Size Selection of Cell-free DNA Increases the Proportion of Tumor Specific Variants in Cancer Patients

The proportion of cell-free DNA (cfDNA) derived from tumor cells in the plasma of patients with cancer is typically less than 2%. To overcome this limitation, methods have been developed to better distinguish the circulating tumor DNA (ctDNA) signal from the noise we encounter in most current methodologies. While these solutions are successful in minimizing the technical noise, it often requires increased sequencing depth and other advanced analytical techniques. Thus, alternative or complimentary approaches, such as liquid biopsies, would be beneficial for improving noninvasive cancer diagnostics. The fragment length of ctDNA may be leveraged to improve this signal-to-noise challenge since the median length of cfDNA in circulation from healthy tissue is typically about 167bp, while ctDNA is shorter on average. We evaluated whether size selecting cfDNA could enrich for shorter ctDNA fragments and thereby enhance signal for the detection of tumor-specific variants. To test this hypothesis, adapter-ligated libraries were size selected using the Yourgene Health LightBench platform for gel-based electrophoresis and size selection, targeting cfDNA fragment sizes up to 142bp (+/- 15bp). The size selected libraries from each patient were first assayed with low-coverage (~0.3X) genome-wide sequencing and analyzed for insert size to ensure proper enrichment of shorter cfDNA fragments. Libraries prior to size selection yielded, on average, 25.9% of reads with cfDNA fragment sizes shorter than 150bp. After size selection, the proportion of cfDNA shorter than 150bp was significantly increased to 96.4% (p<0.001; Wilcoxon Rank Sum). Copy number alterations (CNAs) were identified in the cfDNA data and characterized using analytical methods originally developed for noninvasive prenatal testing and subsequently optimized for ctDNA. The amplitude of a detectable autosomal CNA represents the relative magnitude of the CNA. When evaluating cfDNA from healthy patients, the amplitudes of CNAs before and after size selection were on average within 6%, consistent with a lack of signal enrichment in the absence of disease. Conversely, detectable CNAs in cancer patients were on average 47% greater in amplitude in size selected samples than in the same samples prior to size selection, consistent with an enrichment of signal. These data demonstrate a proof-of-concept for using size selection to enhance signal for the detection of tumor-specific variants in cancer patients.

Learning objectives:

1.    Discuss current methodology for detecting tumor-specific copy number alterations in cancer patients.

2.    Explain how size selection can enhance signal for the detection of tumor-specific variants in cancer patients.