Genomewide Copy Number Alteration Screening of Circulating Plasma DNA

Potential for the Detection of Incipient Tumors

L. Lenaerts; P. Vandenberghe; N. Brison; H. Che; M. Neofytou; M. Verheecke; L. Leemans; C. Maggen; B. Dewaele; L. Dehaspe; S. Vanderschueren; D. Dierickx; V. Vandecaveye; F. Amant; J. R. Vermeesch

Disclosures

Ann Oncol. 2019;30(1):85-95. 

In This Article

Results

A Subset of Elderly People has Chromosomal Imbalances in Plasma cfDNA

Plasma cfDNA of 386 men and 616 women, with a median age of 72 years (range 64–96 years), was subjected to GIPseq analyses. Ninety-five percent of participants had a normal GIPseq profile. For 5% of cases, an aberrant profile was detected and a second blood sample was analyzed (Figure 1A). One participant refused secondary sampling. Eighteen other samples, for which aberrations could not be confirmed upon second sampling, were reclassified as normal. In the remaining 30 cases (20 men and 10 women), representing 3% of the study population, the GIPseq profile of the second blood sample reproduced the CNAs found in the first sample. There was no association between the detection of chromosomal aberrations in plasma cfDNA and the age of the participants (frequency of chromosomal aberrations: 2.7% in people aged 64–69 years; 3.2% in people aged 70–79; and 3.3% in the 80+ group; P > 0.5). Cell-free DNA concentrations did not significantly differ between participants with a reproducible aberrant GIPseq profile and those with a normal score (mean ± SEM: 9.6 ± 0.5 ng/ml plasma for 'normal' cases; 6.3 ± 0.6 ng/ml for 'aberrant' cases; P = 0.66).

Figure 1.

Flow chart of analyses and clinical outcome. (A) Flow chart of plasma sampling and subsequent GIPseq profiling. (B) Clinical follow-up results of the 30 cases with a reproducible GIPseq profile. *For 8 of the 24 cases, for whom no cancer diagnosis was made, clinical follow-up data for either whole-body diffusion weighted MRI examination(s) or general hematological analysis are missing, either due to withdrawal of the participant during the course of the study or because the observed abnormalities in cfDNA were believed not to be cancer-related aberrations. CLL, chronic lymphocytic leukemia; HL, Hodgkin lymphoma; MALT, mucosa-associated lymphoid tissue; MBL, monoclonal B-cell lymphocytosis; MDS, myelodysplastic syndrome; NHL, non-Hodgkin lymphoma; SLL, small lymphocytic lymphoma.

Two types of chromosomal anomalies were observed upon cfDNA profiling: (i) genomewide aberrations with gains and/or losses on multiple chromosomes, affecting whole chromosomes and chromosome arms, characterized by an elevated GIPseq QS-score (n = 12 cases; Figure 2A and Table 1) and (ii) isolated aberrations in which only one chromosome or chromosomal segment was affected, with an overall QS < 2 (n = 18 cases; Figure 2B and Table 2). Isolated anomalies encompassed: a segmental del(3p), del(5q), del(9q), del(20q) or del(22q); segmental gains on 9p and 9q, or 14q, or a trisomy 12 or 15.

Figure 2.

Chromosomal aberrations observed in cell-free plasma DNA. Representative circos plots showing reproducible chromosomal anomalies detectable in cfDNA. Plotting is based on the GIPseq results of the second plasma sample of each case, showing those chromosomal anomalies that are reproducible in the two independent plasma samples and with a chromosomal z-score ≥3.0 (suggesting gain; in green) or ≤3.0 (suggesting loss; in red). Chromosomal regions with clear reproducible amplifications and deletions, resulting in a neutral z-score are displayed as well. For every case, the genomic representation profile of the autosomal chromosomes is shown in clockwise order, aligned with chromosomal ideograms (outer circle). (A) Cases with genomewide chromosomal aberrations (n=12). For some of these cases, high z-scores for almost every chromosome were observed. This indicates that either all chromosomes are indeed affected, or the z-scores of particular individual chromosomes or chromosomal fragments might be skewed due to excessive presentation of other, highly amplified chromosomes or chromosome arms. (B) Cases with single chromosomal aberrations (n=18). For one del(20q) case focal gains on chromosome 5 emerged upon second cfDNA sampling (second last case; see also supplementary Figure S8, available at Annals of Oncology online). Cases are shown from the periphery to the center in the same order as in Tables 1 and 2.

For the 30 participants with aberrant GIPseq profiles, aCGH on peripheral blood cell DNA was carried out to investigate the possibility of an underlying hematological malignancy. Furthermore, general peripheral blood parameters (including cell count and cytology) were examined and subjects were screened via WB-DWI MRI for the presence of potentially malignant lesions. The results are summarized in Table 1 and Table 2 and Figure 1B.

