|Home | About | Journals | Submit | Contact Us | Français|
Clinical therapeutic studies using 225Ac-labeled antibodies have begun. Of major concern is renal toxicity that may result from the three alpha-emitting progeny generated following the decay of 225Ac. The purpose of this study was to determine the amount of 225Ac and non-equilibrium progeny in the mouse kidney after the injection of 225Ac-huM195 antibody and examine the dosimetric consequences. Groups of mice were sacrificed at 24, 96 and 144 h after injection with 225Ac-huM195 antibody and kidneys excised. One kidney was used for gamma ray spectroscopic measurements by a high-purity germanium (HPGe) detector. The second kidney was used to generate frozen tissue sections which were examined by digital autoradiography (DAR). Two measurements were performed on each kidney specimen: (1) immediately post-resection and (2) after sufficient time for any non-equilibrium excess 213Bi to decay completely. Comparison of these measurements enabled estimation of the amount of excess 213Bi reaching the kidney (γ-ray spectroscopy) and its sub-regional distribution (DAR). The average absorbed dose to whole kidney, determined by spectroscopy, was 0.77 (SD 0.21) Gy kBq−1, of which 0.46 (SD 0.16) Gy kBq−1 (i.e. 60%) was due to non-equilibrium excess 213Bi. The relative contributions to renal cortex and medulla were determined by DAR. The estimated dose to the cortex from non-equilibrium excess 213Bi (0.31 (SD 0.11) Gy kBq−1) represented ~46% of the total. For the medulla the dose contribution from excess 213Bi (0.81 (SD 0.28) Gy kBq−1) was ~80% of the total. Based on these estimates, for human patients we project a kidney-absorbed dose of 0.28 Gy MBq−1 following administration of 225Ac-huM195 with non-equilibrium excess 213Bi responsible for approximately 60% of the total. Methods to reduce renal accumulation of radioactive progeny appear to be necessary for the success of 225Ac radioimmunotherapy.
Most radioimmunotherapy (RIT) clinical trials involve the use of antibodies labeled with β-emitting radionuclides (Kaminski et al 2005, Gopal et al 2003, Divgi et al 2004, Witzig et al 2002, Bander et al 2005). Advantages of β-emitters include their relatively long range (e.g. the average range of 90Y β-emission is 3.9 mm in soft tissue); as a result, the radiolabeled antibody need not localize in or on every cancer cell to be effective since substantial radiation dose may be delivered by cross-fire. Other advantages of β-emitters include availability at low cost, convenience and widespread clinical experience in their handling.
There is, however, growing interest in the development of therapies using antibodies labeled with α-particle-emitting radionuclides (Miederer et al 2008, Sgouros 2008, Boskovitz et al 2009, Song et al 2009, Milenic et al 2010, McDevitt et al 2001). Alpha particles have a relatively short range in tissue (<90 μm) and a high linear energy transfer (LET), depositing large amounts of energy per unit track length. The ratio of energy deposited by an α-particle traversal of a small volume, such as a cell nucleus, to that deposited by a β-particle is typically ~400. Moreover, the biological impact per unit dose is further enhanced by a factor of at least 3 due to the higher relative biological effectiveness (RBE) of α- versus β-radiation for the endpoint of cellular reproductive failure (Adelstein et al 2002). The net result of these factors is that the α-particle dose corresponding to D0 (the radiation dose required to reduce surviving fraction by a factor e−1) is reached with only 2 or 3 hits per cell, compared to more than 1000 hits for low LET radiations (Humm 1986). Finally, the shorter range of α-particles will result in less radiation cross-fire to normal tissues.
Until recently, α-particle RIT has focused on radionuclides that emit single α-particles, such as 211At (Akabani et al 2006, Zalutsky et al 2008) and 213Bi (Jurcic et al 2002, Friesen et al 2007). While these studies demonstrated aspects of feasibility and safety of α-particle-based therapies, efficacy was typically limited by the short half-lives of 211At and 213Bi (7.2 h and 46 min respectively). Short-lived radionuclides are pharmacokinetically incompatible with antibody-targeted therapy due to the significant time required (typically >1 day) to achieve a targeting differential, during which the vast majority of activity will decay.
