In the last decade, the advent of immunotherapies with inhibitors of immune checkpoints, such as anti-PD-1 and anti-CTLA-4, has revolutionized the treatment of advanced or metastatic melanoma. However, the clinical benefit remains limited to a subset of patients. Identifying the patients most likely to benefit from these novel therapies (and avoiding unnecessary toxicity in non-responding patients) is therefore critical. Previous studies found a significant link between the high mutational load of a tumor (TMB) and its response to anti-PD-1 monotherapy, regardless of the histological type of cancer. Unfortunately, TMB measurement is expensive, and requires extensive sequencing approaches difficult to implement in clinical practice. I have shown that melanomas known to be secondary to mutagenic ultraviolet rays (UVR) often carry a high TMB. The cumulative UVR damage translates into visible stigmas termed dermatoheliosis on patients' skin, easy to recognize with the naked eye of the clinician around the scar of the primary melanoma. My project proposes to establish, for the first time, dermatoheliosis as a novel predictive factor of response to anti-PD-1 immunotherapy, to be used within multidisciplinary tumor boards as a powerful decision-support tool to select the best treatment option. Specifically, I will 1) develop, validate and test in a prospective manner, an artificial intelligence (AI)-based algorithm, to assess features of pericicatricial dermatoheliosis based on a collection of photographs obtained from patients with unresectable locally advanced or metastatic melanoma 2) demonstrate the link between dermatoheliosis, TMB, immune and treatment response by characterizing pericicatricial skin single cell transcriptomics, as well as tumor DNA, RNA and host immunological profiles of the patients. This directly accessible, non-invasive, surrogate marker for TMB will be a game changer in clinical practice and will subsequently be translated to other skin cancers.