Gene Expression Profiling Identifies 4 Sub-Classes with Distinct Clinical Associations in Hyperdiploid Myeloma.

Hyperdiploid myeloma (H-MM) affects about 50% of MM. These patients exhibit significant karyotypic and clinical heterogeneity. In this study, we addressed possible biological heterogeneity within H-MM by gene expression profiling of CD138-enriched plasma-cell RNA from 53 H-MM patients using the Affymetrix U133A chip (Affymetrix, Santa Clara, CA). Analysis was performed using GeneSpring 7 (Agilent Technologies, Palo Alto, CA). Unsupervised hierarchical clustering performed using the top 100 most variable genes (generated by ANOVA) identifies 4 putative patient clusters. These clusters were reproduced using principal component analysis. Cluster 1 was characterized by high expression of a large group of cancer testis antigens and mitosis/proliferation related genes (TOP2A, NEK, ASPM, CENPA). Cluster 2 was characterized by high expression of genes involved in IL-6 and HGF signaling (MET, IL6R, SOCS3, SPINT1). Cluster 3 was characterized by high expression of genes involved in TNF/NF-KB signaling and anti-apoptosis (NFKB2, CFLAR, BIRC3, TNFAIP3, RRAS2). Cluster 4 was characterized by high expression of genes involve in angiogenesis (protocadherin, EDNRB). Unsupervised hierarchical clustering using the same set of genes in an independent cohort of H-MM patients (n=60) yielded the same 4 patient clusters over-expressing similar genes. The gene expression profile of these clusters were not driven by differences in trisomies (only about 25% of genes that define the clusters were located on the commonly trisomic chromosomes). Association between the 4 clusters and clinical parameters such as survival, chromosome 13 deletion (Δ13), presence of IgH translocations, PCLI, RAS mutation and p16 methylation was analyzed. Cluster 1 tumors were more proliferative (median PCLI 3.8, p<0.05). Cluster 2 was associated with presence of Δ13 (56%, p<0.05) although none of the genes that defined this cluster falls on chromosome 13. p16 methylation was also less common in cluster 2 (13%, p=0.06). Cluster 3 was associated with low PCLI (median 0.2, P<0.05). Most of the patients with p16 methylation also belonged to this cluster (67%, p=0.06). There was a complete absence of Δ13 in Cluster 4. Patients in the 4 clusters also had significantly different survival (log-rank p=0.002.). The difference in survival was independent of Δ13 since Δ13 has no impact on survival in this patient cohort. The molecular profiles of these subclasses may explain some of these associations. CTA expression is associated with proliferative tumors with poor prognosis. IL-6 is an important survival factor and HGF signaling has been associated with poor survival in MM which may explain the relative poor prognosis of this group. Cluster 3 may have better survival as it is associated with an anti-apoptotic signature rather than proliferative signature. In summary, gene expression profiling identified 4 sub-classes within H-MM (validated on an independent dataset and using different clustering techniques) with distinct clinical and biological associations. The gene signature associated with each sub-classes provide a molecular basis for the different clinical behaviour and further insights into deregulated pathways in H-MM.