Update of the CLRP eye plaque brachytherapy database for photon-emitting sources
Habib Safigholi, Zack Parsons, Stephen G. Deering, Rowan M. Thomson
UUpdate of the CLRP eye plaque brachytherapydatabase for photon-emitting sources
Habib Safigholi, Zack Parsons, Stephen G. Deering, and Rowan M. ThomsonCarleton Laboratory for Radiotherapy Physics, Department of Physics,Carleton University, Ottawa, Ontario, K1S 5B6, Canadaemail: a) safi[email protected] and b) [email protected]
AbstractPurpose:
To update and extend the Carleton Laboratory for RadiotherapyPhysics (CLRP) Eye Plaque (EP) dosimetry database for low-energy photon-emittingbrachytherapy sources using egs brachy , an open-source EGSnrc application. Theprevious database, CLRP EPv1, contained datasets for the Collaborative OcularMelanoma Study (COMS) plaques (10-22 mm diameter) with
Pd or
I seeds(
BrachyDose -computed, 2008). The new database, CLRP EPv2, consists of newly-calculated 3D dose distributions for 17 plaques [8 COMS, 5 Eckert & Ziegler BEBIG,and 4 others representative of models used worldwide] for
Pd,
I, and
Cs seeds.
Acquisition and Validation Methods:
Plaque models are developed with egs brachy , based on published/manufacturer dimensions and material data. TheBEBIG plaques (modelled for the first time) are identical in dimensions to COMSplaques but differ in elemental composition and/or density. Previously-benchmarkedseed models are used. Eye plaques and seeds are simulated at the centre of full-scatter water phantoms, scoring in (0.05 cm) voxels spanning the eye for scenarios:(i) ‘HOMO’: simulated TG43 conditions; (ii) ‘HETERO’: eye plaques and seeds fullymodelled; (iii) ‘HETsi’ (BEBIG only): one seed is active at a time with other seedgeometries present but not emitting photons (inactive); summation over all i seeds ina plaque then yields ‘HETsum’ (includes interseed effects). For validation, doses arecompared to those from CLRP EPv1 and published data. Data Format and Access:
Data are available at https://physics.carleton.ca/clrp/eye_plaque_v2 , http://doi.org/10.22215/clrp/EPv2 . The data consist of3D dose distributions (text-based EGSnrc “3ddose” file format) and graphical presen-tations of the comparisons to previously published data. Potential Applications:
The CLRP EPv2 database provides accurate reference3D dose distributions to advance ocular brachytherapy dose evaluations. The fully-benchmarked eye plaque models will be freely-distributed with egs brachy , supportingadoption of model-based dose evaluations as recommended by TG-129, TG-186, andTG-221.
Key words:
CLRP, eye plaque, dose calculation, brachytherapy, egs brachy i a r X i v : . [ phy s i c s . m e d - ph ] F e b LRP EPv2 dosimetry Database page 1 I . Introduction Eye plaque (EP) brachytherapy plays an important role in the treatment of intraocularcancers offering delivery of conformal doses to the tumour, with steep dose gradients sparingcritical organs at risk.
Survival rates with plaque therapy have been reported to becomparable to enucleation but with eye preservation and the possibility of retaining somevisual function.
4, 5
Recent work suggests inferior clinical outcomes with proton therapycompared with eye plaque brachytherapy. The small size of the eye combined with considerable dose gradients for eye plaquebrachytherapy mean that dosimetry is critical.
1, 2
Traditionally, dose calculations for plaquescontaining photon-emitting sources follow the water-based approach of Task Group (TG) 43. Model-based dose calculation algorithms (MBDCAs), including Monte Carlo (MC) simula-tions, promise more accurate dose evaluations by accounting for non-water treatment compo-nents (applicator, sources) and patient non-water tissues. Various MC studies have demon-strated considerable errors incurred with the TG-43 approach for eye plaque brachytherapydose evaluation. For example, MC studies focusing on modelling plaque backing/insertmaterials (in water phantom) report dose decreases relative to TG-43 of 11 to 40% in thetumour, and as large as 90% in organs at risk for the standardized plaques of the Collab-orative Ocular Melanoma Study (COMS) containing
I or
Pd seeds.
Considerabledifferences with TG-43 have been demonstrated for other photon-emitting plaque modelsas well as when modelling non-water patient/ocular anatomy.
10, 12–14
These discrepancieswith TG-43 doses motivated both TG-129 and TG-221 to recommend that dose evaluationsaccounting for the effects of the plaque backing and insert be carried out in parallel withtraditional TG-43 calculations,
1, 2 in accord with the general recommendations of TG-186 onadoption of MBDCAs in brachytherapy. The present work supports widespread adoption of MC dose evaluations for photon-emitting eye plaques. We use egs brachy , a freely-distributed and open-source EGSnrcapplication,
16, 17 to develop 17 eye plaque models: 10-24 mm diameter COMS, 12-20 mmBEBIG (manufactured by Eckert & Ziegler BEBIG, Berlin, Germany), and four 16 mmdiameter plaques representative of various models in use worldwide. Employing bench-marked egs brachy I, Pd, and
Cs seed models, we simulate the eye plaques ina water phantom to generate 3D dose distributions scored in (0.05 cm) voxels in the eyeregion.LRP EPv2 dosimetry Database page 2These 3D dose distributions are used to completely update and extend the Carleton Labo-ratory for Radiotherapy Physics (CLRP) Eye Plaque Brachytherapy database that was orig-inally published in 2008 (version 1, “CLRP EPv1”: contained BrachyDose -calculated dosedistributions for 10-22 mm COMS plaques with
Pd or
I seeds ). The present articledescribes the CLRP EPv2 database https://physics.carleton.ca/clrp/eye_plaque_v2 ( http://doi.org/10.22215/clrp/EPv2 ) that contains 3D dose distributions for diversephoton-emitting eye plaques, and the egs brachy plaque models that will be freely-distributed with the egs brachy distribution ( https://physics.carleton.ca/clrp/egs_brachy/ ). II . Acquisition and Validation Methods II . A . Monte Carlo simulations of eye plaques All MC calculations are performed with EGSnrc application egs brachy (GitHub com-mit hash 8166234, 2020, available at https://github.com/clrp-code/egs_brachy/tree/egs_brachy_2020 ). The benchmarking of egs brachy is documented in previous publica-tions. Transport parameters are generally EGSnrc defaults , using the low-energydefault specifications distributed with egs brachy . Electron transport is not modelled. Thephoton energy transport cutoff is set to 1 keV. Photoelectric absorption, Rayleigh scattering,Compton scattering, and fluorescent emission of characteristic x rays are simulated. Photoncross sections are from the XCOM database. Dose is approximated as collision kerma,scored with a tracklength estimator in voxels with mass energy absorption coefficients (dis-tributed with egs brachy ; previously calculated with EGSnrc application g ). The “un-renormalized” photoelectric cross sections are used, consistent with EGSnrc default (notethat there is ambiguity in whether renormalized or unrenormalized Scofield photoelectriccross sections are in better agreement with experimental data ).The egs++ class library geometry module is used to develop the eye plaque models:COMS plaques with diameters ( D ) from 10 to 24 mm (in 2 mm increments),
2, 23
BEBIGplaques with diameters from 12 to 20 mm (in 2 mm increments; simulated for the firsttime), and four different 16 mm diameter “representative” plaque models.
