Breast cancer genome and transcriptome integration implicates specific mutational signatures with immune cell infiltration

Abstract

A recent comprehensive whole genome analysis of a large breast cancer cohort was used to link known and novel drivers and substitution signatures to the transcriptome of 266 cases. Here, we validate that subtype-specific aberrations show concordant expression changes for, for example, TP53, PIK3CA, PTEN, CCND1 and CDH1. We find that CCND3 expression levels do not correlate with amplification, while increased GATA3 expression in mutant GATA3 cancers suggests GATA3 is an oncogene. In luminal cases the total number of substitutions, irrespective of type, associates with cell cycle gene expression and adverse outcome, whereas the number of mutations of signatures 3 and 13 associates with immune-response specific gene expression, increased numbers of tumour-infiltrating lymphocytes and better outcome. Thus, while earlier reports imply that the sheer number of somatic aberrations could trigger an immune-response, our data suggests that substitutions of a particular type are more effective in doing so than others.

Figure 1. Clustered correlation matrix of 266 breast cancer cases.

The left panel shows the dendrogram and clustered correlation matrix (red is positive, blue negative correlation) of 266 breast cancer cases. The top 5,000 most variable transcripts were used for correlating the samples. For the columns in the right panel, colour codes are as follows: ER: ER-positive dark grey, ER-negative light grey. Subtype: Red, basal; dark blue, Luminal B; light blue, Luminal A; green, normal-like; and dark yellow, her2. Grade: white, NA; light grey, grade 1; grey, grade 2; and black, grade 3. Tissuetype: red, ductal; blue, lobular; light blue, micropapillary; grey, mucinous; dark yellow, papillary; yellow, apocrine; and dark green, other type. # subs: The length of the green bar is proportional to the number of substitutions. Cases with >10,000 substitutions are shown with a soft-red coloured bar of equal length. The remaining nine columns show the status of driver genes. Light grey, wild type; dark yellow, copy-number amplification; blue, homozygous deletion; and mutations (substitution, indels, rearrangements) are dark green if activating and red if inactivating.

Figure 2. Boxplots of CCND3 and GATA3 expression.

All y axes show expression levels in our cohort (n=266, left panels, log2-FPKM) and from TCGA cases (n=960, right panels, RNA Seq V2 RSEM, log2) according to copy-number state of CCND3 (top panel, our cohort n=7 amplified, TCGA n=152 with loss and n=233 amplified) and GATA3 mutation state (bottom panel, our cohort n=25 mutated, TCGA n=95 mutated).The box is bounded by the first and third quartile with a horizontal line at the median, whiskers extend to the maximum and minimum value. The notch shows the 95% CI of the median.

Figure 3. Mitotic cell cycle gene activity related to mutational signatures and outcome.

(a) The average expression of genes (n=409) from Gene-Ontology term Mitotic Cell Cycle (GO:0000278) were used to rank ER-positive samples. The vertical black line indicates the third quartile border. Top panel: heatmap of median centred expression values in log2-FPKM, red indicates above median, blue below median expression. Genes are in rows, samples in columns. Below the heatmap: ‘Average MCC' shows the average expression of the cell cycle genes. Grade: pathological grade; white, NA; light grey, grade 1; grey, grade 2; and black, grade 3. Last five rows: the length of the green bar is proportional to the number of substitutions. # subs: the total number of substitutions, samples with >10,000 substitutions are shown with a soft-red coloured bar and are of equal length. For the columns labelled Sig. 1, 3, 8 and 13, soft-red indicate samples >3,000 of such substitutions. (b,c) Overall and relapse-free survival Kaplan–Meier curves. Blue indicates patients with less than the median number of substitutions, red indicates higher than median. P values are logrank-test values. The x-axis shows time in months, y-axis shows the proportion of surviving patients.

Figure 4. Activity of TIL-signature genes related to mutational signatures.

The average expression of genes (n=116) from a TIL specific RNA-signature was used to rank ER-positive samples. The vertical black line indicates the third quartile border. Top panel: heatmap of median centred expression values in log2-FPKM, red indicates above median, blue below median expression. Genes are in rows, samples in columns. Below the heatmap: the first row shows the average expression of the TIL genes. Last three rows: the length of the green bar is proportional to the number of substitutions of the indicated signatures. Samples with >3,000 substitutions are shown with a soft-red coloured bar and are of equal length.

Figure 5. Combined MCC and TIL-signature genes and outcome.

Overall (a) and relapse-free (b) survival Kaplan−Meier curves of our cohort and (c) metastasis-free survival of independent in-house and public data sets. Green line indicates patients with high expression of TIL genes (top quartile) and low expression of MCC genes (bottom three quartiles). The red line indicates patients with low expression of TIL genes and high expression of MCC genes. Blue indicates the remaining patients. P values are logrank-test for trend values. The x-axis shows time in months, y-axis shows the proportion of patients. For the numbers at risk, 1 indicates the highTIL/lowCC group (green line), 3 indicates the lowTIL/highCC (red line) and 2 the remaining patients.