T.Kanai, K. Yusa, M. Tashiro, H. Shimada, K. Torikai, Gunma University, Gunma, Japan

Transcription

1 The carbon-ion cancer therapy facility at Gunma University T.Kanai, K. Yusa, M. Tashiro, H. Shimada, K. Torikai, J. Koya, T. Ishii, Y. Yoshida, S. Yamada, T. Ohno, T. Nakano Gunma University, Gunma, Japan

2 Content 1. Introduction 2. Accelerator System 3. Design of SOBP 4. Treatment Planning System 5. Clinical experiences 6. Treatment of Moving Targets

3 1. Introduction

4 GHMC Main Features Facility at Gunma Heavy-ion Medical Center (GHMC) is an outcome of the NIRS R&D Program for a compact carbon facility ( ). Compact high-performance linac injector Compact synchrotron lattice Spiral wobbling Layer stacking System Treatment planning system

5 Schedule R&D at NIRS Engineering and Physics design Contract Jan Manuf. Design and Fabrication Installation and Testing Beam Test Clinical Comm st Treatment March 16 th pts Advanced medical technology Scanning Port 1~3 Restart of treatment

6 GHMC Gunma University 3 treatment rooms, 4 beam lines A: Horizontal BL B: Horizontal + Vertical BL C: Vertical BL D: R & D for scanning

7 More compact! Less expensive! NIRS (1994-) 120m 65m Gunma University (2010-) 2001 Concept of the project 2004 Start Collaborations with NIRS 135,000,000 USD 45m HIMAC 1/3 cost/size GHMC 65m

8 2. Accelerator System

9 GHMC Specifications Ion type Energy Maximum range SOBP Maximum aperture Lateral field shaping Beam intensity Dose rate Beam delivery modes :Carbon : MeV/u :25 cm :2 14 cm (with ridge filters) :15 cm x 15 cm :Single radius wobbling Spiral wobbling :1.3 x 10 9 pps (3 sec nozzle :5 GyE/min for φ15 cm 10 cm SOBP :Respiration gating Layer Stacking

10 Accelerator System 66.0m C MeV/u max 1.3 x 10 9 pps 38.4m Synchrotron Extraction RF Cavity ECR C 4+ RFQ APF-IH C 4+ Foil C 6+ 4 MeV/u Injector

11 Injector Linac

12 Synchrotron

13 HEBT Line

14 Generate and accelerate C 4+ Convert to C 6+ by stripping foil ECR Ion Source Injector Linac Plenty of beam > 300 eµa C 6+ at injector exit after stripping No HV deck! RFQ Linac Optimized for C 4+ APF-IH Linac E-field focusing No magnets! Highly efficient RF power 1.3 MW 0.5 MW Stable operation ECR (10 kev/u) No HV Deck! RFQ (0.6 MeV/u) 10.6 m 6.2 m APF-IH (4.03 MeV/ u ) Extensive E, B field calculations! Manufactured by SHI

15 Synchrotron Compact lattice by NIRS (modified FODO) Advanced beam simulation code Know-how to achieve the design beam current Slow extraction process (MELCO process) Features Main dipoles are almost rectangular Very low-noise power supplies Wide-band (non-resonant) RF accelerating cavities Very stable and reproducible system No daily tuning is needed Synchrotron: main parameters Circumference m 63.3 Dipole field T 0.134~1.477 Momentum compaction factor Long straight section m 3.1 Ion C(6+) Injection energy MeV/u 4 Extraction energy MeV/u 140~400 Number of particles pps 1.3 x 10 9 Cycling period sec > 3 Accelerating RF frequency MHz 0.875~6.769 Harmonic number 2 Max. RF voltage kv ~0.3429

16 Synchrotron Commissioning 24 Aug Start of synchrotron commissioning 25 Aug Acceleration up to 400 MeV/u 30 Aug First extraction 5 Sep First beam transported to treatment room-a 30 Sep Beam transport to all treatment rooms 26 Dec Beam transport at all scheduled energies Beam commissioning was conducted in a very short period. Commissioning is based on a systematic use of computer modeling, magnetic field data and beam measurements. And a thorough understanding of the dynamics. Circulating beam current 1 sec Beam spill Computer model Magnetic field measurement (factory data) Systematic beam measurement data Operating parameters Building is also an important part of the performance

17 Synchrotron Commissioning 0.4 sec 1.8 sec Beam gate Beam spill Beam is turned on/off for respiration gating Beam extraction timing is variable for each pulse Circulating beam current Beam gate 0.4 s/div Multiple beam on/off per pulse Beam shut-off time < 1ms Beam spill

