Extended Data Fig. 2: Gene set enrichment analysis of gene expression data from primary human breast tumors and BC cell lines.
a–c, GSEA comparing primary tumors of patients who developed BMs and those who developed metastases in other organs. a, GSEA plot showing downregulation of a p53 expression signature in primary tumors that metastasized to the brain vs. bone (MBC cohort). Oncogenic signature ‘p53_DN.V1_DN’, brain vs. bone p = 0.005. Brain n = 7, bone = 24. b, GSEA plot showing upregulation of a signature indicative of decreased p53 signaling in primary BCs that metastasized to the brain in comparison to other sites in a second dataset (GSE12276). Brain vs. others, oncogenic signature ‘P53_DN.V2_UP’, p = 0.003. Brain n = 7, others n = 121. c, GSEA plots showing downregulation of a signature indicative of increased p53 signaling (left) and upregulation of a signature indicative of decreased p53 signaling (right) in triple-negative primary breast tumors that metastasized to the brain vs. those that did not (data source GSE76714)75. Brain vs. others, oncogenic signature: ‘P53_DN.V1_DN’ p = 0.018, oncogenic signatures: P53_DN.V2_UP p = 0.035, brain n = 21, others n = 38. See Supplementary Table 3. d, GSEA of primary BCs (TCGA cohort), comparing tumors with TP53 inactivation to TP53-WT tumors. ‘Smid_breast_cancer_relapse_in_brain_dn’, p = 0.001. TP53 inactivation: n = 623, TP53-WT: n = 259. e, ssGSEA of brain-relapse signatures in primary BCs with TP53 inactivation compared to TP53-WT tumors (MBC cohort). Tumors with TP53 inactivation showed increased expression of signatures positively correlated with BM. TP53 inactivation vs. TP53-WT, ‘Smid_breast_cancer_relapse_in_brain’ **p = 0.01, ‘Smid_breast_cancer_relapse_in_brain_up’ **p = 0.003, _DN p = 0.07. See Supplementary Table 4. f, Correlation between BM potential and p53 pathway activity levels in human BC. High potential for BM is strongly correlated with low p53 pathway activity (METABRIC cohort). Readout for p53 pathway activity: ssGSEA scores for ‘Kannan_TP53_targets’ gene set. Readout for brain-metastatic potential: ssGSEA scores for ‘Smid_breast_cancer_relapse_in_brain’ gene set. Two-sided Pearson’s correlation r = −0.41, p = 2.9 × 10−54. n = 1334 tumor samples. g, ssGSEA of brain-relapse signatures in primary BCs that metastasized to the brain compared to those that metastasized elsewhere. GSE12276. Brain vs. others, ‘Smid_breast_cancer_relapse_in_brain’ **p = 0.007, rank #1, ‘Smid_breast_cancer_relapse_in_brain_up’ **p = 0.008, rank #8, ‘Smid_breast_cancer_relapse_in_brain_dn’ ***p = 0.01, rank #8. Brain: n = 7, others: n = 121. See Supplementary Table 5. h,i, GSEA of BM gene expression gene sets in tumors with TP53 mutations, hotspot mutations or copy-number loss, compared to TP53-WT tumors, in the METABRIC cohort. GSEA showed an increase in gene expression that is upregulated in BM and a decrease in gene expression that is downregulated in BM, in (h) tumors with monoallelic TP53 inactivation vs. TP53-WT tumors; and (i) tumors with biallelic TP53 inactivation vs. TP53-WT tumors, independent of the mode of p53 inactivation. P-values and q-values indicated in the respective panels. TP53-WT: n = 693, TP53-WT/mut: n = 168, TP53-WT/Hotspot: n = 30, TP53-WT/del: n = 505, TP53-mut/del: n = 405, TP53-Hotspot/del: n = 60, TP53-del/del: n = 5.
