ORIGINAL
ARTICLE
/
Gastrointestinal
imaging
Characterization
of
focal
pancreatic
lesions
using
normalized
apparent
diffusion
coefficient
at
1.5-Tesla:
Preliminary
experience
M.
Barral
a,
D.
Sebbag-Sfez
a,b,
C.
Hoeffel
c,
U.
Chaput
d,
A.
Dohan
a,b,
C.
Eveno
e,
M.
Boudiaf
a,
P.
Soyer
a,∗,b,faDepartmentofAbdominalImaging,hôpitalLariboisière,AssistancePublique—Hôpitauxde
Paris,2,rueAmbroise-Paré,75010Paris,France
bUniversitéParis-Diderot,SorbonneParisCité,10,avenuedeVerdun,75010Paris,France cDepartmentofRadiology,centrehospitalier,hôpitalRobert-Debré,11,boulevardPasteur,
51092Reimscedex,France
dDepartmentofDigestiveDiseases,hôpitalLariboisière,AssistancePublique—Hôpitauxde
Paris,2,rueAmbroise-Paré,75010Paris,France
eSurgicalOncologic&DigestiveUnit,hôpitalLariboisière,AssistancePublique—Hôpitauxde
Paris,2,rueAmbroise-Paré,75475Pariscedex10,France
fUMRINSERM965-Paris7‘‘Angiogenèseetrecherchetranslationnelle’’,2,rue
Ambroise-Paré,75010Paris,France
KEYWORDS Diffusion-weighted MRimaging; Apparentdiffusion coefficient; NormalizedADC; Pancreasimaging; Lesion characterization Abstract
Purpose:Tocomparethecapabilities ofapparentdiffusioncoefficient(ADC)andnormalized ADCusingthepancreaticparenchymaasreferenceorganinthecharacterizationoffocal pan-creaticlesions.
Patients and methods:Thirty-six patients with focal pancreatic lesions (malignant, n=18; benigntumors,n=10;focalpancreatitis,n=8)underwentdiffusion-weightedMRimaging(DWI) at1.5-Teslausing3bvalues(b=0,400,800s/mm2).LesionADCandnormalizedlesionADC
(definedastheratiooflesionADCtoapparentlynormaladjacent pancreas)werecompared betweenlesiontypesusingnonparametrictests.
Results:Significant differences in ADC values were found between malignant (1.150×10−3mm2/s) and benign tumors (2.493×10−3mm2/s) (P=0.004) and between
benign tumors and mass-forming pancreatitis (1.160×10−3mm2/s) (P=0.0005) but not
betweenmalignanttumorsandmass-formingpancreatitis(P=0.1092).UsingnormalizedADC,
∗Correspondingauthor.
E-mailaddress:[email protected](P.Soyer).
2211-5684/$—seefrontmatter©2013Éditionsfrançaisesderadiologie.PublishedbyElsevierMassonSAS.Allrightsreserved.
significantdifferences were found between malignanttumors (0.933×10−3mm2/s), benign
tumors(1.807×10−3mm2/s)andmass-formingpancreatitis(0.839×10−3mm2/s)(P<0.0001). Conclusion:Ourpreliminary resultssuggestthatnormalizingADCoffocalpancreaticlesions withADCofapparently normaladjacentpancreatic parenchymaprovides higherdegreesof characterizationoffocalpancreaticlesionsthantheconventionalADCdoes.
©2013Éditionsfrançaisesderadiologie.PublishedbyElsevierMassonSAS.Allrightsreserved.
Diffusion-weightedmagneticresonanceimaging(DWI)with quantitativemeasurementofapparentdiffusioncoefficient (ADC)valueshasawell-establishedroleforthediagnosisof avarietyofabdominalabnormalities[1—7].Regarding pan-creaticdisease,severalresearchershavedemonstratedthat DWIwithADCmeasurementhelpsdetectandfurther char-acterize focal pancreatic lesions[8—11] aswell asassess the severityof other pancreatic conditions[12,13]. How-ever,individualstudiesfoundconflictingresultswithrespect tothecapabilities of ADC measurement indifferentiating between pancreatic cancers and mass-forming pancreati-tis [14—16], mostly because of an overlap in ADC values betweenthesetwoentities[17].
Tolimitthepossibleinfluenceoftechnicalparameterson theresultingADCvalue,researchershaveusedanormalized ADC to improve characterization of pathologic conditions with DWI [18,19]. However, this approach has not been evaluatedyet for thecharacterizationof focalpancreatic lesions.
Accordingly, we performed this study to compare the capabilities of apparent diffusion coefficient (ADC) with thoseofnormalizedADCusingthepancreaticparenchymaas referenceinthecharacterizationoffocalpancreaticlesions.
Patients
and
methods
Patients
This retrospective study was performed according to the review board guidelines of our institution and informed
Table1 Demographicdataofthreegroupsofpatientswithfocalpancreaticlesions.
