Is a Wider Margin (2 cm vs. 1 cm) for a 1.01–2.0 mm Melanoma Necessary?

Abstract

Background

The current NCCN recommendation for resection margins in patients with melanomas between 1.01–2 mm deep is a 1–2 cm radial margin. We sought to determine if margin width had an impact on local recurrence (LR), disease-specific survival (DSS), and type of wound closure.

Methods

1.01–2.0 mm melanomas were evaluated at a single institution between 2008 and 2013. All patients had a 1 or 2 cm margin.

Results

We identified 965 patients that had a 1 cm (n=302, 31.3%) or 2 cm margin (n=663, 68.7%). Median age was 64 and 592 (61.3%) were male. 32.5% and 48.7% of head and neck and extremity patients had a 1 cm margin vs. 18.9% of trunk pts (p<0.001). LR was 0.6% and 1.5% for a 1 and 2 cm margin, respectively (p=NS). Five-year DSS was 87% for a 1 cm margin and 85% for a 2 cm margin (p=NS). Breslow thickness, melanoma on the head and neck, lymphovascular invasion (LVI) and sentinel lymph node biopsy (SLNB) status significantly predicted LR on univariate analysis, however only location and SLNB status were associated with LR on multivariate analysis. Margin width was not significant for LR or DSS. Wider margins were associated with more frequent graft or flap use only on the head and neck (p=0.025).

Conclusions

Our data show selectively using a narrow margin of 1 cm did not increase the risk of local recurrence or decrease DSS. Avoiding a 2 cm margin may decrease the need for graft/flap use on the head and neck.

Introduction

The incidence of melanoma in the United States continues to rise with an estimated annual percentage increase ranging from 1.5–4.1% in 2015. [1]. Newer modalities of treatment, including immunotherapy and targeted therapies, have been introduced, while the surgical treatment of melanoma has undergone minimal change. The gold standard of treatment for localized melanoma is wide excision (WE) of the lesion with a 1 or 2 cm radial margin based on the thickness of melanoma. There have been several large prospective trials which have studied resection margins in melanomas ranging from 0.8 to 4 mm in thickness as summarized in Table 1 [2–7]. These studies were unable to demonstrate a significant difference in overall survival (OS), disease-specific survival (DSS) or local recurrence when a wide margin of 4 or 5 cm was compared to narrow margin of 2 cm [8]. One of the earliest studies to report on margins of resection and recurrence was done by Veronesi et al. The authors showed a higher rate of recurrence when a 1 cm margin was used compared to a 3 cm margin for melanomas < 2 mm in thickness, although this was not statistically significant. In addition, they reported no difference in OS [9]. The data produced from the aforementioned studies helped form the foundation for the National Comprehensive Cancer Network (NCCN) recommendations for a 1.01–2 mm melanoma [10].

TABLE 1.

Summary of clinical studies assessing breslow thickness and resection margin width

Study	Year	Number of patients	Breslow Thickness (mm)	Margin Width (cm)	Median Follow-up (months)	Local Recurrence Rate	5-year OS
Prospective studies
Cascinelli	1998	612	<2	1 vs. 3	90	No difference	No differencea
Cohn-Cedermark et al.	2000	989	0.8–2	2 vs. 5	132	No difference	No difference
Balch et al.	2001	740	1–4	2 vs. 4	120	No difference	No difference
Khayat et al.	2003	337	<2.1	2 vs. 5	192	No difference	No difference
Thomas et al.	2004	774	≥3	1 vs. 3	60	No differenceb	No difference
Gillgren et al.	2011	936	>2	2 vs. 4	80.4	No difference	No difference
Retrospective studies
Hudson et al.	2012	576	1–2	1 vs. 2	38	Increased with 1 cm marginsc	No difference
Doepker et al.	2015	1024	1–2	1 vs. 2	14.8	No difference	No difference

OS overall survival

^aNo statistical difference was seen in OS up to 12 years

^bA statistical difference was seen in locoregional recurrence between the 1 and 3 cm margin groups

^cLocal recurrence was 3.6% and 0.9% in the 1 and 2 cm margin groups, respectively (p=0.04). This difference was seen on univariate analysis

