MR Imaging Features of Small Solid Pseudopapillary Tumors: Retrospective Differentiation From Other Small Solid Pancreatic Tumors

Transcription

1 Gastrointestinal Imaging Original Research Gastrointestinal Imaging Original Research Mi Hye Yu 1 Jae Young Lee 1 Min Kim 2 Se Hyung Kim 1 Jeong Min Lee 1 Joon Koo Han 1 yung-ihn Choi 1 Yu MH, Lee JY, Kim M, et al. Keywords: MRI, neoplasms, pancreas, solid pseudopapillary tumor DOI: /JR Received February 15, 2010; accepted after revision May 17, Department of Radiology and the Institute of Radiation Medicine, Seoul National University Hospital, 101 Daehangno, Chongno-gu, Seoul, , Republic of Korea. ddress correspondence to J. Y. Lee (leejy@radiol.snu.ac.kr). 2 Department of Pathology, Seoul National University Hospital, Seoul, Republic of Korea. JR 2010; 195: X/10/ merican Roentgen Ray Society MR Imaging Features of Small Solid Pseudopapillary Tumors: Retrospective Differentiation From Other Small Solid Pancreatic Tumors OJECTIVE. The purpose of this study was to describe the MRI features of small solid pseudopapillary tumors useful for differentiating these tumors from other small solid pancreatic tumors. MTERILS ND METHODS. Forty-five patients (18 men, 27 women; mean age, 59 ± 11.7 years) with 45 solid pancreatic tumors smaller than 3 cm in diameter registered from 2000 to 2009 were included. Twenty-two of the tumors were adenocarcinomas, 12 were endocrine tumors, and 11 were solid pseudopapillary tumors. Two radiologists analyzed the MR images for morphologic features, signal intensity of the lesion on unenhanced images, and dynamic enhancement pattern. The contrast-to-noise ratio between the lesion and the pancreas was calculated on T1- and T2-weighted images and on dynamic images. RESULTS. Solid pseudopapillary tumors commonly had the following features: completely well-defined margin (82%), pure solid consistency (82%), low signal intensity on unenhanced T1-weighted images (100%), high signal intensity on T2-weighted images (100%), a strong predominance among women (91%), and early heterogeneous and slowly progressive enhancement (100%). The solid pseudopapillary tumors rarely had a capsule or hemorrhage. characteristic qualitative feature of solid pseudopapillary tumors that was significantly different from the other two tumor types was very high signal intensity on T2-weighted images (p < 0.01). Quantitative analysis showed that the solid pseudopapillary tumors had a lower contrast-to-noise ratio ( ± 5.84) on T1-weighted images and a higher contrast-to-noise ratio (37.35 ± 33.59) on T2-weighted images than did the other two tumor types (p < 0.01). CONCLUSION. Small solid pseudopapillary tumors were predominantly seen as completely well-demarcated pure solid tumors in middle-aged women. They had lower signal intensity on T1-weighted images and higher signal intensity on T2-weighted images than did the other two tumor types and had early heterogeneous and progressive enhancement. S olid pseudopapillary tumors are very uncommon pancreatic neoplasms that predominantly affect young women [1 4]. The lesions usually are large and encapsulated, frequently containing varying amounts of necrosis, hemorrhage, and cystic changes, all of which are known to be the characteristic findings of solid pseudopapillary tumors that allow differentiation from other solid pancreatic tumors on cross-sectional images [5 7]. With the widespread use of cross-sectional imaging techniques for a variety of indications, small pancreatic tumors are encountered more routinely [8, 9]. ecause they are rarely aggressive and have low-grade malignant potential with an excellent prognosis after complete resection [1 3, 10, 11], solid pseudopapillary tumors should be differentiated from other, more aggressive tumors such as adenocarcinoma and endocrine tumors. Follow-up MRI is commonly used for differentiation of small pancreatic tumors that have been identified with ultrasound or CT because MRI has high-contrast resolution between focal pancreatic lesions and pancreatic parenchyma and better depicts fat, hemorrhage, cystic changes, and pancreatic ductal dilatation. To the best of our knowledge, no report has described the MRI features of small solid pseudopapillary tumors, the cross-sectional imaging features of which are different from those of larger solid pseudopapillary tumors. Furthermore, a direct comparison of the MRI features of small solid pseudopapillary tumors and those of other solid pancreatic tumors has not been reported, to our 1324 JR:195, December 2010

