FDG-PET for detection of pulmonary metastases from malignant primary bone tumors: Comparison with spiral CT

Annals of Oncology 12 479-486,2001 ) 2001 Khmer Academic Publishers Printed in the Netherlands

Original article FDG-PET for detection of pulmonary metastases from malignant primary bone tumors: Comparison with spiral CT C. Franzius,1 H. E. Daldrup-Link,2 J. Sciuk,1 E. J. Rummeny,2 S. Bielack,3 H. Jiirgens3 & O. Schober1 Departments of 'Nuclear Medicine, Clinical Radiology, " Pediatric Henuitology and Oncology. University Hospital, Miinster, German\

Summary Background The purpose was the comparison of positron emission tomography using F-18-fluorodeoxy-glucose (FDGPET) and spiral thoracic CT to detect pulmonary metastases from malignant primary osseous tumors Patients and methods In 71 patients with histologically confirmed malignant primary bone tumors (32 osteosarcomas, 39 Ewing's sarcomas) 111 FDG-PET examinations were evaluated with regard to pulmonary/pleural metastases in comparison with spiral thoracic CT Reference methods were the clinical follow-ups for 6-64 months (median 20 months) or a histopathologic analysis Results In 16 patients (23%) reference methods revealed a pulmonary/pleural metastatic disease FDG-PET had a sensitivity of 0.50, a specificity of 0 98, and an accuracy of 0 87 on a

Introduction Malignant primary bone tumors metastasize mainly by hematogenous spread. Staging procedures as presently applied identify 15%-25% of patients as clinically apparent metastatic at diagnosis [1]. Without systemic cytotoxic treatment, 80%-90% of patients die due to metastases. The improved outcome of patients with primary bone tumors in the past two decades has mainly been achieved by the use of adjuvant aggressive chemotherapy to control micrometastases [2, 3] Thus, survival rates have been increased by up to 55%-65% in a localized disease However, outcome is still poor in patients with clinically apparent metastases. In patients with Ewing's sarcomas, therapy studies for pulmonary metastatic disease include total lung irradiation or myeloablative chemotherapy concepts [4]. If the lungs are the only site of metastases in osteosarcoma patients, surgical resection added to standard therapy may be curative. Long-term survival is reported to be at 20%50% after complete resection of all macroscopically evident metastatic tissue. In contrast, when the radical surgical removal of pulmonary metastases is impossible, patients rarely survive longer than two years [5, 6]. Therefore, an accurate and early detection of pulmo-

patient based analysis Comparable values for spiral CT were 0 75, 1.00, and 0.94 It was shown that no patient who had a true positive FDG-PET had a false negative CTscan, nor was a pulmonary metastases detected earlier by FDG-PET than by spiral CT Conclusions There seems to be a superiority of spiral CT in the detection of pulmonary metastases from malignant primary bone tumors as compared with FDG-PET Therefore, at present a negative FDG-PET cannot be recommended to exclude lung metastases However, as specificity of FDG-PET is high, a positive FDG-PET result can be used to confirm abnormalities seen on thoracic CT scans as metastatic Key words Ewing's sarcoma, FDG-PET, malignant primary bone tumors, osteosarcoma. pulmonary metastases, thoracic CT

nary metastases is necessary to further improve outcome. The current diagnostic tool for the detection of pulmonary metastases is spiral CT More recently, positron emission tomography with F-18-fluorodeoxy-glucose (FDG-PET) has been used to improve the differentiation between benign and malignant focal pulmonary abnormalities [7] FDG-PET proved itself to be a good staging tool in a variety of malignant tumor entities. Malignant primary osseous tumors usually show an increased glucose metabolism and thus FDG-PET can be used for grading [8] and therapy monitoring of these tumors [9, 10] Additionally, FDG-PET is useful for detection of osseous metastases of Ewing's sarcoma [11] However, a systematic evaluation of the ability of FDG-PET to detect pulmonary metastases from primary bone tumors has not been reported.

