- Email: [email protected]
Significance of preoperative radiographic pancreatic density in predicting pancreatic fistula after surgery for pancreatic neuroendocrine tumors
The American Journal of Surgery (2015) -, -–-
Significance of preoperative radiographic pancreatic density in predicting pancreatic fistula after surgery for pancreatic neuroendocrine tumors Yasmine Assadipour, M.D.a, Saı¨d C. Azoury, M.D.a,b, Nicholas N. Schaub, M.D.a, Young Hong, M.D.a, Robert Eil, M.D.a, Suzanne M. Inchauste, M.D.a, Seth M. Steinberg, Ph.D.c, Aradhana M. Venkatesan, M.D.d, Steven K. Libutti, M.D., F.A.C.S.e, Marybeth S. Hughes, M.D., F.A.C.S.a,b,* a
Surgery Branch, Center for Cancer Research, National Cancer Institute, Bethesda, MD; bThoracic and Gastrointestinal Oncology Branch, Center for Cancer Research, National Cancer Institute, Bethesda, MD; cBiostatistics and Data Management Section, Center for Cancer Research, National Cancer Institute, Bethesda, MD; dRadiology and Imaging Sciences NIH Clinical Center, National Institutes of Health, Bethesda, MD; eDepartment of Surgery, Albert Einstein College of Medicine, Montefiore Medical Center, Bronx, NY
KEYWORDS: Pancreatic fistula; Neuroendocrine tumors; Pancreaticoduo denectomy; Pancreatic resection; Pancreatic enucleation
Abstract BACKGROUND: Postoperative pancreatic fistula remains the most severe and worrisome complication after surgery. Predictive preoperative assessment remains challenging. The authors examine the role of pancreatic computed tomography density in predicting postoperative pancreatic fistula after surgery for pancreatic neuroendocrine tumors. METHODS: A single institutional retrospective review of pancreatic surgery for neuroendocrine tumors between 1998 and 2010 was conducted. Preoperative contrast-enhanced computed tomography scans were reviewed, with mean region of interest measurements of pancreatic parenchymal density obtained from 10-mm thick axial computed tomography images. RESULTS: A total of 119 patients were identified: 59 with enucleations and 60 with resections. Decreased preoperative pancreatic density was significantly associated with an increased grade of postoperative pancreatic fistula (P , .01). Subgroup analyses revealed that decreased gland density was associated with increased grade of postoperative pancreatic fistula in the resection (P , .01) but not in the enucleation group (P 5 .34).
The authors have no conflicts of interest to disclose. All funding is provided by the Intramural Program of the National Cancer Institute; this did not create a conflict of interest. This study was approved by the appropriate institutional review board and has been performed in accordance with the ethical standards as laid down in the 1964 Declaration of Helsinki and its later amendments or comparable ethical standards. This article does not contain any studies with human participants or animals performed by any of the authors. * Corresponding author. Tel.: 301-594-9341; fax: 301-451-6933. E-mail address: [email protected] Manuscript received January 22, 2015; revised manuscript April 30, 2015
0002-9610/$ - see front matter Published by Elsevier Inc. http://dx.doi.org/10.1016/j.amjsurg.2015.07.031
The American Journal of Surgery, Vol -, No -, - 2015
2
CONCLUSIONS: A significant association between postoperative pancreatic fistula grade and preoperative pancreatic computed tomography density is observed in patients undergoing resection for pancreatic neuroendocrine tumors. Published by Elsevier Inc.
Pancreatic fistula remains a significant complication after surgery, and predictive preoperative assessment remains challenging.1 Pancreatic fistula occurrence after tumor resection adds significant morbidity to a patient’s postoperative course and can be life-threatening. Patients who experience a pancreatic leak are at greatly increased risk of other morbidities, including sepsis, biliary strictures, and mortality.2 Although fistulas have an overall mortality of 5% to 8%, and can generally be managed conservatively with prolonged nil per os (npo) status and parenteral nutrition, mortality for patients with a high-grade fistula can be as high as 28% to 40%.3,4 Patients with higher grade fistulas will require procedural interventions varying from percutaneous drain placement to reoperation. Many factors have been implicated in a higher incidence of postoperative pancreatic fistula, including pancreatic neuroendocrine tumors, hereditary syndromes, elevated body mass index, and a soft pancreas.5–9 A consensus definition of fistula grades developed by the New International Study Group for Pancreatic Surgery has allowed for objective analysis and comparison of patient outcomes across institutions.10 It has been independently validated and found consistently to correlate with clinical outcomes.11 Use of imaging to predict fistula rate has been examined in patients undergoing pancreaticoduodenectomy for pancreatic adenocarcinoma. Studies of preoperative computed tomography scans determined that visceral fat area was more accurate than body mass index in predicting pancreatic fistula.12,13 Pancreatic density and fistula rate or severity has not been evaluated in patients with neuroendocrine tumors, nor for other resection options such as distal pancreatectomy or enucleation. Herein, the authors aim to examine the role of pancreatic computed tomography density in predicting postoperative pancreatic fistula after surgery for pancreatic neuroendocrine tumors.
