Is there a relation between type of primary melanoma treatment and the development of intralymphatic metastasis ? A review of the literature q

Article history: Received 29 October 2015 Received in revised form 19 February 2016 Accepted 24 February 2016

The concept of incidence of locoregional metastases increasing with tumor thickness was recognized decades ago [12][13][14]. Previously, in transit metastases (ITM) and satellite lesions (SL) were considered different entities, but The American Joint Committee on Cancer (AJCC) has classified both ITM and SL in 2002 as intralymphatic metastases (ILM) [15]. Historically, SL have been defined to reside within centimeters of the primary tumor location and ITM in the pathway between primary site and regional lymph node basin. The leading hypothesis is that both originate from tumor cell emboli entrapped in dermal lymphatic vessels between primary tumor and regional lymph node basin [16,17]. ILM automatically upstages a patient's disease into stage IIIB/IIIC, decreasing 5-year survival to 59% and 40%, respectively [1]. Survival rates for patients with SL alone, SL/ITM, or ITM are identical and similar to that of patients with nodal disease [18]. Scar recurrence, 'true local recurrence', differs in pathophysiology, as these develop from residual cells of the initial melanoma, a result of false-negative margins or microsatellites.
Curative treatment for primary melanoma remains surgery (wide local excision, WLE) [2,19]. Four prospective additional elective lymph node dissection (ELND) trials showed no impact on survival [20][21][22][23][24]. ELND has become redundant after the introduction of the sentinel lymph node biopsy (SLNB) in 1992, which preserves its diagnostic advantage with less morbidity [21][22][23][25][26][27]. Patients with a positive SLNB undergo a completion node dissection (CLND). The MSLT-I study showed a small but significant disease-free and melanoma-specific survival benefit in patients with intermediate thickness melanoma (1.2-3.5 mm) and nodal disease following early treatment [28]. Most notably, a melanoma-specific survival improvement of 20% was reported for patients with intermediate thickness melanoma undergoing SLNB as opposed to observation, although the MSLT-I did not show improvement in recurrence free, distant metastasis free and melanoma specific survival for the entire population. The MSLT-II study will answer in the near future whether a CLND is indeed indicated after a positive SLNB [29,30]. Other treatment modalities have included therapeutic lymph node dissection (TLND), for metastatic nodal disease at the time of diagnosis, and delayed lymph node dissection (DLND), for patients developing metastatic nodal disease [31].
SLNB in addition to WLE alone has been suspected of causing ILM by inducing lymphatic stasis or entrapment of melanoma cells [32,33]. Pathophysiology on which this hypothesis is built is that the lymph flow from the skin reaches the nodal basin within minutes, with melanoma cells still in lymphatic channels en route to the lymph node basin at the time of SLNB or nodal dissection [33,34]. Estourgie et al. published a fourfold risk of ITM recurrence in SLNB positive patients as compared to SLNB negative patients, thereby raising the question whether surgical treatment of the regional lymph node basin can be responsible for ITM, although the same research group refuted this finding in a larger population [35,36]. Although various authors have studied this phenomenon, most notably Morton et al. in the aforementioned MSLT-I trial and van Poll et al. using data of the Melanoma Institute Australia, a definite answer as to whether the incidence of ILM should be attributed to unfavorable primary tumor characteristics alone or is increased by the SLNB procedure by means of a review of all available data has not yet been published [10,16,28,37,38].
The objective of this review was to provide an extensive body of evidence, answering the question whether ILM frequencies increase after performing SLNB.

Methods
Pubmed, Embase, Cochrane Library and Medline were searched for articles using the terms 'melanoma' and 'recurrence' or 'in transit metastasis' or 'ITM' or 'SL' or 'intralymphatic metastasis' or 'local recurrence' or 'satellite' or 'sentinel node' or 'survival' between January 1980 and September 2014. Articles were excluded if they had not been written in English, if they did not distinguish between a local recurrence and ILM, if incidence for ILM as a first recurrence (FR) was not reported, if studies exclusively reported on SLNBÀ or SLNB+ or if treatment strategy was unclear. Duplicates, case reports, letters to the editors and case series were excluded. Data regarding ILM as FR derived from our institution's SLNB database (UMCG database) were added to the review.
ITM was classified as recurrent melanoma in the pathway between primary melanoma location and the regional nodal basin, with the lesion more than two or five centimeters from this location, depending on the definition used in the article. All other cutaneous and subcutaneous metastases between the re-excision scar and the location of ITM were classified as SL. As consensus is now that ITM and SL are the same entity, all ITM and SL were combined into one value, 'ILM'.
For all included articles the number of patients with ILM as first recurrence (FR) were calculated per treatment group: for WLE alone, for WLE with ELND, WLE and DLND or TLND and WLE with SLNB. The last group was stratified into tumor-negative SLNB (SLNBÀ) patients and tumor-positive SLNB (SLNB+) patients undergoing CLND. When assessing risk of ILM as FR, WLE was compared to the WLE/SLNBÀ group. WLE/SLNB+ was compared to WLE/DLND, WLE/ELND and WLE/TLND groups. As only SLNB+ patients undergo additional CLND, this division groups together the most similar procedures regarding interruption of lymph flow. Additional study characteristics were collected: study design, number of patients, mean/median Breslow thickness, age at diagnosis, and melanoma ulceration status.

