From Journal of Medical Genetics
Clinical Impact of Unclassified Variants of the BRCA1 and BRCA2 Genes
Posted: 11/28/2011; J Med Genet. 2011;48(11):783-786. © 2011 BMJ Publishing Group
processing....
- Abstract and Introduction
Abstract and Introduction
Abstract
Women who carry a pathogenic mutation in BRCA1 or BRCA2 have high risks of developing breast and ovarian cancers. The functional effect of many missense variants on BRCA1 and BRCA2 protein function is not known. Here, the authors construct a historical cohort of 4030 female first-degree relatives of 1345 unselected patients with ovarian cancer who have been screened for BRCA1 and BRCA2 mutations. The authors compared the risks by the age of 80 years for all cancers combined in female first-degree relatives of women with a pathogenic mutation, women with a variant of unknown significance (unclassified variant) and non-carriers. The cumulative risk of cancer among the relatives of patients with a pathogenic mutation was much higher than the risk in relatives of non-carriers (50.2% vs 28.5%; HR=2.87, p<10−4). In contrast, the cumulative risk of cancer among relatives of patients carrying an unclassified variant was similar to the risk of cancer for relatives of non-carriers (27.6% vs 28.5%; HR=1.08, p=0.79). The authors used three different algorithms to predict the pathogenicity of unclassified variants and compared their penetrance with non-carriers. In this sample, only Align Grantham Variation Grantham Deviation appeared to predict penetrance based on first-degree relatives.Introduction
Identifying women who carry pathogenic mutations in BRCA1 or BRCA2 is important for cancer prevention and individualised treatment.[1] However, there are many different genetic variants in BRCA1 and BRCA2 in mixed populations, and it is challenging to distinguish pathogenic variants from benign ones. In most cases, frame-shift insertions, frame-shift deletions and nonsense mutations result in the introduction of a premature stop codon, which in turn results in a truncated protein. These mutations can be predicted to be pathogenic. Missense variants result in the substitution of one amino acid for another and may or may not impair protein function. Missense variants constitute a large proportion (46%) of the coding variants of BRCA1 and BRCA2 in the Breast Cancer Information Core (BIC) Database (http://research.nhgri.nih.gov/projects/bic/). A few missense variants have been judged to be deleterious based on the observed frequencies in cases and controls and on functional assays. If a missense variant is equally common in case and control populations, it is considered to be a normal non-pathogenic variant. However, if the variant is rare and there is insufficient epidemiological information to classify it one way or the other, it is likely to be categorised as an unclassified variant (ie, potentially deleterious). Deciding on the clinical importance of an individual unclassified variant is a challenge for oncologists and genetic counsellors. In this study, we investigate the clinical importance of unclassified variants as a group by comparing the risks of cancer in the first-degree relatives of carriers of unclassified variants with the risks for known pathogenic mutations and for non-carriers.[2] The study was approved by the ethics review board of Women's College Research Institute. All study subjects provided signed informed consent prior to participation.We recorded detailed family histories of cancer, including the ages and sites of diagnosis of all cancers in the male first-degree relatives and female first-degree relatives of 1345 patients with ovarian cancer (probands) who were unselected for age of diagnosis (56.1±12.3 years) or family history. The germline DNA of the 1345 patients was previously screened for BRCA1 and BRCA2 mutations.[3] All of the patients were residents of Ontario, Canada, and were diagnosed as having invasive ovarian cancer between 1995 and 2004. There were 176 carriers of 110 different known pathogenic mutations. Of the 110 pathogenic mutations, 102 were protein-truncating mutations (including 14 large insertions/deletions), 4 were splicing mutations (BRCA1 IVS4−1G→T, BRCA1 IVS15+1G→A, BRCA1 IVS16+6T→C and BRCA2 9345G A) and 4 were missense mutations (BRCA1 Cys61Gly, BRCA1 Arg1495Met, BRCA1 Arg1699Trp and BRCA2 Glu2663Val). The pathogenicity of the missense variants was based on annotations in the BIC Database and Myriad Genetics Inc. (Salt Lake City, USA) published data.[4] There were also 11 carriers of nine benign polymorphic missense variants (BRCA1 Glu1214Lys, BRCA1 Pro1238Leu, BRCA1 Val1736Ala, BRCA1 Pro1859Arg, BRCA2 Lys2729Asn, BRCA2 Lys2950Asn, BRCA2 Val2969Met, BRCA2 Tyr3098His and BRCA2 Ile3412Val). These were classified as benign according to Myriad Genetics' published data.[4]
In addition, there were 32 carriers of 22 other missense variants (15 in BRCA1 and 7 in BRCA2) with unknown functional effects (unclassified variants; Table 1). The functional effects of these 22 unclassified variants were predicted using the bioinformatic tools Polymorphism Phenotyping (PolyPhen)-2,[5] Sorting Intolerant from Tolerant (SIFT)[6] and Align Grantham Variation Grantham Deviation (Align-GVGD).[7] Of these 22 missense variants, 12 were predicted to be pathogenic by each of the PolyPhen-2 and SIFT (73% concordance with κ statistics of 0.45) tools. PolyPhen-2 predicted 11 variants to be probably pathogenic and only 1 variant to be possibly pathogenic. Only three variants were predicted to be pathogenic by Align-GVGD (class C45 and higher). Class C45 was chosen as a cut-off point for Align-GVGD based on receiver operating characteristic curve analysis using BRCA1 and BRCA2 missense variants with a known functional effect reported in the BIC Database or in Myriad Genetics' published data.[4] The chosen cut-off point corresponded to 90% specificity and 50% sensitivity. Available library alignments from human to fish in the Align-GVGD website were used for predicting the functional effect by Align-GVGD. There was a 59% (κ statistic=0.18) and 55% (κ statistic=0.09) concordance between Align-GVGD and each of the SIFT and PolyPhen-2 tools, respectively.