The NCCN guidelines recommend either a 1 or 2 cm margin width based on evidence extrapolated from the trials summarized in Table 1. Currently, there is no prospective trial that has directly compared a 1 or 2 cm margin width for a 1.01 to 2 mm melanoma. Typically a surgeon will use their best discretion in choosing margin widths when treating melanomas between 1.01–2 mm in thickness. Best discretion refers to the surgeon using a narrow margin to preserve functionality and cosmesis or avoid a potential need for skin graft or flap reconstruction. A more recent single-center, retrospective series directly compared outcomes for a 1 to 2 mm melanoma using a 1 or 2 cm margin width [11]. Hudson et al. retrospectively reviewed 576 patients with 224 (38.9%) having a 1 cm margin and 352 (61.1%) having a 2 cm margin width. The distribution of local recurrence was significant between the two groups, with the 1cm group having a recurrence rate of 3.6% compared to only 0.9% in the 2 cm group (p=0.044). Interestingly, only head and neck location was associated with local recurrence on multivariate analysis (MVA). The authors demonstrated no difference in OS between the two margin groups on MVA [11].

We sought to add to the literature by directly comparing the use of a 1 or 2 cm margin width of resection for a 1.01–2 mm melanoma in a large retrospective series. The goal of the study is to determine if using a narrow margin had an impact on the need for a skin graft or flap reconstruction, local recurrence or DSS in patients diagnosed with a melanomas between 1.01–2 mm in thickness.

Methods

After obtaining Institutional Review Board approval, a retrospective series of consecutive patients diagnosed with melanoma 1.01–2 mm was identified from a single-institution database of patients from 2002–2013. All patients had a WE with a 1 or 2 cm radial margin from the clinically visible edge of the lesion or biopsy scar. Demographic and clinicopathologic characteristics (sex, age at diagnosis, histologic subtype, location of primary tumor, Breslow thickness, and sentinel lymph node biopsy (SLNB) status) along with outcomes data were retrieved. Ulceration, lymphovascular invasion (LVI), and mitotic rate (MR) were also evaluated, although these features were not uniformly recorded for all patients on final pathology (Table 2). Satellitosis, vertical growth phase and regression were not included in the final analysis due to the high number of unavailable data points. Patients that had reexcision for melanoma in-situ or residual disease at the margins after undergoing a 1 or 2 cm margin were also excluded.

TABLE 2.

Patient demographics and tumor characteristics stratified by resection margin

Variable	All patients	1 cm	2 cm	P value
	N = 965	n = 302 (31.3%)	n = 663 (68.7%)
Gender (%)
Male	592 (61.3)	177 (58.6)	415 (62.6)	0.263
Female	373 (38.7)	125 (41.4)	248 (37.4)
Age (years)
Median (range)	64 (15–96)	67 (15–96)	63 (15–90)	0.004
Histologic Type (%)a
Superficial spreading	562 (58.2)	182 (60.3)	380 (57.3)	0.017
Nodular	218 (22.6)	52 (17.2)	166 (25)
Acral lentiginous	22 (2.3)	12 (4.0)	10 (1.5)
Lentigo maligna	44 (4.6)	21 (7.0)	23 (3.5)
Desmoplastic (NOS)	2 (0.2)	1 (0.3)	1 (0.2)
Pure	14 (1.5)	5 (1.7)	9 (1.4)
Mixed	8 (0.8)	2 (0.7)	6 (0.9)
Other	18 (1.9)	5 (1.7)	13 (2.0)
Location of Primary (%)
Head/neck	195 (20.2)	98 (32.5)	97 (14.6)	<0.001
Trunk	330 (34.2)	57 (18.9)	273 (41.2)
Extremities
UE	275 (28.5)	93 (30.8)	182 (27.5)
LE	165 (17.1)	54 (17.9)	111 (16.7)
Breslow Thickness (mm)
Median	1.4	1.3	1.4	<0.001
Ulceration (%)a
Present	195 (20.2)	64 (21.2)	131 (19.8)	0.679
Absent	750 (77.7)	234 (77.5)	516 (77.8)
LVI (%)a
Present	18 (1.9)	3 (1.0)	15 (2.3)	0.225
Absent	907 (94)	289 (95.7)	618 (93.2)
MR (%)a
<1 mm2	253 (26.2)	75 (24.8)	178(26.8)	0.157
1–2 mm2	375 (38.9)	133 (44)	242 (36.5)
>2 mm2	308 (31.9)	90 (29.8)	218 (32.9)
SLNB Pathology (%)b
Negative	751 (77.8)	214 (70.9)	537 (81.0)	0.190
Positive	114 (11.8)	25 (8.3)	89 (13.4)

