In vitro and in vivo pre-clinical analysis of a F(ab')₂ fragment of panitumumab for molecular imaging and therapy of HER1-positive cancers

Monoclonal antibodies (mAb) have been used in medicine for nearly three decades for purposes including imaging and therapy due to their selectivity for specific targets [1]. While intact monoclonal antibody molecules are still most commonly used, they may not necessarily be the most efficient or desired molecular form depending on the application. Because of their relatively large size (approximately 150 kD), intact mAbs tend to have unfavorable imaging kinetics, relatively poor tumor penetration, and present with the potential for eliciting host antibody responses [2–7]. The solution to these myriad obstacles has been to reduce the size of intact antibodies to smaller forms or fragments, achieved either through enzymatic cleavage or by genetic engineering. The latter strategy requires a serious commitment of time and resources while enzymatic methods for generating monovalent or bivalent fragments of a mAb is somewhat facile with a lesser investment incurred.

The bivalent F(ab')₂ antibody fragment can be generated by cleaving the antibody on the carbonyl side of cysteinyl residues, below the disulfide bonds with pepsin [8]. This results in an Fc and an F(ab')₂ fragment [9]. The removal of the Fc portion during digestion also removes the potential of binding with Fc receptors thus reducing non-specific interactions [10]. The average molecular weight of the F(ab')₂ fragment is approximately 110 kD.

Radiolabeled mAbs are utilized in applications that include monitoring of tumor response to therapy, detection of metastatic lesions, dosimetric calculations, and therapy [10, 11]. Again, mAb fragments may be preferable for several reasons. The removal of the Fc segment could reduce the non-specific distribution in vivo of the mAb via the Fc receptors found on normal cells. F(ab')₂ fragments differ in their pharmacokinetic characteristics compared to intact antibodies resulting in distinct blood clearance and tumor localization patterns, clearing faster from the circulation than intact antibody while demonstrating better penetration into tumor sites [7, 12–19]. The rapid clearance from the blood compartment by F(ab')₂ results in a higher signal-to-noise ratio at earlier time points. A more favorable scenario for the imaging of patients is thus provided.

The smaller size and rapid clearance of antibody fragments such as F(ab')₂ should also lower their immunogenicity potential, reducing the risk of patients developing a humoral response against the antibody fragment, and potentially permitting repeated treatment of patients [20]. The ability to administer multiple doses of mAb for either therapy or imaging has not been a trivial consideration in the management of cancer patients.

Panitumumab (ABX-EGF, Vectibix™, Amgen, Thousand Oaks, CA, USA) is a fully human IgG₂ mAb that binds to the epidermal growth factor receptor (EGFR) with high affinity [21]. Panitumumab gained FDA-approval in 2006 for the treatment of patients with EGFR expressing metastatic colorectal carcinoma with disease progression while on or following fluoropyrimidine-, oxaliplatin-, or irinotecan-containing chemotherapy regimens [22]. Panitumumab has been well tolerated in clinical trials and as a result, close observation of patients has not been required nor has pre-medication with antihistamines [23]. The intact antibody has been shown to be successfully radiolabeled with ¹¹¹In in high yields and has demonstrated excellent tumor targeting with low normal tissue uptake [24, 25]. Panitumumab has also been successfully used for positron-emission tomography (PET) imaging using ⁸⁶Y [26, 27].

Extensive studies have been performed on the intact panitumumab; to date, there are no reports utilizing a fragment of panitumumab for either imaging or therapeutic applications. This paper represents the first in vitro and in vivo characterization of panitumumab F(ab')₂ fragment with an emphasis on its evaluation towards both imaging and therapeutic applications.

