Evolution and refinement of magnetically guided sentinel lymph node detection in breast cancer: meta-analysis.

Abstract

Superparamagnetic iron oxide nanoparticles (SPIO) have been used as a tracer for sentinel lymph node (SLN) localization in breast cancer, demonstrating comparable performance to the combination of radioisotope (RI) and blue dye (BD). A systematic literature search and meta-analysis with subgroup and meta-regression analysis were undertaken to update the available evidence, assess technique evolution, and define knowledge gaps. Recommendations were made using the GRADE approach. In 20 comparative studies, the detection rate was 97.5 per cent for SPIO and 96.5 per cent for RI ± BD (risk ratio 1.006, 95 per cent c.i. 0.992 to 1.019; P = 0.376, high-certainty evidence). Neoadjuvant therapy, injection site, injection volume or nodal metastasis burden did not affect the detection rate, but injection over 24 h before surgery increased the detection rate on meta-regression. Concordance was 99.0 per cent and reverse concordance 97.1 per cent (rate difference 0.003, 95 per cent c.i. -0.009 to 0.015; P = 0.656, high-certainty evidence). Use of SPIO led to retrieval of slightly more SLNs (pooled mean 1.96 versus 1.89) with a higher nodal detection rate (94.1 versus 83.5 per cent; RR 1.098, 1.058 to 1.140; P < 0.001; low-certainty evidence). In meta-regression, injection over 24 h before surgery increased the SPIO nodal yield over that of RI ± BD. The skin-staining rate was 30.8 per cent (very low-certainty evidence), and possibly prevented with use of smaller doses and peritumoral injection. The performance of SPIO is comparable to that of RI ± BD. Preoperative injection increases the detection rate and nodal yield, without affecting concordance. Whether skin staining and MRI artefacts are reduced by lower dose and peritumoral injection needs to be investigated.