6562 Background: Oral cavity squamous cell carcinoma (OCSCC) frequently presents as clinically advanced disease with poor prognosis. When diagnosed at early stages, survival rates approach 80%, underscoring the need for validated, cost-effective detection methods. OCSCC is driven by the serial acquisition of genetic alterations. Tumor-defining somatic mutations are attractive biomarkers and hence their presence in saliva may be associated with malignancy as shown in a few proof-of-concept studies, including our previous work. Based on this premise, we present a low-cost, accurate, next generation sequencing (NGS) test with high clinical utility aimed at detecting mutations in the saliva for early diagnosis and potential screening of OCSCC. Methods: We have designed a custom NGS panel that covers exons of 7 most frequently mutated genes in OSCC. This minimal gene set derived from the analysis from 3 public datasets, predicted incidence of at least one somatic aberration in 89% of patients. We recruited 91 treatment-naïve OCSCC patients and profiled DNA from tissue and matched pre-operative saliva using this test. We also tested DNA from 12 subjects with premalignant lesions with high-grade oral dysplasia and matched saliva. Results: Using stringent variant calling criteria, at least one somatic variant was detected in 88 (96%) of the 91 primary tumors. 90.9% of the matched saliva were concordant, with only a minor decrease in early stage disease. Tumor-specific mutations (≥5% AF) in driver genes were detected in 10 (83.3%) dysplastic lesions, suggesting that driving clonal events may occur early in disease development. Interestingly, in 3 matched saliva of the dysplastic samples, the same mutations were detected. To ensure a variant is not a false positive call, we performed a vigorous multistep analytical validation of this saliva-based test: (i) independent re-sequencing of 24 saliva confirmed 94% reproducibility; (ii) no functionally relevant variants were detected in saliva from 12 of 13 healthy subjects without history of tobacco and alcohol usage; (iii) reproducibility, sensitivity, and specificity were confirmed using a positive control with 7 loci at 0.25% AF across 8 independent saliva sequencing runs and a certified negative control and was found to be on par with droplet digital PCR. Conclusions: These data highlight the feasibility of saliva-based testing for early diagnosis of OCSCC and premalignant lesions.