Statistical analysis
For a comprehensive review of the data, all data were summarized in tables and analyzed using version 18 SPSS, (IBM, Chicago, Illinois, USA). Descriptive statistics were used to calculate frequencies of ILM for the different treatment strategies. Chi-square tests were used to check for significant differences.
Subsequently, all studies were assigned a weight based on the amount of included patients and entered into a meta-analysis. Meta-analyses were performed stratified for treatment, SLNB results and anatomical localization of the primary tumor. Proportions of ILM and the corresponding 95% CI were calculated and entered in a datasheet. Meta-analyses were performed with the 'metan' module using STATA/SE version 12.0 (StataCorp, College Station, Texas, USA) with the original data as reported in the studies. Pooled ILM proportions and their 95% CI were calculated using a random effects model.

Study characteristics
19,620 studies were identified and assessed according to the inclusion criteria. 36 studies with a total of 33,622 patients were included for analysis (Table 1), including our ongoing academic medical center database (UMCG database). 6 studies were excluded because they exclusively reported on SLNBÀ or exclusively on SLNB+ patients (n = 684 patients) [11,[39][40][41][42][43]. Median follow-up ranged from >12 months-11 years. Fifteen out of 36 studies reported mean Breslow depth and 6 reported exclusively median Breslow depth. One study reported Breslow depth using incremental depths [44]. Melanoma ulceration status was reported in 23 studies; in 15 of those data were only available for part of the population. Twelve studies provided treatment/recurrence data on WLE (14,729 patients), 5 on WLE/ELND (1682 patients), 1 on WLE/ DLND (362 patients) and 18 on WLE/SLNB (11,201 patients). For the remaining 5648 patients in 7 studies, treatment was not specified. No study reported outcomes exclusively for TLND.
In 23 of the 36 included studies a clear definition of ITM/SL was not provided. ITM was defined as (sub)cutaneous disease recurrence between locoregional lymph node basin and 2, 3 or 5 centimeters from the original scar in n = 5, n = 1 and n = 4 studies, respectively. The remaining 3 studies defined ILM as recurrence within the pathway of lymphatic drainage, between scar and regional nodal basin, and between tumor and nodes, respectively. Seven out of 36 studies distinguished SL from ITM; out of these, 2 studies defined SL and LR as the same entity [13,16,35,[45][46][47][48].
Of the 11,201 patients undergoing SLNB, ILM was split out according to tumor status in 6913 patients. Of the SLNB+ group 153/1330 patients (11.5%) developed an ILM as FR versus 176/5783 patients (3.0%) in the SLNBÀ group. Differences in distribution between the four treatment modalities and differences between SNÀ and SN+ were statistically significant. ILM as FR after WLE was significantly lower than after WLE/SLNB, WLE/ELND and WLE/DLND (all p < 0.001). ILM was significantly lower after WLE/ SLNBÀ compared to WLE/SLNB+ (p < 0.001) ( Table 3).

Background
In this review, 33,622 melanoma patients from 36 studies were analyzed to establish whether performing SLNB on melanoma patients in addition to WLE alone leads to an increase in ILM. This is an ongoing field of discussion in the literature. In fact, Read et al. recently published one of the largest databases so far, (n = 11,614) where 505 patients developed ILM as a recurrence at any time during follow-up [51]. ILM percentages were 4.7% and 21.6% for SLNBÀ and SLNB+ patients, respectively. Numbers were not specified for the 190 patients who developed ILM as FR, which explains partly why the numbers are higher than in our study.
Critics of SLNB have argued that as of yet there is no agreement on adjuvant therapy for node-positive patients and that only 20% of the patients undergoing SLNB will have a positive node [52]. However, nowadays there are new approaches available with tar- 33 Ribero [ geted and/or immunotherapies that may lead to new adjuvant strategies [53,54]. The argument that no randomized controlled studies have shown a survival advantage for SLNB in nodepositive patients has become partly redundant upon publication of the MSLT-I, which shows a (small, but significant) survival advantage for a selective group of patients, i.e. patients with an intermediate thickness melanoma and positive SLNB. Proponents advocate that SLNB is a procedure with a relatively low morbidity and that the current false-negative rate for SLNB performed in reputable institutes is <6%, declining further as experience progresses [55,56].

