Making Melanoma Diagnosis and Treatment More Predictable

Melanoma can be a very humbling disease. Some of the most talented doctors have been challenged over the years when diagnosing and planning treatment for melanoma because of its variability. It can do anything. It can lay dormant for years and reactivate. It can spontaneously disappear. The one rule about melanoma has always been that there are no rules, which can be very frustrating. No physician wants to expect something to behave one way and then see it behave completely differently.

The landscape is changing, however. The time has come when we can collect information that can be used to provide a relatively reliable, precise prognosis. With the 31-GEP (gene expression profiling) test, we can determine with a very high level of accuracy how a patient’s melanoma will behave in terms of node positivity and recurrence—and we can treat accordingly.

As a surgical oncologist, I have been working with GEP testing for almost a decade, and it has been a labor of love. I have learned to be a better physician. I was proud to publish an important study earlier this year demonstrating the ability for the integrated 31-GEP test for sentinel lymph node biopsy (i31-SLNB) to reduce the number of unnecessary SLNB procedures in patients with cutaneous melanoma, and that research has been followed up more recently with additional data reinforcing the utility of this test. In this supplement, we will explain the results of the research and their impact on melanoma treatment capabilities.

This is still a challenging disease to diagnose and treat. However, we have taken a massive step toward making it more predictable and, thus, improving our abilities to provide the best possible personalized treatment for each patient. This is a great time to work in melanoma.

J. Michael Guenther, MD
Surgical Oncologist
St. Elizabeth Hospitals (Kentucky)

A prospective, multicenter analysis of the integrated 31-gene expression profile test for sentinel lymph node biopsy (i31-GEP for SLNB) test demonstrates reduced number of unnecessary SLNBs in patients with cutaneous melanoma

doi.org/10.1186/s12957-024-03640-x 

SUMMARY

Current National Comprehensive Cancer Network (NCCN) guidelines recommend foregoing sentinel lymph node biopsy (SLNB) when the likelihood of finding a positive sentinel lymph node (SLN) is less than 5%, discussing and considering SLNB when the likelihood is 5% to 10%, and offering an SLNB when the likelihood is above 10%.¹ However, 92% to 95% of patients with T1 tumors have a negative node and only 12% of SLNB procedures overall are positive.¹ Given the risk of complications from an SLNB as well as the cost, the authors posited that a tool to help clinicians select patients most likely to have a negative SLNB would improve patient care and decrease healthcare costs.¹

This prospective, multicenter study assessed the accuracy of the integrated 31-gene expression profile for SLNB (i31-SLNB) in predicting SLN positivity among patients with T1–T2 tumors, for whom SLNB guidance would be most impactful.¹ To determine if incorporating the i31-SLNB into decision-making resulted in fewer SLNB procedures performed, propensity score-matching was performed to compare SLNB utility rates for patients in the current study to those in a non-overlapping cohort for whom the 31-GEP was not used for SLNB decision-making.¹

The study, which included 262 patients with T1 tumors and 60 patients with T2 tumors—overall median Breslow thickness 0.8 mm) found that no patients with < 5% i31-SLNB predicted risk had a positive SLNB (0/35); that propensity matching demonstrated an 18.5% reduction in SLNB procedures performed (43.7% vs. 62.2%. p < 0.001); and that the i31-SLNB could have reduced the number of unnecessary biopsies by 25.0% (35/140). SLNB was performed in 140 patients, with a positivity rate of 6.4 (9/140).¹

DR. GUENTHER’S COMMENTARY

When I first saw, approximately 9 years ago, data indicating that 60% of recurrences were in people with negative nodes, that put a cold chill into me, because I had performed so many SLNB procedures over the years and always thought negative nodes meant we were in good shape. As it turns out, the node is an important part of this process, but it is not the only part, and the 31-GEP probably supersedes that. The two of them are additive. Nodal status is not a substitute for a gene expression, and neither is a gene expression a substitute for nodal status. They are complementary pieces of information. It was a misconception that the lymph node being clean meant the patient was off the hook. That is not the case at all.

The results of this study were basically what I expected. The 35 patients who said, “I don’t trust this and I want it done anyway” were an interesting part. Obviously, they all had negative sentinel node biopsies, and if we assume that these 35 patients had also not undergone the SLNB procedure, we have a substantial reduction; we went from an approximately 60% SLNB performance rate down to just under 30% SLNB performace rate. That is about a 50% relative reduction in the number of SLNB procedures being performed.

