Purpose:

Because BRCA1 is a high-risk breast/ovarian cancer susceptibility gene, BRCA1 sequence variants of uncertain clinical significance (VUS) complicate genetic counseling. As most VUS are rare, reliable classification based on clinical and genetic data is often impossible. However, all pathogenic BRCA1 variants analyzed result in defective homologous recombination DNA repair (HRR). Thus, BRCA1 VUS may be categorized based on their functional impact on this pathway.

Experimental Design:

Two hundred thirty-eight BRCA1 VUS—comprising most BRCA1 VUS known in the Netherlands and Belgium—were tested for their ability to complement Brca1-deficient mouse embryonic stem cells in HRR, using cisplatin and olaparib sensitivity assays and a direct repeat GFP (DR-GFP) HRR assay. Assays were validated using 25 known benign and 25 known pathogenic BRCA1 variants. For assessment of pathogenicity by a multifactorial likelihood analysis method, we collected clinical and genetic data for functionally deleterious VUS and VUS occurring in three or more families.

Results:

All three assays showed 100% sensitivity and specificity (95% confidence interval, 83%–100%). Out of 238 VUS, 45 showed functional defects, 26 of which were deleterious in all three assays. For 12 of these 26 variants, we could calculate the probability of pathogenicity using clinical and genetic data, resulting in the identification of 7 (likely) pathogenic variants.

Conclusions:

We have functionally categorized 238 BRCA1 VUS using three different HRR-related assays. Classification based on clinical and genetic data alone for a subset of these variants confirmed the high sensitivity and specificity of our functional assays.

Translational Relevance

Germline loss-of-function (LoF) variants in BRCA1 (MIM 113705) strongly predispose women to hereditary breast and ovarian cancer. Therefore, sequence analysis of BRCA1 is routinely conducted in counselees from families with early-onset breast and/or ovarian cancer. Besides variants that clearly abolish protein function and are known to increase cancer risk, a large number of variants of uncertain significance have been identified for which the associated cancer risk is unknown. Information on functional consequences of these BRCA1 variants is important, because it will: (i) aid genetic counseling, as variants that lead to a functional defect are likely to confer high cancer risk, and (ii) guide the choice for optimal treatment strategies for patients with cancer carrying LoF variants, because BRCA1 deficiency renders tumors hypersensitive to platinum drugs and PARP inhibitors.

A large fraction of all hereditary breast and ovarian cancers (HBOC) is caused by heterozygous pathogenic BRCA1 variants (1), which greatly increase the risk of women to develop breast cancer or ovarian cancer at any point during their lives (2, 3). Consequently, patients with suspected HBOC are generally tested for the presence of germline BRCA1 variants. Besides known pathogenic and benign variants, this yields many variants for which the associated cancer risk is unknown [variants of uncertain clinical significance (VUS)]. This uncertainty complicates clinical management of variant carriers. Despite considerable international efforts to determine pathogenicity for these VUS using a multifactorial likelihood model (MLM) that combines clinical and genetic information (4), for most VUS, this is not successful (5). In the absence of sufficient genetic information, assessment of pathogenicity is depending on in silico analyses and assays that measure effects on gene functions (6). In silico analyses assume that changes in evolutionary conserved amino acids are most likely to be pathogenic, especially when they result in pronounced alterations in the chemical characteristics of these amino acids (7). Although this is true in general, prediction algorithms are inferior to well-validated functional assays (8, 9). Especially for BRCA1 and BRCA2, there are now several functional assays with high sensitivity and specificity that allow accurate assessment of the effects of VUS on gene function (10–13). Many of these assays measure effects of variants on homologous recombination repair (HRR) of DNA double-strand breaks, which is thought to be the main tumor-suppressor activity of BRCA1 and BRCA2. In fact, all known pathogenic variants of BRCA1 and BRCA2 analyzed lead to HRR deficiency. However, when functional assays are well validated, also less direct readouts such as effects on proliferation and transcription activation may be used (14, 15).

We have previously developed an assay system for the functional categorization of BRCA1 VUS (10). The assay system is based on complementation of Brca1-deficient mouse embryonic stem (ES) cells by human BRCA1 cDNA variants. Using this assay system, we have now functionally categorized 238 BRCA1 VUS, including all 232 BRCA1 missense and in-frame deletion VUS reported by the clinical genetic laboratories in the Netherlands and Belgium up to 2015. Specifically, we measured the effects of BRCA1 VUS and a validation set of 25 known benign and 25 known pathogenic BRCA1 variants on HRR using the HRR deficiency–targeted drugs cisplatin and olaparib and the direct repeat GFP (DR-GFP) HRR reporter (16). To investigate if these assays are able to detect deleterious VUS outside the RING and BRCT domains, we included three variants in the coiled-coil domain that are known to disrupt the interaction with PALB2 and attenuate HRR. As functional data alone are currently considered to be insufficient for VUS classification, we have also collected and analyzed available genetic and clinical data for deleterious VUS or VUS occurring in three or more Dutch or Belgian families.

BRCA1 VUS selection

In 2015, we used the LOVD BRCA1 database (https://databases.lovd.nl/shared/genes/BRCA1) to select all 232 BRCA1 missense and in-frame deletion VUS identified in counselees and reported by the clinical genetic laboratories in the Netherlands and Belgium for analysis in our functional assays. In addition, we selected VUS c.4198A>G, c.4220T>C, c.4232T>C, c.4952C>T, c.5359T>A, and c.5363G>A. In silico splice site prediction analysis was performed for all variants using Alamut Interactive Biosoftware (https://www.interactive-biosoftware.com/alamut-visual/).

Generation of human BRCA1 knockin ES cells

Using Quick-Change Lightning site-directed mutagenesis (SDM; Stratagene), tandem AarI restriction sites were introduced to allow excision and replacement of eight consecutive segments of the human BRCA1 RMCE (Recombinase-Mediated Cassette Exchange) vector (10). Of note, the BRCA1 cDNA sequence of this RMCE vector contains the SNPs c.2082C>T p.(Ser694Ser), c.3113A>G p.(Glu1038Gly), c.4308T>C p.(Ser1436Ser), and c.4837A>G p.(Ser1613Gly), which are known to be benign and which are frequently observed in the population (https://brcaexchange.org). Excised segments were replaced by synthetic gBlocks (IDT) containing BRCA1 variants using Gibson Assembly (17). Mutations and assembly boundaries were sequence verified, and introduction of human BRCA1 cDNAs via RMCE was conducted by cotransfection of R26CreERT2/RMCE;Brca1SCo/Δ;Pim1DR-GFP/wt ES cells with RMCE vectors and pCAGGs-Flpo-IRES-puro using Lipofectamine 2000 (Invitrogen; Thermo Fisher Scientific) as described (10). Cells that had successfully undergone RMCE were selected for 11 days using 200 to 800 μg/mL G418 and cryopreserved. Correct RMCE was confirmed by quantitative PCR (see Supplementary Materials and Methods), and expression of human BRCA1 was analyzed by Western blotting using a polyclonal antibody against human BRCA1 (9010; Cell Signaling Technology) as described (10). The ES cells used in this study tested negative for Mycoplasma, as determined by the MycoAlert PLUS Mycoplasma Detection Kit (Lonza).

