- Email: [email protected]
Pancreatic cancer: Lack of association between apparent diffusion coefficient values and adverse pathological features
Clinical Radiology 68 (2013) e191ee197
Contents lists available at SciVerse ScienceDirect
Clinical Radiology journal homepage: www.clinicalradiologyonline.net
Pancreatic cancer: Lack of association between apparent diffusion coefficient values and adverse pathological features A.B. Rosenkrantz a, *, B.W. Matza a, A. Sabach a, C.H. Hajdu b, N. Hindman a a b
Department of Radiology, NYU Langone Medical Center, 550 First Avenue, New York, NY 10016, USA Department of Pathology, NYU Langone Medical Center, 550 First Avenue, New York, NY 10016, USA
art icl e i nformat ion Article history: Received 26 September 2012 Received in revised form 12 November 2012 Accepted 20 November 2012
AIM: To identify retrospectively potential associations between apparent diffusion coefficient (ADC) values of pancreatic adenocarcinoma and tumour grade as well as other pathological features, using histopathological assessment from the Whipple procedure as the reference standard. MATERIALS AND METHODS: Thirty patients with pancreatic adenocarcinoma underwent magnetic resonance imaging (MRI) including diffusion-weighted imaging with b-values of 0 and 500 s/mm2 before the Whipple procedure. Two radiologists independently recorded the ADC values of the tumour and benign pancreas for all cases. ADC values were compared with histopathological findings following the Whipple procedure. RESULTS: The intra-class correlation coefficient was 0.689 for benign pancreas and 0.695 for tumours, indicating good inter-reader agreement for ADC values. The mean ADC value was significantly lower in tumours than in benign pancreas for both readers (reader 1: 1.74 0.34 103 mm2/s versus 2.08 0.48 103 mm2/s, respectively, p ¼ 0.006; reader 2: 1.69 0.41 103 mm2/s versus 2.11 0.54 103 mm2/s, respectively, p < 0.001). However, there was no significant difference in mean ADC between poorly and well/moderately differentiated tumours for either reader (reader 1: 1.69 0.36 103 mm2/s versus 1.78 0.33 103 mm2/s, respectively, p ¼ 0.491; reader 2: 1.62 0.33 103 mm2/s versus 1.75 0.49 103 mm2/s, respectively, p ¼ 0.405). The area under the curve (AUC) for differentiation of poorly and well/moderately differentiated tumours was 0.611 and 0.596 for readers 1 and 2, respectively, and was not significantly better than an AUC of 0.500 for either reader (p 0.306). In addition, ADC was not significantly different for either reader between tumours with stage T3 versus stage T1/T2, between tumours with and without metastatic peri-pancreatic lymph nodes, or between tumours located in the pancreatic head versus other pancreatic regions (p 0.413). CONCLUSION: No associations between ADC values of pancreatic adenocarcinoma and tumour grade or other adverse pathological features were observed. Ó 2012 The Royal College of Radiologists. Published by Elsevier Ltd. All rights reserved.
* Guarantor and correspondent: A.B. Rosenkrantz, Department of Radiology, NYU Langone Medical Center, 550 First Avenue, New York, NY 10016, USA. Tel.: þ1 212 263 0232; fax: þ1 212 263 6634. E-mail address: [email protected] (A.B. Rosenkrantz). 0009-9260/$ e see front matter Ó 2012 The Royal College of Radiologists. Published by Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.crad.2012.11.006
e192
A.B. Rosenkrantz et al. / Clinical Radiology 68 (2013) e191ee197
Introduction
MRI
Pancreatic adenocarcinoma represents the most common malignancy of the pancreas and the fourth most common cause of cancer-related death in men and women in the Unites States.1 Despite its rapidly progressive disease course and overall very poor survival even with aggressive intervention, long-term survival has been shown to be possible in the setting of early detection.2 Furthermore, accurate characterization of the aggressiveness of newly detected cases may have a role in establishing patient prognosis and treatment selection. Although a spectrum of pathological features, including tumour size, grade, stage, and nodal status, significantly influence patient survival, accurate prediction of these features at the time of initial diagnosis remains a substantial challenge.2e5 Diffusion-weighted imaging (DWI) is a quantitative magnetic resonance imaging (MRI) technique that reflects tissue cellularity and has been employed to assist the detection and characterization of aggressiveness of a variety of neoplasms.6 Apparent diffusion coefficient (ADC) values obtained from DWI have generally been shown to be lower in malignant than in benign tissue, and lower in higher than lower-grade tumours.7e9 Indeed, at least nine studies have demonstrated significantly lower ADC in pancreatic adenocarcinoma than in benign pancreas tissue.10e18 Yet, the role of ADC values in predicting adverse pathological features of pancreatic cancer is less established. For instance, to the authors’ knowledge, only a single previous study has shown a significant association between the ADC value and pathological grade of pancreatic cancer.19 However, in our clinical practice, such a relationship has not been observed. Thus, in view of the paucity of data on this topic, the present study was conducted to identify potential associations between the ADC value of pancreatic adenocarcinoma and tumour grade, as well as other adverse pathological features, using histopathological assessment from the Whipple procedure as the reference standard.
MRI was performed using a 1.5 T whole-body system (Magnetom Sonata, Symphony, or Avanto; Siemens Healthcare, Erlangen, Germany) and anterior and posterior torso phased-array coils. Examinations included two-dimensional (2D) axial in- and opposed-phase gradient-echo T1weighted sequences, 2D axial and/or coronal half-Fourier acquisition single-shot turbo-spin echo T2-weighted sequences, 2D axial fat-suppressed turbo spin-echo T2weighted sequences, and dynamic three-dimensional (3D) fat-suppressed gradient-echo T1-weighted sequence, performed before and during the arterial, portal venous, and equilibrium phases after the intravenous administration of 0.1 mmol/kg gadolinium-chelate. These sequences were not formally evaluated as part of this study. In addition, all examinations included an axial single-shot echo-planar imaging DWI sequence with tri-directional motion-probing gradients [1600e3400 ms repetition time (TR), 67e82 ms echo time (TE), 300e400 mm field of view (FOV) with 70e80% rectangular FOV, 144 192 matrix, 7e8 mm section thickness; navigator-trigger technique on the Avanto scanner, breath-hold technique on the Symphony and Sonata machines]. The ADC map was constructed for all cases on a voxel-by-voxel basis using b-values of 0 and 500 s/mm2.
Materials and methods Patients The present retrospective study was Health Insurance Portability and Accountability Act (HIPAA)-compliant and approved by the institutional review board, with a waiver of the requirement for written informed consent. A departmental database was searched to identify patients with pathologically confirmed pancreatic adenocarcinoma who underwent 1.5 T MRI of the pancreas that included a DWI sequence incorporating b-values of 0 and 500 s/mm2 prior to the Whipple procedure. MRI examinations were performed between 1 January 2006 and 31 December 2011. A total of 30 patients (19 male, 11 female; mean age 68 13 years) were identified. The mean delay between MRI and the Whipple procedure was 31 25 days.
