Histological Subtypes of Canine Lymphoma are Characterized by Unique Gene Expression Profiles

Abstract

Introduction: In people, specific histological subtypes of lymphoma have different biological behaviors. Similar differences are recognized for some subtypes of canine lymphoma. Recently, microarray expression analysis has been used to identify unique gene expression profiles that characterize specific subtypes of human lymphoma. In this study, we use high‐density oligonucleotide microarrays to classify canine lymphomas based upon their gene expression profiles.Methods: Total RNA was isolated from 27 lymphoma and 3 normal lymph node canine biopsy samples. Microarray analysis was performed as described (http://www.dnaarrays.org). cRNA was hybridized to GeneChip® Canine Genome Arrays (Affymetrix), that contain over 21,700 probe sets, measuring transcripts across the entire canine genome. Gene expression levels were estimated from GeneChip® PM probe intensities by means of an enhanced version of the Li‐Wong PM‐only algorithm. Exceptional micro‐clusters of gene expression profiles were identified by means of a custom “agglomerative partitioning” algorithm which recursively partitions sets of gene expression profiles into a hierarchy (KTree) of “typical”(largest) and “exceptional”(smaller) subsets.Results: We identified microclusters of gene expression profiles that distinguish normal lymph node, B‐cell lymphoma, T‐cell lymphoma, and diffuse large B‐cell lymphoma phenotypes. We confirmed that, when standard hierarchical cluster analysis is applied to the identified gene expression profiles, tumors consistently cluster in concordance with their histological diagnosis.Conclusions: Histologically defined subtypes of canine lymphoma are characterized by distinct gene expression profiles. Gene expression profiling has great potential for the identification and verification of histological subtypes of lymphoma and for increasing understanding of the molecular pathology of canine lymphoma.