Open Access
Add to BibSonomy
Abstract
Pharmacokinetics and biodistribution studies are essential for characterizing fluorescent agents in vivo. However, few simple methods based on fluorescence imaging are available that account for tissue optical properties and sample volume differences. We describe a method for simultaneously quantifying mean fluorescence intensity of whole blood and homogenized tissues in glass capillary tubes for two fluorescent agents, ABY-029 and IRDye 680LT, using wide-field imaging and tissue-specific calibration curves. All calibration curves demonstrated a high degree of linearity with mean R2 = 0.99 ± 0.01 and RMSE = 0.12 ± 0.04. However, differences between linear regressions indicate that tissue-specific calibration curves are required for accurate concentration recovery. The lower limit of quantification (LLOQ) for all samples tested was determined to be < 0.3 nM for ABY-029 and < 0.4 nM for IRDye 680LT.
© 2024 Optica Publishing Group under the terms of the Optica Open Access Publishing Agreement
1. Introduction
Pharmacokinetics (PK) and biodistribution (BD) of therapeutic and diagnostic (imaging) agents play important roles in both pre-clinical and clinical studies, as they provide vital information on the kinetic behavior of an agent in vivo. For a potential drug candidate or drug delivery system, the study of PK and BD assists rational assessment on its efficacy and toxicity profiles [1–3]. The advancement of molecular imaging has generated a plethora of imaging agents designated for different modalities and numerous applications, including fluorescent imaging agents [4]. The field of fluorescent imaging agents is rapidly growing, with a number of agents in clinical testing for fluorescence guided surgery [5,6], and even recently approved for use in humans [7]. Consistent with drug assessment, the development and characterization of fluorescent imaging agents involves PK and BD studies to evaluate safety profiles, tissue targeting performances, and tissue image contrast [8]. The USA Federal Drug Administration (FDA) requires extensive biological testing for Phase 0 or 1 studies of imaging agents including single-dose acute toxicity testing, PK, and BD.
Furthermore, determining accurate PK and BD is especially important for paired-agent imaging (PAI), where targeted and untargeted agents with unique fluorescence emissions are co-administered and differences in their kinetic profiles are used for quantitative determination of in vivo protein receptor concentrations [9–12]. For this imaging technique, the observed contrast is dependent on the ratiometric fluorescence from both contrast agents. Therefore, it is important for the targeted and the untargeted dyes to be similar in size, charge, lipophilicity, and kinetics. It is vital that the PK of the targeted and the untargeted agents closely match during tissue distribution and clearance for accurate kinetic compartment modeling [10]. If the PK profiles of the two agents do not match, differences need to be mathematically accounted for accurate receptor concentration determination [13]. Moreover, computational kinetic modeling can be used to predict tracer activity and understand experimental results, especially in the case of PAI. PK curves of the imaging agents at hand are essential for producing accurate computational models, which we have demonstrated extensively [14–18].
Multiple approaches have been employed to assess imaging agent PK and BD in whole blood and tissue based on method of detection, lab equipment, and cost [1,19]. Analysis of plasma or tissue homogenate by HPLC or LC-MS/MS; (liquid chromatography with tandem mass spectrometry) is a common methodology and is highly quantitative, but requires chemical extraction and precipitation or filtering to remove particulate matter. For instance, BD and PK studies for the IRDye 800CW drug master file FDA application were performed using HPLC on plasma and tissue homogenate that was chemically extracted, centrifuged, and filtered to remove particulate matter [20]. Similarly, the PK and BD analysis of LUM015 was performed with a combination of LC-MS/MS and HPLC with fluorescence detection, respectively, using the supernatant from chemically extracted and precipitated plasma and tissue homogenates [21]. Finally, the PK analysis of tozuleristide (BLZ-100) was performed using LC-MS/MS on blood serum collected from humans [22]. However, agent recovery efficiency depends on the extraction and homogenization methods applied, and molecular targeted agents that are bound to cells/tissue need to be extracted through protein denaturation to achieve high rates of recovery [23].
The ability to measure agents in whole blood and tissues without chemical modification or particulate removal circumvents the need to separate the agents from their molecular targets. Radiolabeled agents can easily be quantified in whole blood and tissues due to the absence of background signals but require special laboratory conditions and regulatory approvals. On the other hand, fluorescence signal can also be easily recovered from whole tissues and blood and is easily measured in a laboratory setting. Despite the rapid development of fluorescence molecular imaging, the number of methods capable of consistently and accurately quantifying imaging agent PK and BD using fluorescence profiles from whole blood and tissue remains low [24]. BD and PK analysis of the recent FDA approved Cytalux (OTL38) were performed by imaging collected serum and whole organs from mice in the IVIS imaging system [25]. The BD of cetuximab-IRDye800CW in Cynomolgus macaques [26] and ABY-029 in rats [27] were analyzed using whole tissues in the Pearl Impulse instrument. The cetuximab-IRDye800CW study included conversion to concentration using a single, non-linear calibration curve in unspecified matrix, while the ABY-029 study reported raw fluorescence of the whole organs. The PK in both of these studies used collected plasma but cetuximab-IRDye800CW was measured using Western blot and ABY-029 was measured with fluorometry. Additionally, fluorescence quantification of imaging agent BD has been performed with fluorescence molecular tomography, whole organ imaging, and ex vivo tissue homogenization [27–30]. These methods of determining fluorescent agent concentration are challenging due to non-zero background signal (i.e., autofluorescence) and the absorption and scattering properties of tissues, which can alter the observed fluorescence emission intensity [31].
