Paola Nix, Ph.D. at Myriad Genetics discusses an ASCO 2020 abstract entitled Functional RNA Studies Are a Useful Tool in Variant Classification but Must Be Used With Caution: A Case Study of One BRCA2 Variant
Since it can guide health management and cancer care decisions, the use of genetic testing to determine hereditary cancer risk is growing. In BRCA1 and BRCA2, for instance, pathogenic variants are considered to be associated with inherited breast and ovarian cancer syndrome. In one of these genes, a germline pathogenic variant is associated with an increased cancer risk and thus merits altered medical management, including increased surveillance and risk-reducing surgery options.1 In addition, the detection of a pathogenic variant of BRCA1 or BRCA2 may affect treatment decisions and clinical recommendations for women with breast and/or ovarian cancer
While developed guidelines from the American College of Medical Genetics (ACMG) and the Molecular Pathology Association (AMP) provide a structure to assist with classification of variants,5 the available evidence for certain variants may be difficult to interpret.6,7 This includes certain sequence variants that have the potential to affect mRNA splicing. A notable fraction of disease-causing variants in cancer predisposition genes influence splicing.8-10 Several of these occur at the 5′ and 3′ exon-intron boundaries, and the possible effect on splicing is well known, based on the predicted sequence of the canonical splicing donor and acceptor (± 1 , 2 splicing sites).5 However, sequence variants found outside these canonical splicing sites can often alter the sequence. It is more difficult to test pathogenicity for these variants, and additional functional evidence is often needed to assess the actual effect of the variants on splicing.
Clinical laboratories typically use functional studies of RNA to help distinguish variants that can affect splicing. While these studies may provide valuable proof of the pathogenicity of such variants, there are important caveats to note, and both how the studies are conducted and how the data are interpreted need to be taken care of. 11,12 One important concern when using RNA analysis to identify variants that may affect splicing is that finding a splice defect alone may not be sufficient. Aberrant splicing can result in either no normal transcript produced by the variant allele (complete splice defect) or some residual normal transcript produced by the variant allle.13 These partial, or "leaky," splice defects could result in sufficient functional transcript to support normal protein function. Therefore, by using RNA analysis in variant classification, it is important to quantify any aberrant splice product and decide whether the variant allele generates any normal transcript.
Functional RNA studies also include an analysis of RNA derived from a patient blood sample or lymphoblastoid cell line by the reverse transcription polymerase chain reaction ( RT-PCR). Mini-gene assays were also used in cases where samples from patients are not available. A research by Gelli et al14 illustrated the shortcomings of mRNA analysis in addressing ambiguous significance variants (VUSs). When assessing splicing, it is not only important to detect and quantify the aberrant splice product, but also to quantify products that are unique to the alleles. The use of informative single nucleotide polymorphisms ( SNPs) present in the same PCR amplicon as the putative spliceogenic variant being examined (ie, tag SNPs) is one way to differentiate variant-specific transcripts. In the case of a heterozygous variant carrier, these tag SNPs can distinguish between the transcripts provided by the normal allele and the variant allle, particularly when the variant allle is located in the intron and is not present in the processed mRNA. It is possible to clone RT-PCR products representing these transcripts into a vector system or use them directly as sequencing research models.
Laboratories that do not require RNA analysis, either as part of their own studies or from published studies, to provide allele-specific transcript quantification for variant classification can misclassify splice-related variants and potentially report false-positive results. Furthermore, if a variant allele has been shown to produce a functional transcript, caution must be taken in assigning pathogenicity, since it may be uncertain whether the amount of functional transcript will sustain normal protein function. For example, Colombo et al15 indicated that a variant allele could express up to 20 percent aberrant product for splice variants causing BRCA2 exon 3 skipping, but may not be associated with even a moderate cancer risk.
