InterMEL: An international biorepository and clinical database to uncover predictors of survival in early-stage melanoma

**Background:** Melanoma accounts for most skin cancer deaths, with an estimated 7,650 deaths in the USA in 2022. While 5-year survival ranges from 98% for localized disease to <30% for distant metastases, prediction of individual outcomes based on clinicopathologic factors has limited power. The TCGA melanoma study had only ~67 primary melanomas, and these were larger than typical primary melanomas, limiting generalizability. The InterMEL consortium aims to define molecular subtypes of primary cutaneous melanoma (≥1.05 mm, stage IIA-IIID) to improve prognostic stratification and guide adjuvant therapy decisions, particularly given the recent FDA approval of pembrolizumab for stage IIB/C melanoma.

**Methods:** This retrospective multicenter study collects archived FFPE primary melanoma tissues and clinical data from 10 international centers. Eligible patients had primary cutaneous melanoma diagnosed between January 1, 1998 and January 1, 2016, stage IIA-IIID (AJCC 8th edition), thickness >1.05 mm, and no prior adjuvant immunotherapy or targeted therapy. Cases died of melanoma within 5 years of diagnosis; controls survived ≥5 years without recurrence. Tissue sections (seven 10μm unstained sections plus two H&E slides) were shipped to Memorial Sloan Kettering Cancer Center for centralized handling. After dermatopathology review, RNA and DNA were co-extracted from the same cell lysate using the AllPrep DNA/RNA FFPE Kit (Qiagen). Samples were distributed for somatic mutation profiling (MSK-IMPACT NGS assay), genome-wide DNA methylation (Infinium MethylationEPIC arrays), and miRNA expression (Nanostring nCounter Human v3 miRNA Expression Assay). Statistical analyses used Pearson correlation, Fisher's Exact test, Welch t-test, and ANOVA.

**Key Results:** Of 685 eligible melanomas, sufficient material was obtained for miRNA screening in 683/685 (99%), methylation in 467 (68%), and somatic mutations in 560 (82%). In 446/685 (65%) cases, aliquots were sufficient for all three platforms. Among samples evaluated, mean NGS coverage was 249x, with 59/317 (18.6%) samples having coverage below 100x. Methylation QC failed in 41/414 (10%) due to low intensity probes or insufficient normalization. Only 6/683 (1%) RNA samples failed Nanostring QC. Age of FFPE tissue blocks (p<0.001) and time from sectioning to co-extraction (p=0.002) were associated with methylation screening failures. Melanin reduced amplification of fragments ≥200bp (absent/lightly pigmented vs heavily pigmented, p<0.003), but heavily pigmented tumors yielded greater RNA amounts (p<0.001) and more RNA >200 nucleotides (p<0.001). Tumor characteristics significantly associated with nucleic acid quantity/quality included: positive correlation between extracted tissue area and DNA/RNA yield; stage III melanomas yielded more RNA per area than stage II (39.5 vs 32.4 ng/mm², p<0.01); TILs were associated with greater dsDNA amounts and amplifiability; ulceration was associated with better DNA A260/A280 ratios but also with DNA-QC failures (15% failed among ulcerated vs 28% among non-ulcerated, p<0.001). Of 561 melanomas with full pathology review, 327 (58.3%) were controls and 234 (41.7%) were cases. Median age at diagnosis was 64 years, 61.5% were male, 95.2% were white. Median Breslow thickness was 3.8 mm (IQR 2.5-6.2). Stage distribution: 52.2% stage II, 47.8% stage III.

**Clinical Implications:** This study demonstrates that multi-omic profiling from co-extracted DNA/RNA of archival FFPE primary melanomas is feasible in a multicenter setting, with 65% of cases yielding sufficient material for all three platforms. Investigators should plan for 35-45% attrition when designing similar studies. Key modifiable factors affecting success include minimizing time between tissue sectioning and nucleic acid extraction, and accounting for block age. Melanin content differentially affects DNA (impairs amplification) and RNA (may protect against degradation). These findings provide practical guidelines for future large-scale molecular studies using limited archival tumor tissue, particularly for early-stage cancers where tissue availability is constrained.