Annals of Surgical Oncology

, Volume 19, Issue 4, pp 1089–1099

Sentinel Lymph Nodes Containing Very Small (<0.1 mm) Deposits of Metastatic Melanoma Cannot Be Safely Regarded as Tumor-Negative

Authors

    • Tissue Pathology and Diagnostic OncologyRoyal Prince Alfred Hospital
    • Melanoma Institute Australia
    • Discipline of Pathology, Sydney Medical SchoolThe University of Sydney
    • Human Oncology and Pathogenesis Program and Department of PathologyMemorial Sloan-Kettering Cancer Center
  • Chitra DeSilva
    • Melanoma Institute Australia
  • Stanley W. McCarthy
    • Tissue Pathology and Diagnostic OncologyRoyal Prince Alfred Hospital
    • Melanoma Institute Australia
    • Discipline of Pathology, Sydney Medical SchoolThe University of Sydney
  • John F. Thompson
    • Melanoma Institute Australia
    • Discipline of Surgery, Sydney Medical SchoolThe University of Sydney
  • Richard A. Scolyer
    • Tissue Pathology and Diagnostic OncologyRoyal Prince Alfred Hospital
    • Melanoma Institute Australia
    • Discipline of Pathology, Sydney Medical SchoolThe University of Sydney
Melanomas

DOI: 10.1245/s10434-011-2208-z

Cite this article as:
Murali, R., DeSilva, C., McCarthy, S.W. et al. Ann Surg Oncol (2012) 19: 1089. doi:10.1245/s10434-011-2208-z

Abstract

Background

Some authors have suggested that patients with very small (<0.1 mm) deposits of metastatic melanoma in sentinel lymph nodes (SLNs) should be considered SLN-negative, whereas others have reported that such patients can have adverse long-term outcomes. The aims of the present study were to determine whether extensive sectioning of SLNs resulted in more accurate categorization of histologic features of tumor deposits and to assess prognostic associations of histologic parameters obtained using more intensive sectioning protocols.

Methods

From patients with a single primary cutaneous melanoma who underwent SLN biopsy between 1991 and 2008, those in which the maximum size of the largest tumor deposit (MaxSize) in SLNs was <0.1 mm in the original sections were identified. Five batches of additional sections were cut from the SLN tissue blocks at intervals of 250 μm. The 1st batch was cut from the blocks without any trimming; these sections were therefore immediately adjacent to the original sections. Each batch included 5 sequential sections, the 1st and 5th stained with hematoxylin-eosin, and the 2nd, 3rd, and 4th stained immunohistochemically with S-100, HMB-45, and Melan-A, respectively. In each batch of sections, the following histologic features of tumor deposit(s) in the SLNs were evaluated: MaxSize; tumor penetrative depth (TPD) (defined as the maximum depth of tumor deposit(s) from the inner margin of the lymph node capsule), and intranodal location (classified as subcapsular if the tumor deposit(s) were confined to the subcapsular zone or parenchymal if there was any involvement of the nodal parenchyma beyond the subcapsular zone). The measured histologic parameters were compared in each batch of sections. The association of histologic parameters with overall survival was assessed for the parameters measured in each batch of sections.

Results

There were 20 eligible patients (15 females, 5 males, median age 60 years). After a median follow-up duration of 40 months, 4 patients had died from melanoma and 2 patients of unknown causes. Completion lymph node dissection (CLND) was performed in 13 cases (65%) and was negative in all cases. Relative to the measured values on the original sections, all 3 parameters were upstaged in subsequent batches of sections, but no further upstaging of MaxSize, TPD, or location was seen beyond batch 3, batch 4, and batch 2, respectively. Increasing MaxSize was associated with significantly poorer overall survival in batches 1, 2, and 3. Parenchymal involvement was significantly associated with poorer survival in batches 2–5. TPD was not significantly associated with overall survival.