The studies as performed were designed to evaluate the in vitro and in vivo properties of the panitumumab-CHX-A" F(ab')₂ fragment and assess the potential of this molecule for imaging and therapeutic applications. To determine the optimal (digestion) cleavage time, 2% pepsin was added to 250 μg panitumumab and incubated at 37°C. Aliquots were removed at 1, 2, 4, 6, 8, and 18 h. As determined by SDS-PAGE, near complete pepsin digestion of panitumumab to a F(ab')₂ fragment appears to occur after 8 h, evident in Figure 1 by the loss of the higher-molecular-weight band of the intact IgG under non-reducing conditions (Figure 1a) and the transition of the heavy-chain band to a lower molecular weight when subjected to reduction with β-mercaptoethanol (Figure 1b).

Having determined the incubation time for the peptic digestion, 100 mg of panitumumab was digested overnight with 2% pepsin. When analyzed by SDS-PAGE, as expected, major bands were visualized corresponding to a molecular weight (M_r) of 79.4 under non-reducing conditions while two bands were evident at 25.1 and 22.4 kD after reduction (Figure 2a). Lower-molecular-weight (LMW) species at approximately 38 and 22 kD were also evident under non-reducing conditions. These LMW species, with a retention time of 24.2 min on SE-HPLC, comprised 30.5% of the reaction mixture, most likely representing pepsin and the Fc fragment (data not shown). The retention time of the panitumumab F(ab')₂ was 36 min by SE-HPLC, consistent with a M_r of 89.1 kD using tandem TSK2000 and 4000 columns for the analysis. Following buffer exchange to phosphate-buffered saline and subsequent concentration of the F(ab')₂ preparation using an Amicon Centriprep with a MWCO of 50 kD, the final product was again analyzed by SDS-PAGE and SE-HPLC: the LMW was no longer present as shown in Figure 2b. A final yield of 37.3 mg of panitumumab F(ab')₂ was obtained as determined by protein quantitation by the Lowry method.

The peptic digest appears to have a modest effect on the immunoreactivity of the panitumumab F(ab')₂ fragment. When analyzed in a competition radioimmunoassay, depicted in Figure 3, the concentration for 50% inhibition (IC₅₀) for intact panitumumab IgG was 0.5 nM while the IC₅₀ for the panitumumab F(ab')₂ was 1 nM.

Evaluating the potential of the panitumumab F(ab')₂ for clinical imaging and radioimmunotherapy applications would require larger quantities of the F(ab')₂. Therefore, a peptic digestion was performed with 1 g of panitumumab. As with the previous preparation, LMW species were detected by SE-HPLC which comprised approximately 34% of the digest mixture. For this larger preparation, a tangential flow filtration system with a 50 kD MWCO was used to eliminate the LMW species, exchange the buffer, and to also concentrate the F(ab')₂. The final product, analyzed by SDS-PAGE and SE-HPLC, was found to be comprised of a single product consistent with the Mr of a F(ab')₂ (data not shown). The final yield of this preparation was appreciably higher than the digestion of 100 mg with a final yield of 56%.

Each of the immunoconjugate preparations were radiolabeled with ¹¹¹In and their characteristics compared. The specific activities ranged from 1.9 to 14.8 μCi/μg, with the labeling efficiency ranging from 13.9% to 49.4% (Table 1). When the radiolabeled panitumumab F(ab')₂ fragments were incubated with purified hEGFR for 4 h at 37°C, 54.7% to 59.5% of the radioactivity was bound. The addition of 10 μg of unlabeled fragment, which reduced the percentage of bound radioactivity to an average of 3.5%, demonstrated the specificity of the RICS. Specificity of the radioimmunoassay was also demonstrated with the lack of binding (1.8%) with ¹¹¹In- HuM195 F(ab')₂. Based on these data, the preparation with 1.7 chelates, from the 10:1 molar excess reaction, was chosen for the remaining studies. In subsequent assays, specific binding of the RIC with hEGFR was as high as 72%.

Tumor targeting of the panitumumab F(ab')₂ was also validated through the use of two imaging modalities, planar γ-scintigraphy, and positron-emission tomography (PET). The s.c. LS-174T tumors on the rear flank of the mice were clearly visualized by planar γ-scintigraphy (Figure 4a). Over the 4-day period that images were collected, not only does the ¹¹¹In-labeled panitumumab remain in the tumor, but the RIC clears from the body.

