Well-differentiated gastro-entero-pancreatic endocrine

Transcription

1 ORIGINAL ARTICLE Endocrine Care Prediction and Diagnosis of Bone Metastases in Well-Differentiated Gastro-Entero-Pancreatic Endocrine Cancer: A Prospective Comparison of Whole Body Magnetic Resonance Imaging and Somatostatin Receptor Scintigraphy S. Leboulleux, C. Dromain, A. L. Vataire, D. Malka, A. Aupérin, J. Lumbroso, P. Duvillard, D. Elias, D. M. Hartl, T. De Baere, J. Guigay, M. Schlumberger, M. Ducreux, and E. Baudin Departments of Nuclear Medicine and Endocrine Tumors (S.L., J.L., M.S., E.B.), of Radiology (C.D., T.D.B.), of Epidemiology (A.L.V., A.A.), of Digestive Oncology (D.M., M.D.), of Pathology (P.D.), of Oncologic Surgery (D.E., D.M.H.), and of Medical Oncology (J.G.), Institut Gustave Roussy and University Paris Sud, Villejuif, France Purpose: Our purpose was to compare the sensitivity of whole body (WB) magnetic resonance imaging (MRI) and somatostatin receptor scintigraphy (SRS) for the diagnosis of bone metastases (BMs) in patients with well-differentiated gastro-entero-pancreatic endocrine cancer (WD-GEP-EC) and to determine predictive factors of BM. Patients and Methods: WB-MRI and SRS were prospectively performed in 79 patients with bronchial (11), thymic (five), gastric (two), duodeno-pancreatic (24), ileal (26), colic (one), or unknown primary (10) WD-GEP-EC. Results: A total of 36 patients (46%) had 333 BMs involving 119 skeletal segments. WB-MRI and SRS were equally sensitive for detecting patients with BM (86 vs. 81%; P 0.56), with 33% of the patients diagnosed with only one procedure. WB-MRI detected more BMs than SRS (80 vs. 57%; P 0.017). Compared with SRS, WB-MRI detected more spine BMs (96 vs. 45%; P 0.001) and tended to detect more pelvic and lower limb BMs (P and P 0.06, respectively). Compared with WB-MRI, SRS detected more skull BMs (100 vs. 0%; P 0.001) and tended to detect more rib BMs (P 0.08). Sternal and upper-limb BMs were equally detected with WB-MRI and SRS (P 0.32 and P 0.46, respectively). Bone staging with SRS and spine MRI rather than WB-MRI would have detected 92% of the patients with BMs and 83% of all BMs. The extent of liver involvement and bronchial-thymic primary tumors were independent predictive factors for BM. Conclusions: We recommend bone staging with SRS and spine MRI in all patients with bronchialthymic or unknown primary WD-GEP-EC. In case of duodeno-pancreatic or ileal primary, bone imaging may be restricted to patients with liver metastases. (J Clin Endocrinol Metab 93: , 2008) X/08/$15.00/0 Printed in U.S.A. Copyright 2008 by The Endocrine Society doi: /jc Received February 28, Accepted May 27, First Published Online June 3, 2008 Well-differentiated gastro-entero-pancreatic endocrine cancers (WD-GEP-ECs) originate from various organs but share common characteristics such as the expression of somatostatin receptors and a relatively long overall survival in a subset of patients (1). Standardization of imaging staging is warranted to understand the natural history of the disease, assess tumor prognosis and progression, prevent morbidity, and classify patients in homogenous groups for therapeutic trials. Somatostatin receptor scintigraphy (SRS) with indium-radiolabeled pentetreotide is a keystone in the staging of WD-GEP-EC Abbreviations: BM, Bone metastasis; CI, confidence interval; CT, computed tomography; EC, endocrine cancer; MRI, magnetic resonance imaging; PET, positron emission tomography; SRS, somatostatin receptor scintigraphy; WB, whole body; WD-GEP-EC, well-differentiated gastro-entero-pancreatic endocrine cancer. J Clin Endocrinol Metab, August 2008, 93(8): jcem.endojournals.org 3021

2 3022 Leboulleux et al. BMs in WD-GEP-EC J Clin Endocrinol Metab, August 2008, 93(8): (2). Diagnostic sensitivity of SRS depends on primary location, tumor size, hormonal secretions, and tumor differentiation (3 7). SRS is also critical for radiolabeled somatostatin analog therapy with treatment efficacy related to SRS results (8). The reported prevalence of bone metastases (BMs) in patients with WD-GEP-EC ranges from 4 13%, with treatment modifications subsequent to the diagnosis of BM in % of the patients (9 15). Previous preliminary studies suggested that BMs might be more frequent in patients with liver metastases and may depend on the location of the primary tumor (11, 12, 14, 16). For the detection of BM, SRS was found to be more sensitive than bone scintigraphy on a per-patient analysis in two studies and as sensitive as bone scintigraphy on a per-lesion analysis in one study (11, 12, 14). Only one study found bone scintigraphy to be more sensitive than SRS (15). The advantage of SRS over bone scintigraphy is its specificity, with a false-positive rate for bone scintigraphy reaching up to 90% (11). Furthermore, two studies, including eight and 11 patients, respectively, found spine magnetic resonance imaging (MRI) more sensitive than SRS for the diagnosis of spine BMs (11, 14). However, spine MRI has the disadvantage of being restricted to only part of the skeleton. Whole body (WB) MRI allows the exploration of the whole skeleton in a single imaging procedure, with