Cancer is one of the leading causes of death globally and detection, continuous monitoring and effective treatment are critical for improving patient outcomes. Currently, cancer diagnosis and genomic analysis for personalized treatment often require invasive procedures such as biopsies, which can be costly, time-consuming, and carry risks to the patient. An alternative, less invasive method for detecting cancer is the detection of tumor-derived circulating free DNA (cfDNA) in biological fluids such as blood. This method has shown promise as to detect cancer as it reflects the genetic changes, such as mutations present in the tumor tissue. In addition to primary cancer diagnosis, highly sensitive disease monitoring is crucial in the minimal residual disease (MRD) setting, which refers to the low-level presence of cancer cells in the body after surgery and treatment and can predict cancer recurrence and overall survival and is therefore of great interest to clinicians as a biomarker. Saliva is a biological fluid that has been explored as a source of cfDNA and blood cells but has not yet been utilized as a basis for assay development due to a lack of characterization of cfDNA levels and validation of consistent tumor DNA presence. As the efficiency of monitoring residual disease and cancer detection relies on frequent, repetitive sampling of cfDNA from the patient, saliva represents an attractive modality as it is an easy-to-obtain sample type, including in an at-home setting. In this proposal, we aim to develop and evaluate an assay to detect and monitor the level of cfDNA from tumor in saliva and compare the sensitivity of detection to cfDNA obtained from plasma/blood for patients with head and neck squamous cell carcinoma (HNSCC), where we recently described the evolutionary trajectories and early vs. late genomic driver event emergence, in both HPV+ and HPV- HNSCC. Despite its high prevalence as the sixth most frequent cancer worldwide, most HNSCC patients continue to be diagnosed with late-stage cancer, underscoring the need for sensitive early detection as well as disease monitoring for treatment response. To achieve the necessary degree of sensitivity for what is often a low tumor DNA fraction in saliva, we will use duplex sequencing to minimize sequencing errors, develop algorithms and machine learning models for genomic analysis of cfDNA, and increase sensitivity by integrating signal with DNA methylation data. As a result of this Phase I proposal, we will investigate the use of cfDNA from saliva samples, compare it with plasma cfDNA testing, develop the experimental and computational methodology for saliva cancer genomic tests, quantify the cost-effectiveness of this approach, and aim to develop a reliable, non-invasive method for detecting and monitoring cancer.