Fragment Size-Based Enrichment of Viral Sequences in Plasma Cell-Free DNA

Abstract

Sequencing of plasma cell-free DNA (cfDNA) is a promising milieu for broad-based cancer and infectious disease diagnostics. The performance of cfDNA sequencing for infectious disease diagnostics is chiefly limited by inadequate analytical sensitivity. The current study investigated whether the analytical sensitivity of cfDNA sequencing for viral diagnostics could be improved by selective sequencing of short cfDNA fragments, given prior observations of shorter fragment size distribution in microbial and cytomegalovirus-derived cfDNA compared with human-derived cfDNA. It shows that the shorter plasma cfDNA fragment size distribution is a general feature of multiple DNA viruses, including adenovirus [interquartile range (IQR), 87 to 165 bp], herpes simplex virus 2 (IQR, 114 to 195 bp), human herpesvirus 6 (IQR, 145 to 176 bp), and varicella zoster virus (IQR, 98 to 182 bp), compared with human (IQR, 148 to 178 bp). It was used to further optimize a size selection-based cfDNA sequencing method, demonstrating an enrichment of viral sequences up to 16.6-fold, with a median fold enrichment of 6.7×, 4.6×, 2.2×, and 10.3× for adenovirus, herpes simplex virus 2, human herpesvirus 6, and varicella zoster virus, respectively. These findings demonstrate a simple yet scalable method for enhanced detection of DNA viremia that maintains the unbiased nature of cfDNA sequencing.

Figure 1

Experimental overview of this study. Multiple plasma specimens containing various DNA viruses were subjected to paired library preparations comparing double-stranded DNA (dsDNA) versus single-stranded DNA library generation to determine the virus/human ratio at different fragment sizes. dsDNA libraries were then subjected to specific size selection before sequencing, and the on-target virus percentage was compared with libraries without size selection. ADV, adenovirus; cfDNA, cell-free DNA; EBV, Epstein-Barr virus; HHV-6, human herpesvirus 6; HSV-2, herpes simplex virus 2; VZV, varicella zoster virus.

Figure 2

Distribution of virus- and human-derived cell-free DNA (cfDNA) fragment lengths by double-stranded DNA (dsDNA) and single-stranded DNA (ssDNA) library preparation. Density plots for human- and virus-derived cfDNA fragment length from nine clinical plasma specimens matched by virus species after dsDNA (A) or ssDNA (B) library preparation. Human cfDNA fragment length was collectively derived from all nine samples. Cumulative distribution plots for cfDNA fragments derived from nine individual specimens of adenovirus (ADV), Epstein-Barr virus (EBV), or varicella zoster virus (VZV) using dsDNA (C) or ssDNA (D) library preparation. The ratio of virus-derived cfDNA relative to that of human origin at each fragment length using a 1-bp window following dsDNA (E) or ssDNA (F) library preparation. Lines are colored by species.

Figure 3

Effect of size selection on virus- and human-derived cell-free DNA (cfDNA) fragments from 12 clinical plasma specimens separated by species. Viral size profiles were cumulatively derived from all three adenovirus (ADV)–positive, three Epstein-Barr virus (EBV)–positive, one human herpesvirus 6 (HHV-6)–positive, two herpes simplex virus 2 (HSV-2)–positive, and three varicella zoster virus (VZV)–positive samples. Human cfDNA size distribution was derived from human fragments in all 12 samples. Lines are colored by pre–size selection (gold) and post–size selection (purple).

Figure 4

Enrichment of viral sequences in plasma cell-free DNA following size-selecting sequencing approach. A: Fold viral enrichment of five different DNA viruses, ranging from 1× to 16×. B: Viral fractions from 12 specimens before and after size selection are graphed in log scale. Dashed line shows a y = 1 line. Dots are colored by virus type. ADV, adenovirus; EBV, Epstein-Barr virus; HHV-6, human herpesvirus 6; HSV-2, herpes simplex virus 2; VZV, varicella zoster virus.

Figure 5

Profiling of mitochondrial DNA (mtDNA) and human-derived cell-free DNA (cfDNA) across all loci collected from 12 clinical plasma specimens. A: Distribution of human reads on chromosomes from before and after size selection sequencing data. B: Size distributions of fragments derived from all human cfDNA (black) and human mitochondria after single-stranded DNA (ssDNA) preparation (orange) or double-stranded DNA (dsDNA) preparation (green) or dsDNA preparation and post–size selection (purple). C: Cumulative size distribution profile of same data. Chrx, chromosome.