BRCA2 Case Analysis c.426-12 426-8del
An example of a splice variant with a complex interpretation is BRCA2 c.426-12 426-8del — an intronic variant which results in the deletion of five nucleotides from the DNA sequence adjacent to the exon 5 splice acceptor location. Computational splice predictions suggest that the native splice acceptor is impaired by this variant, which could lead to excessive splicing of mRNA and premature truncation of protein. Published functional studies support this prediction, with studies by Zhang et al in 200916 and Sanz et al in 201017 show that this variant causes exon 5, which is out of picture, to be skipped. But both studies also noted that a full-length transcript is formed by the variant allele, suggesting a partial splice defect.
Recently published RNA analyses from one clinical testing laboratory revealed that aberrant splicing was triggered by BRCA2 c.426-12 426-8del, resulting in deletion of exon 5.18,19 However, there was no allele-specific quantification of normal transcripts to determine how much, if any, of the variant allele was produced. However, on the basis of RNA studies showing irregular splicing (evidence category PS3), absence of variant from population controls (evidence category PM2), and silico splicing models predicting native site weakening (evidence category PP3), the authors concluded that this variant could be categorized as probable pathogenic in accordance with ACMG / AMP guidelines5.
Conversely, on the basis of the evidence available at the time, our laboratory initially identified BRCA2 c.426-12 426-8del as a VUS (Fig 1). This evidence included 2 previously published functional studies showing that some normal transcript was provided by the variant allele.16,17 These functional data alone are not sufficient to establish if the variant is associated with an increased risk of cancer since there is ambiguity as to how much normal transcript the variant allele must be expressed to promote normal functioning of BRCA2. Although population databases and in silico modeling can provide useful classification evidence, these tools must also be carefully considered. Benign variations can also be rare (e.g., missing from population controls), and a complete or partial splice defect can not be discriminated against in silico splicing models. The question of whether the variant has a clinical effect is not addressed explicitly by these instruments, provided that the splice defect is not completely penetrating. As a result, to identify BRCA2 c.426-12 426-8del as potentially pathogenic, this body of evidence was not adequate for our laboratory.
Yeah. FIG 1.
Suggested flow for the use of functional RNA studies in classification of variants. SNP, polymorphism by single nucleotide.
Our laboratory was later able to reclassify, using additional clinical data, BRCA2 c.426-12 426-8del from VUS to benign. This included proof from a validated background weighting algorithm (HWA) that scores a variant of multiple individuals carrying the same variant of interest on the basis of personal and family histories.20 The variant-specific score is compared with matched pathogenic controls (individuals with established pathogenic variants in BRCA2) and matched benign controls (individuals with only benign controls). If they interact with the benign controls, variations are considered benign and have no extraordinary interact with pathogenic controls. The HWA was previously validated by Pruss et al20 and showed that it has > 99.5% negative and positive predictive values. The HWA named BRCA2 c.426-12 426-8del benign in this case, suggesting that carriers did not have a personal or family history consistent with a pathogenic variant in BRCA2 (Appendix Fig A1).
Moreover, co-occurrence of a VUS in the same gene with recognized pathogenic variants can provide evidence that a variant is benign based on associated clinical phenotypes. In our laboratory, in individuals with no known symptoms of Fanconi anemia, BRCA2 c.426-12 426-8del was found to co-occur with two distinct pathogenic variants in BRCA2 (c.7719dupA and c.9257-1G > C). Based on haplotype analysis at the time of study, both pathogenic variants were found to be in trans with c.426-12 426-8del. This provides further proof that BRCA2 c.426-12 426-8del is benign. Collectively, the HWA and co-occurrence data are consistent with a benign classification and indicate that adequate functional transcripts are provided by the variant allele to maintain normal BRCA2 function. In compliance with ACMG / AMP guidelines (proof groups BS4 and BP2), our laboratory considered the functional RNA data for partial splicing defect coupled with these additional clinical results as ample evidence to downgrade BRCA2 c.426-12 426-8del from VUS to benign.
An significant component of classification for variants that may affect mRNA splicing is the functional data from the RNA review. Recent use of laboratories conducting their own functional variant classification RNA studies may lead to conclusive, actionable results for more patients; however, it is necessary to maintain a high standard of RNA analysis to ensure that variant classifications also maintain a high level of accuracy