression was found. Both 111In- and 64Cu-DOTA-trastuzumab Fab imaged small (5 to 10 mm) or larger (10 to 15 mm) MDA-MB-361 tumors. Higher blood, liver, and spleen radioactivity were observed for 64Cu-DOTA-trastuzumab Fab than 111In-DOTA-trastuzumab Fab. Conclusions We conclude that 111In-DOTA-trastuzumab Fab was more specific than 64Cu-DOTA-trastuzumab Fab for imaging HER2-positive tumors, especially those with low receptor density. This was due to higher levels of circulating radioactivity for 64Cu-DOTA-trastuzumab Fab which disrupted the relationship between HER2 density and T/B ratios. Use of alternative chelators that more stably bind 64Cu may improve the association between T/B ratios and HER2 density for 64Cu-labeled trastuzumab Fab.


Background
The human epidermal growth factor receptor-2 (HER2) is overexpressed in 20% to 25% of breast cancers (BC) and is the target for treatment with trastuzumab (Herceptin), a humanized IgG 1 monoclonal antibody (mAb) [1,2]. HER2 amplification is normally assessed ex vivo in a primary tumor biopsy by immunohistochemical (IHC) staining for HER2 protein or by fluoresecence in situ hybridization to detect increased HER2 gene copy number [3]. However, discordance in HER2 expression between primary and metastatic BC has been found in 20% to 30% of cases [4,5] and thus, it would be useful to have an imaging technique to assess HER2 phenotype in situ in BC lesions. Several investigators have shown that HER2 expression can be imaged in human BC xenografts in athymic mice by single photon emission computed tomography (SPECT) using trastuzumab or its Fab fragments labeled with 111 In or 99 m Tc [6][7][8][9]. These studies have been extended to imaging HER2-positive BC in patients using 111 In-labeled trastuzumab IgG [2,10]. More recently, positron-emission tomography (PET) using trastuzumab labeled with 89 Zr has shown promise for imaging HER2 expression in tumor xenograft mouse models and in patients with metastatic BC [11,12]. Imaging also offers an opportunity to detect response to HER2-targeted therapies in BC [13]. We previously reported that SPECT with 111 In-labeled pertuzumab (anti-HER2) detected early response to treatment with trastuzumab (Herceptin) in athymic mice bearing s.c. MDA-MB-361 BC xenografts [14]. Smith-Jones et al. demonstrated that PET with 68 Ga-labeled trastuzumab F (ab') 2 fragments identified response of HER2-positive BT-474 human BC tumors in mice to treatment with heat shock protein (Hsp90) inhibitors [15].
PET offers several potential advantages compared to SPECT for imaging tumors because it has higher intrinsic sensitivity, is more easily quantified, and in some instances offers higher spatial resolution. Despite these apparent benefits, few studies have reported a comparison of PET and SPECT for imaging HER2-positive tumors using the same agent labeled with a single photon-emitter or positron-emitter. Dijkers et al. compared 89 Zr-and 111 In-labeled trastuzumab in mice bearing s.c. SK-OV-3 human ovarian cancer xenografts and reported no significant differences in tumor and normal tissue uptake [12]. MicroPET with 89 Zr-labeled trastuzumab visualized these tumors, but the corresponding microSPECT images with 111 In-labeled trastuzumab were not presented.
In this study, we compared microSPECT/CT and microPET/CT for imaging s.c. human tumor xenografts expressing a wide range of HER2 density in athymic mice using trastuzumab Fab fragments modified with 1,4,7,10-tetraazacyclododecane N, N', N″, N'″-tetraacetic acid (DOTA) for complexing 111 In or 64 Cu. 64 [16,17]. 64 Cu complexed to DOTA and linked to mAbs and peptides has been widely studied for PET imaging of tumors [15,[18][19][20][21][22][23]. Fab fragments were selected for these studies because their pharmacokinetics of tumor uptake and elimination from the blood and normal tissues is compatible with the half-lives of 64 Cu and 111 In [24].

