Non-Hodgkin Lymphoma and Hodgkin Disease: Coregistered FDG PET and CT at Staging and Restaging—Do We Need Contrast-enhanced CT?1

University of ZurichZurich Open Repository and Archive

Winterthurerstr. 190

CH-8057 Zurich

http://www.zora.uzh.ch

Year: 2004

Non-Hodgkin lymphoma and Hodgkin disease: coregisteredFDG PET and CT at staging and restaging--do we need

contrast-enhanced CT?

Schaefer, N G; Hany, T F; Taverna, C; Seifert, B; Stumpe, K D M; von Schulthess, GK; Goerres, G W

Schaefer, N G; Hany, T F; Taverna, C; Seifert, B; Stumpe, K D M; von Schulthess, G K; Goerres, G W (2004).Non-Hodgkin lymphoma and Hodgkin disease: coregistered FDG PET and CT at staging and restaging--do we needcontrast-enhanced CT? Radiology, 232(3):823-829.Postprint available at:http://www.zora.uzh.ch

Posted at the Zurich Open Repository and Archive, University of Zurich.http://www.zora.uzh.ch

Originally published at:Radiology 2004, 232(3):823-829.

Schaefer, N G; Hany, T F; Taverna, C; Seifert, B; Stumpe, K D M; von Schulthess, G K; Goerres, G W (2004).Non-Hodgkin lymphoma and Hodgkin disease: coregistered FDG PET and CT at staging and restaging--do we needcontrast-enhanced CT? Radiology, 232(3):823-829.Postprint available at:http://www.zora.uzh.ch

Posted at the Zurich Open Repository and Archive, University of Zurich.http://www.zora.uzh.ch

Originally published at:Radiology 2004, 232(3):823-829.

Non-Hodgkin lymphoma and Hodgkin disease: coregisteredFDG PET and CT at staging and restaging--do we need

contrast-enhanced CT?

Abstract

PURPOSE: To retrospectively compare diagnostic value of coregistered fluorine 18 fluorodeoxyglucosepositron emission tomographic (PET) and computed tomographic (CT) scans obtained with low-dosenonenhanced CT (PET/CT) with those routinely obtained with contrast material-enhanced CT forstaging and restaging of disease in patients with Hodgkin disease or high-grade non-Hodgkinlymphoma. MATERIALS AND METHODS: Sixty patients (mean age, 39.6 years +/- 17.1 [standarddeviation]) with Hodgkin disease (n = 42) or high-grade non-Hodgkin lymphoma (n = 18) were includedin this retrospective study. All patients underwent PET/CT and contrast-enhanced CT within amaximum of 24 days (mean, 9.1 days +/- 7.0) of each other for staging (n = 19) or first follow-upexamination (n = 41). Findings were extracted from original written reports (PET/CT, contrast-enhancedCT) and compared with findings of reference standard, which included biopsy or follow-up withclinical, laboratory, or other imaging findings. For statistical analysis, sensitivity and specificity werecalculated with findings of the reference standard. Agreement of both methods was determined withCohen kappa and McNemar tests on a per-patient basis. RESULTS: For evaluation of lymph nodeinvolvement, sensitivity of PET/CT and contrast-enhanced CT was 94% and 88%, and specificity was100% and 86%, respectively. For evaluation of organ involvement, sensitivity of PET/CT andcontrast-enhanced CT was 88% and 50%, and specificity was 100% and 90%, respectively. Agreementof both methods was excellent (kappa = 0.84) for assignment of lymph node involvement but only fair(kappa = 0.50) for extranodal disease. A difference with P <.05 (McNemar test) was consideredsignificant in regard to exclusion of disease with PET/CT, compared with contrast-enhanced CT.CONCLUSION: PET/CT performed with nonenhanced CT is more sensitive and specific than iscontrast-enhanced CT for evaluation of lymph node and organ involvement, especially regardingexclusion of disease, in patients with Hodgkin disease and high-grade non-Hodgkin lymphoma.

Niklaus G. Schaefer, MD2

Thomas F. Hany, MDChristian Taverna, MDBurkhardt Seifert, PhDKatrin D. M. Stumpe, MDGustav K. von Schulthess,

MD, PhDGerhard W. Goerres, MD

Index terms:Dual-modality imaging, PET/CT,

99.12919, 99.12963Hodgkin disease, CT, 99.12912Hodgkin disease, stagingLymphoma, CT, 99.12912Lymphoma, PET, 99.12963Lymphoma, staging

Published online before print10.1148/radiol.2323030985

Radiology 2004; 232:823–829

Abbreviation:FDG � fluorodeoxyglucose

1 From the Departments of Nuclear Med-icine (N.G.S., T.F.H., K.D.M.S., G.K.v.S.,G.W.G.) and Internal Medicine (C.T.),University Hospital Zurich, Raemistrasse100, CH-8091 Zurich, Switzerland; andDepartment of Biostatistics, University ofZurich, Switzerland (B.S.). Received June23, 2003; revision requested August 29;final revision received February 2, 2004;accepted February 16. Address cor-respondence to G.W.G. (e-mail:[email protected]).

Authors stated no financial relation-ship to disclose.

Current address:2 Department of Internal Medicine,Townhospital Waid, Zurich, Switzer-land.

