A multistep model for pancreatic adenocarcinoma has been proposed recently. In this model, well-defined, noninvasive ductal lesions are recognized as precursors of invasive cancer and have been classified under the nomenclature of pancreatic intraepithelial neoplasia, or PanIN. Increasing evidence suggests that PanINs represent true neoplasms of the pancreatic ductal epithelium, accumulating histologic and genetic abnormalities in their progression toward invasive cancer. We have constructed a tissue microarray containing 55 PanIN lesions of all histologic grades in order to perform a multicomponent analysis of the pancreatic adenocarcinoma progression model. The protein products of 14 genes encompassing a variety of functional classes, such as tumor suppressor genes (p53, Smad4/Dpc4), oncogenes (beta-catenin), cell cycle antigens (p16, cyclin D1), proliferation antigens (Ki-67, topoisomerase II alpha), and epithelial apomucins (MUC1, MUC2, MUC5), as well as "novel" genes described as differentially up-regulated in invasive pancreas cancer by global microarray expression analysis (mesothelin, prostate stem cell antigen, fascin, and 14-3-3varsigma), were analyzed by immunohistochemistry on the PanIN tissue microarray. Comparison of the results from the current study with previously published data performed on routine histologic sections of PanINs demonstrates that tissue microarrays are a valid platform for molecular analysis not only of invasive cancers but of precursor lesions as well. In addition, this study demonstrates that molecular abnormalities in PanINs are not random but can usually be stratified into "early" changes (e.g., expression of MUC5 and prostate stem antigen, or loss of p16), "intermediate" changes (e.g., expression of cyclin D1), and "late" changes (e.g., expression of p53, proliferation antigens, MUC1, mesothelin, and 14-3-3varsigma, or loss of Smad4/Dpc4). Understanding the molecular pathogenesis of precursor lesions of invasive pancreatic adenocarcinomas using a high-throughput tissue microarray-based approach is a valuable adjunct to designing rational strategies for early detection of this lethal neoplasm.