We have reviewed the evidence for CHEK2 as a cause of Li–Fraumeni syndrome (LFS) in view of the continued debate about the cancer spectrum associated with mutations in this gene.1 In 1969 Li and Fraumeni reported four families with an autosomal dominant predisposition to a range of cancer types in children and young adults, including soft tissue sarcoma and breast cancer.2 In 1988, they extended their analysis to 24 families to characterise better the phenotype of the syndrome.3 These families contained greater than expected numbers of soft tissue sarcomas, osteosarcomas, breast cancers, childhood adrenocortical carcinomas, brain tumours and leukaemia. This group of cancers was soon accepted as the formal criteria for the clinical definition of classical LFS (box 1).

In 1990 Malkin et al4 were able to show that the majority of families with classical Li–Fraumeni have mutations in the TP53 gene and this seminal finding was quickly confirmed by others.5^–7 Based on current analyses, which include screening the TP53 gene with multiple ligation-dependent probe amplification (MLPA), the mutation prevalence nears 80%. In 1994, Birch et al8 proposed the definition of a second class of families, which do not meet the stringent criteria for LFS, but nevertheless who were candidates for TP53 screening (box 1). In families of this type (Li–Fraumeni-like, or LFL) the prevalence of TP53 mutations is about 30–40%.

In 1999, Bell and colleagues claimed that the CHEK2 gene might also be a cause of the cancer aggregations in families that meet the criteria for LFS or LFL9 They identified a CHEK2 mutation in one family with classical LFS, previously reported by us as negative for TP53 mutation,8 and in two families with features suggestive of LFS. Sarcomas were present in two of the three families. In the LFS family from our centre with the CHEK2 mutation (1100delC), a previously unaffected individual who did not carry the CHEK2 mutation has since developed premenopausal breast cancer. Her mother and two of her siblings, who were mutation carriers, also had premenopausal breast cancer. This latest event suggests that other, possibly additional, genes may be involved in this family, the only classical LFS family to date reported to carry a CHEK2 mutation.

Over the last 8 years, much data have accumulated on the clinical expression of CHEK2 mutations and we should therefore re-evaluate the original claim. There is one predominant founder mutation in CHEK2 (1100delC) which is carried by 0.7% of the Northern European population. In contrast, the LFS is believed to affect one in 20 000 births—or approximately 100 times less.10 If Bell and colleagues are correct, would we not expect the number of CHEK2 mutations in LFS to greatly exceed the number of TP53 mutations? There are two relevant questions here: (1) do families with CHEK2 mutations manifest the cancers which are associated with the classical LFS (or LFL); and (2) are the cancers found to be associated with CHEK2 mutations also found in LFS? A detailed analysis of 67 CHEK2 families11 failed to identify an increased risk of cancer for any site other than breast cancer—notably, no increases in sarcomas or adrenocortical cancers were seen. A study of 4008 Polish cancer cases found excess risks for cancers of the prostate, breast, colon, kidney and thyroid among carriers of CHEK2 mutations1 12—none of these sites (other than breast) is featured in LFS or LFL.

Based on these and other analyses we believe that CHEK2 mutations are not associated with the cancer types seen in the LFS or LFL (other than breast cancer) and it is no longer reasonable to consider CHEK2 mutations to be a cause of LFS.