Author contributions:Guarantors of integrity of entire study,N.G.S., G.W.G.; study concepts, N.G.S.,T.F.H., C.T., G.W.G.; study design,N.G.S., G.W.G.; literature research,N.G.S., G.W.G.; clinical studies, T.F.H.,K.D.M.S., G.W.G.; data acquisition,N.G.S., C.T.; data analysis/interpretation,N.G.S., T.F.H., G.W.G.; statistical analysis,N.G.S., G.W.G., B.S.; manuscript prepara-tion, N.G.S., T.F.H., G.K.v.S., G.W.G.;manuscript definition of intellectual con-tent and revision/review, all authors;manuscript editing, N.G.S., T.F.H.,G.W.G.; manuscript final version ap-proval, T.F.H., G.W.G., N.G.S., B.S.N.G.S. and T.F.H. contributed equallyto this work.© RSNA, 2004

Non-Hodgkin Lymphoma andHodgkin Disease: CoregisteredFDG PET and CT at Stagingand Restaging—Do We NeedContrast-enhanced CT?1

PURPOSE: To retrospectively compare diagnostic value of coregistered fluorine 18fluorodeoxyglucose positron emission tomographic (PET) and computed tomo-graphic (CT) scans obtained with low-dose nonenhanced CT (PET/CT) with thoseroutinely obtained with contrast material–enhanced CT for staging and restaging ofdisease in patients with Hodgkin disease or high-grade non-Hodgkin lymphoma.

MATERIALS AND METHODS: Sixty patients (mean age, 39.6 years � 17.1[standard deviation]) with Hodgkin disease (n � 42) or high-grade non-Hodgkinlymphoma (n � 18) were included in this retrospective study. All patients under-went PET/CT and contrast-enhanced CT within a maximum of 24 days (mean, 9.1days � 7.0) of each other for staging (n � 19) or first follow-up examination (n �41). Findings were extracted from original written reports (PET/CT, contrast-en-hanced CT) and compared with findings of reference standard, which includedbiopsy or follow-up with clinical, laboratory, or other imaging findings. For statisticalanalysis, sensitivity and specificity were calculated with findings of the referencestandard. Agreement of both methods was determined with Cohen � and McNemartests on a per-patient basis.

RESULTS: For evaluation of lymph node involvement, sensitivity of PET/CT andcontrast-enhanced CT was 94% and 88%, and specificity was 100% and 86%,respectively. For evaluation of organ involvement, sensitivity of PET/CT and con-trast-enhanced CT was 88% and 50%, and specificity was 100% and 90%, respec-tively. Agreement of both methods was excellent (� � 0.84) for assignment oflymph node involvement but only fair (� � 0.50) for extranodal disease. A differencewith P � .05 (McNemar test) was considered significant in regard to exclusion ofdisease with PET/CT, compared with contrast-enhanced CT.

CONCLUSION: PET/CT performed with nonenhanced CT is more sensitive andspecific than is contrast-enhanced CT for evaluation of lymph node and organinvolvement, especially regarding exclusion of disease, in patients with Hodgkindisease and high-grade non-Hodgkin lymphoma.© RSNA, 2004

Positron emission tomography (PET) with fluorine 18 fluorodeoxyglucose (FDG) is used forstaging and follow-up examinations in patients with Hodgkin disease and non-Hodgkinlymphoma. In comparison with morphologic imaging with contrast material–enhancedcomputed tomography (CT), metabolic imaging with FDG PET showed a higher specificityin the staging of disease (1–3). Small lesions, however, may be missed at PET when FDGuptake is low. The addition of metabolic imaging can have a great effect on treatment inpatients with a residual mass at posttreatment evaluation (4). Therefore, FDG PET mainlyhas been performed in addition to contrast-enhanced CT for staging and follow-upexaminations.

With the recent introduction of in-line FDG PET with coregistered nonenhanced CT

823

Ra

dio

logy

(PET/CT) scanners, a combined methodof metabolic and morphologic imaging isavailable. PET and CT data can be ac-quired in the same imaging session, with-out the need to change the patient posi-tion between scanning with one modalityand the other, to obtain coregistered im-ages. Routinely, nonenhanced low-doseCT is used for attenuation correction, aswell as image coregistration (5).

The aim of this study was to retrospec-tively compare the diagnostic value ofcoregistered PET/CT scans obtained withlow-dose nonenhanced CT with thoseroutinely obtained with contrast-en-hanced CT for staging and restaging ofdisease in patients with Hodgkin diseaseor high-grade non-Hodgkin lymphoma.

MATERIALS AND METHODS

Patients

Between May 2001 and October 2002,60 patients (37 men, 23 women; meanage, 39.6 years � 17.1 [standard devia-tion]) who were examined in our PET/CTinstitute at Department of Nuclear Med-icine, University Hospital Zurich, Zurich,Switzerland, were included in this retro-spective analysis. Forty-two had histolog-ically proved Hodgkin disease; of thisnumber, 27 were men (mean age, 37.3years � 14.3) and 15 were women (meanage, 33.1 years � 14.6). Eighteen hadnon-Hodgkin lymphoma; of this num-ber, 10 were men (mean age, 46.2 years �23.8) and eight were women (mean age,51.3 years � 14.3). In Table 1, patientcharacteristics are listed. We consecu-tively included all patients who under-went contrast-enhanced CT and PET/CTwithin 24 days apart of each other with-out treatment between examinations inthis time period. For staging purposes,PET/CT and contrast-enhanced CT wereperformed in 19 patients (11 withHodgkin disease and eight with non-Hodgkin lymphoma). In 41 patients eval-uated for restaging (31 with Hodgkindisease and 10 with non-Hodgkin lym-phoma), PET/CT and contrast-enhancedCT were performed after at least two cy-cles of chemotherapy or after given times(3, 6, or 12 months) after the completionof therapy. The records of all patientswere reviewed (N.G.S.) in accordancewith the ethical guidelines of the hospi-tal institutional review board after signedwritten informed consent was obtained.