Imaging was also performed with ⁸⁶Y-CHX-A"-DTPA-panitumumab F(ab')₂ on the ATLAS. Mice bearing the LS-174T xenografts were injected i.v. with approximately 100 μCi (3 μg) of ⁸⁶Y- panitumumab F(ab')₂. Images were taken at 24 and 48 h; after the 48 h images were collected, the mice were euthanized and the tumor, blood, and normal organs were harvested to obtain direct counts to correlate with the quantitation by imaging. All of the tumors were clearly visualized for both days following injection of the RIC as shown in the maximum intensity image (Figure 4b). The blood pool (heart, lungs, liver) is visible in these images, but it appears to have decreased on the second day while the tumor uptake increased. Specificity is demonstrated by reduction in tumor uptake when 0.1 mg of unlabeled panitumumab was co-injected with the ⁸⁶Y-labeled panitumumab F(ab')₂ (Figure 4c).

Direct quantitation of the distribution of ⁸⁶Y-panitumumab F(ab')₂ fragment in the liver and tumor provided results similar to what was obtained with the ¹¹¹In-labeled fragment (Figure 5a). The percentage of injected dose per cubic centimeter calculations from the images correlated (Table 4) with the ex vivo percentage of injected dose per gram quantitation (r² = 0.91, p = 0.89). Finally, when the specific activity of the ⁸⁶Y-panitumumab F(ab')₂ was lowered by the addition of 15 μg of panitumumab F(ab')₂ (Figure 5b); no mass effect was observed in the level of radioactivity in the blood, tumor, or normal organs as determined by imaging and ex vivo quantitation (Figure 5c and Table 4).

As a prelude to: (1) utilizing the panitumumab F(ab')₂ for monitoring response to radioimmunotherapy and (2) exploiting the panitumumab F(ab')₂ as a targeting vector for radioimmunotherapy of disseminated peritoneal disease, direct quantitation of intraperitoneal (i.p.) tumor xenograft targeting was performed. The targeting of i.p. tumors by ¹¹¹In-panitumumab F(ab')₂ was evaluated using both an i.p. and i.v. injection route, the results of which are presented in Figure 6. Not unexpectedly, i.p. administration of the RIC resulted in excellent targeting of the i.p. tumors (Figure 6a). Peaking at 48 h, the tumor percentage of injected dose per gram was 45.67 ± 3.79 and declined to 8.50 ± 3.63 at 168 h. When mice bearing i.p. tumor xenografts were given an i.v. injection of the RIC, the peak tumor percentage of injected dose per gram was at a similar level to the aforementioned experiment; however, this maximum did not occur until 72 h (Figure 6b). The pattern of normal organ distribution in these last two studies was similar to what was obtained with the i.v. injected ¹¹¹In-labeled panitumumab F(ab')₂ already discussed.

The in vivo and in vitro properties of the intact panitumumab mAb has been previously described by Ray et al. [25] which included imaging by planar γ-scintigraphy. The potential of imaging EGFR-positive tumors for purposes of monitoring disease response to therapy and performing dosimetric calculations was successfully extended to PET imaging with ⁸⁶Y-panitumumab [26, 36]. While the ¹¹¹In-CHX-A"-panitumumab demonstrated tumor uptake in LS-174T xenografts [25], maximal localization of the intact mAb does not occur until 48-96 h post-injection [13]. The objective of this study was to evaluate the in vivo and in vitro properties of panitumumab F(ab')₂ fragment and to assess its utility as a targeting agent in radioimmunodiagnostic and radioimmuntherapeutic protocols. To date, this report appears to be the first such characterization of a fragment generated from panitumumab.

Although intact monoclonal antibodies have been considered candidates for targeted therapy due to their specificity, with their long residence time in the blood of days to weeks, they are not ideal carriers for imaging probes; it is extremely difficult to perform serial imaging of less than several days [37, 38]. Early studies demonstrated that the size of antibody-based imaging agents are inversely related to their blood clearance; the clearance rate of Fab or Fab' > F(ab')₂ > IgG [13, 15]. Covell et al. [15] found that the whole murine IgG was retained in the (mouse) body 17 times longer than F(ab')₂. Consequently, normal tissue exposure is much greater with the intact antibody than it is for the fragment.

