Dose predictions for [177Lu]Lu-DOTA-panitumumab F(ab′)2 in NRG mice with HNSCC patient-derived tumour xenografts based on [64Cu]Cu-DOTA-panitumumab F(ab′)2 – implications for a PET theranostic strategy

Background Epidermal growth factor receptors (EGFR) are overexpressed on many head and neck squamous cell carcinoma (HNSCC). Radioimmunotherapy (RIT) with F(ab')2 of the anti-EGFR monoclonal antibody panitumumab labeled with the β-particle emitter, 177Lu may be a promising treatment for HNSCC. Our aim was to assess the feasibility of a theranostic strategy that combines positron emission tomography (PET) with [64Cu]Cu-DOTA-panitumumab F(ab')2 to image HNSCC and predict the radiation equivalent doses to the tumour and normal organs from RIT with [177Lu]Lu-DOTA-panitumumab F(ab')2. Results Panitumumab F(ab')2 were conjugated to DOTA and complexed to 64Cu or 177Lu in high radiochemical purity (95.6 ± 2.1% and 96.7 ± 3.5%, respectively) and exhibited high affinity EGFR binding (Kd = 2.9 ± 0.7 × 10− 9 mol/L). Biodistribution (BOD) studies at 6, 24 or 48 h post-injection (p.i.) of [64Cu]Cu-DOTA-panitumumab F(ab')2 (5.5–14.0 MBq; 50 μg) or [177Lu]Lu-DOTA-panitumumab F(ab')2 (6.5 MBq; 50 μg) in NRG mice with s.c. HNSCC patient-derived xenografts (PDX) overall showed no significant differences in tumour uptake but modest differences in normal organ uptake were noted at certain time points. Tumours were imaged by microPET/CT with [64Cu]Cu-DOTA-panitumumab F(ab')2 or microSPECT/CT with [177Lu]Lu-DOTA-panitumumab F(ab')2 but not with irrelevant [177Lu]Lu-DOTA-trastuzumab F(ab')2. Tumour uptake at 24 h p.i. of [64Cu]Cu-DOTA-panitumumab F(ab')2 [14.9 ± 1.1% injected dose/gram (%ID/g) and [177Lu]Lu-DOTA-panitumumab F(ab')2 (18.0 ± 0.4%ID/g) were significantly higher (P < 0.05) than [177Lu]Lu-DOTA-trastuzumab F(ab')2 (2.6 ± 0.5%ID/g), demonstrating EGFR-mediated tumour uptake. There were no significant differences in the radiation equivalent doses in the tumour and most normal organs estimated for [177Lu]Lu-DOTA-panitumumab F(ab')2 based on the BOD of [64Cu]Cu-DOTA-panitumumab F(ab')2 compared to those estimated directly from the BOD of [177Lu]Lu-DOTA-panitumumab F(ab')2 except for the liver and whole body which were modestly underestimated by [64Cu]Cu-DOTA-panitumumab F(ab')2. Region-of-interest (ROI) analysis of microPET/CT images provided dose estimates for the tumour and liver that were not significantly different for the two radioimmunoconjugates. Human doses from administration of [177Lu]Lu-DOTA-panitumumab F(ab')2 predicted that a 2 cm diameter HNSCC tumour in a patient would receive 1.1–1.5 mSv/MBq and the whole body dose would be 0.15–0.22 mSv/MBq. Conclusion A PET theranostic strategy combining [64Cu]Cu-DOTA-panitumumab F(ab')2 to image HNSCC tumours and predict the equivalent radiation doses in the tumour and normal organs from RIT with [177Lu]Lu-DOTA-panitumumab F(ab')2 is feasible. RIT with [177Lu]Lu-DOTA-panitumumab F(ab')2 may be a promising approach to treatment of HNSCC due to frequent overexpression of EGFR. Supplementary Information The online version contains supplementary material available at 10.1186/s41181-021-00140-1.