2 knowledge. We performed our study to discern the MRI features of small solid pseudopapillary tumors and to investigate the differential features of these tumors from other common small solid pancreatic tumors. Materials and Methods Institutional review board approval was obtained, and informed patient consent was waived because of the retrospective nature of this study. Subjects We retrospectively reviewed the surgical and pathologic records from January 2000 to July 2009 at our institution to identify pathologically proven small solid pancreatic tumors 3 cm or smaller in longest diameter in patients who underwent unenhanced and dynamic gadolinium-enhanced MRI. We excluded lymphoma and metastasis (one case each) because of their rarity. ecause small adenocarcinomas were much more common than small endocrine tumors and small solid pseudopapillary tumors, the inclusion period for adenocarcinoma ended in September 2007 rather than July Forty-five consecutively registered patients (18 men, 27 women; mean age, 59 ± 11.7 [SD] years) with pathologically proven small solid pancreatic tumors (22 adenocarcinomas, 12 endocrine tumors, 11 solid pseudopapillary tumors) were included in the study. Six of the 22 adenocarcinomas, eight of the 12 endocrine tumors, and seven of the 11 solid pseudopapillary tumors were asymptomatic and detected incidentally on CT scans obtained for various reasons. In the patients with ductal adenocarcinoma (11 men, 11 women; mean age, 66 ± 6.8 years; range, years), the mean size of the ductal adenocarcinomas was 2.15 ± 0.46 cm (range, cm). In the patients with endocrine tumors (six men, six women; mean age, 56 ± 12.0 years; range, years), the mean size of the tumors was 1.57 ± 0.53 cm (range, cm). In the patients with small solid pseudopapillary tumors (10 women, one man; mean age, 48 ± 9.9 years; range, years), the mean size of the tumors was 2.19 ± 0.65 cm (range, cm). ll 45 patients underwent surgery for curative treatment. Pathologic examination showed that 10 of the 11 solid pseudopapillary tumors were benign and one was solid pseudopapillary carcinoma with splenic invasion. ll endocrine tumors were well differentiated, and only one endocrine tumor functioned as an insulinoma. MRI Technique MRI was performed with one of three 1.5-T systems (Sonata, Siemens Healthcare; Signa Excite, GE Healthcare; Signa HDx, GE Healthcare) or a 3-T system (Signa Excite, GE Healthcare). The MRI examination consisted of unenhanced fat-saturated T1-weighted 3D gradient-recalled echo images (TR/TE, / ; slice thickness, mm), in-phase and out-of phase T1- weighted 2D or 3D gradient-recalled echo images (3D, / ; 2D, / ; slice thickness, 5 7 mm), fat-saturated T2-weighted fast spin-echo images (2,000 10,000/ ; slice thickness, 6 7 mm), and ultrafast T2-weighted fast spin-echo (HSTE) images ( ,402.5/57 102; slice thickness, 6 7 mm) in the axial, coronal, or both planes in all patients. Dynamic fat-saturated T1-weighted 3D gradientrecalled echo imaging ( / ; slice thickness, mm) was performed after administration of 0.1 mmol/kg of gadobenate dimeglumine (Multi- Hance, racco) per kilogram of body weight at an injection rate of 2 ml/s. rterial, portal, and delayed phase images were obtained serially 30 seconds, 1 minute, and 3 minutes after contrast injection. Fig. 1 Diagram shows dynamic enhancement patterns of pancreatic tumors. White denotes enhancing area; black, nonenhancing area. Persistent homogeneous enhancement Early heterogeneous and progressive enhancement Early heterogeneous and no progressive enhancement No enhancement Image nalysis Qualitative analysis Two experienced abdominal radiologists (14 and 10 years of experience in abdominal imaging) in consensus and blinded to the surgical and histopathologic results analyzed the MR images at a PCS workstation. The following morphologic features were evaluated: location of the lesion in the pancreas (head, body, or tail), margin of the lesion on fat-saturated unenhanced T1-weighted images (completely well defined, partially well defined, or ill defined), shape of the lesion on fat-saturated unenhanced T1-weighted images (round, ovoid, or irregular), presence of a lesion capsule (rim of low signal intensity on T2-weighted or HSTE images), lesion homogeneity on T1- or T2-weighted images, presence of hemorrhage (area of high signal intensity on T1-weighted images and area of low signal intensity on T2-weighted or HSTE images), presence of an internal cystic portion (area of high signal intensity similar to that of the CSF on HSTE or T2-weighted images without enhancement), presence of pancreatic ductal dilatation, and presence of pancreatic parenchymal atrophy. The signal intensity of a lesion on fat-saturated unenhanced T1-weighted images was compared with that of the surrounding pancreas and was graded as follows: hypointense when the