Abstract
BACKGROUND: The aim of this study was to validate the generalisability of the ECOG-ACRIN A6702 multicenter trial identified mean apparent diffusion coefficient (ADC) cutoff (1.53 × 10(- 3)mm(2)/s) and compare its biopsy reduction efficacy against a data-derived ADC minimum (ADC(min)) cutoff. METHODS: This dual-cohort retrospective analysis included 453 patients with 494 BI-RADS 4/5 lesions (derivation cohort: 288 patients/311 lesions; validation: 165 patients/183 lesions). ADC(min) and mean ADC (ADC(mean)) were measured, the data-derived ADC(min) cutoff was optimised via an ROC analysis (negative likelihood ratio ≤ 0.1) .Performance metrics included biopsy reduction rates, sensitivity, and false-negative rates, stratified by lesion type, size, BI-RADS category, and field strength. RESULTS: In the derivation cohort, the data-derived ADC(min) cutoff (1.39 × 10(- 3)mm(2)/s) showed comparable performance to the A6702 ADC(mean) (ADC(ACRIN)) cutoff in overall (24.7% vs. 24.1%), benign (47.9% vs. 45.1%) biopsy reduction and sensitivity (95.2% vs. 95.2%, all P > 0.05), with its robustness confirmed by the validation cohort (26.2% overall, 57.9% benign reduction, 96.3% sensitivity). Both cutoffs exhibited field strength dependency, with higher biopsy reduction at 1.5T vs. 3.0T (derivation: 30.1% vs. 17.8% for ADC(min), P = 0.01; validation: 28.7% vs.17.5% for ADC(min)). CONCLUSIONS: The data-derived ADC(min) cutoff (1.39 × 10(- 3)mm(2)/s) demonstrated non-inferiority to the ADC(ACRIN) cutoff in biopsy reduction while preserving sensitivity. Observed performance variations between 1.5T and 3.0T suggest that field strength and other technical factors may influence cutoff efficacy. Prospective multicenter studies with standardized protocols are needed to validate its generalisability and clarify the role of field strength.