Background
Head and neck squamous cell carcinoma (HNSCC) is the 7th most prevalent cancer in the world (Bray, 2018). In the United States, HNSCC is responsible for 3% of all cancers and 1.8% of all deaths due to cancer (Siegel, 2020). In Canada, cancers of the oral cavity, esophagus or larynx were predicted to cause 8950 new cases and 4200 deaths from cancer in 2020 (Brenner, 2020). HNSCC is treated by surgery followed by radiation (50-70 Gy) or chemoradiotherapy (CRT) with high dose cisplatin (100 mg/m 2 ) administered every 3 weeks for 3 cycles (Chow, 2020;De Felice et al., 2018;Schüttrumpf et al., 2020). For advanced HNSCC, this is often the definitive treatment (Chow, 2020). However, nephrotoxicity is a dose-limiting toxicity of cisplatin (Hoek et al., 2016) and grade 2 or grade 3-4 nephrotoxicity occur in up to 25% and 5-8% of patients, respectively (Saba et al., 2017). Reduction in the dose of cisplatin (< 50 mg/m 2 ) or substituting carboplatin may reduce these toxicities but is associated with poorer survival.
The epidermal growth factor receptor (EGFR) is overexpressed on 38-47% of HNSCC tumours and is a poor prognostic marker (Kalyankrishna and Grandis, 2006). Combining anti-EGFR chimeric monoclonal antibody (mAb) cetuximab (Erbitux®, Eli Lilly) with platinum-based chemotherapy and 5-fluorouracil (EXTREME regimen) showed modest improvement in overall survival (Vermorken et al., 2008) and this regimen is recommended in the National Comprehensive Cancer Network Guidelines for treatment of HNSCC (Colevas, 2018). Unfortunately, patients with recurrent or metastatic HNSCC resistant to this regimen have a poor prognosis and there is no accepted second line treatment. The fully human anti-EGFR mAb panitumumab (Vectibix, Amgen) has also shown promise for improving the outcome of patients with human papillomarvirus (HPV)negative HNSCC treated with CRT (Ferris et al., 2016).
Radioimmunotherapy (RIT) which links the β-particle-emitting radionuclide 177 Lu [t 1/2 = 6.7 d; Eβ max = 0.5 MeV (78.6%), Eβ max = 0.38 MeV (9.1%), Eβ max = 0.18 MeV (12.2%)] to anti-EGFR mAbs may be a promising approach for HNSCC, considering that EGFR are frequently overexpressed on tumours (Kalyankrishna and Grandis, 2006). RIT may also prove more effective than "naked" anti-EGFR mAbs, since it does not rely on inhibiting tumour growth signaling, but rather on inflicting lethal DNA damage. The feasibility of RIT for HNSCC is suggested by preclinical studies of cetuximab or panitumumab modified with DOTA (1,4,7,10-tetraazacyclododecane-1,4,7,10-tetraacetic acid) to complex [ 177 Lu] Lu, which reported strong tumour growth inhibition of HNSCC xenografts in BALB/c nude mice or athymic mice that were resistant to cetuximab or panitumumab (Liu et al., 2014;Song et al., 2017). We propose that patients with HNSCC could be selected for RIT with [ 177 Lu]Lu-labeled anti-EGFR mAbs by positron emission tomography (PET) with the corresponding [ 64 Cu]Cu-labeled mAbs [Eβ +max = 0.65 MeV (17.4%)] in a "PET theranostic" strategy. An analogous SPECT theranostic strategy that uses [ 177 Lu]Lu-labeled panitumumab to select patients for RIT is not as attractive due to the low abundance γ-photons emitted by 177 Lu [Eγ = 113 keV (6%) and 208 keV (10%)] which combined with the collimation required for SPECT decreases the sensitivity for tumour imaging. In addition, the higher absorbed doses for 177 Lu compared to 64 Cu are not suitable for a theranostic approach in which imaging is performed to select patients for RIT. Although many HNSCC are EGFR-positive, PET with [ 64 Cu]Cu-labeled mAbs would confirm lesions as EGFR-positive