Single-molecule methylation profiles of cell-free DNA in cancer with nanopore sequencing

Abstract

Epigenetic characterization of cell-free DNA (cfDNA) is an emerging approach for detecting and characterizing diseases such as cancer. We developed a strategy using nanopore-based single-molecule sequencing to measure cfDNA methylomes. This approach generated up to 200 million reads for a single cfDNA sample from cancer patients, an order of magnitude improvement over existing nanopore sequencing methods. We developed a single-molecule classifier to determine whether individual reads originated from a tumor or immune cells. Leveraging methylomes of matched tumors and immune cells, we characterized cfDNA methylomes of cancer patients for longitudinal monitoring during treatment.

Keywords: Cell-free DNA; Methylation; Nanopore; Single-molecule sequencing.

Fig. 1

Nanopore sequencing of cfDNA. A An optimized protocol for generating cfDNA sequencing libraries enables high-throughput methylation characterization. The key improvement was the optimization of specific end-repair, a-tailing, and ligation conditions to maximize the number of cfDNA fragments available for nanopore sequencing. B Cell-free DNA library comparison. An optimized workflow enables approximately an order of magnitude increase in sequencing yield versus the conventional protocol. C Sequencing yield correlation with input cfDNA. Fluorometric quantification was performed on cancer patient-derived cfDNA samples and compared to the aligned sequencing yield. Each patient is shown as a separate color in triplicate. Correlation and significance value are annotated on the plot. D Genome-wide methylation quantification. The degree of methylation across the genome was computed for healthy and patient-derived cfDNA. E Nucleosome enrichment analysis. The ratio of mononucleosomes to di-nucleosomes was quantified for each tissue type, using a cutoff of 250 bp between mono- and di-nucleosomes. F Distribution of fragment sizes. Example fragment sizes are shown for healthy and patient-derived cfDNA. Mono- and di-nucleosome size peaks are annotated with dotted lines to be 167 bp and 334 bp. G Methylation profiles of healthy- and patient-derived cfDNA. Gene-level methylation values for each sample were determined, and statistically significant ones (q < 0.01) are plotted as a heatmap with the gene-level methylation percentage as the intensities. The heatmap was clustered by gene-level methylation. H Differential methylation. Statistically significant differences in methylation between sample types are shown for several selected genes

Fig. 2

Single-molecule methylated sequence classification. A Overview of method. For a given patient, matched primary tumor tissue and peripheral leukocytes were obtained as reference samples alongside longitudinal plasma samples. Methylation data from the cfDNA is then classified leveraging the methylation profile of the reference samples. B Classification accuracy. We used GP2D and healthy donor-derived nucleosome mixtures to validate the classification procedure. ROC curves are plotted, where each curve represents a distinct immune threshold score. The curve is plotted by varying the cancer threshold score. C Admixture validation. The proportion of reads classified as belonging to cell line reference is plotted as a function of the actual admixture ratio and sequencing depth. D Longitudinal methylation profiles of patient-derived cfDNA in colorectal cancer. The overall cfDNA sequencing yield (upper panel) is plotted against the number of reads with methylation profiles matching that of the matched tumor with a calculated score of > 0.9 (lower panel). Clinically relevant events are annotated; significant corresponding changes in tumor-specific cfDNA are marked with an asterisk. CT refers to computed tomography imaging

Fig. 3

Longitudinal methylation profiles of patient-derived cfDNA in other malignancies. A Assessing tumor burden in patient P4822 with metastatic pancreatic neuroendocrine carcinoma. The overall cfDNA sequencing yield (top) is plotted against the number of reads with methylation profiles matching the primary tumor with a tumor score of > 0.9 (bottom). Clinically relevant events are annotated. Clinically relevant events are annotated. Everolimus and peptide receptor radionuclide therapy (PRRT) was used for treatment of the metastatic neuroendocrine cancer. Positron-emission tomography (PET) was combined with CT imaging. B Longitudinal gene-level analysis of cfDNA changes in P4822. The number of tumor-specific differentially methylated genes found to be matching in cfDNA is shown for each time point. Differentially methylated genes were identified as those with the largest difference in methylation between the primary tumor and immune cells. Such methylated genes observed in cfDNA are defined as matching the primary tumor when its methylation state (e.g., hypermethylation or hypomethylation) is concordant. Specific time points are annotated with asterisks to denote clinical events with significant changes in methylation. C Assessing tumor burden in patient P6527 with metastatic cholangiocarcinoma. The overall cfDNA sequencing yield (top) is plotted against the number of reads with methylation profiles matching the primary tumor with a tumor score of > 0.9 (bottom). Clinically relevant events are annotated. Treatment included gemcitabine and a chemotherapy combination of 5-fluouracil and oxaliplatin (FOLFOX) was used for treatment of the cholangiocarcinoma. D Longitudinal gene-level analysis of cfDNA changes in P6527. The number of tumor-specific differentially methylated genes found to be matching in cfDNA is shown for each time point. Differentially methylated genes were identified as those with the largest difference in methylation between the primary tumor and immune cells. Such methylated genes observed in cfDNA are defined as matching the primary tumor when its methylation state (e.g., hypermethylation or hypomethylation) is concordant. Specific time points are annotated with asterisks to denote clinical events with significant changes in methylation