1, 14
These rep-resentative plaque models were previously developed by Lesp´erance et al to approximatedifferent plaque models in use worldwide (for which exact dimensions and material specifica-LRP EPv2 dosimetry Database page 3tions are not widely available and/or accurately known), and include: “Short lip-acrylic”(Sla), “COMS-thin acrylic” (Cta), “No lip-Silastic” (NlS), “Stainless steel-acrylic”(Ssa). The plaque models and associated parameters are summarized in table 1 with pa-
Table 1 rameters/dimensions defined in figure 1; the online CLRP EPv2 database contains diagrams
Fig 1 of each plaque model.Table 1:
Summary of plaques modelled: reference notation; characteristics of the backing andinsert; values for diameter ( D ; does not include lip width if present), collimating lip height ( h lip ,if lip present), plaque height ( h ), and radial distance from the centre of the eye to seed centres( R seed ) – see Fig. 1. Dimensions (mm)Plaque model Backing (thickness) Insert Radionuclides
D h h lip R seed Ref.COMS Modulay (0.5 mm) full, Silastic I, Pd,
Cs 10 −
24 2.75 2.7 13.7
BEBIG BioPontoStar (0.5 mm) full, Silastic
I 12 −
20 2.75 2.7 13.7 ∗ COMS - thin acrylic (Cta) Modulay (0.5 mm) thin 0.85 mm, acrylic I, Pd,
Cs 16 — 2.7 13.7
Short lip - acrylic (Sla) Modulay (0.5 mm) full, acrylic I, Pd,
Cs 16 1.8 1.5 12.95
No lip - Silastic (NlS) Modulay (0.5 mm) full, Silastic I, Pd,
Cs 16 1.8 — 12.95
Stainless steel - acrylic (Ssa) stainless steel (1 mm) full, acrylic I, Pd,
Cs 16 2.75 2.1 13.45 ∗ Personal communication, Michael Andr´assy (Eckert & Ziegler BEBIG) June 18, 2019 x z R s eed h li p h eye center (0, 0, 0) . c m C AX D Figure 1:
Schematic diagram depicting the eye plaque. Indicated are parameters de-scribing the eye plaque ( D : diameter; h : height; h lip : collimating lip height) andshowing R seed as the distance from seed centre to the eye centre, as well as the coordi-nate system (origin at the inner sclera, 0.1 cm from idealized eye’s outer surface; CAX:central axis). Each eye plaque model has a backing (with or without collimating lips), and containseither a full insert (that conforms to the outer sclera) or a thin layer of fixative that holdsseeds in place. The COMS plaques have a Modulay backing (elemental composition by mass:77% Au, 14% Ag, 8% Cu, and 1% Pd; mass density, ρ = 15.8 g/cm ) with collimating lipsand a full Silastic insert seed carrier (39.9% Si, 28.9% O, 24.9% C, 6.3% H, 0.005% Pt; ρ =1.12 g/cm ). Compared to the COMS plaques, the BEBIG plaques have identical geometryLRP EPv2 dosimetry Database page 4but with a different gold alloy comprising the backing, “Bio PontoStar” (87% Au, 10.6% Pt,1.5% Zn, 0.2% Rh, 0.2% In, 0.2% Ta, and 0.02% Mn, ρ = 18.8 g/cm ), and the Silastic inserthas a different mass density of ( ρ = 1.09 g/cm [Personal communication, Michael Andr´assy(Eckert & Ziegler BEBIG), June 18, 2019]). The representative plaque models (Cta, Sla,NlS) have Modulay backings, with the exception of the plaque model with the stainless steelbacking (Ssa; 99.05% Fe, 0.005% Mn, 0.003% Si, and 0.0015%, ρ = 7.9 g/cm ); their insertsare either Silastic or acrylic (32% O, 60.1% C, 8.06% H, ρ =1.19 g/cm ). All representativeplaques have collimating lips (but of varying lengths, h lip - see table 1) with the exceptionof “No lip -Silastic” (NlS).All plaques models are based on fitting an idealized eye assumed to be a sphere of radius1.23 cm, i.e. , for the eye plaque models with a full insert (COMS, BEBIG, Sla, NlS, Ssa),the insert conforms to the eye’s outer sclera (radius 1.23 cm). The “eye plaque” coordinatesystem is used
2, 10, 14 and has its origin at the inner sclera on the plaque’s central axis, takento be 0.1 cm from the outer sclera (figure 1). The plaque’s central axis (CAX) defines the z -axis, with x and y coordinates transverse to the plaque. Plaques contain between 5 and 33seeds, with positions and orientations provided in the online CLRP EPv2 database based onearlier publications: TG-129 report (COMS 10 - 22 mm; BEBIG), Cutsinger et al (COMS24 mm), and Lesp´erance et al (representative plaques: Cta, Sla, NlS, Ssa).Three different seeds are used for COMS and representative plaque simulations: the Pd Theragenics TheraSeed ® model 200,
18, 28 125
I Amersham OncoSeed model 6711,
18, 29 and
Cs Isoray model CS-1 Rev2 ® .
18, 30
For BEBIG plaques, the ophthalmic
I BEBIGIsoSeed ® I-125 (I25.S16) is used which is geometrically identical to the I25.S06 but hashigher activity for use in ophthalmologic oncology.
