Annals of Surgical Oncology

, Volume 21, Issue 1, pp 292–299

Patterns of Recurrence and Survival After Lymphadenectomy in Melanoma Patients: Clarifying the Effects of Timing of Surgery and Lymph Node Tumor Burden

Authors

    • Sydney Medical SchoolThe University of Sydney
    • Melanoma Institute Australia
    • The Mater Hospital
    • Royal Prince Alfred Hospital
    • Royal North Shore Hospital
  • Sandro Pasquali
    • Melanoma Institute Australia
  • Lauren E. Haydu
    • Sydney Medical SchoolThe University of Sydney
    • Melanoma Institute Australia
  • John F. Thompson
    • Sydney Medical SchoolThe University of Sydney
    • Melanoma Institute Australia
    • The Mater Hospital
    • Royal Prince Alfred Hospital
Melanomas

DOI: 10.1245/s10434-013-3253-6

Cite this article as:
Spillane, A.J., Pasquali, S., Haydu, L.E. et al. Ann Surg Oncol (2014) 21: 292. doi:10.1245/s10434-013-3253-6

Abstract

Background

Melanoma patients with lymph node (LN) metastases have variable survival after lymphadenectomy. This study investigates whether lymphadenectomy at different times in the course of disease progression influences disease-free survival (DFS; time from primary diagnosis to first recurrence after lymphadenectomy), post recurrence survival (PRS; time from first recurrence after lymphadenectomy to death), and overall survival (OS; time from diagnosis to death).

Methods

Between 1992 and 2010, a total of 1,704 patients underwent lymphadenectomy; 502 underwent immediate completion lymphadenectomy (ICL) after positive sentinel node biopsy (SNB), 214 had delayed completion lymphadenectomy (DCL) for regional recurrence after positive SNB with no ICL or after an earlier false-negative SNB, 709 had no SNB and later required delayed therapeutic lymphadenectomy (DTL) for clinically evident metastasis, and 279 had immediate therapeutic lymphadenectomy (ITL) for clinically positive LNs at primary melanoma diagnosis.

Results

Median DFS for ICL, DCL, DTL, and ITL was 68, 48, 82, and 16 months, respectively (p < 0.001). Median PRS for ICL, DCL, DTL, and ITL was 14, 8, 9, and 9 months, respectively (p < 0.001). Median OS for ICL was not reached whilst for DCL, DTL, and ITL it was 71, 101, and 29 months, respectively (p < 0.001). Extranodal spread and tumor, node, metastasis classification system N stage were the only significant prognostic factors for OS within each group. ICL patients had significantly improved DFS (p = 0.005) and OS (p = 0.012) beyond 5 years compared to DTL patients.

Conclusions

Variable outcomes after lymphadenectomy were observed with different timing of surgery and LN tumor burden. ICL patients had the best outcome.

Lymphadenectomy is currently the standard treatment for melanoma patients with lymph node (LN) metastasis.14 When evaluating patient presentations ranging from low-volume subclinical disease detected by sentinel node biopsy (SNB) to clinically detected LN involvement, the mean 5-year survival is 63 %; however, depending on specific circumstances, it ranges from 90 to 20 %.5 The question arises, is this difference in outcome mostly an issue of lead-time bias, or in some cases does the distant metastatic potential of retained metastatic LNs increase over time, presumably with the acquisition of further genetic and molecular competencies beyond those required to enable LN metastases?

To evaluate these complex issues, the four common presentations of patients with LN involvement were categorized for this study, as follows: (1) patients who have a positive SNB (subclinical disease) and who undergo immediate completion lymphadenectomy (ICL); (2) patients with a delayed completion lymphadenectomy (DCL) for regional recurrence when they did not have completion lymphadenectomy after positive SNB, and patients with negative SNB who later develop regional recurrence (false-negative SNB) and require lymphadenectomy (in this study, these two patient subgroups are combined because both had residual nodal disease after SNB); (3) patients with clinically negative LN and no SNB at the time of primary melanoma diagnosis who later develop regional LN failure and delayed therapeutic lymphadenectomy (DTL); and (4) patients who have regional LN disease at the time of primary melanoma diagnosis and who have immediate therapeutic lymphadenectomy (ITL).

