A comparison of the accuracy of two minimally invasive breast biopsy methods: a systematic literature review and meta-analysis

Objective: The primary objective was to quantify and compare the accuracy and failure rates of directional vacuum assisted biopsy (DVAB) and core needle biopsy (CNB) when used under stereotactic (ST) guidance to biopsy suspicious breast lesions identified with screening mammography. Methods: We performed a systematic review of the literature published from January 1996 to July 2004, reporting all-comers populations in Western-style health care systems (i.e., North America, Europe, Australia or New Zealand), referred after screening mammography for breast biopsy using DVAB or CNB under ST guidance. Meta-analyses were conducted for DVAB and CNB, using open surgical biopsy and/or long-term clinical and/or mammogram follow-up as the diagnostic reference standard. The main outcomes of interest were those of greatest clinical relevance, i.e., miss rates and underestimation rates for malignancy. Also, technical failure rate and non-diagnostic rate were assessed for each biopsy method. Results: Thirty-five studies qualified for the review. There were 12 studies with a DVAB group (n=5,119 patients), and 25 studies with a CNB group (n=6,236). There were no studies including both a DVAB and a CNB group, thus precluding any direct, within-study comparisons of accuracy. Overall agreement rate between DVAB and the reference standard was 97.3%, and between CNB and the reference standard, 93.5%. The frequency of technical failures with CNB was slightly higher than DVAB (5.7 vs. 1.5%), as was the frequency of non-diagnostic samples (2.1 vs. 0%). Of the non-diagnostic CNB samples, 23% were subsequently found to be malignant on reference standard. In multivariate analyses using four covariates (procedure type, geographic location, reference standard, and patient position), there were no significant predictors of agreement rates, but some variables were significant predictors of miss rates. For benign to malignant upgrades, study location was a significant predictor, with more upgrades in non-NA locations. For atypia to malignant upgrades, the type of procedure was a significant predictor, with more underestimations in CNB studies. Conclusion: The best available evidence suggests that, in screening populations referred for minimally invasive breast biopsy biopsy requiring ST guidance, DVAB may provide lower miss and underestimation rates for clinically relevant diagnoses than does CNB.