Design, construction, and in vivo feasibility of a positioning device for irradiation of mice brains using a clinical linear accelerator and intensity modulated radiation therapy

Abstract

Purpose: The goal of this study was to design a positioning device that would allow for selective irradiation of the mouse brain with a clinical linear accelerator.

Methods: We designed and fabricated an immobilization fixture that incorporates three functions: head stabilizer (through ear bars and tooth bar), gaseous anesthesia delivery and scavenging, and tissue mimic/bolus. Cohorts of five mice were irradiated such that each mouse in the cohort received a unique dose between 1000 and 3000 cGy. DNA damage immunohistochemistry was used to validate an increase in biological effect as a function of radiation dose. Mice were then followed with hematoxylin and eosin (H&E) and anatomical magnetic resonance imaging (MRI).

Results: There was evidence of DNA damage throughout the brain proportional to radiation dose. Radiation-induced damage at the prescribed doses, as depicted by H&E, appeared to be constrained to the white matter consistent with radiological observation in human patients. The severity of the damage correlated with the radiation dose as expected.

Conclusions: We have designed and manufactured a device that allows us to selectively irradiate the mouse brain with a clinical linear accelerator. However, some off-target effects are possible with large prescription doses.

Keywords:

• Radiation-induced brain injury
• mouse models
• small animal irradiation
• whole-brain irradiation

Introduction

Radiation is a standard component in the treatment of intra-cranial neoplasms. It is well understood that radiation therapy can cause significant damage to the normal brain and impair quality of life. Late effects in the central nervous system can be subdivided into radiation-induced cognitive impairment which leads to white matter specific non-vascular damage and radiation necrosis which is primarily a vascular-mediated injury with each classification having distinct radiological (MRI) characteristics (Valk and Dillon 1991). Rodent models have been used extensively to study the effects of radiation in the brain. Application of human-like low dose radiation regimes to the rodent brain lead to models which show behavioral deficits akin to radiation-induced cognitive impairment with no corresponding MRI changes (Atwood et al. 2007; Shi et al. 2009). In parallel, high dose (single fraction of 4500 cGy or more) rodent models produce histopathological and radiological findings consistent with radiation necrosis (Hideghéty et al. 2013; Jiang et al. 2015).

Pre-clinical studies on radiation-induced brain injury in rodents primarily make use of cesium-137 (Brown et al. 2005; Greene-Schloesser et al. 2014) or orthovoltage (keV) X-ray (Ford et al. 2011; Burrell et al. 2012) irradiators. In contrast, therapy in human patients is delivered with either cobalt-60 or MV X-ray irradiators. The rationale for the use of these differing photon energies is in part related to the tissue depth at which the maximum dose is deposited (dmax) with dmax being proportional to photon energy. For smaller animals, higher energies can be impractical as the dmax is larger than the animal dimensions. To be able to irradiate small rodents in a 6 MV linear accelerator (LINAC) in a time-effective manner the following are necessary: a method of anesthesia, a method to ensure consistent placement, and radiation bolus so that the target is at a depth slightly larger or equal to dmax. The positioning device described in this report accomplishes all three of these challenges. Our immobilization device incorporates a 3-point head holder for accurate positioning, integrated gaseous anesthesia delivery and scavenging, and tissue equivalent material around the head that serves as radiation bolus. We also produced a baseplate that attaches to the Varian exact lock-bar and couch index point system used for patient positioning and has slots for five individual mouse setups. This positioning device allowed us to irradiate five mice simultaneously with a different radiation dose for each brain using intensity modulated radiation therapy (IMRT) plans.

Materials and methods

Design and construction of the apparatus

The individual immobilization fixtures were machined out of nylon given that it is inexpensive, easy to machine, and has electron density that provides sufficient bolus for MV photons (1.115 relative electron density (Schneider et al. 1996)). A cylindrical opening with diameter of 16 mm and 40 mm depth was drilled into a nylon block sized 56 × 56 × 50 mm (width, length, depth). Additional slots were included in the block for ear bars and a tooth bar to serve as a three-point immobilization device as well as for anesthesia delivery and scavenging as shown in Figure 1. For ease of access when loading the mouse into the holder, there is a removable piece at the top of the block that is 15 mm in depth. If the intent were to perform body irradiations, this piece can be replaced with a larger one that extends to cover the whole body. A baseplate was then constructed with five square slots where the immobilization fixtures can securely be inserted plus additional slots that would allow the baseplate to be mounted on a Varian Exact treatment couch indexed lock bar system (Varian Medical Systems, Palo Alto, CA). This maximizes the reproducibility of placement between imaging and treatment and between treatment sessions. Since the position of the positioning device can be kept constant, a single simulation CT can be used across multiple treatment sessions.