Conclusions

The results of this study indicate that very small (<0.1 mm) deposits of melanoma in SLNs may be associated with adverse clinical outcomes and that this is due, at least in part, to the underestimation of SLN tumor burden in the initial sections. Our evidence does not support clinical decision-making on the assumption that patients with very small melanoma deposits in SLNs have the same outcome as those who are SLN-negative.

Histologic features of metastatic melanoma in sentinel lymph nodes (SLNs), such as increasing size of tumor deposits, increasing (subcapsular) tumor penetrative depth (TPD) of tumor, and nonsubcapsular intranodal location of tumor have been shown in various studies to be predictive of non-SLN tumor involvement and poorer survival.123 Recently, we demonstrated that a combination of several histologic features of metastatic tumor in SLNs can accurately predict non-SLN involvement and survival in a large cohort of SLN-positive melanoma patients.24,25

The prognostic significance of very small (<0.1 mm) deposits of metastatic melanoma in SLNs is controversial.2628 Some authors have suggested that patients with SLNs harboring metastatic melanoma <0.1 mm in maximum size have a prognosis equivalent to that of SLN-negative patients and that these patients should be managed as if they were SLN-negative.29 In contrast, other authors have observed that patients with very small SLN tumor deposits can have adverse long-term outcomes.8 Reported differences in prognosis are likely caused by many other factors besides the size of tumor deposits in SLNs influence prognosis in melanoma patients.24,25,30

Nevertheless, it is well established that tumor burden in SLNs measured using the size of SLN tumor deposits is an important prognostic factor. A major factor influencing the measured size of tumor deposits in SLNs is the histologic protocol used to section SLNs, which varies widely between institutions. Previous studies have shown that more extensive sectioning protocols increase the detection rates of metastatic melanoma in SLNs and also reveal larger deposits.3135 Therefore, it is conceivable that the true size of SLN tumor deposits may be underestimated using protocols that examine less of the cross section of the SLNs than more extensive sectioning protocols. If this is correct, some patients with SLN tumor deposits reported to be <0.1 mm will be found to have tumor deposits >0.1 mm if more sections are examined.

The aim of the present study of patients with SLNs containing very small deposits of melanoma was to determine if more extensive sectioning of SLNs resulted in more accurate categorization of histologic features of the tumor deposits and to assess the prognostic associations of the histologic parameters obtained using more extensive sectioning protocols.

Methods

The study was carried out with the approval of the institutional Ethics Committee. Patients with primary cutaneous melanoma who underwent SLN biopsy (SLNB) between 1991 and 2008 were identified from the database of Melanoma Institute Australia (MIA). Patients with more than 1 primary melanoma and those whose histologic slides of SLNs were not available for review were excluded. Only cases in which the maximum size of the largest tumor deposit in the SLNs was <0.1 mm in the original sections were included in the study. The primary melanoma was reported (or reviewed in cases diagnosed in external laboratories) by pathologists at the Royal Prince Alfred Hospital (RPAH), Sydney, Australia. Clinical and pathologic information on the primary melanomas was retrieved from the databases of MIA and RPAH.

SLNs were initially processed as follows: The SLNs were bisected longitudinally in a paramedian plane and embedded in paraffin blocks with the cut surfaces facing upward. Four consecutive sections were cut; the 1st and 4th sections were stained with hematoxylin-eosin, and the 2nd and 3rd sections were stained immunohistochemically with S-100 and HMB-45. These were the original sections.

Paraffin-embedded tissue blocks from eligible cases were retrieved, and five additional batches of sections were cut from the blocks at intervals of 250 μm. The 1st batch was cut from the blocks without any trimming, and the sections were therefore immediately adjacent to the original sections. Each batch included 5 sequential sections, the 1st and 5th stained with hematoxylin-eosin, and the 2nd, 3rd, and 4th stained immunohistochemically with S-100, HMB-45, and Melan-A, respectively. The additional sections therefore examined the SLNs a further 1 mm from the paramedian plane.

