Pancreas-visceral fat CT value ratio and serrated pancreatic contour are strong predictors of postoperative pancreatic fistula after pancreaticojejunostomy

The probability of postoperative in-hospital mortality after pancreatoduodenectomy (PD) has decreased, especially in high-volume centres, with a mortality rate of less than 4% over recent decades [1, 2]. A recent study using a national clinical database from Japan revealed that the 30-day and in-hospital mortality rates were 1.2 and 2.8%, respectively [3]. Despite the fact that a low mortality rate has been observed, the incidence of clinically relevant postoperative pancreatic fistula (POPF: grade B/C), which most negatively affects patient outcome, has been recently reported to be 11–37% in patients with soft pancreas and 1–6% in patients with hard pancreas [4‐9]. Regarding the risk factors for POPF, previous studies have reported various risk factors, such as age, sex, preoperative jaundice, operative time, intraoperative blood loss, type of pancreatic reconstruction, anastomotic technique, consistency of the pancreatic stump and pancreatic duct diameter [10‐14], but there have been no reports focusing on preoperative computed tomography (CT) configurations, especially the contour of the pancreas, for predicting POPF preoperatively.

The procedures of pancreatoenteral anastomosis have not been standardized, and each institution performs their own preferred procedure, such as pancreaticogastrostomy, pancreaticojejunostomy, external tube drainage, the lost stent method and invagination; this diversity of procedures makes it difficult to evaluate the frequency of POPF [15‐17]. Our institution reported the method of 12 interrupted-stitched duct-to-mucosa pancreaticojejunostomy, named the “pair-watch suturing technique (PWST)”, which allowed us to standardize the method of pancreaticojejunostomy [18‐20]. However, even though the anastomotic technique has progressed, POPF still has yet to be thoroughly prevented after PD, and the incidence of POPF in patients with soft pancreas has been reported to be particularly high; thus, the prevention of POPF in patients with soft pancreas is still under discussion [21‐23].

Recently, Sugimoto M et al. [21] reported that a thick parenchyma, a small main pancreatic duct (MPD), and fatty infiltration determined by postoperative histology were strongly associated with clinically relevant POPF after PD, especially in patients with soft pancreas, which was judged by intraoperative findings, and the study showed the negative impact of fat infiltration into the pancreatic parenchyma. Because pancreatic texture and consistency can be determined only by intraoperative findings or postoperative histological examinations, a high-risk group of POPF patients, especially those with soft pancreas, cannot be identified preoperatively. To solve this problem, Kuwahara T et al. [24] showed the usefulness of preoperative endoscopic ultrasonography-elastography (EUS-EG), which made it possible to objectively assess tissue elasticity preoperatively and predict the development of POPF after PD. However, EUS-EG is still an uncommon procedure for the preoperative assessment of pancreatic consistency, and we consider it indispensable to select true soft pancreas and POPF high-risk patient groups preoperatively based on pancreatic configurations, such as the MPD diameter and parenchymal thickness, and on CT attenuation values of the pancreas, such as visceral fat and other ratios. All these parameters are easily measurable by plain CT images.

In terms of the contour of the pancreas in preoperative CT, the precise cause of significant changes in the irregularity of the borders of the pancreas is unknown. When we analysed CT images to evaluate the precise morphology of the pancreas in order to investigate the type of pancreas that is likely to develop POPF, we noticed that 20 to 30% of patients with a mostly normal pancreas had irregular pancreatic borders, so we called this type of pancreas a serrated pancreas. According to previous research, a serrated pancreatic border is reported to be the result of ageing or acute weight loss after reversal of type 2 diabetes mellitus (DM) treated by a low-calorie diet [25]. This report also showed that the pancreas of patients with type 2 DM obviously had more pancreatic marginal irregularities compared with the pancreas of healthy patients. In the field of diabetic internal medicine, pancreatic contour has been sometimes discussed, but there have been no studies evaluating the relationship between pancreatic outer contour and POPF after PD. Moreover, we hypothesized that a serrated pancreatic contour was associated with intralobular frailty, which results in parenchymal vulnerability during pancreaticojejunostomy. This vulnerability might result in difficulties associated with the anastomosis, with determining the risk of POPF regardless of the MPD size and pancreatic thickness, and with surgeons anastomosis skills.

In the present study, we evaluated 262 patients who underwent PD with PWST and analysed the pre- and intraoperative risk factors for POPF in these patients, focusing on the association between the incidence of POPF and the preoperative CT configuration of the pancreas as well as the CT attenuation values of the pancreatic parenchyma, visceral fat and these ratios. Our aim was to elucidate the true preoperative risk factors for POPF even in patients with soft pancreas, making it possible to select POPF high-risk patients preoperatively regardless of intraoperative pancreatic consistency judged by the surgeon’s hand.