Methods After approval by the Office of Human Subject Research at the National Institutes of Health, a retrospective review was conducted on all patients who underwent resection or enucleation for pancreatic neuroendocrine tumors from March 1998 to April 2010. All participants provided informed written consent and were enrolled on protocols at the National Cancer Institute. Patient variables including age, gender, body mass index, hereditary syndrome type, tumor characteristics, and preoperative serum albumin level were reviewed. Body mass index classification (kg/m2) was
noted as normal % 24.9, overweight 25 to 29.9, and obese R 30.0. Operative reports were reviewed for surgical approach and type of resection. After resection or enucleation, closed-suction drains were placed at the surgical bed. Perioperative octreotide as prophylaxis was not used in this series. Drains were removed once patients were tolerating a regular diet and maintaining a normal drain amylase level with output of less than 100 mL/day. Drain amylase levels were obtained for all patients postoperatively. From strict International Study Group for Pancreatic Surgery criteria, postoperative pancreatic fistula were categorized as follows: grade A, asymptomatic and clinically silent fistula, only demonstrated by drain amylase level greater than 3 times the normal serum level; grade B, a fistula requiring a postoperative drain for greater than 21 days, nonoperative intervention such as total parenteral nutrition, octreotide, percutaneous drainage, or readmission within 30 days of hospitalization; and grade C, a severe fistula requiring surgical intervention, intensive care unit care greater than 48 hour, or death within 30 days of operation. Clinically significant postoperative pancreatic fistula was defined as any grade B fistula requiring an intervention and all grade C fistulas. Intervention was defined as requiring an intra-abdominal drain placement or a reoperation. Additional drains were placed when clinically indicated due to fever, leukocytosis, or increased pain necessitating a computed tomography scan showing an inadequately drained collection. Patient records were reviewed to determine the length of stay, clinical course, and complications. Preoperative contrast-enhanced triple phase computed tomography scans were reviewed. Mean region of interest measurements of pancreatic parenchymal density in Hounsfield units were obtained every 10 mm in axial computed tomography images acquired during the portal venous phase. Measurements were determined by blind evaluators and verified independently by a radiologist who was also blind to the clinical course of the subjects. The pancreas density was determined by calculating the mean of the region of interest measurements for each pancreas (Fig. 1). Continuous parameters were compared between patients with and without clinically significant postoperative pancreatic fistula using an exact Wilcoxon rank sum test. The significance of the trends in continuous parameters across ordered categorical parameters such as increasing fistula grade were determined by a Jonckheere–Terpstra test for trend. All P values are 2 tailed and presented without any adjustment for multiple comparisons. P value less than .05 is considered statistically significant.
Y. Assadipour et al.
Pancreas density predicts PNET fistula
3
Figure 1 An example of the method by which region of interest measurements of density were obtained every 10 mm from the preoperative computed tomography scan of a patient in the axial orientation during the portal venous phase.
Results A total of 119 patients who underwent surgery for a pancreatic neuroendocrine tumor at the National Institutes of Health between 1998 and 2010 were identified. Of this group, 70 (59%) were female and 49 (41%) were male, with an average age of 42.3 6 1.36 years for the entire group. Sixty-three (53%) patients had a heritable genetic mutation predisposing them to neuroendocrine tumors, including 28 patients with multiple endocrine neoplasia type 1, 34 with von Hippel–Lindau syndrome, and 1 with neurofibromatosis. Forty-five tumors were functional, and 74 were nonfunctional. Sixty patients underwent enucleation, and 59 patients underwent resection. In the resection group, 34 patients underwent a distal pancreatectomy, 14 patients underwent a Whipple, and 11 underwent a dual Whipple and distal pancreatectomy for multiple lesions not amenable to enucleation. Ninety-seven (82%) patients underwent an open approach, and 22 (19%) underwent a laparoscopic resection. There were no statistically significant differences in patient characteristics between patients with and without a clinically significant pancreatic fistula (Table 1). The average length of stay for patients was 14 6 .85 days. There were 93 patients with no or clinically insignificant fistula (78%), and 26 patients with a clinically significant fistula (22%). Length of stay for patients with a clinically significant fistula (average 23.4 days) was significantly longer (P ,.01) when compared with clinically insignificant and/or no fistula (average 11.7 days). In the 26 patients with clinically significant fistulas, 21 (81%) had a drain placed, and 5 (19%) required a reoperation; 1 patient (4%) developed
sepsis, and there were no mortalities reported within 30 days of operation. In the 93 patients without clinically significant fistulas, 2 patients required reoperations. One patient required a gastrectomy for prolonged gastroparesis after pancreaticoduodenectomy; the second patient developed sepsis after an internal hernia and incarcerated small intestine, underwent reoperation but did not improve and died. The overall mortality for the cohort was .8%, with 1 death in 119 patients (Table 1). An increasing postoperative pancreatic fistula grade was found to be associated with decreasing pancreatic density, measured in Hounsfield units (Fig. 2). The mean pancreatic density for patients with no fistula was 92.17 6 3.03 (n 5 44), for grade A fistula was 81.43 6 5.45 (n 5 26), for grade B was 76 6 4.46 (n 5 39), and for grade C fistula was 36.1 6 10.28 (n 5 4; P , .01; Fig. 2A). In a subgroup analysis of resections and enucleations, this difference was statistically significant in the resection group, and a similar but insignificant trend was found for the enucleation group. The mean pancreatic density for patients who underwent resection with no fistula was 94.85 6 3.82 (n 5 24), for grade A fistula was 73.33 6 9.62 (n 5 11), for grade B was 65.96 6 6.96 (n 5 18), and for grade C fistula was 13 (n 5 1; P , .01; Fig. 2B). The mean pancreatic density for patients who underwent enucleation with no fistula was 88.94 6 4.84 (n 5 20), for grade A fistula was 88.08 6 6.06 (n 5 15), for grade B was 84.66 6 5.18 (n 5 21), and for grade C fistula was 44.48 6 9.34 (n 5 3; P 5 .34; Fig. 2C). Risk factors for any pancreatic fistula previously reported in this cohort, body mass index and tumor functionality, were evaluated as independent risk factors for a clinically significant postoperative pancreatic fistula. There was no
The American Journal of Surgery, Vol -, No -, - 2015
4 Table 1
Clinical characteristics and 90 day outcomes of study cohort
Variable
No or clinically insignificant fistula (n 5 93)
Clinically significant fistula (n 5 26)
Total (n 5 119)
Age (years) Gender, n (%) Male Female Albumin (g/dL) BMI category, n (%) Normal % 24.9 Overweight 25.0–29.0 Obese R 30.0 Genetic background, n (%) Sporadic Inherited Functionality, n (%) Nonfunctional Functional Surgery, n (%) Enucleation Resection Approach, n (%) Laparoscopic Open Number of lesions, n (%) One Two Three or more Length of stay (days) Morbidities, n (%) Drain placement Reoperation Sepsis Death
42 6 1.5
43.5 6 3.5
42.3 6 1.36
35 (37.6) 58 (62.4) 3.86 6 .04
14 (53.8) 12 (46.2) 4.1 6 .08
49 (41.2) 70 (58.8) 3.9 6 .04
26 (29.5) 28 (31.8) 34 (38.6)
7 (28.0) 6 (24.0) 12 (48.0)
33 (29.2) 34 (30.1) 46 (40.7)
42 (45.2) 51 (54.8)
14 (53.8) 12 (46.2)
56 (47.1) 63 (52.9)
37 (39.8) 56 (60.2)
8 (30.8) 18 (69.2)
45 (37.8) 74 (62.2)
45 (48.4) 48 (51.6)
15 (57.7) 11 (42.3)
60 (50.4) 59 (49.6)
19 (20.4) 74 (79.6)
3 (11.5) 23 (88.5)
22 (18.5) 97 (81.5)
72 (80) 7 (7.8) 11 (12.2) 11.7 6 0.6
21 (80.8) 2 (7.7) 3 (11.5) 23.4 6 2.57
93 (80.2) 9 (7.8) 14 (12.1) 14.23 6 .85
21 5 1 0
21 7 2 1
0 2 1 1
(0) (2.2) (1.1) (1.1)
(80.8) (19.2) (3.8) (0)
(17.7) (5.9) (1.7) (.8)
BMI 5 body mass index.
