Despite the small number of patients in our cohort, our results confirm previous studies on SLN analysis in melanoma [3, 4, 8–11], in terms of SLN identification rate (97%), percentage of SLN positive patients (22%) and percentage of additional positive CLND (25%). We also observed a significant association between positive SLN and primary tumor thickness and microscopic ulceration. Although only one SLN was harvested in 67% of our cases, the mean number of SLN harvested was 1.5 +/− 1 in our study, very similar to those found by previous studies [8, 12]. Furthermore, Gershenwald 7.  found that among the 343 patients who underwent CLND, the majority (72%) had only one positive SLN as compared to 67% in our study. To our knowledge, there are neither consensus nor recommendations on the minimal or maximal number of SLN that should be harvested during the procedure.
As expected, recurrences were significantly more frequent in SLN positive patients (32%) than in SLN negative patients (13%) suggesting a better regional control of melanoma progression after SLNB and immediate CLND. The rate of locoregional lymph node relapses after lymph node excision in SLN positive patients varied between 0 and 20% [4, 13]. Conversely, locoregional relapses after LND of clinically palpable lymph node palpation varied between 20 and 50% [13–15]. It is nonetheless still difficult to know whether better regional lymph node control is related to complementary CLND performed after positive SLN is discovered. Another study  retrospectively compared two groups of SLN positive patients from different hospitals, some of whom underwent additional CLND. The authors found no significant difference in terms of survival between the two groups. In our study, relapse and mortality rates in SLN positive patients were not influenced by the result of additional CLND casting further doubt on the benefits of complementary surgery.
The SLNB was considered false-negative if a primary recurrence developed in the regional lymph node basin from which a tumor-free SLN had been removed. In our study, the number of false-negative (10/153 = 6.5%) was similar (3-8%) to other studies [11, 16, 17]. However, there is ongoing debate on how to correctly calculate the false-negative rate. It should not be expressed as a percentage of the total population, but rather as the number of false-negative results divided by the number of false-negatives and true-positives . In our study, the false negative rate (10/54 = 18.5%) was also similar to the rate calculated in other studies (5.7% to 21%) . Beyond the technical problems associated with the SLN procedure, false negatives may be related to different factors: the time it takes to learn to perform the technique , lymphatic drainage disruption related to primary tumour excision, lymphatic obstruction by tumor cell embolism, the presence of a neck SLN , inadequate histological analysis and hematogenous dissemination.
The risk factors for recurrence after negative SLN are identical to those observed after positive SLN: presence or absence of macroscopic ulceration and tumoral thickness superior to 4 mm . Analysis of all relapses (Table 3) according to SLN status (positive or negative) showed a poor sensitivity (41%) and positive predictive value (32%) for SLN analysis but good specificity (81%) and negative predictive value (87%). Although rarely calculated, these values are similar to those found by Saltman et al. The same analysis focused on mortality confirmed the good specificity (80%) and negative predictive values (91%), i.e. similar to the findings of Morton et al..
Our overall survival rates at 5 years (69.7% and 90.3% for SLN positive and negative patients, respectively) were similar to previous prognostic values of SLN analysis [8, 20] when followed by additional CLND. A multivariate analysis of an international cohort of 2313 stage III patients  showed that the overall survival rate at 5 years was greater in patients presenting with micrometastases than in patients with palpable lymph node metastases (67% vs 43%). However, there were wide variations (23% to 87%) in patients with micrometastases depending on histological characteristics of primary melanoma (ulceration, mitotic index), its anatomical localization, number of SLN affected and patient age . A bayesian analysis was recently carried out of studies with more than 50 patients undergoing SLN between 1993 and 2010 . The authors focused on the prognostic benefit of the SLN analysis in terms of specific survival for melanoma with tumor thickness between 1 and 4 mm. For these patients, the risk of melanoma-related death varied between 26.2% and 31.6% when SLN was positive versus 9.7% and 15.6% when SLN was negative.
All of these results for relapse and survival involve the combination of two surgical procedures: research and analysis of SLN followed by additional CLND in the event of micrometastases in SLN. At the present time, no advantages of lymph node excision have been shown in terms of regional monitoring of metastatic damage or overall survival [13, 15, 21]. However, an additional lymph node excision is more often recommended when the SLN is invaded .
Many attempts have been made to predict non-SLN (NSLN) metastasis in the additional lymph node excision performed after positive SLN based on demographic, primary tumor, and SN features of patients with melanoma. Previous studies [9, 10, 23–25] indicate that overall SLN tumor burden, primary tumor thickness, and number of SLN harvested may be useful in identifying a group at low risk for positive NSLN. It is nonetheless interesting that these two parameters are also predictive for SLN micrometastases [20, 26]. However, marked variability in the correlation of individual features with NSLN metastases and the degree of this correlation has characterized the literature on this issue to date [27, 28]. There is currently no consensus regarding what degree of risk of NSLN involvement indicates that it is safe to forego CLND. Before elimination of CLND can be advocated, prospective clinical trials designed to assess the safety of omitting formal CLND with respect to survival and locoregional control in low-risk groups are needed. The ongoing Multicenter Selective Lymphadenectomy Trial II  which compares CLND versus close observation with sonography and clinical examination for patients with a positive SLN, should provide valuable information about which patients might be spared a CLND.
The frequency of post-operative complications (neuropathic pain, infection and lymphocele) observed in our study was similar to other studies in terms of morbidity related to SLN analysis .