WP1 quantified CAYA melanoma incidence, sex differences and survival using EUROCARE-6 and ECIS-derived registry data, compiling historical registry information for pan-European analyses. Harmonized REDCap datasets/codebooks now include >500 cases, enabling hypothesis-driven epidemiological sub-studies. Exposome modelling was operationalized by stratifying Europe into latitudinal climatic zones and integrating environmental layers to support regression-based risk estimation. Genetics reached sequencing scale: from 465 selected germline samples, 323 were selected for WGS, passed initial QC and entered bioinformatic processing, supporting preliminary variant analyses and CAYA-oriented genetic risk modelling. WP2 advanced harmonized molecular workflows in L/GCMN. CNV profiling across 16 lesions from 14 patients (aCGH, shallow WGS, methylation-derived CNV) showed concordance for broad events while improving detection of focal/segmental CNVs and highlighting the need to update interpretation criteria beyond legacy CGH. Visium-based spatial transcriptomics expanded, generating and integrating 14 L/GCMN datasets and supporting identification of lesion-specific melanocytic states relevant to transformation biology.
WP3 consolidated the pathology network and curated case base: multicenter expert review continued under standardized QC and cutting/handling procedures and a HALO Link second-opinion workflow was advanced and documented, capturing implementation lessons (WSI variability, tissue scarcity) that inform standards for scalable cross-border pathology support. WP4 consolidated and analyzed the most extensive real-world anti-PD-1 dataset in pediatric melanoma (106 patients ≤18 years, stage III–IV) in a GDPR-compliant database with peer-reviewed dissemination. Results indicate adjuvant outcomes and safety comparable to adults, similar outcomes/safety across age strata including <12 years, and apparently lower survival in advanced disease versus adult series, addressing a key evidence gap due to rarity. WP5 progressed AI-based diagnostic support through scaled imaging acquisition/governance and explainable AI, combining Grad-CAM attribution with a pathologist-oriented strategy that decomposes WSIs into interpretable tissue/structure components and extracts human-aligned quantitative descriptors (e.g. melanocyte/nest metrics). WP6 advanced ML models for prognosis/survival and sentinel lymph node positivity prediction through retrospective validation and integration. Non-invasive technologies moved beyond prototyping: breath-analysis workflows were optimized (including simulated breath generation and sensor improvements) and the disposable sensing patch progressed through sensor configuration and device-readiness work, with clinical translation prepared through technical set-up and study preparation.
WP7 finalized an HTA framework for digital health/AI in rare diseases (scoping review, HTA/KOL interviews, proof-testing with MELCAYA technologies) and a transferability tool. ELSI work defined policy priorities (training, over/underdiagnosis balance, autonomy, equitable access, responsible AI), and a Delphi instrument (34 statements) was prepared across awareness/early recognition, governance, assessment/reimbursement and health system organization. WP8 supported technical integration via harmonization exchange and infrastructure-alignment planning to position outputs for long-term accessibility via UNCAN.eu. WP9 strengthened patient-generated evidence through SenseMaker self-ethnography. Training was completed and a pilot survey on access-related burden was designed and tested with patient advocates, improving usability/data quality.
