This copy is for personal use only. To order printed copies, contact [email protected] Original Research  n  Gastrointestinal

Pancreatic Steatosis and Fibrosis: Quantitative Assessment with Preoperative Multiparametric MR Imaging1 Purpose:

To evaluate the diagnostic performance of multiparametric pancreatic magnetic resonance (MR) imaging, including the T2*-corrected Dixon technique and intravoxel incoherent motion (IVIM) diffusion-weighted (DW) imaging, in the quantification of pancreatic steatosis and fibrosis, with histologic analysis as the reference standard, and to determine the relationship between MR parameters and postoperative pancreatic fistula.

Materials and Methods:

This retrospective study was approved by the institutional review board, and the informed consent requirement was waived. A total of 165 patients (93 men, 72 women; mean age, 62 years) underwent preoperative 3-T MR imaging and subsequent pancreatectomy (interval, 0–77 days). Fat fractions, IVIM DW imaging parameters (true diffusion coefficient [D], pseudodiffusion coefficient [D*], and perfusion fraction [f]), pancreas-to-muscle signal intensity ratios on unenhanced T1-weighted images, and pancreatic duct sizes were compared with the fat fractions and fibrosis degrees (F0–F3) of specimens. In 95 patients who underwent pancreatoenteric anastomosis, MR parameters were compared between groups with clinically relevant postoperative pancreatic fistula and those without. The relationship between postoperative pancreatic fistula and MR parameters was evaluated by using logistic regression analysis.

Results:

Fat fractions at MR imaging showed a moderate relationship with histologic findings (r = 0.71; 95% confidence interval: 0.63, 0.78). Patients with advanced fibrosis (F2–F3) had lower D*([39.72 6 13.64] 31023mm2/sec vs [32.50 6 13.09] 31023mm2/sec [mean 6 standard deviation], P = .004), f (29.77% 6 8.51 vs 20.82% 6 8.66, P , .001), and unenhanced T1-weighted signal intensity ratio (1.43 6 0.26 vs 1.21 6 0.30, P , .001) than did patients with F0–F1 disease. Clinically relevant fistula developed in 14 (15%) of 95 patients, and f was significantly associated with postoperative pancreatic fistula (odds ratio, 1.17; 95% confidence interval: 1.05, 1.30).

Conclusion:

Multiparametric MR imaging of the pancreas, including imaging with the T2*-corrected Dixon technique and IVIM DW imaging, may yield quantitative information regarding pancreatic steatosis and fibrosis, and f was shown to be significantly associated with postoperative pancreatic fistulas.

1

 From the Departments of Radiology (J.H.Y., J.M.L., J.K.H., B.I.C.), Pathology (K.B.L.), and Surgery (S.W.K., M.J.K., J.Y.J.), Seoul National University Hospital, 101 Daehak-ro, Jongno-gu, Seoul 110-744, Korea; Department of Radiology, Seoul National University College of Medicine, Seoul, Korea (J.H.Y., J.M.L., J.K.H., B.I.C.); Institute of Radiation Medicine, Seoul National University Medical Research Center, Seoul, Korea (J.M.L., J.K.H., B.I.C.); and Siemens Healthcare, Erlangen, Germany (S.K.). Received September 22, 2014; revision requested November 10; revision received June 12, 2015; accepted July 15; final version accepted August 18. Address correspondence to J.M.L. (e-mail: [email protected] ).

Imaging

Jeong Hee Yoon, MD Jeong Min Lee, MD Kyung Bun Lee, MD Sun-Whe Kim, MD Mee Joo Kang, MD Jin-Young Jang, MD Stephan Kannengiesser, PhD Joon Koo Han, MD Byung Ihn Choi, MD

 RSNA, 2015

q

Online supplemental material is available for this article.

 RSNA, 2015

q

Radiology: Volume 000: Number 0—   2016  n  radiology.rsna.org

1

GASTROINTESTINAL IMAGING: Pancreatic Steatosis and Fibrosis

P

ancreatoduodenectomy remains the only potentially curative therapy for periampullary malignant tumors (1); however, even though surgical morbidity has dramatically decreased in the past several decades (2,3), postoperative morbidity stemming from complications has remained high (4,5). Previous studies have shown that postoperative pancreatic fistula is one of the complications that often results in morbidity after pancreas resection (6–8), with a reported incidence of 5%–40% (9) and a doubled risk of mortality in patients with postoperative pancreatic fistula (7,8). Thus, it is important to preoperatively estimate the risk of postoperative pancreatic fistula to determine a patient’s suitability for surgery and to adjust the postoperative management strategy (10). To date, there have been several attempts to predict postoperative pancreatic fistula based on demographic findings, intraoperative texture assessment, and postoperative histologic findings (6,11). Several pancreatic glandular factors, such as soft-tissue texture,

Advances in Knowledge nn Fat fractions estimated with the T2*-corrected Dixon technique showed moderate correlation with histologically assessed pancreas fat (r = 0.71; 95% confidence interval: 0.63, 0.78). nn In patients with advanced pancreatic fibrosis (F2–F3), perfusionrelated parameters (pseudodiffusion coefficient [D*], [32.50 6 13.09] 31023mm2/sec; perfusion fraction [f], 20.82% 6 8.66) of intravoxel incoherent motion (IVIM) diffusion-weighted (DW) imaging significantly decreased compared with those in patients with no or early pancreatic fibrosis (F0–F1) (D*, [39.72 6 13.64] 31023mm2/sec [P = .004]; f, 29.77% 6 8.51 [P , .001]). nn The odds of developing clinically relevant postoperative pancreatic fistula for a 1% increase of f were 1.17 (95% confidence interval: 1.05, 1.30). 2

Yoon et al

pancreatic steatosis, absence of fibrosis, and small pancreatic duct size, have been reported to be associated with an increased rate of postoperative pancreatic fistula (11,12). Thus, if these factors could be identified at preoperative imaging assessment, it would provide value in the preoperative risk analysis (10,13). However, only a few studies on the preoperative prediction of postoperative pancreatic fistula have been performed (11,13,14) due to the limited capability of imaging modalities to depict pancreatic parenchymal changes, such as steatosis and fibrosis. Relatively recently, magnetic resonance (MR) imaging has been used with increasing frequency to evaluate various pancreatic diseases. In addition, several studies have shown that the multiparametric MR approach, including fat quantification and diffusionweighted (DW) imaging with the intravoxel incoherent motion model (IVIM) may enable quantitative assessment of diffuse parenchymal diseases, such as steatosis and fibrosis, in abdominal solid organs (14–18). On the basis of the promising results of these studies, we hypothesized that multiparametric MR imaging may enable accurate quantification of pancreatic fibrosis and steatosis, which have been shown to be associated with postoperative pancreatic fistula (12,19,20). Thus, the purpose of this study was to (a) evaluate the diagnostic performance of multiparametric pancreatic

Implications for Patient Care nn Fat fractions obtained by using the T2*-corrected Dixon method and perfusion-related IVIM DW imaging parameters (D* and f) may be useful in the noninvasive assessment of pancreatic steatosis and fibrosis. nn The perfusion fraction of IVIM DW imaging may serve as an imaging biomarker that can be used to estimate the increasing risk of postoperative pancreatic fistula in patients who undergo pancreatoenteric anastomosis after pancreatectomy.

