The Melanoma Institute of Australia (MIA) previously developed a risk calculator for sentinel lymph node (SLN) metastasis risk to help clinicians and patients with primary cutaneous melanoma decide whether to proceed with a SLN biopsy (SLNB). A new study published in JAMA Dermatology provides a validation update on the tool based on a larger and more geographically diverse study population.
Not only were the results with a six-factor model using a larger population similar to those from the original dataset but also the calculator showed improved precision with narrowed 95% CIs.
Both the original study and the larger validation analysis compared the accuracy of the calculator with the SLNB results that were available for each patient. The full calculator requires the input of age at diagnosis, Breslow tumor thickness (mm) and melanoma subtype (acral, superficial spreading, nodular, pure desmoplastic, or lentigo maligna melanoma). Clinicians can also include tumor mitotic rate (x/mm2 or mitosis present/absent), ulceration (present/absent) and lymphovascular invasion (present/absent) when this information is available.
The new analysis included data from the National Danish Melanoma Database (N = 8533), three cancer centers in the United Kingdom (N = 2663), two in the United States (N = 1844), one in New Zealand (N = 449), one in Sweden (N = 1215), and one in Brazil (N = 1027). When pooled with the original cohort, 15,732 patients were included in the validation.
What Did the New Study Find?
A decision-curve analysis revealed the differences in clinical decision-making using the six-factor model or one using only Breslow thickness and ulceration. In this context, a given threshold probability reflects the minimum level of risk at which a clinician/patient would choose to proceed with an SLNB.
Using all six MIA parameters and a threshold of 8% resulted in a favorable balance of minimizing unnecessary biopsies, while maintaining confidence in detecting metastasis. This lower-threshold approach is better suited to patients with a more conservative risk tolerance, who are willing to undergo biopsy when they have a lower risk for metastasis rather than miss one.
In contrast, a simpler predictive model using only Breslow thickness and ulceration resulted in a higher net benefit at a 14% threshold for patients with higher risk tolerance. Net benefit refers to the reduced number of unnecessary biopsies (a positive) at the risk of missing some true cases of metastasis (a negative). This may be preferable for patients who want to limit biopsy to a high risk for metastasis. For these patients, avoiding biopsy is a greater concern than missing a true positive. This simpler approach may also be valuable in settings where more detailed information, such as subtype or mitotic rate, is unavailable.
“These findings reinforce the tool’s reliability in predicting the risk a melanoma has spread to the lymph nodes in diverse patient groups, providing clinicians worldwide with greater confidence in its use for everyday practice,” primary investigator Alexander Varey, MD, PhD, said in a press statement.
In addition, with the larger sample size the 95% CIs shrank with a mean reduction of more than 75%. Thresholds — the probability that reflects the minimum level of risk at which a clinician or patient would choose to proceed with an SLNB — of 5% and 10% are generally considered to be clinically relevant. At the 5% threshold, this increased precision shifted clinical interpretation in more than half of the patients (58%) — who previously had lower CI bounds below 5% — now had lower bounds greater than 5% with the inclusion of more data. In clinical terms, this means having greater confidence that a patient’s true risk exceeds the 5% threshold (which reflects a patient’s preference for a more conservative approach) improving the reliability of biopsy decisions in patients near that risk cutoff.
Similarly, among patients whose upper 95% confidence bounds previously exceeded 10%, roughly a quarter (24%) now had upper bounds that fell below 10%, which increases the certainty that these patients are not at high risk. In clinical terms this supports safer biopsy avoidance in that group.
What Makes the Tool Useful in Clinical Practice?
“This calculator is helpful because you can get all of this information just from the pathology report. You don’t have to do another assay or spend thousands of dollars on genetic profiling,” said Mark Faries, MD, a surgical oncologist at Cedars‑Sinai and The Angeles Clinic and Research Institute, Los Angeles. He is also the co-director of the Cutaneous Oncology Program at the Cedars-Sinai and heads surgical oncology at The Angeles Clinic.
The tool also adds easy-to-understand information for patients to improve the discussion of their care. “If a patient comes in newly diagnosed with a melanoma, you would be able to put these parameters into the calculator, and it’ll give you a number to suggest the risk that they have involvement of their lymph node”, said Faries.
“Based on that information, you — together with the patient — can decide whether or not they can just have an excision of the skin site, or if they also need to have a sentinel lymph, node biopsy done.”
He added, “you can use [this calculator] for every patient. Just having that number helps patients understand what they’re looking at. Many times, patients come in with a very pessimistic outlook, based on people’s general feeling about melanoma. So even if they have a fairly substantial risk of having something in the node that justifies doing the node biopsy, you can reassure them that actually the most likely outcome is that everything’s going to be okay. Even from that sort of peace of mind standpoint, it’s useful for every patient.”
What Does the New Study Add?
In this study, “they’ve collected a very large additional number of patients from other centers around the world and have further validated the results of the initial analysis…this new work makes the estimates more precise,” said Faries.
With the earlier version of the calculator, “there was a pretty wide range, where the probability might’ve reasonably fallen for a prediction. Now with these larger numbers [of patients] it’s a much smaller range. So you have more confidence that the number you’re getting is correct.,” he said.
Are There Any Caveats or Room for Improvement?
“There still is some room for improvement at the lowest risk end of the scale,” Faries noted. “The reason for that is that the calculator was developed based on patients who had a sentinel lymph node biopsy done. So they’ve already been selected to some extent. Relatively speaking, the number of patients they have at the very bottom end of the risk scale is not very large because those people don’t get the lymph biopsy done. So I think when we have somebody who we generally wouldn’t recommend doing a sentinel node biopsy for, the calculator is a little bit less definitive. I think there’s more work that could be done to help at that very bottom end.”
Varey reported receiving personal fees from Novartis AG and Merck & Co., Inc. (MSD).
Faries reported serving on the advisory boards of MSD; Bristol-Myers Squibb Company; Regeneron Pharmaceuticals, Inc.; and Replimune Group, Inc.