Dutch National Round Robin Trial on Plasma-Derived Circulating Cell-Free DNA Extraction Methods Routinely Used in Clinical Pathology for Molecular Tumor Profiling.
Author(s): van der Leest P, Ketelaar EM, van Noesel CJM, van den Broek D, van Boerdonk RAA, Deiman B, Rifaela N, van der Geize R, Huijsmans CJJ, Speel EJM, Geerlings MJ, van Schaik RHN, Jansen MPHM, Dane-Vogelaar R, Driehuis E, Leers MPG, Sidorenkov G, Tamminga M, van Kempen LC, Schuuring E
Publication: Clin Chem, 2022, Vol. , Page
PubMed ID: 35616097 PubMed Review Paper? No
Purpose of Paper
The purpose of this paper was to investigate the potential effects of extraction method on cell-free DNA (cfDNA) yield, fragment size, and the number of detected mutations using plasma specimens. Case-matched plasma specimens collected from patients diagnosed with metastatic lung cancer were processed by 14 different laboratories with unique standard workflows that differed from one another in terms of plasma volume (2-4 mL), extraction kit (six different kits used), and elution volume (25-100 µL) to isolate cfDNA from each of four plasma specimens; isolated cfDNA was then mailed to a central laboratory for analysis.
Conclusion of Paper
The majority of extracts with a higher than average (more than one CV above the median) number of mutant molecules detected or a higher than average cfDNA yield were isolated using a silica membrane-based extraction method (QIAamp Circulating Nucleic Acid Kit or Cobas ccfDNA Sample Preparation Kit). In contrast, extracts with a lower than average number of detected mutant molecules or a lower cfDNA yield were more often isolated when a magnetic bead-based extraction method was used (Maxwell RSC LV ccfDNA Kit, MagNa Pure 24 Total NA Isolation Kit, QIAsymphony DNA Mini Kit or the QIAamp MinElute ccfDNA Mini Kit). Specimens extracted with a magnetic-bead based method made up a higher percentage of specimens with both higher and lower than average ratios of long to short cfDNA, indicating no effect of extraction method on integrity. When normalized to the median, a significantly higher number of mutant copies and a higher cfDNA yield was detected in cfDNA isolated with a silica-membrane based kit than a magnetic bead-based kit, but the variant allele frequency (VAF) was not significantly different between extraction methods. Of the specimens with a higher than average number of mutant copies, only a small percentage had a cfDNA yield that was higher than average or a higher than average ratio of long to short fragments, indicating the number of detected mutant molecules is not predictive of cfDNA yield or integrity. However, a higher yield of cfDNA was associated with a higher number of mutant copies detected in multilevel linear regression analysis. There was no effect of cfDNA integrity (size ratio or Tapestation traces) between specimens with a higher versus lower than average number of mutant molecules detected. A higher elution volume and use of 2 mL plasma rather than 3 or 4 mL yielded the highest cfDNA yield/mL plasma, and use of 3 mL of plasma rather than 2 increased the ratio of long to short cfDNA fragments.
Studies
-
Study Purpose
The purpose of this study was to investigate the potential effects of extraction method on cell-free DNA (cfDNA) yield, fragment size, and number of detected mutations. Case-matched plasma specimens were collected from four patients with metastatic non-small cell lung cancer into citrate bags. cfDNA was isolated from matched diagnostic leukapheresis plasma aliquots by 14 different laboratories using workflows that differed from one another in plasma volume (2-4 mL), extraction kit (six different kits used), and elution volume (25-100 µL). Plasma was immediately aliquoted and frozen at -80°C. Patients were selected for inclusion based on having detectable levels of driver mutations in plasma. Plasma was thawed, centrifuged at 1600 g for 10 min, aliquoted for shipment and stored at −80°C. DNA was extracted from some aliquots by the central laboratory using the QIAamp Circulating Nucleic Acid Kit. The remaining aliquots were shipped on dry ice to fourteen different laboratories each of which extracted cfDNA from 2-4 mL of plasma using one of the following methods: QIAamp Circulating Nucleic Acid Kit (4 laboratories), Maxwell RSC LV ccfDNA Kit (4 laboratories), MagNa Pure 24 Total NA Isolation Kit (1 laboratory), Cobas ccfDNA Sample Preparation Kit (2 laboratories), QIAsymphony DNA Mini Kit (1 laboratory), or the QIAamp MinElute ccfDNA Mini Kit (2 laboratories). Isolated cfDNA was stored at -20°C and then mailed on dry ice back to the central laboratory. One laboratory (extraction with the Maxwell RSC LV ccfDNA Kit) stored cfDNA at 4°C after extraction. One shipment experienced longer shipping delays and thawed during shipment and was subsequently stored at 4°C (extraction was with the QIAamp Circulating Nucleic Acid Kit) and cfDNA from two laboratories were not frozen upon arrival (QIAamp Circulating Nucleic Acid Kit and QIAamp MinElute ccfDNA Mini Kit). cfDNA yield was quantified at the central laboratory using Qubit dsDNA HS assays and MassARRAY LiquidIQ. cfDNA fragment size was analyzed by ddPCR amplification of 137, 420, and 1950 bp fragments of β-actin and by TapeStation. Mutations were detected using ddPCR assays for KRAS G12/G13, TP53 p.R273H TP53 p.H179R, and EGFR p. T790M. Lower than average was defined as more than one coefficient of variance (CV) below the median. Higher than average was defined as more than one CV higher than the median.
