Usefulness of the Liver Fibrosis Index (LFI) Measured during Endoscopic Ultrasonography in Evaluating the Pancreatic Parenchyma

Abstract

Endoscopic ultrasound has changed the evaluation of pancreatic diseases and has achieved a histopathological diagnosis (when associated with a puncture); however, this procedure requires training, is not free of complica-tions, and around 25 % of patients may have false negatives. Therefore, quantitative elastography with the strain ratio has been implemented to differentiate benign masses from malignant ones. There is growing but not yet con-clusive evidence, given the heterogeneity of the results (without consensus on its performance). It is necessary to develop other methods that allow for greater diagnostic certainty, such as the liver fibrosis index (LFI) measured by endoscopic ultrasonography. This method is based on artificial intelligence and validated for diagnosing and moni-toring liver fibrosis. Our group considers that it could also be used to assess the pancreatic parenchyma. Aim: To evaluate whether the LFI can differentiate three types of pancreatic tissues: normal pancreas, fatty pancreas, and pancreatic cancer. Materials and methods: Prospective cross-sectional single-center study. We included sixty-six patients over 18 years of age with an indication for endoscopic ultrasonography. Group 1 consisted of patients with an indication other than the biliopancreatic disease (55 patients). The endoscopic ultrasonography (EUS) fatty pancreas classification scale was applied to this group, taking the echogenicity of the spleen (previously validated) as a reference; this group was subdivided into normal pancreatic parenchyma and fatty pancreas. Group 2 (11 patients) included those examined for solid pancreatic lesions with a positive cytological diagnosis of pancreatic carcinoma. We used a Google Form as a data collection tool, available with a shortened address (shorturl.at/pIMWX). It was filled out before and after the procedure by Gastroenterology fellows, previously trained for this purpose. The LFI was measured in the pancreas in real-time using software supplied by the manufacturer (Hitachi Noblus) between January 2019 and January 2020. All patients underwent a complete biliopancreatic echoendoscopy, with a linear Pentax echoendoscope and Hitachi Noblus processor. Then, qualitative and quantitative elastography was per-formed, including LFI measurement. Results: We included a total of 66 patients: 11 with a diagnosis of pancreatic cancer confirmed by cytology and 55 sent for ultrasound endoscopy due to pathologies other than the biliopan-creatic disease. The age range was 23-89, with a mean of 56.75 years. The most frequent history was steatosis or steatohepatitis (n = 14) (25.45 %). The most frequent indication for performing the procedure was subepithelial lesion (n = 29) (52.73 %). The percentages of patients according to pancreatic echogenicity were Grade I (n = 29) (52.73 %); Grade II (n = 5) (9.09 %); Grade III (n = 18) (32.73 %); Grade IV (n = 3) (5.45 %). Grades I and II were taken as a normal pancreas and Grades III and IV as a fatty pancreas, divided into n = 34 patients (61.82 %) for a normal pancreas and n = 21 (38 %) for a fatty pancreas. According to the scale used, there is a fatty pancreas prevalence of 38.18 %. The LFI was measured in three subgroups: those considered endoscopically normal, those classified as fatty pancreas, and patients diagnosed with pancreatic cancer confirmed by cytology taken from the pancreas. The LFI for these groups were, respectively, normal pancreas: LFI 2.60, range 0.97-3.47 (95 % CI 2.17-3.02); fatty pancreas: LFI 3.87, range 2-5.5 (95 % CI 3.44-4.29); pancreatic cancer: LFI 6.35, range 5.8-7.8 (95 % CI 5.92-6.77). Conclusions: This is the first pilot study that applies the LFI to the pancreatic parenchyma. It is useful in differentiating a normal pancreas, a fatty pancreas, and pancreatic carcinoma non-invasively. This finding must be validated in larger and more heterogeneous populations.