Recently, interest has focused on 225Ac as a potential α-emitting radionuclide for RIT (Miederer et al 2008, Song et al 2009). 225Ac has a ten day half-life, long enough to allow differential tumor accumulation and possibly cellular internalization of radiolabeled antibody prior to decay. In addition, the total decay of 225Ac, from original parent through successive progeny, involves the emission of four α-particles. Thus, the decay of 225Ac at a tumor site, especially if the construct were internalized, could result in an extremely radiotoxic event.
Figure 1 summarizes the decay scheme of 225Ac. The first decay results in a 221Fr nucleus with a recoil kinetic energy (mean 120 keV) significantly greater than the chemical bond energies of any chelating agent. The fate of 221Fr and its downstream progeny will thus depend upon the 225Ac decay site. If 225Ac decays within tumor, especially if internalized, some or all of the subsequent decays may also occur in tumor. Other scenarios (e.g. 225Ac decay in the circulation) could result in some or all subsequent decays occurring elsewhere.
In terms of non-invasive imaging, there are two detectable events: (1) a 218 keV gamma-ray emitted with the α-decay of 221Fr to 217At and (2) a 440 keV gamma-ray emitted with the β-decay of 213Bi to 213Po. The latter of these has been successfully imaged in clinical studies (Sgouros et al 1999). Due to its short half-life (32 msec), it can reasonably be assumed that the decay of 217At occurs at the same location as that of 221Fr. However, the half-life of 213Bi is sufficiently long (46 min) for significant translocation to occur prior to its decay. By imaging separately the 218 keV and 440 keV photons, it may be possible to determine the spatial locations of 221Fr and 213Bi atoms. Thus, for non-invasive imaging the best estimate of the location of 225Ac would have to be derived from that of 221Fr. In pre-clinical systems, this situation may be improved by the use of invasive methods that enable the location of 225Ac to be determined.
Previous biodistribution studies with 225Ac-huM195 antibody in mice (Jaggi et al 2005) and non-human primates (Miederer et al 2004) have shown accumulation of 213Bi in kidney suggesting this may be dose limiting for 225Ac-antibody therapy. Early studies on the metabolism of bismuth compounds suggested preferential uptake in the renal cortex (Russ et al 1975, Slikkerveer and Dewolff 1989, Szymanska et al 1977). Speidel et al (1991) showed that while bismuth isotopes were excreted through the kidney, there was non-negligible (3–5%) longer term retention. Furthermore, the kidney is particularly sensitive to radiation damage (Emami et al 1991, O’Donoghue 2004). The focus of this study was to use invasive methods to quantify the contributions to absorbed radiation dose in kidney from 225Ac arriving as radiolabeled intact antibody versus 213Bi arriving in the form of a dissociated radiometal following 225Ac decay elsewhere in the body. Our long-term goal is to better understand kidney dosimetry for 225Ac-labeled antibody therapies.
Experiments were performed using 4–12 week old female BALB/C mice obtained from Taconic (Germantown, NY). All animal studies were conducted according to the NIH Guide for the Care and Use of Laboratory Animals and were approved by the Institutional Animal Care and Use Committee at Memorial Sloan-Kettering Cancer Center.
A total of 12 anesthetized mice were each injected with 225Ac-huM195 (22.2 kBq in 100 μl) in the retro-orbital venous plexus. Animals in groups of 4 were sacrificed at 24, 96 and 144 h post-administration. Immediately following sacrifice, both kidneys were removed from each animal. One kidney was used for nuclear spectroscopy measurements with a HPGe detector. The other was rapidly frozen and sectioned for histology and DAR.
