Genomic origin, fragmentomics, and transcriptional properties of long cell-free DNA molecules in human plasma.

Recent studies have revealed an unexplored population of long cell-free DNA (cfDNA) molecules in human plasma using long-read sequencing technologies. However, the biological properties of long cfDNA molecules (>500 bp) remain largely unknown. To this end, we have investigated the origins of long cfDNA molecules from different genomic elements. Analysis of plasma cfDNA using long-read sequencing reveals an uneven distribution of long molecules from across the genome. Long cfDNA molecules show overrepresentation in euchromatic regions of the genome, in sharp contrast to short DNA molecules. We observe a stronger relationship between the abundance of long molecules and mRNA gene expression levels, compared with short molecules (Pearson's r = 0.71 vs. -0.14). Moreover, long and short molecules show distinct fragmentation patterns surrounding CpG sites. Leveraging the cleavage preferences surrounding CpG sites, the combined cleavage ratios of long and short molecules can differentiate patients with hepatocellular carcinoma (HCC) from non-HCC subjects (AUC = 0.87). We also investigated knockout mice in which selected nuclease genes had been inactivated in comparison with wild-type mice. The proportion of long molecules originating from transcription start sites are lower in Dffb-deficient mice but higher in Dnase1l3-deficient mice compared with that of wild-type mice. This work thus provides new insights into the biological properties and potential clinical applications of long cfDNA molecules.

Fragmentation landscape of cell-free DNA revealed by deconvolutional analysis of end motifs.

Cell-free DNA (cfDNA) fragmentation is nonrandom, at least partially mediated by various DNA nucleases, forming characteristic cfDNA end motifs. However, there is a paucity of tools for deciphering the relative contributions of cfDNA cleavage patterns related to underlying fragmentation factors. In this study, through non-negative matrix factorization algorithm, we used 256 5' 4-mer end motifs to identify distinct types of cfDNA cleavage patterns, referred to as "founder" end-motif profiles (F-profiles). F-profiles were associated with different DNA nucleases based on whether such patterns were disrupted in nuclease-knockout mouse models. Contributions of individual F-profiles in a cfDNA sample could be determined by deconvolutional analysis. We analyzed 93 murine cfDNA samples of different nuclease-deficient mice and identified six types of F-profiles. F-profiles I, II, and III were linked to deoxyribonuclease 1 like 3 (DNASE1L3), deoxyribonuclease 1 (DNASE1), and DNA fragmentation factor subunit beta (DFFB), respectively. We revealed that 42.9% of plasma cfDNA molecules were attributed to DNASE1L3-mediated fragmentation, whereas 43.4% of urinary cfDNA molecules involved DNASE1-mediated fragmentation. We further demonstrated that the relative contributions of F-profiles were useful to inform pathological states, such as autoimmune disorders and cancer. Among the six F-profiles, the use of F-profile I could inform the human patients with systemic lupus erythematosus. F-profile VI could be used to detect individuals with hepatocellular carcinoma, with an area under the receiver operating characteristic curve of 0.97. F-profile VI was more prominent in patients with nasopharyngeal carcinoma undergoing chemoradiotherapy. We proposed that this profile might be related to oxidative stress.

Fragmentomics of urinary cell-free DNA in nuclease knockout mouse models.

