Stefan Hoegerle, MDCarsten Altehoefer, MDNadir Ghanem, MDGabriele Koehler, MDCornelius F. Waller, MDHans Scheruebl, MDErnst Moser, MD, PhDEgbert Nitzsche, MD

Index terms:Carcinoid, 70.316Fluorine, radioactiveGastrointestinal tract, CT, 70.12112,

70.12115Gastrointestinal tract, MR,

70.121411, 70.121412Gastrointestinal tract, neoplasms,

70.316Gastrointestinal tract, PET, 70.12163Gastrointestinal tract, SPECT,

70.12162

Published online: July 19, 200110.1148/radiol.2202001440

Radiology 2001; 220:373–380

Abbreviation:FDG 5 2-[fluorine 18]fluoro-2-deoxy-

D-glucose

1 From the Department of Radiology,Divisions of Nuclear Medicine (S.H.,E.M., E.N.) and Diagnostic Radiology(C.A., N.G.), Department of Pathology(G.K.), and Department of InternalMedicine, Division of Hemato-oncology(C.F.W.), Albert-Ludwigs University,Hugstetterstrasse 55, 79106 Freiburg,Germany; and the Department of Gas-troenterology, University HospitalBenjamin Franklin, Free University Ber-lin, Germany (H.S.). Received August25, 2000; revision requested October17; revision received December 21;accepted February 12, 2001. Addresscorrespondence to S.H.© RSNA, 2001Author contributions:Guarantors of integrity of entire study,S.H., E.N.; study concepts, S.H., E.N.,E.M.; study design, S.H., E.N., C.F.W.;literature research, S.H., H.S., C.F.W.;clinical studies, S.H., E.N., C.A., N.G.,G.K., C.F.W.; data acquisition, S.H.,E.N., C.A., N.G., G.K.; data analysis/interpretation, S.H., E.N., C.A., N.G.,C.F.W.; manuscript preparation, S.H.,E.N.; manuscript definition of intellec-tual content, all authors; manuscriptediting, S.H., E.N.; manuscript revi-sion/review and final version approval,all authors.

Whole-Body 18F Dopa PET forDetection of GastrointestinalCarcinoid Tumors1

PURPOSE: To evaluate fluorine 18 (18F) dopa positron emission tomography (PET)in comparison with established imaging procedures in gastrointestinal carcinoidtumors.

MATERIALS AND METHODS: After evaluation of the normal distribution of 18Fdopa, 17 patients with histologically confirmed tumors were examined with 18Fdopa PET. Results of 2-[fluorine 18]fluoro-2-deoxy-D-glucose (FDG) PET, somatosta-tin-receptor scintigraphy, and morphologic imaging (computed tomographyand/or magnetic resonance imaging) were available for all patients. Results of theprocedures were evaluated by two radiologists and two nuclear medicine specialists,whose consensus based on all available histologic, imaging, and follow-up findingswas used as the reference standard.

RESULTS: Ninety-two tumors were diagnosed: eight primary tumors, 47 lymphnode metastases, and 37 organ metastases. 18F dopa PET led to 60 true-positivefindings (seven primary tumors, 41 lymph node metastases, 12 organ metastases);FDG PET, 27 (two primary tumors, 14 lymph node metastases, 11 organ metasta-ses); somatostatin-receptor scintigraphy, 52 (four primary tumors, 27 lymph nodemetastases, 21 organ metastases); and morphologic imaging, 67 (two primarytumors, 29 lymph node metastases, 36 organ metastases). This resulted in thefollowing overall sensitivities: 18F dopa PET, 65% (60 of 92); FDG PET, 29% (27 of92); somatostatin-receptor scintigraphy, 57% (52 of 92); morphologic procedures,73% (67 of 92). Although the morphologic procedures were most sensitive fororgan metastases, 18F dopa PET enabled best localization of primary tumors andlymph node staging.

CONCLUSION: 18F dopa PET is a promising procedure and useful supplement tomorphologic methods in diagnostic imaging of gastrointestinal carcinoid tumors.

Despite the great number of existing procedures, diagnostic imaging of gastrointestinalcarcinoid tumors is frequently difficult and not infrequently unsuccessful (1). The mainreason is that assessment of the walls of hollow organs in the gastrointestinal tract and oflymph nodes with morphologic imaging procedures (computed tomography [CT] andmagnetic resonance [MR] imaging) is limited. Therefore, diagnosis of primary tumors andlymph node staging is often unsatisfactory (2–6). On the other hand, the sensitivity offunctional imaging procedures (metaiodobenzylguanidine scintigraphy or somatostatin-receptor scintigraphy) is limited because their anatomic resolution is worse (7–10). Thiscan be compensated for only partially by the increased contrast attained with singlephoton emission CT (SPECT) (11). An additional limitation is the lack of tracer uptake dueto deficient catecholamine stores or a lack of receptor expression (12), which may lead tofalse-negative findings. This is especially due to the well-known variability of biologicproperties of this heterogeneous group of tumors (13).