Patients and methods Patient population In this retrospective analysis, we included all patients diagnosed with a malignant primary osseous tumor, who underwent one or more FDGPET between August 1995 and June 1999, and who had an additional

480 thoracic CT scan within four weeks before or after each FDG-PET examination One hundred thirteen consecutive FDG-PET examinations on seventy-two patients with a histologically proven diagnosis of a malignant primary bone tumor were evaluated One patient and a total of three examinations had to be excluded from analysis because of a still unknown reference state The remaining 71 patients, 26 female, 45 male, and of which 32 had osteosarcomas and 39 Ewing's sarcomas, ranged in age from 3 to 42 years (median 14 years) In 42 patients one FDG-PET examination was obtained Of the remaining 29 patients, 21 had 2 examinations, 6 had 3 examinations and 2 had 4 examinations Thirty FDG-PET scans were performed at the time of diagnosis (either of the first tumor manifestation or a recurrence of the disease) whereby no chemotherapy and/or radiotherapy was carried out for at least three months before FDG-PET Thirty-five FDG-PET examinations were performed within the first ten days of chemotherapy, and forty-five scans were obtained at a later stage during therapy (chemotherapy and/or radiotherapy) or just after the conclusion of therapy

Imaging techniques FDG-PET scans were acquired using an ECAT EXACT 921/47 (CTI/ Siemens, Knoxville, Tennessee), allowing simultaneous acquisition of 47 contiguous slices with a 3 375 mm slice thickness (one bed position. 15 86 cm axial field of view) The patients had been fasting for at least five hours before the F-I8-FDG injection No patient was known to have diabetes melhtus or a pathological glucose tolerance Blood glucose levels at the time of injection were less than 6 66 mmol/1 in all patients For muscle relaxation, all patients received 5 mg diazepam orally (dose adjusted for weight in children) one hour before intravenous application of 3 7 MBq/kg body weight F-I8-FDG After F-I8-FDG administration, the patients were hydrated intravenously with 500 ml normal saline solution (0 9% NaCl) over 30 minutes, followed by an intravenous injection of 20 mg furosemide (dose adjusted for weight in children) All patients were asked to urinate immediately before being scanned The first 75 FDG-PET examinations were done without attenuation correction Emission scanning in whole-body technique (5-12 bed positions, acquisition time 5 min/bed position) was started 60 minutes after F-18-FDG application These FDG-PET scans were reconstructed by filtered back projection using a Hanning filter with cut off at the Nyquist frequency The resulting in-plane resolution of transa\ial images was approximately 8 mm fullwidth-half-maximum (FWHM), with an axial resolution of approximately 5 mm FWHM In the remaining 35 FDG-PET examinations, whole-body emission/transmission scanning (6 mm emission, 4 mm transmission/bed position) was performed Images with attenuation correction were reconstructed iteratively (ECAT software version 7 I) Additionally, in all FDG-PET examinations done with transmission scanning emission images without attenuation correction were obtained by filtered back projection as described above CT examinations of the thorax were performed using a Tomoscan AV-scanner (Philips, Best, The Netherlands) with a helical mode and without intravenous contrast medium administration in most patients All CT scans were performed after inspiration with the patient in a supine position using the following parameters 120 kV, 100 mAs, table feed 8 mm, collimation 5 mm, and reconstruction index 4 mm The scan volume extended from the lung base to the thoracic inlet and was acquired during one single breath hold period In the case of suspected lymph node metastases, images were obtained after an intravenous injection of I 5 ml/kg bodyweight nonionic contrast agent with a flow of 2 ml/s and a delay of 45 s Images were reconstructed using the lung and the soft tissue algorithm All patients/parents gave their informed consent to all FDG-PET and CT examinations

Imaging analysis All images were evaluated visually by two experienced observers (consensus readings) describing the number and size of abnormal

thoracic FDG-accumulations and focal lesions on the CT scans respectively FDG-PET scans (emission scans and additionally attenuation corrected scans if available) and spiral thoracic CT scans were interpreted independently and blinded to the results of other imaging studies, and clinical and/or pathological data Only data on the primary tumor (histological type, primary tumor site), age and gender of the patients were available to the observers On FDG-PET scans, any intrathoracic focal uptake above pulmonary background level was considered to be a metastasis Activity ratios between the visually detectable lesions and homologous contralateral normal lung were calculated according to Lowe et al [12] using maximum count rate values within the regions of interest (ROIs) As whole-body FDG-PET was performed, images were also analyzed regarding osseous and softtissue metastases and primary/recurrent tumor sites Assessing the thoracic CT scans, calcified and uncalcified nodules were considered as being metastases in osteosarcoma patients, whereas in Ewing's sarcoma patients only uncalcified nodules were classified as metastatic Aside from the clinically applied threshold (1), another more sensitive threshold (II) was used additionally in all thoracic CT examinations Applying threshold I, any pulmonary/pleural nodule > 10 mm or more than one nodule > 5 mm and <10 mm were considered positive for metastatic disease Applying threshold II, additionally, any solitary nodule > 5 mm and < 10 mm or multiple nodules < 5 mm were categorized as positive CT was not assigned to metastatic disease in case the pattern was typical for another disease (l e , granulomatous inflammation) The relationship between the diagnostic ability of FDG-PET in the detection and the size of the lesions was evaluated. The lesions were classified into three size groups measured on the thoracic CT scan (largest lesion in each examination) > 10 mm, 5-9 mm, and < 5 mm