Urinary cell-free DNA (ucfDNA) is a potential biomarker for bladder cancer detection. However, the biological characteristics of ucfDNA are not well understood. We explored the roles of deoxyribonuclease 1 (DNASE1) and deoxyribonuclease 1-like 3 (DNASE1L3) in the fragmentation of ucfDNA using mouse models. The deletion of Dnase1 in mice (Dnase1-/-) caused aberrations in ucfDNA fragmentation, including a 24-fold increase in DNA concentration, and a 3-fold enrichment of long DNA molecules, with a relative decrease of fragments with thymine ends and reduction of jaggedness (i.e., the presence of single-stranded protruding ends). In contrast, such changes were not observed in mice with Dnase1l3 deletion (Dnase1l3-/-). These results suggested that DNASE1 was an important nuclease contributing to the ucfDNA fragmentation. Western blot analysis revealed that the concentration of DNASE1 protein was higher in urine than DNASE1L3. The native-polyacrylamide gel electrophoresis zymogram showed that DNASE1 activity in urine was higher than that in plasma. Furthermore, the proportion of ucfDNA fragment ends within DNase I hypersensitive sites (DHSs) was significantly increased in Dnase1-deficient mice. In humans, patients with bladder cancer had lower proportions of ucfDNA fragment ends within the DHSs when compared with participants without bladder cancer. The area under the curve (AUC) for differentiating patients with and without bladder cancer was 0.83, suggesting the analysis of ucfDNA fragmentation in the DHSs may have potential for bladder cancer detection. This work revealed the intrinsic links between the nucleases in urine and ucfDNA fragmentomics.

Epigenetic analysis of cell-free DNA by fragmentomic profiling.

Cell-free DNA (cfDNA) fragmentation patterns contain important molecular information linked to tissues of origin. We explored the possibility of using fragmentation patterns to predict cytosine-phosphate-guanine (CpG) methylation of cfDNA, obviating the use of bisulfite treatment and associated risks of DNA degradation. This study investigated the cfDNA cleavage profile surrounding a CpG (i.e., within an 11-nucleotide [nt] window) to analyze cfDNA methylation. The cfDNA cleavage proportion across positions within the window appeared nonrandom and exhibited correlation with methylation status. The mean cleavage proportion was ∼twofold higher at the cytosine of methylated CpGs than unmethylated ones in healthy controls. In contrast, the mean cleavage proportion rapidly decreased at the 1-nt position immediately preceding methylated CpGs. Such differential cleavages resulted in a characteristic change in relative presentations of CGN and NCG motifs at 5' ends, where N represented any nucleotide. CGN/NCG motif ratios were correlated with methylation levels at tissue-specific methylated CpGs (e.g., placenta or liver) (Pearson's absolute r > 0.86). cfDNA cleavage profiles were thus informative for cfDNA methylation and tissue-of-origin analyses. Using CG-containing end motifs, we achieved an area under a receiver operating characteristic curve (AUC) of 0.98 in differentiating patients with and without hepatocellular carcinoma and enhanced the positive predictive value of nasopharyngeal carcinoma screening (from 19.6 to 26.8%). Furthermore, we elucidated the feasibility of using cfDNA cleavage patterns to deduce CpG methylation at single CpG resolution using a deep learning algorithm and achieved an AUC of 0.93. FRAGmentomics-based Methylation Analysis (FRAGMA) presents many possibilities for noninvasive prenatal, cancer, and organ transplantation assessment.

Nuclease deficiencies alter plasma cell-free DNA methylation profiles.

The effects of DNASE1L3 or DNASE1 deficiency on cell-free DNA (cfDNA) methylation were explored in plasma of mice deficient in these nucleases and in DNASE1L3-deficient humans. Compared to wild-type cfDNA, cfDNA in DNASE1L3-deficient mice was significantly hypomethylated, while cfDNA in DNASE1-deficient mice was hypermethylated. The cfDNA hypomethylation in DNASE1L3-deficient mice was due to increased fragmentation and representation from open chromatin regions (OCRs) and CpG islands (CGIs). These findings were absent in DNASE1-deficient mice, demonstrating the preference of DNASE1 to cleave in hypomethylated OCRs and CGIs. We also observed a substantial decrease of fragment ends at methylated CpGs in the absence of DNASE1L3, thereby demonstrating that DNASE1L3 prefers to cleave at methylated CpGs. Furthermore, we found that methylation levels of cfDNA varied by fragment size in a periodic pattern, with cfDNA of specific sizes being more hypomethylated and enriched for OCRs and CGIs. These findings were confirmed in DNASE1L3-deficient human cfDNA. Thus, we have found that nuclease-mediated cfDNA fragmentation markedly affects cfDNA methylation level on a genome-wide scale. This work provides a foundational understanding of the relationship between methylation, nuclease biology, and cfDNA fragmentation.