significant relationship between these factors and a clinically significant fistula. We then compared pancreatic density among patients with no fistula (grade 0), clinically insignificant fistula (either grade A or B not requiring an intervention), and patients with a clinically significant fistula (either grade B requiring an intervention or grade C). The mean pancreatic density for patients with no fistula was 92.17 6 3.03 (n 5 44), with clinically insignificant fistula was 77.06 6 4.32 (n 5 44), and with clinically significant fistula was 73.96 6 6.01 (n 5 25; P 5 .02; Fig. 3A). This observation was again significant for resection (P , .01; Fig. 3B), and not for enucleation (P 5 .69; Fig. 3C).
Comments This study demonstrated an association between increasing postoperative pancreatic fistula grade and decreasing preoperative pancreatic density in patients undergoing surgery for pancreatic neuroendocrine tumors. This association was significant for patients undergoing resection but not for patients undergoing an enucleation. There was also a significant difference in pancreas density
between patients with no fistula, patients with insignificant fistulas, and patients with a clinically significant fistula for patients undergoing resection. Although there was no threshold for pancreas density that could predict by itself if a clinically significant fistula would occur, preoperative assessment of pancreatic density may help to identify highrisk patients, guide informed consent, risk stratification, and subsequent patient selection. In enucleations, the most important risk factor for pancreatic fistula has been identified to be enucleation of a tumor less than 2 mm from the main pancreatic duct.14 Large differences in the techniques of formal resections and enucleations may explain why enucleations did not follow the same relationship of increasing grade of postoperative pancreatic fistula with decreasing pancreatic density observed in resections. In a formal pancreaticoduodenectomy, calcified, denser pancreatic tissue that surrounds the anastomosis provides a natural barrier to prevent leakage of pancreatic fluid, but a fatty ‘‘soft’’ pancreas is more penetrable. If an enucleation leaves an unappreciated large duct branch exposed or has breached the main pancreatic duct, there is no pancreas tissue surrounding the duct to prevent a leak. Since the relevant portion of pancreatic
Y. Assadipour et al.
Pancreas density predicts PNET fistula
5
Figure 2 (A) Postoperative pancreatic fistula grade and pancreatic density for all patients. NA, not applicable (no fistula), n 5 44. A, n 5 26. B, n 5 39. C, n 5 4. P 5 .0038. (B) For resection: NA, not applicable (no fistula), n 5 24. A, n 5 11. B, n 5 18. C, n 5 1. P 5 .0008. (C) For enucleation: NA, not applicable (no fistula), n 5 20. A, n 5 15. B, n 5 21. C, n 5 3; P 5 .34.
tissue was enucleated, it cannot play a role in containing pancreatic output. Although many techniques and sealing agents have been proposed to optimize the integrity of the pancreaticojejunal anastomosis, the most common cause of anastomotic leak after a pancreaticoduodenectomy, none have demonstrated clear superiority.15–23 Prophylactic drain placement has also not been demonstrated to decrease the frequency or severity of fistula nor the future requirement of interventions.15 Although pharmacologic agents such as octreotide have not shown a clear benefit, recently pasireotide, a long-acting somatostatin paralog with broader binding profile, has been shown to reduce incidence of severe fistula after pancreas surgery.16,20,23 Pancreatic density may be a useful selection tool for placing patients on pasireotide prophylactically. In addition, identification of meaningful prophylactic interventions will be the most evident among patients that are at high risk for postoperative pancreatic fistula. Study of patients with low-pancreatic density and high risk for postoperative pancreatic fistula could be incorporated to future clinical trial design. By studying interventions on
patients with low-pancreatic density, new means of reducing fistula incidence can be developed. We acknowledge the limitations of this study. This review is retrospective in nature; further evaluation of the role of pancreatic density in predicting postoperative pancreatic fistula would be strengthened by a prospective study. Also, though this is one of the largest studies reviewing enucleation and resection for pancreatic neuroendocrine tumors, the analysis would benefit from more patients and independent validation of the results. In addition, as there is variation in computed tomography scan protocols among institutions, differences in contrast bolus and timing of the portal venous phase may result in different density values. Therefore, rather than considering the absolute value of pancreatic density, relative densities among patients imaged at a particular institution using a given protocol may be more relevant. This study is unique in that to our knowledge, this is the only study to examine the association of computed tomography preoperative pancreatic density and postoperative pancreatic fistula for pancreatic neuroendocrine tumor pathology.