and importantly predict the tumour and normal tissue uptake of [ 177 Lu]Lu-labeled anti-EGFR mAbs. Moreover, information on the tumour and normal organ uptake can be used to estimate the radiation equivalent doses from RIT which may predict the effectiveness and normal organ toxicity in an individual patient. Since 64 Cu has a short half-life (t 1/2 = 12.7 h), F(ab') 2 fragments of mAbs that accumulate in tumours but are quickly eliminated from the blood and most normal organs in the feasible imaging time of ] would be most attractive for a PET theranostic strategy. F(ab') 2 also offer advantages for RIT because their more rapid elimination compared to IgG quickly decreases circulating radioactivity which is responsible for dose-limiting bone marrow toxicity from RIT due to the cross-fire effect from the 2 mm β-particles emitted by 177 Lu (Aghevlian et al., 2017). Moreover, F(ab') 2 unlike Fab do not exhibit high uptake in the kidneys, which minimizes kidney toxicity from RIT (Behr et al., 1996). We previously reported that panitumumab F(ab') 2 complexed to 64 Cu by NOTA (2,2′,2″-(1,4,7-triazacyclononane-1,4,7-triyl) triacetic acid) imaged s.c. PANC-1 human pancreatic cancer tumours that overexpress EGFR in non-obese diabetic severe combined immunodeficiency (NOD/scid) mice (Boyle et al., 2015). This study further showed that radioactivity in the blood in non-tumour bearing Balb/c mice was much lower at 18 h p.i. of [ 64 Cu]Cu-NOTA-panitumumab F(ab') 2 and Fab than for the intact IgG, but kidney uptake was much higher for Fab than F(ab') 2 , thus F(ab') 2 are preferred for PET and RIT. [ 177 Lu]Lu-DOTA-cetuximab F(ab') 2 have been reported to be effective for RIT of human colorectal xenografts in female SWISS nu/nu mice (Bellaye et al., 2018).
For a PET theranostic approach to be feasible, the tumour and normal organ uptake of [ 64 Cu]Cu-DOTA-panitumumab F(ab') 2 must be very similar to that of [ 177  assessed in BOD studies or by region-of-interest (ROI) analysis of microPET/CT images. We compared these predicted doses based on [ 64 Cu]Cu-DOTA-panitumumab F(ab') 2 to those estimated directly from the BOD of [ 177 Lu]Lu-DOTA-panitumumab F(ab') 2 to assess the feasibility of a PET theranostic strategy. The PDX mouse model of HNSCC used in this study is clinically relevant because these PDX recapitulate the properties of HNSCC tumours in patients (Karamboulas and Ailles, 2019). This strengthens our assessment of the feasibility of a PET theranostic strategy for patients with HNSCC.

Methods
Cell culture and patient-derived HNSCC xenografts MDA-MB-468 human breast cancer cells (1.3 × 10 6 EGFR/cell) (Reilly and Gariepy, 1998) were purchased from the American Type Culture Collection (ATCC, Manassas, VA) and cultured in RPMI 1640 medium (Sigma-Aldrich, St. Louis, MO) supplemented with 10% fetal bovine serum (FBS; Invitrogen, Carlsbad, CA) and 1% penicillin streptomycin (Sigma-Aldrich). A primary tumour specimen (#391) was surgically obtained from a patient with HNSCC under a protocol approved by the Research Ethics Board at the University Health Network . This tumour was dissected into small fragments (~1 mm 3 ) and engrafted subcutaneously (s.c.) on the right flank of NOD-Rag1 null IL2rg null (NRG) mice. These patient-derived tumour xenografts (PDX) were serially propagated in NRG mice following an Animal Care Protocol (No. 1542.28) approved by the Animal Care Committee at the University Health Network and following Canadian Council on Animal Care guidelines. The PDX used in this study were between the 3rd to 5th passage from the initial engraftment of the HNSCC tumour in NRG mice.