Image analysis Images were evaluated independently by two fellowship-trained radiologists with 4 and 6 years of experience, using a picture archiving and communication system (PACS; iSite, Philips Healthcare). The radiologists were aware that all cases had undergone subsequent surgery for pancreatic adenocarcinoma, but were unaware of other pathological findings. The radiologists reviewed the ADC map for all patients in conjunction with the conventional sequences in order to localize each tumour on the ADC map. A free-hand regionof-interest (ROI) was traced along the edge of each lesion on the ADC map, slightly inside its outer margin to avoid partial volume averaging artefact. The ADC of normal pancreatic parenchyma was also measured for each case by placing a circular ROI within the mid-body of the pancreas, taking care to avoid lesions, vessels, and artefacts. All benign and malignant ROIs were placed twice, and the average value of the two ROIs recorded for subsequent data analysis.
Reference standard All patients had undergone a Whipple procedure following MRI. The following histopathological features were then recorded for each case: pathologically determined tumour size; grade (classified in a binary fashion as poorly differentiated versus well/moderately differentiated, defined in terms of most aggressive elements present); T stage (classified in a binary fashion as stage T3 versus stages T1/T2,); and nodal status (classified in a binary fashion in terms of whether metastatic peri-
A.B. Rosenkrantz et al. / Clinical Radiology 68 (2013) e191ee197
pancreatic lymph nodes were identified). Well and moderately differentiated tumours, as well as stage T1 and T2 tumours, were grouped for the analyses of tumour grade and stage, respectively, given the infrequency of well-differentiated and stage T1 tumours. Finally, in view of potential worse survival for pancreatic adenocarcinoma located in the pancreatic head,3 it was recorded whether each tumour was located in the pancreatic head versus the body or tail of the pancreas. The presence of perineural invasion was not evaluated given that nearly the entire sample (28 of 30 cases) exhibited this pathological feature.
e193
Comparison of ADC values between benign and malignant pancreas Table 1 summarizes findings in terms of mean ADC values of benign pancreas and pancreatic adenocarcinoma for each reader. Mean ADC values were significantly lower for pancreatic tumours than benign pancreas for both readers (reader 1: 1.74 0.34 103 mm2/s versus 2.08 0.48 103 mm2/s, respectively, p ¼ 0.006; reader 2: 1.69 0.41 103 mm2/s versus 2.11 0.54 103 mm2/s, respectively, p < 0.001).
Association between ADC values and tumour grade Statistical analysis Statistical analysis was performed using software (Medcalc for Windows, version 11.1.1.0; Medcalc Software, Mariakerke, Belgium). As a measure of intra-reader agreement, the intra-class correlation coefficient (ICC) of measured ADC values was calculated for the two readers for both benign pancreas and pancreatic adenocarcinoma. ICC was interpreted as follows: <0.04, poor agreement; 0.4e0.6, good agreement; 0.6e0.8, moderate agreement; >0.8, excellent agreement. A paired t-test was used to compare ADC values between pancreatic tumours and the benign pancreas. A series of unpaired t-tests were used to compare ADC values between the following groups: poorly and well/moderately differentiated tumours; stage T3 and stage T1/T2 tumours; tumours with and without metastatic peri-pancreatic lymph nodes; and tumours located in the pancreatic head versus body or tail. Receiver operating characteristic curve (ROC) analysis was used to calculate the area under the curve (AUC) of the ADC values for differentiating poorly and well/moderately differentiated cases. Finally, Pearson’s correlation coefficient was used to assess the association between ADC values and pathologically determined tumour size. These tests were performed separately for the two readers. All comparisons were two-sided and considered statistically significant at p < 0.05.
Results Tumours The 30 tumours had a mean size at histopathological analysis of 25 8 mm. Among the 30 tumours, 15 (50%) were poorly differentiated, 25 (83%) were stage T3, 22 (73%) had metastatic peri-pancreatic lymph nodes, and 23 (77%) were located in the pancreatic head. Sample cases are shown in Figs 1 and 2.
Inter-reader agreement ICC between the two readers in terms of ADC values was 0.689 for benign pancreas and 0.695 for tumours, consistent with good inter-reader agreement.
Table 2 summarizes findings in terms of mean ADC of pancreatic adenocarcinoma, stratified by various pathological features. There was no significant difference in mean ADC between poorly and well/moderately differentiated tumours for either reader (reader 1: 1.69 0.36 103 mm2/s versus 1.78 0.33 103 mm2/s, respectively, p ¼ 0.491; reader 2: 1.62 0.33 103 mm2/s versus 1.75 0.49 103 mm2/s, respectively, p ¼ 0.405). At ROC analysis, the AUC for differentiation of poorly and well/ moderately differentiated tumours was 0.611 and 0.596 for readers 1 and 2, respectively. For neither reader was the AUC significantly better than the diagonal line corresponding with random guessing and having an AUC of 0.500 (p 0.306).
Association between ADC values and other prognostic features There was no significant correlation between ADC value and tumour size for either reader (reader 1: r ¼ 0.072, p ¼ 0.706; reader 2: r ¼ 0.024, p ¼ 0.901). In addition, there was no significant difference in ADC value between tumours with stage T3 versus stage T1/T2, between tumours with and without metastatic peri-pancreatic lymph nodes, or between tumours located in the head versus other regions of the pancreas (reader 1: p 0.543; reader 2: p 0.413; Table 2).
Discussion In this study, no associations between ADC values of pancreatic adenocarcinoma and a spectrum of adverse pathological features were found when using findings from the Whipple procedure as the reference standard. Specifically, no association was observed between ADC and any of tumour size, grade, T stage, nodal status, or location. The lack of any apparent association was noted for two independent radiologists who had good inter-reader agreement in terms of recorded ADC values. Thus, based on the present data, it seems that ADC values are currently unlikely to be of clinical value for the non-invasive prediction of adverse pathological features of newly detected cases of pancreatic cancer. It is relevant that both radiologists observed significantly lower ADC values in pancreatic adenocarcinoma than in
e194
A.B. Rosenkrantz et al. / Clinical Radiology 68 (2013) e191ee197
Figure 1 A 75-year-old man with adenocarcinoma of the pancreatic body. (a) Axial half-Fourier acquisition single-shot turbo spin-echo T2weighted image shows a solid mass in the pancreatic body (arrow). (b) Axial fat-suppressed contrast-enhanced gradient-echo T1-weighted image shows hypovascularity of the mass (arrow). (c) Axial diffusion-weighted image with a b-value of 500 s/mm2 shows discrete enlargement of the pancreas in this region (arrow) without clear hyperintensity relative to the remainder of the pancreas. (d) ADC map shows foci of decreased ADC within the lesion (arrow). ADC of the mass, when averaged between the two readers, was 1.28 103 mm2/s, lower than the mean ADC value of all cases of pancreatic adenocarcinoma within the study cohort. At histopathological assessment following a Whipple procedure, the mass was found to be a stage T2 moderately differentiated adenocarcinoma without metastatic peri-pancreatic lymph nodes.
