Impact of mutational profiles on response of primary oestrogen receptor-positive breast cancers to oestrogen deprivation

Abstract

Pre-surgical studies allow study of the relationship between mutations and response of oestrogen receptor-positive (ER+) breast cancer to aromatase inhibitors (AIs) but have been limited to small biopsies. Here in phase I of this study, we perform exome sequencing on baseline, surgical core-cuts and blood from 60 patients (40 AI treated, 20 controls). In poor responders (based on Ki67 change), we find significantly more somatic mutations than good responders. Subclones exclusive to baseline or surgical cores occur in ∼30% of tumours. In phase II, we combine targeted sequencing on another 28 treated patients with phase I. We find six genes frequently mutated: PIK3CA, TP53, CDH1, MLL3, ABCA13 and FLG with 71% concordance between paired cores. TP53 mutations are associated with poor response. We conclude that multiple biopsies are essential for confident mutational profiling of ER+ breast cancer and TP53 mutations are associated with resistance to oestrogen deprivation therapy.

Figure 1. CONSORT diagram and mutational landscape.

Samples were selected in two phases using the same quality criteria. Samples in phase I (a) underwent whole-exome sequencing (WES) at low coverage for mutation detection followed by capture-probe sequencing for validation. Our goal was to select the same number of controls, good responders and poor responders, but due to the availability of samples and exclusion criteria, we were not able to identify 20 poor responders, instead 15 poor and 25 good responders entered the analysis. In phase II (b), samples that failed WES (not shown, see Supplementary Fig. 1) and samples from additional patients without prior WES were sequenced with the same capture-probe panel as in phase I. To balance the number of patients in the responder groups, preferentially poor responders were added. When samples from phase I and II combined, a total of 86 patients entered the downstream analysis, of which 77 are paired samples (see also Table 2). CONSORT diagram is simplified; a more detailed version can be found in Supplementary Fig. 1. (c) Mutation type of all validated mutations in the exome of 59 patients from phase I and (d) number of mutations in each patient by responder groups. Identical mutations found in the baseline and surgery sample of the same patients appear once in this figure only.

Figure 2. Differences of mutation counts and treatment effects.

Analysis on the mutation load of samples with exome-wide mutation profile from phase I. (a–b) Poor responder showed significantly more mutations than good responder on baseline (B) and surgery (S). (c) Also the number of mutations on a per-patient basis (mutations from B and S samples combined, counting identical mutations once only) was significantly higher in poor responders. Median and interquartile ranges are shown as bars. (d) No significant difference between the B and S mutation counts within responder groups between each of the 49 paired samples. (e) Good and poor responders showed a significant, but low reduction of the mean variant allele fractions (VAFs) of single-nucleotide variants between B and S. Whiskers show 95% confidence interval. Significance was tested by Mann–Whitney test.

Figure 3. Intra-tumour heterogeneity.

Five examples with clear intra-tumour heterogeneity are shown (Supplementary Figs 6–8 for plots of all samples). Some patients had clusters present in both samples (P007, P014 and P039), while others had several clusters that were found in either the baseline or surgery sample (P002 and P046). The variant allele fractions of mutations are shown. Whole-exome sequencing was used for copy-number assessment and only mutations in copy-number neutral regions were plotted. Colours indicate assigned clusters by SciClone (Methods). Cancer-related genes listed in Supplementary Table 4 are labelled in the plots.

Figure 4. Frequently mutated genes.

Sample matrix for genes with mutations in 10% or more of the patients. All 163 tumour samples from 86 patients (including 77 pairs) with baseline (B) and surgery (S) sample are shown. For phase I samples, the bottom row shows if the sample successfully underwent whole-exome sequencing and therefore mutations identified in this sample were added to the capture-probe panel. The TP53 mutation of P038 and one of each mutation of MLL3 for P046 were not identical between B and S. The overall concordance between B and S samples of patients was 71%.

Figure 5. Relation of mutations to Ki67.

(a) Correlation of mutations counts to the Ki67 level was highest for treated samples at surgery. (b) On the combined set from phase I and phase II, the Ki67 level of poor responders was significantly higher for patients with mutated TP53 (mut) than wild-type TP53 (WT). This was not seen for good responders, although Ki67 level for TP53-mutated patients was higher on baseline (Supplementary Fig. 11). Significance tested by Mann–Whitney test, red lines show median and interquartile ranges.