Reference methods The results of both studies were compared with the clinical follow-up or histopathological analyses By definition, metastatic disease was considered to be present if either confirmed by hislological examination of the resected tissue or confirmed by the obvious progression in number and/or size of the lesions on follow-up examinations A lesion was defined as being benign after a stable imaging follow-up for at least six months (6-64 months, median 20 months) or. in equivocal cases, by a negative biopsy proof In case of doubt, the reference stale was classified as unknown

Statistical analysis Diagnostic sensitivity, specificity, accuracy, and positive and negative predictive values were calculated on a patient based analysis (first FDG-PET and thoracic CT in each patient) and on an examination based analysis (all examinations) For analysis referring to patients, the 95% confidence interval (CI) for each parameter is given

Results Reference methods Reference methods revealed pulmonary metastatic disease in 16 out of 72 (23%) patients (nine proven by histology, seven proven by progression). In 55 patients reference methods were negative (three histological analyses). In one patient (referring to the first FDGPET and CT examination) the reference state was classified as still unknown On an examination based analysis, pulmonary/pleural metastatic disease was revealed in 24 out of 113 examina-

481 Table I Patient-based analysis (first examination in each patient) Number of patients

Sensitivity 95% CI

Specificity 95% CI

Accuracy 95% CI

Positive predictive value 95% CI

Negative predictive value 95%, CI

FDG-PET

71

0 50(8/16) 0 27-0 73

0 98(54/55) 0 90-1 00

0 87(62/71) 0 77-0 94

0 89(8/9) 0 56-0 99

0 87(54/62) 0 76-0 94

Thoracic CT Threshold I"

71

075(12/16) 0 50-0 90 1 00(16/16) 0 82-1 00

1 00(55/55) 0 94-1 00 0 76(42/55) 0 63-0 87

094(67/71) 0 86-0 98 082(58/71) 071-0 90

1 00(12/12) 0 77-1 00 0 55(16/29) 0 37-0 72

0 93(55/59) 0 3-0 98 1 00(42/42) 0 92-1 00

Threshold II b

a b

71

Treshold I any lesion > 10 mm or more than one lesion ^ 5 mm and < 10 mm were classified as positive Treshold II any lesion ^ 5 mm or more than one lesion < 5 mm were classified as positive

Table 2 FDG-PET examination-based analysis

All examinations Osteosarcoma Ewing's sarcoma Examinations without therapy Examinations at the beginning of therapy" Examinations during/after therapy Examinations with attenuation correction Examinations without attenuation correction

Number of examinations

Sensitivity

Specificity

Accuracy

Positive predictive value

Negative predictive value

110 49 61 30 35 45 35 75

0 54(13/24) 0 50(4/8) 0 56(9/16) 0 14(1/7) 0 71 (5/7) 0 70(7/10) 0 38(3/8) 0 63(10/16)

0 95(82/86) 1 00(41/41) 0 91 (41/45) 091(21/23) 1 00(28/28) 0 94(33/35) 0 89(24/27) 0 98(58/59)

0 86(95/110) 0 92(45/49) 0 82(45/61) 0 73(22/30) 0 94(33/35) 0 89(40/45) 0 77(27/35) 0 91 (68/75)

077(13/17) 1 00(4/4) 0 69(9/13) 0 33(1/3) 1 00(5/5) 0 78(7/9) 0 50(3/6) 091 (10/11)

0 88(82/93) 091 (41/45) 0 85(41/48) 078(21/27) 0 93(28/30) 0 92(33/36) 0 83 (24/29) 0 91 (58/64)