Two studies have demonstrated the use of a 96-well plate fluorescence assay for homogenized whole tissues. Oliveira et al. (2012) applied a single calibration curve created in buffer to determine homogenate concentration, which did not consider optical property differences between tissues and had high variance in several organs, including spleen, liver and tumor [32]. A more recent study by Iaboni et al. (2023) demonstrated a 96-well plate fluorescence assay but used tissue-specific calibration curves to overcome the limitations of quantifying fluorescence signal in media of differing optical properties [24]. Although this technique was demonstrated to be robust with lower limit of quantification (LLOQ) at 800 nm to be 0.5 nM, the calibration was dependent on the Z-position of the plate reader and the concentration of the solution used for determining the position. This dependency suggests that the depth of the solution was not equally sampled by the excitation light (i.e., excitation pathlength) for all solution concentrations and optical properties. Here, we expand on this technique to evaluate an assay that employs wide-field fluorescence imaging of homogenized tissue and whole blood in borosilicate capillary tubes to determine PK and BD parameters of multiple fluorescent agents simultaneously. Application of system- and tissue-specific calibration curves for extrapolating concentration values for whole blood and freshly excised, homogenized tissues allows this analysis to be performed with any wide-field fluorescence imaging system (Fig. 1). The use of borosilicate capillary tubes standardizes the path length of excitation, which should be equally penetrable by both visible and NIR light. To demonstrate viability of this technique, we simultaneously characterize two spectrally and chemically distinct, fluorescent agents, ABY-029 and IRDye 680LT. ABY-029 is an epidermal growth factor receptor (EGFR)-specific Affibody molecule conjugated to a near-infrared dye, IRDye 800CW, and is currently being used in pre-clinical and clinical testing for fluorescence guided surgery. IRDye 680LT is a pharmacokinetically similar dye with no protein conjugation, is frequently used as the untargeted agent in PAI, and shows strong potential for clinical translation [11].
2. Materials and methods
An experimental flow chart overview is provided in Fig. 1 to illustrate the generalized procedures for both PK and BD determination using wide-field fluorescence imaging.
2.1 Imaging agents
ABY-029 was synthesized under Good Laboratory Practice (GLP) protocols by the University of Alabama at Birmingham (UAB) Vector Production Facility as described previously [27]. IRDye 680LT NHS ester was purchased from LI-COR Biosciences, Inc. (Lincoln, NE). IRDye 680LT carboxylate, the untargeted agent in this study, was acquired by dissolving IRDye 680LT NHS ester in phosphate-buffered saline (PBS, pH = 8.5) and stirring at room temperature for 5 hours, as per manufacturer’s instructions. All the test animals were administered with a 1:1 molar ratio mixture of ABY-029 and IRDye 680LT, based on an ABY-029 dose of 0.0487 mg/kg in 200 µL of PBS via tail vein injection. As established previously, this is equivalent to the human 0.00395 mg/kg (180 nmole) dose calculated with the method of Reagan-Shaw et al. [28] and used in human Phase 0 testing.
2.2 Cell lines and culture methods
The human squamous cell carcinoma cell line (HNSCC), FaDu, was purchased from ATCC (Manassas, VA, USA) and cultured in Dulbecco's Modified Eagle Medium (DMEM) with 10% Fetal Bovine Serum (FBS) and 1% penicillin-streptomycin.
2.3 Naïve and xenograft mouse models
All animal procedures were conducted according to a protocol approved by the Dartmouth Institutional Animal Care and Use Committee (IACUC) and conducted according to NIH-OLAW and AAAALAC guidelines. Female athymic nude mice, were purchased from Charles River Laboratories (Wilmington, MA) at 6–8 weeks of age. A total of 11 mice (NPK = 11) were used for the plasma pharmacokinetic study, and 6 mice (NBT = 6) were used for the biodistribution study.
For tumor implantation, all mice (NBT = 6) were inoculated with 1.5 × 106 FaDu cells in 50 µL culture medium by subcutaneous injection on the left flank. The mice were randomly assigned into two groups: one group to establish tissue-specific calibration curves (ncc = 3) and the other group for tissue BD study (nbd = 3). Mice were euthanized and tissue harvested when tumor size reached approximately 600 mm3.
2.4 Fluorescence data collection and analysis
2.4.1 Fluorometry – Fluoromax
Fluorescence emission spectra were collected for plasma solutions using the Fluoromax-3 (Horiba Jobin Yvon, Edison, NJ). Each plasma phantom was placed in a 40 µL micro-cuvette (Starna Cells, Atascadero, CA) and the fluorescence emission spectra were collected for both dyes. For IRDye 680LT, an excitation wavelength of 620 nm and fluorescence emission collection range of 650-800 nm was used. For ABY-029, an excitation wavelength of 720 nm and fluorescence emission collection range of 730-900 nm was used. The resulting fluorescence spectra area under the curve was approximated with the trapezoidal method using MATLAB Version 2022b (MathWorks, Natick, MA). The peak intensity was calculated by determining the maximum intensity for each spectrum.
2.4.2 Wide field fluorescence – Pearl Impulse
For both PK and BD studies, the fluorescence images were acquired using the Pearl Impulse Small Animal Imaging System (LI-COR, Lincoln, NE). For each sample, white light, 700 nm (IRDye 680LT) and 800 nm (ABY-029) channel images were captured simultaneously using 85 µm resolution (PK) or 168 µm (BD). Three types of containers were imaged: 15 mL and 2 mL centrifuge tubes, and borosilicate capillaries. The centrifuge tubes were imaged one at a time on the animal bed, while the capillary tubes were taped with optical black masking tape (ThorLabs, Newton NJ) to a flat surface covered with optical black aluminum foil (ThorLabs, Newton NJ) (Fig. 2(b)). Each capillary tube was imaged one at a time on the Pearl to avoid fluorescence crosstalk and saturation. The mean fluorescence intensities (MFI) were determined using MATLAB Version 2022b (MathWorks, Natick, MA). Firstly, a quadrilateral region of interest (ROI) was drawn to close crop the image to within 2-3 mm of the sample. Then, regions of interest (ROI) were created for each sample as well as the background (Fig. 2(a)-(c)) on the white light images. The ROIs were transferred to the 700 nm or 800 nm fluorescence images and the MFIs for each ROI were recorded for the background and sample. The mean background fluorescence intensity was subtracted from the mean sample fluorescence intensity to account for inconsistent background signal.