Imaging

All data were acquired with a com-bined PET/CT in-line system (Discovery

LS; GE Medical Systems, Waukesha, Wis).This dedicated system integrates a PETscanner (Advance NXi; GE Medical Sys-tems) with a multisection helical CTscanner (LightSpeed Plus; GE MedicalSystems) and permits the acquisition ofcoregistered CT and PET images in onesession.

Patients fasted for at least 4 hours priorto scanning, which started 40–60 min-utes after the injection of a standard doseof 370 MBq of FDG. In addition, an oralCT contrast agent (Micropaque Scanner;Guerbet, Aulnay-sous-bois, France) wasadministered, starting 15 minutes beforethe injection of FDG (6). Patients wereexamined in the supine position. Nointravenous contrast agent was adminis-tered. An oral contrast agent was admin-istered for all CT and PET/CT examina-tions. Initially, starting at the level of thehead, the CT scans were acquired withthe following parameters: 80 mA, 140 kV,0.5-second tube rotation, 4.25-mm sec-tion thickness, 867-mm scan length, and22.5-second data acquisition time. TheCT scans were acquired during breathhold with the normal expiration posi-tion, and scanning included the areafrom the head to the pelvic floor.

Immediately following CT, a PET emis-sion scan was acquired, with an acquisi-tion time of 4 minutes for the emissionscan per cradle position and a one-sec-tion overlap. Acquisition of scans in sixcradle positions from the pelvic floor tothe head resulted in an acquisition timeof approximately 24 minutes. The CTdata were used for attenuation correc-tion, and images were reconstructed byusing a standard iterative algorithm (or-dered subset expectation maximization).The acquired images were viewed withsoftware that provided multiplanar refor-matted images of PET, CT, and fused datawith linked cursors (eNtegra 3.0215; GEMedical Systems).

Contrast-enhanced CT was performedin different hospitals. All were equippedwith helical CT scanners: a single-sectionunit (Sele; Picker/Elscint, Hamburg, Ger-many) and two multisection units (Emo-tion Dual Slice CT, Siemens, Erlangen,Germany; Somatom Volume Zoom, Sie-mens). The CT scanners were either sin-gle- or multisection scanners, and imag-ing of the neck, thorax, abdomen, andpelvis was performed according to theguidelines of the American College of Ra-diology, with acquisition of contiguoussections of 5-mm thickness. In all pa-tients, data acquisition included thesame anatomic regions from the neck tothe pelvic floor as were included in

PET/CT scans. All contrast-enhanced CTexaminations were performed with an in-travenous injection of 120–200 mL ofcontrast medium. All CT images were dis-played as transverse sections.

PET/CT and contrast-enhanced CTdata were acquired within a maximum of24 days (mean, 9.1 days � 7.0) of eachother. The mean clinical follow-up forthe confirmation of findings at PET/CT orcontrast-enhanced CT that were negativewas 9.3 months � 2.5.

Imaging and Pathologic FindingAnalysis

The information about possible or ev-ident pathologic findings described inoriginal written reports of PET/CT andcontrast-enhanced CT results was used toperform statistical analysis (N.G.S., B.S.).All pathologic findings as reported withPET/CT were directly compared withfindings reported with contrast-en-hanced CT for the assignment of the an-atomic localization of lymph node in-volvement and the identification oforgan infiltration by two readers (N.G.S.,G.W.G.). Only when discrepant findingsbetween PET/CT and contrast-enhancedCT were reported were the images viewedagain by two readers in consensus. Bothreaders were board-certified nuclear med-icine physicians and radiologists (T.F.H.,G.W.G.) with more than 2 years of expe-rience in interpretation of PET/CT scansand 8 years of experience in interpreta-tion of contrast-enhanced CT scans. Thisallowed assessment if a lesion was missedwith one imaging method.

All pathologic findings described inthe original written reports were rankedto assign the probability of lymphomainvolvement of a given lesion on PET/CTand contrast-enhanced CT scans, sepa-rately, with the following grading scale: 0(no lymphoma), 1 (possible lymphoma),or 2 (evident lymphoma involvement).In addition, a consensus was reachedabout whether a lesion corresponded tolymph node or organ involvement due tolymphoma. The findings of PET/CT andcontrast-enhanced CT were comparedwith histologic findings, if available (re-staging in seven of 41 patients, staging in13 of 19 patients). Clinical follow-up wasinvestigated in all 60 patients at a meanof 9.3 months � 2.5 after imaging(N.G.S.). The standard of reference waspreferably biopsy, if available, or follow-up, which included clinical, laboratory,or other imaging (plain radiography,magnetic resonance [MR] imaging, bonescintigraphy) findings.

824 � Radiology � September 2004 Schaefer et al

Ra

dio

logy

In regard to change in treatment, ret-rospective analysis of the patient’schart was performed to determinewhether the treatment was based onPET/CT or contrast-enhanced CT find-

ings. In addition, the treating oncolo-gist (N.G.S.) was contacted to deter-mine whether treatment decisions werebased on contrast-enhanced CT orPET/CT findings.

