Single-Cell Genomics-Based Molecular Algorithm for Early Cancer Detection

As one of the prime applications of liquid biopsy, the detection of tumor-derived whole cells and molecular markers is enabled in a noninvasive means before symptoms or hints from imaging procedures used for cancer screening. However, liquid biopsy is not a diagnostic test of malignant diseases per se because it fails to establish a definitive cancer diagnosis. Although single-cell genomics provides a genome-wide genetic alternation landscape, it is technologically challenging to confirm cell malignancy of a suspicious cell in body fluids due to unknown technical noise of single-cell sequencing and genomic variation among cancer cells, especially when tumor tissues are unavailable for sequencing as the reference. To address this challenge, we report a molecular algorithm, named scCancerDx, for confirming cell malignancy based on single-cell copy number alternation profiles of suspicious cells from body fluids, leading to a definitive cancer diagnosis. The scCancerDx algorithm has been trained with normal cells and cancer cell lines and validated with single tumor cells disassociated from clinical samples. The established scCancerDx algorithm then validates hexokinase 2 (HK2) as an efficient metabolic function-associated marker of identifying disseminated tumor cells in different body fluids across many cancer types. The HK2-based test, together with scCancerDx, has been investigated for the early detection of bladder cancer (BC) at a preclinical phase by detecting high glycolytic HK2^high tumor cells in urine. Early BC detection improves patient prognosis and avoids radical resection for enhancing life quality.