Early Post-Operative Pancreatitis and Systemic Inflammatory Response Assessed by Serum Lipase and IL-6 Predict Pancreatic Fistula

Background Post-operative pancreatic fistula (POPF) remains a critical complication after pancreatic resection. This prospective pilot study evaluates perioperative markers of pancreatitis and systemic inflammation to predict clinically relevant grade B/C-POPF (CR-POPF). Methods All patients undergoing pancreatic resection from December 2017 to April 2019 were prospectively enrolled. Surgical procedures and outcomes were correlated with perioperative blood markers. ROC analysis was performed to assess their predictive value for CR-POPF. Cut-offs were calculated with the Youden index. Results In total, 70 patients were analysed (43 pancreatoduodenectomies and 27 distal pancreatectomies). In-hospital/90-d mortality and morbidity were 5.7/7.1% (n = 4/n = 5) and 75.7% (n = 53). Major complications (Clavien–Dindo ≥ 3a) occurred in 28 (40.0%) patients, CR-POPF in 20 (28.6%) patients. Serum lipase (cut-off > 51U/L) and IL-6 (> 56.5 ng/l) on POD3 were significant predictors for CR-POPF (AUC = 0.799, 95%-CI 0.686–0.912 and AUC = 0.784, 95%-CI 0.668–0.900; combined AUC = 0.858, 95%-CI 0.758–0.958; all p < 0.001). Patients with both or one factor(s) above cut-off more frequently developed CR-POPF than cases without (100 vs. 50% vs. 7.5%, p < 0.001). This also applied for overall and severe complications (p = 0.013 and p = 0.009). Conclusions Post-operative pancreatitis and inflammatory response are major determinants for development of POPF. A combination of serum lipase and IL-6 on POD3 is a highly significant early predictor of CR-POPF and overall complications, potentially guiding patient management. Clinical trial registration The study protocol was registered at clinicaltrials.gov (NCT04294797) Electronic supplementary material The online version of this article (10.1007/s00268-020-05768-9) contains supplementary material, which is available to authorized users.


Introduction
Pancreatic resections (PR) represent surgical procedures with considerable rates of mortality and morbidity. Due to constant improvements in surgical technique and complication management, post-operative death has significantly decreased over the last decades currently ranging around 5% in most centres [1][2][3]. In contrast, morbidity following PR remains frequent and gradually increases with the complexity of the procedure performed [4]. While enhanced recovery concepts have resulted in a decline in general post-operative morbidity such as pneumonia or wound infections, specific complications like post-operative pancreatic fistula (POPF), postpancreatectomy haemorrhage (PPH) and delayed gastric emptying (DGE) remain common and often trigger other complications [5]. Early anticipation and treatment of clinically relevant grade B or C POPF (CR-POPF) is of utmost importance to prevent fatal outcome. While most pancreatic fistula is selflimiting without the need of intervention (biochemical leak), persistent uncontrolled and insufficiently drained pancreatic juice leakage can lead to a series of disastrous events including destruction of surrounding tissue and erosion of blood vessels resulting in life-threatening infections, sepsis and haemorrhage [6]. In cancer patients, such major complications may delay the start of adjuvant chemotherapies and influence the patient's prognosis [7,8]. In order to aid with timely detection of potentially severe CR-POPF, evaluation of predictive biomarkers that could be able to differentiate early between non-significant biochemical leaks and CR-POPF is of urgent interest.
The aim of this prospective study was to assess readily available biomarkers of local pancreatic inflammation and systemic inflammatory response in regard to their association with development of CR-POPF, 90-day morbidity and mortality following PR as a tool for post-operative decision-making.

