Comprehensive molecular portraits of human breast tumors

Author

The Cancer Genome Atlas Network

Doi

10.1038/nature11412

PMID: 23000897 · DOI: 10.1038/nature11412 · Journal: Nature (2012)

TL;DR

The Cancer Genome Atlas Network performed a comprehensive multi-platform molecular characterization of 825 primary breast cancers using genomic DNA copy number arrays, DNA methylation, exome sequencing, mRNA arrays, microRNA sequencing, and reverse phase protein arrays. Integrating data across platforms confirmed four main breast cancer subtypes (Luminal A, Luminal B, HER2-Enriched, Basal-like) with distinct mutational spectra, copy number landscapes, and pathway activities. Only three genes (TP53, PIK3CA, GATA3) were mutated at >10% across all breast cancers, and Basal-like breast cancers showed striking molecular commonalities with high-grade serous ovarian carcinomas.

Cohort & data

  • 825 patients with primary breast cancer; 466 tumors assayed across 5 platforms, 348 across all 6 platforms.
  • Cancer type: BRCA.
  • Dataset: brca_tcga_pub.
  • Platforms: Agilent mRNA expression microarrays (n=547), Illumina Infinium DNA methylation (n=802), Affymetrix 6.0 SNP arrays (n=773), microRNA sequencing (n=697), whole exome sequencing (n=507), Reverse Phase Protein Arrays (n=403).
  • PAM50 classifier used for mRNA subtype assignment.

Key findings

  • 35 significantly mutated genes (SMGs) identified by exome sequencing of 510 tumors (30,626 somatic mutations total).
  • TP53 mutations: 80% Basal-like, 72% HER2E, 29% Luminal B, 12% Luminal A.
  • PIK3CA mutations: 45% Luminal A, 39% HER2E, 29% Luminal B, 9% Basal-like.
  • GATA3 mutations enriched in Luminal A; hot-spot 2bp intron 4 deletion exclusively in Luminal A (13/13 mutants).
  • MAP3K1 and MAP2K4 mutations (12% of Luminal A tumors) represent inactivation of contiguous p38/JNK1 pathway steps; mutually exclusive with each other.
  • Novel SMGs identified: TBX3, RUNX1, CBFB, AFF2, PIK3R1, PTPN22, PTPRD, NF1, SF3B1, CCND3.
  • Statistically significant mutual exclusivity among PIK3R1, PIK3CA, PTEN, and AKT1 mutations (P = 0.025).
  • PI3K pathway activation markers (pAKT, pS6, p4EBP1 by RPPA) were highest in Basal-like and HER2E subtypes despite high PIK3CA mutation frequency in Luminal A, suggesting a disconnect between PIK3CA mutation and pathway activation in Luminal A.
  • Basal-like cancers share molecular features with high-grade serous ovarian carcinomas: TP53 mutations, RB1/BRCA1 loss, MYC amplification, high genomic instability, similar copy number landscapes.
  • Two types of clinically HER2-positive disease identified: HER2E-mRNA-subtype/HER2+ (strong HER1/p-HER1/HER2/p-HER2 RPPA signal) vs. Luminal-mRNA-subtype/HER2+ (higher ER/GATA3/BCL2 expression).
  • ~10% of sporadic breast cancers harbored germline predisposing variants (47/507 patients; 9 genes including BRCA1, BRCA2, ATM, CHEK2, BRIP1).
  • ~20% of Basal-like tumors had germline and/or somatic BRCA1/BRCA2 variants.

Genes & alterations

  • TP53: Nonsense/frame-shift mutations predominate in Basal-like; missense in Luminal subtypes.
  • PIK3CA: E545K mutation almost exclusively Luminal A (25/27); amplification in 49% Basal-like.
  • GATA3: Intron 4 hotspot deletion (Luminal A); exon 5 frame-shift mutations (Luminal B).
  • ERBB2: DNA amplification (80% HER2E); 4/8 somatic variants in lobular cancers, 3 in TK domain.
  • CDH1: Mutations common in lobular histology (30/36); associated with lower mRNA/protein expression.
  • MAP3K1: Inactivating mutations in Luminal A (14%); p38/JNK1 pathway.
  • PTEN: Loss more common in Basal-like; activates PI3K pathway.
  • INPP4B: Loss in 30% Basal-like and HER2E; sensitizes to PI3K inhibitors.
  • RB1: Loss/mutation in 20% Basal-like; CCNE1 amplification is mutually exclusive.
  • CCND1: Amplification preferentially in Luminal tumors (29% Luminal A, 58% Luminal B).
  • MYC: Amplification across subtypes; hyperactivation characteristic of Basal-like.
  • KRAS (32%), BRAF (30%), EGFR (23%): Amplified (not typically mutated) in Basal-like.
  • MDM2: Amplification more frequent in Luminal B.
  • SF3B1: 15 non-silent mutations including recurrent K700E (4 cases).

Clinical implications

  • PIK3CA mutations in Luminal/ER+ cancers suggest potential benefit from PI3K/AKT pathway inhibitors, though pathway activation biomarkers are paradoxically low in Luminal A.
  • ~20% of Basal-like patients with BRCA1/2 variants may benefit from PARP inhibitors (e.g., olaparib) and/or platinum compounds.
  • Two subtypes of clinically HER2+ disease may respond differently to anti-HER2 therapies (trastuzumab, pertuzumab); HER2E subtype shows co-activation of HER1/EGFR suggesting dual targeting.
  • Molecular similarity between Basal-like breast and serous ovarian cancers suggests shared therapeutic opportunities (platinum analogs, taxanes).
  • PTEN and INPP4B loss in Basal-like cancers may sensitize to PI3K pathway inhibitors.
  • Copy number amplifications in Basal-like (FGFR1/2, IGF1R, KIT, MET, PDGFRA) represent potential druggable targets.
  • CDK4/Cyclin D1 amplification in Luminal B tumors suggests CDK4/6 inhibitor sensitivity.

Limitations & open questions

  • Short median follow-up (17 months) with few overall survival events (93/818) precluded survival analyses.
  • Normal-like (n=8) and Claudin-low (n=8) subtypes too small for separate analysis.
  • The disconnect between PIK3CA mutation and PI3K pathway activation in Luminal A remains unexplained; functional implications for therapy response are unclear.
  • The two “Reactive” RPPA subgroups may reflect stromal/microenvironmental elements rather than tumor-intrinsic biology.
  • Treatment data and response outcomes were not integrated; therapeutic hypotheses remain to be tested in clinical trials.

Citations from this paper used in the wiki

  • “Somatic mutations in only three genes (TP53, PIK3CA and GATA3) occurred at > 10% incidence across all breast cancers” (Summary).
  • “TP53 mutations occurred in 80% of cases” in Basal-like breast cancers (Results, mRNA-subtype association).
  • “PIK3CA (45%)” in Luminal A; “we observed a statistically significant exclusion pattern among PIK3R1, PIK3CA, PTEN, and AKT1 mutations (P = 0.025)” (SMG analysis).
  • “~20% of Basal-like tumors had a germline (n=12) and/or somatic (n=8) BRCA1 or BRCA2 variant, which suggests 1 in 5 Basal-like patients might benefit from PARP inhibitors and/or platinum compounds” (Basal-like summary).
  • “Comparison of Basal-like breast tumors with high-grade Serous Ovarian tumors showed many molecular commonalities, suggesting a related etiology and similar therapeutic opportunities” (Summary).

This page was processed by paper-compiler on 2026-05-06.