J Huazhong Univ Sci Technol［Med Sci］ 35(6):793-800,2015 DOI 10.1007/s11596-015-1509-z J Huazhong Univ Sci Technol［Med Sci］ 35(6):2015
793
Is There Comparable Morbidity in Pylorus-preserving and Pylorus-resecting Pancreaticoduodenectomy? A Meta-analysis* Qi-jun CHEN (陈琦军)1†, Zhi-qiang HE (何志强)1†, Yan YANG (杨 艳)2, Yu-shun ZHANG (张宇舜)1, Xing-lin CHEN (陈星霖)3, Hong-ji YANG (杨洪吉)4, Shi-kai ZHU (朱世凯)4, Ping-yong ZHONG (钟平勇)4, Chong YANG (杨 冲)4#, He-shui WU (吴河水)1# 1 Pancreatic Disease Institute, Union Hospital, Tongji Medical College, Huazhong University of Science and Technology, Wuhan 430022, China 2 Community Health Service Center, the 9th Hospital of Wuhan, Wuhan 430081, China 3 Key Laboratory of Geriatrics of Health Ministry, Department of Geriatrics, Union Hospital, Tongji Medical College, Huazhong University of Science and Technology, Wuhan 430022, China 4 Organ Transplantation Center, Hospital of the University of Electronic Science and Technology of China and Sichuan Provincial People’s Hospital, Chengdu 610072, China. © Huazhong University of Science and Technology and Springer-Verlag Berlin Heidelberg 2015
Summary: Pancreaticoduodenectomy (PD) is the most effective treatment for patients with pancreatic head or periampullary lesions. Two major strategies exist: pylorus-preserving pancreaticoduodenectomy (PPPD) and pylorus-resecting pancreaticoduodenectomy (PRPD). However, it is yet unclear regarding the morbidity after PPPD and PRPD. This study analyzed the morbidity after PPPD and PRPD to determine the optimal surgical treatment of masses in the pancreatic head or periampullary region. A systematic search of databases identifying randomized controlled trials (RCTs) from the Cochrane Library, PubMed, EMBASE and Web of Science was performed. Outcome was compared by postoperative morbidity including overall morbidity, pancreatic fistulas, wound infections, postoperative bleeding, biliary leakage, ascites and delayed gastric emptying (DGE) rate between PPPD and PRPD. The DGE rate in the PRPD subgroups (conventional PD [CPD] and subtotal stomach-preserving PD [SSPPD], respectively) was also analyzed. The results showed that 9 RCTs including 722 participants were included for meta-analysis. Among these RCTs, 7 manuscripts described PRPD as CPD, and 2 manuscripts described PRPD as SSPPD. There were no significant differences in the overall morbidity, pancreatic fistulas, wound infections, postoperative bleeding, or biliary leakage between PPPD and PRPD. There was a lower rate of DGE with PRPD than that with PPPD (RR=2.15, P=0.03, 95% CI, 1.09–4.23). Further subgroup analysis indicated a comparable DGE rate for the CPD but a lower DGE rate for the SSPPD group than the PPPD group. However, the result did not indicate any difference between CPD and SSPPD regarding the DGE rate (P=0.92). It is suggested that PPPD is comparable to PRPD in overall morbidity, pancreatic fistulas, wound infections, postoperative bleeding and biliary leakage. The current data are not sufficient to draw a conclusion regarding which surgical procedure is associated with a lower postoperative DGE rate. Our conclusions were limited by the available data. Further evaluations of RCTs are needed. Key words: pancreaticoduodenectomy; pylorus; morbidity; meta-analysis
Pancreaticoduodenectomy (PD) is the standard of care for patients with malignant disease of the pancreatic head or periampullary region[1]. The conventional PD (CPD) procedure involves removing the pancreatic head, duodenum, common bile duct, gall bladder and distal portion of the stomach associated with adjacent lymph nodes[2]. Recently, subtotal stomach-preserving PD (SSPPD) surgeries have been widely performed for
Qi-jun CHEN, E-mail:
[email protected]; Zhi-qiang HE, E-mail:
[email protected]; † The authors contributed equally to this work. # Corresponding authors, Chong YANG, E-mail:
[email protected]; He-shui WU, E-mail:
[email protected] * This project was supported by the National Natural Science Foundation of China (No. 81372261).
