Hypertrophic cardiomyopathy (HCM), the most common inherited cardiac disease, with a prevalence of one in 200–500 individuals [1], is characterized by left ventricular (LV) hypertrophy unexplained by secondary causes and a broad clinical spectrum. HCM is associated with considerable morbidity and mortality [2]. For personal and family risk management, molecular testing can confirm the genetic origin of HCM. HCM generally follows Mendelian inheritance, predominantly in an autosomal dominant fashion. However, it can exhibit both incomplete penetrance and variable expressivity, resulting in apparently sporadic cases [3]. HCM is primarily caused by pathogenic variants (PVs) in sarcomeric core genes, followed by variants in non-sarcomeric genes with varying levels of supportive evidence [4]. HCM-causing variants are typically identified using large gene panels based on targeted next-generation sequencing (tNGS), which facilitates differential diagnosis with other heart diseases. The overall detection rate varies depending on patient baseline characteristics (i.e., demographic, family, and clinical history), ranging from 20–30% in sporadic cases to 40–60% in familial HCM. This implies that genetic analysis yields inconclusive or negative results in 40% to >60% of HCM cases [5]. Several clinical (e.g., phenocopies/misdiagnoses) and technical (e.g., unidentified genes and/or variants) factors may contribute to this relatively low diagnostic yield. tNGS panels target exonic regions and are unable to effectively detect intronic positions. Notably, MYBPC3 accounts for most HCM cases, and PVs in this gene are generally loss-of-function (LoF) variants, leading to nonsense-mediated decay (NMD) and subsequent haploinsufficiency. In this scenario, variants that alter RNA splicing (i.e., spliceogenic variants) can produce aberrant transcripts that also result in NMD. They are increasingly recognized as causative of HCM in several cohorts [6^-9], potentially representing >3% of HCM probands [7]. However, these studies screened for previously characterized spliceogenic MYBPC3 variants using NGS covering all MYBPC3 introns (6, 7) or using whole-exome sequencing (WES) including 100-bp intronic regions beyond intron-exon boundaries [8, 9]. Torrado, et al. [6] focused on the discovery and characterization of rare unknown variants using several deep learning-based splicing algorithms and functional assays at the mRNA and protein levels.

Here, we re-evaluated raw data obtained with a cardiac tNGS panel to improve the molecular HCM diagnostic rate. The study was conducted in accordance with the Declaration of Helsinki and approved by the Ethics Committee of Sant’Andrea Hospital (Rome, Italy; approval No.: 42 of September 28, 2007).

Between 2021 and 2023, 142 consecutive unrelated patients diagnosed with HCM had negative genetic testing results. This cohort of HCM patients was characterized by the absence of a family history and a broad spectrum of clinical features. The conventional molecular analysis covered exonic regions of 174 genes associated with hereditary heart diseases, including HCM, and consisted of (i) genomic DNA extraction from peripheral blood (DNeasy Blood & Tissue Kit and QIAamp DNA Blood Mini Kit, Qiagen, Hilden, Germany), (ii) library preparation (TruSight Cardio panel, Illumina, San Diego, CA, USA), (iii) sequencing on a MiniSeq system (Illumina), and (iv) standard bioinformatics analysis using BaseSpace platforms (i.e., BaseSpace Sequence Hub and BaseSpace Variant Interpreter, Illumina). As a default filter setting, the padding size was 20 bp on both sides of exonic regions.

To include the annotation of deeper intronic variants as well, sequencing data were re-analyzed based on an ad-hoc parallel workflow (Fig. 1). Through a first approach for the screening of known disease-causing variants, a list of intronic variants previously described and interpreted as pathogenic in the literature [6] was compiled, considering up to 100 bp of flanking sequenced regions (Supplemental Data Table S1). A corresponding in-silico hotspot for these PVs was built by annotating their genomic coordinates using Integrative Genomics Viewer v2.16.2 [10] and was later applied to raw sequencing data (i.e., BAM files) obtained from the BaseSpace Sequence Hub. PVs thus identified and visualized were validated using Sanger sequencing. In a second approach used for identifying unknown variants, all MYBPC3 intronic variants detected by tNGS were considered. Common intronic variants were filtered out, and rare variants were analyzed using the SpliceAI software [11] (a web-based interface) to evaluate their probability of being splice-altering. In the latter case, a score of >0.8 indicated an increased damaging impact with high precision [12]. Subsequently, mRNA-level evaluation was performed to assess the impact of emerging variants (i.e., SpliceAI score >0.5) on RNA splicing. Upon identifying a potential splice-altering variant, it was classified according to the American College of Medical Genetics and Genomics (ACMG) criteria [13] and validated using Sanger sequencing. By applying the filtering approach, 23 additional intronic variants were identified in our cohort. Considering a frequency threshold of 1% in the general population, 13 were common, and 10 were rare variants. All rare variants had a SpliceAI score <0.5, a more permissive cutoff to discriminate potential splice-altering variants from non-altering variants [12] (Supplemental Data Table S2). Therefore, none of these variants required further investigation.

Figure 1. Parallel multistep workflow for diagnostic setting. Left side (first approach): development and application of an in-silico hotspot for previously reported PVs; right side (second approach): in-silico filtering and in-vitro analysis of unknown variants.
Abbreviations: PVs, pathogenic variants; IGV, Integrative Genomics Viewer; ACMG, American College of Medical Genetics and Genomics.