Within the first 10 days of chemotherapy

tions (13 histological analyses, 11 follow-ups) and excluded in 86 examinations (three histological analyses). The benign lesions in the three patients with histological exclusion of metastatic disease were the following an interstitial inflammatory nodule, three lesions of a septic spread showing partial resorption, multiple granulomatous nodules which revealed to be tuberculosis In three examinations the reference state had to be classified as unknown, one examination being on the patient mentioned above. Two follow-up examinations in another two patients were done after pulmonary metastases had already been removed In the following results, only the 71 patients and the 110 examinations with defined reference states were considered. FDG-PET Pulmonary findings Analyzing the results of the first FDG-PET examination each patient underwent, foci with increased pulmonary/ pleural tracer uptake were found in nine patients. By reference methods, eight of the examinations were classified as true positive, one as false positive. Sixtytwo patients demonstrated no pathological pulmonary/ pleural tracer accumulation in the first FDG-PET examination. Fifty-four were considered to be true negative,

eight false negative. Thus, FDG-PET had a sensitivity of 0.50, a specificity of 0 98, and an accuracy of 0 87 (Table 1) Analyzing all FDG-PET examinations, 13 were categorized as true positive (Figure 1), and 4 as false positive (Figure 2) Ninety-three examinations showed no pathological pulmonary/pleural tracer uptake. Eighty-two were considered to be true negative, eleven false negative (Figure 3). FDG-PET had a sensitivity of 0.54, a specificity of 0.95, and an accuracy of 0.86 (Table 2) on an examination based analysis. Analyzing only the 30 examinations without any chemotherapy, 1 examination was classified as true positive, 2 as false positive, 21 as true negative and 6 as false negative. This results in a sensitivity of 0 14, a specificity of 0.91, and an accuracy of 0 73 (Table 2). The results for different subgroups of patients and /or examinations - osteosarcoma vs Ewing's sarcoma, without therapy vs. during or just after therapy, without vs with attenuation correction - are summarized in Table 2. In examinations performed with emission scanning, we did not find any additional foci in the attenuation corrected images compared with the non attenuation corrected emission images. Twelve of sixteen examinations with lesions ^ 10 mm were correctly identified as positive with FDG-PET (sensitivity 0.75). Only one out of four examinations

483

(a)

(b)

Figure 3 A 42-yeur-old female patient with an osteosarcoma of the right pelvis The pulmonary metastasis seen on the thoracic CT (C) at time of diagnosis (arrow) was missed on the FDG-PET scan (A coronal, B transverse) Follow-up CT examinations 6 and 12 months later revealed an increase in size

with a maximum nodule size between 5 mm and 9 mm was correctly classified as positive FDG-PET was false negative in all examinations with metastases < 5 mm The minimum activity ratio of foci classified as positive was 1.5; the maximum ratio was 4.0 A differentiation between malignant and benign lesions was impossible using this semi-quantitative analysis Osseous and soft I issue findings With regard to other metastatic locations, there were 34 true positive osseous lesions in all examinations, of which 13 had not been detected by the conventional staging methods (CT/MRI of the primary tumor site,

