comparison of genomics and functional imaging from canine ... · sarcomas treated with...

Predictive Biomarkers and Personalized Medicine

Comparison of Genomics and Functional Imaging from CanineSarcomas Treated with Thermoradiotherapy Predicts TherapeuticResponse and Identifies Combination Therapeutics

Jen-Tsan Chi1,2, Donald E. Thrall3, Chen Jiang4, Stacey Snyder3, Diane Fels5, Chelsea Landon5,Linda McCall4, Lan Lan4, Marlene Hauck3, James R. MacFall6, Benjamin L. Viglianti5, and Mark W. Dewhirst5

AbstractPurpose: While hyperthermia is an effective adjuvant treatment to radiotherapy, we do not completely

understand the nature of the response heterogeneity.

Experimental Design: We performed gene expression analysis of 22 spontaneous canine sarcomas

before and after the first hyperthermia treatment administered as an adjuvant to radiotherapy. In parallel,

diffusion-weighted MRI (DWI) was done prior to the treatment course and at the end of therapy.

Results: From the integrative analysis of gene expression and DWI, we identified significant correlation

between tumor responses with genes involved in VEGF signaling, telomerase, DNA repair, and inflamma-

tion. The treatment-induced changes in gene expression identified 2 distinct tumor subtypes with

significant differences in their gene expression and treatment response, as defined by changes in DWI.

The 2 tumor subtypes could also be readily identified by pretreatment gene expression. The tumor

subtypes, with stronger expression response and DWI increase, had higher levels of HSP70, POT1, and

centrosomal proteins, and lower levels of CD31, vWF, and transferrin. Such differential gene expression

between the 2 subtypes was used to interrogate connectivity map and identify linkages to an HSP90

inhibitor, geldanamycin. We further validated the ability of geldanamycin to enhance cell killing of human

tumor cells with hyperthermia and radiotherapy in clonogenic assays.

Conclusions: To our knowledge, this is one of the first successful attempts to link changes in gene

expression and functional imaging to understand the response heterogeneity and identify compounds

enhancing thermoradiotherapy. This study also demonstrates the value of canine tumors to provide

information generalizable to human tumors. Clin Cancer Res; 17(8); 2549–60. �2011 AACR.

Introduction

The clinical benefit of hyperthermia when combinedwith radiation has been proven in randomized humantrials for treatment of melanoma (1), esophageal cancer(2), head and neck cancer (3), cervix cancer (4), andglioblastoma (5). Most recently, neoadjuvant and adjuvantchemotherapy and hyperthermia led to prolonged survivalin human soft tissue sarcoma patients compared to

chemotherapy alone (6). Despite these successes, the bio-logic rationale for combining hyperthermia with radiationor drugs is incompletely defined. While heat can kill cells ina time and temperature dependent manner (7), thermalcytotoxicity is not likely to make a major contribution tothe clinical effect of hyperthermia due to the inability touniformly raise temperature to cytotoxic levels (8). Manyother effects of hyperthermia on a tumor at the cellularlevel, other than the heat shock response (9), remainincompletely defined and are likely playing a role inresponse.

Despite the positive results observed, there are significantvariations in the response to hyperthermia treatmentbetween tumors of the same type (10, 11). Understandinghow the tumor response to hyperthermia treatment variesmay allow the prediction of who will respond to thetreatment as well as the potential to identify means toimprove the treatment efficacy. One way to achieve theseobjectives is through the use of microarrays to profile thegene expression of tumors associated with hyperthermiatreatment. Although widespread genetic alterations, affect-ing many important physiologic pathways, have beenidentified in several studies (12–14), the intratumoral geneexpression changes resulting from thermoradiotherapy and

Authors' Affiliations: 1Institute for Genome Sciences & Policy, 2Depart-ment of Molecular Genetics & Microbiology, Duke University, Durham;3College of Veterinary Medicine, North Carolina State University, Raleigh;and Departments of 4Biostatistics and Bioinformatics, 5Radiation Oncol-ogy, and 6Radiology & School of Medicine, Duke University, Durham,North Carolina

Note: Supplementary data for this article are available at Clinical CancerResearch Online (http://clincancerres.aacrjournals.org/).

Current address for B. L. Viglianti: Department of Radiology, University ofMichigan, Ann Arbor, MI 48103.

Corresponding Author: Jen-Tsan Chi. Duke University, 101 ScienceDrive, DUMC 3382, CIEMAS 2177A, Durham NC 27708. Phone: 919-6684759; Fax: 9196680795; E-mail: [email protected]

doi: 10.1158/1078-0432.CCR-10-2583

�2011 American Association for Cancer Research.

ClinicalCancer

Research

www.aacrjournals.org 2549

Research. on June 15, 2020. © 2011 American Association for Cancerclincancerres.aacrjournals.org Downloaded from

Published OnlineFirst February 3, 2011; DOI: 10.1158/1078-0432.CCR-10-2583

http://clincancerres.aacrjournals.org/

their relationship with tumor response are not well under-stood. Such definition of gene expression can also beanalyzed with bioinformatics tools and other existing genesignatures representing specific biological pathways andprocesses. The simultaneous characterization of geneexpression in concert with other biologic endpoints canuncover relationships to better understand how thesechanges come about as well as identify additional thera-peutic targets. Most importantly, these methods can allowfor unexpected connection in gene expression to providemechanistic insights into response heterogeneity and iden-tify means to further improve therapeutic benefit.

Our purpose in this study was to assess gene changesoccurring in canine spontaneous soft tissue sarcomas as aresult of thermoradiotherapy. In addition, we comparedgene changes to changes in tumor volume and the apparentdiffusion coefficient (ADC) of water, quantified usingdiffusion-weighted MRI (DWI). We applied this analysisto spontaneous soft tissue sarcomas since they are moresimilar than rodent tumors to human cancers (15) in termsof intertumoral heterogeneity, multiple mutations in onco-genes and tumor suppressor genes, and varying environ-mental conditions. Through such integrative analysis, weidentified many important genes which are strongly corre-lated with treatment response. The variation in gene expres-sion response also led to identification of 2 tumor subtypescharacterized by a significant difference in the effectof treatment on the ADC. Analysis of differential gene

expression among these 2 subtypes allowed us to identifygenes associated with them and to identify compoundswhich may enhance the therapeutic efficacy of hyperther-mia. We further validated the ability of one identifiedcompound, geldanamycin, to enhance the cell killing ofhyperthermia and radiotherapy in clonogenic assays.

Materials and Methods

Design and subjectsTwenty-two dogs with a spontaneous soft tissue sarcoma

were studied. Tumor types included hemangiopericytoma(HPC, n¼ 12), undifferentiated sarcoma (Sarcoma, n¼ 6),fibrosarcoma (FSA, n ¼ 3), and peripheral nerve sheathtumor (PNST, n ¼ 1). Canine hemangiopericytoma is amalignant mesenchymal tumor with a suspected pericyteorigin (16). It behaves similar to other malignant mesche-chymal tumors in the dog, being locally invasive andhaving metastatic potential associated with histologicgrade.