Fig. 2. Blood and plasma phantom configuration and linear regressions. (a) and (b) Representative images that display how the large volumes for blood and plasma, respectively, were measured on the Pearl. (c) A representative image that shows the small volume measurements in a capillary tube on the Pearl. The regions of interest (ROI) for the sample fluorescence (blue box) and the background fluorescence (red box) measurements are shown. (d) A schematic of how the blood and plasma phantoms were measured in capillary tubes. The darker opacity corresponds to black aluminum foil used to cover the neighboring samples to avoid fluorescence crosstalk. A similar layout was used for measuring the in vivo blood fluorescence. (e) and (f) Linear regressions of the Pearl florescence versus concentration for each type of blood and plasma phantom measured in the Pearl for IRDye 680LT and ABY-029, respectively. The data was transformed to log10 scale for better visualization. (g) and (h) Linear regressions for IRDye 680LT and ABY-029, respectively, area under the curve and peak intensity versus concentration measured on the fluorometer. All fluorescence and concentration data were transformed to log10 scale for linear analysis.
2.5 Blood and homogenized tissue calibration curves
2.5.1 Blood phantom preparation
Bovine whole blood with sodium citrate anticoagulant was purchased from Lampire Biological Laboratories (Lampire, Pipersville, PA) and stored in 4°C. A stock solution with 1:1 molar ratio of the IRDye 680LT and ABY-029 dyes was created with a concentration of 0.26 µM in bovine blood. The stock solution was serially diluted to create additional blood phantoms with concentration at 0, 0.0026, 0.013, 0.026, 0.13, and 0.26 µM. The serially diluted phantoms were produced, imaged, and quantified to create calibration curves as described in Sections 2.4 and 2.5. The phantom concentrations were chosen based on Samkoe et al., where they found that 0-0.26 µM was within the linear range [30]. For each phantom, the effects of large and small volumes (effectively changing the sample path length) on MFI were examined using four different measurement types. Additionally, the difference between whole blood and plasma fluorescence was also examined.
The blood phantom imaging protocol was previously described in Pannem et al., and the process is summarized here [29]. Briefly, 2000 µL of the whole blood phantom was placed in a 15 mL centrifuge tube and imaged on the Pearl to look at how a large volume effects the MFI (Fig. 2(a)). For small volume effects, approximately 50 µL of the blood phantom solution was pipetted onto a piece of Parafilm (Bemis NA, Neenah, WI, USA) and then wicked into a capillary tube (Kimble Chase, Vineland, NJ). The remaining blood phantom solution was then centrifuged (10 min, 2500 rpm, 4°C) to isolate plasma.
The plasma was collected into a 2.0 mL microcentrifuge tube and imaged on the Pearl (Fig. 2(a)). After imaging, approximately 50 µL of the plasma phantom was placed on a piece of Parafilm and then wicked into a capillary tube. The capillary tubes were measured on the Pearl using the procedure described above for whole blood phantoms. The background subtracted MFI for each phantom concentration, measurement type, and dye were determined as described in Section 2.4.2. The plasma was then transferred to a cuvette and the emission spectra collected on fluorometer, as described in Section 2.4.1. The Pearl and fluorometer measurement techniques were compared using root mean squared error (RMSE), which is a metric that represents the distance between the fit line and the experimentally acquired datapoints.
2.5.2 Homogenized tissue preparation
Under anesthesia, mice with subcutaneous xenograft HNSCC tumors were sacrificed by cervical dislocation, and the following tissue samples were collected immediately after: brain, heart, spleen, lung, tumor, kidneys, liver, a fraction of intestine, skin, and muscle. All tissue samples were rinsed with PBS twice and dried using a Kimwipe (Kimberly-Clark, Dallas, TX, USA). The weight of each organ was recorded. Due to the large size of the liver, a smaller piece from the left lobe was excised and re-weighed. Only the excised liver piece was used for homogenization.
All tissues collected were placed in polystyrene test tubes. The amount of PBS buffer, and time required for homogenization was determined by tissue type and mass. For each tissue sample, PBS was added to the tube based on the homogenization factor (HF), which is the ratio between PBS volume (µL) and tissue mass (mg). For example, a HF of 3 means a 100 mg tissue sample was added with 300 µL PBS for homogenization. The tissue was then homogenized using a VWR 200 Homogenizer (VWR, Radnor, PA) at highest speed. For all homogenization processes, the time interval was kept between 10–20 s followed by resting for 5 s. For tumor, skin, and muscle samples, a fraction of PBS was first added, namely 1, 3, and 3 times of tissue mass; and homogenized for 20, 40 and 40 s, respectively. Then additional PBS with 2, 3, and 3 times of tissue mass were added to the homogenates, respectively; and re-homogenized for 20 s. This process allowed for more complete blending for muscle, skin, and tumor due to their fibrous texture. The average mass, amount of PBS added, and the corresponding time length of homogenization of each tissue type were summarized in Table 1.
Table 1. Summary of experimental variables for tissue homogenization.a
For each type of tissue, a serial dilution of paired agents was performed using tissue homogenate as the solvent. A 1:1 molar ratio mixture of ABY-029 and IRDye 680LT was added to tissue homogenate to generate the initial concentration of 2600 nM. A 10-fold serial dilution was performed to acquire tissue homogenates with concentrations of 260, 26, 2.6, and 0.26 nM. For one mouse, an additional dilution for each tissue (with the exception of heart) was made at 0.026 nM. All tissue homogenates were loaded into heparinized capillary tubes and imaged using Pearl as described in Section 2.4.2.
2.5.3 Calibration curves
All calibration curves were created by performing a log10 transformation on the fluorescence signal (mean fluorescence intensity, MFI, area under the peak, or peak maxima). For each channel, a log10 fluorescence signal versus log10 concentration plot with all measurement types was created using GraphPad Prism version 9 (PK) or version 10 (BD) (GraphPad Software, San Diego, CA, USA). A simple linear regression was performed using GraphPad Prism on the transformed data to determine a relationship between the measure of fluorescence emission and concentration for each measurement type. The linear regression equations followed the following format:
where k represents the slope and b represents the Y-intercept. The coefficient of determination (R2) was calculated for each linear regression. Additional statistical analyses were performed as described in Section 2.7.2.5.4 Lower limit of quantification (LLOQ)
The LLOQ was calculated for blood and tissue sample type using the method described by Iaboni et al. (2023) [24]. Briefly, the autofluorescence of each sample was determined without fluorophore being added. The LLOQ for each sample type was found to be 5-times the autofluorescence signal.
