Specific genomic regions are differentially affected by copy number alterations across distinct cancer types, in aggregated cytogenetic data

Abstract

Background: Regional genomic copy number alterations (CNA) are observed in the vast majority of cancers. Besides specifically targeting well-known, canonical oncogenes, CNAs may also play more subtle roles in terms of modulating genetic potential and broad gene expression patterns of developing tumors. Any significant differences in the overall CNA patterns between different cancer types may thus point towards specific biological mechanisms acting in those cancers. In addition, differences among CNA profiles may prove valuable for cancer classifications beyond existing annotation systems.

Principal findings: We have analyzed molecular-cytogenetic data from 25579 tumors samples, which were classified into 160 cancer types according to the International Classification of Disease (ICD) coding system. When correcting for differences in the overall CNA frequencies between cancer types, related cancers were often found to cluster together according to similarities in their CNA profiles. Based on a randomization approach, distance measures from the cluster dendrograms were used to identify those specific genomic regions that contributed significantly to this signal. This approach identified 43 non-neutral genomic regions whose propensity for the occurrence of copy number alterations varied with the type of cancer at hand. Only a subset of these identified loci overlapped with previously implied, highly recurrent (hot-spot) cytogenetic imbalance regions.

Conclusions: Thus, for many genomic regions, a simple null-hypothesis of independence between cancer type and relative copy number alteration frequency can be rejected. Since a subset of these regions display relatively low overall CNA frequencies, they may point towards second-tier genomic targets that are adaptively relevant but not necessarily essential for cancer development.

Figure 1. The overall frequency of genomic copy number alterations (CNA) differs among cancer types.

Boxplots show the CNA frequency distributions among tumor samples in 10 randomly selected cancer types. The boxplot delineations mark the percentiles 5%, 25%, 75% and 95%. The red lines indicate the mean frequency for each cancer type, whereas the blue line represents the overall mean frequency across all 160 cancer types analyzed here. Frequency values are defined as the ratio of number of samples showing a CNA for a genomic region (i.e., cytogenetic bands) over total number of samples in that cancer type. a) Before normalization b) After normalization. In b) the nominal frequency distribution for each cancer type is re-scaled so that its mean matches the overall mean across all cancer types. (NOS – “not otherwise specified”: high-order classifications, not further assigned to more detailed levels).

Figure 2. The tissue type of a cancer has a strong influence on its CNA likelihood pattern.

a) examples of individual chromosome segments, showing their observed CNA frequencies stratified by cell type. Each dot summarizes all samples classified under one particular ICD type, color-coded by root cell type. In the left panel, three chromosome segments are shown that exhibit strong differences between cell types; on the right, three negative examples without such a signal. All p-values were corrected for multiple testing according to Benjamini-Hochberg. b) the dendrogram (tree) has been obtained using hierarchical Ward clustering on the global frequency-normalized CNA profiles across all 160 genomic regions. Cancer types are again color-coded according to the cell type of origin, with the same legend as in a). Partitioning the tree by cutting at different heights produces multiple clusters; validation of those clusters based on the cancer origin (metric: Random Index) shows that the clustering works significantly better than expected at random.

Figure 3. Examples for non-neutral CNA regions.

a) Heatmap of CNA profiles on genomic regions (same clustering as in Figure 2). Genomic locations are represented with orange color when considering duplications/gains, and in blue when considering deletions/losses. Color intensity shows relative CNA frequencies; the most-affected region in each row is arbitrarily set the to brightest color (1.0) for display purposes. b) Small regions (black rectangles on the heatmap) are zoomed in to show how non-neutral CNAs can differentiate between cancer types. The example shows that 7q is preferentially gained in brain tumors (red labels) whereas it is preferentially lost in germ cell (black labels), myeloid and myeloproliferative cancer types (blue labels). c) Small regions (red rectangles on the heatmap) are zoomed in to show how 8q is preferentially lost in medullublastomas (green labels) and is preferentially gained in epithelial tumors (pink labels). Some chromosomes consist entirely of non-neutral regions (such as chromosomes 18 and 7). Note that the spatial resolution of the CNA data on the chromosome is limited (roughly corresponding to cytogenetic band resolution).

Figure 4. Not only CNA “hotspots” are informative in cancer classification.

Genomic regions (bands) are sorted according to their overall frequency of CNAs observed. Those regions that are informative with respect to cancer type clustering are marked with arrows. a) Considering duplications (gains) b) Considering deletions (losses).

Figure 5. Comparison of non-neutral vs. hot-spot CNA.

Genomic regions affected by CNAs, either more frequently than average (black rectangle), or non-neutrally with respect to cancer-type classifications (grey rectangle). The intersection defines regions that are affected both frequently and non-neutrally. Changes are color-coded (gains in orange and losses in blue).