Background: Actinic keratosis (AK) is considered as precursor lesion of invasive squamous cell carcinoma. Molecular studies on AK are limited because of too small size of the biopsy specimen to obtain enough DNA or RNA.
Methods: Twenty biopsy cases of AK, followed by second same-sited biopsies, were included. Ten cases were diagnosed with total regression (regression group), while the other 10 were diagnosed with invasive carcinoma (progression group) in the follow-up biopsies. Using digital spatial profiling (DSP) technology, whole-gene expression analysis defined by specific regions of interest was performed for all 20 cases. After the clinicopathological features were assessed, separate and integrated analyses of these features and gene expression patterns were performed using machine-learning technology. All analyses were performed on both lesion keratinocytes (KT) and infiltrated stromal lymphocytes (LC).
Results: Among the 18,667 genes assessed, 33 and 72 differentially expressed genes (DEGs) between the regression and progression groups were found in KT and LC respectively. The primary genes distinguishing the two groups were KRT10 for KT and CARD18 for LC. Clinicopathological features were weaker in risk stratification of AK progression than the gene expression patterns. Pathways associated with various cancers were upregulated in the progression group of KT, whereas the nucleotide-binding oligomerization domain (NOD)-like receptor signalling pathway was upregulated in the progression of LC.
Conclusion: Gene expression patterns were effective for risk stratification of AK progression, and their distinguishing power was higher than that of clinicopathological features.
Keywords: actinic keratosis; gene expression; risk stratification.
© 2023 Australasian College of Dermatologists.