Statistical Analysis

Sensitivity and specificity were calcu-lated on the basis of the true-positive andtrue-negative findings as described in the

TABLE 1Patient Characteristics

Stage Sex/Age (y) Type of Disease Primary Site of Disease

IA F/19 HD, nodular sclerosis Mediastinal lymph nodesIA F/31 HD, nodular sclerosis Mediastinal lymph nodesIA F/76 HD, nodular sclerosis Axillary lymph nodesIB F/44 HD, nodular sclerosis Mediastinal lymph nodesIB F/26 HD, nodular sclerosis Mediastinal lymph nodesIIA F/27 HD, nodular sclerosis Paratracheal, mediastinal lymph nodesIIA F/28 HD, nodular sclerosis Subcarinal, paraaortal lymph nodesIIA F/27 HD, nodular sclerosis Mediastinal, pretracheal lymph nodesIIA F/31 HD, nodular sclerosis Mediastinal, axillary lymph nodesIIA F/37 HD, nodular sclerosis Cervical, mediastinal lymph nodesIIAE F/20 HD, nodular sclerosis Lung infiltration, mediastinal lymph nodesIIB F/41 HD, nodular sclerosis Cervical, supraclavicular, mediastinal lymph nodesIIIB F/45 HD, nodular sclerosis Mediastinal, supraclavicular, paratracheal lymph nodes; splenic hilumIVBE F/17 HD, nodular sclerosis Cervical, supraclavicular, mediastinal lymph nodes; bone infiltrationIVBE�S F/28 HD, nodular sclerosis Supraclavicular, infraclavicular, axillary, iliac lymph nodes; spleen and bone infiltrationIA M/64 HD, mixed cell type Cervical lymph nodesIIA M/46 HD, mixed cell type Pharyngeal, cervical lymph nodesIIA M/42 HD, mixed cell type Mediastinal, axillary lymph nodesIIA M/25 HD, mixed cell type Mediastinal, cervical lymph nodesIIB M/35 HD, mixed cell type Cervical, supraclavicular, mediastinal lymph nodesIIBE M/22 HD, mixed cell type Mediastinal, supraclavicular, peribronchial lymph nodes; lung infiltrationIB M/59 HD, nodular sclerosis Retroperitoneal lymph nodesIB M/31 HD, nodular sclerosis Mediastinal lymph nodesIB M/66 HD, nodular sclerosis Iliac lymph nodesIIA M/23 HD, nodular sclerosis Mediastinal, cervical lymph nodesIIA M/24 HD, nodular sclerosis Mediastinal, cervical lymph nodesIIA M/67 HD, nodular sclerosis Cervical, mediastinal lymph nodesIIA M/35 HD, nodular sclerosis Axillary, mediastinal lymph nodesIIA M/27 HD, nodular sclerosis Mediastinal, cervical lymph nodesIIA M/43 HD, nodular sclerosis Supraclavicular, mediastinal lymph nodesIIA M/20 HD, nodular sclerosis Cervical, axillary, mediastinal lymph nodesIIA M/34 HD, nodular sclerosis Iliac, inguinal lymph nodesIIA M/23 HD, nodular sclerosis Supraclavicular, peribronchial lymph nodesIIB M/37 HD, nodular sclerosis Cervical, paratracheal lymph nodesIIB M/42 HD, nodular sclerosis Mediastinal, supraclavicular, paratracheal lymph nodesIIBE M/24 HD, nodular sclerosis Mediastinal lymph nodes, lung infiltrationIIIA M/42 HD, nodular sclerosis Mediastinal lymph nodes, retroperitoneal lymph nodesIIIA M/49 HD, nodular sclerosis Paraaortal, aortopulmonary, pretracheal lymph nodesIIIAS M/47 HD, nodular sclerosis Cervical, supraclavicular, mediastinal, iliac lymph nodes; retroperitoneal lymph nodesIIIBS M/22 HD, nodular sclerosis Spleen infiltration; mediastinal, axillary, supraclavicular lymph nodesIIIBS M/25 HD, nodular sclerosis Spleen infiltration; mediastinal, supraclavicular lymph nodesIIIBS M/32 HD, nodular sclerosis Cervical, supraclavicular, mediastinal lymph nodes; retroperitoneal lymph nodes;

spleen infiltrationIAE F/52 NHL, B-cell lymphoma Lung infiltrationIB F/61 NHL, B-cell lymphoma Mediastinal lymph nodesIB F/34 NHL, B-cell lymphoma Mediastinal lymph nodesIE F/69 NHL, B-cell lymphoma Bone infiltrationIIA F/48 NHL, B-cell lymphoma Mediastinal lymph nodesIIA F/37 NHL, B-cell lymphoma Mediastinal lymph nodesIVB F/70 NHL, B-cell lymphoma Cervical, supraclavicular, axillary, mediastinal, paraaortal, iliac, inguinal lymph nodesIVBE�S F/39 NHL, B-cell lymphoma Liver, spleen infiltration; bone marrowIA M/54 NHL, B-cell lymphoma Cervical lymph nodesIAE M/81 NHL, B-cell lymphoma Duodenal infiltrationIB M/17 NHL, B-cell lymphoma Mediastinal lymph nodesIB M/77 NHL, B-cell lymphoma Mediastinal lymph nodesIIA M/17 NHL, B-cell lymphoma Mediastinal lymph nodesIIAS M/49 NHL, B-cell lymphoma Inguinal, iliac lymph nodes; spleen infiltrationIIB M/61 NHL, B-cell lymphoma Pharyngeal, cervical lymph nodesIVA M/52 NHL, B-cell lymphoma Mesenteric lymph nodesIVA M/17 NHL, B-cell lymphoma Lung infiltration, mesenteric lymph nodesIVBS M/37 NHL, T-cell lymphoma Cervical, axillary, paraaortal, iliac lymph nodes; spleen infiltration

Note.—HD � Hodgkin disease, NHL � non-Hodgkin lymphoma.

Volume 232 � Number 3 PET/CT or Contrast-enhanced CT for Lymphoma Staging � 825

Ra

dio

logy

same anatomic region (N.G.S. and B.S.).Statistical analysis was performed withsoftware (StatView, version 5.0.1; SAS In-stitute, Cary, NC). The agreement ofPET/CT and contrast-enhanced CT find-ings was assessed separately for lymphnode and organ involvement with the �statistic. The agreement was determined,with � values as follows: 0–0.20, verypoor; 0.21–0.40, poor; 0.41–0.60, fair;0.61–0.80, good; and 0.81–1.00, excel-lent.