18, 31
The egs brachy models of theseseeds were recently developed and benchmarked as part of updated 2020 CLRP TG-43v2database. Photons are initialized within the seeds according to the NNDC spectra for Pd and
Cs seeds, and for
I from the NCRP report 58, all consistent with CLRPTG43v2 database. The phase space source, particle recycling and other variance reductiontechnique features of egs brachy (that enhance simulation efficiency) are not used. Themean energy of photons emitted from the seeds calculated by egs brachy are 20.51 keV(
Pd model 200), 27.34 keV (
I model 6711), 30.29 keV (
Cs model CS-1), and 28.16keV (I25.S16). Plaques and seeds are modelled at the centre of a full-scatter water phantom ( ρ =0.998 g/cm ) which extends from -15 cm ≤ x, y, z ≤
15 cm. Dose is scored in a 51 × × voxels spanning the eye region, extending from -1.275 cm ≤ x, y ≤ ≤ z ≤ egs brachy ’s voxel volumecorrection for which 10 random points/cm is used.
16, 18
Different scenarios are simulated, all with seeds fully modelled:1. ‘HOMO’: Simulated TG-43 conditions with the plaque backing/insert modelled aswater and no interseed effects ( egs brachy in ‘superposition’ run mode).2. ‘HETERO’: Eye plaques containing seeds are fully modelled in the water phantom.3. ‘HETsi’ (BEBIG plaques only): One seed is modelled at a time, with other seedgeometries present but the seeds inactive. This produces one 3D dose distribution foreach seed position in a plaque. This is repeated for all seeds to enable superposition(accounting for seeds of possibly differing source strengths) to obtain ‘HETsum’ whichis the sum of all HETsi (includes interseed effects).Each calculation involves simulation of 10 photon histories to ensure type A statisti-cal uncertainties ≤ z = 2 .
26 cm (along the central axis at the oppo-site side of the eye to the plaque). Doses are reported in terms of dose rate per unitseed air kerma strength (Gy h − U − where 1 U=1 cGy cm h − ) by dividing the cal-culated MC dose per history by the seed air kerma strength per history ( S histK ). The S histK values were previously calculated for the NIST WAFAC detector geometry as partof the CLRP TG43v2 database as 6 . × − Gy cm /hist ( Pd model 200),3 . × − Gy cm /hist ( I model 6711), 4 . × − Gy cm /hist ( I I25.S16),and 3 . × − Gy cm /hist ( Cs model CS-1)( https://physics.carleton.ca/clrp/egs_brachy/seed_database_v2 ). As an example of the calculation time for a clin-ical scenario: simulation of a COMS 16 mm plaque fully loaded with 13
I seeds (model6711) requires 200 s to achieve 2% statistical uncertainty at the tumor apex ( z = 0 . https://physics.carleton.ca/clrp/3ddose_tools/ LRP EPv2 dosimetry Database page 6 ), e.g. , to extract doses along different axes (CAX: central axis, z ; transverse x, y for z = 0 . , . II . B . CLRP EPv2 data overview Table 2 and figure 2 summarize CLRP EPv2 data for all 17 plaques loaded with the different
Table 2Fig 2 seeds, with additional data provided online. In general, considerable dose reductions forHETERO relative to HOMO doses are observed ( e.g. , Table 2), due to the attenuation andscattering in the backing/insert material. Doses decrease considerably with position alongthe plaque central axis, with steeper fall-off for lower energy
Pd (20.51 keV) comparedwith higher energy
I [27.34 keV (6711), 28.16 keV (I25.S16)] or
Cs (30.29 keV) seeds.These general observations are consistent with earlier work.
The lowest dose per unitseed air kerma strength along central and transverse axes is for 10 mm COMS which is thesmallest-diameter plaque; conversely, the highest values are for 24 mm COMS, the largest-diameter plaque (figure 2). In general, HOMO and HETERO dose rates per unit seed airkerma strength for BEBIG plaques are greater than those for the corresponding COMSplaques.All representative plaque models generated greater dose rates per unit seed air kermastrength compared to the 16 mm COMS or BEBIG plaques with the lowest dose rate forCta and highest dose rate for Sla plaque. Also, the transverse dose profiles are relatively flat(near the central axis; 0 to 0.8 cm) for plaques with higher seed number capacity (22 and 24mm) due to the seed distribution in the plaque.LRP EPv2 dosimetry Database page 7Table 2:
Summary of CLRP EPv2 HETERO/HOMO dose ratios at depths along the central axisfor each plaque fully-loaded with
I (model 6711 or I25.S16 for BEBIG),
Pd (model 200), or
Cs (CS-1 Rev2) seeds. The zero before the point is omitted ( e.g. , .
