The anatomic landmarks marking the external limits of the dissection are as follows: the sartorius muscle is the lateral limit. The medial border is the adductor magnus muscle. Be aware: this is not the same as the adductor longus muscle, which is further lateral of the adductor magnus (more medial to the adductor magnus is the gracilis muscle). Cranially, the border is formed by the inguinal (Poupart’s) ligament. The caudal limit is where the adductor magnus muscle and sartorius muscle meet. Dorsally, the femoral artery and vein are the borders of the dissection. All of the nodes inferior to the inguinal (Poupart’s) ligament should be removed. While several different incisions may be used, this author considers two types of incision: a curvilinear incision starting laterally over the inguinal ligament and curving medially and inferiorly, ending over the midpoint of the adductor magnus muscle. Or others perform the dissection through a longitudinal incision in the direction of the limb up to the inguinal ligament or even above the ligament, in case of a CGD (to allow deep dissection through one single incision). The lymphatic tissue within the groin fat should be dissected carefully off of the femoral vessels and nerves all the way up to the inguinal canal and for 3 cm superior to the inguinal ligament. This last does not mean that the iliac nodes should be removed, but the subcutaneous nodes ventral from the inguinal (Poupart’s) ligament, but cranially from the external entrance to the femoral canal. Some attempt to spare the saphenous vein to potentially reduce morbidity (lymph edema); however, others resect it to ensure complete excision of the lymph nodes. A sartorius transposition, as commonly used in other operations, can be performed to cover the femoral artery and vein, but this is not mandatory. Some will do this routinely, others will retain this option for redo surgery and/or in case the risk of a postoperative radiotherapy indication is considered high.

The anatomic landmarks marking the borders of the DGD are as follows: the caudal limit is the inguinal (Poupart’s) ligament. The cranial border is the bifurcation of the common iliac artery to the external and internal iliac arteries. Laterally, the iliopsoas muscle is the border; medially, the bladder is the border. Dorsally, the obturator artery and nerve form the border. Again two types of incision can be used: a longitudinal incision in case of a CGD in the direction of the limb, and a separate transverse incision, approximately 3 cm superior to the inguinal (Poupart’s) ligament. This incision is taken down through the external oblique, internal oblique, and transversus muscles, and the surgeon at that point stays extraperitoneally as in the approach to the iliac vessels for renal transplantation. In selected cases, surgeons might even consider an approach through a laparotomy if the disease is bulky and/or if there is extent of disease toward the aortic bifurcation.

The American Joint Committee on Cancer (AJCC) 7th edition considers node-positive disease as stage IIIA–C. This depends on the absence (pT1a–4a) or presence (pT1b–4b) of ulceration of the primary tumor, if the metastasis is microscopic (SN, N1a–N2a) or macroscopic (N1b–N2b), the number of involved nodes (≥4 metastases; N3), and/or the presence or absence of satellite or in-transit metastases [1]. Five-year survival rates differ from 78% for stage IIIA to 40% for stage IIIC [1].

Elective lymph node dissection or prophylactic lymph node dissections have now been abandoned. Originally, Herbert Snow described this approach in 1892, and it was performed routinely by some during the 1950s and 1960s. Four randomized controlled trials have prospectively analyzed the value of ELND in melanoma. Veronesi et al. only examined limb melanomas, because the draining nodal basin at risk is clear in those cases, which can sometimes be difficult to anticipate for trunk melanomas. They found no difference in outcome in 267 ELND patients versus 286 nodal observation patients [2]. Sim et al. found no differences between an immediate ELND (n = 54), a delayed ELND (3 months) (n = 56), and nodal observation (n = 63) [3]. Balch et al. performed the largest RCT on this topic of 383 ELND versus 356 nodal observation patients [4]. However, there were indications that some subgroups might benefit from the elective removal of lymph nodes, such as younger patients (<60 years) (P = 0.042), non-ulcerated melanomas (P = 0.018), and patients with intermediate thickness melanoma (Breslow 1–2 mm) (P = 0.031) [4]. Finally, a WHO study by Cascinelli et al. only analyzed trunk melanomas in 122 ELND versus 130 nodal observation patients and did not find a significant difference with survival rates of 61.7% for ELND versus 51.3% for nodal observation (P = 0.09) [5]. Again subgroup analyses seemed to indicate a benefit; in this case when only looking at those with nodal involvement, the survival was 48.2% in the ELND group versus 26.6% in the nodal observation group (P = 0.04) [5].

