Myeloma is intimately associated with osteolytic bone disease, resulting from myeloma cells' interactions with osteoclasts and osteoblasts and their progenitors, and is dependent on the changes it induces in bone metabolism for progression. Myeloma cell dependence on the bone marrow microenvironment is also evident experimentally, where interaction of primary myeloma plasma cells (MMPC) with osteoclasts (OC) and with mesenchymal stem cells (MSC) support the survival of primary myeloma cells. To understand the molecular mechanisms associated with the survival of MMPC, we used Acuity 4 software to analyze Affymetrix U133 Plus2 chip data and identify changes in gene expression in induced MMPC freshly isolated from 8 patients by interaction with OC and from 8 additional patients with MSC. Expression by MMPC of 675 genes was changed following interaction with OC; 552 genes were upregulated and 123 down regulated. Expression of 296 genes was changed in MSC co cultures (161 upregulated, 135 down regulated). Comparison of the genes whose expression was similarly changed in both co culture systems identified 72 probesets, representing 58 genes, that were commonly changed; 33 were upregulated and 25 down regulated. Ingenuity Pathway Analysis assigned 54 of the 58 genes to 4 distinguished networks of interrelated genes with high probability scores. We next tested the hypothesis that the expression of genes whose expression was commonly changed in the co culture systems has clinical significance. To accomplish this, we used gene expression data available on 127 relapsed patients who had been uniformly treated on our Total Therapy 2 protocol, and for whom gene expression (GEP) data at first relapse (RL) were available. 71 of these patients also had pre treatment (BL) GEP data; for these 71 patients we calculated change in expression of the 72 probesets as the ratio of RL/BL expression signal. We identified 7 genes whose expression changes were significantly (p≤0.05) associated with survival after relapse: These genes were, in order of significance: CCNE2, PECAM1, KLHL21, ICAM1, PLAU, ANPEP, and DUSP1, with p-values ranging from 0.017 to 0.05. Up regulation of PECAM1, ANPEP, DUSP1, and down regulation of CCNE2, KLHL21, ICAM1, and PLAU were associated with longer survival. We further determined whether expression level of these 72 probesets at relapse, defined by signal intensity, correlated with post relapse survival of the 127 patients; 18 genes were significantly (p<0.05)associated with survival: of these, the top 6 genes, sorted in order of p-values of the univariate test were CCNE2 (p<1e-7, HR, defined as ratio of hazard for a twofold increased in signal) =1.83), PECAM1 (p=2e-7, HR=0.64), FOSB (p=2.3e-6, HR=0.76), HMOX1 (p=8.2e-5; HR=0.72), CISH (p<0.0002, HR=0.76), and JUN (p=0.0008, HR=0.78). Eleven other genes associated with survival had p values ranging from <0.002 to 0.047. Although not the purpose of this work, we also tested the ability of the 72 probesets to predict survival with each probeset dichotomized at the median. Using the BRB Array tool, we found that 17 genes predict post relapse survival with at p=0.01 based on log-rank tests in 100 permutations. The percent variability explained by the first 2 principal components = 42.5. Using co culture of myeloma cells with osteoclasts and MSC we identified MMPC genes whose expression is associated with the survival of patients after relapse. These genes define potential targets for improving the survival of relapsed myeloma patients.
No relevant conflicts of interest to declare.
Asterisk with author names denotes non-ASH members.