Concomitant splenic artery ligation has no preventive effect on left‐sided portal hypertension following pancreaticoduodenectomy with the resection of the portal and superior mesenteric vein confluence for pancreatic ductal adenocarcinoma

Abstract Background Left‐sided portal hypertension (LSPH) caused by splenic vein (SV) division in pancreaticoduodenectomy (PD) with portal vein (PV) resection remains challenging. The current study aimed to investigate the efficacy of splenic artery (SA) ligation in preventing LSPH. Methods One‐hundred thirty patients who underwent PD with PV resection for pancreatic ductal adenocarcinoma were classified into SV and SA preservation (SVP, n = 30), SV resection and SA preservation (SVR, n = 59), and SV resection and SA ligation (SAL, n = 41). The postoperative incidence of LSPH was assessed. Results The incidence of variceal formation in SVP, SVR, and SAL were 4.8%, 53.2%, and 46.4% at 3 mo, 13.0%, 71.2%, and 62.5% at 6 mo, and 25.0%, 87.5%, and 87.1% at 12 mo, respectively. The rate was significantly higher in SVR at 3 and 6 mo (P = .001 and P < .001, respectively) and in SVR and SAL (P < .001) at 12 mo. Variceal hemorrhage occurred only in SVR (n = 4). The platelet count ratio at 3, 6, and 12 mo began to significantly decrease from 3 mo in SVR (0.77, 0.67, and 0.60, respectively; P < .001) and 6 mo in SAL (0.91, 0.73, and 0.69, respectively; P < .001). The spleen volume ratio also showed significant increase from 3 mo in SVR (1.24, 1.34, and 1.42, respectively; P < .001) and 6 mo in SAL (1.31, 1.32, and 1.34, respectively; P < .001). SVR and SAL were significant risk factors for variceal formation at 12 mo (odds ratio, 21.0 and 20.3, respectively). Conclusion In PD with PV resection, SAL delayed LSPH but could not prevent its occurrence.


| INTRODUC TI ON
Pancreaticoduodenectomy (PD) for pancreatic ductal adenocarcinoma (PDAC) commonly requires the resection and reconstruction of the portal vein (PV) / superior mesenteric vein (SMV) confluence. The splenic vein (SV) often needs to be resected to achieve tumor clearance or maximize the mobility of PV/SMV and is rarely reconstructed. 1 The division of the SV can lead to left-sided portal hypertension (LSPH), causing gastrointestinal variceal formation, thrombocytopenia, and splenomegaly. [1][2][3][4] To prevent postoperative LSPH, additional concomitant surgical procedures have been reported, with some surgeons claiming to have reconstructed the SV to PV/SMV, inferior mesenteric vein (IMV), or left renal vein. [5][6][7][8] However, this reconstruction is not simple because SV tends to be too long, with its long-term patency being unknown.
Conversely, splenic artery ligation (SAL) is a simple procedure for portal modulation. [9][10][11][12] Previously, we reported the efficacy of concomitant splenic artery resection (SAR) in preventing LSPH after PD with resection of the PV/SMV confluence. 2 Therefore, we performed concomitant SAL to reduce the incidence of LSPH following PD with PV/SMV resection and SV division since September 2016. Here we aimed to determine the effect of PD-SAL on LSPH incidence, paying attention to variceal formation, collateral development, platelet counts, and spleen volume.

| Preoperative characteristics and surgical outcomes
We collected data on various preoperative factors such as age, sex, body mass index (BMI), maximum tumor size on CT, performance status, complete blood count parameters, albumin, carcinoembryonic antigen (CEA) and carbohydrate antigen (CA19-9), spleen volume, the presence of preoperative chemotherapy or chemoradiotherapy , T and N factors according to the Union for International   Cancer Control (UICC) 8th classification, and resectability of the   tumor (classified into resectable, R; borderline resectable, BR; and unresectable, UR) according to The National Comprehensive Cancer Network guideline, 14 based on the findings of MDCT as previously reported. We also evaluated surgical procedures and outcomes, including PD and subtotal stomach-preserving PD (SSPPD), the left gastric vein (LGV) division or preservation including the confluence variant, the IMV division or preservation including the confluence variant, intraoperative blood loss, operative duration, degree of postoperative complications according to the Clavien-Dindo (C-D) classification, 15 pancreatic fistula according to the International Study Group on Pancreatic Fistula, 16 the presence of pathological PV invasion (pPV), the achievement of curative resection (R0 resection), and postoperative hospital stay duration.

| Surgical procedures
An anterior approach to the superior mesenteric artery (SMA) has been the standard resection technique in PD for PDAC. 17 The middle colic vein (MCV) was resected in all cases for lymph node dissection. The anastomosis between PV and SMV was performed using a 6-0 nonabsorbable running suture. An interposition graft was used only when a primary repair was not feasible for the anastomotic tension. When the SV, LGV, or IMV was involved in the tumor, they were divided and not reconstructed. Reconstruction of the digestive tract was performed using a modified Child method, end-to-side pancreaticojejunostomy, end-to-side hepaticojejunostomy, and end-to-side or side-to-side gastrojejunostomy. In SAL, the root of the SA was clumped using Hem-o-lock (Weck Surgical Instruments; Teleflex Medical, Morrisville, NC).