A widespread characteristic of these tumors is the uptake and metabolization of aminoacids, which resulted in the term amine precursor uptake and decarboxylation, or APUD,system being used to denote the diffuse neuroendocrine system (14). This property oftaking up amino acids, transforming them into biogenic amines by means of decarbox-ylation, and storing them in vesicles was already used for imaging of neuroendocrinepancreatic tumors with carbon 11 (11C) dopa (3,4-dihydroxyphenylalanine) positron

Nuclear Medicine

373

emission tomography (PET) (15,16). Inaddition, there is a single case report onthe successful use of fluorine 18 (18F)dopa PET for staging a metastasizingcarcinoid tumor (17). PET offers the high-est resolution among the functionalmethods, and the use of 18F dopa, such as2-[fluorine 18]fluoro-2-deoxy-D-glucose(FDG), enables performance of whole-body examinations in one session.

The purpose of the present study was toevaluate 18F dopa whole-body PET in com-parison with established imaging proce-dures in gastrointestinal carcinoid tumors.

MATERIALS AND METHODS

Patients

After approval by the institutional re-view board, 17 consecutive patients(seven women, 10 men; age range, 19–74years; mean age, 55 years 6 16 [SD]) withhistologically proved gastrointestinal car-cinoid tumors were examined by meansof 18F dopa PET. At least one tumor hadbeen confirmed histologically in patientswith multiple metastases. All patientsgranted informed consent.

None of the patients presented withclinical symptoms of carcinoid syndrome.Urinary 5-hydroxyindoleacetic acid wasmoderately elevated in two patients. In allpatients, the carcinoid tumor was foundunexpectedly during surgical, endoscopic,or imaging procedures. On the basis of theunexpected finding, imaging procedureswere used for staging after surgical resec-tion of the primary tumor in seven cases:three appendiceal tumors (2.2-cm diame-ter, 2.0-cm diameter with solitary lymphnode metastasis, and 1.7-cm diameter withinfiltration of the mesoappendix), two ilealtumors, one colonic tumor within an in-testinal polyp, and one rectal tumor. In theremaining 10 patients, no previous surgeryhad been performed. After staging in allpatients, surgery was performed in six; nosurgery was performed in the other 11 pa-tients.

In all patients, results of morphologicimaging (CT and/or MR imaging), soma-tostatin-receptor scintigraphy, and FDGPET were available as part of clinical rou-tine staging. All of the imaging procedureswere performed within a maximum of 6weeks after the tumor was found.

Histologic and Immuno-histochemical Findings

Carcinoid tumors were diagnosed onthe basis of the histologic findings andadditional immunohistochemical tests inwhich the expression of neuroendocrine

tumor markers such as synaptophysin,neuron-specific enolase, or chromograninwas examined. According to recent classi-fications, all cases were malignant carci-noid tumors that were subdivided intocategories of high, moderate, or lowdifferentiation (13,18). All preparationswere tested by using standard immuno-histochemical procedures for serotoninexpression. Histologic and immunohisto-chemical examinations were performed byone pathologist (G.K.) blinded to the im-aging findings.

FDG and 18F Dopa PET

All patients fasted for 12 hours prior tothe start of the examination to provideoptimal conditions for uptake of the ra-diopharmaceuticals. Blood glucose levelswere controlled in all patients for theFDG PET; none was greater than normal(110 mg/dL [6.1 mmol/L]). For FDG PET,360 MBq 6 30 of FDG was injected intra-venously; for 18F dopa PET, 200 MBq 630 of 18F dopa was injected intrave-nously. Both radiopharmaceuticals wereproduced by using standard procedures(19,20). The uptake time was 90 minutesfor FDG to attain optimization of the tu-mor-to-background ratio. For 18F dopa, theuptake time was 60–90 minutes, which isanalogous to the uptake time for brain ex-aminations (21). Data were acquired with atwo-dimensional ring scanner (Ecat Exact;Siemens/CTI, Knoxville, Tenn) with a rodsource by using postinjection segmentedattenuation correction. Eight to 10 bed po-sitions with an 11-cm transverse field ofview were measured (2 minutes transmis-sion and 8 minutes emission per position).Images were reconstructed by means of aniterative procedure with ordered subsets(ordered subset-expectation maximization,or OSEM, two iterations, eight subsets)(22), no pre- or postfiltering was used, andfinal reconstruction resolution of the im-ages was 6 mm.

Comparison with Control Group

The study group was compared with acontrol group to determine normal dis-tribution of 18F dopa in the body. Thecontrol group consisted of five consecu-tive patients (three men, two women; agerange, 42–73 years; mean age, 57 years)in whom an examination with 18F dopawas clinically indicated for Parkinson dis-ease. None of the control subjects hadmalignant disease.

Somatostatin-Receptor Scintigraphy

After intravenous injection of 170MBq 6 25 of indium 111 pentetreotide (Oc-

treoScan; Mallinckrodt, Petten, the Neth-erlands), planar scintigrams were ob-tained with a large-field-of-view gammacamera (Bodyscan; Siemens, Erlangen,Germany) and a medium-energy collima-tor. Four hours and 24 hours after injec-tion, ventral and dorsal whole-body im-ages were acquired, and target images ofthe abdomen and thorax were also ob-tained. The SPECT examination of thethorax and abdomen was performed 24hours after injection by using a triple-headed camera (Prism XP 3000; PickerMarconi, Cleveland, Ohio) with the fol-lowing parameters: 128 3 128 matrix,120 projections in 3° angle increments,and 40-second acquisition time per pro-jection. Images were reconstructed by us-ing filtered back-projection: no prefilter-ing, reconstruction with ramp filter, andpostprocessing with a low-pass filter witha cutoff frequency corresponding to a 5%noise level, order of five.

