Journal Review H ighlights of recently published research on cutaneous melanoma Development and Validation Study Data Demonstrate Accuracy of i31-GEP SLN Status Prediction, Point to Benefits for its More Precise, Personalized Risk Prediction The integrated 31-gene expression profile (i31-GEP) test combining the continuous 31-GEP score with clinicopathologic features optimizes prediction of sentinel lymph node (SLN) positiv-ity in patients with any stage primary cutaneous melanoma, according to research published in JCO Precision Oncology . Using the i31-GEP to provide precise personalized information could therefore reduce the number of unnecessary SLN biopsies (SLNB) and improve the identification of patients likely to have a pos-itive result who may benefit most, said investiga-tors who developed and validated the tool. The i31-GEP artificial intelligence algorithm for SLN risk prediction was developed using data from a training cohort of 1,398 patients with primary cutaneous melanoma tumors (T1-T4) having known Breslow thickness and a contin-uous 31-GEP test result. Models were generated using the continuous 31-GEP test result, Breslow thickness, and multiple other clinicopathologic features. The continuous 31-GEP score was found to be the most important variable and best pre-dictor of SLN positivity among all covariates. Performance validation of the i31-GEP for predicting SLN positivity was conducted using retrospective data from an independent cohort of 1,674 patients with T1-T4 disease seen across 30 centers. Linear regression analysis showed high concordance between the observed SLN positivity rates and those predicted by the i31-GEP (slope = 0.999). The validation study also assessed accuracy of the i31-GEP by considering a predicted likeli-hood of a positive SLN <5% as a negative test and a ≥5% likelihood of SLN positivity as a positive test. Using these criteria and data from the entire patient population, the i31-GEP demonstrated sensitivity of 95.1% for predicting SLN positivity and a negative predictive value of 98.1% (false negative rate of 1.9%). Further analysis considering the subgroup of patients for which additional guidance on the decision to perform SLNB is needed (i.e., patients with T2 disease or considered to have “high-risk” T1a dis-ease based on their clinicopathologic features) showed the sensitivity and negative predictive value of the i31-GEP remained high (89.8% and 97.4%, respectively). Comparing the i31-GEP risk predictions to those risk classifications based on T stage alone as per National Comprehensive Cancer Net-work (NCCN) guidelines, showed that use of the i31-GEP test would result in significant risk reclassification. The percentage of patients in the overall cohort predicted to have <5% SLN positivity risk was increased threefold using the i31-GEP test versus the NCCN guidelines (from 8.5% to 27.7%). Applying the i31-GEP to the The i31-GEP demonstrated sensitivity of 95.1% for predicting SLN positivity and a negative predictive value of 98.1% (false negative rate of 1.9%). November/December 2024 Supplement | 3
The Dermatology Digest Innovations in Personalized Medicine November/December 2024 Supplement: Page 3
