This paper compared the cell-free DNA (cfDNA) yields obtained from cerebrospinal fluid (CSF) using four different kits and investigated the effects of CSF volume and cfDNA input on sequencing depth. The incidence of genetic alterations was compared among patients with CSF tumors of different types and between patients with positive and negative cytology. When possible, the mutations identified in CSF were compared to those in matched formalin-fixed paraffin-embedded (FFPE) tumor specimens.
Evaluation of the Oncomine Pan-Cancer Cell-Free Assay for Analyzing Circulating Tumor DNA in the Cerebrospinal Fluid in Patients with Central Nervous System Malignancies.
Shah M, Takayasu T, Zorofchian Moghadamtousi S, Arevalo O, Chen M, Lan C, Duose D, Hu P, Zhu JJ, Roy-Chowdhuri S, Riascos RF, Chen H, Luthra R, Esquenazi Y, Ballester LY
Purpose
Conclusion
The cfDNA yield from CSF was highest when extraction was with the QIAamp Circulating Nucleic Acid Kit. As the CSF volume used increased, the cfDNA yield increased non-significantly but there was no effect on sequencing depth. However, the sequencing depth was higher when 20 ng cfDNA was used as opposed to <20 ng. More mutations were identified when sequenced at >10,000X than <10,000X. Based on the presence of low level amplifications (1.3- to 1.65-fold, 0.07% to 0.21%) and low frequency SNVs (0.07% to 0.21%) in the CSF of the patients without CNS tumors, CNV and SNV thresholds were set at ≥1.7-fold and ≥0.3%, respectively. Genetic alteration(s) were identified 15 of the 38 CSF specimens from patients with CNS tumors, including 18 SNVs in 14 patients and six copy number alterations in five breast cancer patients. Genetic alterations were identified in 11 of 17 CSF specimens from patients with metastatic breast cancer, three of eight patients with metastatic lung cancer, and in the one patient with uterine cancer. Importantly, no genetic alterations were identified in the 10 patients with primary CNS tumors. The SNVs identified were found in eight genes, of which, TP53 was most common and the CNVs were found in three genes (CCND1, MYC, and ERBB2). ddPCR confirmed all five mutations evaluated. Matched tumor sequencing data was available for eight of the specimens with genomic alterations observed in the CSF, two of which showed perfect concordance (both cases with ERBB amplification) and partial concordance (some but not all mutation detected) was observed in the remaining six patients. Importantly, 10 of the 15 mutations observed in tumor specimens were also found in CSF cfDNA and all tumor CNVs were found in matched CSF cfDNA. Mutations were found in eight of 10 CSF specimens with positive cytology but only in six of 24 with negative CSF cytology.
Studies (1)
Purpose
This study compared the cfDNA yields obtained from CSF using four different kits and investigated the effects of CSF volume and cfDNA input on sequencing depth. The incidence of genetic alterations was compared among patients with CSF tumors of different types and between patients with positive and negative cytology. When possible, the mutations identified in CSF were compared to those in matched FFPE tumor specimens. CSF was obtained from a total of 38 patients with CNS tumors including 10 with primary CNS tumors, 17 with metastatic breast cancer, eight with lung cancer, and one each with ovarian cancer, ocular melanoma, and uterine cancer and from ten patients without a history of CNS malignancy. The CSF specimens were collected before treatment from eight patients and after treatment or during follow-up from 30 patients. Matched FFPE tumor specimens (no processing details provided) were obtained from seven of the patients. To compare extraction kits, cfDNA was extracted from CSF of 10 patients using the QIAamp Circulating Nucleic Acid, Magmax Cell-free Total Nucleic Acid, QIAsymphony DSP Circulating DNA, and Promega Maxwell RSC Circulating Cell-free DNA kits but all remaining cfDNA extractions were with the QIAamp Circulating Nucleic Acid Kit. DNA was extracted from tumor enriched areas of ten 5 µm Hemo-De deparaffinized sections of FFPE tumors using the AllPrep DNA/RNA FFPE Kit. cfDNA was evaluated using a TapeStation. NGS libraries were constructed using the Oncomine Pan-cancer Cell-free Assay, high-depth sequenced on the Ion S5 XL sequencer and analyzed with Torrent Suite Software v.5.2 and IonReporter versions 5.6 and 5.10. CNV and SNV thresholds were set at ≥1.7-fold and ≥0.3%. Mutations were confirmed by ddPCR using a AKT1 p.E17K and PIK3CA p.H1047R assays, respectively. CSF cytology methodology was not described.