a sensitivity ranging from % in other cancer types (17, 18). Its value has not been studied yet in patients with WD-GEP-EC. The objectives of our single-center prospective study were first, to compare the diagnostic performance of WB-MRI to SRS in the detection of BM in a large series of patients with WD-GEP-EC of various primaries, and second, to identify risk factors for BM to propose an imaging procedure strategy for bone staging. Patients and Methods Patients Patients with WD-GEP-EC referred to our institution between January 2005 and November 2006 were enrolled in this single-center prospective study. Inclusion criteria included a history of WD-GEP-EC confirmed by an experienced pathologist, a planned routine baseline workup, including SRS, thoracic computed tomography (CT), and abdominal CT, or liver MRI to which was added a WB-MRI. The presence of known BM before the initiation of the study was not criterion for exclusion. Our institutional review board approved the study, and all patients gave written informed consent. Imaging techniques WB-MRI was performed in the supine position on a 1.5-T unit (Sigma version 4.8; GE Medical Systems, Milwaukee, WI) with a body coil. The entire skeleton was covered using four contiguous coronal acquisitions with two different MRI sequences: short inversion recovery (inversion time of 150 msec, repetition time of msec depending on the number of sections, echo time of 50 msec); and spin echo T1 (repetition time of msec, echo time of msec). The slice center was started at the inferior border of the mandible, scanned with a slice thickness of 8 mm, an interslice of 2 mm, and a field of view of 48 cm. The matrix was No respiratory gating was performed. No iv contrast agent was administered. The total body MRI was performed within 20 min. SRS was performed after iv injection of MBq indium-111- DTPA-Phe1-octreotide (OctreoScan; Mallinckrodt Medical, Petten, The Netherlands). Digestive artifacts were reduced with an adequate colonic preparation (64 g macrogol 4000 in the evening after the injection and again the next morning before 24-h imaging). Acquisition was performed using both 111 In photopeaks (171 and 245 kev), using a large field of view doublehead -camera equipped with a medium-energy collimator (Axis; Philips Medical Systems, Best, The Netherlands). Four static anterior and posterior spot views covering the abdomen and pelvis were acquired at 4 h, and 16 static anterior and posterior spot views covering the WB were acquired at 24 h, and when needed, at 48 h ( word matrix, at least 10 min per view or 300,000 preset counts for the head and neck and 500,000 for the rest of the body). Abdominal single-photon emission computed tomography was performed at 24 h (64 projections, word matrix, 1 min per projection, and iterative reconstruction). When necessary, additional lateral views of the head and lateral views or single-photon emission computed tomography of the chest was performed. Image analysis Two experts (C.D. and S.L.) unaware of patients clinical, biological, or morphologicalfindingsindependentlyreviewedwb-mriandsrsfindings. Any WB-MRI with a focal or nodular or well-circumscribed low signal on T1 and high signal on short inversion recovery was defined as a BM (17, 18). Any extra-articular bone uptake on SRS was defined as a BM. The skeleton was divided into segments: skull, spine (divided into cervical, thoracic, and lumbar spine); pelvis; lower limbs; upper limbs, including arms, shoulders, scapulae, and claviculae; sternum; and ribs. When more than 10 BMs were depicted in a given segment, the number 11 was arbitrarily considered for statistical analysis. A patient was considered as having BMs when at least one bone lesion meeting the malignancy criteria was depicted on WB-MRI and/or SRS. In patients with discordant imaging findings, the diagnosis of bone involvement was confirmed by a second examination during follow-up and/or MRI with gadolinium injection focused on the skeletal segment involved. The total number of BMs was determined by summing the highest number of BMs of each skeleton segment detected by either WB-MRI and/or SRS. Statistics Sensitivity was defined as the number of BMs detected by one imaging procedure compared with the number of BMs detected by all imaging procedures. Sensitivities of WB-MRI and SRS, and their exact 95% confidence intervals (CIs) were estimated per patient, per skeleton segment, and per number of BMs. Sensitivities of WB-MRI and SRS were compared using the McNemar test for matched proportions. In the perskeleton segment and per number of BM analyses, sensitivities of WB- MRI and SRS were compared using the McNemar test extended to clustered data to account for within-patient correlation (19). Risk factors for BM were studied. Associations between BM and the following characteristics, dichotomized when necessary at the median value, were analyzed by univariate and multivariate logistic regression: sex; age ( 58 vs. 58 yr); primary tumor (bronchial-thymus, duodeno-pancreas, ileum, or unknown); time from diagnosis of WD-GEP-EC ( 54 vs. 54 months); number of visceral metastatic sites, including lymph nodes, and excluding BM, since initial diagnosis of WD-GEP-EC (zero or one vs. two vs. more than or equal to three); number of previous treatments (zero vs. one vs. two vs. more than or equal to three); and extent of liver involvement (0 vs. 25 vs. 25%). Variables associated with BM with a P value lower than 0.10 in the univariate analysis were included in the multivariate regression. All reported P values are two sided. Analyses were performed using SAS statistical software (SAS Institute Inc., Cary, NC). Significance level was 0.05, two sided. Results Patients Among the 169 patients with WD-GEP-EC referred to our institution between January 2005 and November 2006, who