This study aimed to establish how the timing of the lymphadenectomy in the course of the disease related to the interval between the diagnosis of the primary tumor and the first recurrence after lymphadenectomy (disease-free survival, DFS), the survival duration after this recurrence (post recurrence survival, PRS), and overall survival (OS) from the time of diagnosis of the primary lesion (Fig. 1). In addition, the prognostic implications of different patterns of recurrence after lymphadenectomy are reported.
https://static-content.springer.com/image/art%3A10.1245%2Fs10434-013-3253-6/MediaObjects/10434_2013_3253_Fig1_HTML.gif
Fig. 1

Patients in the ICL group underwent an immediate lymphadenectomy [sentinel node (SN) positive]; patients in the DCL underwent a delayed lymphadenectomy (>4 months for positive SN(s) (n = 60) or had regional LN recurrence after a negative SNB (n = 154); patients in the DTL group underwent a delayed lymphadenectomy for clinically positive LN (SNB not performed at the time of primary melanoma diagnosis); patients in the ITL group underwent an immediate lymphadenectomy (clinically positive LN at the time of melanoma diagnosis). Study end points, as defined, are depicted at right. DFS disease-free survival, PRS postrecurrence survival, OS overall survival

Methods

This study received approval from the Melanoma Institute Australia (MIA) Research Committee.

Patients

Between January 1992 and February 2010, patients with a proven single primary cutaneous melanoma managed with lymphadenectomy before any other recurrence events were identified from the MIA Research Database. LN dissections were conducted according to standard MIA policy and national melanoma guidelines.2,6,7 The patients were categorized into the four groups described above (Fig. 1). To allow comparison of the four clinical scenarios in a consistent manner, the definitions of DFS, PRS, and OS were used. Survival time was calculated from the date of primary melanoma diagnosis (for DFS and OS analysis) or first relapse after lymphadenectomy (for PRS analysis) to date of last follow-up, melanoma death, or recurrence. Types of recurrence are categorized as previously described.8

Statistical Analysis

The following clinicopathologic factors were categorized for the survival analysis: age (<50 and ≥50 years), sex, primary site, dissected regional LN field, Breslow thickness (≤4 mm, >4 mm), ulceration, mitotic rate of the dermal invasive melanoma (<1/mm2 vs. ≥1/mm2), number of excised nodes (continuous variable), number of positive nodes [classified according to the 7th edition of the American Joint Committee on Cancer tumor, node, metastasis (TNM) staging manual], extranodal spread (ENS), and site of first relapse after lymphadenectomy.9

Univariate survival analyses were performed by the Kaplan–Meier method and the Cox proportional hazard model for categorical and continuous covariates, respectively. Survival curves were constructed and median survival was estimated according to the Kaplan–Meier method and compared with the Wilcoxon or log rank test where appropriate. Kaplan–Meier OS curves were adjusted for confounding covariates to account for differences in primary tumor features across the four scenarios.10,11 The proportional hazard assumption was assessed using Schoenfeld’s residuals for each covariate. Violation of the proportional hazard assumption indicates that the hazard ratio cannot be reliably reported in a Cox regression model because the hazard ratio for at least one category of a covariate varies over time relative to the reference category.12,13 Two methods were used to account for violation of the proportional hazard assumption: first, separate proportional hazard models were run for each stratum of the violating covariate, and second, univariate and multivariate conditional survival was assessed for patients surviving longer than 5 years (DFS and OS).

Statistical analyses were conducted considering a 95 % confidence interval (CI) and were computed by IBM SPSS Statistic 19.0 and Stata SE/10.0.