Summary of Findings
The cfDNA yield from CSF was highest when extraction was with the QIAamp Circulating Nucleic Acid Kit (mean 47.4 ng) followed by the Maxwell RSC System (mean 32.0 ng) and QIAsymphony DSP Circulating DNA kit (mean 21.2 ng) and lowest with the Magmax Cell-free Total Nucleic Acid Kit (mean 21.0 ng). There was a non-significant trend to increased cfDNA yield with increased CSF volume (P=0.06). While there was no relationship between sequencing depth and CSF volume, the sequencing depth was higher when 20 ng cfDNA was used as opposed to <20 ng (38,321X versus 18,132X). Mutations identified in 26.3% of specimens sequenced at <10,000X and 52.6% of specimens sequenced at >10,000X, clearly showing the importance of starting with sufficient CSF. Based on the presence of low level amplifications (1.3- to 1.65-fold, 0.07% to 0.21%) and low frequency SNVs (0.07% to 0.21%) in the CSF of the patients without CNS tumors, CNV and SNV thresholds were set at ≥1.7-fold and ≥0.3%, respectively. Genetic alteration(s) were identified in 15 of the 38 CSF specimens from patients with CNS tumors including 18 SNVs in 14 patients and six copy number alterations in five breast cancer patients. Genetic alterations were identified in 11 of 17 CSF specimens from patients with metastatic breast cancer, three of the eight patients with metastatic lung cancer, and in the one patient with uterine cancer. Importantly, no genetic alterations were identified in the 10 patients with primary CNS tumors. The SNVs identified were found in eight genes, of which, TP53 was most common and the CNVs were found in three genes (CCND1, MYC, and ERBB2). ddPCR confirmed all five mutations evaluated. Matched tumor sequencing data was available for eight of the specimens with genomic alterations observed in the CSF, two of which showed perfect concordance (both cases with ERBB amplification) and partial concordance (some but not all mutation detected) was observed in the remaining six patients. Importantly, 10 of the 15 mutations observed in tumors were also found in CSF cfDNA and all tumor CNVs were found in matched CSF cfDNA but three additional SNVs and two additional CNVs not found in the matched tumor specimen were found in CSF. Mutations were found in eight of 10 CSF specimens with positive cytology but in only six of 24 with negative CSF cytology.
Biospecimens
FluidCerebrospinal FluidTissueNervePreservation Types
None (Fresh)Diagnoses
NeoplasticCarcinomaNeoplasticNot specifiedNeoplasticMelanomaNot specifiedPlatforms
DNAAutomated electrophoresis/BioanalyzerDNADigital PCRDNANext generation sequencingPre-analytical Factors
Classification Pre-Analytical Factor Value(s) Preaquisition Diagnosis/ patient condition Metastatic breast cancerMetastatic ocular melanomaMetastatic ovarian cancerMetastatic uterine cancerMetastic lung cancerNo history of CNS malignancyPrimary CNS tumorBiospecimen Acquisition Biospecimen location CSFTumorBiospecimen Aliquots and Components Aliquot size/volume 0.7-4.9 mLAnalyte Extraction and Purification Analyte isolation method Magmax Cell-free total Nucleic acid kitPromega Maxwell RSC circulating cell-free DNA KitQIAamp Circulating Nucleic Acid kitQIAsymphony DSP Circulating DNA KitNext generation sequencing Specific Targeted nucleic acid 20 ng cfDNA<20 ng cfDNANext generation sequencing Specific Technology platform ddPCR