Genome-wide search for breast cancer linkage in large Icelandic non-BRCA1/2 families

Abstract

Introduction: A significant proportion of high-risk breast cancer families are not explained by mutations in known genes. Recent genome-wide searches (GWS) have not revealed any single major locus reminiscent of BRCA1 and BRCA2, indicating that still unidentified genes may explain relatively few families each or interact in a way obscure to linkage analyses. This has drawn attention to possible benefits of studying populations where genetic heterogeneity might be reduced. We thus performed a GWS for linkage on nine Icelandic multiple-case non-BRCA1/2 families of desirable size for mapping highly penetrant loci. To follow up suggestive loci, an additional 13 families from other Nordic countries were genotyped for selected markers.

Methods: GWS was performed using 811 microsatellite markers providing about five centiMorgan (cM) resolution. Multipoint logarithm of odds (LOD) scores were calculated using parametric and nonparametric methods. For selected markers and cases, tumour tissue was compared to normal tissue to look for allelic loss indicative of a tumour suppressor gene.

Results: The three highest signals were located at chromosomes 6q, 2p and 14q. One family contributed suggestive LOD scores (LOD 2.63 to 3.03, dominant model) at all these regions, without consistent evidence of a tumour suppressor gene. Haplotypes in nine affected family members mapped the loci to 2p23.2 to p21, 6q14.2 to q23.2 and 14q21.3 to q24.3. No evidence of a highly penetrant locus was found among the remaining families. The heterogeneity LOD (HLOD) at the 6q, 2p and 14q loci in all families was 3.27, 1.66 and 1.24, respectively. The subset of 13 Nordic families showed supportive HLODs at chromosome 6q (ranging from 0.34 to 1.37 by country subset). The 2p and 14q loci overlap with regions indicated by large families in previous GWS studies of breast cancer.

Conclusions: Chromosomes 2p, 6q and 14q are candidate sites for genes contributing together to high breast cancer risk. A polygenic model is supported, suggesting the joint effect of genes in contributing to breast cancer risk to be rather common in non-BRCA1/2 families. For genetic counselling it would seem important to resolve the mode of genetic interaction.

Figure 1

Maximum LOD scores by chromosomal position, and relation to previously suggested candidate loci. Parametric HLOD scores for the nine Icelandic families are shown in (a) for the dominant (dark teal thick line) and the recessive model (plum). NP-LOD scores are shown in (b) using different exponential scoring options in Merlin software: S-all (orange thick line) and S-pairs (indigo). The position of previously published loci is shown in (c), according to GWS studies in red (or black if reported in more than one GWS study) (adapted from Table 5 in [16]), and according to single locus reports [17-19] in blue. A position at 2p indicated by a subset of relatively early-onset multiple-case families in one GWS-study [11] is included in (c) and shown in grey. Parametric LOD scores of family 70234 are shown in (d) with line colours as in (b).

Figure 2

Cosegregation of haplotypes at three chromosomal regions in family 70234. On top, the pedigree of this family is shown with circles denoting females and boxes males, with red filling denoting diagnosis of BC and shaded red also ovarian cancer. The pedigree is somewhat distorted in order to avoid recognition, but preserves the number of male and female meioses. Information about approximate age (in years) at diagnosis of cancer is shown below symbols (Br for breast and Ov for ovarian). One woman inherited a 999del5 BRCA2 mutation from her father (red box), not otherwise blood-related to this family. This woman and one other (denoted by not GWS) were not included in the genome-wide search. One of a pair of identical twins was omitted in linkage calculations. Under the pedigree, genotypes of markers of interest are shown (with chromosomal band position, marker names and genomic distance shown at the left, brackets indicating position between sites of relevant recombination). Colouring of alleles denotes whether they belong to a shared haplotype (allele frequency shown in the rightmost column) or derive from a recombined chromosome (blue). Plain (not coloured) alleles denoted 11 and 3 at D2S367 and D2S2163 respectively, are identical by state, but probably not by descent, to the commonly segregated allele (conclusion supported by fine-genotyping of additional markers, data not shown). The low frequency (0.06) of the shared D6S434 allele was validated by typing in 59 unrelated Icelandic control individuals. This figure was 0.02 in the controls (2/118), which coincides with the published Genethon frequency of the allele.