Imaging the Met Receptor Tyrosine Kinase (Met) and Assessing Tumor Responses to a Met Tyrosine Kinase Inhibitor in Human Xenograft Mouse Models with a [99mTc] (AH-113018) or CY 5** (AH-112543) Labeled Peptide

Developing an imaging agent targeting the hepatocyte growth factor receptor protein (Met) status of cancerous lesions would aid in the diagnosis and monitoring of Met-targeted tyrosine kinase inhibitors (TKIS). A peptide targeting Met labeled with [99mTc] had high affinity in vitro (Kd = 3.3 nM) and detected relative changes in Met in human cancer cell lines. In vivo [99mTc]-Met peptide (AH-113018) was retained in Met-expressing tumors, and high-expressing Met tumors (MKN-45) were easily visualized and quantitated using singlephoton emission computed tomography or optical imaging. In further studies, MKN-45 mouse xenografts treated with PHA 665752 (Met TKI) or vehicle were monitored weekly for tumor responses by [99mTc]-Met peptide imaging and measurement of tumor volumes. Tumor uptake of [“mTc]-Met peptide was significantly decreased as early as 1 week after PHA 665752 treatment, corresponding to decreases in tumor volumes. These results were comparable to Cy5**-Met peptide (AH-112543) fluorescence imaging using the same treatment model. [99mTc] or Cy5**-Met peptide tumor uptake was further validated by histologic (necrosis, apoptosis) and immunoassay (total Met, p Met, and plasma shed Met) assessments in imaged and nonimaged cohorts. These data suggest that [99mTc] or Cy5**-Met peptide imaging may have clinical diagnostic, prognostic, and therapeutic monitoring applications.