6
The American Journal of Surgery, Vol -, No -, - 2015
Figure 3 (A) Postoperative pancreatic clinical fistula and pancreatic density for all patients. No fistula, n 5 44. Clinically insignificant fistula: grade A and grade B not requiring intervention, n 5 44. Clinically significant fistula: grade B requiring intervention or grade C, n 5 25. P 5 .02. (B) For resection: no fistula, n 5 24. Clinically insignificant fistula: grade A and grade B not requiring intervention, n 5 20. Clinically significant fistula: grade B requiring intervention or grade C, n 5 10. P 5 .0023. (C) For enucleation: no fistula, n 5 20. Clinically insignificant fistula: grade A and grade B not requiring intervention, n 5 24. Clinically significant fistula: grade B requiring intervention or grade C, n 5 15; P 5 .69.
Conclusions This study demonstrates a significant association between increasing postoperative pancreatic fistula grade and decreasing pancreatic density. Pancreas density is not able to independently predict if a clinically significant fistula would occur but can provide supportive information regarding overall risk of developing a fistula. This preoperative knowledge of increased fistula risk in pancreatic surgery for neuroendocrine tumors may better guide informed consent, operative planning, and management. Pancreatic density would be a useful component of patient selection when designing trials aimed at prevention or treatment of pancreatic fistulas. Including high-risk patients in these trials will improve the ability to identify an efficacious intervention and improve patient outcomes.
References 1. Yeo CJ, Cameron JL, Sohn TA, et al. Six hundred fifty consecutive pancreaticoduodenectomies in the 1990s: pathology, complications, and outcomes. Ann Surg 1997;226:248–57. 2. Kazanjian KK, Hines OJ, Eibl G, et al. Management of pancreatic fistulas after pancreaticoduodenectomy: results in 437 consecutive patients. Arch Surg 2005;140:849–54. 3. van Berge Henegouwen MI, De Wit LT, Van Gulik TM, et al. Incidence, risk factors, and treatment of pancreatic leakage after pancreaticoduodenectomy: drainage versus resection of the pancreatic remnant. J Am Coll Surg 1997;185:18–24. 4. Fuks D, Piessen G, Huet E, et al. Life-threatening postoperative pancreatic fistula (grade C) after pancreaticoduodenectomy: incidence, prognosis, and risk factors. Am J Surg 2009;197:702–9. 5. DeOliveira ML, Winter JM, Schafer M, et al. Assessment of complications after pancreatic surgery: a novel grading system applied to 633 patients undergoing pancreaticoduodenectomy. Ann Surg 2006; 244:931–7.
Y. Assadipour et al.
Pancreas density predicts PNET fistula
6. House MG, Fong Y, Arnaoutakis DJ, et al. Preoperative predictors for complications after pancreaticoduodenectomy: impact of BMI and body fat distribution. J Gastrointest Surg 2008;12:270–8. 7. Inchauste SM, Lanier BJ, Libutti SK, et al. Rate of clinically significant postoperative pancreatic fistula in pancreatic neuroendocrine tumors. World J Surg 2012;36:1517–26. 8. Lin JW, Cameron JL, Yeo CJ, et al. Risk factors and outcomes in postpancreaticoduodenectomy pancreaticocutaneous fistula. J Gastrointest Surg 2004;8:951–9. 9. Callery MP, Pratt WB, Kent TS, et al. A prospectively validated clinical risk score accurately predicts pancreatic fistula after pancreatoduodenectomy. J Am Coll Surg 2013;216:1–14. 10. Bassi C, Dervenis C, Butturini G, et al. Postoperative pancreatic fistula: an international study group (ISGPF) definition. Surgery 2005; 138:8–13. 11. Tan WJ, Kow AW, Liau KH. Moving towards the New International Study Group for Pancreatic Surgery (ISGPS) definitions in pancreaticoduodenectomy: a comparison between the old and new. HPB (Oxford) 2011;13:566–72. 12. Park CM, Park JS, Cho ES, et al. The effect of visceral fat mass on pancreatic fistula after pancreaticoduodenectomy. J Invest Surg 2012;25:169–73. 13. Tranchart H, Gaujoux S, Rebours V, et al. Preoperative CT scan helps to predict the occurrence of severe pancreatic fistula after pancreaticoduodenectomy. Ann Surg 2012;256:139–45. 14. Brient C, Regenet N, Sulpice L, et al. Risk factors for postoperative pancreatic fistulization subsequent to enucleation. J Gastrointest Surg 2012;16:1883–7.
7 15. Adham M, Chopin-Laly X, Lepilliez V, et al. Pancreatic resection: drain or no drain? Surgery 2013;154:1069–77. 16. Allen PJ, Gonen M, Brennan MF, et al. Pasireotide for postoperative pancreatic fistula. N Engl J Med 2014;370:2014–22. 17. Hashimoto D, Hirota M, Chikamoto A, et al. Short-term outcome of new end-to-side inserting pancreaticojejunostomy without stitches on the pancreatic cut end. Hepatogastroenterology 2014; 61:489–92. 18. Katsaragakis S, Larentzakis A, Panousopoulos SG, et al. A new pancreaticojejunostomy technique: a battle against postoperative pancreatic fistula. World J Gastroenterol 2013;19:4351–5. 19. Lillemoe KD, Cameron JL, Kim MP, et al. Does fibrin glue sealant decrease the rate of pancreatic fistula after pancreaticoduodenectomy? Results of a prospective randomized trial. J Gastrointest Surg 2004;8: 766–72. 20. McMillan MT, Christein JD, Callery MP, et al. Prophylactic octreotide for pancreatoduodenectomy: more harm than good? HPB 2014;16: 954–62. 21. Orci LA, Oldani G, Berney T, et al. Systematic review and metaanalysis of fibrin sealants for patients undergoing pancreatic resection. HPB (Oxford) 2014;16:3–11. 22. Tewari M, Hazrah P, Kumar V, et al. Options of restorative pancreaticoenteric anastomosis following pancreaticoduodenectomy: a review. Surg Oncol 2010;19:17–26. 23. Vanounou T, Pratt WB, Callery MP, et al. Selective administration of prophylactic octreotide during pancreaticoduodenectomy: a clinical and cost-benefit analysis in low- and high-risk glands. J Am Coll Surg 2007;205:546–57.