Results

There were 1,704 eligible patients. Clinicopathologic features according to the four scenarios are reported in Table 1. In the DCL group, there were 60 patients who received a DCL when they experienced relapse after not having an ICL and 154 who experienced a regional LN relapse after an earlier false-negative SNB.
Table 1

Clinicopathologic characteristics of the 1,704 melanoma patients treated by lymphadenectomy for lymph node metastasis according to clinical scenario

Characteristic

Variable

ICL (n = 502)

DCL (n = 214)

DTL (n = 709)

ITL (n = 279)

pa

Age (years)

<50

226 (45.0 %)

67 (31.3 %)

264 (37.2 %)

75 (26.9 %)

<0.001

 

>50

276 (55.0 %)

147 (68.7 %)

445 (62.8 %)

204 (73.1 %)

 

Sex

Male

306 (61.0 %)

130 (60.7 %)

438 (61.8 %)

190 (68.1 %)

0.197

 

Female

196 (39.0 %)

84 (39.3 %)

271 (38.2 %)

89 (31.9 %)

 

Primary tumor site

Head and neck

82 (16.3 %)

34 (15.9 %)

143 (20.2 %)

75 (26.9 %)

<0.001

 

Trunk

198 (39.4 %)

68 (31.8 %)

283 (39.9 %)

82 (29.4 %)

 
 

Extremities

222 (44.2 %)

112 (52.3 %)

283 (39.9 %)

122 (43.7 %)

 

Dissected nodal field

Axilla

225 (44.8 %)

80 (37.4 %)

291 (41.0 %)

94 (33.7 %)

0.007

 

Head and neck

90 (17.9 %)

41 (19.2 %)

173 (24.4 %)

80 (28.7 %)

 
 

Groin

187 (37.3 %)

93 (43.5 %)

245 (34.6 %)

105 (37.6 %)

 

Breslow thickness (161 missing) (mm)

<1.0

25 (5.4 %)

15 (7.9 %)

180 (27.6 %)

11 (4.6 %)

<0.001

 

1.01–2.0

124 (26.6 %)

66 (34.9 %)

209 (32.1 %)

33 (13.9 %)

 
 

2.01–4.0

194 (41.6 %)

73 (38.6 %)

178 (27.4 %)

89 (37.6 %)

 
 

>4

123 (26.4 %)

35 (18.6 %)

84 (12.9 %)

104 (43.9 %)

 

Ulceration (329 missing)

Absent

262 (60.0 %)

99 (56.6 %)

369 (67.7 %)

96 (44.0 %)

<0.001

 

Present

175 (40.0 %)

76 (13.8 %)

176 (32.1 %)

122 (56.0 %)

 

Mitotic rate (280 missing)

0/mm2

16 (3.5 %)

6 (3.3 %)

62 (10.9 %)

12 (5.4 %)

<0.001

 

1/mm2

34 (7.5 %)

12 (6.6 %)

76 (13.3 %)

13 (5.9 %)

 
 

2/mm2

64 (14.2 %)

22 (12.2 %)

97 (17.0 %)

18 (8.1 %)

 
 

3–5/mm2

148 (32.8 %)

47 (26.0 %)

169 (29.6 %)

53 (23.9 %)

 
 

6–10/mm2

111 (24.6 %)

51 (28.2 %)

103 (18.1 %)

55 (24.8 %)

 
 

>10/mm2

78 (17.3 %)

43 (23.8 %)

63 (11.1 %)

71 (32.0 %)

 

AJCC TNM N stageb

N1 (1 LN)

331 (65.9 %)

94 (43.9 %)

340 (48.0 %)

109 (39.1 %)

<0.001

 

N2 (2–3 LNs)

154 (30.7 %)

73 (34.1 %)

199 (28.1 %)

91 (32.6 %)

 
 

N3 (>4 LNs)

17 (3.4 %)

47 (22.0 %)

170 (24.0 %)

79 (28.3 %)

 

No. of excised LNs

Median; IQR

20; 14–28

23; 15–30

19; 13–29

21; 13–32

0.113

Extranodal spread

Absent

491 (97.8 %)

137 (64.0 %)

496 (70.0 %)

189 (67.7 %)

<0.001

 

Present

11 (2.2 %)

77 (36.0 %)

213 (30.0 %)

90 (32.3 %)

 