3 J Clin Endocrinol Metab, August 2008, 93(8): jcem.endojournals.org 3023 underwent SRS, thoracic CT, and abdominal CT, or liver MRI, 79 patients (42 males, 37 females; median age: 57 yr; range: 30 79) were enrolled in the study. WB-MRI availability precluded other patients from inclusion. Patient clinical characteristics are given in the Table 1. Primary WD-GEP-EC sites were: foregut in 42 cases (53%) (bronchial, 11; thymic, five; gastric, two; and duodeno-pancreas, 24); midgut in 26 cases (33%) (ileum, 26); hindgut in one case (1%) (colon); and unknown in 10 cases (13%). The tumor was part of a multiple endocrine neoplasia type 1 in five cases (6%). The median interval between the diagnosis of WD-GEP-EC and the current study was 54 months (range ). A total of 75 patients (95%) harbored visceral metastases with one organ involved in 25 cases (32%), two organs in 23 (29%), and three or more in 27 cases (35%). BM had been diagnosed before the study in 12 patients (15%). TABLE 1. Patient characteristics Foregut Midgut Hindgut Bronchialthymus duodenumpancreas Gastric Ileum Colon Unknown primary Patients 16 (20) 24 (30) 2 (3) 26 (33) 1 (1) 10 (13) 79 No. of males/females 9/7 15/9 0/1 15/11 0/1 3/7 42/37 Median age (yr) range Median time from diagnosis (months) range Median No. of prior 3 (1 6) 2 (0 6) 3 (2 4) 3 (1 5) 3 2 (0 5) 3 (0 6) treatments (range) 0 (%) 0 1 (4) 0 0 (0) 0 1 (10) 2 (3) 1 (%) 2 (13) 6 (25) 0 4 (15) 0 4 (40) 16 (20) 2 (%) 4 (25) 6 (25) 1 6 (23) 0 2 (20) 19 (24) 3 (%) 10 (62) 11 (46) 1 16 (62) 1 3 (30) 42 (53) Surgery of the primary (65) Other surgery (38) Systemic chemotherapy (49) Liver chemo-embolization (30) Somatostatin analogs (49) Radionucleide therapy (6) Radiofrequency ablation (16) External beam radiation (20) Interferon (5) Hormonal hypersecretion 5HIAA 5 (31) 0 a 0 15 (58) 1 2 (22) b 22 (29) Cg A 11 (69) 12 (50) 0 14 (54) 0 5 (50) 42 (53) Other c 9 (56) 15 (63) 1 1 (4) (33) Visceral metastatic sites 0 2 (13) 2 (8) 0 0 (0) (5) 1 5 (31) 8 (33) 1 9 (35) 0 2 (20) 25 (32) 2 2 (13) 9 (38) 1 7 (27) 1 3 (30) 23 (29) 3 7 (44) 5 (21) 0 10 (39) 0 5 (50) 27 (34) Distant organs involved BMs 12 (75) 8 (33) 1 12 (46) 0 3 (30) 36 (46) Liver 12 (75) 22 (92) 2 20 (77) 1 7 (70) 64 (81) Mediastinal, subclavicular 7 (44) 4 (17) 0 7 (27) 0 3 (30) 21 (27) LN Lung 4 (25) 2 (8) 0 1 (4) 0 3 (3) 10 (13) Peritoneal carcinomatosis 0 (0) 2 (8) 1 11 (42) 0 2 (2) 16 (20) Abdominal LN 2 (13) 8 (33) 0 16 (62) 0 6 (6) 32 (41) Adrenals 3 (19) 4 (17) 0 0 (0) 0 0 (0) 7 (9) Ovaries 1 (6) 0 (0) 0 1 (4) 1 0 (0) 7 (9) Other 3 (19) 1 (4) 0 1 (4) 0 4 (40) 5 (6) Median time between WB-MRI and SRS (d) range Unless stated, values are number (%). CgA, Chromogranin A; LN, lymph node. a Five hydroxyindolacetic acid (5HIAA) was available in 22 of the 24 patients with duodeno-pancreatic WD-GEP-EC. b Five hydroxyindolacetic acid was available in nine of the 10 patients with unknown primary WD-GEP-EC. c Other secretions included gastrin in five cases, insulin in three, somatostatin in two, glucagon in one, vasoactive intestinal peptide in one, pancreatic polypeptide in two, glycoprotein hormone -subunit in three, neuron-specific enolase in three, calcitonin in nine, carcinoembryonic antigen in one, adrenocorticotropin hormone in two, GH in one, and serum parathormone-related protein in two cases. The total is above 26 because of multiple secretions in some patients. Total

4 3024 Leboulleux et al. BMs in WD-GEP-EC J Clin Endocrinol Metab, August 2008, 93(8): TABLE 2. Diagnosis of BMs per patient SRS normal SRS abnormal Total WB-MRI normal 43 (54) 5 (6) 48 (61) WB-MRI abnormal 7 (9) 24 (30) 31 (39) Total 50 (63) 29 (37) 79 (100) Values are expressed as number (%). Previous treatment modalities, hormonal patterns, and distant organ involvement are reported in Table 1. The time elapsed between the initial diagnosis of WD- GEP-EC and the present study and the number of visceral metastatic sites were similar in patients with either bronchial, thymic, duodeno-pancreatic, ileal, and unknown primaries (Kruskal-Wallis test, P 0.85, and P 0.39, respectively). Imaging results Per-patient analysis The median interval between WB-MRI and SRS was 1 d (range 0 63). BMs were found in a total of 36 patients (46%). WB-MRI showed BM in 39% (n 31) of the patients, and SRS showed BM in 37% (n 29). There were 24 (67%) patients FIG. 1. Spine metastases diagnosed with both WB-MRI and SRS (A C), with WB-MRI only (D F) (white arrow), and skull metastases diagnosed with SRS only (G I) (black arrow).