Figure 1. Immobilization fixture. Panel A shows a rendering of the intended design. The red arrows indicate the ear bars, the green arrow the tooth bar, and the black arrow the removable top piece. Panels B and C show the front and back view of one the actual irradiation fixtures used, respectively. The color coded arrows correspond to the same structures as in Panel A. Note that the removable block (black arrow) is not shown so that the cavity is visible. Panel C also shows the anesthesia connections (blue arrows) which allow for both delivery and scavenging of anesthetic gas. Thumbscrews are used to secure the ear bars and the tooth bar to securely retrain the mouse. (Color online.)

Treatment planning and irradiation

All animal experiments were approved by the Purdue Animal Care and Use Committee. Male and female 4–5 week old BALB/cAnNHsd mice (Envigo, Indianapolis, IN) were used for this study in cohorts of five mice at a time. Mice were initially anesthetized in an induction box with sevoflurane (Priamal Critical Care, Bethlehem, PA) and then positioned in the restraining device where anesthesia was maintained through sevoflurane administered through the anesthesia ports of the immobilization fixture. Positioning of a cohort of five mice is shown in Figure 2(A). Radiation therapy simulation was performed with computed tomography (CT) using 0.625 mm slices and a 64 slice CT scanner (GE Light Speed VCT, GE Medical Systems, Milwaukee, WI). A Varian exact simulation couch overlay was used to reproduce positioning of the mice (Civco Medical Solutions, Orange City, IA). Mice were then allowed to recover under observation and returned to the mouse facility. Mice were treated using a Varian Clinac 6Ex linear accelerator with a millennium 120 leaf multi-leaf collimator (Varian Medical Systems, Palo Alto, CA). The simulation CT scan was imported into Eclipse v 11 treatment planning system (Varian Medical Systems, Palo Alto, CA). Mice brains were contoured individually and an intensity modulated radiation therapy (IMRT) plan was generated using the planning system at a 400 MU/min dose rate. Doses were assigned to each whole mouse brain of 1000, 1500 2000, 2500, or 3000 cGy, and dose was calculated using the AAA_11031 algorithm, heterogeneity correction, and sliding window leaf motion technique (Varian Medical Systems, Palo Alto, CA). The resultant treatment plan is shown in Figure 2(B).

Figure 2. Irradiation setup. Panel A shows five mice in the custom-made holder held under anesthesia prior to irradiation. Each head-holder securely attaches to the baseplate which then attaches to the treatment couch via the lockbar. Panel B shows a corresponding treatment plan overlaid on the simulation CT where each mouse has a distinct prescribed dose of either 1000, 1500, 2000, 2500, or 3000 cGy in a single fraction. The lockbar slightly raises the baseplate on that particular side of the holder which is why the plane in the brain of the mice in the row of 2 is slightly lower than for those in the row of 3.

The planned composite X-ray fluence was compared to the machine composite X-ray fluence for all fields using a MapCheck 2 diode array (Sun Nuclear, Melbourne, FL). Gamma analysis and distance to agreement analysis were used to compare the planned and output absolute dose with point passing criteria of 3 mm and 3%. Ninety-eight percent of 275 points passed using this analysis. To more stringently evaluate the treatment plan, gamma analysis and distance to agreement analysis were performed using 2 mm and 2% point dose comparison, in this scenario 95% of 275 points passed. After the radiation therapy plan was generated mice were anesthetized using the induction chamber above and positioned on the treatment couch using the positional device. Two orthogonal setup port films were taken using a computed portal radiography system and graticule (Onconcepts, Rochester, NY). The port films were compared to digitally reconstructed radiographs (DRRs) in Offline Review (Varian Medical Systems, Palo Alto, CA). Couch shifts were applied to align the positional device to the planned position in the DRRs. Each treatment beam was then delivered.