Interpretation of the PET Imagesand Somatostatin-ReceptorScintigrams

The reconstructed images were as-sessed by two nuclear medicine special-ists (S.H., E.N.) on a standard computermonitor (Sun Microsystems, San Diego,Calif) at all three levels (transverse, coro-nal, and sagittal views) by using an in-verse gray scale. Somatostatin-receptorscintigrams and FDG and 18F dopa PETscans in each patient were interpreted atdifferent times, with the readers blindedto the results of the other studies. Imageswere interpreted by means of consensusof the two nuclear medicine specialists;there were no discrepancies. Any focaltracer accumulation exceeding normalregional tracer uptake was rated as a patho-logic finding—a tumor. These pathologicfindings were classified as primary tu-mors (lesion within the intestine), lymphnode metastases (tracer accumulations inregional and distant lymph nodes), andorgan metastases (pathologic uptake inparenchymatous organs such as the liveror lungs). Intestinal uptake that was lin-early and nonfocally limited was rated asa nonspecific nonpathologic finding.

Morphologic Imaging

Morphologic imaging was performedaccording to standard protocols. Abdom-inal CT and/or MR images were availablefor all patients; additional thoracic CTexaminations were performed in all pa-tients with proved abdominal tumors todetect potential tumors in the thorax.

374 z Radiology z August 2001 Hoegerle et al

Since the decision whether to use CTand/or MR imaging was made within theclinical routine independent of this study,it can be considered randomized: Therewere nine patients with abdominal CTscans, six patients with abdominal MRimages, and two with both.

CT examinations were performed withtwo scanners (Somatom Plus 4 or Soma-tom Plus S; Siemens). Abdominal helicalCT was performed with 70–120 mL ofintravenously administered iopromide(Ultravist 300; Schering, Berlin, Ger-many) and included the portal venousphase. In addition, the gastrointestinaltract was imaged after oral administra-tion of diluted diatrizoate meglumine(Gastrografin; Schering). Collimation wasset to 8 or 10 mm at a table feed of 12 or15 mm; the reconstruction interval was 8or 10 mm. For thoracic helical CT, 70–100 mL of iopromide was administeredintravenously, and the following scan-ning parameters were used: collimationof 8 or 10 mm, table feed of 12 or 15 mm,and reconstruction interval of 8 or 10mm.

Static abdominal MR images were ac-quired with 1.5-T imagers (Magnetom Vi-sion or Magnetom Symphony; Siemens)by using a body coil or body phased-arraycoil for signal reception. The upper abdo-men was imaged during a single breathhold by using nonenhanced T2-weightedturbo spin-echo sequences (2,800–3,200/120–138 [repetition time msec/echo timemsec]), nonenhanced T1-weighted gradi-ent-echo sequences (fast low angle shot,87–148/4.0–4.8, 70° flip angle), and gad-olinium-enhanced T1-weighted gradient-echo sequences after spectral fat satura-tion (104–168/4.0–4.8). The standarddose of 0.1 mmol of gadopentetate dime-

glumine (Magnevist; Schering) per kilo-gram of body weight was administeredintravenously. Pelvic MR imaging with aphased-array coil and abdominal MR im-aging with a body coil were performedwithout breath hold. T2-weighted turbospin-echo sequences (4,700/120–138) andnonenhanced and gadolinium-enhancedT1-weighted spin-echo sequences (500–750/12–15) were used. The section thick-ness was 6–8 mm for all sequences.

Interpretation of MorphologicImages

Images were interpreted by two radiol-ogists (C.A., N.G.) blinded to the resultsof the scintigraphic and PET examina-tions. If CT and MR imaging had beenperformed in the same patient, both ex-aminations were assessed at the sametime. Images were interpreted by meansof consensus of the two radiologists. Ev-ery contrast medium–enhancing tumorwithin the walls of the hollow organs ofthe gastrointestinal tract was rated as aprimary tumor. Lesions within parenchy-matous organs were rated as organ me-tastases if they were not clearly identifiedas benign lesions (eg, cystic lesions orhemangiomas) according to standard cri-teria (attenuation or signal intensity andenhancement pattern after administra-tion of contrast medium). Lymph nodeswere staged morphologically accordingto the standard criterion of nodal diam-eter. Lymph nodes with a diameter aslarge as 1 cm were rated as tumor free,and lymph nodes with a diameter ofmore than 1 cm in the longest axis wererated as lymph node metastases (5,6,23).No other lymph node criteria were as-sessed.

Data Evaluation

Since not all lesions could be histolog-ically proved and the different imagingmodalities frequently showed discrepan-cies, the results of the individual imagingprocedures were evaluated as follows.The results of the individual procedureswere interpreted by the two radiologistsand the two nuclear medicine specialistswithout knowledge of clinical data orother findings, as described earlier. Forfurther data evaluation, a committeeconsisting of the two radiologists and thetwo nuclear medicine specialists achieved aconsensual diagnosis regarding the pres-ence of tumor and the number and local-ization of tumors in each patient.

