Management of patients with cutaneous melanoma with metastatic sentinel lymph node involvement has changed significantly since the introduction of sentinel lymph node biopsy (SLNB) in the early 1990s. Traditionally, complete lymph node dissection (CLND) was the standard of care for patients with metastatic sentinel node involvement. However, the results of 2 pivotal clinical trials1,2 challenged this practice, showing that CLND did not provide a melanoma-specific survival benefit over active surveillance in patients with positive SLNB.

These results led to changes in melanoma management guidelines. However, most guidelines, such as the NCCN,3 as well as some authors, continue to recommend considering CLND in cases with certain high-risk features of recurrence or non-sentinel nodal metastases, or when adequate ultrasound follow-up is not feasible.4–7 These high-risk features coincide with exclusion criteria or are underrepresented in the aforementioned clinical trials.8 As a result, there is variability in clinical practice, which has been documented in several studies.9–14

The aim of the present study was to evaluate the impact of CLND on melanoma-specific survival (MSS) and overall survival (OS) in a subgroup of patients with melanoma and positive SLNB considered to be at high risk. Secondary objectives were to analyze clinical and pathological factors that may be associated with survival in these high-risk patients.

Of a total of 430 patients with cutaneous melanoma and positive SLNB included in the study, 146 underwent CLND, while 282 were assigned to the observation group. After propensity score matching, the high-risk groups were balanced according to the defined parameters, resulting in 95 patients in each group (Fig. 1).

Fig. 1.

Flow diagram of patients with cutaneous melanoma and positive sentinel lymph node biopsy between 2017 and 2022.

Table 1 shows the study variables. The results showed that the groups were fairly homogeneous with respect to patient characteristics, tumor characteristics, SLNB characteristics, and lymph node metastasis characteristics. However, the presence of microsatellitosis was more common in the observation group, although this difference did not reach statistical significance (15.6% vs 6.6%; P=.06). Extracapsular extension was also more common in the observation group (17.6% vs 6.4%; P=.03).

Median follow-up was 36.3 months. In the survival curve analysis, CLND was not a significant predictor of OS, NRFS, MSS, or LS (Fig. 2). However, a trend toward improved NRFS was observed in the CLND group, although this trend did not reach statistical significance (P=.2).

Fig. 2.

Survival curve analysis comparing overall survival (OS), nodal recurrence-free survival (NRFS), melanoma-specific survival (MSS), and local recurrence-free survival (LS) between high-risk patients who underwent complete lymph node dissection (CLND) and those managed with observation. The observation arm is shown as a blue line, and the CLND arm is shown as a red line.

Multivariable analysis of factors associated with survival in selected patients showed that CLND had no significant association with OS, MSS, NRFS, or LS. The presence of ulceration was significantly associated with worse MSS (HR, 4.18; 95%CI, 1.66–10.54; P=.002), as were ECE (HR, 3.1; 95%CI, 1.12–8.61; P=.03) and ECOG >0 (HR, 3.46; 95%CI, 1.46–8.24; P=.005). For NRFS, an association with LVI was observed (HR, 2.5; 95%CI, 1.11–5.63; P=.02). For OS, the data were similar to the MSS analysis, with the additional association of a high mitotic rate (≥6mitoses/mm2) with OS (HR, 2.54; 95%CI, 1.1–5.87; P=.029) (Table 2).

This multicenter observational study showed that, in a subgroup of patients with melanoma metastasis in the sentinel lymph node and high-risk features, after a median follow-up of 36.3 months, CLND did not offer survival advantages compared with observation. These findings are consistent with those obtained in a previous study by our group that included the global cohort of patients with positive sentinel lymph node biopsy and support the results of the DeCOG and MSLT-II trials.21 In the aforementioned trials, after a median follow-up of 35 and 43 months, respectively, no survival benefit was observed. Therefore, the results of the present study suggest that in patients with high-risk criteria, CLND may also be omitted without compromising oncologic outcomes and while avoiding associated comorbidity.22

The results of the present study are important because these trials had certain limitations. In both studies, patients with ECE and microsatellite disease were excluded. Patients with more than 3 positive sentinel nodes accounted for <1% in MSLT-II, and only 8% had 2 or more positive nodes in DeCOG.23 In addition, two-thirds of patients in these trials had micrometastases <1mm.24 Excluding these factors introduces a selection bias against patients at higher risk of non-sentinel lymph node metastasis, which has also been associated with worse MSS.7,25–27

Having ruled out that CLND provides a survival benefit in this high-risk patient population, it is reasonable to ask whether it provides any benefit in terms of regional disease control or prognostic and staging information.