Magnetic resonance imaging (MRI)

After irradiation, the weight of the mice was evaluated daily to look for potential acute radiation effects (namely oral mucositis). Mice were then followed with longitudinal post-contrast T1 and T2 weighted MRI at 1, 4, 8, 12, and up to 16 weeks. MRI was performed on a 7T Bruker Biospec (Billerica, MA) under isoflurane anesthesia with a 40 mm volume coil. Immediately before each imaging session, mice were injected intraperitoneally with 0.2 mL MultiHance (gadobenate dimeglumine; Bracco Diagnostics Inc, Princeton, NJ) contrast agent diluted 1:10 in sterile saline. The use of contrast was meant to differentiate between focal radiation necrosis which is noted to have increased signal intensity in both post-contrast T1 and T2 imaging in patients (Valk and Dillon 1991) and mice (Perez-Torres et al. 2014). In contrast, radiation induced white matter injury has no post-contrast T1 enhancement though there is white matter specific enhancement on the T2-weighted images (Reddick et al. 2000; Fouladi et al. 2004).

Histology and immunohistochemistry

For the first cohort, mice were sacrificed at one hour post-irradiation, without undergoing MRI, for the evaluation of radiation induced DNA damage via phosphorylated histone H2AX (γ-H2AX) immunohistochemistry (Nowak et al. 2006; Ford et al. 2011). Subsequent cohorts were euthanized at either 4, 8, or 16 weeks post-irradiation for histological evaluation. Additional unirradiated mice of the same sex, age, and strain were included as negative controls. The heads were placed in 4% paraformaldehyde overnight after which the brains were extracted. The tissue was then sent to the Purdue Histology Lab core facility for paraffin processing and embedding and 5 µm sections were produced. One section from each brain was stained with hematoxylin and eosin (H&E) according to standard protocols. Immunohistochemistry was performed for the phosphorylated histone H2AX at serine 139 (clone JBW 301, Millipore, Billerica, MA). Briefly, sections were deparaffinized and rehydrated. Antigen retrieval was performed by overnight incubation in 50 °C citrate buffer at pH 6. Sections were blocked with 10% goat serum for one hour at room temperature, followed by overnight incubation with the γ-H2AX antibody and then one hour incubation at room temperature with goat anti-moue HRP secondary antibody (W402B, Promega, Madison, WI). Sections were then developed with diaminobenzidine (DAB) for 10 min. Staining was visualized and photographed in an EVOS-XL digital light microscope.

Results

Throughout the entire study 75 mice, consisting of 35 males and 40 females were irradiated. There was an 87% survival rate for mice to reach the end of their intended time point. Of those that died prematurely, seven died due to acute weight loss. Acute weight loss was observed for radiation doses higher than 1500 cGy after irradiation that peaked at around day 9 post-irradiation as shown in Figure 2. This weight loss was transient and mice quickly re-gained the weight within a couple of days though they had to be placed on soft food. This weight loss is likely due to extracranial effects likely in the form of damage to the soft tissue in the mouth, esophagus, and/or salivary glands. In fact, salivary gland and mandibular injury models are generated in rodents with doses of at least 1500 cGy (Vissink et al. 1991; Sønstevold et al. 2015). There were also four mice that died due to anesthetic or ear bar complications during irradiation. Another extracranial effects that was seen, but was not problematic was alopecia, though only in mice that received at least 1500 cGy, and can be seen in Supplementary Figure 1.

Gamma H2AX immunohistochemistry

The γ-H2AX staining is performed as a marker of DNA double strand breaks, and therefore an indirect marker of effective radiation dose. This was performed on mice that were sacrificed 1 h post-irradiation. Photomicrographs of the staining are shown in Figure 3 for a representative 1000 and 3000 cGy mouse brain. There was positive staining throughout all the areas examined, which is consistent with DNA double strand breaks due to a whole-brain irradiation treatment plan as is shown in Figure 2. The intensity of the staining and the number of nuclei stained increased as the dose of radiation increased as evidenced in Figure 3, particularly evident in the dentate gyrus (Figure 3(C and H)).

Figure 3. γ-H2AX staining results for the treatment plan shown in Figure 2. Results from the brain irradiated with 1000 cGy is shown on the top row (A–E) with 3000 cGy in the bottom row (F–J). A and F show the left hemisphere in a 1.25× magnification. B and G show the staining in hippocampus at 10× magnification. C–E and H–J show the staining at 20× magnification in dentate gyrus, cortex above the hippocampus and thalamus near the third ventricle, respectively. Scale bar in G and H indicates 400 and 200 µm, respectively.

