Thermal neutron irradiation field design for boron neutron capture therapyof human explanted liver

S. Bortolussia! and S. AltieriDepartment of Nuclear and Theoretical Physics, University of Pavia, INFN Section of Pavia, via Bassi 6,27100 Pavia, Italy

!Received 9 May 2007; revised 18 September 2007; accepted for publication 19 September 2007;published 15 November 2007"

The selective uptake of boron by tumors compared to that by healthy tissue makes boron neutroncapture therapy !BNCT" an extremely advantageous technique for the treatment of tumors thataffect a whole vital organ. An example is represented by colon adenocarcinoma metastases invadingthe liver, often resulting in a fatal outcome, even if surgical resection of the primary tumor issuccessful. BNCT can be performed by irradiating the explanted organ in a suitable neutron field.In the thermal column of the Triga Mark II reactor at Pavia University, a facility was created for thispurpose and used for the irradiation of explanted human livers. The neutron field distribution insidethe organ was studied both experimentally and by means of the Monte Carlo N-particle transportcode !MCNP". The liver was modeled as a spherical segment in MCNP and a hepatic-equivalentsolution was used as an experimental phantom. In the as-built facility, the ratio between maximumand minimum flux values inside the phantom !!max/!min" was 3.8; this value can be lowered to 2.3by rotating the liver during the irradiation. In this study, the authors proposed a new facilityconfiguration to achieve a uniform thermal neutron flux distribution in the liver. They showed thata !max/!min ratio of 1.4 could be obtained without the need for organ rotation. Flux distributionsand dose volume histograms were reported for different graphite configurations. © 2007 AmericanAssociation of Physicists in Medicine. #DOI: 10.1118/1.2795831$

Key words: BNCT, liver irradiation, MCNP, thermal neutrons

I. INTRODUCTION

Boron neutron capture therapy !BNCT" is an experimentalform of binary radiotherapy, based on preliminary enrich-ment of the tumor with 10B, followed by neutron irradiation.The therapeutic action comes from the capture reaction10B!n ,""7Li, which is characterized by a high thermal neu-tron cross section !3837 b". The resulting high linear energytransfer particles from the boron reaction have a range thatcorresponds approximately to the cell diameter and damagethe cells in which the reaction takes place. If a higher boronconcentration can be obtained in the tumor compared to thatin the normal tissues, the irradiation can deliver a lethal doseto the tumor, while keeping the dose absorbed by healthyparenchyma below tolerance levels.1

The first BNCT clinical trials were attempted in the 1950sin the USA at Massachusetts Institute of Technology !MIT"and Brookhaven National Laboratories !BNL", using thermalneutron beams and the boron compound, borax.2 The treatedpatients were affected by glioblastoma multiforme. The out-come of these treatments was poor, mainly due to the lowboron concentration ratio obtained between the tumor andhealthy tissue.3 In the 1960s, new BNCT clinical trials beganin the USA and Japan, thanks to the availability of new com-pounds such as boron phenylalanine !BPA" and sodium bo-rocaptate that were able to ensure a better selective boronuptake in the tumor.4,5 A third phase of BNCT trials wasbegun in the second half of the 1990s, when epithermal neu-tron beams were introduced to better penetrate the cranium,thus obtaining a higher thermal neutron flux inside the tu-

moral target. A new clinical trial for the treatment of glio-blastoma multiforme was started at BNL.6 At the same time,a trial for the treatment of cutaneous melanoma and anotherfor the treatment of intracerebral melanoma and glioblastomawere started at MIT.7,8

