Expression profile of skin papillomas with high cancer risk displays a unique genetic signature that clusters with squamous cell carcinomas and predicts risk for malignant conversion

Chemical induction of squamous tumors in the mouse skin induces multiple benign papillomas: high-frequency terminally benign low-risk papillomas and low-frequency high-risk papillomas, the putative precursor lesions to squamous cell carcinoma (SCC). We have compared the gene expression profile of twenty different early low- and high-risk papillomas with normal skin and SCC. Unsupervised clustering of 514 differentially expressed genes (P<0.001) showed that 9/10 high-risk papillomas clustered with SCC, while 1/10 clustered with low-risk papillomas, and this correlated with keratin markers of tumor progression. Prediction analysis for microarrays (PAM) identified 87 genes that distinguished the two papilloma classes, and a majority of these had a similar expression pattern in both high-risk papillomas and SCC. Additional classifier algorithms generated a gene list that correctly classified unknown benign tumors as low- or high-risk concordant with promotion protocol and keratin profiling. Reduced expression of immune function genes characterized the high-risk papillomas and SCC. Immunohistochemistry confirmed reduced T-cell number in high-risk papillomas, suggesting that reduced adaptive immunity defines papillomas that progress to SCC. These results demonstrate that murine premalignant lesions can be segregated into subgroups by gene expression patterns that correlate with risk for malignant conversion, and suggest a paradigm for generating diagnostic biomarkers for human premalignant lesions with unknown individual risk for malignant conversion.