ABSTRACT
Objective:
Congenital hyperinsulinism (CHI) is the most frequent cause of severe and persistent hypoglycaemia from birth. Understanding the pathophysiology and genetic defects behind hyperinsulinism and its complications provides clues to timely diagnosis and management. The aim of this study was to evaluate the underlying genetic aetiology of a specific Iranian pediatric cohort with CHI.
Methods:
A total of 44 unrelated children, 20 girls and 24 boys, with an initial diagnosis or history of CHI from all regions of Iran were recruited between 2016 and 2019. Targeted next generation sequencing (tNGS) was performed for the genes found in about half of CHI patients.
Results:
Mutations were identified in 24 cases (55%). Patients with a confirmed genetic cause were mainly diagnosed below age of one year old (p=0.01), had fewer other syndromic features, excluding seizure, (p=0.03), were less diazoxide responsive (p=0.04) and were more diazoxide unresponsive leading to pancreatectomy (p=0.007) compared to those with no identified mutations. Among 24 patients with identified genetic mutations, 17 (71%) had a mutation in ABCC8, 3 (12%) in KCNJ11, 3 (12%) in HADH, and 1 patient had a mutation in KMT2D. These included five novel mutations in ABCC8, KCNJ11, and KMT2D.
Conclusion:
This is the biggest genetic study of CHI in Iran. A high frequency of recessive forms of CHI, especially HADH mutations, in our study could be due to a high rate of consanguineous marriage. We recommend tNGS to screen for all the CHI genes.
What is already known on this topic?
It is well known that congenital hyperinsulinism (CHI) is the most frequent cause of severe and persistent hypoglycaemia in the neonatal period, infancy, and childhood. To date, mutations in at least nine different genes have been reported to cause CHI: ABCC8, KCNJ11, GLUD1, GCK, HADH, HNF4A, SLC16A1, HNF1A and UCP2. Data are mainly limited to European populations while the occurrence of the pathogenic mutations underlying CHI are higher in consanguineous families which are more prevalent in Asian societies.
What this study adds?
We report the frequency of causal gene mutations in a cohort of Iranian children with a diagnosis of CHI and add five novel mutations. Based on our findings we recommend screening of HADH gene variants in all patients with diazoxide-responsive CHI if there is no access to targeted next generation sequencing.
Introduction
Congenital hyperinsulinism (CHI) is the most frequent cause of severe and persistent hypoglycaemia and the most common metabolic abnormality in the neonatal period, infancy, and childhood (1,2). There is an unregulated secretion of insulin from pancreatic β-cells in the course of low blood glucose in CHI patients, leading to severe and persistent hypoglycaemia due to genetic defects (3,4). CHI is rare but a very high incidence has been reported in isolated European populations and also communities with a high rate of consanguinity (5,6,7). About 60% of infants with CHI develop hypoglycaemia during the first months of life. This condition typically goes through remission and flare-up cycles due to hypoglycaemia. Age of onset is variable and symptoms can range from asymptomatic and mild to severe symptoms, including medically unresponsive hypoglycaemia (5,6).
The clinical manifestations, histological subtypes and underlying molecular mechanisms of CHI are heterogeneous (4). Based on histological assessments, there are two major subtypes of CHI including diffuse and focal forms (3,8). The differentiation between these two subtypes is important for clinical management. The diffuse form, where abnormality is in all pancreatic β-cells, is inherited in an autosomal recessive or dominant mode, most commonly due to mutations in ABCC8 and KCNJ11. Recessive mutations in ABCC8/KCNJ11 are usually severe and require high concentrations of intravenous glucose to maintain normoglycemia while dominant mutations usually cause milder disease. The focal form is confined to a small region of the pancreas with sporadic inheritance. This subtype results from a heterozygous paternal mutation in ABCC8/KCNJ11 and somatic loss of maternal chromosome 11p15 in the focal pancreatic lesion (9).
Understanding the pathophysiology and genetic defects behind hyperinsulinism and its complications has provided clues to diagnosis and management of the disease. Since CHI causes a set of complex and heterogenous metabolic complications, cases will have different clinical presentation with different age of onset and prognosis. The conservative treatment includes diazoxide as first choice. However, some children, mainly those with the recessive mutations in ABCC8/KCNJ11, develop a form of disease unresponsive to medical therapy and so pancreatectomy is the only option (3,10).
