Improving clinical brachytherapy: dosimetry & verification

Transcription

1 1/XXX Improving clinical brachytherapy: dosimetry & verification Organ Frank motion Verhaegen, in radiotherapy Gabriel Fonseca, Murillo Bellezzo, Shane White, Guillaume Landry, Brigitte Reniers nada* MAASTRO Clinic, Maastricht, the Netherlands

2 Despite much research: Current status brachytherapy of dose calculations (5 yrs after publication of TG186) Model patient Real patient 30 cm water Dose calculations based on TG43 protocol = all water

3 Shortcomings of current dose calculation standards (TG43) Fat is 40% different from water! air water? tissue water? Contrast medium water? source superposition? source shielding? applicators? radiation scatter? How do we verify the dose? From M Rivard

4 6 major improvements needed in brachytherapy 1. Heterogeneities in tissues (tissue water) -problem for low energy photons, is largest source of errors 2. Heterogeneities in sources -source design and attenuation in sources already in TG43 -intersource attenuation, applicators, shielding, metal imaging markers 3. Dose reporting -dose-to-water vs dose-to-medium -known problem in external beam radiotherapy (errors limited to a few %) 4. Dose verification 5. Relative Biological Effectiveness of low energy photons -everyone uses 1.0, but is 1.0 (especially in low-energy sources) 6. Image guidance

5 1. Heterogeneities in tissues An example in breast electronic brachytherapy TARGIT study: Electronic brachytherapy in breast radiotherapy In radiotherapy clinics, physicians (and physicists) think only in terms of dose to water Inter-patient differences are not taken into account Dose-response data is unreliable if you think you give all patients the same dose Dose vs distance from source Inserting x-ray source in breast IntraBeam electronic brachytherapy source (50 kv) Vaidya et al. Lancet 376: , 2010 Vaidya et al. Eur J Surg Oncol 28: 447, 2002

6 Strong dose heterogeneity due to gold fiducial markers in prostate (e.g. LDR boost after EBRT) Severe underdosing near gold markers Clinical relevance? Would we adapt our planning strategies? Methods Monte Carlo simulations using GEANT4 Calculate dose with and without markers Virtual water phantom Single + multiple 125 I sources 1 gold marker, various distances from source Patient cases with many LDR seeds and several gold markers Based on post-implant CT Assess magnitude of dose depression and volume of cold spots

7 OUTCOME: single sources 125 I source Au marker Single source scenarios Up to 95% dose reduction Slight dose enhancement on proximal side of marker Multiple sources: Less reduction

8 OUTCOME: clinical implants Up to 50% dose reduction (90 Gy in this example) Can modify the 100% isodose contour Cold spots (<95%) are observed D 90 was not affected 90Gy Note: dose reduction due to interseed attenuation 5% (Landry et al. Dose reduction in LDR brachytherapy by implanted prostate gold fiducial markers. Med Phys 39, , 2012)

9 Define for each voxel: material and density (segmentation) Single-energy CT: only electron density e (e /cm 3 ) Dual-energy CT: e and effective atomic number Zeff How sensitive is the dose to tissue mis-assignment? How well are tissues known? Average over population In an individual patient? With TG-43 we never have this problem (all water) 15/26

10 30/52 Variation of tissue composition Human tissues vary from one individual to the other Data in literature is scarce and old All refs trace back to: (Woodard&White, BJR 1986) Does any of this matter dosimetrically?

