Aromatase inhibition remodels the clonal architecture of estrogen-receptor-positive breast cancers

Abstract

Resistance to oestrogen-deprivation therapy is common in oestrogen-receptor-positive (ER+) breast cancer. To better understand the contributions of tumour heterogeneity and evolution to resistance, here we perform comprehensive genomic characterization of 22 primary tumours sampled before and after 4 months of neoadjuvant aromatase inhibitor (NAI) treatment. Comparing whole-genome sequencing of tumour/normal pairs from the two time points, with coincident tumour RNA sequencing, reveals widespread spatial and temporal heterogeneity, with marked remodelling of the clonal landscape in response to NAI. Two cases have genomic evidence of two independent tumours, most obviously an ER- 'collision tumour', which was only detected after NAI treatment of baseline ER+ disease. Many mutations are newly detected or enriched post treatment, including two ligand-binding domain mutations in ESR1. The observed clonal complexity of the ER+ breast cancer genome suggests that precision medicine approaches based on genomic analysis of a single specimen are likely insufficient to capture all clinically significant information.

Figure 1. Collision tumours of independent origin and ER status in BRC38.

(a) Clonality plot comparing the VAFs of SNVs in the baseline and surgical samples. (b) Gene fusions and copy number alterations. Outer ring: CN alterations in the baseline sample (amplifications in red, deletions in blue). Inner ring: CN alterations in the surgical sample. Centre: gene fusion events that were specific to the baseline (green) or surgical sample (brown). Expansion: CN alterations on chromosome 1. (c) Immunohistochemistry results indicate the ER status of one baseline (left) and two surgical samples (middle, right). Scale bars, 200 μm.

Figure 2. Simple and stable clonal structure in BRC14.

(a) Clonality plot comparing the variant allele fraction of SNVs in the baseline and surgical samples. (b) Gene fusions and copy number alterations. Outer ring: CN alterations in the baseline sample (amplifications in red, deletions in blue). Inner ring: CN alterations in the surgical sample. Centre: gene fusion events that were specific to the baseline (green) or surgical sample (brown).

Figure 3. Subclonal complexity and response to AI inhibition in BRC15.

Clonality plots derived from four-dimensional clustering of SNV VAFs in distinct core samples. (a) Two samples separated spatially in the baseline tumour. (c,d) The first baseline core sample compared with two cores taken from the surgical sample. (b) Gene fusions and copy number alterations in (from outer ring to inner ring) Baseline core 1, Baseline core 2, Surgical core 1, Surgical core 2 (amplifications in red, deletions in blue). Centre: gene fusion events that are baseline sample-specific (green), surgical sample-specific (brown) or shared (black).

Figure 4. Complex and stable clonal structure in BRC11.

(a) Clonality plot comparing the variant allele fraction of SNVs in the baseline and surgical samples. (b) Gene fusions and copy number alterations. Outer ring: CN alterations in the baseline sample (amplifications in red, deletions in blue). Inner ring: CN alterations in the surgical sample. Centre: gene fusion events that were specific to the baseline (green) or surgical sample (brown).