are gadolinium contrast agents suitable for gdnct

Are gadolinium contrast agents suitable forgadolinium neutron capture therapy

Gelsomina De Stasio Deepika Rajesh Patrizia CasalboreMatthew J Daniels Robert J Erhardt1 Bradley H Frazer Lisa M Wiese

Katherine L Richter1 Brandon R Sonderegger Benjamin Gilbert2Sebastien Schaub Rachel J Cannara John F Crawford Mary K GillesTolek Tyliszczak John F Fowler Luigi M Larocca Steven P Howard

Delio Mercanti11 Minesh P Mehta and Roberto Pallini

University of Wisconsin-Madison Department of Physics and Synchrotron Radiation Center 3731Schneider Drive Stoughton WI 53589 USA

University of Wisconsin Department of Human Oncology Madison WI 53792 USAIstituto di Biologia Cellulare CNR Viale Marx 15 I-00137 Roma Italy

1University of Wisconsin Department of Statistics Madison WI 53706 USABob Jones University Greenville SC 29614 USA

Biomedical Engineering Laboratory PSE-A Swiss Federal Institute of Technology CH-1015 LausanneSwitzerland

Hengelweg 37 CH-5303 Wurenlingen SwitzerlandLawrence Berkeley National Laboratory 1 Cyclotron Rd MS 6-2100 Berkeley CA 94720 USA

Istituto di Anatomia Patologica Universitarsquo Cattolica del Sacro Cuore Largo A Gemelli 8 I-00168Roma Italy

11Istituto di Neurobiologia e Medicina Molecolare CNR Viale Marx 15 00137 Roma ItalyIstituto di Neurochirurgia Universitarsquo Cattolica del Sacro Cuore Largo A Gemelli 8 I-00168 Roma Italy

Objective Gadolinium neutron capture therapy (GdNCT) is a potential treatment for malignantbrain tumors based on two steps (1) injection of a tumor-specific 157Gd compound (2) tumorirradiation with thermal neutrons The GdNC reaction can induce cell death provided that Gd isproximate to DNA Here we studied the nuclear uptake of Gd by glioblastoma (GBM) tumorcells after treatment with two Gd compounds commonly used for magnetic resonance imagingto evaluate their potential as GdNCT agentsMethods Using synchrotron X-ray spectromicroscopy we analyzed the Gd distribution at thesubcellular level in (1) human cultured GBM cells exposed to Gd-DTPA or Gd-DOTA for 0ndash72 hours (2) intracerebrally implanted C6 glioma tumors in rats injected with one or two dosesof Gd-DOTA and (3) tumor samples from GBM patients injected with Gd-DTPAResults In cell cultures Gd-DTPA and Gd-DOTA were found in 84 and 56 of the cell nucleirespectively In rat tumors Gd penetrated the nuclei of 47 and 85 of the tumor cells aftersingle and double injection of Gd-DOTA respectively In contrast in human GBM tumors 61of the cell nuclei contained Gd-DTPADiscussion Efficacy of Gd-DTPA and Gd-DOTA as GdNCT agents is predicted to be low due tothe insufficient number of tumor cell nuclei incorporating Gd Although multiple administrationschedules in vivo might induce Gd penetration into more tumor cell nuclei a search for new Gdcompounds with higher nuclear affinity is warranted before planning GdNCT in animal modelsor clinical trials [Neurol Res 2005 27 387ndash398]

Keywords Gd-DOTA Gd-DTPA gadolinium neutron capture therapy (GdNCT) glioblastomaspectromicroscopy synchrotron

INTRODUCTIONGadolinium neutron capture therapy (GdNCT) is apotential treatment for malignant brain tumors The

GdNCT concept which has not yet been clinicallytested is based on two steps (1) the patient is injectedwith a tumor-specific 157Gd compound (2) the tumor isirradiated with thermal or epithermal neutrons Afterneutron capture the 158Gd excited nucleus decaysdelivering localized irradiation Both the injection of theGd compound and thermal neutron irradiation result inonly limited damage to normal brain while the radia-tion dose is considerably higher in the Gd-containingtumor This is due to the high neutron absorption cross

Correspondence and reprint requests to Gelsomina De Stasio Universityof Wisconsin-Madison Department of Physics and SynchrotronRadiation Center 3731 Schneider Drive Stoughton WI 53589 USA[pupasrcwiscedu] Accepted for publication October 20041Present address Department of Chemistry and Biochemistry Universityof Notre Dame IN 46556 USA2Present address Department of Earth and Planetary Science Universityof California Berkeley CA 94708 USA

2005 W S Maney amp Son Ltd Neurological Research 2005 Volume 27 June 387101179016164105X17206

section (ie the probability that an element may capturea neutron) of the 157Gd isotope The absorption crosssection is 254000 barn for 157Gd compared withvalues of 0ndash2 barn for hydrogen nitrogen and oxygenor any other element physiologically present in thebrain and 3840 barn for boron the element mostcommonly employed to date

Why GdNCTGlioblastoma multiforme (GBM) affects 12000

patientsyear in the US and an estimated 300000patients worldwide The incidence and mortality arealmost equal indicating the uniformly fatal outcome ofthis brain malignancy and the need for new therapeuticapproaches The vast majority of GBM patients suc-cumb within 1 year of diagnosis Conventional therapyconsists of maximal resection followed by post-operative radiotherapy to y60 Gy with a mediansurvival of just under 10 months

Although molecular biology-based therapeuticapproaches are the major focus of current clinicalresearch efforts a more immediate impact might beobtained by developing a safe radiation dose-escalationstrategy GdNCT represents one such novel approach

Although the idea of Gd-based NCT was firstformulated in the 1980s12 as an alternative to a similartherapy based on 10B3 its development has suffered dueto lack of appropriate Gd-containing tumor-selectiveagents4 Conventional Gd compounds localize in andaround the abnormal vasculature surrounding tumorsas shown by MRI To date it has been surmised that thisGd remains extracellular relative to the malignantneoplasm itself Until recently sophisticated techniquesto analyze the microscopic distribution of Gd andcorrelate it with the location of cell nuclei were notavailable We recently showed that a Gd-containingMRI contrast agent (Gd-DTPA) penetrates cancer cellnuclei in vitro5 thereby stimulating a reevaluation ofthe GdNCT approach

The complete neutron capture reactionWhen a neutron is captured by 157Gd a highly

localized nuclear reaction occurs producing fiveAuger and Coster-Kronig (ACK) electrons 07 internalconversion electrons 18 X-rays and 08 X-rays withvarying energies and radiation lengths (for more detailson GdNC nuclear reaction see6ndash11) Among these themost abundant and most biologically relevant are thehigh linear-energy-transfer (LET) ACK electrons Theserange in energy between 0 and 50 keV with an averageof 42 keV and average LET of 03 MeVmm Bycomparison in boron neutron capture therapy theaverage LET is 02 MeVmm for both Li and alphaparticles

However the range (or mean free path) of ACKelectrons is limited to 0ndash150 nm (125 nm for theaverage energy of 42 keV)67 Due to this rather shortpath-length of ACK electrons it is necessary for Gdatoms to be localized in the immediate proximity ofDNA ie within the nucleus1 When an ACK electron is

emitted it most likely will interact with a watermolecule within a radius of a few nanometers produ-cing a hydroxyl radical which in turn locally propa-gates the oxidative damage12 Lethal double-strandDNA breaks occur in proportion to the extent of theradiation-induced oxidative damage