benign pancreas. This identified difference for both radiologists demonstrates an ability of the DWI sequence employed in the present study to distinguish groups of tissues, despite the failure to observe differences between subgroups of pancreatic adenocarcinoma. This difference between benign and malignant pancreatic regions has been consistently reported in previous studies.10,13,16 In addition, Muraoka et al.12 showed correlations between ADC values of pancreatic tumours and the presence of collagenous fibres, thus hypothesizing that such fibres may account for the observed abnormal diffusion behaviour in pancreatic tumours.12 To the authors’ knowledge, one previous study has reported significantly lower ADC in cases of pancreatic adenocarcinoma that are poorly differentiated in comparison with well/moderately differentiated lesions.20 In that study, Wang et al.20 reported mean ADC values in these two groups of 1.46 0.17 103 and 2.10 0.42 103 mm2/s, respectively. As in the present study, these authors also
used a maximal b-value of 500 s/mm2 and field strength of 1.5 T. However, in contrast to the findings of the present study and numerous other prior studies in the literature, Wang et al. failed to observe a significantly lower ADC value in pancreatic cancer than in benign pancreas. In fact, in their study, well- and moderately differentiated pancreatic cancer exhibited a higher ADC value than benign pancreas, although this difference was not statistically significant. This suggests that there may be underlying pathological differences between the lesions included in the cohorts of Wang et al. and the present cohort. Furthermore, the two cases of well- or moderately differentiated pancreatic adenocarcinoma in their study with dense fibrosis exhibited ADC values that overlapped the ADC values of poorly differentiated tumours, rather than overlapping the ADC values of other low-grade tumours. This finding also suggests heterogeneity among subgroups of pancreatic adenocarcinoma and indicates the complexity of efforts to correlate the ADC values of the lesions with these broad
A.B. Rosenkrantz et al. / Clinical Radiology 68 (2013) e191ee197
e195
Figure 2 A 67-year-old man with adenocarcinoma of the pancreatic head. (a) Axial half-Fourier acquisition single-shot turbo spin-echo T2weighted image shows a solid mass in the pancreatic head (arrow). (b) Axial fat-suppressed contrast-enhanced gradient-echo T1-weighted image shows hypovascularity of the mass (arrow). (c) Axial diffusion-weighted image with b-value of 500 s/mm2 shows mild increased signal intensity within the region of the mass (arrow). (d) ADC map shows mild decreased ADC within the lesion (arrow). ADC of the mass, when averaged between the two readers, was 1.90 103 mm2/s, greater than the mean ADC value of all cases of pancreatic adenocarcinoma within the study cohort. At histopathological assessment following a Whipple procedure, the mass was found to be a stage T3 poorly differentiated adenocarcinoma with metastatic peri-pancreatic lymph nodes.
categories. Nonetheless, in comparison with the study of Wang et al.,20 the present study entailed a slightly larger sample size and the reporting of independent results from two separate observers. Given the overall paucity of existing
Table 1 Comparison of apparent diffusion coefficient (ADC) values between benign pancreas and pancreatic adenocarcinoma for two readers. Reader 1 a
Benign pancreas Pancreatic adenocarcinoma a
Reader 2
ADC (103 mm2/s)
pValue
ADCa (103 mm2/s)
pValue
2.08 0.48 1.74 0.34
0.006
2.11 0.54 1.69 0.41
<0.001
Data are mean SD.
data and differences in findings between the two available studies, additional studies seem warranted to further investigate this issue. In addition to previously noted pathological factors that may impact ADC values, it is also important to note the role of the MRI technique itself in influencing ADC measurements. In particular, field strength, strategy for respiratory compensation, number of b-values used, maximal b-value selected, and ADC measurement technique, all warrant consideration. In the present study, all cases were imaged at 1.5 T. Although two previous studies showed no difference in pancreatic ADC values between 1.5 and 3 T,21,22 there may be an increase in artefacts and associated decrease in overall image quality for abdominal DWI at 3 T depending on the specific hardware and DWI sequence used,21e23 such that it is possible that DWI of the abdomen may be more robust at
e196
A.B. Rosenkrantz et al. / Clinical Radiology 68 (2013) e191ee197
Table 2 Comparison of apparent diffusion coefficient (ADC) values between subgroups of pancreatic adenocarcinoma for two readers. Reader 1
Reader 2
a
pADC pADC (103 mm2/s) Value (103 mm2/s) Value Grade Well/moderately differentiated Poorly differentiated T stage T1/T2 T3 Metastatic peripancreatic nodes Present Absent Location Head Elsewhere in pancreas a
1.78 0.33
0.491 1.75 0.49
1.69 0.36
1.62 0.33
1.70 0.42 1.75 0.33
0.425 1.81 0.45 1.66 0.41
0.679
1.71 0.31 1.80 0.43
0.543 1.68 0.43 1.70 0.42
0.912
1.72 0.33 1.80 0.42
0.576 1.65 0.42 1.80 0.41
0.413
0.405
Data are mean SD.
1.5 T at the present time. In addition, ADC maps were calculated based on two b-values of 0 and 500 s/mm2. Given its central location within the abdomen, the signal-to-noise ratio (SNR) within the pancreas may be poor at comparatively high b-values for certain hardware and software arrangements, which in turn can result in imprecise ADC calculations. Thus, a maximal b-value of 500 s/mm2 was selected to achieve sufficient SNR within the pancreas on the high b-value images for all systems and DWI sequences employed. Numerous prior studies of the role of ADC values in pancreatic cancer have also used a similar maximal b-value.12,13,18,20,24 In addition, the use of only two b-values in the present study is consistent with the DWI technique in numerous past studies.12,13,18,20,24 A greater number of b-values would have required longer acquisition time. In addition, a prior study that used Monte Carlo simulations to identify the optimal strategy for b-value selection in the kidneys, observed that increasing the number of b-values did not improve precision of DWI metrics in comparison with increasing the number of averages of the selected b-values.25 Furthermore, patients were imaged using either breath-hold or navigator-triggered DWI technique, depending on the given machine. To the authors’ knowledge, no previous study has compared ADC values within the pancreas in the same patients between these two techniques; thus, this remains an important area for future investigation to assist optimization of pancreatic DWI technique. Finally, some authors have suggested using normalized or relative ADC values, rather than absolute ADC measurements, for optimal tissue characterization.26,27 However, several studies failed to show improved diagnostic performance from this measurement strategy,28e30 and this approach was not explored in the present study. The biggest limitation of this study is uncertainty as to the basis for the differences in findings compared with the study of Wang et al.20 As previously noted, it is possible that histological differences between cases included in the two studies account for the discrepant conclusions. Additional
limitations of this study include its retrospective nature, relatively small sample size (despite larger sample size than in the study by Wang et al.), and assessment of only a single maximal b-value. In conclusion, although ADC values are significantly different between benign pancreas and pancreatic adenocarcinoma, no associations between ADC values and tumour grade or other adverse pathological features were observed. As the present findings differ from those of a previous study on this topic, additional studies are warranted. Until such studies are conducted, caution is warranted in the clinical use of ADC values to attempt to predict the prognosis of newly diagnosed pancreatic adenocarcinoma.