MR imaging, including the T2*-corrected Dixon technique and IVIM DW imaging, in the quantification of pancreatic steatosis and fibrosis, with histologic findings as the reference standard, and (b) determine the relationship between MR parameters and postoperative pancreatic fistula.

Materials and Methods Study Population This retrospective study was approved by the Seoul National University institutional review board, and the requirement for informed consent was waived. A Siemens Healthcare employee (S.K.) provided technical support for implementation and optimization of sequences in our study. The authors not associated with Siemens Healthcare (J.H.Y., J.M.L., K.B.L., S.W.K., M.J.K., J.Y.J., J.K.H., B.I.C.) maintained full control of the data. From June 2011 to December 2012, 166 patients underwent MR imaging of the pancreas and subsequent pancreatectomy at our institution

Published online before print 10.1148/radiol.2015142254  Content code: Radiology 2016; 000:1–11 Abbreviations: ADC = apparent diffusion coefficient AUC = area under the receiver operating characteristic curve CI = confidence interval D = true diffusion coefficient D* = pseudodiffusion coefficient DW = diffusion weighted f = perfusion fraction IVIM = intravoxel incoherent motion ROI = region of interest Author contributions: Guarantors of integrity of entire study, J.M.L., J.K.H.; study concepts/study design or data acquisition or data analysis/interpretation, all authors; manuscript drafting or manuscript revision for important intellectual content, all authors; approval of final version of submitted manuscript, all authors; agrees to ensure any questions related to the work are appropriately resolved, all authors; literature research, J.H.Y., J.M.L., K.B.L., J.K.H.; clinical studies, J.H.Y., J.M.L., K.B.L., S.W.K., M.J.K., J.Y.J.; statistical analysis, J.H.Y.; and manuscript editing, J.H.Y., J.M.L., M.J.K., J.Y.J., S.K., J.K.H., B.I.C. Conflicts of interest are listed at the end of this article.

radiology.rsna.org  n Radiology: Volume 000: Number 0—   2016

GASTROINTESTINAL IMAGING: Pancreatic Steatosis and Fibrosis

(median interval, 2 days; range, 0–77 days). One male patient was excluded from the study owing to previous pancreatectomy and severe atrophy of the remnant pancreas. Finally, 165 patients (mean age, 62.2 years; age range, 29– 86 years) consisting of 73 men (mean age, 62.8 years; age range, 29–81 years) and 92 women (mean age, 61.3 years; age range, 29–86 years) were included in our study. Patients underwent pancreatectomy for the following indications: pancreatic adenocarcinoma (n = 47), pancreatic neuroendocrine tumor (n = 23), intraductal papillary mucinous neoplasm (n = 32), pancreatic cyst (n = 17 [benign cyst, n = 2; serous cystadenoma, n = 8; mucinous cystadenoma, n = 7), common bile duct cancer (n = 22), ampulla of Vater cancer (n = 19), benign biliary stricture (n = 2), and miscellaneous diseases (n = 3). The following types of surgery were performed in the 165 patients: pylorus-preserving pancreatoduodenectomy (n = 72), pancreatoduodenectomy (n = 16), central pancreatectomy (n = 7), distal pancreatectomy (n = 56), total pancreatectomy (n = 6), subtotal pancreatectomy (n = 3), and enucleation (n = 5) (Fig 1).

MR Image Acquisition MR protocol.—MR examinations were performed with a 3-T imager (Magnetom Verio; Siemens Healthcare, Erlangen, Germany) and a 32-channel phased-array coil. Our routine MR protocol consisted of the following sequences: (a) unenhanced T1- and heavily T2-weighted imaging, (b) dual-echo imaging with the fast lowangle shot sequence, (c) MR cholangiopancreatography, (d) DW imaging with multiple b values; (e) fat fraction mapping with chemical shift imaging, and (f) dynamic imaging, including arterial, portal venous, and delayedphase imaging with fat-suppressed T1-weighted imaging using the threedimensional gradient-echo (volumeinterpolated breath-hold examination, or VIBE) sequence after injecting 0.1 mmol/mL gadobutrol. Imaging parameters for unenhanced heavily T2-weighted, T1-weighted, and DW

Yoon et al

Figure 1

Figure 1:  Flow diagram of the study population. CP = central pancreatectomy, DP = distal pancreatectomy, ISGPS = International Study Group of Pancreatic Surgery, PD = pancreatoduodenectomy, P-E = pancreatoenteric, POPF = postoperative pancreatic fistula, PPPD = pylorus-preserving pancreatoduodenectomy, SP = subtotal pancreatectomy, TP = total pancreatectomy.

examinations are summarized in Table E1 (online). DW image acquisition and postprocessing.—DW images were acquired by using the single-shot echo-planar imaging pulse sequence and multiple b values (0, 25, 50, 75, 100, 150, 200, 500, 800, and 1000 sec/mm2) (Table E1 [online]). Diffusion weighting was accomplished by using three orthogonal gradient directions. The apparent diffusion coefficient (ADC) was obtained by using all b values monoexponentially fitted to the following equation: Sb = S0 exp(2b 3 ADC), where Sb is the signal intensity at a given b value and S0 is the signal intensity at a b value of 0 sec/mm2. The data for all b values were postprocessed by using the prototype software (Siemens Healthcare) of a voxel-based nonlinear biexponential fit with least-square estimations, and three parametric maps, including the true diffusion coefficient (D), pseudodiffusion coefficient (D*), and perfusion fraction (f), were generated based on the following equation: Sb/S0 = (12f) 3 exp(2b 3 D) + f 3 exp[2b 3 (D + D*)] (21). Fat fraction maps.—Fat fraction maps were obtained by using a prototype three-dimensional gradient-echo

Radiology: Volume 000: Number 0—   2016  n  radiology.rsna.org

sequence with the T2*-corrected Dixon method (Siemens Healthcare). T2*corrected triple-echo data consisted of in-phase (360°), opposed-phase (540°), and in-phase (1080°) images, and the imaging parameters are described in Table E1 (online). Voxel-wise correction of T2* decay was applied by using T2* values calculated from the signal intensities and echo times of two in-phase images with a linear fit in log space (16). Water-fat separation was performed by using the first T2*corrected in-phase image and the first T2*-corrected opposed-phase image. Reconstructed images are water-only images, fat-only images, a T2* map, and a fat fraction map.