Intranuclear Gd detection with synchrotronspectromicroscopy

Conventional techniques such as fluorescence micro-scopy or autoradiography cannot detect Gd-DTPA orGd-DOTA We therefore used synchrotron spectro-microscopic methods to determine the intracellular andintranuclear localization of Gd Spectromicroscopictechniques have now been used for more than a decadefor elemental and oxidation-state analysis in mag-netic1314 tribological15 environmental16ndash19 and cata-lysis20 systems Experiments in biology performed withhard-X-ray fluorescence imaging21 demonstrate excel-lent elemental sensitivity but limited spatial resolutionwhile full-field soft-X-ray transmission microscopyachieves lateral resolution in the nanometer range butlacks elemental sensitivity2223 Scanning transmissionX-ray microscopy (STXM) and secondary ion massspectrometry (SIMS) have been the only approaches toachieve both high resolution and chemical sensitivity incells and tissues24ndash27 The X-ray PhotoElectron Emission(X-PEEM)28 type of spectromicroscopy has now reacheda state of maturity such that it can be effectivelyintroduced into the biomedical field to address highresolution and elemental analysis problems at theforefront of medical research Here we present the firstimages of physiological elements at the subcellular levelobtained with the SPHINX (Spectromicroscope forPHotoelectron Imaging of Nanostructures with X-rays2930 and MEPHISTO (Microscope a Emission dePHotoelectrons par Illumination Synchrotronique deType Onduleur31 instruments at the WisconsinSynchrotron Radiation Center With these instrumentsit is possible to investigate trace element biodistribu-tions such as gadolinium localization into cell nucleifor GdNCT as well as all other physiological elementsThe SPHINX and MEPHISTO X-PEEM results in vitrowere reproduced validated and confirmed using theSTXM spectromicroscope on beamline 1102 at theBerkeley-Advanced Light Source for Gd analysis2627

MATERIALS AND METHODS

Glioblastoma cell culture and cell exposure to GdcompoundsFor in vitro studies we used the T98G32 and TB10human glioblastoma cell lines TB10 is a new tumor cellline established and characterized in our laboratoryDetails on tumor cell culturing and immuno-phenotyp-ing of TB10 have been described elsewhere53334 Wealso used the C635 rat glioma cell line Cells were grownon silicon wafers for X-PEEM spectromicroscopy on100 nm thick Si3N4 windows for STXM spectromicro-scopy and in Petri dishes for inductively coupled

Biodistribution of Gd in glioblastoma cell nuclei Gelsomina De Stasio et al

388 Neurological Research 2005 Volume 27 June

plasma mass spectrometry (ICP-MS) analysis During theculture and exposure period all cell lines were main-tained in a humidified incubator at 37uC and grown inDMEM-F12 medium containing 10 fetal bovineserum 1 penicillin and streptomycin 1 non-essential amino acids Subconfluent cultures wereexposed to one of the magnetic resonance imaging(MRI) contrast agents gadolinium-diethylene triaminepentaacetic acid (Gd-DTPA Magnevist Schering AGBerlin Germany) or gadolinium-tetraazacyclododecane-tetraacetic acid (Gd-DOTA Dotarem Guerbet RoissyCdG Cedex France) The exposure times were 0 6 1224 48 and 72 hours (six time points) and theconcentrations of Gd compounds are shown inTable 1 Control dishes were treated with the sameamount of the vehicle alone

After exposure to the Gd compounds the cell cultureson silicon substrates for X-PEEM analysis were washedthree times with PBS solution to remove free Gd fixedin 4 para-formaldehyde in PBS for 20 minutes doublewashed in Milli-Q-water air dried and ashed in UVO3

for 140 hours36 Parallel cultures were run and treatedidentically for ICP-MS measurements of the bulk Gdconcentration Each set of six samples (one per timepoint 0 6 12 24 48 and 72 h exposure to 18 mM Gd-DTPA or 18 mM Gd-DOTA) was repeated in quad-ruplicate from different cell passages Cell cultures forSTXM analysis on Si3N4 windows were exposed to100 mM Gd-DTPA or 100 mM Gd-DOTA in the culturemedium for 72 h fixed washed and air dried Eachsample was prepared in quadruplicate although at theend of cell culturing and transportation to Berkeley onlyone sample per type had an intact Si3N4 window andcould be analyzed

INDUCTIVELY COUPLED PLASMA MASSSPECTROMETRY (ICP-MS)ICP-MS is a technique to measure total or bulkconcentrations in any given solution For the in vivoexperiments the volume of each tissue block was

accurately measured followed by digestion in a knownvolume of 1 N HNO3 The total amount of Gd wasmeasured using ICP-MS and the results were normalizedto the tissue volume For in vitro cell cultures thevolume measurement is extremely inaccurate and leadsto 100 errors in the final analysis of Gd concentra-tion in cells We therefore counted the number of cellsin a small aliquot from each culture and calculatedthe total number of cells present in the sample (typically2ndash76105 cellssample) The cell volume was thencalculated knowing the average volume of cells foreach cell line These were 62400 TB10 cellsml 74400T98G cellsml and 98000 C6 cellsml Once theaccurate cell volume in each sample was extractedusing these numbers the ICP-MS results could simply benormalized (typical volume of cells 20ndash60 mlsample)This strategy gave reproducible results across fourrepeated cultures with errors usually within 30(Table 1)

INTRACEREBRAL INJECTION OF C6 CELLS ANDMAGNETIC RESONANCE IMAGINGOF RAT TUMORSThe following experiments were performed in accor-dance with the animal care protocol approved by theEthics Committee of the Catholic University of Romeand of the University of Wisconsin-Madison ResearchAnimal Resources Center (RARC)37 Malignant rat C6glioblastoma cells were used for intracerebral trans-plantation in rats Adult male Wistar rats (200ndash250 g)(Catholic University Breeding Laboratory Rome Italy)were anesthetized with intraperitoneal diazepam(Valium 2 mg100 g Roche Milan Italy) followed byintramuscular ketamine (Ketalar 4 mg100 g Parke-Davis Milan Italy) The animal skulls were immobi-lized in a stereotactic head frame and a burr hole wasmade 3 mm right of the midline and 2 mm behind thecoronal suture The tip of a Hamilton microsyringe wasplaced at a depth of 4 mm from the dura in the anteriorcaudate nucleus and the cells were slowly injected Fivehundred thousand cells in 5 ml medium with 10 fetal

Table 1 Exposure of Gd-compounds

Exposurecompound

Cellline

Exposureconcentration

[Gd] concentration in culturemedium (ppm)

Exposure time(h)

[Gd] in cellsiexclSD(ICP-MS) (ppm)

Gd-DTPA TB10 18 mM 2880 0ndash72 0ndash2260iexcl1213

Gd-DTPA T98G 500 mM 80 0ndash72 0ndash862iexcl338


Gd-DTPA C6 500 mM 80 0ndash72 0ndash674iexcl216


Gd-DOTA TB10 18 mM 2830 0ndash72 0ndash5067iexcl1051

Gd-DOTA T98G 500 mM 80 0ndash72 0ndash136iexcl13


Gd-DOTA C6 500 mM 80 0ndash72 0ndash127iexcl28


The experimental paradigms used for the in vitro study Exposure compound cell line exposure concentrations are reported Six exposure times percell culture were used between 0 and 72 hours The formula weights used for [Gd] exposure calculations for Gd-DTPA and Gd-DOTA areFW5546 gmol and 577 gmol respectively The last column shows Gd concentrations obtained with ICP-MS in cells (1 ppm51 mgmllt1 mgg)All uptake curves were similar therefore results are reported only for the 72 h time points All results were reproduced across three or fourindependent cell cultures and the standard deviation is given in the last column