Significance of preoperative radiographic pancreatic density in predicting pancreatic fistula after surgery for pancreatic neuroendocrine tumors Yasmine Assadipour, M.D.a, Saı¨d C. Azoury, M.D.a,b, Nicholas N. Schaub, M.D.a, Young Hong, M.D.a, Robert Eil, M.D.a, Suzanne M. Inchauste, M.D.a, Seth M. Steinberg, Ph.D.c, Aradhana M. Venkatesan, M.D.d, Steven K. Libutti, M.D., F.A.C.S.e, Marybeth S. Hughes, M.D., F.A.C.S.a,b,* a
Surgery Branch, Center for Cancer Research, National Cancer Institute, Bethesda, MD; bThoracic and Gastrointestinal Oncology Branch, Center for Cancer Research, National Cancer Institute, Bethesda, MD; cBiostatistics and Data Management Section, Center for Cancer Research, National Cancer Institute, Bethesda, MD; dRadiology and Imaging Sciences NIH Clinical Center, National Institutes of Health, Bethesda, MD; eDepartment of Surgery, Albert Einstein College of Medicine, Montefiore Medical Center, Bronx, NY
KEYWORDS: Pancreatic fistula; Neuroendocrine tumors; Pancreaticoduo denectomy; Pancreatic resection; Pancreatic enucleation
Abstract BACKGROUND: Postoperative pancreatic fistula remains the most severe and worrisome complication after surgery. Predictive preoperative assessment remains challenging. The authors examine the role of pancreatic computed tomography density in predicting postoperative pancreatic fistula after surgery for pancreatic neuroendocrine tumors. METHODS: A single institutional retrospective review of pancreatic surgery for neuroendocrine tumors between 1998 and 2010 was conducted. Preoperative contrast-enhanced computed tomography scans were reviewed, with mean region of interest measurements of pancreatic parenchymal density obtained from 10-mm thick axial computed tomography images. RESULTS: A total of 119 patients were identified: 59 with enucleations and 60 with resections. Decreased preoperative pancreatic density was significantly associated with an increased grade of postoperative pancreatic fistula (P , .01). Subgroup analyses revealed that decreased gland density was associated with increased grade of postoperative pancreatic fistula in the resection (P , .01) but not in the enucleation group (P 5 .34).
The authors have no conflicts of interest to disclose. All funding is provided by the Intramural Program of the National Cancer Institute; this did not create a conflict of interest. This study was approved by the appropriate institutional review board and has been performed in accordance with the ethical standards as laid down in the 1964 Declaration of Helsinki and its later amendments or comparable ethical standards. This article does not contain any studies with human participants or animals performed by any of the authors. * Corresponding author. Tel.: 301-594-9341; fax: 301-451-6933. E-mail address: [email protected] Manuscript received January 22, 2015; revised manuscript April 30, 2015
0002-9610/$ - see front matter Published by Elsevier Inc. http://dx.doi.org/10.1016/j.amjsurg.2015.07.031
The American Journal of Surgery, Vol -, No -, - 2015
2
CONCLUSIONS: A significant association between postoperative pancreatic fistula grade and preoperative pancreatic computed tomography density is observed in patients undergoing resection for pancreatic neuroendocrine tumors. Published by Elsevier Inc.
Pancreatic fistula remains a significant complication after surgery, and predictive preoperative assessment remains challenging.1 Pancreatic fistula occurrence after tumor resection adds significant morbidity to a patient’s postoperative course and can be life-threatening. Patients who experience a pancreatic leak are at greatly increased risk of other morbidities, including sepsis, biliary strictures, and mortality.2 Although fistulas have an overall mortality of 5% to 8%, and can generally be managed conservatively with prolonged nil per os (npo) status and parenteral nutrition, mortality for patients with a high-grade fistula can be as high as 28% to 40%.3,4 Patients with higher grade fistulas will require procedural interventions varying from percutaneous drain placement to reoperation. Many factors have been implicated in a higher incidence of postoperative pancreatic fistula, including pancreatic neuroendocrine tumors, hereditary syndromes, elevated body mass index, and a soft pancreas.5–9 A consensus definition of fistula grades developed by the New International Study Group for Pancreatic Surgery has allowed for objective analysis and comparison of patient outcomes across institutions.10 It has been independently validated and found consistently to correlate with clinical outcomes.11 Use of imaging to predict fistula rate has been examined in patients undergoing pancreaticoduodenectomy for pancreatic adenocarcinoma. Studies of preoperative computed tomography scans determined that visceral fat area was more accurate than body mass index in predicting pancreatic fistula.12,13 Pancreatic density and fistula rate or severity has not been evaluated in patients with neuroendocrine tumors, nor for other resection options such as distal pancreatectomy or enucleation. Herein, the authors aim to examine the role of pancreatic computed tomography density in predicting postoperative pancreatic fistula after surgery for pancreatic neuroendocrine tumors.