11 Sensitivity of dose calcs to tissue composition Assess the influence in tissue composition and its variation across the population on dose calculations

12 20/26 Simple test geometry Spherical geometries Look at variations of D with distance for various media

13 RESULTS Simulation - breast tissue (adipose + gland) D can differ from TG43 by >80% in 3cm Difference due to variation in breast composition Different low energy sources behave differently (Landry et al. Sensitivity of low energy brachytherapy Monte Carlo dose calculations to uncertainties in human tissue composition.med. Phys. 37, , 2010)

14 Simulation Dose ratio for a breast case (Pd-103) Left: From water to average breast, 30% largest effect! Right: Compositional uncertainty (1* ) among patients, ±10% This means most of the accuracy will be gained by replacing water average breast tissue (Landry et al. Sensitivity of low energy brachytherapy Monte Carlo dose calculations to uncertainties in human tissue composition.med. Phys. 37, , 2010)

15 Patient Study: electronic brachytherapy for breast AXXENT (Xoft) mini x-ray source Isotope-free brachy source 0-50 kv, low energy x-rays Tunable dose distributions Prototype #001 Source in scintillation liquid Tomato-shaped dose distribution

16 Detailed Monte Carlo modelling of Axxent source EGS++ Monte Carlo model tip water Support (Y) W-Target vacuum water Support (Y) W-Target vacuum Primary electrons Primary e- electronse- e- e- Thin transmission target Geant4 model catheter catheter -bremsstrahlung and characteristic photons produced by electron interactions in target and support -target (W): 87% -support (Y): 7% -rest: 1% -missing 5%: photo-electric characteristic photons in Yttrium support 16 Liu et al. Phys. Med. Biol. 53, 61-75, 2008

17 Electronic brachytherapy for breast S White et al, Comparison of TG-43 and TG-186 in breast irradiation using a low energy electronic brachytherapy source, Med Phys, 41/6, , 2014 PTV adipose applicator gland balloon

18 Dose ratios: heterogeneous/tg43 or heterogeneous/(water-air model) Large differences between Dwm and Dmm Ref=TG43 Loss of backscatter Differences in PTV close to skin Ref=water+air model

19 Conclusions from EBT breast study TG-43 overestimates dose to certain regions e.g. skin may permit dose escalation TG-43 underestimates dose in e.g. ribs These changes are dependent on the dose reporting method (Dwm vs Dmm; see further)

20 Better ways to distinguish tissues? Dual-energy CT (DECT) Use dual energy CT to extract ( e, Z) directly from CT images Many tissues have similar e, but different Z Use Z to distinguish the tissues Breast and prostate phantoms relative electron density theory 0.92 simulation corrected simulation effective atomic number

21 DECT segmentation vs. SECT G Landry et al, Simulation study on potential accuracy gains from dual energy CT tissue segmentation for low energy brachytherapy Monte Carlo dose calculations. Phys. Med. Biol. 56, , 2011 Reference SECT DECT Many tissues missassigned

22 RESULTS Dose errors TG43 = large errors 103 Pd is very sensitive DECT performs the best Green means correct dose

23 2. Heterogeneities in sources/applicators New dose calculation algorithms Model-Based Dose Calculation Algorithms (MBDCA) Stochastic methods: Monte Carlo (none in commercial TPS) Deterministic approaches: GBBS (Acuros, Varian) Primary/Scatter separation: CCC (Elekta-Nucletron) Monte Carlo simulation of an HDR Ir-192 source Monte Carlo simulation of an LDR I-125 seed

24 MC based dose calculation system: AMIGOBrachy (A Medical Image-based Graphical platform, G Fonseca)

25 AAPM Task Group TG-186 (2012) Mandate: Provide user guidance through multiple calculation models, issues on patient geometry, patient heterogeneity: Model-based dose calculation techniques in brachytherapy: Status and clinical requirements for implementation beyond the TG-43 formalism L Beaulieu, A Carlsson, J-F Carrier, S Davis, F Mourtada, M Rivard, R Thomson, F Verhaegen, T Wareing, J Williamson. Report of the Task Group 186 on model-based dose calculation methods in brachytherapy beyond the TG-43 formalism: Current status and recommendations for clinical implementation. Med Phys 39, , 2012

26 Guidelines of Task Group 186 (AAPM) Recommendations in TG186: Define for each voxel: material and density (segmentation) Define the dose scoring medium Provide guidelines for commissioning complex dose calcn algorithms Aims: Must maintain inter-institution consistency (as in TG43) Avoid chaos due to increased complexity Should improve estimates of outcome studies May allow better estimate of radiobiological parameter /

