Most patients with melanoma have a thin (≤1.00 mm) lesion. There is uncertainty as to which patients with thin melanoma should undergo sentinel lymph node (SN) biopsy. We sought to quantify the proportion of SN metastases in patients with thin melanoma and to determine the pooled effect of high-risk features of the primary lesion on SN positivity.

Published literature between 1980 and 2015 was searched and critically appraised. Primary outcome was the proportion of SN metastases in patients with thin cutaneous melanoma. Secondary outcomes included the effect of high-risk pathological features of the primary lesion on the proportion of SN metastases. Summary measures were estimated by Mantel–Haenszel method using random effects meta-analyses.

Sixty studies (10,928 patients) met the criteria for inclusion. Pooled SN positivity was 4.5 % [95 % confidence interval (CI) 3.8–5.2 %]. Predictors of a positive SN were: thickness ≥0.75 mm [adjusted odds ratio (AOR) 1.90 (95 % CI 1.08–3.34); with a likelihood of SN metastases of 8.8 % (95 % CI 6.4–11.2 %)]; Clark level IV/V [AOR 2.24 (95 % CI 1.23–4.08); with a likelihood of 7.3 % (95 % CI 6.2–8.4 %)]; ≥1 mitoses/mm² [AOR 6.64 (95 % CI 2.77–15.88); pooled likelihood 8.8 % (95 % CI 6.2–11.4 %)]; and the presence of microsatellites [unadjusted OR 6.94 (95 % CI 2.13–22.60); likelihood 26.6 % (95 % CI 4.3–48.9 %)].

The pooled proportion of SN metastases in thin melanoma is 4.5 %. Thickness ≥0.75 mm, Clark level IV/V, mitoses, and microsatellites significantly increased the odds of SN positivity and should prompt strong consideration of SN biopsy.