Jean Jacques-Grob, MD
Professor of Dermatology
Aix Marseille University
Service de Dermatologie
Sentinel lymph node biopsy (SLNB) has been a source of debate in melanoma ever since Donald Morton introduced the technique in the early 1990’s. Some continue to consider it a costly, invasive technique that is unjustifiable since it has shown no survival benefit. This attitude has softened in the past 10 years, since hundreds of publications and studies have confirmed the prognostic value of SLNB. Its global use in staging, therapy selection, and trial inclusion criteria has strengthened its place de facto in clinical practice. SLNB proponents accurately consider it the best-ever staging method for melanoma, and argue that SLNB limits the use of radical/complete node dissection (CLND) to patients who need it, although no one truly knows who needs it until nodal disease becomes clinically detectable.
Hopefully, the day could soon be coming when SLNB fades into the sunset, as we keep improving our analytic ability to find molecular and immunological information establishing stage and prognosis in blood tests and the primary tumor itself. Given the negative primary endpoint results from the recent publication of the large Multicenter Selective Lymphadenectomy Trial-I (MSLT-I) by Morton, et al in the New England Journal of Medicine,1 and new research calling into question the survival benefits of CLND in patients with positive sentinel nodes,2 routine use of this controversial procedure could be replaced by other strategies.
The Pivotal Results of MSLT-1
The results of MSLT-I were long awaited to answer the many questions about SLNB and determine whether it should continue to be used. In the trial, 1,661 patients were randomized either to the SLNB arm (with SLNB followed by immediate CLND for a positive biopsy) or to the observation arm, with simple surveillance of the patient (“watch and wait”) and CLND only in the eventuality of clinically detectable nodal recurrence. The primary endpoint was overall survival, with secondary endpoints of recurrence-free or disease-free survival (RFS or DFS), survival with tumor-positive or tumor-negative nodes, and incidence of sentinel node metastases versus clinically detectable nodes. Unfortunately, a close look at the results shows that the trial falls short in answering several major questions about SLNB.
SLNB as a Marker
To be of utility, a marker has to be correlated to prognosis. It also has to show a good positive and negative predictive value for individual selection of high-risk patients.
Is SLNB of prognostic value?
Definitively YES. This has been established in many papers and was confirmed by MSLT-I. It is however only a statistical marker, with rather low individual predictive capability. Indeed, 62 percent of patients with intermediate thickness primaries and positive SLNBs, and 48 percent of thick melanoma patients with positive SLNBs, are alive at 10 years. More important, negative SLNBs also have a low individual predictive value, since 15 percent of intermediate thickness patients with negative nodes and 35 percent of thick melanoma patients with negative nodes will nonetheless die from melanoma within 10 years.
Is a positive SLNB a sound way to identify patients at risk of extensive nodal disease and/or disseminated metastatic disease?
Positive SLNBs may overestimate metastases, producing false positive markers for metastatic progression. There is no way of knowing exactly which micrometastases in the nodes will sooner or later become true nodal metastases. It is dubious that ALL micrometastases will turn into macro-metastases; some will simply disappear under immune system pressure. The authors of MSLT-I argue that since cumulative rates of nodal involvement were ultimately comparable in the two arms of the trial (immediate SLNB vs. delayed CLND), this proves that all tumor micro-deposits in the nodes will sooner or later become true nodal metastases. This is a post-hoc deduction for which the trial was not designed. Comparing cumulative rates of nodal involvement in the two arms of MSLT-1 may be much more difficult to interpret than the authors think, given the possibility of nodal metastases outside the area explored by SLNB, and the fact that time could have a great impact on the respective rate of nodal involvement in the two arms. Which arm patients were in influences the date of detection of nodal disease and the date and probability of nodal recurrence versus extra-nodal recurrence.
