Predicting Treatment Response and Survival of Multiple Myeloma with the Xbp-1 Co-Expression Gene Module

Multiple Myeloma (MM) is a B-cell malignancy characterized by monoclonal expansion and accumulation of abnormal plasma cells in bone marrow. The extensive biological and clinical heterogeneity of MM hinders effective treatment and etiology research. Several molecular classification systems for prognosis have been proposed, but they are not predictive for treatment response. XBP-1 is a critical molecule in coordinating normal plasma cell differentiation and survival because of its role in regulating the unfolded protein response. We identified a module of 100 genes co-expressed with XBP-1 (XBP-1-M) in the dataset GSE2658 by QlucoerOmics Explore 2.2 (Qlucore AB, Lund, Sweden) software. Using unsupervised consensus clustering, we assigned 264 MMs into XBP-1 low or XBP-1 high subtype in the validation dataset GSE9782. Patients in GSE9782 were randomly assigned into bortezomib or high-dose dexamethasone (Dex) treatment arms in the APEX 3 trial (039) (Blood 2007;109:3177-88). We integrated XBP-1-M clustering-based response analysis with the choice of treatment in GSE9782, XBP-1 high MMs had lower response rate (CR+PR) than XBP-1 low group (18.5% vs 59.67%, p<0.01). In the XBP-1 high MMs, significantly different response rate between bortezomib treatment and Dex treatment (36.34% vs 0%, p<0.05) was observed, while in the XBP-1 low MMs, no significantly different response between bortezomib treatment and Dex treatment (55.88% vs 61.18%, p>0.05) was noted. We further integrated XBP-1-M clustering-based survival analysis with the choice of treatment in GSE9782, XBP-1 high MMs showed significantly poorer progression-free survival (PFS) compared with XBP-1 low MMs during the 5-8 years follow-up, using either bortezomib(p=0.0071) or Dex (p=0.0011) regimen. In the patients treated with Dex regimen, XBP-1 high MMs showed significant poorer overall survival (OS) compared with XBP-1 low MMs (p=0.0316), while there were no differences in bortezomib regimen between XBP-1 high MMs and XBP-1 low MMs (p=0.224). Next, we integrated the XBP-1-M clustering-based genetic abnormality in GSE9782, there were higher frequencies of genetic abnormalities in XBP-1 high MMs than those in XBP-1 low MMs (1q gain 49.5% vs 12.6%, p<0.01; 17p del 21.9 % vs 3.2%, p<0.01; 13q del 58.1% vs 28.5%, p<0.05; t(4,14) 22.8% vs 15.8%, p<0.05; t(11,14) 31.4% vs 3.2%, p<0.01). The higher cytogenetic abnormalities may be related to the poorer treatment response and survival in patients with XBP-1 high expression. Therefore, we conclude that XBP-1-M may enable prediction of the treatment response and clinical outcome in multiple myeloma.