| BackgroundPancreaticoduodenectomy (PD) is the standard operative procedure recognized by many phyisicans in the treatment of most benign and malignant disorders of pancreatic head and periampullary region. With the advancements in operative technique, development in surgical instruments and improvements in postoperative clinical outcomes in recent years, despite surgical mortality has decreased to less than 5%, postoperative morbidity remains unacceptably high at approximately 41%-65% even in some special units. Postoperative clinical relevant pancreatic fistula (CR-POPF) is widely considered to be the most common and catastrophic complication after PD. It can have dreadful consequences, particularly abdominal sepsis and hemorrhage, and remains the leading risk factor for longer hospital stay and increased healthcare costs.To decrease the incidence of CR-POPF and determine appropriate management, more and more investigators pay their attention to risk factors related CR-POPF in recent years. A great deal of risk factors of CR-POPF have been reported, including sex, age, drain amylase level, intraoperative blood loss, fatty pancreas, pancreas-to-muscle signal intensity ratio, pancreatic fibrosis and so on. Although most of them can provide reliable information for predicting CR-POPF after PD, but in fact a majority of them can only be identified intra-operatively or postoperatively. Some of them can be identified preoperatively, but it is too complicated for clinical surgeons rather than radiological physicians to follow in daily clinical practice.Although the risk factors of CR-POPF have been studied extensively, however, the soft pancreatic parenchyma is the only validated risk factor. Fatty pancreas is usually characteristic as fatty infiltration which would increase the softness of the gland and usually correlates with a small pancreatic duct, but the extent of pancreatic fat can only be pathological analyzed postoperatively, which limits its widely application. Fatty pancreas is more frequently observed in the obese patients, and it seems logical that the obese patients usually come with a relatively large volume of pancreatic gland.This study was supported by national high technology research and development program "863", the national natural science foundation of China and guang dong province industrial technology research and development plan project fund. Our team combined with those professors concerning hepatopancreatobiliary surgery, computer and imaging diagnostic, developed successfully the abdominal Medical Image Three Dimensional Visualization System (MI-3 DVS) for abdominal viscera synchronous stereoscopic 3 dimensional imaging. We acquired the three-dimensional model of main arteries and veins around the pancreas, the pancreas and the pancreas duct by the MI-3 DVS software. After the simulation of surgical excision, we obtained these quantitative values of the morphological features of pancreatic stump mainly includes residual remnant pancreatic parenchyma volume (Remnant Pancreatic Parenchymal Volume, RPPV), the main pancreatic duct diameter on cross sectional (Main Pancreas Duct Diameter, MPDD) and the area of cut surface on transection section (Area Of the Cut Surface, AOCS), We furtherly combined above pancreas characteristic with the clinical data concerning preoperative, intraoperative, and defined whether CR-POPF occurs or not as the finish line in this study and comprehensively analyses the role of these factors including pancreatic stump morphological features in the prediction of CR-POPF. To sum up, the objective of the present study was trying to assess the role of morphological features of pancreatic stump, preoperative clinical data and intraoperative clinical data in predicting the occurrence of CR-POPF after PD.Objective1. To explore a method of collection high quality submillimeter pancreatic and peri-pancreatic vascular 64 slice CT original data.2. To acquire the morphological features of pancreatic stump based on MI-3 DVS software.3. To analysis comprehensively the role of the pancreatic stump morphological features, preoperative and intraoperative clinical related variables in the prediction of CR-POPF occurred.Methods1. Materials:1.1 Equipments:(1) lenove computer; (2) HP blade server; (3) PHILIPS BRILLIANCE 64-slice helical CT, Image process workstation; (4) binocular tube high pressure injector; (5) DICOM Viewer; (6) Medical Image Three Dimensional Visualization System (MI-3DVS).1.2 Agent:Schering AG, Germany. Ultravist (300 mg I/mL).2. Subject:A total of 90 patients who