Malignanttumors(n=18) Benigntumors(n=10) Mass-formingpancreatitis(n=8)
Genderdistribution(M/F) 11/7 5/5 7/1 Age(years) Median 62 51.5 52.5 Q1—Q3 55—69 48—58 43—55 Range 36—80 23—79 39—63 Lesionlocation Head 9 6 3 Neck 0 1 0 Body 5 2 2 Tail 4 1 3 Lesionsize(mm) Median 32.5 27 33.5 Q1—Q3 26—40 25—35 24—42 Range 16—56 15—80 20—48
Q1:lowerquartile;Q3:upperquartile;M:male;F:female.
consentwasobtainedfromallpatients.FromJanuary2010 throughJanuary2012,theMRimagingdatabasesoftwo Uni-versityHospitalswereretrospectivelyqueriedtoidentifyall adult patientsreferred for MRimagingof thepancreasat 1.5-T.Theelectronicarchivingsystemsofbothinstitutions were then used toretrieve the subgroup of patients who hadfocalpancreaticlesions,asevidencedbytheresultsof histopathological analysiseither aftersurgery, endoscopic biopsyorpercutaneousbiopsy.
Thestudypopulationconsistedof36patients(23males and13 females)withameanageof 56years±12.6years (SD)(range:21—80years)whounderwentDWIexamination ofthepancreasat1.5-T.
Eighteenpatients(50%)hadmalignantpancreatictumors that consisted in 13 adenocarcinomas (including seven well-differentiated, three moderately and three poorly differentiated adenocarcinomas) and five non-secreting neuroendocrine tumors, 10 patients (28%) had benign tumorsofthepancreasthatconsistedinsevenserous cys-tadenomas, twosolid pseudopapillary neoplasms and one mucinouscystadenoma,andeightpatients(22%)hadfocal, mass-formingpancreatitisthatconsistedinfivechronic pan-creatitis andthreeauto-immune pancreatitis.The gender andagedistributionineachgroupofpatientsaredescribed inTable1.
The diagnosis of pancreatic cancer was histopatho-logically confirmed after surgical resection (n=9), endo-scopic ultrasound-guided biopsy (n=3) or percutaneous computed-tomography guidedbiopsy (n=6). Thediagnosis of mass-forming chronic pancreatitis was histopathologi-callyconfirmedaftersurgicalresection(n=2)orendoscopic
ultrasound-guided biopsy and a follow-up of at least 10-months(n=3).Thediagnosisofmass-formingauto-immune pancreatitis was made on the basis of histopathologi-calanalysisafterendoscopicultrasound-guidedbiopsyand a favorable response to steroid therapy. The diagnosis of serous and mucinous cystadenoma was histopathologi-cally confirmed after surgical resection. The diagnosis of solid pseudopapillary neoplasm was made after surgical resection; both were considered non-malignant but with uncertain potential for malignancy,with complete encap-sulation and without atypia [20,21]. Forall patients, the time interval between DWI and histopathological confir-mation was less than 10days (mean, 3.4 days; range, 1—9days).
MR
examination
protocol
All patients underwent MR imaging examination of the abdomenusinga1.5-Tsystem(MagnetomAvanto®,Siemens
Healthcare,Erlangen,Germany,runningsoftwareSyngoMR B15).High-resolutionfree-breathingT2-weightedfast spin-echosequencewithrespiratorytriggeringusingprospective acquisition correction and three-dimensional volumetric interpolatedbreath-holdgradient-echo(3DVIBE)sequence beforeandafterintravenousadministrationofa gadolinium-chelate were obtained in all patients in addition to DWI sequence.Allimagingexaminationswereperformedwitha nine-channelanteriorphased-arraycoilandanine-channel posteriorphased-arraycoil.Patientswereimagedinsupine position.
DWI was performed with a fat-suppressed single-shot spin-echoecho-planardiffusion-weightedtechniquein the axial plane with three gradient factors (b values=0, 400 and 800s/mm2) within the same acquisition. The
diffu-siongradientswereappliedinthreeorthogonaldirections along thethreemain axesof themagnet bore(i.e., fre-quency,phaseandsliceselectdirections).The single-shot echo-planarimagingreadoutwaspreceded bya diffusion-sensitizing block consisting of two 180◦ radiofrequency pulses. Parallel imaging with generalized autocalibrat-ing partially parallel acquisition (GRAPPA) was used with an acceleration factor (or reduction factor) of 2. Fat suppression consisted of a spectral adiabatic inversion-recovery (SPAIR) technique [22]. DWI was obtained using arespiratory-triggeredacquisitionandpriorto gadolinium-chelateadministrationinallpatients.Theotherparameters wereasfollows:repetitiontime/echotime,5300ms/75ms; echospacing,0.69ms;matrixsize, 182×192;6/8 partial Fourier acquisition; section thickness, 6mm; intersection gap, 1mm; voxel size, 2.1×2.0×6.0mm; field of view 380mm; number of signal averages, 4;echo-planar imag-ingfactor,148;receiverbandwidth,1736Hz/pixel;30axial sectionsacquired;acquisitiontime,257s.Imagingdatasets were reconstructed using a GRAPPA-based algorithm. No specific bowel preparation was used before MR exami-nation and no antispasmodic agents were given to the patients.
