Favorable Preoperative Exocrine Function Evaluated by 13C Trioctanoin Breath Test is a Signicant Physiological Predictor of Pancreatic Fistula After Pancreaticoduodenectomy

The association between the pancreatic stula (PF) after pancreaticoduodenectomy (PD) and preoperative exocrine function has yet to be elucidated. The aim of this study is to evaluate the association between the preoperative results of 13 C-trioctanoin breath test and occurrence of PF, showing the clinical relevancy of the breath test to predict the PF. In the present study, the subject were 80 patients who underwent 13 C trioctanoin breath test prior to PD from 2006 to 2018. We conducted the uni- and multivariate analyses to reveal the preoperative predictor of PF, showing the association of the 13 C trioctanoin absorption and incidence of PF.


Introduction
The postoperative mortality after pancreatoduodenectomy (PD) has been reducing especially in highvolume centers because of the advancement of surgical skill and perioperative administration. 30-day and in-hospital mortality rates were reported to be 1.2 and 2.8%, respectively, by national clinical database from Japan 1 . However, postoperative pancreatic stula (PF) is still a large threat to both patients and pancreatic surgeons, because it sometimes causes the postoperative fatal intraabdominal bleeding 2, 3 and abscess 4,5 , and its incidence is still reported to be high (11)(12)(13)(14)(15)(16)(17)(18)(19)(20)(21)(22)(23)(24)(25)(26)(27)(28)(29).4%) in patients with soft pancreas [6][7][8] . Therefore, it is urgent to elucidate the global preoperative risk factor of PF and there has been several articles showing the preoperative risk factors such as obesity, fatty pancreas, narrow pancreatic duct, male and surgical technics et al [9][10][11] . However, most of these predictors might affect the incidence of PF in an indirect fashion. In contrast, we conjectured that preoperative favorable exocrine function, which could be associated with normal pancreas, directly affect the incidence of PF, because excessive secretion of pancreatic juice after PD might cause the disruption of the anastomotic site. However, this assumption remains unclear because it has been still clinically challenging to address the preoperative exocrine function, which is mainly represented by fat absorption.
In terms of evaluating pancreatic exocrine functions, several articles have already reported the relevancy of various pancreatic function tests [12][13][14][15][16][17] . Indeed, BT-PABA (N-benzoyl-L-trypsyl-p-aminobenzoic acid), fecal chymotrypsin, fecal elastase-1testing (FE-1), fecal fat excretion test, and 13 C-trioctanoin breath test has been clinically employed for the evaluation of it. Among these usable testing, 13 C-trioctanoin breath test does not have a necessity of urine or stool collection, and also is not affected by hepato-renal function of subjected patients; thus, we consider possibly more acceptable for evaluating a perioperative pancreatic exocrine function. The aim of this study is to evaluate the association between the results of 13 C-trioctanoin breath test and occurrence of PF, showing the clinical relevancy of the breath test to predict the PF.