EGFR immunoreactivity
A saturation radioligand binding assay was performed to measure the dissociation constant (K d ) and maximum number of binding sites (B max ) for binding of [ 177 Lu]Lu-DOTA-panitumumab F(ab´) 2 to EGFR on MDA-MB-468 human breast cancer cells (1.3 × 10 6 EGFR/cell) (Reilly and Gariepy, 1998). Increasing concentrations of [ 177 Lu]Lu-DOTA-panitumumab F(ab´) 2 (0.098-200 nmoles/L) were incubated with 1 × 10 6 MDA-MB-468 cells in 200 μL of PBS, pH = 7.4 in 1.5 mL Eppendorf tubes at 4°C for 3 h with gentle shaking every 30 mins. The samples were then centrifuged at 1000×g for 2 mins on an Eppendorf Centrifuge 5424 (Thermo Fisher Scientific, Waltham, MA) and the supernatant containing unbound 177 Lu collected. The cell pellets were rinsed with ice cold PBS, pH 7.4, centrifuged again and the supernatant was collected and pooled with the previously collected supernatant. This was repeated twice. Total bound 177 Lu (TB) in the cell pellets and unbound 177 Lu in the supernatant were measured in a γ-counter (Model 1480; PerkinElmer, Waltham, MA). Non-specifically bound 177 Lu (NSB) was assessed by repeating the assay in the presence of 50-fold molar excess of panitumumab IgG. Specifically bound 177 Lu (SB) was calculated by subtracting NSB from TB. [ 177 Lu]Lu-DOTA-panitumumab F(ab´) 2 bound to MDA-MB-468 cells (pmoles) was plotted vs. the concentration of free (unbound) [ 177 Lu]Lu-DOTApanitumumab F(ab´) 2 (nmoles/L) and the curve was fitted to a one-site-receptor-binding model using Prism Ver. 4.0 software (GraphPad, San Diego, CA).

Biodistribution (BOD) studies and microPET/CT
The tumour and normal tissue uptake of [ 64 Cu]Cu-DOTA-panitumumab F(ab´) 2 were determined by BOD and microPET/CT studies in NRG mice with subcutaneous (s.c.) HNSCC PDX. Groups of 3-4 tumour-bearing NRG mice were injected i.v. (tail vein) with [ 64 Cu]Cu-DOTA-panitumumab F(ab´) 2 (5.5-14.0 MBq; 50 μg) and sacrificed at 6 h, 24 or 48 h post-injection (p.i.). The tumour and samples of blood and normal tissues were obtained, weighed and the radioactivity in each measured in a γ-counter using a window (425-640 keV) to include the 511 keV annihilation γ-photon of 64 Cu. Tumour and normal organ uptake were expressed as percent injected dose/g (%ID/g). Micro-PET/CT was performed after i.v. injection (tail vein) of 37 MBq (50 μg) of [ 64 Cu]Cu-DOTA-panitumumab F(ab´) 2 in a group of 4 NRG mice with s.c. HNSCC PDX. Mice were anaesthetized using 2% isoflurane in O 2 and were imaged in a supine position on a NanoScan® SPECT/CT/PET system (Mediso, Budapest, Hungary). PET images were acquired for 10, 20 and 40 mins at 6, 24 and 48 h p.i. of [ 64 Cu]Cu-DOTA-panitumumab F(ab´) 2 , respectively. Images were reconstructed by an ordered subset expectation maximization (OSEM) algorithm and consisted of 4 subsets and 4 iterations with attenuation and scatter correction supported by an isotropic voxel size of 300 μm. Prior to PET, CT images were acquired with 50 kVp X-rays, 980 μA and 300 msec exposure time. CT scans were reconstructed using the medium voxel and slice thickness settings, resulting in an isotropic voxel size of 250 μm. PET and CT images were coregistered by the Mediso Nucline NanoScan 3.00.020.0000 software. PET and CT DICOM files were exported using Mediso's Nucline Acquisition/Reconstruction Software to the Inveon Research Workplace Software 4.0 (Siemens) for analysis and quantification of 64 Cu uptake (%ID/g) as well as estimation of the volume of the liver and tumour. Regions of interest (ROI) were drawn around the tumour and liver on the PET images aided by delineation of the anatomy on the CT images to compare the accuracy of PET for quantifying the tumour and liver uptake compared to BOD studies. The ROI for each organ was drawn two-dimensionally on axial slices of the image. To ensure the coverage of the entire organ, > 5 ROIs were drawn before the slices were convoluted to form a 3D volume that encompassed the entire organ.

