Résumé
Background: In multiple myeloma (MM), deletion of chromosome (chr.) 13q and hypodiploidy are adverse prognostic factors. Prognostic evaluation is frequently based on interphase-fluorescence in-situ hybridization (FISH) with a single probe for chr. 13q14.
Patients and methods: CD138-positive bone marrow cells from 97 patients with newly diagnosed MM (58 training group (TG) and 39 validation group (VG)) were enriched by magnetic-activated cell-sorting (median purity, 95%). Sorted cells were analyzed by interphase-FISH with probes for chr. 13q14, and additionally 9q34, 11q23, 19q13, t(11;14), and t(4;14). GEP was performed with Affymetrix U133A+B (TG) and HGU133-2.0plus (VG) microarrays. Nearest shrunken centroids classification (NSC) was applied to discriminate clones with FISH-detected del(13q14) vs. those without, using VSN-normalized gene expression values. Chromosomal localization of predictor genes was determined using the MapIt program, and functional relationship was established by Gene Ontology (GO) annotation and creation of GO-slims.
Results: A deletion of chr. 13q14 was found in 27/58 patients (47%) of the training and in 21/39 (54%) of the validation group. Frequencies of trisomies were lower (9q: 48 vs. 74%; 11q: 41 vs. 74%; 19q: 44 vs. 84%) and of IgH translocations higher (48 vs. 16%) in patients with del(13q14). NSC resulted in a predictor for del(13q14) of 378 probe-sets with a cross-validated classification error rate of 22%; the VG is under statistical investigation. Of the predictor genes, 18% were localized on chr. 13 (distributed evenly from 13q12 to 13q33), followed by chr. 19, 11 and 3 (10/6/6%). In the 50 probe-sets with the highest scores, the most frequent localizations of the represented genes were chr. 19, 9, and 13 (12/8/6 of 50). In 8/8 incorrectly classified patients with del(13q14), at least 2 of 3 trisomies (9q, 11q, 19q) were present, hinting at hyperdiploidy. Only 1/5 incorrectly classified patients without del(13q14) harbored 3 trisomies. Biological functions (GO level 3) of predictor genes were related to protein and DNA metabolism (43%), cell growth/maintenance (27%), and cell communication (17%). The most frequent GO term for cellular component was ribosome (34%). Sixty-nine of 80 human ribosomal protein (RP) genes were represented in the predictor, and made up 33 of the top-50 probe-sets. RP genes were overexpressed in patients without del(13q14) compared to plasma cells from 7 normal donors. Expression levels of RPL12, RPLP2 and RPL13A (on chr. 9, 11 and 19) correlated with the respective chr. copy numbers.
Conclusion: FISH-detected del(13q14) is associated with non-hyperdiploidy rather than defining an independent subentity of MM. Overexpression of RP-genes in malignancies has been linked to cell growth, disease progression and drug resistance. A possible pathogenetic role of the upregulation of virtually all ribosomal protein genes observed in del(13q)-negative/hyperdiploid MM clones has to be evaluated further.