Analysis of recurrently protected genomic regions in cell-free DNA found in urine.
Author(s): Markus H, Zhao J, Contente-Cuomo T, Stephens MD, Raupach E, Odenheimer-Bergman A, Connor S, McDonald BR, Moore B, Hutchins E, McGilvrey M, de la Maza MC, Van Keuren-Jensen K, Pirrotte P, Goel A, Becerra C, Von Hoff DD, Celinski SA, Hingorani P, Murtaza M
Publication: Sci Transl Med, 2021, Vol. 13, Page
PubMed ID: 33597261 PubMed Review Paper? No
Purpose of Paper
This paper compared the size profile, genomic position, and cellular contributions of cell-free DNA (cfDNA) of plasma and urine specimens and investigated if the genomic localization of urine cfDNA differed between cancer patients and healthy controls. In urine specimens, potential effects of collection timing (first void versus subsequent collection) and delayed addition of EDTA on cfDNA yield, fragment size profile and the percentage of aberrant fragment ends were also investigated.
Conclusion of Paper
Although both plasma and urine cfDNA displayed an approximately 10 bp step pattern, plasma cfDNA had only one predominant peak at 167 bp while urine cfDNA had multiple peaks between 40 and 120 bp. The modal peak of urine cfDNA was 80-81 bp for the majority of specimens,which has previously been shown to be associated with the histone tetramer H32H42. Sequencing of chromosome 12, showed oscillations in cfDNA coverage, with peaks in the same location in urine and plasma cfDNA; however, the observed peak width was narrower in urine than plasma cfDNA. Maps of repeated protected regions (RPRs) across the genome displayed modal distance between peaks that was similar in urine and plasma specimens (177-184 bp) but overlap in peak location was limited to 52-80% between plasma specimens, 49-60% between urine pairs, and 30-62% between urine and plasma pairs. The confidence interval (relative enrichment) was significantly lower for peaks with overlap between the two specimen types than those that did not overlap.. Plasma and urine cfDNA fragments from open chromatin regions were generally shorter (more degraded) than those from closed chromatin regions. Based on DNAse I hypersensitivity, plasma cfDNA was more similar to lymphoid and myeloid cells than other cell types, while urine cfDNA was more similar to epithelial, renal epithelial, and renal cortical cells. Similarly, nucleosome-depleted cfDNA was more strongly correlated when plasma and lymphoblast cell lines were compared and between urine and epithelial cell lines (kidney and endometrial) and a lymphoblast cell line. Mapping of fragment start and end sites showed a mode of 77 bp upstream and downstream of RPRs, respectively. The fragment start and end sites had nucleotide sequence patterns in both plasma and urine specimens, but the patterns differed between them. The authors state this indicated that different enzymes are responsible for the cfDNA fragmentation observed in plasma and urine specimens. A slightly higher percentage of urine cfDNA fragments with ends localized to RPRs were identified in urine from patients with pediatric solid tumors or adenocarcinoma than in healthy controls (36.6% versus 35%), but there was no obvious difference in fragment end nucleotide frequencies between urine from cancer patients and healthy controls. However, multidimensional scaling separated urine specimens from healthy controls and cancer patients based on fragment end sequence, and area under the curve analysis was able to discriminate between healthy control and patients based on the percentage of fragments with ends that localized to RPRs and fragment end nucleotide frequencies alone or in combination. In 4 of the 6 patients who had copy number variants detected in the tumor but not urine specimen, there was a higher percentage of fragments with ends localized to RPRs for regions with gains compared to regions with no changes or losses.
cfDNA yield and fragment size profiles were comparable between the first void urine specimen and the subsequent specimen that EDTA was added to promptly (13-23 min after collection, T0) or after 30 min. However, for four of the five patients higher cfDNA levels were found in the subsequent void specimen stored for ≥ 60 min compared to T0. Fragment size distribution in the first void and subsequent void specimens overlapped regardless of storage before the addition of EDTA. Similarly, the percentage of fragments with abberent ends within recurrently protected regions (RPRs) was comparable in the first void specimen and the subsequently collected specimen regardless of when the EDTA was added.
Studies
-
Study Purpose
This study compared the size profile, genomic position, and cellular contributions in cfDNA from plasma and urine specimens and investigated if the genomic localization of urine cfDNA differed between cancer patients and healthy controls in urine and plasma specimens, respectively. K2EDTA blood was obtained from 15 healthy volunteers. Urine was collected from 30 healthy volunteers, 10 patients with non-metastatic pediatric solid tumors, and 12 patients with pancreatic adenocarcinoma. Plasma was separated by centrifugation at 820 g for 10 min followed by 16,000 g for 10 min and stored at -80°C. EDTA was added to urine prior to centrifugation at 1600 g for 10 min and storage at -80°C. cfDNA was extracted from urine using the MagMAX Cell-Free DNA Isolation Kit and from plasma using the QIAamp Circulating Nucleic Acid Kit. DNA was quantified by digital PCR. Sequencing libraries were prepared from plasma cfDNA using the ThruPLEX Tag-seq Kit and pair-end sequenced on a HiSeq 4000 and from urine cfDNA using ThruPLEX Plasma-seq Kit and pair-end sequenced on a NovaSeq 6000. Fragment size distribution and genomic coverage were analyzed using Picard tools v2.2.1.
