e21566 Background: Since 2018, adjuvant immune checkpoint inhibitors (ICI) have been available for patients with resected stage III/IV melanoma. The Checkmate 238 and Keynote 054 studies showed a survival benefit in patients receiving adjuvant ICI. However, there is no validated biomarker to identify patients at high risk of relapse, despite receiving adjuvant therapy. Here we investigated resistance patterns to immune checkpoint inhibition in the adjuvant setting. Methods: In this cohort study, a discovery cohort of 30 patients from the Department of Dermatology, University Hospital Zürich, was used to evaluate resistance patterns to adjuvant ICI. A second cohort of 107 patients from the Department of Dermatology, University Hospital Tuebingen, was used to validate the findings. We measured epidermal growth factor receptor (EGFR) expression in primary tumor and metastatic tissue by immunohistochemistry and digital pathology using QuPath. The primary endpoint was relapse-free survival. Results: We included 137 patients with melanoma receiving adjuvant ICI: 63 (46%) were females, and 52 (38%) patients harbored a BRAFV600 mutation. The median follow-up was 45 and 39 months in the Zürich and Tuebingen cohorts, respectively (95% confidence interval (CI) 41.8-48.2 and 36.8-41.2, respectively). The two cohorts were homogenous in terms of patients' and tumor characteristics, except for the relapse rate, which was significantly higher in the Tuebingen cohort ( p< 0.002). 71 (52%) therapy naïve in transit or lymph node metastasis and 23 (17%) primary tumors were available to evaluate EGFR expression by immunohistochemistry and digital pathology using QuPath. Surprisingly, in both cohorts, patients with high EGFR expression in metastatic tissue had a significantly higher relapse rate than patients with low EGFR expression ( p= 0.0004 and p= 0.0168 for the Zürich and Tuebingen cohorts, respectively). EGFR expression in primary tumors did not correlate with relapse-free survival ( p= 0.7286). Next-generation sequencing revealed that EGFR expression assessed by immunohistochemistry was not associated with the presence of EGFR mutations and/or amplifications. Conclusions: Currently, there are no validated biomarkers to identify patients with melanoma at high risk of recurrence, despite receiving adjuvant ICI. We found in two independent cohorts that EGFR expression in therapy naive metastatic tissue is positively associated with relapse in patients receiving adjuvant ICI. Our data identified EGFR as a prognostic biomarker that can easily be assessed by immunohistochemistry, allowing us to stratify patients for adjuvant ICI therapy according to their risk of relapse. Stratification according to EGFR expression should be considered in future clinical trials in the adjuvant setting.