Malignant melanoma is an aggressive cancer that is notorious for its abilities to adapt to a wide variety of metastatic sites and to rapidly develop treatment resistance. A proposed explanation for these behaviours is that melanomas are frequently genetically unstable and thus able to rapidly generate genetically distinct clones that are favourably selected in a variety of contexts within patients (e.g. different metastatic sites, different anti-melanoma therapies). It is therefore important to understand the drivers of melanoma progression and therapy resistance via characterisation of the genomic changes occurring throughout disease evolution.
To do this, we employ various genotyping approaches, including SNP arrays to identify variations in copy number profile that occur during melanoma evolution. However, comparisons of SNP data derived from spatiotemporally separated tumor samples raise several bioinformatics challenges. For example, comparisons between samples of copy number changes derived from SNP array data may be very imprecise if tumors are highly aneuploid or genetically heterogeneous.
To address this, we developed a novel statistical method to quantify genomic differences between tumours without relying on allelic copy number estimation. The method uses statistical analysis across the genome to simultaneously compare both log R ratio and B allele frequency. This new method can precisely calculate the level of differences in samples for which allelic copy number estimation is challenging, and detect heterogeneous copy number changes. It also corrects for ploidy differences between samples, which offset the normalization of the signals and create an over-estimation of genomic changes.
This technique is now being applied to understand the nature of copy number changes associated with various experimental and clinical contexts in which evolutionary changes occur in melanoma, such as metastasis and the development of therapy resistance. Methods to improve the accuracy of inter-tumoral comparisons of molecular data are likely to reveal underlying mechanisms that drive the adaptive and evolutionary changes in cancer that ultimately kill patients by spawning of metastatic, therapy resistant disease.