signal intensity of the lesion was lower than that of the back muscle, hypointense when the signal intensity of the lesion was not lower than that of the back muscle but was lower than that of the pancreas, and isointense when the signal intensity of the lesion was similar to that of the pancreas. The signal intensity of a lesion on fat-saturated T2-weighted images was compared with that of the surrounding pancreas and graded as follows: very hyperintense when the signal intensity of the lesion was higher than that of the spleen, hyperintense when the signal intensity of the lesion was not higher than that of the spleen but was higher than that of the pancreas, isointense when the signal intensity of the lesion was similar to that of the pancreas, and hypointense when the signal intensity of the lesion was lower than that of the pancreas. The dynamic enhancement pattern on T1- weighted images was analyzed as follows: persistent homogeneous enhancement, early heterogeneous and progressive enhancement, early heterogeneous and no progressive enhancement, or no enhancement (Fig. 1). Quantitative analysis For quantitative evaluation, working at a PCS workstation we defined regions of interest (ROIs) on fat-saturated T2-weighted images, fat-saturated unenhanced T1-weighted images, and dynamic enhanced T1-weighted images and measured the mean signal intensity of a pancreatic lesion and adjacent pancreatic parenchyma. The signal intensity of the pancreatic lesions was measured in the axial plane where the lesion had the maximum diameter. The signal intensity of the rterial Phase Portal Phase Delayed Phase JR:195, December

3 pancreatic parenchyma was measured in the pancreatic parenchyma with normal signal intensity on both T1- and T2-weighted images, that is, normal high signal intensity on T1-weighted images and normal low signal intensity on T2-weighted images. We measured signal intensity three times by placing an ROI box that was as large as possible and calculating the mean signal intensity by averaging the ROI measurements. The mean ROI sizes were as follows: ± (SD) mm 2 at the pancreatic lesions and ± mm 2 at the pancreatic parenchyma. Noise was measured on each image in ROIs of approximately 250 mm 2 positioned immediately lateral in relation to the abdominal wall. The areas with the most prominent ghost artifacts were not included. We used the following formula to calculate the contrast-to-noise ratio (CNR) of the lesion compared with the pancreas on the MR images: SI lesion SI pancreas ) / N SD, where SI is mean signal intensity and N SD is noise. Pathologic nalysis One pathologist (9 years of experience in pancreatic pathology) blinded to all imaging findings TLE 1: Qualitative Imaging Features of Small Solid Pseudopapillary Tumors and Other Small Pancreatic Tumors Characteristic Solid Pseudopapillary Tumors (n = 11) Endocrine Tumors (n = 12) p a denocarcinoma (n = 22) p a Sex Men Women Location Head ody Tail Margin Completely defined Partially defined Ill-defined Shape Round Ovoid Irregular Capsule Hemorrhage Cystic portion Ductal dilatation Pancreatic atrophy T1 signal intensity Very low Low Isointense 3 1 T2 signal intensity b Low 1 Isointense 4 13 High Very high Enhancement pattern Persistent homogeneous 8 3 Early heterogeneous progressive Early heterogeneous not progressive 3 No enhancement Note Values are numbers of tumors. a etween solid pseudopapillary tumors and other tumors. b T2-weighted images were not available for two patients with endocrine tumors and one patient with adenocarcinoma JR:195, December 2010

4 Fig year-old woman with 1.4-cm solid pseudopapillary tumor., xial fat-saturated T2-weighted MR image shows small completely well-defined homogeneous round lesion (arrow) in pancreatic body. Signal intensity of lesion is much higher than that of both spleen and pancreas. Rim of low signal intensity representing capsule is not present., xial ultrafast T2-weighted MR image shows internal foci of high signal intensity within lesion (arrow), suggesting presence of tiny cystic portions. Pathologic examination showed hemorrhagic necrosis of 10% of tumor area (not shown). C, xial dynamic fat-saturated T1-weighted gradientrecalled echo images in unenhanced and arterial, portal, and delayed phases (left to right) show lesion (arrows) has early heterogeneous and progressive enhancement. Unenhanced image shows predominant signal intensity of lesion is lower than that of spleen. retrospectively reviewed the microscopic slides of the 11 solid pseudopapillary tumors for the presence of a fibrous capsule, hemorrhage or necrosis, and cystic degeneration. If a fibrous capsule was present, hemorrhagic focus or cystic degeneration, the