No Drain,No Problem: Serum-Based Marker Prediction of Clinically Relevant Postoperative Pancreatic Fistula

Abstract

Background

Postoperative day (POD) 1 drain amylase concentration (DAC) is considered the most accurate predictor for the development of a clinically relevant postoperative pancreatic fistula (CR-POPF) after pancreaticoduodenectomy (PD). Recent studies have associated drain placement with negative postoperative outcomes. This study aims to evaluate multiple biochemical markers and their associations with CR-POPF development in order to identify a reliable, non-drain dependent alternative to DAC.

Methods

This is a review of 53 consecutive PD patients between 2021 and 2022. Albumin, C-reactive protein (CRP), C-reactive protein-to-albumin ratio (CAR), DAC, white blood cell count, and procalcitonin values were compared by CR-POPF status. The discriminatory abilities of CAR, CRP, and DAC for CR-POPF were compared using receiver operating characteristic (ROC) curves.

Results

Six of 51 included patients developed a CR-POPF. Receiver operating characteristic curve analysis produced an area under the curve of .977 for POD 1 DAC (cut-off 5131.0 IU/L, sensitivity 100%, specificity 95.5%), .858 for POD 1 CRP (cut-off 52.5 mg/L, sensitivity 100%, specificity 72.7%), and 1.000 for POD 3 CAR (cut-off 99.2, sensitivity and specificity 100%). POD 3 CAR produced a positive and negative predictive value of 100%.

Conclusion

The CAR and CRP provide early and accurate identification of patients with post-PD CR-POPFs. These markers offer a method of safe CR-POPF detection, when the gold standard DAC is unavailable, ultimately allowing for early intervention and patient rescue.

Keywords

pancreatic fistula C-reactive protein-to-albumin ratio pancreaticoduodenectomy C-reactive protein

Key Takeaways

• C-reactive protein-to-albumin ratio and C-reactive protein serum markers identify patients at risk for CR-POPF.

• These serum biomarkers are as accurate as drain amylase concentration for detection of CR-POPF.

• The use of procalcitonin, WBC count, or serum albumin as sole biochemical predictive marker for CR-POPF is not supported.

Introduction

Reported rates of clinically relative postoperative pancreatic fistula (CR-POPF) after pancreaticoduodenectomy (PD) at high volume centers are consistently between 10 and 20% in the United States.^1-3 The sequelae resulting from CR-POPF are divided into two broad categories: sepsis and hemorrhage, both of which can lead to disastrous clinical consequences and ultimately delay life-preserving adjuvant therapy.⁴ At this time, there is no agreed upon technical method for the absolute prevention of CR-POPFs; thus, accurate early identification of patients at risk for a CR-POPF is of critical importance. Early identification and intervention (ie, rescue) may improve the long stagnant morbidity rate associated with PD.

The International Study Group in Pancreatic Surgery (ISGPS) defined a CR-POPF as a drain amylase concentration (DAC) three times the upper limit of normal serum levels on postoperative day (POD) 3, or later, in the setting of a clinically relevant development or condition.⁵ While POD 3 DAC is required for diagnostic CR-POPF confirmation, POD 1 DAC is often considered the most accurate predictor, or screening test, for CR-POPF. Drain amylase concentration (DAC)-based testing is problematic as it requires an intact and functional peri-pancreatic drain.

Traditionally, at least one prophylactic intraperitoneal drain is placed near the pancreaticojejunal anastomosis and maintained through POD 3 or longer, but a global survey of pancreatic surgeons found that 41% reported not regularly placing drains during PDs and significantly more reported practicing early drain removal, often on POD 1.⁶ This change in practice is presumably a result of reports of increased complication rates with prophylactic drainage, including CR-POPF and intraabdominal infection, after PD.⁷ Many pancreatic surgeons now agree that prophylactic drainage is unnecessary in low-risk patients, although its role in moderate to high-risk patients remains debated.⁸

