APPLICATION OF THE ACMG/AMP FRAMEWORK TO CAPTURE EVIDENCE RELEVANT TO PREDICTED AND OBSERVED IMPACT ON SPLICING: RECOMMENDATIONS FROM THE CLINGEN SVI SPLICING SUBGROUP

Abstract

The American College of Medical Genetics and Genomics (ACMG) and the Association for Molecular Pathology (AMP) framework for classifying variants uses six evidence categories related to the splicing potential of variants: PVS1 (null variant in a gene where loss-of-function is the mechanism of disease), PS3 (functional assays show damaging effect on splicing), PP3 (computational evidence supports a splicing effect), BS3 (functional assays show no damaging effect on splicing), BP4 (computational evidence suggests no splicing impact), and BP7 (silent change with no predicted impact on splicing). However, the lack of guidance on how to apply such codes has contributed to variation in the specifications developed by different Clinical Genome Resource (ClinGen) Variant Curation Expert Panels. The ClinGen Sequence Variant Interpretation (SVI) Splicing Subgroup was established to refine recommendations for applying ACMG/AMP codes relating to splicing data and computational predictions. Our study utilised empirically derived splicing evidence to: 1) determine the evidence weighting of splicing-related data and appropriate criteria code selection for general use, 2) outline a process for integrating splicing-related considerations when developing a gene-specific PVS1 decision tree, and 3) exemplify methodology to calibrate bioinformatic splice prediction tools. We propose repurposing of the PVS1_Strength code to capture splicing assay data that provide experimental evidence for variants resulting in RNA transcript(s) with loss of function. Conversely BP7 may be used to capture RNA results demonstrating no impact on splicing for both intronic and synonymous variants, and for missense variants if protein functional impact has been excluded. Furthermore, we propose that the PS3 and BS3 codes are applied only for well-established assays that measure functional impact that is not directly captured by RNA splicing assays. We recommend the application of PS1 based on similarity of predicted RNA splicing effects for a variant under assessment in comparison to a known Pathogenic variant. The recommendations and approaches for consideration and evaluation of RNA assay evidence described aim to help standardise variant pathogenicity classification processes and result in greater consistency when interpreting splicing-based evidence.

Figure 1.

Schematic demonstrating assignment of gene-specific codes to canonical splice sites based on a modified version of the PVS1 framework proposed by Abou Tayoun et al., 2018. It is important to note that each PVS1 assigned weight may be reduced if there is evidence of potential rescue mechanisms. For example, skipping of either exon 4 or 7 may lead to a protein that retains partial function. Annotating gene-specific lists of naturally occurring splicing events can provide greater evidence of potential ‘rescue’ isoforms. Also see Supplemental Information, Box S1, Biologically relevant transcripts and the rescue transcript model.

Figure 2.

Model for optimising thresholds for prediction algorithms of alternative splicing. (A) Schematic demonstrating how collation of three variant datasets (in vitro splicing data, splicing prediction scores, and clinical classification data) enables calibration of splicing prediction algorithms for pathogenicity. While clinically classified variant data is preferable, splicing assay data can be used as an imperfect surrogate for pathogenicity. More extensive annotation of alternative splicing events and level of aberration will lead to an improved correlation of splicing events with variant pathogenicity. The distribution of hypothetical computationally predicted splice scores is illustrated, showing significant overlap of non-spliceogenic/spliceogenic datasets (left side) and Benign/Pathogenic datasets (right side). The low, intermediate and high prediction score used to assign ACMG/AMP code weighting can be determined by calculating likelihood ratios for different score categories, and obtaining consensus on the score thresholds to be applied. (B) Process for calibrating splicing prediction score thresholds for computational tools. A worked example of a likelihood ratio calculation is shown in Table S4. Note: truth datasets exclude splice site variants, which are captured by the PVS1 decision tree process.

Figure 3.

An exemplar PP3, BP4, and BP7 decision tree for maximum SpliceAI splicing prediction scores and calibrated cut-off scores. The analytical process is shown in Figure 2B and data shown in Table 2. BP7 should not be applied for splice motif regions given their higher prior. This may be defined as the standard splice region (a conservative application already implemented by several VCEPs) or the minimal splice region. PP3 may still be applied for missense or insertion-deletion variants that show computational evidence for a deleterious effect for change in protein sequence.

Figure 4 -

Decision tree for application of bioinformatic codes and RNA splicing assay results for variant interpretation. Footnotes: (a) Alternative prediction tools/thresholds may be appropriate for variants that impact sites other than GT-AT donor-acceptor motifs. (b) LP variants at the canonical positions should only be used as evidence if additional supporting clinical evidence is present. (c) Silent (excluding last 3nt of exon and first nt of exon) and intronic variants at or beyond the +7 and −21 positions (conservative designation for splice region) or otherwise as at or beyond the +7 and −4 positions (less conservative designation for minimal splice region). (d) If multiple impacts are observed from a splicing assay, use flowchart for the most conservative application of PVS1 based on experimental data. (e) We recommend that these thresholds be refined and applied in a disease- and gene-specific manner, including advice from Variant Curation Expert Panels. Categorization as complete or near complete needs to consider multiple factors, including assay/technique, RNA source, and validation of assay weights using established controls. Examples of laboratory-specific approaches and suggested operational thresholds have been reported previously.^{; –}