Genomewide Chromosomal Imbalances in Plasma cfDNA may Point to an Underlying (pre)Malignancy

Five cases, with genomewide chromosomal imbalances in cfDNA, were diagnosed with cancer or a premalignant condition upon clinical follow-up (Table 1): one non-Hodgkin lymphoma (NHL)-type chronic lymphocytic leukemia (CLL) (Rai I–Binet A); one classical Hodgkin lymphoma (HL) (stage II); one myelodysplastic syndrome (MDS) with excess blasts (type 1, IPSS 2); one NHL-type mucosa-associated lymphoid tissue (MALT; stage I) and one high-count monoclonal B-cell lymphocytosis (MBL), a premalignant condition at risk for progression to B-cell malignancy. Molecular investigations in biopsies from the indices, confirmed that the CNAs detected in cfDNA were, at least partially, derived from tumor DNA. One additional individual, having an isolated trisomy 12 in cfDNA, was eventually identified with NHL-type small lymphocytic lymphoma (SLL; stage III), with the trisomy 12 present as a low-grade mosaicism in peripheral blood cells. Supplementary document S1, available at Annals of Oncology online, comprehensively describes follow-up investigations in these six cases. Figure 3 depicts the results of the molecular and clinical examinations for the NHL, type CLL. Analogous results for the HL, MDS, MALT, SLL and MBL cases are shown in supplementary Figures S1–S5, available at Annals of Oncology online.

Figure 3.

GIPseq profiles, array CGH and WB-DWI MRI analyses for the case with non-Hodgkin lymphoma-type CLL. (A) GIPseq-profiles of the two consecutive blood samples of case 12012016-28. For every chromosome, the ideogram is shown at the left, together with the genomic representation profile of sample one (immediately at the right) and sample two (far right, taken three months after sample one). The vertical graph represents the actual genome representation profile: dotted lines on either side of the axis, plus or minus ×1.5; red areas, likely deleted regions; green areas, likely duplicated or amplified regions. (B) Array CGH on peripheral blood DNA. The column labels represent the numbered chromosomes plus X and Y; the y-axis, represents the log 2 of the intensity ratios; each graphed point represents an array probe. (C) WB-DWI MRI. Arrows indicate the presence of multiple infra- and supradiaphragmatic adenopathies, indicative of a lymphoproliferative process.

Supplementary Figure S1.

GIPseq profiling and FISH analyses for case 25022016–07, diagnosed with classical Hodgkin lymphoma, stage II.
(A) GIPseq-profiles of the two consecutive blood samples. For every chromosome, the ideogram is shown at the left, together with the genomic representation profile of sample one (immediately at the right) and sample two (far right, taken three months after sample one). For further explanation of the graphic interventions used in a GIPseq profile, see Figure 3. (B) Validation by FISH analyses of Hodgkin/Reed-Sternberg (RS) cells on formalin-fixed paraffinembedded biopsy tissue. In the GIPseq-profiles, arrows indicate the cytoband position of the examined target genes. In the representative FISH images of interphase nuclei, one giant Reed-Sternberg (RS) cell and several other small interphase cells, are shown. FISH analyses point to a gain of BCL6 (3q27) (orange/green) relative to the CEP 8 control probe (blue) in 90% of investigated RS cells (panel 1); loss of ATM (11q22) (orange) over the CEP 12 control probe (green) in 21% of RS cells (panel 2) and loss of TP53 (17p13) (orange probe) over CEP 17 (green probe) (panel 3) in 25% of RS cells, confirming the gain of 3q, and loss of 11q and of 17pter detected by GIPseq profiling of cfDNA.

Supplementary Figure S2.

GIPseq profiling and karyotyping for case 02102015–03 diagnosed with myelodysplastic syndrome with excess blasts type 1, IPSS 2.
(A) The GIPseq-profiles for the two consecutive blood samples are shown. For every chromosome, the ideogram is shown at the left, together with the genomic representation profile of sample one (immediately at the right) and sample two (far right). (B) Karyotyping was performed on bone marrow cells: 45,XX,dic(5;17)(q11;p12~13)[4]/46,XX[12].

Supplementary Figure S3.

GIPseq profiling and low-pass sequencing for case 09022017–08 diagnosed with a gastric leiomyoma and non-Hodgkin lymphoma type mucosa-associated lymphoid tissue (MALT).
(A) GIPseq profiles of the 2 consecutive blood samples. For further explanation of the graphic interventions used in a GIPseq profile, see Figure 3. (B) Low-pass sequencing (0,1x) of biopsy DNA, extracted from formalin-fixed paraffinembedded (FFPE) tumor biopsy material. Copy number profiling was done via the software package QDNAseq (Scheinin I. et al. Genome Research 2014; 24: 2022–2032), using a bin size of 100 kilo base pairs, showing copy number alterations in MALT DNA but not in leiomyoma DNA.

