Genetic Polymorphisms Identify the Likelihood of Bone Disease in Myeloma: Correlations with Myeloma Cell DKK1 Expression and High Risk Gene Signatures.

Background: Bone disease in myeloma occurs as a result of complex interactions between myeloma cells and the bone marrow microenvironment. To date, no studies have evaluated the potential impact of genetic polymorphisms upon and/or within this microenvironment. Patients and

Methods: Peripheral blood DNA from 282 patients enrolled in the Total Therapy 2 (TT2) protocol was studied using the previously reported Affymetrix 3k BOAC custom chip to evaluate relevant genetic polymorphisms. DKK1 gene expression and high risk GEP gene signatures were assessed as previously reported (

Results: The top 200 SNPs were first evaluated based upon univariate P values linked to limited or extensive bone disease. The top 50 SNPs with the smallest P values were then selected (eliminating closely linked genes) for recursive partitioning analysis. The recursive partitioning was conducted both with and without the insertion of known biologically relevant SNPs. The pruned tree developed with recursive partitioning proved to be quite stable and incorporated 4 dominant SNPs: rs 3766934 Epoxide hydrolase (EPHX1); sr3783408 MAP kinase; sr 1062637 RNA helicase DDX18; and sr3181366 TNFSF8-TNF-α. No alternate SNP substituted for EPHX1 in the recursive model. The 4 SNP signature of EPHX1 GG, MAP4K5 AG/AA, DDX18 GG/CC plus TNFSF8 CT/TT was highly correlated with the number of MRI lesions: mean 8.66 lesions versus 3.33 for alternate SNPs (P<.001). In the univariate association of standard prognostic factors, 4 SNP signature and 70 and 17 gene models for high risk, the best correlation with bone disease was with the 4 SNPs (P<.0001) followed by the 17 gene GEP model (P=0.05). Using stepwise multivariate regression analysis the best correlation with bone disease was Log₂DKK-1 expression (P=.0001) but closely followed by SNPs EPHX1 (P=.0009); MAP4K5 (P=.0071) plus TNFSF8 and serum LDH (P=.02). The various analyses training (2/3) and validation sets were used with assessment of sensitivity and specificity. The microenvironmental 4 SNP signature combined with GEP DKK1 provided the best prediction of myeloma bone disease and overall outcome.

Conclusions: This first assessment of genetic polymorphisms linked to myeloma bone disease in myeloma has led to the identification of polymorphisms with both potential biologic relevance and utility in prognostic models of myeloma bone disease including risk stratification for bisphosphonate use. Validation of these findings may allow a search for potential therapeutic targets.