Sentinel lymph node biopsy for cutaneous melanoma: results of 10 years’ experience in two regional training hospitals in the Netherlands
- First Online:
- Cite this article as:
- van den Broek, F.J., Sloots, P.C., de Waard, J.D. et al. Int J Clin Oncol (2013) 18: 428. doi:10.1007/s10147-012-0399-3
- 162 Views
Background and objective
The Multicenter Selective Lymphadenectomy Trial (MSLT-I) demonstrated that the sentinel node (SN) status in cutaneous melanoma affects prognosis and that completion lymphadenectomy in SN-positive patients may improve survival. Our objective was to evaluate sentinel lymph node biopsy (SLNB) in two regional hospitals in the Netherlands.
Patients with localized melanoma were planned for wide excision and SLNB. Completion lymphadenectomy was recommended for positive SN status. Data were compared with the MSLT-I.
A median of 2 (1–7) SNs were identified in 305 patients and complications occurred in 11%. Fifty-four patients (18%) demonstrated SN metastases and 45 underwent completion lymphadenectomy (20% additional metastases). Six patients with initially negative SN developed lymph node metastases (sensitivity 90%). Overall disease-free survival was 83% (SN-negative 91% vs. SN-positive 41%; p < 0.001) and melanoma-specific survival was 93% (SN-negative 97% vs. SN-positive 62%; p < 0.001). Multivariate regression analysis revealed the SN status to be the most significant predictor for recurrence and melanoma-related death.
Our results of SLNB are comparable to data from high-volume centers participating in MSLT-I. From a patient perspective, the false-negative SN rate of 10% and complication rate of 11% should be weighed against being informed about prognosis and having a possible therapeutic benefit from completion lymphadenectomy.
KeywordsMelanomaSentinel lymph node biopsyNeoplasm stagingLymphatic metastasisSurvival analysisSkin neoplasms
Sentinel lymph node biopsy
Lymph node dissection
Over 90% of all patients with a cutaneous melanoma present with clinical stage I/II disease . The prognosis with regard to recurrence and survival is determined by primary tumor thickness (Breslow) or level of skin invasion (Clark), the presence of ulceration and the status of the locoregional lymph nodes draining the primary [2, 3]. Until about 15 years ago, the status of the regional lymph nodes could only be assessed by an elective lymph node dissection plus histopathological evaluation of all resected lymph nodes . This procedure is associated with high morbidity and high costs . Furthermore, elective lymph node dissection only appears to be associated with a rather modest survival benefit in patients who harbor or have an increased risk of nodal (micro)metastases [6, 7].
Since the introduction of lymphatic mapping and sentinel lymph node biopsy (SLNB) for melanoma, it is possible to stage the regional lymph node basin with low morbidity . This technique accurately selects patients with nodal micrometastases (e.g. non-palpable nodal metastases) and the sentinel node (SN) status has proven to be the most important prognostic factor [2, 9]. Over the years it has been incorporated in the management of melanomas in many institutions in order to opt for selective lymph node dissection (LND) in SN-positive patients only [10–12]. Despite these advantages, there is much debate about whether selective LND improves overall survival [13, 14]. Reports on increased incidence of in-transit metastasis have further aroused disagreement, since in-transit metastases are difficult to treat [15, 16]. In 2006, the results of the Multicentre Selective Lymphadenectomy Trial (MSLT-I) were published, concluding that the survival of SN-positive patients with intermediate-thickness melanoma is improved when immediate complete LND is performed .
In 1996, SLNB was introduced in the management of melanomas in our training hospitals. The aims of the present study were to evaluate our SLNB procedures concerning successful identification rate, complications, recurrence and survival and to compare our results with the data from the MSLT-I trial from expert and high-volume centers.