Functional assays

For functional assays, cells were thawed and cultured in batches with 17 to 21 VUS pools, with 3 independently transfected human BRCA1 wild-type and empty RMCE vector control pools per batch. After 10 days, mouse Brca1 was switched off using 1-day (at least 24 hours) 0.5 μmol/L 4-OHT incubation. Subsequently, cells were cultured for 3 days in 50% CM/2i-medium (refs. 10, 18; see Supplementary Materials and Methods) and seeded for cytotoxicity or DR-GFP HRR assays. Three-day cisplatin and olaparib sensitivity assays were performed as described (10), with some modifications: cells were seeded in triplicate at 1,500 cells per well in 96-well plates and cultured in 50% CM/2i-medium. For DR-GFP HRR assays, 500,000 cells were seeded in 12-well plates for transfections with I-SceI-mCherry plasmid (10). Two days after transfection, mCherry/GFP double-positive cells were monitored by flow cytometry on a CyAn (Beckman Coulter) or LSRFortessa (BD Biosciences) cell analyzer. Data were analyzed using FlowJo software, including correction for flow aberrations using FlowAI (FlowJo LLC, BD Biosciences). Experiments were repeated twice.

Functional data analysis and statistics

For the analysis of cytotoxicity data, the raw fluorescence intensities of the cisplatin- or olaparib-treated wells were converted per plate to a cell viability value, with 0 being the mean of the empty wells and 1 being the mean of the no-drug wells. IC50 values were calculated by taking the tested drug concentrations around the IC50 and calculating the IC50 from these points in log-space, because the dose–response curve is estimated to be linear in log-space around the IC50. Samples that had a nonmonotonous curve course were excluded. In the DR-GFP assays, HRR activity was measured as the percentage of GFP-positive transfected cells.

For each series of experiments, data were normalized to three independently transfected wild-type human BRCA1 (WT) and empty RMCE vector (EV) controls. All samples and controls were tested in biological triplicates. Gaussian distributions were estimated for both the EV and WT controls. These two distributions were used to calculate P, the posterior probability for a variant to be deleterious (19) based on its mean-normalized IC50 or GFP percentage over the three biological replicates in the assay. For olaparib, the natural logarithm of the mean-normalized IC50 + 0.05 was used to estimate the distributions and to calculate P values. The prior probability of being deleterious was assumed to be equal (uninformative prior). The VUS were categorized as follows: P > 0.99: deleterious; 0.95 < P ≤ 0.99: likely deleterious; 0.05 < P ≤ 0.95: intermediate; 0.01 < P ≤ 0.05: likely neutral; P ≤ 0.01: neutral. Categorizations were considered uncertain in case the average values ± the SD over the biological repeats resulted in opposite outcomes [i.e., (likely) deleterious vs. (likely) benign]. To establish the sensitivity and specificity of our assays, we included 25 known pathogenic and 25 known neutral variants (20).

Multifactorial likelihood analysis

To increase the chance of an informative variant classification within the timeframe of our study, we selected 43 VUS that had been observed in at least three families and/or shown to impair protein function in the functional assays. Clinical and genetic data were collected as described (5). Information for cosegregation analysis was provided in the form of a deidentified pedigree for families with known carrier status in more than one individual. Pedigree details included gender, breast/ovarian cancer status, and age at cancer diagnosis, or age at interview if unaffected. Unaffected individuals known to have undergone prophylactic surgery (mastectomy or oophorectomy) were censored at age of earliest surgery. Breast tumor pathology information collected for known variant carriers included hormone receptor status (estrogen receptor, progesterone receptor, and human epidermal growth factor receptor 2) and/or tumor grade.

Likelihood ratios (LR) of pathogenicity were established on the basis of cosegregation of the variant with breast cancer in families of mutation carriers (21) and breast tumor data (age of diagnosis, grade, or hormone receptor status; ref. 22). Prior probabilities of pathogenicity were derived from Align GVGD (7) and MaxEntScan splicing predictions (23) and are available on http://priors.hci.utah.edu/PRIORS. To estimate prior probabilities for in‐frame deletion variants, we took the highest prior of the deleted bases (5).

Supplementary Table S4 summarizes the LRs assigned for each component for each variant with at least one data point. Prior probabilities and LRs were combined to calculate posterior probabilities: (Prior Probability × Combined LR)/(Prior Probability × [Combined LR + (1−Prior Probability)]). Where multiple data points were available for a single data type (e.g., cosegregation), LRs were combined multiplicatively. VUS were classified based on the five-tier classification system introduced by an International Agency for Research on Cancer Working Group (24).

Web resources

Generation of human BRCA1 complemented mouse Brca1SCo ES cells

As the role of BRCA1 in HRR appears to be crucial for its tumor-suppressor activity (10, 20, 25–29), we set out to functionally classify all known BRCA1 missense and in-frame deletion VUS reported in the Netherlands and Belgium based on their effect on this DNA repair pathway in Brca1-deficient mouse ES cells. For this purpose, we used cisplatin and olaparib sensitivity assays, as well as the DR-GFP reporter assay to measure HRR by gene conversion (16). In addition to the Dutch/Belgian VUS, we (re)analyzed three VUS, c.4198A>G p.(Met1400Val), c.4220T>C p.(Leu1407Pro), and c.4232T>C p.(Met1411Thr), that are known to attenuate the interaction of BRCA1 with PALB2 (10, 30). To validate our assays, we included 25 known pathogenic and 25 known benign variants for which the clinical significance has previously been established on the basis of clinical and genetic data (Supplementary Table S1; ref. 20 and the NIH ClinVar database). Because variant c.5359T>A was always observed together with variant c.5363G>A (likely in cis) when it was classified as pathogenic based on genetic data (20, 24, 31), we included the c.[5359T>A; 5363G>A] double mutation as a known pathogenic control.

Our previously described assay system (10) made use of SDM to introduce VUS in the BRCA1 cDNA sequence of an RMCE vector. Although this method worked sufficiently well before, the large size of the BRCA1 RMCE vector makes it relatively inefficient. In addition, when SDM is used, the complete BRCA1 cDNA needs to be sequence verified for absence of unwanted additional mutations. Therefore, we created modified BRCA1 RMCE vectors in which we could easily replace part of the BRCA1 cDNA sequence by a synthetic gene block containing a specific variant (Fig. 1). We generated a library of BRCA1 VUS and controls using Gibson assembly (17). As our previous categorizations were based on the cisplatin sensitivity assay only, we also included previously generated and analyzed VUS and controls (ref. 10; Supplementary Tables S1 and S2). The inserted gene blocks of all BRCA1 RMCE vectors were sequence verified before transfection into Brca1SCo ES cells containing the DR-GFP reporter (10). Before performing functional analyses, pools of transfected cells were assayed by qPCR for RMCE efficiency, human BRCA1 protein expression was confirmed by Western blotting, and the presence of the correct BRCA1 variant was verified using Sanger sequencing (Fig. 1; Supplementary Fig S1A and S1B, and data not shown). In addition, we assayed all VUS with Alamut Interactive Biosoftware for potential effects on mRNA splicing, which cannot be evaluated with our cDNA approach. Seven missense VUS are predicted to affect consensus splice sites, which may be deleterious (Supplementary Table S2): c.441G>C, c.670G>C, c.4096G>A, 4675G>A, c.5072C>T, c.5074G>A, and c.5074G>C. For five of these variants, there is also experimental evidence that they result in reduced or even abrogated levels of the normal BRCA1 transcript (Supplementary Table S2; refs. 14, 32–37). In addition, c.4675G>A and c.5074G>C have recently been classified as pathogenic by expert panel review (https://www.ncbi.nlm.nih.gov/clinvar/).

Figure 1.

