Overall Postoperative Morbidity and Pancreatic Fistula Are Relatively Higher after Central Pancreatectomy than Distal Pancreatic Resection: A Systematic Review and Meta-Analysis

Objective To compare the intraoperative and postoperative outcomes of central pancreatectomy (CP) with distal pancreatectomy (DP). Methods A systematic literature search was performed on electronic databases from MEDLINE, Embase, and PubMed from 1998 to 2018. Statistical analysis and meta-analysis were performed using statistics/data analysis (Stata®) software, version 12.0 (StataCorp LP, College Station, Texas 77845, USA). Dichotomous variables were analyzed by estimation of relative risk (RR) with a 95 percent (%) confidence interval (CI) and continuous variables were analyzed by standardized mean differences (SMD) with 95% CI. Results Twenty-four studies with 593 CP and 1226 DP were included in the meta-analysis. CP had significantly longer operation time (SMD: 1.03; 95% CI 0.62 to 1.44; P < 0.001) and lengthier postoperative hospital stay (SMD: 0.63; 95% CI 0.20 to 1.05; P < 0.001) and lengthier postoperative hospital stay (SMD: 0.63; 95% CI 0.20 to 1.05; P < 0.001) and lengthier postoperative hospital stay (SMD: 0.63; 95% CI 0.20 to 1.05; P < 0.001) and lengthier postoperative hospital stay (SMD: 0.63; 95% CI 0.20 to 1.05; P < 0.001) and lengthier postoperative hospital stay (SMD: 0.63; 95% CI 0.20 to 1.05; P < 0.001) and lengthier postoperative hospital stay (SMD: 0.63; 95% CI 0.20 to 1.05; P < 0.001) and lengthier postoperative hospital stay (SMD: 0.63; 95% CI 0.20 to 1.05; P < 0.001) and lengthier postoperative hospital stay (SMD: 0.63; 95% CI 0.20 to 1.05; P < 0.001) and lengthier postoperative hospital stay (SMD: 0.63; 95% CI 0.20 to 1.05; P < 0.01). Estimated blood loss was significantly lower in CP (SMD: −0.34; 95% CI −0.58 to −0.09; P = 0.007). Overall postoperative morbidity (RR: 1.30; 95% CI: 1.13 to 1.50; P < 0.001), overall pancreatic fistula (RR: 1.41; 95% CI: 1.20 to 1.66; P < 0.001), clinically relevant fistula (RR: 1.64; 95% CI: 1.25 to 2.16; P < 0.001), and postoperative hemorrhage (RR: 1.90; 95% CI: 1.18 to 3.06; P < 0.05) were all significantly higher after CP. On long-term follow-up, DP patients were more likely to have postoperative exocrine (RR: 0.56; 95% CI: 0.37 to 0.84; P < 0.05) and endocrine (RR: 0.27; 95% CI: 0.18 to 0.40; P < 0.001) insufficiency. There was no statistically significant difference in transfusion requirement, postoperative mortality, reoperation, and tumor recurrence. Conclusion CP is associated with significantly higher morbidity and clinically relevant pancreatic fistula. CP should only be reserved for selected patients who require postoperative pancreatic function preservation.


