Discrimination of pancreatic and gastrointestinal neuroendocrine tumors by MALDI Imaging Mass Spectrometry

Aim: Neuroendocrine tumors (NETs) are a heterogeneous group of diseases that can arise from neuroendocrine cells throughout the body, with gastrointestinal tract and pancreas being the most common primary sites. In the metastatic situation, primary site of the tumors is of special interest. We use MALDI Imaging Mass Spectrometry (IMS) to identify a molecular signature capable to discriminate pancreatic from gastrointestinal NETs. Methods: Formalin-fixed paraffin-embedded tissue microarrays sections including gastrointestinal and pancreatic NETs tissue core specimens were subjected to in-situ digestion with trypsin. Digested tissues were sprayed with a matrix solution and subsequently analyzed by a Matix-Assisted Laser Desorption-Ionization (MALDI) mass spectrometer instrument (Bruker Daltonik GmbH, Bremen, Germany). Data were acquired at 150µm spatial resolution, and visualized using FlexImaging software (Bruker Daltonik GmbH, Bremen, Germany). Following MALDI analysis, matrix was rinsed off with methanol, and hematoxylin and eosin (HE) staining was then performed on the same tissue for histological examination. Results: A set of 205 samples including 126 pancreatic NET (pNET) and 79 gastrointestinal NET (gasNET) randomized into either training or testing cohorts were analyzed in this study. HE stained tissues were examined by a pathologist who digitally marked adequate area of interest. The annotated regions were superimposed with MALDI data, and mass spectra profiles were exported and loaded into SCiLS Lab software (SCiLS, Bremen) for classification analysis using a linear discriminant analysis (LDA) algorithm. LDA model predicted pNET and gasNET correctly with 100% sensitivity and specificity in the training set, and 95% sensitivity and specificity in the testing set. Conclusions: We provided evidence that MALDI IMS technology can be used to rapidly distinguish pancreatic and gastrointestinal neuroendocrine tumors at the peptide level with high accuracy.