Episignatures in practice: independent evaluation of published episignatures for the molecular diagnostics of ten neurodevelopmental disorders

Variants of uncertain significance (VUS) are a significant issue for the molecular diagnosis of rare diseases. The publication of episignatures as effective biomarkers of certain Mendelian neurodevelopmental disorders has raised hopes to help classify VUS. However, prediction abilities of most published episignatures have not been independently investigated yet, which is a prerequisite for an informed and rigorous use in a diagnostic setting. We generated DNA methylation data from 101 carriers of (likely) pathogenic variants in ten different genes, 57 VUS carriers, and 25 healthy controls. Combining published episignature information and new validation data with a k-nearest-neighbour classifier within a leave-one-out scheme, we provide unbiased specificity and sensitivity estimates for each of the signatures. Our procedure reached 100% specificity, but the sensitivities unexpectedly spanned a very large spectrum. While ATRX, DNMT3A, KMT2D, and NSD1 signatures displayed a 100% sensitivity, CREBBP-RSTS and one of the CHD8 signatures reached <40% sensitivity on our dataset. Remaining Cornelia de Lange syndrome, KMT2A, KDM5C and CHD7 signatures reached 70-100% sensitivity at best with unstable performances, suffering from heterogeneous methylation profiles among cases and rare discordant samples. Our results call for cautiousness and demonstrate that episignatures do not perform equally well. Some signatures are ready for confident use in a diagnostic setting. Yet, it is imperative to characterise the actual validity perimeter and interpretation of each episignature with the help of larger validation sample sizes and in a broader set of episignatures.

Multiple molecular diagnoses in the field of intellectual disability and congenital anomalies: 3.5% of all positive cases

PURPOSE

Wide access to clinical exome/genome sequencing (ES/GS) enables the identification of multiple molecular diagnoses (MMDs), being a long-standing but underestimated concept, defined by two or more causal loci implicated in the phenotype of an individual with a rare disease. Only few series report MMDs rates (1.8% to 7.1%). This study highlights the increasing role of MMDs in a large cohort of individuals addressed for congenital anomalies/intellectual disability (CA/ID).

METHODS

From 2014 to 2021, our diagnostic laboratory rendered 880/2658 positive ES diagnoses for CA/ID aetiology. Exhaustive search on MMDs from ES data was performed prospectively (January 2019 to December 2021) and retrospectively (March 2014 to December 2018).

RESULTS

MMDs were identified in 31/880 individuals (3.5%), responsible for distinct (9/31) or overlapping (22/31) phenotypes, and potential MMDs in 39/880 additional individuals (4.4%).

CONCLUSION

MMDs are frequent in CA/ID and remain a strong challenge. Reanalysis of positive ES data appears essential when phenotypes are partially explained by the initial diagnosis or atypically enriched overtime. Up-to-date clinical data, clinical expertise from the referring physician, strong interactions between clinicians and biologists, and increasing gene discoveries and improved ES bioinformatics tools appear all the more fundamental to enhance chances of identifying MMDs. It is essential to provide appropriate patient care and genetic counselling.

A Solve-RD ClinVar-based reanalysis of 1522 index cases from ERN-ITHACA reveals common pitfalls and misinterpretations in exome sequencing

PURPOSE

Within the Solve-RD project (https://solve-rd.eu/), the European Reference Network for Intellectual disability, TeleHealth, Autism and Congenital Anomalies aimed to investigate whether a reanalysis of exomes from unsolved cases based on ClinVar annotations could establish additional diagnoses. We present the results of the "ClinVar low-hanging fruit" reanalysis, reasons for the failure of previous analyses, and lessons learned.

METHODS

Data from the first 3576 exomes (1522 probands and 2054 relatives) collected from European Reference Network for Intellectual disability, TeleHealth, Autism and Congenital Anomalies was reanalyzed by the Solve-RD consortium by evaluating for the presence of single-nucleotide variant, and small insertions and deletions already reported as (likely) pathogenic in ClinVar. Variants were filtered according to frequency, genotype, and mode of inheritance and reinterpreted.