In response to these changing clinical practices, multiple inflammatory markers have been investigated for their screening and diagnostic capabilities in relation to CR-POPFs. While C-reactive protein (CRP) has shown promise in large studies, other markers being evaluated include procalcitonin and white blood cell (WBC) count.^1,9 One argument against the use of inflammatory marker measurement alone is that they fail to fully account for the patient’s nutritional status. Albumin, a commonly used nutritional marker, has been associated with complication severity after PD.¹⁰

C-reactive protein-to-albumin ratio (CAR) is a novel biochemical marker that accounts for both inflammation and nutritional status. Elevated preoperative CAR values have been shown to be a strong, independent predictor of poor prognosis after pancreatic adenocarcinoma resection, but exceedingly few studies have assessed if CAR is a viable screening or diagnostic marker for the development of CR-POPF.^11,12 Further, only Sakamoto et al¹³ has directly compared CAR to the current standard for CR-POPF screening, POD 1 DAC.

This study aims to evaluate various biochemical markers’ association with CR-POPFs. The authors hypothesize that serum-based biomarkers will allow for accurate and reliable early screening of patients at risk for the development of a CR-POPF after PD. These markers may provide a future avenue for CR-POPF detection without the requirement of intraperitoneal drainage.

Methods

This is an institutional review board (IRB) approved analysis of 53 consecutive PD patients at a single institution with a single surgeon between April 2021 and July 2022. Informed consent was waived by the IRB due to the study’s retrospective nature.

Through extensive chart review, demographic information was collected including age, sex, body mass index, and American Society of Anesthesiologists (ASA) physical status classification. The operative duration, estimated blood loss (EBL), pancreatic texture, and main pancreatic duct size were obtained. Postoperative information regarding serum and drain studies, inpatient hospital course, and the observed development of a CR-POPF was collected.

The 2016 ISGPS terminology is utilized throughout this study. Based on the upper limit of normal serum amylase at this institution, a POD 3 or later DAC greater than 445 IU/L was determined to be a biochemical leak or CR-POPF, as appropriate. Drain amylase concentration values reported as less than 30 IU/L (undetectable) were interpreted as having a value of zero to allow for statistical analysis.

Continuous variables are expressed as the mean with standard deviation (SD) or median with interquartile range (IQR) based on normality, while categorical variables are expressed as counts and proportions. Quantitative data were analyzed using Student’s t test, Mann-Whitney U test, Wilcoxon matched signed-rank test, and Friedman and repeated measures ANOVA where appropriate. All post hoc analysis was completed with Bonferroni adjustment for multiple comparisons. Qualitative data were analyzed using a chi-square or Fisher’s exact test. Point-biserial correlation analysis was completed to establish associations between biochemical markers and CR-POPFs. Receiver operating characteristic (ROC) curve analysis was performed and the results were reported as the area under the curve (AUC) with associated P-value and 95% confidence interval (CI). A cut-off value obtained from the maximum Kolmogorov-Smirnov metric with its sensitivity and specificity is provided for selected results. Reported P-values less than .05 were considered significant. All data analyses were performed using IBM SPSS version 28 (Predictive Analytics Software, Armonk, NY).

Intraoperatively, all patients underwent a Catell-Braasch maneuver and complete Kocherization of the duodenum. Required dissections were completed and resectability was verified. An “SMA first” approach to resection was utilized in most cases. The antrum of the stomach and proximal jejunum were divided using a linear stapling device while the bile duct and pancreatic neck were sharply transected. Once the specimen was removed, the proximal jejunum was brought under the SMA and a side-to-side pancreaticojejunostomy was performed in a modified Blumgart manner. The duct-to-mucosa anastomosis was completed using 5-0 Proline suture while 2-0 silk was used in the evagination of the pancreatic parenchyma. 8 cm of 5 F pediatric feeding tube was utilized as a pancreatic duct stent in most cases. An end-to-side hepaticojejunostomy in a running fashion was performed on the same loop of jejunum followed by a stapled gastrojejunostomy. Rarely, the falciform ligament was utilized as a flap to create a barrier between the pancreatic anastomosis and the gastroduodenal artery stump. A 19 F round drain was placed through the right lower abdominal wall to drain the peripancreatic region in all patients.