Workflow for the functional categorizations of BRCA1 variants in mouse ES cells. Flp RMCE vectors containing a BRCA1 cDNA segment deletion were complemented with synthetic BRCA1 variant containing gene blocks by Gibson Assembly. Sequence verified vectors were transfected into R26CreERT2/RMCE;Brca1SCo/Δ;Pim1DR-GFP/wt mouse ES cells, and pools of cells that underwent correct RMCE were used in functional assays essentially as described (10).

Figure 1.

Workflow for the functional categorizations of BRCA1 variants in mouse ES cells. Flp RMCE vectors containing a BRCA1 cDNA segment deletion were complemented with synthetic BRCA1 variant containing gene blocks by Gibson Assembly. Sequence verified vectors were transfected into R26CreERT2/RMCE;Brca1SCo/Δ;Pim1DR-GFP/wt mouse ES cells, and pools of cells that underwent correct RMCE were used in functional assays essentially as described (10).

Close modal

Cisplatin sensitivity assay

Confirming its specificity, all 25 known benign variants were correctly categorized as functionally neutral in the cisplatin assay (Fig. 2A; Supplementary Table S1). In addition, all 25 known pathogenic variants—including 15 variants we had not analyzed before—were categorized as deleterious. The results from the cisplatin sensitivity assay (Fig. 2A; Supplementary Fig. S2A) confirmed the functional categorization of previously analyzed VUS (Supplementary Table S2; ref. 10). However, technical optimizations now allowed the functional categorization of several VUS and controls that had previously remained uncategorized due to inconsistent data or technological flaws, including the well-known pathogenic truncating founder mutation 5382insC (c.5266dup). Of note, similar to previously published findings (38), variant c.5359T>A by itself only had an intermediate effect on cisplatin sensitivity (Supplementary Table S2). The c.[5359T>A; 5363G>A] double mutation however was clearly deleterious (Supplementary Table S1), in line with the assumption that c.5359T>A is only pathogenic in cis with c.5363G>A. In total, 27 VUS were deleterious in the cisplatin assay (Table 1; Supplementary Table S2). For two additional VUS—c.5207T>C p.(Val1736Ala) and c.5348T>C p.(Met1783Thr)—the mean cisplatin response suggested functional defects, but these variants could not be categorized as neutral or deleterious because of experimental variation (Supplementary Fig. S3A–S3D). Both p.(Val1736Ala) (refs. 10, 39) and p.(Met1783Thr) (refs. 11, 29) have previously been shown to confer partial defects in BRCA1 function.

Figure 2.

Results of the functional analyses of BRCA1 variants. A, Normalized cisplatin IC50 values for BRCA1 wild-type (WT) controls (green), known benign variants (light-green), VUS (blue), known pathogenic variants (pink), and empty RMCE vector (EV) controls (red). Averages and SDs of biological repeats are shown as dots and shades, respectively. Horizontal solid and dotted gray lines mark (from top to bottom) 0.99 and 0.95 probability of being neutral and 0.95 and 0.99 probability of being deleterious. B, Normalized olaparib logeIC50 values, depicted as in A. C, Normalized percentages GFP-positive transfected cells as a measure of HRR, depicted as in A. a.u., arbitrary unit.

Figure 2.

Results of the functional analyses of BRCA1 variants. A, Normalized cisplatin IC50 values for BRCA1 wild-type (WT) controls (green), known benign variants (light-green), VUS (blue), known pathogenic variants (pink), and empty RMCE vector (EV) controls (red). Averages and SDs of biological repeats are shown as dots and shades, respectively. Horizontal solid and dotted gray lines mark (from top to bottom) 0.99 and 0.95 probability of being neutral and 0.95 and 0.99 probability of being deleterious. B, Normalized olaparib logeIC50 values, depicted as in A. C, Normalized percentages GFP-positive transfected cells as a measure of HRR, depicted as in A. a.u., arbitrary unit.

Close modal
Table 1.

BRCA1 VUS with functional defects in the cisplatin, olaparib, or DR-GFP HRR assay.

BRCA1 VUS with functional defects in the cisplatin, olaparib, or DR-GFP HRR assay.
BRCA1 VUS with functional defects in the cisplatin, olaparib, or DR-GFP HRR assay.

Olaparib sensitivity assay

As we observed before (10), the differences in IC50 for the PARP inhibitor olaparib between neutral and deleterious variants were much more pronounced than for cisplatin (Fig. 2B; Supplementary Fig. S2B and S2C; Supplementary Table S2). As the invariably low olaparib IC50 values for deleterious variants are accompanied by a large variation between repeat experiments for (partially) BRCA1-proficient variants, we used log-transformed olaparib IC50 values for functional categorization. This led to the identification of 33 clearly deleterious variants including 7 variants (c.5074G>C, c.5207T>C, c.5238C>G, c.5348T>C, c.5363G>A, c.5419A>T, and c.5512G>A) that were not identified as deleterious by the cisplatin assay. Fifteen variants, including c.134A>C which was deleterious in the cisplatin assay, showed intermediate or variable defects precluding categorization using the olaparib sensitivity assay (Supplementary Fig. S3E–S3H; Table 1; Supplementary Table S2).

Thus, the olaparib assay showed functional impairment for both hypomorphic variants that could not be categorized in the cisplatin assay (c.5207T>C and c.5348T>C). In addition, all deleterious variants identified by the olaparib assay are present in the functionally important RING or BRCT domains. Together, this suggests that the greater sensitivity of the olaparib assay does not come at the cost of reduced specificity. This is confirmed by the correct categorization of all known benign variants. Of note, also the c.5363G>A control for the known pathogenic c.[5359T>A; 5363G>A] double mutant showed loss of function in the olaparib assay.

HRR assay

It is known that the sensitivity of Brca1-deficient mouse ES cells to cisplatin or PARP inhibition is alleviated by restoration of HRR (10, 40). Conversely, it seems likely that variants failing to complement cisplatin or olaparib sensitivity share a negative impact on HRR. We used the well-established DR-GFP reporter (10, 16) to directly measure the effect of BRCA1 variants on HRR. This reporter expresses GFP upon HRR of a DNA double-strand break generated by the meganuclease I-SceI. Analysis by flow cytometry indeed showed that all variants identified as deleterious in the cisplatin or olaparib assays resulted in defective HRR (Table 1; Supplementary Table S2; Fig. 2C; Supplementary Fig. S2D). In addition, there were a number of variants which only scored as (likely) deleterious in the DR-GFP assay: c.41T>C, c.4220T>C, c.4232T>C, c.5002T>C, c.5062G>C, c.5071A>G, c.5074G>A, c.5128G>A, c.5225A>G, c.5300G>C, and c.5435C>G. Also c.134A>C, which was deleterious in the cisplatin assay but could not be categorized in the olaparib assay, scored as deleterious. Thus, the DR-GFP assay confirmed the deleterious nature of all variants with intermediate or variable defects in both drug sensitivity assays. Only two variants could not be categorized due to variable results (Supplementary Fig. S3I).