thoracic CT, and bone scintigraphy) [11]. Additionally, three lesions in soft tissue revealed to be true positive, of which two were not detected before by conventional staging Primary tumor sites In 53 of the 65 examinations performed without any chemotherapy or during the first 10 days of chemotherapy, the primary/recurrent tumor site demonstrated a clear (50 examinations) or moderate (3 examinations) increased glucose metabolism In 11 examinations, the primary tumors had been resected before and therefore were negative on the FDG-PET scans However, in one examination the primary tumor which was located in the skull was not detectable on the FDG-PET scan. Analyzing the primary/recurrent tumor sites in the 45 examinations during or just after the end of chemotherapy, there was a reduced glucose metabolism compared with a pretherapeutic examination in 16 patients, who all were categorized as responders histologically. In 21 examinations, the primary/recurrent tumor sites did not show any increased glucose uptake (14 after resection, 7 after chemotherapy). In five examinations, the glucose uptake was identical or increased in comparison to an examination before chemotherapy. These patients revealed to be non-responders. In three examinations with a moderate increased glucose metabolism of the primary/recurrent tumor site during/after chemotherapy, there was no pretherapeutic FDG-PET and therefore no trend Two patients revealed to be responders, one to be a non-responder [9] Spiral thoracic CT On a patient based analysis (first CT examination) using threshold I, all 12 examinations with at least 1 lesion > 10 mm or more than 1 lesion ^ 5 mm and < 10 mm were classified as true positive Fifty-five out of fifty-nine negative examination were considered as true negative, four as false negative Applying threshold I, spiral thoracic CT had a sensitivity of 0 75, a specificity of 1.00, and an accuracy of 0.94 (Table 1). Using threshold II, in 29 examinations there were at least 1 lesion ^ 5 mm or more than 1 lesion < 5 mm Sixteen examinations were considered to be true positive and thirteen to be false positive. All 42 negative examinations were classified as true negative. Applying threshold II, spiral thoracic CT had a sensitivity of 1.00, a specificity of 0.76, and an accuracy of 0.82 on a patient based analysis (Table 1) Results on an examination based analysis and results for different subgroups of patients and/or examinations - osteosarcoma vs Ewing's sarcoma, without therapy vs. during or just after therapy, without vs. with attenuation correction in the FDG-PET - are summarized in Table 3. Both imaging modalities In one patient (Figure 2) FDG-PET was false positive and CT revealed the typical pattern of a granulomatous inflammation, which was proven histopathologically A

484 Table 3 Thoracic CT examination-based analysis

All examinations Threshold 1" Threshold II b Osteosarcoma Threshold 1" Threshold ll b Ewing's sarcoma Threshold la Threshold Il h Examinations without therapy Threshold 1" Threshold ll b Examinations at the beginning of therapy1' Threshold 1° Threshold ll b Examinations during/after therapy Threshold 1" Threshold ll b Examinations with attenuation correction in the FDG-PET Threshold 1° Threshold ll b Examinations without attenuation correction in the FDG-PET Threshold 1° Threshold Il b

Number of examinations

Sensitivity

Specificity

Accuracy

Positive predictive value

Negative predictive value

110 110

0 83(20/24) 1 00(24/24)

1 00(86/86) 0 79(68/86)

096(106/110) 0 84(92/110)

1 00(20/20) 0 57(24/42)

0 96(86/90) 1 00(68/68)

49 49

0 75(6/8) 1 00(8/8)

1 0 (41/41) 0 81 (33/41)

0 96(47/49) 0 84(41/49)

1 00(6/6) 050(8/16)

0 95(41/43) 1 00(33/33)

61 61

088(14/16) 1 00(16/26)

1 0 (45/45) 0 78(35/45)

0 97(59/61) 084(51/61)

1 00(14/14) 0 62(16/26)

0 96(45/47) 1 00(35/35)

30 30

0 57(4/7) 1 0 (7/7)

1 0 (23/23) 0 78(18/23)

0 9 (27/30) 0 83(25/30)

1 0 (4/4) 058(7/12)

0 88(23/26) 1 0 (18/18)

35 35

1 0 (7/7) 1 0 (7/7)

1 0 (28/28) 0 68(19/28)

1 0 (35/35) 0 74(26/35)

1 0 (7/7) 044(7/16)

1 0 (28/28) 1 0 (9/19)

45 45

0 9 (9/10) 1 0 (10/10)

1 0 (35/35) 0 89(31/35)

0 98(44/45) 0 91 (41/45)

1 0 (9/9) 071 (10/14)

0 97(35/36) 1 0 (31/31)

35 35

0 88(7/8) 1 0 (8/8)

1 0 (27/27) 0 82(22/27)

0 97(34/35) 0 86(30/35)

1 0 (7/7) 062(8/13)

0 96(27/28) 1 0 (22/22)

75 75

081 (13/16) 1 0 (16/16)

1 0 (59/59) 0 78(46/59)

0 96(72/75) 0 83(62/75)

1 0 (13/13) 0 55(16/29)

0 95(59/62) 1 0 (46/46)

