1.

Introduction

Gastric cancer is a common malignant tumor, which is one of the most frequent causes of cancer-related death, especially in developing countries.^1,2 The conventional treatment includes surgery, radiation therapy, and chemotherapy. However, gastric cancer is difficult to cure unless it can be detected at an early stage.^3,4 Therefore, the treatment of gastric cancer needs a unique treatment approach; fortunately, angiogenesis-targeted therapy has emerged as a promising approach for it.⁵ Correspondingly, a cyclic 9-mer peptide named GX1 was identified by Zhi et al.,⁶ which exhibits the specific ability to target vasculature of gastric cancer.^7–9 Furthermore, the GX1 peptide was successfully conjugated with near-infrared (NIR) fluorescent dye Cy5.5 to obtain the Cy5.5-GX1 probe, which could be used for tumor-targeted imaging and detection.¹⁰ Unfortunately, the in vivo pharmacokinetic properties of GX1 have not been fully studied yet.¹¹ The injection dose (ID) of the GX1-based probe used in current studies is often determined according to experience, and the optimal dose is also unexplored. Appropriate ID of targeted probes not only avoids receptor oversaturation and wasting of probes but also can avoid the phenomenon of image overexposure caused by excessive probe injections. Thus, extensive investigations focused on the influence of the probe ID on pharmacokinetics quantification may provide valuable information about the optimal dose of molecular therapeutic agents for preclinical trials. On the other hand, camera integration time (IT) is also an important parameter that can affect the quality of dynamic fluorescence imaging.¹² Therefore, it is necessary to do further in-depth research on the tracer ID and camera IT-dependent pharmacokinetic quantification, which would be valuable for the preclinical application of GX1-based targeted probes.

Because it can provide functional information at cellular and molecular levels for image-guided optimization and management of personalized cancer treatment, noninvasive molecular imaging of angiogenesis may be a method for the early diagnosis of cancer.^13,14 Over the past few decades, NIR fluorescence imaging has attracted more and more attention because of its nonradiation property and low tissue absorption.^15–19 In particular, dynamic fluorescence imaging combined with a kinetics analysis offers multiple advantages in comparison to static images. For example, it can estimate the kinetic parameters, which can be used to calculate the metabolic rate and receptor density.^20,21 Kinetic analysis can also separate a specific signal from a nonspecific one, which further improves the accuracy of measuring the binding potential (Bp) of a targeted probe.^22–24 Some in vivo quantitative studies have also been carried out by dynamic fluorescence imaging combined with the kinetics analysis.^25–30 As described in Ref. 28, using dynamic fluorescence imaging together with an exponential model, Kwon et al. successfully determined the uptake parameters as a function of the ID of Cy5.5-c(KRGDf) in the tumor region. Zhu et al.³⁰ designed and synthesized a 64Cu/ZW-1-based dual-modality imaging probe [c(RGDyK)-C(DOTA)-ZW-1] to demonstrate for the first time that dynamic fluorescence imaging could obtain comparable quantitation results in vivo compared with dynamic positron emission tomography (PET) imaging. In addition, in our previous study, we quantified molecular specificity of Cy5.5-labeled cyclic 9-mer peptide (Cy5.5-GX1) in vivo with dynamic fluorescence imaging, which provides better understanding of its kinetic properties.¹¹

In this study, we investigated the influence of the tracer ID and camera IT on quantifying pharmacokinetics of Cy5.5-GX1 in gastric cancer-bearing nude mice. Based on three factors, including whether or not to inject free GX1, the ID of Cy5.5-GX1, and the camera IT, 32 mice were randomly subdivided into eight test groups and received 60-min dynamic fluorescence imaging. We used the Gurfinkel exponential model (GEXPM) and Lammertsma simplified reference tissue model (SRTM) combined with the singular value decomposition (SVD) analysis to quantitatively analyze the obtained dynamic fluorescent images. Here, the SRTM was selected to estimate the Bp of the injected tracer. The Bp is one of the most important macroparameters for pharmacokinetic quantification of Cy5.5-GX1, which is defined as the product of tracer binding affinity and available receptor density.^30–32 Correspondingly, GEXPM was selected to calculate the sum of the pharmacokinetic rate constants (SKRC), which is another important macroparameter that reflects receptor–ligand affinity.²⁹ In addition, the reason for choosing these two models is because they are able to avoid blood sampling and capture the parameters of interest.³³ Then, the change of Bp and SKRC as a function of the tracer ID and camera IT was drawn upon, whose results would provide valuable references for preclinical applications of the GX1 probe.

2.

Materials and Methods

3.

Results

3.1.

Dynamic Fluorescence Imaging

Figure 1(a) shows one of the representative raw fluorescent images (without background subtraction) at 24 h after the 2-nmol Cy5.5-GX1 injection. The tumor could be identified from the fluorescent images in the process of data acquisition. To confirm the specificity of Cy5.5-GX1 in vivo, dynamic fluorescence imaging was performed on the mice, which received an excess of free GX1 peptide as a blocked agent 1 h prior to the Cy5.5-GX1 injection. The corresponding representative raw fluorescent image is shown in Fig. 1(b). Under the same conditions of camera IT and dynamic display scale, there were no obvious fluorescent signals in the tumor region for the blocked experiment. Especially in the fluorescent image at 24 h after the probe injection, almost no fluorescent signal was observed in the tumor region. The high-intensity fluorescent signal in the kidneys suggested that Cy5.5-GX1 was excreted through the bladder.³⁵

Fig. 1

Representative raw fluorescent images (without background subtraction) generated at 24 h after administration of (a) 1-nmol Cy5.5-GX1 injected alone and (b) excessive free GX1 injected 1 h prior to the Cy5.5-GX1 injection. White arrows indicate the tumor locations.

Figure 2(a) shows the tumor-to-muscle region (TMR) ratio as a function of time after the tracer injection for the experiments with the same tracer ID but different camera ITs. We found that the longer the camera IT, the greater the TMR value. Especially for the time point at 24 h after the tracer injection, there was a significant statistical difference in the TMR for the three groups ($p<0.01$, ANOVA). Figure 2(b) shows the TMR as a function of lapse time after the tracer injection for another four selected groups. Three groups were selected with the same camera IT but different tracer IDs, whose posthoc testing results showed that their TMRs had no statistical difference. The remaining group was the blocked group (ID: 1 nmol, IT: 1 s) and the TMR value was significantly different from that of the former three groups (all $p<0.01$, LSD).

Fig. 2

Ratio of the TMR as a function of time after the tracer injection: (a) groups with the same tracer ID but different camera ITs and (b) three groups with the same IT but different tracer IDs, and a blocked group.

3.2.

Kinetic Analysis

First, we analyzed the Bp value as a function of the camera IT and tracer ID. For the investigation of the camera IT, we used the same tracer ID but different camera ITs. The estimated Bp values are shown in Fig. 3(a). Statistical analysis demonstrated that the Bp values in this investigation experiment had no statistical significance ($p=0.81$, ANOVA). In the investigation of the tracer ID, the camera IT was fixed at 1 s, but the tracer ID was set to be from 0.5 to 3 nmol. In addition, the blocked experiments were also conducted. Figure 3(b) shows the estimated Bp values for this investigation experiment. Posthoc testing results confirmed that there was also no statistical difference for the Bp values in different ID groups. However, if excessive amounts of free GX1 were injected as a blocking agent, the corresponding Bp values were significantly different from the others (all $p<0.01$, LSD).

Fig. 3

Bp values of Cy5.5-GX1 and SKRC values estimated in the tumor and muscle ROIs as a function of the camera IT and tracer ID, respectively. Relationships between (a) Bp values and different camera ITs, (b) Bp values and different tracer IDs, (c) SKRC values and different camera ITs, and (d) SKRC values and different tracer IDs.

Second, we analyzed the relationship of the SKRC values in the tumor and muscle ROIs with the tracer ID and camera IT. In the first step, we fixed the tracer ID as 1 nmol and changed the camera IT to 0.5, 1, and 1.5 s. Figure 3(c) shows the estimated SKRC values in the tumor and muscle ROIs, respectively. We found that the SKRC values exhibited no statistical difference for different camera ITs wherever they were (in the tumor ROIs: $p=0.52$ and in the muscle ROIs: $p=0.50$, ANOVA). In the second step, the camera IT was fixed at 1 s, but the tracer ID was changed from 0.5 to 3 nmol. Similarly, the blocked group that was injected with excessive amounts of free GX1 was also included. Figure 3(d) shows the related statistical analysis results. From Fig. 3(d), some interesting conclusion could be addressed. First, an approximate linear relationship between the SKRC values estimated in the tumor ROIs and the Cy5.5-GX1 ID was observed from 0.5 to 1.5 nmol. The SKRC value became larger as the tracer ID increased. Second, the SKRC values estimated in the tumor ROIs remained unchanged when the tracer ID ranged from 1.5 to 3 nmol, and posthoc testing results indicated no statistical difference between them. However, they were statistically different from those with a 0.5- and 1-nmol tracer ID (all $p<0.01$, LSD). Third, the SKRC values estimated in the muscle ROIs did not show any statistical difference for different tracer IDs ($p=0.53$, ANOVA). Lastly, in the blocked group, there was no statistical difference for the SKRC values between the tumor and muscle ROIs ($p=0.08$, $t$-test). However, the SKRC values of tumor ROIs in the blocked group were significantly different from those estimated in other nonblocked groups (all $p<0.01$, LSD).

4.

Discussion

Angiogenesis is the basis of tumor progression, so vascular suppression therapy is a promising remedy that can bring new hope to cancer patients.³⁶ The identification of molecules with receptor binding affinity is essential for this treatment. It can promote the specific targeting ability of agents, which can further improve the therapeutic effect. In this study, with the help of dynamic fluorescence imaging, we observed an increase and retention of Cy5.5-GX1 uptake in tumor ROIs. Such a specific uptake was abrogated when excess amounts of free GX1 peptide were injected 1 h before the tracer injection. These results demonstrated that Cy5.5-GX1 was a promising probe with receptor binding affinity in vivo and had great potential in application of vascular suppression therapy. In Sec. 3.1, statistical analysis results suggested that the TMR of Cy5.5-GX1 was dependent on the camera IT but was not sensitive to the tracer ID. In other words, we can increase the camera IT to obtain a higher TMR value.

Compared with conventional static imaging, dynamic fluorescence imaging combined with kinetic analysis provides more information on Cy5.5-GX1 in the metabolic process, which may accelerate the clinical application of GX1. One of the most interesting parameters is Bp, which is used to estimate the available receptor density in the ROIs and is also considered as an indicator of the equilibrium state of the tracer.³⁴ From Fig. 3(b), we did not find any significant difference for Bp values when the ID of Cy5.5-GX1 was increased from 0.5 to 3 nmol, indicating that the available receptor of GX1 did not reach its saturation state, which was in this dose range. Similarly, no statistical difference was observed for the SKRC values in the tumor ROIs when the ID of Cy5.5-GX1 was increased from 1.5 to 3 nmol. Such a dose-independent response may be caused by the mechanism of partial self-inhibition at a higher dose rather than saturated receptor binding.^28,37 This is because the SKRC value is equal to the sum of the kinetic parameters, which cannot be used directly to estimate the available receptor density. Because GEXPM describes the kinetic characteristics of the tracer with a one-way process that counts only its continuous accumulation in the tumor ROI, it may be more appropriate to reflect on the kinetics of the tracer during the initial stage of the injection. As a result, the combination of SKRC and Bp values may be more comprehensive in describing the kinetic characteristics of the tracer in vivo.

If we fixed the tracer ID and changed the camera IT, there would be no statistical differences observed for both the SKRC and Bp values. This suggests that the SKRC and Bp values were independent of the camera IT. This conclusion is useful when we want to obtain a high TMR value without compromising the dynamic quantitative results. At a lower ID (0.5 nmol), the tracer may selectively accumulate in the kidneys or other parts rather than in the tumor region, which may reduce the ratio of tumor-to-muscle.³⁸ The resulting Bp values and SKRC values of tumor ROIs are lower, but both of them are still significantly different from those of the blocked group ($p<0.05$, LSD). In addition, when the tracer ID reaches 6 nmol and the camera IT is set at 0.5 s, saturated pixels will appear in the acquired fluorescent images. In this case, one needs to reduce the camera IT to avoid saturated pixels, but this also leads to a decrease in TMR. From this point of view, a higher tracer ID may not be suitable for the quantitative analysis. In conclusion, the Bp and SKRC are essential kinetic parameters when using dynamic fluorescence imaging for the quantitative analysis, and both of them can provide more information on ligand–receptor binding than the TMR obtained from a static fluorescent image.

In this study, we evaluated the influence of the tracer ID and camera IT for quantifying the pharmacokinetics of Cy5.5-GX1. We also compared the performance of the pharmacokinetic parameters obtained by dynamic fluorescence imaging with the TMR obtained from a static fluorescent image in the quantitative analysis. In dynamic fluorescence imaging, the determination of the targeted probe ID could avoid receptor oversaturation and wasting of probes, which makes the obtained quantitative results more accurate and reliable. Hence, our quantitative results may provide valuable information about the optimal dose of molecular therapeutic agents for the preclinical application of GX1-based targeted probes, such as the preclinical study of individual therapy. In addition, the camera IT is also a quantitative factor, which has to be considered, and it is related to the quality of dynamic fluorescence imaging. The determination of the relationship between camera IT and pharmacokinetic parameters is beneficial for a more reasonable design of the experimental program. In summary, our quantitative methods and results will provide a valuable reference for future research of GX1-based targeted probes.

5.

Conclusions

Dynamic fluorescence imaging combined with a kinetic analysis may provide more robust quantitative results than the static fluorescence imaging technique. In particular, it could provide information about the optimal ID for individualized cancer therapy. Our results could provide a reference for future pharmacokinetic studies of GX1 and also promote its future clinical applications.