Universal Targeted Haplotyping by Droplet Digital PCR Sequencing and Its Applications in Noninvasive Prenatal Testing and Pharmacogenetics Analysis.

BACKGROUND: The analysis of haplotypes of variants is important for pharmacogenomics analysis and noninvasive prenatal testing for monogenic diseases. However, there is a lack of robust methods for targeted haplotyping. METHODS: We developed digital PCR haplotype sequencing (dHapSeq) for targeted haplotyping of variants, which is a method that compartmentalizes long DNA molecules into droplets. Within one droplet, 2 target regions are PCR amplified from one template molecule, and their amplicons are fused together. The fused products are then sequenced to determine the phase relationship of the single nucleotide polymorphism (SNP) alleles. The entire haplotype of 10s of SNPs can be deduced after the phase relationship of individual SNPs are determined in a pairwise manner. We applied dHapSeq to noninvasive prenatal testing in 4 families at risk for thalassemia and utilized it to detect NUDT15 diplotypes for predicting drug tolerance in pediatric acute lymphoblastic leukemia (72 cases and 506 controls). RESULTS: For SNPs within 40 kb, phase relation can be determined with 100% accuracy. In 7 trio families, the haplotyping results for 97 SNPs spanning 185 kb determined by dHapSeq were concordant with the results deduced from the genotypes of both parents and the fetus. In 4 thalassemia families, a 19.3-kb Southeast Asian deletion was successfully phased with 97 downstream SNPs, enabling noninvasive determination of fetal inheritance using relative haplotype dosage analysis. In the NUDT15 analysis, the variant status and phase of the variants were successfully determined in all cases and controls. CONCLUSIONS: The dHapSeq represents a robust and scalable haplotyping approach with numerous clinical and research applications.

Genome-wide detection of cytosine methylation by single molecule real-time sequencing.

5-Methylcytosine (5mC) is an important type of epigenetic modification. Bisulfite sequencing (BS-seq) has limitations, such as severe DNA degradation. Using single molecule real-time sequencing, we developed a methodology to directly examine 5mC. This approach holistically examined kinetic signals of a DNA polymerase (including interpulse duration and pulse width) and sequence context for every nucleotide within a measurement window, termed the holistic kinetic (HK) model. The measurement window of each analyzed double-stranded DNA molecule comprised 21 nucleotides with a cytosine in a CpG site in the center. We used amplified DNA (unmethylated) and M.SssI-treated DNA (methylated) (M.SssI being a CpG methyltransferase) to train a convolutional neural network. The area under the curve for differentiating methylation states using such samples was up to 0.97. The sensitivity and specificity for genome-wide 5mC detection at single-base resolution reached 90% and 94%, respectively. The HK model was then tested on human-mouse hybrid fragments in which each member of the hybrid had a different methylation status. The model was also tested on human genomic DNA molecules extracted from various biological samples, such as buffy coat, placental, and tumoral tissues. The overall methylation levels deduced by the HK model were well correlated with those by BS-seq (r = 0.99; P < 0.0001) and allowed the measurement of allele-specific methylation patterns in imprinted genes. Taken together, this methodology has provided a system for simultaneous genome-wide genetic and epigenetic analyses.

Single-molecule sequencing reveals a large population of long cell-free DNA molecules in maternal plasma.

