The Gene Expression Signatures (GEP) of Whole Bone Marrow Biopsies (Bx) from Patients with Multiple Myeloma (MM) in Remission Reflect Disease Risk and Therapy.

Background: We have previously shown that the GEP signatures of whole bone marrow biopsies from patients with MM can be distinguished from those of normal healthy subjects. What is not clear is whether the GEP of myeloma Bx samples returns to normal in clinical remission following tandem stem cell transplantation.

Patients and Methods: Samples included random Bx from 85 newly diagnosed MM patients treated on Total Therapy 2 protocol; 22 also had Bx in remission and 19 at relapse; in addition, Bx were available from 8 subjects with MGUS and 12 normal donors. Labeled cRNA was applied to Affymetrix U133Plus2 microarrays. Significant analysis of microarray (SAM) was performed to identify remission-related genes by comparing normal cases with remission cases. Unsupervised hierarchical cluster analysis was employed on Log2 transformed GEP intensity values of remission-related gene identified with SAM using a false discovery rate of 1%.

Results: Altogether 998 genes were significantly differentially expressed in a comparison of normal and remission samples. Of these, 31 (3%) were higher and 967 (97%) lower in remission samples compared to normal donors. Two-dimensional hierarchical cluster analysis using the 998 genes and 22 remission and 12 normal donor samples provided visual evidence of 2 separate remission signatures. A single sample acquired what appeared to be a completely normal signature with 8 of 22 cases having a normal-like signature and 12 of the 22 cases having a distinct non-normal signature. Two-dimensional hierarchical clustering with remission-related genes using whole biopsies from 85 newly diagnosed MM, 8 MGUS, 22 remissions, and 19 relapses produced 2 major branches: one contained 50 MM (59%), 5 MGUS (63%), 10 remissions (45%), and 4 relapses (21%) with the second branch containing 35 MM (41%), 3 MGUS (37%), 12 remissions (55 %), and 15 relapses (79%). All 12 of the non-normal, while none of the 10 normal-like remissions, clustered in the group containing the majority of relapse cases (p=0.002). A comparison of 16 clinical variables amongst the newly diagnosed cases revealed that the 35 clustering with the non-normal remission cases were significantly associated with higher levels of B2M (P = 0.001), creatinine (P = 0.0056) and LDH (P = 0.024); higher incidence of abnormal (P = .0075) and a lower incidence of hyperdiploid karyotypes (P = 0.0004); higher levels of FISH-1q21 amplification (P = 0.04) and FISH-13 deletion (P = 0.04, 80% cutoff); lower HGB (P = 0.0058), and fewer MRI-defined focal lesions (P = 0.03). There was a strong trend for the presence of high-risk features at diagnosis among the 12 non-normal remission samples compared to the 10 normal-like samples. Of the 10 cases with the normal-like remission signature, 8 were on the thalidomide arm of the trial but only 4 of 12 patients with the non-normal remission signature were on the thalidomide arm (P=0.029).

Conclusion: Despite -obtaining a clinical remission, the gene expression pattern within the marrow of myeloma patients does not return to normal in a substantial subset of patients.