Top

Published in:

07-08-2023 | Pancreatoduodenostomy | Pancreatic Tumors

A Machine Learning Approach to Predict Postoperative Pancreatic Fistula After Pancreaticoduodenectomy Using Only Preoperatively Known Data

Authors: Amir Ashraf Ganjouei, MD, MPH, Fernanda Romero-Hernandez, MD, Jaeyun Jane Wang, MD, Megan Casey, MS, Willow Frye, MS, Daniel Hoffman, MD, Kenzo Hirose, MD, Eric Nakakura, MD, Carlos Corvera, MD, Ajay V. Maker, MD, Kimberly S. Kirkwood, MD, MPH, Adnan Alseidi, MD, EdM, Mohamed A. Adam, MD

Published in: Annals of Surgical Oncology | Issue 12/2023

Abstract

Background

Clinically-relevant postoperative pancreatic fistula (CR-POPF) following pancreaticoduodenectomy (PD) is a major postoperative complication and the primary determinant of surgical outcomes. However, the majority of current risk calculators utilize intraoperative and postoperative variables, limiting their utility in the preoperative setting. Therefore, we aimed to develop a user-friendly risk calculator to predict CR-POPF following PD using state-of-the-art machine learning (ML) algorithms and only preoperatively known variables.

Methods

Adult patients undergoing elective PD for non-metastatic pancreatic cancer were identified from the ACS-NSQIP targeted pancreatectomy dataset (2014–2019). The primary endpoint was development of CR-POPF (grade B or C). Secondary endpoints included discharge to facility, 30-day mortality, and a composite of overall and significant complications. Four models (logistic regression, neural network, random forest, and XGBoost) were trained, validated and a user-friendly risk calculator was then developed.

Results

Of the 8666 patients who underwent elective PD, 13% (n = 1160) developed CR-POPF. XGBoost was the best performing model (AUC = 0.72), and the top five preoperative variables associated with CR-POPF were non-adenocarcinoma histology, lack of neoadjuvant chemotherapy, pancreatic duct size less than 3 mm, higher BMI, and higher preoperative serum creatinine. Model performance for 30-day mortality, discharge to a facility, and overall and significant complications ranged from AUC 0.62–0.78.

Conclusions

In this study, we developed and validated an ML model using only preoperatively known variables to predict CR-POPF following PD. The risk calculator can be used in the preoperative setting to inform clinical decision-making and patient counseling.

Available only for authorised users

McPhee JT, Hill JS, Whalen GF, et al. Perioperative mortality for pancreatectomy: a national perspective. Ann Surg. 2007;246(2):246–53. https://doi.org/10.1097/01.sla.0000259993.17350.3a.CrossRefPubMedPubMedCentral

Burkhart RA, Relles D, Pineda DM, et al. Defining treatment and outcomes of hepaticojejunostomy failure following pancreaticoduodenectomy. J Gastrointest Surg Off J Soc Surg Aliment Tract. 2013;17(3):451–60. https://doi.org/10.1007/s11605-012-2118-9.CrossRef

Gouma DJ, van Geenen RC, van Gulik TM, et al. Rates of complications and death after pancreaticoduodenectomy: risk factors and the impact of hospital volume. Ann Surg. 2000;232(6):786–95. https://doi.org/10.1097/00000658-200012000-00007.CrossRefPubMedPubMedCentral

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(6):702–9. https://doi.org/10.1016/j.amjsurg.2008.03.004.CrossRefPubMed

Bassi C, Dervenis C, Butturini G, et al. Postoperative pancreatic fistula: an international study group (ISGPF) definition. Surgery. 2005;138(1):8–13. https://doi.org/10.1016/j.surg.2005.05.001.CrossRefPubMed

Bassi C, Marchegiani G, Dervenis C, 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):584–91. https://doi.org/10.1016/j.surg.2016.11.014.CrossRefPubMed

Harnoss JC, Ulrich AB, Harnoss JM, Diener MK, Büchler MW, Welsch T. Use and results of consensus definitions in pancreatic surgery: a systematic review. Surgery. 2014;155(1):47–57. https://doi.org/10.1016/j.surg.2013.05.035.CrossRefPubMed

Gaujoux S, Cortes A, Couvelard A, et al. Fatty pancreas and increased body mass index are risk factors of pancreatic fistula after pancreaticoduodenectomy. Surgery. 2010;148(1):15–23. https://doi.org/10.1016/j.surg.2009.12.005.CrossRefPubMed

Lin JW, Cameron JL, Yeo CJ, Riall TS, Lillemoe KD. Risk factors and outcomes in postpancreaticoduodenectomy pancreaticocutaneous fistula. J Gastrointest Surg Off J Soc Surg Aliment Tract. 2004;8(8):951–9. https://doi.org/10.1016/j.gassur.2004.09.044.CrossRef

10.