Actinium-225 was provided by Oak Ridge National Laboratory (Oak Ridge, TN). The humanized anti-CD33 antibody (IgG1) huM195 antibody (Protein Design Labs) was labeled with 225Ac using a two-step labeling method, as described previously (McDevitt et al 2002). Radiochemical purity was 95–99% by instant thin-layer chromatography. The specific activity of 225Ac-huM195 was approximately 3.2 MBq mg−1.
Immediately after sacrifice, one kidney was placed in a vial and positioned 10 cm from the front face of a Princeton Gamma Tech PGT IGC P-Type HPGe (Princeton, NJ 08540) detector connected to a Canberra Industries multi-channel analyzer system (Meriden, CT) and counted for 30 min. The system energy resolution (FWHM) was 1.8 keV for 1336 keV γ-rays. The digitized signal was stored as an energy histogram and total counts in the spectral peaks (i.e. area under the curve) corresponding to 218 keV (221Fr) and 440 keV photons (213Bi) were determined. Thereafter, kidneys were stored in a refrigerator for at least 24 h and the measurements repeated. The second measurement reflected only activity that was originally present in the sample as 225Ac. The system calibration factor was estimated by measuring a 225Ac standard of known activity in equilibrium with its progeny, positioned at the same location as the kidney samples.
The second kidney from each mouse was embedded in OCT (Optimum Cutting Tissue, Miles Inc., Elkhart, IN) and frozen on dry ice in a cryo-mold. Sets of ten contiguous 8 μm thick tissue sections were cut using a Microm HM500 cryostat microtome (Microm International, Walldorf, Germany) and deposited onto poly-L-lysine coated glass microscope slides. Alternate sections were used for DAR and histological staining. Activity standards were constructed by mixing known activities of 225Ac in known volumes of OCT compound, pouring into cryo-molds and freezing on dry ice. Thereafter, sections of 8 μm thickness were cut and collected on glass microscope slides in a manner identical to that used for the kidney specimens.
Digital autoradiograms were obtained of the tissue sections and two 225Ac standards, all of which were covered with a thin Mylar film and placed in a film cassette against a Fuji film BAS-MS2325 imaging plate (Fuji Photo Film Co, Tokyo, Japan). Two exposures were made: (1) a 1 h exposure begun 45–60 min after sacrifice and (2) a 24 h exposure begun at least 24 h post-sacrifice. As for the case of nuclear spectroscopy, the second DAR measurement reflected only activity originally present in the sample as 225Ac.
Exposed phosphor plates were read by a Fujifilm BAS-1800II bio-imaging analyzer (Fuji Photo Film Co, Tokyo, Japan) generating digital images with 50 μm pixel dimensions. DAR image intensity, expressed in the machine readout parameter of photo-stimulated luminescence per square mm (PSL mm−2), was quantified using Multi Gauge software (version 2.2, Fujifilm, Tokyo, Japan). To minimize differences associated with variable plate sensitivity, a single phosphor plate was used for all sample exposures.
Tissue sections for histology were fixed in 10% phosphate-buffered formalin for 5 min, washed twice, air dried and stained with hematoxylin and eosin. Digital images (pixel dimension 12 μm) were obtained with an Olympus BX60 System Microscope (Olympus America Inc, Melville, NY) equipped with a motorized stage (Prior Scientific Inc, Rockland, MA). Subsequently, the H & E images were re-sampled to the same resolution as that of the DAR data. Ten DAR images (early and late exposures of five tissue sections) were manually aligned to the histologic images using rigid planar transforms, consisting of translations and rotations only. Once aligned, the five DAR images for each exposure time were averaged to create a single mean-intensity DAR image. For each kidney, two DAR images corresponding to early and late exposures were thus obtained. Regions of interest (ROI) corresponding to the cortical and medullary regions of each kidney were manually drawn on each histologic image and then overlaid onto the corresponding aligned DAR image. The mean PSL mm−2 in each ROI was then recorded. ROIs were also drawn on the co-exposed 225Ac standards.
The analysis was based on the following rationale and assumptions.