Tumors were graded histologically as either low or highgrade based on the number of mitoses per high power field;there were 5 high grade tumors, 16 low grade tumors, and 1tumor of unknown grade. Fifteen tumors were located onan extremity, 5 on the trunk, and 2 on the head. No tumorwas judged to be resectable completely, without amputa-tion, prior to treatment. Median tumor volume, measuredphysically with calipers and computed as the product of 3orthogonal diameters divided by p/6, was 80 cm3, with arange of 17 to 180 cm3.

Dogs were part of an institutionally-approved rando-mized trial investigating the effect of thermal dose fractio-nation when combined with radiotherapy. This trialinvolved a 5-week treatment of daily fractionated radiationtherapy and 1 of 2 hyperthermia prescriptions. Prescription1 was 1 hyperthermia treatment per week and prescription2 was approximately 3 hyperthermia treatments per week.The total thermal dose was the same in each prescription.

Dogs received local hyperthermia under anesthesia usinga superficial 433 MHz microwave applicator. Thehyperthermia technique has been described elsewhere indetail (17). The radiotherapy prescription was 56.25 Gyadministered to the planning target volume in 25 dailyfractions of 2.25 Gy using 6 MV photons. Dogs receivedeither 1 or 2 radiation fractions prior to the posttreatmentbiopsy (Supplementary Fig. S3).

Outcome variablesThe percent change in tumor volume at the end of the 5-

week treatment was used as themeasure of tumor response.For economic reasons it was not possible to longitudinallyassess animals for tumor regrowth following completion oftherapy. Most tumors remained unresectable after therapy,making histologic assessment of the entire tumor alsoimpossible.

Water diffusion was quantified in 16 of the 22 dogs priorto treatment and at the end of treatment using DWI with a1.5 T unit (Siemens Symphony, Siemens Medical Systems).

Translational Relevance

Our work also has the potential to be applied tofuture practice of cancer medicine. Our research resultsdemonstrate the successful application of both geneexpression and functional imaging in determining thenature and heterogeneity in the treatment response inclinical trials. The use of functional imaging in mon-itoring the tumor physiology and cell killing can beimportant to provide timely quantitative data on howdifferent tumors respond to cancer therapeutics. Theintegrative analysis of treatment response can identifycritical processes and candidate biomarkers genes asso-ciated with the varying degrees of treatment response topredict and individualize the treatment plan. The dif-ferential gene expression between tumors with distinctresponse can also be used to identify compounds whichwill enhance the cancer therapeutics. Therefore, ourstudy has potential to impact the future practice ofclinical trials and cancer medicine. This manuscript alsocomplies with the criteria of original research with aclear application to future practice of medicine. Thisstudy is also hypothesis testing, possesses robust statis-tics, and is biologically based. The gene expression resultfrom canine sarcoma is further tested in human cells.We believe this work will be of considerable interest toreaders of Clinical Cancer Research.

Chi et al.

Clin Cancer Res; 17(8) April 15, 2011 Clinical Cancer Research2550




DWI was performed using a half Fourier acquisition, diffu-sion-weighted, single shot turbo spin echo (HASTE)sequence with b values of 0 and 500 sec/mm2 (TR ¼3,000 msec, TE ¼ 132 msec, echo train length ¼ 256,128� 128matrix, NEX¼ 2 for b¼ 0 sec/mm2 andNEX¼ 6for b¼ 500 sec/mm2, 4.0-mm slice thickness, 0.5 mm gap).Diffusion-weighted images were used to compute theapparent diffusion coefficient (ADC) using proprietarysoftware provided by the manufacturer. T2-weighted spinecho images were also acquired and used for anatomicregistration of the ROI used for quantification of ADC. TheADC was quantified in every pixel in the tumor by drawinga region of interest around the tumor in every ADC image.ADC was quantified using ImageJ (http://rsbweb.nih.gov/ij/). The change in ADC was expressed as the percentchange in mean ADC at the end of treatment comparedto the mean pretreatment ADC value. ADC images and ahistogram depiction of pretreatment and postreatmentpixel values are shown in Supplementary Figures S1 and S2.

Gene expression analysisTumor biopsies for gene expression analysis were

obtained prior to any treatment and 24 hours after thefirst hyperthermia treatment. Tumor biopsies wereobtained with an 18 g Tru-Cut needle and inspected grosslyfor quality. Multiple samples were collected with oneassessed for gene expression; the others were used for otherendpoints being described elsewhere. Some of the otherbiopsy samples were stained with hematoxylin and eosinstain (H&E) and verified that viable tumor was obtained.There was some variation in time between the pretreat-

ment biopsy and the first treatment that was caused byscheduling conflicts. The median time between the pre-treatment biopsy and the first hyperthermia treatment was84 hours. Dogs received either one (n¼ 12) or two (n¼ 10)2.25 Gy radiation fractions prior to the posttreatmentbiopsy. The posttreatment biopsy was always obtained24 hours after the first heat treatment. The timing of allsample collections, treatments, and endpoint evaluations isindicated in Supplementary Figure S3.Tissue used for gene expression analysis was placed

immediately into RNALater for RNA harvesting with Qia-gen RNAeasy and quality was checked by an Agilent Bioa-nalyzer. RNA was labeled with Affymetrix One-CycleLabeling Reagents and then hybridized to AffymetrixCanine 2.0 GeneChips� to measure the expression of18,000 C. familiaris mRNA/EST-based transcripts and over20,000 nonredundant predicted genes.

Statistical analysis of gene expression dataGene expression profiles of the 44 biopsies from the 22

tumors (Gene Expression Omnibus, GEO, accession num-ber: GSE23380) were normalized by the Robust MultichipAverage (RMA) algorithm (18), filtered by indicated criteriaand arranged by hierarchical clustering (Gene Cluster pro-gram and 2-way average linkage clustering based on uncen-tered correlation; ref. 19). For each probe set, the change inexpression was quantified by subtracting the expression

value of the pretreatment from the expression value of thecorresponding posttreatment samples. Given that the RMAalgorithm provides log (base 2) transformed summarymeasures, this difference corresponds to the fold of changescaused by treatments. The binary regression predictionmodel was used to build gene signature predictive model(2 metagenes, 150 probsets) with Matlab from the pre-treatment samples. Briefly, a signature represents a group ofgenes that collectively demonstrate a consistent pattern ofexpression (Supplementary Fig. S4A) in relation to thepretreatment tumor grouping. The gene selection, identi-fication, and regression model is based solely on thetraining data (pretreatment tumors) and maintains statis-tical independence from the validation dataset (posttreat-ment tumors). The predicted probabilities in the leave-one-out validation (pretreatment; Supplementary Fig. S4B) andindependent validation (posttreatment; SupplementaryFig. S4C) were shown.

To test whether the physiologic continuous outcomesvaried between 2 groups, the 2 sample Wilcoxon rank sumtest (20) was used. To identify probe sets expressed differ-entially, the Wilcoxon signed rank test was used. To testwhether the change in expression is associated with con-tinuous outcomes (e.g., tumor volume, ADC), the Spear-man rank correlation test was used. The significance of theidentified association was further verified by permutations(1,000).