thickness of the capsule, the percentage of capsule compared with the total tumor circumference, and the percentage of hemorrhage or necrosis or cystic degeneration compared with the total tumor area were analyzed. Statistical nalysis For statistical analysis, Fisher s exact test was used for qualitative analysis. The Mann-Whitney U test was performed to evaluate the statistical difference between the small solid pseudopapillary tumors and the other two tumor types in the qualitative assessment of the CNR of lesions on T1- and T2-weighted images. Changes in CNR between phases of dynamic enhancement for each type of tumor were compared by use of Wilcoxon s signed rank test. receiver operating characteristic curve of the CNR of lesions on T1- and T2-weighted images was obtained to determine the optimum cutoff value for differentiation of solid pseudopapillary tumors from the other two tumor types. The optimum cutoff value was defined as the value at which the sum of the sensitivity and specificity was maximized. The sensitivity, specificity, and accuracy of each MRI feature of small solid pseudopapillary tumors that were significantly different from either endocrine tumors or adenocarcinomas in the qualitative and quantitative analyses were calculated. ll statistical analyses were performed with SPSS software (version 12.0 for Microsoft Windows, SPSS). value of p < 0.05 was considered to indicate statistical significance. Results ll but one of the patients with small solid pseudopapillary tumors were women. statistically significant difference was found between small solid pseudopapillary tumors and small adenocarcinomas with respect to patient sex (p = 0.024) but not between small solid pseudopapillary tumors and small endocrine tumors (p > 0.05). Patients with small solid pseudopapillary tumors were significantly younger than those with adenocarcinomas (p = 0.001), but no statistical difference between solid pseudopapillary tumors and endocrine tumors was found with respect to patient age (p > 0.05). The MRI findings on 11 small solid pseudopapillary tumors are presented in Table 1. ll solid pseudopapillary tumors had a completely (82%) or partially (18%) well-defined margin. Six solid pseudopapillary tumors (55%) were round (Fig. 2). Six solid pseudopapillary tumors (55%) had homogeneous signal intensity on fat-saturated T1-weighted images, and eight solid pseudopapillary tumors (73%) had heterogeneous signal intensity on fat-saturated T2-weighted images. Only two of the 11 solid pseudopapillary tumors (18%) had focal heterogeneous portions indicating cystic spaces, and only one (9%) had a focal portion indicating intratumoral hemorrhage. On T2-weighted images, only one solid pseudopapillary tumor (9%) had a low-signal-intensity rim representing a capsule. On fat-saturated T2-weighted images, all solid pseudo papillary tumors had high (6 tumors) or very high (5 tumors) signal intensity compared with the pancreatic parenchyma (Figs. 2 and 3). ll of the solid pseudopapillary tumors had early heterogeneous and progressive enhancement (Figs. 2 and 3). The results of the qualitative analysis showed that in comparison with small endocrine tumors and small adenocarcinomas (p < 0.05), small solid pseudopapillary tumors had a significant- C JR:195, December

5 Fig year-old woman with 2.2-cm solid pseudopapillary tumor., xial fat-saturated T2-weighted image shows completely well-defined round lesion (arrow) without capsule in pancreatic head. Signal intensity of lesion is higher than that of pancreas but lower than that of spleen. Pathologic examination revealed lesion had complete thin capsule (1.5 mm thick) (not shown)., xial dynamic fat-saturated T1-weighted gradient-recalled echo images in unenhanced and arterial, portal, and delayed phases (left to right) show lesion (arrows) has early heterogeneous and progressive enhancement. Unenhanced image shows signal intensity of lesion is lower than that of spleen. Lesion does not have high-signalintensity portion, so no evidence of intratumoral hemorrhage is present. Fig year-old man with 1.7-cm endocrine tumor., xial fat-saturated T2-weighted MR image shows lesion (arrow) in tail of pancreas has predominantly higher signal intensity than pancreas but not spleen., xial ultrafast T2-weighted MR image shows slightly higher signal intensity of lesion (arrow) than of pancreas and small areas of hyperintense foci. C, xial dynamic fat-saturated T1-weighted gradientrecalled echo MR images in unenhanced and arterial, portal, and delayed phases (left to right) show lesion (arrows) has early strong and persistent enhancement, which is unusual for solid pseudopapillary tumor. Unenhanced image shows