Genomic Similarities in Actinic Keratosis and Cutaneous Squamous Cell Carcinoma Found

Actinic keratosis
Actinic keratosis – a precursor to squamous cell carcinoma or skin cancer on a man’s hand
The largest AK whole-exsome sequencing study to date is reported on.

Cutaneous squamous cell carcinoma (cSCC) and its precursor, actinic keratosis (AK), are highly similar at the genomic level in terms of tumor mutational burden, driver gene patterns, and copy number aberrations, according to the largest AK genomic dataset to date. These findings were published in the Journal of Investigative Dermatology.

Investigators collected 4mm AK punch biopsies and performed DNA extraction, genetic analysis, and whole exome sequencing (WES) analyses. They conducted clonal analyses and estimated tumor purity on the basis of cellular frequency of the dominant clone, with all somatic variants assigned to their nested clones. They selected 5 sets of expression data of patient samples downloaded from Gene Expression Omnibus.

Investigators analyzed a total of 37 AKs from 37 patients, including 23 immunosuppressed (IS) patients and 14 immunocompromised (IC) patients. There was WES data from cSCC samples available for comparison in 16 patients.

Using WES, investigators found a mean tumor mutational burden (TMB) of 43.5 mutations per megabase DNA, which was similar to the TMB of 50 mutations per megabase DNA that they previously reported in cSCC samples.

AKs from IS patients were significantly more mutated than those from IC patients when comparing total nonsynonymous mutations (median, 1,609 vs 729 mutations, respectively; P =.03), and also had a significantly higher TMB (55.9 mutations per megabase DNA) compared to IC patients (22.3 mutations per megabase DNA; P =.03).

Signature-32 mutations were found in 22 AKs and exclusively in patients exposed to azorthioprine (P <.00001), mirroring what investigators had found in cSCC samples.

A median of 9% (range, 0-62%) of AKs affected by copy number aberrations (CNAs), similar to cSCC (P =.68) and independent of immune status. The most common CNAs in AK were significantly correlated with those in cSCC (r =0.68; P =.003). Loss of chromosome region 9p (43%), 13q (32%), and 5q (24%) were the most frequent copy number losses.

There were 44 AK SMGs confirmed by at least 2 different methods, and they included tumor suppressor genes that are consistently reported as mutated in cSCC: TP53, NOTCH1, NOTCH2, FAT1, and KMT2C. HMCN1 was also altered in 50% of both AKs and cSCC. The SMG frequency in AKs was independent of patients’ immune status. The most common mutational signatures present in the 44 SMGs were signature-32 (40%) and signature-7 (38%).

When comparing the OncoPrint for this study’s 44 AK SMGs and a previous study’s OncoPrint for 22 cSCC SMGs, TP53, NOTCH1 and NOTCH2 were the only shared SMGs, and none of the total 63 SMGs differed significantly by frequency between AK and cSCC when adjusted for multiple comparisons.

There were no significant differences between AK SMGs and well-differentiated cSCC SMGs, but there were 10 moderately and/or poorly differentiated cSCC SMGs that were significantly more altered than in AK. The mutation distribution along the protein domains for AK and cSCC SMGs was similar, except for the PIK3CA hotspot in AK.

Genes previously implicated in cSCC, such as CDKN2A and HRAS, were mutated at similar rates between this study’s AK and cSCC cohorts. There were 2 genes, CACNA1C and KCNK5, that showed significant evidence of nonsynonymous mutations more clonally dominant in AK than cSCC.

Clonal analyses revealed similar patterns of clonality among the 3 SMGs shared by AKs and cSCC, with a higher proportion of clonal mutations occurring in TP53 and NOTCH1 than in NOTCH2.

When comparing the 16 AKs that also had cSCC WES data available, investigators found a significant positive correlation in the mutational signature profiles of 10 (71.4%) of the AK-cSCC pairs.

Several immune system signaling pathways were significantly more mutated in cSCC than AK, as well as TGFB, adipocytokine, GnRH, and insulin signaling. Additionally, several metabolism pathways were significantly more differentially mutated in cSCC thank AK.

Comparing this study’s AK SMGs to existing gene expression profiles of normal skin, AK, and cSCC, investigators 5 SMGs significantly downregulated in at least 2 datasets of AK vs normal skin, suggesting a tumor suppressor role: KIF24, KCNK5, EPB41L, INSIG2, and ABI3BP, and NOTCH1 was the only SMG significantly upregulated. For cSCC vs AK, IMPA1 was the only SMG significantly differentially expressed in at least 2 datasets as upregulated in cSCC, suggesting a tumor promotor role.

Examining expression patterns of the TGFB pathway genes – significantly more mutated in cSCC than AK – in normal skin, skin with AK, and skin with cSCC, investigators found 2 clusters of progression-dependent dysregulation of TGFB signaling: an upregulation of genes in normal skin that became increasingly downregulated, progressing from AK to cSCC, and the reverse pattern. There were 55 genes downregulated in the deleted regions and 2 genes upregulated in the gained regions in AK vs the normal control.

The study was limited by lack of AK grading.

Investigators wrote that their genomic analysis uncovered “significant molecular alterations” shared between AK and cSCC and “which may contribute to evolution from AK to cSCC,” including alterations in key signaling pathways, mutations in specific genes, and differences in intrasample heterogeneity.

Disclosure: Several study authors declared affiliations with biotech, pharmaceutical, and/or device companies. Please see the original reference for a full list of authors’ disclosures. 


Thomson J, Bewicke-Copley F, Anene CA, et al. The genomic landscape of actinic keratosis. J Invest Dermatol. 2021;141(7):1664-1674. doi:10.1016/j.jid.2020.12.024