In vivo Localization of ⁹⁰Y- and ¹⁷⁷Lu-Radioimmunoconjugates Using Cerenkov Luminescence Imaging in a Disseminated Murine Leukemia Model

Abstract

Cerenkov radiation generated by positron-emitting radionuclides can be exploited for a molecular imaging technique known as Cerenkov Luminescence Imaging (CLI). Data have been limited, however, on the use of medium-to-high energy β-emitting radionuclides of interest for cancer imaging and treatment. We assessed the use of CLI as an adjunct to determine localization of radioimmunoconjugates to hematolymphoid tissues. Radiolabeled ¹⁷⁷Lu- or ⁹⁰Y-anti-CD45 antibody (Ab; DOTA-30F11) was administered by tail vein injection to athymic mice bearing disseminated murine myeloid leukemia with CLI images acquired at times afterward. Gamma counting of individual organs showed preferential uptake in CD45+ tissues with significant retention of radiolabeled Ab in sites of leukemia (spleen and bone marrow). This result was confirmed in CLI images with 1.35 × 105 ± 2.2 × 104 p/sec/cm2/sr and 3.45 × 103 ± 7.0 × 102 p/sec/cm2/sr for ⁹⁰Y-DOTA-30F11 and ¹⁷⁷Lu-DOTA-30F11, respectively, compared to undetectable signal for both radionuclides using the non-binding control Ab. Results showed that CLI allows for in vivo visualization of localized β-emissions. Pixel intensity variability resulted from differences in absorbed doses of the associated energies of the β-emitting radionuclide. Overall, our findings offer a preclinical proof of concept for the use of CLI techniques in tandem with currently available clinical diagnostic tools.

INTRODUCTION

The Cerenkov effect is the release of a continuum of ultraviolet and visible light energy resulting from the interaction of particulate emissions from radioactive decay within a di-electric medium (1, 2). The resulting luminescence or radiation is observed when charged particles, such as a β⁻, β⁺, or α, travel through water with a velocity that exceeds the speed of light. Loss of kinetic energy and deceleration occur by polarization of the water molecules which, in turn, relax back to equilibrium through the emission of light (3–6). The well characterized relationship between the Cerenkov effect and the speed of light acts as a threshold velocity and is largely dependent on the medium with which the charged particles interact; as well as a physical property known as the refractive index which accounts for an approximate 25% reduction of the speed of light through water allowing for the observation of emitted Cerenkov radiation. In water, the threshold velocity can be achieved with β⁻ or β⁺ energies of only a few hundred keV; while the mass and mean ionization path-length of an α-particle would require several thousand MeV to reach the same threshold velocity, making it an improbable observation (7, 8).

Early work conducted by Elrick et.al. indicated the possibility of a Cerenkov luminescence imaging (CLI) application for living biological samples; however, experimental limitations were attributed to the resolution capabilities of the imaging and detection equipment available at the time (9). Historically, optical methods of imaging have been relegated to pre-clinical research. This has been due in part to the inherent limitations of available techniques at the human scale such as high-rates of light scattering and poor tissue penetration, both of which increase the difficulty in quantifying collected data suitable for clinical applications (6). In 2009 Robertson and colleagues detailed a method for the in vivo imaging of Cerenkov radiation utilizing ¹⁸F (FDG) in conjunction with a commercially available imaging platform and relevant software (4). Throughput of the technique was shown to be relatively high and allowed for clear visualization of tumor xenografts with image acquisition on the order of seconds to minutes. Since that time, CLI has become increasingly well known as a particulate imaging technique for both β⁺ and β⁻ emitting radionuclides (10–16). Given the current lack of FDA approved theranostic radionuclides (those select few that can serve as therapeutic agents whilst providing an imageable photon), the vast majority of CLI literature has focused more on β⁺ emitting radionuclides that allow for comparisons of the collected data to concurrently run Positron Emission Tomography (PET) imaging studies.

As other investigators previously noted, within the disparity between clinically approved theranostic radionuclides an opportunity exists to exploit CLI as a preclinical imaging approach for real-time monitoring of in vivo radionuclide localization without the need for surrogate isotopes or adjunct imaging such as PET (3, 4, 6, 15, 17). In this report we have assessed the feasibility and potential role of CLI in therapy based studies using medium-to-high energy β-emitters (⁹⁰Y and ¹⁷⁷Lu) in a clinically relevant model of disseminated acute myeloid leukemia (AML). Reported herein are the in vitro phantom and in vivo imaging studies to assess CLI model applicability. Therapeutic feasibility assessments were made by investigating the use of CLI as an adjunct to biodistribution to determine tissue localization of an anti-CD45 radioimmunotherapeutic agent.