The following regions were used forthe anatomic assignment of lymph nodeinvolvement to measure the agreementbetween the methods: neck, mediasti-num, axilla, abdomen, retroperitonealspace, and groin. Organ involvement wasassessed for the lung, liver, spleen, gastro-intestinal tract, and bone marrow.

To evaluate the difference in probabil-ity assignment between the two imagingmodalities, we separately analyzed thedata in regard to the standard of refer-ence (true-positive and true-negative re-sults according to the standard of refer-ence). We applied the McNemar testseparately in these two groups, with aconfidence level of 95% (a differencewith P � .05 was considered significant).Furthermore, the McNemar test accord-ing to sex (male or female patients) andhistologic findings (Hodgkin disease ornon-Hodgkin lymphoma) was used todetermine a significant difference be-tween these subgroups (N.G.S., G.W.G.,B.S.). All tests were performed on a per-patient basis, and hence, dependency orclustering did not occur. With the McNe-mar test, the P value after Bonferroni cor-rection was calculated as .05/6 � .008.

RESULTS

Lymph Node Involvement

At initial staging, all 19 patients in thestaging group had evidence of Hodgkindisease or non-Hodgkin lymphoma atone or more lymph node stations, andthis evidence was correctly demonstratedwith both imaging modalities. In four ofthese 19 patients, PET/CT depicted ad-ditional nodal involvement, which wasnot detected at contrast-enhanced CT.Contrast-enhanced CT did not depictnodal involvement that was not seen atPET/CT.

In a total of 41 patients who under-went imaging with both modalities forrestaging, lymph node involvementwas found in 11 patients with PET/CTand in 13 patients with contrast-en-hanced CT (Table 2). Concordant

lymph node involvement was reportedin seven patients. In four patients,PET/CT depicted additional nodal in-volvement. In two of these patients, in-creased FDG uptake was histologicallyconfirmed as nodal involvement. In athird patient, PET/CT depicted a suspi-cious lymph node that was not histo-logically or cytologically confirmed. Inthe fourth patient, a mesenteric bulkconsidered as stable disease at contrast-enhanced CT (1-year follow-up afterthe completion of chemotherapy) re-vealed increased FDG uptake at PET/CT.Two false-negative lesions were foundat PET/CT: First, a paraaortic lymphnode shown on contrast-enhanced CTimages revealed no FDG uptake and wasmissed on the nonenhanced CT imageat PET/CT. Second, subcarinal lymphnode involvement with Hodgkin dis-ease in a 3.8-cm-diameter enlargedlymph node was not depicted onPET/CT images but was identified oncontrast-enhanced CT scans and wasproved at biopsy. In 24 patients, nonodal involvement was detected withboth imaging modalities. In an addi-tional four patients, contrast-enhancedCT depicted lymph nodes suspicious forlymphoma involvement. All patientsremained without further evidence ofrecurrent lymphoma during clinicalfollow-up, and findings were consid-ered false-positive.

The sensitivity and specificity forlymph node involvement on a per-pa-tient basis was 94% (30 of 32) and 100%(28 of 28) for PET/CT and 88% (28 of 32)and 86% (24 of 28) for contrast-en-hanced CT. The McNemar test for theprobability assignment of suspicious le-sions revealed no significant differencebetween PET/CT and contrast-enhancedCT, with P � .07 for exclusion of patho-logic lymph node involvement and P �.37 for true-positive involvement oflymph nodes.

Organ Involvement

In 15 of 19 patients who underwentPET/CT and contrast-enhanced CT forstaging purposes, no organ involvementwas reported (Figure). In three patients,extranodal disease was reported atPET/CT (one patient each had bone inva-sion, spleen infiltration, and a combina-tion of both). Bone involvement wasconfirmed at bone scintigraphy, MR im-aging, and additional radiography. Thisconfirmation led to an upgrading of thestage of the disease. The spleen involve-ment was confirmed after 10 months atfollow-up PET/CT, which showed pro-gression of the disease. In all three pa-tients, the PET/CT findings were consid-ered to be true-positive (Table 3). In onepatient, at contrast-enhanced CT forstaging, findings indicated additionalsmall-bowel involvement. These findingsled to an adaptation of treatment, andthus, findings were considered true-posi-tive at contrast-enhanced CT and corre-spondingly false-negative at PET/CT. Nofalse-positive findings were reported withPET/CT and contrast-enhanced CT forstaging (Table 3).

At restaging examinations, involve-ment of one or more organs was reportedat PET/CT in four patients and at con-trast-enhanced CT in eight patients. Aconcordance for the presence of organinvolvement was reported with bothmethods in three patients. At PET/CT,additional organ involvement was iden-tified in two patients. In one patient,spleen infiltration was histologically con-firmed, with additional recurrence in aretroperitoneal lymph node and liver in-volvement; this finding was described al-ready at PET/CT and contrast-enhancedCT. This finding therefore did not changethe results on a per-patient basis. In thesecond patient, infiltration of the duode-nal loop was found with PET/CT becauseof highly increased FDG uptake afterchemotherapy. The corresponding con-

TABLE 2Results of Lymph Node Evaluation on a Per-Patient Basis

Finding

Staging Restaging

PET/CT CT* PET/CT CT*

True-positive 19 19 11 9False-positive 0 0 0 4True-negative 0 0 28 24False-negative 0 0 2 4

Note.—Data are numbers of patients.* With contrast enhancement.

826 � Radiology � September 2004 Schaefer et al

Ra

dio

logy

trast-enhanced CT scan showed no wallthickening or other signs of bowel in-volvement with lymphoma. This findingwas not histologically confirmed butnevertheless invoked resumption of che-motherapy. Therefore, the finding wasconsidered true-positive at PET/CT and

false-negative at contrast-enhanced CT(Table 3). In 32 patients, no organ in-volvement was found at both contrast-enhanced CT and PET/CT.

