FDG PET imaging of Ela1-myc mice reveals major biological differences between pancreatic acinar and...

ORIGINAL ARTICLE

FDG PET imaging of Ela1-myc mice reveals major biologicaldifferences between pancreatic acinar and ductal tumours

Ibane Abasolo & Judit Pujal & Rosa M. Rabanal &Anna Serafin & Pilar Navarro & Olga Millán &

Francisco X. Real

Received: 27 July 2008 /Accepted: 23 January 2009 /Published online: 28 February 2009# Springer-Verlag 2009

AbstractPurpose The aim was to evaluate FDG PET imaging inEla1-myc mice, a pancreatic cancer model resulting in thedevelopment of tumours with either acinar or mixed acinar-ductal phenotype.Methods Transversal and longitudinal FDG PET studieswere conducted; selected tissue samples were subjected toautoradiography and ex vivo organ counting. Glucosetransporter and hexokinase mRNA expression was analysedby quantitative reverse transcription polymerase chain

reaction (RT-PCR); Glut2 expression was analysed byimmunohistochemistry.Results Transversal studies showed that mixed acinar-ductal tumours could be identified by FDG PET severalweeks before they could be detected by hand palpation.Longitudinal studies revealed that ductal—but not acinar—tumours could be detected by FDG PET. Autoradiographicanalysis confirmed that tumour areas with ductal differen-tiation incorporated more FDG than areas displaying acinardifferentiation. Ex vivo radioactivity measurements showedthat tumours of solely acinar phenotype incorporated moreFDG than pancreata of non-transgenic littermates despitethe fact that they did not yield positive PET images. To gaininsight into the biological basis of the differential FDGuptake, glucose transporter and hexokinase transcriptexpression was studied in microdissected tumour areasenriched for acinar or ductal cells and validated using cell-specific markers. Glut2 and hexokinase I and II mRNAlevels were up to 20-fold higher in ductal than in acinartumours. Besides, Glut2 protein overexpression was foundin ductal neoplastic cells but not in the surrounding stroma.Conclusion In Ela1-myc mice, ductal tumours incorporatesignificantly more FDG than acinar tumours. This differ-ence likely results from differential expression of Glut2 andhexokinases. These findings reveal previously unreportedbiological differences between acinar and ductal pancreatictumours.

Keywords Pancreatic cancer . 18F-FDG . Positron emissiontomography . Glut proteins . Hexokinase

Introduction

Pancreatic ductal adenocarcinoma (PDAC) is associatedwith the worst 5-year survival rate of any human cancer [1].

Eur J Nucl Med Mol Imaging (2009) 36:1156–1166DOI 10.1007/s00259-009-1083-3

I. Abasolo : J. Pujal : P. Navarro : F. X. RealInstitut Municipal d’Investigació Mèdica-Hospital del Mar,Parc de Recerca Biomèdica de Barcelona,Carrer del Dr. Aiguader, 8808003 Barcelona, Spain

I. Abasolo : F. X. RealDepartament de Ciències Experimentals i de la Salut,Universitat Pompeu Fabra,Parc de Recerca Biomèdica de Barcelona,Carrer del Dr. Aiguader, 8808003 Barcelona, Spain

I. Abasolo :O. MillánInstitut d’Alta Tecnologia - CRC,Parc de Recerca Biomèdica de Barcelona,Carrer del Dr. Aiguader 8808003 Barcelona, Spain

R. M. Rabanal :A. SerafinDepartament de Medicina i Cirurgia Animals,Universitat Autònoma de Barcelona,08193 Bellaterra (Cerdanyola del Valles)Barcelona, Spain

F. X. Real (*)Programa de Patología Molecular,Centro Nacional de Investigaciones Oncológicas,Melchor Fernández Almagro, 328029 Madrid, Spaine-mail: [email protected]

This high mortality is due, in part, to difficulties inestablishing an early and accurate diagnosis: for example,the 2-year survival of patients with a 2-cm tumour is only20%, indicating very early and fast rates of tumourprogression [2].

While screening for early PDAC appears to be animportant goal, there are a number of difficulties. First,except for individuals with a family history of PDAC or apersonal and/or family history of chronic pancreatitis—which constitute a small fraction of those developing thedisease—a population of subjects with a high risk ofdeveloping PDAC cannot be identified. Second, diagnosisof early pancreatic cancer is difficult due to the lack ofspecific symptoms associated with early stages of thedisease, the lack of adequate biomarkers and the need touse invasive techniques in order to make a cyto-histologicaldiagnosis. In addition, there is a risk of inducing pancreatitisas a result of leakage and activation of pancreatic enzymes.For these reasons, it is extremely important to develop non-invasive imaging techniques for the detection of PDAC, andmouse models of the disease are a useful tool to this end.

Positron emission tomography (PET) with 18F-fluoro-deoxyglucose (FDG) has been widely used in recent yearsfor the diagnosis and staging of PDAC [3]. Indeed, FDGPET is regarded to be more accurate in the detection ofprimary tumours and identification of metastases than otherimaging methods [4].

In the last 15 years, several genetic mouse models ofPDAC have been developed which recapitulate the featuresof human PDAC to a variable extent. Among them, theEla1-myc mouse is the only single-transgene model thatgives rise to pancreatic tumours displaying ductal differen-tiation with a relatively short latency [5, 6]. In these mice,the myc oncogene is expressed under the elastase-1 acinarpromoter, leading to the spontaneous development ofpancreatic tumours in all mice. Interestingly, approximately50% of them are acinar cell carcinomas while the remainingare mixed acinar-ductal or ductal tumours displaying histo-logical features typical of human PDAC such as an abundantstroma [5, 6]. In addition, the ductal tumours arising inEla1-myc mice share important biological characteristicswith human PDAC, such as the expression of proteases andreceptors involved in tumour progression [5] or thedevelopment of liver metastasis [7]. The occurrence of bothacinar and ductal tumours in Ela1-myc mice renders thismodel useful not only to study human PDAC but also toassess the biological differences between acinar and ductaltumours arising in a common genetic background.

