Abstract Our understanding of disease progression in multiple myeloma (MM) and its precursor conditions, monoclonal gammopathy of undetermined significance (MGUS) and smoldering multiple myeloma (SMM), is classically founded on bulk analysis studies. Low disease burden at the precursor stages has precluded comprehensive analyses of the transcriptomic events underlying malignant transformation. Here, we use single-cell RNA sequencing data from 29,387 bone marrow plasma cells from 26 patients with MGUS, SMM, or MM and 9 healthy controls to characterize the transcriptional transformation at each step of progression. Due to varying disease burdens, many samples contained a mixture of healthy and neoplastic plasma cells. We leveraged this impurity to perform a patient-specific characterization of the disease, comparing each patient's neoplastic and healthy plasma cells. This approach isolated the disease phenotype in each patient, which is usually confounded by biological and technical variability when comparing tumors to samples from healthy donors. We found that neoplastic cells from patients with MGUS and SMM already exhibit phenotypic changes similar to those of advanced myeloma. We observed upregulation of genes corresponding to known MM subtypes, such as CCND1 in patients with t(11;14) translocations and other known driver genes such as HIST1H1C. We also found universal downregulation of certain genes such as CD27, a member of the tumor necrosis factor receptor family associated with the differentiation of B cells into plasma cells, which may signify a common loss of a normal plasma phenotype across samples with different driver events and stages. Pathway analysis of differentially expressed genes further revealed that biological pathways related to myeloma were altered as early as MGUS. We observed that SMM patients with hyperdiploidy exhibit upregulation of ribosomal proteins, as reported in advanced disease. Upregulated genes in select MGUS and SMM samples were enriched for the eukaryotic translation initiation factor 3 (eIF3) complex, which plays important roles in translation, as well as proteasome activity, a function central to the survival of MM and targeted by therapies such as bortezomib. We observed enrichment of the E2F family of transcription factors in MGUS and SMM samples; these are master regulators of proliferation that have been suggested as therapeutic targets in myeloma. Five samples were enriched for genes associated with extracellular exosomes, which has been reported to play an important role in cancer cell signaling and to contribute to osteolysis and drug resistance in MM. Pathway enrichment of genes downregulated in neoplastic cell populations revealed weakened response to endoplasmic reticulum and oxidative stress, presumably allowing myeloma cells to tolerate high volumes of abnormal protein production without apoptosing. To further identify shared gene expression patterns across samples, we employed a Bayesian Non-Negative Matrix Factorization method to decompose our data into 31 gene signatures that capture its variability. In addition to recovering signatures corresponding to known MM subtypes, demonstrating that our method captures cohesive transcriptional networks, we find signatures that capture disease biology shared across subtypes. Most notably, we identified a signature that is active in healthy plasma cells across disease stages and dramatically lost in MM and precursor cells. The top genes in this signature include CD27 and CD79A, which are associated with the B cell lineage and whose downregulation may signify dedifferentiation of premalignant cells as early as MGUS, and JSRP1, CTSH, and HCST, genes as of yet unreported to be involved in plasma and MM cell biology. This phenotype would be obscured at early disease stages by bulk analysis. We validated the discovery and behavior of this signature in an external single-cell dataset from Ledergor et al. (Nature Medicine 2018). In summary, using single-cell RNA sequencing, we discovered that canonical MM pathways are altered as early as MGUS and identified a signature of genes which distinguishes healthy and neoplastic cells even at early disease stages. Our identification of patient-specific transcriptional changes as early as MGUS paves the way for future work exploring personalized treatment approaches prior to malignant disease. Disclosures Haradhvala: Constellation Pharmaceuticals a MorphoSys Company: Consultancy. Zavidij: Constellation Pharmaceuticals: Current Employment. Sontag: curai health: Current holder of individual stocks in a privately-held company; Takeda Pharmaceuticals: Research Funding; Genentech: Research Funding; IBM: Research Funding. Ghobrial: AbbVie, Adaptive, Aptitude Health, BMS, Cellectar, Curio Science, Genetch, Janssen, Janssen Central American and Caribbean, Karyopharm, Medscape, Oncopeptides, Sanofi, Takeda, The Binding Site, GNS, GSK: Consultancy. Getz: IBM, Pharmacyclics: Research Funding; Scorpion Therapeutics: Consultancy, Current holder of individual stocks in a privately-held company, Membership on an entity's Board of Directors or advisory committees.