Yapay Zeka Hızlı Bakış & Çıkarımlar
AI InsightThese findings support a combined clinical-radiological approach to optimize decision-making in this challenging patient population.
📌 3 Maddede Bu Makale
Abstract
Aims: Breast lesions in women aged ≤40 years present unique diagnostic challenges due to dense breast tissue, lack of routine screening, and overlapping imaging features between benign and malignant conditions. Although the Breast Imaging Reporting and Data System (BI-RADS) is widely used for risk stratification, its performance and the added value of clinical parameters in younger populations remain incompletely defined. Methods: This retrospective, single-center study included 316 women aged ≤40 years who underwent core needle biopsy for breast lesions. After excluding high-risk (B3) lesions and non-diagnostic cases, 285 patients (247 benign, 38 malignant) were analyzed. Clinical, radiological, and pathological data were evaluated. The diagnostic performance of BI-RADS was assessed, and a multivariate logistic regression model was constructed to identify independent predictors of malignancy. Model performance was evaluated using receiver operating characteristic (ROC) analysis and calibration testing. Results: The overall malignancy rate was 13.3% (38/285). Malignant cases were significantly associated with older age within the cohort, palpable mass, axillary lymph node positivity, and suspicious ultrasonographic margin characteristics (all p<0.05). BI-RADS classification demonstrated a strong stepwise association with malignancy, with category-specific positive predictive values rising from 0.7% in BI-RADS 3 and 14.4% in 4A to 90.0% in 4B, with high rates likewise observed in 4C (85.7%) and BI-RADS 5 (77.8%). In multivariate analysis, presentation with a palpable mass (OR 7.41, 95% CI 2.20-24.94; p=0.001), axillary lymph node positivity (OR 3.16, 95% CI 1.09-9.22; p=0.035), and BI-RADS category (OR 25.53, 95% CI 5.60-116.48; p<0.001) remained independent predictors of malignancy, while age lost significance. The predictive model demonstrated excellent discrimination (AUC=0.913) and a higher AUC compared to BI-RADS alone (AUC=0.867), although the difference did not reach conventional statistical significance (DeLong’s test, p=0.050). Conclusion: BI-RADS classification is a strong predictor of malignancy in women aged ≤40 years, demonstrating a clear stepwise increase in risk across categories. The integration of clinical parameters provides incremental predictive value and improves diagnostic performance. These findings support a combined clinical-radiological approach to optimize decision-making in this challenging patient population.