1. ¹Division of Hematology/Oncology, UCLA School of Medicine, Cedars Sinai Medical Center, Los Angeles
2. ²Department of Pathology and Laboratory Medicine, UCLA School of Medicine, Cedars Sinai Medical Center, Los Angeles
3. ³Department of Hematology/Oncology, Cedars Sinai Medical Center, Los Angeles
4. ⁴Medical and Research Services, VASDHS, San Diego, California
5. ⁵Division of Hematology/Oncology and Cancer Center, UCSD School of Medicine, La Jolla, California
6. ⁶Department of Pathology, University of Arizona, Tucson, Arizona
7. ⁷Department of Hematology/Oncology, University of Arizona, Tucson, Arizona

Correspondence: Dr S de Vos, Division of Hematology/Oncology, UCLA Medical Center, 650 Charles E. Young Drive S., 11-266 Factor Building, Los Angeles, CA 90095-1678. E-mail: devos@mednet.ucla.edu

Received 30 October 2002.

Abstract

Follicular lymphoma (FL) is characterized by a continuous rate of relapse and transformation to a high-grade lymphoma, usually diffuse large B-cell lymphoma (DLBCL), associated with a dismal prognosis and a poor response to conventional chemotherapy. The progression of indolent to aggressive FL is accompanied by the successive accumulation of recurrent chromosomal defects, but the resultant alterations of gene expression are largely unknown. To expand the understanding of the pathogenesis of FL transformation, we initially performed oligonucleotide microarray analyses using Affymetrix HuFL chips on five cases with matched snap-frozen lymph nodes before and after transformation. Expression data were analyzed using the Affymetrix Microarray Suite 4.0 and Genespring 4.0. Thirty-six genes with increased expression and 66 genes with decreased expression associated with transformation were identified and functionally classified. The expression of differentially expressed genes was confirmed by real-time quantitative RT-PCR (QRT-PCR) using a total of seven matched pairs and an additional five FL and five unrelated DLBCL. In addition, selected genes were further analyzed by QRT-PCR or immunohistochemistry using a large, unrelated series of FL (grades 1 to 3) as well as transformed and de novo DLBCL (total of 51 samples). The microarray results correlated with the protein expression data obtained from samples at the time of initial diagnosis and transformation. Furthermore, the expression of 25 candidate genes was evaluated by QRT-PCR with a 78% confirmation rate. Some of the identified genes, such as nucleobindin, interferon regulatory factor 4, and tissue inhibitor of metalloproteinases 1, are already known to be associated with high-grade non-Hodgkin's lymphoma. Novel candidate genes with confirmed increased and decreased expression in transformed DLBCL include ABL2 and NEK2, and PDCD1 and VDUP1, respectively. In summary, this study shows that transformation of FL to DLBCL is associated with a distinct set of differentially expressed genes of potential functional importance.