Peripancreatic drains were removed after POD 3 provided the drain amylase was low and the output was serous. All attempts were made to remove the surgical drain prior to hospital discharge, but when unsuccessful the need for drainage was reassessed at each postoperative visit.

Results

Two procedures were aborted prior to completion resulting in exclusion of those patients from analysis. Of the 51 patients who completed a PD during the study period, 34 (66.7%) were without evidence of a biochemical leak postoperatively, 11 (21.6%) biochemical leaks were identified, and 6 (11.8%) patients developed a CR-POPF. Of those who developed a CR-POPF, 5 (9.8%) were grade B and 1 (2%) was a grade C. Table 1 compares the demographic and perioperative results.

Table 1.

Comparison of Demographic and Perioperative Results.

Demographic and Perioperative Results		No CR-POPF (n = 45)	CR-POPF (n = 6)	P-Value
Age at operation, years	Mean ± SD	64 ± 11	63 ± 11	.925
BMI		29.1 ± 6.9	28.2 ± 6.6	.754
Pancreatic duct diameter, mm^a		3.7 ± 3.0	1.33 ± 0.5	<.001
Operative time, min		221.3 ± 36.5	221.3 ± 30.2	.998
EBL, mL	Median (IQR)	350 (300)	675 (700)	.179
Male	Count	23	4	.671
ASA classification	II	4	0	.447
	III	32	3	.295
	IV	9	3	.104
Pancreatic texture^a	Soft	19	5	.188
Pancreatic texture^a	Firm	23	1	.188
Pathology	PDAC	25	1	.099
	Ampullary carcinoma/adenoma	3	1	.404
	Cholangiocarcinoma	1	1	.224
	IPMN	2	0	1.000
	pNET	2	1	.319
	Chronic pancreatitis	5	0	1.000
	Duodenal carcinoma/adenoma	4	1	.480
	Other	3	1	.404

Abbreviations: CR-POPF, clinically relevant postoperative pancreatic fistula; BMI, body mass index; SD, standard deviation; IQR, interquartile range; EBL, estimated blood loss; ASA, American Society of Anesthesia; PDAC, pancreatic ductal adenocarcinoma; IPMN, intraductal papillary mucinous neoplasm; pNET, pancreatic neuroendocrine tumor.

^aData not available for 3 patients.

Biochemical Marker Results

Comparisons of albumin, DAC, CAR, CRP, procalcitonin, and WBC values between patients with and without a CR-POPF can be found in Table 2 with their associated P-value.

Table 2.

Central Tendencies of Biochemical Markers Between Groups.

Biochemical Marker			No CR-POPF (n = 45)	CR-POPF (n = 6)	P-value
Albumin, g/dL	POD 1	Mean ± SD	3.45 ± 0.43	3.20 ± 0.32	< .001
	POD 2		3.15 ± 0.33	2.98 ± 0.17	< .001
	POD 3		3.06 ± 0.36	2.87 ± 0.10	< .001
CAR	POD 1^a	Median (IQR)	13.0 (7.50)	21.7 (40.8)	.013
	POD 2^a		47.1 (35.1)	109.9 (47.2)	< .001
	POD 3^b		44.8 (45.9)	134.0 (48.5)	< .001
CRP, mg/L	POD 1^a	Median (IQR)	48.0 (24.0)	66.0 (108)	.021
	POD 2^a		160.0 (106.0)	333.5 (103.0)	< .001
	POD 3^b		148.0 (150.0)	395.5 (131.0)	< .001
DAC, IU/L	POD 1^c	Median (IQR)	181.5 (1 705)	19 567.0 (139 772)	< .001
	POD 3		100.0 (404.0)	2 384.0 (16 107)	< .001
	POD 5^d		15.5 (98.0)	1 788.0 (17 646)	< .001
Procalcitonin, ƞg/mL	POD 1^e	Median (IQR)	0.93 (2.0)	0.37 (3.0)	.642
	POD 2^f		0.85 (1.0)	0.52 (2.0)	.832
	POD 3^e		0.53 (1.0)	0.73 (1.0)	.409
WBC count, k/µL	POD 1	Mean ± SD	13.2 ± 3.4	16.5 ± 4.4	.037
	POD 2		12.8 ± 3.9	18.7 ± 6.7	.003
	POD 3		11.7 ± 4.0	14.8 ± 5.9	.106

Abbreviations: CR-POPF, clinically relevant postoperative pancreatic fistula; POD, postoperative day; CAR, C-reactive protein-to-albumin ratio; CRP, C-reactive-to-protein; DAC, drain amylase concentration; WBC, white blood cell.

