Exploring the Germline Genetics of In Situ and Invasive Cutaneous Melanoma: A Genome-Wide Association Study Meta-Analysis

Abstract

Importance: It is unknown whether germline genetic factors influence in situ melanoma risk differently than invasive melanoma risk.

Objective: To determine whether differences in risk of in situ melanoma and invasive melanoma are heritable.

Design, setting, and participants: Three genome-wide association study meta-analyses were conducted of in situ melanoma vs controls, invasive melanoma vs controls, and in situ vs invasive melanoma (case-case) using 4 population-based genetic cohorts: the UK Biobank, the FinnGen cohort, the QSkin Sun and Health Study, and the Queensland Study of Melanoma: Environmental and Genetic Associations (Q-MEGA). Melanoma status was determined using International Statistical Classification of Diseases and Related Health Problems codes from cancer registry data. Data were collected from 1987 to 2022, and data were analyzed from September 2022 to June 2023.

Exposure: In situ and invasive cutaneous melanoma.

Main outcomes and measures: To test whether in situ and invasive melanoma have independent heritable components, genetic effect estimates were calculated for single-nucleotide variants (SNV; formerly single-nucleotide polymorphisms) throughout the genome for each melanoma. Then, SNV-based heritability was estimated, the genetic correlation between melanoma subtypes was assessed, and polygenic risk scores (PRS) were generated for in situ vs invasive status in Q-MEGA participants.

Results: A total of 6 genome-wide significant loci associated with in situ melanoma and 18 loci with invasive melanoma were identified. A strong genetic correlation (genetic r = 0.96; 95% CI, 0.76-1.15) was observed between the 2 classifications. Notably, loci near IRF4, KLF4, and HULC had significantly larger effects for in situ melanoma compared with invasive melanoma, while MC1R had a significantly larger effect on invasive melanoma compared with in situ melanoma. Heritability estimates were consistent for both, with in situ melanoma heritability of 6.7% (95% CI, 4.1-9.3) and invasive melanoma heritability of 4.9% (95% CI, 2.8-7.2). Finally, a PRS, derived from comparing invasive melanoma with in situ melanoma genetic risk, was on average significantly higher in participants with invasive melanoma (odds ratio per 1-SD increase in PRS, 1.43; 95% CI, 1.16-1.77).

Conclusions and relevance: There is much shared genetic architecture between in situ melanoma and invasive melanoma. Despite indistinguishable heritability estimates between the melanoma classifications, PRS suggest germline genetics may influence whether a person gets in situ melanoma or invasive melanoma. PRS could potentially help stratify populations based on invasive melanoma risk, informing future screening programs without exacerbating the current burden of melanoma overdiagnosis.

Figure 1.. Miami Plot of Invasive Melanoma and In Situ Melanoma

A Miami plot showing P values from the genome-wide association study meta-analysis of invasive melanoma vs controls on a log₁₀ scale, and in situ melanoma vs controls on a −log₁₀ scale. The plot has been capped at a −log₁₀ P value less than 30. The samples used in the genome-wide association study meta-analysis were the UK Biobank, FinnGen, QSkin Sun and Health Study, and Queensland Study of Melanoma: Environmental and Genetic Associations. To be present in the plot, a given single-nucleotide variant had to be present in at least 2 of the meta-analyzed cohorts. For a Miami plot of the meta-analysis results with all analyzed single-nucleotide variant plotted, regardless of how many cohorts they were present in, please see eFigure 1 in Supplement 1. The main difference between these figures is that eFigure 1 in Supplement 1 shows an additional locus on chromosome 1 (KISS1) for invasive melanoma, which was only present in the FinnGen population.

Figure 2.. Scatterplot of Effect Estimates of Lead Loci of In Situ and Invasive Melanoma

Effect size plot of the lead single-nucleotide variants from the in situ melanoma and the invasive melanoma genome-wide association study meta-analysis. Point colors represent the genome-wide significance of the point (ie, whether the single-nucleotide variant was significant in the in situ melanoma analysis, invasive melanoma analysis, or both analyses). Effect estimates for the IRF4 and KLF4 were significantly different from one another at P < .003. The solid black line represents a correlation of 1, and the crosses represent standard errors of the genome-wide association study effect estimates.

Figure 3.. Manhattan Plot Case-Case Genome-Wide Association Study Meta-Analysis of In Situ Melanoma vs Invasive Melanoma

This Manhattan plot from the genome-wide association study meta-analysis of in situ vs invasive melanoma used data from the UK Biobank, QSkin Sun and Health Study, and Queensland Study of Melanoma: Environmental and Genetic Associations. The red line indicates genome-wide significance (P < 5 × 10⁻⁸) on the log₁₀ scale, and the blue line indicates suggestive genome-wide significance (P < 5 × 10⁻⁷). The single genome-wide significant single-nucleotide variant is rs4566922. This single-nucleotide variant is near to the gene HULC.

Figure 4.. Comparison of the In Situ Melanoma vs Invasive Melanoma Polygenic Risk Score (PRS) Distribution Between Individuals With In Situ and Invasive Melanoma in the Q-MEGA Sample

Comparison of the distribution of the PRS of the Queensland Study of Melanoma: Environmental and Genetic Associations (Q-MEGA) sample. The vertical dotted lines represents the mean PRS for each group. The PRS is weighted using the effect estimates derived from the case-case genome-wide association study meta-analysis of in situ vs invasive melanoma in the UK Biobank, QSkin Sun and Health Study, and part of Q-MEGA.