RESULTS

We identified causal variants in 59 cases (3.9%), 50 of them also raised by other approaches and 9 leading to new diagnoses, highlighting interpretation challenges: variants in genes not known to be involved in human disease at the time of the first analysis, misleading genotypes, or variants undetected by local pipelines (variants in off-target regions, low quality filters, low allelic balance, or high frequency).

CONCLUSION

The "ClinVar low-hanging fruit" analysis represents an effective, fast, and easy approach to recover causal variants from exome sequencing data, herewith contributing to the reduction of the diagnostic deadlock.

OMIXCARE: OMICS technologies solved about 33% of the patients with heterogeneous rare neuro-developmental disorders and negative exome sequencing results and identified 13% additional candidate variants

Purpose: Patients with rare or ultra-rare genetic diseases, which affect 350 million people worldwide, may experience a diagnostic odyssey. High-throughput sequencing leads to an etiological diagnosis in up to 50% of individuals with heterogeneous neurodevelopmental or malformation disorders. There is a growing interest in additional omics technologies in translational research settings to examine the remaining unsolved cases. Methods: We gathered 30 individuals with malformation syndromes and/or severe neurodevelopmental disorders with negative trio exome sequencing and array comparative genomic hybridization results through a multicenter project. We applied short-read genome sequencing, total RNA sequencing, and DNA methylation analysis, in that order, as complementary translational research tools for a molecular diagnosis. Results: The cohort was mainly composed of pediatric individuals with a median age of 13.7 years (4 years and 6 months to 35 years and 1 month). Genome sequencing alone identified at least one variant with a high level of evidence of pathogenicity in 8/30 individuals (26.7%) and at least a candidate disease-causing variant in 7/30 other individuals (23.3%). RNA-seq data in 23 individuals allowed two additional individuals (8.7%) to be diagnosed, confirming the implication of two pathogenic variants (8.7%), and excluding one candidate variant (4.3%). Finally, DNA methylation analysis confirmed one diagnosis identified by genome sequencing (Kabuki syndrome) and identified an episignature compatible with a BAFopathy in a patient with a clinical diagnosis of Coffin-Siris with negative genome and RNA-seq results in blood. Conclusion: Overall, our integrated genome, transcriptome, and DNA methylation analysis solved 10/30 (33.3%) cases and identified a strong candidate gene in 4/30 (13.3%) of the patients with rare neurodevelopmental disorders and negative exome sequencing results.

Genotype-first in a cohort of 95 fetuses with multiple congenital abnormalities: when exome sequencing reveals unexpected fetal phenotype-genotype correlations

PURPOSE

Molecular diagnosis based on singleton exome sequencing (sES) is particularly challenging in fetuses with multiple congenital abnormalities (MCA). Indeed, some studies reveal a diagnostic yield of about 20%, far lower than in live birth individuals showing developmental abnormalities (30%), suggesting that standard analyses, based on the correlation between clinical hallmarks described in postnatal syndromic presentations and genotype, may underestimate the impact of the genetic variants identified in fetal analyses.

METHODS

We performed sES in 95 fetuses with MCA. Blind to phenotype, we applied a genotype-first approach consisting of combined analyses based on variants annotation and bioinformatics predictions followed by reverse phenotyping. Initially applied to OMIM-morbid genes, analyses were then extended to all genes. We complemented our approach by using reverse phenotyping, variant segregation analysis, bibliographic search and data sharing in order to establish the clinical significance of the prioritised variants.

RESULTS

sES rapidly identified causal variant in 24/95 fetuses (25%), variants of unknown significance in OMIM genes in 8/95 fetuses (8%) and six novel candidate genes in 6/95 fetuses (6%).

CONCLUSIONS

This method, based on a genotype-first approach followed by reverse phenotyping, shed light on unexpected fetal phenotype-genotype correlations, emphasising the relevance of prenatal studies to reveal extreme clinical presentations associated with well-known Mendelian disorders.