Notably, CLND did not significantly improve nodal recurrence-free survival in this study, although there was a trend in that direction. MSLT-II showed a 69% reduction in nodal recurrence in the CLND group. However, according to some authors, close ultrasound follow-up of the observation group allows early detection of recurrence and surgical salvage in most cases.8,22,23

We found that in this high-risk subgroup, LVI in the primary tumor was associated with worse NRFS. The presence of this factor may predict nodal recurrence and indicate the need for adjuvant treatment and/or closer nodal surveillance. Mitra et al. were the first to demonstrate that LVI was associated with a significantly higher rate of early nodal recurrence in a group of SLNB-positive patients undergoing clinical and ultrasound follow-up similar to the MSLT-II protocol.28 These findings are consistent with those of Keller et al.29 using MSLT-II follow-up data.

In terms of staging, replacing CLND with surveillance precludes classification of patients with nodal involvement according to the American Joint Committee on Cancer, 8th edition (AJCC 8th), because this system includes the total number of positive lymph nodes obtained by CLND to assign a pathologic N stage.30–32 To fill this gap, models have been proposed to stratify stage III patients into subgroups with different prognoses using characteristics of the primary tumor and sentinel lymph node metastasis, thus facilitating selection for systemic adjuvant treatment.33,34

Our results are consistent with these studies, as we found that ulceration and ECE were significantly associated with MSS and OS. Tumor ulceration is a prognostic factor associated with both an increased risk of positive non-sentinel nodes and worse survival, even in early AJCC stages. ECE indicates a biologic predisposition to nodal invasion28,35,36 and, in the context of microscopic nodal disease, may also be associated with an increased risk of non-sentinel node involvement and worse survival.37–39 Maurichi et al.39 found that ECE was a significant prognostic factor in melanoma patients with positive SLNB, with 5-year overall survival rates of 62.5% and 71.7% for SLNB-positive patients with and without ECE, respectively. The 5-year disease-free survival rates were 54% and 64% for SLNB-positive patients with and without ECE, respectively.

Importantly, we found that baseline ECOG 1 or higher was significantly associated with both OS and MSS. The association between ECOG performance status and survival has been described in patients with advanced metastatic melanoma and other solid tumors, highlighting its role in response to both BRAF inhibitors and immunotherapy.40–45 For example, several studies showed that worse response to immunotherapy was associated with worse baseline ECOG in patients with advanced or metastatic melanoma.44,45 However, we have not found similar studies linking ECOG with prognosis in patients with melanoma and positive SLNB or with poorer response to adjuvant immunotherapy. This may be due to limited experience, as these patients were excluded from adjuvant clinical trials.46–48

Also noteworthy is the association between a high mitotic rate of >6mitoses/mm2 and OS. The impact of mitotic rate on treatment and survival of melanoma patients has been previously documented49 and even as a predictor of sentinel node metastasis.15 However, it is not included in the AJCC 8th classification, possibly because such prognostic information is already captured by other variables. For example, the study by Roach et al.50 using data from the Sunbelt Trial and dichotomizing this variable showed an association between mitotic rate and survival in univariable analysis but not in multivariable analysis. Similar results have been obtained using US Surveillance, Epidemiology, and End Results data,49,51 which appear to be more useful in stage I disease than in more advanced stages.

Limitations of our study include its retrospective nature and the limited sample size of high-risk patients.

Strengths include the fact that the participating centers are referral centers with standardized active follow-up that captures most recurrences in this subgroup of patients.

In conclusion, in our study, we did not find that CLND was associated with a survival benefit in patients with positive SLNB considered to be at high risk. These findings are consistent with the results of previous clinical trials in which this subgroup was not well represented. It seems unreasonable to continue offering CLND in this subgroup of patients.

This research did not receive specific funding from public sector agencies, the commercial sector, or non-profit organizations for the participating centers. However, the research conducted by the Melanoma group at Hospital Clínic de Barcelona was supported by the CIBER of Rare Diseases of the Instituto de Salud Carlos III, Spain; AGAUR 2017_SGR_1134 and the CERCA Program of the Generalitat de Catalunya, Spain; a research grant from the “Fundación Científica de la Asociación Española Contra el Cáncer” (GCB15152978SOEN), Spain; the European Commission under the Sixth Framework Programme, Contract No. LSHC-CT-2006-018702 (GenoMEL); the European Commission under the Seventh Framework Programme, Diagnoptics; the European Commission under the HORIZON2020 Framework Programme, iTobos and Qualitop; and the European Commission under the Horizon Europe Programme, HORIZON-MISS-2021-CANCER-02, MELCAYA (reference 101096667). This research was also partially supported by grants from the Fondo de Investigaciones Sanitarias (FIS) P.I. 18/00419 and 22/01467, Spain. Part of the work was carried out at the Esther Koplowitz Center, Barcelona.

None declared.