Patients and methods
Patients and data acquisition
Characteristics of 305 patients undergoing successful sentinel lymph node biopsy
Mean age in years (±SD)
Median Breslow thickness in mm (IQR)
SLNB and histopathological evaluation
Before surgery, 50–100 MBq 99mTc-nanocolloid albumin (Solco R/Nanocoll, Sorin Biomedica Diagnostics, Vercelli, Italy) was injected intradermally around the primary excision scar. Dynamic lymphatic mapping was performed to identify the lymphatics draining to one or more SN(s) and to mark their exact location on the skin. The marked location on the skin was verified by the surgeon using a gamma detection probe, which was also used during the operation.
Just before starting the SLNB procedure, 1 mL patent blue dye was injected intradermally around the primary excision scar. The SN was defined as a lymph node(s) reached by an afferent lymphatic vessel on dynamic lymphoscintigraphy or by a blue-colored afferent vessel on surgical exploration (i.e. lymph nodes on a direct drainage pathway from the primary tumor). Radioactivity was measured in the SN and compared to the background. All resected SNs were histopathologically examined using conventional hematoxylin-and-eosin staining plus immunohistochemical staining with S-100, Melan-A and HMB-45. Patients with tumor-positive SNs were offered completion LND. Lymph nodes from the LND specimen were analyzed only by hematoxylin-and-eosin staining.
Continuous variables with normal distribution were represented by means ± standard deviation (SD) and in case of skewed distribution by medians ± interquartile range. Testing differences between continuous variables was done by the Student’s t test or Wilcoxon rank test, when appropriate. Differences between proportions were compared with the chi-squared test or Fisher’s exact test.
Multivariate logistic regression analysis was used to define the risk of having a positive SN for the prognostic factors. Disease-free and disease-specific survival were analyzed by the Kaplan–Meier method; differences were tested with the log-rank test. Multivariate Cox regression analysis was used to determine the independent risk of covariates for recurrence or disease-specific death. For multivariate regression analyses, only factors which showed an association (p < 0.1) on univariate analysis were used in the final multivariate model.
SLNB and histopathology
A median of 2 SNs (interquartile range 1–2; range 0–8) could be identified by dynamic lymphoscintigraphy in 304 patients. In one patient no SN was visualized by lymphoscintigraphy but during surgery a blue node was harvested, revealing metastasis on histopathology. In 37 patients (12%) SNs were visualized in 2 or more basins. The median number of harvested SNs at operation was 2 (interquartile range 1–2; range 1–7). Most SNs (n = 340; 60%) were both blue and radioactive, 205 (36%) were only radioactive and 18 (3%) were only blue. In 27 patients (9%) additional non-SNs were excised, but all without metastases.
Results of multivariate logistic regression analysis: prognostic factors and corresponding odds ratio of harboring a positive sentinel node
Breslow thickness (per mm)
Clark level (per level)
Age (per year)
Nine SN-positive patients did not undergo completion LND because their health status did not allow major surgery (n = 2) or patients chose a wait-and-see policy (n = 7). Completion LND was performed in the remaining 45 patients, revealing additional metastases to non-SNs in 9 patients (20%).
After SLNB, complications occurred in 34 patients (11%): wound infection in 5 (1.6%), hematoma formation in 3 (1.0%; requiring operative intervention in one), seroma in 22 (7.2%; requiring surgical exploration and lymph vessel ligation in 4), edema in 4 (1.3%; all lower extremity) and neuralgia in 3 (1.0%).
After completion LND (n = 45), complications occurred in 35 patients (78%): wound infection in 17 (38%), bleeding in 2 (4.4%), seroma formation in 26 (58%) and edema in 5 (11%).
The median time of follow-up for recurrence was 50 months. Recurrence of disease occurred in 24 of the 251 SN-negative patients (9.6%). Six patients (2.4%) had local recurrence, 2 patients (0.8%) developed in-transit metastasis in the leg and 9 patients (3.6%) developed distant metastasis. During follow-up six patients (2.4%) developed regional lymph node metastasis despite a negative SN [false-negative rate of 6/(6 + 54) = 10%; SLNB sensitivity of 90%]. Three additional patients developed regional lymph node metastasis after having local, in-transit or distant metastasis (overall false-negative rate 14.3%).
