The authors stated that a total of 71 eligible studies were identified, of which 69 were included in the analysis (2 studies could not be translated). The details of 69 included studies, which enrolled a total of 10,454 participants, were reported in the article. The intention-to-study population analysed was 8,059.
Study quality.
Patient characteristics were adequately described in 80% of studies; the reasons for patient withdrawal were reported in 29% of studies; a report of test performance characteristics was given in 46% of studies; 19% of studies incorporated measures of variability; and 51% of studies reported the use of radiocolloid for lymphatic mapping. Blinding of those interpreting the tests was not incorporated in the quality score since this was only reported in 2 studies. Fifty per cent of the studies scored 2 or less out of a total possible score of 5 points.
Successful mapping of SLNs.
The results showed that 96.35% of patients completing the planned studies had successfully mapped SLNs. The proportion of successfully mapped patients ranged from 41 to 100% across the included studies.
Test performance measures.
The FNR ranged from 0 to 29.4% across the studies; the pooled estimate using a random-effects model was 7.0% (95% CI: 5.2, 8.8). There was significant between-study heterogeneity (P<0.0001). The observed FNR decreased with increasing study size (P=0.046) and with increasing proportions of successfully mapped patients (P=0.001). Regression analyses indicated that the reporting of patient characteristics, reporting of withdrawals, use of measures of variability, and a proportion of successfully mapped patients greater than 90% were significantly associated with an FNR of less than 10%.
The PPN rate ranged from 0 to 25% across the studies; the pooled estimate using a fixed-effect model was 4.6% (95% CI: 3.8, 5.4). No significant between-study heterogeneity was observed (P=0.29). The observed PPN rate increased with increasing percentage of positive lymph node (P=0.033) and decreased with increasing proportion of successfully mapped patients (P<0.0001). Regression analyses indicated that the reporting of patient characteristics, proportion of successfully mapped patients greater than 90%, and decreasing proportion of positive lymph nodes were significantly associated with a PPN of less than 10%.
There were no statistically significant differences in test performance measures between studies that used radiocolloid mapping techniques and those that did not.