^aData not available for 25 patients.

^bData not available for 28 patients.

^cData not available for one patient.

^dData not available for 14 patients.

^eData not available for 26 patients.

^fData not available for 24 patients.

Mean albumin levels were significantly lower in those with CR-POPFs on all days analyzed. While there was an overall negative trend in the mean albumin for both groups from POD 1 through POD 3, this difference was not large enough to demonstrate statistical significance among the CR-POPF cohort (F (1.0, .33) = 5, P = .073).

While the CR-POPF group was found to have a significantly higher median DAC on all analyzed PODs compared to those without, there was extreme variability in the measured DAC values. This was particularly evident amongst the CR-POPF group and is reflected in an IQR of 139772 IU/L on POD 1.

The median CRP of CR-POPF patients was significantly higher than those without a CR-POPF on all PODs analyzed. For those without a CR-POPF, there was a CRP peak median value on POD 2, while the peak was after POD 2 for those with a CR-POPF. On Friedman ANOVA, it was noted that both study groups had a significant change in median values between POD 1 through 3 ( $χ_{C R - P O P F}^{2} (2) = 6.50, p = 0.039;$ $χ_{N o C R - P O P F}^{2} (2) = 26.00 p < 0.001)$ . Although on further analysis, neither group had a significant change between POD 2 and 3 (T_CR-POPF = .50, P = .480; T_{No CR-POPF} = .21, P = .516).

Like CRP, the median CAR value was significantly greater on all analyzed PODs for the CR-POPF group compared to the group without. Additionally, the peak CAR occurred on POD 2 among those without a CR-POPF while those with had a median peak value that occurred later, but again there was no statistical difference in POD 2 compared to POD 3 values among those with a CR-POPF ( $χ_{C R - P O P F}^{2} (2) = 6.50, p = 0.039;$ T_{POD 2-3} = .50, P = 1.000).

Point-Biserial Correlations

Table 3 displays the correlation analysis results between the biochemical markers and the development of a CR-POPF. Notably, there was a strong positive correlation between DAC on POD 1 and the development of CR-POPF, which weakened to a moderate positive correlation on POD 3 and POD 5. Both CRP and CAR were found to be progressively more positively correlated with CR-POPF with each POD which resulted in strongly positive correlations for both on POD 3 (CAR r = .81, CRP r = .79, both P < .001).

Table 3.

Point-Biserial Correlation Analysis of Biochemical Markers and CR-POPF.

Biochemical Marker			r	P-Value
Albumin, g/dL	POD 1	N = 51	−.20	.171
	POD 2	N = 51	−.17	.234
	POD 3	N = 51	−.018	.207
CAR	POD 1	N = 26	.50	.010
	POD 2	N = 26	.68	<.001
	POD 3	N = 23	.81	<.001
CRP, mg/L	POD 1	N = 26	.48	.014
	POD 2	N = 26	.65	<.001
	POD 3	N = 23	.79	<.001
DAC, IU/L	POD 1	N = 50	.61	<.001
	POD 3	N = 51	.45	<.001
	POD 5	N = 37	.40	.014
Procalcitonin, ηg/mL	POD 1	N = 25	−.05	.817
	POD 2	N = 27	−.07	.723
	POD 3	N = 25	.02	.912
WBC count, k/μL	POD 1	N = 51	.29	.037
	POD 2	N = 51	.41	.003
	POD 3	N = 51	.23	.106

ROC Curve Analysis

Figure 1 displays the results of biochemical marker paired-sample ROC analysis of the 26 patients with complete DAC, CAR, and CRP data on POD 1. No difference was demonstrated between the AUC for POD 1 DAC and CAR (P = .281) or between DAC and CRP (P = .200). Values of 5131.0 IU/L, 15.5, and 52.5 mg/L or greater on for DAC, CAR, and CRP, respectively, were found to be positive screening tests for the future development of a CR-POPF. This produced a sensitivity of 100% for all three markers and a specificity of 95.5% for DAC, 68.2% for CAR, and 72.7% for CRP.