Since the potential benefit from an early or prophylactic lymph node dissection can only occur in patients with (microscopic) nodal involvement, it was suggested that ELND studies could not show this benefit, because it was diluted by the majority of patients with negative nodes. Moreover, only intermediate thickness melanomas were thought to potentially benefit. Therefore, Morton and coworkers developed the concept of lymphatic mapping to detect the sentinel node (SN) and potentially target node-positive patients, who might benefit from an early lymph node dissection [6].

The prognostic value of the SN in the staging of stage I/II melanoma patients to detect occult microscopic disease has been broadly demonstrated, and therefore the procedure is widely accepted as a staging tool [7–9].

The therapeutic value of the SN procedure for intermediate thickness melanoma has been analyzed in one single large prospective randomized controlled trial, the Multicenter Selective Lymphadenectomy Trial-1 (MSLT-1). Despite reporting the 10-year results, there is still an ongoing debate on the potential therapeutic value for a subgroup of patients [10–14].

This trial randomized 60:40 to wide local excision (WLE) + SN versus WLE + nodal observation. In the SN group, the patients with a positive SN underwent an early completion lymph node dissection (CLND). In the nodal observation group, patients underwent a lymph node dissection in case of clinically detected recurrence. The primary endpoint was melanoma-specific survival (MSS) and was not statistically different between the SN group (81.4 ± 1.5%) versus nodal observation group (78.3 ± 2.0%) (HR 0.84 (95% CI 0.64–1.09)) (P = 0.18) [11].

Despite this clear result regarding the main study endpoint, a number of alternative arguments toward a potential benefit have been posed, the first and foremost being the fact that the subgroup analysis of SN-positive patients versus node-positive patients in the nodal observation group shows a clear benefit of 62.1% (± 4.8%) versus 41.5% (± 5.6%) (HR 0.56 (95% CI 0.37–0.84)) (P = 0.006) [11]. This is supported by a new statistical analysis: the accelerated-failure-time latent-subgroup analysis [11, 15, 16]. Another is with respect to an improvement in terms of disease-free survival (DFS) [11].

Opposition to this interpretation for a benefit has proposed many arguments, the foremost of which being that a subgroup analysis is invalid, as, in general, subgroup analyses are post hoc, underpowered, and not preplanned. Interestingly, in the design of the MSLT-1 trial, Morton and colleagues precalculated and powered the study for an a priori subgroup analysis, which largely refutes this argument. Furthermore, as discussed subsequently, a formal statistical model has also validated this subgroup analysis. However, despite this prior planning, there remain criticism and some concern about the validity of its conclusions. This analysis does not take into account patients who were false negative and have a worse survival rate than the SN-positive or even the node-positive patients from the observation arm [14]. Excluding them from the subgroup analysis increases the survival of SN-positive patients, because not all node-positive patients are included in the subgroup analysis; especially those with a poor outcome are excluded [14]. There are also false-positive patients, which artificially improves survival in this group. There are a number of hypothetical reasons for a false-positive result, e.g., benign nevus cells, which have been incorrectly concluded to be melanoma cells [17]. By including these (actually non-metastatic) patients in the SN-positive population, the outcome of the SN-positive group may be falsely improved.

In considering the subgroup analysis, a formal statistical review using a novel technique has been applied which has not been widely utilized. In considering “accelerated-failure-time latent-subgroup analysis,” it can be stated that this is a new statistical hypothesis, which has yet to be validated. It is not yet a widely accepted statistical tool. It was developed on the basis of the interim results of the MSLT-1 in 2006 [10, 15, 16] by the statistician involved in the MSLT-1, among others, and should first be validated on external studies rather than be used as proof for a survival benefit of the MSLT-1 [14].

Finally, the DFS benefit is controversial, since the group undergoing CLND obviously will recur less frequently in the regional nodes (as they are already removed) compared to the nodal observation group [18].

Thus, there is no unequivocal survival benefit for undergoing an SN in melanoma and there is ongoing debate between believers and nonbelievers and pro and con. Both parties do concur on the prognostic information gained by the SN.

Recent developments in effective systemic therapy for stage IV melanoma with targeted therapies (BRAF/MEK inhibitors) and/or immunotherapies (anti-CLTA-4/anti-PD-1) might influence the reason to perform an SN in the future [19–33]. SN-positive patients (treated by CLND) are now able to participate in adjuvant therapy trials with these new agents, which have already proven their efficacy in stage IV disease. The first results of adjuvant ipilimumab showed a significant improvement in relapse-free survival at 3 years (46.5% vs. 34.8%) [34]. Very recently, it was reported that this also turned into an MSS benefit at 5 years (65.4% vs. 54.4%, HR 0.72 (CI 0.58–0.88), P = 0.001) [35]. Other pivotal studies with adjuvant anti-PD-1 (nivolumab, pembrolizumab), high- versus low-dose ipilimumab, combination ipilimumab + nivolumab, and BRAF/MEK inhibitors will report first outcomes between 2017 and 2020. It is expected that this will mandate the routine use of SN staging. The completion lymph node dissection for SN+ disease might change in light of effective adjuvant therapy.