18 Treatment Beam Lines 3 treatment rooms, 4 beam lines A: Horizontal B: Horizontal + Vertical C: Vertical m Beam monitors Wobbler magnets scanning speed 57Hz ~50 mm/msec at isocenter Ridge filters, Range shifters Flatness monitors, X-ray tube MLC 3.75mm leaf pitch 40 pairs X-ray DR 2 orthogonal views Treatment Room B

19 Spiral Wobbling * Advantages compared to the conventional broad beam methods Smaller range loss due to thinner scattering foil More homogeneous lateral distribution Higher beam utilization efficiency Yg [mm] Wobbling Time :20ms Yg [mm] Wobbling Time :200ms Yg [mm] Wobbling Time :1000ms Fast wobbling magnets Pencil beam: σ=2.5cm Relative Dose Xg [mm] 3.00E+00 Wobbling Time : 20ms 2.50E E E E E E Lateral Position Xg [mm] Relative Dose Xg [mm] 1.20E+01 Wobbling Time : 200ms 1.00E E E E E E Lateral Position Xg [mm] Relative Dose Xg [mm] 6.00E E E E E E+01 Wobbling Time : 1000ms 0.00E Lateral Position Xg [mm] Testing has been completed. Preparing for the Government medical approval process. * Optimization of Spiral-Wobbler System for Heavy-Ion Radiotherapy, M. KOMORI, T.FURUKAWA, T.KANAI, K. NODA Journal Title;Jpn J Appl Phys Part 1, VOL.43;NO.9A; pp (2004).

20 Stacked-Layer Conformal Treatment Kanai et al., Med. Phys. 10, (1983) Longitudinally divide the target slices Conform thin layer of SOBP (mini-peak) to each slice variable SOBP Bolus is used to form the distal layer Multi-Leaf Collimator is used to form the proximal part of the target beam Testing to be completed in mm pitch

21 3. Design of SOBP

22 NIRS method for SOBP design Assumption 1 (pragmatic approach) HSG cell represents all of tumor cell MeV/u; SOBP 60mm RBE=1 4 Clinical kev/µ m 3 Biological 4 2 Physical 77keV/µ m D 10 for X-rays (Gy) Depth in Water (mm)

23 SOBP RBE for Neutron therapy Photon _ Eq. D( z) (GyE) = = ( scaling ( scaling _ factor) RBE _ factor) RBE HSG HSG RBE at SOBP center ( d, z) d SOBP ( z) d ( d, center) HSG SOBP HSG dsobp ( s, center) d ( s, z) Bio. Response relative to SOBP center SOBP ( z) Fix to survival level of 10 % GyE (NIRS) HSG HSG d X (0.1) dsobp (0.1; center) 1.45 dsobp ( z) HSG HSG d (0.1; center) d (0.1; z) SOBP SOBP

24 Dose dependence of the biological response relative to SOBP center d HSG SOBP HSG dsobp ( s, center) ( s, z) 1.5 z d HSG SOBP HSG dsobp (0.1; center) (0.1; z) mm +20 mm +10 mm 0 mm -10 mm -20 mm -25 mm Except very near the distal part of the SOBP, the variation is small Physical Dose at Center of SOBP = d HSG of reference carbon beam(gy) SOBP (center)

25 Variation of sensitivity estimated by TCP analysis. TCP = exp N i 1 exp 2π α exp nαd ( α i α ) 2( α ) d ( ) ( T T k ) α / β Fixed Parameters 2 2 T p α=0.75,β=0.076, α=0.15, T k =0, T p =7 1 tentative 1.2 TCP can be analyzed by LQ model introducing the variation of the sensitivity α, α TCP for Non Small Cell Lung Cancer Optimal Dose 25.2 Gy (60 Gy (CRx); T2) 22.2 Gy (52.8 Gy (CRx); T1) O ptim al D ose Gy (72 Gy (CRx)) ~95% TCP Frac. 9 Frac Fx 9Fx Fx 0.2 1Fx Physica l Dose (Gy) Total Physical Dose at center of SOBP 60 mm(gy)

26 Biological uncertainties: We have to introduce variation in radiation sensitivity, (α± α) in TCP analysis with LQ model. There will be large uncertainty for survival fractions below d % 1-5% 0.8 ( s; center) ( s; z) HSG SOBP HSG dsobp d d SOBP 350 MeV/u, SOBP 60mm ( z) ( center) SOBP Biological dose distribution used in TPS is in the uncertainty region of tumor responses (α± α) for various fractionation size F 9F 12F 16F α+0.15 α F 12F 9F 4F alpha-0.15 alpha Depth in Water (mm)

27 At present stage of carbon therapy It can be justified to use fixed SOBP for various fractionation size. Detail analysis of carbon clinical trial should be required in order to step further. Tumor responses and normal tissue responses