27 Work group on TG186 Comparing modelling of applicators for 192 Ir Monte Carlo codes: ALGEBRA, BrachyDose, Geant4, MCNP5, MCNP6, Penelope2008 Grid Based Boltzmann Solver (Varian): BrachyVision ACUROS Collapsed cone superposition/convolution (Elekta): Oncentra Brachy Advanced Calculation Engine (ACE)

28 Results Differences between MC codes are small but not zero! Differences between Acuros and ACE (up to 10%)

29 Recent work of the Work Group Differences Geant4-MNCP6 Differences ACE-MNCP6

30 3. Difference in reporting dose to water or medium MBDCA result in Dose to the real medium Three different ways of dose reporting Transport photons in water, score in water: D w,w (TG43) Transport photons in medium, score in medium: D m,m (natural way for MBDCA) Transport photons in medium, score in water: D w,m In EBRT this is not a problem TG43

31 Difference in reporting dose to water or medium: example of breast brachytherapy Left: Right: D w,m and D m,m in mean adipose tissue ( 103 Pd, 125 I, Axxent 50 kv) Ratio D w,m /D m,m differences up to 70% slightly dependent on source, varies slightly with distance from source (Landry et al. The difference of scoring dose to water or tissues in Monte Carlo dose calculations for low energy brachytherapy photon sources. Med. Phys. 38, , 2011)

32 Dose reporting method may influence clinical practice Why do we need 3 different dose reporting methods (Dw,w ; Dw,m ; Dm,m)? 103 Pd breast implant Most people agree we should transport photons in medium, but score in what? Arguments for D m,m Natural scoring method in Monte Carlo Outcome correlates better with this natural quantity? Arguments for D w,m All clinical practice is based on dose to water Most measurements are water based Cells are mostly made of water (embedded in matrix of other media)

33 Different dose reporting methods and conversion methods Cavity theory G Fonseca et al, Dose specification for 192 Ir high dose rate brachytherapy in terms of dose to water in medium and dose to medium in medium. Phys Med Biol 60, , 2015) Strong variation with energy Weak variation with energy Large Cavity Theory (LCT) uses (μ en / )w,m, assuming charged particle equilibrium (CPE) for the cavity Small Cavity Theory (SCT) uses (S/ )w,m, for Bragg-Gray cavities with dimensions much smaller than the secondary electron ranges

34 A head&neck brachy case G Fonseca et al, Dose specification for 192 Ir high dose rate brachytherapy in terms of dose to water in medium and dose to medium in medium. Phys Med Biol 60, , 2015 Variation of mean photon energy with distance from implant ( 192 Ir) Dm,m / Dw,m(LCT) Dw,m(LCT) / Dw,m(SCT)

35 But what is a large/small cavity? Small Large Burlin cavity theory (mix of small/large cavity) Value of d to be used in the cavity theory, as function of photon energy and cavity chord 125 I 192 Ir

36 4. Dose verification Using an imaging panel to capture photon fluence outside patient

37 Goal - Real Time Dosimetry Imaging Additional Steps Plan AMIGOBrachy Treatment EPID Dosimetry

38 Using a robotic arm with position sensors for accurate brachy experiments

39 Using RAMBO (Robotic Arm for Measurements in Brachytherapy and Other applications) to calibrate EPID panel 39/45

40 Meet RAMBO (Robotic Arm for Measurement in Bachytherapy and Other applications) & CARMEN (Cybernetic Arm for Radiotherapy MEasurements and Novel applications) Gabriel Fonseca

41 Dwell time and dwell position verification G Fonseca et al, Online pretreatment verification of high-dose rate brachytherapy using an imaging panel. Phys Med Biol, under review, 2017 Dwell time errors detected in real time Dwell position errors detected in real time