Lermite E, Pessaux P, Brehant O, et al. Risk factors of pancreatic fistula and delayed gastric emptying after pancreaticoduodenectomy with pancreaticogastrostomy. J Am Coll Surg. 2007;204(4):588–96. https://doi.org/10.1016/j.jamcollsurg.2007.01.018.CrossRefPubMed

11.

Pratt WB, Callery MP, Vollmer CM. Risk prediction for development of pancreatic fistula using the ISGPF classification scheme. World J Surg. 2008;32(3):419–28. https://doi.org/10.1007/s00268-007-9388-5.CrossRefPubMed

12.

Callery MP, Pratt WB, Kent TS, Chaikof EL, Vollmer CM. A prospectively validated clinical risk score accurately predicts pancreatic fistula after pancreatoduodenectomy. J Am Coll Surg. 2013;216(1):1–14. https://doi.org/10.1016/j.jamcollsurg.2012.09.002.CrossRefPubMed

13.

Al Abbas AI, Borrebach JD, Pitt HA, et al. Development of a novel pancreatoduodenectomy-specific risk calculator: an analysis of 10,000 patients. J Gastrointest Surg Off J Soc Surg Aliment Tract. 2021;25(6):1503–11. https://doi.org/10.1007/s11605-020-04725-0.CrossRef

14.

Frozanpor F, Loizou L, Ansorge C, Lundell L, Albiin N, Segersvärd R. Correlation between preoperative imaging and intraoperative risk assessment in the prediction of postoperative pancreatic fistula following pancreatoduodenectomy. World J Surg. 2014;38(9):2422–9. https://doi.org/10.1007/s00268-014-2556-5.CrossRefPubMed

15.

Graham JA, Kayser R, Smirniotopoulos J, Nusbaum JD, Johnson LB. Probability prediction of a postoperative pancreatic fistula after a pancreaticoduodenectomy allows for more transparency with patients and can facilitate management of expectations. J Surg Oncol. 2013;108(2):137–8. https://doi.org/10.1002/jso.23362.CrossRefPubMed

16.

Kim Z, Kim MJ, Kim JH, et al. Prediction of post-operative pancreatic fistula in pancreaticoduodenectomy patients using pre-operative MRI: a pilot study. HPB. 2009;11(3):215–21. https://doi.org/10.1111/j.1477-2574.2009.00011.x.CrossRefPubMedPubMedCentral

17.

Kirks RC, Cochran A, Barnes TE, et al. Developing and validating a center-specific preoperative prediction calculator for risk of pancreaticoduodenectomy. Am J Surg. 2018;216(3):498–505. https://doi.org/10.1016/j.amjsurg.2018.02.019.CrossRefPubMed

18.

Kiyochi H, Matsukage S, Nakamura T, Ishida N, Takada Y, Kajiwara S. Pathologic assessment of pancreatic fibrosis for objective prediction of pancreatic fistula and management of prophylactic drain removal after pancreaticoduodenectomy. World J Surg. 2015;39(12):2967–74. https://doi.org/10.1007/s00268-015-3211-5.CrossRefPubMed

19.

Kosaka H, Kuroda N, Suzumura K, Asano Y, Okada T, Fujimoto J. Multivariate logistic regression analysis for prediction of clinically relevant pancreatic fistula in the early phase after pancreaticoduodenectomy. J Hepato-Biliary-Pancreat Sci. 2014;21(2):128–33. https://doi.org/10.1002/jhbp.11.CrossRef

20.

Lee TK, Kang CM, Park MS, et al. Prediction of postoperative pancreatic fistulas after pancreatectomy: assessment with acoustic radiation force impulse elastography. J Ultrasound Med Off J Am Inst Ultrasound Med. 2014;33(5):781–6. https://doi.org/10.7863/ultra.33.5.781.CrossRef

21.

