Cancer Diagnosis Program | Biorepositories and Biospecimen Research Branch

Search BRD Papers

High-throughput and affordable genome-wide methylation profiling of circulating cell-free DNA by methylated DNA sequencing (MeD-seq) of LpnPI digested fragments.

Author(s): Deger T, Boers RG, de Weerd V, Angus L, van der Put MMJ, Boers JB, Azmani Z, van IJcken WFJ, Grünhagen DJ, van Dessel LF, Lolkema MPJK, Verhoef C, Sleijfer S, Martens JWM, Gribnau J, Wilting SM

Publication: Clin Epigenetics, 2021, Vol. 13, Page 196

PubMed ID: 34670587 PubMed Review Paper? No

Purpose of Paper

Conclusion of Paper

Studies

1. Study Purpose

   This study compared MeD-Sequencing quality metrics and methylation profiles of cfDNA isolated from  blood collected in EDTA and CellSave tubes, among cfDNA isolated using three different kits, and cfDNA eluted into different buffers with and without subsequent vacuum concentration. The authors also established a minimum cfDNA input amount and compared methylation profiles in specimens obtained from patients with metastatic colorectal cancer pre- and post-surgery  and from specimens collected from healthy volunteers. Unless otherwise specified, blood was collected from patients with metastatic colorectal cancer into CellSave tubes, and plasma was separated within 4 days by centrifugation at 1711 g for 10 min followed by 12,000 g for 10 min. Plasma was stored at -80°C until cfDNA isolation using the QIAamp Circulating Nucleic Acid Kit. cfDNA was digested with LpnPI and libraries were constructed using the ThruPLEX DNAseq 96 D Kit and sequenced using an Illumina HiSeq 2500 machine. Mutations in KRAS, TP53, and PIK3CA were detected by digital PCR.  For the comparison of tube type, blood was collected from three patients into both EDTA and CellSave tubes. DNA was extracted using the QIAamp Kit and sequencing libraries were prepared from 10 ng DNA or 8 µL DNA (>30 ng). To identify effects of extraction method, DNA was extracted from the plasma of 37 patients with the semi-automated QiaSymphony DSP Circulating DNA Kit and from the plasma of two patients using the Maxwell RSC ccfDNA Plasma Kit, the QIAamp Kit, and the QiaSymphony Kit. To identify effects of elution buffer choice and the vacuum concentration of cfDNA, cfDNA was extracted from the plasma of a metastatic breast cancer patient using the QIAamp Kit and eluted in water, QIAamp’s AVE buffer, QIAsymphony’s ATE buffer, and the Maxwell Kit buffer and sequenced with or without vacuum concentration. To evaluate the use of Me-D Sequencing for patient monitoring, plasma was obtained from nine healthy volunteers and from 8 patients with metastatic colorectal cancer before and 5-days after surgery.

   Summary of Findings:

   The percentage of duplicate reads and reads that passed the LpnPI filter were comparable in plasma from EDTA and CellSave tubes and specimens clustered by patient source not by tube type in principal component analysis. Further, correlations between specimens from the same patient were higher than correlations with specimens from nine healthy volunteers (P<0.005).  The MeD-Sequencing assay was equally successful when 10 ng or >30 ng DNA was used as input and specimens from the same patient clustered together regardless of input volume. However, in the follow-up analysis of 51 specimens, the 4 specimens with <10 ng DNA produced libraries of insufficient concentration for sequencing. In contrast, only 1 of the 34 specimens with 10 ng DNA and 1 of the 13 specimens with >10 ng DNA failed to yield libraries of sufficient concentration for the assay.

   cfDNA extracted using the automated QIAsymphony instrument had fewer reads that passed the LpnPI filter and more duplicate reads than non-matched cfDNA samples that were extracted using the QIAamp Kit (P<0.001).  Further, most of the cfDNA samples that were extracted using the automated QIAsymphony instrument did not meet the threshold of 10% of reads that passed the LpnP1 filter in the first 2 million reads.  When cfDNA was extracted from two plasma specimens using the Maxwell RSC, QIAamp, and  QiaSymphony platforms, the QIAsymphony Kit did not perform as well as the other two methods (one specimen had insufficient cfDNA yield and the other specimen had a lower percentage of reads that passed the LpnP1 filter and more duplicates).  Methylation profiles were strongly correlated between cfDNA isolated using QIAamp and Maxwell Kits in both plasma specimens (r>0.8) but were only modestly correlated between QiaSymphony and QIAamp Kits (r=0.64) or Maxwell extracted cfDNA (r=0.62) in the sample that yielded sufficient cfDNA for MeD-Sequencing with the QiaSymphony method. When cfDNA was isolated from MCF-7 cells using the QIAamp Kit, the percentage of duplicates and reads that passed filters were comparable when cfDNA was eluted using water, QIAamp’s AVE buffer, QiaSymphony’s ATE buffer, and the Maxwell elution buffer.  Vacuum concentration of cfDNA in the Maxwell elution buffer led to a lower percentage of reads that passed the filter compared to unfiltered cfDNA eluted in Maxwell (or other) buffer, but the percentage of reads that passed filter was not affected by the concentration of DNA eluted in other buffers. Further, correlations between vacuum-concentrated and unconcentrated specimens were strong when cfDNA was eluted in water (r=0.73) or AVE buffer (r=0.72) but modest when eluted in Maxwell buffer (r=0.48). The authors attributed the lower percentage of reads that passed filter and weaker correlations following concentration of cfDNA in Maxwell buffer to an increase in EDTA concentration.

   MeD-Sequencing profiles of cfDNA isolated from plasma that was obtained from eight patients with colorectal liver metastases pre-surgery clustered separately from specimens obtained from healthy volunteers, but the methylation patterns of cfDNA obtained from the cancer patients five days after surgery were more like those of healthy volunteers. Similarly, Z scores were higher in cfDNA from pre-surgical blood specimens from colorectal cancer patients relative to cfDNA isolated from the blood of healthy volunteers (average 31.52), although Z-scores were much lower in blood specimens collected from colorectal cancer patients 5 days post-surgery (0.06). The summarized Z-scores for cfDNA obtained before surgery were strongly correlated with the variant allele frequencies for KRAS, TP53, and PIK3CA mutations (r=0.90, P=0.006). Finally, methylation of EYA4, MSC, and ITGA4 were all significantly higher in cfDNA isolated from plasma obtained pre-surgery than post-surgery (P<0.01, all).

   Biospecimens

   • Bodily Fluid - Plasma
   • Bodily Fluid - Blood

   Preservative Types

   • Frozen

   Diagnoses:

   • Normal
   • Neoplastic - Carcinoma

   Platform:

   Analyte	Technology Platform
   DNA	Digital PCR
   DNA	DNA sequencing

   Pre-analytical Factors:

   Classification	Pre-analytical Factor	Value(s)
   Preaquisition	Diagnosis/ patient condition	Metastatic colorectal cancer Healthy
   Biospecimen Acquisition	Time of biospecimen collection	Pre-surgery 5 days post-surgery
   Analyte Extraction and Purification	Analyte purification	Vacuum concentrated Not vacuum concentrated
   Analyte Extraction and Purification	Analyte isolation method	QiaSymphony Kit QIAamp Kit Maxwell RSC kit
   Analyte Extraction and Purification	Rehydration of dried sample/specimen	Water QIAamp’s AVE buffer QIAsymphony’s ATE buffer Maxwell kit buffer
   Biospecimen Acquisition	Type of collection container/solution	CellSave tube EDTA tube

   View more