P42: Clinical Benefit of Sentinel Lymph Node Biopsy Nomograms for Head and Neck Melanoma

Thursday, March 23, 2023

3:36 PM – 3:41 PM

Location: SSO HUB: Melanoma & Sarcoma Zone

Poster Presenter(s)

Harrison M. Drebin, BA

Research Coordinator
Memorial Sloan Kettering Cancer Center, United States

Disclosure information not submitted.

Author(s)

Harrison M. Drebin, BA

Research Coordinator
Memorial Sloan Kettering Cancer Center, United States

Disclosure information not submitted.
NK

Nicholas R. Kurtansky, BS

Data Analyst
Dermatology Service, Department of Medicine, Memorial Sloan Kettering Cancer Center, United States

Disclosure information not submitted.
Jennifer R. Cracchiolo, MD

Assistant Professor, Head and Neck Service
Memorial Sloan Kettering Cancer Center
New York, New York, United States

Disclosure information not submitted.
MM

Michael A. Marchetti, MD

Assistant Attending
Dermatology Service, Department of Medicine, Memorial Sloan Kettering Cancer Center, United States

Disclosure information not submitted.
Edmund Bartlett, MD

Assistant Attending, Gastric and Mixed Tumor Service
Memorial Sloan Kettering Cancer Center
New York, New York, United States

Disclosure(s): Excite International Inc: Other (Terminated, January 3, 2022); SkylineDx: Research Grant (Ongoing)

Disclosure(s):

Edmund Bartlett, MD: Excite International Inc: Other (Terminated, January 3, 2022); SkylineDx: Research Grant (Ongoing)

Introduction: Selection of melanoma patients for sentinel lymph node biopsy (SLNB) is based on the risk of lymph node metastasis. This study aims to assess the performance and clinical benefit of three nomograms for patients with primary cutaneous head and neck melanoma.

Methods: Eligible patients from 2000 to 2021 at a single center were retrospectively identified from a prospectively maintained database. Predicted probability of SLNB positivity using the Melanoma Institute of Australia (MIA), Memorial Sloan Kettering (MSK), and Friedman nomograms were calculated from published nomogram functions. Missing values for histologic subtype, Clark level, and mitotic rate were imputed using multiple imputations using chained equations. Missing values for ulceration, desmoplastic subtype, lymphovascular invasion were assumed negative. Receiver operator characteristic (ROC) and calibration curves were generated. Decision curve analysis was used to compute the net benefit and net avoidable interventions for each model.

Results: 450 patients with head/neck melanoma underwent SLNB, of whom 72 (16.0%) had a positive sentinel lymph node. The area under the ROC curves (AUC) of the MIA, MSK, and Friedman models were 0.684 (95% CI 0.620-0.749), 0.648 (95% CI 0.580-0.716), and 0.625 (95% CI 0.560-0.690), respectively. For all models, calibration was best at predicted SLNB positivity rates below 20%, with the MSK model tending to overpredict risk (Fig 1a-c). Compared to SLNB for all, model-guided selection provided no clear net benefit at SLNB thresholds below 10% (Fig 1d). At the 10% threshold, use of the MIA or Friedman models provided added net benefit and would lead to a net reduction in the number of avoidable SLNBs (Fig 1d,e).

Conclusions: The MIA, MSK, and Friedman SLNB nomograms had comparable AUC among these patients with head and neck melanoma. None of the models clearly provided added net benefit compared to SLNB for all at low thresholds. At risk thresholds of 10% or greater, clinically corresponding to patients with stage T2a melanoma or greater, use of both the MIA and Friedman models provided added net benefit and would reduce the number of negative SLNBs. The MSK model had lower net benefit compared to other models and SLNB for all, perhaps due to miscalibration in this cohort.

Learning Objectives:

Understand decision curve analysis to compare prognostic models
Discuss the strengths and weaknesses of three published nomograms in patients with head and neck melanoma
Determine which nomograms they prefer to use to select patients with head and neck melanoma for sentinel lymph node biopsy