Multi-ancestry meta-analysis of keloids uncovers novel susceptibility loci in diverse populations

Abstract

Keloids are raised scars that grow beyond original wound boundaries, resulting in pain and disfigurement. Reasons for keloid development are not well-understood, and current treatment options are limited. Keloids are more likely to occur in darker-skinned individuals of African and Asian descent than in Europeans. We performed a genome-wide association study (GWAS) examining keloid risk across and within continental ancestry groups, incorporating 7837 cases and 1,593,009 controls. We detected 26 loci in the multi-ancestry analysis, 12 of which replicated in an independent dataset. Heritability estimates were 6%, 21%, and 34% for the European, East Asian, and African ancestry analyses, respectively. Genetically predicted gene expression and colocalization analyses identified 27 gene-tissue pairs, nine in skin and fibroblasts. Pathway analyses implicated integrin signaling and upstream regulators involved in cancer, fibrosis, and sex hormone signaling. This investigation nearly quintuples the number of keloid-associated risk loci, illuminating biological processes in keloid pathology.

Fig. 1. Summary of populations for each meta-analysis.

EUR European ancestry meta-analysis, AFR African ancestry meta-analysis, EAS East Asian ancestry analysis, Multi-ancestry Multi-ancestry meta-analysis, combining all data sources. Cohort-specific information may be found in Table 1.

Fig. 2. Multi-ancestry meta-analysis of keloids replicates known genes and identifies previously unidentified keloids genomic risk loci.

Manhattan plot. The top five most significant loci are labeled with the nearest gene. The red line signifies the traditional GWAS significance threshold of p < 5 × 10^-8 while the blue dotted line signifies the suggestive threshold (p < 1 × 10^-5). Logistic regression statistical tests; multiple testing correction p-value threshold used (5 × 10⁻⁸). Plot created using the fastman R library.

Fig. 3. Ancestry-specific keloid meta-analyses display consistency of major results across ancestry groups.

Manhattan plots. a European ancestry meta-analysis; b East Asian ancestry analysis; c African ancestry meta-analysis. Logistic regression statistical tests; multiple testing correction p-value threshold used (5 × 10⁻⁸). Plots created using the fastman R library.

Fig. 4. SNP-based heritability (h²) estimates are highest in the African ancestry population and lowest in the European ancestry population.

Estimates were derived from LDSC analyses using ancestry-specific GWAS summary statistics. Data are presented as mean values +/- SE.

Fig. 5. Multi-ancestry GPGE analysis, restricted to keloid-relevant tissues.

Nine genes across four tissues are predicted to exhibit increased gene expression in response to keloids risk alleles. Two-sided Wald test; multiple testing correction significance threshold for the four keloid-relevant tissues = p < 1.5 × 10^-6.