28 Physical Aspect of SOBP Design LET distribution in water phantom should be taken into account for SOBP design Fragmentations produced by aluminum ridge filter should be taken into account. C 290 M ono-energy m easured by 0.2 m m P lastic S cintillator Ridge Filter Design Shift Aluminium (Ridge Filter) C hanneln um ber Depth in Water (mm)

29 Measurements of PDD Depth dose distributions; 350 MeV/n Roos Chamber Monte 350 MeV/n; Geant4 Pencil Beam Calro Al0mm2 Al10mm Al20mm Al30mm Al40mm Al50mm Al60mm total total-10mm total-20mm total-30mm total-40mm total-60mm Depth in Water (mm) Depth in Water(mm) Aluminum Ridge Filter Position isocenter Wobbler M. SAD 15x15cm: Full Open 20cm water We adjusted fragmentation yield.

30 Physics to biology 3.5 LQ model parameter, α and β OER (He) OER (C) a(hsg)forc b(hsg)forc a(hsg)forhe b(hsg)forhe OER(HSG)forC OER(HSG)forHe α (He) All fragmented particles are assumend to be alpha particles 1 α (C) LET (kev/µm) β

31 5 Phys.Dose(350) Bio.Dose(350) Phys.Dose(380) Bio.Dose(380) 4Phys.Dose (290) Bio.Dose (290) SOBP 80, 290,350,380 MeV/n 3 2 SOBP Center E Phys.D Bio.D Depth in Water (mm)

32 1 Survival Fraction of HSG cells (for carbon 350 MeV/n, SOBP 80 mm beam) Survival curve of HSG cells in GHMC (350 MeV/n, SOBP 80 mm) Surviving Fraction Surviving Fraction [%] mm +30 mm Center Proximal Distal Physical Dose (at different Dose [Gy] SOBP points) [Gy]

33 4. Treatment Planning System, Xio-N

34 System structure Treatment Planning plat form (Xio) XiO-N system Dose calculation code Connection software Calc. Engine K2Dose System Data DICOM-ION Server

35 Specification in calculation Pencil Beam algorithm Heavy-ion specific modeling QA Plan Planning for new wobbling methods Single Circle Conventional wobbling Layerstacking Spiral wobbling

36 Pencil beam method Algorithms - Broad beam downstream of aperture is divided small gauss shape beams which are transported through the inhomogeneous structure - Scattering at each thickness of ridge filter is also considered - Dose at a grid is summed up and distribution results in couple of minutes

37 Heavy Ion specific modeling RBE modeling -From LET (or depth) dependence of α & β of LQ model, clinical dose distributions are calculated so as to give 10 % survival level of HSG cell at the center of the SOBP. β[gy -2 ] α[gy -1 ] Range[mmH 2 O] HSG d X (0.1) GyE( NIRS) = 1.45 dsobp ( z) HSG d (0.1; z) SOBP

38 QA Plan TPS for patients TPS for QA Plan Comparison Measure Dose distributions in water phantom

39 Beam Modeling Check SAD check Lateral penumbra dependence -Degrader thickness -Collimator position etc.

40 Dose calculation check 1.5 distal-xion SOBPcenter-XioN proximal-xion DistalC380SOBP80-PinPoint CenterC380SOBP80-PinPoint ProximalC380SOBP80-PinPoint C380SOBP80-PinPoint XioN Relative Dose Relative Dose Lateral (mm) depth (mm)

41 Dose distribution check Check using QA Phantom and QA plan 41

42 5. Clinical Experiences

43 Initiation of the clinical protocols at Gunma University Many of phase I/II dose escalation and phase II studies for various tumor sites have been carried out at NIRS since Although promising clinical outcomes have been reported from NIRS, it is of interest whether the efficacy of carbon ion radiotherapy from a single institution can be reproduced in other facilities when optimal doses and fractionations are used for a similar patient population. At GHMC, the efficacy and safety of carbon ion radiotherapy were reviewed for each tumor type, and then the currently best available dose and fractionation schedules determined at NIRS were adopted for our clinical protocols.

44 Dose & fractionation at Gunma University Sites Total dose (GyE) Fractionation (Fr) Duration (Week) Head & Neck Lung (stage I) 52.8 or Liver Prostate Sarcoma 64 or Rectum (pelvic rec.)