Neurological Research 2005 Volume 27 June 389

calf serum (Gibco BRL Life Technologies Italia MilanoItaly) were injected in each rat Ten days after surgerythe rats were anesthetized and treated with Gd-DOTA(Dotarem 461024 molkg Guerbet Roissy CdGCedex France) via tail vein injection The dose of Gd-DOTA used in this experiment is two-fold the maximumrecommended dosage when using this compound asMRI contrast agents in patients with brain tumors38Fourteen rats were injected with Gd-DOTA once ortwice with the second injection 12 hours after the firstOne hour after the last Gd injection the rats were killedwith an overdose of barbiturate In six rats a region thatincluded the tumor and 2 mm of surrounding tissue wasdissected immersed in OCT and flash-frozen in liquidnitrogen For X-PEEM spectromicroscopy analysis thefrozen tissues were microtomed to 4 mm thick sectionsdeposited on a silicon wafer fixed overnight inparaformaldehyde vapors and ashed for 108 hoursAdjacent sections were deposited on a glass slidestained with hematoxylin and eosin (HampE) and ana-lyzed for histology The remaining portion of each tissueblock was dissolved in 1 N HNO3 and analyzed withICP-MS

In eight additional rats magnetic resonance imagingof formaldehyde-fixed brains was obtained 48 hoursafter brain removal using a 15-tesla whole-body MRscanner (Horizon Echospeed GE Medical SystemMilwaukee WI) The brains were kept in a plastic tubecontaining formaldehyde solution and maintained in thesame position throughout imaging Multi-slice coronalT1-weighted spin-echo images (TR 500 milliseconds TE17 milliseconds NEX 3 field of view 1269 cm2 slicethickness 15 mm space interval 02 mm) wereacquired Then a T2-weighted spin-echo MR imagewas obtained (TR 4000 ms TE 104 ms NEX 3 field ofview 1269 cm2 slice thickness 15 mm space interval02 mm) in the same slice positions After MRI thebrains were embedded in paraffin sectioned at 10 mmon the coronal plane and stained with HampE

Human glioblastoma tumor sample preparationSix GBM patients scheduled for craniotomy and

GBM resection at the Institute of Neurosurgery CatholicUniversity School of Medicine Rome gave informedconsent39 to the injection of Gd-DTPA (161024 molkgMagnevist Schering Milan Italy) 1ndash2 hours beforetumor excision The tumor was immediately embeddedin OCT frozen in isopentane and sectioned on a cryo-microtome at 4 mm Two adjacent tissue sections werecollected from each tumor specimen One section wasdeposited on a glass slide fixed and stained with HampEfor light microscopy The adjacent section was depos-ited on a gold-coated silicon wafer fixed in para-formaldehyde vapors ashed and studied with theMEPHISTO spectromicroscope

X-PEEM spectromicroscopy analysis of cells and tissuesX-PEEM spectromicroscopy was performed both on

C6 tumor specimens of rats injected with Gd-DOTAand on GBM tumor tissues from patients injected with

Gd-DTPA The spectromicroscopy data on physiologi-cal elements as well as Gd distribution were obtainedwith either the SPHINX or MEPHISTO X-PEEM spectro-microscopes installed on the HERMON beamline at theSynchrotron Radiation Center and extracted accordingto the trace-element analysis strategy described at lengthelsewhere29 SPHINX (Elmitec GmbH ClausthalGermany) and MEPHISTO (which we developed) havean optimum lateral resolution of 10 nm and 20 nmrespectively3031 and acquire X-ray absorption spectrawith a resolving power up to 15000 (EDE) in the 60ndash1300 eV energy range For elemental distributions thephosphorus map was obtained by digital ratio of imagesat 139 and 132 eV on-peak and pre-peak of the P2pedge29 A 5ndash10 mm or 10ndash20 mm diameter circular orelliptical region with high P signal was identified as thecell nucleus in rat and human tissues respectively andoutlined for Gd intra-nuclear analysis The nucleus is infact rich in P due to the high density of phosphategroups in DNA The Ca map was obtained dividing animage at 349 eV by one at 346 eV (Ca2p on- and pre-peak respectively) For C1s we ratioed 287 and 282 eVimages for K2p 297 and 295 eV and for Na1s 1068and 1059 eV images For trace concentration Gdanalysis Gd-containing-pixel maps were obtained fromstacks of 81 images across the Gd3d absorption edge at1175 eV and Gd peak areas were extracted from eachpixel This approach is preferable to the digital imageratio described for physiological elements to producedistribution maps of elements occurring in traceamounts with lower noise and artifacts2930 Stacks ofimages or lsquoGd-moviesrsquo acquired between 1220 and1140 eV were binned 464 (therefore the resolution ofthe Gd map is four times lower than in the images Gdpixel size51461452 mm2) to enhance the Gd signalnoise ratio and a Gd3d spectrum was extracted fromeach pixel We then masked the Gd peak region fittedthe spectrum to a 7th order polynomial divided thespectrum by the fitted curve and obtained the Gd peakarea which is proportional to the local Gd concentra-tion We then displayed the Gd peak area in spectrumcolors and superimposed this map on a direct image ofthe cells or tissues

Each Gd-movie was repeated twice and only thereproducible Gd-pixels were retained as Gd-containingQuantitative Gd concentration analysis with SPHINX orMEPHISTO is not possible at present However bymaintaining all experimental and digital analysis para-meters constant the relative concentration of different Gdcompounds in cells and tissues can be directly comparedFrom reference standards we estimate that this spectro-microscopy trace-element analysis has a minimum detec-tion limit of 1000 ppm Cell and tissue ashing increases[Gd] by a factor of 10 therefore the detection limit is onthe order of 100 ppm or lower in all samples analyzedhere Since this detection limit is greater than the Gdconcentration necessary for GdNCT (24 ppm or 13 ppmsee Discussion) in our analysis we accepted one Gd-containing pixelnucleus as a sufficient amount forGdNCT29 For both in vitro and in vivo experiments weidentified the cell nuclei from the P distribution map



counted the number of cell nuclei containing at least 1 Gdpixel and evaluated the effectiveness of each Gdcompound accordingly Due to the time-consumingnature of the X-PEEM Gd data collection and processingand also due to its low sensitivity for in vitro experimentsonly TB10 cells exposed to the two compounds for 0ndash72 hat the greatest concentrations were analyzed For thesethe bulk concentration reached 2000 ppm for Gd-DTPAand 5000 ppm for Gd-DOTA (see Table 1) Ashing inUVO3 selectively removes carbon a major element incells and tissues and consequently enhances the relativeconcentration of the other elements Ashing takes place atair pressure and temperature and slowly (100ndash200 hours)lsquoflattensrsquo the cell morphology36 It is particularly usefulwhen the element to be localized by X-PEEM is present intrace concentrations otherwise undetectable

STXM spectromicroscopy analysis of in vitro cellcultures

Due to the use of an undulator source and theoptimization of beamline 1102 at the ALS for the Gd3dabsorption energy (y1200 eV) the STXM spectro-microscope is much faster than X-PEEM for Gddetection and much more sensitive This instrumentcan in fact detect a single monolayer of Gd atomsAshing of the cell cultures for STXM analysis thereforewas not required and the exposure concentrationscould be kept low at 100 mM Gd-DTPA and Gd-DOTA In addition higher resolution in Gd and cellimaging was also possible on this novel instrumentdesigned and developed by one of us (TT) and histeam2627 For Gd detection in cells we acquired imageson- and off-peak at 1183 eV and 1178 eV respectivelythen ratioed these two images to obtain the Gddistribution map and superimposed it upon one of thesample images Although this instrument does not allowthe detection of P for nuclei identification the nucleiappear thicker and denser in transmitted X-rays andtherefore their outlines are easily located