Nassour I, AlMasri S, Hodges JC, Hughes SJ, Zureikat A, Paniccia A. Novel calculator to estimate the risk of clinically relevant postoperative pancreatic fistula following distal pancreatectomy. J Gastrointest Surg Off J Soc Surg Aliment Tract. 2022;26(7):1436–44. https://doi.org/10.1007/s11605-022-05275-3.CrossRef

22.

Roberts KJ, Hodson J, Mehrzad H, et al. A preoperative predictive score of pancreatic fistula following pancreatoduodenectomy. HPB. 2014;16(7):620–8. https://doi.org/10.1111/hpb.12186.CrossRefPubMed

23.

Cameron JL, He J. Two thousand consecutive pancreaticoduodenectomies. J Am Coll Surg. 2015;220(4):530–6. https://doi.org/10.1016/j.jamcollsurg.2014.12.031.CrossRefPubMed

24.

Ball CG, Pitt HA, Kilbane ME, Dixon E, Sutherland FR, Lillemoe KD. Peri-operative blood transfusion and operative time are quality indicators for pancreatoduodenectomy. HPB. 2010;12(7):465–71. https://doi.org/10.1111/j.1477-2574.2010.00209.x.CrossRefPubMedPubMedCentral

25.

Tolles J, Meurer WJ. Logistic regression: relating patient characteristics to outcomes. JAMA. 2016;316(5):533–4. https://doi.org/10.1001/jama.2016.7653.CrossRefPubMed

26.

Kamiński B, Jakubczyk M, Szufel P. A framework for sensitivity analysis of decision trees. Cent Eur J Oper Res. 2018;26(1):135–59. https://doi.org/10.1007/s10100-017-0479-6.CrossRefPubMed

27.

Ho TK. The random subspace method for constructing decision forests. IEEE Trans Pattern Anal Mach Intell. 1998;20(8):832–44. https://doi.org/10.1109/34.709601.CrossRef

28.

Chen T, Guestrin C. XGBoost: A Scalable Tree Boosting System. In: Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. ACM; 2016:785–794. https://doi.org/10.1145/2939672.2939785

29.

Lundberg S, Lee SI. A Unified Approach to Interpreting Model Predictions. Published online November 24, 2017. Accessed 21 Oct 2022. http://arxiv.org/abs/1705.07874

30.

Ellis RJ, Brock Hewitt D, Liu JB, et al. Preoperative risk evaluation for pancreatic fistula after pancreaticoduodenectomy. J Surg Oncol. 2019;119(8):1128–34. https://doi.org/10.1002/jso.25464.CrossRefPubMedPubMedCentral

31.

Krell RW, McNeil LR, Yanala UR, Are C, Reames BN. Neoadjuvant therapy for pancreatic ductal adenocarcinoma: propensity-matched analysis of postoperative complications using ACS-NSQIP. Ann Surg Oncol. 2021;28(7):3810–22. https://doi.org/10.1245/s10434-020-09460-z.CrossRefPubMed

32.

van Dongen JC, Wismans LV, Suurmeijer JA, et al. The effect of preoperative chemotherapy and chemoradiotherapy on pancreatic fistula and other surgical complications after pancreatic resection: a systematic review and meta-analysis of comparative studies. HPB. 2021;23(9):1321–31. https://doi.org/10.1016/j.hpb.2021.04.027.CrossRefPubMed

33.

Lapshyn H, Petruch N, Thomaschewski M, et al. A simple preoperative stratification tool predicting the risk of postoperative pancreatic fistula after pancreatoduodenectomy. Pancreatol Off J Int Assoc Pancreatol IAP Al. 2021;21(5):957–64. https://doi.org/10.1016/j.pan.2021.03.009.CrossRef

34.

Shi Y, Gao F, Qi Y, et al. Computed tomography-adjusted fistula risk score for predicting clinically relevant postoperative pancreatic fistula after pancreatoduodenectomy: Training and external validation of model upgrade. EBioMedicine. 2020;62:103096. https://doi.org/10.1016/j.ebiom.2020.103096.CrossRefPubMedPubMedCentral

35.

Kolbinger FR, Lambrecht J, Leger S, et al. The image-based preoperative fistula risk score (preFRS) predicts postoperative pancreatic fistula in patients undergoing pancreatic head resection. Sci Rep. 2022;12(1):4064. https://doi.org/10.1038/s41598-022-07970-2.CrossRefPubMedPubMedCentral

36.

House MG, Fong Y, Arnaoutakis DJ, et al. Preoperative predictors for complications after pancreaticoduodenectomy: impact of BMI and body fat distribution. J Gastrointest Surg Off J Soc Surg Aliment Tract. 2008;12(2):270–8. https://doi.org/10.1007/s11605-007-0421-7.CrossRef

37.