In this work we have analysed the FDG uptakeproperties of tumours occurring in Ela1-myc mice andfound that ductal tumours display a greater glucose uptakethan acinar tumours. The marked overexpression offacilitative glucose transporters and hexokinases observed

in ductal tumours may account for the different behavioursobserved upon FDG PET evaluation.

Material and methods

Animals

Ela1-myc transgenic mice were kindly provided by E.P.Sandgren (Department of Pathobiological Sciences, Schoolof Veterinary Medicine, University of Wisconsin-Madison,USA) and maintained at the Animal Facility of ourInstitution. All animal handling was done in accordancewith the institutionally approved Animal ExperimentationEthics Committee guidelines. Overall, 76 mice were studied(37 male and 39 female animals) with ages ranging from 10to 35 weeks.

MicroPET imaging

Two different studies were performed: transversal andlongitudinal. In the transversal study, 13 Ela1-myc micebearing palpable tumours and 13 non-transgenic littermatecontrols were subjected to a single microPET scan and thensacrificed. In the longitudinal study, 15 transgenic mice and17 control littermates underwent repeated microPET imagingevery 2–3 weeks starting at the age of 11 weeks. Animalswere fasted for 4 h and then anaesthetized with 1.5–2%isoflurane (FORANE, Baxter Healthcare Corp., Deerfield,IL, USA) immediately before the intravenous injection of9.25–11.1 MBq 18F-FDG. Anaesthesia was maintainedthroughout the FDG uptake time (40–60 min). Mice set ina prone position were imaged for 10 min within a microPETR4 scanner (Concorde Microsystems, Knoxville, TN, USA),using an energy window of 350–750 keV and a coincidencewindow of 6 ns. The resulting list-mode data were sortedinto 2D histograms by Fourier re-binning and transverseimages reconstructed by an ordered subset expectationmaximization (OSEM-2D) algorithm (4 iterations, 16 sub-sets) into a 128×128×63 (0.82×0.82×1.3 mm) matrix. Theimage data were corrected for non-uniformity of response ofthe microPET, dead time count losses and physical decay tothe time of injection; no attenuation, scatter or partial-volumeaveraging correction was applied. In order to facilitate visualcomparison among images, in each case the heart’s signalwas set as 50% of the maximum intensity of the wholeimage. The same procedures were applied for the acquisitionof all ex vivo PET images.

Autoradiography

Tissue samples were embedded in OCT (Tissue Tek, SakuraFinetek, Torrance, CA, USA) and snap-frozen in 2-

Eur J Nucl Med Mol Imaging (2009) 36:1156–1166 1157

methylbutane (−80°C). For each tumour sample, fiveconsecutive 10-μm sections were obtained at two differentlevels. Two sections were reserved for histopathologicalexamination and the remaining sections were exposed ontoa BAS SR imaging plate and then analysed at 100-μmspatial resolution using the BAS 5000 system (Fuji PhotoFilm, Tokyo, Japan).

Ex vivo radioactivity measurement

Immediately after microPET imaging, normal pancreas orpancreatic tumours of nine transgenic mice and nine controllittermates were dissected, weighed and measured in anautomated Wizard gamma counter (Wizard 1470, Perki-nElmer, Boston, MA, USA). Blood and muscle sampleswere also collected and used as internal controls. FDGuptake was expressed as the percentage of injected dose pergram of tissue (%ID/g).

Pathological examination and Glut2immunohistochemistry

After macroscopic examination, tumour volumes werecalculated from fresh tissue using the formula V = (w2×l)/2,as described elsewhere. Tissue samples not destined forautoradiography were fixed in 4% buffered paraformalde-hyde and embedded in paraffin. Tissue sections were stainedwith haematoxylin-eosin (H&E) for evaluation by a veteri-nary pathologist.

Glut2 immunohistochemistry was performed by incu-bating a rabbit antiserum raised against a peptide encom-passing residues 512–523 of Glut2 (a kind gift of B.Thorens, University of Lausanne, Switzerland) [8] withsections of paraffin-embedded tissues (three acinar tumours,four ductal tumours and two normal pancreata). Antibodybinding was visualized using the EnVision kit (Dako,Carpenteria, CA, USA), 1:2 in phosphate-buffered saline(PBS) during 30 min at room temperature, and developedwith diaminobenzidine. Sections were counterstained withhaematoxylin, dehydrated and mounted. Immunostainingwas visualized in an Olympus AX70 microscope (Tokyo,Japan), and images were acquired using Studio Lite 1.0software (Olympus, Tokyo, Japan).

Quantitative reverse transcription polymerase chain reaction(RT-PCR) analysis

Tumour areas enriched in acinar or ductal components weremanually microdissected from OCT-frozen tumour sectionsafter microscopic examination of a consecutive H&E-stained section. RNA was extracted using Trizol (Invitro-gen, Carlsbad, CA, USA), treated with DNase (Ambion,Foster City, CA, USA) and reverse transcribed using

Taqman Reverse Transcription Reagents (Applied Biosys-tems, Foster City, CA, USA). Quantitative PCR wasconducted using SYBR Green (Qiagen, Alameda, CA,USA) in an ABI PRISM 7900HT Sequence DetectionSystem (Applied Biosystems, Foster City, CA, USA) withan annealing temperature of 55°C. The acinar or ductalnature of the regions from which RNA was extracted wasconfirmed by RT-PCR analysis using appropriate differen-tiation markers: amylase (Amyl) and carboxypeptidase A(Cpa) as acinar markers and keratin 7 (Krt7) and cysticfibrosis transmembrane regulator (Cftr) as ductal markers.To assess glucose metabolism, glucose transporters (Glut1to Glut5) and hexokinases I and II (HkI and HkII) mRNAlevels were analysed. All primers were designed tospecifically amplify the corresponding cDNAs and notgenomic DNA (Table 1). Quantitative PCRs were repeatedat least three times for each sample. In all cases, thresholdcycle (Ct) values were transformed to obtain the RNA copynumber ratio for the selected gene and the housekeepingtranscript (Hprt).

Statistical analysis

To compare gene expression levels between acinar andductal tumours, the Mann-Whitney non-parametric U testwas applied. One tailed p values were considered for geneto Hprt ratios of acinar and ductal markers and two-tailedp values for the Gluts and Hk to Hprt ratios of thequantitative RT-PCR and for the comparison of the %ID/gvalues obtained after gamma activity measurements. Differ-ences were considered significant (*) at the 95% confidencelevel (p<0.05).

Results

Pancreatic tumours in Ela1-myc mice are detectableby FDG PET

To determine whether tumours of Ela1-myc mice could bedetected by FDG PET, 23- to 25-week-old Ela1-myctransgenic mice bearing palpable tumours (n=13) and sex-and age-matched non-transgenic littermates (n=13) werestudied. Immediately after imaging, animals were sacrificedand the main abdominal organs, including tumours, weredissected and imaged ex vivo. Findings are summarized inTable 2. Control mice failed to display an abnormalabdominal FDG uptake (Fig. 1a). On the contrary, alltransgenic littermates showed an increased FDG uptake(Figs. 1b–e). As shown in Figs. 1f–g, ex vivo microPETimaging of the main organs and tumours confirmed that theincreased FDG uptake areas corresponded to tumours.However, no correlation was found between tumour size

1158 Eur J Nucl Med Mol Imaging (2009) 36:1156–1166

and FDG uptake measured by drawing 3D regions ofinterest over the FDG PET images (data not shown).Despite the homogeneous genetic background of the miceused, there was wide variability in tumour development andprogression in Ela1-myc mice, and tumour volume did notcorrelate with the age of the animal in this transversal study(mean ± SE: 1.518±0.440 cm3). When big tumours werepresent, no remaining normal pancreatic tissue was ob-served, whereas small tumours were often embedded withinnormal pancreatic tissue. Histological examination of thetumours excised from all the transgenic mice included inthis transversal analysis showed mixed regions of acinar

and ductal differentiation (see Table 2 and Figs. 1h–j), aspreviously reported [6, 7].

Pancreatic ductal tumours show higher FDG uptakethan acinar tumours

Analysis of all in vivo and ex vivo PET images revealed wideinter- and intra-tumoural variability in FDG uptake, as shownin Fig. 1f (compare FDG uptake in tumours H, I and J). Inorder to elucidate whether these differences might resultfrom the variable differentiation features of the tumours,autoradiography and histological studies were conducted in15 tumours showing a non-uniform FDG uptake pattern.Tumour areas showing ductal differentiation (Figs. 1i–j)accumulated more FDG than the tumour regions rich inacinar cells (Fig. 1h).

Early detection of pancreatic tumours by FDG PET

To determine whether FDG PET was able to detect earlypancreatic tumours, a longitudinal study including Ela1-myc transgenic mice (n=15) and control littermates (n=17)was conducted. The first FDG PET scans were performed atthe age of 11 weeks, when it was expected that most micewould display regions of acinar dysplasia but would lacktumours. From that age on, PET imaging was repeatedevery 2–3 weeks. Six mice were sacrificed at the age of19 weeks to study tumours at early stages of progression;the remaining mice were kept under study until theirsacrifice was advisable for ethical reasons. The age atwhich an increased FDG uptake was first visualized tendedto be vary among subjects but always preceded detection byhand palpation: tumours were detected by PET at 18.1±4.3 weeks and by hand palpation at 23.9±4.1 weeks(Table 3). Detection of the tumours with a ductal componentby FDG PET imaging preceded their becoming palpable by

Table 1 Primers used for qRT-PCR analysis

Symbol Gene ID Forward Reverse Amplicon size (bp)

Glut1 20525 GCGGGAGACGCATAGTTACA GTAGCGGTGGTTCCATGTTT 234

Glut2 20526 GCCTGTGTATGCAACCATTG GAAGATGGCAGTCATGCTCA 205

Glut3 20527 CCAGGAGATCCAGGAGATGA GCTCCAATCGTGGCATAGAT 226

Glut4 20528 ATCTTGATGACCGTGGCTCT CGGCAAATAGAAGGAAGACG 271

Glut5 56485 CTCGTGGTTGGAAACTTGGT GAGCCCCTCGTAGGTTTTTC 226

HkI 15275 GTCAAGATGCTGCCAACCTT TCAGCGACGTGATCAAAAAG 215

HkII 15277 GGGCTAGGAGCTACCACACA ATTCGGCAATGTGGTCAAAC 250

Amyl 11723 CAGGCAATCCTGCAGGAACAA CACTTGCGGATAACTGTGCCA 483

Cpa 109697 GCATCCATTCTAGGGAGTGG GAAAGAGTACTTGATGCCCTG 589

Cftr 12638 ACAAAACCTGGATCCCAATG GAGCTGTCCAGGAAACTGCT 103

Krt7 110310 CACGAACAAGGTGGAGTTGGA TGTCTGAGATCTGCGACTGCA 114

Hprt 15452 TGCGCTCATCTTAGGCTTTGT AGAGGTCCTTTTCACCAGCA 180

Table 2 Results of the transversal FDG PET imaging study

Mouse # Age atsacrifice(weeks)