LVI lymphovascular invasion, MR mitotic rate, SLNB sentinel lymph node biopsy

^aData unavailable or missing for histological type in 77 cases, for ulceration in 20 cases, for LVI in 40 cases, and for mitotic rate in 29 cases; unavailable or missing data has been removed prior to testing

^bSLNB was not performed in 100 patients

All cases were reviewed and confirmed by a board-certified dermatopathologist. All available original tissue biopsies performed at an outside institution were reexamined prior to clinical evaluation or surgery. Not all patients had a SLNB, which was due to patient preference, comorbid conditions preventing the use of general anesthesia or failure to map on preoperative lymphosctinigraphy. All patients with SLN metastases were offered completion lymph node dissection (CLND) as standard of care. Recurrence during follow-up was categorized as local, regional nodal/in-transit or distant. Local recurrence was defined as recurrence within 2 cm of the scar or graft.

Statistical Analysis

Chi-square tests were performed on categorical variables. Wilcoxon rank sum tests or analysis of variance tests were used for continuous variables to test for differences between 1 and 2 cm margins of resection. Five-year OS and DSS rates were calculated using the Kaplan-Meier (KM) method of estimation. Survival curves were compared with the log-rank test. Statistical significance was determined by a p-value of <0.05. Hazard ratios were estimated by Cox proportional hazards model. All analysis was done in R, version 3.1.0 (a statistical computing environment).

Results

Patient characteristics

A total of 965 patients met study criteria and were included for review and analysis. The median age of the cohort was 64 years with a range of 15–96 years and 592 (61.3%) were male. The median Breslow thickness for the entire cohort was 1.4 mm. The predominant histologic subtype was superficial spreading 562 (58.2%). Of the 965, 302 (31.3%) had a 1 cm margin and 663(68.7%) had a 2 cm margin (Table 2). There was no time trend seen regarding the use of 1 or 2 cm margin widths. One cm margins were more frequently employed for extremity and head and neck locations compared to the trunk. Of the 302 who had a 1 cm margin, 245 (81.1%) were located on the extremities and head and neck compared to 57 (18.9%) on the trunk; however 41.2% of those with truncal melanomas had 2 cm margins (p<0.001). No difference was seen in the histologic prognostic parameters between the two margin groups (Table 2).

Wound closure

Patients included in our study had their wounds closed primarily, with a split-thickness skin graft (STSG), full-thickness skin graft (FTSG) or flap reconstruction. The rate of primary closure and skin graft or flap reconstruction between the two margin groups are shown in Table 3. Of the 965 patients, 660 (68.4%) were closed primarily; while 305 (31.6 %) had either a skin graft or flap reconstruction. Primary closure was most frequent (88.2%) in truncal melanomas (Table 3). Of the 965 patients, 94 had both a 1 cm margin and a skin graft or flap/graft closure. Of those 94, 67 patients (71.3%) had a primary lesion located on the nose, near the eye or temple. Ninety-seven of 195 (49.7%) patients with head and neck melanomas had a 2 cm margin, with 80 (82.5%) of those needing a skin graft or flap for closure (p=0.025). Of the 440 with extremity lesions 293 (67.9%) had a 2 cm margin, and of those, 94 (32.1%) needed a graft or flap reconstruction compared to 25 (17.0%) with 1 cm margins.

TABLE 3.

Distribution of wound closure

Wound closure by margin (%)	All patients N = 965	1 cm n = 302	2 cm n = 663	P value
Primary	660 (68.4)	208 (68.9)	452 (68.2)	0.885
Skin graft/flap	305 (31.6)	94 (31.1)	211 (31.8)
Wound closure by location (%)
Head/neck	195	98	97	0.025
Primary	48 (24.6)	31 (31.6)	17 (17.5)
Skin graft/flap	147(75.4)	67 (68.4)	80 (82.5)
Trunk	330	57	273	0.056
Primary	291 (88.2)	55 (96.5)	236 (86.4)
Skin graft/flap	39 (11.8)	2 (3.5)	37 (13.6)
Extremities	440	147	293	0.001
Primary	321 (73.0)	122 (83.0)	199 (67.9)
Skin graft/flap	119 (27.0)	25 (17.0)	94 (32.1)

Recurrence

The median follow-up of the entire cohort was 15 months with a range of 0.2–129 months. Of the 965 patients, 20 (2.1%) experienced local recurrence, and of those, 6 (2.0%) experienced a local recurrence with a 1 cm margin and 14 (2.1%) with a 2 cm margin, respectively (p=0.791) (Table 4). There was no statistical difference in local recurrence between the margin widths. On univariate analysis (UVA), breslow thickness, LVI, melanoma on the head and neck and a positive SLNB were significant for local recurrence. However, only head and neck location (p=0.004) and SLNB (p<0.001) retained significance on MVA (Table 5). Margin width demonstrated no impact on local recurrence, regional nodal/in-transit or distant recurrence on either UVA or MVA.

