作者: Ralph H. Hruban , Kornelia Polyak , Victor E. Velculescu , Alberto Bardelli , Patrice J. Morin
DOI:
关键词:
摘要: Identifying the primary site in cases of metastatic carcinoma unknown origin has profound clinical importance managing cancer patients. Although transcriptional profiling promises molecular solutions to this challenge, simpler and more reliable methods for purpose are needed. A training set 11 serial analysis gene expression (SAGE) libraries was analyzed using a combination supervised unsupervised computational select small group candidate genes with maximal power discriminate carcinomas different tissue origins. Quantitative real-time PCR used measure their levels an independent validation 62 samples ovarian, breast, colon, pancreatic adenocarcinomas normal ovarian surface epithelial controls. The diagnostic evaluated cluster methods. From 21,321 unique SAGE transcript tags derived from libraries, five were identified patterns that distinguished four types adenocarcinomas. data obtained clustered tumor manner, generating self-organized map distinctive site-specific domains. Eighty-one percent (50 62) correctly allocated corresponding regions. Metastases tightly tumors. classification generated based on selected database. This may provide practical approach determine type clinically origin.