Comparison of the results from the three functional assays

In total, we analyzed 238 BRCA1 VUS and identified functional defects in 45 of them, 43 VUS in the RING and BRCT domains and 2 in the coiled-coil domain (Table 1; Supplementary Table S2). With all 25 known pathogenic variants and all 25 known benign variants correctly categorized in the cisplatin, olaparib, and DR-GFP HRR assays, each assay has a sensitivity and specificity of 100% [95% confidence interval (CI), 83%–100%]. Although the results of our assays were highly concordant (Fig. 3), a subset of variants was only scored as deleterious in the olaparib and/or DR-GFP assays (Fig. 4; Table 1; Supplementary Table S2). In most cases, these variants appear to be hypomorphic, conferring partial defects which were only significant in the HRR assay (Table 1; Supplementary Table S2; Supplementary Fig. S4A). Interestingly, and in line with our previous observations (10), the coiled-coil variants c.4220T>C p.(Leu1407Pro) and c.4232T>C p.(Met1411Thr) only affect HRR and have no significant effect on the (short-term) drug sensitivities of our complemented ES cells (Table 1; Supplementary Table S2; Supplementary Fig. S4B). In contrast to deleterious RING or BRCT variants, both deleterious coiled-coil variants are also able to support proliferation when mouse Brca1 is switched off (ref. 10; Supplementary Fig. S5A). These data indicate that the proliferation defect and drug sensitivity of BRCA1-deficient mouse ES cells are not caused by a defect in the completion of HRR per se.

Figure 3.

Comparison of the results of the three functional assays. The average normalized (n) outcome of the cisplatin (nIC50), olaparib (logenIC50), and DR-GFP (%GFP+) assays for each BRCA1 variant is plotted in a three-dimensional space. Data are shown for all VUS (blue) and known benign (green) and known pathogenic (red) variants analyzed.

Figure 3.

Comparison of the results of the three functional assays. The average normalized (n) outcome of the cisplatin (nIC50), olaparib (logenIC50), and DR-GFP (%GFP+) assays for each BRCA1 variant is plotted in a three-dimensional space. Data are shown for all VUS (blue) and known benign (green) and known pathogenic (red) variants analyzed.

Close modal
Figure 4.

Overlap between functional categorizations of BRCA1 VUS. Venn diagram showing the number of functionally deleterious BRCA1 VUS for each assay and the overlap between them. In total, 27 VUS showed a functional defect in the cisplatin assay, the olaparib assays identified 7 additional deleterious VUS (adding up to 34 in total), and 11 more deleterious VUS were identified by the DR-GFP assay (adding up to 45 in total).

Figure 4.

Overlap between functional categorizations of BRCA1 VUS. Venn diagram showing the number of functionally deleterious BRCA1 VUS for each assay and the overlap between them. In total, 27 VUS showed a functional defect in the cisplatin assay, the olaparib assays identified 7 additional deleterious VUS (adding up to 34 in total), and 11 more deleterious VUS were identified by the DR-GFP assay (adding up to 45 in total).

Close modal

Comparison with other functional assays

Our functional assays are well validated by the correct functional categorization of the 25 known pathogenic and 25 known benign variants and measure effects on the clinically relevant BRCA1 function in HRR. In addition, RMCE constructs as well as transfected ES cells were sequence verified and assayed for expression of human BRCA1 protein (Supplementary Fig. S1). Nevertheless, the possibility of miscategorization can never be completely excluded, also because of the intrinsically artificial nature of any functional assay system. It is therefore useful to compare the results of different well-validated functional assays. The recent publication of a saturation mutagenesis screen of BRCA1 RING and BRCT single-nucleotide variants in haploid LIG4-knockout HAP1 chronic myeloid leukemia cells (14) offers an excellent opportunity for such comparison. This high-throughput mutagenesis screen makes use of CRISPR/Cas9-assisted targeting of endogenous BRCA1 and is based on the finding that HAP1 cells do not tolerate BRCA1 loss (41). The results of the HAP1 proliferation assay are highly concordant with those of our assays, with all overlapping 16 known pathogenic variants and 6 known benign variants categorized correctly (Supplementary Table S3). Six known pathogenic variants and three VUS in the RING and BRCT encoding regions were not present in the HAP1 dataset. Most of the 64 shared VUS in the RING and BRCT encoding regions are categorized the same in all three of our assays and in the HAP1 screen. Ten discrepant cases may largely be attributed to a partial or intermediate loss of function that is only detected in the olaparib and/or DR-GFP assay(s) (Supplementary Fig. S4C; Supplementary Table S3). For two of these VUS, c.5074G>A and c.5074G>C, also defective mRNA splicing (Supplementary Table S2) may contribute to their deleterious nature in HAP1 cells. Two other variants show opposite categorization in all three of our assays compared to the HAP1 screen. Variant c.19C>T p.(Arg7Cys) appeared to have no effect on HAP1 proliferation, but was defective in all three of our assays. For c.5258G>C p.(Arg1753Thr), we noticed the opposite. Also in these cases, it seems likely that the variants have intermediate defects, that may vary depending on cell type or assay specific factors. Of note, upon transient transfection of these VUS in our ES cell system, DR-GFP HRR activity was only mildly affected by c.19C>T, whereas c.5258G>C leads to a strong HRR defect (Supplementary Fig. S5B). Cleary, there is a discrepancy between the results with stably integrated BRCA1 c.5258G>C versus transient expression, but sequence and Western blot analyses revealed no straightforward explanation (data not shown).

Interestingly, none of the 11 variants that were only identified as deleterious in the DR-GFP assay were picked up by the HAP1 proliferation screen. Of these, coiled-coil variants c.4220T>C and c.4232T>C were not included in the analysis. However, both variants supported mouse ES cell proliferation, suggesting that they would not have been deleterious in the HAP1 assay. Despite the well-described impact of coiled-coil domain inactivating variants on HRR (25, 30), it remains to be determined if variants that are only identified in our DR-GFP assay affect tumor suppression.

As observed before (10), in general the results of our assay system also correlate well with those of various other assays in which BRCA1 germline variants are analyzed individually for functional effects (25, 42, 43). However, multiplexing as used in the HAP1 assay clearly allows higher throughput and avoids batch or experimenter-dependent differences (14, 26). In principle, there are no barriers for parallel RMCE transfections and multiplexed analysis of BRCA1 variants in our assay system. To illustrate this, we performed parallel RMCE for 30 variants in the BRCA1 RING encoding region. Using high-throughput sequencing, we measured the relative abundance of individual BRCA1 variants before and after switching off endogenous mouse Brca1 and with or without adding olaparib. The results are highly concordant with those from the one-by-one analyses (Supplementary Fig. S6). Of note, variant c.134A>C showed some functional activity in one out of three individual olaparib assays but was clearly deleterious in the pooled proliferation and olaparib analyses (Fig. 6A–C; Supplementary Fig S3E–S3H).

VUS classification based on genetic and clinical data

Although the functional assays show high sensitivity and specificity, clinical classification requires multiple independent data sources. We set out to collect genetic and clinical data for VUS shown to impair protein function in the functional assays and VUS that had been observed in at least three families (Supplementary Table S4). In total, we collected data for 35 variants from 98 families. The recent publication of a large-scale MLM analysis of BRCA1 and BRCA2 VUS by the ENIGMA consortium (5) allowed us to add information for 18 of these variants, and obtain results for an additional set of 51 variants (Supplementary Table S4).

The multifactorial likelihood analysis method was used to calculate the posterior probability of pathogenicity associated with VUS (4, 44). This method incorporates data from, for example, cosegregation analysis and breast tumor pathology with a prior probability based on evolutionary conservation and chemical properties of the altered amino acids (7) or predicted effects on mRNA splicing (23). For the 35 VUS for which we collected clinical and/or genetic data, an LR for pathogenicity was calculated. A combined LR between 0.5 and 2.0, especially when derived from a limited number of data points, is assumed to provide insufficient data to perform reliable integrated analysis (23). Therefore, posterior probabilities of pathogenicity were not calculated for 9 of the 35 variants. As also 17 of the 51 variants only analyzed by Parsons and colleagues (5) yielded unreliable combined LRs, posterior probabilities for pathogenicity could be calculated for a total of 60 variants (Fig. 5; Supplementary Table S4). Nine variants are classified as (likely) pathogenic and 40 as (likely) benign using the five-tier classification scheme described in Plon and colleagues (24). For the 11 variants that remained VUS, posterior probability ranged from 0.948 to 0.071 (Fig. 5; Supplementary Table S4).