In the field of circulating cell-free DNA, most of the studies have focused on short DNA molecules (e.g., <500 bp). The existence of long cell-free DNA molecules has been poorly explored. In this study, we demonstrated that single-molecule real-time sequencing allowed us to detect and analyze a substantial proportion of long DNA molecules from both fetal and maternal sources in maternal plasma. Such molecules were beyond the size detection limits of short-read sequencing technologies. The proportions of long cell-free DNA molecules in maternal plasma over 500 bp were 15.5%, 19.8%, and 32.3% for the first, second, and third trimesters, respectively. The longest fetal-derived plasma DNA molecule observed was 23,635 bp. Long plasma DNA molecules demonstrated predominance of A or G 5' fragment ends. Pregnancies with preeclampsia demonstrated a reduction in long maternal plasma DNA molecules, reduced frequencies for selected 5' 4-mer end motifs ending with G or A, and increased frequencies for selected motifs ending with T or C. Finally, we have developed an approach that employs the analysis of methylation patterns of the series of CpG sites on a long DNA molecule for determining its tissue origin. This approach achieved an area under the curve of 0.88 in differentiating between fetal and maternal plasma DNA molecules, enabling the determination of maternal inheritance and recombination events in the fetal genome. This work opens up potential clinical utilities of long cell-free DNA analysis in maternal plasma including noninvasive prenatal testing of monogenic diseases and detection/monitoring of pregnancy-associated disorders such as preeclampsia.

Plasma DNA Profile Associated with DNASE1L3 Gene Mutations: Clinical Observations, Relationships to Nuclease Substrate Preference, and In Vivo Correction.

Plasma DNA fragmentomics is an emerging area in cell-free DNA diagnostics and research. In murine models, it has been shown that the extracellular DNase, DNASE1L3, plays a role in the fragmentation of plasma DNA. In humans, DNASE1L3 deficiency causes familial monogenic systemic lupus erythematosus with childhood onset and anti-dsDNA reactivity. In this study, we found that human patients with DNASE1L3 disease-associated gene variations showed aberrations in size and a reduction of a "CC" end motif of plasma DNA. Furthermore, we demonstrated that DNA from DNASE1L3-digested cell nuclei showed a median length of 153 bp with CC motif frequencies resembling plasma DNA from healthy individuals. Adeno-associated virus-based transduction of Dnase1l3 into Dnase1l3-deficient mice restored the end motif profiles to those seen in the plasma DNA of wild-type mice. Our findings demonstrate that DNASE1L3 is an important player in the fragmentation of plasma DNA, which appears to act in a cell-extrinsic manner to regulate plasma DNA size and motif frequency.

Detection and characterization of jagged ends of double-stranded DNA in plasma.

Cell-free DNA in plasma has been used for noninvasive prenatal testing and cancer liquid biopsy. The physical properties of cell-free DNA fragments in plasma, such as fragment sizes and ends, have attracted much recent interest, leading to the emerging field of cell-free DNA fragmentomics. However, one aspect of plasma DNA fragmentomics as to whether double-stranded plasma molecules might carry single-stranded ends, termed a jagged end in this study, remains underexplored. We have developed two approaches for investigating the presence of jagged ends in a plasma DNA pool. These approaches utilized DNA end repair to introduce differential methylation signals between the original sequence and the jagged ends, depending on whether unmethylated or methylated cytosines were used in the DNA end-repair procedure. The majority of plasma DNA molecules (87.8%) were found to bear jagged ends. The jaggedness varied according to plasma DNA fragment sizes and appeared to be in association with nucleosomal patterns. In the plasma of pregnant women, the jaggedness of fetal DNA molecules was higher than that of the maternal counterparts. The jaggedness of plasma DNA correlated with the fetal DNA fraction. Similarly, in the plasma of cancer patients, tumor-derived DNA molecules in patients with hepatocellular carcinoma showed an elevated jaggedness compared with nontumoral DNA. In mouse models, knocking out of the Dnase1 gene reduced jaggedness, whereas knocking out of the Dnase1l3 gene enhanced jaggedness. Hence, plasma DNA jagged ends represent an intrinsic property of plasma DNA and provide a link between nuclease activities and the fragmentation of plasma DNA.

Comparison of Single Molecule, Real-Time Sequencing and Nanopore Sequencing for Analysis of the Size, End-Motif, and Tissue-of-Origin of Long Cell-Free DNA in Plasma.

