RESUMO
BACKGROUND: To improve patient selection for sentinel node (SN) biopsy, the Melanoma Institute of Australia (MIA) created a predictive model based on readily available clinicopathologic factors. OBJECTIVES: Validation of the MIA nomogram using the National Cancer Database (NCDB), a nationwide oncology outcomes database for >1500 Commission-accredited cancer programs in the United States. METHODS: A total of 60,165 patients were included in the validation. The probability of SN positivity was calculated for each patient. Using calculated probabilities, a receiver operating characteristic curve was generated to assess the model's discrimination ability. RESULTS: At baseline, the NCDB cohort had different clinicopathologic characteristics compared with the original MIA data set. Despite these differences, the MIA nomogram retained high-predictive accuracy within the NCDB dataset (C-statistic, 0.733 [95% CI, 0.726-0.739]), although calibration weakened for the highest risk decile. LIMITATIONS: The NCDB collects data from hospital registries accredited by the Commission on Cancer. CONCLUSIONS: In conclusion, this study validated the use of the MIA nomogram in a nationwide oncology outcomes database collected from >1500 Commission-accredited cancer programs in the United States, demonstrating the potential for this nomogram to predict SN positivity and reduce the number of negative SN biopsies.