Supplementary Figure S4.

GIPseq profiling and FISH analysis for case 15092016–04, diagnosed with a non-Hodgkin lymphoma type small lymphocytic lymphoma (SLL), stage III.
(A) The GIPseq-profiles for the two consecutive blood samples are shown. For further explanation of the graphic interventions used in a GIPseq profile, see Figure 3. (B) Validation by FISH analyses on freshly isolated peripheral blood cells, showing a gain of the centromeric region of chromosome 12 (12p11.1-q11) (green) over ATM (11q22) (orange probe), confirming the presence of a trisomy 12 in 6% of investigated cells. In the GIPseq profiles, arrows indicate the cytoband position of the examined target gene. (C) WB-DWI MRI. Arrows point to axillary adenopathies; hyper intense at b1000 images (panel 1) corresponding to low apparent diffusion coefficient (ADC) of 0.82 × 10−3 mm2/sec (panel 2).

Supplementary Figure S5.

GIPseq profiling and FISH analysis for case 16032017–13, diagnosed with a high-count monoclonal B-cell lymphocytosis.
(A) The GIPseq-profiles for the two consecutive blood samples are shown. For further explanation of the graphic interventions used in a GIPseq profile, see Figure 3. (B) Validation by FISH analyses on freshly isolated peripheral blood cells. In the GIPseq profiles, arrows indicate the cytoband position of the examined target genes. FISH analyses point to a loss of RUNX1 (21q22) (green) relative to RUNX1T1 (8q21.3) (orange) in 21% of investigated cells (panel 1); loss of TP53 (17p13) (orange probe) over CEP 17 (17p11.1-q11) (green) in 26% of cells (panel 2) and gain RARA (17q21) (green) over PML (15q24) (orange) pointing to an isochromosome of the long arm of chromosome 17 in 10% of cells (panel 3).

In the remaining 24 individuals, of whom 7 presenting with multiple and 17 having isolated chromosomal anomalies in cfDNA, no definitive cancer diagnosis was made (Table 1 and Table 2 and Figure 1B). For eight of them, no complete clinical follow-up was carried out, either due to the participant's voluntary withdrawal, and hence preventing a final diagnosis, or because the abnormalities in cfDNA were believed not to be cancer-related (indicated with an asterisk, Table 1 and Table 2). For one such case (case 22022016–03), again having genomewide CNAs in cfDNA but not in peripheral blood DNA, WB-DWI MRI revealed supraclavicular adenopathies, being suggestive of a lymphoma. No confirmatory diagnosis was made since this woman refused subsequent clinical investigations because of her age. Additionally, one male (case 30032017–02) having genomewide aberrations in cfDNA, was independently of our investigations, diagnosed with a lip cancer, but no correlation was found between the imbalances in cfDNA and the copy number profile in tumor DNA (data not shown).

About One-third of Chromosomal Imbalances in Plasma cfDNA of Asymptomatic Individuals Originates From Peripheral Blood Cells

In 9 of the 24 (37.5%) individuals without a cancer diagnosis, the aberrant cfDNA species were shown, upon aCGH, FISH and/or low-pass sequencing of peripheral blood DNA, to originate from a low-grade mosaicism in peripheral blood cells, but without fulfilling the criteria for a hematological diagnosis. It concerned cases with an isolated anomaly on chromosome 3p, 5q, 15, 20q or 22q (Table 2). GIPseq profiles of each of these imbalances, and confirmatory analyses in peripheral blood cells, are shown in supplementary Figures S6–S8, available at Annals of Oncology online. For 15 out of the 24 undiagnosed cases (62.5%), the chromosomal imbalances observed in cfDNA were not detected in peripheral blood DNA.

Supplementary Figure S6.

GIPseq profiling and confirmatory analyses in peripheral blood cells for cases with an isolated deletion on chromosomes 3, 5 and 22.
(A) GIPseq-profile of case 08122015–04, pointing to a segmental 3p deletion, and confirmatory array CGH (aCGH) analysis showing that 40% of peripheral blood cells carry this 3p deletion. (B) GIPseq-profile of a representative case with an isolated segmental 5q deletion (case 16062016–33) and FISH showing loss of EGR1 (5q31) (orange) relative to D5S23, D5S721 (5p15.2) (green) in 10% of the investigated peripheral blood cells. In the GIPseq profiles, the arrow indicates the cytoband position of the examined target gene. (C) GIPseq-profile of case 23012017–28, pointing to a segmental 22q deletion, and confirmatory aCGH analysis showing a low-grade mosaisicme in about 20% of peripheral blood cells. For every case, the GIPseq profile of the second plasma sample is shown.

