Utility of the Minimum Apparent Diffusion Coefficient in Decreasing Unnecessary Biopsy Rates for Suspected Breast Magnetic Resonance Imaging Lesions: A Single-Centre, Retrospective Study

Abstract

Abstract Purpose: Apparent diffusion coefficient (ADC) helps identify malignant breast lesions and reduce unnecessary breast biopsies. However, previous research has primarily focused on the mean ADC (ADCmean) rather than the minimum ADC (ADCmin). Therefore, this study aimed to evaluate ADCmin’s ability to decrease unnecessary biopsy rates and determine the threshold of the ADCmin for clinical implementation. Methods: In this single-centre, retrospective study, we included 497 patients with 548 magnetic resonance imaging (MRI)-detected Breast Imaging Reporting and Data System (BI-RADS) 3, 4, or 5 lesions. Using a negative likelihood ratio of 0.1, we selected the ADC value with the highest specificity as the ADC threshold (ADCnl, including ADCnl-min and ADCnl-mean). Diagnostic performance indices were calculated using the receiver operating characteristic curve analysis. Differences in the area under the curves were assessed using the Hanley and McNeil test. Contingency table analysis was employed to compare the biopsy rate reduction efficacy. Results: The ADCmin showed comparable diagnostic performance to the ADCmean. The ADCmin and ADCmean values of 1.39×10−3 mm2/s and 1.52×10−3 mm2/s, respectively, were determined as thresholds. Upon applying the ADCnl-min and ADCnl-mean to the validation cohort, unnecessary biopsies of BI-RADS 4 and 5 lesions were reduced by 26.2% and 26.7%, respectively. The reduction in biopsies did not differ significantly between the ADCnl-min and ADCnl-mean. Conclusion: The ADCmin has the same ability to reduce unnecessary biopsies as the ADCmean. An ADCnl-min of 1.39×10−3 mm2/s or an ADCnl-mean of 1.52×10−3 mm2/s may help avoid one-quarter of unnecessary biopsies classified as BI-RADS 4 and 5 lesions.