lesion has heterogeneous low signal intensity in comparison with spleen. ly greater tendency to occur in middle-aged women, occur in the pancreatic tail, have completely well-defined margins, be round, have very high signal intensity on fat-saturated T2- weighted images, and have early heterogeneous and progressive enhancement (Table 1). In the qualitative comparison of small solid pseudopapillary tumors and small endocrine tumors, the enhancement pattern and the T2 signal intensity of the tumors were significant differentiating features. ll small solid pseudopapillary tumors had early heterogeneous progressive enhancement, but most of the endocrine tumors (8/12, 67%) had persistent homogeneous enhancement (Table 1 and Fig. 4). In the qualitative comparison of small solid pseudopapillary tumors and small adenocarcinomas, the following MRI features differed significantly between small solid pseudopapillary tumors and small adenocarcinomas (p < 0.05) (Table 1 and Fig. 5): age and sex; lesion location, margin and shape; presence of ductal dilatation; and presence of pancreatic atrophy. However, the enhancement pattern of small solid pseudopapillary tumors was not significantly different from that of small adenocarcinomas. In the quantitative analysis, the CNR between the lesion and the pancreas on fatsaturated unenhanced T1- and T2-weighted images was significantly higher for solid C 1328 JR:195, December 2010

6 Fig year-old woman with 2.5-cm adenocarcinoma., xial fat-saturated T2-weighted MR image shows ill-defined irregular lesion (arrow) in pancreatic head. Signal intensity of lesion is slightly higher than that of adjacent pancreatic parenchyma., xial ultrafast T2-weighted MR image shows dilatation of pancreatic duct (arrowhead) caused by lesion (not shown). C, xial dynamic fat-saturated T1-weighted gradientrecalled echo images in unenhanced and arterial, portal, and delayed phases (left to right) show lesion (arrows) has slow and gradual progressive enhancement similar to that of solid pseudopapillary tumor. Unenhanced image shows ill-defined isointense lesion. TLE 2: Contrast-to-Noise Ratio etween Lesion and Pancreatic Parenchyma Sequence Solid Pseudopapillary Tumors pseudopapillary tumors than for both ductal adenocarcinomas and endocrine tumors (p < 0.05) (Table 2 and Fig. 6). significant difference between solid pseudopapillary tumors and endocrine tumors also was found with respect to the CNR between the lesion and the pancreas on arterial phase T1- weighted images (Table 2). With a cutoff value of 10 for the CNR between the lesion and the pancreas on fat-saturated T2-weighted images, the sensitivity and specificity for solid pseudopapillary tumors were 90.9% and 74.2%. With a cutoff value of 30 on fat-saturated T2-weighted images, the sensitivity and specificity for solid pseudopapillary tumors were 45.5% and 90.3%. With a cutoff value of 5.6 on fat-saturated unenhanced T1-weighted images, the sensitivity and specificity were 100% and 64.7%, and with a cutoff value of 19, the values were 63.6% and 91.2%. Regarding the change in CNR between lesion and pancreas on dynamic T1-weighted images, both small solid pseudopapillary tumors and small adenocarcinomas had slow and progressive enhancement compared with the enhancement of the pancreatic parenchyma (Figs. 2, 3, 5, 6, and 6). Small endocrine tumors, however, had early enhancement similar to that of the pancreatic parenchyma and afterward had relatively steady enhancement (Figs. 4 and 6C). The sensitivity, specificity, and diagnostic accuracy of the findings that had statistically significant differences between small solid pseudopapillary tumors and either endocrine tumors or adenocarcinomas in the diagnosis of small solid pseudopapillary tumors were Endocrine Tumors p denocarcinoma p Fat-saturated T2-weighted ± ± ± Unenhanced T1-weighted ± ± ± Contrast-enhanced T1-weighted rterial phase ± ± ± Portal phase 8.47 ± ± ± Delayed phase 2.30 ± ± ± Note Values are mean ± SD. old type indicates statistically significant difference. calculated (Table 3). The sensitive findings were absence of ductal dilatation or pancreatic atrophy, female sex, completely well-defined tumor margins, low signal intensity on fat-saturated T1-weighted images and high signal intensity on fat-saturated T2-weighted images, early heterogeneous and progressive enhancement, and CNR between lesion and pancreas greater than 10 on fat-saturated T2-weighted images and less than 5.6 on fat-saturated unenhanced T1-weighted imag- C JR:195, December