MATERIALS AND METHODS

Mice

Female SJLB6F1/J and SJL/J mice, 8–12 weeks old, were purchased from Jackson Laboratories (Bar Harbor, ME); female athymic mice, 8–12 weeks old, were purchased from Harlan Laboratories (Livermore, CA). All mice were housed at the Fred Hutchinson Cancer Research Center (FHCRC) in a pathogen-free environment under protocols approved by the FHCRC Institutional Animal Care and Use Committee. Mice were placed on alfalfa-free irradiated chow (Animal Specialties, Richmond, IN) at least 4 days before imaging to prevent non-specific signal.

Cell lines, antibodies, and production and labeling of DOTA-Ab

Murine myeloid SJL leukemia cells were obtained and maintained as described previously (18). Leukemia was established in study mice as previously described (19–21). Polyclonal rat IgG antibody (negative control) was purchased from Sigma Aldrich (St Louis, MO). The rat IgG_2b anti-murine CD45 Ab 30F11 was purified as previously described (18). DOTA-Ab conjugates were produced as described previously (22). DOTA-Ab was labeled with ⁹⁰Y or ¹⁷⁷Lu from Perkin Elmer Life Sciences (Waltham, MA) under metal-free conditions using a process of radiometal chelation as previously described (22, 23). Labeling efficiencies were greater than 90% as determined by thin-layer chromatography and radiolabeled DOTA-Ab was purified via size exclusion chromatography employing a PD10 column as described previously (22, 23).

Biodistribution Studies

Groups of 5 mice were injected intravenously with 1 × 10⁵ SJL leukemia cells. Two days after injection mice were given 100 μg (0.67 nmol) of DOTA-30F11 or DOTA-rat IgG labeled with 100 μCi of either ⁹⁰Y or ¹⁷⁷Lu via tail-vein injection. Mice were euthanized at 4, 24, 48, and 72 hours post-injection for resection of tissues, followed by gamma counting using a Packard Cobra counter (Packard Instrument Company, Meriden, CT). Correction was made for radioactive decay and counts were used to determine the percentage of injected dose per gram of tissue (% ID/g).

Blood Clearance Studies

Groups of 5 mice without disease were co-injected with 100 μg (0.67 nmol) of 30F11 labeled with 10 μCi ¹²⁵I and Rat IgG labeled with 10 μCi ¹³¹I. At T = 5, 15, 30, and 45 minutes, and 1, 2, 20, 24, and 48 hours post-injection, 10 μL of whole blood was removed from the retro-orbital plexus of each mouse utilizing a calibrated pipet (VWR, Radnor, PA). This was followed by gamma counting on a Packard Cobra counter (Packard Instrument Company, Meriden, CT). Values were corrected for radioactive decay and the percentage of the injected dose per gram (%ID/g) of blood was calculated. The % ID/g in the blood was analyzed via a non-linear regression curve fit for exponential two-phase decay utilizing GraphPad Prism 6 software (Graph Pad, La Jolla, CA).

Dosimetry

Radioactivity was measured in organ and tissue samples obtained from euthanized mice. The ⁹⁰Y or ¹⁷⁷Lu uptake, retention, and clearance patterns were determined for each tissue by constructing time-activity curves for groups of mice, and fit to mathematical functions. These biodistribution data were then integrated, and standard medical internal radiation dose methods (24) for localized beta dosimetry were applied, together with the appropriate nuclear transmission decay data, to calculate radiation absorbed doses for blood and each organ as described previously (25, 26). These methods account for the specific organ self-dose absorbed fractions as well as the beta-particle cross-organ dose contributions in the small organs of the animal.