A negative SLNB does not exclude metastatic progression. We are not speaking here of the technically false negative for SLNB (where a positive SLNB is simply missed due to technical or anatomical reasons). We are speaking about a truly negative SLNB in patients who will nonetheless develop metastatic disease. One reason this can occur is that in patients with negative SLNBs, nodal micrometastases outside the sentinel node might develop into macro-metastases. Second, the sentinel node concept is an anatomical criterion, just like tumor thickness or ulceration. As such, there can be delay before biological aggressiveness can have clear anatomical expression. In other words, a biologically aggressive primary melanoma detected very early (a situation that increasingly occurs) might have a low Breslow thickness not typically calling for SLNB, or might express so little nodal tumor deposit that it results in a negative SLNB, even if it already has a high potential to kill. Early detection of melanoma before its aggressiveness can be picked up by anatomical criteria such as SLNB may explain why more and more patients who ultimately die from melanoma are coming from AJCC Stage IIA and IIB groups, and not from Stage III (positive SLNB).3
Although statistically correlated to prognosis, SLNB status is not a reliable marker to predict the course of an individual patient’s disease.
The Issue of Survival Benefits
SLNB was adopted by the medical community based on the hypothesis that it was not only a marker, but a marker that could directly or indirectly benefit the patient by improving survival or at least quality of life. The answers to a few simple questions tell us whether SLNB truly offers such benefits.
Does SLNB improve surgical management of melanoma in a way that offers an overall survival (OS) benefit? More specifically, does early intervention (SLNB + CLND for a positive sentinel node) offer a melanoma-specific survival benefit?
Even the trial authors acknowledge that MSLT-1 did not demonstrate clear utility in improving overall survival, which happened to be their primary endpoint. When comparing survival between patients who have CLND after positive SLNBs are found against patients who have therapeutic LND only after clinically detectable nodal metastases are found, there is no significant advantage for the SLNB arm in 10-year melanoma-specific survival. Nonetheless, attempting to show some 10-year-survival advantage, the authors of MSLT-1 found it only among those with intermediate tumor thickness – 62 percent 10-year survival in the SLNB arm vs. 41 percent in the observation arm. The advantage did not hold true for patients with thick melanomas. This survival comparison by subgroups is post-hoc, not randomized. Comparability would suppose that all patients from the SLNB arm with a positive SLNB would eventually have had invasive, palpable nodal disease if they had been in the observation arm. But this is uncertain.
Does CLND add a survival benefit to SLNB alone?
Another problem with MSLT-1 from the start was taking for granted that any patient with a positive sentinel node would definitely benefit from CLND. The utility of CLND after SLNB, however, is still an open question. To test this hypothesis properly, after finding positive sentinel nodes in patients, researchers would have to perform CLND on one group of patients while leaving the rest of the nodal basin intact in a comparable control group until the nodes became palpable, and then observe which group lived longer. The results of a recent German trial reported at the 2015 American Society of Clinical Oncology annual meeting argue against the survival value of CLND after a positive SLNB. With 483 patients studied and a median follow-up of 35 months, patients randomly assigned to undergo or not undergo CLND after a positive SLNB were statistically indistinguishable with respect to distant metastasis-free survival, recurrence-free survival, and melanoma-specific survival.2
On the heels of this small study, we now await the results of Morton, et al’s far longer, larger follow-up study to MSLT-1 – MSLT-II4 – which is ongoing with nearly 2,000 subjects. It will hopefully provide some definitive answers. Similar in design to the German study, this Phase III trial is examining the survival value of CLND vs. SLNB alone in patients found to have SLN metastases. The primary outcome measure will be melanoma-specific survival, and secondary outcome measures will be disease-free survival (DFS) and recurrence over 10 years of follow-up.
Does SLNB offer any quality-of-life benefit via increased DFS?
This question is relevant in itself, though it is uncertain how much a patient will value delayed recurrence at the price of an additional initial nodal procedure when there is no OS benefit. The DFS benefit reported in MSLT-1 is artificial and ambiguous in the context of this trial, since DFS means different things in the two arms. In the SLNB (+ CLND if positive) arm, DFS mainly represents time before any recurrence after the early nodal surgery, while in the observation arm DFS mainly represents time to first clinical expression of nodal disease. This is in no way comparable. In other words, when the authors describe a DFS benefit in favor of the immediate SLNB arm, it actually means that any distant or nodal recurrence in a patient who had an early nodal surgery usually occurs later than the first nodal surgery in a patient who never previously had nodal surgery. This supposed advantage probably means nothing clinically relevant – and certainly nothing that translates to increased overall survival.