7 40 p = p = p = p = p = p = Contrast-to-Noise Ratio Unenhanced rterial es. The specific findings were age less than 50 years, tumor location in the pancreatic tail, presence of hemorrhage or capsule, very high signal intensity on fat-saturated T2- weighted images, and CNR between lesion and pancreas greater than 30 on fat-saturated T2-weighted images and less than 19 on fatsaturated unenhanced T1-weighted images. fter pathologic review of the 11 solid pseudopapillary tumors, only two tumors, measuring 2.9 and 2.2 cm in diameter (Fig. 3), had complete capsules, which were 2 4 mm and 1.5 mm thick. The former was depicted at MRI, but the latter was not. These two tumors also had approximately 30% and 10% hemorrhage and necrosis of the tumor volume. Five other small solid pseudopapillary tumors (1.3, 1.6, 1.7, 1.8, and 2.7 cm in diameter) did not have any capsule, and two of these lesions (1.8 and 2.7 cm in diameter) had approximately 5 10% hemorrhage and necrosis. The other four solid pseudopapillary tumors (1.4, 2.6, 2.9, and 3 cm in diameter) had sparse capsules occupying less than 5% (three tumors) and approximately 10% (one tumor) of the tumor circumference. ll four lesions had capsules thinner than 1 mm, and one lesion (1.4 cm in diameter) had approximately 10% hemorrhage and necrosis. s for the cystic portion, only two tumors (2.9 and 2.6 cm in diameter) had prominent cystic portions. Portal Delayed Fig. 6 Change in contrast-to-noise ratio between lesion and pancreas during dynamic enhancement., Graph shows small solid pseudopapillary tumors of pancreas have slow progressive enhancement relative to adjacent pancreas., Graph shows small adenocarcinomas of pancreas have slow progressive enhancement relative to adjacent pancreas. C, Graph shows small endocrine tumors of pancreas have rapid persistent enhancement. Contrast-to-Noise Ratio Contrast-to-Noise Ratio Unenhanced Discussion ccording to our qualitative and quantitative results, small solid pseudopapillary tumors can be characterized as completely or partially well-defined pure solid tumors with the following features: low or very low signal intensity on fat-saturated unenhanced T1-weighted images, high or very high signal intensity on T2-weighted images, strong predominance among women, slight predilection for the pancreatic tail, and early heterogeneous and slowly progressive enhancement. The small solid pseudopapillary tumors in our study were similar to the usual previously reported solid pseudopapillary tumors [4, 5, 7] in some respects, that is, completely well-defined margins, strong predominance among women, and slow and progressive enhancement. The tumors in our study, however, were different from the usual solid pseudopapillary Unenhanced rterial 20 p = 0.09 p = rterial Portal Portal p = Delayed Delayed tumor in other respects, that is, rarely having a capsule, having hemorrhagic and cystic changes, and occurring in patients older than those with the usual solid pseudopapillary tumor. Whereas the usual solid pseudopapillary tumor is known to be found in young women in the second and third decades of life [4], the small solid pseudopapillary tumors in our study occurred mainly in women in the third and fourth decades of life. ecause they are rare cross-sectional imaging findings with other pancreatic neoplasms, both a capsule and intratumoral hemorrhage have been known to be important in the diagnosis of solid pseudopapillary tumors. s our study showed, however, MR images did not depict a capsule around most of the small solid pseudopapillary tumors (10/11, 91%). Pathologic examination also showed that most of the small solid pseudopapil- C 1330 JR:195, December 2010

8 TLE 3: Value of Statistically Significant Findings in Diagnosis of Small Solid Papillary Tumors Finding lary tumors (9/11, 82%) had a sparse or no capsule. In addition, most of the small solid pseudopapillary tumors (10/11, 91%) did not exhibit intratumoral hemorrhage on MR images. This finding correlated with the pathologic finding that all but one of the small solid pseudo papillary tumors had intratumoral hemorrhage of 10% or less of the tumor area. These findings are supported by those of a recent study [9] in which all six small ( 3 cm) solid pseudopapillary tumors were both radiologically and pathologically unencapsulated masses without hemorrhagic areas and with rare cystic changes. Even though they are specific findings, both a capsule and intratumoral hemorrhage are difficult to use as clues in the diagnosis of small solid pseudopapillary tumors because of their paucity. It is important to find clues other than the presence of a capsule and hemorrhage in the diagnosis of small solid pseudopapillary tumors. In our study, the features of small solid pseudopapillary tumors that enabled differentiation from small endocrine tumors and adenocarcinomas were age less