In five patients, seven extranodal le-sions were depicted with contrast-en-hanced CT that were not depicted at PET/

CT. In four patients, pathologic lunginfiltration was found with contrast-en-hanced CT, and in one patient, involve-ment of the liver, spleen, and bone mar-row was described. These lung lesions,usually described as patchy areas of opac-ity, were considered to correspond tolymphoma. However, at clinical evalua-tion with follow-up contrast-enhancedCT, findings were normal in all four pa-tients, thus confirming the results withtrue-negative PET/CT scans. In the pa-tient with liver, spleen, and bone marrowinvolvement depicted with contrast-en-hanced CT, findings at PET/CT were neg-ative and clinical follow-up with con-trast-enhanced CT after 4 monthsrevealed no pathologic findings or clini-cal signs of recurrence. In these five pa-tients with erroneously suspected extra-nodal lymphoma at contrast-enhanced

TABLE 3Results of Organ Evaluation on a Per-Patient Basis

Finding

Staging Restaging

PET/CT CT* PET/CT CT*

True-positive 3 1 4 3False-positive 0 0 0 5True-negative 15 15 37 32False-negative 1 3 0 1

Note.—Data are numbers of patients.* With contrast enhancement.

(a–c) PET/CT images in 17-year-old adolescent girl with swollen supraclavicular lymph nodes. Contrast-enhanced CT scan acquired 2 days earlierrevealed presence of enlarged lymph nodes in neck, thorax, and abdomen but did not identify osseous involvement. (a) Coronal maximumintensity projection image shows intense uptake bilaterally in enlarged lymph nodes at the neck/supraclavicular level (upper arrow) and uptake inmediastinum (lower arrow). There is no increased FDG uptake at abdominal lymph node sites and there is a normal appearance of the upperabdominal organs, especially the spleen. In contrast, there is an asymmetric appearance at the ischial tuberosity (arrowhead). There is FDGcontamination (*) at the right arm after intravenous tracer injection. (b) Scans show extensive involvement of lymph nodes in the mediastinum,with highly increased FDG uptake (arrowheads). Top: Transverse CT scan. Middle: Transverse PET scan. Bottom: Coregistered PET/CT image.(c) Transverse scans obtained at level of the pelvic floor reveal presence of a bone metastasis at the ischial tuberosity (arrow). This lesion has asclerotic margin and was missed on the conventional contrast-enhanced CT image. Top: Transverse CT scan. Middle: PET scan. Bottom:Coregistered PET/CT image.

Volume 232 � Number 3 PET/CT or Contrast-enhanced CT for Lymphoma Staging � 827

Ra

dio

logy

CT, the treatment did not change, and allcontrast-enhanced CT findings were con-sidered false-positive (Table 3).

The sensitivity and specificity for ex-tranodal involvement on a per-patientbasis were 88% (seven of eight) and 100%(52 of 52) for PET/CT and 50% (four ofeight) and 90% (47 of 52) for contrast-enhanced CT, respectively. The McNe-mar test for the probability assignment ofsuspicious lesions revealed no significantdifference between both imaging meth-ods, with P � .07 for the exclusion ofpathologic organ involvement and P �.37 for the assignment of organ involve-ment with Hodgkin disease and high-grade non-Hodgkin lymphoma.

On the basis of findings at PET/CT forstaging, treatment was changed in three(16%) patients, and on the basis of con-trast-enhanced CT findings, it waschanged in one (5.2%) patient. For therestaging examinations, PET/CT revealedadditional findings that led to a changeof treatment in six (15%) patients, andcontrast-enhanced CT revealed addi-tional findings that led to an adaptationof treatment in one (2.4%) patient.

Overall, the sensitivity and specificityfor lymph node and organ involvementon a per-patient basis were 93% (37 of 40)and 100% (80 of 80) for PET/CT and 80%(32 of 40) and 89% (71 of 80) for con-trast-enhanced CT, respectively. The Mc-Nemar test for the comparison of theprobability assignment of suspicious le-sions between PET/CT and contrast-en-hanced CT indicated a significant differ-ence for the exclusion of pathologicorgan or lymph node involvement (P �.004) and no significant difference for theassignment of pathologic involvement(P � .11).

There were no significant differencesin regard to assignment between the im-aging modalities according to sex (malepatients, P � .58; female patients, P �1.0) or histologic findings (Hodgkin dis-ease, P � .42; non-Hodgkin lymphoma,P � .68).

The � value as a measure for agreementbetween PET/CT and contrast-enhancedCT revealed that the agreement betweenthe methods was excellent for lymphnode staging and restaging examina-tions. However, only a poor to fair agree-ment was found for evaluation of extra-nodal disease (Table 4).

DISCUSSION

The data of this retrospective study showthat PET/CT with low-dose nonenhanced

CT for coregistration has a superior diag-nostic value compared with contrast-en-hanced CT alone for staging and restag-ing of disease in patients with high-gradenon-Hodgkin lymphoma or Hodgkin dis-ease.

