A mathematical model of cell-free DNA fragment size reveals cancer-specific fragmentomic patterns

Analysis of cell-free DNA (cfDNA) fragmentomic features shows promise for minimally invasive cancer diagnostics. The size distribution of cfDNA molecules has emerged as a novel analytical dimension enhancing ctDNA detection. Here, we propose a mathematical model of cfDNA length distribution across multiple biodiy flulids, deconvoluted into a series of peaks (components) represented by the Cauchy–Lorentz distribution with an approximate 10 bp periodicity. This model provides a new analytical framework for investigating the biological mechanisms of cfDNA fragmentation and release, leading to the identification of a previously uncharacterized 159 bp fragment length, that may indicate a demarcation between intra- and inter-nucleosome-derived cfDNA. Furthermore, we demonstrate that tumor-specific ctDNA shortening and phagocytosis-associated cfDNA shortening can be differentiated through analysis of the amplitudes and scales of intra- and inter-nucleosomal components. Moreover, leveraging ctDNA-specific alterations revealed by cfDNA size profile deconvolution analysis could significantly enhance cancer detection.