This consensus, serving as the refer-ence standard against which the resultsof the individual procedures were mea-sured, was based on histologic findings insurgical specimens in 31 lesions or theresults of all of the imaging procedures,with inclusion of serial follow-up mor-phologic imaging, in 61 lesions. On thebasis of these data, no discrepancies be-tween radiologists and nuclear medicinespecialists were observed. All lesions de-tected solely by means of 18F dopa PETwere accepted for the reference standardonly on condition that the specific lesionwas histologically verified or that a corre-late in morphologic imaging was presentinitially or disclosed at follow-up examina-tions (eg, a lymph node initially not fulfill-ing the morphologic criterion of malig-nancy [,1 cm] but showing 18F dopaaccumulation and increasing size at fol-low-up examinations). This consensus pro-cedure resulted in a set of data for eachpatient with respect to primary tumor,lymph node status, and organ metastases.Sensitivity and specificity of the individualprocedures were calculated from thesedata.

RESULTS

Control Group

With the exception of intestinal up-take, all of the persons in the controlgroup showed a uniform distributionpattern of 18F dopa in the body (Fig 1).Although no tracer accumulations wereobserved in the neck and thorax, the bileducts, especially the gallbladder, and theurogenital system (renal pelvis, ureter,and urinary bladder) showed distincttracer uptake in all patients. In the intes-tinal tract, slight contrasting of the duo-denum and parts of the pancreas was ob-served in all subjects. In three patients,there were linear nonfocal limited accu-

Figure 1. Possible Parkinson disease in a 53-year-old woman. Representative coronal 18F dopaPET images (sequence from left to right is ventral to dorsal) show normal distribution of theradiopharmaceutical in the trunk. Physiologically, the gallbladder (long thick black arrow), partsof the pancreas (short thick black arrows), duodenum (thin black arrow) and renal collectingsystem (long white arrows), right-side ureter (short white arrow), and urinary bladder (curvedarrows) are depicted.

Volume 220 z Number 2 18F Dopa PET for Detection of Carcinoid Tumors z 375

mulations in the intestine. These wereapparently nonpathologic normal variants,most likely bile within the intestine.

Study Group

The results of the various imaging pro-cedures for each patient are presented inTable 1. Patient 10 could not be included

in the final assessment, since the originalhistologic findings were later revised. Thispatient had been referred with histologi-cally confirmed multiple hepatic metasta-ses of a gastrointestinal carcinoid tumorwith unknown primary localization. Bothsomatostatin-receptor scintigraphy and 18Fdopa PET showed completely normal find-

ings, whereas FDG PET showed pro-nounced metastatic hepatic disease withmultiple other tumors in the lungs, medi-astinum, and pelvic skeleton. Because ofthis pattern of findings, which is unusualfor a carcinoid tumor, the histologic find-ings were reviewed a second time; the re-vised finding was a pulmonary adenocarci-

TABLE 1Number of Tumors Diagnosed with the Different Imaging Procedures

Patient No. Previous SurgerySubsequent

SurgeryHistologic

DifferentiationImmunohistochemicalFinding of Serotonin F18 Dopa PET FDG PET SRS*

CT and/orMR Imaging Consensus

1 No No High Positive 9† 0 3 8 9†

2 Yes (appendix) No Moderate Positive 0 0 0 0 03 No Yes High Positive 15† 5 7 7 15†

4 Yes (rectum) No Moderate Negative 1† 2† 5† 6† 6†

5 Yes (colon polyp) Yes High Positive 2† 0 1† 0 16 Yes (ileum) No Moderate Positive 2 2 4 5 57 No Yes High Positive 1 0 2† 1 2†

8 No No Moderate Negative 1 1 8 8 89 Yes (appendix) No Moderate Positive 0 0 0 0 0

10 No No Adenocarcinoma Negative 0 28† 0 27† 28†

11 No No High Negative 0 1 3 4 412 No Yes Moderate Positive 4† 2† 3† 2 4†

13 Yes (ileum) Yes High Positive 1† 0 0 1 1†

14 No No High Positive 23† 7 13† 16† 24†

15 No Yes Moderate Positive 2† 1 1 1 2†

16 Yes (appendix) No Moderate Positive 0 0 0 0 017 No No Moderate Negative 0 6 3 11 11

* SRS 5 somatostatin-receptor scintigraphy.† Primary tumor, residual tumor, or local recurrence.

TABLE 2Results of the Imaging Procedures

Lesion

F18 Dopa PET FDG PET SRS*CT and/or MR

Imaging

CT and/or MRImaging with F18

Dopa PET

ConsensusTP TN FP FN TP TN FP FN TP TN FP FN TP TN FP FN TP TN FP FN

Primary tumor 7 7 1 1 2 8 0 6 4 7 1 4 2 8 0 6 7 7 1 1 8Lymph node metastases 41 8 0 6 14 8 0 33 27 8 0 20 29 8 0 18 47 8 0 0 47Organ metastases 12 8 0 25 11 8 0 26 21 8 0 16 36 8 2 1 37 8 2 0 37Overall 60 25 1 32 27 24 0 65 52 23 1 40 67 24 2 25 91 23 3 0 92

Note.—Data are the number of lesions and exclude patient 10, in whom the original histologic findings were later revised. FN 5 false-negative finding,FP 5 false-positive finding, TN 5 true-negative finding, TP 5 true-positive finding.

* SRS 5 somatostatin-receptor scintigraphy.

