Sentinel-lymph-node procedure in colon and rectal cancer: a systematic review and meta-analysis

Summary

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 (χ²=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.

Introduction

Colorectal cancer is the second most common malignancy in developed countries and is the fourth leading cause of cancer-related death worldwide, accounting for 610 000 deaths in 2008.¹ Diagnosis is made by endoscopic biopsy or polypectomy. For malignant disease, segmental resection with en-bloc resection of lymph nodes is standard treatment. Although submucosal resection of early colorectal cancer is now proposed as a sufficient treatment, the resulting uncertainty about undetected lymph-node metastases is a serious drawback.²

Regional lymph-node metastasis represents one of the most important indications for adjuvant chemotherapy.³ 5-year survival is 90% for patients with stage I disease and 75% for those with stage II disease.⁴ By definition, patients with stage I and II disease are node-negative; however, up to 30% will develop distant metastases and eventually die from colorectal cancer.⁴ Understaging during the initial procedure probably contributes to this mortality. Such understaging is most probably a result of occult tumour cells (ie, tumour-cell deposits <0·2 mm) and micrometastasis (ie, tumour-cell deposits of 0·2–2·0 mm) in the lymph nodes, which are easily missed by routine histopathological examination.⁵ Currently, only one 4 μm thick haematoxylin-and-eosin-stained section per 5 mm of each lymph node is assessed for metastasis. Serial sectioning and additional immunohistochemistry or reverse transcriptase (RT)-PCR could diagnose lymphatic spread more accurately. Ideally, all regional lymph nodes should be examined with these techniques, but this would be too expensive and time consuming and therefore not feasible in everyday practice. The sentinel-lymph-node concept could offer a solution.

Since Cabanas⁶ showed the existence of a so-called sentinel lymph node, this concept has been used effectively to detect lymph-node metastasis on the basis of selective lymph-node biopsy in patients with penile carcinoma. He confirmed this lymph node to be the first site of metastasis, and showed that in clinically non-suspicious nodes it was frequently the only lymph node affected.

The main advantage of the sentinel-lymph-node procedure in colorectal cancer is that metastases, occult tumour cells, and micrometastases could be detected by careful analyses of a restricted number of lymph nodes, thus improving cancer staging.7, 8 The term upstaging is typically used to describe patients with a positive sentinel lymph node identified with histopathological techniques other than the conventional method (single-section haematoxylin and eosin staining), in the absence of positive non-sentinel lymph nodes assessed by conventional haematoxylin and eosin staining. These patients might benefit from adjuvant chemotherapy; however, there is currently a paucity of evidence to support such tailored treatment because most studies have insufficient follow-up data. In patients with breast cancer, targeted removal of sentinel nodes in combination with improved adjuvant therapy leads to low rates of axillary recurrence without a formal axillary dissection.⁹

We did a meta-analysis to assess the diagnostic performance of the sentinel-lymph-node procedure in terms of sensitivity and detection rate. We requested individual patient data from each included study, which enabled detailed analyses in addition to the main analysis. We stratified for tumour localisation, because colon and rectal cancer show different biological behaviour, and T stage, because studies of melanoma and breast cancer suggest that stage is inversely related to the diagnostic performance of the procedure.¹⁰