Excess of de novo variants in genes involved in chromatin remodelling in patients with marfanoid habitus and intellectual disability

PURPOSE

Marfanoid habitus (MH) combined with intellectual disability (ID) (MHID) is a clinically and genetically heterogeneous presentation. The combination of array CGH and targeted sequencing of genes responsible for Marfan or Lujan-Fryns syndrome explain no more than 20% of subjects.

METHODS

To further decipher the genetic basis of MHID, we performed exome sequencing on a combination of trio-based (33 subjects) or single probands (31 subjects), of which 61 were sporadic.

RESULTS

We identified eight genes with de novo variants (DNVs) in at least two unrelated individuals (ARID1B, ATP1A1, DLG4, EHMT1, NFIX, NSD1, NUP205 and ZEB2). Using simulation models, we showed that five genes (DLG4, NFIX, EHMT1, ZEB2 and ATP1A1) met conservative Bonferroni genomewide significance for an excess of the observed de novo point variants. Overall, at least one pathogenic or likely pathogenic variant was identified in 54.7% of subjects (35/64). These variants fell within 27 genes previously associated with Mendelian disorders, including NSD1 and NFIX, which are known to be mutated in overgrowth syndromes.

CONCLUSION

We demonstrated that DNVs were enriched in chromatin remodelling (p=2×10^-4) and genes regulated by the fragile X mental retardation protein (p=3×10^-8), highlighting overlapping genetic mechanisms between MHID and related neurodevelopmental disorders.

Increased diagnostic and new genes identification outcome using research reanalysis of singleton exome sequencing

In clinical exome sequencing (cES), the American College of Medical Genetics and Genomics recommends limiting variant interpretation to established human-disease genes. The diagnostic yield of cES in intellectual disability and/or multiple congenital anomalies (ID/MCA) is currently about 30%. Though the results may seem acceptable for rare diseases, they mean that 70% of affected individuals remain genetically undiagnosed. Further analysis extended to all mutated genes in a research environment is a valuable strategy for improving diagnostic yields. This study presents the results of systematic research reanalysis of negative cES in a cohort of 313 individuals with ID/MCA. We identified 17 new genes not related to human disease, implicated 22 non-OMIM disease-causing genes recently or previously rarely related to disease, and described 1 new phenotype associated with a known gene. Twenty-six candidate genes were identified and are waiting for future recurrence. Overall, we diagnose 15% of the individuals with initial negative cES, increasing the diagnostic yield from 30% to more than 40% (or 46% if strong candidate genes are considered). This study demonstrates the power of such extended research reanalysis to increase scientific knowledge of rare diseases. These novel findings can then be applied in the field of diagnostics.

Secondary actionable findings identified by exome sequencing: expected impact on the organisation of care from the study of 700 consecutive tests

With exome/genome sequencing (ES/GS) integrated into the practice of medicine, there is some potential for reporting incidental/secondary findings (IFs/SFs). The issue of IFs/SFs has been studied extensively over the last 4 years. In order to evaluate their implications in care organisation, we retrospectively evaluated, in a cohort of 700 consecutive probands, the frequency and burden of introducing the search for variants in a maximum list of 244 medically actionable genes (genes that predispose carriers to a preventable or treatable disease in childhood/adulthood and genes for genetic counselling issues). We also focused on the 59 PharmGKB class IA/IB pharmacogenetic variants. We also compared the results in different gene lists. We identified variants (likely) affecting protein function in genes for care in 26 cases (3.7%) and heterozygous variants in genes for genetic counselling in 29 cases (3.8%). Mean time for the 700 patients was about 6.3 min/patient for medically actionable genes and 1.3 min/patient for genes for genetic counselling, and a mean time of 37 min/patients for the reinterpreted variants. These results would lead to all 700 pre-test counselling sessions being longer, to 55 post-test genetic consultations and to 27 secondary specialised medical evaluations. ES also detected 42/59 pharmacogenetic variants or combinations of variants in the majority of cases. An extremely low metabolizer status in genes relevant for neurodevelopmental disorders (CYP2C9 and CYP2C19) was found in 57/700 cases. This study provides information regarding the need to anticipate the implementation of genomic medicine, notably the work overload at various steps of the process.

