Prospective validation of the prognostic 31-gene expression profiling test in primary cutaneous melanoma

Background Gene expression profiling (GEP) has been integrated into cancer treatment decision-making in multiple neoplasms. We prospectively evaluated the prognostic utility of the 31-GEP test (DecisionDx-Melanoma, Castle Biosciences, Inc) in cutaneous melanoma (CM) patients undergoing sentinel node biopsy (SNB). Methods One hundred fifty-nine patients (age 26-88) diagnosed with melanoma between 01/2013 and 8/2015 underwent SNB and concurrent GEP testing. GEP results were reported as low-risk Class 1 (subclasses 1A and 1B) or high-risk Class 2 (subclasses 2A and 2B). Statistical analyses were performed with chi-square analysis, t tests, log-rank tests, and Cox proportional hazard models. Recurrence-free survival (RFS) and distant metastasis-free survival (DMFS) were estimated using Kaplan-Meier method. Results Median follow-up was 44.9 months for event-free cases. Median Breslow thickness was 1.4 mm (0.2-15.0 mm). There were 117 Class 1 and 42 Class 2 patients. Gender, age, Breslow thickness, ulceration, SNB positivity, and AJCC stage were significantly associated with GEP classification (P 0.05 for all). Recurrence and distant metastasis rates were 5% and 1% for Class 1 patients compared with 55% and 36% for Class 2 patients. Sensitivities of Class 2 and SNB for recurrence were 79% and 34%, respectively. Of 10 SNB-positive/Class 2 patients, 9 recurred. By multivariate analysis, only SNB result and GEP class were statistically associated with both RFS (P = 0.008 and 0.0001) and DMFS (P = 0.019 and 0.001). Conclusions Gene expression profiling Class 2 result and SNB positivity were independently associated with recurrence and distant metastasis in primary CM patients. GEP testing may have additive prognostic utility in initial staging work-up of these patients.