SAM (21) and PAM (22) were used to identify genesassociated with treatment or subgroups. The selected geneswere then merged with the GoodMatch list provided byAffymetrix to convert the corresponding human probe setsfor gene ontology (GO) enrichment analysis in gather (23)and connectivity map (24). One thousand, five hundredand eighty one probe sets were identified by SAM betweenthe 2 tumor groups with a false discovery rate (FDR) of0.16% (Fig. 6A). These probe sets were merged with Good-Match list from Affymetrix website using filemerger (http://filemerger.genome.duke.edu/) to the correspondinghuman probe sets in U133A2 for analysis using connec-tivity map (ref. 24; http://www.broadinstitute.org/cmap/).For the PAM selection of the genes separating the 2 tumorsubtypes, a threshold of 3 (Supplementary Fig. S7A) wasselected to yield the cross-validation error rate of 18%(Supplementary Fig. S7B).

SYBR I green based qPCRThe NCBI Dog Genome Resource was used to identify

the mRNA sequence of genes of interest for primersdesigned by primer 3 (http://frodo.wi.mit.edu/primer3/input.htm). One microgram of total RNA was reversetranscribed into cDNA using a Qiagen QuantiTect ReverseTranscription Kit (Qiagen, Inc.). Fifty nanogram of cDNAwas used as a template for qPCR. In addition to thetemplate cDNA, the qRCR reaction contained forwardand reverse primers, each at a concentration of 300nmol/L, and 1 � SYBR green I fluorescent dye, per man-ufacturers’ instructions (Power SYBR Green PCR MasterMix, Life Technologies Corp.). qPCR was performed with

Integrative Gene Expression Analysis with Functional Imaging

www.aacrjournals.org Clin Cancer Res; 17(8) April 15, 2011 2551




an ABI StepOne real-time PCRmachine (Life TechnologiesCorp.) with thermal cycling conditions of 15 minutes at95�C and 40 cycles of 15 seconds at 95�C plus 1 minute at60�C. Relative quantification was determined using thePfaffl method standardized to the gene for canine riboso-mal protein RPLP0, accession number XM_853158.1 (F:CAGCAAGTGGAAAGGTGTAATCAG, R: CCCATTCTAT-CATCAATGGATACAA).

PECAM1: F: GCAGCCATCTTGCGGTGGGATT, R: CTG-AGGTGGAGGGTGGGTGTGT.

VWF: F: CCCAGTGCTCCCAGAAGCCCT, R: TCGCCC-CGTGGTGAACCGAT.

ENO II: F: GATACAGCCCCTCCTGCTCTCCC, R: TCC-ACCGCCCGCTCAATACG.

Transferrin: F: TGACCAGCAGCTTTTGTTTG, R: ACAT-TGTGTGTCGTCCCTGA.

BBX: F: TACCACCCAGCCTATCAGGA, R: GCACCTTTG-CAAGACCTCTC.

HSP70: F: ATCACCATCACCAACGACAA, R: GAAGGCG-TAGGACTCCAGG.

Clonogenic survival assaysHCT116 cells were plated in 6-well plates at 2 different

densities (300 and 900 cells per well) in 3 mL of DMEMmedia and allowed to attach for 24 hour at 37�C. Gelda-namycin (5 or 10 nmol/L) or the vehicle (DMSO) wasadded to the cells for 1 hour at 37�C. Then the plates werewrapped with foil before being placed in temperature-controlled water baths for heat treatments at either 37�Cor 42�C for 1 hour. For the radiation treatments, cellsreceived 2 Gy using a Cesium-137 Irradiator immediatelypost heating. The media containing drugs was removed 6hours after radiation or hyperthermia treatment and thecells incubated in fresh media at 5% CO2 and 37�C for 7 to10 days. After incubation, cell colonies were fixed with 10%methanol–10% acetic acid and stained with a 0.4% crystalviolet solution. Colonies with more than 50 cells werecounted using a ColCounter (Oxford Optronix). Platingefficiencies were determined for each treatment and nor-malized to the vehicle controls. The data shown is themeanof 3 independent experiments.

Results

The functional imaging studies of treatment responseWe measured tumor volume prior to and at the end of a

fractionated hyperthermia-radiation prescription. Appar-ent diffusion coefficient of water (ADC)was also quantifiedin 16 of the 22 dogs prior to treatment and at the end oftreatment using DWI. Significant variation in the percentADC change was observed between tumors. ADC decreasedin 8 of 16 dogs and increased in 8 of 16 dogs at the 5-weekmeasurement point compared to the pretreatment mea-surement. The range of change was from �49.9% toþ68.9% (Fig. 1A). Tumor volume decreased in 21 of 22dogs and increased slightly in 1 of 22 dogs at the end oftherapy compared to the pretreatment tumor volume. Therange of volume change was from �94.3% to þ3.6%

(Fig. 1B). There was a statistically significant associationbetween the percent change in tumor volume and thepercent change in ADC (Fig. 1C).

The gene expression studies of treatment responseTumor biopsies were obtained from all 22 subjects prior

to any treatment and 24 hours after the first hyperthermiatreatment. Gene expression profiles of the 44 samples werenormalized by RMA and deposited into GEO (accessionnumber: GSE23380). To reduce the confounding factorvariation between different of individual tumors, we usedzero-transformation to characterize treatment-inducedgene expression by comparing gene expression for the sametumors before and after treatment. This analytic approach,which has been used successfully in several studies (25–29), involves the subtraction of the pretreatment expres-sion value of a probe set from the posttreatment expressionvalue of the same probe set. From the treatment-responsivegene expression, we selected 2712 probe sets having at leasta 2-fold change in at least 2 samples and arranged byhierarchical clustering (Fig. 2).

When all 22 samples were grouped using hierarchicalclustering in an unsupervised analysis based on changes ingene expression in response to thermoradiotherapy, therewere 2 distinct groups of tumors with 15 tumors in 1 group(group I, marked by a red bar in Fig. 2A) and the remaining7 tumors in the second group (group II, marked by a bluebar in Fig. 2A). The separation of these 2 groups of tumorswas mostly due to several large gene clusters which wereonly induced in group I tumors (Fig. 2A and B) and wasrobust by various filtering criteria. The group I tumorsexhibited a much stronger gene expression response asmanifested by larger number of genes which were inducedonly in the group I tumors. The separation of these 22tumors was not due to histopathologic types or treatmentregimes (Fig. 2A). Importantly, the clustering pattern ofgene expression changes after thermoradiotherapy wassignificantly (P ¼ 0.0275) associated with varying amountof changes in ADC at the end of treatment, 5 weeks later(Fig. 2C).

Large clusters of genes were induced and repressed inresponse to thermoradiotherapy (Fig. 2A and B). Thesetreatment-altered genes can be classified as either alteredin most tumors (common response) or seen in only asubgroup of tumors (group I specific response). Com-monly induced genes included many genes in the heat-responsive (heat-responsive protein 12), DNA damageresponse (Mdm2), and cell cycle inhibition (p21/Cip;Fig. 2B). Genes commonly repressed by thermoradiother-apy included genes (Fig. 2B) involved in cell cycle pro-gression, confirming a reduction in cellular proliferationin response to treatment. Cell cycle inhibition is known tobe associated with both radiation and hyperthermiaresponses. Hyperthermia has been shown to induce bothG1 and G2 cell cycle delays (30, 31), whereas the majorcell cycle delay seen after radiation treatment is a G2delay (31).