The contribution of any radioactive species to a measurement is proportional to the total number of decays of the species that occurs over the time of measurement. If a measurement of duration, δ, begins at a time, τ, after sacrifice, then the total number of decays during measurement, Ã, is given by
where Asac is the activity in the sample at the time of sacrifice and λ is the appropriate decay constant (= ln(2)/T1/2), i.e.
For 225Ac and its progeny in equilibrium, the appropriate decay constant is that of 225Ac as the activities of all equilibrium progeny are equal to that of 225Ac. For non-equilibrium 213Bi, the appropriate decay constant is that of 213Bi.
The total number of decays is related to the measured value, N, by
where κ is a calibration factor, estimated using standards of known activity.
The activity at the time of sacrifice can therefore be calculated from N by
The activity of 225Ac and its progeny in equilibrium at the time of sacrifice was estimated based on the late measurement using equation (4). For the late measurement the ratio of the 218 (221Fr) to 440 keV (213Bi) spectral peak counts was equal to that of the 225Ac standard. Independent estimates of Asac were made for each of the 218 keV and 440 keV spectral peaks. These were used to generate expected values for the number of counts in the 440 keV peak at the early measurement due to 225Ac and progeny in equilibrium. The differences between the expected values and the actual measured values (ΔN) represented the contribution of non-equilibrium, excess 213Bi at the time of measurement. Equation (4) was then used with N replaced by ΔN to derive estimates of non-equilibrium (i.e. excess) 213Bi activity at the time of sacrifice.
The metric used for DAR analysis was mean PSL mm−2 value for each ROI. Values of Asac per unit volume for 225Ac and progeny in equilibrium were estimated from the late DAR using equation (4) and calibration factors (in units of Bq s ml−1 per PSL mm−2) derived from the activity standards. The calculated Asac for 225Ac were then used to generate expected values of PSL mm−2 for the early measurement. The differences between expected and actual values for the early measurements were taken to represent the contribution of non-equilibrium excess 213Bi. Analysis of the output of the DAR data provided estimates of the relative contributions of 225Ac (and equilibrium progeny) and non-equilibrium excess 213Bi to the cortex and medulla, defined using co-registered H&E-stained tissue sections. The partitioning of the whole kidney dose, estimated by spectroscopy as described above, into separate cortex and medulla doses was performed using this relative uptake data.
The absorbed radiation dose to the kidney is determined by two principal components.
Using the methods described above, we separately estimated the two components of kidney-absorbed dose. The activity concentrations of equilibrium 225Ac and non-equilibrium excess 213Bi were evaluated at each sacrifice time to generate two activity concentration-time profiles. Cumulated activity concentrations, x, in the kidney for each radionuclide (x) were then derived by fitting mono-exponential functions to the activity concentration-time data and integrating to infinity. The absorbed dose to the kidney for radionuclide (x), Dx, was then given by
where Δx is the absorbed dose constant for the non-penetrating α and β emissions of 225Ac and progeny (4.40 × 10−12 J) or 213Bi (1.33 × 10−12 J) as appropriate. The penetrating photon emissions which contribute <1% of the total radiation dose were ignored.
In order to derive a projected estimate of human kidney-absorbed dose it was assumed that the fractional uptake of administered activity in kidney (Ak/A0) adjusted for relative fractional organ mass is the same in mice and humans. Specifically,
where A0 and Ak denote administered activity and activity in kidney respectively. Standard values of whole body and kidney masses for human (Stabin et al 2005) and mouse (Stabin et al 2006) were used.
In terms of cumulated activity concentration (i.e. per unit mass) in kidney, x, the relationship reduces to
Projected absorbed doses were then calculated using equation (5).