In Pavia, BNCT was applied to multiple dispersed livermetastases, using the autotransplant method.9,10 In the period2001–2003, two patients were treated, both affected by livermetastases !from colon and rectum carcinoma". Both patientshad a very poor prognosis due to the condition of the liverfollowing earlier successful resection of the primary tumor.The procedure, developed for the first time, consisted of ex-planting the liver, which had been previously infused with aBPA-fructose complex, and irradiating it in the thermal col-umn of the University’s TRIGA Mark II reactor. After BPAinfusion and before irradiation, the liver was washed to re-move blood from the vessels to prevent vein wall damage.The irradiation of the isolated liver kept the rest of the pa-tient’s body free from radiation damage. After the irradiation,the liver was reimplanted into the patient. A computed to-mography scan of both patients, made ten days after treat-ment, demonstrated the selective destruction of the me-tastases. Furthermore, no healthy liver damage wasdiagnosed in the 44 months of survival time of the first pa-tient, who enjoyed a good quality of life during the entirepost-treatment period. The second patient had the same earlyperioperative course, but he died after 33 days from cardiacfailure due to worsening of dilated cardiomyopathy, fromwhich he was suffering.11

4700 4700Med. Phys. 34 „12…, December 2007 0094-2405/2007/34„12…/4700/6/$23.00 © 2007 Am. Assoc. Phys. Med.

The treatment plan for irradiation of an explanted organaffected by metastases must deliver a lethal dose to the tumorregardless of its spatial distribution and keep the dose ab-sorbed by the normal tissue as low as possible, or at leastunder the tolerance level. Irradiation with thermal neutronsmust also destroy the small tumoral nodules that cannot bediagnosed before treatment. To fix the irradiation time in thetreatment plan, a small tumor is assumed to be located at thepoint at which the thermal neutron flux is at a minimum. Atthat point, the neutron fluence must deliver a total dose suf-ficient to destroy the tumoral nodule. With this assumption,the tumor will absorb a lethal therapeutic dose whenever it islocated inside the organ. The total dose absorbed by the nor-mal tissue is thus minimal near the point of minimum flux,but increases at the points at which the thermal neutron fluxis higher. For this reason, a considerable effort is being madeto design a new graphite configuration in order to ensure amore uniform neutron flux distribution in the explanted or-gan. In this article, the results of the new experimental andMonte Carlo studies devoted to this goal are presented.

II. MATERIALS AND METHODS

II.A. Monte Carlo calculations

The reactor structure was simulated by MCNP.12 The ge-ometry and the materials are as described in the Final SafetyReport13,14 of Pavia University’s Triga Mark II reactor. Fig-ure 1 shows the vertical and horizontal sections of the reactorgeometry.

The source was simulated by sampling the neutrons witha uniform distribution in the volume of each fuel element,with energy distributed as the Watt spectrum.15 The tallieswere normalized to the power of the reactor using the factor1.9#1016 s!1 !neutrons per second emitted from a reactorcore working at 250 kW".

The explanted human liver placed in the irradiation posi-tion was modeled as a spherical segment, 6 cm high and witha circular base radius of 15 cm. The phantom volume was

divided into cubic voxels of 1#1#1 cm3. Along the verti-cal axis !z axis", five meshes of voxels at z=!6, !5, !4, !3,and !2 cm, starting from the center of the coordinate systemwhich is at the center of the core !Fig. 2", were used.

In the simulations, the neutron flux was obtained for eachvoxel !F4:n tally". Two neutron energy bins were used: onethermal !En$0.2 eV" and one in the energy range 0.2 eV–0.5 MeV.

II.B. Experimental measurements

The Monte Carlo calculations were validated by compar-ing experimental data with the results of the simulations. Thethin foil activation method16 was used for an experimentalmeasurement of the thermal neutron flux. The experimentalphantom consisted of a Teflon spherical segment with 3-mmthick walls. To study the neutron flux distribution in the liver,a hepatic-equivalent water solution was created and used tofill the Teflon phantom. The composition and the elementalpercentages by weight of the hepatic solution are reported inTables I and II. The hepatic solution was similar to the livercomposition reported in ICRU 46 !Ref. 17" and the onlysignificant differences were the percentages of carbon andoxygen. Nevertheless, the sum of these two element percent-ages was equal to the sum of the same element percentagesin the ICRU 46 report recommendations. This made the so-lution suitable for liver simulation because carbon and oxy-gen cross sections are very similar, at least up to 20 MeV.16

The Teflon phantom was equipped with thin copper wiresalong the x, y, and z axes, as shown in Fig. 2, and wasirradiated at the liver position. After irradiation, the copperwires were cut into pieces 0.5 cm long and a Ge detector wasemployed to measure the 64Cu activity by gamma spectrom-etry. The measured activity was then used to evaluate thethermal neutron flux using the Westcott formula.18,19

III. MONTE CARLO VALIDATION RESULTS

As a first step in the validation of the Monte Carlo results,measurements and calculations were carried out in a phan-tom filled with air. The results are reported in Fig. 3 andshow agreement between the simulation and experimentalmeasurements of better than 10%.