Phantom and Imaging Studies

For phantom studies quantities of 10, 20, 40, or 100 μCi of ⁹⁰Y or ¹⁷⁷Lu were placed into clear, colorless 1.5 mL polypropylene tubes. These tubes were subsequently placed into a 12 × 75 mm polypropylene culture tube that was then placed into a plastic beaker of distilled H₂O such that the activity source was surrounded by liquid for image collection (Supplemental Figure 1). For CLI imaging studies, groups of 5 SJL/J, SJLB6F1/J, or athymic mice were injected via tail-vein with 100 μg (0.67 nmol) of DOTA-30F11 or DOTA-rat IgG labeled with 100 μCi or 300 μCi of ⁹⁰Y or ¹⁷⁷Lu. Animals were anesthetized using 2.5% isofluorane in O₂ and imaged individually on a Xenogen IVIS Spectrum (Caliper Life Sciences; Hopkinton, MA) system using Living Image software version 4.2 (Caliper Life Sciences, Hopkinton, MA). All images of phantoms, mice, and murine tissues were acquired as photons/sec with an exposure time of 90 seconds, an f-stop of 1, a binning setting of “medium” (corresponding to a binning of 4), at a 6.5 cm Field of View (FOV) with 2.5x magnification (corresponding to a FOV setting of “B”) on the fluorescent image detection channel with the excitation lamp turned off (6). Images were corrected and normalized for each comparison figure using default settings for cosmic rays, flat field correction, and background subtraction. Optical data were overlaid with a photograph acquired with a binning of 2, an f-stop of 8, a 0.2-second exposure, and the same corrections applied as for optical data. Optical data are reported as the maximum pixel values for each region with units of photons/second/square centimeters/steridian (p/sec/cm²/sr).

RESULTS

Phantoms and Initial CLI Assessment

Initial water phantom studies conducted in polypropylene tubes allowed for an assessment of signal detection in an anatomically unrestricted in vitro environment. The ⁹⁰Y and ¹⁷⁷Lu radionuclides were each tested using 10, 20, 40, and 100 μCi (0.37, 0.74, 1.48, and 3.7 MBq respectively). Measured radiance values increased from 7.48 × 10⁶ to 7.42 × 10⁷ p/sec/cm³/sr for ⁹⁰Y, and 7.65 × 10⁴ to 4.25 × 10⁵ p/sec/cm³/sr for ¹⁷⁷Lu, with the tested quantity of activity irrespective of the radionuclide (Fig. 1A, Supplemental Fig. 1). Differences in the intensity of the measured radiance between radionuclides was attributed to differences in both decay energy and ionization path length (⁹⁰Y: E_β-avg = 0.933 MeV, maximum path length = 12.0 mm; ¹⁷⁷Lu: E_β-avg = 0.050 to 0.150 MeV, maximum path length = 2.1 mm)(27).

Fig 1. In vitro phantoms at varying levels of added activity as measured by Cerenkov Luminescence and effects of hair on ability to image mice via CLI.

300 μCi/100 μg ⁹⁰Y-DOTA-30F11 given via tail-vein to non-disease bearing mice. A) As expected, in vitro phantoms exhibited clear signal that increased in a dose-dependent manner. Differences seen in signal intensity were likely attributed to differences in both decay energy and path length between the two isotopes. B) Signal was not detected in fur-bearing mice, but significant splenic signal was clearly and measurably detected in athymic nude mice 1 hour after tail-vein injection of radiolabeled DOTA-Ab.

Initial in vivo imaging studies were conducted using 300 μCi (11.1 MBq)/100 μg of ⁹⁰Y-DOTA-30F11 in groups of disease-free female SJL/J, SJLB6F1/J, and athymic mice. High radiance values in organs of hematopoiesis (spleen and bone marrow) were taken as positive visual confirmation of the specific uptake of the radiolabelled Ab. No radiance was detected in images among SJL/J and SJLB6F1/J mice, while the spleens from the hairless athymic mice (Fig. 1B) displayed radiance values consistent with specific uptake in that organ (1.6 × 10⁶ ± 3.2 × 10⁵ p/sec/cm³/sr). Previously published work has indicated that the presence of fur (or hair) on a given surface results in the blurring and scattering of light rays that negatively impact the resulting CLI image quality (28). In order to circumvent these optical properties athymic nude mice were exclusively used in the remaining studies.

Biodistribution and Blood Clearance Studies

Biodistribution studies were conducted using 100 μCi (3.7 MBq) of ⁹⁰Y or ¹⁷⁷Lu as a radioactive label on 100 μg for a specific activity of 1 μCi/μg (0.037 MBq/μg) of DOTA-30F11 or DOTA-rat IgG (Fig. 2A–2D). A minimal, yet immediate, increase in the circulating concentration of either DOTA-30F11 conjugate was observed in the blood. At 4 hours after injection the values for ⁹⁰Y-DOTA-30F11 and ¹⁷⁷Lu-DOTA-30F11 were 7.45 ± 2.8 and 7.62 ± 0.7% ID/g, respectively. In both cases, radiolabeled Ab rapidly cleared the blood and had circulating concentrations approaching zero (0.63 ± 0.2 and 1.25 ± 0.7% ID/g for ⁹⁰Y-DOTA-30F11 and ¹⁷⁷Lu-DOTA-30F11, respectively) 24 hours after radiolabeled Ab injection (Fig. 2A and Fig. 2B).