ZMIZ1 Variants Cause a Syndromic Neurodevelopmental Disorder

ZMIZ1 is a coactivator of several transcription factors, including p53, the androgen receptor, and NOTCH1. Here, we report 19 subjects with intellectual disability and developmental delay carrying variants in ZMIZ1. The associated features include growth failure, feeding difficulties, microcephaly, facial dysmorphism, and various other congenital malformations. Of these 19, 14 unrelated subjects carried de novo heterozygous single-nucleotide variants (SNVs) or single-base insertions/deletions, 3 siblings harbored a heterozygous single-base insertion, and 2 subjects had a balanced translocation disrupting ZMIZ1 or involving a regulatory region of ZMIZ1. In total, we identified 13 point mutations that affect key protein regions, including a SUMO acceptor site, a central disordered alanine-rich motif, a proline-rich domain, and a transactivation domain. All identified variants were absent from all available exome and genome databases. In vitro, ZMIZ1 showed impaired coactivation of the androgen receptor. In vivo, overexpression of ZMIZ1 mutant alleles in developing mouse brains using in utero electroporation resulted in abnormal pyramidal neuron morphology, polarization, and positioning, underscoring the importance of ZMIZ1 in neural development and supporting mutations in ZMIZ1 as the cause of a rare neurodevelopmental syndrome.

Autosomal recessive variations of TBX6, from congenital scoliosis to spondylocostal dysostosis

Proximal 16p11.2 microdeletions are recurrent microdeletions with an overall prevalence of 0.03%. In patients with segmentation defects of the vertebra (SDV), a burden of this microdeletion was observed with TBX6 as a candidate gene for SDV. In a published cohort of patients with congenital scoliosis (CS), TBX6 haploinsufficiency was compound heterozygous with a common haplotype. Besides, a single three-generation family with spondylocostal dysostosis (SCD) was reported with a heterozygous stop-loss of TBX6. These observations questioned both on the inheritance mode and on the variable expressivity associated with TBX6-associated SDV. Based on a national recruitment of 56 patients with SDV, we describe four patients with variable SDV ranging from CS to SCD associated with biallelic variations of TBX6. Two patients with CS were carrying a proximal 16p11.2 microdeletion associated with the previously reported haplotype. One patient with extensive SDV was carrying a proximal 16p11.2 microdeletion associated with a TBX6 rare missense change. One patient with a clinical diagnosis of SCD was compound heterozygous for two TBX6 rare missense changes. The three rare variants were affecting the chromatin-binding domain. Our data illustrate the variable expressivity of recessive TBX6 ranging from CS to SCD.

The first familial case of inherited 2q37.3 interstitial deletion with isolated skeletal abnormalities including brachydactyly type E and short stature

Albright hereditary osteodystrophy (AHO)-like syndrome is also known as brachydactyly-mental retardation syndrome (BDMR; OMIM 60040). This disorder includes intellectual disability in all patients, skeletal abnormalities, including brachydactyly E (BDE) in approximately half, obesity, and facial dysmorphism. Patients with 2q37 microdeletion or HDAC4 mutation are defined as having an AHO-like phenotype with normal stimulatory G (Gs) function. HDAC4 is involved in neurological, cardiac, and skeletal function. This paper reports the first familial case of 2q37.3 interstitial deletion affecting two genes, HDAC4 and TWIST2. Patients presented with BDE and short stature without intellectual disability, showing that haploinsufficiency of the HDAC4 critical region may lead to a spectrum of phenotypes, ranging from isolated brachydactyly type E to BDMR.