pancreatic, esophageal, cervical, and, prostate cancer; astrocytoma; and adult T-cell leukemia. 1,2 The endogenous ligand for Met, hepatocyte growth factor (HGF) or scatter factor (SF), binds to Met and heparin sulfate proteoglycans on the cell surface, resulting in receptor dimerization, phosphorylation, and activation of the Met signaling pathway. 1,3,4 Met signaling is required for normal early development (embryogenesis) and homeostasis in adulthood and promotes tissue repair and regeneration. [4][5][6][7] The over-expression or amplification of Met or mutations in HGF in cancer causes Met signaling to become dysregulated, resulting in increased proliferation and angiogenesis, inhibition of apoptosis, and progression to metastatic disease. 1,8,9 Patients whose tumors overexpress Met or HGF have a poorer prognosis and are more likely to exhibit drug resistance. 10,11 For instance, in non-small cell lung cancer resistant to erlotinib and gefitinib, amplification of Met was observed, suggesting a role for Met in the development of drug resistance. Since activation of Met and its signaling pathways appears to be a survival pathway implemented by tumor cells in response to treatment, the ability to determine Met expression without resorting to tissue biopsy could be useful in determining when the addition of a Met-targeted therapeutic might be beneficial for the patient.
As a consequence of its important role in cancer growth, numerous Met-directed therapeutics have been developed. Some of these inhibit activation of the Met signaling pathway by antagonizing HGF binding to the Met extracellular domain or adenosine triphosphate (ATP) binding to the Met intracellular kinase domain. 9,[12][13][14][15][16][17] As these Metdirected therapeutics move forward to the clinic, the development of imaging agents specifically targeting Met would be useful for diagnosis, drug selection, and monitoring patient responses to specific molecularly targeted therapies. Furthermore, in vivo imaging of Met in human studies could assist drug development by monitoring real-time changes in Met status with Met-targeted therapeutics.
Although several potential Met imaging agents using monoclonal antibody (mAb), mAb fragments, or peptides radiolabeled with positron emission tomography (PET) or singlephoton emission computed tomography (SPECT) radionuclides have been evaluated in preclinical mouse models, clinical translation of these agents has yet to be reported. [18][19][20][21][22] In this work, we describe the preclinical evaluation of imaging agents targeting Met with the end goal toward development of Met imaging in a clinical setting. To be appropriate for clinical translation, such an agent will require not only a suitable ligand with high affinity for Met but also reliable synthetic methods that could easily be adapted for widespread clinical use. A small-molecular-weight cyclic peptide (AH-111972, GE Healthcare Life Sciences, Oslo, Norway; DX-1662, Dyax Corp., Cambridge, MA; 26 amino acids) has been identified through vector searching that binds to the extracellular domain of human Met with low nanomolar affinity determined by surface plasmon resonance technology (Biacore; GE Healthcare Life Sciences, Oslo, Norway) and fluorescence polarization techniques. 23 This peptide has been labeled with 99m Tc ([ 99m Tc]-Met peptide [AH-113018]) or a fluorescent polymethine cyanine dye (Cy5 ⋆⋆ -Met peptide [AH-112543]; similar fluorescence to Cy5, 653/675 nm) at the C-terminal end (lysine) for SPECT imaging or fluorescent surgical and endoscopic imaging, respectively ( Figure 1). 24 Using fluorescence polarization techniques (developed by GE Healthcare Life Sciences), the [ 99m Tc]-Met peptide or the Cy5 ⋆⋆ -Met peptide demonstrated half-maximal inhibitory concentrations (IC 50 ) of ~ 200 nM, which was comparable to the unlabeled Met peptide (90 nM). 23,25 Both of these labeled peptides have been formulated into a kit, which can easily be prepared for clinical application.
In this article, we further assess the ability of [ 99m Tc]-Met peptide and Cy5 ⋆⋆ -Met peptide to image Met status noninvasively and demonstrate that Met expression can be modulated by Met-targeted therapeutics in preclinical mouse xenograft models. 26,27 Preliminary in vitro binding studies with the [ 99m Tc]-Met peptide were performed to determine the biological activity (% immunoreactivity), K d, and Met concentrations (B max ) in MKN-45 (human gastric carcinoma, high Met expressor), SNU-16 (human gastric carcinoma, moderate Met expressor), and U87-MG (human glioblastoma, low Met expressor) cells. 28,29 These same cell lines were used in mouse xenograft models to determine the biodistribution and imaging characteristics of these agents. For the Cy5 ⋆⋆ -Met peptide, preliminary studies included only imaging in MKN-45 xenograft mice to determine the optimal dose and time to image after dosing. For the treatment studies, MKN-45 xenograft mice were treated with PHA 665752, a known Met tyrosine kinase inhibitor (TKI), and imaged with either Cy5 ⋆⋆ or [ 99m Tc]-Met peptide. [30][31][32] The PHA 665752 treatment paradigm was further validated by examining changes in tumor volume, necrosis, and apoptosis between treated and untreated mice, and this was correlated to imaging results. Additionally, plasma and tumor samples were collected to evaluate shed Met and tumor Met abundance, respectively. 28,33 As shed Met may interfere with the binding of the labeled Met peptide, we investigated the relationship between shed Met, tumor Met abundance, and tumor mass. These studies demonstrate the potential utility of Cy5 ⋆⋆ -and [ 99m Tc]-Met peptide to image Met and determine the efficacy of a targeted Met TKI therapeutic.

Cell Lines and Reagents
MKN-45, a human gastric carcinoma cell line (high Met expressor), was grown in RPMI-1640 (with 2 mM L-glutamine). SNU-16, a human gastric carcinoma cell line (moderate Met expressor), was grown in RPMI-1640 with 2 mM L-glutamine, 10 mM HEPES, 1 mM sodium pyruvate, 4,500 mg/L glucose, and 1,500 mg/L sodium bicarbonate. U87-MG, a human glioblastoma cell line (low Met receptor expressor), was grown in Dulbecco's Modified Eagle's Medium with 2 mM L-glutamine, 0.1 mM nonessential amino acids (Thermo Fisher Scientific, Grand Island, NY), and 1 mM sodium pyruvate. All media were supplemented with 10% fetal bovine serum and penicillin/streptomycin/amphotericin B. All cell lines were grown at 37°C in a 5% CO 2 humidified atmosphere.
Both the unlabeled Met peptide (AH-111972 trifluoracetate salt [DX-1662], 26 amino acids, Cys4-16, 6-14; Ac-Ala-Gly-Ser-Cys-Tyr-Cys-Ser-Gly-Pro-Pro-Arg-Phe-Glu-Cys-Trp-Cys-Tyr-Glu-Thr-Glu-Gly-Thr-Gly-Gly-Gly-Lys) and Cy5 ⋆⋆ -Met peptide (AH-112543; a kit formulation containing the lyophilized Met-peptide [AH-111972] conjugated to Cy5 ⋆⋆ [tetra-alkyl-sulfonated active ester cyanine dye, peak excitation/emission: 653/675 nm) were provided by GE Healthcare Life Sciences. [5-[[(2,6-dichlorophenyl)  peptide; the B/F at each concentration of cold antibody was determined. Using the same B/F (y) from the self-displacement (saturation) and the cold displacement curve, the associated radioactivity (x, cpm) and the amount of unlabeled Met peptide required to displace the radiolabel (x, ng), respectively, were determined. The corresponding values (cpm versus ng) were then plotted and fitted with the line of best fit, in which the reciprocal of the slope represents the actual "biological" (immunoreactive) specific activity. The ratio of the immunoreactive specific activity to the theoretical specific activity (assumes total mass of antibody added during radiolabeling was recovered) represents the immunoreactive fraction.