Patients And Methods
Among the 133 patients who underwent 13 C-labeled trioctanoin breath test before and after pancreatectomies in our institution from 2006 to 2018, the subjects were 80 patients who underwent PD and perioperative course and factors associating PF could be precisely evaluated. 13 C-labeled trioctanoin breath test, which directly and objectively re ects the ability of fat absorption, was performed one to three before the surgery.
All of the patients were fasted overnight prior to the breath test. Breath samples were collected in 100ml bags with a one-way check valve. Samples were obtained 15 minutes prior to the test, and 0, 5, 10, 15, 20, In the present study, we retrospectively compared pre-and postoperative fat absorption levels and their change rate during pancreatectomies between the groups with PF and non-PF. The medical ethics committee approved the study protocol of Fujita Health University School of Medicine (HM17165). In terms of surgical procedure of PD, we employed the inferior pancreaticoduodenal artery (IPDA) rst approach to reduce the intraoperative blood loss 18 . A drain was removed until postoperative day (POD) 5 to 7 as long as drain discharge was clear and drain amylase level was not as three times high as the upper limit of serum amylase level (132U/ml). In all patients, amylase level of abdominal drainage uid were measured until day 7 after PD. Pancreatic stula was de ned and graded according to the International Study Group on Pancreatic Fistula classi cation 19 . In the present study, we divided the subject into the patients with clinically relevant PF of Grade B or C and those with non-PF or biochemical leak. To identify pre-and intra-operative risk factors of pancreatic stula, we compared various factors between these two groups.
In terms of surgical procedure of PD, we employed the IPDA (inferior pancreatoduodenal artery)-rst approach from 2007 18 . Brie y, IPDA is encircled and ligated before pancreatic resection, aiming the reduction of intraoperative blood loss (Fig. 1). For pancreatojejunostomy, the rst-layer anastomosis was done by duct-to-mucosa anastomosis with 6-8 interrupted sutures by 5-0 PDS II (Ethicon, Inc.Somerville, NJ, USA). The second-layer anastomosis was done by the 6 to modi ed Kakita procedure using 3 − 0 proline 20 . A 5F external pancreatic stent tube were inserted in the remnant main pancreatic duct in all 80 patients.
All statistical analyses were done by the statistical software package SPSS for Macintosh (version 24.0, IBM, Armonk, NY, USA). The results of the continuous variables were expressed as median and range, and statistical signi cance was evaluated by the Mann-Whitney U test. Discrete variables were evaluated by χ2 analysis or Fisher's exact test, as appropriate. Pre-and intraoperative risk factors associated with POPF were analyzed using univariate and multivariate analysis (logistic regression analysis). Only variables with p-values less than 0.05, as determined by univariate analysis, were included in the multivariate analysis. Results were considered signi cant when P values were less than 0.05. Receiver operating characteristic (ROC) curve analysis were employed to estimate the best cut-off points for the 13 C-trioctanoin breath test to predict PF. In the present study, the pancreatic con guration based on the imaging study, pancreatic texture and diameter of pancreatic duct intraoperatively evaluated were excluded by the items of univariate analysis because these factor might be strong confounding factors and the aim of this study is to nd the objective, quantitative and physiological predictor of PF.

Results
Preoperative background of the 80 patients is shown in In terms of preoperative lab data, the detail of blood cell counts, and several nutritional markers were described also in Table.

Pre-and intraoperative risk factors of POPF
As shown in Table 2, univariate analysis by comparing preoperative risk factors between the PF group and non-PF group identi ed the pancreatic ductal adenocarcinoma (PDAC) (p=0.009) and favorable results of 13 C-trioctanoin breath test (p=0.005) as the signi cant risk factor of PF. Indeed, when we compared the levels of 13 C trioctanoin absorption between PF and non-PF group, preoperative fat absorption level is signi cantly higher than in the PF group than in the non-PF group (40.2 vs. 34.4, p=0.05). Moreover, optimal cut-off value of the preoperative fat absorption level to predict PF was 38.0 (sensitivity:90%, speci city:74%, AUC:0.78, p=0.005). Indeed, the incidence of PF was extremely higher in the patients whose value of breath test was greater than 38.0 (33%, 9/27) compared to the patients with those less than 38.0 (1.8%, 1/53). By multivariate analysis as shown in Table.3, preoperative 13 C trioctanoin breath test >38.0% dose/h were selected as the most independent risk factors for PF (p=0.001, Odd's ratio:16.7).
Next, we focused on the association between the incidence of PF and the level of 13 C trioctanoin breath test >38.0% dose/h in only non-PDAC patients because we considered that the prediction of PF in non-PDAC cases, that is mostly soft pancreas, is urgent issue to be solved.
As the same manner of gure 2A, ROC curve revealed that the cut-off value was 37.9% dose/h (Area under the curve:AUC:0.78) ( Figure 3A). As shown in gure3B, the incidence of PF was markedly high (39.0%, 9/23) in the patients with preoperative favorable fat absorption, whereas it was 3.7% (1/27) in the patients with unfavorable absorption (Figure 3b).