2.6 In vivo mouse studies
2.6.1 Plasma pharmacokinetic study
The fluorophore excretion of IRDye 680LT and ABY-029 were determined by monitoring the blood over an 8 h period. Naïve nude mice (NPK = 11) were intravenously administered with 200 µL of 1:1 molar mixture of IRDye 680LT and ABY-029 via tail vein (0.7 nmol each based on an average mouse weight of 20 g) [28]. Whole blood was collected at 1 min, 2 min, 4 min, 8 min, 15 min, 30 min, 1 h, 2 h, 3 h, 4 h and 8 h post administration. Every mouse had a maximum of 4 blood draws, with one blood sample collected at 1 h post administration for normalization to minimize errors due to variations in volume. For the remaining time points, blood was collected from three unique mice (Fig. S1). At each timepoint, the tail artery was nicked with a scalpel, approximately 50 µL of blood was collected with a heparinized capillary tube (Kimble Chase, Vineland, NJ), and the bleeding was stopped with light pressure using sterile gauze pad. After collection, the capillary tubes containing blood were imaged using the Pearl and the MFI was determined for each sample as described in Section 2.4.2 and concentrations determined from calibration curves determined in Section 2.5.3. MATLAB’s Curve Fitting toolbox was used with the bisquare weight fitting method to determine the bi-exponential decay of IRDye 680LT and ABY-029.
2.6.2 Arterial and venous blood comparison
During the in vivo plasma excretion experiment, we noticed visible variations in the color of the blood at different timepoints. To ensure that there weren’t significant differences between different levels of blood oxygenation, we compared the MFI between arterial (oxygenated) and venous (deoxygenated) blood. Three nude mice were injected with 200 µL a 1:1 molar ratio (0.7 nmol) of ABY-029 and IRDye 680LT via tail vein. At 0.25 h, 0.5 h, 1 h and 4 h, the tail artery and vein were nicked with a scalpel and approximately 50 µL of blood was collected using a heparinized capillary tube. Samples were measured using the Pearl (Section 2.4.2) and concentration quantification was performed using the corresponding calibration curves, as described in Section 2.5.3.
2.6.3 Biodistribution (BD) study
Three mice with subcutaneous xenograft of HNSCC tumor were administered with a 1:1 molar ratio mixture of 0.9 nmol (based on average mouse body weight of 25 g) ABY-029 and IRDye 680LT in 200 µL of phosphate buffered solution (PBS) via tail vein injection [28]. Mice were sacrificed 5 h after paired agent administration. Then mice underwent the same tissue collection, weighing, homogenization, and imaging process as described in Section 2.5.2. Tissue-specific concentrations of ABY-029 and IRDye 680LT were extrapolated by tissue-specific 800 and 700 nm MFIs of the homogenate from the corresponding calibration curves determined in Section 2.5.3.
2.7 Statistical analysis
Linear regression, graphing and further statistical analyses were performed using GraphPad Prism 9.0 (PK) or 10.0 (BD). For PK calibration curves, the slopes of the linear regression results from different measurement types were compared, in 700 and 800 nm channels individually, using the “test whether the slopes and intercepts are significantly different” option in GraphPad Prism (F-test). This option further compared the intercepts when the slopes were not significantly different. For BD calibration curves, a similar comparison was performed to determine whether homogenates from different types of tissue share the same optical properties by slope comparison using the same approach, in both 700 and 800 nm channels. For the arterial and venous blood comparison, a repeated measures ANOVA was performed in GraphPad Prism to compare the arterial and venous blood at each timepoint. Statistical significance was based on p < 0.05. All data were presented as mean ± standard deviation otherwise specified.
3. Results
3.1 Calibration curves for blood and tissue homogenates
3.1.1 Blood phantom quantification and calibration curves
For the blood and plasma phantoms measured on the Pearl, the MFI demonstrated a strong, positive correlation with concentration for both imaging agents and all measurement types with all R2 > 0.97 (Fig. 2(e)-(f), Table S1). The slopes of the linear fits were compared using a F-test and the results indicate that there is no significant difference between the slopes of the various measurement types (IRDye 680LT: p = 0.9138; ABY-029: p = 0.7665). The area under the curve fluorescence and the peak fluorescence intensity plasma phantom measurement performed on the fluorometer demonstrated there was a strong, positive correlation with mean R2 = 0.98 ${\pm} $ 0.01 and RMSE = 0.10 ${\pm} $ 0.02 (Fig. 2(g)-(h), Table S1).
