Diagnostic efficacy for predicting intraductal papillary mucinous neoplasms of the pancreas with high grade dysplasia or invasive carcinoma based on the surgery indications in different guidelines

Abstract

Objective: To evaluate the performance of the European Evidence-based Guidelines on Pancreatic Cystic Neoplasms (EEGPCN)(2018) and International Association of Pancreatology(IAP) Guideline(Version 2017) in predicting high grade dysplasia/invasive carcinoma-intraductal papillary mucinous neoplasm(HGD/INV-IPMN). Methods: A retrospective analysis of 363 patients,who underwent surgical resection in Changhai Hospital affiliated to Navy Medical University from January 2012 to December 2018 and were pathologically identified as (intraductal papillary mucinous neoplasm, IPMN),was performed. The patients,including 230 males and 133 females,aging (61.7±10.1) years(range:19 to 83 years). The proportion of HGD/INV-IPMN who met with the absolute indication(AI) of EEGPCN and high risk stigma(HRS) of IAP were compared. The binary Logistic regression analysis was used to find the independent risk factors of HGD/INV-IPMN.Eight combinations of risk factors derived from relative indication/worrisome feature or risk factors in this study,were made to evaluate the diagnostic efficacy. The area under curve(AUC) of receiver operating characteristics was used to evaluate the the cutoff value of risk factors(①CA19-9≥37 U/ml,②diameter of main pancreatic duct 5.0-9.9 mm,③enhancing mural nodule<5 mm,④(acute) pancreatiti,⑤acyst diameter ≥40 mm,⑤bcyst diameter ≥30 mm, ⑥thickened or enhancing cyst walls,⑦neutrophile granulocyte to lymphocyte ratio(NLR)≥2, ⑧cyst located in head, uncinate or neck,⑨carcinoembryonic antigen(CEA) ≥5 μg/L) number for predicting HGD/INV-IPMN.The accuracy,sensitivity,specificity,positive predictive value,negative predictive value,true positive,true negative,false positive,false negative,positive likelihood ratio,negative likelihood ratio,Youden index and F1 score were calculated. Results: Ninety-two patients(49.5%) of 186 ones who met AI and 85 patients(48.3%) of 176 ones who met HRS were respectively confirmed as HGD/INV-IPMN. In those patients who were not met AI,tumor location,thickened/enhancing cyst wall,CA19-9 elevated,NLR≥2 and CEA elevated were significantly (P<0.05) correlated with HGD/INV-IPMN. And tumor location(head/uncinate/neck vs. body/tail,OR=3.284,95%CI:1.268-8.503,P=0.014),thickened/enhancement cyst wall (with vs.without,OR=2.713,95%CI:1.177-6.252,P=0.019),CA19-9(≥37 U/L vs.<37 U/L, OR=5.086,95%CI:2.05-12.62,P<0.01) and NLR(≥2 vs.<2,OR=2.380,95%CI:1.043-5.434,P=0.039) were the independent risk factors of HGD/INV-IPMN. Patients with ≥4 risk factors of 9 in combination Ⅷ(①②③④⑤b⑥⑦⑧⑨) were diagnosed as HGD/INV-IPMN with the moderate accuracy(71.0%),moderate sensitivity (62.0%) and moderate specificity (73.0%). Patients with ≥4 risk factors of 9 in Combination Ⅶ(①②③④⑤a⑥⑦⑧⑨) were diagnosed as HGD/INV-IPMN with the highest specificity(83.0%) and patients with ≥3 risk factors of 8 in combination Ⅵ(①②③④⑤b⑥⑧⑨) were diagnosed as HGD/INV-IPMN with the highest sensitivity(74.0%). The AUC for diagnosis of HGD/INV-IPMN in combination Ⅵ,Ⅶ and Ⅷ were 0.72,0.75 and 0.75,respectively. Older patients and younger patients could respectively refer to combination Ⅶ and combination Ⅵ to improve the management of IPMN. Conclusions: Patients who meet AI of EEGPCN should undertake resection, otherwise the method we explored is recommended. The method of improvement for diagnosis of HGD/INV-IPMN is relatively applicable and efficient for decision-making of surgery, especially for younger patients with decreasing of missed diagnosis and elder patients with decreasing of misdiagnosis.