The Fab fragments, smaller in size, clear even faster than an F(ab')₂ fragment and have been developed predominantly as imaging agents [13, 15, 39]. Their limitations pertain to the fact that they are monovalent and their molecular weight subjects them to efficient glomerular filtration. Monovalency often results in the loss of functional affinity and reduces the binding strength of the Fab or Fab' fragment as compared to the F(ab')₂ or IgG [39, 40]. In 1983, Wahl and colleagues reported on a direct comparison of three radio-iodinated mAb forms, anti-CEA IgG, F(ab')₂ and Fab using γ-scintigraphy. Interestingly, the F(ab')₂ exhibited the fastest tumor localization [14]. At 2 days post-injection of the ¹³¹I-anti-CEA-F(ab')₂ fragment tumor was clearly visualized. By the third day the images were equivalent to those obtained at 11 days with the intact mAb. The authors concluded that Fab fragments were not an optimal vector for imaging due to their rapid clearance, low accumulation in tumor and high renal accumulation. Similar observations have been noted with other mAbs. As reviewed by Tolmachev [40], ¹¹¹In-DTPA-trastuzumab-Fab tumor uptake as compared to that of ¹¹¹In-DTPA-trastuzumab-F(ab')₂ is considerably lower. In contrast, an early clinical imaging study conducted by Delaloye et al. [19] compared ¹²³I-labeled Fab and F(ab')₂ fragments of an anti-carcinomembryonic antigen mAb in colorectal carcinoma patients for the detection of disease using emission-computed tomography. The Fab fragment was reported to have clearer images than those of the F(ab')₂ fragment with a higher overall detection of tumor lesions. The authors postulate that the success of these studies was the result of careful selection and matching of the target, the targeting vehicle, and the radionuclide.

Based on these earlier reports and the importance of retaining affinity/avidity, the F(ab')₂ fragment was selected for this investigation as the targeting vehicle to be exploited in imaging modalities, e.g., PET, planar γ-scintigraphy, MRI, and optical. In this study, F(ab')₂ fragments were successfully generated from panitumumab by peptic digest and the protocol developed was readily scaled-up. The in vitro analysis, SDS-PAGE and SE-HPLC, indicated that the final product has a Mr of 79.4 and 89.1 kD, respectively, and retained immunoreactivity for HER1. Once the F(ab')₂ fragment was generated, conjugation with the bifunctional chelate, acyclic CHX-A"-DTPA was performed for radiolabeling with medically relevant radionuclides such as ¹¹¹In and ⁸⁶Y which is also appropriate for radiolabeling with therapeutic radionuclides such as ¹⁷⁷Lu, ²¹³Bi, and ²¹²Bi [41–43]. The conjugation was performed at three different molar ratios of chelate to panitumumab F(ab')₂. The different ratios had minimal effect on the immunoreactivity of the panitumumab F(ab')₂ as demonstrated by a competition radioimmunoassay. Furthermore, the specific binding of the immunoconjugate preparations with hEGFR was comparable following radiolabeling with ¹¹¹In. Based on the results, the 10:1 molar excess ratio (1.7 chelate/antibody) was chosen for the remaining studies.