The MFIs measured on the Pearl for the various phantom types were compared against plasma phantoms measured using the fluorometer as we have previously published [6,27]. Figure 2(f) and (g) show that there is a strong positive relationship between the area or maxima and the concentration for both dyes. The RMSE values for each phantom measurement type and imaging agent is shown in Table S1. For samples measured in the Pearl, the mean RMSE was 0.08 ${\pm} $ 0.01 for IRDye 680LT and 0.115 ${\pm} $ 0.007 or ABY-029. For the fluorometer measurements, the mean RMSE was 0.11$\pm $ 0.02 for IRDye 680LT and 0.128 ${\pm} $ 0.005 for ABY-029. The mean RMSE was 0.097 ${\pm} $ 0.02 for the wide field imaging and 0.12 ${\pm} $ 0.02 for the fluorometer measurements. The fluorometer determined calibration curves had a lower R2 and a higher RMSE when compared to the wide-field imaging (mean R2 = 0.95 ${\pm} $ 0.01 and mean RMSE = 0.12 ${\pm} $ 0.02, Fig. 2(g)-(h), Table S1). The combined lowest RMSE for both IRDye 680LT and ABY-029 was for wide field imaging of whole blood in the capillary tube. Plasma measured in capillary tube also demonstrated a low RMSE for IRDye 680LT (0.073) and ABY-029 (0.12). Based on these results, and the fact that the whole blood capillary measurements were the most time efficient, they were used to create a calibration curve that describes the relationship between MFI and concentration to use in the in vivo study. The relationships between IRDye 680LT and ABY-029 fluorescence and the concentrations are described by the Eq. (2) and Eq. (3) (Fig. 2(g)-(h), Table S1),
3.1.2 Calibration curves of tissue homogenates
To quantitatively characterize the relationship between MFI and imaging agent concentration in each type of tissue, calibration curves were established separately. The log10 transformed fitted curves were plotted, separated by type of imaging agent and HF in Fig. 3. Tissue-specific calibration curves of ABY-029 and IRDye 680LT were co-plotted in Fig. S2. Regression parameters were summarized in Table S2. Simple linear regressions for all tissues measured yielded R2 ranging from 0.9575 to 0.9973 (mean 0.98 ${\pm} $ 0.01) for ABY-029 and 0.9810 to 0.9948 (mean 0.990 ${\pm} $ 0.005) for IRDye 680LT. RMSE ranged from 0.07004 to 0.2018 (mean 0.13 ${\pm} $ 0.05) for ABY-029 and 0.08353 to 0.1700 (0.12 ${\pm} $ 0.03) for IRDye 680LT. The overall mean R2 and RMSE for all tissues and agents were 0.99 ${\pm} $ 0.01 and 0.13${\pm} $ 0.04, respectively.
Fig. 3. Calibration curves from simple linear regression of log10 transformed MFI on log10 transformed imaging agent concentration. (a) Calibration curves of ABY-029 for tissues with HF = 3, (b) calibration curves of ABY-029 for tissues with HF = 6, (c) calibration curves of IRDye 680LT for tissues with HF = 3 and (d) calibration curves of IRDye 680LT for tissues with HF = 6.
To compare optical properties of homogenates from different types of tissues, an F-test comparing linear regression slopes was performed, among tissues homogenized with HF of 3 or 6, and in 700 and 800 nm channels, respectively. The F-test same as blood and plasma phantom comparison showed statistically significant differences in regression slopes for ABY-029 among tissues homogenized with HF = 3 (brain, heart, intestine, kidney, liver, lung, spleen, tumor) and HF = 6 (skin, muscle), and for IRDye 680LT among tissues homogenized with HF = 3 (Fig. 3 a-c). It is interesting to note that neither slopes nor intercepts of IRDye 680LT calibration curves for skin or muscle were statistically significant (Fig. 3(d)).
3.1.3 LLOQ
The LLOQ for each sample type and calibration curve are summarized in Table 2. In general, the LLOQ for ABY-029 was lower than that of IRDye 680LT, except in brain and tumor. This outcome was especially notable in the wide-field LLOQ for all blood and plasma samples, where the LLOQ for ABY-029 was 1-2 orders of magnitude less than that found for IRDye 680LT. Of the ABY-029 blood and plasma samples, the larger volume specimens had the smallest LLOQ, followed by the capillary samples, and the plasma measured on the fluorometer had the highest values. The IRDye 680LT LLOQs were virtually the same for all wide-field measured blood and plasma samples, while the fluorometer measured samples were substantially higher. When considering the homogenized tissues, intestine, kidney, liver, and spleen all had the largest LLOQs for both ABY-029 and IRDye 680LT.
Table 2. Summary of LLOQ for each sample type and calibration curve.
3.2 Quantifying blood and tissue distribution of paired agents
3.2.1 In vivo blood pharmacokinetics of paired agents
The pharmacokinetic clearance of the paired agents over time was determined by measuring the whole blood fluorescence collected in capillary tubes. The calibration curves determined in the phantom study [Eq. (2) & (3)] were used to convert the MFI values to concentrations. The concentration over time for IRDye 680LT and ABY-029 are shown in Fig. 4(a) and 4(b) respectively. A bi-exponential decay curve [Eq. (4)] was fit to the data and are described by Eq. (5) and Eq. (6) for IRDye 680LT and ABY-029, respectively.
Fig. 4. In vivo blood PK for IRDye 680LT and ABY-029. The bi-exponential decay fits of IRDye680LT (a) and ABY-029 (b) in a live mouse over an 8-hour time-period. The mean fluorescence intensity variation of IRDye 680LT (c) and ABY-029 (d) in individual mouse arterial and venous blood over a 4-hour time-period.
While performing the in vivo plasma excretion study, it was noticed that there were variations in the color of the blood that was likely due to variation in oxygenation. To account for this variation, the difference in MFI between arterial and venous blood over a four-hour period was examined. The results show the expected decrease in fluorescence over time, and a repeated measures ANOVA test showed that there is no significant difference in MFI between the two blood types for imaging agent (IRDye 680LT: p = 0.4959; ABY-029: p = 0.9065). Note that at no time, did either of the ABY-029 or IRDye 680LT concentrations fall below the determined LLOQ.
3.2.2 In vivo tissue biodistribution of paired agents
Tissue-specific concentrations of ABY-029 and IRDye 680LT were measured 5 h after intravenous injection of both agents, using the same protocol for tissue homogenization, capillary tube-based sample loading and imaging. Quantification of tissue homogenate MFIs in both 700 and 800 nm channel are shown in Fig. 5 (a & b). A numerical summary of the data is shown in Table S3. Strong signals from both ABY-029 and IRDye 680LT were observed in kidney and liver. IRDye 680LT MFI in the kidney was notably higher than other tissue types; while ABY-029 MFI in the kidney was also higher than other tissues, the divergence was less significant. Lowest MFI for both imaging agents were detected from brain homogenate, followed by muscle and heart. Another notable difference in inter-tissue fluorescence tendency between the two agents lies in the tumor tissue, as being the third highest tissue for ABY-029 fluorescence while tumor MFI for IRDye 680LT was ranked sixth among all ten types of tissue.
