Significant positive impact of duodenum‐preserving pancreatic head resection on the prevention of postoperative nonalcoholic fatty liver disease and acute cholangitis

Abstract Aim This study aimed to compare the incidence of postoperative nonalcoholic fatty liver disease (NAFLD), postoperative cholangitis, and fibrosis‐4 (FIB)‐4 index in patients who underwent duodenum‐preserving pancreatic head resection (DPPHR) and pancreaticoduodenectomy (PD) for low‐grade malignant tumors and verify the usefulness of DPPHR in preventing the occurrence of these disorders. Methods This retrospective study included 70 patients who underwent PD (n = 39) and DPPHR (n = 31) between 2006 and 2018 for benign or low‐grade malignant tumors. The present study compared the preoperative background, cumulative incidence of postoperative NAFLD and cholangitis, and other biochemical markers, including the FIB‐4 index. Subanalysis by propensity score matching (PSM) analysis was conducted to minimize treatment selection bias. Results In terms of the cumulative incidence of NAFLD, the 5‐y incidence was significantly lower in the DPPHR group than in the PD group both before (10% vs 38%, P = .002) and after (13% vs 38%, P = .008) matching. Multivariate analyses identified DPPHR as the only independent preventive factor for postoperative NAFLD (hazard ratio: 0.160, 95% confidence intervals: 0.034–0.76, P = .021). The 5‐y cumulative incidence of postoperative cholangitis was significantly higher in the PD group than in the DPPHR group before (51% vs 3%, P < .001) and after (49% vs 4%, P < .001) matching. The FIB‐4 index at 12 mo postoperatively was significantly better in the DPPHR group than in the PD group (1.45 vs 2.35, P = .006) before matching. Conclusion Preservation of the duodenum and bile duct may contribute to preventing long‐term postoperative NAFLD and cholangitis, and liver fibrosis for benign or low‐grade malignant pancreatic head tumors.


| INTRODUC TI ON
With the development of surgical techniques in recent y, short-term complications and mortality after pancreaticoduodenectomy (PD) have decreased. 1,2 Conversely, long-term complications, such as postoperative malnutrition, resulting in fatty liver disease and postoperative cholangitis, greatly reduce the quality of life (QOL) of patients with low-grade malignant tumors whose prognosis is favorable and whose long-term survival is anticipated. Recently, laparoscopic surgical techniques have improved, and laparoscopic PD and robotic PD have been widely performed for low-grade malignant pancreatic tumors. [3][4][5] Even if these advancements allow PD to be performed with smaller wounds and less invasiveness, the incidence of complications, such as fatty liver disease and postoperative cholangitis, is not theoretically reduced. In addition, low-grade malignant tumors, such as solid pseudopapillary neoplasm (SPN) and pancreatic neuroendocrine neoplasm (NEN), are expected to have long-term survival in most cases and are often detected at a young age, making it a dilemma for surgeons as to whether PD is indeed necessary and can be indicated. This is because the pancreatic head is anatomically adjacent to the bile duct and duodenum, and therefore resected to conveniently remove the tumor.
Postoperative nonalcoholic fatty liver disease (NAFLD) develops in 30%-40% of patients after PD. 6 Although it is usually a benign disease and has been mostly ignored, recently there have been reports of cases of long-term progression to nonalcoholic steatohepatitis (NASH), leading to cirrhosis and death, 7,8 which is a problem that must be solved. Pancreatic exocrine function replacement is believed to be an effective treatment for NAFLD/NASH, but its effect remains controversial. 9 Recently, the usefulness of the fibrosis (FIB)-4 index as an indicator of liver fibrosis in patients with NAFLD has been widely reported, 10,11 and its value is strongly correlated with pathological liver fibrosis. Therefore, the clinical relevance of the FIB-4 index as an indicator of liver fibrosis in patients with NAFLD after pancreatectomy remains unclear.
The development of postoperative acute cholangitis after PD may be caused by hepaticojejunostomy, and one possible cause is the reflex of intestinal fluid and intestinal gas into the intrahepatic bile duct, resulting in increased intraductal pressure and bacteremia. 12 Anastomotic stenosis is also a major cause of cholangitis, and patients with a preoperative narrow bile duct diameter in benign disease are at high risk of recurrent cholangitis. 13 Therefore, patients with almost no jaundice, no bile duct dilatation, and lowgrade tumors, such as primitive neuroectodermal tumor, solidpseudopapillary neoplasm, or intraductal papillary mucinous neoplasm, are likely to have a high risk of postoperative cholangitis, and their postoperative QOL might be greatly impaired. Moreover, postoperative cholangitis is recognized as the second-most common cause of NAFLD in PD, which might trigger the transition to NASH from NAFLD. 8 Duodenum-preserving pancreatic head resection (DPPHR) was first described by Beger et al 14 as a procedure for severe chronic pancreatitis, and its indications were expanded to include pancreatic head tumors. However, DPPHR requires careful preservation of the pancreatic head arterial arcades and is not widely used compared with PD because of the difficulty of its indication, especially when the tumor is close to the intrapancreatic bile duct or arterial arcades. Therefore, there are very few studies on the postoperative nutritional status and complications after DPPHR and no reports comparing the frequency of postoperative NAFLD and postoperative cholangitis with those of PD. We previously reported that exocrine pancreatic function, that is, fat absorption function, was significantly preserved in patients who underwent DPPHR compared with those who underwent PD. 15 Therefore, we hypothesized that patients who underwent DPPHR would have significantly less postoperative hepatic damage caused by NAFLD or cholangitis, even though the pancreatic head was resected as much as in PD.
This study aimed to compare the incidence of postoperative NAFLD, postoperative cholangitis, and the FIB-4 index in patients treated with DPPHR and PD for low-grade malignancy and prove the usefulness of DPPHR in preventing the development of these disorders. To the best of our knowledge, this is the first study to prove the importance of preservation of the bile duct and duodenum to prevent postoperative NAFLD, acute cholangitis, and liver fibrosis. As for the surgical indication for DPPHR, we can adapt DPPHR as long as the arterial arcade and bile duct are considered preservable. However, DPPHR often cannot be indicated if the tumor is either adjacent to the bile duct or is too large to preserve the arterial arcade in the pancreatic head, regardless of the degree of its malignancy. If a lymph node dissection is considered necessary due to swelling of the peripancreatic lymph node, DPPHR is not indicated, and PD remains the only option.