Figure 2 shows representative early and late γ-ray spectra for kidneys removed at 24, 96 and 144 h after injection of 225Ac-huM195. It is apparent that the heights of the peaks of interest (218 keV and 440 keV) are dependent on the time of sacrifice and that the ratios of these heights are dependent on the time of measurement. The average count ratios (440/218) for all excised kidneys were 4.8 (SD 0.6) and 1.2 (SD 0.1) for early and late measurements respectively, with no apparent dependence on the time of sacrifice. Figure 3 plots the mean (±SD) values of the activity concentrations of 225Ac and 213Bi in kidney at the times of sacrifice together with the best-fitting mono-exponential clearance curves. These results are consistent with the continual replenishment of 213Bi in the kidney by new atoms generated by 225Ac decays in the rest of body, leading to an effective half-life (4.6 day) similar to that of 225Ac (6.7 day). The results also indicate that the majority of activity in the kidneys at the time of sacrifice was in the form of non-equilibrium 213Bi.
Figure 4 shows a set of images (H&E stained section with early and late DAR) of three kidneys removed at 24, 96 and 144 h after injection of 225Ac-huM195 illustrating the placement of cortical and medullary ROI on the H&E sections and their transfer to the DAR. In qualitative terms, it is apparent that the late DAR show enhanced cortical uptake, whereas the early DAR show similar uptake in both cortex and medulla. This was a universal finding for all kidney samples examined and indicates that 225Ac was primarily present in the renal cortex, probably in the form of antibody conjugate. In contrast, non-equilibrium-free 213Bi was more likely to be present in the renal medulla.
Using the previously described methods, the average cortex:medulla ratios for all excised kidneys were 1.9 (SD 0.2) and 0.4 (SD 0.3) for equilibrium 225Ac and non-equilibrium 213Bi respectively. Consistent with the nuclear spectroscopy results there was no systematic dependence of these ratios on the time of sacrifice. These results indicate that, of the 225Ac-huM195 present in the kidneys, the concentration was approximately two times greater in the cortex than in the medulla; whereas activity arriving in the form of 213Bi was preferentially located in the renal medulla at a concentration approximately two times greater than in the cortex.
The spectroscopic data enabled estimation of the average absorbed radiation dose from the combination of equilibrium 225Ac and non-equilibrium 213Bi to the entire kidney. Based on this analysis, the estimated absorbed dose delivered to the mouse kidney by the administration of 22.2kBq of 225Ac-huM195 was 17.4 (SD 4.8) Gy (i.e. 0.78 Gy kBq−1). Of this total, the contribution due to activity arriving in the kidney in the form of 213Bi was estimated to be 10.4 (SD 3.7) Gy. Incorporation of the differing relative concentrations of 225Ac and 213Bi in cortex and medulla derived from DAR analysis led to differential absorbed dose estimates for the two renal components. Assuming relative volumes of 70% and 30% for cortex and medulla (including renal pelvis) as per MIRD pamphlet 19 (Bouchet et al 2003), the absorbed dose estimates were 15 (SD 4) Gy to the cortex and 23 (SD 7) Gy to the medulla. It is noteworthy that a large component (18.2 (SD 6) Gy) of the estimated absorbed dose to the renal medulla was delivered by non-equilibrium excess 213Bi generated from extra-renal decays of 225Ac. Dosimetry estimates for mouse kidney are summarized in table 1.
The projected absorbed dose to kidney for human patients derived as described in methods was 0.28 Gy MBq−1 of which 0.11 Gy MBq−1 was due to activity arriving in the form of 225Ac and 0.17 Gy MBq−1 to non-equilibrium excess 213Bi. Human dosimetry projections are summarized in table 1. The separate projections for cortex and medulla were derived by proportionality with the dose estimates for mouse kidney.