BACKGROUND: Recent studies using single molecule, real-time (SMRT) sequencing revealed a substantial population of analyzable long cell-free DNA (cfDNA) in plasma. Potential clinical utilities of such long cfDNA in pregnancy and cancer have been demonstrated. However, the performance of different long-read sequencing platforms for the analysis of long cfDNA remains unknown. METHODS: Size biases of SMRT sequencing by Pacific Biosciences (PacBio) and nanopore sequencing by Oxford Nanopore Technologies (ONT) were evaluated using artificial mixtures of sonicated human and mouse DNA of different sizes. cfDNA from plasma samples of pregnant women at different trimesters, hepatitis B carriers, and patients with hepatocellular carcinoma were sequenced with the 2 platforms. RESULTS: Both platforms showed biases to sequence longer (1500 bp vs 200 bp) DNA fragments, with PacBio showing a stronger bias (5-fold overrepresentation of long fragments vs 2-fold in ONT). Percentages of cfDNA fragments 500 bp were around 6-fold higher in PacBio compared with ONT. End motif profiles of cfDNA from PacBio and ONT were similar, yet exhibited platform-dependent patterns. Tissue-of-origin analysis based on single-molecule methylation patterns showed comparable performance on both platforms. CONCLUSIONS: SMRT sequencing generated data with higher percentages of long cfDNA compared with nanopore sequencing. Yet, a higher number of long cfDNA fragments eligible for the tissue-of-origin analysis could be obtained from nanopore sequencing due to its much higher throughput. When analyzing the size and end motif of cfDNA, one should be aware of the analytical characteristics and possible biases of the sequencing platforms being used.

Droplet digital PCR is a cost-effective method for analyzing long cell-free DNA in maternal plasma: Application in preeclampsia.

OBJECTIVE: Long cell-free DNA (cfDNA) can be found in the plasma of pregnant women and cancer patients. We investigated if droplet digital PCR (ddPCR) can analyze such molecules for diagnostic purposes using preeclampsia as a model. METHOD: Plasma samples from ten preeclamptic and sixteen normal pregnancies were analyzed. Two ddPCR assays targeting a single-copy gene, VCP, and one ddPCR assay targeting LINE-1 repetitive regions were used to measure the percentages of long cfDNA >533, 1001, and 170 bp, respectively. The LINE-1 assay was developed as guided by in silico PCR analyses to better differentiate preeclamptic and normal pregnancies. RESULTS: Preeclamptic patients had a significantly lower median percentage of long cfDNA than healthy pregnant controls, as determined by the LINE-1 170 bp assay (28.9% vs. 35.1%, p < 0.0001) and the VCP 533 bp assay (6.6% vs. 8.7%, p = 0.014). The LINE-1 assay provided a better differentiation than the VCP 533 bp assay (area under ROC curves, 0.94 vs. 0.79). CONCLUSION: ddPCR is a cost-effective approach for unlocking diagnostic information carried by long cfDNA in plasma and may have applications for the detection of preeclampsia. Further longitudinal studies with larger cohorts are required to assess the clinical utility of this test.

Identification and characterization of extrachromosomal circular DNA in maternal plasma.

We explored the presence of extrachromosomal circular DNA (eccDNA) in the plasma of pregnant women. Through sequencing following either restriction enzyme or Tn5 transposase treatment, we identified eccDNA molecules in the plasma of pregnant women. These eccDNA molecules showed bimodal size distributions peaking at ∼202 and ∼338 bp with distinct 10-bp periodicity observed throughout the size ranges within both peaks, suggestive of their nucleosomal origin. Also, the predominance of the 338-bp peak of eccDNA indicated that eccDNA had a larger size distribution than linear DNA in human plasma. Moreover, eccDNA of fetal origin were shorter than the maternal eccDNA. Genomic annotation of the overall population of eccDNA molecules revealed a preference of these molecules to be generated from 5'-untranslated regions (5'-UTRs), exonic regions, and CpG island regions. Two sets of trinucleotide repeat motifs flanking the junctional sites of eccDNA supported multiple possible models for eccDNA generation. This work highlights the topologic analysis of plasma DNA, which is an emerging direction for circulating nucleic acid research and applications.

Orientation-aware plasma cell-free DNA fragmentation analysis in open chromatin regions informs tissue of origin.