than 50 years, female sex, tumor location in the pancreatic tail, completely well-defined tumor margin, Sensitivity (%) Specificity (%) ccuracy (%) ge < 50 y Female sex Location in tail of pancreas Completely well-defined margin Round shape Capsule Hemorrhage No ductal dilatation No pancreatic atrophy Low T1, high T2 signal intensity a Very high T2 signal intensity b Early heterogeneous and progressive enhancement c Contrast-to-noise ratio between lesion and pancreas on fat-saturated images T2-weighted, > T2-weighted, > T1-weighted, < T1-weighted, < Note old type indicates 80% or greater. a Lower signal intensity than that of pancreas on unenhanced fat-saturated T1-weighted images and higher signal intensity than that of pancreas on fat-saturated T2-weighted images. b Higher signal intensity than that of spleen on fat-saturated T2-weighted images. no pancreatic ductal dilatation, no pancreatic atrophy, very high signal intensity on fatsaturated T2-weighted images, and lower T1 CNR and higher T2 CNR between lesion and pancreas. n ill-defined tumor margin and the presence of pancreatic ductal dilatation and atrophy strongly favored adenocarcinoma. Early homogeneous and persistent enhancement favored endocrine tumor. Most of the benign solid pseudopapillary tumors (9/11, 82%) had either a well-defined margin or were round. The only malignant solid pseudopapillary tumor in our study, however, had both a partially defined margin and an irregular shape. This finding is consistent with reports [12, 13] that malignant solid pseudopapillary tumors have a higher tendency toward focal lobulated margins than do benign solid pseudopapillary tumors. Pancreatic ductal dilatation was not observed in most of the solid pseudopapillary tumors (10/11, 91%) in this study. This finding is consistent with the observation that solid pseudopapillary tumors rarely cause bile duct or pancreatic ductal dilatation because of the softness of the lesions [14]. In contrast, most of the adenocarcinomas (21/22, 95%) were associated with pancreatic ductal dilatation, as has been previously established [15, 16]. Interestingly, compared with other tumors, small solid pseudopapillary tumors more frequently had higher signal intensity than the spleen on fat-saturated T2-weighted images. Similarly, the CNR of the small solid pseudopapillary tumors was significantly lower on T1-weighted images and significantly higher on T2-weighted images than was the CNR of other tumors. The reason for these distinctions can be found in the pathologic characteristics. solid pseudopapillary tumor is characterized by solid areas alternating with a pseudopapillary pattern and cystic spaces resulting from degenerative changes in the solid portion [11, 17]. Whereas large tumors contain a mixture of solid, cystic, and pseudopapillary patterns with a high hemorrhagic tendency, smaller tumors are primarily organized in solid sheets and nests of cells, which have an abundant cytoplasm with variable degenerative changes [17 19]. The pathologic features, such as abundant cytoplasm, pseudopapillary arrangement, and degenerative changes, can contribute to the high T2 signal intensity of small solid pseudopapillary tumors on MR images. Regarding dynamic enhancement, solid pseudopapillary tumors are known to exhibit mild and gradual increases in contrast enhancement [15, 19, 20]. Nakatani et al. [21] described solid pseudopapillary tumors as having minimal contrast enhancement in the early arterial phase that gradually increases in the late phase. In the qualitative and quantitative analyses in our study, as in the previous studies, all of the small solid pseudopapillary tumors had slow and progressive enhancement, whereas most of the endocrine tumors (8/12, 66%) had early homogeneous and prolonged enhancement up to the delayed phase [22]. This difference between solid pseudopapillary tumors and endocrine tumors with respect to enhancement pattern may be an important clue to the correct diagnosis. Solid pseudopapillary tumors and adenocarcinomas, however, had similar dynamic enhancement patterns. This finding is presumably related to the large interstitial spaces in both of these tumor types. There were several limitations to our study. The first was the small number of cases because of the rarity of the tumor and the retrospective design. Second, we shortened the inclusion period for adenocarcinoma to reduce the overall number of adenocarcinomas. JR:195, December