TABLE 3Calculated Sensitivities of the Imaging Procedures

Lesion F18 Dopa PET FDG PET SRS* CT and/or MR ImagingCT and/or MR Imaging

with F18 Dopa PET

Primary tumor (n 5 8) 88 (7) 25 (2) 50 (4) 25 (2) 88 (7)Lymph node metastases (n 5 47) 87 (41) 30 (14) 57 (27) 62 (29) 100 (47)Organ metastases (n 5 37) 32 (12) 30 (11) 57 (21) 97 (36) 100 (37)Overall (n 5 92) 65 (60) 29 (27) 57 (52) 73 (67) 99 (91)

Note.—Data are percentages. Numbers in parentheses are the number of lesions found with each modality.* SRS 5 somatostatin-receptor scintigraphy.

376 z Radiology z August 2001 Hoegerle et al

noma with only slight neuroendocrinesubpopulation.

The results of the individual imagingprocedures compared with the referencestandard are listed in Table 2; data referto the remaining 16 patients. The great-est number of true-positive findings wasobtained with morphologic imaging, fol-lowed by 18F dopa PET, somatostatin-re-ceptor scintigraphy, and FDG PET.

After subdivision of the results into thecategories of primary tumors, lymphnode metastases, and organ metastases,the morphologic procedures were mostsensitive in the detection of organ metas-tases, whereas diagnosis of primary tu-mor and lymph node staging were con-siderably less reliable. The best resultswith respect to these latter two evalua-tion categories were obtained with 18F

dopa PET. By using a combination ofmorphologic imaging and 18F dopa PET,seven of eight primary tumors, all lymphnode metastases, and all organ metasta-ses were correctly identified.

Table 3 shows the sensitivities of theindividual procedures calculated on thebasis of the applied reference standard,which cannot yet be considered optimal(see Discussion). Since all of the proce-dures led to only isolated false-positiveresults that were confirmed with histo-logic or follow-up findings, specificitywas greater than 90% for all of the imag-ing methods.

In 38% (six of 16) of the patients, 18Fdopa PET produced additional informa-tion that was obtained with none of theother imaging procedures. By using 18Fdopa PET, it was possible in three of thesepatients to localize the previously un-known primary tumors (Fig 2) that wereconfirmed by means of histologic exam-ination in two patients and follow-up inone patient. In one patient, 18F dopa PETenabled proof of a local tumor recurrencethat was histologically confirmed. In fourpatients, previously unknown lymphnode or organ metastases were diag-nosed. Of the 18 lymph node metastasesdetected solely by means of 18F dopa PET,seven were confirmed histologically, andthe remaining 11 nodes initially showeda morphologic correlate (lymph nodes,1 cm) or were confirmed at follow-upexamination. The one organ metastasiswas also confirmed by means of fol-low-up morphologic imaging.

In 31% (five of 16) of the patients, theresults of 18F dopa PET resulted in modi-fication or even complete change intherapeutic strategy. In three of these pa-tients, the results led to surgical interven-tion with the aim of potentially curativeoutcome. In two asymptomatic patients,the extent of metastasis became clear (Fig3), so surgical reduction of the tumor wasnot performed. In one patient, however,palliative surgery was performed laterowing to symptoms of intestinal stenosis.

At all of the functional imaging proce-dures, considerable variability of traceruptake was observed in the individualtumors (Table 1, Figs 4, 5). Comparisonof histologic and immunohistochemicalfindings showed that 84% (27 of 32) ofthe total number of false-negative 18Fdopa PET findings were observed in pa-tients with non–serotonin-expressing tu-mors, whereas only 16% (five of 32) offalse-negative findings were in patientswith serotonin-expressing tumors. In allseven of the patients in whom 18F dopaPET showed all tumors in agreement with

Figure 2. Known hepatic metastasis of a neuroendocrine tumor in a 62-year-old man. Coronal18F dopa PET image (A) clearly shows the small primary tumor (short arrow) in a loop of theileum, in addition to the known hepatic metastasis (long arrow). Coronal FDG PET image (B) doesnot depict the primary tumor, and even the hepatic metastasis (straight arrow) shows only aslightly elevated glucose metabolism. Despite a 12-hour fast, pronounced FDG accumulation isfound as a normal variant in the myocardium of the right and left ventricles (curved arrows).

Figure 3. Lymphogenic metastasizing ileal tumor in a 64-year-old man. Coronal 18F dopa PETimages (sequence from left to right is ventral to dorsal) show the primary tumor (thick blackarrow) and multiple lymph node metastases (thin black arrows). In addition, there are physio-logic 18F dopa accumulations in the gallbladder (short straight white arrows), renal collectingsystem (long straight white arrows), and urinary bladder (curved arrows).

Volume 220 z Number 2 18F Dopa PET for Detection of Carcinoid Tumors z 377

the reference standard, the tumors werehighly (n 5 5) or moderately (n 5 2)differentiated serotonin-expressing tu-mors.

DISCUSSION

The results of this study help confirmthat 18F dopa PET enables diagnostic im-aging of gastrointestinal carcinoid tu-mors. Initial experience shows that theresults with respect to primary tumor lo-calization and lymph node staging arebetter than those of the established im-aging procedures; serotonin-expressingtumors are apparently especially well de-picted with 18F dopa PET. Moreover, thedata demonstrate that carcinoid tumorsrequire extensive diagnostic testing, sinceadequate diagnostic certainty is not at-tained with any single imaging proce-dure. The best possible results were ob-tained with a combination of 18F dopaPET and morphologic imaging with CTand/or MR imaging, since these proce-dures complement one another to anearly optimal degree.

