Unclassified Variants in the Breast Cancer Susceptibility Genes BRCA1 and BRCA2

Genetic testing for mutations in the breast cancer susceptibility genes BRCA1 and BRCA2 was underway by the mid-1990 s and is now commonly performed. Important decisions regarding the clinical management of individuals from high-risk families are often made based on whether the proband carries a pathogenic variant or not. However, test results are often confounding. In addition to sequence variants that are highly likely to cause disease (for example, protein-truncating mutations that result from a number of different kinds of underlying sequence alterations), clinical mutation screening often reveals missense substitutions, potential splicing variants, and/or small in-frame insertion–deletion variants (indels) that are initially classified as variants of uncertain clinical significance (variously referred to as unclassified clinical variants, UCVs; variants of uncertain significance, VUSs; and unclassified variants

Appendix

We discuss classes of sequence variants and how they would be classified a priori in 11 examples.

• (1) If the proximal promoter of BRCA1 or BRCA2 is completely deleted, little or no mRNA would be produced. Such a deletion would be recognized as pathogenic. In contrast, the functional consequences of small insertions or deletions (indels), rearrangements, or point mutations within the promoter would be difficult to predict a priori; consequently, these types of variants would initially be reported as UVs.

• (2) In general, if one or more exons are entirely deleted, the mRNA splicing machinery will typically fuse the last exon before the deletion to the first exon after the deletion. If the length of the deleted exons is not a multiple of 3 bp, the resulting splicing event will create a shift in the mRNA reading frame; consequently, the deletion would be recognized as pathogenic. Conversely, if the length of the deleted exons is a multiple of 3 bp, the reading frame would be maintained and the resulting structural aberration would probably be reported as an UV. One can also reason through the analogous situation for duplications of internal exons.

• (3) The functional consequence of a smaller deletion that destroys either the splice acceptor at the beginning of an exon or the splice donor at the end of an exon is very difficult to predict. The splicing apparatus may use nearby sequences that resemble a bonafide splice acceptor or splice donor, or splicing may skip one or more exons. Accordingly, such variants are typically reported as UVs. Although exceedingly rare, duplication of a splice acceptor or splice donor would also be reported as an UV.

• (4) Indel mutations located entirely within an individual exon are more tractable. If the length of an indel is not a multiple of 3 bp, the consequence will most likely be a shift in the reading frame with the indel usually reported as pathogenic. Again, if the length of the indel is a multiple of 3 bp, the resulting mRNA reading frame will not be shifted and the indel would be reported as an UV.

• (5) Single-nucleotide substitutions occurring deep within introns are usually considered innocuous and are often not reported even when they are observed. However, there is always the possibility that such a variant will create or activate a splice junction; if sequence analysis indicated that this was the case, such a sequence variant would sometimes be reported as an UV.

• (6) Single-nucleotide substitutions falling within the splice acceptor consensus sequence, usually considered to extend from ˜ 20 bp before the exon to 3 bp into the exon, may well interfere with splicing. Software exists to predict whether such substitutions will affect the targeted splice acceptor (for review, see [75]). However, even if splicing to the targeted acceptor is completely abrogated, it is difficult to predict the final consequence to the mRNA structure and thus the protein. Consequently, if splice acceptor analysis algorithms predict that a substitution in the donor consensus is very unlikely to affect splice junction usage, the variant could be considered innocuous. If the software reports a modest or more severe probability that the substitution will alter splice junction usage the substitution would typically be reported as an UV.

• (7) Single-nucleotide substitutions falling within the splice donor consensus sequence, usually considered to extend from 3 bp before the end of the exon to 6 bp into the intron, may well interfere with splicing. Analytically, the situation is essentially the same as for splice acceptor sequence variants.

• (8) The cores of the splice donor and acceptor consensus sequences are the nearly invariant GT and AG dinucleotides that mark the first two and last two nucleotides of most introns. Nucleotide substitutions to the canonical GT–AG dinucleotides will almost always disrupt splice junction usage. These are typically either reported as being UVs with an explanation that they are quite likely to be pathogenic and that the patient should submit a new sample from which an mRNA splicing experiment can be done, or else they are reported as pathogenic.

• (9) Single-nucleotide substitutions falling within an exon may be silent, missense, or nonsense. Silent substitutions are usually considered innocuous and are generally not reported unless they fall very near the ends of the exon, where they may interfere with either the splice acceptor or the splice donor. Nonsense substitutions are generally considered pathogenic. The main exception is if they fall in the last coding exon downstream of the relevant functionally important C-terminal protein domains. Indeed, BRCA2 harbors a common nonsense substitution, K3326X, located downstream of the nuclear localization signal, which has been shown in an association study to be essentially neutral with respect to breast cancer risk [12]. Missense substitutions, on the other hand, have highly variable effects. The missense allele by definition directs synthesis of a protein that differs from the canonical sequence at just one amino acid. There are regions in the BRCA1 and BRCA2 proteins where almost any missense substitution is well tolerated. Likewise, there are other functionally important domains where an appreciable fraction of missense substitutions are apparently pathogenic [26,33]. However, to date, it has not been possible to classify missense substitutions at the time of their initial observation with sufficient certainty for clinical purposes; consequently, they are usually first reported as UVs. Moreover, the nucleotide substitution underlying a missense amino acid substitution can interfere with mRNA splicing as described for silent substitutions. Thus, this possibility must also be considered.

• (10) Exonic nucleotide substitutions or in-frame indels can create de novo splice donors or splice acceptors. Splice donor and acceptor consensus sequences are well enough understood that it should be possible to recognize such sequence variants with reasonable sensitivity, though the capacity to predict the impact on in vivo splicing of such an analysis is likely to be low. When a variant that could create a de novo splice junction is observed, it would typically be reported as an UV with the annotation that it could effect splicing.

• (11) In addition to splice junction consensus sequences, there are also exonic splice enhancers that act to improve the efficiency of nearby splice junctions, exonic splice silencers that act to block usage of nearby potential splice junctions, and intronic analogs of both of these sequence elements. At this time, software that could be used to predict these elements does not have the capacity to predict those elements that are functionally important, let alone sufficient sensitivity and specificity to identify sequence variants that could alter the activity of such an element (for review, see [75]). Consequently, sequence variants that could alter the activity of such elements typically go unrecognized.