Introduction
Complete surgical resection is the only potentially curative treatment for pancreatic cancer. However, only 15-20% of patients are amenable to resection on initial diagnosis [1][2][3]. The distal pancreatectomy (DP) is considered as a standard surgical procedure for lesions located in the pancreatic neck and body [2]. Unfortunately, during the resection of benign and low-malignant lesions, normal pancreatic parenchyma is resected in the DP and may result in loss of pancreatic function and possible postoperative exocrine and endocrine impairment. After the introduction of the first central pancreatectomy (CP) with reconstruction by Dagradi and Serio in 1982, this procedure has been used as a parenchyma-preserving surgical procedure for resection of benign and low-malignant lesions of neck and proximal body of pancreas. After that, the procedure has been advanced gradually from open surgery to laparoscopic and robotic approaches [4,5].   Current literature has reported a relatively higher incidence of postoperative new-onset diabetes mellitus after DP than CP and pancreaticoduodenectomy (PD) [6,7]. Kang et al. have reported that resected pancreatic volume was an independent risk factor for postoperative endocrine impairment [8]. In CP, the volume of remnant pancreas is responsible for maintaining postoperative endocrine and exocrine function, but there are still controversies regarding the management of additional pancreatic stump. The morbidity following CP is comparatively higher than other standard pancreatic resections, pancreatic fistula being major morbidity [9,10]. CP has relative benefits of preserving normal pancreatic parenchyma and spleen, but potential challenges to reconstruct additional pancreatic stump and high incidence of postoperative pancreatic fistula create a dilemma to choose appropriate surgical procedure.

Methods
The systematic review and meta-analysis were performed according to the Preferred Reporting Items for the Systematic Review and Meta-Analysis Protocols (PRISMA-P) guideline [11].

Search Strategy.
A systematic literature search was performed on electronic databases from Ovid MEDLINE (R), Embase, and PubMed from 1 January 1998 to 31 December 2017. Search headings used were "(central pancreatectomy OR medial pancreatectomy OR middle pancreatectomy OR segmental pancreatectomy) and (distal pancreatectomy OR left pancreatectomy)". Searches were performed without any restrictions and all the abstracts, studies, and citations were reviewed. 2181 studies were found after the comprehensive search of the database. 1988 articles were excluded after screening the title abstracts and duplicated materials. 193 studies were evaluated in detail and ultimately 24 eligible studies were included in systematic review and meta-analysis ( Figure 1).

Study Selection and Quality Assessment.
Two reviewers independently screened all the selected citations independently. Any disagreement between the two reviewers was resolved by discussion with the corresponding author. All the retrospective and prospective matched pairs and nonmatched pairs comparing CP and DP were extracted.
Inclusion criteria were (1) Original English articles; (2) Patients with benign pathology of low-malignant tumors of the pancreatic neck or proximal body; (3) Studies comparing the clinical outcomes between CP and DP; (4) Studies that provided adequate information about demographic characteristics and intraoperative and postoperative outcomes.
Quality assessment was done by the Newcastle Ottawa Scale (NOS). Studies selected had a score above 5. Scoring criteria were based on the selection of study groups, the comparability of groups, and the ascertainment of either exposure or outcome. One study had 9, 3 had 8, 17 had 7, and 3 had 6 out of possible 9 scores.

Data Extraction.
Data were extracted for (a) demographic characteristics, (b) intraoperative outcomes (operation time, intraoperative blood loss, and transfusion requirement), (b) short-term postoperative outcomes (postoperative hospital stay, overall morbidity, pancreatic fistula, clinically relevant pancreatic fistula, postoperative hemorrhage, reoperation, and 30-day mortality), and (c) long-term outcomes (overall endocrine function, Insulindependent diabetes mellitus, exocrine function, and tumor recurrence). If two articles were published by the same authors or from the same institution, a comparatively more informative study with the maximum population was selected. When data was found in median and range, the mean and standard deviation was estimated as described by Wan et al. and Luo et al. [12,13].

Statistical Analysis.
Statistical analysis and meta-analysis were performed by statistics/data analysis (Stata ® ) software, version 12.0 (StataCorp LP, College Station, Texas 77845 USA). Dichotomous variables were analyzed by estimation of relative risk (RR) with a 95 percent (%) confidence interval (CI) and continuous variables were analyzed by standardized mean differences (SMD)/weighted mean differences (WMD) with 95% CI. P value <0.05 was considered a statistically significant difference between the two groups. Heterogeneity was defined as low, moderate, and high based on I square value (<25%: low; 25-75%: moderate; >75%: high). Heterogeneity with a high I square value >30% and P value <0.05 was considered statistically significant. Fixed effect (Mantel-Haenszel) model was used when there was no significant heterogeneity and the random effects (DerSimonian and Laird) model was used for those with significant heterogeneity. Publication bias was examined in a funnel plot using Begg's and Egger's tests. Publication bias was considered to be present when the P value was <0.1.