Methods After approval by the Office of Human Subject Research at the National Institutes of Health, a retrospective review was conducted on all patients who underwent resection or enucleation for pancreatic neuroendocrine tumors from March 1998 to April 2010. All participants provided informed written consent and were enrolled on protocols at the National Cancer Institute. Patient variables including age, gender, body mass index, hereditary syndrome type, tumor characteristics, and preoperative serum albumin level were reviewed. Body mass index classification (kg/m2) was
noted as normal % 24.9, overweight 25 to 29.9, and obese R 30.0. Operative reports were reviewed for surgical approach and type of resection. After resection or enucleation, closed-suction drains were placed at the surgical bed. Perioperative octreotide as prophylaxis was not used in this series. Drains were removed once patients were tolerating a regular diet and maintaining a normal drain amylase level with output of less than 100 mL/day. Drain amylase levels were obtained for all patients postoperatively. From strict International Study Group for Pancreatic Surgery criteria, postoperative pancreatic fistula were categorized as follows: grade A, asymptomatic and clinically silent fistula, only demonstrated by drain amylase level greater than 3 times the normal serum level; grade B, a fistula requiring a postoperative drain for greater than 21 days, nonoperative intervention such as total parenteral nutrition, octreotide, percutaneous drainage, or readmission within 30 days of hospitalization; and grade C, a severe fistula requiring surgical intervention, intensive care unit care greater than 48 hour, or death within 30 days of operation. Clinically significant postoperative pancreatic fistula was defined as any grade B fistula requiring an intervention and all grade C fistulas. Intervention was defined as requiring an intra-abdominal drain placement or a reoperation. Additional drains were placed when clinically indicated due to fever, leukocytosis, or increased pain necessitating a computed tomography scan showing an inadequately drained collection. Patient records were reviewed to determine the length of stay, clinical course, and complications. Preoperative contrast-enhanced triple phase computed tomography scans were reviewed. Mean region of interest measurements of pancreatic parenchymal density in Hounsfield units were obtained every 10 mm in axial computed tomography images acquired during the portal venous phase. Measurements were determined by blind evaluators and verified independently by a radiologist who was also blind to the clinical course of the subjects. The pancreas density was determined by calculating the mean of the region of interest measurements for each pancreas (Fig. 1). Continuous parameters were compared between patients with and without clinically significant postoperative pancreatic fistula using an exact Wilcoxon rank sum test. The significance of the trends in continuous parameters across ordered categorical parameters such as increasing fistula grade were determined by a Jonckheere–Terpstra test for trend. All P values are 2 tailed and presented without any adjustment for multiple comparisons. P value less than .05 is considered statistically significant.
Y. Assadipour et al.
Pancreas density predicts PNET fistula
3
Figure 1 An example of the method by which region of interest measurements of density were obtained every 10 mm from the preoperative computed tomography scan of a patient in the axial orientation during the portal venous phase.
Results A total of 119 patients who underwent surgery for a pancreatic neuroendocrine tumor at the National Institutes of Health between 1998 and 2010 were identified. Of this group, 70 (59%) were female and 49 (41%) were male, with an average age of 42.3 6 1.36 years for the entire group. Sixty-three (53%) patients had a heritable genetic mutation predisposing them to neuroendocrine tumors, including 28 patients with multiple endocrine neoplasia type 1, 34 with von Hippel–Lindau syndrome, and 1 with neurofibromatosis. Forty-five tumors were functional, and 74 were nonfunctional. Sixty patients underwent enucleation, and 59 patients underwent resection. In the resection group, 34 patients underwent a distal pancreatectomy, 14 patients underwent a Whipple, and 11 underwent a dual Whipple and distal pancreatectomy for multiple lesions not amenable to enucleation. Ninety-seven (82%) patients underwent an open approach, and 22 (19%) underwent a laparoscopic resection. There were no statistically significant differences in patient characteristics between patients with and without a clinically significant pancreatic fistula (Table 1). The average length of stay for patients was 14 6 .85 days. There were 93 patients with no or clinically insignificant fistula (78%), and 26 patients with a clinically significant fistula (22%). Length of stay for patients with a clinically significant fistula (average 23.4 days) was significantly longer (P ,.01) when compared with clinically insignificant and/or no fistula (average 11.7 days). In the 26 patients with clinically significant fistulas, 21 (81%) had a drain placed, and 5 (19%) required a reoperation; 1 patient (4%) developed
sepsis, and there were no mortalities reported within 30 days of operation. In the 93 patients without clinically significant fistulas, 2 patients required reoperations. One patient required a gastrectomy for prolonged gastroparesis after pancreaticoduodenectomy; the second patient developed sepsis after an internal hernia and incarcerated small intestine, underwent reoperation but did not improve and died. The overall mortality for the cohort was .8%, with 1 death in 119 patients (Table 1). An increasing postoperative pancreatic fistula grade was found to be associated with decreasing pancreatic density, measured in Hounsfield units (Fig. 2). The mean pancreatic density for patients with no fistula was 92.17 6 3.03 (n 5 44), for grade A fistula was 81.43 6 5.45 (n 5 26), for grade B was 76 6 4.46 (n 5 39), and for grade C fistula was 36.1 6 10.28 (n 5 4; P , .01; Fig. 2A). In a subgroup analysis of resections and enucleations, this difference was statistically significant in the resection group, and a similar but insignificant trend was found for the enucleation group. The mean pancreatic density for patients who underwent resection with no fistula was 94.85 6 3.82 (n 5 24), for grade A fistula was 73.33 6 9.62 (n 5 11), for grade B was 65.96 6 6.96 (n 5 18), and for grade C fistula was 13 (n 5 1; P , .01; Fig. 2B). The mean pancreatic density for patients who underwent enucleation with no fistula was 88.94 6 4.84 (n 5 20), for grade A fistula was 88.08 6 6.06 (n 5 15), for grade B was 84.66 6 5.18 (n 5 21), and for grade C fistula was 44.48 6 9.34 (n 5 3; P 5 .34; Fig. 2C). Risk factors for any pancreatic fistula previously reported in this cohort, body mass index and tumor functionality, were evaluated as independent risk factors for a clinically significant postoperative pancreatic fistula. There was no
The American Journal of Surgery, Vol -, No -, - 2015
4 Table 1
Clinical characteristics and 90 day outcomes of study cohort
Variable
No or clinically insignificant fistula (n 5 93)
Clinically significant fistula (n 5 26)
Total (n 5 119)
Age (years) Gender, n (%) Male Female Albumin (g/dL) BMI category, n (%) Normal % 24.9 Overweight 25.0–29.0 Obese R 30.0 Genetic background, n (%) Sporadic Inherited Functionality, n (%) Nonfunctional Functional Surgery, n (%) Enucleation Resection Approach, n (%) Laparoscopic Open Number of lesions, n (%) One Two Three or more Length of stay (days) Morbidities, n (%) Drain placement Reoperation Sepsis Death
42 6 1.5
43.5 6 3.5
42.3 6 1.36
35 (37.6) 58 (62.4) 3.86 6 .04
14 (53.8) 12 (46.2) 4.1 6 .08
49 (41.2) 70 (58.8) 3.9 6 .04
26 (29.5) 28 (31.8) 34 (38.6)
7 (28.0) 6 (24.0) 12 (48.0)
33 (29.2) 34 (30.1) 46 (40.7)
42 (45.2) 51 (54.8)
14 (53.8) 12 (46.2)
56 (47.1) 63 (52.9)
37 (39.8) 56 (60.2)
8 (30.8) 18 (69.2)
45 (37.8) 74 (62.2)
45 (48.4) 48 (51.6)
15 (57.7) 11 (42.3)
60 (50.4) 59 (49.6)
19 (20.4) 74 (79.6)
3 (11.5) 23 (88.5)
22 (18.5) 97 (81.5)
72 (80) 7 (7.8) 11 (12.2) 11.7 6 0.6
21 (80.8) 2 (7.7) 3 (11.5) 23.4 6 2.57
93 (80.2) 9 (7.8) 14 (12.1) 14.23 6 .85
21 5 1 0
21 7 2 1
0 2 1 1
(0) (2.2) (1.1) (1.1)
(80.8) (19.2) (3.8) (0)
(17.7) (5.9) (1.7) (.8)
BMI 5 body mass index.