5 J Clin Endocrinol Metab, August 2008, 93(8): jcem.endojournals.org 3025 TABLE 3. Diagnosis of BMs per skeleton segment Skeletal segments involved (n) No. of BMs % WB-MRI only (no.) % BMs detected with SRS only (no.) % WB-MRI and SRS (no.) Skull (8) 13 0 (0) 100 (13) 0 (0) Cervical spine (7) 8 25 (2) 25 (2) 50 (4) Thoracic spine (22) (38) 2 (2) 51 (42) Lumbar spine (20) (45) 3 (2) 27 (17) Pelvis (24) (35) 8 (6) 47 (36) Lower limbs (12) (11) 9 (2) 41 (9) Upper limbs (14) (7) 48 (13) 26 (7) Sternum (4) 9 44 (4) 56 (5) 0 (0) Ribs (8) 31 0 (0) 65 (20) 35 (11) Total (119) (142) 20 (65) 38 (126) diagnosed with both WB-MRI and SRS, seven (19%) with WB- MRI only, and five (14%) with SRS only (Table 2 and Fig. 1). Therefore, WB-MRI and SRS were equivalent for the diagnosis of bone involvement, identifying 86% (95% CI 71 95) and 81% (95% CI 64 92) of the patients, respectively (P 0.56). Of note, SRS showed visceral metastases in 54 (68%) of the whole cohort of patients. SRS was performed during treatment with a long-acting somatostatin analog in 21 (27%) of the patients and was abnormal in all cases, showing BMs and/or visceral metastases. BMs were present in all patients with bronchial endocrine cancer (EC) (n 11), in 20% (n 1) of the patients with thymic EC, in one of the two patients with gastric EC, in 33% (n 8) of the patients with duodeno-pancreatic EC, in 46% (n 12) of the patients with ileal EC, and in 30% (n 3) of the patients with unknown primary. In the seven patients with BMs shown on WB-MRI but not on SRS, the diagnosis of BMs was confirmed in five patients either by a focused MRI with gadolinium injection performed 3 22 months after baseline WB-MRI (n 4), or a second WB-MRI performed 1 yr later (n 1), and is still pending in the remaining two patients. In the five patients with normal WB-MRI but BMs shown on SRS, the diagnosis of BMs was confirmed by the persistence of bone uptake on a SRS performed 8 12 months later in all five and by a focused MRI with gadolinium injection in two patients. Per-skeleton and per-lesion analysis Among the 119 skeletal segments involved, 58 (49%) were detected with both WB-MRI and SRS, 29 (25%) with WB-MRI, only and 32 (26%) with SRS only. Therefore, WB-MRI and SRS detected 73% (95% CI 64 80) and 76% (95% CI 67 83) (P 0.74) of skeletal segments involved, respectively. Regarding BM location, involved skeleton segments are reported in Table 3. BMs were limited to spine in six (17%) patients and limited to other skeleton segments in eight (22%) cases. In the remaining 22 (61%) patients, they were located in both spine and other skeleton segments. Among the 333 BMs, 126 (38%) were detected with both WB-MRI and SRS, 142 (43%) with WB-MRI only, and 65 (20%) with SRS only. Therefore, WB-MRI and SRS detected 80% (95% CI 76 85) and 57% (95% CI 52 63) of the BMs, respectively, (P 0.017). The number of BMs detected per skeleton segment is reported in Table 3. WB-MRI and SRS detected 0 and 100% of skull BMs, respectively (P 0.001), 96 and 45% of spinal BMs (P 0.001), 92 and 55% of pelvis BMs (P 0.054), 91 and 50% of lowerlimb BMs (P 0.06), 52 and 74% of upper-limb BMs (P 0.46), 44 and 56% of sternal BMs (P 0.32), and 35 and 100% of rib BMs (P 0.08) (Table 3). Among the 36 patients with BM, WB-MRI detected more BMs than SRS in 23 (64%) cases, fewer BMs than SRS in eight (22%) cases, and the same number of BMs as SRS in five cases (14%). Combination of imaging procedures A combination of SRS and WB-MRI limited to the spine (coronal spine MRI) would have detected 92% of the patients with BM, 91% of the skeletal segments involved, and 83% of all BMs, detecting fewer BMs (P 0.001) and fewer skeletal segments involved (P 0.01) but as many patients with BMs (P 0.24) than the combination of SRS and WB-MRI. The same combination (SRS and coronal spine MRI) would have detected more individual BMs, more skeletal segments involved, and more patients with BM than coronal spine MRI alone (P , , and 0.02, respectively) or SRS alone (P , 0.013, and , respectively) (Table 4). Finally, this combination (SRS and coronal spine MRI) would also have detected fewer BMs (P 0.02) and fewer skeletal segments in- TABLE 4. Combinations of imaging procedures for the diagnosis of patients with BMs, skeletal segments involved, and BMs WB-MRI SRS alone WB-MRI and SRS Coronal spine MRI Coronal spine MRI and SRS Coronal spine and pelvis MRI Coronal spine and pelvis MRI and SRS Patients (n 36) 86 (31) 81 (29) 100 (36) 72 (26) 92 (33) 81 (29) 94 (34) Skeletal segments involved (n 119) 73 (87) 76 (90) 100 (118) 37 (44) 91 (108) 53 (63) 95 (113) BMs (n 333) 81 (268) 57 (191) 100 (332) 45 (148) 83 (276) 66 (219) 93 (311) Values are expressed as percentage (no.).