Numberof tumourmasses

Biggestdiameter (cm)

Total tumourvolume (cm3)

1 23 2 1.12 1.15

2 25 >8 2 4.85

3 25 1 + smalltumoursa

1.3 0.09

4 25 4 1.511 2.30

5 25 1 + smalltumoursa

1.28 1.61

6 25 3 0.92 0.51

7 27 2 1.8 0.67

8 27 >4 1.5 2.28

9 29 3 0.77 0.27

10 29 4a <0.02 0.01

11 29 2 2.1 4.40

12 30 2 1.9 0.17

13 35 3 1.26 1.4

a Small nodules embedded within normal pancreas

All Ela1-myc transgenic animals developed tumours with mixedacinar and ductal differentiation


at least 2 weeks, whereas mice with pure acinar tumours oracinar dysplasia failed to show increased uptake.

Initially, tumours appeared as a small, FDG uptaking dotclose to the medial line of the animal (see Figs. 2a–c and d–f).Size and intensity of the first positive FDG PET signal wasnot predictive of the tumour outcome in terms of growth rate.

Indeed, tumour growth rates also showed marked inter-individual variation: Fig. 2 shows PET images of a mousebearing a slow growing tumour (which increased its size 4.3-fold over 6 weeks, Figs. 2a–c) in contrast with those of amouse bearing a fast growing tumour (whose size augmented57-fold over 3 weeks, Figs. 2d–f).

To further analyse the differential FDG uptake pattern oftumours according to their progression stage, an additionalcohort of 18 mice (11 Ela1-myc and 7 control littermates)were administered FDG, subjected to microPET imagingand then sacrificed. Of 11 transgenic mice, 3 weresacrificed at the age of 10 weeks (young), whereas 8 micewere >20 weeks old by the time of sacrifice (aged). Blood,muscle and pancreas or pancreatic tumour samples weredissected and measured in a gamma counter. Youngtransgenic mice did not show any ductal neoplastic area,while 20-week-old animals showed pancreatic tumourswith mixed acinar-ductal phenotype. In all of them, FDGaccumulation was quantified as the %ID/g. Muscle wasconsidered as a reference tissue, since its FDG content wassimilar in control and transgenic animals (0.680±0.258%ID/gand 0.875±0.102%ID/g, respectively). As expected, pancre-atic tumours showed a highly significant increase in the%ID/g when compared to the pancreas of non-transgeniccontrol animals. Those tumours that upon subsequenthistological analysis displayed mixed ductal-acinar differ-entiation showed the highest %ID/g (27.95±14.12), where-as FDG uptake in pure acinar tumours was significantlylower (19.045±1.96%ID/g) (p=0.025). Nevertheless, bothtypes of tumours showed significantly higher uptakelevels than normal pancreata (2.08±0.46%ID/g) (p=0.007). It is worth noting that tumour uptake wasmeasured prior to further dissection and, therefore, the%ID/g obtained for ductal tumours corresponds to thecombined uptake of the ductal and acinar components.

Because tumour development in Ela1-myc mice ispreceded by a variable degree of acinar dysplasia, we alsoperformed FDG PET data with 10-week-old mice; nodifferential or abnormal FDG uptake was observed at theabdominal level (see Figs. 3a, b). However, dissection ofthe pancreata and histological analysis revealed the pres-ence of early neoplastic foci within normal tissue. Radio-activity measurements of the whole pancreas (including theneoplastic foci) showed that the pancreas of Ela1-myc miceincorporated a significantly higher amount of FDG thantheir control littermates (11.153±3.299%ID/g vs 1.960±0.176%ID/g, respectively; p=0.014) (Fig. 3c).

Levels of the glucose transporter Glut2 and hexokinasesI and II are higher in ductal than in acinar tumours

The increased FDG uptake of tumours is often due to theincreased expression of glucose transporters (Gluts) and the

Fig. 1 Representative FDG PET images of mice included in thetransversal study. a Twenty-six-week-old non-transgenic mouseshowing basal FDG uptake in the abdominal area. b–g Twenty-five-week-old Ela1-myc transgenic mouse with pancreatic adenocarcino-ma. b Coronal microPET image at the tumour level. High FDG uptakecan be seen in abdominal tumours as well as in the heart and thebladder. c Photograph showing tumour localization of the animal in b.d, e Transverse views of the microPET image at the positionsindicated in b. f Ex vivo microPET projection image of tumours anddissected tissues of the main abdominal organs and (g) theirmacroscopic appearance. The letters H,I and J indicate the tumoursthat were further analysed by autoradiography and histology, shown inh, i and j, respectively. h–j Autoradiography and H&E-stainedsections of the tumours indicated in f. White squares within theautoradiograph picture show the tumour region from which the H&Esections were taken. FDG uptake was increased in tumour areasdisplaying ductal differentiation (i, j) compared to areas showingacinar differentiation (h). Scale bar: 1 cm (g); 25 μm (h, i, j)


enzymes responsible for FDG cell trapping, the hexo-kinases (Hks) [9]. We therefore investigated whether thehigher FDG uptake observed in ductal tumours was due tothe differential expression of these genes. RNAwas isolatedfrom manually microdissected tumour areas displayingacinar or ductal histology. The acinar or ductal purity ofthe obtained samples was assessed using qRT-PCR of acinar-specific (amylase and carboxypeptidase A) and ductal-specific (Krt7 and cystic fibrosis transmembrane regulator)markers. Hprt was used as housekeeping gene to normalizethe analysed transcripts. As shown in Fig. 4, relative levelsof amylase and carboxypeptidase mRNA were 27- and 134-fold higher in acinar than in ductal tumour areas, respectively(Fig. 4a). On the other hand, relative Krt7 and CftrmRNA levels were 65.9- and 65.2-fold higher in ductalcompared to acinar tumour areas (Fig. 4b).