" Threshold I any lesion > 10 mm or more than one lesion > 5 mm and < 10 mm were classified as positive b Threshold II any lesion > 5 mm or more than one lesion < 5 mm were classified as positive c Within the fist 10 days of chemotherapy

microbiological examination confirmed tuberculosis In the other three examinations with false positive FDGPET results there were no morphologic correlations in the thoracic CT, and the follow-ups (clinical and imaging) for more than six months (6, 11, and 15 months) did not reveal metastatic disease. In two patients, both FDG-PET and CT (using threshold I) detected pulmonary metastases for the first time in a follow-up examination. Additionally, CT (using threshold I) revealed pulmonary metastatic disease in a follow-up examination in three other patients with negative FDG-PET examinations. If threshold II would have been applied, two of these five patients could have be identified earlier (on the first CT examination). No patient had lung metastases detected earlier by FDGPET than by thoracic CT Also no patient in our series had totally normal CT results and a true positive FDGPET scan

Discussion The lungs are a common site of metastases in primary malignancies of the bone Early pulmonary involvement is usually asymptomatic. Long-term survival of patients with pulmonary metastatic disease has improved over the last decades due to aggressive surgical resection (in

osteosarcoma patients), bilateral lung irradiation (in Ewing's sarcoma patients), and effective multiagent systemic therapy [4, 5, 13]. A factor that is more difficult to assess is the impact of improved imaging techniques Since CT became a routine part of diagnosis and follow-up, the detection rate of pulmonary metastases at diagnosis has increased [14, 15] However, it is known from animal studies that there are still limitations in the detection of pulmonary metastases even using helical CT, because micrometastases could be missed [16]. Another limitation of CT is that small pulmonary nodules are often not specific for metastases and can also represent benign abnormalities (i.e., granuloma or atypical vessels presenting as a nodule) [16]. In this study, CT revealed pulmonary lesions ascribed to benign conditions in 13 patients (first examinations) if threshold II (any lesion ^ 5 mm or more than 1 lesion < 5 mm were classified as positive) was applied. This is reflected in a low specificity of 0.75 on a patient based analysis. Applying the clinically used threshold I (only lesions ^ 10 mm or more than 1 lesion ^ 5 mm and < 10 mm were classified as positive) all 13 examinations with small benign abnormalities were correctly categorized as negative and specificity rose to 1 00. However, in another four examinations small lesions, which would have been considered benign if classified according to threshold I, revealed to be metastatic disease and thus,

485 sensitivity decreased from 100 (threshold II) to 0.76 (threshold I) Detection of pulmonary/pleural metastases from primary osseous tumors by FDG-PET has been documented Schulte et al. [10] published four cases (in a total of 28 patients) with lung metastases from osteosarcoma that accumulated FDG. The aim of our study was to evaluate the clinical usefulness of FDG-PET in the detection of pulmonary metastases compared with CT The results show FDG-PET to be inferior to CT in the detection of metastases, and especially of small metastases Only one out of four examinations with metastatic lesions between 5 mm and 9 mm was correctly identified as positive with FDG-PETand none of the examinations with metastases < 5 mm. In our series, no patient had a completely normal CT scan and a true positive FDGPET. These results are comparable with results from a FDG-PET study in patients with soft-tissue sarcomas, showing that sensitivity for detection of pulmonary metastases of FDG-PET was lower in comparison to spiral thoracic CT [17]. Whereas spiral CT scan is performed during one single breath hold period, FDG-PET acquisition takes several minutes per bed position causing respiration blurredness. One of the main reasons for the limited sensitivity of FDG-PET is the size of pulmonary metastases from sarcomas. In our patient group, most of the metastatic lesions had a diameter below two times the full-width-half-maximum of the PET scanner ( 2 x 8 mm). Therefore true activity concentration in these lesions was obviously underestimated by FDG-PET (partial volume effect) [18]. However, it is known that even lesions smaller than the FWHM of a scanner are visible if the signal/ background ratio is high enough [19, 20]. There are studies describing cases where small pulmonary metastases from osteosarcomas missed on thoracic CT were detectable on Tc-99m-MDP bone scintigraphy (planar and/or SPECT) which has a much lower spatial resolution than PET and the same respiration blurredness [21, 22]. Therefore, another explanation for the limited sensitivity of FDG-PET seems to be a low tumor/background signal in pulmonary metastases from sarcomas. Almost all primary/recurrent tumor sites showed a clearly increased glucose uptake before or at the beginning of chemotherapy if no resection had been performed. There might be a different behavior of pulmonary metastases and the primary/recurrent tumor sites with regard to the expression of glucose transporters, as is demonstrated for primary lung cancer and liver metastases [23]. A further reason for the failure of FDG-PET may be the location of the metastases. A few lesions were located close to the myocardium with its physiologically high glucose metabolism which might have resulted in these lesions being missed on the FDG-PET scan In this analysis, attenuation correction did not result in an improvement of the detectability of pulmonary metastases. Nevertheless, further studies are necessary to analyze the role of attenuation correction in FDG-PET examinations performed in pediatnc and young adult