Clinical Data Collection Intraoperative data collection.—All surgical procedures were performed by one of two pancreatobiliary surgeons (S.W.K., J.Y.J.; each with more than 15 years of experience). During the procedure, the texture was classified into one of three levels depending on its hardness: soft, firm, or hard (22,23). In addition, the surgeons measured and recorded pancreatic duct size (in millimeters) at the pancreatic resection margin. 3

GASTROINTESTINAL IMAGING: Pancreatic Steatosis and Fibrosis

Postoperative data collection.—In all patients, volumes of fluid from surgical drains were checked daily, and fluid and serum amylase levels were measured on postoperative days 1, 3, 5, 7, and 10 (13). On postoperative day 7, patients underwent computed tomography (CT) to investigate possible complications. The diagnosis of postoperative pancreatic fistula was made if we detected a drain output of any measurable volume of fluid on or after postoperative day 3 with an amylase content greater than three times the serum amylase activity, in accordance with the criteria of the International Study Group for Pancreatic Surgery (9). The severity of postoperative pancreatic fistula was categorized into three levels (A, B, and C) according to the International Study Group for Pancreatic Surgery criteria (9) during the hospital stay: Level A indicates mild transient fistula; level B, fistula requiring adjustment of the clinical management strategy; and level C, a clinically important fistula causing deviation of the clinical management strategy (9).

Image Analysis One radiologist (J.H.Y., 8 years of clinical experience) who was blinded to the histologic and clinical findings drew regions of interest (ROIs) on fat fraction maps (mean area, 208.6 mm2 6 124.9 [standard deviation]), unenhanced T1weighted images (mean area, 128.8 mm2 6 101.1), and parametric maps (ADC, D, D*, and f) of the pancreas (mean area, 225.9 mm2 6 172.6). Two additional round ROIs, approximately 100 mm2 in area, were drawn on the bilateral paraspinal muscles of unenhanced T1-weighted images. Information regarding the type of surgery performed was provided, and the radiologist was aware of the common surgical procedures and postoperative findings after typical pancreatectomy. ROIs were placed in the pancreatic parenchyma near the resection margin on the fat map, unenhanced T1weighted images, and four parametric maps obtained with DW imaging. The pancreatic duct was carefully avoided with reference to heavily T2-weighted 4

Yoon et al

images. In patients who underwent total pancreatectomy (n = 6) and those who underwent enucleation (n = 5), signal intensity was measured in the proximal pancreatic parenchyma next to the tumor. In addition, the size of the pancreatic duct was measured on axial heavily T2-weighted images. Small pancreatic parenchyma and small pancreatic ducts were measured after magnifying the images at the workstation.

Histologic Analysis One pathologist (K.B.L., 11 years of experience) who was unaware of the clinical and MR findings reviewed pancreas specimens. Specimens were stained with hematoxylin-eosin, and the pathologist assessed fat deposition and fibrosis in the radial margin. The area of intraparenchymal fat was measured in square millimeters on a slide that contained at least 1 3 1 cm of pancreatic parenchyma, and the degree of pancreatic steatosis (fat fraction percentage) was defined as the ratio between the area of intraparenchymal fat and the total area of the pancreatic parenchyma on the slide (13,24). Fibrosis was graded as follows: F0, normal parenchyma; F1, mild fibrosis with thickening of periductal fibrous tissue; F2, moderate fibrosis with marked sclerosis of interlobular septa and no evidence of architectural changes; and F3, severe fibrosis with detection of architectural destruction (25). Statistical Analysis Spearman correlation coefficients were obtained for each pair for clinical, histologic, and MR parameters. Correlation coefficients were interpreted as follows: weak, 0.2; moderate, 0.5; and strong, 0.8 (26). Bland– Altman 95% limits of agreement were obtained between the fat fraction at MR imaging and the specimen. The x2 test was performed to compare categorical variables, and either the Student t test or the Mann-Whitney test was used for continuous variable comparison between two groups. All MR imaging parameters and fat amounts

were compared between fibrosis stages (F0–F3) by using the Kruskal-Wallis test with pairwise comparison. Nonparametric receiver operating characteristic curve analysis was performed, and area under the receiver operating characteristic curve (AUC) was calculated to assess diagnostic performance in the detection of advanced fibrosis, and AUC was compared among the parameters (27). Cutoff values were determined to maximize the Youden index. We also performed univariate logistic regression analysis to evaluate the relationship between each parameter and advanced fibrosis, as well as clinically relevant postoperative pancreatic fistula. Thereafter, multivariate analyses were performed by using a logistic regression model for all variables with a P value lower than .2 at univariate analysis to determine associated factors of advanced fibrosis and postoperative pancreatic fistula. Statistical analyses were performed by using commercially available software (SPSS, version 21, SPSS, Armonk, NY; Medcalc, version 12, Medcalc Software, Mariakerke, Belgium), and a P value of less than .05 indicated a significant difference. The Bonferroni correction was applied to adjust for the number of comparisons (P , .017 in three comparisons, P , .0125 in four comparisons).

Results Preoperative MR Assessment of Pancreatic Fat Fraction and Duct Size In the specimens, the measured amount of fat ranged from 0% to 70.0% (median, 5.0%; 25th to 75th percentile, 1.0%–10.0%). Measured fat fraction at MR imaging ranged from 1.7% to 39.1% (median, 6.5%; 25th to 75th percentile, 4.6%–10.8%). Measured fat fractions at MR imaging showed a moderate correlation with the amount of fat in the specimens (r = 0.71: 95% confidence interval [CI]: 0.63, 0.78). Bland-Altman 95% limits of agreement between fat fractions at MR imaging and the specimen were 216.3% to

radiology.rsna.org  n Radiology: Volume 000: Number 0—   2016

GASTROINTESTINAL IMAGING: Pancreatic Steatosis and Fibrosis

16.8% (mean difference, 0.25; 95% CI: 21.06, 1.56) (Fig 2). Intraoperatively assessed mean pancreatic duct size was 3.0 mm 6 2.4, and mean pancreatic duct size at MR imaging was 2.9 mm 6 2.4, with moderate correlation (r = 0.50; 95% CI: 0.3, 0.61). Fat fraction and fibrosis were not significantly different between men and women (P = .79 and P = .72, respectively), but age was weakly to moderately correlated with the fat fraction (r = 0.39;

Yoon et al

95% CI: 0.25, 0.51) and fibrosis (r = 0.17; 95% CI: 0.01, 0.31).

