Preoperative splenic artery embolism followed by splenectomy is safe and effective in patients with sinistral portal hypertension

Background
 Although preoperative splenic artery embolism (SAE) has been widely used for splenomegaly, the efficiency and safety of preoperative SAE in patients with sinistral portal hypertension (SPH) is unknown. Methods We designed a retrospective cohort of SPH patients who received preoperative SAE in our hospital (February 2018 to September 2020) and compared to those who received splenectomy only, in terms of intraoperative and postoperative outcomes. Results In all, 59 patients (18 patients received preoperative SAE) were analyzed. The median age was 44.7 years. Preoperative SAE reduced the intraoperative blood loss (637.0 vs. 420.3 ml, P = 0.041) and operation time (174.0 vs. 141.5 min, P = 0.012). The incidence of complications including postoperative pancreatic fistula (POPF), bleeding, and thromboembolism was comparable. Multivariate analysis showed that SAE was a protective factor for intraoperative blood loss and operation time, while prior pancreatic pseudocyst/abscess was a risk factor. Conclusions Preoperative SAE could reduce intraoperative blood loss and operation time in SPH patients without increasing the incidence of complications compared to splenectomy only.


Introduction
Sinistral portal hypertension (SPH), also named left-sided portal hypertension, is a rare clinical syndrome. Unlike common causes of portal hypertension, occlusion of the splenic vein results in hypertension in the drainage system for splenic blood, which further leads to the hypertension of gastric varices or esophageal varices [1]. The pathogenesis of occlusion can vary, and pancreatic disorders play a critical role [2][3][4].
Although most patients are asymptomatic, upper gastrointestinal bleeding has been reported in 4-17% of cases [5], and these patients might develop hemorrhagic shock and even die [2]. Thus, many treatment options including splenectomy, splenic artery embolism (SAE), and splenic vein stenting (SVS) have been proposed for fatal bleeding events [6][7][8][9]. However, in light of the common background shared by these patients-namely, a history of pancreatitis-patients with SPH usually present with severe fibrosis adjoining the spleen, which could increase the surgical difficulty and the risk of excessive blood loss. Thanks to the development of interventional therapy, a preoperative SAE method that could reduce blood flow to the spleen and consequently lower the possibility of intraoperative blood loss has been proposed [10]. Nevertheless, no studies have compared the effectiveness and safety of this new method with conventional splenectomy in SPH patients to date; thus, we were motivated to conduct this study.
We designed a retrospective cohort study to compare splenectomy only with SAE combined with splenectomy in patients with SPH to determine whether preoperative SAE could reduce intraoperative blood loss and the operation time. Secondary outcomes such as postoperative complications, expenses, and length of stay (LOS) were also analyzed. Zihe Wang and Mao Li serve as co-authors.

Patients
The study had a retrospective cohort design, and patients were grouped into two cohorts on the basis of the proce-

Interventional and surgical procedures
SAE procedures were performed as previously described [11]. A 2.5-or 5-Fr microcatheter via femoral artery access was advanced into the splenic artery, and the splenic artery was embolized at the level of the splenic hilum distal to the left gastroepiploic artery. Within 24 h, three experienced surgeons in our department performed open splenectomies, and they also performed the surgical procedures in the other group. Because most of the patients in both groups had a history of pancreatitis, those patients also underwent distal pancreatectomy.

Intraoperative and postoperative outcomes
This study focused on the effect of preoperative SAE on decreasing the amount of blood loss during surgery. Considering the difficulty in precise measurement of intraoperative blood loss, we chose the equation proposed by Mercuriali et al. [12], and the blood loss in this equation was red blood cell loss per se. Therefore, intraoperative blood transfusion in this study reflected packed red cells. Complications including postoperative pancreatic fistula (POPF), bleeding (including bleeding from gastric varices and surgical sites), thromboembolism, postoperative abdominal fluid collection, pleural effusion, and adhesive ileus were all recorded during follow-up. POPF was defined by the International Study Group on Pancreatic Fistula (ISGPF) criteria [13]. Cost data represented the use of all expense categories including medications, supplies, cost of surgery, ICU staff, and cost of interventional therapy if applied.