On POD 1, DAC values were available for 50 patients in this study. Of those, 10 (20%) patients were found to have a POD 1 DAC value of at least 5131.0 IU/L. Based on this cut-off, all 6 patients who developed a CR-POPF were correctly identified, but 4 patients were incorrectly classified as being at risk. This resulted in a positive predictive value (PPV) of 60% and a negative predictive value (NPV) of 100%. The CAR values were available for 26 patients on POD 1. Of those, 11 (42.3%) were found to have a POD 1 CAR value of at least 15.5. All 4 patients who developed a CR-POPF with a known POD 1 CAR value were correctly identified based on this cut-off value, but 7 were incorrectly classified as being at risk (PPV 36.4%, NPV 100%). CRP values were available for 26 patients on POD 1. Of those, 10 (38.5%) were found to have a POD 1 CRP value of at least 52.5 mg/L. All 4 patients were correctly identified based on this cut-off value, although 6 patients were incorrectly identified as being at risk (PPV 40%, NPV 100%).

Receiver operating characteristic analysis of the 23 patients with complete CAR data for POD 1 through 3 is presented in Figure 2. The POD 3 was the best predictor of CR-POPF, of the days analyzed, for CAR with an AUC of 1.000 (P = .002; 95% CI 1.000–1.000), although there was no statistical difference in the reported AUC for each POD (all P > .156). A cut-off value of 99.2 was identified on POD 3 and associated with a sensitivity and specificity of 100%.

On POD 3, 23 patients had available CAR values. 4 (17.4%) patients were identified to have a CAR value greater than 99.2, 100% of which developed a CR-POPF (PPV 100%, NPV 100%). This is compared to the results of DAC with a cut-off of 445 IU/L, as suggested by the ISGPS guidelines, which determined 14 (27.5%) patients to be at risk for CR-POPF development. This DAC cut-off accurately identified all 6 patients who would develop a CR-POPF, but incorrectly identified 8 patients to be at risk (PPV 42.9%, NPV 100%).

Ultimately of the 4 patients identified by CAR to have a CR-POPF, all of which were grade B fistulas. 1 (25%) required interventional radiology (IR) for arterial coil embolization and 3 (75%) required extended hospitalization times (range 8-24 days), prolonged surgical drainage, and antibiotics. Of the 2 patients without available CAR data, 1 (50%) died during the index hospitalization on POD 29. The remaining 1 (50%) patient required additional IR intraperitoneal drain placement. While not statistically significant (P = .294), the mean length of hospital admission was 10 days amongst patients without a CR-POPF while it was 15 days for those with.

Discussion

Serum-based biomarkers, CAR and CRP, each demonstrate a remarkable ability to screen for CR-POPF after PD on POD 1. In fact, both were found to be equivalent to the often-touted POD 1 DAC predictive ability. Furthermore, CAR shows potential for diagnostic use on POD 3 with 100% accuracy in identifying patients at risk for future CR-POPF development. The results of this study support that reliable, early identification of patients who will develop a post-PD CR-POPF can be made without a peritoneal drain. A heightened level of clinical scrutiny and low-threshold for cross-sectional imaging in the so-identified POD 3 patients will allow for early intervention with antibiotics, percutaneous drain placement, and surgical intervention as necessary. This may ultimately result in improved patient outcomes and shorter hospital stays.

These results additionally support the widely accepted associations between CR-POPF and small duct size, soft pancreatic texture, and higher intraoperative EBL. The data reflect a POD 1 DAC of 5131.0 IU/L or greater to be indicative of an at-risk patient for the future development of CR-POPF. This is consistent with multiple previously reported cut-off values near 4000 to 5000 IU/L on POD 1.^14-17 Together, these consistent results provide validity for the dataset and support our novel biochemical marker findings.