Preoperative MR Assessment of Pancreatic Fibrosis Pancreatic fibrosis and parameters at MR imaging.—Values of IVIM DW imaging parameters in each fibrosis stage are summarized in Table 1. As fibrosis progressed, all parameters tended to decrease, but only D* and f showed significant differences among different

Figure 2

fibrosis stages (P = .021 and P , .001, respectively). With increasing degrees of fibrosis, pancreatic parenchyma showed a higher fat fraction (r = 0.35; 95% CI: 0.21, 0.48) and larger pancreatic duct size (r = 0.36; 95% CI: 0.21, 0.48) at MR imaging. Unenhanced T1weighted image signal intensity ratio (r = 20.29; 95% CI: 20.42, 20.14) decreased as fibrosis progressed. Differentiation of advanced pancreatic fibrosis (F2).—In patients with advanced fibrosis, D*, f, and unenhanced T1-weighted image signal intensity ratio were significantly decreased, whereas fat fraction and duct size were increased (Tables E1, E2 [online]). Fat fraction, f, and unenhanced T1-weighted image signal intensity ratio were correlated with fibrosis stage according to multivariate analysis (AUC, 0.83; 95% CI: 0.77, 0.89), and the odds ratios are summarized in Table 2.

Correlation between MR Imaging, Histology, and Intraoperative Evaluation Intraoperative texture was assessed, and 99, 43, and 15 specimens were classified as soft, firm, and hard, respectively. In eight patients, intraoperative texture was not assessed. Among the three pancreatic textures, there were no significant differences in either the fat amount in specimens (soft texture: mean, 8.8% 6 12.5; range, 0%–70%; firm texture:

Figure 2:  Bland-Altman plot of fat fractions obtained with the T2*-corrected Dixon method and histologic analysis. The difference between the two methods (MR fat fraction minus histologic fat fraction, y-axis) was plotted against average values of two methods (x-axis).

Table 1 Parameters of DW Imaging, Fat Fraction, and Pancreatic Duct Size at Each Fibrosis Stage Parameter 2

ADCtotal (310 mm /sec) D (31023mm2/sec) D* (31023mm2/sec) f (%) Fat fraction on MR image (%) Fat fraction in specimen (%) Unenhanced T1-weighted image signal intensity ratio Pancreatic duct size on MR image (mm) Pancreatic duct size in surgical field (mm) 23

F0 (n = 41)

F1 (n = 86)

F2 (n = 21)

F3 (n = 17)

P Value

1.37 6 0.29 (1.03–2.58) 1.31 6 0.33 (0.96–2.95) 41.95 6 15.04 (13.5–73.3) 31.00 6 9.20 (10.4–62.2) 6.02 6 2.92 (1.8–15.1) 3.98 6 5.32 (0.0–20.0) 1.44 6 0.26 (0.83–1.95)

1.39 6 0.28 (0.93–2.55) 1.32 6 0.27 (0.72–2.42) 38.67 6 12.87 (9.5–71.8) 29.19 6 8.16 (11.6–53.8) 8.24 6 5.55 (1.7–30.5) 8.11 6 11.45 (0.0–70.0) 1.43 6 0.25 (1.05–2.52)

1.34 6 0.22 (1.07–1.94) 1.22 6 0.15 (1.02–1.54) 32.02 6 12.57 (4.0–51.1) 21.61 6 8.54 (8.9–40.4) 11.03 6 6.06 (4.5–29.5) 13.91 6 12.68 (0.0–45.0) 1.32 6 0.26 (0.78–1.85)

1.30 6 0.18 (1.02–1.64) 1.22 6 0.14 (0.99–1.52) 33.10 6 14.06 (5.5–65.3) 19.86 6 8.97 (8.0–39.6) 13.10 6 9.21 (3.6–39.1) 18.12 6 17.71 (0.0–50.0) 1.09 6 0.32 (0.7–1.72)

.704 .280 .021 ,.001 ,.001 ,.001 ,.001

1.99 6 0.99 (0.99–5.60)

2.70 6 2.51 (0.99–21.0)

3.76 6 2.80 (0.99–10.62)

4.21 6 2.48 (0.99–10.1)

,.001

1.82 6 1.11 (0.8–4.0)

2.97 6 2.18 (0.8–10.0)

3.98 6 3.07 (1.0–10.0)

4.50 6 3.14 (1.0–10.0)

,.001

Note.—Data are mean 6 standard deviation. Data in parentheses are the range. P , .05 indicates a significant difference among the groups. ADCtotal = apparent diffusion coefficient obtained by using multiple b values, SI = signal intensity.

Radiology: Volume 000: Number 0—   2016  n  radiology.rsna.org

5

GASTROINTESTINAL IMAGING: Pancreatic Steatosis and Fibrosis

mean, 9.1% 6 11.5; range, 0%–50%; hard texture: 9.2% 6 12.1; range, 0%–45%) (P = .94) or fat fractions at MR imaging (soft texture: mean, 8.2% 6 6.0; range, 1.8%–39.1%; firm texture: 9.1% 6 5.9; range, 2.2%–30.5%; hard texture: 9.6% 6 6.8; range, 4.7%– 29.5%) (P = .67). In the comparison of intraoperatively assessed pancreatic texture and fibrosis stage, 17 (46%) of 37 specimens with advanced fibrosis (F2– F3) were classified as having a soft pancreas, and 38 (32%) of 120 specimens with normal parenchyma or mild fibrosis (F0–F1) were classified as having a firm or hard pancreas.

