Enhanced Classification of Protein-Truncating Variants Reveals the Need for Improved Predictive Tools

ASHG 2024 Annual Meeting, Colorado, Amerika Birleşik Devletleri, 5 - 09 Kasım 2024, ss.730-731, (Özet Bildiri)

Classifying genetic variants as pathogenic or benign requires diverse information, varying by

variant type, and often focuses on those most likely to be pathogenic. In ClinVar, ~42% of

pathogenic/likely pathogenic (P/LP) variants are protein-truncating (PTC-variants) with ~16% of

these classified as variants of uncertain significance (VUS). The functional impact of PTC-variants

depends on whether the mRNA undergoes nonsense-mediated RNA decay (NMD), which can

lead to loss-of-function (LoF) (NMD-triggering); or, mRNA can escape NMD (NMD-escape).

allowing truncated proteins that may have LoF, gain-of-function (GoF), or neutral effects.

Current classification methods inadequately utilize NMD information, despite its importance.

NMD targets transcripts with PTC-variants in the final exon or within the last 50-55 base pairs of

the penultimate exon (canonical rule), with additional rules indicating that PTC-variants <150

nucleotides from the start codon (start-proximal) or in long exons (long-exon) are likely to

escape NMD.Using these rules, we annotated ~42K P/LP PTC-variants from ClinVar, finding

~32.3% predicted as NMD-escapees across 2,459 genes. In contrast, ~47% of ~19K VUS PTC-

variants are predicted as NMD escapees (p-value < 2.2e-16), highlighting the classification

challenge for these ambiguous variants (LoF, GoF, neutral). This underscores the need to

improve classification methods for NMD-escape PTC-variants. We hypothesized a depletion of

PTC-variants in gene regions predicted to escape NMD in control databases (i.e., in healthy

individuals) could reveal loci associated with disease susceptibility. To test this, we annotated

~752K high-quality PTC-variants from gnomAD v4, finding 45.7% as NMD-escapees (~19.5%

canonical, ~7% start-proximal, and ~20.2% long exon, referring to different NMD escape

annotation rules). We identified ~18% of 14,650 canonical transcripts depleted for NMD-escape

variants in control individuals, with 90% tolerant to NMD-triggering PTC-variants (pLI <0.9 or

LOEUF >= 0.6), and ~26.4% associated with disease traits in OMIM. Of these, 402 genes had ≥ 1

NMD-escape P/LP variant, with 70 genes showing clustering in NMD-escape regions, including

19 known disease genes with NMD-escape mechanisms, such

as DVL1 (MIM:601365), MN1 (MIM:156100), NOVA2 (MIM:601991),

and PPM1D (MIM:605100).In sum, there is a need to enhance the classification of NMD-

escaping PTC-variants to determine their pathogenicity. This could be achieved by improving

prediction tools that analyze the depletion of these variants in regions predicted to escape NMD

in control individuals, offering insights into disease susceptibility.