The relative expression of Glut1–5 and hexokinases Iand II (HkI and HkII) mRNAs was analysed. Among thesegenes, only Glut2 and both Hks were found to beupregulated in ductal compared to acinar tumours (Figs. 4c,d). Relative Glut2 mRNA expression was 20-fold higher(p=0.025) in tumours showing ductal differentiation than inthose showing acinar differentiation. Glut1, Glut3 andGlut4 mRNA levels were not statistically different inacinar and ductal tumours (Fig. 4c). As expected, theGlut5/Hprt ratio was found to be very low orundetectable in all samples (data not shown) since itsexpression is very limited outside the central nervoussystem. Regarding hexokinase expression, relative mRNAlevels of both HkI and HkII were up to 65-fold higher inductal than in acinar tumours (p=0.025) for both genes(Fig. 4d).

Unlike acinar tumours, PDAC display an importantdesmoplastic reaction, both in humans and mice [10]. Todetermine whether Glut2 was mainly expressed intumour cells or in the stroma, immunohistochemistry withanti-Glut2 antibodies was performed. In normal pancreas,Glut2 was strongly expressed at the plasma membrane ofislet cells (Fig. 5a), weakly in ductal cells (Fig. 5a, inset),but it was undetectable in acinar cells. In tumours, acinarcells also showed undetectable levels of Glut2 (Fig. 5b),thus confirming the results obtained by RT-PCR. Incontrast, moderate to high Glut2 expression was foundin ductal tumour cells but not in the stroma (Fig. 5c).Immunohistochemical analysis showed that HkII expres-sion was restricted to epithelial cells, rather than thestroma, and was already detectable in acinar dysplasia.Unlike Glut2, HkII expression in tumours was similar inareas displaying acinar vs ductal differentiation (notshown).

Altogether, these findings suggest that the increasedFDG uptake observed in ductal tumour areas results fromthe distinct glucose metabolism associated with ductaldifferentiation of transformed cells and not to the contribu-tion of the stroma.

Discussion

Despite an increasing interest in the development ofspecific probes for the detection of preneoplastic lesionsand tumours, FDG is the most commonly applied PETstrategy. The use of FDG for cancer detection relies on theincreased glucose uptake associated with the exacerbated

Table 3 Results of the longitudinal FDG PET imaging study of Ela1-myc mice

Mouse # Age at sacrifice(weeks)

Age at first positive PETstudy (weeks)

Age at first palpabletumour (weeks)

Advantage of PET detectionover hand palpation (weeks)

Tumour histology

1 14.4 12.6 Undetectable – Acinar-ductal

2 18.7 Undetectable Undetectable – Acinar

3 18.7 Undetectable 18.7 – Acinar

4 18.9 Undetectable Undetectable – Acinar atypia




8 19.9 13.0 16.9 3.9 Acinar-ductal

9 22.7 16.9 19.9 3.0 Acinar-ductal

10 22.7 16.9 22.7 5.9 Acinar-ductal

11 23.0 16.9 22.7 5.9 Acinar-ductal

12 26.6 16.9 25.7 8.9 Acinar-ductal

13 28.7 22.7 25.7 3.0 Acinar-ductal

14 31.3 22.7 24.7 2.0 Acinar-ductal

15 32.4 24.7 30.1 5.4 Acinar-ductal


glycolytic metabolism of most malignant cells [11].Accelerated glucose consumption can be the consequenceof a variety of factors, including the overexpression ofglucose transporters and hexokinases [12, 13].

Intensive efforts have been made to explore imagingmethods for the detection, staging and differential diagnosisof PDAC. FDG PET has been widely used in the clinicalsetting in the last few years [3] and it has been proposed tobe a more accurate tool in the detection of primary tumoursand metastases than other imaging methods [4]. Standarduptake values calculated for FDG have been proposed to bean important prognostic factor in PDAC [14, 15]. However,FDG may lack both the sensitivity to detect premalignantchanges [16] and the specificity required to distinguishbenign from malignant tumours [17].

In the last few years, transgenic and homologousrecombination-based strategies have led to the developmentof a large number of mouse models of human cancer with avariable level of resemblance to human tumours. Suchmodels may be useful to better understand the biologicalfeatures of early stage preneoplastic and neoplastic lesions.

In this study, we have used FDG PET to evaluate tumourprogression in mice in which the myc oncogene is activatedunder the elastase-1 promoter. All Ela1-myc transgenicmice develop pancreatic tumours with a short latency(2–7 months), 50% of which present ductal elementsresembling human PDAC. This characteristic renders theEla1-myc model a useful system for testing new therapeuticor diagnostic agents. In addition, this model has additionalinterest since it has recently been shown that K-ras candrive the development of PDAC in mice upon its activationusing the Ela1 promoter [18], indicating that tumours withductal histology can arise from acinar initiation events.

Using Ela1-myc mice, we show that detection oftumours by FDG PET precedes by several weeks theirdetection by hand palpation. In addition, we have found celldifferentiation—but not tumour size—to be the majordeterminant of FDG uptake: Ductal tumours showedsignificantly higher FDG uptake than acinar tumours.Indeed, small acinar tumours were not detectable by FDGPET imaging although they showed an increased FDGuptake compared to normal pancreas, as shown by the exvivo FDG radioactivity measurements. Similarly, acinardysplasia occurring in 10-week-old mice was associatedwith increased FDG uptake—as detected ex vivo—but thiswas not evidenced by imaging methods. The inability todetect these lesions may result from their small size, belowthe volumetric detection limits of the microPET scannerused in this study. The theoretical volumetric resolution ofmicroPET R4 PET has been established as less than 16 μl[19], which is approximately the size of the smallesttumours found in this report (those of 10-week-old mice).The small size of the purely acinar tumours found in thisstudy, the relatively high abdominal background of Ela1-myc mice and the phenomenon known as the partialvolume effect (by which small regions with low activityappear to have even lower activities) [20] make early acinar

Fig. 2 FDG PET images of tumours from mice included in thelongitudinal study. a–c Representative coronal images obtained atdifferent time points of an Ela1-myc mouse bearing a slow growingtumour. The arrow in a indicates the area where the FDG uptake wasconsidered abnormal for the first time. d–f Representative coronalimages of another Ela1-myc mouse bearing a fast growing tumour.The arrow in f indicates the area where the FDG uptake wasconsidered abnormal for the first time. The white asterisk indicates anacinar tumour with low FDG uptake; the arrowhead shows a ductaltumour with high FDG uptake. g Ex vivo microPET projection imageof tumours and main abdominal organs. The white asterisk andarrowhead correspond to the acinar and ductal tumours indicated in f.h FDG accumulation (%ID/g) as measured ex vivo in a gammacounter in acinar-only (n=5) and acinar-ductal mixed tumours (n=3)was found to be significantly higher than in normal pancreas ofcontrol transgenic mice (n=4). Data are represented as mean andstandard deviation. Please note that the height of the section shown isnot the same in both animals given the distinct location of the tumourin the mice


tumours undetectable by means of FDG PET imaging. Adiffuse increase in uptake in the pancreas of transgenicmice may also contribute to the lack of detection of earlylesions. An increased uptake of FDG in dissected pancreaswas also reported in the N-nitrosobis(2-oxopropyl)amine-induced chemical carcinogenesis model in hamsters, whereacinar dysplastic regions were found to incorporate moreFDG than the acinar-ductal tumours [21]. Despite theseexplanations, we hypothesized that additional factorsshould contribute to the lack of detection of acinar tumoursbecause large acinar tumours also failed to displayincreased FDG uptake.

To further investigate the biological basis of the distinctFDG uptake of acinar and ductal tumours, we have studiedthe differential expression of glycolytic enzymes. Theenhanced anaerobic glycolysis of tumour cells requiresboth glucose transporter overexpression and enhancedhexokinase activity. Interestingly, our results demonstratehigher HkI, HkII and Glut2 mRNA expression in ductalthan in acinar tumours. Among these genes, both hexoki-nases (HkI and HkII) have been reported to be overex-pressed in human PDAC [22, 23]. Indeed, alteredexpression of facilitative glucose transporters and glucose

phosphorylating enzymes is a common feature of manytumours. Other genes, such as Glut1, may also have anincreased expression in acinar and ductal tumours com-pared to normal tissues, but their expression is similar inboth types of tumours and, thus, cannot account for thedifferential FDG uptake seen in ductal vs acinar tumours.Regarding the tumours whose expression is altered, HkI isconsidered a “housekeeping” gene, not altered by metabo-lism or glucose levels, but its expression has been found tobe increased in certain tumours, including breast andthyroid cancers [24, 25], whereas HkII expression hasbeen reported to respond to glucose deprivation andhypoxia [26, 27]. Glut2 is usually expressed in pancreaticislets, very low levels of Glut2 transporters being foundin pancreatic ducts, and its overexpression has not beenassociated with PDAC. Using immunohistochemistry, wedemonstrate that Glut2 protein levels are clearly increasedin neoplastic ducts compared to normal ducts. Recently,other glycolytic enzymes, such as α-enolase andglyceraldehyde-3-phosphate dehydrogenase (Gapdh),have been found to be overexpressed in PDAC comparedto paired non-cancerous tissues [28]. Glucose uptake isoften correlated with the proliferative index in tumours.

Fig. 3 FDG uptake in young Ela1-myc mice. a, b Representativecoronal FDG PET images of a C57Bl6 control mouse (a) and of anEla1-myc mouse (b). c Relative FDG accumulation (%ID/g) in thepancreas of 10-week-old mice (n=6) as measured ex vivo in a gammacounter. Although not detectable by FDG PET, the pancreas of the

Ela1-myc mice showed a significantly higher FDG uptake comparedto control littermates (p=0.021). Data are represented as mean andstandard deviation. d, e Histological aspect of the pancreas of an 11-week-old wild-type (d) and Ela1-myc mouse (e). The latter micro-photograph displays an acinar tumour (e)


Fig. 5 Immunohistochemical analysis of Glut2 expression in normaland neoplastic pancreas tissues. a Glut2 expression in normalpancreas is restricted mainly to endocrine cells (asterisk), with verylow levels being found in ductal cells. The arrow indicates a duct,shown at a higher magnification in the inset. b Acinar pancreatictumour showing no Glut2 labelling in neoplastic cells. c An area of

the same tumour showing ductal differentiation displays Glut2expression within neoplastic ducts but not in the stroma. The arrowindicates a neoplastic duct, shown at a higher magnification in theinset. Glut2 expression is higher in the membrane of positive cells (seearrows in the inset). Scale bar: 50 μm (a, b and c), 25 μm (for theinsets)

Fig. 4 Quantitative RT-PCR analysis of RNA isolated from tumourareas rich in acinar or ductal components. The acinar or ductalphenotype of the selected regions was confirmed by RT-PCR usingappropriate differentiation markers, taking the expression of ahousekeeping gene (Hprt) as a reference. a The amylase/Hprt andCpa/Hprt ratios were significantly higher in microdissected acinartumour areas (n=5) than in ductal ones (n=3). b In contrast,

significantly higher Krt7/Hprt and Cftr/Hprt ratios were observed inmicrodissected ductal tumour areas. c Analysis of the Glut trans-porters expression revealed that Glut2 had a significantly higherexpression in ductal than in acinar tumour areas, whereas othertransporters such as Glut1 and Glut3 showed similar relative mRNAlevels in both tumour types. d Ductal tumours also showed a higherexpression of HkI and HkII than acinar tumours


However, differences in proliferation are unlikely to accountfor the higher FDG uptake and Glut2 and HkII expressionobserved in ductal tumours: we have previously shownslightly higher—though not statistically significant—Ki67labelling in tumours displaying predominant acinar differen-tiation [5]. Altogether, a variety of mechanisms contribute toenhanced glycolysis in pancreatic cancer, and a betterunderstanding of the mechanisms of FDG accumulation inPDAC and its precursors is needed.