Incidence of Postoperative Pancreatic Fistula and Clinical, MR Imaging, and Histologic Findings Among the 95 patients (59 men, 36 women) who underwent pancreaticoduodenectomy and central pancreatectomy, 36 had postoperative pancreatic fistula (class A, n = 22; class B, n = 13; class C, n = 1); the remaining 59 patients did not show signs of postoperative pancreatic fistula. There were no significant differences when we compared patients with clinically relevant fistula with those without fistula or with class A fistula (Table 3; Figures 3, 4). Among the MR parameters, f was the only parameter significantly associated with the development of clinically relevant postoperative pancreatic fistula (Table 4); D*, pancreatic duct size at MR imaging, fibrosis, fat, sex, and age were not significantly associated with the development of clinically relevant postoperative pancreatic fistula. The odds of developing clinically relevant postoperative pancreatic fistula for a 1% increase in f were 1.17 (95% CI: 1.05, 1.30), and the AUC was 0.87 (95% CI: 0.79, 0.94) for f and 0.92 (95% CI: 0.85, 0.97) for the model including all significant variables at univariate analysis (fat, D*, f, and pancreatic duct size at MR imaging). Discussion Our study showed that pancreas steatosis and clinically important pancreatic fibrosis can be quantified at 6

Yoon et al

Table 2 Results of Univariate and Multivariate Analyses of Parameters for Advanced Fibrosis Univariate Analysis Variable Male sex Age (y) Fat fraction on MR image (%) ADCtotal* D (31023mm2/sec) D* (31023mm2/sec) f (%) Unenhanced T1-weighted image signal intensity ratio Duct size on MR image (mm)

Odds Ratio

95% CI

1.12 1.04 1.12 0.46 0.16 0.96 0.88 0.04 1.29

Multivariate Analysis P Value

Odds Ratio

95% CI

P Value

0.54, 2.33 .76 1.00, 1.08 .025 1.05, 1.20 .0002 0.10, 2.11 .3 0.02, 1.00 .03 0.93, 0.99 .0035 0.83, 0.92 ,.0001 0.009, 0.20 ,.0001

… 1.02 1.14 … 0.21 0.99 0.89 0.12

… 0.97, 1.06 1.05, 1.23 … 0.02, 1.97 0.95, 1.03 0.84, 0.95 0.02, 0.76

… .52 .0012 … .17 .64 .0005 .025

1.09, 1.53

1.06

0.90, 1.26

.47

.0012

Note.—ADCtotal = apparent diffusion coefficient obtained by using multiple b values, SI = signal intensity.

Table 3 Comparison of Histologic and MR Findings between Groups with and without Postoperative Pancreatic Fistula in 95 Patients Who Underwent Pancreatoduodenectomy or Central Pancreatectomy Finding Clinical   Sex (male:female)*   Age (y)   Disease (PAC:others)* Histologic findings   Fat amount (%)   Fibrosis (F0–F1:F2–F3)*   Fibrosis (F0–F2:F3)* Intraoperative   Texture (soft:firm or hard)*†   Pancreatic duct size (mm) MR imaging   Pancreatic duct size (mm)   Fat fraction (%)   f (%)   D* (31023mm2/sec)   D (31023mm2/sec)  ADCtotal (31023mm2/sec)   Unenhanced T1-weighted   image signal intensity ratio

No or Grade A POPF (n = 81)

Grade B or C POPF (n = 14)

49:32 62.7 6 9.5 (39, 77) 20:61

10:4 60.4 6 11.9 (29, 75) 1:13

.63 .23 .18

8.23 6 11.7 (0, 70) 64:17 73:8

8.14 6 12.5 (0, 45) 11:3 14:0

.68 .97 .41

48:32 3.14 6 2.50 (0.8, 10.0) 3.19 6 2.06 (0.99, 10.1) 8.15 6 5.24 (1.7, 25.3) 25.27 6 8.37 (8.0, 47.7) 35.11 6 12.91 (4.0, 69.5) 1.27 6 0.24 (0.72, 2.42) 1.35 6 0.25 (1.03, 2.55) 1.31 6 0.29 (0.70, 1.95)

9:5 1.93 6 1.10 (1.0, 4.0) 2.30 6 0.33 (1.78, 2.97) 10.45 6 8.35 (2.9, 30.5) 37.99 6 9.20 (30.0, 62.2) 47.72 6 10.89 (27.0, 71.2) 1.27 6 0.13 (1.04, 1.43) 1.32 6 0.26 (1.02, 1.56) 1.40 6 0.23 (1.07, 1.85)

P Value

.76 .004 ,.001 .17 ,.001 .001 .92 .77 .28

Note.—Unless otherwise indicated, data are mean 6 standard deviation. Data in parentheses are the minimum and maximum. P , .05 indicates a significant difference between the two groups. ADCtotal = apparent diffusion coefficient obtained by using multiple b values, PAC = pancreatic adenocarcinoma, POPF = postoperative pancreatic fistula. * Data are number of patients. †

Pancreas texture was not available in one patient without postoperative pancreatic fistula.

multiparametric MR imaging, as fat fractions obtained by using the T2*-corrected Dixon technique showed moderate correlation with the histologic fat

amount, and perfusion-related IVIM DW imaging parameters (D* and f) were significantly decreased in patients with advanced pancreatic fibrosis.

radiology.rsna.org  n Radiology: Volume 000: Number 0—   2016

GASTROINTESTINAL IMAGING: Pancreatic Steatosis and Fibrosis

Yoon et al

Figure 3

Figure 3:  Preoperative MR and postoperative CT images in a 58-year-old man who underwent pancreatoduodenectomy for common bile duct cancer. (a) Heavily T2-weighted image shows the pancreatic duct (arrowheads) was not dilated. (b) Unenhanced T1-weighted image shows the pancreas-to-muscle signal intensity ratio was 1.16. (c) A free-hand ROI was placed in the pancreas (arrowheads) on the calculated fat map, and the estimated fat fraction was 4.8%. (d) Signal decay curve of the ROI on IVIM DWI parametric maps. Slope of signal intensity decay for lower b values was steep, and the overall curve showed a hockey stick appearance. Pancreatic parenchyma was diagnosed as F0 and 2% of pancreas fat. (e) Postoperative follow-up CT image shows pancreatojejunal anastomosis site leakage and loculated fluid collection (arrows). The patient underwent insertion of a percutaneous catheter for drainage and received antibiotics. He was discharged without sequelae on postoperative day 21.