Although morphologic imaging proce-dures showed the greatest sensitivitywith respect to the total number of tu-mors, it was again confirmed that local-ization of primary tumor and lymphnode staging are not satisfactory. The rea-sons for this may lie in the fact that pri-mary tumors are often small and situatedin the walls of hollow organs (1,13). Theweakness of morphologic lymph nodestaging is the known lack of reliable cri-teria, since assessment can be made onlyon the basis of size (5,6,23). Owing to thisuncertainty, the size limits vary slightly,and some examiners use the longestlymph node axis, whereas others use theshortest axis.

Because of the disadvantages cited, sev-eral functional imaging procedures suchas metaiodobenzylguanidine scintigraphyand somatostatin-receptor scintigraphyhave been developed (7,8), and somatosta-tin-receptor scintigraphy is currently thenuclear medical procedure of choice (10).Sensitivities of 44%–84% (cumulative sen-sitivity, 70%) have been reported for meta-iodobenzylguanidine scintigraphy, andsensitivities of 71%–100% (cumulativesensitivity, 86%) have been reported forsomatostatin-receptor scintigraphy (10).On the basis of these results alone, the pro-cedures cannot be considered optimal.

Since small tumors (,1 cm) were fre-quently not imaged at somatostatin-re-ceptor scintigraphy in the present studyand even larger tumors showed no soma-

tostatin-receptor expression in many cases,the results for somatostatin-receptorscintigraphy in this study (overall sensi-tivity, 57%) were even lower than thosein the literature (9,10). This is probably

because the sensitivity of somatostatin-receptor scintigraphy was frequently de-termined by means of comparison withmorphologic imaging, which is limited,as discussed earlier. If only the estab-

Figure 4. Ileal tumor with multiple metastases in a 61-year-old man. Whereas transverse CT (A)and transverse 18F dopa PET (B) images both show a retroperitoneal lymph node metastasis (blackarrows) and a hepatic metastasis (white arrows), the corresponding transverse somatostatin-receptor image obtained by means of SPECT (C) depicts only the lymph node metastasis (arrow).The hepatic metastasis apparently shows no receptor expression.

Figure 5. Metastases of an unknown primary tumor in a 59-year-old man. Coronal somatosta-tin-receptor scintigram (A) and CT scan (not shown) both revealed all known tumors—multipleserotonin-negative lymph node metastases (long black arrow) and multiple hepatic metastases(short black arrow)—whereas the coronal 18F dopa PET image (B) showed only moderate 18F dopaaccumulation in one of the lymph node metastases (black arrow) and normal uptake in thegallbladder (white arrow).

378 z Radiology z August 2001 Hoegerle et al

lished imaging procedures had beencompared in the present study, all lesionsdetected solely by means of 18F dopa PETwould not have been detected (n 5 22).This would have resulted in a sensitivityof 74% (52 of 70) for somatostatin-recep-tor scintigraphy, which is within therange of previously published data (10).

The main advantage of lymph nodestaging with 18F dopa PET is that, partic-ularly in serotonin-expressing carcinoidtumors (patients 1, 3, and 14 in Table 1),the benign or malignant nature of evensmall lymph nodes (,1 cm) can be clas-sified. These nodes are hardly detectableat planar or SPECT imaging and are notunequivocally assessable at morphologicimaging. This sensitive detection of smalltumors leads to an improvement of thereference standard, which is identical tothe verified 18F dopa PET findings insome patients. Despite this advanced ref-erence standard, criticism of severalpoints is appropriate. The sensitivity of100% (47 of 47) for lymph node stagingin the combination of 18F dopa PET andmorphologic imaging is not a realisticvalue, and it is only because all lymphnode metastases negative at 18F dopa PEThad a diameter considerably greater than1 cm. However, it is plausible or evenprobable that individual lymph node me-tastases showed no 18F dopa accumula-tion and also were not enlarged. Sincethe other functional methods also didnot show additional lymph node metas-tases, and surgical exploration with his-tologic analysis was not possible in a ma-jority of cases, such findings would nothave been detected.

Little experience has been gained thusfar with respect to PET of neuroendocrinetumors. To date, FDG, 11C hydroxytryp-tophan, and 11C dopa have been used asradiopharmaceuticals (15,24,25).

Although 18F dopa has been used thusfar almost exclusively for brain examina-tions (21,26), it also enables whole-bodyexaminations for neuroendocrine tumorssuch as carcinoid tumors (17). However,similar to 11C dopa uptake (15), there is aconsiderable variability in tracer uptake.This variability can be explained with theheterogeneous nature of neuroendocrinegastrointestinal tumors, which presentconsiderable differences in biologic, his-tologic, and clinical characteristics. Thevariation ranges from classic highly dif-ferentiated carcinoid tumors to poorlydifferentiated neuroendocrine carcino-mas (small cell carcinomas) (13). Becauseof this heterogeneity, there are variousclassifications that differ in a number ofcriteria (18). Among these criteria are the

silver affinity, histologic growth pattern,localization of the primary tumor, tumordiameter, metastasis, and hormone pro-duction. In the current study, the tumorswere classified only according to biologicbehavior (benign or malignant), degreeof differentiation, and serotonin expres-sion.