Quantitative Data Synthesis
Results of meta-analysis are included in Table 3.

Heterogeneity and Publication Bias
Significant heterogeneity (high I square value >30% and P value <0.05) was observed in three continuous variables (operation time, estimated blood loss, and length of hospital stay). All three outcomes were pooled on the random-effects model. Therefore, a sensitivity analysis was done by omitting 1 study at a time and the pooled RR was calculated for the remaining studies to identify the potential source of heterogeneity between studies, but no single study significantly Yamaguchi et al. [17] Balzano et al. [29] Su et al. [31] Shibata et al. [15] Müller et al. [28] Crippa et al. [36] Ocuin et al. [22] Hirono et al. [18] Cataldegirman et al. [27] DiNorcia et al. [7] Shikano et al. [16] Kang et al. [24] Dumitrascu et al. [34] Du et al. [19] Zhan et al. [21] Zureikat et al. [23] Song et al. [26] Mise et al. [14] Herrera-Cabezón et al. [32] Jilesen et al. [33] Dokmak et al. [35] Boggi et al. [30]  affected the primary outcome or heterogeneity. This may be due to a difference in surgical skills among surgeons and postoperative management strategy. Publication bias was considered to be present when the P value was <0.1. Assessment of publication bias of dichotomous data using the funnel plot showed symmetrical distribution and the Egger's test did not show any statistical significance.

Discussion
This meta-analysis included 24 studies involving 1819 patients and assessed the intraoperative and postoperative outcomes after CP or DP. Long operation time in the CP is due to its distinct anatomical location and complex surgical procedure. But one [29] study reported a comparatively shorter duration of operation in the CP group. In our study, on comparing intraoperative outcomes between the two procedures, estimated blood loss was statistically higher during distal pancreatectomy. Thus, a relatively higher number of patients required transfusion after distal pancreatectomy. In CP, the method of reconstruction of the distal stump was recorded in 17 studies with 359 patients.

Conclusion
Our study suggests that pancreatic neck resection requires longer operation time, high rate of postoperative pancreatic fistula, and high morbidity and mortality, but less amount of normal parenchyma is resected in contrast to DP. The incidence of postoperative endocrine and exocrine insufficiency is relatively less after CP. Few previously published meta-analyses have shown that CP can be feasible for benign and low-malignant lesions of the pancreatic neck and proximal body. In contrast to those studies, our study showed that the incidence of serious postoperative morbidity (i.e., clinically relevant pancreatic fistula) was significantly high after CP. In our study, the cumulative incidence of postoperative endocrine insufficiency was relatively lower in Asians compared to the western population (19.5% versus 26%). We believe that postoperative diabetes can be well controlled with oral hypoglycemic drugs and insulin. However, regarding higher morbidity and mortality after CP, it is still questionable for patient safety.
Although CP has the advantage of postoperative pancreatic function preservation, due to lengthy operation time, high rate of complications, and higher incidence of postoperative fistula. We conclude DP is a comparatively safe procedure compared to CP. Therefore, for tumors in pancreatic body and tail, DP is the safest, most feasible, and accepted procedure unless pancreatic parenchyma preservation is of utmost importance.

Data Availability
The data types used to support the findings of this study are included in the article and supplementary information files.

Additional Points
Significant heterogeneity existed in three continuous variables (operation time, blood loss, and length of hospital stay), which indicates the difference is surgical skills between surgeons, studies conducted in different countries, different sample sizes, and so on. This study lacks data for postoperative gastroparesis and only a few studies recorded data on tumor recurrence. Moreover, only prospective observational and retrospective studies were identified during the literature search, so this study may not be as reliable as metaanalysis was performed on clinical trials and international multicenter studies.