Differences in the proteomic pattern of colonic and pancreatic carcinoma in multi tissue assays by MALDI imaging mass spectrometry (IMS). Tumortypisierung von Kolon- und Pankreasgewebeproben durch MALDI Imaging-Massenspektrometrie

Aim: Adenocarcinomas represent the most common metastatic tumors of unknown primary site. Among these, liver metastases of colonic and pancreatic adenocarcinoma are of special importance. Classification of the metastatic tumor concerning tissue of origin is required. The goal of this study was to identify a proteomic signature in FFPE tissue to discriminate between colon and pancreatic carcinoma using MALDI IMS. Methods: A total of 85 needle core biopsies of colonic and 86 core biopsies of pancreatic carcinomas collected in multi tissue assays were on-tissue trypsin digested and subsequently analyzed by MALDI imaging mass spectrometry. Data collected from each tissue section were correlated with a serial hematoxylin and eosin-stained sections. A training set, including 39 colon and 32 pancreas primary carcinoma biopsies, was used by a support vector machine (SVM) algorithm to select a group of candidate peptide signals to discriminate carcinomas of the two different tissue origins. A classification model was generated and validated with two testing sets, one including 23 colonic and 35 pancreatic carcinomas, and one including 23 different colonic and 19 pancreatic carcinoma core biopsies. Statistical analysis was performed with ClinProTools 3.0 (Bruker Daltonik GmbH, Bremen). Results: From the training set, twenty-five ion peptides were selected (through the statistical criteria of a Wilcoxon/Kruskal-Wallis test) by the SVM classification model discerning colonic from pancreatic primary carcinoma. These discriminatory features correctly classified the first test set with a sensitivity of 88.8% and a specificity of 91.5%, and the second test set with a sensitivity of 83.9% and a specificity of 98.3%. Six m/z discriminant signatures (530.1, 543.2, 788.4, 836.3, 852.4, 1562.8) exhibited a statistical significance with a PWKW< 10-6 and a minimum 2 fold intensity difference threshold. Conclusion: A proteomic pattern to distinguish colonic from pancreatic carcinoma was established using MALDI IMS technology. We have demonstrated that a histological classification of colonic and pancreatic carcinoma is possible by applying MALD-Imaging on FFPE-tissues. This proof-of-principle study has shown that proteomic classification by MALDI represent a valuable approach for histopathological diagnostics without applying immunohistochemisty or further molecular techniques.