This review concluded that the estimated risk of nodal recurrence after a negative sentinel lymph node biopsy was less than 5% and supported use of this technology for staging patients with melanoma. The possibility of missing studies and limitations in the quality assessment and analysis mean that these conclusions should be interpreted with caution.

To perform a meta-analysis of all published studies of sentinel lymph node biopsy for staging patients with melanoma.

MEDLINE and EMBASE were searched from 1990 to December 2009 without language restrictions. Search terms were reported. References of selected articles were screened. Abstracts for two relevant conferences were searched for the preceding five years. The authors stated that experts were contacted to locate unpublished studies. The review was restricted to published studies.

Studies that evaluated the accuracy of sentinel lymph node biopsy in patients with melanoma were eligible for inclusion. Studies had to report on the number of patients in whom sentinel lymph node biopsy was attempted and the number who were successfully mapped. Continuous follow-up of those with a negative biopsy was required. Studies with fewer than 50 patients, those restricted to patients with a positive biopsy and those that referred to a very specific population or location were excluded.

Across studies, 53.4% of patients were male and the median age was 54 years (range two to 97). The mean Breslow thickness was 2.17mm. An average of 22% of patients had ulcerated melanomas. All studies used scintigraphy. Most studies used blues dies to map the sentinel lymph node.

The authors did not state how studies were selected for inclusion.

Two reviewers independently assessed study quality using the eight-item Methodological Index for Non-Randomised Studies to assign a score out of 16 points.

Two reviewers independently extracted data on the proportion of patients successfully mapped, the false negative rate and the probability of nodal recurrence in patients with positive and negative biopsies. Disagreements were resolved through discussion or referral to a third reviewer.

Summary estimates were calculated for each of the outcomes of interest. The Mantel-Haenszel fixed-effect model was used where there was no heterogeneity and a random-effects model was used where heterogeneity was detected. Heterogeneity was assessed using the Q statistic and investigated using meta-regression. The authors stated that subgroup analyses were planned a priori but the variables assessed were not specified in the methods.

Seventy-one studies were included (25,240 participants, range 54 to 1,599). Summary quality scores ranged from 7 to 15.5. Median duration of follow-up was 33 months (range seven to 72 months).

Nineteen studies were restricted to patients in whom at least one sentinel lymph node was removed. In the other studies the proportion of patients who had at least one sentinel lymph node extracted was 97.5%. The proportion of patients in whom the sentinel lymph node was successfully mapped ranged from 88% to 100% (52 studies) with a pooled estimate of 98.1% (95% CI 97.3% to 98.6%). The proportion was higher in more recent studies (p<0.001), studies with higher quality scores (p=0.02) and for women (p=0.03) and was associated with mean age (p<0.001) and the proportion of ulcerated tumours (p=0.001).

The false negative rate ranged from zero to 34% (69 studies) with a pooled estimate of 12.5% (95% CI 11% to 14.2%). The false negative rate was inversely associated with the proportion of patients successfully mapped (p=0.001) and positively associated with duration of follow-up (0=0.003) and study quality score (p=0.02).

The probability of nodal recurrence in patients with a negative biopsy ranged from zero to 10.4% with a summary estimate of 3.4% (95% CI 3% to 3.8%) across studies. There was a negative association with the proportion successfully mapped (p<0.001) and positive association with length of follow-up (p=0.02), younger age (p=0.02), proportion of females (p=0.008), mean Breslow thickness (p=0.02) proportion of ulcerated tumours (p=0.007) and the false negative rate (p<0.001).

The probability of nodal recurrence in patients with a positive biopsy ranged from zero to 35% with a summary estimate of 7.5% (95% CI 5.9% to 9.4%). There was a positive association with older age (p=0.03) and greater false negative rate (p=0.02) and inverse association with the proportion of patients with ulceration (p=0.004).

Results stratified based on distant and all recurrences were reported.

The estimated risk of nodal recurrence after a negative sentinel lymph node biopsy was less than 5%, which supported the use of this technology for staging patients with melanoma.

The review addressed a clear question. Inclusion criteria were defined, but some aspects lacked clarity (such as the exclusion of studies in "very specific populations or locations"). The literature search was adequate for published studies. Some steps were reported to locate unpublished data, but the review was restricted to published studies and so publication bias was a possibility. Appropriate steps were taken to minimise bias and errors when extracting data and assessing quality; methods used to select studies for inclusion were not reported. Study quality was formally assessed, but the criteria were designed for non-randomised studies and were not listed; criteria for diagnostic accuracy studies may have been more appropriate. Results of the assessment were reported only as summary quality scores (which are problematic) and so the quality of the included studies remained unclear.

Methods used to pool data were not based on the most sophisticated models. Prevalence of the target condition was not considered when pooling predictive values. The results of the heterogeneity assessment were not reported. It appeared that the subgroup analyses were appropriate, but not all items considered were listed and results were presented only for variables found to show significant associations.

The possibility of missing studies and limitations in the quality assessment and analysis mean that the authors' conclusions should be interpreted with caution.

Practice: The authors stated that this review supported use of sentinel lymph node biopsy for staging patients with melanoma.

Research: The authors did not state any implications for research.

Not stated.

This is a critical abstract of a systematic review that meets the criteria for inclusion on DARE. Each critical abstract contains a brief summary of the review methods, results and conclusions followed by a detailed critical assessment on the reliability of the review and the conclusions drawn.