Mouse Tumor Models and Drug Treatment
Athymic female nude mice (Ncr-nu/nu, National Cancer Institute, Frederick, MD) were injected subcutaneously on the right flank or shoulder with MKN-45, SNU-16, or U87-MG cells ( Additional "naïve" imaging cohorts (not injected with the labeled Met peptide or imaged) from vehicle (n = 3) or treated (n = 3) groups were euthanized after 1, 2, 3, or 4 weeks of treatment from which blood and tumors were removed. The blood samples were collected in ethylenediaminetetraacetic acid (EDTA)-coated tubes, and after centrifugation, the separated plasma samples were fast frozen and stored at −70°C until assayed for shed Met. Tumors were weighed and then apportioned into separate samples as follows: (1) Tc]-Met peptide (10 to 20 μCi per mouse, 0.02-0.04 nmol) and euthanized (CO 2 inhalation) at the appropriate time. The blood and various tissues were excised from each animal, weighed, and assayed for radioactive content by gamma counting (Wizard 1480, PerkinElmer, Shelton, CT). The total activity in the blood and each tissue was expressed as the % injected dose per gram normalized to a 20 g mouse [%ID/g = (cpm tissue × body weight)/(cpm total injected ) × 20) 100]. An additional blood sample was also taken for determination of shed Met, which was processed as described above and assayed after the radioactivity had decayed. Immediately following counting, whole tumors and/or tumor samples were either fast frozen or fixed and assayed as described above.
For the blocking studies with unlabeled Met peptide, MKN-45 tumor-bearing mice were divided into two groups, with one group receiving [ 99m Tc]-Met peptide (~ 10 μCi per mouse, 0.02 nmol associated mass) only and the other group receiving a coinjection of the same amount of [ 99 Tc]-Met peptide + 24 nmol of unlabeled Met peptide. The mice were euthanized after 2 hours, and the processing of the blood and tissues was performed as described above. Statistical analysis of the differences between the two groups was done using the Welch t-test, with p < .05 as significant (GraphPad InStat version 3.00 for Windows).

MicroSPECT Imaging Studies
Tumor-bearing mice were anesthetized using isoflurane/O 2 (1.5-3% v/v) and imaged at various times after intravenous injection (tail vein) of [ 99m  . After the final imaging session, the mice were euthanized, whereupon blood and tumors were collected and analyzed as described above after the radioactivity had decayed.
Using the InVivoScope software (Bioscan, Poway, CA), the images were reconstructed by an ordered subset expectation maximization iterative multiplexing multipinhole reconstruction algorithm from which ROI were drawn and standardized uptake values (SUVs) were determined. The SUVs (μCi/cc) normalized to a 20 g mouse were calculated as follows: (SUV × body weight)/(injected dose × 20) 100. Statistical analysis of the differences between the two groups was done using the Welch t-test, with p < .05 as significant (GraphPad InStat version 3.00 for Windows).

Cy5 ⋆⋆ -Met Peptide (AH-112543) Spectral Fluorescence Imaging Studies
In preliminary imaging studies, MKN-45 xenograft mice were injected intravenously with 0.5, 1.0, or 2 nmol of Cy5 ⋆⋆ -Met peptide and imaged at various times. For the treatment studies, MKN-45 tumor-bearing mice were injected with 1 nmol of Cy5 ⋆⋆ -Met peptide and imaged after 60 minutes once a week for 4 to 5 weeks following a baseline study. Images were acquired using the Maestro In Vivo Imaging System (PerkinElmer, Waltham, MA) using a red light filter for excitation (615-665 nm) and a true yellow filter (630-680 nm) for emission, and then the signal was "unmixed" with a Cy5 ⋆⋆ spectral library using the Maestro software and corrected for autofluorescence. After the final imaging session, the mice were euthanized and blood and the tumors were collected and analyzed as described above.
ROI were drawn over the tumor and a nontarget tissue (lower left hind side away from the tumor) from which total signal intensities (scaled counts/s) were determined. The tumor total signal intensity was corrected for background by subtracting the nontarget contribution. Statistical analysis of the differences between the two groups was determined with the Student t-test (p < .05).