Figure 5.

Classification of BRCA1 VUS using multifactorial likelihood analysis. Posterior probability of pathogenicity was calculated based on MLM for variants functionally deleterious in all assays (red), only deleterious in the olaparib and/or DR-GFP assays (gray), or neutral (green). For variants scoring above the dotted red line, the posterior probability is larger than 0.95, indicating that these are (likely) pathogenic. Below the dotted green line variants are predicted to be (likely) benign. For most variants, the posterior probability of pathogenicity is too low to be visible on the graph. *, affects mRNA splicing; †, c.134A>C could not be classified in the individual olaparib assay, but was deleterious in the pooled olaparib assay; ‡, c.5258G>C was HRR defective upon transient transfection; 1, MLM results from this study; 2, MLM results from this study combined with Parsons and colleagues (5); 3, MLM results from Parsons and colleagues (5).

Figure 5.

Classification of BRCA1 VUS using multifactorial likelihood analysis. Posterior probability of pathogenicity was calculated based on MLM for variants functionally deleterious in all assays (red), only deleterious in the olaparib and/or DR-GFP assays (gray), or neutral (green). For variants scoring above the dotted red line, the posterior probability is larger than 0.95, indicating that these are (likely) pathogenic. Below the dotted green line variants are predicted to be (likely) benign. For most variants, the posterior probability of pathogenicity is too low to be visible on the graph. *, affects mRNA splicing; †, c.134A>C could not be classified in the individual olaparib assay, but was deleterious in the pooled olaparib assay; ‡, c.5258G>C was HRR defective upon transient transfection; 1, MLM results from this study; 2, MLM results from this study combined with Parsons and colleagues (5); 3, MLM results from Parsons and colleagues (5).

Close modal

Thirty-seven out of the 40 variants classified as (likely) benign were functionally neutral in all three of our assays. For three likely benign variants, the functional categorization was not consistent for the three assays. For c.134A>C, the categorization was not clear in the olaparib assay, whereas the variant was sensitive to cisplatin and defective in HRR (Supplementary Table S2; Fig. 4EH), whereas variants c.5062G>C and c.5419A>T only showed defects in the olaparib and/or HRR assays. Of note, for 20 variants predicted (likely) benign—including c.134A>C, c.5062G>C, and c.5419A>T—no cosegregation data were available and MLM classification was solely based on tumor pathology and prior probabilities (Supplementary Table S4).

Seven out of the nine variants predicted to be (likely) pathogenic were deleterious in all three of our assays. Both discrepant variants, c.4675G>A and c.5074G>C, are reported to affect mRNA splicing (33–37) and can therefore not be evaluated well in our cDNA assay. However, the amino acid change alone is already deleterious for c.5074G>C in the olaparib and DR-GFP assays. It should be noted that also c.5258G>C, for which we only observed an HRR defect upon transient transfections, almost reached the threshold for classification as likely pathogenic. For five unambiguously functionally deleterious variants, the clinical and genetic data were not sufficient to provide a clinical classification and those variants remained VUS (Fig. 5; Supplementary Table S4).

In total, the combined MLM approaches classified 40 VUS as (likely) benign and nine as (likely) pathogenic. These results correlated well with classifications based on our functional assays. Although concordance may be less for variants with functional defects in only the olaparib and/or DR-GFP assays, we did not analyze sufficient numbers of this category to allow firm conclusions.

In this work, we present the functional categorization of 238 BRCA1 VUS using complementation assays for cisplatin and olaparib sensitivity and homologous recombination. All three assays allow the identification of defects in DNA repair via HRR, either directly by the DR-GFP reporter or indirectly by HRR-defect targeted drugs. Loss-of-function variants in BRCA1 or BRCA2 confer high risk to breast and/or ovarian cancer, and the essential role of BRCA1 and BRCA2 in HRR is thought to be at the heart of their tumor-suppressor function. This is illustrated by the fact that thus far all known (likely) pathogenic variants affecting BRCA1, BRCA2, and their shared interaction partner PALB2 result in defective HRR (10, 13, 25–29, 45–47). We further optimized and validated our previously published mouse ES cell–based assay system, thereby increasing its robustness. All three assays achieved 100% sensitivity and specificity (95% CI, 83%–100%) based on the validation sets of 25 known pathogenic and 25 known benign variants.

Of the 238 BRCA1 VUS analyzed, 27 showed a clear functional defect in the cisplatin sensitivity assay we previously used for BRCA1 variant categorization (10). Confirming the tight relation between the three assays, these 27 variants were also identified as deleterious in the olaparib sensitivity and/or DR-GFP HRR assays. However, a subset of functionally deleterious VUS was only identified in the olaparib and/or DR-GFP assays and not in the cisplatin assay. In most of these cases, the functional defects seemed to be less severe than for variants that were deleterious in all three of the assays. Although there are indications that hypomorphic variants confer moderate risks of breast cancer (48), functional assays for BRCA1 VUS are currently not validated for their identification. Thus far, only BRCA1 c.5096G>A p.(Arg1699Gln) is known to confer intermediate loss of tumor suppression activity based on clinical and genetic data (49, 50), and this variant was deleterious in all of our assays.

It is also possible that some functionally deleterious BRCA1 variants are not identified in all three assays because they confer a different type of defect. In contrast to the deleterious RING and BRCT variants we tested, coiled-coil variants c.4220T>C p.(Leu1407Pro) and c.4232T>C p.(Met1411Thr) are defective in HRR but can still complement ES cells in proliferation and cisplatin or olaparib response. These coiled-coil variants lead to a specific type of HRR defect by disrupting the interaction with PALB2, which links BRCA1 and BRCA2 in the response to DNA damage (30). Apparently, this type of HRR defect has different consequences in different experimental settings, because p.(Leu1407Pro) and p.(Met1411Thr) do affect cisplatin and olaparib response of MDA-MB-436 breast cancer cells (25). At present, it is not clear if deleterious BRCA1 coiled-coil variants affect the tumor-suppressive capacity of BRCA1 to the same extent as deleterious RING or BRCT variants. In summary, all three of our functional assays are equally well validated. The cisplatin assay is most stringent, and it remains to be investigated if variants that are only deleterious in the olaparib and/or DR-GFP assays have milder effects on tumor predisposition.

Given the large number of BRCA1 VUS (currently over 2,000 listed in the NIH ClinVar database) and the current technical possibilities to modify endogenous genes using CRISPR/Cas9, it is clear that multiplexed saturated mutagenesis approaches as shown in HAP1 cells (14) hold great promise for future functional annotations. Although neither the HAP1 assay nor our mouse ES cell assays are likely to recapitulate all aspects of BRCA1-associated tumorigenesis, they show a near-perfect overlap in functional categorizations. The few discrepancies may be attributed to partial loss-of-function variants, of which the consequences for tumor suppression are not yet known.