Comparison with the histologic andimmunohistochemical findings showsthat serotonin-expressing tumors espe-cially take up 18F dopa. Immunohisto-chemically, general neuroendocrine tu-mor markers such as the neuron-specificenolase, synaptophysin, and chromogra-nin, and the cell-specific markers such aspeptide hormones and biogenic amines(dopamine, serotonin), can be differenti-ated (1,27). Because 11C dopa is decar-boxylated in pancreatic tumors (16), theuptake mechanism is attributed to thelong-known fact that neuroendocrine tu-mors are capable of taking up amino ac-ids, transforming them by means of de-carboxylation to biogenic amines, andthen storing them in vesicles (14). Thisgives rise to the hypothesis that onlyamine-producing tumors can be imagedwith 18F dopa PET. However, this hy-pothesis has to be proved in further stud-ies.

In a majority of cases, the tumors are ofhigh and moderate differentiation, sothey are difficult to image with FDG PETbecause of slow growth and largely nor-mal glucose metabolism (24). This phe-nomenon was demonstrated especially inlymphomas and in differentiated thyroidcarcinomas, which lose their capacity tostore radioiodine with increasing malig-nancy and undifferentiation, whereas gly-colysis and thus FDG storage increase atthe same time (28,29). The poor results ofFDG PET in the present study can be ex-plained with this hypothesis.

Potential limitations of 18F dopa PETare the substantial physiologic uptake inthe duodenum and pancreas, whichmight mask tumors in these sites, and theunspecific accumulations within the in-testine, which might lead to false-posi-tive results.

Despite the promising results of 18Fdopa PET, no basic recommendations toreplace the established somatostatin-re-ceptor scintigraphy with 18F dopa PETcan be made. Rather, somatostatin-recep-tor scintigraphy will retain its validity,since apparently some carcinoid tumorsmanifest with somatostatin-receptor ex-pression but do not take up 18F dopa.Moreover, somatostatin-receptor expres-sion is frequently tested if somatostatinanalogues, for which both nonradioac-

tive substances and substances emittingb radiation are available, are used pallia-tively (30,31).

Since both 18F dopa PET and soma-tostatin-receptor scintigraphy are expen-sive procedures and cost is increasinglyimportant, routine use of both proce-dures in combination with morphologicimaging procedures cannot be realized inall cases. However, additional use of 18Fdopa PET appears reasonable as part ofstaging when there is an immunohisto-chemical finding from a tumor with se-rotonin expression. In the current study,the results of 18F dopa PET had an impor-tant influence on further therapy innearly one-third of the patients. A sav-ings potential arises thanks to improvedcoordination of therapeutic measureswith the individual tumor spread. On theone hand, unnecessary surgery, which isstressful for the patient, is avoided; onthe other hand, the localization of theprimary tumor can support potentiallycurative surgical therapy when there islittle or no metastasis. This is especiallyimportant in the case of carcinoid tu-mors, since complete resection is theonly potentially curative therapeutic ap-proach (1,32) and should thus lead to areduction in cost later.

Basic limitations of this study werethat the number of patients was smalland that the diagnostic imaging proce-dures were performed only after the initialsurgery in some patients. This situationwas unavoidable because of unexpectedintraoperative findings. Moreover, onlyone-third of the lesions could be verifiedhistologically, which would have beenthe only real reference standard. Thislimitation results in a certain unreliabil-ity of all statistical values calculated inthis study. However, this weakness wasunavoidable because in a majority ofpatients, postoperative staging eitherbrought no evidence of additional tu-mors or surprisingly numerous tumorswere diagnosed, which made surgicaltherapy with histologic assessment im-possible.

In conclusion, 18F dopa PET is a prom-ising imaging procedure and useful sup-plement to morphologic imaging meth-ods, since it improves localization ofprimary tumors and lymph node stagingof gastrointestinal carcinoid tumors.

References1. Vinik AI, McLeod MK, Fig LM, Shapiro B,

Lloyd RV, Cho K. Clinical features, diag-nosis, and localization of carcinoid tu-mors and their management. Gastroen-terol Clin North Am 1989; 18:865–896.

2. Picus D, Glazer HS, Levitt RG, Husband

Volume 220 z Number 2 18F Dopa PET for Detection of Carcinoid Tumors z 379

JE. Computed tomography of abdominalcarcinoid tumors. AJR Am J Roentgenol1984; 143:581–584.

3. Laurent F, Raynaud M, Biset JM, Boisserie-Lacroix M, Grelet P, Drouillard J. Diagno-sis and categorization of small bowel neo-plasms: role of computed tomography.Gastrointest Radiol 1991; 16:115–119.

4. Sugimoto E, Lorelius LE, Eriksson B, ObergK. Midgut carcinoid tumours: CT appear-ance. Acta Radiol 1995; 36:367–371.

5. Bollen EC, Goei R, van’t Hof-GrootenboerBE, Versteege CW, Engelshove HA, Lam-ers RJ. Interobserver variability and accu-racy of computed tomographic assess-ment of nodal status in lung cancer. AnnThorac Surg 1994; 58:158–162.

6. Gdeedo A, Van Schil P, Corthouts B, VanMieghem F, Van Meerbeck, Van Marck E.Comparison of imaging TNM [(i)TNM]and pathological TNM [pTNM] in stagingof bronchogenic carcinoma. Eur J Cardio-thorac Surg 1997; 12:224–227.

7. Feldman JM, Blinder RA, Lucas KJ,Coleman RE. Iodine 131 metaiodobenzyl-guanidine scintigraphy of carcinoid tu-mors. J Nucl Med 1986; 27:1691–1696.