The main aim of this study was to differentially estimate the amount of 225Ac-labeled antibody and non-equilibrium progeny in the kidney, likely to be the dose limiting normal tissue for 225Ac RIT (Jaggi et al 2005). This was primarily achieved by γ-ray spectroscopy, with additional sub-regional information provided by DAR. The key features of the methodology were the ability to perform measurements very rapidly (within 10 min in the case of γ-ray spectroscopy) after animal sacrifice and repeat them on the same samples after any non-equilibrium excess had decayed. These requirements constrained the possible techniques that could be used. In particular, the consequential alteration or destruction of the biological specimen precluded techniques such as emulsion autoradiography or liquid scintillation counting. Although DAR is a suitable method it currently lacks the spatial resolution to identify uptake at a cellular or sub-cellular level and is, in that sense, sub-optimal. Comparison of γ-ray spectra at early and late times post-sacrifice enabled the amount of non-equilibrium excess 213Bi to be estimated. It was shown that non-equilibrium 213Bi (resulting from the extra-renal decay of 225Ac) was a major contributor to kidney-absorbed dose and in the renal medulla was responsible for 80% of the total dose.
Mechanistically, our understanding is that the initial decay of 225Ac dissociates the first progeny 221Fr from the antibody-chelate. The blood plasma will contain a steady-state equilibrium mixture of (antibody-associated) 225Ac and (non-antibody-associated) progeny. 213Bi in blood plasma which may be bound to transferrin (Sun and Szeto 2003, Miquel et al 2004) is of particular significance because of its preferential kidney uptake, possibly mediated by bismuth-binding metallothioneins (Sun et al 1999, Garrett et al 1999). The level of non-equilibrium 213Bi in the kidney is continuously replenished by the plasma steady state and is greater than the amount of 213Bi produced by the in situ decay of 225Ac in kidney. Our data show an almost constant ratio (approx 7:1) between 225Ac (and by inference equilibrium 213Bi) and non-equilibrium 213Bi. These findings are highly significant for clinical applications of 225Ac-labeled antibody therapy. In principle, 225Ac is a good pharmacokinetic match with antibodies that require significant time to achieve a targeting differential. However, the downside is that such molecules remain present for a protracted time, primarily in the blood plasma, and, if labeled with 225Ac, will be continuous generators of alpha-emitting progeny with essentially no tumor selectivity. Given the apparently high level of renal uptake of 213Bi, the radiobiological potency of alpha particles and the clinical significance of renal toxicity, it will be essential to develop methods to reduce kidney accumulation of radioactive progeny. Possible approaches include inhibition of renal absorption or acceleration of renal (or other) clearance. Preliminary studies (Jaggi et al 2006) have explored the use of metal chelators (DMSA, DMPS and Ca-DTPA) to scavenge 213Bi atoms and of non-radioactive Bi atoms to block the renal binding sites of 213Bi. Renal uptake of 213Bi was reduced by these interventions (14–45% by metal chelation; 11–16% by cold Bi). These authors also found that the use of diuretics (with or without metal chelation) further reduced 213Bi accumulation in the kidney. Approaches such as these may improve sparing of this potentially dose limiting organ and will probably be mandatory for the successful clinical implementation of 225Ac-antibody therapy.
We recently initiated a phase I clinical trial using 225Ac-huM195 antibody at MSKCC. Although the initial administered activity (18.5 kBq kg−1/0.5 μCi kg−1) is insufficient to permit imaging, such studies are planned at higher administered activities. It will be of considerable interest to see if clinical non-invasive imaging is capable of accurately assessing the biodistribution of 225Ac and its troublesome progeny.
In this study, we quantified in mice the renal uptake of non-equilibrium 213Bi produced by 225Ac decays in the rest of body and assessed its relative contribution to kidney-absorbed dose following administration of 225Ac-labeled antibody. We found that non-equilibrium 213Bi is responsible for a significant fraction of the total radiation dose to kidney. For the renal medulla, the dosimetric contribution of non-equilibrium 213Bi was as high as 80%. The absorbed dose to the mouse kidney per kBq of administered 225Ac-huM195 was estimated to be 0.77 Gy. For human patients we project a kidney-absorbed dose of 0.28 Gy MBq−1 following administration of 225Ac-huM195 with non-equilibrium excess 213Bi responsible for approximately 60% of this total. These findings reinforce the importance of developing methods to reduce renal accumulation of radioactive progeny in clinical applications of 225Ac radioimmunotherapy.