Predicting sentinel node positivity in patients with melanoma: external validation of a risk-prediction calculator (the Melanoma Institute Australia nomogram) using a large European population-based patient cohort

Abstract

BACKGROUND: A nomogram to predict sentinel node (SN) positivity [the Melanoma Institute Australia (MIA) nomogram] was recently developed and externally validated using two large single-institution databases. However, there remains a need to further validate the nomogram's performance using population-based data. OBJECTIVES: To perform further validation of the nomogram using a European national patient cohort. METHODS: Patients with cutaneous melanoma who underwent SN biopsy in the Netherlands between 2000 and 2014 were included. Their data were obtained from the Dutch Pathology Registry. The predictive performance of the nomogram was assessed by discrimination (C-statistic) and calibration. Negative predictive values (NPVs) were calculated at various predicted probability cutoffs. RESULTS: Of the 3049 patients who met the eligibility criteria, 23% (691) were SN positive. Validation of the MIA nomogram (including the parameters Breslow thickness, ulceration, age, melanoma subtype and lymphovascular invasion) showed a good C-statistic of 0·69 (95% confidence interval 0·66-0·71) with excellent calibration (R 2 = 0·985, P = 0·40). The NPV of 90·1%, found at a 10% predicted probability cutoff for having a positive SN biopsy, implied that by using the nomogram, a 16·3% reduction in the rate of performing an SN biopsy could be achieved with an error rate of 1·6%. Validation of the MIA nomogram considering mitotic rate as present or absent showed a C-statistic of 0·70 (95% confidence interval 0·68-0·74). CONCLUSIONS: This population-based validation study in European patients with melanoma confirmed the value of the MIA nomogram in predicting SN positivity. Its use will spare low-risk patients the inconvenience, cost and potential risks of SN biopsy while ensuring that high-risk patients are still identified.