Methods
Following approval by the local ethics committee (study number 1081/2017), all patients undergoing pancreatic head resection or distal pancreatectomy at the Department of Visceral, Transplant and Thoracic Surgery, Medical University of Innsbruck, Austria, between December 2017 and May 2019 were enrolled in this prospective study, and written informed consent was obtained. Exclusion criteria comprised total pancreatectomy, duodenum preserving pancreatic head resection, enucleation and cases with unresectability after exploration. The study was conducted in accordance with the Helsinki declaration 2013 and the STROBE checklist [9], and the protocol was registered at clinicaltrials.gov (NCT04294797).
POPF was defined according to the 2016 update of the International Study Group for Pancreatic Surgery (ISGPS), [10] postpancreatectomy haemorrhage (PPH) and delayed gastric emptying (DGE) according to the respective 2007 ISGPS definitions [11,12]. Post-operative pancreatitis (POAP) was defined by Connor's proposal, with serum amylase/lipase values being increased above the upper limit of normal (53 and 60 U/L, respectively, according to our local laboratory) between surgery (skin closure) and end of POD1 [13]. Failure to rescue (FTR) was the rate of deaths in the total number of patients experiencing complications. All complications were assessed within 90 days after surgery, graded according to the Clavien-Dindo classification [14] and recorded prospectively through our surgical units' auditable database (ChiBase).
Preoperatively, routine laboratory parameters were assessed on the day before surgery (white blood cell count, amylase, lipase, albumin, C-reactive protein, IL-6, IL-8, procalcitonin, TNF-alpha). The same markers were evaluated on the morning of POD1 and POD3 with additional measurement of drain fluid amylase and lipase levels.

Statistical analysis
Data are reported as mean (SD), median (range) or numbers with percentages. ROC analysis was performed to assess laboratory values regarding prediction of CR-POPF, and optimal cut-offs were calculated with the Youden index. Risk groups were compared regarding outcome parameters with appropriate two-tailed contingency tests for categorical variables and with the Kruskal-Wallis test for continuous data with non-normal distribution. P values B 0.05 were considered statistically significant, and analyses were performed using SPSSÒ version 23 (IBM, Armonk, New York, USA). Perioperative dynamics of serum markers were displayed with GraphPad Prism 8.1.2 (GraphPad Software Inc., La Jolla, California, USA).

Subgroup analysis of pancreatic head and distal resections
The combination of IL-6 and lipase showed a strong correlation with CR-POPF independently of the performed procedure with an AUC of 0.880 (95%-CI 0.754-1.000) following pancreatic head resections (p \ 0.001) and an AUC of 0.826 (95%-CI 0.641-1.000) following distal pancreatectomies (p = 0.008). However, in distal resections, IL-8 in particular showed a higher AUC than IL-6. Supplemental Table 1 portrays ROC subgroup analysis for all markers.

