Oral potentially malignant disorders (OPMDs) affect approximately 2% of the global population and carry risk of progression to oral squamous cell carcinoma (OSCC). Current risk assessment relies on invasive biopsy with histopathologic grading, a method limited by inter- and intra-observer variability. Although genomic alterations, including loss of heterozygosity (LOH) at 3p14, 9p21, and 17p13 and TP53 mutations, are strongly associated with malignant progression, their clinical adoption has been hindered by the need for tissue acquisition and complex assays. We previously demonstrated that brush biopsy enables non-invasive detection of copy number alterations and somatic mutations with performance comparable to tissue-based whole exome sequencing (WES). In this study, we evaluated whether brush biopsy could identify high-risk genomic alterations and prospectively stratify progression risk in patients with OPMD. A prospective cohort of 110 patients underwent brush biopsy sampling and was followed for a median of 5.5 years. Samples were analyzed using 1000× WES. Allele-specific copy number analysis was performed using ASCAT v3, and somatic mutations were identified with Mutect2. LOH at 9p21, 17p13, and 3p14 was identified in 51%, 35%, and 15% of patients, respectively, while TP53 mutations were detected in 33%. LOH at 9p21 (p=0.05), LOH at 17p13 (p=0.001), and TP53 mutations (p=0.03) were associated with shorter time to malignant transformation. The absence of these alterations yielded sensitivity of 82% and negative predictive value of 91%. Risk increased with accumulation of genomic events (p=0.003). Notably, in a subset of 40 patients with available pathology, histologic grading did not significantly stratify progression risk, whereas genomic alterations did. These findings support non-invasive genomic profiling via brush biopsy as an objective, scalable approach for OPMD risk stratification, with potential for use beyond specialists, including primary care settings, to expand access for at-risk patients.