Fig. 2. Biodistribution of radioactivity in leukemic athymic mice.

Labeled DOTA-30F11 conjugates demonstrate specific binding in CD45-positive tissues and sites of hematopoiesis. Bone marrow and spleen uptake peaked at: A) 8.45 ± 1.0 % and 129 ± 18% ID/g for ⁹⁰Y-DOTA-30F11, respectively, and B) 22.2 ± 8.9 % and 237 ± 50% ID/g for ¹⁷⁷Lu-DOTA-30F11, respectively. As expected, negative control Ab (DOTA-rat IgG) showed minimal non-specific binding with either ⁹⁰Y (C) or ¹⁷⁷Lu (D).

CD45⁺ hematopoietic tissues displayed specific uptake and targeting as expected with the use of an anti-CD45 Ab. Specifically, splenic uptake for ⁹⁰Y-DOTA-30F11 at 4 hours was 48.9 ± 3.5 % ID/g and remained elevated, with a value of 129 ± 18% ID/g at 72 hours. Similarly, the values for ¹⁷⁷Lu-DOTA-30F11 were found to be 102 ± 10% ID/g at 4 hours and remained elevated at 72 hours (237 ± 50% ID/g). Bone marrow (BM) uptake, assessed via 1 mL normal saline femur flushes, for ⁹⁰Y-DOTA-30F11 increased to a maximum of 8.45 ± 1.0% ID/g at 24 hours and decreased to 5.37 ± 2.2% ID/g at 72 hours. BM uptake for the ¹⁷⁷Lu-DOTA-30F11 cohorts were found to be significantly higher, yet followed the same trend, as uptake increased to a maximum of 22.2 ± 8.9% ID/g at 24 hours and decreased to 15.3 ± 5.0% ID/g at 72 hours. Uptake in the non-specific normal organs was minimal for both the ⁹⁰Y- and ¹⁷⁷Lu-DOTA-30F11 cohorts, with only characteristic transient uptake in the liver (maximum observed values were 6.04 ± 0.7% ID/g for ⁹⁰Y-DOTA-30F11 and 11.0 ± 2.0% ID/g for ¹⁷⁷Lu-DOTA-30F11) and kidneys (maximum observed values were 4.93 ± 0.8% ID/g with ⁹⁰Y-DOTA-30F11 and 6.03 ± 0.7% ID/g for ¹⁷⁷Lu-DOTA-30F11) 4 hours after the injection of the radiolabeled conjugate (Fig. 2A and 2B). All organ uptake in cohorts that received control DOTA-rat IgG Ab-conjugate showed only transient, non-specific uptake in the blood and normal organs, independent of radionuclide used (Fig. 2C and 2D).

Splenic uptake was further analyzed to determine the percentage of the injected dose at each time point for DOTA-30F11 and DOTA-Rat IgG radiolabeled with ⁹⁰Y or ¹⁷⁷Lu. Both ⁹⁰Y- and ¹⁷⁷Lu-DOTA-Rat IgG demonstrated negligible uptake in the spleen, with a peak value of 0.261 ± 0.12% ID at 48 hours for ⁹⁰Y-DOTA-Rat IgG and 0.598 ± 0.15% ID at 4 hours for ¹⁷⁷Lu-Rat IgG. ⁹⁰Y-DOTA-30F11 demonstrated elevated, specific percent injected dose by 4 hours with 4.94 ± 0.90% ID, peaking at 24 hours with a value of 5.59 ± 1.1% ID and remaining elevated at 4.02 ± 0.71% ID at 72 hours post-injection. Similarly, ¹⁷⁷Lu-DOTA-30F11 showed a value of 9.83 ± 0.71% ID at 4 hours, peaked with 11.7 ± 2.4% ID at 24 hours, and remained elevated in the spleen at 9.46 ± 1.5% ID at 72 hours post-injection (Fig. 3A).

Fig. 3. Splenic uptake from biodistribution expressed as (A) the % ID, as well as (B) blood clearance of 30F11 and Rat IgG Abs, and (C) effect of radioactivity on splenic mass over time.