significant relationship between these factors and a clinically significant fistula. We then compared pancreatic density among patients with no fistula (grade 0), clinically insignificant fistula (either grade A or B not requiring an intervention), and patients with a clinically significant fistula (either grade B requiring an intervention or grade C). The mean pancreatic density for patients with no fistula was 92.17 6 3.03 (n 5 44), with clinically insignificant fistula was 77.06 6 4.32 (n 5 44), and with clinically significant fistula was 73.96 6 6.01 (n 5 25; P 5 .02; Fig. 3A). This observation was again significant for resection (P , .01; Fig. 3B), and not for enucleation (P 5 .69; Fig. 3C).
Comments This study demonstrated an association between increasing postoperative pancreatic fistula grade and decreasing preoperative pancreatic density in patients undergoing surgery for pancreatic neuroendocrine tumors. This association was significant for patients undergoing resection but not for patients undergoing an enucleation. There was also a significant difference in pancreas density
between patients with no fistula, patients with insignificant fistulas, and patients with a clinically significant fistula for patients undergoing resection. Although there was no threshold for pancreas density that could predict by itself if a clinically significant fistula would occur, preoperative assessment of pancreatic density may help to identify highrisk patients, guide informed consent, risk stratification, and subsequent patient selection. In enucleations, the most important risk factor for pancreatic fistula has been identified to be enucleation of a tumor less than 2 mm from the main pancreatic duct.14 Large differences in the techniques of formal resections and enucleations may explain why enucleations did not follow the same relationship of increasing grade of postoperative pancreatic fistula with decreasing pancreatic density observed in resections. In a formal pancreaticoduodenectomy, calcified, denser pancreatic tissue that surrounds the anastomosis provides a natural barrier to prevent leakage of pancreatic fluid, but a fatty ‘‘soft’’ pancreas is more penetrable. If an enucleation leaves an unappreciated large duct branch exposed or has breached the main pancreatic duct, there is no pancreas tissue surrounding the duct to prevent a leak. Since the relevant portion of pancreatic
Y. Assadipour et al.
Pancreas density predicts PNET fistula
5
Figure 2 (A) Postoperative pancreatic fistula grade and pancreatic density for all patients. NA, not applicable (no fistula), n 5 44. A, n 5 26. B, n 5 39. C, n 5 4. P 5 .0038. (B) For resection: NA, not applicable (no fistula), n 5 24. A, n 5 11. B, n 5 18. C, n 5 1. P 5 .0008. (C) For enucleation: NA, not applicable (no fistula), n 5 20. A, n 5 15. B, n 5 21. C, n 5 3; P 5 .34.
tissue was enucleated, it cannot play a role in containing pancreatic output. Although many techniques and sealing agents have been proposed to optimize the integrity of the pancreaticojejunal anastomosis, the most common cause of anastomotic leak after a pancreaticoduodenectomy, none have demonstrated clear superiority.15–23 Prophylactic drain placement has also not been demonstrated to decrease the frequency or severity of fistula nor the future requirement of interventions.15 Although pharmacologic agents such as octreotide have not shown a clear benefit, recently pasireotide, a long-acting somatostatin paralog with broader binding profile, has been shown to reduce incidence of severe fistula after pancreas surgery.16,20,23 Pancreatic density may be a useful selection tool for placing patients on pasireotide prophylactically. In addition, identification of meaningful prophylactic interventions will be the most evident among patients that are at high risk for postoperative pancreatic fistula. Study of patients with low-pancreatic density and high risk for postoperative pancreatic fistula could be incorporated to future clinical trial design. By studying interventions on
patients with low-pancreatic density, new means of reducing fistula incidence can be developed. We acknowledge the limitations of this study. This review is retrospective in nature; further evaluation of the role of pancreatic density in predicting postoperative pancreatic fistula would be strengthened by a prospective study. Also, though this is one of the largest studies reviewing enucleation and resection for pancreatic neuroendocrine tumors, the analysis would benefit from more patients and independent validation of the results. In addition, as there is variation in computed tomography scan protocols among institutions, differences in contrast bolus and timing of the portal venous phase may result in different density values. Therefore, rather than considering the absolute value of pancreatic density, relative densities among patients imaged at a particular institution using a given protocol may be more relevant. This study is unique in that to our knowledge, this is the only study to examine the association of computed tomography preoperative pancreatic density and postoperative pancreatic fistula for pancreatic neuroendocrine tumor pathology.