ICL immediate completion lymphadenectomy, DCL delayed completion lymphadenectomy, DTL delayed therapeutic lymphadenectomy, ITL immediate therapeutic lymphadenectomy, AJCC American Joint Committee on Cancer, TNM tumor, node, metastasis classification system, LN lymph node, IQR interquartile range

aDifferences of each independent variable across the clinical scenarios

bAJCC stage is considered at the time of lymphadenectomy

The median number of excised LNs was 20 [interquartile range (IQR) 14–29], 22 (IQR 17–28), 34 (IQR 25–47), and 14 (IQR 10–19) in all patients and in those who had an axilla, a neck, and a groin dissection, respectively. The number of excised LNs did not differ between the four scenarios (p = 0.113).

The median follow-up after melanoma diagnosis was 69 months (95 % CI 66–73). Multivariate survival analysis for DFS, PRS, and OS (Table 2) was performed with significant features at univariate analysis (Supplementary Material). The variable for clinical scenario was found to violate the proportional hazard assumption in DFS (ICL vs. DTL, p < 0.001) and OS (overall, p < 0.001) analyses, but not in PRS (overall, p = 0.098 or ICL vs. DTL, p = 0.473).
Table 2

Multivariate analysis for prognostic factors according to clinical scenarioa

Scenario

Characteristic

DFS

PRS

OS

  

HR (95 %CI)

p

HR (95 %CI)

p

HR (95 %CI)

p

ICL

Age >50 years

1.64 (1.18–2.27)

0.003

NS

NS

 

Sex male

NS

NS

1.56 (1.06–2.30)

0.023

 

Thickness >4.0 mm

1.65 (1.20–2.26)

0.002

NS

NS

 

Ulceration present

2.09 (1.53–2.85)

<0.001

NS

2.61 (1.79–3.80)

<0.001

 

Mitoses present

NS

3.1 (1.26–7.76)

0.014

NS

 

TNM N2 stage

1.49 (1.09–2.05)

0.013

NS

1.20 (0.81–1.77)

0.366

 

TNM N3 stage

5.51 (2.90–10.46)

<0.001

NS

13.28 (6.50–27.14)

<0.001

 

ENS present

4.97 (2.32–10.66)

<0.001

NS

5.27 (2.45–11.34)

<0.001

 

Regional LN recurrence

NA

NS

NA

 

Distant recurrence

NA

2.91 (1.87–4.55)

<0.001

NA

DCL

Age >50 years

NS

NS

NS

 

Sex male

NS

NS

NS

 

Thickness > 4.0 mm

2.17 (1.35–3.51)

0.001

NS

1.99 (1.17–3.39)

0.011

 

Ulceration present

NS

NS

1.75 (1.08–2.84)

0.022

 

Mitoses present

NS

NS

NS

 

TNM N2 stage

NS

1.84 (1.02–3.31)

0.042

1.49 (0.86–2.59)

0.152

 

TNM N3 stage

NS

NS

2.27 (1.26–4.10)

0.007

 

ENS present

1.90 (1.25–2.91)

0.003

NS

1.60 (1.00–2.56)

0.05

 

Regional LN recurrence

NA

NS

NA

 

Distant recurrence

NA

NS

NA

DTL

Age >50 years

NS

NS

1.52 (1.13–2.06)

0.006

 

Sex male

1.62 (1.24–2.13)

<0.001

NS

1.66 (1.24–2.22)

0.001

 

Thickness >4.0 mm

1.80 (1.28–2.52)

0.001

NS

2.07 (1.48–2.92)

<0.001

 

Ulceration present

NS

NS

NS

 

Mitoses present

NS

NS

1.52 (1.08–2.13)

0.017

 

TNM N2 stage

2.15 (1.32–2.41)

<0.001

NS

2.64 (1.86–3.74)

<0.001

 

TNM N3 stage

2.00 (1.27–2.36)

<0.001

NS

2.92 (2.05–4.15)

<0.001

 

ENS present

1.45 (1.10–1.91)

0.008

NS

1.45 (1.08–1.94)

0.013

 

Regional LN recurrence

NA

NS

NA

 

Distant recurrence

NA

NS

NA

ITL

Age > 50 years

NS

NS

NS

 

