A locus segregating with familial Alzheimer's disease (AD) has been mapped to chromosome 21, close to the amyloid precursor protein (APP) gene. Recombinants between the APP gene and the AD locus have been reported which seemed to exclude it as the site of the mutation causing familial AD. But recent genetic analysis of a large number of AD families has demonstrated that the disease is heterogeneous. Families with late-onset AD do not show linkage to chromosome 21 markers. Some families with early-onset AD show linkage to chromosome 21 markers, but some do not. This has led to the suggestion that there is non-allelic genetic heterogeneity even within early onset familial AD. To avoid the problems that heterogeneity poses for genetic analysis, we have examined the cosegregation of AD and markers along the long arm of chromosome 21 in a single family with AD confirmed by autopsy. Here we demonstrate that in this kindred, which shows linkage to chromosome 21 markers, there is a point mutation in the APP gene. This mutation causes an amino-acid substitution (Val----Ile) close to the carboxy terminus of the beta-amyloid peptide. Screening other cases of familial AD revealed a second unrelated family in which this variant occurs. This suggests that some cases of AD could be caused by mutations in the APP gene.

We have analyzed a panel of 14 cases of childhood adrenocortical tumors unselected for family history and have identified germline TP53 mutations in >80%, making this the highest known incidence of a germline mutation in a tumor-suppressor gene in any cancer. The spectrum of germline TP53 mutations detected is remarkably limited. Analysis of tumor tissue for loss of constitutional heterozygosity, with respect to the germline mutant allele and the occurrence of other somatic TP53 mutations, indicates complex sequences of genetic events in a number of tumors. None of the families had cancer histories that conformed to the criteria for Li-Fraumeni syndrome, but, in some families, we were able to demonstrate that the mutation had been inherited. In these families there were gene carriers unaffected in their 40s and 50s, and there were others with relatively late-onset cancers. These data provide evidence that certain TP53 alleles confer relatively low penetrance for predisposition to the development of cancer, and they imply that deleterious TP53 mutations may be more frequent in the population than has been estimated previously. Our findings have considerable implications for the clinical management of children with andrenocortical tumors and their parents, in terms of both genetic testing and the early detection and treatment of tumors.

We have constructed a physical map of the region homozygously deleted in the U2020 cell line at 3p12, including the location of putative CpG islands. Adjacent to one of these islands, we have identi®ed and cloned a new gene (DUTT1) and used probes from this gene to detect two other homozygous deletions occurring in lung and breast carcinomas: the smallest deletion is within the gene itself and would result in a truncated protein. The DUTT1 gene is a member of the neural cell adhesion molecule family, although its widespread expression suggests it plays a less specialized role compared to other members of the family.

Comparative genomic hybridisation has been used to map copy number changes in nine cases of ductal carcinoma in situ of the breast obtained from waxembedded archive material. A wide variety of abnormalities were detected including gain of regions of 1q, 17q, 19q, 20p and 20q and loss on 13q, 14q, 17p, 16q and 22q. Ampli®cation of areas on 10p, 8q and 20q were also observed. Chromosomal alterations were more frequent in higher grade DCIS and closely resemble those previously detected in invasive breast cancer using the same technique. These data provide strong molecular support for the view that DCIS is a precursor lesion of invasive breast carcinoma.