Platform development
The RARAF FLASH irradiator is based on a retired Clinac 2100C (Varian Medical Systems, Palo Alto, CA), recently installed in the accelerator hall at RARAF (Fig. 1). Due to the tight space, the bed was discarded, and the gantry rotation was disabled with the beam permanently pointing vertically up.
For most FLASH irradiations, the Clinac is operated in service mode, using the 9 MeV electron setting (FLASH mode), providing the best penetration into tissue below the 11 MeV photoneutron threshold of 56Fe [15]. Higher dose rates, with a degraded dose depth profile are achievable when operating in 6 MV photon mode, with the target retracted (SuperFLASH mode). In this mode, the electron current is increased by several orders of magnitude, nominally to overcome the low photon yield at 6 MeV, allowing very high electron dose rates.
To minimize beam divergence, the foil assembly was removed from the carousel. The target piston was also decoupled from the Clinac control hardware and was set to be permanently in the “out” position. In this way the only material in the beam is the beam exit window and the integrated ionization chamber assembly.
We have modified the timer interface card (Fig. 2), replacing the Varian generated GDLY CNT signal with our own signal. Under normal Clinac operation, this signal controls the delay between the electron gun firing and the klystron. When GDLY CNT is 0, the two are synchronized and radiation is delivered; when it is +12V the electron beam is generated out of phase with the klystron and no radiation is delivered.
To generate the control pulse train (Fig. 3) we utilized a USB-CTR08 card (Measurement Computing, Norton, MA). The card accepts as input the KLY I signal from the Clinac controller and turns the output on (+5V) after a predetermined number of KLY I pulses for a predetermined number of pulses and then either stops or repeats. The output signal is used to actuate a solid-state relay (VOR1121A6; Mouser Electronics, Mansfield, TX) which grounds TP21 on the timer Interface card (beam on) or allows it to be pulled up to +12V (beam off).
Dosimetry
In 9 MeV electron mode, beam intensity could be monitored using the built-in ion chamber. Specifically, test Point TP1 on the ion chamber control card was hooked up to a multichannel analyzer, and the average pulse height and number of pulses recorded for each irradiation. In superFLASH mode, the ion chamber is severely saturated and cannot be used (Fig. 4). An alternative beam monitor is under development.
NIST traceable dosimetry was performed using an Advanced Markus ion chamber (AMIC) and UNIDOS E electrometer (PTW, Freiburg, Germany). The ion chamber was calibrated to absorbed dose in water, using a 60Co source at the MD Anderson Accredited Dosimetry Calibration Laboratory (ADCL). Measured dose was corrected for temperature, pressure, and radiation quality (Table 1).
However, the AMIC only provides reliable dosimetry for dose rates of up to about 1.5 kGy/sec and for radiation fields >1 cm in diameter. For higher dose rates and smaller fields, we used EBT3 Radiochromic film (Ashland Specialty Chemicalsf, Wayne NJ) at doses below 20 Gy and OC-1 OrthoChromic film (Orthochrome Inc., Hillsborough, NJ) at doses above 5 Gy. Each batch of film was calibrated by irradiating film and the AMIC at a dose rate of approximately 1 Gy/sec, using a large field size (Fig. S1).
Film was scanned using an Epson V700 scanner in transmission mode (EBT-3) or reflection mode (OC-1). The scanner was operated several times to warm up the lamp and then films (including an unirradiated control film) were placed on the scanner bed (EBT-3 under a sheet of glass) and scanned at 300 DPI and a bit depth of 48 bit. Only the red channel was used for dose reconstruction. Optical density was taken as negative log 10 of the ratio of red color value in the irradiated and unirradiated films. Dose vs. optical density was calibrated using a first order rational function \(D=\frac{a-b\times OD}{OD-c}\). A matlab script was then used to convert scanned film images to dose maps and to generate dose histograms in selected regions of the image.
Safety
Our accelerator hall was originally built of interlocking 27” thick reinforced concrete shield blocks. As this is not strictly sufficient shielding in conventional clinical operations [16], we have mapped the leakage radiation and lock off high radiation rooms (specifically the room directly above the Clinac head) during operation. Lower radiation areas are cordoned off as “no linger” zones. Similar to clinical installations, the DOOR interlock is used to ensure that the accelerator hall is vacated and locked during use. A Geiger-Muller based area monitor (Luldlum Measurements Inc, Sweetwater, TX) is located adjacent to the Clinac with a lighted sign visible from outside the accelerator hall. Additional monitors are placed in the rooms adjacent to the vault, including at the console.
As we are only irradiating using electrons and have seen excessive stray radiation outside the accelerator hall only when generating photon beams, we have disabled the target piston, such that the target is always in the out position.