Statistical analysis
Continuous data that followed the normal distribution are presented as means with standard deviations (SDs), while nonnormally distributed data are presented as medians with interquartile ranges (IQRs). Categorical data are presented as frequencies and percentages. Pearson's χ 2 test or Fisher's exact test were chosen to analyze categorical data. Two-tailed unpaired t-tests were employed to compare normally distributed data, whereas Mann-Whitney's U tests were used for nonnormally distributed data or ranked data. Owing to the nonnormal distribution of the estimated blood loss data and operation time data, we transformed the continuous variables into binary variables based on cutoff values, defined as median blood loss and operation time in the splenectomy cohort, to perform the Firth logistic regression analysis [14]. A univariate analysis was performed to identify the potential risk factors (P < 0.1 for entry), and then a multivariate analysis was performed.
A P < 0.05 denoted statistical significance. All statistical analyses were performed using RStudio 1.4 (Boston, MA).

Demographic and clinical characteristics
A total of 59 patients were analyzed (

Risk factors for excessive intraoperative blood loss in patients with sinistral portal hypertension who underwent splenectomy
We conducted univariate and multivariate analyses using demographic and clinical characteristics and intraoperative variables to identify risk factors for intraoperative blood loss (Table 3). In the univariate analysis, factors associated with excessive intraoperative blood loss included prior abdominal surgery, prior pancreatic pseudocyst/abscess,

Risk factors for prolonged operation time in patients with sinistral portal hypertension who underwent splenectomy
We also analyzed the risk factors for prolonged operation time (