889 = . ). Statisticaluncertainties are less than 0.2%. Values of HOMO and HETERO dose rate per unit seed airkerma strength (Gy U − h − ) are provided in the CLRP EPv2 webpage. COMS (mm) BEBIG (mm) RepresentativeNuc z (cm) 10 12 14 16 18 20 22 24 12 14 16 18 20 Cta Sla Ssa NlS0.0 .889 .875 .867 .852 .853 .848 .854 .844 .879 .871 .857 .857 .852 1.00 1.02 .964 .9220.1 .897 .887 .880 .870 .866 .862 .863 .857 .885 .878 .869 .864 .861 .995 1.01 .963 .9340.2 .897 .890 .884 .878 .872 .870 .869 .864 .884 .879 .873 .867 .864 .988 1.00 .961 .938 I Pd Cs L R PEP v d o s i m e tr y D a t a b a s e p ag e z / cm do s e / a i r k e r m a / G y h - U - CLRP_EPv2, HETERO, CAX value mm mm
14 mm12 mm 20 mm z = . c m a)
18 mm22 mm16 mm
Solid lines:
I(125.S16)_BEBIGDash/dot lines:
I(6711)_COMS z / cm do s e / a i r k e r m a / G y h - U - CLRP_EPv2, HETERO, CAX value
NISSsaCtaSla z = . c m b) I(6711)_Representative x / cm do s e / a i r k e r m a / G y h - U - CLRP_EPv2, HETERO, x profile, z = 0.5 cm
10 mm 14 mm16 mm mm mm
22 mm mm mm c) x / cm do s e / a i r k e r m a / G y h - U - CLRP_EPv2, HETERO, x profile, z = 0.5 cm
I(6711)_Representative
NlSCtaSsa
Sla d) Figure 2: Overview of CLRP EPv2 HETERO results (for
I): doses along the central axis (CAX; z -axis) for (a) COMS and BEBIGand (b) representative plaques; doses along the transverse axis ( x , restricted to x ≥ z = 0 . ≤ II . C . Data validation For validation, dose distributions (from 3ddose files: dose rates per unit seed air kermastrength, see section II . A .) are compared to previously-published doses computed with BrachyDose data from CLRP EPv1 for 10 - 22 mm COMS with I, Pd
10, 11 and 16 mmCOMS with
Cs (Ref. ); MCNP data
8, 9 for 10 - 22 mm COMS plaques ( I, Pd,
Cs);
BrachyDose data for representative plaques. It is not possible to directly compare resultsfor BEBIG and 24 mm COMS plaques because the present study is the first published MCsimulation of these plaques (but results for these plaques are compared to the other plaquesizes/types). The following subsections summarize some of the comparisons, including dosesalong central and transverse axes, as well as at points of interest in the eye (on the centralaxis: sclera, optic disc, posterior pole, fovea, eye center, lens, lacrimal gland for 8 differ-ent plaque positions of the COMS 16 mm plaque, following the work of Rivard et al ).Doses are compared directly and also via the the percent difference of doses for egs brachy ( D egs brachy ) and another code ( D code , from BrachyDose or MCNP ):%∆( egs brachy , code) = D egs brachy − D code D egs brachy × . (1)The following subsections provide a sample of the comparisons carried out, demonstrat-ing overall that egs brachy dose distributions are in excellent agreement with previouslypublished results.LRP EPv2 dosimetry Database page 10 II . C . . COMS plaques Dose distributions generated with egs brachy for COMS plaques compare well with pub-lished
BrachyDose and
MCNP data, as demonstrated by the overview of comparison resultspresented in this section.Figure 3 presents data for some COMS plaques along the central and transverse ( x ) axis, Fig 3 focusing on HETERO results; Figure 4 summarizes HETERO dose differences over all COMS
Fig 4 plaque sizes, comparing egs brachy with
BrachyDose
10, 14 and
MCNP . Across all plaque sizes, egs brachy and
BrachyDose central axis HETERO doses are within 1.2% (
Pd), 1.3%(
I), and 1.4% (
Cs ; 16 mm COMS only), and median discrepancies are much smaller(Fig. 4); agreement is comparable along transverse axes. Comparing egs brachy and
MCNP central axis HETERO doses, percent differences are at most 3.7% (
Pd), 2.2% (
I), and1.8% (
Cs) for HETERO.In considering the results in figure 4, there are different trends for %∆ with the dif-ferent radionuclides and codes. For example, all median %∆ values for egs brachy and
BrachyDose are very near zero for
Pd but slightly higher and positive (near 0.5%) for
I. The values of %∆ are generally largest for
MCNP and egs brachy , with the largest andpositive values for
Pd, followed by
Cs , whereas the values are more generally negativefor
I. The fact that the %∆ comparisons lie above and below zero for different radionuclidessuggests good agreement in the plaque models.Overall, doses along the central axis are in good agreement between egs brachy andpreviously-published
BrachyDose and
MCNP results, noting that 1 σ statistical uncertaintiesare <
1% on for
BrachyDose
10, 13, 14 and order of 2% for
MCNP (exception: 7.6% for 22 mmplaque). Discrepancies between results for the different codes may be due to a combinationof factors including cross sections, transport parameters, code versions, source spectra, mass-energy absorption coefficient values. Melhus and Rivard estimated that the total uncertaintyon their results is 5% at a depth of 5 mm from the plaque (all three radionuclides and COMSplaque sizes, including non-statistical sources of uncertainty).LRP EPv2 dosimetry Database page 11 z / cm do s e / a i r k e r m a / G y h - U - egs_brachyBrachyDoseMCNP CLRP_EPv2, COMS EPs,
Pd, HETERO CAX value mm mm
22 mm a) mm x / cm do s e / a i r k e r m a / G y h - U - egs_brachyBrachyDose CLRP_EPv2, COMS EPs,
Pd, HETERO, x profile, z =0.5 cm mm b) mm
22 mm16 mm z / cm do s e / a i r k e r m a / G y h - U - egs_brachyBrachyDoseMCNP CLRP_EPv2, COMS EPs,
I, HETERO CAX value mm mm
22 mm c) mm x / cm do s e / a i r k e r m a / G y h - U - egs_brachyBrachyDose CLRP_EPv2, COMS EPs,
I, HETERO, x profile, z =0.5 cm
10 mm d)
18 mm22 mm18 mm z / cm do s e / a i r k e r m a / G y h - U - egs_brachyBrachyDoseMCNP CLRP_EPv2, COMS EPs,
Cs, HETERO CAX value mm mm
22 mm e) mm x / cm do s e / a i r k e r m a / G y h - U - egs_brachy CLRP_EPv2, COMS EPs,
Cs, HETERO, x profile, z =0.5 cm
10 mm f)
18 mm22 mm16 mm
Figure 3:
COMS HETERO doses: (a, c, e) along the central axis; (b, d, f) transverse x axis (at 0.5 cm depth), for egs brachy (lines), BrachyDose (‘ + ’ symbol) and MCNP (circle symbol) for nuclides: Pd [model 200; panel (a, b)];
I [model 6711; panel(c,d)];
Cs [Cs-1 Rev 2; panel (e,f)]. The 10 mm, 16 mm, 18 mm, and 22 mm EPsare shown in black, green, blue, and red colors, respectively.