Pecorelli N, Carrara G, De Cobelli F, et al. Effect of sarcopenia and visceral obesity on mortality and pancreatic fistula following pancreatic cancer surgery. Br J Surg. 2016;103(4):434–42. https://doi.org/10.1002/bjs.10063.CrossRefPubMed

38.

Li B, Pu N, Chen Q, et al. Comprehensive diagnostic nomogram for predicting clinically relevant postoperative pancreatic fistula after pancreatoduodenectomy. Front Oncol. 2021;11:717087. https://doi.org/10.3389/fonc.2021.717087.CrossRefPubMedPubMedCentral

39.

Griffin BR, Bronsert M, Reece TB, et al. Creatinine elevations from baseline at the time of cardiac surgery are associated with postoperative complications. J Thorac Cardiovasc Surg. 2022;163(4):1378–87. https://doi.org/10.1016/j.jtcvs.2020.03.174.CrossRefPubMed

40.

Mungroop TH, van Rijssen LB, van Klaveren D, et al. Alternative fistula risk score for pancreatoduodenectomy (a-FRS): design and international external validation. Ann Surg. 2019;269(5):937–43. https://doi.org/10.1097/SLA.0000000000002620.CrossRefPubMed

41.

Bertsimas D, Dunn J, Velmahos GC, Kaafarani HMA. Surgical risk is not linear: derivation and validation of a novel, user-friendly, and machine-learning-based predictive optimal trees in emergency surgery risk (POTTER) calculator. Ann Surg. 2018;268(4):574–83. https://doi.org/10.1097/SLA.0000000000002956.CrossRefPubMed

42.

Senders JT, Staples PC, Karhade AV, et al. Machine learning and neurosurgical outcome prediction: a systematic review. World Neurosurg. 2018;109:476-486.e1. https://doi.org/10.1016/j.wneu.2017.09.149.CrossRefPubMed

Title: A Machine Learning Approach to Predict Postoperative Pancreatic Fistula After Pancreaticoduodenectomy Using Only Preoperatively Known Data
Authors: Amir Ashraf Ganjouei, MD, MPH
Fernanda Romero-Hernandez, MD
Jaeyun Jane Wang, MD
Megan Casey, MS
Willow Frye, MS
Daniel Hoffman, MD
Kenzo Hirose, MD
Eric Nakakura, MD
Carlos Corvera, MD
Ajay V. Maker, MD
Kimberly S. Kirkwood, MD, MPH
Adnan Alseidi, MD, EdM
Mohamed A. Adam, MD
Publication date: 07-08-2023
Publisher: Springer International Publishing
Keywords: Pancreatoduodenostomy
Pancreatoduodenostomy
Published in: Annals of Surgical Oncology / Issue 12/2023
Print ISSN: 1068-9265
Electronic ISSN: 1534-4681
DOI: https://doi.org/10.1245/s10434-023-14041-x

Keynote webinar | Spotlight on medication adherence

Springer Medicine

A Machine Learning Approach to Predict Postoperative Pancreatic Fistula After Pancreaticoduodenectomy Using Only Preoperatively Known Data

Abstract

Background

Methods

Results

Conclusions

Keynote webinar | Spotlight on medication adherence

Springer Medicine

Abstract

Background

Methods

Results

Conclusions

Please log in to get access to this content

Other articles of this Issue 12/2023

ASO Visual Abstract: Satisfaction and Well-Being After Contralateral Prophylactic Mastectomy Among BRCA Mutation Carriers and Noncarriers—A Longitudinal Analysis of BREAST-Q Domains

2022 PSOGI Consensus on HIPEC Regimens for Peritoneal Malignancies: Diffuse Malignant Peritoneal Mesothelioma

Does a Positive Appendiceal Resection Margin in Low-Grade Appendiceal Mucinous Neoplasms, Warrant Additional Surgery? Our Institution Experience and Literature Review

ASO Visual Abstract: Exploration of Malignant Characteristics in Neoadjuvant Chemotherapy-Resistant Rectal Cancer, Focusing on Extramural Lesions

ASO Author Reflections: What the COVID-19 Pandemic Could Teach—It’s All About Building Healthcare Systems Resilience

Identification of Pediatric Tumors Intraoperatively Using Indocyanine Green (ICG)