DOTA conjugation and radiolabeling of Fab fragments
Trastuzumab or rituximab Fab fragments were modified with DOTA for complexing 111 In or 64 Cu by reaction of 1.5 mg of Fab in 300 μL of NaHCO 3 buffer (pH 7.5) with a 60-or 90-fold excess, respectively, of the Nhydroxysuccinimidyl ester of 1,4,7,10-tetraazacyclododecane tetraacetic acid (NHS-DOTA; Macrocyclics, Dallas, TX, USA). The conjugation reaction was performed at 4°C for 18 h. DOTA-conjugated Fab were purified from excess DOTA by transferring to an Amicon Ultracel 30 K device, diluting to 12.0 mL with 1 M CH 3 COONH 4 buffer, pH 6.0 and centrifuging at 4,000 × g for 15 min. This purification step was repeated six times. Finally, purified DOTA-Fab were recovered and the concentration determined spectrophotometrically [E 280 nm = 1.45 (mg/mL) -1 cm -1 ]. The final concentration was adjusted to 5 mg/mL with 1 M CH 3 COONH 4 buffer, pH 6.0.
Radiolabeling was performed by incubating 50 μg of DOTA-Fab in 10 μL of CH 3 COONH 4 buffer, pH 6.0 with 360 MBq of 111 InCl 3 (> 7 GBq/mL; MDS-Nordion, Kanata, ON, Canada) or 216 MBq of 64 CuCl 2 (> 4 GBq/ mL; MDS-Nordion) for 3 h at 46°C. 111 In-or 64 Cu-labeled DOTA-Fab were purified on an Amicon Ultracel 30 K device. The final radiochemical purity was measured by instant thin layer-silica gel chromatography (ITLC-SG; Pall Life Sciences, Ann Arbor, MI, USA) developed in 100 mM sodium citrate, pH 5.0 or by size-exclusion HPLC using a flow-through radioactivity detector (FSA; PerkinElmer). The R f values for 111 In-or 64 Cu-DOTA-Fab on ITLC were 0.0 and those for 111 In-or 64 Cu-DOTA or free radionuclides were 1.0. The DOTA substitution level of the Fab fragments (chelators/molecule) was measured by labeling a 10 μL aliquot of the unpurified conjugation reaction with 111 In, then determining the proportion of 111 In-DOTA-Fab vs. free 111 In-DOTA by ITLC-SG and multiplying this fraction by the molar ratio used in the reaction [26]. The HER2 binding affinity of 111 In-and 64 Cu-DOTAtrastuzumab Fab was determined by direct (saturation) radioligand binding assays using SKBR-3 human BC cells (1.3 × 10 6 HER2/cell) [9]. Briefly, increasing concentrations (0 to 600 nmol/L) of 111 In-or 64 Cu-DOTAtrastuzumab Fab were incubated with 1 × 10 5 cells in 24-well plates at 4°C for 3 h. Unbound radioactivity was removed and the dishes were rinsed two times with phosphate-buffered saline. The cells were dissolved in 100 mM NaOH, recovered, and the total cell-bound radioactivity (TB) was measured in a γ-counter (Perki-nElmer Wizard 3). The assay was repeated in the presence of 16 μmol/L of unlabeled trastuzumab IgG to measure non-specific binding (NSB). Specific binding (SB; nanomoles per liter) was calculated by subtracting NSB from TB and was plotted vs. Mice were euthanized by cervical dislocation under general anaesthesia. Tumor and normal tissue uptake of radioactivity was measured in a γ-scintillation counter (Wizard 3, PerkinElmer, Waltham, MA) was expressed as percent injected dose per gram and as tumor/normal tissue (T/NT) ratios. The relationship between tumor/blood (T/B) ratios and HER2 density was examined. The uptake of 111 In-or 64 Cu-DOTA-trastuzumab Fab fragments in small (5 to 10 mm diameter) vs. larger (10 to 15 mm diameter) tumor xenografts was compared.

MicroSPECT and microPET imaging
MicroSPECT was performed at 24 h p.i. of 70 MBq (10 μg) of 111 In-DOTA-trastuzumab Fab or 111 In-DOTArituximab Fab in athymic mice with s.c. HER2-positive tumor xenografts. Anaesthesia was induced and maintained by inhalation of 2% isoflurane in O 2 . MicroSPECT was performed on a NanoSPECT/CT tomograph (Bioscan, Washington, DC, USA) equipped with four NaI scintillation detectors fitted with 1.4-mm multi-pinhole collimators [full-width half-maximum (FWHM) = 1.2 mm]. A total of 24 projections were acquired in a 256 × 256 matrix with a minimum of 80,000 counts per projection. Micro-SPECT image acquisition time was 85 to 120 mins. Micro-SPECT images were reconstructed using an orderedsubset expectation maximization (OSEM) algorithm (nine iterations). Prior to microSPECT imaging, cone-beam CT images were acquired (180 projections, 1 s/projection, 45 kVp) on the NanoSPECT/CT system. Co-registration of microSPECT and CT images was performed using Invivo-Scope software (Bioscan).
