The performance and role of BI-RADS classification in predicting malignancy in non-palpable breast lesions.

INTRODUCTION: Breast cancer continues to be a substantial global health concern, constituting a substantial proportion of new cancers diagnosed. Breast Imaging Reporting and Data System (BI-RADS) is a standardised instrument that facilitates the appraisal of lesions in mammographic reporting. A hook wire localisation biopsy (HWLB) is frequently employed to diagnose suspicious microcalcifications or non-palpable lesions. The study aims to ascertain the probability of detecting malignancies in non-palpable breast lesions by examining the malignancy rate detected by BI-RADS classification and HWLB results. MATERIALS AND METHODS: This retrospective study utilised data from patients who underwent HWLB at a tertiary hospital from January 2016 to December 2021. RESULTS: Among the 65 patients analysed, 21.5% exhibited malignant lesions, predominantly detected through routine mammogram screening. There was a statistically significant association (p=0.007) between BI-RADS classification and the malignancy status. Using BI-RADS 4A as the threshold ensures 100% sensitivity but leads to a low specificity of 13.73%, Positive Predictive Value (PPV) of 24.1% and Positive Likelihood Ratio (LR+) of 1.2. While increasing the threshold to BI-RADS 4B improves specificity to 80.4%, reducing false positives. This suggests that many BI-RADS 4A cases are actually mostly benign. CONCLUSION: BI-RADS 4A notably exhibited a high false-positive rate, possibly due to benign lesions mimicking cancer radiologically and the operator-dependent nature of radiology. Despite these constraints, findings affirm the crucial role of hook wire localisation biopsy and BI-RADS classification in evaluating non-palpable breast lesions.