The characteristics of the subjects are shown in Table 1. In these 262 patients, the median age (range) was 67.6 years old, and the male/female ratio was 158/104. The primary diseases were pancreatic adenocarcinoma (PDAC) in 118 patients, intraductal papillary mucinous neoplasm (IPMN) in 52 patients, bile duct carcinoma in 53 patients and others in 39 patients. The results of the CT analysis were as follows: MPD diameter (mm) = 3.0, pancreatic parenchymal thickness (mm) = 13.5, pancreatic parenchymal CT value (Hounsfield Unit: HU) = 38.2, visceral fat CT value (HU) = − 98.2, and pancreas-visceral fat CT value ratio = − 0.39, and the pancreatic texture judged intraoperatively was soft in 123 (46.9%) and hard in 139 patients (53.1%).

In terms of the type of pancreatic outer contour, the smooth type was observed in 228 (87%) and the serrated type in 34 patients (13%).

A pancreatic stent was placed in 140 (53.4%) patients. The operation time (min) and intraoperative blood loss (ml) were 526 and 863, respectively. In terms of the incidence of POPF, among the 262 patients, clinically relevant POPF, that is, POPF grade B/C, was found in 27 patients (10.3%): grade B in 23 (8.8%) and grade C in 4 (1.5%). For reference, BL was found in 20 patients (7.6%). In terms of the treatment of POPF grade B, CT-guided drainage was performed in 10 patients, re-initiation of antibiotics in 4, wound drainage in 3, drain exchange in 3, angiography for bleeding in 2 and persistent drainage for more than 3 weeks in one. In the 4 patients with POPF grade C, open laparotomy was performed in 3 patients, and mechanical ventilation for the treatment of acute lung injury was performed in one patient.

In the present study, using a homogeneous cohort of patients who underwent pancreaticojejunostomy with PWST, we newly revealed that the pancreas-visceral fat CT value ratio, serrated-type pancreatic contour, and a history of chemoradiotherapy were strong preoperative predictors of POPF after PD. Furthermore, soft pancreatic texture was selected as the only intraoperative risk factor for POPF, and this result has been widely accepted among pancreatic surgeons until now. Among these risk factors, we considered that the pancreas-visceral fat CT value ratio and serrated-type pancreatic contour were closely associated with the degradation of pancreatic parenchymal quality characterized by parenchymal fat deposition, which was evidenced by histological evaluation of the parenchymal stump.

Since the effective management of POPF has proven to be a difficult challenge despite recent improvements in postoperative patient care, the early identification of POPF high-risk groups has made a paradigm shift among pancreatic surgeons from a reactive and passive approach that begins to treat POPF when it becomes apparent to a proactive approach that depends on early anticipation and timely prevention through prophylactic measures. However, this approach is predicated on the assumption that POPF high-risk groups can actually be predicted. To predict POPF more precisely, a clinical scoring system by Callery MP et al. [28] was considered to be very useful because the incidence of clinically relevant POPF reached more than 67% in patients with scores greater than seven, which consisted of gland texture, pathology (pancreatic adenocarcinoma or pancreatitis or others), pancreatic duct diameter and the amount of intraoperative blood loss. However, this score cannot be assessed preoperatively because the scoring systems include several intra- and postoperative variables, such as gland texture, pathological diagnosis and blood loss. In particular, as pancreatic gland texture is judged by the surgeon’s hand intraoperatively, this approach is a very subjective method; therefore, preoperative indicators for predicting POPF should be identified as an alternative to conducting the proactive approach. When we treat high-risk groups of patients, we should pay much more attention to the occurrence of POPF from the preoperative setting so that preoperative, careful informed consent can be provided to patients and their families. Postoperatively, we should carefully check the amylase levels of the drainage discharge and the appearance of fluid. If inflammatory reactions are unexpectedly escalated or continue in high-risk groups, an early follow-up CT scan should be conducted to detect the presence or absence of pancreatic fistula.

Among the various preoperative risk factors, the pancreas-visceral fat CT value ratio was selected as the most independent factor for predicting POPF rather than the CT value of the pancreatic parenchyma itself. We considered that the pancreas-visceral fat CT value ratio represents the quality of the pancreatic parenchyma. Kitajima Y et al. [29] measured intramuscular adipose tissue content (IMAC) using CT, and this measure has recently attracted much attention for evaluating the quality of skeletal muscle because several studies have revealed that increased IMAC is positively linked to worse survival after resection of PDAC [30] and to an increased complication rate after hepatectomy for hepatocellular carcinoma [31]. Nevertheless, there have been no reports regarding pancreatic parenchymal quality. In this study, we hypothesized that a lower quality of the pancreatic parenchyma might result in vulnerability of pancreaticojejunal anastomosis due to severe fat infiltration. Since IMAC is calculated by the ratio of the multifidus muscle to the subcutaneous fat CT attenuation value, we analysed whether the pancreas-visceral fat CT value ratio represented the quality of the pancreas by comparing the incidence of POPF, and this is the first report regarding the estimation of pancreatic parenchymal quality using plain CT images.

