Predictive value of peripheral blood cell-free DNA breast cancer gene mutation profiling for postoperative pathological malignancy in BI-RADS 4 breast nodules.

To investigate the predictive value of peripheral blood cfDNA breast cancer gene mutation profiling for postoperative pathological malignancy in BI-RADS 4 breast nodules. Clinical data from 212 patients with BI-RADS 4 breast nodules at our hospital from January 2020 to January 2024 were retrospectively collected. Patients were divided into benign group (128 cases) and malignant group (84 cases) according to postoperative pathological results. Peripheral venous blood was collected within 1 week before surgery, cfDNA was extracted and targeted next-generation sequencing (NGS, coverage depth 500-1,000×) of 49 breast cancer-related genes was performed. cfDNA concentration, tumor mutation burden (TMB), mutation allele frequency (MAF), gene-specific mutation detection rates (BRCA1, TP53, ESR1, ERBB2, PIK3CA, etc.), and a composite Gene Mutation Score (GMS) were analyzed. Multivariate logistic regression analysis was used to identify independent predictors of postoperative pathological malignancy, and ROC curves were plotted to evaluate predictive efficacy. The malignant group had higher cfDNA concentration (14.3 ± 5.8 vs. 8.6 ± 3.2 ng/mL, P < 0.001) and TMB [2.86 (1.54,4.72) vs. 0.42 (0.28,0.68) mut/Mb, P < 0.001] than the benign group, while overall mutation detection rate (2.18% ± 0.63% vs. 2.84% ± 0.52%, P < 0.001), mutation coverage in promoter/regulatory regions of key driver genes (36.2 ± 9.5 vs. 45.6 ± 8.3 RPK, P < 0.001), and GMS (0.95 ± 0.74 vs. 1.82 ± 0.68, P < 0.001) were lower in the malignant group, indicating distinct mutation patterns. Mutation rates in BRCA1, TP53, ESR1, and ERBB2 in the malignant group were significantly higher than those in the benign group (P < 0.001). G3 grade patients had higher TMB and lower GMS, BRCA1/TP53 MAF, and regulatory region coverage than G1/G2 grade patients (P < 0.05). Multivariate logistic regression analysis showed that age (OR = 1.058, 95%CI: 1.021-1.096), BI-RADS classification (OR = 2.874, 95%CI: 1.643-5.027), TMB (OR = 4.326, 95%CI: 2.214-8.452), and GMS (OR = 0.178, 95%CI: 0.082-0.388) were independent predictors of postoperative pathological malignancy (P < 0.05). The combined model (age + BI-RADS classification + TMB + GMS) had an AUC of 0.923 (95%CI: 0.884-0.962), sensitivity of 88.1%, specificity of 89.1%, positive predictive value of 84.1%, and negative predictive value of 91.8%, superior to single indicators (P < 0.001). cfDNA breast cancer gene mutation profiling has good predictive value for postoperative pathological malignancy in BI-RADS 4 breast nodules, with TMB and GMS being independent predictors. The combined model integrating clinical, imaging, and gene mutation features demonstrates excellent predictive efficacy and can provide a non-invasive, accurate assessment tool for preoperative risk stratification of BI-RADS 4 lesions.