LRP EPv2 dosimetry Database page 12
COMS plaque diameter -2-1012345 % D fova (eb, BD): open boxes, solid lines
10 mm 14 mm
CLRP_EPv2, COMS EPs, HETERO CAX value differences
Pd: black
I: blue(eb, MCNP): filled boxes, dashed lines
Cs: red
12 mm 16 mm 18 mm 20 mm 22 mm
Figure 4:
Percent difference ( %∆ ; Eq. (1)) for HETERO doses along the central axis ofCOMS plaques for egs brachy compared with published BrachyDose (BD)
2, 10 (openboxes, solid lines) and
MCNP (filled boxes, dashed lines). Results are shown for all plaquediameters with Pd (black),
I (blue), and
Cs (red;
BrachyDose only for 16 mmbecause data were not published for other sizes). Whiskers represent the total range of %∆ values, the boxes indicate the inner quartiles (extending between the first and thirdquartiles), and the horizontal line indicates the median. Statistical uncertainties are ≤ egs brachy , ≤
1% for
BrachyDose , and order of 2% (7.6% for 22 mm)for MCNP . Figure 5 provides further comparison of egs brachy results with
BrachyDose and
MCNP
Fig 5 for 16 mm COMS. Fig. 5a presents the ratio of doses HETERO/HOMO along the centralaxis for egs brachy , BrachyDose ,
10, 13 and
MCNP : comparing with egs brachy , BrachyDose ratios are within 1.5% (
Pd), 0.5% (
I), and those for
MCNP are within 3.0% (
Pd),and 3.5% (
I). The fact that dose ratios are systematically higher for
MCNP relative to egs brachy (and
BrachyDose ) may be partially attributed to the fact that HOMO
MCNP simulations include interseed attenuation (those for egs brachy and
BrachyDose do not),thus lower the dose in the denominator and increasing the HETERO/HOMO ratio. Previousresearch assessed the magnitude of interseed effects, reporting that doses differed by less than0.5% (HETERO) and by 1 to 2% for seeds in water (HOMO). For all codes, the largest discrepancies are observed at the opposite side of the eye to theLRP EPv2 dosimetry Database page 13plaque where the statistical uncertainties are largest. Figure 5b summarizes comparisons ofdoses at organs at risk (positioned off the central axis), considering eight different plaquepositions as done by Rivard et al : differences are ≤ egs brachy and BrachyDose and ≤
4% for egs brachy and
MCNP with the exception of the lacrimal gland HETERO dosefor two (of the eight) plaque positions considered. For those positions, the lacrimal glandHETERO doses are very low (3.5 Gy for
Pd and 6.6 Gy for
I, compared with 85 Gy asthe prescription dose) and
MCNP differs by 12.5% (
Pd) and 17% (
I) from egs brachy .This large discrepancy in the dose at the lacrimal gland for these plaque positions has beenobserved previously in comparing
BrachyDose and
MCNP doses and elsewhere in comparing egs brachy , BrachyDose , and
MCNP results.
8, 17
The results shown in this subsection demonstrate that there is generally closer agree-ment between the two EGSnrc-based code, egs brachy and
BrachyDose , than between egs brachy and
MCNP , consistent with previous work. Overall, these comparisons validatethe new egs brachy models and 3D dose distributions for COMS plaques.LRP EPv2 dosimetry Database page 14 z / cm H E T E R O / H O M O egs_brachyBrachyDoseMCNP CLRP_EPv2, COMS16mm, HETERO/HOMO CAX value I Pd a) organs at risk -4-3-2-10123456 % D fova (eb, BD): open boxes, dashed lines HOMO fovea disk lacrimal glandlens
CLRP_EPv2, COMS16 mm, HOMO and HETERO
Pd : black
I : blue (eb, MCNP): filled boxes, solid lines
HETERO HETERO HOMO HETERO HOMO HETERO HOMO b) Figure 5:
Comparisons of results for 16 mm COMS plaques for (a) HETERO/HOMOratio of doses along the central axis (for
Pd (200),
I (6711),
Cs (CS-1)) and(b) box-and-whisker summary of percent dose differences ( egs brachy (eb) comparedwith
BrachyDose (BD) and
MCNP ) over eight different plaque positions consideringfour organs at risk and
Pd (200) and
I (6711) seeds.
8, 10
LRP EPv2 dosimetry Database page 15 II . C . . BEBIG plaques The present work offers the first MC simulations of the BEBIG model plaques, and socomparisons to previously published results are not possible. However, the BEBIG plaquesare of the same (geometric) dimensions to the COMS plaques, and thus the validation of our egs brachy models for those plaques (section II . C . .) supports the new egs brachy BEBIGmodels. The dose distributions are expected to be different for BEBIG compared with COMSdue to differences in plaque backing medium (elemental composition and density), insertdensity, and seed model (
I: IsoSeed ® I25.S16 for BEBIG, model 6711 for COMS) on thebasis of earlier work investigating backing composition and seed type; this is confirmedby the results ( e.g. , Table 2).Figure 6 provides a subset of dose results for a few BEBIG plaque diameters, contrasted Fig 6 with corresponding COMS plaques. Fig. 6a demonstrates that when the doses are normalizedto the dose at z = 0 . z < . z / cm do s e / do s e a t . c m ( t u m o r ape x ) BEBIG (I25.S16)COMS (6711)
CLRP_EPv2,
I-COMS and -BEBIG, HETERO, Normalized dose mm mm
16 mm mm mm
16 mm a) z / cm H E T E R O / H O M O BEBIG (I25.S16)COMS (6711) CLRP_EPv2,
I-COMS and -BEBIG, HETERO/HOMO CAX value
12 mm mm
20 mm b) Figure 6:
Summary of results for BEBIG plaques with
I model I25.S16 seeds (dashedlines) compared with corresponding COMS plaques (containing
I model 6711 seeds)for 12, 16, and 20 mm plaque diameters for doses along the central axis (a) normalized tounity at z = 0 . cm and (b) ratio HETERO/HOMO. Combined statistical uncertaintiesfor HETERO/HOMO dose ratios are ≤ LRP EPv2 dosimetry Database page 17Figure 6b presents the ratio of HETERO/HOMO doses for a subset of BEBIG and COMSplaque sizes (
I seed models: I25.S16 for BEBIG, 6711 for COMS), characterizing theheterogeneity effect due to the plaque. Over all plaque sizes, the dose ratio HETERO/HOMOis larger for BEBIG relative to COMS very near the plaque, but further from the plaquethe dose decrease is more substantial for BEBIG than COMS. These differences betweenBEBIG and COMS dose distributions result from a combination of effects due to plaquebacking materials of different elemental compositions and densities, different insert massdensities, and distinct seed models.The lower mass density of Silastic combined with the slightly higher average energy of theBEBIG I25.S16 seeds results in a slightly higher HETERO/HOMO dose ratio very near theplaque. Further away, the higher atomic number of predominant plaque components (Au,Pt) for BEBIG plaques results in larger dose decreases compared with COMS plaques (noPt; Ag with Au, among other elements), in accord with previous results comparing pure goldbackings with Modulay. Overall, the differences between HETERO/HOMO dose ratios forBEBIG and COMS are 2% or less, consistent with previous comparisons of plaque backings and seed models. In a separate set of simulations, interseed attenuation is investigated for the BEBIGplaques (containing I25.S16 seeds) by comparison of HETERO simulations with and with-out interseed effects ( egs brachy ‘normal’ and ‘superposition’ run modes, respectively).Interseed attenuation reduces doses by 0 .
