Document resource
Background Although immune checkpoint inhibitors (ICIs) have transformed melanoma management, about half of patients do not benefit and predictive biomarkers remain limited. DetermaIO is a 27-gene expression classifier of the tumor microenvironment (TME) that identifies a ‘hot’ (IO+) versus ‘cold’ (IO-) TME phenotype. We have previously demonstrated that DetermaIO is predictive of selective benefit to ICIs and we have performed clinical validations in triple-negative breast cancer, non-small cell lung cancer, colorectal cancer, and other tumor types. In this study, we evaluated its clinical utility in a single institution pilot cohort of melanoma patients.Methods Thirty-one archival metastatic melanoma specimens (cutaneous n = 10, uveal n = 4, other/unknown primary n = 17) from patients treated with ipilimumab 3mg/kg and nivolumab 1mg/kg at Thomas Jefferson University were analyzed using the DetermaIO RT-qPCR assay. The prespecified IO+ cutpoint (IO Score ≥ 0.09) was compared against disease control rate (DCR = complete response + partial response + stable disease) based on RECIST v1.1. Odds ratios (ORs), Fisher’s exact tests, and subgroup analyses (treatment naïve vs previously treated) were performed. AUC-guided threshold optimization was explored, but the 0.09 cutpoint provided the most balanced sensitivity/specificity.Results Seventeen of 31 samples (55%) were IO+ based on the DetermaIO assay. DCR was 61% (19/31) and 13 of those 19 responders were IO+ (68%). Of the 12 patients who had progressive disease as best response, 8 were IO- (67%). IO+ samples had over 4-fold increased odds of disease control (OR = 4.33, p = 0.075) versus IO-. In the treatment naïve subgroup (n = 22), the OR was 4.0 and the responders had higher IO Scores (mean +0.09) than the non-responders (–0.11). In the pre-treated subgroup (n=9), all responses occurred in IO+ cases, but elevated IO Scores were also observed in non-responders (mean +0.13), suggesting therapy-induced TME activation that does not necessarily translate to sensitivity. Of the four uveal melanoma tumors, two were classified as IO+ and two as IO-. One of the two IO+ cases achieved stable disease (50%), while both IO- cases developed progressive disease.Conclusions In this preliminary series, DetermaIO stratified melanoma patients by likelihood of ICI response, with the strongest predictive signal in treatment naïve disease and promising activity in uveal melanoma. Incorporating a predictive biomarker could improve therapeutic sequencing in cutaneous melanoma and help inform individual treatment decisions for uveal melanoma. These data support prospective validation of DetermaIO in melanoma for treatment decision-making.Ethics Approval The study was reviewed and approved by the Institutional Review Board at Thomas Jefferson University (IRB#iRISID-2024-1601). The participants gave informed consent prior to tumor tissue collection and its potential use in research.