Principal Component Analysis Of Seven Skin-Ageing Features Identifies Three Main Types Of Skin Ageing

Background The underlying phenotypic correlations between wrinkles, pigmented spots (PS), telangiectasia and other related facial-ageing subphenotypes are not well understood. Objectives To analyse the underlying phenotypic correlation structure between seven features for facial ageing: global wrinkling, perceived age (PA), Griffiths photodamage grading, PS, telangiectasia, actinic keratosis (AK) and keratinocyte cancer (KC). Methods This was a cross-sectional study. Facial photographs and a full-body skin examination were used. We used principal component analysis (PCA) to derive principal components (PCs) of common variation between the features. We performed multivariable linear regressions between age, sex, body mass index, smoking and ultraviolet radiation exposure and the PC scores derived from PCA. We also tested the association between the main PC scores and 140 single-nucleotide polymorphisms (SNPs) previously associated with skin-ageing phenotypes. Results We analysed data from 1790 individuals with complete data on seven features of skin ageing. Three main PCs explained 73% of the total variance of the ageing phenotypes: a hypertrophic/wrinkling component (PC1: global wrinkling, PA and Griffiths grading), an atrophic/skin colour component (PC2: PS and telangiectasia) and a cancerous component (PC3: AK and KC). The associations between lifestyle and host factors differed per PC. The strength of SNP associations also differed per component with the most SNP associations found with the atrophic component [e.g. the IRF4 SNP (rs12203592); P-value = 1 center dot 84 x 10(-22)]. Conclusions Using a hypothesis-free approach, we identified three major underlying phenotypes associated with extrinsic ageing. Associations between determinants for skin ageing differed in magnitude and direction per component.