9 ecause these tumors are the most common, we believe that including the true number might have distorted the real difference between small solid pseudopapillary tumor and adenocarcinoma. We also believe, however, that the difference, if any, would not be great because adenocarcinomas were consecutively included for 7 years. nother limitation was that CNR is highly dependent on differences in technique and that without rigid standardization, CNR results cannot be applied to clinical practice at other centers. Therefore, further prospective study with rigid standardization of CNR should be performed. Despite the limitations of the study, we believe our results are valuable because small solid pseudopapillary tumors are very uncommon and, to our knowledge, no report has specifically described the MRI features of these tumors. We conclude that small solid pseudopapillary tumors are predominantly seen as completely well-demarcated pure solid tumors in middle-aged women and that they have lower signal intensity on T1-weighted images and higher signal intensity on T2-weighted images in comparison with the other two tumor types and have early heterogeneous and progressive enhancement. These MRI features may be useful in differentiating small solid pseudopapillary tumors from other small pancreatic tumors. References 1. Lam KY, Lo CY, Fan ST. Pancreatic solid-cysticpapillary tumor: clinicopathologic features in eight patients from Hong Kong and review of the literature. World J Surg 1999; 23: Mergo PJ, Helmberger TK, uetow PC, Helmberger RC, Ros PR. Pancreatic neoplasms: MR imaging and pathologic correlation. RadioGraphics 1997; 17: Yang F, Jin C, Long J, et al. Solid pseudopapillary tumor of the pancreas: a case series of 26 consecutive patients. m J Surg 2009; 198: Coleman KM, Doherty MC, igler S. Solidpseudopapillary tumor of the pancreas. Radio- Graphics 2003; 23: Choi JY, Kim MJ, Kim JH, et al. Solid pseudopapillary tumor of the pancreas: typical and atypical manifestations. JR 2006; 187:481; [web] W178 W Choi I, Kim KW, Han MC, Kim YI, Kim CW. Solid and papillary epithelial neoplasms of the pancreas: CT findings. Radiology 1988; 166: Cantisani V, Mortele KJ, Levy, et al. MR imaging features of solid pseudopapillary tumor of the pancreas in adult and pediatric patients. JR 2003; 181: Machado MC, Machado M, acchella T, Jukemura J, lmeida JL, Cunha JE. Solid pseudopapillary neoplasm of the pancreas: distinct patterns of onset, diagnosis, and prognosis for male versus female patients. Surgery 2008; 143: Yao X, Ji Y, Zeng M, Rao S, Yang. Solid pseudopapillary tumor of the pancreas: cross-sectional imaging and pathologic correlation. Pancreas 2010; 36: Papavramidis T, Papavramidis S. Solid pseudopapillary tumors of the pancreas: review of 718 patients reported in English literature. J m Coll Surg 2005; 200: Tang LH, ydin H, rennan MF, Klimstra DS. Clinically aggressive solid pseudopapillary tumors of the pancreas: a report of two cases with components of undifferentiated carcinoma and a comparative clinicopathologic analysis of 34 conventional cases. m J Surg Pathol 2005; 29: Chung YE, Kim MJ, Choi JY, et al. Differentiation of benign and malignant solid pseudopapillary neoplasms of the pancreas. J Comput ssist Tomogr 2009; 33: Lee JH, Yu JS, Kim H, et al. Solid pseudopapillary carcinoma of the pancreas: differentiation from benign solid pseudopapillary tumour using CT and MRI. Clin Radiol 2008; 63: Koizumi J, Kodera K, Kaneda S, et al. Two cases of solid and cystic tumor of the pancreas (including one case of MRI). Nippon Igaku Hoshasen Gakkai Zasshi 1990; 50: Elsayes KM, Narra VR, bou El bbass H, ly TS, Radwan SM, Chen ZM. Pancreatic tumors: diagnostic patterns by 3D gradient-echo post contrast magnetic resonance imaging with pathologic correlation. Curr Probl Diagn Radiol 2006; 35: Rosewicz S, Wiedenmann. Pancreatic carcinoma. Lancet 1997; 349: Santini D, Poli F, Lega S. Solid-papillary tumors of the pancreas: histopathology. JOP 2006; 7: Pelosi G, Iannucci, Zamboni G, resaola E, Iacono C, Serio G. Solid and cystic papillary neoplasm of the pancreas: a clinico-cytopathologic and immunocytochemical study of five new cases diagnosed by fine-needle aspiration cytology and a review of the literature. Diagn Cytopathol 1995; 13: Vargas-Serrano, Dominguez-Ferreras E, Chinchon-Espino D. Four cases of solid pseudopapillary tumors of pancreas: imaging findings and pathological correlations. Eur J Radiol 2006; 58: Kalb, Sarmiento JM, Kooby D, dsay NV, Martin DR. MR imaging of cystic lesions of the pancreas. RadioGraphics 2009; 29: Nakatani K, Watanabe Y, Okumura, et al. MR imaging features of solid-pseudopapillary tumor of the pancreas. Magn Reson Med Sci 2007; 6: uetow PC, Parrino TV, uck JL, et al. Islet cell tumors of the pancreas: pathologic-imaging correlation among size, necrosis and cysts, calcification, malignant behavior, and functional status. JR 1995; 165: JR:195, December 2010