The second step consisted in studying the distortionscaused by the liver tissue to the neutron flux distribution. TheTeflon phantom was filled with the solution and irradiated inthe thermal column. The thermal neutron flux distributionsalong the x, y, and z axes are reported in Fig. 4.

The agreement between the Monte Carlo calculations andthe experimental measurements was still good but the he-patic solution drastically changed the behavior of the neutronflux distribution, mainly along the longitudinal x axis. The

FIG. 1. Vertical !left" and horizontal !right" section of Pavia University’sTriga Mark II reactor !MCNP geometry". The coordinate system is centeredin the reactor core.

TABLE I. Compounds used for the hepatic-equivalent solution with 50 ppm of 10B.

Compound H2O CH4N2O NaHSO4 KCl H3PO4 H3BO3 Total

Mass !g" 909.2 64.3 11.2 4.2 9.5 1.6 1000

4701 S. Bortolussi and S. Altieri: Thermal neutron field for explanted liver BNCT 4701

Medical Physics, Vol. 34, No. 12, December 2007

ratio between maximum and minimum thermal neutron fluxvalues !!max/!min" was %4. To make the flux distributionmore uniform, the organ was rotated by 180° halfwaythrough the irradiation time. The effect of this rotation isshown in Fig. 5, in which the thermal neutron flux distribu-tion is shown along the x axis for the different meshes at z=!6, !4, and !2 cm. Thanks to the symmetry of the spheri-cal segment around the z axis, the Monte Carlo calculationgave exactly the same results as if the phantom had beenrotated. For this reason, the distributions shown were ob-tained by reversing the MCNP results and adding them to theoriginal values. The !max/!min ratio was thus lowered to2.31.

IV. EFFECT OF GRAPHITE CONFIGURATION ONNEUTRON FLUX DISTRIBUTION

After validating the Monte Carlo simulation, new con-figurations of the graphite surrounding the phantom to ensurea more uniform distribution of the thermal neutron flux in theliver were tested. The different configurations are shown inFig. 6.

In the original configuration #Fig. 6!a"$, the thermal neu-tron flux was higher in the parts of the liver facing the core,therefore, attempts were made to lower these values by in-creasing the distance between the graphite and the anteriorparts of the liver. In particular, some graphite blocks wereremoved from around the irradiation position #Fig. 6!b"$. Theresults of this simulation are shown in Fig. 7. The !max/!minratio was lowered to 1.83.

To further lower this ratio, the distance between the liverand the graphite in the posterior part of the liver, farther fromthe core, was reduced. In this new configuration #Fig. 6!c"$,the anterior half of the liver was placed into the larger irra-diation room and the posterior half in a channel that main-tained the original dimensions. For this setup, the ratio wasreduced to 1.75.

The distributions obtained in the calculations describedabove still required rotation of the liver halfway through the

treatment time. It was felt, however, that a solution withoutrotation should be investigated; in fact, being a patient treat-ment, the system must be as safe as possible; a system thatdoes not rotate is simpler and safer. Some channel configu-rations with graphite backscattering to enrich the neutronfluxes in the liver parts not facing the source were studied.Different configurations of backscattering !planar, parabo-loid, and conical shaped" were tested. The flux distributionsobtained by these cases without liver rotation were compa-rable to those described previously with rotation. Despite thebenefit of no rotation, the uniformity, r=1.94, was slightlyworse when compared to that of the existing channel. More-over, it is very difficult to model graphite blocks into a coni-cal shape. For these reasons, this type of solution was dis-carded, and we investigated whether simpler configurationscould be obtained.