Image
analysis
MR images were analyzed using a commercially avail-ableworkstation(MMWPwiththeSyngoSoftware,Siemens
Healthcare)bytwoobservers (onefourth-yearresident in radiology and one radiologist with21years of experience ininterpretingabdominalMRimages)workinginconsensus, blindedto the histologicalnature of thefocal pancreatic lesions.Foreachfocalpancreaticlesion,largestaxial diam-eterandlocationwererecorded.
The two observers placed regions of interest (ROIs) to encompass as much as possible of each focal pancre-aticlesiononthediffusion-weightedimagesobtainedwith
b=0s/mm2. However, a 1-mm peripheral margin of focal
pancreaticlesionwasleftoutsidetheROItoavoidincluding adjacentpancreaticparenchymawithintheROI.Duringthe samesession,circularROIswithaminimumsizeof100 pix-elswere placedoneachof thefourpancreaticsegments. Special care was given to avoid pancreatic vessels, pan-creaticducts,calcificationsandartifactsinROIplacement. TheROIswerethentransferredfromtheb0imagestothe
ADCmaps by usingthe‘‘copy-and-paste’’ functionof the workstation.
The ADC maps were generatedautomatically fromthe sourcedata usingthe integrated Syngo softwareand ADC valueswerecalculated withthreeb values,includingthe
b=0,400and800valuess/mm2usingamono-exponential
fittingalgorithm[23].Becausetheb=0valuewasincluded for ADC calculation, the resulting ADC was the ADCto-tal and we did not separate perfusion and true diffusion effects[16,23].ADCandnormalizedADCwerecalculated. NormalizedADC wasdefinedasthe ratioof focal pancre-atic lesion ADC to apparently normal adjacent pancreas ADC.
The fourpancreaticsegments weredefinedasfollows: the head was defined as the pancreatic segment located betweenthesuperiormesentericveinandthe gastroduode-nalartery,thatliestotherightofthesuperiormesenteric artery;theneck(oristhmus)wasthethinsectionbetween theheadandthebodyoftheglandthatliesanteriortothe confluenceofthesuperiormesentericveinandsplenicvein, which grooves itsposterior aspect; the body wasdefined asthelongestportionofthepancreas,extendingfromthe neckandpassingtothetail,lyingtotheleftofthesuperior mesentericvessels;thetailwasdefinedasthefinalportion of theleft pancreas,that lies anteriortothe left kidney adjacenttothesplenichilum[24].
Statistical
analysis
Lesion sizes and results of ADC measurements for each focal pancreatic lesion and pancreatic parenchyma were expressed asmedians, first quartiles (q1), third quartiles (q3)andranges.Lesionsizes,ADCandnormalizedADC val-ueswerecomparedbetweensubgroupsoffocalpancreatic lesions with the Kruskal-Wallis test for overall compari-son and the Wilcoxon signed rank test was used when overall comparison was significant. Pairwise comparisons of the ADC values obtained for the different pancre-aticsegments were made using the Wilcoxon signed rank test.
R software (version 2.8, R Foundation, http://www. r-project.org/)wasusedforstatistical analysis.All statis-ticaltestsweretwo-tailedandstatisticalsignificance was consideredatP<0.05.
Results
Largestaxialdiametersoffocalpancreatic lesionsaswell aslesionlocation withrespecttospecific pancreatic seg-mentarereportedinTable1.Nosignificantdifferencesin lesionsize werefound between thethree groups offocal pancreaticlesions(P=0.775).
No significant differences in ADC values were found betweenthefourpancreaticsegments(Tables2&3). Sim-ilarly,no differences in ADC valuesof apparently healthy pancreatic parenchyma were found between the three groupsoffocalpancreaticlesions(Table4).
SignificantdifferencesinADCvalueswerefoundbetween malignanttumors(1.150×10−3mm2/s) (Fig.1)andbenign
tumors (2.493×10−3mm2/s) (Figs. 2 & 3) (P=0.004) and
between benign tumors and mass-forming pancreatitis (1.160×10−3mm2/s) (Fig.4)(P=0.0005) butnotbetween
malignanttumorsandmass-formingpancreatitis(P=0.1092) (Table5).In addition,overlapin ADCvalues betweenthe threesubgroups of focal pancreatic lesionswas observed (Fig.5).
UsingnormalizedADC,significantdifferenceswerefound between malignant tumors (0.933×10−3mm2/s), benign
tumors(1.807×10−3mm2/s)andmass-formingpancreatitis
(0.839×10−3mm2/s) (P<0.0001). Table 6 shows
normal-izedADCofpancreaticlesionsaccordingtolesiontype.By
Figure1. Sixty-year-oldmanwithpoorlydifferentiated adeno-carcinoma of the pancreatic head. Axial ADC map shows focal pancreaticlesion(arrow).ADCofthelesionis1.201×10−3mm2/s
andADCofadjacentpancreaticparenchymais1.289×10−3mm2/s. TheresultingnormalizedADCofthepancreatictumoris0.931.
