Multiomics in primary and metastatic breast tumors from the AURORA US network finds microenvironment and epigenetic drivers of metastasis

Authors

Garcia-Recio S

Hinoue T

Wheeler GL

Kelly BJ

Garrido-Castro AC

Pascual T

De Cubas AA

Perou CM

Doi

10.1038/s43018-022-00491-x

PMID: 36585450 · DOI: 10.1038/s43018-022-00491-x · Journal: Nature Cancer (2023)

TL;DR

The AURORA US Metastasis Project performed multiplatform genomic profiling (RNA-seq, exome sequencing, low-pass WGS, and DNA methylation arrays) on 55 females with metastatic breast cancer (51 primary tumors and 102 metastases). The study found that expression subtype switching occurs in approximately 30% of cases, often coincident with DNA clonality shifts involving HER2. It identified epigenetic suppression of HLA-A through DNA hypermethylation and focal deletions in 17% of metastases, leading to reduced immune cell infiltrates especially in brain and liver metastases, with implications for immunotherapy and HER2-targeting therapies.

Cohort & data

  • 55 female participants with metastatic breast cancer; 51 primary cancers and 102 metastases assayed.
  • Cancer type: BRCA (including TNBC/basal-like and luminal/HER2E subtypes).
  • Dataset: brca_aurora_2023.
  • Assays: whole-exome-seq, whole-genome-seq (low-pass 5x), rna-seq (rRNA depletion), epic-methylation-array.
  • 88/153 specimens had all four assays; 141/153 had three of four.
  • Median age at primary diagnosis: 49 years; 18% African American, 7% Hispanic.
  • Metastatic sites: liver (n=28), lung (n=13), lymph nodes (n=12), brain (n=11), other (n=16).
  • 20 participants had samples collected at autopsy.

Key findings

  • Expression subtype switching between primary and metastasis in 13/39 (33%) of paired cases; basal-like was the most stable subtype (15/16 concordant), while luminal/HER2E subtypes switched in 8/19 individuals.
  • TP53 and FLG were more frequently mutated in metastases than primaries (66% vs 33%, P=0.006; 28% vs 3%, P=0.003), though not significant after FDR adjustment.
  • 11 DNA segments were more frequently amplified in metastases (q<0.05), including regions overlapping MYC and MDM4.
  • ESR1 mutations identified in metastases from 4 individuals, consistent with endocrine therapy resistance.
  • HLA-A promoter DNA hypermethylation found in 23 tumors (12 individuals); more frequent in metastases than primaries (P=0.035).
  • HLA-A focal deletions in 23 samples from 8 participants; 13 samples from 3 participants had deletions near an HLA gene (<40 kb).
  • Metastases with HLA-A methylation or focal deletion showed reduced immune signatures and higher predicted neoantigens (P=0.002 in basal-like tumors).
  • MHC class I metagene signature was consistently and significantly downregulated only in basal-like metastases across AURORA and RAP cohorts.
  • Liver metastases had the lowest immune cell features in 9/14 individuals with multiple metastases profiled.
  • Brain metastases showed 48 signatures lower than matched primaries, mostly immune/stromal.
  • DNA hypomethylation at distal ESR1/FOXA1 binding sites led to downregulation of cell-adhesion genes including JAM3 and FOXF1 in metastases.
  • In TCGA-BRCA, HLA-A methylation in primary tumors was associated with worse overall survival even after adjusting for stage and PAM50 subtype.

Genes & alterations

  • TP53 — most frequently mutated gene (43% of cohort); enriched in metastases (66% vs 33%).
  • ESR1 — activating mutations in 4 metastatic cases, linked to endocrine therapy resistance; distal ESR1 binding site hypomethylation associated with suppression of cell-adhesion genes.
  • ERBB2 — subtype switching involving HER2 amplification gain (participant AFR3: E668Q activating mutation) or loss (participant AFE4: deletion of HER2 amplification under trastuzumab pressure).
  • PIK3CA — mutated in 13% of cohort.
  • KMT2C — mutated in 13% of cohort.
  • PTEN, NF1, BRCA2 — each mutated in 10% of cohort.
  • HLA-A — epigenetically silenced via promoter hypermethylation and focal deletions in metastases; correlated with reduced HLA-A protein expression (Spearman rho=0.6, P=0.0001) and lower immune cell infiltrates.
  • B2M — focal deletions in 3 tumors from 3 individuals; methylation rare (1 tumor).
  • MYC, MDM4 — copy number amplifications enriched in metastases.
  • FOXA1 — binding sites at distal enhancers hypomethylated in metastases; implicated in estrogen-mediated suppression of cell-adhesion genes.
  • JAM3 — downregulated in metastases via both distal ESR1 binding site hypomethylation and promoter hypermethylation; validated in TCGA (P=4.1e-15).
  • GATA3 — mutated in 7% of cohort.

Clinical implications

  • HLA-A-low tumors (via methylation or deletion) may not respond to immune checkpoint inhibitors (ICIs) due to inability to present neoantigens to CD8+ T cells, despite high neoantigen burdens.
  • DNA demethylating agents combined with ICIs proposed as a biomarker-driven strategy for HLA-A-methylated tumors.
  • Liver biopsies may underestimate immune therapy biomarker positivity due to intrinsically lower immune cell features at this metastatic site.
  • Subtype switching (especially HER2 gain/loss) has direct implications for targeted therapy selection; biopsy of metastatic disease is recommended as molecular features may differ substantially from the primary tumor.
  • Trastuzumab resistance observed in participant AFE4, where the tumor evaded therapy by losing HER2 amplification.
  • Lower MHC class I gene expression observed after increased lines of metastatic therapy (P=0.047), suggesting treatment-associated immune evasion.

Limitations & open questions

  • Small sample size (n=55) likely underpowered for detecting somatic mutation frequency differences; TP53/FLG enrichment did not survive FDR correction.
  • Integration of fresh-frozen and FFPE specimens introduces potential technical confounders.
  • Participants received multiple lines of therapy (median 3, range 0-20), and treatment effects could not be directly evaluated.
  • Whether HLA-A methylation is a cause or consequence of immune pressure remains unresolved.
  • Lung metastases yielded no significant expression signatures even in the combined cohort, suggesting further sample size limitations.
  • The functional consequences of distal ESR1 binding site hypomethylation on cell adhesion and metastatic capacity require experimental validation.

Citations from this paper used in the wiki

  • “HLA-A methylation in 23 tumors (12 individuals), and HLA-A promoter methylation was significantly more frequent in metastases than in primary tumors when comparing unpaired data (P = 0.035)”
  • “In 9 of 14 individuals, the lowest levels of the GP2-immune-metagene signature were in liver metastases”
  • “TP53 and FLG genes were more frequently mutated in metastases than in primary tumors (66% versus 33% (P = 0.006) and 28% versus 3% (P = 0.003), respectively)”
  • “Of the 39 RNAseq cases tested, 13 of 39 showed subtype ‘switching’ between a primary tumor and its metastasis”
  • “In 17% of metastases, DNA hypermethylation and/or focal deletions were identified near HLA-A and were associated with reduced expression and lower immune cell infiltrates”
  • “ICIs may have little effect on these HLA-A-low tumors, as these cannot be recognized by CD8+ T cells (noting these HLA-A-methylated tumors tend to have high neoantigen burdens)”

This page was processed by crosslinker on 2026-05-05.