Gene Expression Profile Analysis According to Recurrent Gene Copy Number Abnormalities Defines a Diffuse Large B-Cell Lymphoma Subgroup Characterized by 9p21 Locus Deletion, Ribosome Machinery Deregulation and Poor Prognosis. A GELA Study

Genomic gains and losses play a crucial role in the development and progression of DLBCL. Some gains or losses are associated with particular morphologic or clinical manifestations and correlate with the “germinal center B-cell like” (GCB)/non-GCB phenotype, as defined by gene expression profiles (GEP). We previously developed a reliable and routinely single applicable PCR assay, which provided information regarding gain/loss of relevant genes, and prognosis in DLBCL, termed QMPSF (Multiplex PCR of Short Fluorescent Fragments). Here, we combined GEP and QMPSF approaches to delineate molecular pathways related to recurrent gene copy number abnormalities (GCNA) and assess their prognosis value in patients treated by R-CHOP. For this purpose a series of 69 newly diagnosed DLBCL, included in the 98–5 GELA trial with available tumor DNA was studied (median age = 69 years [59–79], IPI2–3: 64%; 4–5: 36%, 40 treated by R-CHOP and 29 by CHOP). A single QMPSF assay, validated by CGH array, to detect GCNA of 8 relevant genes including SIM1 (6q16), MYC (8q24), CDNK2A (9p21), RB1 (13q14), REL (2p13), BCL2 (18q21), TP53 (17p13), and CDKN1B (12p13) was performed. In addition a dedicated QMPSF assay that provides a “bar code” of the 9p21 locus containing CDKN2A (p16INK4a and p14ARF) and CDKN2B (p15INK4b) was designed. To delineate specific gene expression profile according to recurrent GCNA a subset of 52 patients were studied by both GEP (Affymetrix U133A) and QMPSF technologies. Gains of MYC, BCL2, and REL were observed in 13, 28 and 20 % respectively. DNA copy losses of TP53, CDNK2A, RB1 and SIM1 were observed in 9, 40, 6 and 17 % of cases respectively. Using supervised analysis, we delineated specific GEP according to the most frequent GCNA detected by QMPSF. Interestingly, a signature related to 9p21 locus (CDKN2A/CDKN2B) deletion was associated with an overexpression of several ribosome machinery coding genes and the involvement of distinct antiapoptotic molecular mechanisms. Subsequent genomic analysis with the dedicated assay indicated that in most of cases deletions were homozygous and abolished simultaneously p14arf and p16INK4a expression. With a median follow-up of 81 months, CDKN2A deletion, strongly correlates to a poor outcome in the entire cohort (5y OS=25% respectively vs.60% for patients in germline configuration, p=.003) and in the subgroup of patients treated by R-CHOP (5y OS=40% vs.70%, p=.04). Furthermore, prognosis impact of GCNA involving CDKN2A was validated in an independent set of 35 patients treated by R-CHOP. To conclude, combination of QMPSF and GEP may constitute a powerful approach to delineate new genomic pathways with prognosis impact in DLBCL. Notably, CDKN2A/CDKN2B loss, detected in more than one third of DLBCL patients constitutes a strong factor of chemoresistance that is not overcome by R combination. GEP indicates that this may be a consequence of an independent p14arf/p53 pathway, involving the well-established p14arf related ribosome regulation function.