Since severe hypoglycaemia has a negative effect on neural system function, especially during the early years of life, early diagnosis and treatment are important. In recent decades there has been a substantial expansion in information of genetic defects leading to CHI. To date, mutations in at least nine different genes have been reported to cause CHI: ABCC8, KCNJ11, GLUD1, GCK, HADH, HNF4A, SLC16A1, HNF1A and UCP2 (4,9). These genes are involved in regulating insulin secretion from β-cells (4). Data about CHI are mainly limited to European populations. However, the occurrence of the pathogenic mutations underlying CHI is increased in consanguineous families compared with non-consanguineous families (4). Therefore, investigating the aetiologies of this disorder is even more momentous in Asian countries with highly consanguineous populations, including Iran.
The aim of this study was to assess genetic mutations underlying CHI by recruiting individuals diagnosed with CHI. All the genetic and clinical data was combined to provide an Iranian CHI database for patients and specialists. We also report the frequency of causal gene mutations and describe a number of novel mutations.
Methods
Study Participants
Unrelated participants diagnosed with CHI were recruited to the study. Patients were from all regions of Iran and who had been referred to two centres in Iran, Imam Reza Hospital, Mashhad, Iran and the Division of Endocrinology and Metabolism in the Department of Paediatrics at the Children’s Medical Centre in Tehran, Iran (Table 1). Clinical information was supplied by the referring clinicians. Informed consent was obtained from parents on behalf of their children. Peripheral blood samples were collected from affected participants and their parents at the time of referral and used to perform genetic testing.
Clinical Data
CHI was defined as fasting hypoglycaemia [glucose <50 mg/dL (2.8 mmol/l)] occurring simultaneously with an inappropriately detectable plasma insulin (>2.0 μU/mL) (11). Paediatric subjects whose hyperinsulinemic hypoglycemia (HH) did not remit after at least three months follow-up were eligible to be enrolled in this study. A detailed clinical and demographical history, along with EDTA blood sample from the affected individual, both parents and affected siblings, were obtained. In the clinical history, diazoxide-responsiveness was defined as the ability to achieve elevated intravenous glucose and maintain normoglycemia (12). Low birth weight was determined based on birth weight adjusted by the gestational age (13).
Genetic Analysis
The genetic testing and variant interpretation were performed by the Exeter Molecular Genetics Laboratory (Exeter, UK). Briefly, DNA was extracted using standard methods and the samples were analysed for coding and flanking intronic regions of the KCNJ11 (NM_000525.3) and ABCC8 (NM_001287174.1) genes by Sanger sequencing. If no mutation was found, targeted next generation sequencing (tNGS) (Agilent custom capture v5.3/Illumina NextSeq500) for the coding regions and exon/intron boundaries of the genes found in about half of CHI patients (ABCC8, KCNJ11, AKT2, GLUD1, GCK, GPC3, HADH, HNF4A, KDM6A, KMT2D, SLC16A1, CACNA1D, PMM2, TRMT10A and HNF1A) (14) was performed. This assay can also detect partial/whole gene deletions and duplications (15). For a patient with mosaicism in KMT2D, a confirmatory dosage analysis of exons 51-54 of the KMT2D genes (NM_003482.3) by Droplet Digital PCR using EvaGreen was performed.
Variants were classified according to the American College of Medical Genetics and Genomics and the Association for Molecular Pathology standards and guidelines for the interpretation of sequence variants (16). The frequencies of the identified variants were checked in GnomAD [>120 000 individuals (http://gnomad.broadinstitute.org)] and in human variant and mutation databases, such as ClinVar and Human Gene Mutation Database, as well as in the literature via PubMed and Google searches. The in silico tools SIFT, PolyPhen-2 and Align-GVGD were used to assess the pathogenicity of missense variant effects, and the prediction of variant effect on mRNA splicing was made using SpliceSiteFinder-like, MaxEntScan, GeneSplice, NNSPLICE and Human Splicing Finder. All in silico programs were accessed through the ALAMUT Visual software version 2.7.1 (Interactive Biosoftware, Rouen, France). Conservation of amino acids and nucleotides across multiple species was performed using the University of California Santa Cruz genome browser (http://genome.ucsc.edu).
Ethical Considerations
The study was approved by the Ethical Committee of the Endocrinology and Metabolism Research Institute (ethical code: IR.TUMS.EMRI.REC.1397.009, date: 18.07.2018). The consent form was signed by all participants, or in the case of minors, the consent form was signed by their parents or legal guardian. A signed, written consent form was separately obtained for genetic testing. All procedures performed in this study were in accordance with the ethical standards of the 1964 Helsinki Declaration and its later amendments or comparable ethical standards.
Statistical Analysis
We used the chi-square test to assess the differences between patients with confirmed genetic cause and those with no identified mutations groups.