Discussion
Portal hypertension is a well-discussed topic, and its causes can be groups as prehepatic, hepatic, and posthepatic while hepatic (mainly cirrhosis) is the most common one. However, for patients with SPH, the etiology evidently differed from those with hepatic portal hypertension. Considering that clinical manifestations of SPH mainly come from primary diseases and splenic vein occlusion, splenectomy is preferred for patients who present with gastrointestinal bleeding. Based on our experience, a history of pancreatitis could give rise to a greater risk of adhesions in the abdominal area, making splenectomy for SPH patients much more difficult than that for patients with other causes of splenomegaly. Thus, a safer, more effective procedure must be developed. With surgeons at the crossroads of treatment, interventional therapy has come to the center stage. Wang and colleagues conducted a study of 14 patients with SPH, and reported no severe complications, such as splenic abscess or rupture of the spleen [15]. However, they also admitted that postembolization syndrome was the most frequent complication, although they did not report the exact number of events. Splenic vein stenting was also proposed to be an effective and safe operation for SPH patients with gastrointestinal bleeding, and it was proven to be more effective than SAE in the control of rebleeding [16]. Notably, 17.4% (4/23) and 7.1% (1/14) of patients in the SAE group and the SVS group, respectively, required subsequent splenectomy due to rebleeding. The study by Fernandes et al. also reported that a high proportion (3/5) of patients who received SAE were submitted to splenectomy finally, even though the sample size was small [3]. In fact, splenic artery embolism was also reported to cause various kinds of complications-not only postembolization syndrome, but also splenic abscess, splenic vein thrombosis, and pancreatitis. Splenectomy, as a surgical approach, could efficiently reduce the risk of further hemorrhage by decreasing venous outflow through the collateral circulation and associated varices. Notably, surgeons may also need to address primary pancreatic diseases when performing splenectomy. In light of that, splenectomy might still be preferable due to its effectiveness; meanwhile, for those patients in poor conditions and cannot bear surgery, the choice of interventional therapies might be preferred. Under these circumstances, the combination approach of surgical and interventional therapy finally drew the attention of surgeons. Preoperative SAE was first proposed by Poulin et al. in 1993 [10]. This method was meant to control arterial flow and lower the risk of intraoperative bleeding, although it might also cause thrombosis in the venous vessels. In addition, it has been reported to decrease the splenic volume, thereby improving the surgical view and intraoperative exposure [17]. Reso and colleagues performed 19 laparoscopic splenectomies after SAE in patients with massive splenomegaly and reported a median estimated blood loss of 200 ml and a median operation time of 130 min [18]. In addition, the results of one comparative study implied a shorter operation time, less estimated blood loss, and shorter postoperative hospital stay for patients who received preoperative SAE [11]. It might raise the question of why the intraoperative blood loss in those studies was less than ours (637.0 ml for the splenectomy-only group and 420.3 ml for the preoperative SAE group). We believed that differences in the primary disease and choice of open surgery could account for this discrepancy. As we mentioned, 98.3% (58/59) of patients in our study suffered from pancreatitis, which resulted in severe fibrosis in the abdominal cavity and might lead to more intraoperative blood loss and a prolonged operation time. In addition, we performed open splenectomy on all patients in this study since portal hypertension was a contraindication for laparoscopic splenectomy. It was widely believed that open splenectomy could cause more intraoperative blood loss than laparoscopic splenectomy.
There was one notably significant difference observed between the patients: more patients in the splenectomyonly group received distal pancreatectomy (39/41 vs. 13/18, P = 0.023), which might raise questions regarding whether this discrepancy could lead to a prolonged operation time and more blood loss. In our hospital, the diseased pancreas is divided with a stapler so that the tail of the pancreas and the spleen can be resected at almost the same time. In other words, distal pancreatectomy did not take a longer time, which was supported by the results of logistic regression analysis. Furthermore, we also conducted an analysis with patients who received distal pancreatectomy, and proved the effectiveness of preoperative SAE in reducing operation time (174 (145-212) vs. 148 (125-167.5) min, P = 0.046)) and blood loss (662.2 (421.4-1120.7) ml vs. 428.5 (326.8-600) ml, P = 0.048).
The incidence of some complications reported in this study notably differed from others. Compared to studies that performed preoperative SAE and splenectomy, this research reported that more patients developed postoperative abdominal fluid collection (9/59), but fewer developed pleural effusion (2/59) [18]. We thought differences in the pathological backgrounds might still be a crucial reason for these differences among studies. Some of our patients underwent drainage of pancreatic pseudocysts due to a history of pancreatitis, and the fibrosis in abdominal cavity might increase the risk of fluid collection postoperatively. In our study, the incidence of postoperative bleeding was 9.8% (4/41) and 5.6% (1/18) in the splenectomy-only group and preoperative SAE group, respectively. Two patients presented with bleeding due to stress ulcers and bleeding from the surgical site, which should not be interpreted as rebleeding from gastric varices. The occurrence of rebleeding among patients with SPH in other studies varied from 0 to 17.4% [3,6,15,16,[19][20][21], while the choice of therapy included observation, splenectomy, SAE, and SVS. It should be clarified that many studies had a relatively small sample size and included patients without gastrointestinal bleeding, which increased the difficulty in comparing the effectiveness.
There were some limitations to this study. First, it was retrospectively designed, and thus, selection bias may exist. However, there was no notable difference in the demographic and clinical characteristics, except for the ASA score, with more patients in preoperative SAE group seeming to have higher scores. Since preoperative SAE could not remain concealed, it was also challenging to implement blinding. Second, the size of the sample is small de facto. Only 18 patients received preoperative SAE, which might have caused relatively high heterogeneity, but we also have to point out that, to our knowledge, this study is the first to compare preoperative SAE to splenectomy only in SPH patients. Finally, we excluded patients with pancreatic or splenic malignancy because these diseases would require other surgical methods that might interfere with the analysis of outcomes such as intraoperative blood loss and operation time. Further studies with a larger sample size and stratified analysis might provide more supportive evidence.

Conclusions
This study demonstrated that preoperative SAE could efficiently reduce intraoperative blood loss and the operation time for patients with sinistral portal hypertension when undergoing splenectomy. Preoperative SAE did not increase the incidence of complications (POPF, bleeding, abdominal fluid collection, pleural effusion, and thromboembolism) or mortality among these patients. Since the LOS and surgery costs were comparable between the two groups, patients who received preoperative SAE had higher overall healthcare expenses.