Nevertheless, because of their inherent limitations, at this moment functional assays alone are not considered sufficient for the classification of VUS. Therefore, we collected available clinical and genetic data for a subset of variants. Furthermore, we took advantage of a recently published dataset from the ENIGMA consortium (5). For 86 VUS, we could calculate a combined LR for pathogenicity based on genetic data and tumor pathology, and for 60 of these VUS, the magnitude of the combined LRs was sufficient to perform a valid integrated evaluation (23). With the MLM, 49 VUS were classified as either (likely) pathogenic (9) or (likely) benign (40), and 11 variants remained of uncertain significance. Our functional data correlated well with the MLM classifications, although three variants that may be hypomorphic variants (i.e., they were not categorized as deleterious in all assays) were predicted likely benign by the MLM approach. It should be noted that the MLM is designed to distinguish high-risk variants from variants with no clinical significance and thus is less suitable to identify hypomorphic variants. Moreover, because there were no cosegregation data, MLM classification of these specific variants and 17 other predicted (likely) benign VUS depended heavily on prior probabilities derived from in silico analysis (7, 23). Although these are predictive, cosegregation analysis would have provided more direct evidence for the probability of disease association. Unfortunately, for the majority of the variants, we could not collect sufficient genetic and clinical data to classify them using the MLM.

As our analyses show, international data sharing through networks such as the ENIGMA consortium is important for the classification of BRCA1 VUS. In addition, the use of well-validated functional assays is incorporated in the guidelines of the American College of Medical Genetics and Genomics (6) and expected to become essential in the classification of the majority of BRCA1 VUS. We are convinced that in the near future the use of validated functional assays will allow the classification of the large majority of BRCA1 VUS, and thereby aid clinical decision-making. Besides facilitating genetic counseling, VUS classification may also guide personalized cancer therapy, as BRCA1-associated breast and ovarian tumors often show loss of the wild-type allele and consequent hypersensitivity to platinum drugs and PARP inhibitors.

L.F.A. Wessels reports project funding for his group (not for any aspect of the submitted work) from Genmab BV. No potential conflicts of interest were disclosed by the other authors.

Karin Segers, Center of Genetics, CHU de Liège, Liège, Belgium.

Cindy Badoer and Isabelle Vandernoot, Department of Genetics, Hôpital Erasme, ULB Center of Human Genetics, Université Libre de Bruxelles, Brussels, Belgium.

Pascale Hilbert, Center of Human Genetics, Institute of Pathology and Genetics, Gosselies, Belgium.

Katrien Storm and Bettina Blaumeiser, Center of Medical Genetics, University and University Hospital of Antwerp, Antwerp, Belgium

Kathleen Claes, Center for Medical Genetics, Ghent University Hospital, Ghent, Belgium

Sara Seneca, Centre for Medical Genetics, University Hospital Brussels, Vrije Universiteit Brussel (VUB), Brussels, Belgium.

Geneviève Michils, Centre for Human Genetics, KU Leuven, Leuven, Belgium.

Ans M.W. van den Ouweland and J.M. Collée, Department of Clinical Genetics, Erasmus University Medical Center, Rotterdam, The Netherlands

Nienke van der Stoep, Department of Clinical Genetics, Leiden University Medical Center, Leiden, The Netherlands.

Rien Blok, Department of Clinical Genetics, Maastricht University Medical Center, Maastricht, The Netherlands.

Fonnet E. Bleeker, Family Cancer Clinic, Netherlands Cancer Institute, Amsterdam, The Netherlands.

Frans B.L. Hogervorst, Family Cancer Clinic and Department of Pathology, Netherlands Cancer Institute, Amsterdam, The Netherlands.

Arjen R. Mensenkamp, Department of Human Genetics, Radboud University Nijmegen Medical Center, Nijmegen, The Netherlands.

Annemieke H. van der Hout and Jan C. Oosterwijk, Department of Genetics, University of Groningen, University Medical Center Groningen, Groningen, The Netherlands.

Rob van der Luijt* and Marco Koudijs, Department of Clinical Genetics, University Medical Center Utrecht, University of Utrecht, Utrecht, The Netherlands.

*Current affiliation: Department of Clinical Genetics, Leiden University Medical Center, Leiden, The Netherlands

P. Bouwman: Conceptualization, data curation, formal analysis, supervision, funding acquisition, validation, investigation, visualization, methodology, writing-original draft, project administration, writing-review and editing. I. van der Heijden: Data curation, formal analysis, investigation, project administration, writing-review and editing. H. van der Gulden: Data curation, formal analysis, investigation, project administration. R. de Bruijn: Data curation, software, formal analysis, visualization, writing-original draft. M.E. Braspenning: Data curation, formal analysis, investigation. S. Moghadasi: Formal analysis, funding acquisition, investigation. L.F.A. Wessels: Supervision, writing-review and editing. K. Segers: Resources. C. Badoer: Resources. I. Vandernoot: Resources. P. Hilbert: Resources. K. Storm: Resources. B. Blaumeiser: Resources. K. Claes: Resources. S. Seneca: Resources. G. Michils: Resources. A.M.W. van den Ouweland: Resources, writing-review and editing. J.M. Collée: Resources. N. van der Stoep: Resources. R. Blok: Resources, writing-review and editing. F.E. Bleeker: Resources. F.B.L. Hogervorst: Resources. A.R. Mensenkamp: Resources. A.H. van der Hout: Resources, writing-review and editing. J.C. Oosterwijk: Resources, writing-review and editing. R. van der Luijt: Resources. M. Koudijs: Resources. M.P.G. Vreeswijk: Conceptualization, data curation, formal analysis, supervision, funding acquisition, validation, investigation, methodology, writing-original draft, project administration, writing-review and editing. J. Jonkers: Conceptualization, supervision, funding acquisition, validation, visualization, methodology, writing-original draft, project administration, writing-review and editing. acknowledgments.

The NKI flow cytometry and genomics core facilities are gratefully acknowledged for expert help.

This work was supported by a grant from the Dutch Cancer Society (NKI 2015-7877), to P. Bouwman, J. Jonkers, and M.P.G. Vreeswijk; the Dutch Research Council (NWO) VICI 91814643, to J. Jonkers; Oncode Institute, which is partly financed by the Dutch Cancer Society; NWO research program Mosaic (Grant 017.008.022); and Van de Kamp Fonds from Leiden University Medical Center (Grant 30.925) to S. Moghadasi.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

Note: Supplementary data for this article are available at Clinical Cancer Research Online (http://clincancerres.aacrjournals.org/).