This represents the first report on the use of a panitumumab fragment for imaging applications. As such, comparisons to other mAb fragments will have to suffice in the discussion of the data reported herein. Smith-Jones (2004) and co-workers conjugated the macrocyclic ligand DOTA to trastuzumab F(ab')₂. reporting an average of 6.3 chelates per F(ab')₂ fragment and immunoreactivity of 81% [37]. No other in vitro assay data was reported, i.e., assessing the effect of the fragmentation of the mAb, or the chelate number, on the overall immunoreactivity of the mAb. A number of studies have investigated the use of F(ab')₂ fragments with radio-iodines for imaging and radioimmunotherapy, making a direct comparison with a metallic radionuclide difficult [12, 18, 19, 44, 45]. For example, an F(ab')₂ of the mAb14C5 was evaluated for the radioimmunodetection of non-small cell lung tumor xenografts [45]. Some targeting was observed; however, tumors were poorly visualized by γ-scintigraphy. More promising results were reported for the ¹²⁵I-labeled F(ab')₂ fragment of mAb B6.2 [12]. In this case, excellent tumor targeting was demonstrated by direct quantitation studies as well as γ-scintigraphy.

In contrast to panitumumab IgG, the F(ab')₂ attains a higher percentage of injected dose per gram in the LS-174T s.c. tumor (21.42 ± 7.67) at an earlier time point (24 h) than the intact IgG (13.27 ± 8.40) following i.v. administration [25]. The F(ab')₂ fragment may not achieve the same level of targeting, it is, however, retained in the tumor. In contrast to the intact IgG, the F(ab')₂ is rapidly cleared from the blood compartment, evident by the biodistribution and pharmacokinetic data presented. At 24 h, the blood percentage of injected dose per gram of panitumumab F(ab')₂ is approximately one-half of that reported for panitumumab IgG. The differential becomes even greater by 7 days. The blood pharmacokinetics of the panitumumab F(ab')₂ in tumor-bearing mice reported here is consistent with that of other F(ab')₂ fragments described in the literature [13]. Interestingly, when compared to panitumumab IgG, there is a "reversal" of the values for the T_1/2α and β values of the tumor and non-tumor bearing mice. In the absence of a tumor burden, the T_1/2α and β for IgG was approximately 5.2 and 3.8 times faster, respectively, than in the presence of tumor. In contrast, the T_1/2α and β of the blood clearance for the ¹¹¹In-panitumumab F(ab')₂ was approximately two times faster in the tumor-bearing mice vs. in mice without a tumor burden. One explanation for this phenomenon is that murine FcRn cross-reacts with and will bind human IgG which results in prolonging its half-life in the mouse [38, 46]. The F(ab')₂, lacking the Fc region, is likely cleared by phagocytic cells of the reticuloendothelial system.

The results reported herein are comparable to studies performed with F(ab')₂ of other mAbs. In a study with ¹¹¹In-DOTA-Herceptin F(ab')₂, the tumor percentage of injected dose per gram was 20.4 ± 6.8 after 24 h. At the same time point, a tumor percentage of injected dose per gram of 21.42 ± 7.67 was obtained with ¹¹¹In-CHX-A"-DTPA-panitumumab F(ab')₂. Differences between the two studies are evident in renal uptake of the RICs. The kidney percentage of injected dose per gram of uptake ¹¹¹In-CHX-A"-DTPA-panitumumab F(ab')₂ was only 13.13% at 24 h compared to approximately 65% with ¹¹¹In-DOTA -Herceptin F(ab')₂ [37]. A F(ab')₂ fragment of the recombinant anti-L1 CAM mAb, chCE7, has also been evaluated for PET imaging when radiolabeled with ⁶⁴Cu and for radioimmunotherapy using ¹⁷⁷Lu [47]. Maximal tumor uptake of the ¹⁷⁷Lu-labeled RIC was observed at 24 h with an ID/g of 14.43 while maximum uptake for the ^67/64Cu-labeled RIC occurred at 8 h. Unfortunately, kidney uptake with either of these radiolabeled chCE7 F(ab')₂ preparations was greater than what was observed in the tumor at all time points of the study.

The panitumumab F(ab')₂ also appears to be a potential vehicle for targeting disseminated intraperitoneal disease. Whether administered i.v. or i.p., excellent tumor targeting of the i.p. tumor xenografts was observed, with a tumor percentage of injected dose per gram as high as 45.66, 2.1-fold greater than the highest value calculated for the s.c. tumor following i.v. administration. Interestingly, the peak tumor percentage of injected dose per gram of both the s.c. and i.p. tumors following i.v. injection occurred at 72 h whereas the peak i.p. tumor value occurred at 48 h following i.p. injection of the panitumumab F(ab')₂. Imaging studies, γ-scinitigraphy and PET, of the i.p. tumor model using these two injection routes are pending.

