A comprehensive analysis of metastatic disease following surgery for clinically localized cutaneous melanoma

Abstract

Background: The National Comprehensive Cancer Network considers "baseline staging" (whole body computed tomography or positron emission tomography scans with or without brain magnetic resonance imaging scans) for all patients with asymptomatic melanoma who had a positive sentinel lymph node biopsy result. The true yield of these workups is unknown.

Methods: We created cohorts of adult patients with malignant melanoma using the National Cancer Database (2012-2020) to mimic 3 common scenarios: 1) clinically node-negative disease, with positive sentinel lymph node biopsy results; 2) clinically node-negative disease, with negative sentinel lymph node biopsy results; and 3) clinically node-positive disease, with confirmed lymph node metastases. Multivariable regression, supervised decision trees, and nomograms were constructed to assess the risk of metastases based on key features.

Results: In total, 10 371 patients were in scenario 1, 55 172 were in scenario 2, and 4012 were in scenario 3. The proportion of patients with any metastatic disease (brain metastases) were as follows: 1.4% (0.3%) in scenario 1, 0.3% (<0.1%) in scenario 2, and 11.6% (1.6%) in scenario 3. On multivariable regression, Breslow depth greater than 4, ulceration, and lymphovascular invasion were associated with greater risk of metastatic disease. A supervised decision tree for patients in scenarios 1 and 2 found that the only groups with more than 2% risk of metastases were groups with T4 tumors or T2/T3 tumors with ulceration and lymphovascular invasion. Most groups had a negligible risk (<0.1%) of brain metastases.

Conclusion: This study is the first large analysis to guide the use of imaging for cutaneous melanoma. Among patients with clinically node negative disease, metastatic disease is uncommon, and brain metastases are exceedingly rare. Further investigation could promote a tailored approach to metastatic workups guided by individual risk factors.