Sex male

NS

NS

NS

 

Thickness > 4.0 mm

NS

NS

1.49 (1.02–2.18)

0.04

 

Ulceration present

NS

NS

NS

 

Mitoses present

NS

NS

NS

 

TNM N2 stage

1.12 (0.73–1.73)

0.606

NS

0.93 (0.57–1.51)

0.765

 

TNM N3 stage

1.86 (1.19–2.91)

0.006

NS

2.33 (1.47–3.69)

<0.001

 

ENS present

1.81 (1.22–2.67)

0.003

1.89 (1.22–2.92)

0.004

2.09 (1.39–3.15)

<0.001

 

Regional LN recurrence

NA

NS

NA

 

Distant recurrence

NA

1.81 (1.07–3.07)

0.027

NA

DFS disease-free survival, PRS postrecurrence survival, OS overall survival, HR hazard ratio, CI confidence interval, LN lymph node, TNM tumor, node, metastasis classification system, ENS extranodal spread, NS not significant, NA not applicable (not tested)

aReference categories: sex (female), age (<50 years), thickness (≤4.0 mm), mitotic rate (<1/mm2), ulceration (absent), TNM N stage (N1), site of recurrence (local/in-transit), ENS (absent)

Patterns of First Recurrence after Lymphadenectomy and DFS

When we initially combined all patients from all groups, we noted that recurrence occurred in 819 patients (48 %) at a median time of 57 months (95 % CI 49–65) after primary diagnosis. Local, in-transit, nodal, and distant metastases as a first event after lymphadenectomy were observed in 65 (3.8 %), 127 (7.4 %), 125 (7.3 %), and 502 (29.5 %) patients, respectively. Considering patients who experienced relapsed after lymphadenectomy, the median time to local, in-transit, nodal, and distant metastasis (calculated from the time of primary melanoma diagnosis), was 12 (95 % CI 11–13), 16 (95 % CI 12–19), 16 (95 % CI 12–20), and 25 months (95 % CI 23–27), respectively. The 90th percentile (i.e., the time until 90 % of relapsing patients had experienced relapse) for local, in-transit, nodal, and distant metastasis was 28, 57, 52, and 75 months, respectively.

DFS was significantly different among the four patient groups (p < 0.001; Fig. 2a). Median DFS times for ICL, DCL, DTL, and ITL groups were 68 (95 % CI not reached), 48 (95 % CI 39–56), 82 (95 % CI 66–97), and 16 months (95 % CI 14–19), respectively. On stratified multivariate analysis (Table 2), ENS was the only independent prognostic factor retained in all the multivariate models performed for each clinical scenario. TNM N stage was a significant independent predictor of DFS in all scenarios except the DCL group.
https://static-content.springer.com/image/art%3A10.1245%2Fs10434-013-3253-6/MediaObjects/10434_2013_3253_Fig2_HTML.gif
Fig. 2

Kaplan–Meier survival curves for a DFS and b PRS

In those patients who had survived that long, DFS after 5 years was significantly different when comparing the ICL (n = 113) and DTL (n = 283) groups (p = 0.005). This difference was still significant after adjustment in multivariate analysis [hazard ratio (HR) 2.57; HR 1.14–5.85; p = 0.023], as was TNM N stage (N2 vs. N1, HR 2.20; 95 % CI 1.75–5.88; p < 0.001; N3 vs. N1, HR 3.16; 95 % CI 1.69–5.92; p < 0.001). This analysis did not violate the proportional hazard assumption (p = 0.286).

Postrecurrence Survival

Combining patients from all groups, the median PRS in those who experienced relapse after lymphadenectomy was 9 months (95 % CI 7–10). In patients with first relapse as local/in-transit, nodal, and distant relapse, median PRS was 18 (95 % CI 14–21), 18 (95 % CI 11–24), and 7 (95 % CI 6–8) months, respectively (p < 0.001).