Biodistribution (BOD) studies and microSPECT/CT
The tumour and normal tissue uptake of [ 177 Lu]Lu-DOTA-panitumumab F(ab´) 2 was determined by BOD and microSPECT/CT studies. [ 177 Lu]Lu-DOTA-panitumumab F(ab´) 2 (6.5 MBq; 50 μg) were injected i.v. (tail vein) in NRG mice with s.c. HNSCC PDX and groups (n = 3-4) of mice were sacrificed at 6 h, 24 h or 48 h p.i. The tumour, and samples of blood and other tnormal issues were obtained, weighed and counted in a γ-counter using a window (130-470 kev) to include the γ-photons of 177 Lu [Eγ = 113 keV (6.6%) and Eγ = 208 keV (11%)]. Tumour and normal organ uptake was expressed as %ID/g. The excised tumours were subjected to immunohistochemical (IHC) staining with anti-human EGFR antibodies (Invitrogen, Carlsbad, CA; Cat. No. 28-8763) to confirm EGFR positivity. In addition, the tumour and normal tissue BOD of irrelevant [ 177 Lu]Lu-DOTA-trastuzumab F(ab´) 2 (2.90 MBq; 50 μg) were determined in a separate group of 5 NRG mice bearing HNSCC PDX at 24 h p.i. The excised tumours in these mice were stained for HER2 using anti-human HER2 antibodies (Invitrogen Cat. No. MA5-14509). Representative mice were anaesthetized using 2% isoflurane in O 2 and microSPECT/CT images were acquired in a supine position, at 6 h, 24 h and 48 h p.i. of [ 177 Lu]Lu-DOTA-panitumumab F(ab´) 2 on a NanoScan® SPECT/CT/PET system (Mediso). Images were acquired in a 256 × 256 matrix. A window (± 10%) was set around each of the γ-photopeaks (208.4 keV; 112.9 keV; 56.1 keV) of 177 Lu. A Mediso APT62 collimator (WB-HS standard) was affixed to each of the 4 detector NaI (TI) detector heads. Images were reconstructed by Monte Carlo methods with three subsets of data undergoing 48 iterations using the Mediso Nucline NanoScan acquisition and reconstruction software (ver 3.00.020.0000). Prior to SPECT, imaging CT images were acquired with 50 kVp X-rays, 980 μA and a 300 msec exposure time. CT scans were reconstructed using the medium voxel and slice thickness settings resulting in an isotropic voxel size of 250 μm. SPECT and CT were co-registered by the Mediso Nucline acquisition/reconstruction software. MicroSPECT/CT images were similarly obtained for NRG mice with HNSCC PDX at 24 h p.i. of [ 177 Lu]Lu-DOTA-trastuzumab F(ab´) 2 .
All animal studies were conducted under a protocol (AUP 2843.8) approved by the Animal Care Committee at the University Health Network following Canadian Council on Animal Care guidelines.

Tumour and normal organ dosimetry
The radiation equivalent doses in NRG mice with s.c. HNSCC PDX after i.v. injection of [ 64 Cu]Cu-DOTA-panitumumab F(ab´) 2 or [ 177 Lu]Lu-DOTA-panitumumab F(ab´) 2 were estimated from the tumour and normal organ uptake of activity in source organs measured in BOD studies. The mean equivalent dose (D) was calculated as D = Ã s × S × W R , where Ã s is the time-integrated activity in the source organs or tumour and S are the Snyder values for mice (Bitar et al., 2007;Xie and Zaidi, 2013) and W R is the radiation weighting factor. W R is 1 for x-rays, γ rays and β-particles, thus taken as 1 in this study. Ã s in the source organs or in the tumour were estimated using Prism Ver. 4.0 software (GraphPad) from the area-under-the-curve (AUC) up to 48 h p.i. of [ 64 Cu]Cu-DOTA-panitumumab F(ab´) 2 or [ 177 Lu]Lu-DOTA-panitumumab F(ab´) 2 (AUC 0-48 h; Bq × sec) derived from the activity vs. time curves. The activity/source organ at each time point t (s) was calculated using the formula %ID/g (t) × ID/100 × organ weight × exp.(−kt), where %ID/g was obtained from the BOD studies, ID was the injected dose in Bq, and k was the decay constant for 64 Cu (1.52 × 10 − 5 s − 1 ) or 177 Lu (1.21 × 10 − 6 s − 1 ). The time integrated activity from 48 h p.i. to infinity (A 48 h -∞ ; Bq × sec) was calculated by dividing the activity at 48 h p.i. by the decay constant for 64 Cu or 177 Lu, assuming further elimination of activity from source organs only by radioactive decay. The S-value for the tumour was estimated using the sphere model in OLINDA/ EXM software based on the measured tumour mass (Stabin et al., 2005). This was repeated with the estimated uptake in the tumour and liver determined by ROI analysis of the microPET/CT images at 6 h, 24 h and 48 h p.i. of [ 64 Cu]Cu-DOTA-panitumumab F(ab´) 2 . To predict the doses in normal organs in humans, the activity in source organs of a human adult female with a 2 cm diameter spherical tumour in the neck at the different times up to 48 h p.i. of the RICs were estimated based on proportional extrapolation of the activities in mice using the % kg/g method, i.e. (%ID/organ) human = [(%ID/ organ) mouse × (mouse body weight/human body weight)] (Kirschner et al., 1973). Mouse and human body weight used in the calculation were set to 30 g and 56,900 g, respectively. The time-integrated activity from 0 to 48 h p.i. (AUC 0-48 h; Bq × sec) and from 48 h p.i. to infinity (A 48 h -∞ ; Bq × sec) were calculated as described earlier for mice, and the combined A 0h to ∞ for each human source organ was used to predict the equivalent doses in human female adults (mSv/Bq) using OLINDA/EXM 1.0 software (Stabin et al., 2005).

Statistical analysis
All results were reported as mean ± SEM. Statistical comparisons for normal organ and tumour uptake of [ 64 Cu]Cu-DOTA-panitumumab F(ab´) 2 and [ 177 Lu]Lu-DOTA-panitumumab F(ab´) 2 were made using a Mann-Whitney U test (P < 0.05) and Prism 4.0 software (GraphPad). Statistical comparisons for radiation equivalent dose were made using a two-tailed unpaired Student's t-test (P < 0.05) using PrismVersion 4.0 software (GraphPad).

