Background and Aims : Through an exploratory proteomic approach based on typical hepatocellular adenomas (HCAs), we previously identified a diagnostic biomarker for a distinctive subtype of HCA with high risk of bleeding, already validated on a multicenter cohort. We hypothesized that the whole protein expression deregulation profile could deliver much more informative data for tumor characterization. Therefore, we pursued our analysis with the characterization of HCA proteomic profiles, evaluating their correspondence with the established genotype/phenotype classification and assessing whether they could provide added diagnosis and prognosis values. Approach and Results : From a collection of 260 cases, we selected 52 typical cases of all different subgroups on which we built a reference HCA proteomics database. Combining laser microdissection and mass-spectrometry–based proteomic analysis, we compared the relative protein abundances between tumoral (T) and nontumoral (NT) liver tissues from each patient and we defined a specific proteomic profile of each of the HCA subgroups. Next, we built a matching algorithm comparing the proteomic profile extracted from a patient with our reference HCA database. Proteomic profiles allowed HCA classification and made diagnosis possible, even for complex cases with immunohistological or genomic analysis that did not lead to a formal conclusion. Despite a well-established pathomolecular classification, clinical practices have not substantially changed and the HCA management link to the assessment of the malignant transformation risk remains delicate for many surgeons. That is why we also identified and validated a proteomic profile that would directly evaluate malignant transformation risk regardless of HCA subtype. Conclusions : This work proposes a proteomic-based machine learning tool, operational on fixed biopsies, that can improve diagnosis and prognosis and therefore patient management for HCAs.< Réduire