Detecting hidden threats using longitudinal liquid biopsies

Multicancer early detection (MCED) assays aim to detect cancers while still asymptomatic and, hopefully, more likely to be curable. To understand how long an MCED assay is interpretable before the emergence of symptoms and how frequently the test should be repeated after a negative result, we mined a surrogate dataset from a longitudinal liquid biopsy program monitoring patients with prostate cancer (PC). Retrospective analysis identified 9 patients who, while receiving hormonal therapy for PC, were diagnosed with a 2 ^nd primary solid malignancy for which they also underwent tissue NGS. Tumor-informed bioinformatic analysis revealed that somatic variants found in the 2 ^nd primary cancer could be detected in the cell-free DNA as early as 32.3 months prior to clinical diagnosis of the 2 ^nd malignancy, with a median lead time of 9.2 months. Due to several features of this cohort, this may actually be an underestimate of the true lead time. Furthermore, the lead time was positively correlated to the number of somatic variants found in the 2 ^nd primary cancer. The majority of these somatic variants were located outside of protein-encoding exonal regions. Our results shed light on the natural history of DNA shedding from the 2 ^nd primary cancers that emerged in this PC patient cohort. We propose that our observations in this surrogate cohort are also relevant to the use of genomic liquid biopsy assays for cancer screening in individuals without cancer history.