Ductal tumours, both in humans and mice, are charac-terized by a strong desmoplasia. We also show here that thehigh levels of Glut2 mRNA are associated with proteinoverexpression in neoplastic cells rather than in theactivated stroma. Because the Ela1-myc model does notrecapitulate the progression of PanIN progression modelassociated with human PDAC, it is not possible here toestablish when—during the progression of intraductallesions—Glut2 overexpression takes place. Other morerecent and sophisticated mouse models recapitulate moreclosely the genetic events associated with human PDAC[18, 29–31]. In this regard, Guerra et al. [18] and Hingoraniet al. [32] have reported that selective expression of mutantK-ras in fetal pancreatic cells results in PanINs andinvasive PDAC, thus making these models most suitablefor future studies. In addition, Guerra et al. have shownthat, in adult mice, cooperation of chronic pancreatitis andK-ras activation is required for PDAC development. Sinceclinical work indicates a substantial overlap in the semi-quantitative uptake values observed in patients with chronicpancreatitis vs PDAC [33], the model of Guerra et al. mayalso contribute to develop better imaging strategies todistinguish between chronic pancreatitis and PDAC [18].This model will also be more appropriate to determine thepotential of FDG PET in the detection of early PanINlesions. However, it must be emphasized that there is stilldebate as to the sequence of events regarding PanINevolution and the risk of different PanINs in theirprogressing to PDAC [2, 34].

Our work does not have direct implications regarding theearly detection of PDAC in humans. However, our findingssupport the notion that the study of the FDG PET imagingproperties of pancreatic ductal precursor lesions may revealimportant clues as to the basis of the biological differencesbetween acinar and ductal tumours and may contribute todevelop better imaging methods to detect preneoplasticlesions.

Acknowledgements We are very grateful to Dr. Eric P. Sandgren forproviding Ela1-myc mice and Dr. Chaitanya R. Divgi for criticalreading of the manuscript. We also thank S. Mancilla, T. Lobato andA. Doutres for technical support and the personnel of the Radiochem-istry Laboratory of IAT for FDG manufacture. This study was partiallysupported by the RTICC (Instituto de Salud Carlos III) and by grantsfrom Plan Nacional de I + D, Ministerio de Educación y Ciencia

(SAF2005-00704), Fundació La Marató TV3 (grant 051110) andGeneralitat de Catalunya (2005SGR00729) to PN and from PlanNacional de I + D, Ministerio de Educación y Ciencia (SAF2004-01137) and VIth Framework WU Programme (Biomed Programme(LSHB-CT-2006-018771 MOLDIAG-PaCa) to FXR. IA was partiallysupported by a postdoctoral fellowship from the Basque Government.

References

1. DiMagno EP, Reber HA, Tempero MA. AGA technical review onthe epidemiology, diagnosis, and treatment of pancreatic ductaladenocarcinoma. American Gastroenterological Association. Gas-troenterology 1999;117:1464–84.

2. Real FX. A “catastrophic hypothesis” for pancreas cancerprogression. Gastroenterology 2003;124:1958–64.

3. Chin BB, Chang PP. Gastrointestinal malignancies evaluated with(18)F-fluoro-2-deoxyglucose positron emission tomography. BestPract Res Clin Gastroenterol 2006;20:3–21.

4. Delbeke D, Pinson CW. Pancreatic tumors: role of imaging in thediagnosis, staging, and treatment. J Hepatobiliary Pancreat Surg2004;11:4–10.

5. Aguilar S, Corominas JM, Malats N, Pereira JA, Dufresne M,Real FX, et al. Tissue plasminogen activator in murine exocrinepancreas cancer: selective expression in ductal tumors andcontribution to cancer progression. Am J Pathol 2004;165:1129–39.

6. Sandgren EP, Quaife CJ, Paulovich AG, Palmiter RD, Brinster RL.Pancreatic tumor pathogenesis reflects the causative genetic lesion.Proc Natl Acad Sci USA 1991;88:93–7.

7. Liao DJ, Wang Y, Wu J, Adsay NV, Grignon D, Khanani F, et al.Characterization of pancreatic lesions from MT-tgf alpha, Ela-mycand MT-tgf alpha/Ela-myc single and double transgenic mice. JCarcinog 2006;5:19.

8. Thorens B. Molecular and cellular physiology of Glut-2, a high-Kmfacilitated diffusion glucose transporter. Int Rev Cytol 1992;137:209–38.

9. Smith TAD. The rate-limiting step for tumor [18F]fluoro-2-deoxy-D-glucose (FDG) incorporation. Nucl Med Biol 2001;28:1–4.

10. Hernández-Muñoz I, Skoudy A, Real FX, Navarro P. Pancreaticductal adenocarcinoma: cellular origin, signaling pathways andstroma contribution. Pancreatology 2008;8:462–9.

11. Pauwels EK, Ribeiro MJ, Stoot JH, McCready VR, Bourguignon M,Mazière B. FDG accumulation and tumor biology. Nucl Med Biol1998;25:317–22.