[ 64 Cu]cu-or [ 177 Lu]Lu-DOTA-panitumumab F(ab´) 2
Panitumumab F(ab´) 2 were produced by proteolytic digestion of the IgG using immobilized pepsin. SDS-PAGE analysis (Fig. 1a) under non-reducing conditions revealed one major band for F(ab´) 2 corresponding to the expected M r~1 10 kDa, while panitumumab IgG migrated as a band with M r~1 50 kDa. Under reducing conditions, F(ab´) 2 migrated as one major band at M r~2 4 kDa, corresponding to the variable and the constant region of the heavy chain (V H -C H ) and the light chain (V L -C L ). Panitumumab IgG migrated as two major bands under reducing conditions with M r = 50 kDa and 24 kDa, corresponding to the dissociated heavy and light chains, respectively. SE-HPLC with UV detection at 280 nm demonstrated a single peak for panitumumab F(ab´) 2 with a retention time (t R ) = 15.21 mins (Fig. 1b). Panitumumab F(ab´) 2 were conjugated to 3.8 ± 0.4 DOTA per F(ab´) 2 . DOTA-panitumumab F(ab´) 2 were labeled with 64 Cu or 177 Lu to a RCP of 95.6 ± 2.1% and 96.7 ± 3.5%, respectively, measured by ITLC-SG developed in 100 mM sodium citrate, pH 5.5. SE-HPLC confirmed high RCP for these RICs with a single peak a t R = 15.38 mins for [ 64 Cu]Cu-DOTA-panitumumab F(ab´) 2 (Fig. 1c) and [ 177 Lu]Lu-DOTA-panitumumab F(ab´) 2 (not shown). [ 177 Lu]Lu-DOTA-panitumumab F(ab´) 2 exhibited saturable binding to EGFR on MDA-MB-468 cells that was displaced by a 50-fold molar excess of panitumumab IgG (Fig. 2). Fitting of the SB curve to a one-site receptor binding model revealed a K d = 2.9 ± 0.7 × 10 − 9 mol/L indicating high affinity binding to EGFR. The B max for MDA-MB-468 cells was 1.1 ± 0.1 × 10 6 EGFR/cell which was similar that reported for binding of EGF or anti-EGFR mAb 528 labeled with 111 In to these cells (1.3 × 10 6 EGFR/cell) (Reilly et al., . Trastuzumab F(ab´) 2 was produced similarly and derivatized with 3.3 ± 0.02 DOTA which enabled labeling with 177 Lu to high RCP (97.4 ± 0.1%) measured by ITLC-SG. SDS-PAGE analysis (Fig. S1) revealed a major protein band with M r~1 00 kDa for trastuzumab F(ab´) 2 and a major band with the expected M r~1 50 kDa for trastuzumab IgG under non-reducing conditions. Under reducing conditions, trastuzumab F(ab´) 2 migrated as a single band at M r~2 5 kDa, corresponding to the variable and constant region of the heavy chain (V H -C H ) and the light chain (V L -C L ) while trastuzumab IgG migrated as two main bands with M r~5 0 kDa and 25 kDa, corresponding to the dissociated heavy and light chains, respectively.  DOTA-trastuzumab F(ab´) 2 (27.4 ± 2.7%ID/g) was significantly higher than [ 64 Cu]Cu-DOTA-panitumumab F(ab´) 2 (4.7 ± 0.5%ID/g; P = 0.04) or [ 177 Lu]Lu-DOTA-panitumumab F(ab´) 2 (5.9 ± 0.0%ID/g; P = 0.04). Liver uptake of [ 177 Lu]Lu-DOTA-trastuzumab F(ab´) 2 (8.0 ± 1.0%ID/g) was significantly lower than [ 177 Lu]Lu-DOTA-panitumumab F(ab´) 2 (12.1 ± 1.1%ID/g; P = 0.04) but not significantly different than [ 64 Cu]Cu-DOTA-panitumumab F(ab´) 2 (7.5 ± 1.0%ID/g; P > 0.05). A comparison of the tumour and liver uptake of [ 64 Cu]Cu-DOTA-panitumumab F(ab´) 2 measured by ROI analysis of microPET/CT images or BOD studies is shown in Table 2. There were no significant differences in the measured tumour uptake but the ROI analysis modestly overestimated the liver uptake of [ 64 Cu]Cu-DOTA-panitumumab F(ab´) 2 at 48 h p.i.