Another strategy to flatten the flux distribution was to usethe open configuration shown in Fig. 6!b", but by adding athermal neutron absorber. A solid material containing smallamounts of 6Li was a good candidate as an absorber becauseit did not produce % radiation. An easy solution was formedby positioning the neutron absorber on the external surfaceof the liver holder, as shown in Fig. 8, in which a 1-mm thickLiF foil was placed around the surface facing the core. Theresults are shown in Fig. 9.

With this relatively simple change of the original setup, aconsiderable improvement in uniformity was obtained. The rvalue was lowered to 1.36, with no rotation. In addition, inthe middle of the phantom, the flux attenuation was $20%,when compared to the configuration without an absorber.

V. DOSE DISTRIBUTION

In order to compare the different graphite configurationsfrom the point of view of dose distribution in the organ, dosevolume histograms !DHVs" were calculated. To define crite-ria to compare the spatial dose distribution in the liver ob-tained in different graphite configurations, the experimentalsituation in the clinical treatment was taken as a reference.

TABLE II. Elemental composition !% by weight" of the hepatic-equivalent solution compared to ICRU 46 liver composition !Ref. 14".

Element O C H N P K S Cl Na

Liver ICRU 46 71.6 13.9 10.2 3.0 0.3 0.3 0.3 0.2 0.20Hepatic solution 83.86 1.29 10.6 3.0 0.3 0.22 0.3 0.2 0.22

FIG. 2. MCNP geometry of the liverphantom inside the Teflon holderplaced in the irradiation position, ver-tical !left" and horizontal !right" sec-tions. The 1 cm3 voxels and the cop-per wires for thermal neutronmeasurement are visible.

4702 S. Bortolussi and S. Altieri: Thermal neutron field for explanted liver BNCT 4702

Medical Physics, Vol. 34, No. 12, December 2007

For both patients, the measured boron concentration inhealthy liver and in tumor was of the order of 8 and 50 ppm,respectively. In the two treatment plans, the irradiation timewas fixed to deliver a neutron fluence of 4#1012 cm!2 to areference point in the explanted organ. With these criteria,necrosis of all metastases was obtained, while the normalparenchyma was mostly spared. In fact, despite the presenceof diffuse hepatic fibrosis, in the first patient, the liver func-tion kept improving after BNCT, and reached normal values,when compared with controls several months later.10 Usingthe thermal neutron flux in each voxel of the phantom, thecontributions were calculated from reactions with nitrogen#14N!n , p"14C$ and boron #10B!n ,""7Li$ with the absorbeddose. The epithermal and fast neutron dose contributioncoming from the elastic scattering on hydrogen was ne-

FIG. 3. Thermal neutron flux distribution in the phantom filled with air:along the x !a", y !b", and z !c" axes. For each axis, the comparison betweenMCNP calculations !triangle" and experimental measurements !circle" isalso shown !see Fig. 2".

FIG. 4. Thermal neutron flux distribution in the phantom filled with thehepatic-equivalent solution: along the x !a", y !b", and z !c". For each axis,the comparison between MCNP calculations !triangle" and experimentalmeasurements !circle" is also shown.

FIG. 5. Calculated thermal neutron flux distribution along the longitudinal xaxis after rotation of 180° at different positions along the z axis; see Fig. 2for rotation axis position.

FIG. 6. The different configurations tested were: !a" the existing configura-tion used in patient treatment; !b" larger open configuration; !c" smaller openconfiguration; and !d" closed conical-shaped channel configuration.

4703 S. Bortolussi and S. Altieri: Thermal neutron field for explanted liver BNCT 4703

Medical Physics, Vol. 34, No. 12, December 2007

glected because it was evaluated to be at least one order ofmagnitude less than that of the reaction 14N!n , p"14C.

For calculations, the following conditions were assumed:

!1" 10B concentration of 8 ppm in the healthy liver;!2" 10B concentration of 50 ppm in the tumor;!3" an irradiation time Tirr to deliver a minimum thermal

neutron fluence, &=4#1012 cm!2, to the tumor. It wasassumed that the tumor was located in voxels in whichthe thermal flux was at a minimum !!min", thus: Tirr=4#1012/!min.