Figure2. Forty-nine year-oldmanwithbenignserous cystade-nomaofthepancreas.Axialapparent diffusioncoefficient(ADC) map showsfocalpancreaticlesion (arrow). ADCofthe lesion is 2.663×10−3mm2/s andADCofadjacent pancreaticparenchyma
is1.253×10−3mm2/s. TheresultingnormalizedADCofthe
pan-creatictumoris2.125.
Figure 3. Twenty-three year-old man with benign solid pseu-dopapillary tumor of the pancreas. Axial apparent diffusion coefficient(ADC)mapshowsfocalpancreaticlesion(arrow).ADC ofthelesionis1.562×10−3mm2/sandADCofadjacentpancreatic
parenchymais1.285×10−3mm2/s.TheresultingnormalizedADC
ofthepancreatictumoris1.215.
Table2 ADC(×10−3mm2/s)ofapparentlydisease-freepancreaticsegmentsin36patientswithfocalpancreaticlesions.
ADCValue Head Neck Body Tail
Median 1.218 1.276 1.299 1.291
Q1—Q3 1.056—1.405 1.177—1.443 1.160—1.418 1.113—1.484
Range 0.656—1.814 0.921—1.642 0.446—1.976 0.580—2.176
Q1:lower quartile;Q3:upper quartile.Apparent diffusioncoefficient (ADC)indicates apparentdiffusioncoefficient.Nosignificant differencesinADCvalueswerefoundbetweenthefourpancreaticsegments(Kruskall-Wallistest).
Table3 P values for comparisons of ADC values of apparently disease-free parenchyma between the four pancreaticsegmentsin36patientswithfocalpancreatic lesions. Segments Pvalue Headvs.Neck 0.0553 Headvs.Body 0.0854 Headvs.Tail 0.0758 Neckvs.Body 0.6771 Neckvs.Tail 0.4747 Bodyvs.Tail 0.2998
ADC: apparent diffusioncoefficient. Comparisons weremade usingtheWilcoxonsignedrankedtest.
comparisonwithADC,lessdegreesofoverlapinnormalized ADCwereobserved(Fig.6).
Discussion
Asstressedrecently,onemajorlimitationofDWIisthe diffi-cultytodifferentiatebetweenpancreaticadenocarcinoma and mass-forming pancreatitis because of overlap in ADC values[17]. Ourpreliminary resultsshow that normalized ADChelpscharacterizefocalpancreaticlesionsandfurther discriminatebetweenpancreaticcancersandmass-forming pancreatitis.Inourstudy,theuseoftheconventionalADC waslessdiscriminating becauseofmarked overlapinADC valuesbetweenthesetwoentities,andthiswasconsistent withthe results of other researchers [23]. The results of ourstudyshowthatADCmeasurementsusinganormalized
Figure4. Sixty-threeyear-oldmanwithchronicpancreatitisof the pancreatic head. Axial apparent diffusion coefficient (ADC) mapshowsfocalpancreatic lesion (arrow).ADCof thelesion is 0.940×10−3mm2/sand ADCofadjacent pancreaticparenchyma
is1.113×10−3mm2/s.TheresultingnormalizedADCofthe
mass-formingpancreatitisis0.847.
ADCismorediscriminatingthanthemorecommonADCto differentiate between focal pancreatic lesions,and more specificallybetweenmalignantpancreatictumorsand mass-formingpancreatitis.
Recent studies have evaluated the capabilities of ADC measurement in discriminating between malignant and benignpancreaticconditionsandotherhavedeterminedto whatextentADCmeasurementhelpsgradetheseverityof chronicpancreatitis[10—12].Inthis regard,someauthors foundthatmalignantpancreatictumorshaveADCvalues sig-nificantlylowerthanthatofnormalpancreaticparenchyma asobservedinourstudy[11,16].
Table4 ADC(×10−3mm2/s)ofapparentlydisease-freepancreaticparenchymainthreegroupsofpatientswithfocal
pancreaticlesions.
Subgroup Head Neck Body Tail
Malignanttumors Median 1.238 1.260 1.264 1.277 Q1—Q3 1.048—1.465 1.138—1.449 1.101—1.460 1.110—1.487 Range 0.656—1814 0.921—1.642 0.446—1.976 0.580—2.176 Mean 1.158 1.240 1.265 1.161 Benigntumors Median 1.215 1.258 1.273 1.298 Q1—Q3 1.163—1.296 1.209—1.349 1.160—1.326 1.113—1.525 Range 0.898—1.479 1.060—1.600 1.048—1.725 1.018—1.885 Mean 1.213 1.308 1.304 1.347 Mass-formingpancreatitis Median 1.291 1.277 1.334 1.343 Q1—Q3 0.924—1.405 1.194—1.443 1.298—1.388 1.214—1.456 Range 0.696—1.530 0.938—1.470 0.953—1.653 1.031—1.469 Mean 1.214 1.245 1.329 1.316
Q1:lowerquartile;Q3:upperquartile.Apparentdiffusioncoefficient(ADC) indicatesapparent diffusioncoefficient.No significant differencesinADCvaluesofpancreaticsegmentswerefoundbetweenthethreegroupsofpatients(0.7739<P<0.9405;Kruskal-Wallis test).