A second question regarding the potential therapeutic effect of the SN procedure in melanoma is with respect to the CLND. Perhaps the SN alone is therapeutic or perhaps the CLND is the therapeutic part of the procedure. This has been the subject of the MSLT-2, which randomized SN-positive patients to CLND or nodal observation with periodic ultrasound of the (positive SN) lymph node basin [36]. The results of this MSLT-2 study are pending and expected no sooner than 2020.

In the meantime, the German DECOG-SLT study has presented its initial results. This study screened 5547 patients with melanoma of whom 1269 had a micrometastasis in the SN (23%). Four hundred and eighty-three patients agreed to be randomized between CLND and nodal observation. After a median follow-up of 35 months, the distant metastasis-free survival was 77.0% in the observation group versus 74.9% in the CLND group (HR 1.03 (90% CI 0.71–1.50)) (P = 0.87) [37]. Although this study has less power compared to the MSLT-2, the follow-up is not yet mature, and there were a large proportion of patients with lesser SN tumor burden; this strongly suggests that the routine use of CLND might not be of benefit for SN-positive patients.

Finally, a large number of studies have examined SN tumor burden as a potential tool to determine which patients might and might not benefit from CLND [8, 38–73]. All these (retrospective) studies have demonstrated the heterogeneous prognosis of SN-positive melanoma patients. Moreover, despite differences in ways to measure the SN tumor burden, all these different factors with different cutoffs have demonstrated their prognostic value with respect to predicting the chance of additional non-SN involvement in the CLND and/or survival. However, there are differences in the interobserver reproducibility of these respective SN tumor burden factors. Murali et al. showed that the maximum diameter and the tumor penetrative depth were the most reproducible [74]. Currently, the EORTC 1208 (Minitub) study is examining if patients with minimal SN tumor burden can safely be managed without CLND.

Unlike in occult microscopic disease, most (if not all) surgeons will agree that a therapeutic lymph node dissection (TLND) is indicated in case of palpable (macroscopic) disease. Although these patients are at high risk for future disease relapse and also distant (visceral) metastasis, which can potentially become fatal, there is still a reasonable chance of cure of 40–59% at 5 years and 20–40% at 10 years [1].

There is no consensus between surgeons worldwide on the required extent of surgery for the groin. Some propose to always perform a combined superficial and deep groin dissection (CGD) in all cases. Others always perform a superficial groin dissection (SGD) only.

Hughes et al. reported a summary of a number of studies on elective and/or therapeutic groin lymph node dissections for melanoma [75]. There was a large spread of pelvic nodal involvement at TLND histology of 17–45% [75]. Five-year estimated survival ranged from 0 to 40% [75].

Badgwell et al. described 235 patients undergoing SGD and 97 undergoing CGD. Five-year overall survival was 42% for patients with positive deep nodes compared to 51% for those with negative deep nodes (P = 0.11) [76]. On multivariate analysis, positive deep nodes, male gender, and extracapsular extension were independent prognostic factors influencing survival [76]. The authors concluded that patients with involved pelvic nodes should be considered stage III and not stage IV and should be surgically treated with intended curation.

van der Ploeg et al. described 121 CGD and 48 SGD [77]. Five-year overall survival was 39.7% for patients without pelvic nodal involvement on CGD and 12.5% with pelvic nodal involvement of CGD [77]. No survival differences were seen between SGD and CGD, especially not after correcting for other prognostic factors [77].

Allan et al. demonstrated that preoperative CT scans have 60% sensitivity and an 86.2% negative predictive value, which means that a negative preoperative CT scan will be false negative in 40% of cases with respect to pelvic nodal involvement [78]. Patients with involved pelvic nodes had a higher risk of disease relapse compared to negative pelvic nodes, although this was not significant in this small series of 72 patients with 22 with pelvic nodal involvement [78].

