Milea Timbergen

70 mutated tumours with WT tumours and found that CCND1 was expressed more frequently in β-catenin mutated DTF 16 . Saito et al. found a statistically significant correlation between nuclear staining of β-catenin and CCND1 overexpression 16, 40 . Although the literature on Wnt target expression in desmoids is limited and somewhat contradictory, both AXIN2 and CCND1 seem to be overexpressed in DTF. Interestingly, by studying genotype-phenotype correlations of CTNNB1 mutations in liver cancer, others have found that T41 mutations were associated with moderate activity whereas S45 mutations led to a weak β-catenin activation which could be compensated by S45 mutant allele duplication 29 . None of our hierarchical cluster analyses, based on the expression of different Wnt target gene sets, discriminated CTNNB1 mutated (S45 and T41) from WT DTF samples, neither were S45F and T41A clearly separated. A possible confounder in our analyses is that the WT group, now defined by the absence of mutations in exon 3 of CTNNB1 , needs to be screened more thoroughly for CTNNB1 mutations. It was demonstrated that a fair part of the DTF, designated as WT by Sanger sequencing, contain low frequency CTNNB1 mutations that are only detected by whole exome sequencing or contain novel intra-genic deletions of the CTNNB1 gene 19, 41 . Moreover, in the WT group, other genomic alterations occurred like APC loss, chromosome 6 loss and BMI1 mutations which are all linked to Wnt/β-catenin activation 19 . In our analyses, we could not distinguish between WT and S45F/T41A mutant DTF corresponding to observations of Crago et al. 19 . In contrast, Colombo et al. reported that CTNNB1 -mutated and WT DTF had different gene expression profiles 42 . In our cluster analyses, distinct clusters of samples were discerned based on the expression levels of putative Wnt target genes that were not related to CTNNB1 mutational status. Unfortunately, we were not able to identify the origin of the cluster pattern as it could not be explained by factors like sex, age at the time of diagnosis, tumour size and tumour status (primary or recurrent tumour). Other clinicopathological factors, not included in this study, for example immune cell infiltration could potentially explain the observed cluster pattern and could be the subject of future studies. Interestingly in this respect is a recent publication from Colombo et al. describing that T41A and S45F mutated DTF displayed a different expression of inflammation related genes 42 . No indications were found in this study for a differential Wnt/β-catenin signalling activity between the T41A and S45F mutants thereby failing to explain the reported different clinical behaviour of the two DTF mutant groups. A more thorough investigation into the molecular consequences of different CTNNB1 mutations is necessary, focusing on protein-protein interactions and identifying their genomic binding sites. Irrespective of the specific CTNNB1 3