Determination of Plasma Shed Met and Total Tumor Met
Met protein content and plasma soluble Met ectodomain (shed Met) levels were measured using a two-site electrochemiluminescent immunoassay developed for use with a SectorImager 2400 plate reader.5 (Meso Scale Discovery [MSD], Rockville, MD). Detection antibodies were tagged with a ruthenium chelate (MSD Sulfo-Tag), which, in the presence of appropriate redox reagents, emits light when current is applied to the multiwell plate; this light is measured by the SectorImager's charge-coupled device (CCD) camera. Purified recombinant Met ectodomain-IgG-fc fusion protein was used as a reference standard for quantitation of Met mass per mass total extracted cell protein. Tumor tissue extracts were prepared by physically disrupting tissue samples before clearing by centrifugation and analysis of Met content as described above.
Correlations of shed Met, total tumor Met, tumor [ 99m Tc], or Cy5 ⋆⋆ Met peptide content to tumor mass or total tumor mass were performed using the Spearman rank correlation coefficient using Graphpad InStat version 3.0 for Windows or Graphpad Prism version 3.02 for Windows).

In Vitro Cell Binding Studies with 99m Tc-Met Peptide
From in vitro saturation studies, [ 99m Tc]-Met peptide exhibited high specific binding for Met ranging from 79 to 99% and high affinity with a K d of 3.31 nM ± 0.58 (n = 5). The percent immunoreactivity (immunoreactive fraction) from these batches was 84.9% ± 4.0 (n = 5), indicating that most of the biological activity was preserved after radiosynthesis. The Met receptor concentrations for MKN-45 cells (B max ) were high, with 7.72 × 10 5 ± 1.59 × 10 5 (n = 5) receptors per cell. In similar saturation studies with the low Met-expressing cell line, U87-MG cells, specific binding was substantially lower (< 40%), with a B max of 0. 085 × 10 5 ± 0.014 × 10 5 (n = 4), which was ~ 90-fold lower than the Met concentrations found for MKN-45 cells. 28,29 These data compare favorably with the expected levels of Met expression for these human cancer cell lines, indicating that this tracer would be appropriate for imaging tumors with high Met expression (> 600,000 receptors per cell), but at low Met expression levels (< 9,000 receptors per cell), nonspecific interactions would hamper the detection of specific Met binding.

In Vivo Biodistribution Studies with [ 99m Tc]-Met Peptide
The biodistribution of [ 99m Tc]-Met peptide was determined in MKN-45 xenografts after 30, 60, 120, and 180 minutes ( Figure 2 and Table 1). Uptake (%ID/g) of the [ 99m Tc]-Met peptide was rapid with the highest uptake observed in the tumor and kidney for all time points. The most rapid clearance (> 70%) of the [ 99m Tc]-Met peptide was from the blood, heart, lungs, spleen, and muscle from 30 to 180 minutes, whereas slower clearances were observed in the liver, kidney, and tumor, with decreases of 49%, 63%, and 52%, respectively, over the same time course. These pharmacokinetics and the high radioactive content in the kidney would suggest renal clearance as the primary route of excretion for [ 99m Figure 2), with the highest uptake always observed in the kidney.
By examining the ratios of the tissue uptakes (%ID/g) to muscle uptakes, a further estimation of imaging feasibility can be assessed. High tumor to muscle ratios of 9.8 ± 2.1, 30.1 ± 3.3, 49.6 ± 11.3, and 55.2 ± 13.4 were observed at 30, 60, 120, and 180 minutes, respectively, in the MKN-45 xenografts, whereas in comparison, the tumor to muscle ratios of SNU-16 and U87-MG xenografts were reduced by 2-to 4-fold and 10-to 16-fold, respectively, indicative of reduced Met densities, as expected for these cell lines ( Figure 3). The continual increase in the tumor to muscle ratios over time is reflective of the increased rate of [ 99m Tc]-Met peptide clearance from the muscle compared to the tumors rather than an increase in tumor uptake. The highest tissue to muscle ratios were observed in the kidneys, where ratios were 30 ± 4.5, 54 ± 12.8, 101 ± 13.8, and 89 ± 8.1 at 30, 60, 120, and 180 minutes, respectively, which exceeded the tumor to muscle ratios over the same time course by 1.6-to 3-fold. These high ratios in the kidneys suggest that it will be difficult to image tumors in the kidneys and urinary tract.
Blocking studies were performed in mice with MKN-45 xenografts in which one group was injected with [ 99m Tc]-Met peptide only and another group was coinjected with the same amount of [ 99m Tc]-Met peptide and 24 nmol of unlabeled Met peptide (~ 1,200 times the mass associated with the radiolabel). After 120 minutes, the uptake (%ID/g) was determined in the blood and other tissues and found comparable for both groups in all nontarget tissues except the spleen, which was significantly increased ~ 170% (p < .02), and the tumor, which was significantly decreased 62% (p = .0067) in the blocked group compared to the [ 99m Tc]-Met peptide-only group. The significant decrease in tumor uptake of the blocked group strongly suggests that the [ 99m Tc]-Met peptide is competing for the same site as the unlabeled peptide. Similar results were obtained for tumor to muscle ratios in which the blocked group showed lower tumor to muscle ratios (53%, p = .0028) compared to the unblocked group, again providing further proof of specific Met binding in vivo (see Figure  3). It should be noted that the tumor to muscle ratios in these studies were ~ 1.8-fold lower than the tumor to muscle ratio obtained from the first biodistribution study because there was a considerable disparity in the size of the tumors between the studies as the tumors from the blocking studies were much larger (0.1 to 1.0 g versus < 0.3 g) compared to the first study.

MicroSPECT Imaging Studies
Initial In these studies, xenografts were imaged prior to treatment (baseline) and then once a week for 3 weeks after beginning treatment; one group was treated intraperitoneally daily with vehicle (saline, 5% DMSO) and another group was treated with PHA 665752 (15 mg/kg, saline, 5% DMSO). Representative coronal images of a single PHA 665752-treated and vehicle xenograft mouse pair obtained at baseline and each subsequent week are shown in Figure 4B, from which ROI were drawn for quantitation of the tumor uptake. The tumor uptake (%ID/g) of [ 99m Tc]-Met peptide in the treated group compared to the vehicle group was decreased 43% (p = .005), 42% (p = .026), and 25% at days 8, 14, and 21, respectively, with days 8 and 14 achieving statistical significance ( Figure  4C). The tumor volumes determined by 3D ultrasonography were found to be decreased in the treated group by 25%, 36% (p = .036), and 55% (p = .016) at days 8, 14, and 21, respectively, compared to the vehicle group ( Figure 4D). Although these tumor volumes decreased over the entire time course, significant reductions were only observed later at days 14 and 21, whereas [ 99m Tc]-Met peptide tumor uptake was significantly reduced by day 8.
After the last imaging session, tumors were excised and sectioned for determinations of necrosis and apoptosis by H&E staining and immunohistochemistry (IHC; ApopTag), respectively, in the vehicle-and PHA 665752-treated groups ( Figure 4E). The tumors from the vehicle group were larger tumors (typically > 0.5 g), and 78% were severely necrotic (+4). Similarly, the fraction of severely necrotic tumors from the PHA 665752-treated group was also 78%; however, only 44% of the tumors were > 0.5 g. Differences in the severity of apoptosis were less pronounced between the vehicle and treated groups, with only 22% of tumors from the treated group exhibiting moderately severe apoptosis compared to the vehicle tumors. Furthermore, although the total [ 99m Tc]-Met peptide uptakes per tumor (%ID/tumor) highly correlated to tumor weight, a 2-to 3-fold reduction was observed in the slopes (%ID/g) of these larger (> 0.5 g) control tumors (slope = 1.3) compared to smaller (< 0.2 g) tumors (slope = 4.2), as determined from linear regression analysis ( Figure 4F). The decreased uptake in these large tumors is consistent with highly necrotic tumors with limited central blood flow, hindering delivery of the [ 99m Tc]-Met peptide to the tumor tissue, which was visually observed in some of the SPECT images of larger tumors (> 0.5 g). In particular, this heterogeneous distribution was seen in the SPECT image of the vehicle xenograft at 3 weeks (see Figure 4B), in which only the outer surface of this large necrotic tumor (0.74 g, histologic grade = +4, severe) was visualized with intense [ 99m Tc]-Met peptide uptake, whereas the interior of the tumor was barely discernible, indicating a high degree of necrosis.

Correlation of [ 99m Tc]-Met Peptide Tumor Uptakes with Shed Met, Tumor Volumes, and Met Content in Vehicle-Treated and PHA 665752-Treated MKN-45 Xenografts
Plasma and tumor samples obtained from naïve and imaged MKN-45 xenografts treated with vehicle or PHA were analyzed for shed Met ectodomain, total tumor Met, and total tumor phosphorylated Met (pMet) content, respectively, and then correlated to tumor mass and/or the total tumor radioactive content (% injected dose [%ID] per tumor) ( Figure 5). In the naïve (nonimaged) vehicle xenografts, the correlation of the tumor masses with shed Met (see Figure 5A), total tumor Met content (see Figure 5B), and total tumor pMet (see Figure  5C) was extremely significant (p < .0001), but in the case of the PHA-treated xenografts (21 and 27 days), only the correlation of the tumor masses with total tumor Met and total tumor pMet was extremely significant (p < .0003), with the shed Met to tumor mass exhibiting a lower but still significant correlation of p < .0265 (see Figure 5A). The tumor uptake (%ID) of [ 99m Tc]-Met peptide of the imaged MKN-45 xenografts that had been treated for 21 days with either vehicle or PHA highly correlated with the tumor mass, total tumor Met, and total tumor pMet content (see Figure 5, D-F), indicating that [ 99m Tc]-Met peptide tumor uptakes reflect Met density. Furthermore, the slopes derived from these correlations in the imaged treated and vehicle-treated groups were similar, whereas the slopes of the nonimaged vehicle-treated cohorts were increased compared to the treated cohort (see Figure 5, B and C). Most likely, this lack of difference in the slopes in the imaged groups is due to lower slopes in the vehicle-treated group as a consequence of the considerable number of larger tumors. Shed Met in plasma samples correlated to tumor mass in vehicle-treated (not shown, r = .7857; p = .048, n = 7) and PHA-treated (r = .8571; p = .0238, n = 7) xenografts that had undergone imaging with slopes of 43.7 and 77.7 plasma shed Met (ng/mL)/tumor mass (g), respectively. These slopes were comparable to those obtained with nonimaged vehicle-and PHA-treated xenografts of 43.8 and 77.5 plasma shed Met (ng/mL)/tumor mass (g), respectively (see Figure 5A), indicating that the presence of [ 99m Tc]-Met peptide does not appear to alter shed Met plasma concentrations.

Comparisons of PHA 665752 Treatment Effects in Imaged and Nonimaged Mice
To further validate the effects of PHA 665752 treatment observed in the longitudinal imaging studies conducted with [ 99m Tc]-Met peptide in MKN-45 xenografts (see Figure 4), naïve (nonimaged) cohorts were euthanized at 1, 2, 3, or 4 weeks posttreatment to coincide with the imaging time points. In these euthanized groups, tumor growth was measured weekly by caliper, and then after euthanasia, tumors were excised, weighed, and sampled for assessment of necrosis (H&E staining), apoptosis (ApopTag), total Met content, and pMet; in addition, blood samples were taken for plasma shed Met determinations. PHA 665752 treatment significantly decreased tumor volumes and weights by 65 to 88% in treated groups at 7, 14, 21, and 28 days posttreatment compared to vehicle groups ( Figure 6A). These results compared favorably to the tumor volume decreases observed with the groups receiving [ 99m Tc]-Met peptide (see Figure 4D) except at the 7-day time point, which, although decreased (27%) with treatment, was not significant compared to the imaged vehicle group. This result might be expected considering that the majority of the tumors of the imaging vehicle group were larger necrotic tumors.
Histologic analysis of necrosis in PHA-treated and vehicle-treated tumors appeared similar, with the majority of vehicle (74%) and treated (82%) tumors exhibiting moderate to severe (3 + , 4 + ) necrosis over a wide range of tumor weights (0.1-3.0 g) ( Figure 6B). Although the degree of necrosis was quite variable in tumors < 1 g, tumors > 1 g most often exhibited necrosis, suggesting that tumor size may influence total tumor Met and total tumor pMet content, whereby levels may be decreased in larger tumors due to necrosis rather than a specific treatment effect. Similarly, the level of apoptosis in the treated tumors appeared comparable to that in the vehicle tumors, although the majority of the vehicle (74%) and treated (85%) tumors were minimally apoptotic (1+, 2+).

Levels of plasma shed Met (ng/mL per g [tumor]
) were significantly increased ~ 2-fold at 1, 2, and 3 to 4 weeks after treatment compared to the vehicle controls ( Figure 6C). This ~ 2fold increase in shed Met plasma concentrations in the treated groups was in accordance with the respective increased Met concentrations per tumor mass of 77.5 and 43.8 plasma shed Met (ng/mL) for treated and vehicle groups, respectively, derived from the slopes in Figure 4A.
Tumor Met (ng/g [tumor]) and tumor pMet (SI/g [tumor]) levels were always decreased in treated groups compared to vehicle groups, although there was some tumor-size dependence ( Figure 6, D and E). In particular, only in the smaller tumors (< 0.6 g) were tumor Met levels significantly reduced, 47 to 48% (1-4 weeks, p = .00201, n = 18; 3-4 weeks, p = .022, n = 8), compared to vehicles; these reduced Met levels in the treated groups remained relatively constant regardless of the time on treatment (see Figure 6D). Similarly, tumor pMet levels were reduced in all the treated groups compared to the corresponding vehicle groups, with significant decreases observed in the smaller tumors (< 0.6 g) at all weeks posttreatment and tumors < 1g after 3 to 4 weeks of treatment (see Figure 6E). The pMet levels of the smaller tumors (< 0.6 g) from all treatment groups had the greatest reductions, ranging from 52 to 58%, compared to the corresponding vehicle group (< 0.6 g; p = . treatment did not cause further increases in pMet inhibition. Most likely, the significant differences between treated and vehicle groups with smaller tumors may be attributed in part to less necrosis, increased tumor homogeneity, and a concomitant increase in the accuracy of tumor sampling. Therefore, tumor samples derived from smaller tumors may more accurately assess Met/pMet levels per gram of tumor compared to larger tumors with a greater degree of necrosis-induced heterogeneity. Although, for the most part, the results of the nonimaged groups treated for 3 to 4 weeks agreed with the results of the imaged groups, it is important to note that the tumors of the imaged vehicle group are larger necrotic tumors and, as such, most comparable to the nonimaged vehicle groups with the larger tumors, < 3 g and < 1 g (see Figure 6, D and E, and Figure 4E).