| ME THODS
Regarding the surgical procedure of DPPHR, a precise surgical technique was described in a previous study. 15 After laparotomy, both the gastric and duodenocolic ligaments were dissected without Kocher's maneuver to preserve the venous drainage from the duodenum. Thereafter, the common hepatic artery and gastroduodenal artery were exposed. The pancreas was divided above the portal vein, and a polyvinyl tube was inserted into the main pancre- All statistical analyses were conducted using SPSS for Macintosh (v. 24.0; IBM, Armonk, NY, USA). Continuous variables were expressed as medians and ranges. The statistical significance of the continuous variables was determined using the Mann-Whitney U test. Categorical variables were compared using Pearson's chisquared test. The risk factors associated with NAFLD were analyzed using univariate and multivariate (logistic regression) analyses. Only variables with a P-value <.1 on univariate analysis were included in the multivariate analysis. The overall cumulative incidence of postoperative NAFLD and acute cholangitis were calculated using the Kaplan-Meier method, and the Kaplan-Meier curves were compared using the log-rank test. To minimize treatment selection bias, propensity score matching subanalysis was used. Propensity scores were developed by logistic regression analysis using the preoperative patient age. This was used because it was the only preoperative factor in which univariate analysis showed a statistically significant difference between the PD and DPPHR groups. The PD and DPPHR patients were then paired 1:1 on these propensity scores using exact matching. A standard caliper size of 0.2 × standard deviation of the propensity score was used. Table 1 shows a comparison of perioperative characteristics between the DPPHR and PD groups. The patients who underwent DPPHR were significantly younger than those who underwent PD (57.5 vs 68.2; P = .003). Otherwise, there were no significant differences in terms of sex, primary disease, preoperative FIB-4 index, operative time, and blood loss. In Table 2 we compare the backgrounds of the two groups after adjusting for age using propensity score matching. All perioperative backgrounds, including age, were comparable after matching. As shown in Table 3, the CT attenuation value at 12 mo postoperatively was significantly higher in the DPPHR group than that in the PD group (60.0 vs 53.1; P = .005), and the FIB-4 index at 12 mo postoperatively was significantly better in the DPPHR group than that in the PD group (1.45 vs 2.35; P = .006) before matching. Concurrently, the serum albumin level was also significantly better in the DPPHR group than in the PD group (4.2 vs 3.9; P = .003), and the prognostic nutrition index tended to be better in the DPPHR group, although the difference was not significant.

| RE SULT
However, these statistically significant differences disappeared in the albumin value and the FIB-4 index after matching, and only the CT attenuation value remained as a significant factor after matching.
When the cumulative incidence of NAFLD was compared between the two groups, the 5-y cumulative incidence was significantly lower in the DPPHR group than in the PD group (10% vs 38%; P = .002) ( Figure 2A). Even after propensity score matching, the cumulative incidence was significantly lower in the DPPHR group than in the PD group (13% vs 38%; P = .008) ( Figure 2B). Moreover, based on the results of univariate and multivariate analyses to identify the perioperative preventive factor of NAFLD in the cohort after matching, DPPHR was identified as the only independent preventive factor for postoperative NAFLD (hazard ratio: 0.160; P = .021) ( Table 4).
The 5-y cumulative incidence of postoperative cholangitis was significantly higher in the PD group than in the DPPHR group before (51% vs 3%; P < .001) and after (49% vs 4%; P < .001) propensity score matching ( In terms of local recurrence of tumors, for reference there were no obvious recurrent cases in 70 cases; however, three patients (one in the DPPHR group and two in the PD group) whose primary tumor was an intraductal papillary mucinous adenoma developed remnant pancreatic cancer.