Discussion
In the present study, we newly elucidated the insight in which 13 C trioctanoin absorption (> 38.0% dose/h) were strong preoperative physiological predictor of PF after PD in not only total cohort, but also in non-PDAC patients whose pancreatic parenchyma could be soft. 13 C trioctanoin breath tests have been employed on the clinical settings to evaluate the pancreatic exocrine de ciency by detecting fat malabsorption through the gut after pancreatectomies 21,22 . Until now, however, there has been few studies evaluating whether preoperative exocrine function test affect the incidence of PF after PD 12,23 .
To predict the development of PF preoperatively, a lot of researchers seek to nd the relevant risk factors such as high BMI 24,25 , fatty pancreas 10, 24, 26 , male 27, 28 and untreated jaundice 29 , and also developed the way of its prediction using various imaging modalities such as CT con gurations (narrow main pancreatic duct 10 , thick pancreatic parenchyma 10 , pancreatic border 6 , CT attenuation value 30 ), MRI ndings 31 , and pancreatic ultrasound elastography 32, 33 et al. Even though these predictors might be clinically useful, most of these factors are strongly associated with the soft parenchymal condition, which cause a technical di culty of anastomosis. Therefore, whether these risk factors are reproducible or not is also depending on the type of pancreato-enteral anastomosis and maturity of those procedures.
On the other hands, the data obtained from 13 C trioctanoin breath test is more objective and quantitative, and especially in patients with Aa > 38.0 %dose/h, the PF incidence is extremely high regardless of parenchymal condition. Previous article mentioned the recovery of the 13 C-labeled trioctanoin absorption after PD positively associated with output of pancreatic enzymes such as lipase, amylase, and chymotrypsin 15 . Thus, we speculated the active production of pancreatic juice might be one of the major causes of PF after PD, and considered that outcome of the present study represented this aspect. In fact, our speculation is supported by the result showing that postoperative maximum drain amylase level (U/L) is signi cantly higher in patients with Aa > 38.0% dose/h than that in those with Aa < 38.0 (Fig. 4).
According to the previous reports regarding the association between PF and results of pancreatic exocrine function test, several reports revealing that higher preoperative level of FE-1 was positively associated with the developments of postoperative PF 12,30 . However, there has been no study revealing the association between the result of 13 C-labeled trioctanoin and occurrence of PF. Therefore, to the best of our knowledge, Aa > 38.0% dose/h is considered to be rst physiological quantitative predictor of PF.
Clinical application of this study is challenging, because prevention of PF is quite di cult even if risky patients were identi ed preoperatively. the administration of octreotide or somatostatin analogs is wellaccepted pharmacological treatment with PF targeting the secretion of pancreatic juice 34,35 . The effect of somatostatin analogs is to reduce the volume of stula output, thereby potentially alleviating the PF 36 . Octreotide also has been considered to reduce the volume and potency of both pancreatic exocrine secretions and hormone production 37 . Since our study demonstrate that the favorable preoperative exocrine function, which in turn high output of pancreatic juice, is regarded as the risk factor of PF, administration of these drugs might become a key treatment of PF in these risky patients. However, prospective or randomized control study should be needed to show this hypothesis.
The present research has several limitations. The rst is that this study included only small number of patients. The second limitation is that the present analysis was a retrospective analysis, and we could not identify the precise mechanism how favorable pancreatic exocrine function cause the PF postoperatively. Therefore, the study is regarded as only an exploratory research. Nonetheless, our study could draw signi cant attention to the association between PF and preoperative pancreatic exocrine function. In conclusion, favorable pancreatic exocrine function evaluated by 13 C trioctanoin breath test preoperatively, is a feasible and objective predictor of PF after PD, paying attention to the development of PF in such high-risk patients.