In each batch of sections, the following histologic features of tumor deposit(s) in the SLNs were evaluated: maximum size of the largest tumor deposit (MaxSize); TPD, defined as the maximum depth of tumor deposit(s) below the lymph node capsule; and intranodal location, classified as subcapsular if the tumor deposit(s) were confined to the subcapsular zone or parenchymal if there was any involvement of the nodal parenchyma beyond the subcapsular zone. We also applied our recently described Non-Sentinel Node Risk Score (N-SNORE)24 to these cases.

The measured histologic parameters were compared for each batch of sections to determine the degree of change with each additional batch of sections. The association of histologic parameters with overall survival was assessed for the parameters as measured for each batch of sections. Survival analysis was carried out using the Kaplan-Meier method, and differences in survival curves were analyzed using the log-rank (Mantel-Cox) test. A P value <0.05 was considered statistically significant. Statistical analysis was carried out using Prism 5.0d software (Graphpad Software Inc., La Jolla, CA).

Results

There were 409 patients with positive SLNs in whom slides were available for review. Twenty-three patients (5.6%) had SLNs in which MaxSize was <0.1 mm. Of these, paraffin blocks of the involved SLNs from 20 patients were available. There were 15 females and 5 males, with a median age of 60 years (range 24–78 years). After a median follow-up duration of 40 months (range 8–83 months), 13 patients were alive with no sign of melanoma recurrence, 1 patient was alive with melanoma, 4 patients had died of melanoma, and 2 patients had died of unknown causes. Completion lymph node dissection (CLND) was performed in 13 patients (65%) and was negative in all cases (Table 1). Regional nodal recurrence was not observed in the 7 patients who did not undergo CLND, although 1 patient (case No. 14) developed lung, liver, and brain metastases 8 months following primary melanoma diagnosis.
Table 1

Clinical and pathologic features

Case No.

Age/

sex

Primary melanoma

SLNs: No. positive/

No. removed

Completion lymph node dissection finding

Status at last follow-up

Anatomic Site

Histologic type

Thickness (mm)

Clark level

Mitotic rate (per mm2)

Ulceration

1

72 M

Lower limb

ALM

3.10

V

0

Not seen

1/2

Negative

Alive NSR

2

65 F

Lower limb

SSM

1.10

IV

1

Not seen

1/1

Not performed

Alive NSR

3

62 F

Lower limb

Not classified

1.15

IV

5

Not seen

1/2

Negative

Alive NSR

4

67 F

Trunk

SSM

1.30

IV

4

Present

1/3

Negative

Alive NSR

5

61 F

Upper limb

SSM

1.10

III

0

Not seen

1/3

Negative

Alive NSR

6

60 F

Lower limb

SSM

1.10

IV

2

Not seen

1/3

Not performed

Alive NSR

7

55 F

Trunk

SSM

1.85

IV

2

Not seen

1/1

Not performed

Alive NSR

8

48 F

Trunk

SSM

1.05

IV

2

Not seen

1/4

Negative

Alive NSR

9

43 F

Lower limb

SSM

0.95

III

1

Not seen

1/3

Negative

Alive NSR

10

39 F

Trunk

NM

0.65

IV

2

Not seen

1/1

Negative

Alive NSR

11

34 F

Lower limb

SSM

0.95

IV

2

Not seen

1/3

Negative

Alive NSR

12

34 F

Lower limb

Spitzoid

1.10

IV

3

Not seen

1/3

Negative

Alive NSR

13

24 F

Lower limb

Spitzoid

1.90

III

4

Not seen

1/2

Not performed

Alive NSR

14

60 F

Head/neck

NM

3.50

V

5

Not seen

1/4

Not performed

Alive with melanoma

15

65 F

Lower limb

ALM

3.50

IV

3

Not seen

1/1

Negative

Dead, cause unknown

16

62 M

Trunk

SSM

1.90

IV

4

Present

1/3

Not performed

Dead, cause unknown

17

78 M

Trunk

NM

3.70

III

18

Present

1/2

Negative

Dead of melanoma

18

62 M

Trunk

NM

1.30

IV

5

Not seen

1/4

Not performed

Dead of melanoma

19

50 F

Trunk

Not classified

2.50

IV

10

Present

1/2

Negative

Dead of melanoma

20

54 M

Trunk

NM

6.00

IV

21

Present

1/5

Negative

Dead of melanoma

M male, F female, ALM acral lentiginous melanoma, Alive NSR alive, no sign of recurrence, SSM superficial spreading melanoma, NM nodular melanoma