5% (12 mm), 0.4% (14 mm), 0.3% (16 mm), 1.0%(18 mm), and 1.6% (20 mm) at the inner sclera and by 0 .
3% (12 mm), 0.2% (14 mm), 0.2%(16 mm), 0.2% (18 mm), and 0.5% (20 mm) at z = 1 cm. These relatively small differencesbetween HETERO simulations with and without interseed effects are in accord with previouswork that considered other seed models.
10, 11
Figure 7 summarizes the HETsi and HETsum results in comparison with HETERO
Fig 7 and HOMO. The figures demonstrate that superposition of the HETsi results to determineHETsum provides dose distributions in excellent agreement (as expected) with HETEROresults.LRP EPv2 dosimetry Database page 18 z / cm ( do s e / a i r k e r m a ) / G y h - U - HOMOHETEROHETsumHETs1HETs8HETs13
CLRP_EPv2, BEBIG16mm,
I (I25.S16) a) z = . c m -1.2 -1.0 -0.8 -0.6 -0.4 -0.2 0.0 0.2 0.4 0.6 0.8 1.0 1.2 x / cm ( do s e / a i r k e r m a ) / G y h - U - HOMOHETEROHETsumHETs1HETs8HETs13
CLRP_EPv2, BEBIG16mm,
I (I25.S16), z= 0.5 cm b) Figure 7:
Summary of 16 mm BEBIG (I25.S16) doses per unit seed air kerma strengthfor HOMO, HETERO, HETsum, and HETsi scenarios for (a) central axis and (b)transverse ( x ) axis at z = 0 . cm. Note that central axis curves are identical for HETs1to HETs7, HETs8 to HETs11, and HETs12 to HETs13, because the radial distancefrom the seed centre to the z axis is the same in each case. Statistical uncertaintiesare ≤ LRP EPv2 dosimetry Database page 19 II . C . . Representative plaques Figure 8 provides HETERO/HOMO dose ratios along the central axis for the representa-
Fig 8 tive plaques, comparing results to those from
BrachyDose from previous work,
1, 14 as wellas 16 mm COMS and BEBIG results for comparison. The egs brachy and
BrachyDose results are in good agreement, with percent differences of 0.1% to 1.1% for z ≤ . z ≤ . z ≤ . z . The BrachyDose results for the HETERO/HOMO dose ratio are generally observed to fluctuate about thesmoother lines presented by the egs brachy results, in accord with the larger statisticaluncertainties on the
BrachyDose results (sub-1% for 1 σ ) in comparison with egs brachy (sub-0.2%). Apart from differences in the number of histories used to generate results for egs brachy and BrachyDose , there are also differences in the spectra of initialized parti-cles, small differences in plaque dimensions (due to rounding in intermediate calculations tospecify plaque geometries), and differences in the voxel scoring grid, as well as differencesassociated with the different codes as seen previously.
8, 17
Near the plaque (and sclera), the “Short lip - acrylic” plaque is observed to increasedoses relative to HOMO doses, in accord with Lesp´erance et al who attributed this to thecombined effects of fluorescence photons from the plaque and less photon attenuation throughthe (short) acrylic insert (for HETERO simulations) in comparison with water (HOMO). Forall other representative plaques, the HETERO/HOMO dose ratio is ≤ LRP EPv2 dosimetry Database page 20 z / cm H E T E R O / H O M O Line : egs_brachySymbol : Brachydose CLRP_EPv2,
Pd (200), Representative EPs
COMS16mmNlSSla a) Cta Ssa0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4 z / cm H E T E R O / H O M O Line : egs_brachySymbol : Brachydose CLRP_EPv2,
I (6711), Representative EPs C O M S mm NlS Sla b) CtaSsa
BEB I G mm ( I . S ) z / cm H E T E R O / H O M O Line : egs_brachySymbol : Brachydose CLRP_EPv2,
Cs (CS-1), Representative EPs C O M S mm NlS Sla c) Cta Ssa
Figure 8:
Comparison of calculated HETERO/HOMO doses along the central axisfor egs brachy (lines) and
BrachyDose
1, 14 (symbols) for fully-loaded representativeplaques (Cta: COMS-thin acrylic; Sla: Short lip-acrylic; Ssa: Stainless steel-acrylic;NlS: No lip-Silastic) with 16 mm COMS and BEBIG results shown for comparison for(a)
Pd (model 200), (b)
I (model 6711; I25.S16 for BEBIG only), and (c)
Cs(CS-1 Rev2) seeds. Statistical uncertainties on the HETERO/HOMO dose ratio are atmost 0.2% for egs brachy and for BrachyDose . LRP EPv2 dosimetry Database page 21
III . Data Format and Access
The CLRP EPv2 website is hosted at Carleton University, Ottawa, Ontario, Canada. Thedatabase is available online at https://physics.carleton.ca/clrp/eye_plaque_v2 . Themain page of the database lists the 17 plaques which for the online datasets are available,as well as details about plaque and seed models, MC simulation scenarios and normaliza-tion, and a spreadsheet of HOMO and HETERO doses along the central axis with 0.05 cmresolution. The CLRP EPv2 database includes results from more than 100 independent egs brachy simulations. For each plaque model, the following information is available onthe CLRP EPv2 database: • A to-scale image showing the eye plaque cross-section fully loaded with seeds in eachof the Cartesian planes intersecting the origin, created using egs view images of actualegs++ model of the plaque and seeds. • A description of the plaque model, and seed coordinates according to publications andmanufacturer information, as implemented in egs++. These models will be releasedfor use with the open-source egs brachy application. • Files containing 3D dose distributions (.3ddose) in a zipped format for HOMO andHETERO scenarios, as well as HETsi for BEBIG plaques. Doses are provided in unitsof dose rate per unit seed air kerma (Gy h − U − ). Total dose (in Gy) in each voxelmay be obtained by multiplying the data by the air kerma per seed (in units of U;1 U = 1 cGy cm h − ), and integrated over the treatment time, taking the exponentialdecay of the sources into account. Statistical uncertainties (1 sigma) on the dose ineach voxel is also included in the 3ddose file (as fractions of the local dose). • A figure for the depth dose curve along the plaque central axis compared to the available
BrachyDose
1, 2, 10, 11, 14 and
MCNP
8, 9 data for HOMO and HETERO scenarios in units ofdose/air kerma (Gy h − U − ) for different seeds (where data are available). • A figure comparing x, y transverse dose profiles at different depths ( z = 0 . , BrachyDose
1, 2, 10, 11, 14 (where data are available). For BEBIG plaques,data for HETsi scenario are presented.LRP EPv2 dosimetry Database page 22 • For the 16 mm COMS plaque, a figure that summarizes percent dose differences of egs brachy with
BrachyDose or MCNP along the central axis and at some organs atrisk ( e.g. , fovea, optic disk, lens, and lacrimal gland). • A figure comparing the ratio of HETERO/HOMO doses along the central axis withcorresponding
BrachyDose
1, 14 data for each Representative plaques ( e.g. , Cta, Sla, Ssa,and NlS) and different nuclide ( e.g. , Pd ,
I ,
Cs ). IV . Potential Impact The most recent AAPM task group reports pertaining to eye plaque brachytherapy, namelyTG-129 and TG-221, recommend that HETERO doses be calculated (or estimated) andreported in parallel with the traditional water-based (HOMO) TG-43 doses. These rec-ommendations are consistent with the recommendations of AAPM-ESTRO-ABG TG-186,which recommends adoption of model-based dose calculations where possible and report-ing of their doses alongside TG-43. Thus, the publicly-available 3D dose distributionsand associated data provided in CLRP EPv2 for each plaque and seed supports TG-129,221, and 186 recommendations, enabling HETERO dose estimations and MC dose evalu-ations. Furthermore, the plaque models developed herein will be distributed freely with egs brachy , as will the seed models that are documented within the CLRP TG43v2 database ( https://physics.carleton.ca/clrp/egs_brachy/seed_database_v2 ). Themodels may be used to carry out custom calculations within user-specified phantoms, in-cluding virtual patient models derived from patient (CT) datasets. In addition, the eyeplaque models may then be modified by users to develop models representative of their ownpractice. Overall, the dose distributions and freely-distributed egs brachy plaque modelssupport state-of-the art, advanced dose evaluations for ocular brachytherapy. V . Conclusion The CLRP EPv2 database offers accurate 3D dose distributions for more plaque modelsand radionuclides, plus lower statistical uncertainties than CLRP EPv1. The CLRP EPv2database contains new datasets for 17 plaques (8 COMS, 5 BEBIG, and 4 representativeplaques) and 3 radionuclides [
Pd,
I (2 seed models), and
Cs], including both HOMOand HETERO (also HETsi for BEBIG) scenarios. The CLRP EPv2 data are validated byLRP EPv2 dosimetry Database page 23comparison with
BrachyDose and
MCNP published data, for the plaque models for whichsuch data exist: COMS plaque data is in good accord with previous
BrachyDose and
MCNP results, and representative plaques agree with previous
BrachyDose results. For example,on the plaque central axis HETERO doses agree within 2% for
BrachyDose and 5% for
MCNP for COMS plaques, while doses agree within 4.5% with
BrachyDose for the repre-sentative plaques. The BEBIG plaques, modelled for the first time, differ only in media(elemental composition, density) from the COMS plaques, and differences in the relativeHETERO/HOMO dose ratios may be understood on the basis of the higher-atomic num-ber content of the BEBIG plaque backings in comparison with COMS, the lower density ofthe seed-carrier insert, as well as the differences in emitted photon spectra [
I: I125.S16for BEBIG, 6711 for COMS]. The HETsi dose distributions offer the potential for users totally dose distributions with different weights for custom-loading plaques with seeds of dif-ferent activities. The CLRP EPv2 database provides reference 3D dose distributions thatsupport the recommendations of AAPM TG-129 and TG-221. The reference 3D dose dis-tributions and benchmarked egs brachy (MC) models of plaques and seeds will be freelydistributed at ( https://physics.carleton.ca/clrp/eye_plaque_v2 ), enabling advancesin ocular brachytherapy research, dosimetry, and clinical practice. VI . Acknowledgements Chris Melhus and Mark Rivard are thanked for providing their MCNP5 data directly tofacilitate comparisons. The authors acknowledge the Natural Sciences and Engineering Re-search Council of Canada (NSERC), the Canada Research Chairs program, the Ministry ofResearch and Innovation of Ontario, and a Compute Canada National Resource Allocation.
VII . Conflict of interest statement
This work was partially supported by Eckert & Ziegler BEBIG GmbH of Berlin, Germany.LRP EPv2 dosimetry Database page 24
References R. M. Thomson, K. M. Furutani, T. W. Kaulich, F. Mourtada, M. J. Rivard, C. G.Soares, F. M. Vanneste, and C. S. Melhus, AAPM recommendations on medical physicspractices for ocular plaque brachytherapy: Report of task group 221, Med. Phys. , e92– e124 (2020). S.-T. Chiu-Tsao, M. A. Astrahan, P. T. Finger, D. S. Followill, A. S. Meigooni, C. S.Melhus, F. Mourtada, M. E. Napolitano, R. Nath, M. J. Rivard, D. W. O. Rogers, andR. M. Thomson, Dosimetry of
I and
Pd COMS eye plaques for intraocular tumors:Report of Task Group 129 by the AAPM and ABS, Med. Phys. , 6161 – 6184 (2012). ABS. OOTF. Committee, The American Brachytherapy Society consensus guidelines forplaque brachytherapy of uveal melanoma and retinoblastoma, Brachytherapy , 1 – 14(2014). COMS Group, The COMS randomized trial of iodine 125 brachytherapy for choroidalmelanom. III. Initial Mortality findings. COMS report no. 18, Arch. Ophthalmol. ,969–982 (2001). COMS Group, The COMS randomized trial of iodine 125 brachytherapy for choroidalmelanom. V. Twelve-year mortality rates and prognostic factors: COMS report no. 28,Arch. Ophthalmol. , 1684–1693 (2006). A. J. Lin, Y. J. Rao, S. Acharya, J. Schwarz, P. K. Rao, and P. Grigsby, Patterns of careand outcomes of proton and eye plaque brachytherapy for uveal melanoma: review of theNational Cancer Database, Brachytherapy , 1225–1231 (2017). M. J. Rivard et al. , Supplement 2 for the 2004 update of the AAPM Task Group No. 43Report: Joint recommendations by the AAPM and GEC-ESTRO, Med. Phys. , e297– e338 (2017). M. J. Rivard et.al. , Comparison of dose calculation methods for brachytherapy of intraoc-ular tumors, Med. Phys. , 306 – 316 (2011). C. S. Melhus and M. J. Rivard, COMS eye plaque brachytherapy dosimetry simulationsfor
Pd,
I, and
Cs, Med. Phys. , 3364 – 3371 (2008). R. M. Thomson, R. E. P. Taylor, and D. W. O. Rogers, Monte Carlo dosimetry for
Iand
Pd eye plaque brachytherapy, Med. Phys. , 5530 – 5543 (2008). R. M. Thomson and D. W. O. Rogers, Monte Carlo dosimetry for
I and
Pd eyeplaque brachytherapy with various seed models, Med. Phys. , 368 – 376 (2010). R. M. Thomson, K. M. Furutani, J. S. Pulido, S. L. Stafford, and D. W. O. Rogers,Modified COMS plaques for
I and
Pd iris melanoma brachytherapy, Int. J. Radiat.Oncol. Biol. Phys. , 1261 – 1269 (2010). M. Lesperance, M. Inglis-Whalen, and R. M. Thomson, Model-based dose calculationsfor COMS eye plaque brachytherapy using an anatomically realistic eye phantom, Med.Phys. , 021717 (12pp) (2014). M. Lesperance, M. Martinov, and R. M. Thomson, Monte Carlo dosimetry for
Pd,
I, and
Cs ocular brachytherapy with various plaque models using an eye phantom,Med. Phys. , 031706 (14pp) (2014).LRP EPv2 dosimetry Database page 25 L. Beaulieu, A. C. Tedgren, J.-F. Carrier, S. D. Davis, F. Mourtada, M. J. Rivard, R. M.Thomson, F. Verhaegen, T. A. Wareing, and J. F. Williamson, Report of the TaskGroup 186 on model-based dose calculation methods in brachytherapy beyond the TG-43formalism: Current status and recommendations for clinical implementation, Med. Phys. , 6208 – 6236 (2012). M. Chamberland, R. E. P. Taylor, D. W. O. Rogers, and R. M. Thomson, egs brachy: aversatile and fast Monte Carlo code for brachytherapy, Phys. Med. Biol. , 8214 – 8231(2016). R. M. Thomson, R. E. P. Taylor, M. J. P. Chamberland, and D. W. O. Rogers, Replyto Comment on ’egs brachy: a versatile and fast Monte Carlo code for brachytherapy’,Phys. Med. Biol. , 038002(5pp) (2018). H. Safigholi, M. J. P. Chamberland, R. E. P. Taylor, C. H. Allen, M. P. Martinov, D. W. O.Rogers, and R. M. Thomson, Update of the CLRP TG-43 parameter database for low-energy brachytherapy sources, Med. Phys. , 4656–4669 (2020). E. Mainegra-Hing, D. Rogers, F. Tessier, and B. Walter, The EGSnrc Code System:Monte Carlo Simulation of Electron and Photon Transport,NRCC Report PIRS-701,Technical report, National Research Council Canada, Ottawa, Canada. https://nrc-cnrc.github.io/EGSnrc/, (2020). M. J. Berger and J. H. Hubbell, XCOM: Photon cross sections on a personal computer,Report NBSIR87–3597, National Institute of Standards Technology (NIST), Gaithers-burg, MD 20899, U.S.A., 1987. E. Mainegra-Hing, D. W. O. Rogers, R. Townson, B. R. B. Walters, F. Tessier,and I. Kawrakow, The EGSnrc g application Technical Report PIRS-3100, Tech-nical report, National Research Council Canada, Ottawa, Canada. https://nrc-cnrc.github.io/EGSnrc/, (2020). ICRU, Report 90: Key data for ionizing-radiation dosimetry: measurement standardsand applications, J ICRU , 1–110 (2014). S. E. M. Cutsinger, K. M. Furutani, R. M. Forsman, and S. M. Corner, Seed coordinatesof a new COMS-like 24 mm plaque verified using the FARO Edge, J of App Clin MedPhys , 293 – 301 (2015). D. P. Potter, Treatment of intraocular melanoma: new concepts, Bull Mem Acad R MedBelg , 103–111 (2003). P. T. Finger, A. Berson, and A. Szechter, Palladium-103 plaque radiotherapy for choroidalmelanoma, Ophthalmology , 606 – 613 (1999). I. Puusaari, J. Heikkonen, and T. Kivela, Effect of radiation dose on ocular complicationsafter iodine brachytherapy for large uveal melanoma: empirical data and simulation ofcollimating plaques, Invest Ophthalmol Vis Sci , 3425–3434 (2004). D. Granero, J. P´erez-Calatayud, F. Ballester, E. Casal, and J. De Frutos, Dosimetricstudy of the ROPES eye plaque, Med. Phys. , 3330–3336 (2004). J. I. Monroe and J. F. Williamson, Monte Carlo-aided dosimetry of the TheragenicsTheraSeed Model 200
Pd interstitial brachytherapy seed, Med. Phys. , 609 – 621(2002).LRP EPv2 dosimetry Database page 26 J. Dolan, Z. Li, and J. F. Williamson, Monte Carlo and experimental dosimetry of an
I brachytherapy seed, Med. Phys. , 4675 – 4684 (2006). M. J. Rivard, Brachytherapy dosimetry parameters calculated for a
Cs source, Med.Phys. , 754 – 762 (2007). H. Hedtj¨arn, G. A. Carlsson, and J. F. Williamson, Monte Carlo-aided dosimetry of thesymmetra model I25.S06 I , interstitial brachytherapy seed, Med. Phys. , 1076–1085(2000).33