8. Krenning EP, Bakker WH, Breeman WAP,et al. Localisation of endocrine-related tu-mours with radioiodinated analogue ofsomatostatin. Lancet 1989; 1:242–244.

9. Kwekkeboom DJ, Krenning EP, BakkerWH, Oei HY, Kooij PPM, Lamberts SWJ.Somatostatin analogue scintigraphy incarcinoid tumours. Eur J Nucl Med 1993;20:283–292.

10. Hoefnagel CA. Metaiodobenzylguanidineand somatostatin in oncology: role in themanagement of neural crest tumours. EurJ Nucl Med 1994; 21:561–581.

11. Schillaci O, Scopinato F, Angeletti S, et al.SPECT improves accuracy of somatosta-tin receptor scintigraphy in abdominalcarcinoid tumors. J Nucl Med 1996; 37:1452–1456.

12. Reubi JC, Kvols LK, Waser B, et al. Detec-tion of somatostatin receptors in surgicaland percutaneous needle biopsy samples

of carcinoids and islet cell carcinomas.Cancer Res 1990; 50:569–577.

13. Kloeppel G, Heitz PU, Capella C, Solcia E.Pathology and nomenclature of humangastrointestinal neuroendocrine (carci-noid) tumors and related lesions. WorldJ Surg 1996; 20:132–141.

14. Pearse AGE. The cytochemistry and ultra-structure of polypeptide hormone pro-ducing cells of the APUD series and theembryologic, physiologic and pathologicimplications of the concept. J HistochemCytochem 1969; 17:303–313.

15. Ahlstrom H, Eriksson B, Bergstrom M,Bjurling P, Langstrom B, Oberg K. Pancre-atic neuroendocrine tumors: diagnosiswith PET. Radiology 1995; 195:333–337.

16. Bergstrom M, Eriksson B, Oberg K, et al.In vivo demonstration of enzyme activityin endocrine pancreatic tumors: decar-boxylation of carbon-11-DOPA to car-bon-11-dopamine. J Nucl Med 1996; 37:32–37.

17. Hoegerle S, Schneider B, Kraft A, Moser E,Nitzsche EU. Imaging of a metastatic gas-trointestinal carcinoid by F-18-DOPApositron emission tomography. Nuklear-medizin 1999; 38:127–130.

18. Capella C, Heitz PU, Hoefler H, Solcia E,Kloeppel G. Revised classification of neu-roendocrine tumors of the lung, pancreasand gut. Digestion 1994; 55(suppl 3):11–23.

19. Luxen A, Perlmutter M, Bida GT, et al.Remote, semiautomated production of6-[F-18]fluoro-L-dopa for human studieswith PET. J Appl Radiat Isot 1990; 41:275–281.

20. Mulholland GK. Simple rapid hydrolysisof acetyl protecting groups in FDG syn-thesis using cation exchange resins. NuclMed Biol 1995; 22:19–23.

21. Huang SC, Yu DC, Bario JR, et al. Kineticsand modeling of L-6-[18F]fluoro-dopa inhuman positron emission tomographicstudies. J Cereb Blood Flow Metab 1991;11:898–913.

22. Hudson HM, Larkin RS. Accelerated im-

age reconstruction using ordered subsetsof projection data. IEEE Trans Med Imag-ing 1994; 13:601–609.

23. Stomper P. Cancer imaging manual. Phil-adelphia, Pa: Lippincott, 1993; 51–61.

24. Adams S, Baum R, Rink T, Schumm-Drager PM, Usadel KH, Hor G. Limitedvalue of fluorine-18 fluorodeoxyglucosepositron emission tomography for theimaging of neuroendocrine tumours. EurJ Nucl Med 1998; 25:79–83.

25. Eriksson B, Bergstrom M, Lilja A, Ahl-strom H, Langstrom B, Oberg K. Positronemission tomography (PET) in neuroen-docrine gastrointestinal tumors. Acta On-col 1993; 2:189–196.

26. Martin WRW, Palmer MR, Patlak CS, CalneDB. Nigrostriatal function in humans stud-ied with positron emission tomography.Ann Neurol 1989; 26:535–542.

27. Memon MA, Nelson H. Gastrointestinalcarcinoid tumors: current managementstrategies. Dis Colon Rectum 1997; 40:1101–1118.

28. Okada J, Oonishi H, Yoshikawa K, et al.FDG-PET for predicting the prognosis ofmalignant lymphoma. Ann Nucl Med1994; 8:187–191.

29. Feine U, Lietzenmayer R, Hanke JP, HeldJ, Wohrle H, Mueller-Schauenburg W.Fluorine-18-FDG and iodine-131-iodideuptake in thyroid cancer. J Nucl Med1996; 37:1468–1472.

30. Kvols LK, Moertel CG, O’Connell MJ,Schutt AJ, Rubin J, Hahn RG. Treatmentof the malignant carcinoid syndrome:evaluation of a long-acting somatostatinanalogue. N Engl J Med 1986; 315:663–666.

31. Otte A, Herrmann R, Heppeler A, et al. Yt-trium-90 DOTATOC: first clinical results.Eur J Nucl Med 1999; 26:1439–1447.

32. Stinner B, Kisker O, Zielke A, RothmundM. Surgical management for carcinoid tu-mors of small bowel, appendix, colonand rectum. World J Surg 1996; 20:183–188.

380 z Radiology z August 2001 Hoegerle et al