In our study, we found that the estimated pancreatic fat fraction at MR imaging showed a moderate correlation

with histologic results. We believe that it would be quite important because the results validated the capability of the

Radiology: Volume 000: Number 0—   2016  n  radiology.rsna.org

noninvasive MR technique to enable quantification of pancreatic steatosis and also provided supporting evidence 7

GASTROINTESTINAL IMAGING: Pancreatic Steatosis and Fibrosis

Yoon et al

Figure 4

Figure 4:  Preoperative MR and postoperative CT images in a 71-year-old woman who underwent pancreatoduodenectomy for pancreatic head cancer. (a) Heavily T2-weighted image shows the pancreatic duct (arrowheads) was slightly prominent, measuring 2 mm in diameter. (b) Unenhanced T1-weighted image shows the pancreas-to-muscle signal intensity ratio was 0.95 (arrows indicate pancreatic parenchyma). (c) A free-hand ROI was placed in the pancreas (arrows) near the splenic vein (arrowheads) on the fat map, and the estimated fat fraction was 13.3%. (d) Signal decay curve of the ROI on IVIM DW parametric maps. Slope of signal intensity decay for lower b values was gradual, and the overall curve lost its hockey stick appearance, which was shown in the patient with normal parenchyma (F0) (d). The pancreatic parenchyma was classified as F2 and 15% of pancreas fat. (e) Postoperative follow-up CT scan shows the pancreatojejunal anastomosis site (arrows) was patent, and no evidence of postoperative pancreatic fistula was found at clinical evaluation.

for prior studies on pancreatic steatosis using advanced MR fat quantification techniques (28–30). Furthermore, 8

the T2*-corrected Dixon technique has advantages over ultrasonography, CT, or dual-echo chemical shift imaging

(13,24,31), as it enables more reliable and more accurate fat quantification than do the other techniques. However,

radiology.rsna.org  n Radiology: Volume 000: Number 0—   2016

GASTROINTESTINAL IMAGING: Pancreatic Steatosis and Fibrosis

Yoon et al

Table 4 Results of Univariate and Multivariate Analyses of Parameters for Clinically Relevant Postoperative Pancreatic Fistula in Patients Who Underwent Pancreatoenteric Anastomosis Univariate Analysis Variable Sex (male) Age (y) Fibrosis grade (F0–F1) Fibrosis grade (F3) Disease type (adenocarcinoma) Fat fraction at MR imaging (%) D* (31023mm2/sec) f (%) Unenhanced T1-weighted image signal intensity ratio Duct size at MR imaging (mm)

Multivariate Analysis

Odds Ratio

95% CI

P Value

Odds Ratio

95% CI

P Value

1.63 0.98 1.03 0 0.23 1.06 1.08 1.19 3.15

0.47, 5.65 0.92, 1.03 0.26, 4.09 … 0.03, 1.91 0.97, 1.16 1.03, 1.13 1.08, 1.30 0.39, 25.18

.43 .42 .97 .1 .23 .19 .0011 ,.0001 .27

… … … … … 1.1 1.1 1.17 …

… … … … … 0.97, 1.25 1.0, 1.13 1.05, 1.30 …

… … … … … .13 .05 .006 …

0.69

0.43, 1.11

.068

0.68

0.37, 1.25

.22

when compared with the results for hepatic steatosis obtained by using the same technique (16), our results showed slightly lower correlation with histologic findings (r = 0.886 vs r = 0.71). The reason for the relatively low correlation is unclear; thus, further studies in this regard may be warranted. In addition, IVIM DW imaging parameters (D* and f) showed a significant decrease in advanced pancreatic fibrosis (F2–F3), which is consistent with the previous literature on hepatic fibrosis (18,32). Decreased f in patients with advanced fibrosis may represent decreased blood volume, which is consistent with findings in previous studies that showed decreased microvascular density in patients with pancreatic fibrosis (33) and altered perfusion parameters at MR imaging (34,35). Furthermore, in many of the study patients, pancreatic fibrosis was accompanied by pancreatic steatosis; fat fraction was significantly higher in patients with advanced fibrosis (F2–F3) than in patients with no or mild fibrosis (F0–F1). Thus, it is plausible that the increased extracellular matrix via pancreatic stellate cell activation and fat accumulation in the pancreatic fibrosis (36) may disturb pancreatic perfusion. However, the ADC obtained by using multiple b values (hereafter, ADCtotal) in our study could not be used to discriminate

patients with advanced fibrosis from those with no or early fibrosis; this finding was inconsistent with the findings of a previous study (14). Indeed, the inconsistency between ADCtotal, D, and pancreatic fibrosis has been reported (37). We do not know why ADCtotal and D did not show significant differences between F0–F1 and F2–F3 classification. However, we assume that fat accumulation in patients with no fibrosis or early fibrosis as well as that in patients with advanced fibrosis could be attributed to diffusion restriction and may counteract the ADCtotal or D values as a confounder (38,39). In our study, we also attempted to elucidate the relationship between the alteration of pancreatic parenchyma and postoperative pancreatic fistula at preoperative MR imaging. Most known associated factors of postoperative pancreatic fistula (12,20,40,41) can be assessed only after surgery. In our study, f at preoperative MR imaging was shown to be a parameter significantly associated with the development of clinically relevant postoperative pancreatic fistula. Thus, we believe that multiparametric MR imaging would be potentially valuable in the preoperative risk analysis of postoperative pancreatic fistula (10,13), and we cautiously suggest that IVIM DW imaging parameters, especially f, could be used

Radiology: Volume 000: Number 0—   2016  n  radiology.rsna.org

to assess risk of postoperative pancreatic fistula, as a surrogate marker for pancreatic exocrine function producing digestive enzymes, which was the proposed mechanism for postoperative pancreatic fistula (41). Although further studies are needed, preoperatively assessed f might be used to plan perioperative management strategies, including prophylactic somatostatin analog administration (22,42,43). However, pancreas fat fraction estimated with MR imaging showed a limited role in the prediction of postoperative pancreatic fistula in our study. Indeed, there have been some controversies regarding the value of pancreatic steatosis as a predictor in the development of postoperative pancreatic fistula (10,13,20), and in our study, fat fraction was shown to be significantly higher in patients with advanced fibrosis (F2–F3) than in those with no or mild fibrosis (F0–F1) (P = .003). Thus, we assume that pancreatic fibrosis may accompany steatosis and that this fibrofatty change in the pancreas may be the reason for the indistinctive role of pancreatic steatosis in the prediction of postoperative pancreatic fistula in our study. We also believe that fat accumulation in patients with advanced fibrosis may have led to the inconsistent perception of pancreatic texture during surgery and that this might have been the reason for the discrepancy between pancreatic texture and fibrosis degree and for the limited capability of pancreatic texture to enable prediction of postoperative pancreatic fistula. On the basis of our observation of fat fraction and f, we believe that pancreatic fibrosis may be more important than steatosis in the development of postoperative pancreatic fistula, which is related to pancreatic exocrine function (42,44), and that the causal relationship between pancreatic fibrosis and steatosis should be investigated in further studies. Our study had several limitations. Because our study was retrospective, there may have been inevitable selection bias. Second, a relatively small number of patients with postoperative pancreatic fistula or advanced fibrosis were included. Third, our fat 9