RESULTS

X-PEEM spectromicroscopy physiological elements incultured GBM cellsDespite the extensive manipulation including fixationair drying and ashing the samples analysis of themorphology and elemental composition of cells demon-strated that we induced no elemental displacement atthe detectable X-PEEM resolution and that micro-analysis of cells and tissues prepared with this modalityis feasible Figure 1 shows the subcellular distribution ofselected physiological elements in the TB10 humanGBM cell line Results indicate a higher concentrationof P in cell nuclei as expected given the high density ofphosphate groups in nucleic acids The calcium mapshows that the highest Ca concentration surrounds thenuclei as anticipated in the endoplasmic reticulum andGolgi apparatus We observed a low concentration ofcarbon in the central region of each cell whichrepresents residual carbon not completely removed by

ashing Potassium and sodium maps exhibited highernoise corresponding to the lower concentrations ofthese elements compared with P and Ca Neither P norCa were observed on the substrate supporting thenotion that elements were not re-distributed duringashing

X-PEEM and STXM spectromicroscopy results Gddistribution in cultured GBM cells

Both Gd-DTPA and Gd-DOTA are widely used MRIcontrast agents that transiently lsquoenhancersquo GBM tumorsas a result of a disrupted bloodndashbrain barrier Werecently analyzed several GBM cell culture samplesexposed to Gd-DTPA5 and found that Gd actuallypenetrates the nucleus in vitro Encouraged by theseresults we extended our research to include Gd-DOTAand analyzed more extensively the Gd-DTPA biodis-tribution in the attempt to optimize the choice of Gdcompounds for GdNCT

In Figure 2 we show representative results ofthe intracellular distribution of Gd-DTPA and Gd-DOTA in cultured tumor cells Following exposure toGd-DTPA Gd was found both in the nucleus and in thecytoplasm (Fig 2a) but the Gd concentration washigher in nuclei (predominantly green vs yelloworange and red pixels) After exposure to Gd-DOTAintracellular Gd was present at higher concentrationas compared with Gd-DTPA (magenta and blue pixelsin Figure 2B) It is strikingly apparent from the Gdlocation map of Figure 2b that Gd-DOTA is oftenexcluded from the nuclei at least within the sensitivitylimit of SPHINX The pixel value (color) indicates alocally higher Gd signal for the Gd-DOTA sample inagreement with the bulk concentrations reported inTable 1

Evaluation of Gd maps for the presence of at least oneGd pixel in the nucleus showed that at the 72-hour timepoint Gd-DTPA was found in 84 of the nuclei in 76cells analyzed and Gd-DOTA in 56 of the 91 nucleianalyzed Both results were collected across fourindependent cell cultures from different passages

STXM analysis of TB10 and T98G cells exposed to100 mM Gd-DTPA and 100 mM Gd-DOTA confirmedand validated these results Figure 3 shows the STXMimage of two TB10 cells exposed to Gd-DOTA andimaged at 1183 eV and the corresponding Gd distribu-tion map The distribution map shows most Gd pixelssurrounding the cell cytoplasm but also in the nucleusparticularly in the nucleoli Overall we found Gd ineight of the 16 TB10 cells and in 12 of the 20 T98Gcells exposed to Gd-DOTA and analyzed with STXMThese correspond to 50 and 60 respectively of thenuclei containing Gd after exposure to Gd-DOTAThese numbers closely resemble the results found withX-PEEM (56) STXM data (not shown) on TB10 cellsexposed to Gd-DTPA showed Gd always present inhigher concentration in the nucleus than in thesurrounding cytoplasm Twenty-four of the 28 cellsimaged contained Gd (86) which closely matches theX-PEEM result (84) For both compounds therefore



X-PEEM and STXM spectromicroscopy data validate andcorroborate each other

Gd-DOTA in cells analyzed with both spectro-microscopies X-PEEM and STXM shows that Gdaccumulates with higher concentration and higher pixeldensity in the cytoplasm than in the nucleus (Figures 2band 3b) Exceptions to this observation are nucleolithese are visible in 10 GBM cells but when theyare they always contain a higher concentration of Gdcompared with the rest of the nucleus but still lowerthan in the cytoplasm (eg Figure 3b)

The highest Gd pixel density and concentration in thecytoplasm is observed in discrete locations or spots inthe perinuclear region These could be subcellular organ-elles or simply molecular aggregates of Gd-DOTA We

Figure 1 (a) Image of human GBM cells acquired by X-PEEM spectromicroscopy usingthe SPHINX instrument with 1200 eV photon energy illumination (b) Phosphorus (c) cal-cium (d) carbon (e) potassium and (f) sodium distribution maps in the same cells In alldistribution maps black or dark color indicate higher element concentration white corre-sponds to no element

Figure 2 Gd-pixel location maps acquired with the MEPHISTO (a) and SPHINX (b) spectromi-croscopes from human GBM TB10 cells exposed to (a) Gd-DTPA and (b) Gd-DOTA for72 hours The spectrum-color intensity scale indicates the highest Gd concentration inmagenta the lowest in red Both Gd maps (pixel size 2 mm2) are fused on SPHINX grayscaleimages of the same cells Nuclei are darker

Figure 3 (a) Two TB10 cells exposed to Gd-DOTA and imagedwith the STXM spectromicroscope at 1183 eV (b) Gd map in thesame cells obtained by ratio of 1178 and 1183 eV images Colorbar values are estimated from Gd density and X-ray absorptionmeasurements (in gcm2) The color Gd map is superimposed onthe STXM image recorded at 1183 eV



currently have no experimental proof supporting eitherinterpretation Analyzing only the nuclei found to containGd we find very few of these Gd spots occurring at a muchlower density than in the surrounding cytoplasm It isconceivable that these are above the nucleus in thecytoplasm and the STXM experiment which works intransmission detects them in the nucleus Similarly in theX-PEEM experiment on ashed and flattened cells if Gd wasabove the nucleus before ashing it would appear in thenucleus after ashing The cytoplasmic layer above thenucleus is considerably thinner than the portion surround-ing the nucleus which partially explains the lower numberand density of spots we observe

We note that the observation of highest Gd in thecytoplasm for Gd-DOTA and in the nucleus forGd-DTPA eliminates the possibility of experimentalartifacts or misinterpretations such as cell surface-bound Gd

MRI ICP-MS and X-PEEM analysis of rat tumor tissuesAll rats injected with Gd-DOTA showed intense

signal enhancement in the brain identified as tumor inT1-weighted MR images However the area of signalenhancement was not restricted to the tumor itself butalso included the surrounding brain tissue whereperitumoral edema was identified on histologicalsections (Figure 4ab) In addition those rats receivingdouble Gd-DOTA injection showed presence of Gd inthe cerebro-spinal fluid (CSF) as demonstrated by thehyperintense signal in the lateral ventricles on T1-weighted MR images (Figure 4e) This finding suggests

that diffusion of Gd to the extracellular space of thebrain which is in continuity with the CSF did occurafter repeated Gd-DOTA injections

The ICP-MS results of Gd bulk concentration in rattumors after administration of Gd-DOTA are reportedin Table 2 Comparison of SPHINX images with theadjacent histological sections confirmed that denserdarker regions (P-rich) corresponded to hematoxylinstained nuclei (Figure 5) On SPHINX analysis Gdwas homogeneously distributed throughout the tumortissue and could be identified both in and betweennuclei (Figure 5) Note that the distribution map of Gdappears so homogeneously random that it could bemisinterpreted as noise We carefully verified that eachpixel corresponded to well-resolved Gd3d peaks andthat it was reproducible across two subsequent Gdmovies

Table 2 reports the overall results of the spectro-microscopy study at different sites that included thetumor itself the peritumoral edema region andnormal brain tissue (Figure 6) In the tumor region wefound that Gd is present in 54 and 85 of cell nucleiafter single and double Gd-DOTA injection respec-tively (Table 2) In the case of double Gd-DOTAinjection however we notice that Gd is present morefrequently in cell nuclei of cells in the edema tissue thanin tumor Twelve hours after the second Gd-DOTAinjection there are still more Gd-containing nuclei inthe region of brain edema than in the tumor itself Thesefindings suggest that in this rat the clearance of Gd-DOTA from edematous brain tissue may have beeninhibited

Figure 4 Histological and MRI studies of intracerebrally implanted C6 glioma tumor in rats C6 cells were stereotacticallyinjected in the right striatum The upper panel (a b c) refers to a rat injected with a single dose of Gd-DOTA the lowerpanel (d e f) to a rat injected with double dose of Gd-DOTA (a) and (d) Coronal histological sections through the tumorepicenter stained with HampE (original magnification 6125) (b) and (e) T1-weighted MRI images of the same specimensshown in (a) and (d) The tumors appear hyperintense after Gd-DOTA administration however some Gd enhancement isalso localized in the edema region surrounding the tumor In addition the specimen with double Gd injection (e) shows ahyperintense signal which highlights the profile of the cerebral ventricles (c) and (f) T2-weighted MRI images of the sametissue specimen showing the typical hyperintense signal of the CSF in the ventricles



X-PEEM spectromicroscopy of Gd distribution in humanGBM tissues

Figure 7 shows the results of the spectromicroscopicGd analysis in human GBM tumor tissue The Gd-pixellocation was obtained using the same parameters as inFigures 2ab and 6b and the Gd concentrations can bedirectly compared The same navigation system illu-strated in Figure 5 allowed us to identify the areas in thetwo adjacent sections so that the same portion of tissuecould be examined for histology first (Figure 7a) and thenimaged by MEPHISTO (Figure 7b) The phosphorusdistribution map was superimposed on the MEPHISTOimage to locate the cell nuclei which are identified by thehigher P concentration (Figure 7c) Systematic analysis ofthe tissue area in Figure 7 and a few more cells around itshowed that after a single injection of Gd-DTPA Gdentered the nucleus of 12 of tumor cell nuclei (15 out of123 cell nuclei) This is one of the most Gd-rich areas weidentified In Table 3 we report the results on all tissuesanalyzed From Table 3 it is evident that the percentage ofnuclei with Gd does not correlate with time betweeninjection and tumor excision but it correlates with the Gd

concentration measured by ICP-MS This indicates asexpected that the greater the concentration in GBMtissue the greater the number of nuclei reached by Gd

Overall we analyzed 60 different circular areas withdiameter 200ndash300 mm from all six patient tissues andfound that 61 of the cancer cell nuclei contained atleast one Gd pixel (135 out of 2217) These datademonstrate that Gd injected in the blood stream in theform of Gd-DTPA reaches glioblastoma cells andpenetrates the cell and nuclear membranes Howeverfrom the perspective of GdNCT the nuclear localizationof Gd from Gd-DTPA was unsatisfactory because only61 of the nuclei analyzed contained Gd

Interestingly this in vivo value falls on the curveobtained from in vitro studies as shown in Figure 8

DISCUSSION

Theoretical requisites for GdNCT efficacyAlthough many authors pointed out that the long-rangeeffects of the GdNC reaction should not be neglected40

Table 2 ICP -MS and SPHINX analysis of rat tissues

nuclei containing Gd (total no of nuclei analyzed)

Sample ID and description [Gd] in tissue (ppm) Tumor Edema Normal brain

C6 1a (Gd1 extr after 1 h) 56

C6 1b (Gd1 extr after 1 h) 42 54 (287 nuclei) 39 (67 nuclei) 24(146 nuclei)

C6 2 (Gd1 extr after 1 h) 38

C6 3 (Gd2 extr after 12 h) 61 47 (177 nuclei) 66 (32 nuclei) Not available




ICP-MS results obtained in tumors from six rats injected with Gd-DOTA Samples 1a and 1b were extracted from the same rat Gd1 indicates 1injection Gd2 indicates two injections 12 hours apart The three samples analyzed with SPHINX and the corresponding results divided by tissueregions are reported in the three columns on the right

Figure 5 (a) Sample C61b on silicon imaged in visible light after ashing and (b) the adjacent sectionHampE stained The areas analyzed in SPHINX are indicated with circles Black dots indicate distinctivefeatures used to identify micro-locations and determine the correspondence of visible and SPHINXimages The resulting frame is also given in which the X and Y coordinates accurately correspond tothe SPHINX sample positioning micrometers



intranuclear presence of Gd is key to the developmentof GdNCT Before performing the present experimentson Gd microdistribution our hypothesis was that one ormore Gd-contrast agents might be identified as usefulNCT agents as well contingent upon uptake in at least90 of the tumor cell nuclei This criterion is based on aprevious demonstration by Fowler et al that ifcytotoxicity is dependent on intracellular incorporation

of a test agent the factor limiting its success is theproportion of tumor cells not incorporating the testagent and therapeutic benefit is unlikely to be achievedif more than 10 of the cells remain unlabeled41 ForGdNCT one or two neutron capture events close toDNA would be sufficient to kill a cell however enoughGd molecules should be present in the nucleus ofvirtually every malignant cell in order to achieve a

Figure 6 Two adjacent tumor tissue sections from a rat injected with Gd-DOTA and killed 1h later (C6-1b inTable 2) (a) A visible light image of the HampE stained section shows the density of nuclei in the GBM tumor(b) SPHINX image of the same area in the adjacent section after ashing which shows similar featuresSuperimposed on this image is the map of Gd-containing pixels and relative concentrations displayed with thesame spectral color palette of Figure 2 Bars520 mm

Figure 7 (a) Light microscopy image of a human glioblastoma tissue section stained with HampE from sam-ple 35 in Table 3 (b) MEPHISTO image of the same region as in (a) obtained at 130 eV photon energy(a) and (b) acquired on two adjacent tissue sections are similar but not identical because the tissue sec-tions are 4 mm thick while the size of the nuclei is 5ndash10 mm (c) Phosphorus distribution map in greento outline the cell nuclei (d) Gd-location map superimposed on the outlined cell nuclei



therapeutically relevant effect The amount of Gdnecessary for GdNCT efficacy can be calculated fromthe formula

GdNC event~n W s

where GdNC events5number of Gd neutron captureeventsnucleus n5number of Gd atomsnucleusW5neutron fluence5109 ncm2s63h51161013 ncm2which is clinically feasible42ndash43 and s5157Gd absorp-tion cross-section for thermal neutrons5254000 barn5025610218 cm2

One neutron capturenucleus can be achieved if46105 Gd atomsnucleus are present which corre-sponds to [Gd]501 ppm Two of us (JFF and JFC)independently calculated that assuming a Poissondistribution of Gd molecules in cell nuclei as a resultof uptake from the vasculature there must be anaverage of at least 24 GdNC events per nucleus

([Gd]gt24 ppm in the nucleus or using the naturalGd isotopic mixture with s549000 barn [Gd]gt13 ppm) in order to be sure that less than one in the1010 cellstumor has zero events

Although quantitative X-PEEM spectromicroscopicanalysis is not yet available from a comparison withthe ICP-MS bulk concentrations we estimate that thedetection limit is on the order of 1000 ppm in ashedcells Therefore if we detect Gd in one pixel that pixelarea (2 mm2) must have contained at least 100 ppm Gdin the cell before ashing which is sufficient for GdNCTand well above the modeled average of 13 ppm forGdNCT efficacy Our goal is therefore the detection ofat least one Gd spot in the nucleus of at least 90 of thetumor cells

In the human in vivo experiments however wefound that only 61 of the cell nuclei contain Gdpixels after one single injection of Gd-DTPA Cellculture experiments might offer a way to compare invivo and in vitro data Figure 8 shows the nuclearuptake of Gd in cultured GBM cells exposed to Gd-DTPA as a function of exposure time In this diagramthe percentage of Gd nuclear uptake in the in vivocondition falls onto the uptake curve obtained in vitro asa function of time when one assumes that (a) the timebetween injection and surgical tumor removal corre-sponds to the exposure time (b) the blood concentrationof Gd remains constant over this time Although thecalculation of [Gd] in the blood supplying the tumormight suffer from inaccurate assumptions we estimatethis value to be y124ndash132 ppm In the in vitro studythe exposure concentration for cells was 18 mM or10 mgml Gd-DTPA which corresponds to[Gd]52880 ppm Therefore in the in vivo conditionsthe [Gd] in blood is well below the [Gd] in the culturemedium

Hypothetically by extending the time of Gd expo-sure the nuclear uptake in vivo might significantlyincrease Longer exposure times can be reproduced invivo by continuous infusion or multiple administrationsof Gd compounds to patients The experience withGBM and brain metastases patients demonstrates thatmultiple administrations of a different Gd compound

Table 3 In vivo human cases injected with 30 mgkg Gd-DTPA

SampleID

Time between injectionand tumor excision (h)

ICP-MS [Gd]iexclSt Dev(ppm)

MEPHISTO analysisareas with Gdanalyzed

areas ()

MEPHISTO analysisnuclei with Gdanalyzed

nuclei ()

35 1 100iexcl20 411 (36) 48406 (118)

37 25 90iexcl20 17 (14) 28287 (97)

36 2 47iexcl09 16 (17) 17182 (93)

40 25 18iexcl04 513 (38) 21433 (48)

39 3 15iexcl03 414 (28) 14458 (3)

38 2 06iexcl01 29 (22) 7451 (16)

Average 1760 (28) 1352217 (61)

Gadolinium concentration in tumor tissue extracted from six GBM patients injected pre-operatively with Gd-DTPA measured by ICP-MS andvisualized by MEPHISTO spectromicroscopy The MEPHISTO results of the fourth column represent the ratio of the 200ndash300 mm-diameter tissueareas in which Gd was observed over the total number of sections analyzed The last column gives the number of nuclei in which Gd was found overthe total number of nuclei analyzed for each patient (tissue section)

Figure 8 Diagram showing the relationship between the percen-tages of nuclei in which Gd was detected as in Figure 2a and thetime of exposure of GBM cells to Gd-DTPA (solid circles) In thehuman GBM cases injected with Gd-DTPA (empty circle atthe average 22 hour 61 nuclei) the value falls onto the in vitrocurve



Motexafin-gadolinium (MGd) lead to intratumoralaccumulation and retention even 2 months after thelast administration44 If the in vitrondashin vivo correspon-dence of Figure 8 is reliable one might extrapolate thatafter a 72-hour continuous infusion of Gd-DTPA topatients with GBM 84 of the tumor cell nuclei wouldcontain Gd a value which approaches but still does notreach the goal of 90 In general if a Gd compounddoes not target a sufficient number of tumor cell nucleiin vitro it is unlikely to do so in vivo Furthermorecontinuous or repeated Gd-infusions may lead to anexcess of Gd in the blood stream at the time ofneutron exposure and undesirable side effects andcomplications

Our in vitro studies with Gd-DOTA showed that thepotential of Gd-DOTA as GdNCT agent is not greaterthan for Gd-DTPA After cell exposure to 18 mM Gd-DOTA for 72 hours only 56 of the cell nucleicontained Gd In vivo the percentage of nucleicontaining Gd varied between 47 and 85 aftersingle or double injection of Gd-DOTA respectivelyThus with repeated Gd-DOTA injections a higherpercentage of nuclei did contain Gd Howeverrepeated Gd-DOTA injections resulted in Gd extravasa-tion in the CSF as well as in the edematous white mattersurrounding the tumor as demonstrated both by MRIand spectromicroscopy This may have relevant patho-logical effects following neutron irradiation such asependymitis or radionecrosis

Comparison of X-PEEM and STXM in vitro datareveals an unexpected result the percentage of nucleiwith Gd may not depend on the exposure concentra-tion The latter was18 mM and 100 mM for X-PEEM andSTXM experiments respectively for both Gd com-pounds Despite this significant difference the percen-tage of nuclei taking up Gd after 72 hours did not varysignificantly across the two experiments However thisobservation may be hindered by the difference insensitivity of the two techniques and cannot at thistime lead to any further conclusions

In conclusion GdNCT had been previously aban-doned because evidence for intracellular and intra-nuclear Gd localization was lacking There is now astrong motivation for revisiting GdNCT since it becamepossible to analyze the intracellular Gd localization indetail and to evaluate the uptake and biodistribution ofGd compounds before clinical testing Our unprece-dented in vitro and in vivo Gd spectromicroscopy datain human and rat glioma tumors provide evidencethat neither Gd-DTPA nor Gd-DOTA are suitableagents for GdNCT since these compounds do nottarget a sufficient number of tumor cell nuclei Despitethe rejection of these agents our results validate theauthenticity reproducibility and versatility of thespectromicroscopy approach to determine the biodis-tribution of Gd in vitro and in vivo In the future whenanother GdNCT agent targeting 90 of tumor cellnuclei is identified with spectromicroscopy pre-clinical animal and early clinical testing with thermaland epithermal neutron irradiation should proceed todetermine the efficacy of GdNCT

ACKNOWLEDGEMENTSThis work was supported by the University of Wisconsin ComprehensiveCancer Center UW-Graduate School Romnes Award and Department ofPhysics X-PEEM spectromicroscopy experiments were conducted at theUW-Synchrotron Radiation Center supported by the NSF under AwardNo DMR-0084402 The ALS and work on the beamline 1102 STXM issupported by the Director Office of Science Office of Basic EnergySciences Division of Materials Sciences and the Division of ChemicalSciences Geosciences and Biosciences of the US DOE at LBNL underContract No DE-AC03-76SF00098

REFERENCES1 Martin RF DrsquoCunha G Pardee M et al Induction of DNA double-

strand breaks by 157Gd neutron capture Pigment Cell Res 1989 2330ndash332

2 Brugger RM Shih JA Evaluation of gadolinium-157 as a neutroncapture therapy agent Strahlenther Onkol 1989 165 153ndash156

3 Smith DR Chandra S Coderre JA et al Ion microscopy imaging of10B from p-boronophenylalanine in a brain tumor model for boronneutron capture therapy Cancer Res 1996 56 4302ndash4306

4 Masiakowski JT Horton JL Peters LJ Gadolinium neutron capturetherapy for brain tumors a computer study Med Phys 1992 191277ndash1284

5 De Stasio G Casalbore P Pallini R et al Gadolinium in humanglioblastoma cells for gadolinium neutron capture therapy CancerRes 2001 61 4272ndash4277

6 Goorley T Nikjoo H Electron and photon spectra for threegadolinium-based cancer therapy approaches Radiat Res 2000154 556ndash563

7 Powell CJ Jablonski A NIST Electron Inelastic-Mean-Free-PathDatabase ndash Version 11 2000 National Institute of Standards andTechnology Gaithersburg MD

8 Greenwood RC Reich CW Baader HA et al Collective and two-quasiparticle states of 158Gd observed through study of radiativeneutron capture in 157Gd Nucl Phys 1978 A304 327ndash428

9 Gierga DP Yanch JC Shefer RE An investigation of the feasibilityof gadolinium neutron capture synovectomy Med Phys 2000 271685ndash1692

10 Miller GA Hertel NE Wehring WB et al Gadolinium neutroncapture therapy Nucl Technol 1993 103 320ndash331

11 X- and gamma-ray mass attenuation coefficients and distances inbrain were retrieved from the NIST tables available at httpphysicsnistgovPhysRefDataXrayMassCoefComTabbrainhtml

12 Hall EJ DNA strand breaks and chromosomal aberrations InRadiobiology for the Radiologist 5th edn Philadelphia LippincottWilliams amp Wilkins 2000 pp 19ndash20

13 Scholl A Stohr J Luning J et al Observation of antiferromagneticdomains in epitaxial thin films Science 2000 287 1014ndash1016

14 Stohr J Wu Y Hermsmeier BD et al Element-specific magneticmicroscopy with circularly polarized x-rays Science 1993 25658ndash661

15 Canning GW Suominen Fuller ML Bancroft GM et alSpectromicroscopy of tribological films from engine oil additivesPart I Films from zddprsquos Tribology Lett 1999 6 159ndash169

16 Chan CS De Stasio G Banfield JF et al Microbial polysaccharidestemplate assembly of nanocrystal fibers Science 2004 303 1656ndash1658

17 Labrenz M De Stasio G Banfield JF et al Formation of sphalerite(ZnS) deposits in natural biofilms of sulfate-reducing bacteriaScience 2000 290 1744ndash1747

18 Beard BL Johnson CM Cox L et al Iron isotope biosignaturesScience 1999 285 1889ndash1892

19 Feng X Fryxell GE Wang L-Q et al Functionalized monolayerson ordered mesoporous supports Science 1997 276 923ndash926

20 Rotermund HH Engel W Kordesch M Ertl G Imaging ofspatiotemporal pattern evolution during carbon-monoxide oxida-tion of platinum Nature 1990 343 355ndash357

21 Pickering IJ Prince RC Salt DE et al Quantitative chemicallyspecific imaging of selenium transformation in plants Proc NatlAcad Sci USA 2000 97 10717ndash10722

22 Larabell CA Le Gros MA X-ray tomography generates 3-Dreconstructions of the yeast Saccharomyces cerevisiae at 60 nmresolution Mol Biol Cell 2003 15(3) 957ndash62



23 Myneni SCB Brown JT Martinez GA et al Imaging of humicsubstance macromolecular structures in water and soils Science1999 286 1335ndash1337

24 Osanna A Jacobsen C Kalinovsky A et al X-ray microscopypreparations for studies of frozen hydrated samples ScanningMicrosc 1996 10 (suppl) 349ndash356

25 Smith DR Chandra S Barth RF et al Quantitative imaging andmicrolocalization of boron-10 in brain tumors and infiltrating tumorcells by SIMS ion microscopy relevance to neutron capture therapyCancer Res 2001 61 8179ndash8187 and the references therein

26 Kilcoyne ALD Tyliszczak T Steele WF et al Interferometer-controlled scanning transmission X-ray microscopes at theadvanced light source J Synchrotron Rad 2003 10 125ndash136

27 Tyliszczak T Warwick T Kilcoyne ALD et al Soft X-ray scanningtransmission microscope working in an extended energy range atthe advanced light source AIP Conference Proceedings 2004705 1356ndash1359

28 Bauer E Photoelectron microscopy E J Phys Cond Matter 200113 11391ndash11404

29 Frazer BH Sonderegger BR De Stasio G et al Mapping ofphysiological and trace elements with X-PEEM Proceedings of the2002 X-ray Microscopy Conference J Phys Fr IV 2003 104 349ndash352

30 Frazer BH Girasole M De Stasio G et al Spectromicroscope forthe PHotoelectron Imaging of Nanostructures with X-rays(SPHINX) performance in biology medicine and geologyUltramicroscopy 2004 99 87ndash94

31 De Stasio G Perfetti L Gilbert B et al The MEPHISTOspectromicroscope reaches 20 nm lateral resolution Rev SciInstrum 1999 70 1740ndash1742

32 Stein GH T98G An anchorage-independent human tumor cellline that exhibits stationary phase G1 arrest in vitro J Cell Physiol1979 99 43ndash54

33 Pallini R Casalbore P Mercanti D et al Phenotypic change ofhuman cultured meningioma cells J Neuro-Oncol 2000 49 9ndash17

34 Falchetti ML Pierconti F Pallini R et al Glioblastoma inducesvascular endothelial cells to express telomerase in vitro CancerRes 2003 63 3750ndash3754

35 Albasanz JL Ros M Martin M Characterization of metabotropicglutamate receptors in rat C6 glioma cells Eur J Pharmacol 1997326 85ndash91

36 Gilbert B Perfetti L Hansen R et al UV-ozone ashing of cells andtissues for spatially resolved trace element analysis Front Biosci2000 5 10ndash17 On-line at httpwwwbioscienceorg2000v5agilbertfulltexthtm

37 University of Wisconsin-Madison Research Animal ResourcesCenter (RARC) Animal Protocol A-48-6700-M01613-4-03-02approved 04082002

38 wwwbiam2orgwwwSpe29085html39 University of Wisconsin Institutional Review Board Human

Subjects Committee approved Protocol 2000-212740 Stalpers L Stecher-Rasmussen F Kok T et al Radiobiology of

gadolinium neutron capture therapy In Sauerwein S Moss RWittig A eds Research and Development in Neutron CaptureTherapy Bologna Italy Monduzzi 2002 pp 825ndash830

41 Fowler JF Kinsella TJ The limiting radiosensitisation of tumours byS-phase sensitisers Br J Cancer 1996 74 (Suppl XXVII) S294ndashS296

42 Riley KJ Binns PJ Harling OK Performance characteristics of theMIT fission converter based epithermal neutron beam Phys MedBiol 2003 48 943ndash958

43 Verbakel WFAR Sauerwein W Hideghety K et al Boronconcentrations in brain during boron neutron capture therapy Invivo measurements from the Phase I trial EORTC 11961 using agamma-ray telescope Int J Radiat Oncol Biol Phys 2003 55 743ndash756

44 Mehta MP Shapiro WR Glantz MJ et al Lead-in phase torandomized trial of motexafin gadolinium and whole-brainradiation for patients with brain metastases centralized assessmentof magnetic resonance imaging neurocognitive and neurologicend points J Clin Oncol 2002 20 3445ndash3453



section (ie the probability that an element may capturea neutron) of the 157Gd isotope The absorption crosssection is 254000 barn for 157Gd compared withvalues of 0ndash2 barn for hydrogen nitrogen and oxygenor any other element physiologically present in thebrain and 3840 barn for boron the element mostcommonly employed to date

Why GdNCTGlioblastoma multiforme (GBM) affects 12000

patientsyear in the US and an estimated 300000patients worldwide The incidence and mortality arealmost equal indicating the uniformly fatal outcome ofthis brain malignancy and the need for new therapeuticapproaches The vast majority of GBM patients suc-cumb within 1 year of diagnosis Conventional therapyconsists of maximal resection followed by post-operative radiotherapy to y60 Gy with a mediansurvival of just under 10 months

Although molecular biology-based therapeuticapproaches are the major focus of current clinicalresearch efforts a more immediate impact might beobtained by developing a safe radiation dose-escalationstrategy GdNCT represents one such novel approach

Although the idea of Gd-based NCT was firstformulated in the 1980s12 as an alternative to a similartherapy based on 10B3 its development has suffered dueto lack of appropriate Gd-containing tumor-selectiveagents4 Conventional Gd compounds localize in andaround the abnormal vasculature surrounding tumorsas shown by MRI To date it has been surmised that thisGd remains extracellular relative to the malignantneoplasm itself Until recently sophisticated techniquesto analyze the microscopic distribution of Gd andcorrelate it with the location of cell nuclei were notavailable We recently showed that a Gd-containingMRI contrast agent (Gd-DTPA) penetrates cancer cellnuclei in vitro5 thereby stimulating a reevaluation ofthe GdNCT approach

The complete neutron capture reactionWhen a neutron is captured by 157Gd a highly

localized nuclear reaction occurs producing fiveAuger and Coster-Kronig (ACK) electrons 07 internalconversion electrons 18 X-rays and 08 X-rays withvarying energies and radiation lengths (for more detailson GdNC nuclear reaction see6ndash11) Among these themost abundant and most biologically relevant are thehigh linear-energy-transfer (LET) ACK electrons Theserange in energy between 0 and 50 keV with an averageof 42 keV and average LET of 03 MeVmm Bycomparison in boron neutron capture therapy theaverage LET is 02 MeVmm for both Li and alphaparticles

However the range (or mean free path) of ACKelectrons is limited to 0ndash150 nm (125 nm for theaverage energy of 42 keV)67 Due to this rather shortpath-length of ACK electrons it is necessary for Gdatoms to be localized in the immediate proximity ofDNA ie within the nucleus1 When an ACK electron is

emitted it most likely will interact with a watermolecule within a radius of a few nanometers produ-cing a hydroxyl radical which in turn locally propa-gates the oxidative damage12 Lethal double-strandDNA breaks occur in proportion to the extent of theradiation-induced oxidative damage

Intranuclear Gd detection with synchrotronspectromicroscopy

Conventional techniques such as fluorescence micro-scopy or autoradiography cannot detect Gd-DTPA orGd-DOTA We therefore used synchrotron spectro-microscopic methods to determine the intracellular andintranuclear localization of Gd Spectromicroscopictechniques have now been used for more than a decadefor elemental and oxidation-state analysis in mag-netic1314 tribological15 environmental16ndash19 and cata-lysis20 systems Experiments in biology performed withhard-X-ray fluorescence imaging21 demonstrate excel-lent elemental sensitivity but limited spatial resolutionwhile full-field soft-X-ray transmission microscopyachieves lateral resolution in the nanometer range butlacks elemental sensitivity2223 Scanning transmissionX-ray microscopy (STXM) and secondary ion massspectrometry (SIMS) have been the only approaches toachieve both high resolution and chemical sensitivity incells and tissues24ndash27 The X-ray PhotoElectron Emission(X-PEEM)28 type of spectromicroscopy has now reacheda state of maturity such that it can be effectivelyintroduced into the biomedical field to address highresolution and elemental analysis problems at theforefront of medical research Here we present the firstimages of physiological elements at the subcellular levelobtained with the SPHINX (Spectromicroscope forPHotoelectron Imaging of Nanostructures with X-rays2930 and MEPHISTO (Microscope a Emission dePHotoelectrons par Illumination Synchrotronique deType Onduleur31 instruments at the WisconsinSynchrotron Radiation Center With these instrumentsit is possible to investigate trace element biodistribu-tions such as gadolinium localization into cell nucleifor GdNCT as well as all other physiological elementsThe SPHINX and MEPHISTO X-PEEM results in vitrowere reproduced validated and confirmed using theSTXM spectromicroscope on beamline 1102 at theBerkeley-Advanced Light Source for Gd analysis2627

MATERIALS AND METHODS

Glioblastoma cell culture and cell exposure to GdcompoundsFor in vitro studies we used the T98G32 and TB10human glioblastoma cell lines TB10 is a new tumor cellline established and characterized in our laboratoryDetails on tumor cell culturing and immuno-phenotyp-ing of TB10 have been described elsewhere53334 Wealso used the C635 rat glioma cell line Cells were grownon silicon wafers for X-PEEM spectromicroscopy on100 nm thick Si3N4 windows for STXM spectromicro-scopy and in Petri dishes for inductively coupled










Exposurecompound

Cellline



Exposure time(h)




























RESULTS

































DISCUSSION























GdNC event~n W s








SampleID




areas ()


nuclei ()

35 1 100iexcl20 411 (36) 48406 (118)

37 25 90iexcl20 17 (14) 28287 (97)

36 2 47iexcl09 16 (17) 17182 (93)

40 25 18iexcl04 513 (38) 21433 (48)

39 3 15iexcl03 414 (28) 14458 (3)

38 2 06iexcl01 29 (22) 7451 (16)

Average 1760 (28) 1352217 (61)


































































Exposurecompound

Cellline



Exposure time(h)




























RESULTS

































DISCUSSION























GdNC event~n W s








SampleID




areas ()


nuclei ()

35 1 100iexcl20 411 (36) 48406 (118)

37 25 90iexcl20 17 (14) 28287 (97)

36 2 47iexcl09 16 (17) 17182 (93)

40 25 18iexcl04 513 (38) 21433 (48)

39 3 15iexcl03 414 (28) 14458 (3)

38 2 06iexcl01 29 (22) 7451 (16)

Average 1760 (28) 1352217 (61)








































































RESULTS

































DISCUSSION























GdNC event~n W s








SampleID




areas ()


nuclei ()

35 1 100iexcl20 411 (36) 48406 (118)

37 25 90iexcl20 17 (14) 28287 (97)

36 2 47iexcl09 16 (17) 17182 (93)

40 25 18iexcl04 513 (38) 21433 (48)

39 3 15iexcl03 414 (28) 14458 (3)

38 2 06iexcl01 29 (22) 7451 (16)

Average 1760 (28) 1352217 (61)


















































































DISCUSSION























GdNC event~n W s








SampleID




areas ()


nuclei ()

35 1 100iexcl20 411 (36) 48406 (118)

37 25 90iexcl20 17 (14) 28287 (97)

36 2 47iexcl09 16 (17) 17182 (93)

40 25 18iexcl04 513 (38) 21433 (48)

39 3 15iexcl03 414 (28) 14458 (3)

38 2 06iexcl01 29 (22) 7451 (16)

Average 1760 (28) 1352217 (61)


































































GdNC event~n W s








SampleID




areas ()


nuclei ()

35 1 100iexcl20 411 (36) 48406 (118)

37 25 90iexcl20 17 (14) 28287 (97)

36 2 47iexcl09 16 (17) 17182 (93)

40 25 18iexcl04 513 (38) 21433 (48)

39 3 15iexcl03 414 (28) 14458 (3)

38 2 06iexcl01 29 (22) 7451 (16)

Average 1760 (28) 1352217 (61)



























































are gadolinium contrast agents suitable for gdnct

Documents