Labeled DOTA-30F11 conjugates displayed specific, high uptake in splenic tissue compared to control. The % ID for DOTA-30F11 was elevated at 4 hours, peaked at 24 hours, and continued to remain stable at 72 hours post-injection. Splenic mass decreased over time in groups given radiolabeled DOTA-30F11, and remained stable in groups given radiolabed DOTA-Rat IgG.

A blood clearance study was performed for both 30F11 and Rat IgG. Mice were co-injected with equal molar doses of ¹²⁵I-30F11 and ¹³¹I-Rat IgG and bled at serial time points. ¹³¹I-Rat IgG had an initial value of 46.5 ± 6.2% ID/g 5 minutes post-injection, with a rapid initial serum clearance phase giving an estimated t_½ of 7.68 minutes. The secondary clearance phase, resulting in a plateau of 13.3 ± 0.93% ID/g at 48 hours post-injection, gave an estimated t_½ of 10.5 hours. ¹²⁵I-30F11 also cleared rapidly from the blood, with a value of 34.0 ± 4.6% ID/g at 5 minutes post-injection, giving an estimated t_½ of 18.8 minutes for the initial serum clearance phase; this went on to plateau in the secondary clearance phase with a value of 6.70 ± 0.24% ID/g at 48 hours post-injection to give a t_½ of 6.93 hours (Fig. 3B).

Parallel CLI

CLI studies were performed in parallel with the biodistribution experiments using the same mice in order to identify any possible correlation to the localization results. Athymic mice were imaged in three positions (dorsal, ventral, and left-lateral) to reduce the possibility of anatomically obscuring a detectable signal. Mice injected with DOTA-30F11 produced images of measurable radiance in the spleen irrespective of radiolabel (⁹⁰Y or ¹⁷⁷Lu), compared to mice injected with DOTA-rat-IgG, which displayed only diffuse abdominal radiance at the earliest time point when using ¹⁷⁷Lu as the radionuclide (Fig. 4A–4D). Splenic radiance was measured in regions of interest on images from mice in the DOTA-30F11 cohorts; CLI images of marrow spaces in the long bones were unobtainable. The cohorts receiving ⁹⁰Y-DOTA-30F11 or ¹⁷⁷Lu-DOTA-30F11 displayed a 10-fold difference in the initial accumulation of activity as measured by the maximum splenic radiance at 4 hours (1.01 × 10⁵ ± 3.4 × 10⁴ p/sec/cm³/sr and 1.12 × 10⁴ ± 1.9 × 10³ p/sec/cm³/sr for ⁹⁰Y-DOTA-30F11 and ¹⁷⁷Lu-DOTA-30F11, respectively). Subsequent measurements of each conjugate over time revealed a gradual decrease in radiance to final measured values of 7.53 × 10⁴ ± 2.5 × 10⁴ p/sec/cm³/sr for ⁹⁰Y-DOTA-30F11 and no detectable signal for ¹⁷⁷Lu-DOTA-30F11 at 72 hours.

Fig. 4. ⁹⁰Y-DOTA-Ab (A, B) and ¹⁷⁷Lu-DOTA-Ab (C, D) Cerenkov Luminescence Imaging in parallel.

Mice bearing AML were imaged at the described time points prior to sacrifice for tissue harvest. Images were corrected for background and normalized to each other. Data shown are images from each group that best show differences in uptake between DOTA-30F11 (A, C) and DOTA-rat IgG (B, D) groups for each isotope.

The CLI images and associated radiance measurements of the cohorts that received control ⁹⁰Y- or ¹⁷⁷Lu-DOTA-rat IgG displayed values consistent with blood pool clearance and washout, and no specific targeting as expected. The initial (4 hour) accumulated activity provided the only observed radiance for either conjugate, 5.70 × 10⁴ ± 9.3 × 10³ p/sec/cm³/sr and 5.42 × 10³ ± 8.7 × 10² p/sec/cm³/sr respectively for ⁹⁰Y-DOTA-30F11 and ¹⁷⁷Lu-DOTA-30F11. Subsequent measurements of either labeled conjugate over time revealed no detectable radiance values on which to base measurements.

