Welcome to another novel blog post!
In this post, we’ll explore Copy Number Variants (CNVs) and why they matter in genetics. We will also discuss how classification criteria are implemented within AION to assess them.
Why CNV Interpretation is Important
CNVs have consistently been associated with rare diseases of high complexity, making their interpretation vital, yet challenging. CNV interpretation enhances the capabilities of genetic labs, leading to more confirmed diagnoses and better patient care. It has been reported that 12% of diagnosed cases are caused by pathogenic CNVs (Rare Disease Whole-Genome Sequencing, n.d.). While CNVs can be more complex to interpret than other genetic variations, AION’s advanced CNV feature expedites your diagnosis process with automated ranking and clinical evidence integration, setting it apart from other filtering solutions that may lack CNV interpretation capabilities.
What are CNVs?
CNVs represent a type of structural variation in which genome regions, spanning a considerable number of basepairs, can be deleted or duplicated. In contrast, Single Nucleotide Polymorphisms (SNPs) and Insertion/Deletions (indels) only alter one or relatively few base pairs. While the reference genome in autosomal chromosomes typically contains 2 copies of each region, CNVs present different copy numbers, with deletions having 0 or 1 copies and duplications having more than 2 copies. Reliable detection of CNVs is feasible from short-read sequencing technologies, with several efficient bioinformatic approaches developed based on read alignment or coverage analysis (Gabrielaite et al., 2021). Additionally, the development and decreasing price of long read sequencing technologies will also pave the way in increasing sensitivity and precision of CNV detection and other types of structural variants that may go undetected with current short-read sequencing tests like Whole Exome Sequencing (WES) or Whole Genome Sequencing (WGS).
ACMG/ ClinGen Criteria for CNVs
In light of recent advancements in CNVs and other types of structural variant detection, genetics experts have long requested a solid classification framework similar to the one used for small variants (Richards et al., 2015). CNV/SVs have unique characteristics that differ from small variants; for example, a CNV can span more than one gene, complicating gene-phenotype associations. Moreover, it is very difficult to observe exactly the same CNV/SV twice, as each variant has its own unique breakpoints, and that’s why consideration of CNV/SV overlap with other variants/genes is so important and a differentiating feature compared to SNPs.
In 2020, a publication on technical standards for the interpretation and reporting of constitutional CNVs was published to address this need (Riggs et al., 2020). While other articles also made efforts in that direction, Riggs et al. platform-agnostic system, created by the American College of Medical Genetics (ACMG) and Clinical Genome Resource (ClinGen), provides a set of consensus rules that enables standardised interpretation of CNVs. This system aligns with SNV guidelines (roughly applying criteria as if they had several strengths of evidence, as in the 2015 SNP/indel system), and is planned to become an international standard with time.
The ACMG/ClinGen technical standards for CNV interpretation consist of a semi-quantitative, evidence-based scoring system that is organised into 5 different sections, based on different types of evidence:
1) Initial assessment of genomic content
2) Overlap with dosage sensitive regions/genes
3) Evaluation of number of overlapping genes
4) Overlap with other cases in literature/databases
5) Segregation and inheritance pattern
These 5 sections remain the same whether evaluating a deletion or a duplication, although the specific criteria within each section might change. While sections 1 and 3 are quite straightforward, focusing on the number of genes affected by the CNV/SV and whether they are coding or overlapping functional elements, sections 2 and 4 are related to overlap with known data in databases /literature and show some challenges for automatic interpretation. Section 5 refers to CNV/SV segregation and phenotype assessment, two features that require access to patient data that might not be easily available from the start.
Determining Pathogenicity
To evaluate pathogenicity for a given CNV/SV, all criteria must be reviewed, with each applied criteria adding or subtracting points to determine a final score. These criteria are modelled to imitate the strength system used in small variants AMCG/AMP guidelines. Specifically, 0.9 points would be equivalent to “very strong” evidence, 0.45 is “strong”, 0.30 is “moderate”, and 0.15 is “supporting”. If the final score is higher than 0.99, the CNV/SV will be evaluated as Pathogenic, while a score lower than -0.99 indicates a Benign classification. This system underwent iterative benchmarking using a dataset of 114 CNVs (58 deletions, 56 duplications), evaluated by several independent reviewers to ensure robustness across several laboratories.
AION’s Implementation and Validation
At Nostos Genomics, we have recently implemented tertiary CNV prioritisation from WGS and WES VCF files. Users can now go beyond the analysis of SNPs and Indels and also focus on CNVs by leveraging a variant ranking and filtering tools.
Within our user-friendly interface, we also automated the analysis of CNV according to ACMG/ClinGen standards. We provide accurate explanations for each applied criterion and prioritise the detected variants accordingly. Importantly, Nostos’ CNV analysis solution automates a significant amount of the criteria compared to other solutions. Finally, we aid the clinical interpretation by relating each CNV with possible phenotypes.
To validate ACMG/ClinGen criteria for CNV analysis, we collaborated with our customers and utilised data extracted from several articles in literature where several experts reached a conclusion on the pathogenicity/ benignancy of several CNV datasets used previously (Riggs et al., 2020; Fan et al., 2021; Testard et al., 2022; among others). Using these datasets, we were able to validate and tune implemented criteria for technical assessment, reaching an average ranking of the causative CNV of 8.7.
Ready to Analyse CNV Data?
If you’re interested in interpreting CNV data, sign up for a free trial of AION today! Unlock the power of CNV interpretation and diagnose more patients with precision and confidence.
References:
