Background: Tumor cells exhibit elevated ROS levels and cancer stem cells have lower ROS levels than their normal counterparts [Kobayashi et al., 2011]. However, little is known about the effect of ROS levels on multiple myeloma (MM) plasma cell (PC) chemosensitivity. Previously we found that ROS levels are variable in CD138-selected MMPCs from different cases at diagnosis. However, ROS levels in MMPCs from focal lesions (FL, n=42) are lower than PCs from random bone marrow (BM) aspirates (n=192, p=0.02). ROS levels also vary among MM molecular subgroups, with MF and CD2 subgroups having low ROS levels, suggesting that the microenvironment and genetic factors affect their oxidative capacities. We hypothesized that ROS level may reflect the cancer stemness properties of MM cells. We first sought to establish a model to investigate MM cells with different ROS levels and then determine if this property confers cancer stemness and chemoresistance.Methods: We analyzed gene expression profiles (GEP) of primary myeloma cells from newly diagnosed patients with low (n=48) and high (n=48) ROS level by gene set enrichment analysis. We compared viability, proliferation, cell cycle status, drug efflux ability, chemosensitivity, and clonogenic capacity of MM cell lines H929 and U266 with high and low ROS levels, sorted by FACS using H2DCFDA. We also investigated whether MMPCs with low or high ROS levels would differ in their sensitivity to BBI-608, a drug that is reported to target STAT3 driven gene transcription and cancer stemness properties in solid cancers [Li et al., 2015] and is currently being tested in hematologic cancers [BBI608-103HEME].Results: Pathways enrichment analysis revealed that primary MM cases with low ROS had increased oxidative phosphorylation, proteasome activity, and nucleotide excision repair compared to cases with high ROS that had increased glycolysis, and chemokine and calcium signaling pathways. To explore these differences we separated MMCLs into cells not sorted by ROS level (non-ROS sorted), ROSLo, and ROSHi groups. After sorting, H929 cells retained significant differences in ROS levels for 48-72 hr and U266 cells returned to basal ROS levels within 24hr. ROSLo H929 cells had significantly reduced cell number (p<0.0001) compared to non-ROS sorted and ROSHi cells. Similarly, ROSLo cells had reduced and ROSHi cells had significantly increased proliferation index compared to non-ROS sorted cells (H929-p<0.05, U266-p<0.001). ROSLo cells also have a significantly larger percentage of cells in S phase (p <0.01) and reduced percentage in G2 (p< 0.001) compared to non-ROS sorted cells. While ROSHi cells (but not ROSLo cells) were less sensitive to melphalan compared to non-ROS sorted cells, there was no difference in sensitivity to bortezomib. Examination of ABC transporter efflux activity revealed that ROSHi cells had increased drug efflux capacity via the multidrug resistance protein MRP1 (p<0.01). ROSLo cells showed greater colony forming capacity, however after serial replating, ROSHi cells increase colony forming capacity after each passage eventually matching that of ROSLo cells. Interestingly, if clonogenic cells are re-sorted into ROSLo and ROSHi groups between passages, ROSLo cells retain superior colony forming capacity compared to ROSHi cells. ROSLo (IC50-0.656μM) and ROSHi (IC50-0.56μM) cells had significantly increased sensitivity to BBI-608 compared to non-ROS-sorted cells (IC50-0.991 μM). Cell number was significantly reduced in all 3 groups without significantly affecting viability. The STAT3 target gene, CCND1 is significantly elevated in ROSHi cells and effectively reduced by BBI-608. After low dose BBI-608 (0.6μM) treatment, the remaining cells were subjected to colony forming assays. While BBI-608 reduced colony forming capacity to some extent in all 3 groups, ROSHi cells were most sensitive.Conclusions: Characterization of MM cells with variable ROS levels revealed that ROSHi cells have greater drug efflux capacity and are more proliferative and chemoresistant, whereas ROSLo cells have reduced proliferation and increased clonogenic capacity. ROSLo MMPCs may represent a subpopulation of tumor cells with stemness properties that are enriched in the hypoxic BM niche and in MM focal lesions. DisclosuresMalaviarachchi:University of Arkansas for Medical Sciences: Employment. Turner:University of Arkansas for Medical Sciences: Employment. Stein:University of Arkansas for Medical Sciences: Employment. Barlogie:University of Arkansas for Medical Sciences: Employment. Morgan:Celgene: Honoraria, Membership on an entity's Board of Directors or advisory committees; MMRF: Honoraria; Bristol Myers Squibb: Honoraria, Membership on an entity's Board of Directors or advisory committees; Takeda: Honoraria, Membership on an entity's Board of Directors or advisory committees; University of Arkansas for Medical Sciences: Employment; CancerNet: Honoraria; Weismann Institute: Honoraria. Epstein:University of Arkansas for Medical Sciences: Employment. Yaccoby:University of Arkansas for Medical Sciences: Employment. Johnson:University of Arkansas for Medical Sciences: Employment.