Castle Biosciences Publishes DecisionDx-Melanoma Study Results

Castle Biosciences announced the publication of results from the largest prospective, multicenter study to date evaluating DecisionDx-Melanoma's integrated sentinel lymph node biopsy test result. The paper, available in Future Oncology, confirms that DecisionDx-Melanoma's i31-SLNB identifies patients below the 5% National Comprehensive Cancer Network threshold for forgoing sentinel lymph node biopsy and outperforms traditional staging criteria and other predictive gene expression profile tests. Current NCCN Cutaneous Melanoma Guidelines recommend forgoing SLNB when the likelihood of SLN positivity is less than 5%, considering SLNB when risk is between 5-10% and offering the procedure when risk exceeds 10%. DecisionDx-Melanoma's i31-SLNB integrates the independently validated 31-GEP score with established clinicopathologic factors, including Breslow thickness, ulceration, mitotic rate and age, to generate a personalized likelihood of SLN positivity that supports decision-making aligned with these guideline thresholds. In this prospective, multicenter study of 912 patients with T1-T4 cutaneous melanoma enrolled across 30 U.S. centers, 430 patients underwent SLNB and 482 did not, allowing for evaluation of both nodal positivity and recurrence outcomes. Among patients who underwent SLNB: Patients with less than 5% predicted risk of SLN positivity by DecisionDx-Melanoma's i31-SLNB had an actual SLN positivity rate of 2.6%. Patients with greater than 10% predicted risk had an SLN positivity rate of 21.4%, an 8.2-fold greater likelihood. In early-stage disease where SLNB decision-making is often most nuanced: Among patients with T1-T2a tumors, SLN positivity was only 1.8% in those with an i31-SLNB less than 5% predicted risk. In contrast, SLN positivity was 16.7% in those with greater than 10% predicted risk by i31-SLNB, a 9.3-fold greater likelihood. All patients with a low-risk i31-SLNB result who had at least two years of follow up demonstrated a 97.8% three-year RFS rate, indicating a very low risk of recurrence. Beyond nodal positivity rates, the study also evaluated performance relative to established guideline benchmarks and other predictive tests. Table 3 from the manuscript reports true-negative to false-negative ratios comparing the standard established by NCCN guidelines which uses American Joint Committee on Cancer staging criteria, DecisionDx-Melanoma's i31-SLNB and other predictive gene expression profile tests, across tumor stages. A TN:FN ratio of 19:1 corresponds to a 5% miss rate, meaning that for every 19 true-negative SLNBs, one positive SLNB would have been missed, consistent with NCCN guideline thresholds. Ratios greater than 19:1 indicate performance exceeding AJCC/NCCN guidance, while ratios below 19:1 indicate higher miss rates. In this study, the i31-SLNB demonstrated a TN:FN ratio of 55:1 in T1-T2a patients and 73:1 in the clinically important T1b-T2a subgroup, exceeding the 19:1 guideline benchmark. By comparison, previously reported data for other predictive GEP tests, including CP-GEP in T1-T2 disease, have shown lower TN:FN ratios. These findings highlight the ability of the i31-SLNB to more precisely identify patients at low risk of SLN positivity while minimizing missed positive nodes, particularly in early-stage tumors where accurate identification below the 5% threshold is critical to avoid unnecessary procedures. Overall, the published data confirm that DecisionDx-Melanoma's i31-SLNB can identify patients at sufficiently low risk of nodal positivity to safely forgo SLNB, reducing unnecessary procedures, procedure-related complications and healthcare costs while supporting risk-aligned management.