Based on the results of this study, the use of procalcitonin, WBC count, or serum albumin as a standalone biochemical predictive marker for CR-POPF is not supported. Procalcitonin was found to have no discernable relationship with fistula development after PD. While WBC count and albumin appear to have some association with CR-POPF development, the results suggest they lack a large enough effect size or consistent enough trend to be reliable markers.

Similar to other available studies, CRP was found to be an excellent predictor of CR-POPF development on POD 1.^1,9 It is theorized that many postoperative complications are a result of an excessive immune mediated inflammatory response when technical or judgment-based errors are not at fault.¹⁸ This, in addition to the fact that CRP levels rise and fall rapidly with the introduction or removal of an inflammatory stimulus, likely account for CRP’s CR-POPF predictive prowess.¹⁹ This is contrasted with other acute phase reactants, such as albumin, that are significantly slower to reflect changes in inflammation. CRP is likely the primary driver for the success of CAR in CR-POPF prediction, but the effect of hypoalbuminemia cannot be understated in postoperative healing. Further study will be necessary to evaluate if CAR demonstrates a substantial improvement over CRP in screening and diagnostic identification of CR-POPF after PD.

One of the strongest arguments against reliance on any inflammatory marker for prediction and detection of CR-POPF after PD is related to the potentially elevated baseline levels of inflammation in patients with active autoimmune diseases, such as Lupus. Ansorge et al⁹ sought to resolve this problem by determining that CRP values peaked after POD 2 among those who developed a CR-POPF. This study’s results are also superficially suggestive of these findings, although upon further analysis there was not a statistically significant difference between the median CRP value on POD 2 and POD 3. An alternative method of dealing with these patients may include a higher cut-off value for test positivity. This too will require further study prior to widespread use of CRP or CAR-based CR-POPF testing.

This study has several obvious limitations which are primarily a result of its retrospective nature, very limited number of patients who developed a CR-POPF, and overall sample size. Additionally, as there was not a standardized protocol for obtaining biochemical markers postoperatively there is undoubtedly unaccounted for bias in data availability for patients. Due to these factors, future prospective studies with a large sample size will be required to validate this study’s findings and further define the ideal cut-off value for CR-POPF identification for each POD.

In conclusion, the use of biochemical markers to predict a CR-POPF after PD is possible in the absence of prophylactic peri-pancreatic drainage. With further study, it is conceivable that an alternative definition of CR-POPF could exist that utilizes CAR or CRP values instead of DAC in the setting of a clinically relevant development or condition. Multiple, prospective studies will be needed to affirm the results of this data, delineate the superiority or equivalence of CAR and CRP, and to further clarify result interpretation among patients with systemic inflammatory diseases.

Footnotes

Declaration of Conflicting Interests

The author(s) declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: While the authors have no relevant conflicts of interest to this study, Dr D. Rohan Jeyarajah is a consultant for Ethicon Inc., a consultant for Angiodynamics, a consultant for SirTex Medical, and on the Angiodynamics safety monitoring board. Dr Houssam Osman, Dr Jessica Heard, Mr. Glen Smith, Mr. Alex Tolman, Mr. Jashwanth Karumuri have no disclosures of any kind.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

IRB Protocol

Methodist Health System: 036. HPB.2018.R.

References

Mintziras

Maurer

Kanngiesser

Bartsch

. C-reactive protein and drain amylase accurately predict clinically relevant pancreatic fistula after partial pancreaticoduodenectomy. Int J Surg. 2020;76:53-58. doi:10.1016/j.ijsu.2020.02.025

Schorn

Demir

Vogel

, et al. Mortality and postoperative complications after different types of surgical reconstruction following pancreaticoduodenectomy-a systematic review with meta-analysis. Langenbecks Arch Surg. 2019;404(2):141-157. doi:10.1007/s00423-019-01762-5

Callery

Pratt

Kent

Chaikof

Vollmer

. A prospectively validated clinical risk score accurately predicts pancreatic fistula after pancreatoduodenectomy. J Am Coll Surg. 2013;216(1):1-14. doi:10.1016/j.jamcollsurg.2012.09.002