Magnetic resonance imaging (MRI)

Standard anatomical MRI with contrast was performed at times ranging from 1 week to 16 weeks post-irradiation. MRI results for all radiation doses at 4 weeks post-irradiation are shown in Figure 4 (top row). Unfortunately, we were unable to detect any abnormalities for any dose at any time point post-irradiation.

Figure 4. MRI and histology at 4 weeks post irradiation for all doses. Top row shows normal anatomical MRI scans for all doses. The middle and bottom rows show two magnifications of standard hematoxylin and eosin staining at the level of the hippocampus. There is a dose-dependent change in the white matter tract identifiable by the tissue getting lighter in color due to degeneration and vacuolation. Arrows indicate the white matter tract and in particular the areas that changed after irradiation. Scale bar indicates 400 and 200 µm, respectively for 10× and 20× magnification, respectively.

Histological results

H&E sections were evaluated at 4, 8, and 16 weeks post-irradiation. Figure 4 shows representative sections of the external capsule at 4 weeks post-irradiations in females. Subtle white matter specific damage is observed that appears to worsen with dose. Specifically, vacuolar changes and white matter degeneration led to small holes appearing in the white matter tract which also broadens. In Figure 4, this is seen as the appearance of white dots and eventually a general whitening of the white matter track is evident at 3000 cGy. This dose dependent effect, which is expected, was observed at all time points for both males and females. Further H&E staining for male and female samples at other timepoints can be found in Supplementary Figure 2.

Discussion and conclusions

Pre-clinical mouse models are important for understanding mechanism of disease and initial evaluation of therapeutics. Ideally, the mouse model should be as close to the human condition as possible. For this reason, we decided to design and construct a device that would allow us to irradiate the mouse brain with a standard clinical 6 MV LINAC in a fashion similar to the human treatment. There are a few examples in the literature of mice being irradiated on clinical radiotherapy devices, namely either with a Gamma Knife (Perez-Torres et al. 2014) or a CyberKnife (Kim et al. 2014) unit. In these prior reports, their positioning systems did not included gaseous anesthesia delivery (opting for injectable anesthesia, instead) or tissue-equivalent bolus.

Our work presented here describes an integrated positioning device that also serves as anesthesia delivery and tissue-equivalent bolus. The device seamlessly interfaces with standard clinical positioning systems and could therefore be used at virtually any radiation oncology department. We were able to apply an IMRT treatment plan with five separate radiation doses such that each mouse brain got a unique dose. DNA damage immunohistochemistry confirmed our ability to deliver unique doses to each brain.

There are two major challenges with this device. First, because of how tight the ‘bolus’ material must be placed it can be challenging to properly place the ear bars. If done incorrectly, the ear bars may impinge on the trachea leading to death during irradiation. Second, because of the characteristic dose depth curves of 6 MV photons it can be difficult to perfectly spare tissues that are very close to the brain or target organ. A similar approach could be implemented for rats that could overcome this second challenge or at least reduce the effect.

Notes on contributors

Dr. Nicholas J. Rancilio, DVM is a Clinical Assistant Professor of Radiation Oncology in the Department of Veterinary Clinical Sciences in the Purdue University College of Veterinary Medicine. He is a diplomate of the ACVR in Radiation oncology.

Mr. Shaun Dahl, MS was a trainee in the Medical Physics program in the School of Health Sciences at Purdue University. He has since graduated and has now entered residency for Medical Physics.

Dr. Ilektra Athanasiadi, DVM is a resident in Radiation Oncology in the Department of Veterinary Clinical Sciences in the Purdue University College of Veterinary Medicine.

Dr. Carlos J. Perez-Torres, is an Assistant Professor of Radiological Health Sciences within the School of Health Sciences at Purdue University. His research focuses on the use of MRI to track radiation-induced pathologies.

Acknowledgements

The authors thank and acknowledge Steve Powers from Purdue Research Machining Services for helping design and construct the positioning device. The authors acknowledge the assistance of the Purdue University Histology Laboratory, a core facility of the NIH-funded Indiana Clinical and Translational Science Institute, for the processing of the tissue sections and H&E staining.

Disclosure statement

No potential conflict of interest was reported by the authors.