Cell-free DNA (cfDNA) in human plasma is a class of biomarkers with many current and potential future diagnostic applications. Recent studies have shown that cfDNA molecules are not randomly fragmented and possess information related to their tissues of origin. Pathologies causing death of cells from particular tissues result in perturbations in the relative distribution of DNA from the affected tissues. Such tissue-of-origin analysis is particularly useful in the development of liquid biopsies for cancer. It is therefore of value to accurately determine the relative contributions of the tissues to the plasma DNA pool in a simultaneous manner. In this work, we report that in open chromatin regions, cfDNA molecules show characteristic fragmentation patterns reflected by sequencing coverage imbalance and differentially phased fragment end signals. The latter refers to differences in the read densities of sequences corresponding to the orientation of the upstream and downstream ends of cfDNA molecules in relation to the reference genome. Such cfDNA fragmentation patterns preferentially occur in tissue-specific open chromatin regions where the corresponding tissues contributed DNA into the plasma. Quantitative analyses of such signals allow measurement of the relative contributions of various tissues toward the plasma DNA pool. These findings were validated by plasma DNA sequencing data obtained from pregnant women, organ transplantation recipients, and cancer patients. Orientation-aware plasma DNA fragmentation analysis therefore has potential diagnostic applications in noninvasive prenatal testing, organ transplantation monitoring, and cancer liquid biopsy.

Single-Molecule Sequencing Enables Long Cell-Free DNA Detection and Direct Methylation Analysis for Cancer Patients.

BACKGROUND: Analysis of circulating tumor DNA has become increasingly important as a tool for cancer care. However, the focus of previous studies has been on short fragments of DNA. Also, bisulfite sequencing, a conventional approach for methylation analysis, causes DNA degradation, which is not ideal for the assessment of long DNA properties and methylation patterns. This study attempted to overcome such obstacles by single-molecule sequencing. METHODS: Single-molecule real-time (SMRT) sequencing was used to sequence plasma DNA. We performed fragment size and direct methylation analysis for each molecule. A methylation score concerning single-molecule methylation patterns was used for cancer detection. RESULTS: A substantial proportion of plasma DNA was longer than 1 kb with a median of 16% in hepatocellular carcinoma (HCC) patients, hepatitis B virus carriers, and healthy individuals. The longest plasma DNA molecule in the HCC patients was 39.8 kb. Tumoral cell-free DNA (cfDNA) was generally shorter than nontumoral cfDNA. The longest tumoral cfDNA was 13.6 kb. Tumoral cfDNA had lower methylation levels compared with nontumoral cfDNA (median: 59.3% vs 76.9%). We developed and analyzed a metric reflecting single-molecule methylation patterns associated with cancer, named the HCC methylation score. HCC patients displayed significantly higher HCC methylation scores than those without HCC. Interestingly, compared to using short cfDNA (area under the receiver operating characteristic [ROC] curve, AUC: 0.75), the use of long cfDNA molecules greatly enhanced the discriminatory power (AUC: 0.91). CONCLUSIONS: A previously unidentified long cfDNA population was revealed in cancer patients. The presence and direct methylation analysis of these molecules open new possibilities for cancer liquid biopsy.

Dnase1l3 deletion causes aberrations in length and end-motif frequencies in plasma DNA.

Circulating DNA in plasma consists of short DNA fragments. The biological processes generating such fragments are not well understood. DNASE1L3 is a secreted DNASE1-like nuclease capable of digesting DNA in chromatin, and its absence causes anti-DNA responses and autoimmunity in humans and mice. We found that the deletion of Dnase1l3 in mice resulted in aberrations in the fragmentation of plasma DNA. Such aberrations included an increase in short DNA molecules below 120 bp, which was positively correlated with anti-DNA antibody levels. We also observed an increase in long, multinucleosomal DNA molecules and decreased frequencies of the most common end motifs found in plasma DNA. These aberrations were independent of anti-DNA response, suggesting that they represented a primary effect of DNASE1L3 loss. Pregnant Dnase1l3-/- mice carrying Dnase1l3+/- fetuses showed a partial restoration of normal frequencies of plasma DNA end motifs, suggesting that DNASE1L3 from Dnase1l3-proficient fetuses could enter maternal systemic circulation and affect both fetal and maternal DNA fragmentation in a systemic as well as local manner. However, the observed shortening of circulating fetal DNA relative to maternal DNA was not affected by the deletion of Dnase1l3 Collectively, our findings demonstrate that DNASE1L3 plays a role in circulating plasma DNA homeostasis by enhancing fragmentation and influencing end-motif frequencies. These results support a distinct role of DNASE1L3 as a regulator of the physical form and availability of cell-free DNA and may have important implications for the mechanism whereby this enzyme prevents autoimmunity.