As a selection criterion for adjuvant therapy, does knowing SLNB status improve medical management in a way that offers an OS benefit?
SLNB began being used long ago to select melanoma candidates for therapy with interferons, which proved to have a limited adjuvant benefit. Support for SLNB has taken on a new dimension in recent years with the development of effective targeted therapy and immunotherapy in advanced metastatic melanoma: many researchers believe that the successful FDA-approved checkpoint blockade therapies (anti-CTLA-4 and anti-PD-1) for stage IV patients might save even more lives if used earlier as adjuvant therapies. The prognostic capabilities of SLNB make it a potentially useful tool in indicating these treatments at earlier stages, for two main reasons: 1) It can target the highest-risk patients, and 2) it can limit useless toxicity in patients at low risk.
However, use of SLNB in selecting candidates for new adjuvant therapies is relevant only if most patients who will eventually develop systemic metastases are initially detected by a positive SLNB; and as we have noted, this is far from a given, since more and more patients who die from melanoma are coming from AJCC Stage IIA and IIB groups3 because of earlier diagnosis. Thus, even a new immune or targeted drug with a demonstrated strong adjuvant impact on OS in a population of SLNB-positive patients may not ultimately have an impact on melanoma mortality, since more and more of the future victims of multimetastatic melanoma may not be patients with a positive SLNB.
Could SLNB be not only useless but deleterious? Could it reduce survival?
There are obvious drawbacks to performing any invasive procedure that is ultimately of limited or no benefit. The question of reducing survival is not irrelevant, since tumor cells in the sentinel node might well present a pattern for priming the immune system against melanoma, and it conceivably could be damaging to destroy this pattern. However, to date, there is no argument to support this. In MSLT-I, no survival disadvantage was found in the SLNB arm, suggesting that any deleterious immune effect was negligible.
For all our arguments, SLNB is still probably inevitable for now, mainly because it is a pillar of the current universally accepted AJCC staging system required for all patient trials. But while SLNB is the best statistical prognostic marker for melanoma populations, this does not mean that using it as a marker is of any benefit or relevance to patients; it is still uncertain whether using it to assign adjuvant therapy with new potent drugs will impact on melanoma mortality. To better select patients for early adjuvant treatment, we need very early (biological) and reliable individual predictive markers, whereas SLNB is a rather late (anatomical), merely statistical marker with poor positive and negative predictive value. Our need for individual biological markers will become more and more crucial, since patients who ultimately die from melanoma are coming more and more from so-called “low-” or “intermediate-risk” populations, and less and less from “high-risk” groups; the absolute number of “low-risk” and “intermediate-risk” patients (AJCC stages IIA, IIB, and IIIA) is simply much higher than the number of patients with “high-risk” melanoma (AJCC stages IIIB and C).
In the end, SLNB is a costly and invasive intervention. In the next several years, if it does not show some definitive benefit, it will inevitably be replaced in staging by tumoral and/or circulating molecular markers, and/or patient immune profiling, which will provide more individual predictive information with less morbidity.
- Morton D, Thompson JF, Cochran A, et al. Final trial report of sentinel node biopsy or nodal observation in melanoma. New Eng J Med 2014; 370:599-609.
- London S. ASCO: German trial argues against complete nodal dissection for SLN-positive melanoma. OncologyPractice.com, May 31, 2015. http://www.oncologypractice.com/news/single-view/asco-german-trial-argues-against-complete-nodal-dissection-for-sln-positive-melanoma/a6bfdf9b16c9a96293bb6339697cbb99.html
- Methods for predicting risk of metastasis in cutaneous melanoma, WO 2014158696 A1, Inventors Cook RW, Maetzold D, Oeschlager K. Filing date 2 28 14. http://www.google.com/patents/WO2014158696A1?cl=en
- Multicenter Selective Lymphadenectomy Trial II (MSLT-II), ClinicalTrials.gov, https://clinicaltrials.gov/ct2/show/NCT00297895. Reviewed June 29, 2015.