Risk Prediction Models for Sentinel Node Positivity in Melanoma: A Systematic Review and Meta-Analysis.

Area of Science:

• Oncology
• Medical Informatics
• Biostatistics

Background:

• Melanoma staging and treatment decisions often rely on sentinel lymph node biopsy (SLNB) positivity.
• Accurate risk prediction models are crucial for identifying patients who would benefit from SLNB.

Purpose of the Study:

• To systematically review and assess the performance of existing risk prediction models for melanoma SLNB positivity.
• To identify the best-performing models for clinical application.

Main Methods:

• A comprehensive literature search of Embase and MEDLINE databases was conducted up to May 1, 2024.
• Included studies developed or validated risk prediction models for melanoma SLNB positivity, with model discrimination as a key metric.
• Data extraction and synthesis followed established guidelines (DSReview, TRIPOD, PRISMA), with meta-analysis used for pooled estimates.

Main Results:

• Twenty-one distinct risk prediction models were identified, with 20 external validations for 8 models.
• The overall pooled weighted C-statistic for all models was 0.78, indicating moderate discriminative ability, though significant heterogeneity was noted.
• The Memorial Sloan Kettering Cancer Center and Melanoma Institute of Australia models demonstrated strong and comparable external validation performance (C-statistics of 0.73 and 0.70, respectively).
• Models incorporating gene expression profiles did not show significantly better discrimination than those using only clinicopathologic features.

Conclusions:

• Several externally validated risk prediction models for melanoma SLNB positivity exhibit strong discriminative performance.
• These validated models offer valuable tools for preprocedural risk assessment in clinical practice.
• Further research is warranted to evaluate the impact of implementing these models on patient care and outcomes.