6 3026 Leboulleux et al. BMs in WD-GEP-EC J Clin Endocrinol Metab, August 2008, 93(8): TABLE 5. Risk factors of BM Variable Prevalence of BM (%) Univariate analysis Odds ratio 95% CI P value Multivariate analysis a Odds ratio 95% CI P value Sex Male (n 42) 50 1 Female (n 37) Age (yr) 58 (n 38) (n 41) Primary a Bronchial-thymus (n 16) Duodeno-pancreas (n 24) Ileum (n 26) Unknown (n 10) Time from diagnosis (months) 54 (n 40) (n 39) No. of prior treatments 1 (n 18) (n 19) (n 42) No. of visceral metastatic sites 1 (n 29) (n 23) (n 27) Extent of liver involvement (%) 0(n 15) (n 43) (n 21) a Colonic and gastric primary were excluded from the analysis (analysis based on 76 patients). volved (P 0.03) but as many patients with BMs (P 1.0) than the combination of coronal spine and pelvis MRI. Predictive factors for BMs Multivariate analyses showed that the presence and extent of liver involvement and bronchial-thymic primaries were independently associated with a higher risk of BM (Table 5). Age, sex, time from diagnosis, and the number of prior treatments were not related to the presence of BM. All bronchial ECs harbored both BMs and liver metastases. Of note, all patients with bronchial primary harbored both BMs and liver metastases, and BMs were always present, even when the extent of liver involvement was less than 25%. However, patients with duodeno-pancreatic and ileal primaries never harbored BMs in the absence of liver metastases. Furthermore, BMs were present in less than 50% of the patients with liver involvement of less than 25% and in more than 50% of the patients with liver involvement exceeding 25% (Table 6). Clinical impact of BMs Symptoms related to BM were present in 16 (44%) patients and consisted of bone pain (n 14, 88%), fracture (n 1), and neurological symptoms (n 1). No hypercalcemia occurred. Specific treatment was required in 10 patients (external beam radiation therapy in seven, radiofrequency ablation in two, and surgery in one) because of bone pain in nine cases and a fracture in the last case. Discussion Tumor stage is the most significant prognostic parameter in patients with WD-GEP-EC. Imaging procedures are increasingly complex in this group of patients. We previously demonstrated that in patients with WD-GEP-EC, liver MRI with hepatic arterial phase is the standard imaging procedure for the explora- TABLE 6. Frequency of BM according to primary and extent of liver involvement Liver metastases Bronchial (n 11) Thymus (n 5) Duodeno-pancreas (n 24) Ileum (n 26) Unknown (n 10) Total Absent 0 (0/0) 52 (1/4) 0 (0/2) 0 (0/6) 33 (1/3) 13 (2/15) Present with liver involvement under 25% 100 (9/9) 0 (0/1) 15 (2/13) 46 (6/13) 20 (1/5) 44 (18/41) Present with liver involvement of 25% or more 100 (2/2) 0 (0/0) 67 (6/9) 86 (6/7) 50 (1/2) 75 (15/20) Values are expressed as percentage (no./total no.). Colonic and gastric primaries (n 3) were excluded from this table.