Table5 ADC(×10−3mm2/s)valuesofpancreaticlesionsinthreegroupsofpatientswithfocalpancreaticlesions.
ADCvalue Malignanttumors Benigntumors Mass-formingpancreatitis
Median 1.150 2.493 1.160
Q1—Q3 0.994—1.350 1.434—2.760 1.047—1.231
Range 0.673—1.596 1.121—2.804 0.780—1.222
Q1:lowerquartile;Q3:upperquartile.Apparentdiffusioncoefficient(ADC)indicatesapparentdiffusioncoefficient.Significant differ-encesinADCvalueswerefoundbetweenthethreegroupsofpatients(P=0.0014;Kruskal-Wallistest)becauseofsignificantdifference betweenmalignantandbenigntumors(P=0.004)andbetweenbenigntumorsandmass-formingpancreatitis(P=0.0005).Conversely, nodifferenceswerefoundbetweenmalignanttumorsandmass-formingpancreatitis(P=0.1092).
Table6 NormalizedADC(×10−3mm2/s)ofpancreaticlesionsinthreegroupsofpatientswithfocalpancreaticlesions.
ADCvalue Malignanttumors Benigntumors Mass-formingpancreatitis
Median 0.933 1.807 0.839
Q1—Q3 0.907—0.954 1.248—1.919 0.759—0.878
Range 0.895—0.985 1.131—2.17 0.708—0.890
Q1:lowerquartile;Q3:upperquartile.Apparentdiffusioncoefficient (ADC)indicatesapparentdiffusioncoefficient.Significant dif-ferencesinADCvalueswerefoundbetweenthethreegroupsofpatients(P<0.0001;Kruskal-Wallistest)withasignificantdifference betweenmalignantandbenigntumors(P<0.0001),betweenbenigntumorsandmass-formingpancreatitis(P<0.0001)andbetween malignanttumorsandmass-formingpancreatitis(P=0.0144).
Several groups have investigated the potential role of ADC measurement in discriminating between malig-nant pancreatic tumors and mass-forming pancreatitis
[14—16,25,26]. As reported recently by Vermoolen et al.
[27],inconsistencieswerefoundbetweenstudies.Whereas Leeetal.[16]andTakeuchietal.[25]foundlowerADC val-uesinbenignpancreaticlesionsbycomparisonwiththose observedinmalignantones,reversedresultswerereported byFattahietal.[14],Kartalisetal.[15]andYamashitaetal.
[26].Ourresults mirror thosereportedby Lee etal. who foundthatADCofpancreaticcancersandmass-forming pan-creatitisobtainedateitherb=500s/mm2orb=1000s/mm2
Figure5. Boxplotsofapparentdiffusioncoefficient(ADC) val-uesoffocalpancreaticlesions,whichdifferedsignificantlybetween benignlesionsandtheothertwosubtypesbutnotbetween mass-formingpancreatitis and malignanttumors. Boxesstretchacross interquartilerange(IR),i.e., fromlowerquartile (Q1)toupper quartile(Q3).Bluedotsindicateoutliers.
weresignificantlylowerthanthatofpancreaticparenchyma of control patients without pancreatic disease [16]. Con-versely, the same group observed lower ADC values for mass-forming pancreatitis by comparison with pancreatic cancer,althoughwedidnotfindsuchdifferenceinourstudy usingADC.Bycontrast,wefoundsuchdifferencesbetween thesetwoconditionsusingnormalizedADConly.
Previous studies have reported ADC values of the normal pancreas using parallel imaging at 1.5-T and marked variations were found among studies. Using a free-breathing technique without respiratory triggering and that included b0 for ADC calculation
(0s/mm2≤b≤800s/mm2),Rosenkrantzetal.found mean
Figure6. Boxplotsofnormalizedapparentdiffusioncoefficient (ADC)valuesoffocalpancreaticlesions,whichdifferedsignificantly betweenthethreetypesoffocalpancreaticlesions.Boxesstretch acrossinterquartilerange(IR),i.e.,fromlowerquartile (Q1)to upperquartile(Q3).Bluedotsindicateoutliers.
ADC values of 1.26×10−3mm2/s at 1.5-T for normal
pancreatic parenchyma, which are close to the overall ADC values we found, irrespective to the type of focal pancreatic lesion being present [22]. Conversely, using a breath-hold technique at 1.5-T with two b values of 50-and500-s/mm2,Wiggermannetal.foundaverylowADCof
0.17×10−3mm2/s [10]. Using a free-breathing technique
at 1.5-T and threeb valuesof 0-,500- and 1000-s/mm2,
another group of researchers found ADC values ranging from 1.59×10−3mm2/s to 1.68×10-3mm2/s for normal
pancreatic parenchyma, with no significant differences between the three pancreatic segments (head, body and tail)[28].IthasbeenassumedthatvariationsinADCvalues may be the results of differences in patient population, imaging sequences,selection of specific b valuesfor ADC calculation,orothertechnicalacquisitionparameterssuch asslicethickness[10,28,29].Inaddition,calculationofADC valuesmaybeinfluencedbytheinclusionoflowbvaluesas explainedbytheintravoxelincoherentmotion(IVIM)theory
[30,31].Inourstudy,foragivenROI,weobtained atotal ADC valuethat consistedin theaddedresults ofdiffusion and microperfusion effects. The effect of microperfusion onthe resultingtotal ADCis more prominent usinglow b
values[32].