Patients in the ICL group had a significantly longer PRS than patients in the other clinical scenarios (log rank test, p < 0.001; Fig. 2b). PRS times for ICL, DCL, DTL, and ITL groups were 14 (95 % CI 9.6–18.4), 8 (95 % CI 6.3–9.7), 9 (95 % CI 7.2–10.7), and 9 (95 % CI 6.7–11.3) months, respectively. Between groups, the log rank test found the following results: ICL vs. DCL, p < 0.001; ICL vs. DTL, p < 0.001; ICL vs. ITL, p < 0.001; DCL vs. DTL, p = 0.424; DCL vs. ITL, p = 0.769; DTL vs. ITL, p = 0.179. On multivariate analysis performed according to the four clinical scenarios (Table 2), distant site of first recurrence was a significant independent prognostic factor for PRS in all scenarios, except for the DCL group, where no independent predictors were identified.

Overall Survival

Combining all patients from all groups, there were 675 deaths due to melanoma (39.6 %), and the median survival from the time of primary melanoma diagnosis for all patients was 91.7 months (95 % CI 80.7–102.9). OS was significantly different across clinical scenarios (p < 0.001; Fig. 3a). Median survival for patients in ICL group was not reached, while for DCL, DTL, and ITL groups it was 71.1 (95 % CI 45.8–96.4), 101.3 (95 % CI 86.1–116.0), and 29.2 (95 % CI 22.7–35.8) months, respectively. In Table 2, analyses of prognostic factors according to the four different scenarios are reported. ENS and TNM N stage were the only two factors that significantly influenced OS in all the clinical scenarios. Adjusted OS curves, allowing for primary tumor differences between groups, are shown in Fig. 3b.
https://static-content.springer.com/image/art%3A10.1245%2Fs10434-013-3253-6/MediaObjects/10434_2013_3253_Fig3_HTML.gif
Fig. 3

Kaplan–Meier survival curves for a cumulative and b adjusted OS for scenarios described in Fig. 1

For patients who survived that long, OS after 5 years was significantly different when comparing the ICL (n = 141) and DTL (n = 332) groups (p = 0.012). However, on multivariate analysis, TNM N stage was the only predictor of OS in patents who survived >5 years (N2 vs. N1, HR 2.37, 95 % CI 1.35–4.14, p = 0.002; N3 vs. N1, HR 4.15, 95 % CI 2.38–7.24, p < 0.001). This analysis did not violate the proportional hazard assumption (p = 0.124).

Discussion

The treatment of choice for melanoma patients with LN metastasis is lymphadenectomy; however, the survival of patients undergoing this treatment ranges widely, depending on the timing of detection and the volume of metastatic disease.5,1416 After surgery, there is no current standard adjuvant therapy. Newer therapeutic agents, such as immune checkpoint drugs, BRAF inhibitors, and MEK inhibitors have been demonstrated to be effective in advanced melanoma, and these drugs are in the process of being evaluated in adjuvant therapy clinical trials.1722 These clinical trials will be recruiting high-risk patients with a predominance of those resected LN disease. Optimizing patient stratification to facilitate better design and more rational interpretation of results of these trials is important.

The current study highlights differences in DFS, PRS, and OS for the patients, depending on the clinical scenario of their lymphadenectomy using a definition of DFS that compensates for the lead-time bias introduced by SNB. Patients having ICL after a positive SNB had superior long-term DFS, progression-free survival, and OS despite the primary tumor characteristics of that group being indicative of a worse prognosis than the main comparator of clinical interest, the DTL group.

First Relapse After Lymphadenectomy

Overall, combining all groups, 48 % of patients experienced a recurrence after lymphadenectomy. The overall incidence of recurrence after lymphadenectomy reported in the literature ranges between 15 and 87 %.2328 Previous studies are consistent with this study and have reported local/in-transit, regional node field, and distant recurrence occurring in 7.8–15.5, 2.8–25, and 32–68 %, respectively.2328

Disease-free Survival

DFS was significantly different among the four clinical scenarios (Fig. 2a). ENS was the only independent predictor of shorter DFS that achieved significance in all four clinical scenarios. In the earlier part of the DFS curve, ICL patients had worse DFS than DTL patients. Most likely this is related to the higher incidence of high-risk tumors in the ICL group as well as the absence of early events attributable to patients in the DTL group, because they had a delayed procedure (Fig. 2a). Therefore, we selectively assessed patients who survived longer than 5 years, and those in the ICL group had a better long-term outcome than those in the DTL group (p = 0.005), even after adjusting for known prognostic factors (HR 2.57; p = 0.023). This observation is only hypothesis generating, but it suggests that SNB may prevent late recurrence events. However, because we did pairwise comparisons, the unadjusted p value is an overestimate and is vulnerable to type I error.