Summary of Findings:
When cfDNA was extracted by the central laboratory using the QIAamp Circulating Nucleic Acid Kit the mean coefficient of variance (CV) for the number of mutant molecules detected was 40% and the mean CV for variant allele frequency was 33%. Twelve of the 56 extractions had a higher than average number of mutant molecules detected and the majority (9 of 12) of these were isolated using a silica membrane-based method (QIAamp Circulating Nucleic Acid Kit or Cobas ccfDNA Sample Preparation Kit). In contrast, of the 15 extractions with a lower than average number of molecules detected, 11 were obtained using a magnetic bead- based method (Maxwell RSC LV ccfDNA Kit, MagNa Pure 24 Total NA Isolation Kit, QIAsymphony DNA Mini Kit or the QIAamp MinElute ccfDNA Mini Kit). When normalized to the plasma specimen’s median number of mutant copies detected, a significantly higher number of mutant copies was detected in cfDNA isolated using silica-membrane-based kits than magnetic bead-based kits (mean of 1.29 versus 0.76, P<0.01), but the VAF was not significantly different between extraction methods. Significantly lower numbers of mutant cfDNA copies were detected in cfDNA extracted using the Maxwell RSC LV ccfDNA Kit than the QIAamp Circulating Nucleic Acid Kit (P<0.001). All 10 of the extracts with a higher than average total cfDNA yield were obtained with a silica membrane-based method and 7 of 8 specimens with cfDNA yields that were below average were extracted with magnetic-bead based methods. When normalized to the plasma specimens median yield, a significantly higher yield was obtained using silica-membrane based kits than magnetic bead-based kits (mean of 1.21 versus 0.86, P<0.0001). The cfDNA yield was significantly higher when the QIAamp Circulating Nucleic Acid Kit was used compared to any of the other kits evaluated, with the exception of the QIAamp MinElute ccfDNA Mini Kit. While 67% (6 of 9 specimens) of the specimens with a ratio of long (1950 bp) to short (137 bp) cfDNA fragments that were greater than one CV above the median, 67% (6) were obtained with a magnetic bead-based method, 57% (8 of 14) of the specimens with a ratio that was lower than average were also obtained using magnetic bead-based methods. Importantly, these percentages were comparable to the total percentage of specimens extracted with a magnetic bead-based method (57%) and no significant effect of extraction method on the ratio of long to short cfDNA was found. Interestingly, of the specimens with a higher than average number of mutant copies, only 27% had a cfDNA yield that was higher than average and 25% had a higher than average ratio of long to short fragments. Only 21% of specimens with a lower than average number of mutant copies had lower than average cfDNA yield. However, a higher yield of cfDNA was associated with a higher number of detected mutant copies in multilevel linear regression analysis (P=0.003). There was no effect of cfDNA integrity (size ratio or Tapestation traces) between specimens with a higher versus lower than average number of detected mutant molecules. A higher elution volume (P=0.004) and use of 2 mL plasma rather than 3 (P=0.003) or 4 mL (P=0.032) resulted in the highest cfDNA yield/mL plasma. A smaller long to short ratio was found when extraction was from 3 mL plasma rather than 2 mL (P=0.042).
Biospecimens
Preservative Types
- Frozen
Diagnoses:
- Neoplastic - Carcinoma
Platform:
Analyte Technology Platform DNA MS/MS DNA Digital PCR DNA Capillary electrophoresis-MS DNA Fluorometry DNA Automated electrophoresis/Bioanalyzer Pre-analytical Factors:
Classification Pre-analytical Factor Value(s) Biospecimen Aliquots and Components Aliquot size/volume 2 mL
3 mL
4 mL
Analyte Extraction and Purification Analyte isolation method QIAamp Circulating Nucleic Acid Kit
Maxwell RSC LV ccfDNA Kit
MagNa Pure 24 Total NA Isolation Kit
Cobas ccfDNA Sample Preparation Kit
QIAsymphony DNA Mini Kit
QIAamp MinElute ccfDNA Mini Kit
Analyte Extraction and Purification Rehydration of dried sample/specimen 25-100µL