Of the 54 SN-positive patients, 29 (54%) had relapse of the disease. Three patients (5.6%) had a local recurrence, 4 patients (7.4%) developed in-transit metastasis and 3 patients (5.6%) had regional lymph node metastasis (2 of whom did not undergo completion LND after positive SN; 1 patient developed lymph node metastasis at a second lymph node basin). The remaining 19 patients (35%) developed distant metastasis.
Results of multivariate Cox regression analysis: prognostic factors and corresponding hazard ratios for disease-free survival (DFS) and disease-specific survival (DSS)
Hazard ratio DFS (95% CI)
Hazard ratio DSS (95% CI)
Breslow thickness (per mm)
Age (per year)
Clark level (per level)
Non-radical excision primary
The median time of follow-up for disease-specific death was 52 months. The DSS curves of SN-positive and SN-negative patients are outlined in Fig. 1. The 5-year DSS rate of SN-negative patients was 97% compared to 62% in SN-positive patients (p < 0.001). There was no difference in 5-year DSS among SN-positive patients who did and did not undergo completion LND. Cox regression analysis revealed SN metastasis as the only remaining independent variable for survival (Table 3).
Subgroup with Breslow 1.2–3.5 mm, comparison with MSLT-I trial data
Comparison of data from the MSLT-I trial (sentinel lymph node biopsy patients only; n = 796) and the data on patients with intermediate Breslow thickness (1.2–3.5 mm) in the present study (n = 177)
MSLT-I (n = 796)
Present study (n = 177)
Breslow thickness (mm, mean ± SD)
1.98 ± 0.63
2.00 ± 0.65
Sentinel node positive
17.5% (26/26 + 122)
9.8% (4/4 + 37)
Overall 5-year DFS (%)
Overall 5-year DSS (%)
Since the introduction of lymphatic mapping and SLNB for melanoma by Morton in 1990, this technique has been incorporated in many institutions worldwide. Despite the Dutch national guidelines at that time, stating that the SLNB procedure was not standard care and should only be performed within trials, we and many other institutions in the Netherlands introduced this procedure in our training hospitals in 1996 for the care of melanoma patients. At that time short-term reports were conflicting regarding which approach was superior: SLNB or wait-and-see [6, 18, 19]. In addition, reports on increased in-transit metastasis after SLNB raised concerns about this technique [15, 16]. Our revised national guidelines in 2004 still recommended not to use the SLNB technique as a standard procedure in cutaneous melanoma . We therefore felt the need to evaluate the results of our SLNB procedures.
The results of the present study show that SLNB can successfully be performed in regional training hospitals in the Netherlands with a high SN identification rate of 98.4% and a complication rate of 11% of minor complications, which are both comparable to data from high-volume centers [21, 22]. We could not confirm the reports about increased frequency of in-transit metastases after SLNB, since in-transit metastasis only occurred in 6 patients (2.0%). The higher frequency of in-transit metastasis in patients with a positive SN (7.4%) merely reflects the worse prognosis in these patients. Others have already stressed this issue but definite proof was delivered by the MSLT-I study, which showed the frequency of local/in-transit metastasis to be 7.7% after SLNB and 8.4% after nodal observation [17, 23–25]. Thus, from a patient perspective, only complication rates should be weighed against the benefit of being informed about prognosis. As shown by others, our study confirmed the SN status to be the single most important prognostic factor for both recurrence and survival [2, 9, 26–28].