Supplementary Figure S7.

GIPseq profiling and FISH analysis for the 6 cases with an isolated amplification of chromosome 15 in cfDNA.
(A) GIPseq profiles for chromosome 15 in the two consecutive cfDNA samples of every case. For further explanation of the graphic interventions used in a GIPseq profile, see Figure 3. (B) FISH on peripheral blood of the indices using the Vysis dual color probe PML (orange spectrum) (15q22)/RARA (green spectrum) (17q21). For case 06122016–12 and case 21062016–02 respectively 21% and 16% of interphase nuclei show a trisomy 15. For cases 14032016–08 and 16022017–08, no fresh peripheral blood sample was available for FISH analyses. (C) Low-pass sequencing (0,1x) of peripheral blood DNA and copy number profiling via QDNAseq software, confirming the presence of +15 for cases 06122016–12 and 21062016–02, and suggesting a gain of chromosome 15 for case 30032017–02. For case 16022017–08, low-pass sequencing was not performed.

Supplementary Figure S8.

GIPseq profiling and array CGH for the 4 cases with an isolated segmental deletion on chromosome 20q in cfDNA.
(A) GIPseq profiles for chromosome 20 in the 2 consecutive blood samples of every case. For further explanation of the graphic interventions used in a GIPseq profile, see Figure 3. (B) Array CGH on peripheral blood DNA, confirming the presence of a del(20q) as a low-grade mosaicism (20 to 30%) in peripheral blood DNA of all cases. For case 29042016–02, gains on chromosome 5 emerged in the second plasma cfDNA sample, which were not detectable by aCGH on peripheral blood DNA.

Four Incidental Cancer Diagnoses Were Made Despite a Normal GIPseq Profile

Four out of the 953 cases, with a normal GIPseq profile, were incidentally diagnosed with one of the following cancers: prostate adenocarcinoma (stage pT3aN0M0; diagnosed 5 months after GIPseq analysis), bronchus adenocarcinoma (stage cT3N3M1b; diagnosed 35 days after GIPseq analysis), liver-metastasized colon adenocarcinoma (stage pT4N1cM1; 1 month after GIPseq analysis) or multiple myeloma IgG kappa (stage 2; 27 days after GIPseq testing). For the prostate adenocarcinoma, additional plasma analysis at the time of diagnosis again showed a normal GIPseq profile (supplementary Figure S9A, available at Annals of Oncologyonline). Whole-genome shallow sequencing was carried out on FFPE biopsy DNA of the solid tumors, and FISH analysis was done on bone marrow plasma cells from the multiple myeloma patient. The genomes of the prostate, bronchus and colon adenocarcinomas and that of the plasma cell tumor displayed noticeable focal and/or broad CNAs (supplementary Figures S9–S11, available at Annals of Oncology online).

Supplementary Figure S9.

GIPseq profiling and low-pass sequencing for the case identified with prostate adenocarcinoma after a normal GIPseq profile.
(A) GIPseq profiles of two plasma cfDNA samples; the first sample being taken at inclusion and the second plasma sample taken after cancer diagnosis and before surgery. For further explanation of the graphic interventions used in a GIPseq profile, see Figure 3. (B) Low-pass sequencing (0,1x) of FFPE-derived prostate adenocarcinoma DNA and copy number analysis via QDNAseq software, showing the presence of copy number alterations in tumor DNA.

Supplementary Figure S10.

GIPseq profiling and low-pass sequencing for the case identified with bronchus carcinoma 35 days after normal GIPseq profiling.
(A) GIPseq profile of the plasma sample taken at inclusion. No plasma sample could be taken after cancer diagnosis as this women had passed away. For further explanation of the graphic interventions used in a GIPseq profile, see Figure 3. (B) Low-pass sequencing (0,1x) of FFPE-derived bronchus carcinoma DNA and copy number analysis via QDNAseq software, showing the presence of copy number alterations in tumor DNA.

Supplementary Figure S11.

GIPseq profiling and low-pass sequencing for the case identified with metastasized colon cancer 1 month after normal GIPseq profiling.
(A) GIPseq profile of the plasma sample taken at inclusion. For further explanation of the graphic interventions used in a GIPseq profile, see Figure 3. (B) Low-pass sequencing (0,1x) of FFPE-derived colon cancer biopsy DNA and copy number analysis via QDNAseq software, showing the presence of copy number alterations in tumor DNA.

Comments

3090D553-9492-4563-8681-AD288FA52ACE

processing....