Relative to the measured values on the original sections, all 3 parameters of nodal metastasis were upstaged in subsequent batches of sections, but no further upstaging of MaxSize, TPD, or location was seen beyond batch 3, batch 4, and batch 2, respectively (Fig. 1a). Examples of the variation seen in these parameters in different batches of sections are shown in Fig. 1b–g.
https://static-content.springer.com/image/art%3A10.1245%2Fs10434-011-2208-z/MediaObjects/10434_2011_2208_Fig1_HTML.gif
Fig. 1

a Number of cases upstaged at each deeper batch of sections for maximum size of largest tumor deposit (MaxSize), tumor penetrative depth (TPD), and intranodal location of deposit (parenchymal considered higher stage than subcapsular). be Case 6, showing variation in histologic parameters: MaxSize, TPD (indicated by black bars), and location, respectively (all S-100 immunostains with red chromogen). b Original sections: 0.05 mm, 0.04 mm, and subcapsular (the inner limit of the fibrous capsule of the lymph node is indicated by a thin black line). c Batch 1: 0.20 mm, 1.03 mm, and parenchymal. d Batch 2: 0.40 mm, 0.40 mm, and parenchymal. e Batch 3: 0.04 mm, 0.02 mm, and subcapsular (deposit adjacent to a connective tissue septum, indicated by arrows, which extends into the nodal parenchyma from the fibrous capsule). fg Case 8, showing variation in MaxSize. f Original sections: 0.06 mm. g Batch 1: 0.25 mm

In 8 cases (40%), MaxSize was greater in the additional sections than in the original sections. MaxSize after examination of the additional sections was >0.1 mm in all 8 cases, >0.2 mm in 5 cases, >0.5 mm in 3 cases, and >1 mm in 2 cases (Fig. 1a). Increased MaxSize was detected in batch 1 in 5 cases, in batch 2 in 2 other cases, and in batch 3 in 1 additional case (Fig. 2a). The highest MaxSize was found in batch 1 in 1 case, in batch 2 in 5 cases, and in batch 3 in 2 cases. No further increase in MaxSize was detected in batches 4 and 5.
https://static-content.springer.com/image/art%3A10.1245%2Fs10434-011-2208-z/MediaObjects/10434_2011_2208_Fig2_HTML.gif
Fig. 2

a Distribution of maximum size of largest tumor deposit at each deeper batch of sections. b 5 Year survival. cf Association of maximum size of largest tumor deposit at each deeper batch of sections with overall survival. c Original batch. d Batch 1 (P = 0.0003). e Batch 2 (P = 0.002). f Batches 3–5 (P = 0.002)

We examined the association of the final MaxSize measurement (after assessment of all batches of sections) with clinical outcomes at last follow-up. In patients with MaxSize <0.1 mm, 8 (67%) were alive with no sign of melanoma recurrence, 1 (8%) was alive but had suffered melanoma recurrence, and 3 (25%) had died of melanoma. In patients with MaxSize 0.1–1.0 mm, 5 (83%) were alive with no sign of recurrence, and 1 (17%) had died of unknown causes. Both patients with MaxSize >1.0 mm died, 1 from melanoma, and the other of unknown cause. Increasing MaxSize stratified into batches of <0.1 mm, 0.1–1.0 mm, and >1.0 mm was associated with significantly poorer overall survival in batches 1, 2, and 3 (Fig. 2b–e).

TPD was greater in the additional sections than in the original sections in 8 cases (40%). Increased TPD was detected in batch 1 in 2 cases, in batch 2 in a further 2 cases, in batch 3 in 2 additional cases, and in batch 4 in 2 more cases (Figs. 1a, 3a). TPD was not significantly associated with overall survival (Fig. 3b–d).
https://static-content.springer.com/image/art%3A10.1245%2Fs10434-011-2208-z/MediaObjects/10434_2011_2208_Fig3_HTML.gif
Fig. 3

a Distribution of TPD at each deeper batch of sections. b 5 Year survival. ce Association of TPD at each deeper batch of sections with overall survival (TPD stratified into SI <0.1 mm; SII = 0.31–1.0 mm; SIII >1.0 mm). c Original batch and batch 1 (P = 0.59). d Batches 2 and 3 (P = 0.54). e Batches 4 and 5 (P = 0.39)

In the original sections, tumor deposits were confined to the subcapsular zone in 19 cases (95%) and involved the parenchyma in 2 cases (10%). Parenchymal involvement was identified in 4 additional cases in batch 1 and in 2 further cases in batch 2 (Fig. 1a). Parenchymal involvement was present in 8 cases (40%) after examination of the additional sections (Fig. 4a). Parenchymal involvement was significantly associated with poorer survival in batches 2–5 (Fig. 4b–d).
https://static-content.springer.com/image/art%3A10.1245%2Fs10434-011-2208-z/MediaObjects/10434_2011_2208_Fig4_HTML.gif
Fig. 4

a Distribution of intranodal location of deposits at each deeper batch of sections. b 5 Year survival. ce Association of intranodal location at each deeper batch of sections with overall survival. c Original batch (P = 0.30). d Batch 1 (P = 0.13). e Batches 2–5 (P = 0.048)

The distribution of N-SNORE worsened (more cases in the 4–8 N-SNORE group) with more extensive sectioning (Fig. 5a). Higher N-SNORE was significantly associated with poorer survival in the later batches of sections (batches 3–5), but not in the original sections (Fig. 5b–e).
https://static-content.springer.com/image/art%3A10.1245%2Fs10434-011-2208-z/MediaObjects/10434_2011_2208_Fig5_HTML.gif
Fig. 5

a Distribution of N-SNORE at each deeper batch of sections. b 5 Year survival. cf Association of N-SNORE at each deeper batch of sections with overall survival. c Original batch (P = 0.30). d Batch 1 (P = 0.21). e Batch 2 (P = 0.08). f Batches 3–5 (P = 0.01)

Discussion

At MIA, and in most melanoma treatment centers worldwide, only patients with positive SLNs (i.e., when melanoma deposits of any size are present in SLNs) are offered CLND, as several studies have demonstrated that SLN status is an accurate predictor of the tumor-harboring status of the entire regional lymph node field. However, it is well documented that the CLND will be negative in the great majority (~80%) of SLN-positive patients. It is therefore possible that CLND may be unnecessary in at least some of these patients and that they could be safely spared the morbidity associated with this procedure. This has led to several attempts to better predict the status of non-SLNs in SLN-positive patients, so that those with a negligible risk of non-SLN involvement may be spared potentially unnecessary CLND.36,9,1215,18,2124 Although several clinical characteristics and features of the primary tumor and SLN tumor deposits have been found to be predictive of non-SLN positivity, the findings are highly variable in these studies. This variability is probably due to differences in methods used for SLN identification, sectioning, and histologic processing.36,14,2124 In addition to non-SLN positivity, several studies have also found that histologic characteristics of SLN tumor deposits are predictive of patient survival, in particular that increasing tumor size in SLNs is associated with poorer survival.46,8,1214,18,3641 Some authors consider SLN tumor of any size to be predictive of an adverse clinical outcome, while others have suggested that patients with SLNs containing very small metastases should be considered SLN-negative.8,29 Our results indicate that it is inappropriate to consider patients with very small (<0.1 mm) deposits in SLNs as being SLN-negative. Such an interpretation may potentially result in adverse clinical outcomes because of failure to adequately treat biologically important disease in these patients.

Many studies have shown that the sectioning protocol used to evaluate SLNs influences SLN positivity rates and the histologic features of SLN deposits, and the use of more extensive sectioning protocols results in detection of a greater proportion of SLN metastases.3133,35,42 However, it has been demonstrated that deposits of melanoma in SLN are distributed throughout the lymph node, and therefore the highest accuracy in evaluation of tumor deposits in SLNs requires comprehensive sequential sectioning through the entirety of SLNs and examination of all sections.43 Obviously, given the large number of SLNB procedures performed, such a comprehensive protocol is impractical and not cost-effective. Less extensive protocols have been implemented in different centers around the globe, but at this time, conclusive superiority of any one protocol over others in terms of prognostic accuracy has not been unequivocally demonstrated.

A key finding of this study is that even in those patients in whom further sectioning of SLNs failed to identify any metastases with MaxSize >0.1 mm, the outcome was still poorer than anticipated for patients with “sentinel node negative” disease. SLN tumor burden, measured in many ways, such as MaxSize or TPD, and intranodal location of tumor deposits have been shown to be predictive of non-SLN status. It has been suggested that in patients with low tumor burden in SLNs (defined as MaxSize <0.1 mm), the risk of non-SLN positivity and adverse clinical outcomes is akin to that of patients with negative SLNs.29 On the one hand, it is possible that patients with a low SLN tumor burden may not progress, possibly because of regression of the tumor or immune-mediated destruction of the tumor cells. On the other hand, it is conceivable that very small tumor deposits may represent dormant micrometastases, and that given sufficient time and additional genetic hits, they will escape from dormancy and progress to macrometastasis and disseminated disease;8 this may explain the eventual death due to melanoma of 3 of the patients in the present study (cases 17, 18, and 20) in whom, despite extensive examination, the MaxSize was <0.1 mm. Alternatively, they may simply indicate that the melanoma has the capacity for dissemination and metastatic colonization, and that the latter may occur at any anatomic site in which a permissive microenvironment exists.

A very pertinent issue is the accuracy of measurement of tumor burden. Given that only a minor proportion of any SLN is examined in routine pathologic assessment, even with extensive protocols in routine use in some centers, one would expect two things. Firstly, some small deposits may be missed altogether (if they are small enough to lie between examined sections); in other words, truly positive SLNs would be classified as negative. Evidence for this comes from studies in which, following nodal recurrence, negative SLNs were sectioned further and were shown in some cases to contain tumor that was missed in the initial evaluation.44,45 Secondly, some deposits that are measured as being very small in one plane of section may in fact be substantially larger in subsequent sections (akin to the “tip of an iceberg” where only a small proportion of the ice mass is present above the sea surface compared with its much greater size beneath the surface). It has been demonstrated that deposits of melanoma in SLNs are usually distributed throughout the lymph node, and a small deposit in one region of the SLN may be sampled,43 while a larger deposit in another region may not be represented in the sections. Furthermore, aggregations of tumor cells are often irregularly shaped, which increases the likelihood of tangential sections of deposits at the edges of deposits being present in examined sections. Some possible scenarios are illustrated in Fig. 6.
https://static-content.springer.com/image/art%3A10.1245%2Fs10434-011-2208-z/MediaObjects/10434_2011_2208_Fig6_HTML.gif
Fig. 6

Possible scenarios to explain variability in tumor burden indices measured in sections from different regions of a lymph node. A bivalved lymph node is illustrated, with horizontal lines indicating the positions of the original and subsequent batches of sections

That SLNs may be falsely negative as a result of failings of pathologic assessment has been previously demonstrated.46 The goal of this study was to focus on the accuracy of measurement of SLN tumor burden in patients with very small (<0.1 mm) tumor deposits on initial evaluation. We found that the measured values of all 3 parameters of SLN tumor burden, MaxSize, TPD, and intranodal location, increased (or in other words, patients were upstaged) with additional sectioning of the SLNs. Therefore, upon identification of very small melanoma deposits in SLNs, we recommend that pathologists should cut and examine additional histologic sections of the involved node(s) in an attempt to establish the histologic features of the deposit(s) with greater accuracy.

TPD and intranodal location, 2 additional prognostically predictive SLN tumor burden parameters, were identified more often with additional sectioning of the SLNs. It is significant that in 5 cases, a higher MaxSize was detected in batch 1, immediately adjacent to the original sections. These are important findings, which suggest that regardless of the histologic protocol used, deposits of melanoma measured as being very small in SLNs may in fact be larger and therefore may be associated with a poorer prognosis.

No further upstaging on the basis of these parameters was seen beyond batch 4 (Fig. 1). At first glance, this suggests that with our protocol, addition of 4 batches of sections as described previously may be sufficient for accurate measurements of SLN tumor burden parameters. However, despite the additional batches of sections cutting a further 1 mm into each of every bisected lymph node, there was always a large amount of additional nodal tissue (potentially harboring tumor deposits) in the block that was not examined. Also, the steps of 250 μm between the batches of sections are considerable gaps within which might reside tumor deposits larger than those seen in the examined sections (Fig. 6). Determination of the optimal approach will require further studies involving large numbers of SLNs from patients with long-term follow-up, in which various protocols are compared with respect to their accuracy of estimation of SLN tumor burden parameters and their correlations with clinical outcomes. One possible approach might involve serial sectioning of SLNs, computerized three-dimensional (3D) reconstruction of the entire node (including any metastases) and application of various sectioning protocol computer simulations to the reconstructed models to determine the minimum number of sections (and their spacing) that will ensure detection of all metastases and accurate quantification of their characteristics such as MaxSize, TPD and intranodal location.

Although we described N-SNORE as a risk score to allow stratification of risk of regional non-SLN involvement,24 and although completion lymphadenectomy in the current cohort was negative in all cases in which it was performed, we chose to evaluate N-SNORE, as the status of the regional lymph nodes (which N-SNORE predicts) is an important prognostic factor in melanoma patients. Interestingly, although N-SNORE calculated from parameters measured in the original sections or in batches 1–2 was not significantly predictive of survival, once the parameters had been maximally upstaged (in batches 3–5), higher N-SNORE values were significantly associated with poorer survival.

Finally, CLND was negative in all cases in which it was performed, and on this basis, one might argue that this supports withholding CLND in patients with very small SLN tumor deposits. However, the true tumor positivity rate in CLND specimens is likely to be an underestimation as these specimens usually contain several lymph nodes, which are usually examined with 1 or 2 hematoxylin-eosin-stained sections per node. Deeper levels and immunohistochemistry for melanocytic markers are generally not used unless there are concerning features on initial evaluation.47 Until there is conclusive evidence from currently accruing prospective randomized clinical trials (e.g., the second Multicenter Selective Lymphadenectomy Trial [MSLT-2]) of a lack of clinical benefit for CLND in patients with (truly) very small SLN tumor deposits, it seems imprudent to withhold CLND in this setting.

In summary, we have shown that key histologic features of metastatic melanoma deposits in SLNs are upstaged with increasing extent of examination of SLNs, and patients cannot be considered to have tumor burden <0.1 mm until multiple sections are examined. More importantly, we have shown that very small (<0.1 mm) deposits of melanoma in SLNs may be associated with adverse clinical outcomes (including death) and that this could be due, at least in part, to the underestimation of SLN tumor burden in the initial sections. Our results do not support the proposal that patients with very small melanoma deposits in SLNs should be considered to be SLN-negative and managed on this basis.

Acknowledgment

We acknowledge the support of the Cancer Institute New South Wales, the Australian National Health and Medical Research Council, and colleagues at Melanoma Institute Australia and the Department of Tissue Pathology and Diagnostic Oncology, Royal Prince Alfred Hospital.

Disclosure

Professor Scolyer is a Cancer Institute New South Wales ClinicalResearch Fellow

Copyright information

© Society of Surgical Oncology 2012