Results
Based on the results of our meta-analysis, the overall incidence of ILM as FR was 3.4%. Patients who did not undergo any lymph node dissection had the lowest incidence, with 1.9% of patients having ILM recurrence after WLE and 3.4% after SLNBÀ. ILM occurrence after WLE/DLND and WLE/ELND was slightly higher (4.7% and 5.5%, respectively), but incidence spiked after SLNB+/CLND at 13.2%. For TLND, insufficient data were available.
Differences in ILM occurrence between WLE and WLE/SLNB groups were statistically significant, leading to the conclusion that   Figs. 1 and 2. b Total number of ILM in the treatment groups for the initial treatments and stratified for SNÀ and SN+, review data. P-value for differences in distribution (Chi2). a sentinel lymph node biopsy alone is associated with an increase in the risk of ILM (from 1.9% to 3.4%, p = 0.01).
To test the stasis hypothesis, the most comparable treatment modalities regarding lymph flow disruption are WLE vs. WLE/ SLNBÀ and WLE/SLNB+/CLND vs. WLE/ELND. As metastasis already has occurred in WLE/DLND groups, this is not a good comparator. As ILM incidence according to meta-analysis doubled between WLE vs. WLE/SLNBÀ and increased almost threefold from 4.7% to 13.2% between WLE/ELND and WLE/SLNB+/CLND groups, (p < 0.001), the increase of ILM is unlikely to be due to the increase in lymph stasis. CLND and ELND are comparable in their amount of lymph flow disruption. This suggests that an aggressive tumor behavior is the main reason for ILM, a statement that is supported by the spike in incidence after SLNB+, which is the patient group with the most aggressive tumor biology.

Limitations
Inevitable to any review, authors use different definitions and inclusion criteria. The level of heterogeneity is considerable, as illustrated in Table 1, where data on patient and tumor characteristics are shown. The inconsistent and varied application of terms  as ITM, SL and LR complicate comparisons among trials. Recently some authors have even abandoned the concept of a true local recurrence, merging ITM, SL and local recurrence into locoregional metastasis, leading to considerable data loss [57]. Also, data on mitosis index, Breslow thickness and ulceration status were inconsistent, thus complicating comparisons, necessitating interpreting the results with caution. In general, patients included in SLNB studies have less favorable primary tumor characteristics than patients who undergo WLE alone [58]. Moreover, before introduction of the SLNB technique, patients with less favorable tumor characteristics were to undergo ELND and would therefore not be included in WLE studies. These limitations may account for the difference between this review and the MSLT-I, a prospective study, in which no increase in ILM or local metastasis was reported between biopsy and observation groups (7.7 ± 1.0% and 8.4 ± 1.3%, respectively; p = 0.38). As we included WLE patients before introduction of SLNB our WLE population would differ from the MSLT-I population.
The percentage of ILM after DLND in our study is lower than expected. This may be due to the small sample size and also due to bias as we only included ILM as FR after DLND. Since these patients have aggressive disease, they may more often progress to distant metastasis instead of locoregional disease.

Summary
This review showed an increase in ILM of 1.5% after only performing a SLNB procedure (ILM 1.9% for WLE vs. 3.4% for SLNBÀ).
Taking into account the patient groups traditionally included in WLE studies it is difficult to say whether this increase represents an actual increase in ILM recurrence or a selection bias.
The SLNB procedure is the most important prognostic tool in clinical practice, providing a survival benefit in selected SLNB+ patients undergoing CLND and potentially serving as a marker to identify patients for adjuvant therapy. Sentinel lymph node biopsy has been suspected of causing to increase intralymphatic metastasis by restricting lymph flow. This review demonstrates this increase, but this result has to be interpreted with caution due to possible selection bias. As the stasis hypothesis seems to be incorrect based on the data in this study, aggressive tumor characteristics are likely the cause of this increase. We therefore advocate performing SLNB procedures, but to proceed with caution, adhere to the guidelines and not extend the indication area.

Conflict of interest statement
The authors declare no conflict of interest.