7 J Clin Endocrinol Metab, August 2008, 93(8): jcem.endojournals.org 3027 tion of liver metastases (5, 20). To our knowledge, this is the first study to assess WB-MRI in a large number of patients with WD- GEP-EC, and to compare it with SRS. Our study has several limitations. First, we did not use cytological or histological verification of every lesion as a standard. This was neither feasible nor ethical because of the number of BMs detected in each patient. In fact, the absence of systematic pathological proof of all lesions is a well-known limitation of this kind of study, precluding true sensitivity and specificity analysis (11, 21 23). Furthermore, the use of progression on a subsequent examination cannot be taken as a standard in WD- GEP-EC because they are not, in most cases, fast growing tumors. Therefore, we compared WB-MRI to SRS that has the advantage over bone scintigraphy to be more specific (11, 24). Therefore, the methodology applied is not ideal but the best that can be achieved. Second, due to the availability of WB-MRI (with one procedure per week available), patients were not consecutive. Finally, 27% of the whole cohort was imaged while being on a long-acting form of somatostatin analogs. This may have decreased the sensitivity of SRS in our study due to a downregulation of SRS receptors (25). However, the exclusion of these patients from the analysis did not increase SRS sensitivity. Furthermore, long-acting somatostatin analogs are now widely used in WD-GEP-EC, and their systematic withdrawal several months before every SRS may be regarded as a limitation in the field of WD-GEP-EC imaging. It is, in fact, generally admitted that SRS can be performed during treatment with long-acting forms of somatostatin analogs, even if it might interfere with SRS results (26). Of note, all SRSs performed under a long-acting form of somatostatin analogs in this study were abnormal, disclosing either BM or visceral metastasis. Our study was not designed to determine the rate of bone involvement. However, we found a higher incidence of bone involvement (46%) than previously reported, suggesting that bone involvement is often underestimated in patients with WD- GEP-EC (9 14). Even after exclusion of patients with known BM at inclusion, the observed 36% rate of bone involvement is still well above the 4 13% previously reported (9 14). This may be due to a selection bias, owing to the study s setting in a tertiary care center with patients with more advanced disease. However, our results are concordant with recent studies using fluorodihydroxyphenylalanine positron emission tomography (PET) or 68 Gal-DOTA-Tyr 3 -octréotide PET (21, 22, 27 29). Symptoms due to BM were present in 44% of the patients with BM, in accordance with previous studies (11, 12). Severe bone events (e.g. fractures, neurological compression, or hypercalcemia) were uncommon (6%), but the need for specific bone treatment was high (10 of 36 patients), suggesting the need for a systematic bone workup. We showed that in the patients under study, WB-MRI and SRS were equivalent for the diagnosis of BM in each patient, with 86 and 81% of the patients being diagnosed, respectively. WB- MRI detected a higher number of BMs than SRS (80 vs. 57%, respectively). However, WB-MRI and SRS were complementary imaging modalities because 43% of the BMs were detected with WB-MRI only and 20% with SRS only. This result can be explained by the variation in the performance of each imaging procedure according to the skeletal segment in question. For example, spine BMs that represented 46% of all BMs and affected 80% of the patients were almost all detected with WB- MRI (96%), whereas SRS only detected 45% of them. On the other hand, skull BMs that affected approximately 10% of the patients were only detected with SRS. This result confirms the lack of sensitivity of WB-MRI for the diagnosis of skull, due to a limited field of view and coronal acquisition, as previously shown in other types of cancer (17, 18, 30). In the same manner, WB-MRI only detected 38% of the rib metastases that were all, however, detected on SRS. Due to the low number of patients with rib metastases, the difference was not statistically significant, and a tendency was noted (P 0.08). We demonstrated for a large series of patients that the presence and extent of liver involvement and bronchial-thymus primary were two independent risk factors for BM. Indeed, BMs were found in 13% of patients with no liver involvement, 42% with limited involvement ( 25%), and 76% of patients with extensive liver involvement ( 25%). BMs were found in all patients with bronchial EC, but in only 30 46% of patients with duodeno-pancreatic, ileal, or unknown primary. A limitation of our study is the absence of patients with bronchial primary without liver metastases. However, our data not only confirmed preliminary studies but also showed for the first time that liver involvement and foregut primary were two independent factors (11, 12, 16). In our study, midgut primaries were not a risk factor for BM, contrary to what was suggested by a previous study on a small number of patients with BM that did not explore liver involvement (14). Based on these results, we recommend routine bone staging with SRS and MRI in all patients with WD-GEP from a bronchial, thymic, or unknown primary EC, whereas in case of duodeno-pancreatic or ileal primary bone staging may be limited to the patients with liver metastases. Because BMs were rarely clinically severe, because WB-MRI availability may be limited, and because the combination of SRS and coronal spine MRI allowed the diagnosis of 92% of the patients with BM and 83% of the BMs, bone staging in routine practice may rely on spine MRI plus SRS. Furthermore, spine MRI has the advantages of including sagittal image acquisition for an easier visualization of BM and contrast agent injection for a better analysis of the BM. There is no doubt that, when more widely available, new PET-dedicated SRS such as 68 gallium radiolabeled somatostatin analogs will replace indium-radiolabeled SRS as first-line imaging (22, 27, 29). However, indium-radiolabeled pentetreotide is, at the moment, the only SRS approved by European and American agencies for staging of WD-GEP-EC. Other tracers, including 18 fluorodihydroxyphenylalanine and 11 C-5-hydroxy-tryptophan PET, may be considered for second-line imaging, in subgroups of patients with carcinoid tumors, or islet cell tumors (21, 23, 28). Fluorodeoxyglucose PET is not routinely recommended in patients with WD-GEP-ECs, which are often slow-growing tumors and rarely show 2-fluoro-2-deoxy-D-glucose uptake (6). However, there are few studies comparing scintigraphic methods to standardized conventional imaging procedures (5, 23). Whether these imaging modalities can replace MRI for bone staging will have to be determined.