There was also a 2-year median follow-up in which no one recurred if they had a low-risk lesion by i31-SLNB. So, we were able to identify a population of patients who, if they had a node biopsy, they did not have a positive node, and if they did not have a node biopsy, they were very safe to not have a recurrence. This allows us to substantially reduce the number of patients having unnecessary SLNB, which was the primary goal of the study.

Eventually, we will have a recurrence there. We will not go zero-for-infinity. However, the reality is that we now have prospective data that should give people the confidence to start de-escalating surgery for patients. The recurrence rate is still another story, but the de-escalation of surgery and putting the lymph node in its position is the point we have reached.

From the surgical oncology point of view, another goal involves patients with shallow lesions who have positive nodes. Approximately 15% under 1 mm are metastasizing lesions.² Those, in a way, are the most important ones to find because that is how we, as surgeons, can potentially save lives.

We also want to look not just at the negative predictive value of this test but at the positive predictive value. We want to identify patients who are likely to have positive nodes that we might not otherwise have thought would. This is a very specific test, but it also has a very high negative predictive value. In my practice, I order this on almost everybody who has a lesion over 0.3 mm for whom it will change their treatment.

We are getting ready to move away from looking at the lymph node as the most important element of the process, because it is a staging procedure. It helps us decide who is high-risk and who is not. It helps us guide systemic therapies. Now, however, we can obtain a risk of recurrence that is individualized to every patient, whether the node is negative or the node is positive. That is really where the rubber hits the road for melanoma.

With the data we have collected over the past few years, we have been able to fill in the pieces of the mosaic of melanoma care, and we are starting to see that early diagnosis of recurrence translates to high salvage rates. My own experience has been that almost two-thirds of our patients who recur are salvaged by a combination of therapies, and the data are telling us that if these people are watched closely—clinical follow-up as well as cross-sectional imaging and, potentially, even blood work in the future—that we have a high salvage rate.

The recurrence risk of a patient is also a precious piece of information because it guides their follow-up. If they have low risk, which many of them do, they go back to the dermatologist for routine follow-up. They do not need to see surgical oncologists, and they do not need expensive testing, time away from work, copays, etc. The relief from that anxiety is palpable.

Meanwhile, patients who have higher-risk lesions stay with the surgical oncologist, and if we have a problem, we tend to find it early enough that it is salvageable. We have evidence of adjuvant therapy that works. We have evidence that close follow-up translates to high survival rates.

The key is matching each patient’s care to the risk that they have, which is a very efficient way to practice medicine, especially with melanoma.

Criticisms of many tests like the 31-GEP have included a lack of prospective data and the fact that they are not part of NCCN guidelines. Now, we have several of these studies with prospective data showing the test producing results. This should give doctors confidence that they can select patients who do not need SLNB. For surgeons, there may be reluctance based on restricting earnings, but we need to get far past that. We need to really look at what’s beneficial for patients. These things will work themselves out. For every SLNB that we do not do on someone who is low-risk, we can better focus on the higher-risk patients.

1. Guenther JM, Ward A, Martin BJ, et al. A prospective, multicenter analysis of the integrated 31-gene expression profile test for sentinel lymph node biopsy (i31-GEP for SLNB) test demonstrates reduced number of unnecessary SLNBs in patients with cutaneous melanoma. World J Surg Onc. 2025;23(1). doi:10.1186/s12957-024-03640-x.

2. Kalady MF, White RR, Johnson JL, Tyler DS, Seigler HF. Thin melanomas: predictive lethal characteristics from a 30-year clinical experience. Ann Surg. 2003 Oct;238(4):528-35; discussion 535-7. doi:10.1097/01.sla.0000090446.63327.40.