treatment of pancreatic and periampullary carcinomas[3–7]. Relative to the CPD, the major modification in the SSPPD is the stomach resection margin. The stomach is divided 3 cm above the pylorus ring, and more than 95% of the stomach is preserved[3, 8]. As the pylorus is removed in both CPD and SSPPD, these two procedures were classified as pylorus-resecting PD (PRPD) in our previous study[9]. Theoretically, SSPPD preserves almost the entire stomach, which may have an advantage over CPD in postoperative mortality and long-term nutritional status. However, there is no high-quality study comparing the effects of CPD and SSPPD for the treatment of pancreatic and periampullary carcinomas. Relative to PRPD, pylorus-preserving PD (PPPD) preserves the pylorus and first portion of the duodenum, and the gastrointestinal reconstruction is restored through
794 J Huazhong Univ Sci Technol［Med Sci］ a duodenojejunostomy[10]. Currently, the theory of comparable operation safety between PPPD and PRPD has been accepted widely[11–13]. The initial intention of PPPD is to avoid excessive resection and to decrease the incidence of post-gastrectomy complications. However, recent studies have indicated comparable morbidity rates between PPPD and PRPD for the treatment of pancreatic and periampullary carcinomas[14, 15] and even higher delayed gastric emptying (DGE) rates for the PPPD procedure[16, 17]. Our previous meta-analysis indicated that PRPD had advantages over PPPD in DGE for periampullary and pancreatic carcinomas[9], but a recent randomized control trial (RCT) reported a comparable DGE occurrence rate and long-term nutritional effects between SSPPD and PPPD[3]. So, it is still unclear whether there is a difference in morbidity between PPPD and PRPD in cases with periampullary and pancreatic carcinomas. Therefore, we initiated an updated meta-analysis to evaluate the impact of the PPPD procedure on morbidity for periampullary and pancreatic carcinoma patients. Additionally, we initiated a subgroup analysis (CPD versus PPPD, SSPPD versus PPPD, respectively) to determine the difference. 1 MATERIALS AND METHODS 1.1 Literature Search Two investigators (He ZQ and Chen QJ) searched the Cochrane Library, EMBASE, PubMed and Web of Science for relevant articles published before June 20, 2014 without date and language restrictions. A sensitive search was performed using terms related to PPPD, Whipple’s PD, conventional PD, SSPPD, PRPD for random control trials (RCTs) on pancreatic or periampullary carcinoma patients. The following keywords and medical subject headings were used for search: ‘‘Whipple operation’’, ‘‘pylorus’’, ‘‘pancreaticoduodenectomy’’, ‘‘pylorus-preserving’’ and ‘‘pancreatic/periampullary tumor’’. The search strategy also used terms such as ‘‘standard pancreaticoduodenectomy’’, ‘‘classic duodenopancreatectomy’’, “conventional pancreaticoduodenectomy”, ‘‘subtotal stomach-preserving pancreaticoduodenectomy”, ‘‘duodenopancreatectomy’’, ‘‘Whipple procedure’’, and ‘‘pylorus preserving pancreaticoduodenectomy’’ to identify relevant information. We entered Boolean operators (AND, OR, NOT) to combine or exclude search terms. The search was limited initially to publications of human RCTs. Additionally, we screened the references of selected papers to identify additional potentially eligible studies. Investigators and experts in the field of pancreatic surgery were contacted to ensure that all the relevant studies were identified. The results were hand-searched for eligible trials. The results were verified and arbitrated by a third investigator (Wu HS). 1.2 Data Extraction The primary goal of this study was to determine the morbidity of PPPD and PRPD in patients with pancreatic or periampullary cancer. Only RCTs reporting quantitative data for at least one of the following outcomes were selected for data extraction: overall morbidity, pancreatic fistulas, wound infections, postoperative bleeding, biliary leakage, ascites and DGE rate. Two authors (Yang Y and Chen XL) independently extracted data from the trials and subsequently compared the results. Any discrepan-
35(6):2015
cies between the authors were resolved by consensus. All the data were evaluated for internal consistency, and disagreements were resolved via a discussion with a third author (Wu HS). The quality was assessed using criteria such as adequate blinding of randomization, completeness of follow-up, and objectivity of outcome measurements as previous description[18]. 1.3 Selection and Exclusion Criteria Only RCTs that directly compared the morbidity of patients with pancreatic or periampullary cancer who underwent a PPPD or a PRPD were included. The exclusion criteria included the follows: (1) papers lacking a control group; (2) studies that were not published as an original paper, such as conference abstracts and letters to the editor; and (3) duplicate publications: if a paper occurred more than once in one of the databases, only the original manuscript was included. 1.4 Subgroup Analysis We were also interested in the potential difference between CPD and SSPPD in PRPD procedures and thus included a subgroup analysis (CPD versus PPPD and SSPPD versus PPPD, respectively). 1.5 Statistical Analysis This meta-analysis was performed using STATA 11.0 (STATA Corp, College Station, USA) and RevMan 5 (http://ims.cochrane.org/revman/download). All the P-values were two-sided. We analyzed the dichotomous variables by calculating the risk ratio (RR) and corresponding 95% confidence interval (CI) as the summary statistic. For the meta-analysis, we used a fixed-effects (weighted with inverse variance) or a random-effects model based on the heterogeneity of the included studies[19]. For each meta-analysis, Cochran’s Q statistic and I2 statistics were first calculated to assess the heterogeneity among the proportions of the included trials. If the P-value was less than 0.10, the assumption of homogeneity was deemed invalid, and the random-effects model was utilized after exploring the causes of heterogeneity[20]. When the results of the two models were substantially different, the random-effects model was presented. For each outcome with data from five or more studies, we began the analysis by creating a funnel plot, comparing the magnitude of the relative risk on the horizontal axis with the standard error of the log relative risk on the vertical axis. We used Begg’s and Egger’s tests[21] to detect possible publication biases. Data were considered significant if the probability of a chance occurrence was less than 5% (P