Discussion
We report here for the first time radiation equivalent dose predictions for [ 177   BOD studies in NRG mice with s.c. HNSCC PDX at 6 or 24 p.i. showed similar uptake of these two RICs (Table 1) with only modest but significant differences in tissue localization at 48 h p.i. between [ 64 Cu]Cu-DOTA-panitumumab F(ab´) 2 and [ 177 Lu]Lu-DOTA-panitumumab F(ab′) 2 . At 6 h p.i., uptake of [ 64 Cu]Cu-DOTA-panitumumab F(ab´) 2 was lower than [ 177 Lu]Lu-DOTA-panitumumab F(ab′) 2 in the intestine, spleen, kidneys and skin but higher in the blood. At 24 h p.i., lung uptake of [ 64 Cu]Cu-DOTA-panitumumab F(ab´) 2 was higher than [ 177 Lu]Lu-DOTA-panitumumab F(ab′) 2 but liver uptake was lower. At 48 h p.i., the uptake of [ 64 Cu]Cu-DOTA-panitumumab F(ab´) 2 was higher than [ 177 Lu]Lu-DOTA-panitumumab F(ab′) 2 in the lungs, stomach, pancreas, intestines and blood but lower in the liver and skin (P < 0.05). These modest differences in BOD may be due to experimental variability in tissue uptake in BOD studies. It is possible that the lower stability constant of DOTA for complexing Cu 2+ (log K = 23.3)  vs. Lu 3+ (log K = 26.7) may cause some differences in BOD between the two RICs (Majkowska and Bilewicz, 2007). The lower uptake of [ 64 Cu]Cu-DOTA-panitumumab F(ab´) 2 in the liver than [ 177 Lu]Lu-DOTA-panitumumab F(ab′) 2 at 24 h p.i. was unexpected, since release of 64 Cu from DOTA should increase liver uptake (Hausner et al., 2009). There were no significant differences in tumour uptake of [ 64 Cu]Cu-DOTA-panitumumab F(ab´) 2 and [ 177 Lu]Lu-DOTA-panitumumab F(ab′) 2 at any time point. In patients, the tumour and normal organ uptake of [ 64 Cu]Cu-DOTA-panitumumab F(ab´) 2 would be measured by ROI analysis to predict doses from [ 177 Lu]Lu-DOTA-panitumumab F(ab′) 2 . Thus, we compared the uptake of the RICs into the tumour and liver measured by ROI analysis of microPET/CT images with that measured in BOD studies (Supplementary Information  Table S1). There were no significant differences in quantification of tumour uptake by these two methods, but ROI analysis overestimated the liver uptake. This may be due to imprecise delineation of the liver ROI resulting in some inaccuracy in measuring liver radioactivity. Manual segmentation in ROI analysis on PET images is subject to variability in delineation, contributing to differences in the measuring the radioactivity (Foster et al., 2014).
The radiation equivalent doses in the tumour and normal organs were calculated as D = Ã s × S × W R , where Ã s is the time-integrated activity in the tumour or normal source organs and S are the Snyder values for mice (Bitar et al., 2007;Xie and Zaidi, 2013) and W R is the radiation weighing factor. The total Ã s was calculated by summing the area-under-the-curve in source organs from time zero to 48 h p.i. (AUC 0-48 h; Bq × sec) and the Ã 48 h -∞ from 48 h p.i. to infinity (Bq × sec). The calculation of Ã s tends to minimize variability in BOD between [ 64 Cu]Cu-DOTA-panitumumab F(ab´) 2 and [ 177 Lu]Lu-DOTA-panitumumab F(ab′) 2 at individual time points. Consequently, we found no significant differences in the doses estimated for [ 177 Lu]Lu-DOTA-panitumumab F(ab′) 2 based on the BOD of [ 64 Cu]Cu-DOTA-panitumumab F(ab´) 2 compared to those based on the BOD of [ 177 Lu]Lu-DOTA-panitumumab F(ab′) 2 except for the liver and whole body, which were modestly underestimated by [ 64 Cu]Cu-DOTA-panitumumab F(ab´) 2 (Table 3). The higher absorbed dose in the liver for [ 177 Lu]Lu-DOTA- Ku et al. EJNMMI Radiopharmacy and Chemistry (2021)