References 1. Jemal A, Siegel R, Ward E, et al. Cancer statistics, 2009. CA Cancer J Clin 2009;59:225e49. 2. Cleary SP, Gryfe R, Guindi M, et al. Prognostic factors in resected pancreatic adenocarcinoma: analysis of actual 5-year survivors. J Am Coll Surg 2004;198:722e31. 3. Sohn TA, Yeo CJ, Cameron JL, et al. Resected adenocarcinoma of the pancreasd616 patients: results, outcomes, and prognostic indicators. J Gastrointest Surg 2000;4:567e79. 4. Perini MV, Montagnini AL, Jukemura J, et al. Clinical and pathologic prognostic factors for curative resection for pancreatic cancer. HPB (Oxford) 2008;10:356e62. 5. Niwa T, Ueno M, Ohkawa S, et al. Advanced pancreatic cancer: the use of the apparent diffusion coefficient to predict response to chemotherapy. Br J Radiol 2009;82:28e34. 6. Koh DM, Collins DJ. Diffusion-weighted MRI in the body: applications and challenges in oncology. AJR Am J Roentgenol 2007;188:1622e35. 7. Guo AC, Cummings TJ, Dash RC, et al. Lymphomas and high-grade astrocytomas: comparison of water diffusibility and histologic characteristics. Radiology 2002;224:177e83. 8. Rahbar H, Partridge SC, Demartini WB, et al. In vivo assessment of ductal carcinoma in situ grade: a model incorporating dynamic contrastenhanced and diffusion-weighted breast MR imaging parameters. Radiology 2012;263:374e82. 9. Vargas HA, Akin O, Franiel T, et al. Diffusion-weighted endorectal MR imaging at 3 T for prostate cancer: tumour detection and assessment of aggressiveness. Radiology 2011;259:775e84. 10. Matsuki M, Inada Y, Nakai G, et al. Diffusion-weighed MR imaging of pancreatic carcinoma. Abdom Imaging 2007;32:481e3. 11. Lee SS, Byun JH, Park BJ, et al. Quantitative analysis of diffusion-weighted magnetic resonance imaging of the pancreas: usefulness in characterizing solid pancreatic masses. J Magn Reson Imaging 2008;28:928e36. 12. Muraoka N, Uematsu H, Kimura H, et al. Apparent diffusion coefficient in pancreatic cancer: characterization and histopathological correlations. J Magn Reson Imaging 2008;27:1302e8. 13. Fattahi R, Balci NC, Perman WH, et al. Pancreatic diffusion-weighted imaging (DWI): comparison between mass-forming focal pancreatitis (FP), pancreatic cancer (PC), and normal pancreas. J Magn Reson Imaging 2009;29:350e6. 14. Lemke A, Laun FB, Klauss M, et al. Differentiation of pancreas carcinoma from healthy pancreatic tissue using multiple b-values: comparison of apparent diffusion coefficient and intravoxel incoherent motion derived parameters. Invest Radiol 2009;44:769e75. 15. Fukukura Y, Takumi K, Kamimura K, et al. Pancreatic adenocarcinoma: variability of diffusion-weighted MR imaging findings. Radiology 2012;263:732e40. 16. Wiggermann P, Grutzmann R, Weissenbock A, et al. Apparent diffusion coefficient measurements of the pancreas, pancreas carcinoma, and mass-forming focal pancreatitis. Acta Radiol 2012;53:135e9. 17. Kamisawa T, Takuma K, Anjiki H, et al. Differentiation of autoimmune pancreatitis from pancreatic cancer by diffusion-weighted MRI. Am J Gastroenterol 2010;105:1870e5.
A.B. Rosenkrantz et al. / Clinical Radiology 68 (2013) e191ee197 18. Kartalis N, Lindholm TL, Aspelin P, et al. Diffusion-weighted magnetic resonance imaging of pancreas tumours. Eur Radiol 2009;19:1981e90. 19. Wang Y, Miller FH, Chen ZE, et al. Diffusion-weighted MR imaging of solid and cystic lesions of the pancreas. RadioGraphics 2011;31:E47e64. 20. Wang Y, Chen ZE, Nikolaidis P, et al. Diffusion-weighted magnetic resonance imaging of pancreatic adenocarcinomas: association with histopathology and tumour grade. J Magn Reson Imaging 2011;33:136e42. 21. Rosenkrantz AB, Oei M, Babb JS, et al. Diffusion-weighted imaging of the abdomen at 3.0 Tesla: image quality and apparent diffusion coefficient reproducibility compared with 1.5 Tesla. J Magn Reson Imaging 2011;33:128e35. 22. Dale BM, Braithwaite AC, Boll DT, et al. Field strength and diffusion encoding technique affect the apparent diffusion coefficient measurements in diffusion-weighted imaging of the abdomen. Invest Radiol 2010;45:104e8. 23. Saremi F, Jalili M, Sefidbakht S, et al. Diffusion-weighted imaging of the abdomen at 3 T: image quality comparison with 1.5-T magnet using 3 different imaging sequences. J Comput Assist Tomogr 2011;35:317e25. 24. Momtahen AJ, Balci NC, Alkaade S, et al. Focal pancreatitis mimicking pancreatic mass: magnetic resonance imaging (MRI)/magnetic resonance
25.
26.
27.
28.
29.
30.
e197
cholangiopancreatography (MRCP) findings including diffusion-weighted MRI. Acta Radiol 2008;49:490e7. Zhang JL, Sigmund EE, Rusinek H, et al. Optimization of b-value sampling for diffusion-weighted imaging of the kidney. Magn Reson Med 2012;67:89e97. Park SO, Kim JK, Kim KA, et al. Relative apparent diffusion coefficient: determination of reference site and validation of benefit for detecting metastatic lymph nodes in uterine cervical cancer. J Magn Reson Imaging 2009;29:383e90. Lin G, Ho KC, Wang JJ, et al. Detection of lymph node metastasis in cervical and uterine cancers by diffusion-weighted magnetic resonance imaging at 3T. J Magn Reson Imaging 2008;28:128e35. Rosenkrantz AB, Kopec M, Kong X, et al. Prostate cancer vs. post-biopsy hemorrhage: diagnosis with T2- and diffusion-weighted imaging. J Magn Reson Imaging 2010;31:1387e94. Ginat DT, Mangla R, Yeaney G, et al. Diffusion-weighted imaging for differentiating benign from malignant skull lesions and correlation with cell density. AJR Am J Roentgenol 2012;198:W597e601. Yin B, Liu L, Zhang BY, et al. Correlating apparent diffusion coefficients with histopathologic findings on meningiomas. Eur J Radiol 2012;81:4050e6.