1.
Nielsen
FC
,
van Overeem Hansen
T
,
Sørensen
CS
.
Hereditary breast and ovarian cancer: new genes in confined pathways
.
Nat Rev Cancer
2016
;
16
:
599
612
.
2.
Kuchenbaecker
KB
,
Hopper
JL
,
Barnes
DR
,
Phillips
K-A
,
Mooij
TM
,
Roos-Blom
M-J
, et al
.
Risks of breast, ovarian, and contralateral breast cancer for BRCA1 and BRCA2 mutation carriers
.
JAMA
2017
;
317
:
2402
16
.
3.
Howlader
N
,
Noone
AM
,
Krapcho
M
,
Miller
D
,
Brest
A
,
Yu
M
, et al
.
editors.
SEER Cancer Statistics Review, 1975–2017
.
Bethesda
:
National Cancer Institute
;
2020 [cited 2019 May 28]. Available from
: https://seer.cancer.gov/csr/1975_2017/.
4.
Easton
DF
,
Deffenbaugh
AM
,
Pruss
D
,
Frye
C
,
Wenstrup
RJ
,
Allen-Brady
K
, et al
.
A systematic genetic assessment of 1,433 sequence variants of unknown clinical significance in the BRCA1 and BRCA2 breast cancer-predisposition genes
.
Am J Hum Genet
2007
;
81
:
873
83
.
5.
Parsons
MT
,
Tudini
E
,
Li
H
,
Hahnen
E
,
Wappenschmidt
B
,
Feliubadaló
L
, et al
.
Large scale multifactorial likelihood quantitative analysis of BRCA1 and BRCA2 variants: An ENIGMA resource to support clinical variant classification
.
Hum Mutat
2019
;
40
:
1557
78
.
6.
Richards
S
,
Aziz
N
,
Bale
S
,
Bick
D
,
Das
S
,
Gastier-Foster
J
, et al
.
Standards and guidelines for the interpretation of sequence variants: a joint consensus recommendation of the American College of Medical Genetics and Genomics and the Association for Molecular Pathology
.
Genet Med
2015
;
17
:
405
24
.
7.
Tavtigian
SV
,
Byrnes
GB
,
Goldgar
DE
,
Thomas
A
.
Classification of rare missense substitutions, using risk surfaces, with genetic- and molecular-epidemiology applications
.
Hum Mutat
2008
;
29
:
1342
54
.
8.
Ernst
C
,
Hahnen
E
,
Engel
C
,
Nothnagel
M
,
Weber
J
,
Schmutzler
RK
, et al
.
Performance of in silico prediction tools for the classification of rare BRCA1/2 missense variants in clinical diagnostics
.
BMC Med Genomics
2018
;
11
:
35
.
9.
Hart
SN
,
Hoskin
T
,
Shimelis
H
,
Moore
RM
,
Feng
B
,
Thomas
A
, et al
.
Comprehensive annotation of BRCA1 and BRCA2 missense variants by functionally validated sequence-based computational prediction models
.
Genet Med
2019
;
21
:
71
80
.
10.
Bouwman
P
,
van der Gulden
H
,
van der Heijden
I
,
Drost
R
,
Klijn
CN
,
Prasetyanti
P
, et al
.
A high-throughput functional complementation assay for classification of BRCA1 missense variants
.
Cancer Discov
2013
;
3
:
1142
55
.
11.
Lee
MS
,
Green
R
,
Marsillac
SM
,
Coquelle
N
,
Williams
RS
,
Yeung
T
, et al
.
Comprehensive analysis of missense variations in the BRCT domain of BRCA1 by structural and functional assays
.
Cancer Res
2010
;
70
:
4880
90
.
12.
Guidugli
L
,
Pankratz
VS
,
Singh
N
,
Thompson
J
,
Erding
CA
,
Engel
C
, et al
.
A classification model for BRCA2 DNA binding domain missense variants based on homology-directed repair activity
.
Cancer Res
2013
;
73
:
265
75
.
13.
Mesman
RLS
,
Calléja
FMGR
,
Hendriks
G
,
Morolli
B
,
Misovic
B
,
Devilee
P
, et al
.
The functional impact of variants of uncertain significance in BRCA2
.
Genet Med
2019
;
21
:
293
302
.
14.
Findlay
GM
,
Daza
RM
,
Martin
B
,
Zhang
MD
,
Leith
AP
,
Gasperini
M
, et al
.
Accurate classification of BRCA1 variants with saturation genome editing
.
Nature
2018
;
562
:
217
22
.
15.
Carvalho
MA
,
Marsillac
SM
,
Karchin
R
,
Manoukian
S
,
Grist
S
,
Swaby
RF
, et al
.
Determination of cancer risk associated with germ line BRCA1 missense variants by functional analysis
.
Cancer Res
2007
;
67
:
1494
501
.
16.
Pierce
AJ
,
Johnson
RD
,
Thompson
LH
,
Jasin
M
.
XRCC3 promotes homology-directed repair of DNA damage in mammalian cells
.
Genes Dev
1999
;
13
:
2633
8
.
17.
Gibson
DG
,
Young
L
,
Chuang
R-Y
,
Venter
JC
,
Hutchison
CA
,
Smith
HO
.
Enzymatic assembly of DNA molecules up to several hundred kilobases
.
Nat Methods
2009
;
6
:
343
5
.
18.
Ying
Q-L
,
Wray
J
,
Nichols
J
,
Batlle-Morera
L
,
Doble
B
,
Woodgett
J
, et al
.
The ground state of embryonic stem cell self-renewal
.
Nature
2008
;
453
:
519
23
.
19.
Guidugli
L
,
Shimelis
H
,
Masica
DL
,
Pankratz
VS
,
Lipton
GB
,
Singh
N
, et al
.
Assessment of the clinical relevance of BRCA2 missense variants by functional and computational approaches
.
Am J Hum Genet
2018
;
102
:
233
48
.
20.
Millot
GA
,
Carvalho
MA
,
Caputo
SM
,
Vreeswijk
MPG
,
Brown
MA
,
Webb
M
, et al
.
A guide for functional analysis of BRCA1 variants of uncertain significance
.
Hum Mutat
2012
;
33
:
1526
37
.
21.
Mohammadi
L
,
Vreeswijk
MP
,
Oldenburg
R
,
van den Ouweland
A
,
Oosterwijk
JC
,
van der Hout
AH
, et al
.
A simple method for co-segregation analysis to evaluate the pathogenicity of unclassified variants; BRCA1 and BRCA2 as an example
.
BMC Cancer
2009
;
9
:
211
.
22.
Spurdle
AB
,
Couch
FJ
,
Parsons
MT
,
McGuffog
L
,
Barrowdale
D
,
Bolla
MK
, et al
.
Refined histopathological predictors of BRCA1 and BRCA2 mutation status: a large-scale analysis of breast cancer characteristics from the BCAC, CIMBA, and ENIGMA consortia
.
Breast Cancer Res
2014
;
16
:
3419
.
23.
Vallée
MP
,
Di Sera
TL
,
Nix
DA
,
Paquette
AM
,
Parsons
MT
,
Bell
R
, et al
.
Adding in silico assessment of potential splice aberration to the integrated evaluation of BRCA gene unclassified variants
.
Hum Mutat
2016
;
37
:
627
39
.
24.
Plon
SE
,
Eccles
DM
,
Easton
D
,
Foulkes
WD
,
Genuardi
M
,
Greenblatt
MS
, et al
.
Sequence variant classification and reporting: recommendations for improving the interpretation of cancer susceptibility genetic test results
.
Hum Mutat
2008
;
29
:
1282
91
.
25.
Anantha
RW
,
Simhadri
S
,
Foo
TK
,
Miao
S
,
Liu
J
,
Shen
Z
, et al
.
Functional and mutational landscapes of BRCA1 for homology-directed repair and therapy resistance
.
Elife
2017
;
6
:
e21350
.
26.
Starita
LM
,
Islam
MM
,
Banerjee
T
,
Adamovich
AI
,
Gullingsrud
J
,
Fields
S
, et al
.
