Animals and weekly observation:
All animal procedures were approved by the Purdue Animal Care and Use Committee under protocol number 1712001655. Four male 3-month old Yucatan mini-pigs were obtained from Premier BioSource (formerly S&S Farms). All pigs were housed in pairs in a facility designated for large animal research. Water was provided ad libitum and a commercial feed ration was made available twice daily. Pigs were observed at least weekly for any overt neurological impairment. Within the first week after irradiation they were observed every 24 h to ensure no acute side effects. Weights were tacked weekly by animal care staff and showed normal weight gain. Pigs were assessed in two sets of two with the first receiving 25 Gy and the second set receiving 15 Gy. There was no explicit control group, instead the contralateral hemsphere was intended to serve as an internal control for each animal. Animal procedures were performed with the help of the Purdue Pre-Clinical Research Laboratory, a core facility of the Purdue Center for Comparative Translational Research with ample experience with pig models.
All pigs were anesthetized using a combination of tiletamine-zolazepam (3 mg/kg), detomidine (0.18 mg/kg), and butorphanol (0.12 mg/kg) administered intramuscularly. After attaining lateral recumbency, pigs were intubated with an appropriately sized endotracheal tube as determined using body weight. General anesthesia was maintained with isoflurane (1–2% inhaled) and oxygen as delivered using mechanical ventilation. Vital signs were monitored and logged throughout all procedures. All pigs received intravenous fluids (PlasmaLyte®; 5–10 mL/kg) via an intravenous catheter in the auricular vein. Butorphanol (0.2 mg/kg IV) was dosed as needed. Pigs were monitored after each procedure to ensure proper recovery from anesthesia until they were capable of standing and walking on their own.
Immobilization devices and CT simulation
Each anesthetized pig was positioned in sternal recumbency and immobilized using an individualized bite plate  and thermoplastic mask on an indexable frame (Uniframe Baseplate, Civco Medical Solutions, Orange City, IA) for radiation therapy simulation CT. The CT simulation treatment couch was positioned in the gantry and the reference isocenter was determined using CT lasers. Crosshair marks were applied to the mask using cloth tape and permanent marker over the intersection of the CT laser at three points. Radiopaque fiducial markers were affixed to the mask at the 3 laser intersection points (Suremark, Vision Line Premium Labels, V-25, Van Arsdale, Innovative Products, Pensacola, FL). Scans were acquired without contrast using a 64-slice CT scanner and 0.625 mm slice thickness (VCT 64-Slice, GE Healthcare, Milwaukee, WI).
Subsequent to the CT scan, the pigs were imaged using a 3 T MRI unit (MAGNETOM® Prisma, Siemens Medical Solutions, Malvern, PA) using a 64-channel head coil with the pigs in sternal recumbency under general anesthesia. MR images of the brain were acquired 1 week pre-irradiation, 3 months post-irradiation, and either 4 months (P2) or 6 months (P3 and P4) post-irradiation with a consistent protocol. Included in the protocol were T1-weighted and T2-weighted images acquired using a three-dimensional Magnetization Prepared Rapid Acquisition Gradient Recalled Echo (3D MP-RAGE; TE = 4.7 ms, TR = 2080 ms, averages = 1) sequence and three-dimensional Fast Spin Echo (3D FSE; TE = 410 ms, TR = 2800 ms, averages = 1) sequence, respectively. All scans were acquired with 0.7 mm isotropic resolution with the same geometry. The animals were then given an intravenous injection of 0.2 mL/kg of MultiHance. A period of 11 min was allotted to allow the contrast enough time to accumulate within the intracranial space before acquiring the post-contrast T1-weighted images.
Radiation treatment planning
CT and MRI images were imported and co-registered using the Varian Eclipse treatment planning system (Varian Eclipse v. 11.0, Varian Medical Systems, Palo Alto, CA). Transverse MRI images and CT images were used for manual brain tissue contouring. The contoured structures included brain, right and left cerebral hemispheres, cerebellum, left and right cerebellum, brainstem, cervical spinal cord, optic nerves (right and left), optic chiasm, eyes, and lenses. Diencephalon was contoured as part of the hemispheres. The planning target volume (PTV) for the first pair of pigs (P1 and P2) included the left cerebral hemisphere and left cerebellum. The PTV for the second pair of pigs (P3 and P4) included the left cerebral hemisphere only. In addition, the structures “brain minus PTV” (brain-PTV) and “brain minus PTV minus 2 mm” were created for plan evaluation and optimization, respectively.