The hotspot approach helped detect the c.1224-80G>A variant in MYBC3 in three patients (2.1%; 95%CI: 0.7–6.0%). No additional intronic variants were identified. According to the ACMG criteria (PM2 and PP5), the c.1224-80G>A variant is classified as likely pathogenic. This variant has been documented previously in four HCM cohorts: N=1/93 [7], N=1/720 [8], N=3/1644 [9], and N=6/3437 [6]. The frequency observed in our cohort corresponds with those reported in the literature, albeit influenced by the population considered.

From a functional perspective, RNA splicing requires the recognition of splice donor and acceptor sites located at the boundaries of introns, generally represented by the canonical dinucleotides “GT” and “AG,” respectively. Briefly, the c.1224-80G>A variant potentially generates a novel splice acceptor site (“AG”) in intron 13 of MYBPC3, leading to the retention of 78 nucleotides upstream of exon 14 in the transcript (Fig. 2). The splice-altering property of this variant, predicted in silico and confirmed in functional studies, includes a SpliceAI Δscore of 0.94 [6] and results in an in-frame insertion of 26 codons, diminishing MYBPC3 transcripts by 50% [7]. Additionally, the splicing of micro-exons, protein-coding regions spanning 3 to 30–50 bp, warrants further attention [14]. Particularly, MYBPC3 contains three micro-exons, including the conserved exon 14 [15, 16]. Micro-exons, with their weak splicing signals, rely on robust signals from flanking intron regions [17, 18]. Therefore, these regions prove critical as several damaging variants have been identified [19]. The c.1224-80G>A variant establishes a new, stronger acceptor splice site in intron 13, potentially favored over the canonical site [7], thus altering the splicing of micro-exon 14 and resulting in an unstable transcript.

Figure 2. MYBPC3 exon 14 splicing. (A) Schematic representation of the nucleotide sequences of exon 13, intron 13, and exon 14. Canonical splicing of intron 13 with the juxtaposition of the 3′-exon 13/5′-exon 14 junction is shown above (green lines). The variant c.1224-80G>A (in red) creates a cryptic splice acceptor site in intron 13 (dinucleotide “AG”), which leads to the juxtaposition of 3′-exon 13 with a locus upstream of 5′-exon 14 (bottom, red lines), inserting 78 intronic nucleotides into the transcript. (B) Representation of aligned NGS reads (top) and Sanger electropherogram (bottom) corresponding to the heterozygous variant c.1224-80G>A (arrow).
Abbreviation: NGS, next-generation sequencing.

Clinical data, such as comprehensive cardiological assessments and familial/personal histories, were collected from the three individuals carrying the c.1224-80G>A variant (Table 1). Their ages ranged from 56–74 years, with two reporting a familial history of HCM. Maximal LV wall thickness varied from 13–27 mm; patient #1 (P1) exhibited systolic anterior motion of the mitral valve (SAM), and P2 experienced left ventricular outflow tract (LVOT) obstruction with late onset. These patients did not share notable phenotypic characteristics. Overall, this variant might be associated with mild hypertrophy, as supported by prior reports. Janin, et al. [7] described a 34-year-old man with sporadic asymptomatic HCM and an LV wall thickness of 17 mm, whereas Singer, et al. [8] reported an asymptomatic 28-year-old man with a family history of HCM.

Table 1 Clinical characteristics of patients with HCM carrying the spliceogenic c.1224-80G>A variant in MYBPC3

Characteristics	Patient 1	Patient 2	Patient 3
Age at first observation (yrs)	56	74	60
Sex	Female	Male	Male
Familial history	No	Yes	Yes
Symptoms
NYHA class	-	II	-
Imaging
LV thickness (mm)	13	27	14
Obstruction (type)	No (SAM)	Yes (LVOT gradient=50 mmHg)	No

Abbreviations: HCM, hypertrophic cardiomyopathy; NYHA, New York Heart Association; LV thickness, left ventricular thickness; SAM, systolic anterior motion of the mitral valve; LVOT gradient, left ventricular outflow tract gradient.

To the best of our knowledge, we report the first-known application of a strategy using a multigene panel for detecting pathogenic intronic MYBPC3 variants, previously identified only through intron sequencing or WES methodologies. Specifically, the filtering approach based on both frequency and in-silico predictions helped exclude new potential intronic variants. The design and application of a hotspot for previously described intronic PVs yielded additional evidence for their prevalence, which is crucial for pathogenicity interpretation. Notably, this facilitated accurate genetic diagnoses in some patients with HCM who had shown gene elusiveness in standard analyses, increasing the detection rate of the cardio panel from 20.2% (N=36/178) to 21.9% (N=39/178) in our study population. The clinical utility of our findings would have been enhanced by complete intronic sequencing of MYBPC3 and other HCM-associated genes, and clinical validity could be supported further by mRNA sequencing to assess the functional impact of identified intronic variants. However, these analyses were impractical owing to practical limitations, such as sample availability, quality, cost, and resource constraints.

Collectively, our findings are consistent with the evidence in the literature and support broader MYBPC3 analysis to provide health professionals and manufacturers with better information for the systematic recognition of additional splice-disrupting variants at non-canonical splice sites.

None.

Conceptualization: Rubattu S, Piane M, Petrucci S. Methodology: Caroselli S, Fabiani M, and Savio C. Investigation: Micolonghi C, Tini G, Musumeci B, Pagannone E, Germani A, and Libi F. Data curation: Caroselli S. Visualization: Caroselli S. Project administration: Rubattu S and Piane M. Supervision: Visco V, Pizzuti A, and Autore C. Writing – original draft: Caroselli S. Writing – review & editing: Rubattu S, Piane M, and Petrucci S.

None declared.

None declared.

Supplementary materials can be found via https://doi.org/10.3343/alm.2024.0201