In our study, we found that the apparently normal parenchymashowedhomogeneousdistributionofADC val-uesamongthefourpancreaticsegments,inaccordancewith theresultsofotherstudies[33].However,weareawareof astudyinwhichthepancreatictailhadlowerADCvalueby comparisonwiththeheadandthebody[34].
Pancreaticadenocarcinomaisusuallyassociatedwithlow ADC values by comparison with those of healthy pancre-atic parenchyma because of the presence of fibrosis and increasedcellularity,which areassociatedwithrestricted waterdiffusion[35].However,necrosis,whichisafrequent componentofpancreaticadenocarcinoma,isresponsiblefor increased ADC valuesdue to increasedrandommotion of watermolecules[35].Consequently,variationsinADCfound amongindividualstudiesmaybeduetomarkeddifferences intherelativeproportionsoffibrosisandnecrosisanddegree ofcellularitywithinthetumors.
Differentiation between mass-forming pancreatitis and pancreatic cancer with conventional ADC measurement is not so straightforward because of inconsistencies and conflicting results between published studies [17]. Some studies reported greater ADC values for mass-forming pancreatitis than for pancreatic cancers [14,15], others reported greater ADC values for pancreatic cancers than formass-formingpancreatitis[16,25]whereasweand oth-ers did not find any significant differences in ADC values betweenthesetwoconditions[10,23].Onereasonmaybe that mass-forming pancreatitis may contain variable pro-portions of fibrosis and inflammation, which may explain variationsamongstudiesandoverlapinADCvaluesbetween mass-formingpancreatitisandpancreaticcancers[36].
Inourpreliminarystudy,wehave defineda normalized ADCusingthepancreaticparenchymaasreference.ADC nor-malizationhasbeendefinedalreadyintheabdomenusing thespleenasareferenceorgan[18].Wepreferredusingthe adjacentpancreaticparenchymafornormalizationbecause measurementsweremadeeasierwithROIsusedfor calcu-lation placed on the same level of slice and because we
assumed that the adjacent parenchyma wassubjected to thesamefieldheterogeneityandsusceptibilityeffectsthan thelesion.However,weagreeuponthefactthatthe appar-entlyhealthypancreaticparenchymausedfornormalization maybeinvolvedatsomedegreesbytheunderlying pancre-aticdisease. Inthis regard, Momtahenet al.found lower ADCvaluesforthepancreatic parenchymaofdisease-free patientsthan for that ofpatients withmass-forming pan-creatitis[36].
Ourstudyhasseverallimitations.First,ourresultswere obtainedfromalimitedcohortstudy,reflectingour prelim-inaryexperience.Second,we usedarespiratory-triggered technique for DWI so that our results may apply only for thisspecific acquisition technique[37,38]. A third limita-tionis that we only calculated ADC total. We agree that furtherstudiesshouldbedonetoaddressthisconcernand that the IVIM model should be appliedto investigate the discriminatingcapabilitiesoftheperfusionfraction(f)and theperfusionfreediffusionparameter(D)[39,40].Similarly, furtherstudiesshouldbedonetoinvestigateatwhatextent thenumberofbvaluesmaymodifynormalizedADC[1,5].
Inconclusion, ourpreliminary resultssuggest that nor-malizing ADC of focal pancreatic lesions with ADC of apparentlynormaladjacentpancreaticparenchymaallows todiscriminatebetweendifferenttypesoffocalpancreatic lesions.Furtherstudies,however,areneededtofully eval-uatetowhat extent normalizedADC can beusedtofully characterizefocalpancreaticlesions.Inaddition,our pre-liminary results obtained in a relatively small population shouldwarrant furtherconfirmation by largerprospective trials.
TAKE-HOMEMESSAGES
• significantdifferencesinnormalizedADCvaluesexist betweenfocalpancreaticlesionsubtypes;
• normalized ADCimproves characterization offocal pancreaticlesions;
• normalizedADCshouldbepreferredtoconventional ADC.
Disclosure
of
interest
Theauthorsdeclarethattheyhavenoconflictsofinterest concerningthisarticle.
References
[1]PadhaniAR,LiuG,KohDM,ChenevertTL,ThoenyHC,Takahara T,etal.Diffusion-weightedmagneticresonanceimagingasa cancerbiomarker:consensusandrecommendations.Neoplasia 2009;11:102—225.
[2]TaouliB.Diffusion-weightedMRimagingforliverlesion char-acterization:acriticallook.Radiology2012;262:378—80. [3]Soyer P, Corno L, Boudiaf M, Aout M, Sirol M, Placé V,
et al. Differentiation between cavernous hemangiomas and untreatedmalignantneoplasmsoftheliverwithfree-breathing diffusion-weightedMRimaging:comparisonwithT2-weighted fastspin-echoMRimaging.EurJRadiol2011;80:316—24.
