Introduction

Hyperinsulinemic hypoglycaemia (HH) is the most frequent cause of persistent hypoglycaemia in neonates and infants. The most severe forms of HH are inherited and the term congenital hyperinsulinism (CHI) refers to this form. HH can be secondary to various risk factors like perinatal asphyxia, maternal diabetes, intrauterine growth restriction and maternal medications. Various developmental syndromes like Beckwith-Wiedemann, Kabuki or Sotos syndrome have also been associated with HH [1, 2].

CHI compromises a group of different genetic disorders with the common finding of recurrent episodes of hypoglycaemia due to inappropriate insulin secretion. The estimated incidence of CHI is 1/50,000 live births and up to 1/2500 in areas with high rates of consanguinity [3]. Mutations in key genes, which play a role in the regulation of insulin secretion, constitute the underlying molecular genetics of CHI. Mutations in 12 different genes (ABCC8, KCNJ11, GLUD1, GCK, HADH, SLC16A1, UCP2, HNF4A HNF1A, HK1, PGM1 and PMM2), that lead to dysregulated insulin secretion, had been described [2]. Recently, mutations in two novel genes have been linked to CHI: calcium voltage-gated channel subunit alpha1 D (CACNA1D) and forkhead box A2 transcription factor (FOXA2) genes [4].

In the last decade, more than 200 different mutations in ABCC8 compared to 30 distinct KCNJ11 mutations, which respectively encode for the SUR1 and Kir6.2 subunits of the K-ATP channel, were estimated to account for 40–45% of CHI cases (82% of diazoxide-unresponsive patients) [5], whereas mutations have been identified in other genes in approximately 5 to 10% of the cases. The aetiology for the remaining 40% of patients was still unknown [6].

Hypoglycaemia is the main feature of CHI with a high risk of seizure and neurological damage. Symptoms of hypoglycaemia are diverse and vary according to the severity and age of patient. They range from asymptomatic, incidental finding on routine glucose monitoring to life threatening apnea, coma or status epilepticus [3]. The clinical presentation is most severe in the new-born period and may be quite subtle later in infancy and childhood. Even within same family, the severity of the disease can vary substantially [7]. The early diagnosis and immediate meticulous management are the corner stones for preventing brain injury in patients with CHI.

Diazoxide is the mainstay of medical treatment and first-line therapy. It is a ligand of the K-ATP which activates intact K-ATP channels reversing glucose-induced channel closure. Diazoxide is ineffective in diffuse CHI due to recessive inactivating mutations of ABCC8 and KCNJ11 and in patients with focal CHI [8].

Recent advances in molecular genetic analysis and 18-fluro L-dihydroxy phenylalanine (18F-DOPA) positron emission tomography (PET CT) imaging have fundamentally changed the clinical approach to patients with CHI and paternally inherited mutations, especially in ABCC8 gene, who more likely harbour a focal lesion on the scan [9, 10].

The aim of the study was to assess the management and outcome of patients with CHI referred to Royal Hospital for Children, Greater Glasgow and Clyde.

Methods

This is a retrospective study of children referred to or attending our tertiary endocrine service with refractory hypoglycaemia and documented CHI between 2009 and 2017.

Diagnosis was based on documented hypoglycaemia, in the presence of an intravenous glucose infusion rate of > 8 mg/kg/min, and/or detectable insulin or C-peptide levels with low free fatty acids and low ketone bodies.

Gestational age, age at presentation, perinatal risk factors, family history, birth weight, maximum glucose infusion rate and initial laboratory investigations were all recorded. Diazoxide responsiveness and side effects, together with the use of other medications and surgical intervention, were also documented. Genetic analysis was performed in the Exeter Molecular Genetic Laboratory using custom designed Agilent SureSelect Target Enrichment which allows detection of base substitutions (SNVs), small insertions and deletions (indels) and partial/whole gene deletions and duplications (CNVs), after obtaining written informed consent from parents.

Patients with developmental delay or behavioural disorders have been assessed through Child Development services.

Statistics

Chi-square test and Mann-Whitney test were used for data analyses, using IBM SPSS (version 22.0). Significant level was set as p < 0.05.

