Extended Data Fig. 2: ER Transcriptional Activity by ChIPseq.
A) Normalised coverage density across core ER sites (as defined in Ross-Innes et al. 2012) in baseline and EOT samples. Paired results for individual patients are joined by dashed black lines. Solid lines are the least squares line summarising the data distribution within each trial arm, as shown in Fig. 3c. B) Box plot showing distribution of median normalised coverage density at core ER sites per patient grouped by trial arm, at baseline and EOT. Box = IQR, centre = median, whiskers = min/max (excluding outliers, defined as <Q1 – 1.5*IQR or > Q3 + 1.5*IQR). C) ER ChIPseq – heatmap of read coverage at core ER binding sites in baseline samples (n = 22) and after letrozole (n = 7) or megestrol combination treatment (arm B n = 9, arm C n = 6). A window of ±5 kb region flanking the tag midpoint is shown. D) Normalised ER coverage intensity over a ± 2.5 kb region flanking the tag midpoint is shown for each trial arm, and for pre-treatment samples. E and F) One sample pair included in the Arm A ChIP-seq cohort was excluded from the trial primary/secondary endpoint analysis due to late reporting of HER2 status (borderline amplified). HER2 amplification was not identified by whole genome sequencing for this patient. ChIP-seq analysis excluding this patient is presented - the pattern of decreased ER binding is maintained whether this sample pair was included or not (A vs B + C, p = 0.067, Mann Whitney U test). E) log transformed fold change in binding intensity across core ER sites (as defined in Ross-Innes et al 2012) for paired baseline and end of treatment samples, grouped by trial arm (n = 21 sample pairs). Box = IQR, centre = median, whiskers = min/max (excluding outliers, defined as <Q1 – 1.5*IQR or > Q3 + 1.5*IQR). F) Least squares lines summarising the distribution of baseline and EOT ER binding for patients in each trial arm (n = 21). G, H) Scatterplots showing the correlation of log2 transformed fold change in ER binding at core ER binding sites between Arm A vs Arm B (G) and Arm A vs Arm C (H). PCC = Pearson’s correlation coefficient, two-sided.