The dose, Dvox!i, was then calculated for each voxel usingthe equation

Dvox!i = !DN + DB # CH" # !vox!i # Tirr,

where DN and DB are the doses per unit fluence from nitro-gen !3% in weight" and from boron !1 ppm", respectively.These contributions were evaluated in MCNP using the Fmtally multiplier card for tally type 4. CH is the healthy liverboron concentration, assumed to be 8 ppm. Finally, differen-tial DVHs were determined by weighting the dose contribu-tion in each voxel with the factor: wi=Volvox!i /Volliver. The

cumulative function of this histogram represented the inte-gral DVH. The results of these calculations are shown in Fig.10, which compares the original thermal column with liverrotation #Fig. 6!a"$, and the configuration shown in Fig. 6!b",with no rotation and with the neutron absorber. In the newconfiguration, the maximum absorbed dose was loweredfrom %7 to %5 Gy. In this case, for a minimum neutronfluence of 4#1012 cm!2, the tumor absorbed %17 Gy,while the healthy tissues absorbed $5 Gy.

VI. CONCLUSIONS

The aim of this work was to study by means of MonteCarlo calculations, the neutron flux distribution inside a liverphantom irradiated at different setups of the thermal columnof the Triga Mark II reactor. First of all, the code was vali-dated by comparing calculations and measurements in theactual irradiation facility, in air and in a hepatic-equivalentsolution. Different graphite configurations were then testedfollowing this scheme: modifications of thermal column withan open channel, modifications with a closed channel, andthe use of a neutron absorber in an open-channel configura-tion. The closed configurations were not a good practicalsolution as they were very expensive and difficult to create,and the improvement in uniformity compared to the existingsituation was poor. An improvement was easy to obtain byopening the graphite in front of the liver and creating an

FIG. 7. Thermal neutron flux distribution in the phantom along the x axis,for the configuration shown in Fig. 6!b". The !max/!min ratio was 1.83.

FIG. 8. Position of the LiF absorber around the circular liver holder towardthe reactor core.

FIG. 9. Thermal neutron flux distribution along the x axis in a modifiedthermal column #Fig. 6!b"$, with a LiF neutron absorber !Fig. 8" and withoutrotation.

FIG. 10. !a" Differential DVH and !b" integral DVH for graphite configura-tion in the actual facility shown in Fig. 6!a" !a1, b1", and for the configu-ration shown in Fig. 6!b", with the LiF absorber around the liver holder andwithout rotation !a2, b2".

4704 S. Bortolussi and S. Altieri: Thermal neutron field for explanted liver BNCT 4704

Medical Physics, Vol. 34, No. 12, December 2007

irradiation room into which the anterior half of the organ wasplaced. The best results were obtained with a neutron ab-sorber placed around the Teflon liver holder. This alsoavoided rotation during the treatment, using a broader irra-diation room. This solution was considered the best option.The uniformity of the thermal neutron flux distribution in-creased and the !max/!min ratio was reduced from 2.31 withliver rotation to 1.36 without liver rotation. The maximumhealthy liver absorbed dose !only charged particles contrib-uting" was lowered from %7 to %5 Gy. This type of solu-tion was straightforward to create, with regard to both thegraphite configuration and the positioning of the LiF foilalong the Teflon holder surface. The DVHs for these twosolutions were compared. The histogram widths and the cu-mulative distribution slopes showed that the latter configura-tion ensured a lower dose to the healthy tissue.

For future developments, it is planned to implement aliver model that better fits the real human explanted organ;thereby optimizing the dose distribution according to the di-mensions of the organ and changing the position and theextension of the neutron absorber around the liver holder.

a"Electronic mail: silva.bortolussi@pv.infn.it1J. A. Coderre and G. M. Morris, “The radiation biology of boron neutroncapture therapy,” Radiat. Res. 151, 1–18 !1999".