[4]Thoeny HC, De Keyzer F. Extracranial applications of diffusion-weighted magnetic resonance imaging. Eur Radiol 2007;17:1385—93.
[5]Kim SY, Lee SS, Byun JH, Park SH, Kim JK, Park B, et al. Malignant hepatic tumors: short-term reproducibil-ity of apparent diffusion coefficients with breath-hold and respiratory-triggered diffusion-weightedMR imaging. Radiol-ogy2010;255:815—23.
[6]SoyerP,LagadecM,SirolM,DrayX,DuchatF,VignaudA,etal. Free-breathingdiffusion-weightedsingle-shotecho-planarMR imagingusing parallelimaging(GRAPPA2)and high bvalue for thedetectionofprimaryrectaladenocarcinoma.Cancer Imaging2010;10:32—9.
[7]Koh DM, Collins DJ. Diffusion-weighted MRI in the body: applications and challenges in oncology. Am J Roentgenol 2007;188:1622—35.
[8]IchikawaT,ErturkSM,MotosugiU,SouH,IinoH,ArakiT,etal. High-bvaluediffusion-weightedMRIfordetectingpancreatic adenocarcinoma: preliminary results. AJR Am J Roentgenol 2007;188:409—14.
[9]BakirB,Salmaslio˘gluA,PoyanliA,RozanesI,AcunasB. Diffu-sionweightedMRimagingofpancreaticisletcelltumors.Eur JRadiol2010;74:214—20.
[10]Wiggermann P, Grützmann R, Weissenböck A, Kamusella P, Dittert DD, Stroszczynski C. Apparent diffusion coefficient measurementsofthepancreas,pancreascarcinoma,and mass-formingfocalpancreatitis.ActaRadiol2012;53:135—9. [11]Matsuki M, Inada Y, Nakai G, Tatsugami F, Tanikake M,
NarabayashiI,etal.Diffusion-weighedMRimagingof pancre-aticcarcinoma.AbdomImaging2007;32:481—3.
[12]Akisik MF, Aisen AM, Sandrasegaran K, Jennings SG, Lin C, ShermanS,et al.Assessmentofchronic pancreatitis:utility ofdiffusion-weightedMRimagingwithsecretinenhancement. Radiology2009;250:103—9.
[13]ErturkSM,IchikawaT,MotosugiU,SouH,ArakiT. Diffusion-weightedMRimagingintheevaluationofpancreaticexocrine functionbeforeandaftersecretinstimulation.AmJ Gastroen-terol2006;101:133—6.
[14]FattahiR,BalciNC,PermanWH,HsuehEC,AlkaadeS,Havlioglu N,etal.Pancreaticdiffusion-weightedimaging(DWI): compar-isonbetweenmass-formingfocalpancreatitis(FP),pancreatic cancer (PC), and normal pancreas. J Magn Reson Imaging 2009;29:350—6.
[15]Kartalis N, Lindholm TL, Aspelin P, Permert J, Albiin N. Diffusion-weighted magnetic resonance imagingof pancreas tumours.EurRadiol2009;19:1981—90.
[16]Lee SS, Byun JH, Park BJ, Park SH, Kim N, Park B, et al. Quantitative analysis of diffusion-weighted magnetic reso-nanceimagingofthe pancreas:usefulnessin characterizing solidpancreaticmasses.JMagnResonImaging2008;28:928— 36.
[17]Wang Y,MillerFH,ChenZE, MerrickL, MorteleKJ,Hoff FL, etal.Diffusion-weightedMRimagingofsolidandcysticlesions ofthepancreas.Radiographics2011;31:E47—64.
[18]DoRK,ChandaranaH,FelkerE,HajduCH,BabbJS,KimD,etal. Diagnosisofliverfibrosisandcirrhosiswithdiffusion-weighted imaging: value of normalizedapparent diffusion coefficient using the spleen as reference organ. AJR Am J Roentgenol 2010;195:671—6.
[19]PapanikolaouN,GourtsoyianniS,YarmenitisS,MarisT, Gourt-soyiannis N. Comparison between two-point and four-point methodsforquantificationofapparentdiffusioncoefficientof normalliverparenchymaandfocallesions:valueof normaliza-tionwithspleen.EurJRadiol2010;73:305—9.
[20]Yin Q, Wang M, Wang C, Wu Z, Yuan F, Chen K, et al. Differentiation between benign and malignant solid
pseu-dopapillarytumorofthepancreasbyMDCT.EurJRadiol2012.
http://dx.doi.org/10.1016/j.ejrad.2012.03.013
[21]Kim CW, Han DJ, Kim J, Kim YH, Park JB, Kim SC. Solid pseudopapillarytumorofthepancreas:canmalignancybe pre-dicted?Surgery2011;149:625—34.