patients, as the benefit of attenuation correction in clinical oncological FDG-PET is still discussed controversially in literature [24-26]. Another important factor to analyze is the influence of treatment on the FDGPET results, as osteosarcomas and Ewing's sarcomas are highly aggressive tumors and neoadjuvant chemotherapy is usually initiated immediately after confirmation of diagnosis Therefore, many FDG-PET examinations have been performed shortly after initiation of chemotherapy. However, the subgroup analysis of examinations without any chemotherapy demonstrated the same superiority in sensitivity of spiral CT in comparison to FDGPET. Additionally, sensitivity of FDG-PET was not worse in patients who had the FDG-PET examination during or after chemotherapy than it was in patients who had not undergone any treatment Altogether, sensitivity of FDG-PET in the detection of pulmonary metastases of both, osteosarcomas and Ewing's sarcomas, is unsatisfactory and, thus, FDG-PET cannot be recommended as a single imaging tool to diagnose or exclude pulmonary metastatic disease. However, the false positive rate of FDG-PET is very low, although glucose metabolism is known to be increased in inflammatory processes. In our patient group, only one false positive FDG-PET result was caused by a granulomatous inflammation (tuberculosis) With a specificity of 0 98, FDG-PET can be used to hasten the way to a surgical resection of small CT abnormalities, if FDG-PET results are positive. Additionally, wholebody FDG-PET exposed further metastases outside the lungs which were not known before from the conventional staging. As another analysis demonstrated, FDGPET revealed to be superior to bone scintigraphy in the detection of osseous metastases from Ewing's sarcoma [11] Furthermore, FDG-PET is a noninvasive method to measure the metabolic activity of the primary/recurrent tumor site during/after chemotherapy and therefore can be used in the monitoring of treatment [9, 10].

Conclusion In summery, sensitivity of FDG-PET in the detection of pulmonary metastases from malignant primary bone tumors seems to be lower as compared to spiral thoracic CT. Therefore, at present FDG-PET cannot be recommended to exclude pulmonary metastatic disease from primary osseous tumors. However, as specificity of FDG-PET is high, a positive FDG-PET result can be used to confirm abnormalities seen on thoracic CT scans as metastatic.

Acknowledgements The authors gratefully acknowledge B. Frohhch (Department of Pediatnc Hematology and Oncology) for many helpful suggestions and their friendly cooperation and A. Heinecke (Department of Medical Computer Science

486 and Biomathematics) for his expert assistance with the statistical analysis.