Chi et al.





Since these 2 tumor subtypes were identified based ontheir differential gene expression response to treatments,we explored the possibility to predict their response basedsolely on pretreatment gene expression. We used the binaryregression model to build a gene signature predictivemodel based on the differential expression of 150 genesamong the expression of tumors in the group I vs. II beforetreatment (Supplementary Fig. S4A). In the leave-one-outvalidation experiments, we found that this model success-fully predictedmost of the tumor grouping (accuracy: 95%,Supplementary Fig. S4B), revealing the consistency of thegrouping and validity of the prediction models. We alsoused the posttreatment samples, not used in the building ofthe predictive models, to further validate the predictionmodels. By using a probability threshold of 0.6, the modelwas able to classify the tumors as group I vs. II with accuracyof 82% (Supplementary Fig. S4C). If the probability thresh-old was set at 0.5, the predictive accuracy dropped to 72%.These results suggest the possibility to predict the treatmentresponse based on the gene expression prior to treatmentfor individual tumors.

The tumor heterogeneity revealed by gene expressionresponse

Therewas a large group of geneswhichwere only inducedin group I tumors (Fig. 2B). These genes encode proteinsinvolved in tissue remodeling (e.g., MMP-1, MMP-2A,MMP-3, TGF-b, CD44, SERPINA1, SEPINB6, and SRF)and inflammatory and immune responses [CCR1, interleu-kin 1b (IL1b), IL6, IL8, IL10, MARCO]. The changes in geneexpression of tissue remodeling are entirely consistent withthe higher changes in ADC in this group (Fig. 2C). Theseresults demonstrate our ability to use microarrays to eluci-date the biological effects and mechanistic insights into theresponse and heterogeneity of the hyperthermia treatment.

Integrative analysis of gene expression and functionalimaging

The quantitative measurements of the response of tumorvolume and ADC by functional imaging allowed us toidentify genes whose expression response was associatedwith these parameters on a linear and continuous scale(Supplementary Table S1). Among the genes selected to be

Figure 1. The variation in thetreatment response amongdifferent tumors. A, change inapparent diffusion coefficient ofwater (ADC) and B, tumor volume,comparing end of therapy tobaseline. Data are shown in rankorder and not subject-specificorder. C, percent change in ADCplotted as a function of percentchange in tumor volume. Theregression equation is % changein ADC ¼ 66 þ 1.12 (% change involume). R ¼ 0.73. P ¼ 0.0014.






statistically significant from this supervised analysis, weidentified several genes with biological interest. For exam-ple, the expression response of BRCA1 was significantlynegatively associated with the change in ADC (Fig. 3A). Inaddition, the expression response of Flt1 and KDR, whichencodes 2 receptors for VEGF, were strongly positivelyassociated with changes in ADC (Fig. 3B and C). Increasedexpression of Flt1 and KDR 24 hours after the firsthyperthermia treatment were also associated with anincrease in ADC at 5 weeks. We also identify a strongpositive correlation between the changes in ADC with bothStat5 (encodes a mediator of inflammatory response) andACCN3 (encodes an acid-sensing channel protein). Tofurther verify the statistical significance, we recalculatedthe P values for these probe sets with 1,000 sample per-mutations (Supplementary Table S3).

We also applied similar analysis to the response of tumorvolume (Fig. 4). We identified strong positive associationbetween the percent change in tumor volume at the end oftherapy and expression of telomerase reverse transcriptase(TERT). Tumors with reduced expression of TERT followingthe first hyperthermia treatment had greater decreases intumor volume at the end of therapy (Fig. 4A) which isconsistent with the known role of the telomerase in tumor

growth. We also found an inverse relationship betweentumor volume response and expression of RAD23A andBRIP1 (BRCA1 interacting protein C-terminal helicase 1;Fig. 4B and C), 2 genes encoding proteins important forDNA damage repair. RAD23A encodes a human homo-logue of Saccharomyces cerevisiae Rad23, a protein involvedin nucleotide excision repair (NER). BRIP1 encodes aprotein member of the RecQ DEAH helicase family andinteracts with the BRCT repeats of BRCA1 in a proteincomplex which is important in the normal double-strandbreak repair function of BRCA1. We have also noted asignificant positive correlation of tumor volume withHMGCR (3-hydroxy-3-methylglutaryl-coenzyme A reduc-tase), RPS11 (Ribosomal protein S11; Fig. 4D and E), andEEFSEC (eukaryotic elongation factor, selenocysteine-tRNA-specific). To further verify the statistical significance,we recalculated the P values for these probe sets with 1,000sample permutations (Supplementary Table S3).

Supervised analysis of genes and compoundsassociated with thermoradiotherapy

To identify the genes whose expression was consistentlyassociated with untreated and treated tumors, we used asupervised method, Prediction Analysis of Microarray

Figure 2. Heterogeneity in gene expression response to hyperthermia treatment. A, the transcriptional response of individual canine tumors tothermoradiotherapy was derived by zero-transformation by subtracting the pretreatment expression values from the posttreatment expression values of thesame patients. Two thousand, seven hundred and twelve probe sets were selected using the criteria of at least 2 observations with at least 2-fold changes andarranged by hierarchical clustering. The vertical red bar marks a cluster of genes induced in almost all samples, the vertical yellow and orange bars mark acluster of genes induced only in the samples in group I, and the vertical green bar marks a cluster of genes repressed in almost all samples. The samples werecolor-coded to indicate their histopathology diagnosis. HPC¼ hemangiopericytoma, FSA¼ fibrosarcoma, PNST¼ peripheral nerve sheath tumor. B, the geneclusters are expanded to show the names of representative genes on the right side. C, the difference in the apparent diffusion coefficient between the 2subgroups with different gene expression response to thermoradiotherapy.

Chi et al.





(PAM) to prioritize the differentially expressed mRNAsfrom array analysis (22). PAM is a class prediction toolbased on the shrunken centroids of gene expression, whichcomputes a standardized centroid for each class and selectsthe number of genes required to characterize each assignedclass with a defined error rate. PAM analysis showed thatonly 35 probes were needed to achieve 100% accuracy ofprediction of treatment effects. The change in the geneexpression of these 35 probes (Supplementary Fig. S5)showed the downregulation of topoisomerase andRAD23 and the upregulation of mdm2 and p21 in bothgroups. Among the top genes induced in the treated sam-ples are the Cyclin-dependent kinase inhibitor 1A (p21,Cip1) and MDM2. Our PAM analysis was consistent withthese processes of cell cycle regulation as well-knownconsequences of heat and radiation (30). Taken together,these results suggest that a prominent and consistent effectof thermoradiotherapy is the upregulation of regulators ofcell cycle progression and the resulting arrest of cell cycle.

Supervised analysis of gene expression andcompounds associated with the 2 tumor subtypesWe also used PAM to prioritize the differentially

expressed mRNAs associated with the 2 tumor subgroups

identified by varying gene expression responses and foundthat only 110 probes were needed to achieve more than90% accuracy of prediction (Fig. 5 and SupplementaryFig. S6). The level of gene expression of these 110 probesamong these 2 subgroups is shown in Figure 5A. Thedifferential expression of the hsp70, TOP1, CD31, andVWF among these 2 tumor subtypes was further testedstatistically in nonpaired t-test and found to be significant(Fig. 5B and C). The tumors in group I had a consistentlyhigher expression level of hsp70, POT1 [protection oftelomeres 1 homologue (S. pombe)], CDC23 [a compo-nent of anaphase-promoting complex (APC)] and severalgenes comprising the centrosome (centrin, centrosomalprotein 70 and 192 kDa). The higher level expression ofhsp70, BBX, and ENOII in group I tumor was verified byRT-PCR (Supplementary Fig. S7). Higher level of hsp70may contribute to the exaggerated response to thehyperthermia treatments, consistent with previous reportsusing immunohistochemical staining (32).

In contrast, in group II tumors there was a higherexpression level of many genes known to be expressedin mature blood vessels, including endothelial and smoothcells. These genes include CD31, VWF, melanoma celladhesion molecule (MCAM), myosin, heavy chain 11,

Figure 3. Selected genes with significant correlation with the response in diffusion coefficient of water (ADC). The correlation between the expressionof BRCA1 (A), Flt-1 (B), KDR (C), Stat5a (D), and ACCN3 (E) with the apparent diffusion coefficient (ADC). The changes in the indicated genes(in log 2) are shown with the change in ADC at the end of treatment. For each correlation, the P and R values are shown.






smoothmuscle, andmyosin light chain kinase (Fig. 5A andC). The highermRNA expression level of transferrin, CD31,and vWF in group II tumors was also verified by real-timeRT-PCR (Supplementary Fig. S7). The high level ofendothelial and smooth muscle-specific genes may suggesta higher level ofmature vasculature and could contribute tothe modest gene expression response to thermoradiother-apy. This may limit the temperatures achieved during heattreatment and/or reduce the gene expression to the appliedheat stresses, consistent with a previous report (33).

Identification of compounds which enhance theefficacy of hyperthermia

We then investigated the potential of using the differ-ential gene expression among the 2 groups and connectiv-ity map to identify compounds with the potential toenhance the therapeutic efficacy of hyperthermia treat-ment. We used Significant Analysis of Microarrays (SAM;ref. 21) to identify probe sets whose expression variedconsistently between the 2 clusters (Fig. 6A). These differ-entially expressed probe sets were used to interrogate theconnectivity map (24) to establish connection with com-pounds inducing similar or opposite expression. Thesecompounds included TSA and MS-275, both known to

be inhibitors of histone deacetylase (HDAC; Supplemen-tary Table S5). In addition, geldanamycin, a known inhi-bitor of hsp90, was also found to be significantly enriched(Supplementary Table S5). These compounds are likely toaffect how cells respond to the hyperthermia because theirability to impact the heat shock proteins and heat shockresponse.

To test the ability of compounds identified to synergizewith heat treatment, we performed clonogenic survivalassays using HTC-116 colon carcinoma cells treated at37�C or 42�C for 30 and 60 minutes in the presence ofdifferent concentrations of geldanamycin (GA). Radiationwas administered immediately after hyperthermia treat-ment was completed. Geldanamycin was removed 6 hourafter heat treatment. Clonogenic assays of human HCT116colorectal carcinoma cells pretreated 1 hour with the indi-cated concentrations of geldanamycin, then exposed toeither 37�C or 42�C for 1 hour alone (Fig. 6B) or imme-diately followed by 2 Gy radiation (Fig. 6B and C). Analysisof survival data indicated that geldanamycin significantlyenhanced cytotoxicity of hyperthermia and radiation aloneas well as hyperthermia combined with radiation (P <0.0001; Fig. 6B and C). Collectively, these results demon-strate that analysis of differential gene expression between

Figure 4. Selected gene with significant correlation with the response in tumor volume. The correlation between the expression of TERT (A), RAD23a (B),BRIP1 (C), HMGCR (D), and RPS11 (E) with the changes in tumor volume. The changes in the expression of indicated genes (in log 2) are shown withthe change in tumor volume at the end of treatment. For each correlation, the P and R values are shown.

Chi et al.





canine tumors with different treatment response allowedthe identification of compounds which enhanced the ther-apeutic efficacy of thermoradiotherapy of human cells.

Discussion

The integrative analysis of gene expression andfunctional imagingIn this study, we performed an integrative analysis of

both gene expression and functional imaging associatedwith thermoradiotherapy of spontaneous canine sarcomas.The functional imaging of tumors with DWI providedquantitative analysis of changes in the tumor related towater content and cell density, which may be related totreatment outcome. The association we found betweenreduction in ADC and reduction in tumor volume supportsthe association of ADC and treatment outcome (Fig. 1C).Interestingly, most solid tumor studies where ADC hasbeen assessed as a predictor of outcome associate increasesin ADC with better outcome (34). In this study, we foundthe opposite, assuming that reduction in volume is a sign ofbetter outcome. However, there are no studies of changes inADC during therapeutic hyperthermia where tumor con-trol duration has been quantified therefore, whetherincreases or decreases in ADC as a response of tumorheating signify improved response is not known. Somedata regarding ADC change are available from thermalablation studies (35, 36) and in that setting a decreasedADC was associated with better tumor response. The tem-peratures used in the hyperthermia treatments in this study

were lower than those used for thermal ablation butperhaps there is another effect of heating on the tumorthat leads to reduced water diffusion in favorably respond-ing tumors. More work is needed on the associationbetween ADC and outcome in tumors undergoinghyperthermia.

Imaging data, coupled with gene expression analysisallowed us to link sets of genes associated with tumorphysiologic function. Further, gene expression responsesrevealed subtypes of tumors exhibiting varying degrees oftreatment response. The differences in ADC betweengroups I and II also demonstrate the potential clinicalrelevance of these 2 subtypes revealed by gene expressionanalysis. Thus, the availability and reciprocal flow betweenthese 2 levels of quantitative information offers uniqueadvantages in understanding how physiologic changes intumors after thermoradiotherapy are linked to gene expres-sion changes. The differential gene expression among these2 subtypes both enables the development of gene expres-sion predictive of putative subtypes and discovers com-pounds which enhance the therapeutic efficacy ofthermoradiotherapy.

The identification of genes associated with tumorresponse from such integrative analysis also has the poten-tial to provide biological insights into treatment response.For example, the changes in expression of 2 VEGF recep-tors, Flt1 and KDR, with treatment response suggested anassociation between VEGF signaling and effects of thermo-radiotherapy treatment on solute transport. The results areconsistent with our preclinical data showing that both

Figure 5. The identification ofgenes associated with the 2tumor subtypes. A, theheatmap of the treatment-induced changes in expressionof 110 probe sets identified byPAM with the names ofselected genes shown. (B andC) The differences of 4 selectedgenes (hsp70, POT1, CD31,and vWF) among the 2 tumorsubtypes were testedstatistically with the P valueshown.






radiation and hyperthermia increase HIF-1 levels, whichdrives VEGF signaling (37, 38). In addition, one wouldexpect that water diffusion will increase in response tomore cell killing (39). Thus, the change in ADCmost likelyreflects both types of effects. The association of Stat 5 withADC response is also of interest since Stat5 encodes aprotein that serves the dual function as a signal transducerand activator of transcription in cells exposed to manyinflammatory cytokines. Such positive correlation was con-sistent with the induction of several inflammatory genes(Fig. 2B) in the subgroup I tumors with a higher response inADC (Fig. 2C). The association of tumor volume responsewith genes involved in various biological processes suggestsa rationale for combining many agents targeting HMGCR,telomerase (40) and VEGF with hyperthermia.

The advantages of studying spontaneous caninesarcoma

Spontaneous canine soft tissue sarcomas exhibit inter-tumoral heterogeneity due to multiple mutations in onco-genes and tumor suppressor genes, varying environmentalconditions, and inherited germline variations (15). Thecombination of these factors leads to immense naturalheterogeneity in tumor phenotypes, disease outcomes,and response to therapies. This is certainly the case forthe varying response to hyperthermia treatment. We are notstating that the canine sarcoma is an identical model ofhuman sarcoma, only that it is a spontaneous tumorcharacterized by many of the same heterogeneities thatcharacterize various human tumors. And, the physics of

hyperthermia treatment, although not discussed in detailhere, is identical to that of human tumor thermal therapyadding more validity to the model.

The advent of DNA microarray technology allows thestudy of gene activity and gene function at a genome levelas a means to analyze the molecular phenotypes andclinical heterogeneity of tumors. Such analysis in thiscohort of canine sarcomas revealed that the gene expressionof tumors is affected significantly, but not uniformly, bythermoradiotherapy. While certain genes are affected inmost tumors, the majority of the affected genes are foundonly in 1 subgroup of the tumors. Such heterogeneityrevealed from the parallel analysis of functional imagingand genomic profiling may also be extended and general-ized to the hyperthermia treatment of human cancers. Thispotential is supported by the ability of compounds identi-fied from the gene expression of canine tumors to enhancethe thermoradiotherapy-induced killing in human cancercell line.

Common and subgroup-specific response genesaffected by thermoradiotherapy

Among the genes affected by thermoradiotherapy inmost tumors, we have found the induction of MDM2,p21, and the reduction in the genes involved in cellularproliferation. The induction of these genes implicates thep53 as important regulator in the thermoradiotherapyresponse, consistent with the previous reports on theimportant role of p53 in hyperthermia and thermora-diotherapy treatment responses (41–45). We also

Figure 6. The identification of compounds associated with the gene expression variations among tumor subtypes. A, a total of 1581 probe sets showingbetween 2 groups of canine sarcoma exhibiting different response to hyperthermia that are selected by SAM with a FDR of 0.16%. (B and C) clonogenicsurvival assays of human HCT116 colorectal carcinoma cells pretreated 1 hour with the indicated concentrations of GA, then exposed to either 37oC or 42oCfor 1 hour alone (B) or immediately followed by 2 Gy radiation (C). GA was removed 6 hour post-HT. Results are a mean of 3 independent experiments. ANOVAanalysis determined significant interactions (P < 0.0001) for HT:GA (*), HT:xRT (**) and HT:GA:xRT (***).

Chi et al.





examined genes affected in only the subgroup I of thetumors as a means to predict and improve the treatmentresponse. The compounds identified in the connectivitymap through the differential gene expression among sub-types may also lead to mechanistic insights into thermo-radiotherapy response. This connection with the hsp90inhibitors suggests their role in determining tumorresponse to thermoradiotherapy. Hsp90 is an importantchaperone responsible for the stability of many proteinsinvolved in oncogenesis (46). Under the hyperthermiatreatments, Hsp90 is likely to play an even bigger role inthe survival and growth in response to heat and oxidativestresses caused by the treatments. In addition, hsp90 is alsofunctionally regulated by acetylation by HDAC6/HDAC4.Thus, our gene expression analysis has suggested the criticalrole of the HDAC-hsp90 pathway for the adaptive responseunder the stresses of thermoradiotherapies. Given the largenumber of compounds that are under development totarget HSP90; these agents are likely to further enhancethe therapeutic benefits of thermoradiotherapies.

The potential of gene expression to predict tumorresponse to thermoradiotherapyHere, we found that gene expression of tumors before

treatment can be used to build a gene expression modelpredictive of the likelihood for response to thermora-diotherapy. These selected genes can be further validatedfor their predictive power in independent cohorts oftumors undergoing thermoradiotherapy. Based on theanalysis of gene expression data, we found that tumorswith high expression levels of hsp70 and centrosomalproteins are likely to have stronger response to thermo-radiotherapy. Our findings suggest that changes in hsp70level in tumor after hyperthermia may be useful to predicttumor response. This result is consistent with previousstudies of the ability of hsp70 to predict hyperthermiaresponse (32) and clinical outcomes (47). It is also inter-esting to note there is a higher expression level of endothe-lial and smooth muscle associated genes expressed in thesubgroup II with less ADC response. This link of tumorblood vessel maturity with hyperthermia treatmentresponse has previously been recognized (48) and suggeststhe value of combination treatment with many antiangio-genesis treatments.

Comparison with other gene expression studies ofhyperthermia

Gene expression analysis has been previously used toanalyze hyperthermia treatment response in several differ-ent contexts (12–14, 49, 50). Our findings of the relevanceof the DNA damage cell cycle (14) and angiogenesis path-ways (13) are consistent with several other studies. Pre-vious studies also point to the changes in the inflammatorypathways associated with thermoradiotherapy (12, 50), asseen in our subgroup-specific gene expression response.However, our studies also differ from the previous studiessince we used functional imaging to directly measure thepathophysiologic response to treatment. In addition, wehave also used heterogeneity in response to develop pre-dictive gene signatures and identify geldanamycin enhan-cing the cell killing by thermoradiotherapy. Such use ofgene expression may allow the codevelopment of bothsynergistic therapeutics and predictive biomarkers to iden-tify patients who are likely to benefit from such therapeu-tics. Taken together, the gene expression studies ofhyperthermia-treated spontaneous canine tumors shedmechanistic insights into the heterogeneity of thermora-diotherapy response. To our knowledge, this is one of thefirst successful attempts to link genomic and functionalimaging changes together and use such associations tounderstand the connection between cellular and physio-logic responses to treatment.

Disclosure of Potential Conflicts of Interest

No potential conflicts of interest were disclosed.

Acknowledgments

The authors appreciate the helpful discussions with members ofDewhirst and Chi lab.

Grant Support

The research support was provided from the NIH (P01CA42745, years 19–23to M.W. Dewhirst, NCI R01CA125618 to J.T. Chi) and Komen Foundation grant(KG090869 to J.T. Chi).

The costs of publication of this article were defrayed in part by the paymentof page charges. This article must therefore be hereby marked advertisementin accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

Received September 27, 2010; revised January 4, 2011; accepted January18, 2011; published OnlineFirst February 3, 2011.

References1. Overgaard J, Gonzalez Gonzalez D, Hulshof MC, Arcangeli G, Dahl O,

Mella O, et al. Randomised trial of hyperthermia as adjuvant toradiotherapy for recurrent or metastatic malignant melanoma. Eur-opean Society for Hyperthermic Oncology. Lancet 1995;345:540–3.

2. Sugimachi K, Kitamura K, Baba K, Ikebe M, Morita M, Matsuda H,et al. Hyperthermia combined with chemotherapy and irradiation forpatients with carcinoma of the oesophagus–a prospective rando-mized trial. Int J Hyperthermia 1992;8:289–95.

3. Valdagni R, Amichetti M. Report of long-term follow-up in a rando-mized trial comparing radiation therapy and radiation therapy plushyperthermia to metastatic lymph nodes in stage IV head and neckpatients. Int J Radiat Oncol Biol Phys 1994;28:163–9.

4. Van Der Zee J, Gonzalez Gonzalez D, van Rhoon GC, van Dijk JD, vanPutten WL, Hart AA. Comparison of radiotherapy alone with radio-therapy plus hyperthermia in locally advanced pelvic tumours: aprospective, randomised, multicentre trial. Dutch Deep HyperthermiaGroup. Lancet 2000;355:1119–25.

5. Sneed PK, Stauffer PR, McDermott MW, Diederich CJ, Lamborn KR,Prados MD, et al. Survival benefit of hyperthermia in a prospectiverandomized trial of brachytherapy boost þ/� hyperthermia for glio-blastoma multiforme. Int J Radiat Oncol Biol Phys 1998;40:287–95.

6. Issels R, Lindner L, Verweij J, Wust P, Reichardt P, Schem B, et al.Neo-adjuvant chemotherapy alone or with regional hyperthermia forlocalised high-risk soft tissue sarcoma: a randomised phase III






multi-centre study of the EORTC-STBSG and the ESHO. LancetOncol 2010;11:561–70.

7. Westra A, Dewey WC. Variation in sensitivity to heat shock during thecell-cycle of Chinese hamster cells in vitro. Int J Radiat Biol Relat StudPhys Chem Med 1971;19:467–77.

8. Horsman MR, Overgaard J. Hyperthermia: a potent enhancer ofradiotherapy. Clin Oncol (R Coll Radiol) 2007;19:418–26.

9. Skitzki J, Repasky E, Evans S. Hyperthermia as an immunotherapystrategy for cancer. Curr Opin Investig Drugs 2009;10:550–8.

10. Vaupel PW, Kelleher DK. Pathophysiological and vascular character-istics of tumours and their importance for hyperthermia: heterogeneityis the key issue. Int J Hyperthermia 2010;26:211–23.

11. Liu FF, Hill RP. Potential role of HSP70 as an indicator of response toradiation and hyperthermia treatments for recurrent breast cancer. IntJ Hyperthermia 1996;12:301–2.

12. Borkamo ED, Schem BC, Fluge O, Bruland O, Dahl O, Mella O. cDNAmicroarray analysis of serially sampled cervical cancer specimensfrom patients treated with thermochemoradiotherapy. Int J RadiatOncol Biol Phys 2009;75:1562–9.

13. Harima Y, Togashi A, Horikoshi K, Imamura M, Sougawa M, SawadaS, et al. Prediction of outcome of advanced cervical cancer tothermoradiotherapy according to expression profiles of 35 genesselected by cDNA microarray analysis. Int J Radiat Oncol Biol Phys2004;60:237–48.

14. Borkamo ED, Dahl O, Bruland O, Fluge O. Global gene expressionanalyses reveal changes in biological processes after hyperthermia ina rat glioma model. Int J Hyperthermia 2008;24:425–41.

15. Khanna C, London C, Vail D, Mazcko C, Hirschfeld S. Guiding theoptimal translation of new cancer treatments from canine to humancancer patients. Clin Cancer Res 2009;15:5671–7.

16. Perez J, Bautista MJ, Rollon E, de Lara FC, Carrasco L, Martin de lasMulas J. Immunohistochemical characterization of hemangiopericyto-mas and other spindle cell tumors in the dog. Vet Pathol 1996;33:391–7.

17. Thrall DE, LaRue SM, Yu D, Samulski T, Sanders L, Case B, et al.Thermal dose is related to duration of local control in canine sarcomastreated with thermoradiotherapy. Clin Cancer Res 2005;11:5206–14.

18. Irizarry RA, Hobbs B, Collin F, Beazer-Barclay YD, Antonellis KJ,Scherf U, et al. Exploration, normalization, and summaries of highdensity oligonucleotide array probe level data. Biostatistics2003;4:249–64.

19. Eisen MB, Spellman PT, Brown PO, Botstein D. Cluster analysis anddisplay of genome-wide expression patterns. Proc Natl Acad SciU S A 1998;95:14863–8.

20. Hajek J, Sidak Z, Sen PK. Theory of Rank Tests. San Diego, CA:Academic Press; 1999.

21. Tusher VG, Tibshirani R, Chu G. Significance analysis of microarraysapplied to the ionizing radiation response. Proc Natl Acad Sci U S A2001;98:5116–21.

22. Tibshirani R, Hastie T, Narasimhan B, Chu G. Diagnosis of multiplecancer types by shrunken centroids of gene expression. Proc NatlAcad Sci U S A 2002;99:6567–72.

23. Chang JT, Nevins Jr. GATHER: a systems approach to interpretinggenomic signatures. Bioinformatics 2006;22:2926–33.

24. Lamb J, Crawford ED, Peck D, Modell JW, Blat IC, Wrobel MJ, et al.The Connectivity Map: using gene-expression signatures to connectsmall molecules, genes, and disease. Science 2006;313:1929–35.

25. Chan DA, Kawahara TL, Sutphin PD, Chang HY, Chi JT, Giaccia AJ.Tumor vasculature is regulated by PHD2-mediated angiogenesis andbone marrow-derived cell recruitment. Cancer Cell 2009;15:527–38.

26. Chen JL, Lucas JE, Schroeder T, Mori S, Wu J, Nevins J, et al. Thegenomic analysis of lactic acidosis and acidosis response in humancancers. PLoS Genet 2008;4:e1000293.

27. Chen JL, Merl D, Peterson CW, Wu J, Liu PY, Yin H, et al. Lacticacidosis triggers starvation response with paradoxical induction ofTXNIP through MondoA.PLoS Genet 2010;6:e1001093.

28. Chi JT, Rodriguez EH, Wang Z, Nuyten DS, Mukherjee S, van de RijnM, et al. Gene expression programs of human smooth muscle cells:tissue-specific differentiation and prognostic significance in breastcancers. PLoS Genet 2007;3:1770–84.

29. Chi JT, Wang Z, Nuyten DS, Rodriguez EH, Schaner ME, Salim A, et al.Gene expression programs in response to hypoxia: cell type speci-

ficity and prognostic significance in human cancers. PLoS Med2006;3:e47.

30. Higashikubo R, White RA, Roti JLR. Flow cytometric BrdUrd-pulse-chase study of heat-induced cell-cycle progression delays. Cell Prolif1993;26:337–48.

31. Zolzer F, Streffer C. G(2)-phase delays after irradiation and/or heattreatment as assessed by two-parameter flow cytometry. RadiationRes 2001;155:50–6.

32. Liu FF, Miller N, Levin W, Zanke B, Cooper B, Henry M, et al. Thepotential role of HSP70 as an indicator of response to radiation andhyperthermia treatments for recurrent breast cancer. Int J Hyperther-mia 1996;12:197–208; discussion 9–10.

33. LyngH,MongeOR,Bohler PJ,RofstadEK. The relevanceof tumour andsurrounding normal tissue vascular density in clinical hyperthermia oflocally advanced breast carcinoma. Int J Radiat Biol 1991;60:189–93.

34. Afaq A, Andreou A, Koh DM. Diffusion-weighted magnetic resonanceimaging for tumour response assessment: why, when and how?Cancer Imaging 2010;10:S179–88.

35. Chen J, Daniel BL, Diederich CJ, Bouley DM, Van Den Bosch MA,Kinsey AM, et al. Monitoring prostate thermal therapy with diffusion-weighted MRI. Magn Reson Med 2008;59:1365–72.

36. Jacobs MA, Herskovits EH, Kim HS. Uterine fibroids: diffusion-weighted MR imaging for monitoring therapy with focused ultrasoundsurgery–preliminary study. Radiology 2005;236:196–203.

37. Moeller BJ, Cao Y, Li CY, Dewhirst MW. Radiation activates HIF-1 toregulate vascular radiosensitivity in tumors: role of reoxygenation, freeradicals, and stress granules. Cancer Cell 2004;5:429–41.

38. Moon EJ, Sonveaux P, Porporato PE, Danhier P, Gallez B, Batinic-Haberle I, et al. NADPH oxidase-mediated reactive oxygen speciesproduction activates hypoxia-inducible factor-1 (HIF-1) via the ERKpathway after hyperthermia treatment. Proc Natl Acad Sci U S A2010;107:20477–82.

39. Ma B, Meyer CR, Pickles MD, Chenevert TL, Bland PH, Galban CJ,et al. Voxel-by-voxel functional diffusion mapping for early evaluationof breast cancer treatment. In:Prince JL, Pham DL, Myers KJ, editors.Information Processing in Medical Imaging, Proceedings. Berlin,Germany: Springer-Verlag; 2009. p. 276–87.

40. Harley CB. Telomerase and cancer therapeutics. Nat Rev Cancer2008;8:167–79.

41. Harima Y, Nagata K, Harima K, Oka A, Ostapenko VV, Shikata N, et al.Bax and Bcl-2 protein expression following radiation therapy versusradiation plus thermoradiotherapy in stage IIIB cervical carcinoma.Cancer 2000;88:132–8.

42. Sakakura C, Koide K, Ichikawa D, Wakasa T, Shirasu M, Kimura A,et al. Analysis of histological therapeutic effect, apoptosis rate andp53 status after combined treatment with radiation, hyperthermia and5-fluorouracil suppositories for advanced rectal cancers. Br J Cancer1998;77:159–66.

43. Ohnishi K, Ohnishi T. Heat-induced p53-dependent signal transduc-tion and its role in hyperthermic cancer therapy. Int J Hyperthermia2001;17:415–27.

44. Guan J, Stavridi E, Leeper DB, Iliakis G. Effects of hyperthermia on p53protein expression and activity. J Cell Physiol 2002;190:365–74.

45. van Bree C, Van Der Maat B, Ceha HM, Franken NA, Haveman J,Bakker PJ. Inactivation of p53 and of pRb protects human colorectalcarcinoma cells against hyperthermia-induced cytotoxicity and apop-tosis. J Cancer Res Clin Oncol 1999;125:549–55.

46. Whitesell L, Lindquist SL. HSP90 and the chaperoning of cancer. NatRev Cancer 2005;5:761–72.

47. Vargas-Roig LM, Fanelli MA, Lopez LA, Gago FE, Tello O, Aznar JC,et al. Heat shock proteins and cell proliferation in human breast cancerbiopsy samples. Cancer Detect Prev 1997;21:441–51.

48. Song CW. Effect of local hyperthermia on blood flow and microenvir-onment: a review. Cancer Res 1984;44:4721s-30s.

49. Klopp AH, Jhingran A, Ramdas L, Story MD, Broadus RR, Lu KH, et al.Gene expression changes in cervical squamous cell carcinoma afterinitiation of chemoradiation and correlation with clinical outcome. Int JRadiat Oncol Biol Phys 2008;71:226–36.

50. Borkamo ED, Dahl O, BrulandO, FlugeO. Kinetics study onmarkers ofthe immune system by gene expression profiling of an in vivo heatedtumor. Int J Hyperthermia 2009;25:41–6.

Chi et al.





2011;17:2549-2560. Published OnlineFirst February 3, 2011.Clin Cancer Res Jen-Tsan Chi, Donald E. Thrall, Chen Jiang, et al. TherapeuticsTherapeutic Response and Identifies CombinationCanine Sarcomas Treated with Thermoradiotherapy Predicts Comparison of Genomics and Functional Imaging from

Updated version

10.1158/1078-0432.CCR-10-2583doi:

Access the most recent version of this article at:

Material

Supplementary

http://clincancerres.aacrjournals.org/content/suppl/2011/04/14/1078-0432.CCR-10-2583.DC1Access the most recent supplemental material at:

Cited articles

http://clincancerres.aacrjournals.org/content/17/8/2549.full#ref-list-1

This article cites 47 articles, 7 of which you can access for free at:

Citing articles

http://clincancerres.aacrjournals.org/content/17/8/2549.full#related-urls

This article has been cited by 2 HighWire-hosted articles. Access the articles at:

E-mail alerts related to this article or journal.Sign up to receive free email-alerts

Subscriptions

Reprints and

[email protected] at

To order reprints of this article or to subscribe to the journal, contact the AACR Publications

Permissions

Rightslink site. (CCC)Click on "Request Permissions" which will take you to the Copyright Clearance Center's

.http://clincancerres.aacrjournals.org/content/17/8/2549To request permission to re-use all or part of this article, use this link



http://clincancerres.aacrjournals.org/lookup/doi/10.1158/1078-0432.CCR-10-2583

http://clincancerres.aacrjournals.org/content/suppl/2011/04/14/1078-0432.CCR-10-2583.DC1

http://clincancerres.aacrjournals.org/content/17/8/2549.full#ref-list-1

http://clincancerres.aacrjournals.org/content/17/8/2549.full#related-urls

http://clincancerres.aacrjournals.org/cgi/alerts

mailto:[email protected]

http://clincancerres.aacrjournals.org/content/17/8/2549


comparison of genomics and functional imaging from canine ... · sarcomas treated with...

Documents