Contents lists available at SciVerse ScienceDirect
Clinical Radiology journal homepage: www.clinicalradiologyonline.net
Pancreatic cancer: Lack of association between apparent diffusion coefficient values and adverse pathological features A.B. Rosenkrantz a, *, B.W. Matza a, A. Sabach a, C.H. Hajdu b, N. Hindman a a b
Department of Radiology, NYU Langone Medical Center, 550 First Avenue, New York, NY 10016, USA Department of Pathology, NYU Langone Medical Center, 550 First Avenue, New York, NY 10016, USA
art icl e i nformat ion Article history: Received 26 September 2012 Received in revised form 12 November 2012 Accepted 20 November 2012
AIM: To identify retrospectively potential associations between apparent diffusion coefficient (ADC) values of pancreatic adenocarcinoma and tumour grade as well as other pathological features, using histopathological assessment from the Whipple procedure as the reference standard. MATERIALS AND METHODS: Thirty patients with pancreatic adenocarcinoma underwent magnetic resonance imaging (MRI) including diffusion-weighted imaging with b-values of 0 and 500 s/mm2 before the Whipple procedure. Two radiologists independently recorded the ADC values of the tumour and benign pancreas for all cases. ADC values were compared with histopathological findings following the Whipple procedure. RESULTS: The intra-class correlation coefficient was 0.689 for benign pancreas and 0.695 for tumours, indicating good inter-reader agreement for ADC values. The mean ADC value was significantly lower in tumours than in benign pancreas for both readers (reader 1: 1.74 0.34 103 mm2/s versus 2.08 0.48 103 mm2/s, respectively, p ¼ 0.006; reader 2: 1.69 0.41 103 mm2/s versus 2.11 0.54 103 mm2/s, respectively, p < 0.001). However, there was no significant difference in mean ADC between poorly and well/moderately differentiated tumours for either reader (reader 1: 1.69 0.36 103 mm2/s versus 1.78 0.33 103 mm2/s, respectively, p ¼ 0.491; reader 2: 1.62 0.33 103 mm2/s versus 1.75 0.49 103 mm2/s, respectively, p ¼ 0.405). The area under the curve (AUC) for differentiation of poorly and well/moderately differentiated tumours was 0.611 and 0.596 for readers 1 and 2, respectively, and was not significantly better than an AUC of 0.500 for either reader (p 0.306). In addition, ADC was not significantly different for either reader between tumours with stage T3 versus stage T1/T2, between tumours with and without metastatic peri-pancreatic lymph nodes, or between tumours located in the pancreatic head versus other pancreatic regions (p 0.413). CONCLUSION: No associations between ADC values of pancreatic adenocarcinoma and tumour grade or other adverse pathological features were observed. Ó 2012 The Royal College of Radiologists. Published by Elsevier Ltd. All rights reserved.
* Guarantor and correspondent: A.B. Rosenkrantz, Department of Radiology, NYU Langone Medical Center, 550 First Avenue, New York, NY 10016, USA. Tel.: þ1 212 263 0232; fax: þ1 212 263 6634. E-mail address: [email protected] (A.B. Rosenkrantz). 0009-9260/$ e see front matter Ó 2012 The Royal College of Radiologists. Published by Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.crad.2012.11.006
e192
A.B. Rosenkrantz et al. / Clinical Radiology 68 (2013) e191ee197
Introduction
MRI
Pancreatic adenocarcinoma represents the most common malignancy of the pancreas and the fourth most common cause of cancer-related death in men and women in the Unites States.1 Despite its rapidly progressive disease course and overall very poor survival even with aggressive intervention, long-term survival has been shown to be possible in the setting of early detection.2 Furthermore, accurate characterization of the aggressiveness of newly detected cases may have a role in establishing patient prognosis and treatment selection. Although a spectrum of pathological features, including tumour size, grade, stage, and nodal status, significantly influence patient survival, accurate prediction of these features at the time of initial diagnosis remains a substantial challenge.2e5 Diffusion-weighted imaging (DWI) is a quantitative magnetic resonance imaging (MRI) technique that reflects tissue cellularity and has been employed to assist the detection and characterization of aggressiveness of a variety of neoplasms.6 Apparent diffusion coefficient (ADC) values obtained from DWI have generally been shown to be lower in malignant than in benign tissue, and lower in higher than lower-grade tumours.7e9 Indeed, at least nine studies have demonstrated significantly lower ADC in pancreatic adenocarcinoma than in benign pancreas tissue.10e18 Yet, the role of ADC values in predicting adverse pathological features of pancreatic cancer is less established. For instance, to the authors’ knowledge, only a single previous study has shown a significant association between the ADC value and pathological grade of pancreatic cancer.19 However, in our clinical practice, such a relationship has not been observed. Thus, in view of the paucity of data on this topic, the present study was conducted to identify potential associations between the ADC value of pancreatic adenocarcinoma and tumour grade, as well as other adverse pathological features, using histopathological assessment from the Whipple procedure as the reference standard.
MRI was performed using a 1.5 T whole-body system (Magnetom Sonata, Symphony, or Avanto; Siemens Healthcare, Erlangen, Germany) and anterior and posterior torso phased-array coils. Examinations included two-dimensional (2D) axial in- and opposed-phase gradient-echo T1weighted sequences, 2D axial and/or coronal half-Fourier acquisition single-shot turbo-spin echo T2-weighted sequences, 2D axial fat-suppressed turbo spin-echo T2weighted sequences, and dynamic three-dimensional (3D) fat-suppressed gradient-echo T1-weighted sequence, performed before and during the arterial, portal venous, and equilibrium phases after the intravenous administration of 0.1 mmol/kg gadolinium-chelate. These sequences were not formally evaluated as part of this study. In addition, all examinations included an axial single-shot echo-planar imaging DWI sequence with tri-directional motion-probing gradients [1600e3400 ms repetition time (TR), 67e82 ms echo time (TE), 300e400 mm field of view (FOV) with 70e80% rectangular FOV, 144 192 matrix, 7e8 mm section thickness; navigator-trigger technique on the Avanto scanner, breath-hold technique on the Symphony and Sonata machines]. The ADC map was constructed for all cases on a voxel-by-voxel basis using b-values of 0 and 500 s/mm2.
Materials and methods Patients The present retrospective study was Health Insurance Portability and Accountability Act (HIPAA)-compliant and approved by the institutional review board, with a waiver of the requirement for written informed consent. A departmental database was searched to identify patients with pathologically confirmed pancreatic adenocarcinoma who underwent 1.5 T MRI of the pancreas that included a DWI sequence incorporating b-values of 0 and 500 s/mm2 prior to the Whipple procedure. MRI examinations were performed between 1 January 2006 and 31 December 2011. A total of 30 patients (19 male, 11 female; mean age 68 13 years) were identified. The mean delay between MRI and the Whipple procedure was 31 25 days.