2L. E. Farr, W. H. Sweet, J. S. Robertson, S. G. Foster, H. B. Locksley, D.L. Sutherland, M. L. Mendelsohn, and E. E. Stickey, “Neutron capturetherapy with boron in the treatment of glioblastoma multiforme,” AJR,Am. J. Roentgenol. 71, 279–291 !1954".

3W. H. Sweet, in Workshop of Neutron Capture Therapy, Report BNL-51994, edited by R. G. Fairchild and V. P. Bond !Brookhaven NationalLaboratory, Upton, NY, 1986", pp. 173–177.

4H. Hatanaka, “Clinical results of boron neutron capture therapy,” BasicLife Sci. 54, 15–21 !1990".

5H. Hatanaka and Y. Nakagawa, “Clinical results of long-surviving braintumour patients who underwent boron neutron capture therapy,” Int. J.Radiat. Oncol. Biol. Phys. 28, 1061–1066 !1994".

6A. D. Chanana et al., “Boron neutron capture therapy for glioblastomamultiforme: Interim results from the phase I/II dose-escalation studies,”Neurosurgery 44, 1182–1193 !1999".

7H. Madoc-Jones, R. Zamenhof, G. Solares, O. Harling, C.-S. Yam, K.Riley, S. Kiger, D. Wazer, G. Rogers, and M. Atkins, in Cancer NeutronCapture Therapy, edited by Y. Mishima !Plenum, New York, 1996".

8P. M. Busse, O. K. Harling, M. R. Palmer, W. S. Kiger, J. Kaplan, I.Kaplan, C. Chuang, J. Y. Goorley, K. Riley, and T. H. Newton, G. A.Santa Cruz, X.-Q. Lu, and R. G. Zamenhof, “A critical examination of theresults from the Harward-MIT NCT program phase I clinical trial onneutron capture therapy for intracranial disease,” J. Neurooncol. 62, 111–121 !2003".

9T. Pinelli, S. Altieri, F. Fossati, A. Zonta, C. Ferrari, U. Prati, L. Roveda,S. Ngnitejeu Tata, S. Barni, P. Chiari, R. Nano, and D. M. Ferguson, inFrontiers in Neutron Capture Therapy, edited by M. F. Hawthorne et al.!Kluwer Academic, New York, 2001", Vol. II, pp. 1427–1440.

10T. Pinelli et al., “TAORMINA: From the first idea to the application tothe human liver,” in Proceedings: Research and Development in NeutronCapture Therapy, Essen, September 2002, edited by W. Sauerwein, R. L.Moss, and A. Wittig !Monduzzi, Bologna, 2002".

11A. Zonta et al., “Clinical lessons from the first applications of BNCT onunresectable liver metastases,” J. Phys.: Conf. Series 41, 484–495 !2006".

12MCNP: A General Monte Carlo N-Particle Transport Code, Version 4c,edited by J. F. Briesmeister, LA-13709-M, April 2000.

13AA.VV., Rapporto finale di sicurezza del reattore Triga Mark II, Pavia,1964.

14AA.VV., Technical Foundation of Triga General Atomics, San Diego,1958.

15B. E. Watt, “Energy spectrum of neutrons from thermal fissions of 235U,”Phys. Rev. 1037, 87 !1952".

16K. H. Beckurts and K. Wirtz, Neutron Physics !Springer-Verlag, Berlin,1964".

17ICRU Report 46, Photon Electron, Proton and Neutron Interaction Datafor Body Tissues, International Commission on Radiation Units and Mea-surements, Bethesda, MD, 1992.

18C. H. Westcott, “Effective cross section values of well moderated thermalreactor spectra,” Report AECL nr 1101, 1960.

19C. H. Westcott, W. H. Walker, and T. K. Alexander, “Effective crosssection and cadmium ratios for the neutron spectra of thermal reactors,”in Proceedings of the International Conference on the Peaceful Uses ofAtomic Energy, Geneva, 1958 !United Nation, New York, 1959", P/20270.

4705 S. Bortolussi and S. Altieri: Thermal neutron field for explanted liver BNCT 4705

Medical Physics, Vol. 34, No. 12, December 2007