Melanoma Risk Models Identify Sentinel Node Biopsy Candidates

Multiple risk prediction models appear to aid in the identification of patients with a melanoma metastasis risk warranting sentinel lymph node biopsy (SLNB), according to findings from a systematic review and meta-analysis published March 12 in JAMA Dermatology.

Researchers identified 23 articles describing 21 different risk prediction models, with 9 models judged to have sufficient information to help guide clinicians in SLNB referral, and the Memorial Sloan Kettering Cancer Center model having the best overall performance.

The study findings may also, "guide subsequent comparative studies to ensure that these models have adequate discrimination and calibration for recommendation as part of routine preoperative clinical assessment," wrote researcher Clair Temple-Oberle, MD, of the Arthur Child Comprehensive Cancer Centre, Calgary, Canada, and colleagues.

The researcher noted that pre procedural risk stratification is key in patients with "thin melanomas where the likelihood of nodal metastasis begins to approximate the procedural risks associated with SLNB."

"In these cases, risk nomograms have a key role to play in helping physicians determine which patients would most benefit from nodal sampling vs routine follow-up surveillance, especially when considering that SLNB itself does not improve melanoma-specific mortality," they wrote. "Consequently, the results of this study set the scene for subsequent testing and implementation of risk prediction models in clinical practice."

The analysis follows a 2024 study which showed significant differences in the ability of various risk prediction models to assess melanoma metastasis risk.

Temple-Oberle and colleagues wrote that in that study calculations diverged by more than 10% in some estimates for younger patients and those with thick melanomas.

They wrote that, "this is clinically relevant given that the benefits of SLNB are often considered to outweigh procedural risks (such as lymphedema, seroma, and local paresthesia) beyond a 5% to 10% likelihood of nodal spread, so the choice of which model is used may change whether a patient receives SLNB."

Their goal in conducting the systematic review and meta-analysis was to, "characterize the clinical populations in which existing SLNB risk prediction models have been derived, the variables included in these calculators, predictive performance, and the potential utility of each model in routine, pre-procedural evaluation of patients with melanoma."

Their systematic review identified 97 studies, which yielded 23 studies included in the meta-analysis based on inclusion criteria. A total of 21 risk prediction models were identified, with 15 unique predictor variables identified.

The researchers wrote that the most commonly identified variable was the Breslow depth measurement of melanoma thickness (15 studies), followed by ulceration (14 studies), patient age (12 studies), Clark level (8 studies), mitoses (8 studies) and anatomic location of the primary melanoma (7 studies).

The 23 articles described the development of 21 different risk prediction models for SLNB positivity, 20 external validations of 8 different risk prediction models, and 9 models that "included sufficient information to obtain individualized patient risk estimates in routine preprocedural clinical practice."

"Among all risk prediction models, the pooled weighted C statistic was 0.78 (95% CI, 0.74-0.81) with significant heterogeneity (I² = 97.4%) that was not explained in meta-regression," the researchers wrote.

Among the other main findings, the researchers wrote:

• "The Memorial Sloan Kettering Cancer Center (MSKCC) and Melanoma Institute of Australia (MIA) models were most frequently externally validated with both having strong and comparable discriminative performance" (pooled weighted C statistic, 0.73; 95% CI, 0.69-0.78 versus pooled weighted C statistic, 0.70; 95% CI, 0.66-0.74).
• "Discrimination was not significantly different between models that included gene expression profiles (pooled C statistic, 0.83; 95% CI, 0.76-0.90) and those that only used clinicopathologic features (pooled C statistic, 0.77; 95% CI, 0.73-0.81) (P =0.11).
• "There was significant heterogeneity in validation studies for both the MIA (I ² = 96.9%) and MSKCC models (I 2 = 94.5%) that was not resolved with meta-regression."

"The MSKCC and MIA models are widely utilized and have been the most extensively validated," Temple-Oberle et al. wrote. "We reaffirm that these models have strong and comparable performance in external validation cohorts despite numerous differences between the development cohorts of these models."

The study represents the first systematic review and meta-analysis of risk prediction models for SLNB positivity in primary cutaneous malignant melanoma.

Study limitations included the restricted assessment of model discrimination, which was largely limited to the C statistic, and significant heterogeneity between risk prediction models "that were not fully explained in meta-regression."

The researchers reported no funding source nor other disclosures related to this study.

Ma B, et al "Risk prediction models for sentinel node positivity in melanoma: a systematic review and meta-analysis" JAMA Dermatol 2025; DOI:10.1001/jamadermatol.2025.0113.