8 3028 Leboulleux et al. BMs in WD-GEP-EC J Clin Endocrinol Metab, August 2008, 93(8): In conclusion, BMs are likely to be underestimated in patients with WD-GEP-EC. WB-MRI and SRS with indium-111-dtpa- Phe1-octreotide were complementary for diagnosing BM, their respective performances varying according to the skeletal segment involved. The extent of liver metastases and bronchial primary were the only two independent risk factors for BM. Based on these results, we recommend systematic bone staging, including SRS and spine MRI, in all patients with WD-GEP with bronchial, thymic, and unknown primary EC, whereas bone staging may be limited to patients with liver metastases in case of duodeno-pancreatic or ileal primary. Acknowledgments Address all correspondence and requests for reprints to: Sophie Leboulleux, M.D., Department of Nuclear Medicine and Endocrine Oncology, Institut Gustave Roussy, 39 Rue Camille Desmoulins, Villejuif Cedex, France. leboulleux@igr.fr. Disclosure Statement: The authors have no conflicts of interest to declare. References 1. Baudin E 2007 Gastroenteropancreatic endocrine tumors: clinical characterization before therapy. Nat Clin Pract Endocrinol Metab 3: Krenning EP, Kwekkeboom DJ, Oei HY, de Jong RJ, Dop FJ, Reubi JC, Lamberts SW 1994 Somatostatin-receptor scintigraphy in gastroenteropancreatic tumors. An overview of European results. Ann NY Acad Sci 733: Cadiot G, Lebtahi R, Sarda L, Bonnaud G, Marmuse JP, Vissuzaine C, Ruszniewski P, Le Guludec D, Mignon M 1996 Preoperative detection of duodenal gastrinomas and peripancreatic lymph nodes by somatostatin receptor scintigraphy. Groupe D etude Du Syndrome De Zollinger-Ellison. Gastroenterology 111: Gibril F, Reynolds JC, Doppman JL, Chen CC, Venzon DJ, Termanini B, Weber HC, Stewart CA, Jensen RT 1996 Somatostatin receptor scintigraphy: its sensitivity compared with that of other imaging methods in detecting primary and metastatic gastrinomas. A prospective study. Ann Intern Med 125: Dromain C, de Baere T, Lumbroso J, Caillet H, Laplanche A, Boige V, Ducreux M, Duvillard P, Elias D, Schlumberger M, Sigal R, Baudin E 2005 Detection of liver metastases from endocrine tumors: a prospective comparison of somatostatin receptor scintigraphy, computed tomography, and magnetic resonance imaging. J Clin Oncol 23: Reisinger I, Bohuslavitzki KH, Brenner W, Braune S, Dittrich I, Geide A, Kettner B, Otto HJ, Schmidt S, Munz DL 1998 Somatostatin receptor scintigraphy in small-cell lung cancer: results of a multicenter study. J Nucl Med 39: Tzannou IA, Karapanagiotou EM, Charpidou A, Dilana K, Alamara C, Christakopoulou I, Georgiou E, Syrigos KN 2007 The use of radiolabeled somatostatin analog scintigraphy in the staging of small cell lung cancer patients. Am J Clin Oncol 30: Kwekkeboom DJ, Teunissen JJ, Bakker WH, Kooij PP, de Herder WW, Feelders RA, van Eijck CH, Esser JP, Kam BL, Krenning EP 2005 Radiolabeled somatostatin analog [177Lu-DOTA0,Tyr3]octreotate in patients with endocrine gastroenteropancreatic tumors. J Clin Oncol 23: Nocaudie-Calzada M, Huglo D, Carnaille B, Proye C, Marchandise X 1996 Comparison of somatostatin analogue and metaiodobenzylguanidine scintigraphy for the detection of carcinoid tumours. Eur J Nucl Med 23: Frilling A, Malago M, Martin H, Broelsch CE 1998 Use of somatostatin receptor scintigraphy to image extrahepatic metastases of neuroendocrine tumors. Surgery 124: Gibril F, Doppman JL, Reynolds JC, Chen CC, Sutliff VE, Yu F, Serrano J, Venzon DJ, Jensen RT 1998 Bone metastases in patients with gastrinomas: a prospective study of bone scanning, somatostatin receptor scanning, and magnetic resonance image in their detection, frequency, location, and effect of their detection on management. J Clin Oncol 16: Lebtahi R, Cadiot G, Delahaye N, Genin R, Daou D, Peker MC, Chosidow D, Faraggi M, Mignon M, Le Guludec D 1999 Detection of bone metastases in patients with endocrine gastroenteropancreatic tumors: bone scintigraphy compared with somatostatin receptor scintigraphy. J Nucl Med 40: Raderer M, Kurtaran A, Leimer M, Angelberger