Fig. 5. Tissue BD of ABY-029 and IRDye 680LT 5 h after simultaneous injection in mouse xenograft. Tissue homogenates were measured to acquire MFI of (a) ABY-029 and (b) IRDye 680LT. Corresponding tissue-specific concentrations of (c) ABY-029 and (d) IRDye 680LT were extrapolated from the calibration curves. Both MFI and concentration values were plotted in log10 scale for better visualization. Dot: individual value, bar: mean value.
Tissue BD of ABY-029 and IRDye 680LT were characterized by extrapolating tissue concentration using the MFI and corresponding calibration curve. Tissue-specific concentrations were summarized in Table S3 and plotted in Fig. 5 (c & d). The trend in tissue-specific values were consistent between MFI and concentration plots for both imaging agents with minor alterations in tissue concentration ranking between fourth and ninth highest for IRDye 680LT. The relative consistency led to relatively similar shape in both plots. Consistent with MFI signals, kidney, liver, and tumor remained the three tissues with highest ABY-029 concentrations, namely 33 ± 2, 20 ± 8 and 1.5 ± 0.3 nM; and brain, muscle and heart were the three tissues with least ABY-029 distribution. IRDye 680LT also exhibited similar trend in tissue distribution as MFI plots, with brain, muscle and heart being the lowest in concentration and kidney, liver and skin having the highest retention (16 ± 5, 0.9 ± 0.3 and 0.4 ± 0.2 nM, respectively). In three cases, the recovered concentration was less than the determined LLOQ: brain for ABY-029, and brain and intestine for IRDye 680LT.
4. Discussion
We proposed that a robust method to collect PK and BD data would be to quantify MFI from whole blood and whole homogenized tissues, respectively, using wide-field imaging. Wide-field imaging is one of the fastest ways to collect fluorescence intensity from samples, but variations in fluorescence signal from the concentration of the fluorophore, optical properties of the tissues, fluorophore-tissue interactions, and sample volume are difficult to separate. Therefore, we have tested loading whole blood or tissue homogenate in uniform glass capillaries to standardize the sample volume and imaging pathlength, while calibration curves created for each unique blood or tissue medium account for differing optical properties and fluorophore-tissue interactions. Our method, summarized in Fig. 1, can be utilized with any wide-field fluorescence imaging system with the appropriate optics. The Pearl Impulse was used here to demonstrate the workflow, but any system could be inserted provided system-specific calibration curves are applied.
Our proposed method of imaging whole blood in capillary tubes was found to be both quantitative and efficient. This was demonstrated by comparing the MFI of whole blood and plasma in large-volume tubes versus capillary tubes using wide-field imaging to the area under the curve and peak intensity measurements collected from plasma using a fluorometer. It was determined that the wide-field imaging capillary methodology volumes, had higher R2 and lower RMSE than all other methods tested, including fluorometry of plasma. All wide-field values of blood and plasma had lower LLOQ than the fluorometer plasma samples, although the large volume samples were lower than their capillary counterparts. Taken together, these results demonstrate that wide-field imaging is more sensitive than fluorometry. Considering the capillary method can be used to acquire blood samples directly from the mouse without having to centrifuge and isolate blood plasma, which can lead to the lysing of red blood cells that could alter the autofluorescence of plasma, cause variations in the plasma fluorescence, and reduce the risk of losing some of the fluorophore in the red blood cell fraction [29]. Additionally, measuring MFI directly from the capillary used to directly collect the blood is fast, thus saving time when multiple time points and animals are being considered, and requires a low volume of blood (∼ 50 $\mu $L), which maximizes the animal health and integrity of the plasma curve. We have previously published methods of obtaining PK plasma clearance from live mice by imaging the exposed carotid artery, which may be argued is the most direct way of obtaining plasma excretion curves, but this methodology is technically challenging and requires optical isolation of the artery to avoid tissue-cross talk [33].
The excretion rates of IRDye 680LT and ABY-029 we determined using the whole blood capillary method closely match previous studies. We determined excretion rates of 0.22 min-1 and 0.0042 min-1 for IRDye 680LT and 0.23 min-1 and 0.0026 min-1 for ABY-029. Several studies from our group examined plasma excretion rates for ABY-029, and a similar (but not identical) anti-EGFR Affibody Imaging Agent (Affibody AB) conjugated to IRDye 800CW, and epidermal growth factor (EGF) and are summarized in Table S4 [14,27,33,34]. Although these agents are not structurally identical to ABY-029, the molecular size and properties should be similar. It can be noted that the fast kinetic rate constant (tissue distribution) is variable between all these agents and ABY-029 but is on the same order of magnitude. Comparatively, the slow elimination rate constants are stable between the studies. No other studies have reported the plasma excretion rate constants for IRDye 680LT but several have reported them for IRDye 700DX, which are consistent with what we found here. Overall, the excretion rates calculated for this work most closely match the rates determined by Samkoe et al. [14], which were also reported by Elliott et al. using the blood sampling method [33], but were re-fit using the curve fitting toolbox in MATLAB (Eq. (4)) [14]. The largest difference observed was in the fast excretion rate for ABY-029 in mice, determined here, and for rats [27]. Therefore, the small differences in the excretion rates could be species-dependent or could also be attributed to the fitting equation used. Overall, the whole blood capillary method was demonstrated to be a robust method that was comparable to previously reported values. Additionally, we demonstrated that there was no difference in imaging agent concentration in blood collected from the tail arteries or veins, which increases the number of available collection points and decreases errors from using the incorrect blood vessel.
Consistent with PK, the BD of the agent in tissue remains an important theme of drug and imaging agent characterization [1–3]. In this study, we combined conventional ex vivo tissue homogenization and calibration curve extrapolation with fluorescence quantification for tissue BD characterization. Similar to Iaboni et al. (2023), statistically significant differences in the calibration curve slopes were observed between different tissue types [24]. This result re-states the necessity of establishing specific calibration curve for each type of tissue. Additionally, the LLOQs were similar between the 96-well plate method (0.5 nM) and our capillary method (0.3 nM) for 800 nm fluorescence emission. Quantitative characterization of tissue-specific BD of ABY-029 and IRDye 680LT showed a similar trend between MFI levels and corresponding tissue concentrations. Major differences in tissue MFI and concentration between the two imaging agents were in the tumor, which was ranked third in both fluorescence and agent concentration for ABY-029 while being ranked lower for IRDye 680LT. The difference is likely due to the overexpression of EGFR in head and neck cancer cell line FaDu, which would retain ABY-029 longer due to high levels of receptor binding. Comparatively, IRDye 680LT has no affinity for EGFR and would clear more rapidly, as we have extensively demonstrated in our pre-clinical paired agent imaging studies [9,11,35].
