Work

Distribution and molecular drivers of the 21-gene recurrence score in early breast cancer with low, intermediate, and high estrogen receptor expression

Year: 2026

Type: article

Abstract: Background Estrogen receptor (ER)-low human epidermal growth factor receptor 2 (HER2)-negative breast cancer shares similar pathological and molecular features with triple-negative breast cancer, yet the utility of the 21-gene assay in this population remains uncertain. This study aimed to evaluate the distribution and molecular drivers of the 21-gene recurrence score (RS) in early breast cancer with low, intermediate, and high ER expression. Methods ER-positive, HER2-negative early breast cancer patients with 21-gene testing results from Shanghai Jiao Tong University Breast Cancer Database (2009-2022) were included. Eligible patients were categorized into three groups: ER-low (1-9%), ER-intermediate (10-50%), and ER-high (>50%). Chi-square and Kruskal-Wallis tests were conducted to compare the 21-gene RS and expression profiles among these groups. Spearman correlation and linear variance decomposition were performed to assess the associations between RS and its constituent gene modules. Results A total of 4754 patients were included (ER‑low: 70, 1.5%; ER‑intermediate: 129, 2.7%; ER‑high: 4555, 95.8%). Mean RS values were 39, 32, and 24 in the three groups, with RS >25 observed in 72.9%, 62.8%, and 42.5% of patients, respectively (P < 0.001).. Among patients with ER-low tumors, the rates of RS >25 were 62.5% and 81.6% for those ≤50 or >50 years, 66.7% and 83.3% for those with low or high clinical risk, and 61.6% and 87.1% for those with PR-positive or PR-negative diseases, respectively. Notably, 90.6% younger patients had RS >15 in the ER-low group. Expression of ESR1 ( P < 0.001), PGR ( P < 0.001), BCL2 ( P = 0.035), SCUBE2 ( P < 0.001) from the ER module, and STMY3 ( P = 0.001) from the invasion module were significantly lower in ER-low than in ER-high. Only, the proliferation module had a modest correlation with RS among patients with ER-low tumors (R = 0.34, P = 0.0076). Variance decomposition showed that the four modules explained only 8.83% of RS variance in ER‑low tumors (7.38% from proliferation), compared with 48.74% (42.93% from ER module) in ER‑high and 28.97% (21.94% from ER module) in ER‑intermediate tumors. Conclusions The 21-gene RS was higher in ER-low breast cancer patients, particularly among those ≤50 years, with high clinical risk or PR-negative diseases. RS was only moderately correlated with proliferation features and minimally determined by any of the four module components in ER-low tumors. These findings are exploratory and require further validation in larger cohorts.

Source: Frontiers in Immunology

DOI: 10.3389/fimmu.2026.1653477