Background: No consensus exists on the validity of the sentinel-lymph-node procedure for assessment of nodal status in patients with colorectal cancer. We aimed to assess the diagnostic performance of this procedure.
Methods: We searched Embase and PubMed databases for studies published before March 20, 2010. Eligible studies had a prospective design, a sample size of at least 20 patients, and reported the rate of sentinel-lymph-node positivity. Individual patient data were requested for localisation and T-stage stratification. A subset of reports with high methodological quality was selected and analysed.
Findings: We identified 52 eligible studies, which included 3767 sentinel-lymph-node procedures (2961 [78·6%] colon and 806 [21·4%] rectal carcinomas). Most tumours 2339 (62·1%) were stage T3 or T4. 1887 (50·1%) of patients were male, 1880 (49·9%) female. Mean overall weighted-detection rate was 0·94 (95% CI 0·92-0·95), at a pooled sensitivity of 0·76 (0·72-0·80) with limited heterogeneity (χ(2)=286·08, degrees of freedom=51; p=0·003). A mean weighted upstaging of 0·15 (95% CI 0·12-0·19) was noted. Individual patient data were available from 19 studies that included 1168 patients. Analysis of these data showed no significant difference in sensitivity between colon (0·86 [95% CI 0·83-0·90]) and rectal cancer (0·82 [0·77-0·88]; p=0·23). Also, there was no dependency of sensitivity on T stage for both colon (pT1: 0·79 [95% CI 0·73-0·84], pT2: 0·76 [0·62-0·90], pT3: 0·73 [0·59-0·87], pT4: 0·73 [0·53-0·93]) and rectal cancer (T1 or T2: 0·81 [0·52-0·94] vs T3 or T4: 0·80 [0·51-0·93]). The subgroup of eight studies with high methodological quality showed a mean detection rate of 0·96 (95% CI 0·90-0·99) for colonic tumours and 0·95 (0·75-0·99) for rectal tumours, and a mean sensitivity of 0·90 (95% CI 0·86-0·93) for colonic tumours and 0·82 (0·60-0·93) for rectal tumours.
Interpretation: The sentinel-lymph-node procedure shows a low sensitivity, regardless of T stage, localisation, or pathological technique. For every patient diagnosed with colon or rectal cancer without clinical evidence of lymph-node involvement or metastatic disease, this procedure in addition to conventional resection should be considered, since the prognostic information provided by this technique could be clinically significant.
Funding: Cancer Center Amsterdam Foundation, Amsterdam, Netherlands.
Copyright © 2011 Elsevier Ltd. All rights reserved.