Strengths and Limitations
This study has several strengths. First, a meticulous screening was carried out to ensure that the data for analyzing did not contain any reduplicative or ineligible individual. On the other hand, we had done our best to re-analyze the existing data, collect unpublished studies and contact the authors to obtain precise diagnostic information for each included individual (supplementary p10–16). Each step of data synthesis was detailed to make our results reproducible. Besides, most of the included studies were prospective, used multiple tracers for SLNB and ultrastaging for DPE, indicating low risk of bias in the results. Furthermore, the L and E-protocol can be clearly distinguished and were evenly used in previous studies, suggesting the generalizability for both. In addition, the studies in which FSE was selectively performed (usually for SLNs with suggestive appearance) were excluded because there were high risks of bias (supplementary). Finally, the sensitivity analyses demonstrated the robustness of the main findings (supplementary).
This study also has several limitations. First, only nine studies provided the sizes of TP and the question how many MIM/ITC were detected under different protocols has not been satisfactorily answered. Furthermore, due to the multidisciplinary nature of SLNB technique, the sample size of single study was usually small. We could not eliminate the impact of pathologists’ experience although all involved institutions were highly specialized. This also poses a further question whether our results can be validated in hospitals without enough experience in SLNB. Besides, due to the length of time elapsed, some authors could not be contacted or provide requested information, including the diagnostic data, protocols of FSE and DPE, and the metastatic types.