Nahm

Connor

Samra

Mittal

. Postoperative pancreatic fistula: a review of traditional and emerging concepts. Clin Exp Gastroenterol. 2018;11:105-118. doi:10.2147/CEG.S120217

Bassi

Marchegiani

Dervenis

, et al. The 2016 update of the International Study Group (ISGPS) definition and grading of postoperative pancreatic fistula: 11 years after. Surgery. 2017;161(3):584-591. doi:10.1016/j.surg.2016.11.014

McMillan

Malleo

Bassi

Sprys

Vollmer

. Defining the practice of pancreatoduodenectomy around the world. HPB (Oxford). 2015;17(12):1145-1154. doi:10.1111/hpb.12475

Zhou

Zhang

. Evidence-based value of prophylactic intraperitoneal drainage following pancreatic resection: A meta-analysis. Pancreatology. 2014;14(4):302-307. doi:10.1016/j.pan.2014.04.028

Wang

Jiang

, et al.

Is routine drainage necessary after pancreaticoduodenectomy?

World J Gastroenterol. 2014;20(25):8110-8118. doi:10.3748/wjg.v20.i25.8110

Ansorge

Nordin

Lundell

, et al. Diagnostic value of abdominal drainage in individual risk assessment of pancreatic fistula following pancreaticoduodenectomy. Br J Surg. 2014;101(2):100-108. doi:10.1002/bjs.9362

10.

Winter

Cameron

Yeo

, et al. Biochemical markers predict morbidity and mortality after pancreaticoduodenectomy. J Am Coll Surg. 2007;204(5). doi:10.1016/j.jamcollsurg.2007.01.026

11.

Ikeguchi

Hanaki

Endo

, et al. C-Reactive protein/albumin ratio and prognostic nutritional index are strong prognostic indicators of survival in resected pancreatic ductal adenocarcinoma. J Pancreat Cancer. 2017;3(1):31-36. doi:10.1089/pancan.2017.0006

12.

Liu

Wang

Peng

. Predictive value of drain pancreatic amylase concentration for postoperative pancreatic fistula on postoperative day 1 after pancreatic resection: An updated meta-analysis. Medicine (Baltimore). 2018;97(38):e12487. doi:10.1097/MD.0000000000012487

13.

Sakamoto

Yagyu

Uchinaka

, et al. Predictive significance of c-reactive protein-to-albumin ratio for postoperative pancreatic fistula after pancreaticoduodenectomy. Anticancer Res. 2019;39(11):6283-6290. doi:10.21873/anticanres.13838

14.

Partelli

Pecorelli

Muffatti

, et al. Early postoperative prediction of clinically relevant pancreatic fistula after pancreaticoduodenectomy: usefulness of C-reactive Protein. HPB (Oxford). 2017;19(7):580-586. doi:10.1016/j.hpb.2017.03.001

15.

Israel

Rettammel

Leverson

, et al.

Does postoperative drain amylase predict pancreatic fistula after pancreatectomy?

J Am Coll Surg. 2014;218(5):978-987. doi:10.1016/j.jamcollsurg.2014.01.048

16.

Molinari

Bassi

Salvia

, et al. Amylase value in drains after pancreatic resection as predictive factor of postoperative pancreatic fistula: Results of a prospective study in 137 patients. Ann Surg. 2007;246(2):281-287. doi:10.1097/SLA.0b013e3180caa42f

17.

Nissen

Menon

Puri

Annamalai

Boland

. A simple algorithm for drain management after pancreaticoduodenectomy. Am Surg. 2012;78(10):1143-1146.

18.

Plas

Rutgers

van der Wal-Huisman

, et al. The association between the inflammatory response to surgery and postoperative complications in older patients with cancer; a prospective prognostic factor study. J Geriatr Oncol. 2020;11(5):873-879. doi:10.1016/j.jgo.2020.01.013

19.

Nehring

Goyal

Patel

. C Reactive Protein. In: StatPearls. Tampa, FL: StatPearls Publishing; 2022. http://www.ncbi.nlm.nih.gov/books/NBK441843/. Accessed July 27, 2022.