Currently, contrast-enhanced CT is thefirst-line imaging modality in Hodgkindisease and non-Hodgkin lymphoma,which allows the detection of morpho-logic abnormalities such as lymph nodeenlargement or changes in contrast en-hancement that suggest organ manifesta-tion. However, disease in normal-sizedlymph nodes and in spleen and bonemarrow is less well depicted (7). In sev-eral studies, the role of FDG PET for pre-treatment as well as posttreatment eval-uation of Hodgkin disease and non-Hodgkin lymphoma has been evaluated(1–4,8–12). PET imaging with FDGshowed no difference in depiction ofnodal and extranodal disease comparedwith conventional morphologic stagingmethods (3,10). In addition, reliable as-sessment of bone marrow disease wasachieved. On the basis of these findings,we have translated these procedures intoclinical routine procedures, with FDGPET imaging as an adjunct to morpho-logic imaging with contrast-enhancedCT in staging and restaging of lym-phoma. Correspondingly, a combinationof both methods as a single-step exami-nation would simplify diagnostic imag-ing in patients with lymphoma.

In our study, the � statistic revealedexcellent agreement between PET/CTand contrast-enhanced CT for the delin-eation of lymph node involvement, inde-pendent of the purpose of the examina-tion for staging or restaging. Basically,pathologic uptake in lymph nodes of anysize was regarded as lymph node involve-ment in PET/CT images. With coregistra-tion of PET and CT images, correct local-ization of uptake is easily achievable eventhough only a low-dose nonenhancedCT scan was used for coregistration.These findings confirm previous resultsfrom CT studies, which did not revealany significant difference in the detec-

tion of mediastinal lymph nodes withnonenhanced and enhanced CT (13–15).

Several false-positive contrast-en-hanced CT findings were reported in re-gard to lymph node involvement at re-staging examinations. Particularly, theremaining soft-tissue densities on con-trast-enhanced CT images that were re-garded as persistent disease reflect the de-layed morphologic response to successfultherapy, and follow-up CT scans in thesepatients revealed normal results after atime. In these cases, PET/CT findingsconfirmed the absence of active tumortissue with delineation of soft-tissuemasses and without evidence of FDG up-take corresponding to scar tissue. At PET/CT, no false-positive results in regard tolymph node involvement were de-scribed. This is in contrast to findings inprevious reports with descriptions of arather high rate of false-positive findingsat PET, since inflammatory lesions maydemonstrate increased FDG uptake (16).However, the use of coregistered struc-tural information increases the specificityof lesion characterization (17).

In contrast, only a fair agreement wasfound for the assessment of organ in-volvement. The main cause for this dis-agreement could be found in the ratherlarge number of false-positive findings atcontrast-enhanced CT, when lung opaci-ties were interpreted as lymphoma. Infour patients, pulmonary involvementwas reported with contrast-enhanced CTbut was considered a false-positive find-ing, since spontaneous resolution ofthese opacities without therapy was seenat follow-up examinations. Fundamen-tally important for staging is the detec-tion of bone marrow involvement, whichwas described in three patients. In twopatients, this involvement was seen ex-clusively on PET/CT images and was con-firmed with additional imaging. In con-trast, at contrast-enhanced CT, boneinfiltration was incorrectly reported inone patient. This confirms the limita-tions of contrast-enhanced CT to identifylimited skeletal involvement (7).

On a per-patient basis, overall perfor-

TABLE 4Agreement between Methods

Involvement Staging Restaging Total

Lymph node 0.85 0.84 0.84Organ 0.38 0.55 0.50

Note.—Data are � values. The agreement with � values was as follows: 0–0.20, very poor;0.21–0.40, poor; 0.41–0.60, fair; 0.61–0.80, good; and 0.81–1.00, excellent.

828 � Radiology � September 2004 Schaefer et al

Ra

dio

logy

mance of both imaging modalities wasnot different in the assignment of dis-ease. PET/CT seems to perform better inexclusion of disease, which represents ahigher diagnostic confidence to excludeinitial disease or recurrence.

A limitation of PET is that small lesionscan be missed. In one patient, the thick-ening of a small-bowel loop indicatedlymphoma involvement on contrast-en-hanced CT images. This lesion was notvisible on FDG PET images. Correct inter-pretation of abdominal FDG uptake canbe difficult, because FDG uptake can bephysiologically increased in small andlarge bowel (18).

PET/CT had a greater effect on thera-peutic decision making in our patientswith Hodgkin disease and high-gradenon-Hodgkin lymphoma than did con-ventional contrast-enhanced CT. In ninepatients, PET/CT revealed relevant addi-tional lesions in lymph nodes and organsat staging and restaging examinations. Incontrast, only in two patients did con-trast-enhanced CT reveal additional le-sions, and this revelation led to a changeof clinical management. In one of thesetwo patients, a mediastinal mass wasidentified on a contrast-enhanced CTscan but was not reported with PET/CT,although retrospectively this mass wasclearly identified on the nonenhancedCT image at PET/CT. This case illustratesthe importance of careful interpretationof the nonenhanced CT scan indepen-dent from information of the PET scan.

An essential advantage of PET/CT im-aging over PET imaging alone is the abil-ity to use CT data for attenuation cor-rection of emission data in thereconstruction of PET images. This allowsreduction of total data acquisition timeto less than 25 minutes per patient (19).If necessary, after completion of scan-ning, contrast-enhanced CT can be per-formed in any desired location and canbe used for image fusion. However, ourdata suggest that PET/CT performed withlow-dose CT and without intravenouscontrast media is sufficient in staging andrestaging of disease in patients withHodgkin disease and high-grade non-Hodgkin lymphoma.

This study had several shortcomings.Mainly, the retrospective nature of thisstudy may have introduced a bias in thedata. The patient population was hetero-geneous and rather small. Furthermore,the technique of contrast-enhanced CTwas not standardized, since data acquisi-tion was performed at different institu-

tions. In addition, the time sequence ofPET/CT and contrast-enhanced CT dataacquisition was not randomized, andnonstandardized written reports as thebasis of analysis could have reduced theclinical importance of our results. Ourstandard of reference included clinicalpatient follow-up information obtainedat biopsy, from clinical and/or laboratoryreports, and at imaging with other tech-niques. Only a few patients had histo-logic verification of the specific suspectedpathologic findings. This is a known lim-itation from previous studies in whichthe performance of imaging methods wascompared in lymphoma patients. How-ever, the aim of this study was to deter-mine whether the two selected imagingmodalities were able to depict viable tu-mor tissue and to give correct informa-tion for staging and restaging, thus rep-resenting the setting in clinical routine.