Postrecurrence Survival

Combining all groups, after lymphadenectomy a recurrence at any site was associated with median survival of 9 months. PRS differed according to site of first relapse. Patients with local/in-transit, regional LN, and distant metastases had a median survival after recurrence of 18, 18, and 7 months, respectively (p < 0.001). Patients in the ICL group had a significantly longer PRS (14 months) than the DCL group (8 months, p < 0.001) and the DTL and ITL groups (9 months, p < 0.001). In a study of recurrence in patients with positive and negative sentinel nodes, Dalal et al.24 reported a median PRS after in-transit, regional, and distant metastasis of 37, 20, and 9 months, respectively, with a median PRS of 16 months.

Overall Survival

There were significant differences in OS according to the clinical scenario. In the ICL group, median survival was not reached. In comparison, it was 71, 101, and 29 for patients in the DCL, DTL, and ITL groups, respectively (Fig. 3a, p < 0.001). ENS and TNM N stage were independent prognostic factors for OS within all scenarios. One striking finding from this study was that the OS of patients in the ICL group plateaued above 50 % in the long term, whereas the other clinical scenarios seemed to converge toward 30 % long-term OS. Importantly, the difference only started to become apparent beyond 7 years after diagnosis. Again, as with DFS, in the earlier part of the OS curve, ICL patients were doing worse than DTL patients for the reasons discussed above. Because there is a time-dependent higher LN disease burden observed in DTL patients, adjusted OS curves (Fig. 3b) were generated for each scenario, which indicated that ICL patients had the best prognosis.14 Whether this finding underlies a lower tumor burden (Table 1) or a possibly effect of an immediate dissection can only be clarified by a randomized study such as the MSLT-2, which is comparing lymphadenectomy and observation in sentinel node–positive patients.

There are clearly limitations to the interpretation of the results of this study. It was not a controlled study, and there are significant differences in tumor features across the four clinical scenarios. In addition, adjuvant treatment factors have not been reported and may well be different between groups, and even though to date they are likely to have a limited impact on OS outcome, radiotherapy would affect regional control.

Using this definition of DFS to avoid lead-time bias is not without other selection biases. For instance, potential patients who did not have SNB who were later included in the DTL group would have been selected for improved prognosis by patients with early, aggressive biology who experienced relapse with distant disease before manifesting regional recurrence. In contrast, similar patients with aggressive biology in the ICL group would have completed their lymphadenectomy, then experienced early relapse and been included in the ICL analysis.

In conclusion, this study provides substantial additional information and understanding of the patterns and timing of recurrence and prognostic factors affecting DFS, PRS, and OS for the four different clinical scenarios of lymphadenectomy for metastatic melanoma. These variations should be taken into account in adjuvant therapy clinical trial design, as patients with LN involvement will constitute a major component of the high-risk patients evaluated. Patients who had ICL for positive SNB had an improved progression-free survival as well as better long-term DFS and OS compared to other clinical scenarios.

Acknowledgment

A.J.S.’s research is supported by the Friends of the Mater Foundation, North Sydney, NSW, Australia. S.P. was a research fellow at Melanoma Institute Australia and was supported by the University of Padova, Padova, Italy. J.F.T.’s research is supported by the Melanoma Foundation of the University of Sydney.

Disclosure

The authors declare no conflict of interest.

Supplementary material

10434_2013_3253_MOESM1_ESM.docx (15 kb)
Supplementary material 1 (DOCX 15 kb)

Copyright information

© Society of Surgical Oncology 2013