Dosimetry Estimates of Absorbed Doses of Radiation from ⁹⁰Y and ¹⁷⁷Lu

Because we desired to correlate in vivo visualization of radionuclide delivery with biodistribution data, we also calculated the radiation absorbed doses to correlate with CLI results. Dosimetry estimates showed that the BM absorbed dose per μCi of ¹⁷⁷Lu -DOTA-30F11 injected was more than 5-fold greater than the absorbed radiation dose delivered by ⁹⁰Y-DOTA-30F11 (6.82 versus 1.33 cGy/μCi, respectively; Table 1). Similarly, the spleen had a radiation absorbed dose that was more than 2.5-fold higher for ¹⁷⁷Lu- than ⁹⁰Y-DOTA-30F11 (158 versus 60.7 cGy/μCi, respectively). All other organs from mice treated with ¹⁷⁷Lu -DOTA-30F11 had similar absorbed dose per unit injected activity compared to mice treated with ⁹⁰Y-DOTA-30F11; the absorbed doses per μCi injected for lung, liver and kidney were 0.60. 3.09, and 1.28 cGy/μCi compared to 0.41, 3.81, and 1.65 cGy/μCi for ¹⁷⁷Lu and ⁹⁰Y, respectively. These differences in dosimetry for targeted organs were not predicted by in vivo CLI, which did not show radioactivity accumulation in organs other than spleen, and actually showed higher maximum splenic radiance at 4 hours for ⁹⁰Y 1.01 × 10⁵ ± 3.4 × 10⁴ p/sec/cm³/sr and 1.12 × 10⁴ ± 1.9 × 10³ p/sec/cm³/sr for ⁹⁰Y-DOTA-30F11 and ¹⁷⁷Lu-DOTA-30F11, respectively), followed by a gradual decrease (7.53 × 10⁴ ± 2.5 × 10⁴ p/sec/cm³/sr for ⁹⁰Y-DOTA-30F11 and no detectable signal for ¹⁷⁷Lu-DOTA-30F11 at 72 hours)(Fig. 4).

Table 1.

Absorbed tissue dose calculated from biodistribution data.^*

Tissue	90Y cGy/uCi	177Lu cGy/uCi
spleen	60.7	158
marrow	1.33	6.82
blood	0.833	0.356
lung	0.405	0.599
liver	3.81	3.09
stomach	0.544	0.373
kidney	1.65	1.28
sm. int.	1.83	1.87
colon	1.26	1.00

^*Tissue dosimetry estimates were calculated via the standard Medical Internal Radiation Dose method, employing tissue gamma counts from biodistribution studies.

Spleen CLI Signal and Conjugate Uptake

Given that the images obtained for either DOTA-30F11 conjugate in the parallel CLI study failed to identify any significant uptake in any tissue other than the spleen it was prudent to determine if the uptake in other organs was masked by either the quantity of splenic uptake or size constraints due to the proximity of that organ to the spleen. This was accomplished by imaging the harvested organs prior to counting tissues during biodistribution. The images were then analyzed in the same manner as previously utilized for whole animal imaging, with radiance measured in regions of interest for organs displaying localization of labeled Ab conjugate. In the cohort that received ⁹⁰Y-DOTA-30F11 (Fig. 5A, Supplemental Fig. 2) CLI data obtained from ex vivo organs revealed initial (4 hour) accumulated hepatic plus small and large intestinal activity (6.71 × 10⁴ ± 1.4 × 10⁴, 9.62 × 10⁴ ± 1.8 × 10⁴, and 9.95 × 10⁴ ± 1.7 × 10⁴ p/sec/cm³/sr, respectively) that quickly dissipated and was immeasurable at 24 hours. Additionally, ex vivo splenic radiance values were obtained and showed an initial accumulation reaching a maximum of 1.35 × 10⁵ ± 2.2 × 10⁴ p/sec/cm³/sr at 24 hours followed by an observed decrease in radiance to 9.71 × 10⁴ ± 1.9 × 10⁴ p/sec/cm³/sr at 72 hours.

Fig. 5. Ex vivo Cerenkov Luminescence Imaging of individual resected organs.

Mice were given 300 μCi/100 μg ⁹⁰Y-DOTA-30F11 or ¹⁷⁷Lu-DOTA-30F11 (A and B, respectively). Corresponding to images collected in Figure 4 (prior to the tissue resection and also to the data shown in Table 1), ⁹⁰Y groups demonstrated a 10-fold higher pixel intensity than mice given ¹⁷⁷Lu.