References 1

Huvos AG Bone Tumors. Diagnosis, Treatment and Prognosis Montreal, Sydney, Tokyo W B Saunders 1991 2 Winkler K, Beron G, Delhng G et al Neoadjuvant chemotherapy of osteosarcoma Results of a randomized cooperative trial (COSS-82) with salvage chemotherapy based on histological tumor response J Clin Oncol 1988, 6 329-37 3 Jurgens H. Exner U. Gadner H et al Multidisciplinary treatment of primary Ewing's sarcoma of bone in children and young adults Six months of intensive combined modality therapy without maintenance J Clin Oncol 1988, 88 484-90 4 Paulussen M, Ahrens S, Burdach S et al Primary metastatic (stage IV) Ewing tumor Survival analysis of 171 patients from E1CESS studies European Intergroup Cooperative Ewing Sarcoma Studies Ann Oncol 1998.9 275-81 5 Bruland OS. Pihl A On the current management of osteosarcoma A critical evaluation and a proposal for a modified treatment strategy Eur J Cancer 1997. 33 1725-31 6 Meyers PA. Heller G, Healey JH et al Osteogemc sarcoma with clinically detectable metastasis at initial presentation J Clin Oncol 1993, II 449-53 7 Lowe VJ, Fletcher JW, Gobar Let al Prospective Investigation of Positron Emission Tomography in Lung Nodules J Clin Oncol 1998, 16 1075-84 8 Kole AC. Nieweg OE, Hoekstra HJ et al Fluonne-18-fluorodeoxyglucose assessment of glucose metabolism in bone tumors J Nucl Med 1998. 39 810-5 9 Franzius C, Sciuk J, Bnnkschmidt C et al Evaluation of chemotherapy response in primary bone tumors with F-I8-FDG-PET in comparison with histologically assessed tumor necrosis Chn Nucl Med 2000. 25 874-81 10 Schulte M. Brecht-Krauss D.Werner M et al Evaluation of neoadjuvant therapy response of osteogemc sarcoma using FDGPET J Nucl Med 1999. 40 1637-43 11 Franzius C. Sciuk J. Daldrup-Link HE et al FDG-PET for detection of osseous melastases from malignant primary bone tumors Comparison with bone scintigraphy Eur J Nucl Med 2000,27 1305-11 12 Lowe VL, Hoffman JM. DeLong DM et al Semiquantitative and visual analysis of FDG-PET images in pulmonary abnormalities J Nucl Med 1994.35 1771-6 13 Tabone MD, Kalifa C. Rodary Cet al Osteosarcoma recurrences in pediatric patients previously treated with intensive chemotherapy J Chn Oncol 1994, 12 2614-20 14 Manna NM, Pratt CB. Rao BN et al Improved prognosis of children with osleosarcoma melastatic to the lung(s) at the time of diagnosis Cancer 1992, 70 2722-7 15 Vanel D. Henry-Amar M. Lumbroso J et al Pulmonary evaluation

of patients with osteosarcoma Role of standard radiography, tomography, CT, scintigraphy, and tomoscintigraphy. Am J Roentgenol 1984, 143 519-23 16 Waters DJ, Coakley FV, Cohen MD et al The detection of pulmonary metastases by Helical CT A clinicopathologic study in dogs J Comput Assist Tomogr 1998, 22 235-40 17 Lucas JD, O'Doherty MJ, Wong JCH et al Evaluation of fluorodeoxyglucose positron emission tomography in the management of soft-tissue sarcoma J Bone Joint Surg(Br) 1998, 80-B 441-7 18 Hoffman EJ, Huang S-C, Phelps ME Quantitation in positron emission computed tomography 1 Effect of object size J Comp Assist Tomogr 1979, 3 229-308 19 Gupta NC, Maloof J, Gunel E Probability of malignancy in solitary pulmonary nodules using fluorine-18-FDG and PET J Nucl Med 1996, 37 943-8 20 Prauer HW, Weber WA, Romer Wet al Evaluation of pulmonary nodules by positron emission tomography using the glucose analogue fluorine-18-fluorodeoxyglucose (FDG), a controlled prospective study in 50 patients Br J Surg 1998, 85 1506-11 21 Brady AP, Ennis JT The scintigraphic detection of ossific mediastinal and pulmonary metastases in osteosarcoma Br J Radiol 1990,63 978-80 22 Pevarski DJ. Drane WE, Scarborough MT The usefulness of bone scintigraphy with SPECT images for detection of pulmonary metastases from osteosarcoma Am J Roentgenol 1998, 170 319-22 23 Kurata T, Ogun T, Isobe Tel al Differential expression of facilitative glucose transporter (GLUT) genes in primary lung cancers and their liver metastases Jpn J Cancer Res 1999, 90 1238-43 24 Bengel FM, Ziegler SI, Avnl N et al Whole-body positron emission tomography in clinical oncology Comparison between attenuation-corrected and uncorrected images Eur J Nucl Med 1997, 24 1091-8 25 Bedigian MP, Benard F, Smith RJ et al Whole-body positron emission tomography for oncology imaging using singles transmission scanning with segmentation and ordered subsets-expectation maximization (OS-EM) reconstruction Eur J Nucl Med 1998,25 659-61 26 Kotzerke J, Guhlmann A, Moog F et al Role of attenuation correction for fluonne-18 fluorodeoxyglucose positron emission tomography in the primary staging of malignant lymphoma Eur J Nucl Med 1999, 26 31-8 Received 14 July 2000, accepted 29 November 2000

Correspondence to C Franzius, MD Department of Nuclear Medicine Munster University Albert-Schweitzer-Str 33 48149 Munster Germany E-mail franziu@uni-muenster de