TABLE 4.

Patterns of recurrence

Recurrence (%)	All recurrences N = 105	1 cm n = 29 (27.6%)	2 cm n = 76 (72.4%)	P value
Local	20 (2.1)	6 (2.0)	14 (2.1)	0.791
Regional nodal/in-transit	44 (4.6)	12 (4.0)	32 (4.8)	0.946
Distant	41 (4.2)	11 (3.6)	30 (4.5)	0.885

TABLE 5.

Univariate and multivariate analysis of predictive factors for local recurrence

Variable	Local Recurrence
	Univariate		Multivariate
	Odds ratio 95% CI	P value	Odds ratio 95% CI	P value
Breslow Thickness	4.01 (1.01–15.91)	0.048	3.23 (0.64–16.14)	0.154
Ulceration	1.40 (0.50–3.91)	0.519	1.16 (0.39–3.47)	0.786
LVI	7.98 (1.66–38.44)	0.010	3.71 (0.65–21.32)	0.142
Head and Neck location	2.83 (1.14–7.03)	0.025	4.68 (1.62–13.51)	0.004
Margin Width	1.09 (0.41–2.85)	0.868	1.02 (0.34–3.04)	0.971
SLNB status	7.30 (2.93–18.16)	<0.001	8.26 (3.05–22.37)	<0.001

SLNB Sentinel lymph node biopsy, LVI lymphovascular invasion

Survival analysis

At the time of last follow-up, 862 (89.3%) patients were alive (829 were alive without evidence of melanoma and 33 with evidence of recurrent melanoma). Overall, 103 (10.7%) patients died. Of those 41 (39.8%) died of disease and the remaining 62 (60.2%) died of unknown or other causes. Of those who had a 1 cm and 2 cm margin of resection, 38 (12.6%) and 65 patients (9.8%) died, respectively. There were 10 patients (3.3%) with a 1 cm margin and 31 (4.7%) with a 2cm margin that died of melanoma. The median OS for the entire cohort was 92.2 months, while the median DSS was not reached. The 5-year OS and DSS for the entire cohort were 69.3% and 84.7%, respectively (Fig. 1).

FIG. 1.

Overall survival (OS). Kaplan-Meier estimates of OS by margins of resection (2 vs. 1 cm). A significant difference was seen between the different margins.

Overall survival

The 5-year OS for a 1 and 2 cm margin was 61.9% and 71.2 % (HR=0.52, 95 % CI=0.35–0.78 p=0.004), as demonstrated by KM survival analysis (Fig. 1). Margin width, ulceration, age and sex were found to be significant predictors of OS on UVA. Margin width (HR=0.66, CI=0.43–1.04, p=0.050), ulceration (HR=1.67, CI=1.08–2.58, p=0.021) and age (HR=1.03, CI=1.00–1.04, p=0.002) remained significant predictors of OS on MVA.

Disease-specific survival

The 5-year DSS between the two margin groups was not statistically different (1 cm-87%, 2 cm-85%, p=0.758) (Fig. 2). On UVA Breslow thickness, ulceration, LVI and increased MR (>2/mm²) were statistically significant predictors of DSS. However, only LVI (HR=3.14, CI=1.25–7.88, p=0.010) retained significance on MVA. The KM curve demonstrated overlapping curves for both margin groups with a median follow-up of 15 months.

FIG. 2.

Disease-specific survival (DSS). Kaplan-Meier estimates of DSS by margins of resection (2 vs. 1 cm). No significant difference was seen between the different margins. Median survival was not reached.