Result
Patient Characteristics
Forty-four unrelated children, 20 (45.45%) girls, with an initial diagnosis or history of CHI from all regions of Iran, diagnosed between 2016 and 2019 were recruited. Of the 44, 21 children (48%) came from consanguineous families (Table 1). Age of diagnosis varied but 36 (81.8%) were diagnosed before their first birthday. Only two (4.5%) patients had a positive family history of hypoglycaemia and 13 (29.5%) had a family history of diabetes. Low birth weight was reported for 10 (22.7%) patients. Insulin level ranged from 7.2 to 147 mU/L within our population and pancreatectomy had already been performed in eight (18.2%). Other symptoms, including hypokalaemia, autism, renal failure, seizure, oesophageal atresia, and failure to thrive were observed in 14 (31.8%). Nine out of 44 patients (20.45%) were Diazoxide-unresponsive including eight patients who had a history of pancreatectomy.
Disease causing mutations were identified in 24 cases (55%). Patients with a confirmed genetic cause in any of the known genes, were mainly diagnosed below the age of one year old (p=0.01), had fewer other syndromic features, excluding seizure (p=0.03), were less diazoxide responsive (p=0.04) and were more diazoxide unresponsive leading to pancreatectomy (p=0.007) compared to those with no identified mutations. Birth weight, hypoglycaemia and consanguinity rate was similar between the two groups (Tables 1, 2, 3).
Genetic Findings
Among 24 patients with identified genetic mutations, 17 (71%) had a mutation in ABCC8, 3 (12%) in KCNJ11, 3 (12%) in HADH, and one patient had a mutation in KMT2D (Table 1). These included five novel mutations in ABCC8, KCNJ11, and KMT2D. Details of the mutations and the clinical features of the patients are described below and in Table 2.
Patients with ABCC8 mutations. A total of 17 probands, including eight males and nine females were found with mutations in ABCC8 gene, of whom two patients were carrying novel mutations; a homozygous in-frame deletion (c4724-4732del) and a heterozygous missense de novo mutation (c.1109G>C; p.Arg370Thr). Both of these patients were diazoxide responsive. In total, five patients had heterozygous mutations, one compound heterozygous and 11 patients had homozygous mutations. Ten patients (59%) were from consanguineous families. The median (interquartile range) birth weight was 3550 (3200-4200) g. Eight patients (47%) were diazoxide unresponsive and had undergone pancreatectomy. Other clinical features included seizure in two (11.8%) patients.
Patients with KCNJ11 mutations. All three patients were homozygous, including two with novel missense mutations (c.362T>G; p.Phe121Cys and c.370T>A; p.Ser124Thr) who were both diazoxide responsive. Birth weight was between 3800 g and 4880 g. The patient with c.287_288delinsTG mutation (p.Ala96Val) had a history of seizure, was diazoxide unresponsive and underwent pancreatectomy.
Patients with HADH mutations. Three cases were detected with HADH gene mutations, of whom one case was homozygous for a frameshift variant (c.617del; p.Lys206fs) and the other two were homozygous for a nonsense variant (c.706C>T; p.Arg236Ter). Two patients were from consanguineous families and had seizures. Birth weight ranged from 3250g to 4800g. The response to diazoxide was good in all the patients.
Patient with KMT2D mutation. One patient was mosaic for a KMT2D partial gene deletion of exons 51-54. The level of mosaicism within his leukocyte DNA was estimated to be at least 20%, consistent with a post-zygotic origin. Genetic testing in the parents indicated a de novo change. The patient was from a non-consanguineous family and was diagnosed with hyperinsulinism at the age of two years. Birth weight was 4500g and the response to diazoxide was good. The patient had a history of seizure and did not have any facial dysmorphism.
Diazoxide Responsiveness and Pancreatectomy
Diazoxide unresponsiveness was seen in nine (38%) of the patients with genetic mutations. These included eight patients (89%) with mutations in ABCC8 and one with KCNJ11 mutation. This figure means 45% of those with KATP-channel mutations were diazoxide-unresponsive. Seven out of eight (88%) patients who had had pancreatectomy had ABCC8 gene mutations and one had KCNJ11 mutation. Pathogenic mutation was invariably found in diazoxide-unresponsive patients (9/9; 100%), although more than half of cases in the diazoxide-responsive group (20/35; 57%) had no genetic variant identified in the genes investigated (Figure 1).