underwent PD from April 2012 to May 2014 Zhu Jiang Hospital of Southern Medical University were included in this study, male 48 cases, female 42 cases, the patients with an average age of 61.3±9.4 years old. All patients were divided into postoperative clinically relevant pancreatic fistula groups (group A,18 cases) and the others (group B,72 cases) according to the occurrence of CR-POPF. The indications for PD were summarized as follows:55 patients with carcinoma of head of pancreas (61.1%),10 with duodenal cancer (11.1%),7 with bile duct carcinoma (7.8%),5 with adenocarcinoma of the ampulla of vater (5.6%), 3 with endocrine carcinoma (3.3%),2 with intraductal papillary mucinous neoplasm (IPMN) (2%),3 with cholangitis (3.3%),2 with cystadenoma (2%),2 with solid and pseudopapillary tumor (2%), and 1 with other benign neoplasms (1.1%).3. The acquisition of high quality upper abdomen 64 multi slice CT image data: The scanning of the patients in supine, the scanning parameters as fellowing:voltage is 120 kv, current is 300 mas; the 0.625 x 64 detector combination. Scanning layer thickness is 5 mm, Pitch (Pitch) is 0.984, the ball swivel week is 0.5 s. The range of whole scanning starts from the top of the diaphragm to the pelvic cavity in cephalopod direction.The delay time of arterial scanning is 20 to 25 s after injection, the delay time of venous phase scanning is 50 to 55 s. The images data of all patients were transmitted to Mxview workstation after scanning. A total of four stage of all patients 64 row helical CT scanning:the routine scanning, the upper abdomen stage, the pancreas stage, the enhancement scanning of vein stage. All four stage image data deposited by Mxview image post-processing workstation. The datas were exported as DICOM file and saved in light disk.4. The three dimensional reconstruction of original CT images of all patients: The original 64 mutle-slice thin data was inputted into the personal lenove computer and the format of these data was DICOM file. The adjusted images of arterial stage and pancreas stage and vein stage were all imported into Medical Image Three Dimensional Visualization System (MI-3DVS) for undergoing procedure segmentation and three-dimensional reconstruction. The reconstructed models were exported as STL format; The MI-3 DVS software is capable of automatic registration, coloring and other processing. The three-dimensional models of pancreas, pancreatic duct and the main peri-pancreatic venous were displayed in three dimensional stereoscopic model successfully.5. The simulation operation and the quantitative acquisition of pancreatic stump morphological features preoperatively based on MI-3DVS software:In the MI-3DVS software, the pancreatic essence and the pancreatic duct were transected referring to the left edge of the superior mesenteric vein in all patients. The acquisition of remnant pancreatic parenchyma volume (RPPV), the main pancreatic duct diameter (MPDD) and the area of cut surface (AOCS) with inherent software volume rendering algorithm, the length and area measurement tool, respectively.6. Preoperative and intraoperative clinical data and main indicators related to the operation:The demographic data, preoperative laboratory test results, intra-operative variables related to the operation and postoperative laboratory test results related to recovery and main postoperative complications..7. Perioperative management:All operations were performed by the chief physicians in hepatopancreatobiliary surgery department helped by a dedicated team. The majority of patients underwent PD with Child reconstruction, pylorus-preserving pancreaticoduodenectomy (PPPD) was performed only in some rare instances of early-stage ampullary cancer. The pancreas was transected referring to the left edge of superior mesenteric vein in all patients; The surgery margin was checked by fast frozen section examination in all patients for pathological analysis.. All specimens were confirmed by pathological examination after surgery resection. All patients received prophylactic antibiotics and H2 Blockers for 3 days postoperatively, and somatostatin analogues until oral feeding recovery. All drain fluids were analyzed for amylase concentration on postoperative days1,3,5 and 7.8. Definition of pancreatic fistula and other outcomes:CR-POPF was determined by the guideline of the International Study Group on Pancreatic Fistula (ISGPF) which released in 2005. Combined grade B+C was defined as "clinical relevant pancreatic fistula" in this study. All other complications were classified