Discussion

The effect of margin width on recurrence and survival with a melanoma (1–4 mm) has been well studied and documented in the literature. A recent meta-analysis by Lens et al. pooled and analyzed data from 5 prospective trials, which looked at wide versus narrow margins in melanomas ranging in thickness from 1 to 4 mm. The authors concluded there was no statistical difference in local recurrence or overall mortality [12]. The limitations of this study are the inherent flaws seen with a meta-analysis, which included clinical heterogeneity associated with varied follow-up periods, different methods for determining local recurrence, and varying thicknesses. Even though there are discrepancies between those studies, the NCCN has used the data to create guidelines for the recommended excision width of melanoma from 1–4 mm [10]. A randomized, prospective trial comparing a 1 or 2 cm margin width for a 1.01–2 mm is lacking. There have been several smaller, non-randomized, retrospective trials trying to answer the question of which margin is optimal and how will it affect recurrence and survival. We analyzed our extensive database and sought to determine if margin width truly had an impact on local recurrence, DSS and wound closure.

The local recurrence rate in our study was not shown to differ significantly between the two margin widths (p=0.791). We evaluated other factors and found head and neck location, SLNB status and LVI to be significant on UVA. When included in the MVA, only head and neck location and SLNB status retained significance for local recurrence, a finding similar to one published by Hudson et al. [11]. Eight more patients experienced a local recurrence after having a wider margin width compared to the narrow margin. We speculated the surgeon used their best judgment and may have performed a wider margin for lesions with aggressive features, such as ulceration or LVI. When looking at our data more closely, we found patients with LVI have a higher rate of local recurrence with a 2 cm margin width compared to a 1 cm margin width (p=0.010). When we evaluated other aggressive biologic features, such as ulceration, MR and satellitosis, there was no difference seen between the groups. Based on this, we are unable to conclusively state that if a lesion has these specific aggressive features then a wider margin should be utilized, though a surgeon may choose to proceed with a wider margin based on personal preference in their practice.

According to the 7^th edition of the AJCC staging system and the Intergroup Trial, ulceration and head and neck location were associated with a worse OS [4, 13]. Margin width, ulceration and age were significant predictors of OS on MVA in our study; however ulceration was only significant on UVA for DSS. Though our data was similar, our study demonstrated no difference in DSS when accounting for margin width or location. These conclusions may become very important when the surgeon is trying to decide on whether to preserve cosmesis and functionality by avoiding a wider margin, especially on the head or neck or distal extremity.

If there is a choice to perform a narrow margin and avoid a potential skin graft or flap reconstruction, the surgeon will likely gravitate to that option. The use of skin grafts or a flap reconstruction can be cosmetically unappealing and potentially increase the cost of care and the overall morbidity the patient is subjected to [14, 15]. Our study is the one of the first to directly compare the use of skin grafts and flap reconstruction versus primary closure in those receiving a 1 or 2 cm margin width. The majority of wounds were closed primarily, but our study does indicate surgeons favored using a narrow margin in areas where a skin graft or flap reconstruction would likely be needed if a wider margin was used, such as on the head and neck or extremity. Avoiding a 2 cm margin may decrease the need for a skin graft or flap, especially on areas where functionality and cosmesis are more important, such as on the head and neck. We recognize this conclusion is drawn from a retrospective analysis and a larger prospective trial will be needed to accurately answer this question.

Some of the recognized limitations in our study are derived from the inherent flaws of a retrospective study with missing data points for some patients and selection bias. One limitation in this study is the short median follow-up seen with our patients. Most local recurrences occur within the first 2 years [7]. After 2 years of established follow-up in the surgical clinic, the patient is discharged and referred to a local dermatologist. The dermatologist is informed to contact the clinic if there were any signs of recurrence. Longer median follow-ups of 38 and 46 months have been reported in the literature with no difference being seen with OS, DSS or local recurrence [11, 15].

Conclusion

In our experience, utilizing a narrow margin of 1 cm did not increase the risk of local recurrence in our patient population. We demonstrated no difference in DSS between the two margin widths. If surgeons are able to avoid a wider margin of 2 cm, they may be able to reduce the need for use of a graft or flap reconstruction on the head and neck or extremity without potentially increasing the chance of a recurrence or decreasing survival. There is currently a multinational, multicenter, prospective randomized control trial (MelmarT) comparing 1 versus 2 cm margins of excision in patients with ≥ 1 mm melanomas with an estimated completion date of December 2029 [16]. The primary outcome of the trial will evaluate local recurrence and melanoma-specific survival.

Synopsis.

A 1 or 2 cm radial margin is recommended for a 1.01–2 mm melanoma. We sought to determine in a large single institution series if margin width had an impact on local recurrence, disease-specific survival and type of wound closure.