| Assessment of LSPH
To assess the development of LSPH, newly developed digestive varices and collateral routes were evaluated at postoperative 3, 6, and 12 mo using enhanced MDCT by the radiologist (N.M.), who was not informed of patient characteristics or outcomes. Esophageal, gastric, pancreatic, and colonic varices were diagnosed when the dilated and beaded veins were detected within the submucosal layer of each organ compared with preoperative. 2 Development of collateral pathways from the divided SV were diagnosed when the diameter of collateral routes became 1.5 times larger than the preoperative. Spleno-renal and gastro-renal shunts were evaluated as spleno-systemic routes. Superior and inferior routes were evaluated as spleno-portal routes according to a previous study. 18 The superior route was defined as a pathway starting in the divided SV, following a superior and rightward direction through perigastric veins and LGV, and finally ending in the PV. The inferior route was defined as a pathway starting from the divided SV, joined to venous routes in the mesocolon through the omental arcade and/or the IMV, and proceeding in an inferior and rightward direction to end in the SMV. 3,18 Blood supply to the spleen after PD-SAL was evaluated based on enhanced MDCT within postoperative 14 d.
Platelet count data were collected preoperative and at postoperative 3, 6, and 12 mo. The platelet count ratio was calculated as postoperative count divided by preoperative count. Postoperative thrombocytopenia of grade 2 or higher (less than 75,000/mL) based on the Common Terminology Criteria for Adverse Events Version 5.0 was also evaluated. The total spleen volume was estimated by tracing the spleen on each transverse CT image obtained at 2.0-mm intervals. Spleen volume was measured preoperative and at postoperative 3, 6, and 12 mo. The spleen volume ratio was calculated as the postoperative volume divided by the preoperative volume. When patients did not undergo enhanced CT or a blood test at that time (with a margin of 1 mo at 3 and 6 mo, and 2 mo at 12 mo), they were excluded from each analysis. We

| Statistical analyses
All continuous values are presented as median (range). Continuous  Table 1 shows comparisons of patient background and surgical outcomes among the groups. The preoperative CEA level, the ratio of CRT to upfront surgery, blood loss level, and postoperative hospital stay were significantly highest in the PD-SVR group. LGV or IMV preservation was significantly higher in the PD-SVP group than in the other groups. There was no significant difference in postoperative complications of C-D grade IIIa or higher and pancreatic fistula.    (Table S3). in PD-SVR was significantly higher than that in PD-SVP at 3, 6, and 12 mo (P = .001, P < .001, and P < .001, respectively). The ratio in PD-SAL was also higher than that in PD-SVP at 3, 6, and 12 mo (P = .011, P = .001, and P = .006, respectively).

| Variceal formation at 12 mo postoperative according to venous drainage routes from the divided SV
The LGV and IMV were preserved more frequently in PD-SVP (Table 1). Collateral routes rarely developed in PD-SVP ( Table 2).
Therefore, the effect of venous drainage routes in LSPH was evalu-

| DISCUSS ION
In this study, we investigated the efficacy of splenic artery (SA) ligation in preventing LSPH and elucidated the following: (a) SAL delayed the occurrence of LSPH but could not prevent it in terms of digestive variceal formation, thrombocytopenia, and splenomegaly; (b) Variceal hemorrhage within 12 mo occurred only in PD-SVR but did not occur in PD-SAL; (c) The arterial supply to the spleen after SAL was mainly from the left gastric (100%) and subphrenic (93%) arteries through or around the stomach without causing any major complications; and (d) The preservation of LGV-SV or IMV-SV confluence had no effect in preventing variceal formation.
In PD with the PV/SMV confluence resection, the divided SV is rarely reconstructed. In a study, the SV was not reconstructed in more than 95% of the 251 patients who underwent PD with SV resection between 2005 and 2014. 1 Most surgeons pay no attention to LSPH after SV division because they believe venous blood from the divided SV will successfully drain to the venous circulation of the stomach and colon without causing any major problems. However, the accumulated blood in the spleen due to SV division causes LSPH, which then manifests as variceal formation, collateral development, thrombocytopenia, and splenomegaly. 1,2 Particularly, variceal formation represented by esophageal, gastric, pancreatic, and colonic varices has a risk of lethal hemorrhage requiring emergency treatment. 1-3 LSPH is a major problem that needs to be addressed.
Preservation of the LGV or IMV prevents postoperative LSPH in a small number of patients. 5,6,18,19 In this study, however, the preservation of the LGV-SV or IMV-SV confluence had no effect on the occurrence of LSPH. The accumulated blood in the spleen due to SV division, flowed to the perigastric circulation via the pre-   LGV preservation (n = 8) LGV division (n = 63) P LGV-PV (n = 3) LGV-SV (n = 5) In conclusion, although SAL could not prevent LSPH, it could delay the occurrence of LSPH without causing variceal hemorrhage after PD with PV/SMV confluence resection.

ACK N OWLED G M ENTS
We thank Toru Ogura (clinical research support center, Mie University Hospital) for his support for data analysis. We also thank Editage (www.edita ge.com) for English language editing.