The study included 90 children with DD referred to the regional center for genetic evaluation. The research consisted of two cohorts. The first cohort included 40 children with DD enrolled from 1 October 2020, to 31 September 2021 (1 year), under a grant received from the 3billion (South Korea) for WES. The second cohort consisted of 50 patients enrolled from 7 December 2022, to 7 April 2023 (4 months), as part of a second grant from the 3billion. All children were under the care of a neurologist or psychiatrist due to DD and were referred for genetic testing. The inclusion criteria were as follows: phenotypes and symptoms in children, including intellectual disability or developmental delay, neurodevelopmental delay, neuromuscular disorder or white matter disease, inborn error of metabolism, and age up to 18 years. The study followed the principles of the 1975 Declaration of Helsinki (amended in 2000) and was approved by the Ethics Committee of I. Horbachevsky Ternopil National Medical University. Informed consent was obtained from all participants’ guardians.

High molecular weight genomic DNA was extracted from patient’s whole blood-EDTA, or buccal swab samples using QIAamp DNA Blood Mini Kit (Qiagen) and AccuBuccal DNA Preparation Kit (Accugene), respectively. WES was performed following the CAP/CLIA validated standard operating protocol. Exome capture was performed using xGen Exome Research Panel v2 (Integrated DNA Technologies, Coralville, Iowa, United States). Sequencing was performed using the NovaSeq 6,000 platform (Illumina, San Diego, CA, United States) as 150 bp paired-end reads. Sequencing data were aligned to either GRCh37/hg19 human reference genome using BWA-MEM and processed for variant calling by GATK v4.2.14 (Van der Auwera et al., 2013; Li and Durbin, 2009). Copy number variants (CNVs) were called using 3bCNV, and tool developed by 3billion that uses depth-of-coverage information of each exon. Variants were annotated, filtered, and classified using EVIDENCE v4 which incorporates Ensembl Variant Effect Predictor (VEP) for annotation and the American College of Medical Genetics and Genomics (ACMG) guideline for variant classification (Richards et al., 2015; Seo et al., 2020; McLaren et al., 2016). The splicing prediction was performed using spliceAI (Jaganathan et al., 2019). The filtered and classified variant list was than manually reviewed by medical geneticists and physicians. The most likely variants that can explain the patient’s phenotype were selected for reporting.

3billion has internally validated that variants (SNV/INDEL) with ‘quality score > 250 and allele fraction > 0.3 (heterozygous), > 0.95 (homozygous) and read depth > 10’ does not require additional validation by Sanger sequencing. This validation process is approved by CAP/CLIA. For variants that do not meet the criteria, Sanger sequencing was performed.

3billion uses internally developed software, GEBRA. The variants are associated with genes/diseases that are relevant. The software leverages OMIM, and other publicly available disease databases including publications from PUBMED to provide up-to-date disease information.

The variants are sorted and arranged by: classification by EVIDENCE and the similarity between the given symptoms/phenotypes of the patient and the disease that the variant is associated with. Then, geneticists review, other details. Other additional features (e.g., gene panel) are included in the software. Multiple panels from different sources (e.g., Genomics England PanelApp) are included.

Qualitative variables are presented as absolute frequencies and percentages, while quantitative variables are expressed as mean and standard deviation (SD). Statistical comparisons between cohorts and groups were conducted using chi-squared tests and t-tests, with the significance level set at p < 0.05. The odds ratio (OR) and 95% confidence intervals were calculated to assess the influence of significant factors on positive WES results. Only statistically significant features were included in this analysis. Statistical analysis was performed using STATISTICA 10.

The cohort consisted of 90 children, with a mean age of 6.33 ± 4.69 years, 61 (67.8%) of whom were male (Table 1). There was no difference in age between the patient cohorts, although the second cohort had a higher proportion of boys. All patients had DD of varying degrees and in different areas of development. The most common issue was speech and language delay (83.3%), followed by intellectual disability of varying severity (72.2%). Behavioral problems were observed less frequently (44.4%), including attention deficit hyperactivity disorder (ADHD) in 24.4% of children. Autism spectrum disorders (ASD) were present in 46.6% of patients, and epilepsy or seizures in 27.8%. Facial dysmorphism was observed in 38.9%. Other symptoms were present in 61.1% of the cohort, with the most common being skeletal deformities (in 17 children – 18.9%), hydrocephalus (in 7 children – 7.8%), eczema (in 8 children – 8.9%), hypotonia (in 17 children – 18.9%), and recurrent infections (in 6 children – 6.7%). A positive family history of DD was noted in one-third of the patients.

WES identified pathogenic or likely pathogenic variants in 25 of the 90 patients (27.8%). In addition, 7 patients (7.8%) had inconclusive variants, while 58 patients (64.4%) had no pathogenic variants identified. Among the positive findings, 21 (84.0%) corresponded to the main clinical manifestations in patients, and 4 (16.0%) secondary findings provided new insights into the patient’s conditions. All identified variants and their clinical relevance are shown in Supplementary file, Supplementary Table S1. Among the positive results, there were 4 homozygous variants inherited in an autosomal recessive manner, while the rest were heterozygous variants with autosomal dominant inheritance. One child was found to have two variants - SMC3 and MT-TL1 in mitochondrial DNA, which are associated with Cornelia de Lange syndrome 3 and MELAS. The child exhibited symptoms of both syndromes. The cohort included two siblings who were found to have a very rare condition, interferon regulatory factor 2 binding protein-like–related disorder (IRF2BPL–related disorder) (Shelkowitz et al., 2019), with the boy having severe impairments and the girl having much milder symptoms. In two patients, a pathogenic and a new likely pathogenic variant were identified in the ZMYND11 gene, which is associated with intellectual developmental disorder.