Cy5 ⋆⋆ -Met Peptide (AH-112543) Spectral Fluorescence Imaging Studies
Preliminary Cy5 ⋆⋆ -Met peptide (Cy5 ⋆⋆ -Met) imaging studies were performed in MKN-45 xenograft mice to determine the optimal imaging parameters and the relationship between tumor size and signal intensity. MKN-45 xenograft mice were injected intravenously with 0.5, 1, or 2 nmol of Cy5 ⋆⋆ -Met and imaged at 30, 60, 90, and 120 minutes postinjection; the best results were obtained with the dose of 1 nmol after 60 minutes of uptake in which tumor fluorescent signal intensities were 2-to 5-fold greater than background ( Figure 7A). The kidneys were the only other organs visualized in the images, indicating that renal clearance was the primary route of excretion. MKN-45 xenograft mice were imaged once a week for 4 weeks, and Cy5 ⋆⋆ -Met fluorescence (signal intensity, scaled counts/s) was quantitated. The Cy5 ⋆⋆ -Met tumor uptake increased linearly and highly correlated with tumor volumes < 350 mm 3 (r = .965) and < 100 mm 3 (r = .985), whereas tumor volumes that exceeded 600 mm 3 had decreased uptakes in relation to their volumes, causing a 4-to 5-fold decrease in slope and a poorer linear correlation (r = .859; Figure 7B). These results confirm that tumors > 600 mm 3 have significant necrosis limiting Cy5 ⋆⋆ -Met uptake. Additionally, larger lesions have reduced uptake due to limitations in detecting the fluorescent signal because of tissue depth, absorption, and light scattering. 36 It is important to note that these limitations have resulted in complicating accurate image reconstruction, so the fluorescence signal (signal intensity, scaled counts/s) represents a relative measure of Cy5 ⋆⋆ Met concentrations in the tumors, whereas the Met concentrations determined from the SPECT imaging with the [ 99m Tc] Met peptide uptake represent an absolute measure. The 1 nmol dose required for the fluorescent imaging was 3 to 10 times higher than the dose required for SPECT imaging, which may also be attributable to this variable tissue attenuation of the fluorescent signal. 24,37 The potential of the Cy5 ⋆⋆ -Met peptide to assess changes in Met status during treatment with the targeted Met TKI PHA 665752 was evaluated in MKN-45 xenograft mice. The xenograft mice were injected with Cy5 ⋆⋆ -Met (1 nmol) and imaged after 1 hour of uptake to obtain baseline images before treatment. Mice were then treated with either vehicle or PHA 665752 (5 days on/2 days off) for the next 3 weeks and imaged weekly; after each imaging session, tumor volumes were determined from caliper measurements. In Figure 7B, representative longitudinal images from a single PHA 665752-or vehicle-treated mouse show a decrease in signal intensity in the treated tumor compared to the vehicle control. Fluorescent signal intensity in the treated tumors was significantly decreased 54% and 55% (p < .05) at 11 and 18 days, respectively, compared to the vehicle controls ( Figure 7C). Similarly, tumor volumes (% increase from baseline) at 11 and 18 days significantly decreased 49% and 64% (p < .05), respectively, in treated versus vehicle controls ( Figure  7D). At 25 days, although both the fluorescent signal intensity and tumor volumes were decreased in the treated xenografts, these decreases were not significant, which may be explained by an increase in variability attributable to necrosis in the larger tumors as well as a decrease in the number of samples.
After the last imaging session (18 or 25 days), plasma and tumors were removed from treated and vehicle xenograft mice for analysis of plasma for shed Met and tumors for necrosis, apoptosis, total Met, and pMet. The tumors from these vehicle and treated xenografts were large, with 89% and 78% > 0.5 g, respectively. Again, moderate to severe necrosis was found in 89% of the tumors from both groups, whereas apoptosis was mostly minimal to mild, with a trend of less severity in the vehicle controls (78%, 1+) compared to the treated xenografts (33%, 1+) ( Figure 7E).  Tc] or Cy5 ⋆⋆ -Met peptide tumor uptake and similarly exhibited decreased levels resulting from PHA 665752 treatment. In addition, tumor proliferation and growth were slowed due to treatment, which corresponded to decreased levels of tumor Met and pMet. Together these studies would suggest that Met imaging may be useful in detecting tumors with Met expression and may also provide a pharmacodynamic readout of tumor responses to a Met-targeted TKI.

Discussion
The significant inhibition in MKN-45 tumor growth observed with PHA 665752 treatment in these studies was consistent with published reports in which PHA 665752 inhibition of Met tyrosine kinase activity resulted in a reduction of tumorigenicity accompanied with increases in apoptosis and necrosis. 30,31 In our studies, similar treatment responses were observed in which MKN-45-treated tumors decreased in volume and exhibited substantial necrosis and mild-moderate apoptosis. Additionally, substantial necrosis was observed in larger tumors due to the combined effects of treatment and inadequate blood supply. Necrosis leads to decreased [ 99m Tc] or Cy5 ⋆⋆ -Met peptide tumor uptakes and decreased total tumor Met and pMet levels. Increased necrosis likely explains the lack of significant difference in uptake between treated and control groups in larger tumors. These necrotic areas would contribute to the overall mass of the tumor but would lack cellular Met expression, thus lowering Met levels per gram of tumor tissue. Likewise in larger control tumors, Met and pMet levels were decreased but exhibited a greater degree of variability dependent on the extent of the necrosis. Greater variability might be expected with these large mass tumors because these determinations are made from a small sample of the tumor, which may not be entirely representative of the Met, pMet, or necrotic status of the entire tumor. In contrast, quantitation of the entire tumor volume is possible with imaging and would seem to be especially beneficial in quantifying these large tumors with heterogeneously distributed necrotic areas. Collectively, these results would suggest that the reduced [ 99m Tc]  Met status is needed to identify those patients most likely to benefit from a Met-targeted therapeutic. These Met imaging agents could provide such an assessment not only in the case of primary disease but also, more importantly, could detect the emergence of Metdriven tumors in patients undergoing treatment, indicating "when" a Met-targeted therapy should be implemented.        Mol Imaging. Author manuscript; available in PMC 2020 December 02.