Discussion
The present study prospectively assessed perioperative biomarkers and their association with complications including POPF after PR. It establishes a clear link between markers of local pancreatic inflammation (serum lipase), measurable systemic response (IL-6) and risk of development of CR-POPF. These markers are detectable early post-operatively, clinically well established and most importantly predict CR-POPF as well as overall morbidity with significant accuracy in both pancreatoduodenectomy and distal resections, although with higher diagnostic yield in pancreatoduodenectomies.
First, we have confirmed recent findings from others, showing that PR remain procedures associated with significant post-operative morbidity and mortality even in high-volume units. Four patients died during hospital stay with two of them due to surgical complications and two due to cardiac infarction after an initially uneventful postoperative course. To prevent future cardiac complications in our institution, preoperative work-up for major abdominal procedures now includes routine cardiac evaluation with cardio-pulmonary exercise testing and-if indicatedcoronary artery angiography. Surgical complication related in-hospital mortality in this series was 2.8%, which is in line with that of other high-volume centres [1,2] and comparable to previously published data of our centre [3]. The high number of vascular reconstructions and concurrent other organ resections shows the progressive approach recently applied in our institution, potentially suggesting a need for optimization in patient selection as well as perioperative management. Overall complication rate in our series reached 75.7%, which also seems high compared to retrospective series with complication rates between 30 and 60% [17]. Since 35% of our post-operative complications are classified as Clavien-Dindo 1-2 with no relevant impact on the further post-operative course, this observed difference is most certainly a result of meticulous prospective auditable documentation. Major complications  occurred in 40% of patients, which is comparable to results reported by other prospective series or randomized trials [18][19][20]. Rates of pancreatic fistula highly differ in the literature ranging from 2-63% [6,21,22]. In our cohort, 28.6% of patients experienced CR-POPF. Early recognition of patients potentially developing severe fistula allows a personalized approach in post-operative management. While high-risk cases might benefit from early initiation of diagnostic and therapeutic steps, patients at low risk for CR-POPF can be allocated to early oral feeding pathways and undergo timely removal of their perianastomotic drains [23]. The post-operative systemic response to a local inflammatory stimulus is strongly related to complications after gastrointestinal surgery [24]. Van Hilst et al. showed higher IL-6 levels in patients with major complications and CR-POPF in 38 patients in the LEOPARD-2 trial [20,25]. IL-6 is a cytokine that induces the production of acutephase proteins such as CRP in the liver. IL-6 levels at 24 h post-operatively in abdominal surgery have been previously shown as accurate in predicting complications as CRP at 72 h [20,24,25]. In our analysis, IL-6 at POD3 was superior to IL-6 at POD1 and CRP at any time point in terms of accuracy to predict CR-POPF (Table 3).
Furthermore, the measurable response to pancreatitis (serum amylase and lipase) has previously been assessed for the prediction of CR-POPF [26,27]. While both markers on POD3 were strongly associated with CR-POPF in our cohort, their predictive value on POD1 was of borderline significance. The presence of POAP as defined by Connor [13] was only associated with further development of POPF when biochemical (not clinically relevant) leaks were included. This suggests a limited power in our cohort with a considerably low number of patients developing POAP (35.7%) compared to 55.8% in a previous Italian study involving 292 patients [28]. In another retrospective analysis, serum lipase at POD1 was assessed in 98 patients undergoing pancreatoduodenectomy, reporting that patients with levels below a cut-off of 44.5 U/L had a substantially lower probability to develop CR-POPF [29]. Intriguingly, our calculated cut-off of 51 U/L was quite comparable, despite measurement on POD3.
Attempting to incorporate both the local pancreatic remnant inflammatory state and the systemic response to improve early post-operative risk stratification, we combined serum lipase and IL-6 on POD3. This ultimately resulted in three groups of patients with significantly different risks for the development of CR-POPF. While all patients with both markers high developed CR-POPF, the rate was only 7.5% with no factor raised above cut-off. The resulting high NPV (92.5%) was markedly superior to that of drain amylase (78.1%). This could strongly impact clinical management, enhanced recovery programs and drainage removal strategies [30]. All patients with both markers increased should be observed with great vigilance to timely initiate further diagnostics and therapeutics and prevent fatal complications of insufficiently drained and persistent fistulas. Importantly, although risk factors for CR-POPF differ between pancreatic head and distal resections, the correlation of IL-6 and serum lipase with later CR-POPF was highly predictive in both subgroups.
Limitations of our study include the single-centre design, inclusion of head and distal resections, application of different techniques for pancreatojejunostomy  ([10%), which needs to be taken into consideration regarding external validity. Further prospective, international validation with a larger sample size and detailed subgroup analysis should be performed. Particularly for pancreatoduodenectomies, our score should be compared to others readily available such as the Fistula Risk Score [31]. Moreover, its applicability in differing perioperative strategies (anastomotic reconstruction techniques, drain management) requires validation.
In conclusion, post-operative complications and POPF remain a major issue even in high-volume centres [21]. The local pancreatic and systemic inflammatory response appears to be decisive, showing a strong association of POD3 serum IL-6 and lipase with CR-POPF and severe complications. The resulting risk groups according to proposed marker cut-off levels allow for improved stratification compared to established criteria such as drain amylase. In case of confirmative validation, these results might foster the development of a new approach in predicting and grading CR-POPF.
Funding Open access funding provided by Medical University of Innsbruck.

Compliance with ethical standards
Conflict of interest All authors declare no conflict of interest for this article.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons. org/licenses/by/4.0/.