The effects of optical properties and non-standard sample volumes on relative MFI collected in the Pearl for ABY-029 are illustrated by comparing our single-acute toxicity study using whole rat organs [27] with the mouse organs homogenized in capillary tubes collected here. Both studies demonstrate the same pattern of MFI but the whole rat organ MFIs varied over a single order of magnitude for normal tissues, while the same tissues for mice varied over nearly three orders of magnitude (Fig. S3). However, the recovered signals of both agents for brain and IRDye 680LT for intestine were below the determined LLOQ. A much higher MFI is particularly notable for liver and kidney, which is likely subject to non-linear effects due to high concentrations of ABY-029 in these organs and the lack of dilution, which was possible with homogenization. However, the brain is substantially lower in this study and is just below our calculated tissue-specific LLOQ. This could be due to ABY-029 being too large to cross the blood-brain-barrier, or simply that it had cleared from this tissue at the 5-hour measurement point. Further study would be required to confirm this.
Overall, we proposed a method to measure concentrations of fluorescent agents from blood and tissue samples, allowing for quantitative PK and BD profiling. This method would be an addition to commonly used HPLC and LC-MS with a special focus on fluorescent agent PK/BD characterization. The usage of capillary tube sampling may minimize the variation in sample thickness, compared to sample loading onto multi-well plates; and allow complete illumination of the sample during imaging. However, additional studies will be required to validate the technique for other imaging agents utilizing fluorophores with visible wavelength emission. Additionally, the PK study demonstrated the advantages of whole blood characterization over more complicated measurement on plasma, while tissue homogenate fluorescence results re-emphasized the necessity of tissue-specific calibration curves for BD characterization.
5. Conclusion
In the methods presented here for both BD and PK assessment, employment of wide-field fluorescent imaging of capillary tubes standardizes the size of the sample with uniform tube diameter, allowing for precise determination of linear calibration curves (overall mean R2 = 0.99 ${\pm} $ 0.01 and RMSE = 0.12 ${\pm} $ 0.04) over 4-5 orders of magnitude and LLOQs for ABY-029 < 0.3 nM and IRDye 680LT < 0.4 nM. Despite the drawback of intensive homogenization and serial dilution process done for each type of tissue in the BD study, tissue- and blood-specific calibration curves avoids inaccuracies in fluorescence concentration recovered due to divergence in tissue optical properties. Therefore, this method allows for precise quantification of tissue BD and may have the potential of application in future preclinical and clinical characterization for fluorescence imaging agents. In addition to clinical translation, kinetic computational modeling of fluorescent imaging agents, for both pharmacological and imaging applications such as paired-agent imaging, could greatly benefit from quantified values of tissue concentrations over time.
Funding
National Cancer Institute (R01 CA167413, R37 CA212187).
Acknowledgements
We kindly thank Brian Pogue for the use of the VWR 200 Homogenizer for this study.
Disclosures
ABY-029 was produced under a National Cancer Institute funded academic-industrial partnership between Dartmouth College, Affibody Medical AB, and LI-COR Biosciences, Inc.
Data availability
Data underlying the results presented in this paper are available upon request from the corresponding author.
Supplemental document
See Supplement 1 for supporting content.
References
1. K. Ball, G. Bruin, E. Escandón, et al., “Characterizing the pharmacokinetics and biodistribution of therapeutic proteins: an industry white paper,” Drug Metabolism and Disposition 50(6), 858–866 (2022). [CrossRef]
2. B. Silva Lima and M. A. Videira, “Toxicology and biodistribution: the clinical value of animal biodistribution studies,” Mol. Ther. Methods Clin. Dev. 8, 183–197 (2018). [CrossRef]
3. Y. Kamiyama, Y. Naritomi, Y. Moriya, et al., “Biodistribution studies for cell therapy products: current status and issues,” Regener. Ther. 18, 202–216 (2021). [CrossRef]
4. F. A. Jaffer and R. Weissleder, “Molecular imaging in the clinical arena,” JAMA 293(7), 855–862 (2005). [CrossRef]
5. S. E. L. Craig, J. Wright, A. E. Sloan, et al., “Fluorescent-guided surgical resection of glioma with targeted molecular imaging agents: a literature review,” World Neurosurg 90, 154–163 (2016). [CrossRef]
6. J. T. Senders, I. S. Muskens, R. Schnoor, et al., “Agents for fluorescence-guided glioma surgery: a systematic review of preclinical and clinical results,” Acta Neurochir. 159(1), 151–167 (2017). [CrossRef]
7. J. L. Tanyi, L. M. Randall, S. K. Chambers, et al., “A phase III study of pafolacianine injection (OTL38) for intraoperative imaging of folate receptor–positive ovarian cancer (Study 006),” J. Clin. Oncol. 41(2), 276–284 (2023). [CrossRef]
8. M. A. Pysz, S. S. Gambhir, and J. K. Willmann, “Molecular imaging: current status and emerging strategies,” Clin. Radiol. 65(7), 500–516 (2010). [CrossRef]
9. V. C. Torres, S. Hodge, J. J. Levy, et al., “Paired-agent imaging as a rapid en face margin screening method in Mohs micrographic surgery,” Front Oncol 13, 1 (2023). [CrossRef]
10. K. M. Tichauer, N. Sadeghipour, Y. W. Wang, et al., “Quantitative drug target imaging using paired-agent principles,” in Quantitative Drug Target Imaging Using Paired-Agent Principles (Springer, 2021), pp. 275–304.
