Cell-Free DNA Variant Sequencing Using CTC-Depleted Blood for Comprehensive Liquid Biopsy Testing in Metastatic Breast Cancer

Abstract

Liquid biopsy analytes such as cell-free DNA (cfDNA) and circulating tumor cells (CTCs) exhibit great potential for personalized treatment. Since cfDNA and CTCs are considered to give additive information and blood specimens are limited, isolation of cfDNA and CTC in an "all from one tube" format is desired. We investigated whether cfDNA variant sequencing from CTC-depleted blood (CTC-depl. B; obtained after positive immunomagnetic isolation of CTCs (AdnaTest EMT-2/Stem Cell Select, QIAGEN)) impacts the results compared to cfDNA variant sequencing from matched whole blood (WB). Cell-free DNA was isolated using matched WB and CTC-depl. B from 17 hormone receptor positive/human epidermal growth factor receptor 2 negative (HR+/HER2-) metastatic breast cancer patients (QIAamp MinElute ccfDNA Kit, QIAGEN). Cell-free DNA libraries were constructed (customized QIAseq Targeted DNA Panel for Illumina, QIAGEN) with integrated unique molecular indices. Sequencing (on the NextSeq 550 platform, Illumina) and data analysis (Ingenuity Variant Analysis) were performed. RNA expression in CTCs was analyzed by multimarker quantitative PCR. Cell-free DNA concentration and size distribution in the matched plasma samples were not significantly different. Seventy percent of all variants were identical in matched WB and CTC-depl. B, but 115/125 variants were exclusively found in WB/CTC-depl. B. The number of detected variants per patient and the number of exclusively detected variants per patient in only one cfDNA source did not differ between the two matched cfDNA sources. Even the characteristics of the exclusively detected cfDNA variants in either WB or CTC-depl. B were comparable. Thus, cfDNA variants from matched WB and CTC-depl. B exhibited no relevant differences, and parallel isolation of cfDNA and CTCs from only 10 mL of blood in an "all from one tube" format was feasible. Matched cfDNA mutational and CTC transcriptional analyses might empower a comprehensive liquid biopsy analysis to enhance the identification of actionable targets for individual therapy strategies.

Keywords: cell-free DNA; circulating tumor cells; liquid biopsy; metastatic breast cancer; next-generation sequencing.

Conflict of interest statement

C.K., M.T., P.M., O.H., and R.K. declare that they have no competing interests. M.S., S.H., P.H., and M.S.H. are employees at QIAGEN, Hilden, Germany. S.K.B. is a consultant for QIAGEN, Hilden, Germany.

Figures

Figure 1

Inter-individual variability of cell-free DNA (cfDNA) concentration. Plasma from circulating tumor cell (CTC)-depleted (depl.)_blood (by AdnaTest EMT-2/StemCell Select) and matched plasma from whole blood were used in the same volume for cfDNA isolation with a QIAamp MinElute ccfDNA Kit, and cfDNA was quantified using an Agilent High Sensitivity Chip (fragments between 100–700 bp).

Figure 2

Size distribution of cfDNA. Matched cfDNA samples isolated from CTC-depl. blood (B,D; CTC isolation using AdnaTest EMT-2/StemCell Select) and from whole blood (A,C) of two exemplary patients (A + B; C + D) displayed a large mononucleosomal fraction and, in general, a similar size distribution without high-molecular-weight DNA (700–10,000 bp). Capillary electrophoresis was performed with an Agilent High Sensitivity Chip. (A) Cell-free DNA eluate from whole blood of patient 2, diluted 1:40; (B) cfDNA eluate from CTC-depl. blood of patient 2, diluted 1:40; (C) cfDNA eluate from whole blood of patient 3, diluted 1:5; (D) cfDNA eluate from CTC-depl. blood of patient 3, diluted 1:20.

Figure 3

Venn diagram (A) and cross table (B) of matched cfDNA variants isolated from whole blood (light gray) and CTC-depleted blood (dark gray). Variants in all exonic regions of 17 genes were examined in 12 metastatic breast cancer (MBC) patients with hormone receptor positive/human epidermal growth factor receptor 2 negative (HR+/HER2−) primary tumor. Depletion of CTCs was conducted using AdnaTest EMT-2/StemCell Select. Variant calling by verification of unique molecular indices and filter of the Ingenuity Variant Analysis were described previously [25]. The proportional Venn diagram was computed using the tool BioVenn [26] and displays a great overlap of identical variants found in both cfDNA sources.

Figure 4

Boxplots describing the number of (A) detected cfDNA variants, and (B) the number of exclusively detected cfDNA variants per patient in whole blood (gray) and CTC-depleted blood (black, using AdnaTest EMT-2/StemCell Select). Means and standard deviations are also displayed, and the Wilcoxon signed-rank test indicated no significant difference between both cfDNA sources.

Figure 5

Boxplots describing the number of exclusively detected cfDNA variants per patient with specific characteristics in whole blood (gray) and CTC-depleted blood (black, using AdnaTest EMT-2/StemCell Select). The averages and standard deviations are indicated. Depicted characteristics are the major parameters of the categories: translation impact (frameshift (A)), classification (pathogenic and likely pathogenic (B), uncertain significance (C)), gene (AR (D), BRCA2 (E), and MUC16 (F)), gene region (exonic (G)), and allele frequency (<1% (H), 1–5% (I)). The Wilcoxon signed-rank test indicated no significant difference between the 12 matched cfDNA sources regarding any depicted characteristic.

Figure 6

Distribution of called cfDNA variants isolated from whole blood (A–D) and CTC-depleted blood (E–H) of HR+/HER2− MBC patients (n = 12) according to their gene location (A,D,E,H), allele frequency (B,F), and classification (C,G). (A,E) Distribution of (exclusive (D,H)) variants according to their gene location. Most variants were found in the MUC16 gene, while pathogenic or likely pathogenic variants were mostly located in the AR or BRCA2 gene. (B,F) Allele frequency of all variants; 90% of all variants showed AFs <5%. (C,G) Classification of variants according to their known impact (benign, likely benign, uncertain significance, likely pathogenic and pathogenic) done by IVA. Nearly 25% of all variants are known to be likely pathogenic or pathogenic. No significant different distribution was detected for cfDNA variants from matched whole blood and CTC-depl. blood.