Objective To explore the ARID1A mutation profile and its correlation with clinical prognosis in Chinese breast cancer patients.

Methods A total of 1 944 primary breast cancer patients from the METABRIC database and 1068 breast cancer patients were enrolled in the METABRIC cohort and The Cancer Genome Atlas (TCGA) cohort, respectively. Additionally, 874 female patients with invasive breast cancer from Guangdong Provincial People's Hospital (GDPH) were included in the GDPH cohort. A gene mutation waterfall plot was generated for the GDPH cohort, and ARID1A mutation status across different cohorts were compared.

Results The proportions of patients carrying ARID1A mutations in the GDPH, TCGA, and METABRIC cohorts were 5.8%, 3.7%, and 4.9%, respectively. Nonsense, missense, and frameshift mutations were the most common types of ARID1A mutations across all three cohorts. In the GDPH cohort, 6.3% of estrogen receptor (ER)-positive patients exhibited ARID1A mutations, which was higher than the proportion of ER-positive patients with ARID1A mutations in other cohorts, and 10 novel mutations were identified. Among the four molecular subtypes in the METABRIC cohort, luminal B breast cancer patients with ARID1A mutations accounted for the highest proportion (P=0.022). In the METABRIC cohort, the median survival time of HR(+) breast cancer patients with ARID1A mutations was shorter than that of wild-type patients, with a statistically significant difference (P=0.023). However, univariate and multivariate analyses revealed that ARID1A mutation was not an independent risk factor for survival in HR(+) breast cancer patients.

Conclusion There may be differences in ARID1A mutations among the GDPH, TCGA, and METABRIC cohorts patients, and ARID1A mutation may not be a risk factor for prognosis in breast cancer patients.