Somatic copy number profiling from hepatocellular carcinoma circulating tumor cells

Abstract

Somatic copy number alterations (SCNAs) are important genetic drivers of many cancers. We investigated the feasibility of obtaining SCNA profiles from circulating tumor cells (CTCs) as a molecular liquid biopsy for hepatocellular carcinoma (HCC). CTCs from ten HCC patients underwent SCNA profiling. The Cancer Genome Atlas (TCGA) SCNA data were used to develop a cancer origin classification model, which was then evaluated for classifying 44 CTCs from multiple cancer types. Sequencing of 18 CTC samples (median: 4 CTCs/sample) from 10 HCC patients using a low-resolution whole-genome sequencing strategy (median: 0.88 million reads/sample) revealed frequent SCNAs in previously reported HCC regions such as 8q amplifications and 17p deletions. SCNA profiling revealed that CTCs share a median of 80% concordance with the primary tumor. CTCs had SCNAs not seen in the primary tumor, some with prognostic implications. Using a SCNA profiling model, the tissue of origin was correctly identified for 32/44 (73%) CTCs from 12/16 (75%) patients with different cancer types.

Keywords: Molecular medicine; Prognostic markers.

Conflict of interest statement

Competing interestsH.R.T has ownership in the intellectual property used to isolate circulating tumor cells in this study (NanoVelcro CTC Assay), which has been licensed to CytoLumina Technologies Corp. H.R.T. and S.X.L. have financial interests in CytoLumina Technologies Corp. given their role in the company. All other authors report no conflicts of interest.

© This is a U.S. government work and not under copyright protection in the U.S.; foreign copyright protection may apply 2020.

Figures

Fig. 1. Experimental design of the study.

a Workflow used in the study for CTC isolation and SCNA profiling using the NanoVelcro assay. Following gradient centrifugation, HCC CTCs are selectively captured on the NanoVelcro surface for identification and subsequent isolation by LMD. CTC samples undergo WGA and QC prior to low-resolution whole-genome sequencing for SCNA profiling. b A total of four specimens are obtained from each patient. From peripheral blood, germline genomic DNA is obtained from the bulk buffy coat layer. Circulating tumor cells are obtained and sequenced as described in (a). From the surgically resected specimen, a sample of both the primary tumor and the peritumoral normal liver are obtained and sequenced.

Fig. 2. Global copy number profiles for blood, peritumoral normal liver, CTCs, and primary tumor (n = 8) or solitary metastasis (n = 1, H167) demonstrates the recapitulation of SCNAs from the tumor in CTC samples, supporting the potential of CTCs to act as a liquid biopsy of the tumor.

Patient H169’s CTC (shown in Supplementary Fig. 1) STR analysis did not match that of the tumor or blood sample, suggesting contamination.

Fig. 3. Heatmap showing copy number profiles at 59 loci frequently amplified or lost in HCC. Gain (red) or loss (blue) at each of the 59 cytobands (identified on the right of the heatmap) in both CTC and tumor samples from all nine patients that passed STR analysis are shown.

Unsupervised hierarchical clustering demonstrates clustering of CTC–tumor pairs for all nine patients. CTC—circulating tumor cell, 1°—Primary tumor.

Fig. 4. Comparison of low-resolution whole-genome copy number profiles for primary tumor, CTCs 1 and 2, normal liver, and whole blood reveals reproduction of the majority of the SCNAs from the primary tumor in both CTC samples.

a Whole genome SCNA profiles for the primary tumor as well as the two CTC samples. b Chromosome 17p loss is seen in the primary tumor and both CTC samples, but not the normal liver or whole blood. The location of the tumor suppressor TP53 gene, the most frequently mutated or lost gene in HCC, is indicated by the blue triangle. c Chromosome 20 amplification was seen in both CTC samples but not in the primary tumor. Chromosome 20 amplifications are a recurrent somatic copy number alteration in HCC associated with 2 important oncogenes. Overexpression of AIB1 (orange triangle) is frequently described in HCC, and has previously been associated with invasiveness. In addition, recent research has demonstrated that the oncogenic effects of MYC dysregulation, a common occurrence in HCC, require overexpression of AURKA (green triangle) for stabilization. Furthermore, in p53-altered HCC patients, the MYC-AURKA complex is an actionable drug target based on preclinical studies.

Fig. 5. Copy number alterations at all genes for the 32 cancer types were transformed using t-SNE into a 2-dimensional space.

The mean of all samples for each cancer type is plotted as well as the individual samples for the best clustering (TGCT—Blue) and worst clustering (ESCA—Pink) cancer types. All individual samples are plotted by cancer type in Supplementary Fig. 5.

Fig. 6. Cancer class prediction based on SCNA profile for each of the 15 HCC CTC samples using the classification model.

9/15 (60%) of CTCs were correctly classified as being HCC, while 6/15 (40%) were classified as being from a GI source. 5/9 (56%) patients had at least one CTC classified as being from a HCC tumor.