Image analysis Images were evaluated independently by two fellowship-trained radiologists with 4 and 6 years of experience, using a picture archiving and communication system (PACS; iSite, Philips Healthcare). The radiologists were aware that all cases had undergone subsequent surgery for pancreatic adenocarcinoma, but were unaware of other pathological findings. The radiologists reviewed the ADC map for all patients in conjunction with the conventional sequences in order to localize each tumour on the ADC map. A free-hand regionof-interest (ROI) was traced along the edge of each lesion on the ADC map, slightly inside its outer margin to avoid partial volume averaging artefact. The ADC of normal pancreatic parenchyma was also measured for each case by placing a circular ROI within the mid-body of the pancreas, taking care to avoid lesions, vessels, and artefacts. All benign and malignant ROIs were placed twice, and the average value of the two ROIs recorded for subsequent data analysis.
Reference standard All patients had undergone a Whipple procedure following MRI. The following histopathological features were then recorded for each case: pathologically determined tumour size; grade (classified in a binary fashion as poorly differentiated versus well/moderately differentiated, defined in terms of most aggressive elements present); T stage (classified in a binary fashion as stage T3 versus stages T1/T2,); and nodal status (classified in a binary fashion in terms of whether metastatic peri-
A.B. Rosenkrantz et al. / Clinical Radiology 68 (2013) e191ee197
pancreatic lymph nodes were identified). Well and moderately differentiated tumours, as well as stage T1 and T2 tumours, were grouped for the analyses of tumour grade and stage, respectively, given the infrequency of well-differentiated and stage T1 tumours. Finally, in view of potential worse survival for pancreatic adenocarcinoma located in the pancreatic head,3 it was recorded whether each tumour was located in the pancreatic head versus the body or tail of the pancreas. The presence of perineural invasion was not evaluated given that nearly the entire sample (28 of 30 cases) exhibited this pathological feature.
e193
Comparison of ADC values between benign and malignant pancreas Table 1 summarizes findings in terms of mean ADC values of benign pancreas and pancreatic adenocarcinoma for each reader. Mean ADC values were significantly lower for pancreatic tumours than benign pancreas for both readers (reader 1: 1.74 0.34 103 mm2/s versus 2.08 0.48 103 mm2/s, respectively, p ¼ 0.006; reader 2: 1.69 0.41 103 mm2/s versus 2.11 0.54 103 mm2/s, respectively, p < 0.001).
Association between ADC values and tumour grade Statistical analysis Statistical analysis was performed using software (Medcalc for Windows, version 11.1.1.0; Medcalc Software, Mariakerke, Belgium). As a measure of intra-reader agreement, the intra-class correlation coefficient (ICC) of measured ADC values was calculated for the two readers for both benign pancreas and pancreatic adenocarcinoma. ICC was interpreted as follows: <0.04, poor agreement; 0.4e0.6, good agreement; 0.6e0.8, moderate agreement; >0.8, excellent agreement. A paired t-test was used to compare ADC values between pancreatic tumours and the benign pancreas. A series of unpaired t-tests were used to compare ADC values between the following groups: poorly and well/moderately differentiated tumours; stage T3 and stage T1/T2 tumours; tumours with and without metastatic peri-pancreatic lymph nodes; and tumours located in the pancreatic head versus body or tail. Receiver operating characteristic curve (ROC) analysis was used to calculate the area under the curve (AUC) of the ADC values for differentiating poorly and well/moderately differentiated cases. Finally, Pearson’s correlation coefficient was used to assess the association between ADC values and pathologically determined tumour size. These tests were performed separately for the two readers. All comparisons were two-sided and considered statistically significant at p < 0.05.
Results Tumours The 30 tumours had a mean size at histopathological analysis of 25 8 mm. Among the 30 tumours, 15 (50%) were poorly differentiated, 25 (83%) were stage T3, 22 (73%) had metastatic peri-pancreatic lymph nodes, and 23 (77%) were located in the pancreatic head. Sample cases are shown in Figs 1 and 2.
Inter-reader agreement ICC between the two readers in terms of ADC values was 0.689 for benign pancreas and 0.695 for tumours, consistent with good inter-reader agreement.
Table 2 summarizes findings in terms of mean ADC of pancreatic adenocarcinoma, stratified by various pathological features. There was no significant difference in mean ADC between poorly and well/moderately differentiated tumours for either reader (reader 1: 1.69 0.36 103 mm2/s versus 1.78 0.33 103 mm2/s, respectively, p ¼ 0.491; reader 2: 1.62 0.33 103 mm2/s versus 1.75 0.49 103 mm2/s, respectively, p ¼ 0.405). At ROC analysis, the AUC for differentiation of poorly and well/ moderately differentiated tumours was 0.611 and 0.596 for readers 1 and 2, respectively. For neither reader was the AUC significantly better than the diagonal line corresponding with random guessing and having an AUC of 0.500 (p 0.306).
Association between ADC values and other prognostic features There was no significant correlation between ADC value and tumour size for either reader (reader 1: r ¼ 0.072, p ¼ 0.706; reader 2: r ¼ 0.024, p ¼ 0.901). In addition, there was no significant difference in ADC value between tumours with stage T3 versus stage T1/T2, between tumours with and without metastatic peri-pancreatic lymph nodes, or between tumours located in the head versus other regions of the pancreas (reader 1: p 0.543; reader 2: p 0.413; Table 2).
Discussion In this study, no associations between ADC values of pancreatic adenocarcinoma and a spectrum of adverse pathological features were found when using findings from the Whipple procedure as the reference standard. Specifically, no association was observed between ADC and any of tumour size, grade, T stage, nodal status, or location. The lack of any apparent association was noted for two independent radiologists who had good inter-reader agreement in terms of recorded ADC values. Thus, based on the present data, it seems that ADC values are currently unlikely to be of clinical value for the non-invasive prediction of adverse pathological features of newly detected cases of pancreatic cancer. It is relevant that both radiologists observed significantly lower ADC values in pancreatic adenocarcinoma than in
e194
A.B. Rosenkrantz et al. / Clinical Radiology 68 (2013) e191ee197
Figure 1 A 75-year-old man with adenocarcinoma of the pancreatic body. (a) Axial half-Fourier acquisition single-shot turbo spin-echo T2weighted image shows a solid mass in the pancreatic body (arrow). (b) Axial fat-suppressed contrast-enhanced gradient-echo T1-weighted image shows hypovascularity of the mass (arrow). (c) Axial diffusion-weighted image with a b-value of 500 s/mm2 shows discrete enlargement of the pancreas in this region (arrow) without clear hyperintensity relative to the remainder of the pancreas. (d) ADC map shows foci of decreased ADC within the lesion (arrow). ADC of the mass, when averaged between the two readers, was 1.28 103 mm2/s, lower than the mean ADC value of all cases of pancreatic adenocarcinoma within the study cohort. At histopathological assessment following a Whipple procedure, the mass was found to be a stage T2 moderately differentiated adenocarcinoma without metastatic peri-pancreatic lymph nodes.