One of the shortcomings of our SLNB procedure was that the sensitivity of the technique was only 90%; 6 patients developed lymph node metastases despite a negative SN. These patients are incorrectly being informed of having a good prognosis. Two additional patients developed lymph node metastases after first having local and in-transit metastasis. We therefore excluded these patients from the calculation of our false-negative rate, since the finding of a positive node in such a setting might be the consequence of the local recurrence. This relatively high rate of false-negative SLNB may be explained by the inclusion of patients with melanomas >4 mm (who harbor an increased risk of recurrence) in one of our hospitals. Indeed, 2 out of 6 patients with false-negative SLNB had a melanoma of >4 mm thickness. However, when applying the same inclusion criteria for intermediate thickness as in the MSLT-I study, the false-negative rate was still 9.8%. In the MSLT-I study, a false-negative SLNB rate was reported in 26 out of 764 and therefore incomprehensibly called 3.4% . In our opinion, however, for a correct reflection of sensitivity, by definition all patients with true nodal metastasis should comprise the denominator, leading to a false-negative rate of 17.6% [26/(26 + 122)] in the MSLT-I, which is about double that of our results. We thus conclude that, although the false-negative rate is still relatively high, our performance of SLNB in melanoma patients is at least as good as experienced high-volume centers. Additionally, to illustrate whether a learning effect may have influenced our relatively high false-negative rate, we found that 3 out of 6 false-negative SNs were harvested during the last 3 years of our cohort. A learning effect therefore does not seem to have influenced the false-negative rate in this study.
Although the SN status has proven to be the most significant predictor for both recurrence and survival, the MSLT-I study failed to show an overall and melanoma-specific survival benefit of the SLNB procedure . Subgroup analysis, however, showed that SLNB significantly improved survival amongst those patients who had tumor-positive sentinel nodes undergoing ‘immediate’ LND compared to ‘delayed’ LND for patients with later detectable nodes. Although there was concern about the role of early detected small nodal micrometastases (i.e. sub-micrometastases) in the SLNB group and their influence on these significant conclusions, 10 years’ data show that the cumulative incidence of regional node metastases in the observation group and biopsy group of 20.5 ± 2.6 and 20.8 ± 1.7%, respectively, were equal. The MSLT-II trial, in which patients with SN metastasis are randomized to completion LND or intensive nodal observation, will further provide answers on this specific topic .
When comparing our data to the SLNB arm of the MSLT-I study (applying the same inclusion criteria), we saw some remarkable differences. The 5-year DFS of patients with a negative SN in our study was 92 versus 44% in SN-positive patients, compared to 83 and 53%, respectively, in the MSLT-I. The same difference was observed for the 5-year DSS which was 99% for SN-negative and 69% for SN-positive patients in our study, compared to 90 and 72%, respectively, in the MSLT-I. Furthermore, the overall DFS and DSS were slightly higher in our study (83 and 93% vs. 78 and 90%). Possible explanations for these somewhat better overall DFS and DSS rates might be the relatively low proportion of males (40 vs. 58%) and lower prevalence of ulceration in our patients (18 vs. 26%). Both factors are known to be associated with poorer outcome, which was also shown in our multivariate analysis (Table 2). The reason why SN-positive patients in our study showed a slightly poorer DFS and DSS might be due to the fact a completion LND was not performed in 17% of our SN-positive patients. We recognize that comparing survival data of a retrospective cohort to a prospective trial must be done with caution. However, the proportion of our patients who did not adhere to the appointed follow-up scheme (n = 33; 11%) had a mean follow-up time of 39 months (interquartile range 18–62) after all, demonstrating the completeness of our data.
Based on this comparison, we conclude that the results of our training hospitals are at least as good as the data from high-volume centers participating in the MSLT-I trial. When counseling a melanoma patient, however, a false-negative sentinel node rate of about 10% and an 11% complication rate should be weighed against being informed about the prognosis and having a possible therapeutic benefit. We agree with the statement that SLNB is “standard of care” for staging of cutaneous melanoma of intermediate thickness and for treatment planning. Early surgical intervention in the regional lymph nodes in patients with positive biopsy may result in improved survival and provides relevant staging information. Therefore, the Dutch guidelines should be adapted and should acknowledge the already widespread practice of SLNB procedure in melanoma patients.
Conflict of interest
None to declare.