ORIGINAL ARTICLE, p 797
The stratum corneum (SC) is continually shed from the skin surface and renewed from underlying keratinocytes that terminate growth and digest their DNA. While it would be reasonable to assume that most DNA and RNA would be broken down into nucleotides and returned to the body, an adequate amount of RNA molecules remains in the SC to allow for molecular characterization.1 The article by Wachsman et al.2 reveals that RNA isolated from the SC overlying skin lesions includes sufficient information to diagnose melanoma. In their study, the SC was isolated by removing adhesive tape affixed over skin lesions. The RNA isolated from the SC was studied on gene expression arrays to determine relative gene expression levels. Wachsman et al. showed that 312 genes were differentially expressed between melanomas, naevi and normal skin specimens. Hierarchical clustering revealed two major branches: one including melanomas and the other including normal skin. The RNA expression patterns from the naevi studied were split, with 13 lesions segregating with melanoma and 55 lesions with normal skin. Genes segregating with the melanomas included melanocytic differentiation genes such as tyrosinase, tyrosinase-related protein 1 and dopachrome tautomerase. Further to optimize segregation of the lesions, melanomas were compared directly with naevi to develop a classifier containing 168 genes with a sensitivity of 100% and a specificity of 88% on a test set of 39 melanomas and 89 naevi.
The classifier was further reduced to a set of 17 genes that maintained the 100% sensitivity and 88% specificity. The melanocytic differentiation markers noted above were not included. Upregulated genes in this 17-gene set included: KIT and EDNRB, genes associated with melanocytic development and proliferation; PRAME, a gene preferentially expressed in melanoma, breast cancer and ovarian cancer; TRIB2, a gene that promotes melanoma growth and survival;3 and NAMPT, a gene overexpressed and implicated in the survival of prostate cancer cells.4 This reduced classifier includes genes more likely to be associated with malignancy and appears to give better segregation of melanomas from benign naevi.
How the RNA markers get into the SC is not known but it seems likely that the source is pagetoid cells extruded into the SC. Pagetoid cells are common over melanoma but they may also be present over benign naevi.5 It is likely that this represents a normal melanocytic elimination pathway. The difference between the moles that segregated with normal skin vs. melanoma with the first 312-gene cluster including melanocytic differentiation markers may have been due to differences in growth phase and elimination of excess melanocytic cells into the SC. The problem with overcalling the benign naevi was reduced in the 17-gene classifier presumably due to selection of genes more likely to be associated with malignant melanocytic cells.
The findings of Wachsman et al. open a new area for the molecular diagnosis of melanoma. It is clear that larger prospective studies with higher naevus to melanoma ratios are needed, but the technology holds significant promise. The ability to use SC RNA for diagnostic purposes may also have utility in a vast number of other dermatological conditions. While this technology may hold great utility in the clinic, it also lends itself to self diagnostics. It is possible that at some point in the future, a worried patient will be able to place the tape over a skin lesion, draw around the lesion location, and send the tape to a central laboratory for diagnosis. The patient might even have the result faster than the appointment with the dermatologist.