Abstract
It has been proposed that women carrying heterozygous mutations of the ATM gene could be at increased risk of developing breast cancer. However, data in the literature are contrasting and no firm conclusion has been reached. Our aim was to verify whether ATM inactivation could play a role in breast tumor development. Following the classical tumor suppressor inactivation scheme, tumors showing loss of heterozygosity (LOH) at the ATM locus should present an increased proportion of mutated ATM forms. We screened a cohort of 173 nonselected primary breast tumors for LOH in a 4 cM region at 11q23 spanning the ATM gene. We analyzed 25 tumors presenting LOH within the ATM locus for mutations in the ATM coding sequence using an RT-PCR-SSCP approach. Five patients were found to bear a coding missense variant, out of which four corresponded to a frequent polymorphism in exon 39. One patient presented a previously unreported variant in exon 19 (2614C\textgreaterT) resulting in a nonconservative change (Pro\textgreaterSer) at aa 872. This variant was not found in any of the other 172 patients nor in 63 healthy controls tested, indicating that it is a rare ATM variant. LOH involved the ATM wild-type allele in the tumor presenting variant 2614. However, because the ATM gene presents a relatively large number of rare coding polymorphism it is difficult, in the absence of familial data, to be conclusive on the significance of this variant. Searching for further variants in exons 19 and 39 in the whole set of 173 breast tumors, we found one tumor showing an acquired deletion of four bases in the ATM gene. Somatic mutations affecting the ATM gene thus seem rare in breast cancer. In our cohort of breast cancer patients, tumors presenting LOH at the ATM locus did not show an increased frequency of sequence variants. Furthermore, allelic imbalance profiles in a 4-cM region of chromosome arm 11q spanning the ATM locus revealed that hot spots of LOH were more likely to correspond to a region localized telomeric to the gene. Therefore, these data suggest that other target genes for genetic inactivation exist in the 11q23 region.