MicroPET was performed at 24 h p.i. of 22 MBq (10 μg) of 64 Cu-DOTA-trastuzumab Fab or 64 Cu-DOTA-rituximab Fab on a Focus 220 microPET tomograph (Siemens Preclinical Solutions, Knoxville, TN, USA). Images were acquired for 20 mins and reconstructed using OSEM, followed by a maximum a posteriori probability reconstruction algorithm with no correction for attenuation or partial-volume effects. The FWHM resolution of the microPET tomograph was 1.6 mm. Immediately after imaging, CT was performed on an eXplore Locus Ultra Preclinical CT scanner (GE Healthcare, Mississauga, ON, Canada) with routine acquisition parameters (80 kVp, 70 mA, and voxel size of 150 × 150 × 150 mm). MicroPET and CT images were coregistered using Inveon Research Workplace software (Siemens). All animal studies were conducted under a protocol (no. 989.9) approved by the Animal Use Committee at the University Health Network following Canadian Council on Animal Care guidelines.

Statistical analyses
Statistical significance of comparisons were assessed by Student's t test (P < 0.05).
expression. The range of HER2 expression examined (5.4 × 10 4 to 1.2 × 10 6 receptors/cell) corresponded to HER2 scores of 0 to 3+ assessed clinically in BC specimens by IHC staining [9]. There was no apparent increased ability of microPET/CT with 64 Cu-DOTAtrastuzumab Fab compared to microSPECT/CT with 111 In-DOTA-trastuzumab Fab to visualize MDA-MB-231 tumors with low HER2 density (1.6 × 10 5 receptors/ cell; Figures 4a and 5a). In addition, there was no increased ability of microPET/CT using 64 Cu-DOTAtrastuzumab Fab to image small (5 to 10 mm diameter) or larger (10 to 15 mm diameter) MDA-MB-361 tumors with intermediate HER2 expression (5.1 × 10 5 HER2/ cell; Figure 6). The intensity of the tumor signal was dependent on HER2 expression with tumors with intermediate (MDA-MB-361) or high (SK-OV-3) HER2 density most readily imaged by microSPECT/CT (Figure 4) or microPET/CT ( Figure 5). However, a threefold higher dose of radioactivity was administered for microSPECT/ CT than microPET/CT (70 vs. 22 MBq) and image acquisition times were up to sixfold longer for micro-SPECT/CT (85 to 20 vs. 20 min, respectively). Thus, the photon detection efficiency (i.e., intrinsic sensitivity) was much higher for microPET/CT than microSPECT/CT. Nonetheless, our results revealed that provided that the administered dose of radioactivity was sufficient and image acquisition times were long enough to yield good counting statistics, microSPECT/CT with 111 In-DOTAtrastuzumab Fab was able to image tumors with the similar HER2 density and size as microPET/CT with 64 Cu-DOTA-trastuzumab. These results agree with those reported by Cheng et al. who noted that s.c. HER2-positive SUM190 tumor xenografts were imaged by either microSPECT or microPET using trastuzumab conjugated to biotinylated 99 m Tc-or 18 F-labeled phosphodiamidate morpholinos (MORFs) through a streptavidin linker [28]. The doses of 99 m Tc or 18 F used in their study were 13 and 0.22 MBq, respectively. Phantom studies revealed that microPET was 15-fold more sensitive in terms of photon detection, but the spatial resolution of microSPECT was superior to that of microPET (1.2 vs. 2.4 mm). The results are also in concordance with those reported by Wong et al. [29], who showed that s.c. epidermal growth factor receptor-positive LS174T human colon cancer xenografts could be imaged using panitumomab F (ab') 2 fragments labeled with 111 In or 86 Y. However, they compared low resolution planar γ-camera imaging with microPET. In our study, we used similar quality high resolution and high sensitivity small animal imaging technologies, namely microSPECT/CT (NanoSPECT; Bioscan) and microPET (Siemens Focus 220) systems for these comparisons.