6
The American Journal of Surgery, Vol -, No -, - 2015
Figure 3 (A) Postoperative pancreatic clinical fistula and pancreatic density for all patients. No fistula, n 5 44. Clinically insignificant fistula: grade A and grade B not requiring intervention, n 5 44. Clinically significant fistula: grade B requiring intervention or grade C, n 5 25. P 5 .02. (B) For resection: no fistula, n 5 24. Clinically insignificant fistula: grade A and grade B not requiring intervention, n 5 20. Clinically significant fistula: grade B requiring intervention or grade C, n 5 10. P 5 .0023. (C) For enucleation: no fistula, n 5 20. Clinically insignificant fistula: grade A and grade B not requiring intervention, n 5 24. Clinically significant fistula: grade B requiring intervention or grade C, n 5 15; P 5 .69.
Conclusions This study demonstrates a significant association between increasing postoperative pancreatic fistula grade and decreasing pancreatic density. Pancreas density is not able to independently predict if a clinically significant fistula would occur but can provide supportive information regarding overall risk of developing a fistula. This preoperative knowledge of increased fistula risk in pancreatic surgery for neuroendocrine tumors may better guide informed consent, operative planning, and management. Pancreatic density would be a useful component of patient selection when designing trials aimed at prevention or treatment of pancreatic fistulas. Including high-risk patients in these trials will improve the ability to identify an efficacious intervention and improve patient outcomes.
References 1. Yeo CJ, Cameron JL, Sohn TA, et al. Six hundred fifty consecutive pancreaticoduodenectomies in the 1990s: pathology, complications, and outcomes. Ann Surg 1997;226:248–57. 2. Kazanjian KK, Hines OJ, Eibl G, et al. Management of pancreatic fistulas after pancreaticoduodenectomy: results in 437 consecutive patients. Arch Surg 2005;140:849–54. 3. van Berge Henegouwen MI, De Wit LT, Van Gulik TM, et al. Incidence, risk factors, and treatment of pancreatic leakage after pancreaticoduodenectomy: drainage versus resection of the pancreatic remnant. J Am Coll Surg 1997;185:18–24. 4. Fuks D, Piessen G, Huet E, et al. Life-threatening postoperative pancreatic fistula (grade C) after pancreaticoduodenectomy: incidence, prognosis, and risk factors. Am J Surg 2009;197:702–9. 5. DeOliveira ML, Winter JM, Schafer M, et al. Assessment of complications after pancreatic surgery: a novel grading system applied to 633 patients undergoing pancreaticoduodenectomy. Ann Surg 2006; 244:931–7.
Y. Assadipour et al.
Pancreas density predicts PNET fistula
6. House MG, Fong Y, Arnaoutakis DJ, et al. Preoperative predictors for complications after pancreaticoduodenectomy: impact of BMI and body fat distribution. J Gastrointest Surg 2008;12:270–8. 7. Inchauste SM, Lanier BJ, Libutti SK, et al. Rate of clinically significant postoperative pancreatic fistula in pancreatic neuroendocrine tumors. World J Surg 2012;36:1517–26. 8. Lin JW, Cameron JL, Yeo CJ, et al. Risk factors and outcomes in postpancreaticoduodenectomy pancreaticocutaneous fistula. J Gastrointest Surg 2004;8:951–9. 9. Callery MP, Pratt WB, Kent TS, et al. A prospectively validated clinical risk score accurately predicts pancreatic fistula after pancreatoduodenectomy. J Am Coll Surg 2013;216:1–14. 10. Bassi C, Dervenis C, Butturini G, et al. Postoperative pancreatic fistula: an international study group (ISGPF) definition. Surgery 2005; 138:8–13. 11. Tan WJ, Kow AW, Liau KH. Moving towards the New International Study Group for Pancreatic Surgery (ISGPS) definitions in pancreaticoduodenectomy: a comparison between the old and new. HPB (Oxford) 2011;13:566–72. 12. Park CM, Park JS, Cho ES, et al. The effect of visceral fat mass on pancreatic fistula after pancreaticoduodenectomy. J Invest Surg 2012;25:169–73. 13. Tranchart H, Gaujoux S, Rebours V, et al. Preoperative CT scan helps to predict the occurrence of severe pancreatic fistula after pancreaticoduodenectomy. Ann Surg 2012;256:139–45. 14. Brient C, Regenet N, Sulpice L, et al. Risk factors for postoperative pancreatic fistulization subsequent to enucleation. J Gastrointest Surg 2012;16:1883–7.
7 15. Adham M, Chopin-Laly X, Lepilliez V, et al. Pancreatic resection: drain or no drain? Surgery 2013;154:1069–77. 16. Allen PJ, Gonen M, Brennan MF, et al. Pasireotide for postoperative pancreatic fistula. N Engl J Med 2014;370:2014–22. 17. Hashimoto D, Hirota M, Chikamoto A, et al. Short-term outcome of new end-to-side inserting pancreaticojejunostomy without stitches on the pancreatic cut end. Hepatogastroenterology 2014; 61:489–92. 18. Katsaragakis S, Larentzakis A, Panousopoulos SG, et al. A new pancreaticojejunostomy technique: a battle against postoperative pancreatic fistula. World J Gastroenterol 2013;19:4351–5. 19. Lillemoe KD, Cameron JL, Kim MP, et al. Does fibrin glue sealant decrease the rate of pancreatic fistula after pancreaticoduodenectomy? Results of a prospective randomized trial. J Gastrointest Surg 2004;8: 766–72. 20. McMillan MT, Christein JD, Callery MP, et al. Prophylactic octreotide for pancreatoduodenectomy: more harm than good? HPB 2014;16: 954–62. 21. Orci LA, Oldani G, Berney T, et al. Systematic review and metaanalysis of fibrin sealants for patients undergoing pancreatic resection. HPB (Oxford) 2014;16:3–11. 22. Tewari M, Hazrah P, Kumar V, et al. Options of restorative pancreaticoenteric anastomosis following pancreaticoduodenectomy: a review. Surg Oncol 2010;19:17–26. 23. Vanounou T, Pratt WB, Callery MP, et al. Selective administration of prophylactic octreotide during pancreaticoduodenectomy: a clinical and cost-benefit analysis in low- and high-risk glands. J Am Coll Surg 2007;205:546–57.