Summary of Findings:
Plasma cfDNA had one predominant peak at 167 bp, but urine cfDNA had multiple peaks between 40 and 120 bp with the modal peak occurring at 80-81 bp for 23 of the 30 urine specimens and at 111-112 bp for the remaining 6 specimens. In both plasma and urine specimens, an approximately 10 bp step pattern (mean 10.8 bp in plasma and 9.9 bp in urine) was observed, but the authors report this was more pronounced in urine than plasma. All four nucleosomal histones were also detected in urine cfDNA. Chromosome 12 sequencing showed oscillations in cfDNA coverage, with peaks in the same location in urine and plasma cfDNA; however, the observed peak width was narrower in urine than plasma cfDNA. Maps of repeated protected regions (RPRs) across the genome also displayed a modal distance between peaks that was similar between urine and plasma specimens (177-184 bp) but overlap in peak location was limited to 52-80% between plasma pairs, 49-60% between urine pairs, and 30-62% between urine and plasma pairs. The confidence interval (relative enrichment) was significantly lower for peaks that overlapped between the two specimen types than those that did not overlap. (P<0.001). Plasma and urine cfDNA fragments from open chromatin regions were generally shorter (more degraded) than those from closed chromatin regions (P< 2x 10-16, both). Based on DNAse I hypersensitivity, plasma cfDNA was more similar to lymphoid and myeloid cells, while urine cfDNA was more similar to epithelial, renal epithelial, and renal cortical cells. Importantly, there was more between-patient variability in similarity scores when urine and cell lines were compared, which the authors state indicates more variability in the cellular contribution of cfDNA among urine specimens. Like DNAseI hypersensitivity analysis, coverage based on nucleosome-depleted regions was most strongly correlated between plasma and lymphoblast cell lines and between urine and epithelial cell lines (kidney and endometrial) and a lymphoblast cell line. When correlations with cell lines were ranked by strength, the largest differences in rank between plasma and urine occurred for monocytes (27 positions higher in plasma), and urinary bladder and renal cortical origin cell lines (33 positions higher in urine). Mapping of fragment start and end sites showed a mode of 77 bp upstream and downstream of repeated protected regions, but the slope of the distribution of ends was steep in plasma but more moderate in urine; the authors state this is consistent with the shorter mean fragment size observed in urine. Nucleotide patterns consistently occurred 10 bp upstream and 10 bp downstream of fragment start and end sites in both plasma and urine specimens, but the patterns differed between plasma and urine. The authors state this indicated that different enzymes are responsible for the cfDNA fragmentation observed in plasma and urine specimens. The percentage of fragments with ends localized to RPRs was 35% in the urine of healthy controls, but 36.6% in urine from patients with pediatric solid tumors and in patients with adenocarcinoma (P<0.01, both). While the authors report no obvious difference in fragment end nucleotides frequencies between the urine from cancer patients and healthy controls, multidimensional scaling separated urine specimens from healthy controls and cancer patients in the third dimension. Importantly, area under the curve analysis could distinguish cancer patients from healthy controls based on the percentage of fragments with ends localized to RPRs and fragment end nucleotide frequencies (0.81 and 0.85, alone respectively and 0.89 combined). Interestingly, in 4 of the 6 patients who had copy number variants detected in the tumor but not urine, there was a higher percentage of fragments with ends localized to RPRs for regions with gains compared to regions with no changes or losses (P<0.05).
Biospecimens
Preservative Types
- Frozen
Diagnoses:
- Normal
- Neoplastic - Pediatric
- Neoplastic - Carcinoma
Platform:
Analyte Technology Platform DNA Next generation sequencing Pre-analytical Factors:
Classification Pre-analytical Factor Value(s) Preaquisition Diagnosis/ patient condition Healthy
Pediatric solid tumors
Adenocarcinoma
Biospecimen Acquisition Biospecimen location Plasma
Urine
-
Study Purpose
This study investigated potential effects of collection timing (first void versus subsequent collection) and delayed addition of EDTA on cfDNA yield, fragment size profile and the percentage of aberrant fragment ends in urine specimens. First void urine specimens were collected from three healthy males and two healthy females at home in a urine cup containing EDTA. Each volunteer then collected a second specimen without an additive which was immediately divided into 5 aliquots. EDTA was added to the first aliquot after 12-23 min (T0), and to the other four aliquots after an additional 30 min, 60 min, 120 min and 240 min. Urine specimens were centrifuged at 1600 g for 10 min and stored at -80°C until cfDNA extraction. cfDNA was extracted from urine using the MagMAX Cell-Free DNA Isolation Kit. Sequencing libraries were prepared from urine cfDNA using ThruPLEX Plasma-seq kit and pair-end sequenced on NovaSeq 6000. Fragment size distribution and genomic coverage were analyzed using Picard tools v2.2.1.
Summary of Findings:
cfDNA yield and fragment size profiles were comparable between the first void urine specimen and the subsequent specimen that EDTA was added to promptly (13-23 min after collection, T0) or after 30 min. However, for four of the five patients higher cfDNA levels were found in the subsequent void specimen stored for ≥ 60 min compared to T0. Fragment size distribution in the first void and subsequent void specimens overlapped regardless of storage before the addition of EDTA. Similarly, the percentage of fragments with abberent ends within recurrently protected regions (RPRs) was comparable in the first void specimen and the subsequently collected specimen regardless of when the EDTA was added.
Biospecimens
Preservative Types
- Frozen
Diagnoses:
- Normal
Platform:
Analyte Technology Platform DNA Next generation sequencing Pre-analytical Factors:
Classification Pre-analytical Factor Value(s) Storage Storage duration 0 h
30 min
60 min
120 min
240 min
Biospecimen Acquisition Time of biospecimen collection First void
Subsequent void