This dissertation investigates the implications of genome-wide 5-hydroxymethylcytosine (5hmC) modifications in circulating cell-free DNA (cfDNA) for outcomes of multiple myeloma (MM). MM is a clinically heterogeneous hematologic malignancy with disparities between European American (EA) and African American (AA) patients, likely driven by multifactorial causes. 5hmC, an epigenetic alteration with distinct regulatory functions relative to the more widely studied 5-methylcytosine, has been associated with survival in MM and therapeutic response of other hematological and solid tumors. Leveraging a racially diverse cohort of MM patients integrating clinical, laboratory, patient-reported data, and genome-wide 5hmC profiles, this work applies machine learning to identify 5hmC marker genes associated with survival and treatment response. Building on prior work linking cfDNA-based 5hmC modifications and survival in MM, we first identified distinct marker genes predictive of overall and progression-free survival among EA and AA patients (Chapter 2). We then extended this investigation to treatment response, highlighting population differences in 5hmC marker genes associated with response to initial therapy. Finally, to explore the potential role of diet in modulating 5hmC and its impact on MM survival, we examined whether fruit and vegetable intake is associated with survival and whether this association is mediated by genome-wide 5hmC modifications (Chapter 5).