| DISCUSS ION
In this study the following new insights were obtained: (a) preservation of the duodenum and bile duct by DPPHR significantly NAFLD that develops after PD differs from ordinary NAFLD in terms of cause, and its major characteristic is that it develops without insulin resistance due to worsening nutritional status postoperatively. NAFLD after PD is thought to be caused by malnutrition induced by impaired fat absorption due to pancreatic exocrine insufficiency or deficiency of duodenal hormones or an eating disorder due to complicated gastrointestinal reconstruction, leading to increased conversion of carbohydrates to fats in the liver, similar to that in fatty liver disease caused by starvation, such as kwashiorkor. 6 Regarding the treatment for NAFLD after PD, pancreatic exocrine enzyme replacement is considered the most promising treatment, with a recent randomized control study showing that it also significantly prevents NAFLD after PD. 25 Therefore, pancreatic exocrine enzyme replacement therapy is routinely administered after a pancreatectomy at our institution.
Given that the cause of NAFLD is malnutrition, the lower incidence of NAFLD after DPPHR could be explained by the preservation of the duodenal and bile duct in the DPPHR group, thus preserving postoperative nutrition by maintaining normal gastroduodenal function and physiological bile secretion. Lue et al 26 revealed that the incidence of NAFLD is significantly lower in DP than in PD, even though the amount of resection of the pancreatic parenchyma is greater in DP than in PD. The absence of gastrointestinal reconstruction and preservation of the duodenum in DP may be essential for the low incidence of NAFLD. Beger et al 27 recently compared the incidence of postoperative pancreatic exocrine insufficiency and new-onset diabetes mellitus in the DPPHR and PD groups using a systematic review approach and reported significantly lower rates in the DPPHR group, which supports our hypothesis. Furthermore, they reported that the postoperative secretion of cholecystokinin, which is secreted from the duodenum and stimulates the secretion of pancreatic and bile juices, was significantly higher in the DPPHR group than that in the PD group.
Although these articles may indirectly support our hypothesis, we considered the comparison of the incidence of NAFLD between DPPHR and PD to be crucial to clinically prove this hypothesis.
To the best of our knowledge, this is the first study to show that preservation of the duodenum and bile duct plays a significant protective role in the development of NAFLD. which is consistent with the results obtained in our study.
The role of postoperative cholangitis in the development of fatty liver disease and transition from NAFLD to NASH remains controversial. However, it has been assumed that the influx of enteric bacteria into the intrahepatic bile ducts activates Kupffer cells in the liver and promotes the uptake of fat droplets into hepatocytes. 31,32 Bacterial translocation due to cholangitis or bacterial enteritis is also considered one of the factors of the second hit that leads to NASH. 33 Whether postoperative cholangitis is associated with the development of postpancreatectomy NAFLD requires further investigation.
The FIB-4 index is a scoring system that combines blood test data to evaluate the degree of liver fibrosis. 34,35 In early detection and follow-up, it is important to evaluate the degree of progression of liver fibrosis, which is considered to have the strongest correlation to life prognosis, rather than the degree of fatty liver disease.
Liver biopsy is essential for the definitive diagnosis of NASH, which has the risk of progression to cirrhosis or liver cancer; however, it is not practical to perform liver biopsy in all patients with NAFLD.
In addition, liver biopsy is associated with hospitalization and sampling error; therefore, the FIB-4 index is a very useful tool to more easily assess the degree of liver fibrosis. In fact, the results of this study showed that the FIB4-index at 12 mo postoperatively was significantly lower in the DPPHR group, indicating that preservation of the duodenum and bile ducts contributes to the suppression of liver fibrosis. However, this significant difference disappeared in the cohort after propensity score matching. The reasons for this may be associated with the fact that age was included in the formula for calculating the FIB-4 index. As the age between the two groups was TA B L E 4 Uni-and multivariate analyses for identifying the influential factor of postoperative NAFLD after propensity score matching adjusted to be equal and the number of cases decreased from 70 to 50 cases by matching, the statistical significance was abolished.
Therefore, further case accumulation is needed to verify whether DPPHR reduces the degree of liver fibrosis when compared with PD.
This study had several limitations. First, this was a single-center retrospective study; therefore, it might be considered only as an exploratory investigation. Therefore, a large-scale multicenter study is desirable to demonstrate the reproducibility of the results of this study. Second, DPPHR is difficult to perform in patients with tumors close to the bile duct or in cases where the pancreatic arterial arcades cannot be preserved, and it is not indicated for all lesions located in the pancreatic head. Depending on the size and location of the tumor, PD may have to be performed in several cases.
Nonetheless, based on the results of this study we believe that patients will obtain a large benefit in the long term if DPPHR is performed rather than PD.
In conclusion, preservation of the duodenum and bile duct may contribute to preventing the development of NAFLD for up to 1 y postoperatively and long-term postoperative cholangitis for benign or low-grade pancreatic head tumors. As there were no cases of recurrence caused by the procedures, DPPHR should be considered for these tumors whenever possible.

D I SCLOS U R E
Funding: The present study was not funded by any organization.