A multiplex homology-directed DNA repair assay reveals the impact of more than 1,000 BRCA1 missense substitution variants on protein function
.
Am J Hum Genet
2018
;
103
:
498
508
.
27.
Ransburgh
DJR
,
Chiba
N
,
Ishioka
C
,
Toland
AE
,
Parvin
JD
.
Identification of breast tumor mutations in BRCA1 that abolish its function in homologous DNA recombination
.
Cancer Res
2010
;
70
:
988
95
.
28.
Towler
WI
,
Zhang
J
,
Ransburgh
DJR
,
Toland
AE
,
Ishioka
C
,
Chiba
N
, et al
.
Analysis of BRCA1 variants in double-strand break repair by homologous recombination and single-strand annealing
.
Hum Mutat
2013
;
34
:
439
45
.
29.
Lu
C
,
Xie
M
,
Wendl
MC
,
Wang
J
,
McLellan
MD
,
Leiserson
MDM
, et al
.
Patterns and functional implications of rare germline variants across 12 cancer types
.
Nat Commun
2015
;
6
:
10086
.
30.
Sy
SMH
,
Huen
MSY
,
Chen
J
.
PALB2 is an integral component of the BRCA complex required for homologous recombination repair
.
Proc Natl Acad Sci USA
2009
;
106
:
7155
60
.
31.
Goldgar
DE
,
Easton
DF
,
Deffenbaugh
AM
,
Monteiro
ANA
,
Tavtigian
SV
,
Couch
FJ
, et al
.
Integrated evaluation of DNA sequence variants of unknown clinical significance: application to BRCA1 and BRCA2
.
Am J Hum Genet
2004
;
75
:
535
44
.
32.
Gelli
E
,
Colombo
M
,
Pinto
AM
,
De Vecchi
G
,
Foglia
C
,
Amitrano
S
, et al
.
Usefulness and limitations of comprehensive characterization of mRNA splicing profiles in the definition of the clinical relevance of BRCA1/2 variants of uncertain significance
.
Cancers (Basel)
2019
;
11
.
33.
Wappenschmidt
B
,
Becker
AA
,
Hauke
J
,
Weber
U
,
Engert
S
,
Köhler
J
, et al
.
Analysis of 30 putative BRCA1 splicing mutations in hereditary breast and ovarian cancer families identifies exonic splice site mutations that escape in silico prediction
.
PLoS One
2012
;
7
:
e50800
.
34.
Wangensteen
T
,
Felde
CN
,
Ahmed
D
,
Mæhle
L
,
Ariansen
SL
.
Diagnostic mRNA splicing assay for variants in BRCA1 and BRCA2 identified two novel pathogenic splicing aberrations
.
Hered Cancer Clin Pract
2019
;
17
:
14
.
35.
Thomassen
M
,
Blanco
A
,
Montagna
M
,
Hansen
TVO
,
Pedersen
IS
,
Gutiérrez-Enríquez
S
, et al
.
Characterization of BRCA1 and BRCA2 splicing variants: a collaborative report by ENIGMA consortium members
.
Breast Cancer Res Treat
2012
;
132
:
1009
23
.
36.
Ahlborn
LB
,
Dandanell
M
,
Steffensen
AY
,
Jønson
L
,
Nielsen
FC
,
Hansen
TVO
.
Splicing analysis of 14 BRCA1 missense variants classifies nine variants as pathogenic
.
Breast Cancer Res Treat
2015
;
150
:
289
98
.
37.
Houdayer
C
,
Caux-Moncoutier
V
,
Krieger
S
,
Barrois
M
,
Bonnet
F
,
Bourdon
V
, et al
.
Guidelines for splicing analysis in molecular diagnosis derived from a set of 327 combined in silico/in vitro studies on BRCA1 and BRCA2 variants
.
Hum Mutat
2012
;
33
:
1228
38
.
38.
Woods
NT
,
Baskin
R
,
Golubeva
V
,
Jhuraney
A
,
De-Gregoriis
G
,
Vaclova
T
, et al
.
Functional assays provide a robust tool for the clinical annotation of genetic variants of uncertain significance
.
NPJ Genom Med
2016
;
1
:
16001
.
39.
Domchek
SM
,
Tang
J
,
Stopfer
J
,
Lilli
DR
,
Hamel
N
,
Tischkowitz
M
, et al
.
Biallelic deleterious BRCA1 mutations in a woman with early-onset ovarian cancer
.
Cancer Discov
2013
;
3
:
399
405
.
40.
Bouwman
P
,
Aly
A
,
Escandell
JM
,
Pieterse
M
,
Bartkova
J
,
van der Gulden
H
, et al
.
53BP1 loss rescues BRCA1 deficiency and is associated with triple-negative and BRCA-mutated breast cancers
.
Nat Struct Mol Biol
2010
;
17
:
688
95
.
41.
Blomen
VA
,
Májek
P
,
Jae
LT
,
Bigenzahn
JW
,
Nieuwenhuis
J
,
Staring
J
, et al
.
Gene essentiality and synthetic lethality in haploid human cells
.
Science
2015
;
350
:
1092
6
.
42.
Jhuraney
A
,
Velkova
A
,
Johnson
RC
,
Kessing
B
,
Carvalho
RS
,
Whiley
P
, et al
.
BRCA1 Circos: a visualisation resource for functional analysis of missense variants
.
J Med Genet
2015
;
52
:
224
30
.
43.
Monteiro
AN
,
Bouwman
P
,
Kousholt
AN
,
Eccles
DM
,
Millot
GA
,
Masson
J-Y
, et al
.
Variants of uncertain clinical significance in hereditary breast and ovarian cancer genes: best practices in functional analysis for clinical annotation
.
J Med Genet
2020 Mar 9 [Epub ahead of print]
.
44.
Lindor
NM
,
Guidugli
L
,
Wang
X
,
Vallée
MP
,
Monteiro
ANA
,
Tavtigian
S
, et al
.
A review of a multifactorial probability-based model for classification of BRCA1 and BRCA2 variants of uncertain significance (VUS)
.
Hum Mutat
2012
;
33
:
8
21
.
45.
Boonen
RACM
,
Rodrigue
A
,
Stoepker
C
,
Wiegant
WW
,
Vroling
B
,
Sharma
M
, et al
.
Functional analysis of genetic variants in the high-risk breast cancer susceptibility gene PALB2
.
Nat Commun
2019
;
10
:
5296
.
46.
Foo
TK
,
Tischkowitz
M
,
Simhadri
S
,
Boshari
T
,
Zayed
N
,
Burke
KA
, et al
.
Compromised BRCA1-PALB2 interaction is associated with breast cancer risk
.
Oncogene
2017
;
36
:
4161
70
.
47.
Park
J-Y
,
Singh
TR
,
Nassar
N
,
Zhang
F
,
Freund
M
,
Hanenberg
H
, et al
.
Breast cancer-associated missense mutants of the PALB2 WD40 domain, which directly binds RAD51C, RAD51 and BRCA2, disrupt DNA repair
.
Oncogene
2014
;
33
:
4803
12
.
48.
Shimelis
H
,
Mesman
RLS
,
Von Nicolai
C
,
Ehlen
A
,
Guidugli
L
,
Martin
C
, et al
.
BRCA2 hypomorphic missense variants confer moderate risks of breast cancer
.
Cancer Res
2017
;
77
:
2789
99
.
49.
Spurdle
AB
,
Whiley
PJ
,
Thompson
B
,
Feng
B
,
Healey
S
,
Brown
MA
, et al
.
BRCA1 R1699Q variant displaying ambiguous functional abrogation confers intermediate breast and ovarian cancer risk
.
J Med Genet
2012
;
49
:
525
32
.
50.
Moghadasi
S
,
Meeks
HD
,
Vreeswijk
MP
,
Janssen
LA
,
Borg
Å
,
Ehrencrona
H
, et al
.
The BRCA1 c. 5096G>A p.Arg1699Gln (R1699Q) intermediate risk variant: breast and ovarian cancer risk estimation and recommendations for clinical management from the ENIGMA consortium
.
J Med Genet
2018
;
55
:
15
20
.

Supplementary data