45 Clinical schedule at Gunma University In 2010, 92 patients have been treated (10-15 patients per day) March 2010.June 2010.June 2010.July 2010.September 2010.September Prostate Lung (stage I peripheral type) Head & Neck cancer (non-squamous) Hepatocellar carcinoma Pelvic recurrence of rectal cancer Bone & Soft tissue sarcoma

46 6. Moving Target

47 Strategy for Mobile Organs Use of 4D-CT Definition of margins (PTV) for each patient Confirmation of dose distributions in the gating period (Gate-In, Exhale Peak, Gate-Out) Range Compensator (Bolus) design Smearing and/or IGTV (GTV + IM) methods To do the above, we utilize our present systems for routines. (Xio-N, Focal4D)

48 Picture of Lung Treatment (Gating) Respiration (Respiratory Wave) Beam Beam Beam Beam Motion of the skin at thoraco-abdomen is monitored by a laser sensor as respiration (respiratory wave), and carbon-ion beam is irradiated at the gated timing corresponding to the exhalation.

49 Picture of Lung Treatment(4 Ports) Roll +20 Roll -20

50 Treatment Flow CT Simulation Room Acquire Resp. Wave Treatment Planning Room Gated CT 4D-CT X-ray Fluoroscopy Temporal TP (Temp. Iso-Center ) Gated X-ray (Reference images for daily setup) (This can be done after TP approved as Rehearsal. ) Definition of IM (Focal 4D) Treatment Planning (Xio-N) Irradiation Room Positioning using X-ray images (Confirm reproducibility) Gated Irradiation

51 Treatment Planning Target Volume Definition (Lung) GTV IGTV PTV 8mm CTV IM+SM (each direction) Bolus Design Smearing method Smearing = TM (Perpendicular to beam) CTV = GTV + 8mm (exclude the outside of lung) PTV = CTV + TM Total Margin 2 TM = IM + SM In case of IGTV method (SM = 3mm) After PTV was made GTV is extended IGTV = GTV + IM (force CT=1 for IM region) From the measured motion at 0-30%Lv, 1/3 of the motion is added for IMs. Beam TM 2 Smearing method IGTV method IGTV(GTV + IM) ForceCT = 1 (tumor value) Smearing = SM After bolus design, the IGTV CT value is returned to the original (raw CT values). Beam IGTV method IGTV=GTV+IM

52 Treatment Plan (Lung) 52.8GyE/4Frac

53 Smearing & IGTV Smearing method (1) IGTV method (2) (1) - (2)

54 Dosimetric Validation QA Plan 1,8E-04 1,6E-04 1,8E-04 1,6E-04 Plan1Beam2 Z=80mm Plan2Beam1 z=60mm Plan2Beam1 z=50mm 1,4E-04 1,4E-04 Dose (Gy/Count) 1,2E-04 1,0E-04 8,0E-05 6,0E-05 Dose (Gy/Count) 1,2E-04 1,0E-04 8,0E-05 6,0E-05 4,0E-05 2,0E-05 0,0E+00 Plan1Beam2 Plan2Beam Depth (cm) 4,0E-05 2,0E-05 0,0E Off Center (cm)

55 Analysis of Patient Positioning From the values of the differences, set-up errors (positioning) and internal motion errors are evaluated for the validation of our strategy of margins & TP.

56 Error Distributions of Interest Points (7 Lung Pat.) X Y Z Y X Y Z Y (mm) (mm) Set-up Errors (Positioning) Errors of Metal Markers (Internal Motion Errors) Frontal Lateral Frontal Lateral X(LR) Y(SI) Z(AP) Y(SI) Y All X(LR) Y(SI) Z(AP) Y(SI) Y All Average Median Standard Deviation (s) Range Minimum Maximum Number of Sample σ (Unit: mm) Exclude Set-up Errors ->

57 Error Distributions of Interest Points (7 Lung Pat.) Set-up Error (Positioning) is about 1 mm at 2σ. Marker Errors does not always correspond to the internal motion errors of tumors mm Max & 7 mm at 2σ for Inferior Direction. Excluding 1 patient having large errors, 6 mm Max & 3 mm at 2σ. Additional Margin (1/3 * IM) is max 3 mm.

58 Summary(1) 1. Commissioning (beam and clinical) was accomplished in a very short time. 2. Treatments have been safely started from March Machine is operating very smoothly and we treated 92 patients last year. 4. We designed therapeutic carbon beams based on NIRS experiences of clinical trials. 5. It can be justified to use fixed ridge filter for various fractionation sizes. 6. We introduced Geant 4 Monte Carlo simulation for calculating LET distributions of fragmented particles produced in the ridge filter and water phantom.

59 Summary(2) 7. A new treatment planning system Xio-N has been used at Gunma University after careful commissioning of the system. 8. Motion is measured for each patient by 4DCT to determine margins. 9. From the analysis of X-ray images of patient positioning, set-up and Internal Margins (adding 1/3 of motion) are considered to be almost reasonable. 10. Further investigation is required.

60 Thank you for your attention 60