Oude Ophuis et al. showed that 35% of CGD had involvement of the pelvic nodes [79]. Preoperative imaging, by CT scan, PET scan, or PET/CT scan, has limited sensitivity and by itself could not safely exclude pelvic nodal involvement [79]. An algorithm with negative imaging, fewer inguinal involved nodes, no extracapsular extension, and a low lymph node ratio (LNR) has low risk of pelvic nodal involvement and might safely be spared a DGD [79].

Van Wissen et al. reported on 70 stage IIIB/C melanoma patients with an indication for a groin dissection. All patients underwent preoperative PET/CT scans. The sensitivity for deep groin (iliac) involvement was 67%, specificity 91%, positive predictive value 73%, and negative predictive value of 81%. Thus the false-negative rate was 33%. Therefore the authors concluded that PET/CT alone was insufficient to safely limit the extent of surgery to the superficial groin only.

Currently, the Australia and New Zealand Trials Group is performing a prospective randomized study: Inguinal or Ilioinguinal Lymphadenectomy for Patients with Metastatic Melanoma to Groin Lymph Nodes and No Evidence of Pelvic Disease on PET/CT Scan – A Randomized Phase III Trial (EAGLE FM) (NCT02166788).

Complications can be divided into two categories: short term and long term . Short-term complications include surgical site/wound infections, seroma, skin necrosis, and fistulas. Long-term complications are mostly chronic lymph edema and rarely nerve damage.

A recent meta-analysis by Soderman et al. summarized 20 studies (including two randomized trials, two prospective cohort, and 16 retrospective series) [80]. In total the complication rate was 52% (44–60%). Infection was seen in 21% (15–27%), wound breakdown in 14% (8–21%), necrosis in 10% (6–15%), seroma in 23% (18–29%), and lymph edema in 33% (25–42%) [80].

The spread of the complication rates is most likely to be caused by the nature of the respective studies; as very few were prospective and most were retrospective, the reliability of registering all respective items is dubious. It is likely that there has been an underreporting of complications in the retrospective studies. Moreover, the methods of assessment of lymph edema differed across studies as patient reported, physician reported, and requiring treatment or measured. In general all surgeons agree that complication rates are high, and this is most frequently caused by the superficial groin dissection and not the deep groin dissection. Although the deep groin dissection part of a combined groin dissection does increase the chance of lymph edema slightly, it is usually not associated with the same problems in terms of surgical site/wound infections, seroma, skin necrosis, and fistulas as with a superficial groin dissection. A deep groin dissection (without superficial groin dissection) for an isolated iliac recurrence is usually associated with far fewer complications than a superficial or combined groin dissection.

Stuiver et al. analyzed potential covariates that might influence the chance of developing complications after groin dissections [81]. All these factors, body mass index (BMI), diabetes, other comorbidity, type of incision, use of sartorius muscle transposition, sparing of saphenous vein, palpable versus SN disease, skin excision, and bed rest did not influence the chance of complications [81]. The only significant factor was older age [81]. Adjuvant radiotherapy has been demonstrated to significantly increase the chance of chronic lymph edema [82, 83].

Attempts have been made to reduce short- and long-term complications of groin dissections. Bartlett et al. have analyzed the use of a sartorius transposition in 381 patients, but did not find any difference in surgical site/wound infections (10% vs. 14%, P = 0.39) [84].

Others have tried the use of fibrin sealant. Weldrick et al. have performed a systemic review of six prospective randomized controlled trials and did not find any difference in surgical site/wound infections (32% vs. 34%, P = 0.90) [85].

Finally, Faut et al. have analyzed different mobilization protocols (1 day of bed rest vs. 5 days vs. 10 days. vs. ≥10 days of bed rest with a Bohler-Braun splint) and did not find any differences [86].

In recent years, minimally invasive approaches to groin dissection have been developed. Sommariva et al. have analyzed ten case series of 168 patients, who have been treated by video-assisted inguinal lymphadenectomy (VEIL) for melanoma or other solid tumors [87]. They reported conversion rates between 0 and 7.7% [87]. Wound-related complications occurred in 0–13.3%; seroma was seen in 4–38.4% of cases [87]. The duration of use of the low-vacuum drain might influence these rates, since longer drain duration will reduce seroma rates but increase wound infection rates. Although these rates might be promising, the quality and cost of the procedure need to balance these reductions in complication rates. In general the mean operation time was 245 min, which is considerably longer than the classical open procedure. At the same time, the median number of harvested lymph nodes was seven, and local recurrence rates were 6.6% [87]. Therefore, this technique should be evaluated in a prospective trial.

A first report has been published on the use of robot-assisted video endoscopic inguinal lymphadenectomy, but this was concerning one single case [88]. This too should be evaluated by a prospective trial.