GASTROINTESTINAL IMAGING: Pancreatic Steatosis and Fibrosis

quantification technique did not correct all confounding factors, including the T1 effect and the multipeak fat spectrum. Fourth, our results were based on our own retrospective data, and this might have resulted in overestimation of the diagnostic performance of the parameters. Further studies are warranted to validate our result in this regard. In addition, we did not consider the surgeons’ experience as an associated factor of postoperative pancreatic fistula. However, as both surgeons were sufficiently experienced in hepatobiliary pancreas surgery (ie, more than 15 years of experience) and as the incidence of postoperative pancreatic fistula in our study was lower than that in previous studies (7,8), we believe that the effect of surgeons’ clinical experience would be minimal. Fifth, measurement was performed by one observer who did not assess interor intraobserver variability; this needs to be investigated in the pancreas in future studies. Sixth, we obtained all data with the same MR imager. Further studies that include multiple imagers from several vendors are warranted. In conclusion, our results suggest that multiparametric MR imaging of the pancreas, including the T2*-corrected Dixon technique and IVIM DW imaging may yield quantitative information on pancreatic steatosis and fibrosis, and f was shown to be significantly associated with postoperative pancreatic fistula. Acknowledgments: We thank In Seong Kim (Siemens Healthcare, Seoul Korea) and Berthold Kiefer (Siemens Healthcare, Erlangen, Germany) for their technical advice and Chris Woo, BA, for his editorial assistance. We appreciate the statistical advice from the Medical Research Collaborating Center at Seoul National University Hospital and Seoul National University College of Medicine. Disclosures of Conflicts of Interest: J.H.Y. Activities related to the present article: none to disclose. Activities not related to the present article: none to disclose. Other relationships: received lecture honorarium from Bayer Healthcare, GE Healthcare, and Philips Healthcare. J.M.L. Activities related to the present article: none to disclose. Activities not related to the present article: received grants from Bayer Healthcare, GE Healthcare, Guerbet, Dong Seo Medical, RF Medical, Starmed, and Toshiba Healthcare; received an honorarium and a traveling fee from Bayer

10

Yoon et al

Healthcare; is an advisory board member for Bayer Healthcare; received technical support for research software from GE Healthcare and Philips. Other relationships: none to disclose. K.B.L. disclosed no relevant relationships. S.W.K. disclosed no relevant relationships. M.J.K. disclosed no relevant relationships. J.Y.J. disclosed no relevant relationships. S.K. Activities related to the present article: none to disclose. Activities not related to the present article: none to disclose. Other relationships: is an employee of Siemens Healthcare. J.K.H. disclosed no relevant relationships. B.I.C. disclosed no relevant relationships.

pancreatic fistula: is it possible? Pancreatology 2013;13(4):423–428. 11. Kirihara Y, Takahashi N, Hashimoto Y, et al. Prediction of pancreatic anastomotic failure after pancreatoduodenectomy: the use of preoperative, quantitative computed tomography to measure remnant pancreatic volume and body composition. Ann Surg 2013; 257(3):512–519. 12. Mathur A, Pitt HA, Marine M, et al. Fatty pancreas: a factor in postoperative pancreatic fistula. Ann Surg 2007;246(6):1058–1064.

1. Freelove R, Walling AD. Pancreatic cancer: diagnosis and management. Am Fam Physician 2006;73(3):485–492.

13. Lee SE, Jang J-Y, Lim C-S, et al. Measurement of pancreatic fat by magnetic resonance imaging: predicting the occurrence of pancreatic fistula after pancreatoduodenectomy. Ann Surg 2010;251(5):932–936.

2. Gouma DJ, van Geenen RC, van Gulik TM, et al. Rates of complications and death after pancreaticoduodenectomy: risk factors and the impact of hospital volume. Ann Surg 2000; 232(6):786–795.

14. Watanabe H, Kanematsu M, Tanaka K, et al. Fibrosis and postoperative fistula of the pancreas: correlation with MR imaging findings—preliminary results. Radiology 2014;270(3):791–799.

3. Cameron JL, Riall TS, Coleman J, Belcher KA. One thousand consecutive pancreaticoduodenectomies. Ann Surg 2006;244(1):10– 15.

15. Patel NS, Peterson MR, Brenner DA, Heba E, Sirlin C, Loomba R. Association between novel MRI-estimated pancreatic fat and liver histology-determined steatosis and fibrosis in non-alcoholic fatty liver disease. Aliment Pharmacol Ther 2013;37(6):630–639.

References

4. Büchler MW, Wagner M, Schmied BM, Uhl W, Friess H, Z’graggen K. Changes in morbidity after pancreatic resection: toward the end of completion pancreatectomy. Arch Surg 2003;138(12):1310–1314; discussion 1315. 5. Balcom JH 4th, Rattner DW, Warshaw AL, Chang Y, Fernandez-del Castillo C. Ten-year experience with 733 pancreatic resections: changing indications, older patients, and decreasing length of hospitalization. Arch Surg 2001;136(4):391–398. 6. Brient C, Regenet N, Sulpice L, et al. Risk factors for postoperative pancreatic fistulization subsequent to enucleation. J Gastrointest Surg 2012;16(10):1883–1887. 7. Butturini G, Daskalaki D, Molinari E, Scopelliti F, Casarotto A, Bassi C. Pancreatic fistula: definition and current problems. J Hepatobiliary Pancreat Surg 2008;15(3): 247–251. 8. Vin Y, Sima CS, Getrajdman GI, et al. Management and outcomes of postpancreatectomy fistula, leak, and abscess: results of 908 patients resected at a single institution between 2000 and 2005. J Am Coll Surg 2008; 207(4):490–498.

16. Hwang I, Lee JM, Lee KB, et al. Hepatic steatosis in living liver donor candidates: preoperative assessment by using breath-hold triple-echo MR imaging and 1H MR spectroscopy. Radiology 2014;271(3):730–738. 17. Guiu B, Petit J-M, Loffroy R, et al. Quantification of liver fat content: comparison of triple-echo chemical shift gradient-echo imaging and in vivo proton MR spectroscopy. Radiology 2009;250(1):95–102. 18. Taouli B, Tolia AJ, Losada M, et al. Diffusion-weighted MRI for quantification of liver fibrosis: preliminary experience. AJR Am J Roentgenol 2007;189(4):799–806. 19. Al Sharaf K, Ihse I, Dawiskiba S, Andrén Sandberg Å. Characteristics of the gland remnant predict complications after subtotal pancreatectomy. Dig Surg 1997;14(2):101–106. 20. Gaujoux S, Cortes A, Couvelard A, et al. Fatty pancreas and increased body mass index are risk factors of pancreatic fistula after pancreaticoduodenectomy. Surgery 2010;148(1): 15–23.