benign pancreas. This identified difference for both radiologists demonstrates an ability of the DWI sequence employed in the present study to distinguish groups of tissues, despite the failure to observe differences between subgroups of pancreatic adenocarcinoma. This difference between benign and malignant pancreatic regions has been consistently reported in previous studies.10,13,16 In addition, Muraoka et al.12 showed correlations between ADC values of pancreatic tumours and the presence of collagenous fibres, thus hypothesizing that such fibres may account for the observed abnormal diffusion behaviour in pancreatic tumours.12 To the authors’ knowledge, one previous study has reported significantly lower ADC in cases of pancreatic adenocarcinoma that are poorly differentiated in comparison with well/moderately differentiated lesions.20 In that study, Wang et al.20 reported mean ADC values in these two groups of 1.46 0.17 103 and 2.10 0.42 103 mm2/s, respectively. As in the present study, these authors also
used a maximal b-value of 500 s/mm2 and field strength of 1.5 T. However, in contrast to the findings of the present study and numerous other prior studies in the literature, Wang et al. failed to observe a significantly lower ADC value in pancreatic cancer than in benign pancreas. In fact, in their study, well- and moderately differentiated pancreatic cancer exhibited a higher ADC value than benign pancreas, although this difference was not statistically significant. This suggests that there may be underlying pathological differences between the lesions included in the cohorts of Wang et al. and the present cohort. Furthermore, the two cases of well- or moderately differentiated pancreatic adenocarcinoma in their study with dense fibrosis exhibited ADC values that overlapped the ADC values of poorly differentiated tumours, rather than overlapping the ADC values of other low-grade tumours. This finding also suggests heterogeneity among subgroups of pancreatic adenocarcinoma and indicates the complexity of efforts to correlate the ADC values of the lesions with these broad
A.B. Rosenkrantz et al. / Clinical Radiology 68 (2013) e191ee197
e195
Figure 2 A 67-year-old man with adenocarcinoma of the pancreatic head. (a) Axial half-Fourier acquisition single-shot turbo spin-echo T2weighted image shows a solid mass in the pancreatic head (arrow). (b) Axial fat-suppressed contrast-enhanced gradient-echo T1-weighted image shows hypovascularity of the mass (arrow). (c) Axial diffusion-weighted image with b-value of 500 s/mm2 shows mild increased signal intensity within the region of the mass (arrow). (d) ADC map shows mild decreased ADC within the lesion (arrow). ADC of the mass, when averaged between the two readers, was 1.90 103 mm2/s, greater than the mean ADC value of all cases of pancreatic adenocarcinoma within the study cohort. At histopathological assessment following a Whipple procedure, the mass was found to be a stage T3 poorly differentiated adenocarcinoma with metastatic peri-pancreatic lymph nodes.
categories. Nonetheless, in comparison with the study of Wang et al.,20 the present study entailed a slightly larger sample size and the reporting of independent results from two separate observers. Given the overall paucity of existing
Table 1 Comparison of apparent diffusion coefficient (ADC) values between benign pancreas and pancreatic adenocarcinoma for two readers. Reader 1 a
Benign pancreas Pancreatic adenocarcinoma a
Reader 2
ADC (103 mm2/s)
pValue
ADCa (103 mm2/s)
pValue
2.08 0.48 1.74 0.34
0.006
2.11 0.54 1.69 0.41
<0.001
Data are mean SD.
data and differences in findings between the two available studies, additional studies seem warranted to further investigate this issue. In addition to previously noted pathological factors that may impact ADC values, it is also important to note the role of the MRI technique itself in influencing ADC measurements. In particular, field strength, strategy for respiratory compensation, number of b-values used, maximal b-value selected, and ADC measurement technique, all warrant consideration. In the present study, all cases were imaged at 1.5 T. Although two previous studies showed no difference in pancreatic ADC values between 1.5 and 3 T,21,22 there may be an increase in artefacts and associated decrease in overall image quality for abdominal DWI at 3 T depending on the specific hardware and DWI sequence used,21e23 such that it is possible that DWI of the abdomen may be more robust at
e196
A.B. Rosenkrantz et al. / Clinical Radiology 68 (2013) e191ee197
Table 2 Comparison of apparent diffusion coefficient (ADC) values between subgroups of pancreatic adenocarcinoma for two readers. Reader 1
Reader 2
a
pADC pADC (103 mm2/s) Value (103 mm2/s) Value Grade Well/moderately differentiated Poorly differentiated T stage T1/T2 T3 Metastatic peripancreatic nodes Present Absent Location Head Elsewhere in pancreas a
1.78 0.33
0.491 1.75 0.49
1.69 0.36
1.62 0.33
1.70 0.42 1.75 0.33
0.425 1.81 0.45 1.66 0.41
0.679
1.71 0.31 1.80 0.43
0.543 1.68 0.43 1.70 0.42
0.912
1.72 0.33 1.80 0.42
0.576 1.65 0.42 1.80 0.41
0.413
0.405
Data are mean SD.
1.5 T at the present time. In addition, ADC maps were calculated based on two b-values of 0 and 500 s/mm2. Given its central location within the abdomen, the signal-to-noise ratio (SNR) within the pancreas may be poor at comparatively high b-values for certain hardware and software arrangements, which in turn can result in imprecise ADC calculations. Thus, a maximal b-value of 500 s/mm2 was selected to achieve sufficient SNR within the pancreas on the high b-value images for all systems and DWI sequences employed. Numerous prior studies of the role of ADC values in pancreatic cancer have also used a similar maximal b-value.12,13,18,20,24 In addition, the use of only two b-values in the present study is consistent with the DWI technique in numerous past studies.12,13,18,20,24 A greater number of b-values would have required longer acquisition time. In addition, a prior study that used Monte Carlo simulations to identify the optimal strategy for b-value selection in the kidneys, observed that increasing the number of b-values did not improve precision of DWI metrics in comparison with increasing the number of averages of the selected b-values.25 Furthermore, patients were imaged using either breath-hold or navigator-triggered DWI technique, depending on the given machine. To the authors’ knowledge, no previous study has compared ADC values within the pancreas in the same patients between these two techniques; thus, this remains an important area for future investigation to assist optimization of pancreatic DWI technique. Finally, some authors have suggested using normalized or relative ADC values, rather than absolute ADC measurements, for optimal tissue characterization.26,27 However, several studies failed to show improved diagnostic performance from this measurement strategy,28e30 and this approach was not explored in the present study. The biggest limitation of this study is uncertainty as to the basis for the differences in findings compared with the study of Wang et al.20 As previously noted, it is possible that histological differences between cases included in the two studies account for the discrepant conclusions. Additional
limitations of this study include its retrospective nature, relatively small sample size (despite larger sample size than in the study by Wang et al.), and assessment of only a single maximal b-value. In conclusion, although ADC values are significantly different between benign pancreas and pancreatic adenocarcinoma, no associations between ADC values and tumour grade or other adverse pathological features were observed. As the present findings differ from those of a previous study on this topic, additional studies are warranted. Until such studies are conducted, caution is warranted in the clinical use of ADC values to attempt to predict the prognosis of newly diagnosed pancreatic adenocarcinoma.