T/B ratios were used to compare the tumor localization of 111 In-and 64 Cu-DOTA-trastuzumab Fab vs. HER2 density because we previously found that there are differences in perfusion between different tumor xenografts which can affect the uptake of radioimmunoconjugates [9]. Use of T/B ratios minimizes these effects by normalizing for blood concentrations which then reveals the relationships between HER2 density and tumor accumulation. Moreover, the T/B ratios are important for discriminating tumors that have different HER2 expression on the images. There was a strong and direct association between the T/B ratios for 111 In-DOTA-trastuzumab and tumor HER2 density (Figure 3a). In addition, the T/B ratios for 111 In-DOTA-trastuzumab Fab were significantly greater than those of irrelevent control 111 In-DOTA-rituximab  (Table 1). Thus, T/B ratios were threefold lower for 64 Cu-than 111 In-DOTA-trastuzumab (3.6:1 vs. 10:1; Table 2) in mice with MDA-MB-361 tumors. The increased circulating radioactivity for 64 Cu-DOTA-trastuzumab Fab may be due to kinetic instability of the 64 Cu-DOTA complex with transchelation of released 64 Cu to copper binding proteins (e.g., albumin, ceruloplasmin, or superoxide dismutase) [26]. These 64 Cu-labeled proteins may non-specifically localize in tumors, disrupting the association between T/B ratios and HER2 density, especially for tumors with low HER2 expression (i.e., MDA-MB-231 and BT-20).
DOTA forms thermodynamically stable complexes with copper (K d = 10 23 M -1 ) but kinetic instability of 64 Cu-DOTA complexes in vivo can lead to loss of radiometal resulting in high blood radioactivity and liver and spleen uptake [17]. In addition to the higher levels of blood radioactivity, we found that the liver and spleen uptake for 64 Cu-DOTA-trastuzumab Fab were three-and twofold greater, respectively, than 111 In-DOTA-trastuzumab Fab (Table 1). In order to improve the kinetic stability of 64 Cu complexes, more thermodynamically stable cross-bridged (CB-DO2A) or sarcophagine (SarAr) chelators have been synthesized [30,31] DOTA or CB-DO2A (but not conjugated to mAbs) showed fourfold lower radioactivity in the blood and twofold lower liver accumulation at 24 h p.i. in rats [30]. Voss et al. noted that ch14.18 mAbs labeled with 64 Cu through the extremely stable SarAr chelator for PET imaging of neuroblastoma or melanoma xenografts in mice exhibited low liver uptake (5% to 10% i.d./g) but no comparison with other chelators was provided [31]. Dearling et al. recently compared the tumor and normal tissue distribution of these same 64 Cu-labeled ch14.18 mAbs using a variety of chelators including DOTA and SarAr in mice bearing M21 melanoma xenografts [17]. Unexpectedly, no significant differences in tumor or liver uptake were found for ch14.18 labeled with 64 Cu using DOTA or the much more stable SarAr chelator. They suggested that in addition to 64 Cu-chelator stability, factors such as the net charge on the chelators may play an important role in sequestration of radioactivity by tissues. In our study, tumor uptake was not significantly different between 111 In-and 64 Cu-DOTA-trastuzumab Fab in mice with MDA-MB-361 tumors, despite the apparent instability of 64 Cu-DOTA-trastuzumab Fab as evidenced by higher levels of radioactivity in the blood, liver, and spleen ( Table  1). The use of more stable chelators such as CB-DO2A or SarAr may diminish blood radioactivity and improve the association between tumor HER2 density and T/B ratios for 64 Cu-labeled trastuzumab Fab. The CB-DO2A and SarAr chelators are unfortunately not yet commercially available in a chemically reactive form for conjugation to mAbs for 64 Cu labeling.

Conclusion
Provided that administered doses of radioactivity and acquisition times were sufficient to yield good counting statistics, we conclude that either microSPECT/CT with 111 In-DOTA-trastuzumab Fab or microPET/CT with 64 Cu-DOTA-trastuzumab Fab visualized small (5 to 10 mm diameter) or larger (10 to 15 mm diameter) s.c. tumor xenografts with low, intermediate, or high HER2 expression in athymic mice. However, due to the higher levels of circulating radioactivity for 64 Cu-DOTA-trastuzumab Fab, no association between HER2 density and T/B ratios was established. In contrast, there was a strong direct association between T/B ratios and HER2 density of these tumors for 111 In-DOTA-trastuzumab Fab. Thus, 111 In-DOTA-trastuzumab Fab was more specific than 64 Cu-DOTA-trastuzumab Fab for imaging HER2-positive tumors with low HER2 density. The use of more stable CB-DO2A or SarAr chelators for 64 Cu may potentially diminish blood radioactivity, provide a stronger association between T/B ratios and tumor HER2 density, and improve the specificity of imaging with 64 Cu-labeled trastuzumab Fab.