Increased midkine expression correlates with desmoid tumour recurrence: a potential biomarker and therapeutic target.

Desmoid tumours (DTs) are soft tissue monoclonal neoplasms exhibiting a unique phenotype, consisting of aggressive local invasiveness without metastatic capacity. While DTs can infrequently occur as part of familial adenomatosis polyposis, most cases arise sporadically. Sporadic DTs harbour a high prevalence of CTNNB1 mutations and hence increased β-catenin signalling. However, β-catenin downstream transcriptional targets and other molecular deregulations operative in DT inception and progression are currently not well defined, contributing to the lack of sensitive molecular prognosticators and efficacious targeted therapeutic strategies. We compared the gene expression profiles of 14 sporadic DTs to those of five corresponding normal tissues and six solitary fibrous tumour specimens. A DT expression signature consisting of 636 up- and 119 down-regulated genes highly enriched for extracellular matrix, cell adhesion and wound healing-related proteins was generated. Furthermore, 98 (15%) of the over-expressed genes were demonstrated to contain a TCF/LEF consensus binding site in their promoters, possibly heralding direct β-catenin downstream targets relevant to DT. The protein products of three of the up-regulated DT genes: ADAM12, MMP2 and midkine, were found to be commonly expressed in a large cohort of human DT samples assembled on a tissue microarray. Interestingly, enhanced midkine expression significantly correlated with a higher propensity and decreased time for primary DT recurrence (log-rank p = 0.0025). Finally, midkine was found to enhance the migration and invasion of primary DT cell cultures. Taken together, these studies provide insights into potential DT molecular aberrations and novel β-catenin transcriptional targets. Further studies to confirm the utility of midkine as a clinical DT molecular prognosticator and a potential therapeutic target are therefore warranted. Raw gene array data can be found at: http://smd.stanford.edu/