Discussion
We described the spectrum of genetic mutations in CHI in an Iranian population, as well as the frequency of each mutation and their related clinical features. Genetic mutations were found in 55% of our patients, consistent with previous studies which identified mutations in 12 out of 19 (63%) patients in Turkey (10) or in only 47% (56 of 118) of the diazoxide-responsive cases in the US (17). This figure is lower than some studies, which could reflect differences in referral patterns and inclusion or exclusion criteria (18).
Consistent with previous studies (17,19,20), mutations in KATP-channel including SUR1 (ABCC8) and Kir6.2 (KCNJ11), accounted for 83% of all CHI-causing mutations in our study. Mutations in the ABCC8 gene with 71% occurrence was the most frequent cause of CHI in our population. In addition, eleven participants were homozygous for a mutation in ABCC8 which confirmed a diagnosis of autosomal recessive CHI. One patient was homozygous for a novel, in-frame deletion in ABCC8, which requires further investigation to determine its clinical significance.
The majority of recessive mutations in the KATP-channel has been shown to lead to medically unresponsive CHI (21,22), but in our study, six out of 11 patients with homozygous ABCC8 mutations and two out of three patients with homozygous KCNJ11 mutations were diazoxide-responsive. Although Salomon-Estebanez et al (23) had suggested that this may be due to the reduction of severity of disease over time, it is not reasonably likely and may be due to variable criteria used to determine diazoxide responsiveness. In contrast, ABCC8 and KCNJ11 heterozygous mutations were characterized by various presentations and treatment responses; three out of six (50%) patients with heterozygous mutations were diazoxide-unresponsive which is in agreement with other studies (21). In our study, all heterozygous mutations were in the ABCC8 gene, which may indicate a dominant pattern of inheritance for certain mutations in the ABCC8 gene.
Although, a subtotal pancreatectomy is necessary in most cases of CHI (24), the vast majority of our participants were diazoxide-responsive (79%). Previous studies have reported different rates of diazoxide responsiveness in their cohorts. In studies of de Lonlay et al (8) and Snider et al (17) the ratio of diazoxide-responsive participants were 28% and 34%, respectively. In a study from Turkey 59% (13/22) of patients were diazoxide-responsive (10). In 27 infants who were born small-for-gestational age in the UK and developed HH, diazoxide-response was 100% (25). Diazoxide-responsiveness in two studies of Iranian patients has been reported to be 18/23 and 3/6 (18,26). This discrepancy is probably due to the number of cases in some studies, variable criteria used to define diazoxide responsiveness, differences in sensitivity of the methods used for detecting the mutations and/or occurrence of mutations in combination with a second mutant allele that complicated the prediction of consequences. Our results, however, are consistent with previous studies (8,27), and indicate that ABCC8 gene defects are the most important cause of diazoxide-unresponsive CHI (8/9 of diazoxide-unresponsive patients) in Iranian children.
The three patients with HADH homozygous mutations were all diazoxide-responsive. This is consistent with other studies showing hyperinsulinism due to HADH gene mutations responds relatively well to diazoxide (12,18,24). Mutations in the HADH gene result in a diffuse form of CHI (28). Interestingly, the rate of observed HADH mutations in our study was the same as for KCNJ11 mutations. This result is in concordance with some other studies (18,26). A high frequency of HADH mutations in our study could be due to a high rate of consanguineous marriage. Based on our results, screening for HADH gene variants is recommended in all patients with diazoxide-responsive CHI.
One patient in our cohort was mosaic for a KMT2D partial gene deletion. Pathogenic variants in the KMT2D gene cause Kabuki syndrome (29). People with Kabuki syndrome also have facial and some other specific congenital anomalies, including growth delays, mental retardation and skeletal abnormalities, which were not observed in our participant.
In nearly 20% of patients, no mutation was detected using tNGS, which may miss large deletions and chromosomal rearrangements. Furthermore, negative results do not exclude a monogenic aetiology and further unidentified mutations in other unidentified genes should be considered as new aetiologies.
Study Limitations
The strength of our study is that we have collected genetic data on the largest cohort of CHI in an Iranian population to date and identified novel variants causing the disease. Our samples were referred from different parts of Iran. Study limitations include the retrospective collection of clinical data and the lack of follow-up data to assess detailed clinical significance of novel findings.
Conclusion
Most novel mutations identified in this study were inherited in a homozygous fashion. Variation in reported rates of diazoxide responsiveness suggest a need for revision of the criteria used to define diazoxide responsiveness. All novel mutations that we report for the first time were medically responsive, and yet should now be considered in CHI analysis. More studies on the molecular basis of CHI are necessary for societies with highly consanguineous families. We recommend tNGS for all CHI patients but screening of HADH gene variants in patients with diazoxide-responsive CHI seems to be required if there is no access to tNGS.