Characteristics of Fetal Extrachromosomal Circular DNA in Maternal Plasma: Methylation Status and Clearance.

BACKGROUND: Although the characterization of cell-free extrachromosomal circular DNA (eccDNA) has gained much research interest, the methylation status of these molecules is yet to be elucidated. We set out to compare the methylation densities of plasma eccDNA of maternal and fetal origins, and between small and large molecules. The clearance of fetal eccDNA from maternal circulation was also investigated. METHODS: We developed a sequencing protocol for eccDNA methylation analysis using tagmentation and enzymatic conversion approaches. A restriction enzyme-based approach was applied to verify the tagmentation results. The efficiency of cell-free fetal eccDNA clearance was investigated by fetal eccDNA fraction evaluations at various postpartum time points. RESULTS: The methylation densities of fetal eccDNA (median: 56.3%; range: 40.5-67.6%) were lower than the maternal eccDNA (median: 66.7%; range: 56.5-75.7%) (P = 0.02, paired t-test). In addition, eccDNA molecules from the smaller peak cluster (180-230 bp) were of lower methylation levels than those from the larger peak cluster (300-450 bp). Both of these findings were confirmed using the restriction enzyme approach. We also observed comparable methylation densities between linear and eccDNA of both maternal and fetal origins. The average half-lives of fetal linear and eccDNA in the maternal blood were 30.2 and 29.7 min, respectively. CONCLUSIONS: We found that fetal eccDNA in plasma was relatively hypomethylated compared to the maternal eccDNA. The methylation densities of eccDNA were positively correlated with their sizes. In addition, fetal eccDNA was found to be rapidly cleared from the maternal blood after delivery, similar to fetal linear DNA.

Single Cell and Plasma RNA Sequencing for RNA Liquid Biopsy for Hepatocellular Carcinoma.

BACKGROUND: Human plasma contains RNA transcripts released by multiple cell types within the body. Single-cell transcriptomic analysis allows the cellular origin of circulating RNA molecules to be elucidated at high resolution and has been successfully utilized in the pregnancy context. We explored the application of a similar approach to develop plasma RNA markers for cancer detection. METHODS: Single-cell RNA sequencing was performed to decipher transcriptomic profiles of single cells from hepatocellular carcinoma (HCC) samples. Cell-type-specific transcripts were identified and used for deducing the cell-type-specific gene signature (CELSIG) scores of plasma RNA from patients with and without HCC. RESULTS: Six major cell clusters were identified, including hepatocyte-like, cholangiocyte-like, myofibroblast, endothelial, lymphoid, and myeloid cell clusters based on 4 HCC tumor tissues as well as their paired adjacent nontumoral tissues. The CELSIG score of hepatocyte-like cells was significantly increased in preoperative plasma RNA samples of patients with HCC (n = 14) compared with non-HCC participants (n = 49). The CELSIG score of hepatocyte-like cells declined in plasma RNA samples of patients with HCC within 3 days after tumor resection. Compared with the discriminating power between patients with and without HCC using the abundance of ALB transcript in plasma [area under curve (AUC) 0.72)], an improved performance (AUC: 0.84) was observed using the CELSIG score. The hepatocyte-specific transcript markers in plasma RNA were further validated by ddPCR assays. The CELSIG scores of hepatocyte-like cell and cholangiocyte trended with patients' survival. CONCLUSIONS: The combination of single-cell transcriptomic analysis and plasma RNA sequencing represents an approach for the development of new noninvasive cancer markers.

