Blood-based early detection of viral precancers

Blood-based (liquid biopsy) cancer screening holds enormous promise to decrease cancer morbidity and mortality, yet exisGng approaches do not reliably detect the earliest stages of disease, when benefit would be greatest. Current liquid biopsy methods typically target mutated or methylated cell free (cf)DNA, which is in low abundance at the earliest stages of cancer development, leading to weak or non-existent signals. Thus, there is a criGcal unmet need for ultrasensiGve and specific liquid biopsy screening methods to detect cancers at their incepGon, when they are most therapeuGcally vulnerable and before regional spread. 10-15% of all cancers are caused by oncogenic viruses. TargeGng viral cfDNA leads to improved sensiGvity for cancer early detecGon, due to the high copy number of viral DNA compared to human DNA per cell. Human papillomavirus (HPV) is the most common oncogenic virus and is responsible for 5% of all cancers worldwide. HPV+ cancers develop over decades, beginning with HPV infecGon and progressing through precancer stages, providing a significant period in which cancer and precancer early detecGon is feasible and permiWng secondary prevenGon. HPV cfDNA is detectable both at HPV+ cancer diagnosis and years prior to diagnosis, with high sensiGvity and specificity, using HPV whole genome sequencing approaches, supporGng the feasibility of blood-based HPV+ cancer and precancer screening. Fragmentomics is an emerging field of study which has shown promise for improving the sensiGvity and specificity of early cancer detecGon, including precancer detecGon. Here we propose to uGlize HPV+ anal cancer and precancer as a model system to test the overarching hypothesis that viral cancers and high grade precancers requiring therapeuGc intervenGon can be accurately detected in the blood and differenGated from low grade precancers and infecGon, which can be safely observed. To do so we will combine two powerful approaches, an ultrasensiGve HPV cfDNA whole genomes sequencing liquid biopsy, HPV-DeepSeek, and a fragment length and end-moGf signature computaGonal analysis framework, fragmentTopics. In Aim 1 we will determine how viral cfDNA release and fragmentaGon pa\erns change across cancer developmental stages. In Aim 2 we will train and validate a mulG-modal machine learning liquid biopsy and determine the diagnosGc accuracy for idenGfying cancers and precancers which require therapeuGc intervenGon. This innovaGve proposal integrates two novel, custom approaches to address a criGcal unmet need in bloodbased cancer screening– the detecGon and classificaGon of precancers. The work is significant because successful compleGon would lead to a minimally invasive screening approach with the potenGal to improve survival and quality of life for 5% of all cancer paGents, with broad implicaGons for blood-based precancer detecGon across all cancers. Project Number: 1R01CA311332-01 | Fiscal Year: 2026 | NIH Institute/Center: National Cancer Institute (NCI) | Principal Investigator: Daniel Faden | Institution: MASSACHUSETTS EYE AND EAR INFIRMARY, BOSTON, MA | Award Amount: $642,406 | Activity Code: R01 | Study Section: Molecular Cancer Diagnosis and Classification Study Section[MCDC] View on NIH RePORTER: https://reporter.nih.gov/project-details/11340418