42 5. Relative Biological Effectiveness for low energy photons Pd-103 I-125 Axxent mini x-ray source (40-50 kv) Ir-192 AXXENT Yield of di-centric chromosomal aberrations in human lymphocytes (Hill, Rad Prot Dosim 112, , 2004) increased RBE for lower energy photons

43 Track structure calculations from microdosimetry =Monte Carlo at micron scale (or smaller)... from single collisions of 30 kev photons 1 m... from single collisions of 1.3 MeV photons (Lappa et al. TRION track structure Monte Carlo code) 1 m

44 50/52 RBE estimate from Monte Carlo modelling Step 1: MC model AXXENT Step 3: Use DNA damage model MCDS (R Stewart) Primary Electron Spectrum (AE) (a) Auger Compton water, 50 kvp photo-electric effect Step 2: Calculate photon & secondary e- spectra in media Step 4: Calculate frequency of SSB and DSB in DNA Step 5: Analyze complex DNA damage SSB yield (Gy -1 Gbp -1 ) Electron Energy (kev) DSB yield (Gy -1 Gbp -1 ) Primary Electron Energy (kev) Primary electron spectrum (kev) Step 6: Fold photon/electron spectra with damage distributions Step 7: Calculate RBE

45 Calculate SSB and DSB frequencies: MCDS (Monte Carlo Damage Simulation; Semenenko and Stewart, 2004, 2006) I) In a DNA segment: randomly distribute expected # lesions in a cell Gy -1 1) Select nucleotide pair at random from segment: {1,nseg} 2) Select one of 2 DNA strands randomly. If not damaged yet, record damage. Else goto 1 3) sb = sb -1; if sb > 0 goto 1 4) repeat 1)-3) for Bp base damages II) Subdivide lesions in segments in clusters Score SSB, DSB, complex breaks, base damage, SSB yield (Gy -1 Gbp -1 ) Electron Energy (kev) DSB yield (Gy -1 Gbp -1 )

46 Calculate RBE for Axxent source Medium SSB (Gy -1 Gbp -1 ) DSB (Gy -1 Gbp -1 ) RBE SSB RBE DSB Water 161 (189) 14.9 (8.2) 0.89 (0.88) 1.42 (1.49) Muscle 162 (188) 14.7 (8.4) 0.90 (0.89) 1.40 (1.47) Breast 161 (188) 15.0 (8.4) 0.89 (0.88) 1.43 (1.50) Breast+Ca 160 (188) 15.1 (8.4) 0.89 (0.88) 1.44 (1.51) Bone (0.1 cm) 178 (188) 11.1 (8.4) 0.99 (0.98) 1.06 (1.11) Bone (1 cm) 176 (188) 11.9 (8.2) 0.98 (0.97) 1.13 (1.19) RBE with 192 Ir ( 60 Co) as reference radiation 50 kv AXXENT source has 40% higher RBE DSB than 192 Ir (50% higher compared to 60 Co) Reniers et al. Calculation of relative biological effectiveness of a low-energy electronic brachytherapy source. Phys. Med. Biol. 53, 7125, 2008

47 Breast electronic brachytherapy study: RBE S White et al, A comparison of the relative biological effectiveness of low energy brachytherapy sources in breast tissue: A Monte Carlo study. Phys Med Biol, 61, , 2016 Density map Material map Mean photon energy map RBE map

48 Summary How to switch from water sphere to real patient geometry Low energy brachytherapy dose calcs very sensitive to tissue composition Recommendations on tissue segmentation needed Recommendations on tissue composition/assignment needed D m,m and D w,m can be very different Be aware of dose prescription Conversion needs cavity theory Recommendations on further research: tissue typing, imaging modalities (Dual Energy CT, quantitative MRI), Relative Biological Effectiveness (RBE>1.0) should be taken into account Effective dose to target may be underestimated Sensitive organ sparing may be overestimated (e.g. Axxent compared to 192 Ir) Measurement of RBE and related physical quantities (e.g. LET) needed Brachytherapy needs input from metrology!