Comparing two gene expression profile tests to standard of care for identifying patients with cutaneous melanoma at low risk of sentinel lymph node positivity

doi.org/10.21873/cdp 

SUMMARY

Most patients undergoing sentinel lymph node biopsy (SLNB) have a negative result, indicating that reliance upon American Joint Committee on Cancer (AJCC) T-stage alone is less than ideal. Gene expression profile (GEP) tests have been developed to identify patients at low risk of node positivity who may consider avoiding SLNB. This study analyzed the accuracy of two tests—the CP-GEP and the 31-GEP— in identifying patients with <5% risk of SLN positivity across the five validation studies of the CP-GEP and four validation studies of the 31-GEP and integrated 31-GEP (i31-SLNB) in T1 and T2 tumors.³

The authors calculated the true-to-false negative (TN:FN) ratios for patients with T1 and T2 tumors included in five reported validation cohorts using the CP-GEP, one cohort using the 31-GEP, and three cohorts using the i31-SLNB. They compared the TN:FN ratio and false negative rates calculated for patients considered low-risk by the CP-GEP or 31-GEP/i31-SLNB tests to the 19:1 ratio and 5% false negative rate based on staging.³

The overall weighted-average performance of the CP-GEP test in these studies was an overall ratio of 15:1 (6.2% false-negative rate), which the authors noted is inferior to AJCC staging to rule out an SLNB.³

Meanwhile, the overall weighted-average performance of the 31-GEP and i31-SLNB tests in these studies was a ratio of 34:1 (2.8% false-negative rate), which the authors noted is superior to  AJCC staging.³

Chi-squared analysis comparing the false-negative rates of the tests found that the 31-GEP/i31-SLNB had a significantly lower false negative rate than CP-GEP (p=0.012).³

DR. GUENTHER’S COMMENTARY

Guidelilnes based on staging provide us with a 5% SLN positivity risk cutoff point to perform an SLNB. So, we really should not perform SLNB for nodes under 5%. We should consider it from 5% to 10%, and we should probably do it for higher than 10% predicted risk if the patient is medically capable. If 5% indicates a one-in-20 chance of having a positive node—a 19-to-one true-negative-to-true-positive ratio—then any worthwhile test should be better than pathology alone.

This research evaluated the two primary competing assays, looking at their predictive value for node positivity and identifying the accuracy of that test. If you pull together all the published studies, the CP-GEP gene assay did not fare very well. It was not able to identify low-risk patients; some came back as high-risk, and the ratio never reached 19-to-one. Conversely, the data supporting the 31-GEP test is far more robust, reproduced, and documented—with thousands of patients. There is a substantial difference in the predictive value of node positivity between the two tests. The other test failed to meet its own endpoints in most of these publications, and we have seen them suggest that perhaps 10% should be the cutoff. That assay did not turn out to be any better than pathology alone; in fact, it was inferior to pathology alone. It is just not there yet.

The study includes a graph showing that the 31-GEP had a 34-to-one true-negative-to-false-negative ratio, with a false-negative rate of 2.8%. The competing essay had a true-negative-to-false-negative ratio of 15-to-one, with a 6.2% false-negative rate. That is a dramatic difference. That 6.2% false negative rate is higher than what we see with pathology alone. 

Meanwhile, a 25-to-one ratio was the lowest we saw for any study with 31-GEP. That is better than pathology’s 19-to-one. So, every single manifestation of the 31-GEP performed better than pathology alone. That was the definition of a test that was worth its weight.

Moving forward, there are some interesting observations. One involves anecdotal evidence that some of these low-risk lesions may recur but still do extremely well. We have all had patients with metastatic disease who survived for 7 or 8 years without aggressive measures. We would like to see what some of these low-risk lesions are,  if they have favorable biology. That will be an interesting subset. Do higher-risk lesions predict a benefit from immunotherapy? That would go from prognostic to predictive, which will be where things get very interesting. These are expensive drugs with a lot of side effects, so if we can figure out who benefits from them and prove that, it could create so many benefits. I think that is coming.

There also will be more research like this study. Again, there is very little contradictory data for the 31-GEP. This assay has proven its worth. The mathematics involved are ingenious, and the refinement of it when we add the clinical pathologic factors was like creating the next generation of iPhones. We struck gold when we were able to get individualized risks instead of just pooled information. Instead of zip codes, we now get an exact address.

It is a fun time to work in this field. We have tools we never dreamed of, and they continue to get better. 

3.  Prieto PA, Ferris LK, Guenther JM. Comparing Two Gene Expression Profile Tests to Standard of Care for Identifying Patients With Cutaneous Melanoma at Low Risk of Sentinel Lymph Node Positivity. Cancer Diagn Progn. 2025;5(3):261-267. doi:10.21873/cdp.10438.

Sponsored by Castle