Jagged Ends of Urinary Cell-Free DNA: Characterization and Feasibility Assessment in Bladder Cancer Detection.

BACKGROUND: Double-stranded DNA in plasma is known to carry single-stranded ends, called jagged ends. Plasma DNA jagged ends are biomarkers for pathophysiologic states such as pregnancy and cancer. It remains unknown whether urinary cell-free DNA (cfDNA) molecules have jagged ends. METHODS: Jagged ends of cfDNA were detected by incorporating unmethylated cytosines during a DNA end-repair process, followed by bisulfite sequencing. Incorporation of unmethylated cytosines during the repair of the jagged ends lowered the apparent methylation levels measured by bisulfite sequencing and were used to calculate a jagged end index. This approach is called jagged end analysis by sequencing. RESULTS: The jagged end index of urinary cfDNA was higher than that of plasma DNA. The jagged end index profile of plasma DNA displayed several strongly oscillating major peaks at intervals of approximately 165 bp (i.e., nucleosome size) and weakly oscillating minor peaks with periodicities of approximately 10 bp. In contrast, the urinary DNA jagged end index profile showed weakly oscillating major peaks but strongly oscillating minor peaks. The jagged end index was generally higher in nucleosomal linker DNA regions. Patients with bladder cancer (n = 46) had lower jagged end indexed of urinary DNA than participants without bladder cancer (n = 39). The area under the curve for differentiating between patients with and without bladder cancer was 0.83. CONCLUSIONS: Jagged ends represent a property of urinary cfDNA. The generation of jagged ends might be related to nucleosomal structures, with enrichment in linker DNA regions. Jagged ends of urinary DNA could potentially serve as a new biomarker for bladder cancer detection.

Convolutional neural network for discriminating nasopharyngeal carcinoma and benign hyperplasia on MRI.

OBJECTIVES: A convolutional neural network (CNN) was adapted to automatically detect early-stage nasopharyngeal carcinoma (NPC) and discriminate it from benign hyperplasia on a non-contrast-enhanced MRI sequence for potential use in NPC screening programs. METHODS: We retrospectively analyzed 412 patients who underwent T2-weighted MRI, 203 of whom had biopsy-proven primary NPC confined to the nasopharynx (stage T1) and 209 had benign hyperplasia without NPC. Thirteen patients were sampled randomly to monitor the training process. We applied the Residual Attention Network architecture, adapted for three-dimensional MR images, and incorporated a slice-attention mechanism, to produce a CNN score of 0-1 for NPC probability. Threefold cross-validation was performed in 399 patients. CNN scores between the NPC and benign hyperplasia groups were compared using Student's t test. Receiver operating characteristic with the area under the curve (AUC) was performed to identify the optimal CNN score threshold. RESULTS: In each fold, significant differences were observed in the CNN scores between the NPC and benign hyperplasia groups (p < .01). The AUCs ranged from 0.95 to 0.97 with no significant differences between the folds (p = .35 to .92). The combined AUC from all three folds (n = 399) was 0.96, with an optimal CNN score threshold of > 0.71, producing a sensitivity, specificity, and accuracy of 92.4%, 90.6%, and 91.5%, respectively, for NPC detection. CONCLUSION: Our CNN method applied to T2-weighted MRI could discriminate between malignant and benign tissues in the nasopharynx, suggesting that it as a promising approach for the automated detection of early-stage NPC. KEY POINTS: • The convolutional neural network (CNN)-based algorithm could automatically discriminate between malignant and benign diseases using T2-weighted fat-suppressed MR images. • The CNN-based algorithm had an accuracy of 91.5% with an area under the receiver operator characteristic curve of 0.96 for discriminating early-stage T1 nasopharyngeal carcinoma from benign hyperplasia. • The CNN-based algorithm had a sensitivity of 92.4% and specificity of 90.6% for detecting early-stage nasopharyngeal carcinoma.