Inverse planning for intensity modulated radiation therapy (IMRT) was used in all pigs. All treatment plans were corrected for tissue heterogeneity using a calculation algorithm (Anisotropic Analytical Algorithm, version 11.0.31, Varian Medical Systems, Palo Alto, CA). For steep dose gradient, the normal tissue objectives were applied using a distance from the target border of 0.1 cm, start dose 100%, end dose 60%, and fall off 0.9 cm. Coplanar, isocentric, non-parallel opposed beams were used with a sliding window technique. Nine angles of radiation beams were distributed entering the left hemisphere (350°, 346°, 330°, 307°, 282°, 270°, 230°, 198°, and 180°). A single dose of 25 Gy for the first pair of pigs (P1 and P2) and 15 Gy for the second pair (P3 and P4) was prescribed to the PTV, while the right side was spared as a control. The single dose of 25 Gy was selected based upon the previous pig model reports [8, 9]. The dose of 15 Gy was chosen based on matching the biological effective dose of one of the most common fractionated whole brain radiotherapy prescriptions (2 Gy × 30 fractions) under the assumption that the alpha–beta ratio of the brain is 3. The plans were evaluated for pre-treatment quality assurance using the MapCheck 2 diode array (Sun Nuclear Corporation, 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%. The plan was considered acceptable for therapy when at least 95% of all points matched. The evaluation of the plan quality included dose volume histograms (DVHs) and dose color wash for PTV coverage and doses to organs at risk (OARs). The doses to OARs were evaluated according to QUANTEC . RadCalc software (LifeLine Software Inc.) was used as an independent method for verification of the monitor units (MUs). The plans were approved by a veterinary Radiation Oncologist.
The treatment parameters are reported as recommended by the ICRU [12,13,14]. Briefly, reported treatment parameters for the PTV included maximum (D2%), minimum (D98%), mean (Dmean), and median (D50%) dose. Homogeneity Index (HI = (D2%–D98%)/D50%), Conformity Index (CI, described below) and Gradient Index (GI = brain volume receiving 50% of prescription dose divided by brain volume receiving 100% of prescription) were used to assess plans retrospectively and were not used in the process of plan approval. An HI close to 0 (zero) shows a homogeneous absorbed dose in the PTV. The CI defines how adequately a target is covered by treatment without irradiation of any tissue outside the PTV. Specifically we calculated the Paddick CI  defined as CI = PTVPIV2/ (PTV × PIV), where PTVPIV is the volume of the PTV that is covered by 100% of the prescription dose and PIV is the brain volume receiving 100% of the prescription dose. A perfect plan has a CI score of 1. The GI is an objective tool to assess how rapidly the dose falls off outside of the PTV. A lower GI indicates steeper dose gradient and a value of < 3 could be ideal. Reported treatment parameters for the OARs (brainstem, cerebellum, spinal cord, optic nerves (right and left), and optic chiasm) included maximum (D2%), mean (Dmean), median (D50%), and Volume of Accepted Tolerance Dose (VATD = dose/volume limit). The maximum point dose (Dmax) was recorded for the lenses. Treatment parameters for the cerebellum were reported only for P3 and P4, since the left side of the cerebellum was included in the PTV for P1 and P2.
Each pig was positioned with the same individualized device used in the CT simulation and aligned to the marked reference isocenter in the radiation therapy vault using room lasers and mask crosshair marks prior to irradiation. Cardinal direction shifts generated in the treatment planning software were applied to align the pig to the plan isocenter. Orthogonal portal MV radiographs were taken to verify the position. A computed portal radiography system was used to develop each portal image (KODAK ACR—2000i, Onconcepts, Rochester, NY). DICOM portal images were imported into the treatment planning system, scaled, and aligned to the digital graticule in the treatment plan’s digitally reconstructed radiographs. The registered images were compared using the offline review program (Varian Medical Systems, Palo Alto, CA). Images were compared for perfect visual alignment of bony structures to the digitally reconstructed radiographs created from CT images used for the IMRT planning. Position was adjusted if alignment differed by greater than 1 mm, and portal radiographs were repeated to document final positioning.
Radiation was delivered with a 6 MV linear accelerator (Varian 6EX, Varian Medical Systems, Inc. Palo Alto, CA) with a 120-leaf multileaf collimator (Millennium 120 MLC, Varian Medical Systems, Palo Alto, CA) using photons with a dose rate of 400 MU/min.
After the final MRI, pigs were euthanized by intravenous injection with pentobarbital (100–200 mg/kg). Due to neurological deficits, P1 was euthanized at 4 months and P2 at 3 months post irradiation. P3 and P4 were euthanized at 6 months as we had originally planned for all pigs. Brains were extracted by veterinary staff of the Indiana Animal Disease Diagnostic Laboratory and left in 10% neutral buffered formalin for at least 24 h. Coronal gross sections were generated to match areas of interest on the MRI datasets, embedded in paraffin, and stained with hematoxylin and eosin (H&E) and Luxol Fast Blue (LFB). The former was utilized for general pathological examination of the sections while the latter was used to evaluate white matter integrity of the irradiated hemisphere.