A historical cohort, consisting of the 4030 female first-degree relatives of all of the patients, including 502 female first-degree relatives of carriers of pathogenic mutations, 103 female first-degree relatives of carriers of unclassified variants and 3425 female first-degree relatives of non-carriers (including those with benign polymorphic missense variants), was constructed. Each female first-degree relative was analysed as if she were a study subject and was followed from birth until the occurrence of any type of cancer, death from another cause or the date of the study interview. The cumulative incidence of cancer among the female first-degree relatives of carriers of unclassified variants was compared with that of the relatives of probands with pathogenic mutations and with that of the relatives of non-carriers, using the Cox proportional hazards model. Of 4030 total female first-degree relatives, 565 (14%) had been diagnosed as having any cancer, and 220 (5.5%) of them had breast, ovarian or fallopian tube cancer. The cumulative hazard functions of the female first-degree relatives for the three subcohorts are shown in figure 1. The cumulative risks of being affected by any cancer, as well as by only breast, ovarian or fallopian tube cancer, by the age of 80 years for each cohort group are given in Table 2 with the corresponding HRs (compared to the female first-degree relatives of non-carriers). The cumulative risk of getting cancer by the age of 80 years for the female first-degree relatives of carriers of pathogenic mutations was 50.2%, compared to 28.5% for the female relatives of non-carriers (HR 2.87, p<10−4). The cumulative risk of cancer for the female first-degree relatives of carriers of any of the 22 unclassified variants was 27.6%, which was similar to the risk for the relatives of non-carriers (28.5%; HR 1.08, p=0.79). The same pattern was seen for the female first-degree relatives with only breast, ovarian or fallopian tube cancer (Table 2). Cumulative incidence functions were constructed for male first-degree relatives (4012 individuals with 385 cancers), and there was no difference between the three subcohorts of relatives of carriers of pathogenic mutations and carriers of unclassified variants with non-carriers (data not shown).
 | Figure 1. The cumulative risk of cancer in female first-degree relatives of ovarian cancer patients based on their BRCA1 and BRCA2 mutation status. |
[ CLOSE WINDOW ]
Figure 1.
The cumulative risk of cancer in female first-degree relatives of ovarian cancer patients based on their BRCA1 and BRCA2 mutation status.
We then calculated the HR of getting cancer among the female first-degree relatives of carriers of unclassified variants predicted to be pathogenic and non-pathogenic for each of the three algorithms, compared to that of non-carriers (Table 3). In our sample, only Align-GVGD appeared to discriminate penetrance based on first-degree relatives. Align-GVGD predicted three of the 22 unclassified variants (Table 1) to be dysfunctional. The cumulative risk of breast, ovarian or fallopian tube cancer among the female first-degree relatives (figure 2) of carriers of the three predicted pathogenic variants by the age of 80 years (53.3%; HR 5.24, p value vs controls=0.005; Table 3) was comparable to that for carriers of truncating mutations (40.2%; HR 5.08, p value vs controls <10−4; Table 2).
 | Figure 2. The cumulative risk of breast, ovarian or fallopian tube cancer in female first-degree relatives of ovarian cancer patients based on their BRCA1 and BRCA2 mutation status. The functional effect of unclassified variants was predicted by the Align Grantham Variation Grantham Deviation algorithm. |
[ CLOSE WINDOW ]
Figure 2.
The cumulative risk of breast, ovarian or fallopian tube cancer in female first-degree relatives of ovarian cancer patients based on their BRCA1 and BRCA2 mutation status. The functional effect of unclassified variants was predicted by the Align Grantham Variation Grantham Deviation algorithm.
These analyses suggest that the majority of the unclassified variants of the BRCA1 and BRCA2 genes, particularly those not involving splice site alterations, are unlikely to increase the risk of cancer. It is possible that some unclassified variants are dysfunctional, with a penetrance similar to that associated with truncating mutations. Our analysis suggests that in silico prediction algorithms such as Align-GVGD[6] with using a stringent cut-off point could potentially be used for distinguishing pathogenic mutations from the rest of the unclassified missense variants in BRCA1 and BRCA2. However, these tools should be used cautiously in clinics until their accuracy has been confirmed in other studies. Our data also call into question the value of commonly used tools such as PolyPhen and SIFT in predicting the functional effect of unclassified variants. However, we should note that the better performance of Align-GVGD in predicting the functional effect of unclassified variants in BRCA1 and BRCA2 genes is probably due to its development by a group of BRCA investigators using a manually curated alignment with a broad phylogenetic scope for these two genes. Align-GVGD may or may not function better than the other algorithms when used for other genes.
Our data suggest that only a very small proportion of actual deleterious mutations in BRCA1 or BRCA2 fall into the category of 'unclassified variant' and should be largely reassuring to women who have been found to have an unclassified missense variant of BRCA1 or BRCA2. If we accept the Align-GVGD classifier as accurate, then the number of deleterious mutations in this sample of 1345 women increases from 176 to 179, and an unclassified missense mutation is probably the cause of cancer in 3 of 1345 patients with ovarian cancer (or 0.2% of the entire sample).
No hay comentarios:
Publicar un comentario