Overall, the sensitivity and specificityfor lymph node and organ infiltration ofnon-Hodgkin lymphoma and Hodgkindisease analyzed on a per-patient basiswere better for PET/CT compared withcontrast-enhanced CT. The agreement ofboth methods was excellent in regard tothe status of disease of lymph nodes (� �0.84) but only fair for extranodal disease(� � 0.50). A significant difference wasfound in regard to exclusion of diseasewith PET/CT compared with contrast-en-hanced CT. Because of identified short-comings, future prospective studies areneeded for evaluation of the role ofPET/CT in the diagnostic work-up andtreatment of patients with Hodgkin dis-ease and high-grade non-Hodgkin lym-phoma.

References1. Newman JS, Francis IR, Kaminski MS,

Wahl RL. Imaging of lymphoma with PETwith 2-[F-18]-fluoro-2-deoxy-D-glucose:correlation with CT. Radiology 1994; 190:111–116.

2. Moog F, Bangerter M, Diederichs CG, etal. Extranodal malignant lymphoma: de-tection with FDG-PET versus CT. Radiol-ogy 1998; 206:475–481.

3. Stumpe KD, Urbinelli M, Steinert HC, etal. Whole-body positron emission tomog-raphy using fluorodeoxyglucose for stag-ing of lymphoma: effectiveness and com-parison with computed tomography. EurJ Nucl Med 1998; 25:721–728.

4. Jerusalem G, Beguin Y, Fassotte MF, et al.Whole-body positron emission tomogra-phy using 18F-fluorodeoxyglucose forposttreatment evaluation in Hodgkin’sdisease and non-Hodgkin’s lymphomahas higher diagnostic and prognosticvalue than classical computed tomogra-phy scan imaging. Blood 1999; 94:429–433.

5. Beyer T, Townsend DW, Brun T, et al. Acombined PET/CT scanner for clinical on-cology. J Nucl Med 2000; 41:1369–1379.

6. Dizendorf E, Hany TF, Buck A, von Schul-thess GK, Burger C. Cause and magnitudeof the error induced by oral CT contrastagent in CT-based attenuation correctionof PET emission studies. J Nucl Med 2003;44:732–738.

7. Carr R, Barrington SF, Madan B, et al.Detection of lymphoma in bone marrowby whole-body positron emission tomog-raphy. Blood 1998; 91:3340–3346.

8. Moog F, Bangerter M, Kotzerke J, et al.18-F-fluorodeoxyglucose-positron emis-sion tomography as a new approach todetect lymphomatous bone marrow.J Clin Oncol 1998; 16:603–609.

9. Moog F, Bangerter M, Diederichs CG, etal. Lymphoma: role of whole-body 2-de-oxy-2-[F-18]fluoro-D-glucose (FDG) PETin nodal staging. Radiology 1997; 203:795–800.

10. Jerusalem G, Beguin Y, Fassotte MF, et al.Whole-body positron emission tomogra-phy using 18F-fluorodeoxyglucose com-pared to standard procedures for stagingpatients with Hodgkin’s disease. Haema-tologica 2001; 86:266–273.

11. Jerusalem G, Beguin Y, Fassotte MF, et al.Persistent tumor 18F-FDG uptake after afew cycles of polychemotherapy is predic-tive of treatment failure in non-Hodgkin’s lymphoma. Haematologica2000; 85:613–618.

12. Jerusalem G, Warland V, Najjar F, et al.Whole-body 18F-FDG-PET for the evalu-ation of patients with Hodgkin’s diseaseand non-Hodgkin’s lymphoma. NuclMed Commun 1999; 20:13–20.

13. Haramati LB, Cartagena AM, Austin JH.CT evaluation of mediastinal lymphade-nopathy: noncontrast 5 mm vs postcon-trast 10 mm sections. J Comput AssistTomogr 1995; 19:375–378.

14. Cascade PN, Gross BH, Kazerooni EA, etal. Variability in the detection of enlargedmediastinal lymph nodes in staging lungcancer: a comparison of contrast-en-hanced and unenhanced CT. AJR Am JRoentgenol 1998; 170:927–931.

15. Patz EF Jr, Erasmus JJ, McAdams HP, et al.Lung cancer staging and management:comparison of contrast-enhanced andnonenhanced helical CT of the thorax.Radiology 1999; 212:56–60.

16. Strauss LG. Fluorine-18 deoxyglucose andfalse-positive results: a major problem inthe diagnostics of oncological patients.Eur J Nucl Med 1996; 23:1409–1415.

17. Hany TF, Steinert HC, Goerres GW, BuckA, von Schulthess GK. PET diagnosticaccuracy: improvement with in-linePET-CT system—initial results. Radiology2002; 225:575–581.

18. Engel H, Steinert H, Buck A, et al. Whole-body PET: physiological and artifactualfluorodeoxyglucose accumulations.J Nucl Med 1996; 37:441–446.

19. Burger C, Goerres G, Schoenes S, et al.PET attenuation coefficients from CT im-ages: experimental evaluation of thetransformation of CT into PET 511-keVattenuation coefficients. Eur J Nucl MedMol Imaging 2002; 29:922–927.

Volume 232 � Number 3 PET/CT or Contrast-enhanced CT for Lymphoma Staging � 829

Ra

dio

logy