according to the criteria proposed by Clavien and Dindo.9. Data processing:Statistical analyses were performed using SPSS 20.0 software for Windows. The values were expressed as means±SD or percentage. The chi-square test, fisher exact test and Mann-Whitney U test were appropriately used. Variables with P<0.100 were entered into a logistic regression model to determine independent risk factors of CR-POPF. The independent risk factors of the variables were expressed as odds ratios with their 95% confidence intervals. The optimal cut-off levels of the significant risk factors between the group with and without CR-POPF were sought by constructing receiver operating characteristics (ROC) curves. A P value< 0.05 (two-tailed) was considered to be statistically significant.Results1. Patient characteristics and morphological features of pancreatic stump:No significant differences were observed between patients with and without CR-POPF concerning age, coexistent diseases and hepato-renal function test. The patients in the CR-POPF group were dominantly male (P=0.020), had a greater BMI (25.1±1.5 kg/m2 vs 23.8±1.4 kg/m2, P=0.001).2. The outcome of three dimensional reconstruction and pancreatic stump morphological features:The 3D model reconstruction of 90 patients were all established successfully. The indication of PD in this study as fellowing:55 patients with carcinoma of head of pancreas,10 with duodenal cancer,7 with bile duct carcinoma,5 with adenocarcinoma of the ampulla of vater,3 with endocrine carcinoma,2 with intraductal papillary mucinous neoplasm (IPMN),3 with cholangitis,2 with cystadenoma,2 with solid and pseudopapillary tumor and 1 with benign neoplasms. The image of individualized three-dimensional model was clear and distinct, with strongly stereo sense. The size, shape, range and large blood vessels around the pancreas were all displayed clearly. The models of three dimensional reconstruction can be rotated in full dimension, scaled and combinated arbitrarily, also can be hyalinized or hided target organs. After the simulation of operation, the RPPV, MPDD and AOCS of all patients were successfully obtained. The RPPV (30.2±4.9 mL vs 25.6±3.7 mL, P<0.001) and AOCS (223.8±27.9 mm2 vs 195.6±27.1 mm2, P<0.001) were all greater in group A. In contrast, the MPDD was significantly less in group A (3.2±0.4 mm vs 3.5±0.5 mm, P= 0.020).3. Comparison of intraoperative data and postoperative outcomes between two groups:There were no significant differences between 2 groups associated with types of procedures, the operative time, intraoperative blood loss and intraoperative blood transfusion. The mean hospital stay was 51.2±10.8 days (P<0.001) in group A compared to 30.4±8.6 days in group B. The mean drain placement of group A compared to the group B was 40.2±9.4 days vs 21.0±8.2 days (P<0.001), indicating that the patients in the group A have a trend of higher hospital costs and delayed resume.4. Uni-and multivariate analysis of preoperative predictive factors of CR-POPF:In univariate analysis, male gender (P=0.026), BMI>25.3kg/m2 (P=0.002), RPPV>27.8 mL (P<0.001), MPDD<3.1mm (P=0.005) and AOCS >222.3mm2 (P<0.001) were all significantly associated with an increased risk of CR-POPF. In multivariate analysis, RPPV>27.8 mL (OR= 12.907,95% CI 1.602-104.004, P=0.016) and BMI>25.3 kg/m2 (OR=12.238,95% CI 1.822～82.215, P=0.010) was the only independent predictive factors of CR-POPF. RPPV, measured preoperatively by MI-3DVS software, was the strongest single predictive factor of CR-POPF. We calculated the cut-off value of RPPV with respect to the presence of CR-POPF. Receiver-operating characteristic curve analysis revealed that a cut-off value of 27.8 mL for RPPV, with a sensitivity and specificity of 77.8% and 86.1%, respectively. The area under the ROC curve of RPPV was 0.770 (95% CI 0.629～0.911, P<0.001), indicating that the RPPV was discriminating as values ranging from 0.7 to 0.8 represent reasonable discrimination.Conclusions1. We collected the high quality submillimeter 64 row CT data of pancreatic and peri-pancreatic vascular successfully;2. MI-3DVS provides a novel method for acquiring preoperatively pancreatic stump morphological features quickly and accurately;3. The BMI>25.3kg/m2 and RPPV> 27.8 mL are the independent risk factors in predicting of pancreatic fistula after PD in this study;4. RPPV> 27.8 mL is the single most effective predictor, preoperative measuring RPPV, combined with the patient’s BMI index may help to optimize the perioperative management of patients with PD in this study. |
PDF Full Text Request