[22]RosenkrantzAB,OeiM,BabbJS,NiverBE,TaouliB. Diffusion-weighted imaging of the abdomen at 3.0 Tesla: image qualityandapparentdiffusioncoefficientreproducibility com-pared with 1.5 Tesla. J Magn Reson Imaging 2011;33:128— 35.
[23]KlaussM,LemkeA,GrünbergK,SimonD,ReTJ,WenteMN, et al.Intravoxel incoherentmotion MRIfor the differentia-tionbetweenmassformingchronicpancreatitisandpancreatic carcinoma.InvestRadiol2011;46:57—63.
[24]FukushimaH,ItohS,TakadaA,MoriY,SuzukiK,SawakiA,etal. Diagnosticvalueofcurvedmultiplanarreformattedimagesin multisliceCTforthedetectionofresectablepancreaticductal adenocarcinoma.EurRadiol2006;16:1709—18.
[25]TakeuchiM,Matsuzaki K, Kubo H, NishitaniH. High-b-value diffusion-weighted magnetic resonance imaging of pancre-aticcancerandmass-formingchronicpancreatitis:preliminary results.ActaRadiol2008;49:383—6.
[26]YamashitaY, NamimotoT, Mitsuzaki K, UrataJ,Tsuchigame T,TakahashiM,etal.Mucin-producingtumorofthepancreas: diagnosticvalueofdiffusion-weightedecho-planarMRimaging. Radiology1998;208:605—9.
[27]Vermoolen MA, Kwee TC, Nievelstein RA. Apparent diffu-sioncoefficientmeasurementsinthedifferentiationbetween benignand malignant lesions: a systematic review. Insights Imaging2012;3:395—409.
[28]Kilic¸kesmez O, Yirik G, Bayramo˘glu S, Cimilli T, Aydin S. Non-breath-hold high b-value diffusion-weighted MRI with parallel imaging technique: apparent diffusion coefficient determinationinnormalabdominalorgans.DiagnIntervRadiol 2008;14:83—7.
[29]SasakiM,YamadaK, WatanabeY, MatsuiM,IdaM,Fujiwara S, et al. Variability in absolute apparent diffusion coeffi-cient valuesacross different platforms may be substantial: amultivendor,multi-institutionalcomparisonstudy.Radiology 2008;249:624—30.
[30]LeBihanD,BretonE,LallemandD,AubinML,VignaudJ, Laval-JeantetM.Separationofdiffusionandperfusioninintravoxel incoherent motion MR imaging. Radiology 1988;168:497— 505.
[31]Luciani A, Vignaud A, CavetM, NhieuJT, MallatA, Ruel L, etal.Livercirrhosis:intravoxelincoherentmotionMRimaging —pilotstudy.Radiology2008;249:891—9.
[32]GuiuB,CercueilJP.Liverdiffusion-weightedMRimaging:the towerofBabel?EurRadiol2011;21:463—7.
[33]Braithwaite AC, Dale BM, Boll DT, Merkle EM. Short- and midterm reproducibility of apparent diffusion coefficient measurements at 3.0-T diffusion-weighted imaging of the abdomen.Radiology2009;250:459—65.
[34]Yoshikawa T, Kawamitsu H, Mitchell DG, Ohno Y, Ku Y, Seo Y, et al. ADC measurement of abdominal organs and lesionsusingparallelimagingtechnique.AJRAmJRoentgenol 2006;187:1521—30.
[35]Muraoka N, Uematsu H, Kimura H, Imamura Y, Fujiwara Y, MurakamiM.Apparentdiffusioncoefficientinpancreatic can-cer: characterization and histopathological correlations. J MagnResonImaging2008;27:1302—8.
[36]Momtahen AJ, Balci NC, Alkaade S, Akduman EI, Burton FR.Focalpancreatitis mimickingpancreaticmass:magnetic resonance imaging (MRI)/magnetic resonance cholangiopan-creatography(MRCP)findingsincludingdiffusion-weightedMRI. ActaRadiol2008;49:490—7.
[37]DaleBM,Braithwaite AC,Boll DT,MerkleEM. Fieldstrength anddiffusionencodingtechniqueaffecttheapparentdiffusion
coefficientmeasurementsindiffusion-weightedimagingofthe abdomen.InvestRadiol2010;45:104—8.
[38] Kwee TC, Takahara T, Koh DM, Nievelstein RA, Luijten PR. Comparison and reproducibility of ADC measurements in breathhold, respiratory triggered, and free-breathing diffusion-weightedMRimagingoftheliver.JMagnReson Imag-ing2008;28:1141—8.
[39] Patel J, SigmundEE, Rusinek H, Oei M, Babb JS, TaouliB. Diagnosisofcirrhosiswithintravoxelincoherentmotion dif-fusionMRIanddynamiccontrast-enhancedMRIaloneand in
combination:preliminary experience.JMagn ResonImaging 2010;31:589—600.
[40]LemkeA,LaunFB,KlaussM,ReTJ,SimonD,DelormeS,etal. Differentiation ofpancreas carcinoma from healthy pancre-atic tissue using multiple b-values:comparison of apparent diffusioncoefficientandintravoxelincoherentmotionderived parameters.InvestRadiol2009;44:769—75.