A history of chemoradiotherapy reduced the incidence of POPF in the present study. In general, preoperative chemoradiotherapy is considered to reduce POPF after pancreatectomy because radiation induces intralobular fibrosis and exacerbates its exocrine function [32, 33]. Moreover, in our institution, most candidates who undergo preoperative chemoradiotherapy are patients with advanced PDAC, in whom exocrine function is generally ruined due to MPD obstruction. As a result, preoperative chemotherapy significantly reduced POPF after PD in our cohort.

Serrated-type pancreatic contour determined by preoperative plain CT was also selected as another independent risk factor for POPF. In fact, the incidence of POPF was significantly higher in patients with a serrated-type pancreas (29.4%, 10/34) than in patients with a smooth pancreas (7.4%, 17/228). Indeed, serrated-type contour was mostly seen in patients with soft pancreas, whereas this type of contour was barely found in patients with a hard pancreatic texture. The precise aetiology of significant changes in the irregularity of the borders of the pancreas is unknown, but serrated-type pancreatic contour has been reported to be the result of ageing or acute weight loss after the reversal of type 2 DM treated by a low-calorie diet [25]. When we focused on basic research regarding pancreatic exocrine architectures, ghrelin, a hunger-stimulating hormone produced by the fundus of the stomach, increased exocrine pancreatic fractal dimensions and textural entropy and decreased the lacunarity of the acinar cell architecture in rats, regardless of age [34]. In the clinical setting, Sasaki K et al. [35] reported that the individual ghrelin ratio (POD1/prior to operation) was significantly lower in patients who developed complications, especially POPF and intraabdominal abscess, than in those who did not. The lack of ghrelin exertion and weight loss in patients with type 2 DM might be related to the formation of a serrated-type pancreas and its parenchymal frailty, which significantly accelerate the incidence of POPF. In our present study, however, the incidence of a serrated-type pancreas was comparable between diabetic and non-diabetic patients, and the level of ghrelin in the blood was not measured. Therefore, further research is needed to elucidate the cause of serrated-type pancreatic contour.

Soft pancreas is generally characterized by a narrow pancreatic duct and vulnerable parenchymal texture, resulting in a high risk of POPF after PD. However, as shown in our 2 × 2 contingency table analysis (Table 4), the incidence of POPF grade B/C was 0% (0/57) in patients with a pancreas-visceral fat CT value ratio of less than − 0.4 and smooth-type contour, even though these patients were categorized as patients with soft pancreas intraoperatively. The high-risk group showed an obviously high incidence of POPF (9/20). Moreover, these pancreases were characterized by obvious fat infiltration pathologically, as shown in Fig. 5. Indeed, fatty pancreas is considered to be a high-risk factor for POPF, and Mathur A et al. [36] demonstrated that patients with increased fat and decreased fibrosis had a higher risk of POPF after PD. Gaujoux S et al. [37] also showed that an increased body mass index, fatty pancreas, and the absence of fibrosis were associated with a risk of POPF after PD. In these reports, the amount of fat deposition and degree of fibrosis were examined by histological findings, and therefore, these factors could not be assessed preoperatively. To assess the amount of pancreatic fat and its influence on POPF preoperatively, Lee SE et al. [38] suggested the usefulness of relative signal intensity decreases in preoperative dual-gradient-echo magnetic resonance imaging (MRI) since preoperative measurements of pancreatic fat by MRI offer noninvasive prediction of the occurrence of POPF. In an attempt to predict pancreatic texture preoperatively, Kuwahara T et al. [24] analysed the usefulness of EUS-EG in 59 patients, 26 with soft pancreas and 33 with hard pancreas, and revealed that the mean elasticity of the pancreas measured by EUS-EG (> 70) was an independent predictor of POPF. Although MRI and EUS-EG are considered useful tools, the usage of these modalities is still limited for the preoperative assessment of PD; therefore, our predictors measurable by plain CT are considered feasible and reasonable for predicting POPF.

Taken together, the high pancreas-visceral fat CT value ratio, serrated-type pancreas and these combinations could be associated with the infiltration of adipose tissue into the pancreas, resulting in parenchymal frailty. The frailty of the parenchyma makes it difficult to perform pancreaticojejunostomy and induces the vulnerability of anastomosis regardless of the surgeon’s anastomosis technique. Therefore, we established a new strategic schema (Fig. 6) for the precise prediction of POPF preoperatively, which might contribute to the proactive approach for POPF.

The pancreas-visceral fat CT value ratio > = − 0.40 and serrated-type pancreas allowed us to preoperatively identify a true POPF high-risk group regardless of pancreatic texture. Preoperative identification of a POPF high-risk group enabled us to develop a proactive strategy, such as the administration of somatostatin analogues and early follow-up CT scans for preventing POPF or its aggravation and for early detection. To establish the new strategy for preventing POPF for high-risk patients is imperative to improve a surgical outcome of PD.