Proteogenomic characterization of invasive breast tumors in young women

Abstract

Breast cancer in women <40, accounting for ~5% of all breast cancer cases diagnosed in the U.S., is more aggressive and associated with worse outcomes compared to breast cancer in older women. We performed a first-ever integrated proteogenomic study from a matched cohort of laser-microdissected tumors of 34 young (<40 years) and 34 older (≥60 years) women to identify molecular features that may underlie the worse outcomes in young women. Progression-free interval was shorter in young women, and their tumors were enriched for more aggressive molecular subtypes. Our multi-omic analysis identified distinct clusters between age groups in luminal but not basal-like cancers. Notably, GATA3 mutations were enriched in luminal tumors from young women while TP53 and PIK3CA mutations were more common in luminal tumors from older women. Young women's tumors exhibited lower estrogen receptor (ER) expression yet paradoxically enhanced ER response pathways and increased expression of tamoxifen-resistance-associated genes (IRS1, FERMT1). Immune pathway activity and immune scores were lower in tumors from young women, whereas proliferative and MYC pathways were notably elevated, identifying potential therapeutic targets. Transcriptomic data from TCGA and METABRIC confirmed our findings, with 10 of 11 observed pathways corroborated. Finally, differential expression of four immune-related surface proteins also suggested the potential of age-specific responses to immune-based therapies. Together, these findings may contribute to the understanding of the molecular mechanisms underlying worse outcomes in young women and offer new insight to therapeutic strategies.

Fig. 1. Clinicopathologic characteristics and survival plots in matched young vs. older women from APOLLO, TCGA, and METABRIC.

A Characteristic table. B K–M plots of progression-free interval. p_adj and HR_adj, p-value and hazard ratio adjusted for PAM50 subtype and histology type.

Fig. 2. Multi-omic features clustering in breast cancer tumors from matched young and older women in APOLLO.

The multi-omic clustering based on differentially enriched features using MOVICS. The age category and PAM50 subtype are annotated at the top. The differentially enriched features between tumors of young and older women include: 12 differentially mutated genes (Somatic Mutation, Firth logistic regression p < 0.1), with variant classes and hotspot mutations annotated in the plot; three Indels (Somatic Signature) exhibiting significant differences (GLM p < 0.05) or trends (GLM p < 0.1); Somatic copy number alteration (SCNA) of six regions exhibiting significant differences (GLM p < 0.05) or trends (GLM p < 0.1); 307 DEGs (Transcriptomics) (|logFC| > 1 i.e. FC of 2 and FDR < 25%); 17 DEPs (Proteomics) (|logFC| > 0.263 i.e. FC of 1.2 and FDR < 25%); 11 pathways (Gene Set Score) (GLM p < 0.1) based on their gene set enrichment scores identified from multi-omics gene-set analysis (MOGSA). Each SCNA region was annotated by a single cytoband by GISTIC, although some of them contained more than one cytoband, and the exact genomic coordinates of SCNA are in Table S2C.

Fig. 3. Kinase activity analyses using APOLLO data.

A and B The kinase activity determined by KSEA using phosphoproteomic and RPPA data, respectively. Red and blue indicate increased and decreased kinase activity in tumors from young women. C and D Networks of significantly positively associated kinase–substrate pairs were identified for tumors from young and older women, respectively, using global proteomic (for kinase), and phosphoproteomic and/or RPPA (for substrate) data.

Fig. 4. Multi-omic analysis of ESR1/ER expression in tumors from matched young and older women.

A–E and H are results from APOLLO. A Somatic copy number alteration (SCNA) of ESR1. B RNA expression of ESR1. C–E ER protein expression quantified by MS-proteomics, RPPA, and IHC, respectively. F and G RNA expression of ESR1 from TCGA and METABRIC data, respectively. H APOLLO ER phosphopeptide S118 data normalized by total ER protein from RPPA.

Fig. 5. Immune and stromal scores in tumors from matched young and older women by all the cases and PAM50 subtypes.

A–E and H–K are results from APOLLO. A Immune score comparisons. B–E 4 Immune cell types (activated dendritic cells, basophils, class-switched memory B-cells, and type 2 T-helper cells). F and G Immune score comparisons from TCGA and METABRIC data, respectively. H Stromal score comparisons. I–K 3 stromal cell types (smooth muscle, endothelial cells, and mesenchymal stem cells). L and M Stromal score comparisons from TCGA and METABRIC data, respectively.