Statistics from Altmetric.com
We refer to the recent paper by Durno et al “Family history and molecular features of children, adolescents, and young adults with colorectal carcinoma” (Gut 2005;54:1146–50).
Among other patients with early onset colorectal cancer (CRC), the authors discuss a girl with CRC onset at the age of 12 years, and a subsequent second primary tumour (glioblastoma). This patient appears to be the sister of a male Turcot syndrome patient described in 1995.1 However, although referred to in that original report, she was not explicitly stated to carry the same PMS2 mutation (R134X) as her brother, and so whether her mutation was assumed or verified by Durno et al is unclear.
More importantly, the authors have overlooked the fact that both of these siblings have since been shown to be compound heterozygotes for the PMS2 mutations R134X and 2184delTC.2 The fact that these patients have germline mutations on both alleles offers an explanation for the earlier observation of a high degree of microsatellite instability (MSI) in their constitutional DNA, not just in tumour DNA.3
We have previously pointed out that in the childhood cancer families so far described, PMS2 mutations behave recessively, the parents of the homozygous or compound heterozygous patients being, by and large, tumour free.2
While other recent publications do suggest that heterozygous PMS2 mutations can predispose to CRC,4–7 these mutations appear, for unknown reasons, to be less penetrant than “classical” hereditary non-polyposis colorectal cancer alleles of, for example, MLH1 or MSH2. Certainly, on the present evidence, a heterozygous PMS2 mutation is unlikely to account for a childhood cancer presentation.
Indeed, given that one of the other cases described by Durno et al was also homozygous for an MLH1 mutation, we would like to emphasize again that a childhood (as opposed to early adult) presentation of CRC or other MSI positive tumour should signal the possibility of a biallelic mismatch repair defect. Demonstration of MSI in constitutional DNA would in such circumstances point directly to the likelihood of biallelic mutation of one of the mismatch repair genes, with important implications for genetic management.
Conflict of interest: None declared.