Conversely, in the cohort that received ¹⁷⁷Lu-DOTA-30F11 (Fig. 5B) CLI images of ex vivo organs revealed a completely different uptake profile. The 4 hour accumulated activity provided an initial splenic radiance value of 3.10 × 10³ ± 4.9 × 10² p/sec/cm³/sr and increased to 4.55 × 10³ ± 7.7 × 10² p/sec/cm³/sr at the final time point of 72 hours. Additionally, uptake in the stomach was measured as early as 48 hours (1.14 × 10⁴ ± 1.6 × 10³ p/sec/cm³/sr) and decreased at 72 hours to 6.91 × 10³ ± 1.2 × 10³ p/sec/cm³/sr when small intestinal uptake was measured at 1.07 × 10³ ± 2.1 × 10² p/sec/cm³/sr.

DISCUSSION

Robertson and colleagues provided in vivo proof of principle and validation of CLI as a promising optical imaging modality in 2009 with ¹⁸F-based studies (4). Since that time several other groups have added to the depth of knowledge in the field mainly utilizing PET radionuclides in xenograft models of disease (4–6, 10–14, 16). Studies presented herein explored the use of CLI as an adjunct technique to a classical biodistribution study in a more clinically relevant disseminated model of AML. In this murine model, as in human AML, the primary sites of disease are the bone marrow and other hematolymphoid organs; however, in the murine model the largest measurable uptake of anti-CD45 Ab was found in splenic tissue as expected. These studies showed the feasibility of following the real-time localization of a radioimmunotherapeutic agent in vivo using medium to high-energy β⁻ emitting radionuclides without sacrificing the animal.

We have shown that initial in vitro phantom studies produced results comparable to those previously reported by other groups (4–6, 10–12, 15, 29–32). Increases in measured radiance positively correlated with increases in the quantity of activity irrespective of the radionuclide measured (⁹⁰Y or ¹⁷⁷Lu; Fig. 1A, Supplemental Fig. 1). Linear regression analyses of the curves generated for each radionuclide indicated semi-logarithmic relationships with r² values of 0.97 and 0.99 for ⁹⁰Y and ¹⁷⁷Lu, respectively. This was expected given that the magnitude of Cerenkov luminescence, while indirectly related to the quantity of radionuclide present, was due to interaction of β-particles with the water medium as opposed to a direct measurement of radionuclide decay. Thus, the observed increases in measured radiance coincide with, but do not directly mirror, the magnitude of the increase in the activity of the phantoms. Additionally, the observed differences in the magnitude of the measured radiance between the radionuclides are related to the approximate 6-fold difference in the emission energy of the β⁻ particles between ⁹⁰Y and ¹⁷⁷Lu.

However, the results of the initial in vivo studies were unexpected. Of the three strains of mice initially studied, only athymic nude mice and not SJL/J or SJLB6F1/J mice were successfully imaged (Fig. 1B). This result was subsequently attributed to presence or absence of fur on the mice and thus the SJL/J or SJLB6F1/J strains were not investigated in subsequent CLI studies. Work by Marschner et al. described the optical properties of hair and application of this work to our studies explained the correlation between the fur and a lack of imageable CLI photons. Briefly, when incident light rays interacted with a smooth cylinder, portions of the light were expected to either reflect or be transmitted through the cylinder. However, two properties of fur resulted in the blurring of each randomly scattered light ray to the point of signal loss prior to detection (Fig. 1B). First, fur is made of overlapping scales of keratin that have a characteristically rough surface causing random light scattering. Second, the overlapping nature of the cuticle scales causes a tilted or tapered surface resulting in a blurring of the light surrounding the hair shaft (28). Consequently, athymic mice were used in our remaining studies to circumvent the interference generated by fur while imaging mice with models of disseminated disease. In future studies it may be appropriate to depilate the area of a solid tumor xenograft in a furred mouse model; likewise, patients undergoing imaging of subdermal masses from an external detector may require depilation prior to the procedure.

Biodistribution studies of the radiolabeled anti-CD45 Ab-conjugate DOTA-30F11 revealed specific targeting of hematopoietic tissues (spleen and BM) with only characteristic transient uptake in normal organs, indicative of the observed rapid blood clearance for both 30F11 and rat IgG irrespective of the radionuclide used (Fig. 2A and 2B; Fig. 3B). As expected, the control DOTA-rat-IgG Ab-conjugate showed no indication of specific targeting or uptake (Fig. 2C and 2D). Apparent increase in splenic uptake over time as measured by the % ID/gm of tissue was confounded by the reduction in mass due to the administered radionuclide dose (Fig. 3C) in mice receiving DOTA-30F11; reduction was not seen in groups receiving DOTA-Rat IgG. To clarify this we further analyzed splenic uptake; % ID for both radionuclides peaked at 24 hours and remained stable through 72 hours post-injection (Fig. 3A). Furthermore, we found that the images obtained by CLI appeared to correlate with the biodistribution results in so far as splenic uptake was clearly depicted irrespective of the radionuclide used, despite differences in the measured radiance between ⁹⁰Y and ¹⁷⁷Lu. However, these images fall short of being able to follow the general trend of the graphically represented biodistribution data. Specifically, the images provided visual confirmation of the initial accumulation of the radiolabeled Ab-conjugate in hematopoietic target tissues as measured by pixel radiance, followed by decreasing image intensity or pixel radiance. A model describing decreasing signal intensity due to isotope decay or biological excretion over time was not corroborated by the biodistribution gamma count data in the spleen using either radionuclide; in fact splenic uptake peaked for both isotopes by 24 hours post-injection and remained consistent through later time points. We attribute this phenomenon to the difference in overall sensitivity of the charge-couple device (CCD) chip, the digital camera or detector in the Xenogen IVIS Spectrum that converts photons (light energy) into a digital signal (electrons) via application of Einstein’s photoelectric effect, as compared to the sensitivity of the Gamma counter. An additional explanation for this observation may relate to the change in splenic mass over time as commonly seen in our biodistribution studies.

In CLI the measured radiance of a region of interest is dependent upon, among other things, the number of photons reaching the detector. It was therefore reasonable to assume that an organ of extreme radiance could obscure detection of those organs of lesser radiance, especially if the responsible β⁻ emission was close to the Cerenkov radiation threshold energy. Therefore, by extension, the observed differences in organ intensity when combined with the nearly four-fold difference in β⁻ decay energy between ⁹⁰Y and ¹⁷⁷Lu may explain the observed differences in overall image intensity between images generated with either of these radionuclides. To further explore this theory, organs were imaged ex-vivo for biodistribution to determine if the imaging results would be different without the carcass encasing them or from close proximity to organs with higher measured radiance compared to those of lesser measured radiance (Supplemental Fig. 2). The imaged organs revealed uptake that was otherwise missed with in vivo imaging (liver and small as well as large intestines with ⁹⁰Y-DOTA-30F11; stomach, small and large intestines with ¹⁷⁷Lu-DOTA-30F11). We therefore determined that the limit of detection was necessarily established by the organ of primary uptake (in this case the spleen). However, in some cases since the software simply analyzed pixels with relatively high signal in relation to a calibrated dark field image, organs with lower radiance were effectively masked by organs with higher radiance when imaged in vivo.

In summary, these studies show that in this disseminated model of leukemia CLI was capable of providing preliminary insight into the general success or failure of the targeting of a radioimmunoconjugate in primary tissues of uptake, making CLI a useful technique to follow biodistribution or therapy studies in real-time without animal sacrifice. CLI was not yet sensitive enough, however, to provide reliable data on organs of lesser uptake as the signal was masked by organs with higher uptake. Despite the limitation of sensitivity, CLI should continue to be explored in the clinic as a means to confirm localization of radionuclides in target organs. Groups are optimizing the use of CLI to augment clinical imaging modalities with ¹⁸F-fluorodeoxyglucose (FDG) via endoscopy, which reduces or removes interference from tissue between the source and the camera, as well as any reduction in signal caused by hair (13). At current sensitivity levels, it has been possible to detect as low as 1 μCi of ¹⁸F in mice. Additionally, Thorek and colleagues recently reported direct visualization, with an external CCD, for CLI of nodal disease in primary and contralateral lymph nodes of human patients with head and neck malignancies using 455 MBq of ¹⁸F-FDG (17). If detection limits can be standardized and fully translated to the human scale it may increase the overall clinical efficacy of CLI techniques. Other applications for CLI include the use of high-energy particle emitters for the in situ excitation of injected fluorochromes. As recently reported by Orbay and colleagues, the intent of this application would allow for the excision of tumor free surgical margins, thus increasing the overall chances for survival and prolonged disease-free intervals (33) Further, careful consideration of emission energy and path-length with respect to radionuclide choice and target organs may be essential in order to optimize the signal-to-noise ratio responsible for high-quality CLI imaging, especially in our disseminated model of disease. Differences in measured pixel intensities and thus radiance values resulted from differences in absorbed doses and emission energies for ⁹⁰Y and ¹⁷⁷Lu, suggesting higher image contrast with the use of ⁹⁰Y in these studies or potentially at equal absorbed doses.