Sample positioning
To allow for reproducible sample positioning both inside and outside the Clinac head, we have constructed a scaffold, using 25 mm extruded aluminum optical construction rails (Thorlabs, Newton, NJ). A graduated optical dovetail rail (Thorlabs) was bolted on the construction rail with the sample holder mounted on a dovetail rail carrier (Fig. 5b). Test tubes were placed in a 1.5” bored acrylic cube (McMaster-Carr, Chicago, IL; Fig. 5a/c). In addition, a custom holder was designed and 3D printed (Protolabs Inc, Maple Plains, MN) using PA12 Nylon (Fig. 5d). This holder allowed irradiation of larger objects, Petri dishes or a mouse anesthesia jig (Precision X ray Irradiation Inc, North Branford, CT; Fig. 5e). Adjustable ¼” lead screens were added for beam collimation.
To facilitate sample alignment to the beam we have mounted a 5mW red cross projection laser (Laserglow Technologies, North York, ON) to the top of the scaffold. The laser was aligned to the beam using in situ irradiated film.
Dicentric analysis
This study was approved by Columbia University’s Institutional Review Board (IRB) protocol IRB-AAAR0643. Peripheral blood was obtained after informed consent from two healthy volunteers (male, 32 yr. and female, 54 y.o.) with no recent history of exposure to ionizing radiation or clastogenic agents. Freshly-drawn blood was heparinized (vacutainers with sodium heparin anticoagulant (Becton Dickinson, NJ)), aliquoted into Matrix Storage Tubes (Fisher), and exposed to 3 Gy or 8 Gy 9-MeV electrons at a wide range of the dose-rates (3 x 105 Gy/sec (3 Gy only), 600 Gy/sec, 50 Gy/sec, 5 Gy/sec, and 1 Gy/min). The control samples were sham-irradiated and received 0 Gy dose.
After irradiation, the blood (0.5 mL) was stimulated with a mitogen (phytohemagglutinin, PB-MAX karyotyping medium) and incubated at 37°C, 5% CO2 for 48 hrs. 3 hrs before fixation, cells were arrested in metaphase with 0.1 mg/ml colcemid (Gibco). Following the treatment, cells were swollen for 20 min in hypotonic solution (0.075 M KCl) and fixed with methanol:acetic acid (3:1). 20 µl of fixed cells were dropped onto glass slides, dried at an ambient temperature of 25°C and relative humidity of 55%. Dried slides were then fixed in 4% formaldehyde and dehydrated in ethanol (70%-85%-100% for 2 min. each). Prepared chromosome spreads were stained with a PNA probe hybridization cocktail (FITC-labelled centromere and Cy3-labelled telomere probes (PNABio) in buffer), covered by a cover glass, denatured for 3 min. at 80oC, and left at a room temperature in the dark for 2 hrs. After that, the slides were washed twice in 70% formamide, then twice in TBST (with 0.05% Tween™-20), and counterstained with Vectashield® Antifade Mounting Medium with DAPI ( 4′,6-diamidino-2-phenylindole).
Images were acquired on a Metafer 4 Scanning System (MetaSystems) equipped with a Carl Zeiss Axioplan Imager Z1, a CoolCube 1 Digital High-Resolution CCD Camera, and a Zeiss Plan-Apochromat 63x oil immersion objective. Metaphases were located using an automated classifier comprised of two acquisition modules, MSearch and AutoCapt. Captured images were analyzed using Isis software (MetaSystems). Metaphase cells were selected at low magnification (x10) and examined for quality under higher magnification (x63) to exclude cells with overlapped chromosomes or not clearly distinguishable chromatids. Only complete metaphases (46 centromere spots) were used for analysis. The samples were blinded to the scorer and decoded afterward. For scoring, 50 cells per donor were analyzed (QuickScan method [17]) using the morphological criteria specified by IAEA cytogenetic dosimetry manual [18]. The dicentric yields were subsequently adjusted by conversion of multicentric aberrations into the dicentric equivalent (tricentric chromosome is equal to two dicentrics, tetracentric chromosome is equal to three dicentrics, etc.). The yields of dicentrics and their distribution among cells have been used to calculate the dispersion index (σ2/y) and the normalized unit of this index (u) using the equation recommended by IAEA [18].
For dose-response curves, the yields of dicentrics (Y) was used to calculate the coefficients of the linear-quadratic mathematical function (\(Y = \alpha D + \beta {D}^{2}\), where D is a dose and α and β are the linear and quadratic coefficients, respectively) in CABAS v.2 software [19]. The uncertainties for \(\alpha /\beta\)quotients were estimated by propagating the relative errors in quadrature: \({\sigma }_{\alpha /\beta }=\frac{\alpha }{\beta }\sqrt{{\left(\frac{{\sigma }_{\alpha }}{\alpha }\right)}^{2}+{\left(\frac{{\sigma }_{\beta }}{\beta }\right)}^{2}}\), where \(\sigma\) represents the standard error.