9. Bassi C, Dervenis C, Butturini G, et al. Postoperative pancreatic fistula: an international study group (ISGPF) definition. Surgery 2005; 138(1):8–13.

21. Le Bihan D, Breton E, Lallemand D, Aubin ML, Vignaud J, Laval-Jeantet M. Separation of diffusion and perfusion in intravoxel incoherent motion MR imaging. Radiology 1988;168(2):497–505.

10. Roberts KJ, Storey R, Hodson J, Smith AM, Morris-Stiff G. Pre-operative prediction of

22. Yeo CJ, Cameron JL, Lillemoe KD, et al. Does prophylactic octreotide decrease the

radiology.rsna.org  n Radiology: Volume 000: Number 0—   2016

GASTROINTESTINAL IMAGING: Pancreatic Steatosis and Fibrosis

rates of pancreatic fistula and other complications after pancreaticoduodenectomy? results of a prospective randomized placebo-controlled trial. Ann Surg 2000;232(3): 419–429. 23. Rosso E, Casnedi S, Pessaux P, et al. The role of “fatty pancreas” and of BMI in the occurrence of pancreatic fistula after pancreaticoduodenectomy. J Gastrointest Surg 2009; 13(10):1845–1851. 24. Kim SY, Kim H, Cho JY, et al. Quantitative assessment of pancreatic fat by using unenhanced CT: pathologic correlation and clinical implications. Radiology 2014;271(1):104–112. 25. Wellner UF, Kayser G, Lapshyn H, et al. A simple scoring system based on clinical factors related to pancreatic texture predicts postoperative pancreatic fistula preoperatively. HPB (Oxford) 2010;12(10): 696–702. 26. Zou KH, Tuncali K, Silverman SG. Correlation and simple linear regression. Radiology 2003;227(3):617–622. 27. DeLong ER, DeLong DM, Clarke-Pearson DL. Comparing the areas under two or more correlated receiver operating characteristic curves: a nonparametric approach. Biometrics 1988;44(3):837–845. 28. Heni M, Machann J, Staiger H, et al. Pancreatic fat is negatively associated with insulin secretion in individuals with impaired fasting glucose and/or impaired glucose tolerance: a nuclear magnetic resonance study. Diabetes Metab Res Rev 2010;26(3):200– 205. 29. Hu HH, Kim HW, Nayak KS, Goran MI. Comparison of fat-water MRI and singlevoxel MRS in the assessment of hepatic and

pancreatic fat fractions in humans. Obesity (Silver Spring) 2010;18(4):841–847. 30. Li J, Xie Y, Yuan F, Song B, Tang C. Noninvasive quantification of pancreatic fat in healthy male population using chemical shift magnetic resonance imaging: effect of aging on pancreatic fat content. Pancreas 2011;40(2):295–299. 31. Al-Haddad M, Khashab M, Zyromski N, et al. Risk factors for hyperechogenic pancreas on endoscopic ultrasound: a case-control study. Pancreas 2009;38(6):672–675. 32. Yoon JH, Lee JM, Baek JH, et al. Evaluation of hepatic fibrosis using intravoxel incoherent motion in diffusion-weighted liver MRI. J Comput Assist Tomogr 2014;38(1):110– 116. 33. Bali MA, Metens T, Denolin V, et al. Tumoral and nontumoral pancreas: correlation between quantitative dynamic contrastenhanced MR imaging and histopathologic parameters. Radiology 2011;261(2):456– 466. 34. Coenegrachts K, Van Steenbergen W, De Keyzer F, et al. Dynamic contrast-enhanced MRI of the pancreas: initial results in healthy volunteers and patients with chronic pancreatitis. J Magn Reson Imaging 2004;20(6):990–997. 35. Patel AG, Reber PU, Toyama MT, Ashley SW, Reber HA. Effect of pancreaticojejunostomy on fibrosis, pancreatic blood flow, and interstitial pH in chronic pancreatitis: a feline model. Ann Surg 1999; 230(5):672–679. 36. Apte MV, Wilson JS. Mechanisms of pancreatic fibrosis. Dig Dis 2004;22(3):273–279.

Radiology: Volume 000: Number 0—   2016  n  radiology.rsna.org

Yoon et al

37. Klauss M, Gaida MM, Lemke A, et al. Fibrosis and pancreatic lesions: counterintuitive behavior of the diffusion imagingderived structural diffusion coefficient d. Invest Radiol 2013;48(3):129–133. 38. Nonomura Y, Yasumoto M, Yoshimura R, et al. Relationship between bone marrow cellularity and apparent diffusion coefficient. J Magn Reson Imaging 2001;13(5):757–760. 39. Bülow R, Mensel B, Meffert P, Hernando D, Evert M, Kühn J-P. Diffusion-weighted magnetic resonance imaging for staging liver fibrosis is less reliable in the presence of fat and iron. Eur Radiol 2013;23(5):1281–1287. 40. Kim Z, Kim MJ, Kim JH, et al. Prediction of post-operative pancreatic fistula in pancreaticoduodenectomy patients using preoperative MRI: a pilot study. HPB (Oxford) 2009;11(3):215–221. 41. Israel JS, Rettammel RJ, Leverson GE, et al. Does postoperative drain amylase predict pancreatic fistula after pancreatectomy? J Am Coll Surg 2014;218(5):978–987. 42. Allen PJ, Gönen M, Brennan MF, et al. Pasireotide for postoperative pancreatic fistula. N Engl J Med 2014;370(21):2014–2022. 43. Rahbour G, Siddiqui MR, Ullah MR, Gabe SM, Warusavitarne J, Vaizey CJ. A metaanalysis of outcomes following use of somatostatin and its analogues for the management of enterocutaneous fistulas. Ann Surg 2012;256(6):946–954. 44. Tran TC, van ’t Hof G, Kazemier G, et al. Pancreatic fibrosis correlates with exocrine pancreatic insufficiency after pancreatoduodenectomy. Dig Surg 2008;25(4):311–318.

11