Among the inconclusive variants, 4 were VUS, 1 was likely pathogenic (KDM5C gene, c.1762C > T, p. Gln588Ter variant), 1 was pathogenic (TNFRSF13B, c.542C > A, p. Ala81GLu), and there was a 15q11.2q12q13.1 duplication as the test was not validated to identify copy number variants. All VUS were consistent with the clinical manifestations of the diseases. Patients with Coffin-Siris syndrome were described in a recently published article (Boyarchuk et al., 2024). The likely pathogenic variant in the KDM5C gene was consistent with Intellectual disability, X-linked, syndromic, Claes Jensen type, which is an X-linked recessive disorder. A heterozygous female might be affected due to skewed X-chromosome inactivation or X-chromosomal abnormalities. Additional genetic testing was considered to establish a genetic diagnosis. A heterozygous pathogenic variant identified in the TNFRSF13B gene is associated with autosomal recessive common variable immunodeficiency (OMIM: 240500). Only one heterozygous variant was identified, and the molecular diagnosis is challenging because most of these diseases are inherited in an autosomal recessive manner. Other genetic testing may identify a second variant, such as a deletion, duplication, or deep intronic variant, that is not detectable by this test. However, there is growing evidence suggesting that the disease can manifest even as a heterozygous (Castigli et al., 2005). Chromosomal abnormalities were identified in 3 children (2 duplications of chromosome 15 and 1 deletion of chromosome 22).

Overall, in 81.3% of positive and inconclusive cases, the identified genetic variants correlated with the primary clinical features, leading to a revision of the diagnosis or management plan in 11 out of 32 cases (34.4%). Additionally, in 15 out of 32 cases (46.9%), the genetic diagnosis partially influenced patient management (enabling treatment adjustments, avoiding unnecessary medications and procedures), although it did not significantly impact the disease outcomes. Secondary findings (12.5%) implicated future health monitoring. The WES results showed a significantly higher percentage of positive findings in the first cohort (47.5% vs 12.0%, p < 0.0001). Therefore, we compared the symptoms present in children from the first and second cohorts (Table 1). In the first cohort, failure to thrive (25.0% vs 8.0%, p = 0.0270) and other symptoms (77.5% vs 48.0%, p = 0.0043) were more common. A trend toward more frequent occurrences of epilepsy or seizures (p = 0.0655) and facial dysmorphism (p = 0.0531) also was observed. Conversely, in the second cohort, speech and language delay (96.0% vs 67.6%, p = 0.0003), intellectual disability (82.0% vs 60.0%, p = 0.0206), ASD (62.1% vs 22.5%, p < 0.0001), and ADHD (34.0% vs 12.5%, p = 0.0184) were significantly more frequent. No differences were observed for other signs.

To identify the features that may have most influenced positive outcomes, we compared the observed signs in patients with positive results and inconclusive results (positive cases) correlated with clinical data, as well as in children with negative results (negative cases) (Table 2).

Children with positive cases were slightly older (7.54 vs 5.66 years), although the difference was not significant and only showed a trend (p = 0.0696). Positive cases were observed twice as often in girls (p = 0.0271). Among the 42 children with ASD, only 6 (14.3%) had positive or inconclusive results, which was significantly lower than in children with negative results (p < 0.0001). On the other hand, epilepsy or seizures were more frequently observed in children with positive cases (p = 0.0005). The risk of a positive genetic result in children with DD is higher in the presence of epilepsy or seizures (odds ratio – 5.4444; 95% CI 2.0176 to 14.6918; p = 0.0008). More than 3 dysmorphic features were also more frequently observed in positive cases (p = 0.0022). The risk of positive cases in children with DD is higher in the presence of more than 3 dysmorphic features (odds ratio – 7.1739; 95% CI 1.7791 to 28.9282; p = 0.0056). Although family history was more common in positive cases, the difference was not significant.

Thus, the comprehensive use of WES for diagnosing DD allowed us to identify a range of genetic variants, including very rare syndromes. Additionally, we identified deletion and duplication syndromes, as well as mitochondrial DNA disorder. The study also provided valuable insights into the clinical utility of WES in improving diagnostic yield, particularly in cases with epilepsy or multiple dysmorphic features. Furthermore, the findings highlighted the importance of early genetic testing in guiding clinical management and treatment decisions, which could lead to improved patient outcomes.

The study has some limitations. Firstly, regarding the interpretation of VUS, we conducted only careful clinical correlation without performing additional tests due to limited resources. While WES provided a definitive diagnosis in many cases, the interpretation of some findings remained challenging due to the lack of clear phenotypic-genotypic correlations at the time of testing. The use of WES trio could have improved the detection of new variants or other genes that play a role in the development of DD.