Studies from this laboratory and others have demonstrated the potential of panitumumab for the non-invasive monitoring of HER1-positive tumors and quantitating HER1 expression using modalities such as γ-scintigraphy, PET, and optical imaging [24, 25, 36, 48, 49]. By evidence of the tumor targeting, direct quantitation of tumor targeting suggests that the ¹¹¹In-labeled panitumumab F(ab')₂ is stable in vivo. This observation is also supported by the imaging studies related herein. This attribute may be due in part to the fact that panitumumab is of the IgG₂ immunoglobulin subclass. In a series of dual-labeled experiments, Buchegger et al. [44] compared F(ab')₂ fragments generated from chimeric mAb of three subclasses (IgG₁, IgG₂, and IgG₄) directed against carcinoembryonic antigen. The in vivo studies revealed that the IgG₂ F(ab')₂ fragment had superior tumor targeting and possessed the longest biological half-life as well. The poorest results were obtained with the F(ab')₂ of the IgG₄. The radio-iodinated F(ab')₂ fragment of chimeric 81C6, another IgG₂ subclass mAb, has also been found stable and suitable for imaging applications [18].

The images presented here, both planar γ-scintigraphy and PET, indicate that there is excellent uptake of the panitumumab F(ab')₂ by 24 h with decreasing background over the course of the imaging sessions, consistent with the studies conducted by Wahl and colleagues [14]. The ex vivo quantitation of tumor and tissue uptake (percentage of injected dose per gram) correlated with the quantitation (percentage of injected dose per cubic centimeter) performed from the PET images. These data were comparable to those obtained with panitumumab IgG, providing confidence in the suggestion that the panitumumab F(ab')₂ would be useful in assessing tumor responses to therapy (i.e., estimating HER1 expression) and would provide accurate data for performing dosimetric calculations for radioimmunotherapy [36]. The caveat to translating the imaging of HER1 with panitumumab, either the intact IgG or a fragment, is whether or not human hepatocytes will be targeted. Clinical studies using ¹¹¹In- or ^99mTc-labeled cetuximab reported visualization of tumor burden in patients with squamous-cell carcinoma of the lung, head, and neck with significant hepatic uptake of radioactivity [50, 51]. Panitumumab may prove superior in this aspect. In pre-clinical studies, cetuximab has been observed to have a higher hepatic uptake of radioactivity than panitumumab [25, 29].

Alternative to enzymatically generated fragments of monoclonal antibodies are a variety of genetically engineered forms (i.e., sfv, minibody, diabody, domain-deleted) that are under evaluation for imaging and therapeutic applications. The valency and molecular weight of these mAb forms have been tailored to alter such biological properties as their blood residence time, tumor penetration, tumor residence time, renal clearance, and catabolic susceptibility [11, 39, 52–54]. Renal and hepatic uptake of many of these forms remains an obstacle when labeled with a metallic radionuclide limiting their potential for imaging to the radio-iodines or ¹⁸F.

This study demonstrates that F(ab')₂ fragments of the anti-EGFR mAb panitumumab can be generated efficiently and quickly. The immunoreactivity of the panitumumab F(ab')₂ fragment is comparable to the IgG. The immunoreactivity is retained on further modification with the chelating agent, acyclic CHX-A"-DTPA and subsequent radiolabeling. The study also demonstrated tumor targeting with LS-174T xenografts. Because of the bivalency, fast blood clearance, and deep tumor penetration, the panitumumab F(ab')₂ fragment is a good candidate for imaging. Studies are continuing with the panitumumab F(ab')₂ to further evaluate the role of panitumumab F(ab')₂ in radiological imaging, SPECT and PET, and also for its potential role in targeted MRI.