P, Niederle B, Vierhapper H, Vorbeck F, Hejna MH, Scheithauer W, Pidlich J, Virgolini I 2000 Value of peptide receptor scintigraphy using (123)I-vasoactive intestinal peptide and (111)In-DTPA-D-Phe1-octreotide in 194 carcinoid patients: Vienna University Experience, 1993 to J Clin Oncol 18: Meijer WG, van der Veer E, Jager PL, van der Jagt EJ, Piers BA, Kema IP, de Vries EG, Willemse PH 2003 Bone metastases in carcinoid tumors: clinical features, imaging characteristics, and markers of bone metabolism. J Nucl Med 44: Zuetenhorst JM, Hoefnageli CA, Boot H, Valdes Olmos RA, Taal BG 2002 Evaluation of (111)In-pentetreotide, (131)I-MIBG and bone scintigraphy in the detection and clinical management of bone metastases in carcinoid disease. Nucl Med Commun 23: Janson ET, Holmberg L, Stridsberg M, Eriksson B, Theodorsson E, Wilander E, Oberg K 1997 Carcinoid tumors: analysis of prognostic factors and survival in 301 patients from a referral center. Ann Oncol 8: Steinborn MM, Heuck AF, Tiling R, Bruegel M, Gauger L, Reiser MF 1999 Whole-body bone marrow MRI in patients with metastatic disease to the skeletal system. J Comput Assist Tomogr 23: Lauenstein TC, Goehde SC, Herborn CU, Goyen M, Oberhoff C, Debatin JF, Ruehm SG, Barkhausen J 2004 Whole-body MR imaging: evaluation of patients for metastases. Radiology 233: Obuchowski NA 1998 On the comparison of correlated proportions for clustered data. Stat Med 17: Dromain C, de Baere T, Baudin E, Galline J, Ducreux M, Boige V, Duvillard P, Laplanche A, Caillet H, Lasser P, Schlumberger M, Sigal R 2003 MR imaging of hepatic metastases caused by neuroendocrine tumors: comparing four techniques. AJR Am J Roentgenol 180: Koopmans KP, de Vries EG, Kema IP, Elsinga PH, Neels OC, Sluiter WJ, van der Horst-Schrivers AN, Jager PL 2006 Staging of carcinoid tumours with 18F- DOPA PET: a prospective, diagnostic accuracy study. Lancet Oncol 7: Gabriel M, Decristoforo C, Kendler D, Dobrozemsky G, Heute D, Uprimny C, Kovacs P, Von Guggenberg E, Bale R, Virgolini IJ Ga-DOTA-Tyr3- octreotide PET in neuroendocrine tumors: comparison with somatostatin receptor scintigraphy and CT. J Nucl Med 48: Koopmans KP, Neels OC, Kema IP, Elsinga PH, Sluiter WJ, Vanghillewe K, Brouwers AH, Jager PL, de Vries EG 2008 Improved staging of patients with carcinoid and islet cell tumors with 18F-dihydroxy-phenyl-alanine and 11C-5- hydroxy-tryptophan positron emission tomography. J Clin Oncol 26: Gibril F, Reynolds JC, Chen CC, Yu F, Goebel SU, Serrano J, Doppman JL, Jensen RT 1999 Specificity of somatostatin receptor scintigraphy: a prospective study and effects of false-positive localizations on management in patients with gastrinomas. J Nucl Med 40: Hofland LJ, van Koetsveld PM, Waaijers M, Zuyderwijk J, Breeman WA, Lamberts SW 1995 Internalization of the radioiodinated somatostatin analog [125I-Tyr3]octreotide by mouse and human pituitary tumor cells: increase by unlabeled octreotide. Endocrinology 136: Oberg K, Eriksson B 2005 Nuclear medicine in the detection, staging and treatment of gastrointestinal carcinoid tumours. Best Pract Res Clin Endocrinol Metab 19: Becherer A, Szabo M, Karanikas G, Wunderbaldinger P, Angelberger P, Raderer M, Kurtaran A, Dudczak R, Kletter K 2004 Imaging of advanced neuroendocrine tumors with (18)F-FDOPA PET. J Nucl Med 45: Montravers F, Grahek D, Kerrou K, Ruszniewski P, de Beco V, Aide N, Gutman F, Grange JD, Lotz JP, Talbot JN 2006 Can fluorodihydroxyphenylalanine PET replace somatostatin receptor scintigraphy in patients with digestive endocrine tumors? J Nucl Med 47: Buchmann I, Henze M, Engelbrecht S, Eisenhut M, Runz A, Schafer M, Schilling T, Haufe S, Herrmann T, Haberkorn U 2007 Comparison of (68)Ga- DOTATOC PET and (111)In-DTPAOC (Octreoscan) SPECT in patients with neuroendocrine tumours. Eur J Nucl Med Mol Imaging 34: Giraudet AL, Vanel D, Leboulleux S, Auperin A, Dromain C, Chami L, Tovo NN, Lumbroso J, Lassau N, Bonniaud G, Hartl D, Travagli JP, Baudin E, Schlumberger M 2007 Imaging medullary thyroid carcinoma with persistent elevated calcitonin levels. J Clin Endocrinol Metab 92: