Correspondence The Royal Marsden Hospital, Fulham Road, London SW3 6JJ, UK

Email meirion@roseway.demon.co.uk

Introduction

All positive sentinel lymph nodes are assumed to inevitably progress to palpable nodal recurrence if not removed in the sentinel lymph node biopsy (SLNB) procedure. It is considered permissible to compare the survival of patients with positive sentinel nodes who undergo early lymphadenectomy with patients who have delayed lymphadenectomy for palpable regional node metastasis, providing that survival is calculated from the date of wide local excision (WLE) of the primary tumor.^{1, 2, 3, 4} The opinion prevails that SLNB should be considered the 'standard of care' as a staging procedure in the management of melanoma,⁵ and should be used routinely in patients at risk of nodal recurrence, even in the absence of a survival advantage.⁶ The suggestion has even been made that "...to deny the patient the opportunity of having a successful SLNB procedure may even have medico-legal consequences".⁷ Indeed, it has been proven beyond reasonable doubt that sentinel node status is the most important prognostic factor for recurrence and survival when compared with other prognostic factors individually.^{4, 8} Algorithms of other prognostic factors might be almost as powerful prognostic indicators as sentinel node status.⁹

The SLNB procedure was established before it was understood that ultrasonography was the best method of imaging lymph-node basins.¹⁰ Deposits of melanoma as small as 4 mm in the groin and neck, and 4.5 mm in the axilla, can be identified according to morphological criteria and the diagnosis proven by ultrasound-guided biopsy. This approach identifies 24% of patients who are eventually proven to have positive sentinel nodes.¹¹ Deposits of this size represent occult clinical disease, but are still defined as micrometastases and classified as N1a or N2a according to the most recent American Joint Committee on Cancer staging system.¹² For this reason the at-risk nodal basins should be screened by ultrasonography before proceeding to SLNB. Sentinel node status has, however, never been shown to have prognostic importance in patients in whom ultrasound is negative. The SLNB procedure might, in part, be patient-driven and it is understandable that patients require the most accurate prognostic information at the time of diagnosis of the primary tumor. Notably, however, up to 14% of patients with negative sentinel nodes will develop tumor recurrence by 3 years.^{8, 13, 14} Furthermore, there are concerns that some deposits of melanoma within the sentinel nodes are destined for dormancy or destruction and are not prognostically relevant. This phenomenon is referred to as prognostic false-positivity and is the subject of this Review.

Indirect evidence for prognostic false-positivity comes from several sources. First, many authors suggest or conclude that if a tumor is below a certain size, a melanoma micrometastasis in a sentinel node is of no adverse prognostic importance, but there is no agreement about the cut-off point.^{15, 16, 17, 18, 19} For example, Spanknebel and coauthors¹⁸ concluded that the prognosis for patients with micrometastases that can be detected only by immunohistochemistry is similar to that of patients with negative sentinel nodes. Similarly, van Akkooi et al.¹⁹ state that patients with micrometastases smaller than 0.1 mm diameter in the sentinel nodes should be judged to have negative sentinel nodes. These observations, however, depend on the sentinel nodes being removed for histology, which could be construed as contributing to any therapeutic benefit.

Secondly, there is the paradox that younger patients with melanoma have a significantly better prognosis than older patients despite having a higher incidence of positive sentinel nodes,^{20, 21} a situation that has been described as "intriguing" and "thought provoking".²² On all continents, the mortality from melanoma rises steeply with advancing age.²³ Despite this observation, Chao et al.²¹ reported in a study of 3,076 patients that the incidence of positive sentinel nodes declines with increasing decades of age, being 23.1% in patients younger than 30 years and only 12% in patients aged between 61 and 70 years (P <0.001). Statius Muller et al.²⁰ found that the incidence of positive sentinel nodes was 37% among patients aged 18–30 years and 17% among patients aged 71–84 years. How can these observations be explained? In the absence of evidence that melanoma spreads more commonly by the bloodstream in older patients, Chao and coauthors suggest that younger patients may have a generally more competent immune system that eliminates small micrometastases before they manifest clinically.²¹ Despite this paradox, two groups have recommended that the increased incidence of positive sentinel nodes in young patients (those under 35 years) should be one of the prognostic factors that justifies routine performance of SLNB in young people²⁴ even when the tumor thickness is less than 1 mm.²⁵

Thirdly, before the publication of the results of the Multicenter Selective Lymphadenectomy Trial (MSLT-1), two retrospective studies indicated that early lymphadenectomy in patients with positive sentinel nodes would result in a survival advantage.^{1, 2} Both studies compared the survival of patients who had positive sentinel nodes and who underwent early lymphadenectomy with that of patients who had delayed lymphadenectomy for palpable nodal recurrence. Kretschmer et al.¹ claimed a 12% survival advantage at 5 years whereas Morton et al.,² who did a matched pair analysis, claimed a survival advantage of 22%, 32% and 37% at 5, 10 and 15 years, respectively, in favor of early lymphadenectomy. Prognostic false-positivity in the patients with positive sentinel nodes is the most likely explanation of why these results differ from those reported in MSLT-1.

Analysis of the data

Direct evidence of prognostic false-positivity comes from the results of MSLT-1.⁴ This trial entered 2,001 patients who were randomized (40% versus 60%) either to WLE of the primary tumor with delayed lymphadenectomy if and when metastatic regional nodes became palpable (the observation arm) or to WLE and SLNB with immediate (early) lymphadenectomy if the sentinel node or nodes contained metastatic melanoma (the biopsy arm). The published results⁴ refer only to 1,269 patients with intermediate-thickness tumors (1.2–3.5 mm; Figure 1). There was no difference in melanoma-specific survival at 5 years (87% versus 86%) and only a small difference of 5.2% in disease-free survival (DFS) at 5 years, in favor of early lymphadenectomy. No survival or other data are available on the 732 patients with tumors that were less than 1.2 mm or more than 3.5 mm thick, who were entered and randomized into MSLT-1 but excluded from this publication.

Figure 1 A flow diagram of patients with intermediate thickness melanoma entered and randomized to MSLT-1 with a mathematical explanation of how prognostic false-positivity was calculated.

Abbreviations: MSLT-1, Multicenter Selective Lymphadenectomy Trial; SLNB, sentinel lymph node biopsy; SN, sentinel node; WLE, wide local excision. Based on data from Morton DL et al. (2006) N Engl J Med 355: 1307–1317.

Full figure and legend (23K)Figures & Tables indexDownload Power Point slide (227K)

The evidence to support prognostic false-positivity comes from a subgroup analysis in MSLT-1 and was reported prominently as one of that study's two conclusions. The authors compared the survival of the 122 patients with positive sentinel nodes in the biopsy arm who underwent early lymphadenectomy with that of the 78 patients in the observation arm who required delayed lymphadenectomy and claimed a 20% survival advantage at 5 years in favor of early lymphadenectomy (Figure 1). A definitive survival comparison between subgroups selected after randomization is not valid. Within a randomized, controlled trial, a survival difference can only be calculated from the point of randomization on an intention-to-treat basis, either for all patients entered and randomized or for patients identified and stratified before randomization. Furthermore, the 122 to 78 survival comparison might be biologically invalid because it assumes that all positive sentinel nodes will inevitably progress to palpable nodal recurrence. Crucially, the authors of this study failed to consider the 26 false-negative patients, who developed palpable nodal recurrence during follow-up and were treated by delayed lymphadenectomy, and who had the same prognosis as the 78 patients in the observation arm who were treated similarly. The most likely explanation is that these patients also had positive sentinel nodes but were not identified by the SLNB procedure.

This information allows us to calculate the approximate incidence of prognostic false-positivity within the sentinel nodes. We know that there was no survival difference in the MSLT-1 study as calculated from the point of randomization and it is accepted that palpable nodal recurrence in melanoma is a surrogate for poor survival. Therefore, it follows that the same proportion of patients in the observation and biopsy arms of MSLT-1 would have been expected to progress to palpable nodal recurrence. As shown in Figure 1, 15.6% of patients in the observation arm progressed to palpable nodal recurrence and, therefore, only 119 patients in the biopsy arm would have been expected to progress in a similar fashion, 26 of whom have already been identified. Therefore, only 93 of the 122 patients with positive sentinel nodes would have been expected to progress to palpable nodal recurrence, giving an incidence of prognostic false-positivity of approximately 24%. In other words, 24% of patients that had positive sentinel nodes and intermediate thickness tumors in MSLT-1 were wrongly up-staged, were given incorrect prognostic information and underwent unnecessary completion lymphadenectomy and possibly unnecessary adjuvant therapy. This calculation indicates that the 20% survival advantage claimed in the 122 to 78 comparison is due to a prognostic difference in the two subgroups compared and is not the result of any therapeutic benefit from early lymphadenectomy. Similarly, for all strata,³ the incidence of prognostic false-positivity is 34% (Figure 2), partly because the rate of false-negativity almost doubles. These simple mathematical calculations provide compelling evidence that prognostic false-positivity is a reality.

Figure 2 A flow diagram of all patients entered and randomized into MSLT-1 with a mathematical explanation of how prognostic false-positivity was calculated.

Abbreviations: MSLT-1, Multicenter Selective Lymphadenectomy Trial; SLNB, sentinel lymph node biopsy; SN, sentinel node; WLE, wide local excision. Based on data from Morton DL et al. (2005) Ann Surg 242: 302–311.

Full figure and legend (22K)Figures & Tables indexDownload Power Point slide (226K)

The incidence of prognostic false-positivity will decrease as more patients in the observation arm develop nodal recurrence. It is unlikely that nodal recurrence will increase sufficiently to affect substantially the calculated incidences of prognostic false-positivity, because the median time to nodal recurrence in the observation arm of MSLT-1 was 16 months, and the median follow-up was 60 months. Nevertheless, if it is assumed that nodal recurrences in the observation arm and among false-negative patients in the biopsy arm will increase at the same rate, then 10% and 20% increases, respectively, in palpable nodal recurrence among patients with intermediate-thickness tumors would reduce the incidence of prognostic false-positivity to 13% and 8.4%, respectively. A 30% increase in nodal recurrence would be required to reduce the incidence to less than 1%. For all strata, 10% and 20% increases in nodal recurrence would reduce the incidence of prognostic false-positivity to 27% and 20%, respectively, and it would take a 50% increase in nodal recurrence to reduce prognostic false positivity to less than 1%.

Discussion of the data from MSLT-1

Pivotal statistical calculations that claim a survival advantage for early lymphadenectomy in patients with positive sentinel nodes assume that all positive sentinel nodes, if not removed, will progress to palpable nodal recurrence. There is, however, no evidence to support this assumption. Morton et al.²⁶ have predicted that eventually the number of patients in the observation arm of MSLT-1 who developed nodal recurrence will equal the number of patients with positive sentinel nodes in the biopsy arm, thereby disproving the hypothesis of prognostic false-positivity. Confirmation of this prediction will not, however, be available for many years. In the meantime, the evidence available for prognostic false-positivity must be considered carefully because it has implications for patient management. Many publications have reported that tiny deposits of melanoma in the sentinel nodes are of no prognostic importance^{15, 16, 17, 18, 19} but there is great reluctance to extrapolate this observation to the hypothesis that these deposits are destined for destruction.

The development of a clinically relevant metastasis requires a complex series of biological events and there is no reason to assume that all melanoma cells detected in the sentinel nodes will acquire the necessary phenotype to complete the metastatic cascade and progress into such entities.²⁷ Proliferative activity and angiogenesis are important in the establishment of lymph-node metastases. Barnhill et al.²⁸ showed that melanoma micrometastases lacked high rates of proliferation, apoptosis and neovasculature compared with larger tumor deposits. These parameters are hallmarks of rapid tumor growth and their absence supports the idea that these tumors are dormant. It has been suggested that tumor-derived angiogenic or lymphangiogenic factors might be suppressed, possibly by host immune responses, so that micrometastases are limited to obtaining nutrients by diffusion, and that this leads either to dormancy²⁹ or to death and metastatic inefficiency. The proliferation marker Ki-67 has been associated with tumor progression in primary melanoma but has not been extensively studied in metastases. Ki-67 is an antigen that is known to be present at all phases of the cell cycle except G0 or G1.³⁰ The activity of Ki-67 can be assessed on paraffin-preserved specimens, and cells displaying Ki-67 immunoreactivity are considered to be replicating, thereby giving an objective technique for estimating the proliferative activity of cancer cells. This might form the basis of a risk analysis for patients with positive sentinel nodes.³¹

Conventional analysis of the results of MSLT-1 shows there is little evidence that patients with positive sentinel nodes benefit from early lymphadenectomy. The difference in DFS of 5.2% at 5 years in favor of the biopsy arm means that 100 patients have to undergo SLNB for around five patients in the biopsy arm to have a DFS advantage. In the absence of any difference in melanoma-specific survival, this could simply mean delay rather than prevention of metastasis. This is a small advantage with which to compensate patients for the additional surgery involved and is of no real palliative benefit. In several clinical situations, such as in trials of adjuvant chemotherapy, a prolongation of DFS can be of palliative advantage if the onset of troublesome symptoms from metastatic disease can be delayed, but there is no similar argument for SLNB in melanoma. The operations of immediate lymphadenectomy in patients with positive sentinel nodes and delayed lymphadenectomy for palpable nodal recurrence are identical. In the context of MSLT-1 there is no evidence that the outcome is different according to the timing of the operations. The analysis presented here, however, suggests that a proportion of patients with positive sentinel nodes in MSLT-1 were prognostically false-positive and, therefore, underwent unnecessary immediate lymphadenectomy. The possibility of prognostic false-positivity within the sentinel node can no longer be ignored and should be explained to patients during informed consent. The incidence of prognostic false-positivity is likely to be higher in MSLT-2 than in MSLT-1 because sentinel node status can be determined by reverse transcription-polymerase chain reaction criteria alone. The presence of benign nevus inclusions in the sentinel nodes represents an important potential source of false-positivity in sentinel node assays based on nonmorphological methods.³²

Conclusions

I do not advise SLNB for patients with melanoma. Relevant regional node basins are screened by ultrasonography at the time of diagnosis. If no suspicious lymph nodes are found and after WLE of the primary tumor, patients embark on an ultrasound-based surveillance program of the regional lymph nodes with the intention of performing delayed lymphadenectomy when nonpalpable, clinically occult nodal recurrence is detected sonographically and proven by ultrasound-guided cytology. Extrapolating from the results of MSLT-1, there is no risk that patients are being disadvantaged by this method of management. On the contrary, patients whose sentinel node status is prognostically false-positive are protected from unnecessary lymphadenectomy because such disease will not progress to occult or palpable nodal recurrence.

References

1. Kretschmer L et al. (2004) Patients with lymphatic metastasis of cutaneous melanoma benefit from lymphonodectomy and early excision of their nodal disease. Eur J Cancer 40: 212–218 | Article | PubMed | ChemPort |
2. Morton DL et al. (2003) Lymphatic mapping and sentinel lymphadenectomy for early-stage melanoma. Ann Surg 238: 538–550 | PubMed |
3. Morton DL et al. (2005) Sentinel node biopsy for early-stage melanoma. Ann Surg 242: 302–311 | PubMed | ISI |
4. Morton DL et al. (2006) Sentinel node biopsy or nodal observation in melanoma? N Engl J Med 355: 1307–1317 | Article | PubMed | ChemPort |
5. Balch CM and Cascinelli N (2006) Sentinel-node biopsy in melanoma. N Engl J Med 355: 1370–1371 | Article | PubMed | ChemPort |
6. Gershenwald JE and Ross IM (2007) Is sentinel-node biopsy superior to nodal observation in melanoma. Nat Clin Pract Oncol 4: 278–279 | Article | PubMed |
7. Thompson JF et al. (2007) Melanoma management in 2007. Aust Fam Physician 36: 487–488 | PubMed |
8. Gershenwald JE et al. (1999) Multi-institutional melanoma lymphatic mapping experience: the prognostic value of sentinel lymph node status in 612 stage I or II melanoma patients. J Clin Oncol 17: 976–983 | PubMed | ISI | ChemPort |
9. Kruper L (2005) Predicting sentinel lymph node status in stage I/II melanoma. J Clin Oncol 23: 710s
10. Bafounta M-L (2004) Ultrasound or palpation for detecting melanoma nodal invasion: a meta-analysis. Lancet Oncol 5: 673–680 | Article | PubMed |
11. Thompson JF and Shaw HM (2006) Benefits of sentinel node biopsy for melanoma. ANZ J Surg 76: 100–103 | Article | PubMed | ChemPort |
12. Balch CM (2001) Prognostic factor analysis of 17,600 melanoma patients. Validation of the American Joint Committee on Cancer Melanoma Staging System. J Clin Oncol 19: 3622–3634 | PubMed | ISI | ChemPort |
13. Berk DR (2005) Sentinel lymph node biopsy for cutaneous melanoma: the Stanford experience, 1997-2004. Arch Dermatol 141: 1016–1022 | Article | PubMed |
14. Vuylsteke RJ (2003) Clinical outcome of Stage I/II melanoma patients after selective sentinel lymph node dissection: long-term follow-up results. J Clin Oncol 21: 1057–1065 | Article | PubMed | ChemPort |
15. Ranieri JM et al. (2002) Importance of lymph node tumour burden in melanoma patients staged by sentinel node biopsy. Ann Surg Oncol 9: 975–981 | Article | PubMed | ISI |
16. Carlson GW et al. (2003) The amount of metastatic melanoma in a sentinel lymph node. Does it have prognostic significance? Ann Surg Oncol 10: 575–581 | Article | PubMed | ISI |
17. Starz H et al. (2005) Sentinel lymphonodectomy and S-classification: a successful strategy for better prediction and improvement of outcome of melanoma. Ann Surg Oncol 11 (3 Suppl): 162S–168S
18. Spanknebel K et al. (2005) Characterisation of micrometastatic disease in melanoma sentinel lymph nodes by enhanced pathology: recommendations for standardizing pathological analysis. Am J Surg Pathol 29: 412–414 | Article | PubMed |
19. van Akkooi ACJ et al. (2006) Clinical relevance of melanoma micrometastases (<0.1 mm) in sentinel nodes: are these nodes to be considered negative? Ann Oncol 17: 1578–1585 | Article | PubMed | ChemPort |
20. Statius Muller MG et al. (2001) The sentinel lymph node status is an important factor for predicting clinical outcome in patients with Stage I or II cutaneous melanoma. Cancer 91: 2401–2408 | Article | PubMed | ChemPort |
21. Chao C et al. (2004) Correlation between prognostic factors and increasing age in melanoma. Ann Surg Oncol 11: 259–264 | Article | PubMed |
22. Carlson GW (2004) Age and the incidence of sentinel lymph node metastases in melanoma. Ann Surg Oncol 11: 236–237 | Article | PubMed |
23. Giblin A-V and Thomas JM (2007) Incidence, mortality and survival in cutaneous melanoma. J Plast Reconstr Aesthet Surg 60: 32–40 | Article | PubMed |
24. Sondak VK et al. (2004) Mitotic rate and younger age are predictors of sentinel lymph node positivity: lessons learned from the generation of a probabilistic mode. Ann Surg Oncol 11: 247–258 | Article | PubMed | ISI |
25. Thompson JF and Shaw HM (2004) Should tumour mitotic rate and patient age, as well as tumour thickness, be used to select melanoma patients for sentinel node biopsy? Ann Surg Oncol 11: 233–235 | Article | PubMed |
26. Morton DL et al. (2007) Sentinel node biopsy in melanoma. N Engl J Med 356: 419–421 | ChemPort |
27. Fidler IJ and Ellis LM (1994) The implications of angiogenesis to the biology and treatment of cancer metastasis. Cell 79: 185–188 | Article | PubMed | ISI | ChemPort |
28. Barnhill RL (2001) The biology of melanoma micrometastases. Recent Results Cancer Res 158: 3–13 | PubMed | ChemPort |
29. Malafa MP et al. (2002) Inhibition of angiogenesis and promotion of melanoma dormancy by vitamin E succinate. Ann Surg Oncol 9: 1023–1032 | PubMed | ISI |
30. Gerdes J et al. (1984) Cell cycle analysis of a cell proliferation-associated human nuclear antigen defined by the monoclonal antibody Ki-67. J Immunol 133: 1710–1715 | PubMed | ISI | ChemPort |
31. Pearl RA et al. (2007) Ki-67 expression in melanoma. A potential method for risk assessment for patients with positive sentinel nodes. J Exp Clin Cancer Res 26: 109–115 | PubMed | ChemPort |
32. Abrahamsen H (2004) Sentinel lymph nodes in melanoma. Extended histopathological evaluation improves diagnostic precision. Cancer 100: 1683–1691 | Article | PubMed |

Contact the journal about this article or read the Article Responses associated with this article.

Subject areas under which this article appears: Surgical Oncology