Copy number profiles of paired primary and metastatic colorectal cancers

Abstract

Liver metastasis is the major cause of death following a diagnosis of colorectal cancer (CRC). In this study, we compared the copy number profiles of paired primary and liver metastatic CRC to better understand how the genomic structure of primary CRC differs from the metastasis. Paired primary and metastatic tumors from 16 patients and their adjacent normal tissue samples were analyzed using single nucleotide polymorphism arrays. Genome-wide chromosomal copy number alterations were assessed, with particular attention to 188 genes known to be somatically altered in CRC and 24 genes that are clinically actionable in CRC. These data were analyzed with respect to the timing of primary and metastatic tissue resection and with exposure to chemotherapy. The genomic differences between the tumor and paired metastases revealed an average copy number discordance of 22.0%. The pairs of tumor samples collected prior to treatment revealed significantly higher copy number differences compared to post-therapy liver metastases (P = 0.014). Loss of heterozygosity acquired in liver metastases was significantly higher in previously treated liver metastasis samples compared to treatment naive liver metastasis samples (P = 0.003). Amplification of the clinically actionable genes ERBB2, FGFR1, PIK3CA or CDK8 was observed in the metastatic tissue of 4 patients but not in the paired primary CRC. These examples highlight the intra-patient genomic discrepancies that can occur between metastases and the primary tumors from which they arose. We propose that precision medicine strategies may therefore identify different actionable targets in metastatic tissue, compared to primary tumors, due to substantial genomic differences.

Keywords: chemotherapy; colorectal cancer; copy number alterations; liver metastasis; loss of heterozygosity.

Figure 1. Clinical characteristics of the study cohort

The course of treatment for the 16 CRC patients is indicated schematically. Samples were obtained at surgical resection time points indicated by yellow blocks for primary CRC and red blocks for the liver metastases. The survival status of the patient is shown, through observations made at 1–6 month intervals until death or December 2015.

Figure 2. Differences between the copy number profiles of each tumor and metastasis pair including the influence of chemotherapy exposure

The proportion of tumor genomes affected by copy number alteration and loss of heterozygosity (LOH) is shown for paired primary tumors (T) and metastases (M) from the 16 cases. The level of loss or gain of DNA is indicated by different colors (A). Whole genome duplication (WGD) is indicated by black triangles above the bar and is defined as >70% of at least half the chromosomes display copy numbers between 3 and 4 with both parental alleles present. The proportion of tumor genome with LOH is plotted (B).

Figure 3. Genomic copy number and loss of heterozygosity (LOH) differences identified between all 16 primary tumor and metastasis pairs

The copy number and LOH was assayed every 10Kb for 287921 sites per metastatic genome and compared with the corresponding region in the paired primary tumor sample with a correction for whole genome duplication applied to all sample pairs. The proportion of regions with copy number that was different between the paired samples is indicated in (A) by the height of the green bars. The proportion of regions for which LOH was observed only in the metastasis sample is indicated by the height of the dark blue bars and those with LOH only in the primary tumor sample by the height of the light blue sections (B). In both charts the grey indicates concordant copy number or LOH states between the paired samples. ^*P < 0.05, ^**P < 0.01 (t-test).

Figure 4. Differences in candidate driver genes between primary and metastasis samples

We analysed the copy number status for 188 candidate CRC genes in the paired primary CRC and metastases (A). The copy number status of 123 genes (123/188 65.4%) was shared between primary and metastasis. KRAS mutation was present in the primary cancers but was not detectable in the metastatic sample following chemotherapy (B). The copy number profile of the genomic region containing KRAS indicates the locus was heterozygous (2 copies; AB) in the primary tumor but showed copy neutral LOH (2 copies; AA) in the paired metastasis (B).

Figure 5. Copy number status private to the metastasis sample affecting clinically actionable targets

The degree of copy number difference was calculated by subtraction of the primary tumor copy number state from that of the liver metastasis. The copy number status of 24 clinically actionable genes was assessed for alterations occurring specifically in metastatic but not primary tissue. ERBB2, a known target receptor of trastuzumab (anti-HER2), was specifically amplified only in the metastatic tissue of C6. Similarly, FGFR1, a known target receptor of regorafenib was specifically amplified only in the metastatic tissue of C8. Black arrow indicates specific gene amplification of C6 (ERBB2) and C8 (FGFR1) respectively.