MADAM, Gene expression analysis is becoming a useful tool for a better definition of neoplasms at diagnostic, prognostic and predictive levels. An example of these applications appears in a recent issue of the BJD, which focuses the attention on a noninvasive procedure for this purpose. 1 This attractive proposal takes us to some biological and practical considerations.As the samples for analysis are taken from the corneal layer overlying the lesions, two biological aspects are essential for this process: how the tumour messenger RNA (mRNA) gets there and the functionality of this mRNA. Most cases of malignant melanomas will not show a sufficient number of superficial tumour cells to explain the positive findings, as confirmed by the authors in their results and discussion. 1 In this context, the RNA cannot be directly provided by the tumour cells but by keratinocytes through a transfer process similar to what happens with melanin. Although the paper does not provide experimental documentation on this issue, microvesicles (exosomes) containing mRNA and microRNA (miRNA) can be taken up by normal host cells, such as keratinocytes or endothelial cells, and translated by recipient cells. 2 Exosomes are specialized membranous nano-sized vesicles derived from endocytic compartments that are released by many cell types. Tumour-derived microvesicles therefore serve as a means of delivering genetic information and proteins to recipient cells in the tumour environment. It has been suggested that microvesicles shed by certain tumour cells hold functional mRNA that may promote tumour progression. Purified exosomes contain functional miRNAs and small RNA, but little mRNA is detected. Exosomes are specialized in carrying small RNA including the class 22-25 nucleotide regulatory miRNAs. Both the evidence provided from studies on exosomes and the lack of the expected phenotypic changes in keratinocytes from the expression of putative melanoma markers in the study support a transfer of predominant short RNA rather than functional mRNA. 1,2 No experimental data are provided on electron microscopic analysis of microvesicles ⁄exosomes or on the comparative gene expression between superficial epidermal samples and direct tumour samples from the same cases. These experimental data will clearly evaluate these possibilities. The biological heterogeneity in melanomas and dysplastic melanocytic lesions would potentially make these assessments more difficult. [3][4][5] We only have to consider that deep tumour compartments provide more accurate prognostic information in melanomas, and dermal compartments rather than the junctional compartments better define dysplastic melanocytic lesions.From a practical perspective, the data clearly segregate malignant from benign melanocytic lesions. However, the analysis needs a 17-gene classifier to achieve acceptable specificity and sensitivity, a complexity that is far greater than the standard histopathological evaluation. Using this approach 13 of 89 lesions classified as malignant by the test did ...