11. C. Wang, X. Xu, S. Hodge, et al., “Identification of a suitable untargeted agent for the clinical translation of aby-029 paired-agent imaging in fluorescence-guided surgery,” Mol. Imaging Biol. 25(1), 97–109 (2023). [CrossRef]
12. C. Wang, X. Xu, M. Folaron, et al., “Improved discrimination of tumors with low and heterogeneous EGFR expression in fluorescence-guided surgery through paired-agent protocols,” Mol. Imaging Biol. 25(1), 110–121 (2023). [CrossRef]
13. K. M. Tichauer, M. Diop, J. T. Elliott, et al., “Accounting for pharmacokinetic differences in dual-tracer receptor density imaging,” Phys. Med. Biol. 59(10), 2341–2351 (2014). [CrossRef]
14. K. S. Samkoe, K. Sexton, K. M. Tichauer, et al., “High vascular delivery of EGF, but low receptor binding rate is observed in AsPC-1 tumors as compared to normal pancreas,” Mol. Imaging Biol. 14(4), 472–479 (2012). [CrossRef]
15. K. M. Tichauer, K. S. Samkoe, K. J. Sexton, et al., “In vivo quantification of tumor receptor binding potential with dual-reporter molecular imaging,” Mol. Imaging Biol. 14(5), 584–592 (2012). [CrossRef]
16. K. M. Tichauer, K. S. Samkoe, K. J. Sexton, et al., “Improved tumor contrast achieved by single time point dual-reporter fluorescence imaging,” J. Biomed. Opt. 17(6), 066001 (2012). [CrossRef]
17. K. M. Tichauer, K. S. Samkoe, W. S. Klubben, et al., “Advantages of a dual-tracer model over reference tissue models for binding potential measurement in tumors,” Phys Med Biol 57(20), 6647–6659 (2012). [CrossRef]
18. B. Meng, M. R. Folaron, R. R. Strawbridge, et al., “Noninvasive quantification of target availability during therapy using paired-agent fluorescence tomography,” Theranostics 10(24), 11230–11243 (2020). [CrossRef]
19. H. Ding and F. Wu, “Image guided biodistribution and pharmacokinetic studies of theranostics,” Theranostics 2(11), 1040–1053 (2012). [CrossRef]
20. M. V. Marshall, D. Draney, E. M. Sevick-Muraca, et al., “Single-dose intravenous toxicity study of IRDye 800CW in Sprague-Dawley rats,” Mol. Imaging Biol. 12(6), 583–594 (2010). [CrossRef]
21. M. J. Whitley, D. M. Cardona, A. L. Lazarides, et al., “A mouse-human phase 1 co-clinical trial of a protease-activated fluorescent probe for imaging cancer,” Sci. Transl. Med. 8(320), 1 (2016). [CrossRef]
22. C. G. Patil, D. G. Walker, D. M. Miller, et al., “Phase 1 safety, pharmacokinetics, and fluorescence imaging study of tozuleristide (BLZ-100) in adults with newly diagnosed or recurrent gliomas,” Neurosurgery 85(4), E641–E649 (2019). [CrossRef]
23. H. Gao and J. Williams, “Tissue sample preparation in LC-MS bioanalysis,” in Sample Preparation in LC-MS Bioanalysis (Wiley, 2019), pp. 201–216.
24. M. Iaboni, A. Coppo, D. Remotti, et al., “Fluorescence-based absolute quantification of near-infrared probes in tissue extracts for biodistribution analyses,” Anal. Biochem. 677, 115251 (2023). [CrossRef]
25. S. M. Mahalingam, S. A. Kularatne, C. H. Myers, et al., “Evaluation of novel tumor-targeted near-infrared probe for fluorescence-guided surgery of cancer,” J. Med. Chem. 61(21), 9637–9646 (2018). [CrossRef]
26. K. R. Zinn, M. Korb, S. Samuel, et al., “IND-directed safety and biodistribution study of intravenously injected cetuximab-IRDye800 in cynomolgus macaques,” Mol. Imaging Biol. 17(1), 49–57 (2015). [CrossRef]
27. K. S. Samkoe, J. R. Gunn, K. Marra, et al., “Toxicity and pharmacokinetic profile for single-dose injection of ABY-029: a fluorescent Anti-EGFR synthetic affibody molecule for human use,” Mol. Imaging Biol. 19(4), 512–521 (2017). [CrossRef]
28. S. Reagan-Shaw, M. Nihal, and N. Ahmad, “Dose translation from animal to human studies revisited,” The FASEB Journal 22(3), 659–661 (2008). [CrossRef]
29. S. Pannem, C. Wang, D. Ravi, et al., “Deriving clinical parameters for paired agent imaging in head and neck squamous cell carcinomas: efficient methodology for the measurement of plasma fluorescence,” Proc. SPIE12361, 1236107 (2023).
30. K. S. Samkoe, K. Sexton, K. Tichauer, et al., “Determination of blood plasma fluorescence extinction coefficients for dyes used in three-compartment binding model,” Proc. SPIE7886, 78860A (2011).
31. B. W. Pogue, X. Zhou, S. Gibbs, et al., “Comparison of techniques for quantification of fluorescence from tissue,” Proc. SPIE6139, 61390N (2006).
32. S. Oliveira, R. Cohen, M. S. van Walsum, et al., “A novel method to quantify IRDye800CW fluorescent antibody probes ex vivo in tissue distribution studies,” EJNMMI Res. 2(1), 1–9 (2012). [CrossRef]
33. J. T. Elliott, K. M. Tichauer, K. S. Samkoe, et al., “Direct characterization of arterial input functions by fluorescence imaging of exposed carotid artery to facilitate kinetic analysis,” Mol. Imaging Biol. 16(4), 488–494 (2014). [CrossRef]
34. K. Sexton, K. Tichauer, K. S. Samkoe, et al., “Fluorescent affibody peptide penetration in glioma margin is superior to full antibody,” PLoS One 8(4), e60390 (2013). [CrossRef]
35. C. Wang, S. Hodge, D. Ravi, et al., “Rapid and quantitative intraoperative pathology-assisted surgery by paired-agent imaging-derived confidence map,” Mol. Imaging Biol. 25(1), 190–202 (2023). [CrossRef]
