Artificial neural networks versus LASSO regression for the prediction of long-term survival after surgery for invasive IPMN of the pancreas

Prediction of long-term survival in patients with invasive intraductal papillary mucinous neoplasm (IPMN) of the pancreas may aid in patient assessment, risk stratification and personalization of treatment. This study aimed to investigate the predictive ability of artificial neural networks (ANN) and LASSO regression in terms of 5-year disease-specific survival. ANN work in a non-linear fashion, having a potential advantage in analysis of variables with complex correlations compared to regression models. LASSO is a type of regression analysis facilitating variable selection and regularization. A total of 440 patients undergoing surgical treatment for invasive IPMN of the pancreas registered in the Surveillance, Epidemiology and End Results (SEER) database between 2004 and 2016 were analyzed. The dataset was prior to analysis randomly split into a modelling and test set (7:3). The accuracy, precision and F1 score for predicting mortality were 0.82, 0.83 and 0.89, respectively for ANN with variable selection compared to 0.79, 0.85 and 0.87, respectively for the LASSO-model. ANN using all variables showed similar accuracy, precision and F1 score of 0.81, 0.85 and 0.88, respectively compared to a logistic regression analysis. McNemar´s test showed no statistical difference between the models. The models showed high and similar performance with regard to accuracy and precision for predicting 5-year survival status.

MiR-10a in Pancreatic Juice as a Biomarker for Invasive Intraductal Papillary Mucinous Neoplasm by miRNA Sequencing

New biomarkers are needed to further stratify the risk of malignancy in intraductal papillary mucinous neoplasm (IPMN). Although microRNAs (miRNAs) are expected to be stable biomarkers, they can vary owing to a lack of definite internal controls. To identify universal biomarkers for invasive IPMN, we performed miRNA sequencing using tumor-normal paired samples. A total of 19 resected tissues and 13 pancreatic juice samples from 32 IPMN patients were analyzed for miRNA expression by next-generation sequencing with a two-step normalization of miRNA sequence data. The miRNAs involved in IPMN associated with invasive carcinoma were identified from this tissue analysis and further verified with the pancreatic juice samples. From the tumor-normal paired tissue analysis of the expression levels of 2792 miRNAs, 20 upregulated and 17 downregulated miRNAs were identified. In IPMN associated with invasive carcinoma (INV), miR-10a-5p and miR-221-3p were upregulated and miR-148a-3p was downregulated when compared with noninvasive IPMN. When these findings were further validated with pancreatic juice samples, miR-10a-5p was found to be elevated in INV (p = 0.002). Therefore, three differentially expressed miRNAs were identified in tissues with INV, and the expression of miR-10a-5p was also elevated in pancreatic juice samples with INV. MiR-10a-5p is a promising additional biomarker for invasive IPMN.

RNF43 Mutations in IPMN Cases: A Potential Prognostic Factor

An intraductal papillary mucinous neoplasm (IPMN) is a common pancreatic precursor lesion, and it often harbors mutations in KRAS, GNAS, and RNF43. To clarify the molecular profiles of IPMNs, we conducted mutation analysis of KRAS, GNAS, and RNF43 in 61 IPMN formalin-fixed, paraffin-embedded (FFPE) specimens. The mutation rates of codons 12, 13, and 61 in KRAS and codon 201 in GNAS were detected by Sanger sequencing. Next-generation sequencing was performed on RNF43, and the results were further verified by Sanger sequencing. We identified KRAS and GNAS mutations in 35 (57%) and 40 (66%) IPMN cases, respectively. GNAS mutations were significantly correlated with the morphologic subtype (P<0.001) and were more prevalent in the intestinal subtype (93%) than in the gastric (55%) and pancreatobiliary subtypes (44%) but were absent in the oncocytic subtype. RNF43 mutations were found in 5 cases (8%), all of which occurred in high-grade dysplasia and invasive lesions (2/5 and 3/5). All 5 cases harboring RNF43 mutations also exhibited GNAS mutations. RNF43 mutations were associated with a worse prognosis in invasive IPMN patients (P=0.002), while KRAS and GNAS mutations did not affect the prognosis of patients.

Independent predictors of secondary invasive pancreatic remnant tumors after initial resection of an intraductal papillary mucinous neoplasm: a nationwide large-scale survey in Japan

Purpose There is no standardized surveillance protocol after intraductal papillary mucinous neoplasm (IPMN) resection. We report the findings of a large-scale survey in Japan, investigating the independent predictors of secondary invasive tumors by analyzing the epidemiology of secondary tumors of the remnant pancreas after initial IPMN resection. Methods An institutional questionnaire about the remnant pancreas after pancreas resection was distributed at the 41st Annual Meeting of the Japanese Society of Pancreatic Surgery in Tokyo. We retrospectively analyzed the patient data including pathological diagnosis, postoperative outcomes, and evaluation methods. Results Redo pancreatectomy was performed for secondary disease in 213 (1.4%) of a total 15,777 patients. Eighty-eight of these 213 patients had undergone initial resection of IPMN. The types of secondary tumors after IPMN resection significantly depended on those of the primary tumors. Through short-interval and long-term follow-up, most of the secondary tumors were detected within 1–4 years. Logistic regression analysis revealed that the initial pathological diagnosis of invasive IPMN was an independent predictor of secondary invasive tumors in the remnant pancreas. Conclusion Primary invasive IPMN proved to be a significant predictor of secondary invasive IPMN. Both short-interval and long-term follow-up may help to determine the prognosis of patients after IPMN resection.