References 1. Jemal A, Siegel R, Ward E, et al. Cancer statistics, 2009. CA Cancer J Clin 2009;59:225e49. 2. Cleary SP, Gryfe R, Guindi M, et al. Prognostic factors in resected pancreatic adenocarcinoma: analysis of actual 5-year survivors. J Am Coll Surg 2004;198:722e31. 3. Sohn TA, Yeo CJ, Cameron JL, et al. Resected adenocarcinoma of the pancreasd616 patients: results, outcomes, and prognostic indicators. J Gastrointest Surg 2000;4:567e79. 4. Perini MV, Montagnini AL, Jukemura J, et al. Clinical and pathologic prognostic factors for curative resection for pancreatic cancer. HPB (Oxford) 2008;10:356e62. 5. Niwa T, Ueno M, Ohkawa S, et al. Advanced pancreatic cancer: the use of the apparent diffusion coefficient to predict response to chemotherapy. Br J Radiol 2009;82:28e34. 6. Koh DM, Collins DJ. Diffusion-weighted MRI in the body: applications and challenges in oncology. AJR Am J Roentgenol 2007;188:1622e35. 7. Guo AC, Cummings TJ, Dash RC, et al. Lymphomas and high-grade astrocytomas: comparison of water diffusibility and histologic characteristics. Radiology 2002;224:177e83. 8. Rahbar H, Partridge SC, Demartini WB, et al. In vivo assessment of ductal carcinoma in situ grade: a model incorporating dynamic contrastenhanced and diffusion-weighted breast MR imaging parameters. Radiology 2012;263:374e82. 9. Vargas HA, Akin O, Franiel T, et al. Diffusion-weighted endorectal MR imaging at 3 T for prostate cancer: tumour detection and assessment of aggressiveness. Radiology 2011;259:775e84. 10. Matsuki M, Inada Y, Nakai G, et al. Diffusion-weighed MR imaging of pancreatic carcinoma. Abdom Imaging 2007;32:481e3. 11. Lee SS, Byun JH, Park BJ, et al. Quantitative analysis of diffusion-weighted magnetic resonance imaging of the pancreas: usefulness in characterizing solid pancreatic masses. J Magn Reson Imaging 2008;28:928e36. 12. Muraoka N, Uematsu H, Kimura H, et al. Apparent diffusion coefficient in pancreatic cancer: characterization and histopathological correlations. J Magn Reson Imaging 2008;27:1302e8. 13. Fattahi R, Balci NC, Perman WH, et al. Pancreatic diffusion-weighted imaging (DWI): comparison between mass-forming focal pancreatitis (FP), pancreatic cancer (PC), and normal pancreas. J Magn Reson Imaging 2009;29:350e6. 14. Lemke A, Laun FB, Klauss M, et al. Differentiation of pancreas carcinoma from healthy pancreatic tissue using multiple b-values: comparison of apparent diffusion coefficient and intravoxel incoherent motion derived parameters. Invest Radiol 2009;44:769e75. 15. Fukukura Y, Takumi K, Kamimura K, et al. Pancreatic adenocarcinoma: variability of diffusion-weighted MR imaging findings. Radiology 2012;263:732e40. 16. Wiggermann P, Grutzmann R, Weissenbock A, et al. Apparent diffusion coefficient measurements of the pancreas, pancreas carcinoma, and mass-forming focal pancreatitis. Acta Radiol 2012;53:135e9. 17. Kamisawa T, Takuma K, Anjiki H, et al. Differentiation of autoimmune pancreatitis from pancreatic cancer by diffusion-weighted MRI. Am J Gastroenterol 2010;105:1870e5.
A.B. Rosenkrantz et al. / Clinical Radiology 68 (2013) e191ee197 18. Kartalis N, Lindholm TL, Aspelin P, et al. Diffusion-weighted magnetic resonance imaging of pancreas tumours. Eur Radiol 2009;19:1981e90. 19. Wang Y, Miller FH, Chen ZE, et al. Diffusion-weighted MR imaging of solid and cystic lesions of the pancreas. RadioGraphics 2011;31:E47e64. 20. Wang Y, Chen ZE, Nikolaidis P, et al. Diffusion-weighted magnetic resonance imaging of pancreatic adenocarcinomas: association with histopathology and tumour grade. J Magn Reson Imaging 2011;33:136e42. 21. Rosenkrantz AB, Oei M, Babb JS, et al. Diffusion-weighted imaging of the abdomen at 3.0 Tesla: image quality and apparent diffusion coefficient reproducibility compared with 1.5 Tesla. J Magn Reson Imaging 2011;33:128e35. 22. Dale BM, Braithwaite AC, Boll DT, et al. Field strength and diffusion encoding technique affect the apparent diffusion coefficient measurements in diffusion-weighted imaging of the abdomen. Invest Radiol 2010;45:104e8. 23. Saremi F, Jalili M, Sefidbakht S, et al. Diffusion-weighted imaging of the abdomen at 3 T: image quality comparison with 1.5-T magnet using 3 different imaging sequences. J Comput Assist Tomogr 2011;35:317e25. 24. Momtahen AJ, Balci NC, Alkaade S, et al. Focal pancreatitis mimicking pancreatic mass: magnetic resonance imaging (MRI)/magnetic resonance
25.
26.
27.
28.
29.
30.
e197
cholangiopancreatography (MRCP) findings including diffusion-weighted MRI. Acta Radiol 2008;49:490e7. Zhang JL, Sigmund EE, Rusinek H, et al. Optimization of b-value sampling for diffusion-weighted imaging of the kidney. Magn Reson Med 2012;67:89e97. Park SO, Kim JK, Kim KA, et al. Relative apparent diffusion coefficient: determination of reference site and validation of benefit for detecting metastatic lymph nodes in uterine cervical cancer. J Magn Reson Imaging 2009;29:383e90. Lin G, Ho KC, Wang JJ, et al. Detection of lymph node metastasis in cervical and uterine cancers by diffusion-weighted magnetic resonance imaging at 3T. J Magn Reson Imaging 2008;28:128e35. Rosenkrantz AB, Kopec M, Kong X, et al. Prostate cancer vs. post-biopsy hemorrhage: diagnosis with T2- and diffusion-weighted imaging. J Magn Reson Imaging 2010;31:1387e94. Ginat DT, Mangla R, Yeaney G, et al. Diffusion-weighted imaging for differentiating benign from malignant skull lesions and correlation with cell density. AJR Am J Roentgenol 2012;198:W597e601. Yin B, Liu L, Zhang BY, et al. Correlating apparent diffusion coefficients with histopathologic findings on meningiomas. Eur J Radiol 2012;81:4050e6.