12. Macheda ML, Rogers S, Best JD. Molecular and cellularregulation of glucose transporter (Glut) proteins in cancer. J CellPhysiol 2005;202:654–62.

13. Smith TA. Mammalian hexokinases and their abnormal expressionin cancer. Br J Biomed Sci 2000;57:170–8.

14. Nakata B, Nishimura S, Ishikawa T, Ohira M, Nishino H, Kawabe J,et al. Prognostic predictive value of 18F-fluorodeoxyglucose positronemission tomography for patients with pancreatic cancer. Int J Oncol2001;19:53–8.

15. Zimny M, Fass J, Bares R, Cremerius U, Sabri O, Buechin P, et al.Fluorodeoxyglucose positron emission tomography and the progno-sis of pancreatic carcinoma. Scand J Gastroenterol 2000;35:883–8.

16. Seitz U, Wagner M, Vogg AT, Glatting G, Neumaier B, Greten FR,et al. In vivo evaluation of 5-[(18)F]fluoro-2′-deoxyuridine as tracerfor positron emission tomography in a murine pancreatic cancermodel. Cancer Res 2001;61:3853–7.

17. Buck AC, Schirrmeister HH, Guhlmann CA, Diederichs CG, Shen C,Buchmann I, et al. Ki-67 immunostaining in pancreatic cancer andchronic active pancreatitis: does in vivo FDG uptake correlate withproliferative activity? J Nucl Med 2001;42:721–5.


18. Guerra C, Schuhmacher AJ, Canamero M, Grippo PJ, Verdaguer L,Perez-Gallego L, et al. Chronic pancreatitis is essential for inductionof pancreatic ductal adenocarcinoma by K-Ras oncogenes in adultmice. Cancer Cell 2007;11:291–302.

19. Knoess C, Siegel S, Smith A, Newport D, Richerzhagen N,Winkeler A, et al. Performance evaluation of the microPET R4PET scanner for rodents. Eur J Nucl Med Mol Imaging2003;30:737–47.

20. Phelps ME, editor. PET molecular imaging and its biologicalapplications. New York: Springer; 2004.

21. van KouwenMC, Laverman P, van Krieken JH, OyenWJ, Jansen JB,Drenth JP. FDG-PET in the detection of early pancreatic cancer in aBOP hamster model. Nucl Med Biol 2005;32:445–50.

22. Higashi T, Tamaki N, Torizuka T, Nakamoto Y, Sakahara H,Kimura T, et al. FDG uptake, Glut-1 glucose transporter andcellularity in human pancreatic tumors. J Nucl Med 1998;39:1727–35.

23. von Forstner C, Egberts JH, Ammerpohl O, Niedzielska D,Buchert R, Mikecz P, et al. Gene expression patterns and tumoruptake of 18F-FDG, 18F-FLT, and 18F-FEC in PET/MRI of anorthotopic mouse xenotransplantation model of pancreatic cancer.J Nucl Med 2008;49:1362–70.

24. Bos R, van Der Hoeven JJ, van Der Wall E, van Der Groep P,van Diest PJ, Comans EF, et al. Biologic correlates of (18)fluorodeoxyglucose uptake in human breast cancer measured bypositron emission tomography. J Clin Oncol 2002;20:379–87.

25. Hooft L, van der Veldt AA, van Diest PJ, Hoekstra OS, Berkhof J,Teule GJ, et al. [18F]fluorodeoxyglucose uptake in recurrentthyroid cancer is related to hexokinase i expression in the primarytumor. J Clin Endocrinol Metab 2005;90:328–34.

26. Natsuizaka M, Ozasa M, Darmanin S, Miyamoto M, Kondo S,Kamada S, et al. Synergistic up-regulation of Hexokinase-2,glucose transporters and angiogenic factors in pancreatic cancercells by glucose deprivation and hypoxia. Exp Cell Res2007;313:3337–48.

27. Yoon DY, Buchler P, Saarikoski ST, Hines OJ, Reber HA,Hankinson O. Identification of genes differentially induced byhypoxia in pancreatic cancer cells. Biochem Biophys ResCommun 2001;288:882–6.

28. Mikuriya K, Kuramitsu Y, Ryozawa S, Fujimoto M, Mori S, Oka M,et al. Expression of glycolytic enzymes is increased in pancreaticcancerous tissues as evidenced by proteomic profiling by two-dimensional electrophoresis and liquid chromatography-mass spec-trometry/mass spectrometry. Int J Oncol 2007;30:849–55.

29. Hingorani SR. New pathways to pancreatic cancer. Cancer BiolTher 2004;3:170–2.

30. Schmid RM. Genetic basis of pancreatic cancer. Best Pract ResClin Gastroenterol 2002;16:421–33.

31. Tuveson DA, Hingorani SR. Ductal pancreatic cancer in humansand mice. Cold Spring Harb Symp Quant Biol 2005;70:65–72.

32. Hingorani SR, Petricoin EF, Maitra A, Rajapakse V, King C,JacobetzMA, et al. Preinvasive and invasive ductal pancreatic cancerand its early detection in the mouse. Cancer Cell 2003;4:437–50.

33. Nitzsche EU, Hoegerle S, Mix M, Brink I, Otte A, Moser E, et al.Non-invasive differentiation of pancreatic lesions: is analysis ofFDG kinetics superior to semiquantitative uptake value analysis?Eur J Nucl Med Mol Imaging 2002;29:237–42.

34. Real FX, Cibrián-Uhalte E, Martinelli P. Pancreatic cancerdevelopment and progression: remodeling the model. Gastroen-terology 2008;135:724–8.


FDG PET imaging of Ela1-myc mice reveals major biological differences between pancreatic acinar and...

Documents

Transcript of FDG PET imaging of Ela1-myc mice reveals major biological differences between pancreatic acinar and...