Results

A total of 39 patients (25 males) were identified during study period (Fig. 1). The median of their current age is 2.42 years (range 0.16–20.36).

Fig. 1
figure 1

Patient referred to hyperinsulinism clinic during study period

Full size image

All had documented HH. Six patients had risk factors (intrauterine growth restriction, small for gestational age (SGA), maternal beta-blockers), required limited diazoxide treatment (< 3 months), had no genetic analysis performed and therefore classified as secondary hyperinsulinism. One patient with Beckwith-Wiedemann syndrome was also excluded.

Clinical characteristics

Regarding the other 32 patients, 26 (81.2%) presented within the first 72 h of life and 27 (84.3%) were born at term. Birth weight ranged between 1916 and 4610 g with a median of 3201 g. Of the 32 patients, 20 (62.5%) were born with an appropriate weight for gestational age (AGA), 5 (15.6%) were large for gestational age (LGA), and one being macrosomic (birth weight > 4500 g (as defined by American College of Obstetricians and Gynecologists)).

Maximum glucose requirement ranged between 6.7 and 18.8 mg/kg/min (median 13.1 mg/kg/min). Insulin and /or C-peptide levels were measured in 28, with 25 having insulin and three with C-peptide alone. Insulin or C-peptide levels were not available in four patients. In these four patients, CHI was diagnosed based on high glucose requirement, low FFAs and low serum ketones in the presence of hypoglycaemia. Insulin levels ranged from 1.3 to 110 mIU/L (median of 11.95 mIU/L).

Management

Diazoxide was started in all patients with most (27) having a baseline echocardiogram and most responding well to treatment. Seven did not respond to diazoxide and required octreotide and/or percutaneous endoscopic gastrostomy (PEG) tube feeding. Six of them ended up requiring surgery.

Adverse effects secondary to diazoxide therapy were observed in four patients, mainly pulmonary hypertension/fluid overload in 3/4, with ileus in the other patient. Diazoxide was discontinued in patients who developed pulmonary hypertension and they were managed with octreotide and/or high caloric intake.

Genetic mutations were detected in 12/32 (37.5%) (KCNJ11-5, ABCC8-4 and GLUD 1-3) (Table 1).

Table 1 Characteristics of patients with genetic mutations
Full size table

K-ATP channel mutations were heterozygous (mono-allelic) except in two siblings of consanguineous parents who had a novel homozygous missense mutation (KCNJ11-A161V).

Patients with genetic mutations were mainly males 8/12 (66.6%) and born AGA 9/12 (75%) compared to 10/20 (50%) and 11/20 (55%) respectively in those with no gene mutation with no significant difference in birth weight between both groups (p = 0.223) (Fig. 2).

Fig. 2
figure 2

Distribution of birth weight among patients with and without genetic mutations

Full size image

18F-DOPA PET CT scan was performed in all seven patients with paternally inherited mutations of the ABCC8 and KCNJ11 genes and two diazoxide-unresponsive patients requiring PEG and/or octreotide and having no identifiable mutation.

CHI resolved within 12 months in 10/32 (31.25%) including three patients with paternally inherited KCNJ11 mutations, all of which had undetectable focal lesions on 18F-DOPA PET CT and did not require surgery.

14/32 (43.75%) required more than 12 months of medication with 10/14 having no identifiable mutation.

There was no correlation between the presence of gene mutation and treatment duration (p = 0.687). Also, maximum glucose infusion rate and insulin levels at diagnosis had no significant correlation with the duration of medical treatment (p = 0.423 and 0.341 respectively).

A total of seven patients underwent pancreatectomy, three had subtotal/near-total pancreatectomy in early infancy and still required medical treatment post pancreatectomy. The one who required near-total pancreatectomy had diffuse disease due to homozygous KCNJ11 novel missense mutation and still required octreotide postoperative, developing diabetes at age of 9 years (his sibling with same mutation was managed conservatively on combination of octreotide subcutaneous injections and overnight frequent bolus feeding through PEG tube). Four patients had focal pancreatectomies, three had paternally inherited mutations (ABCC8) and one with focal lesion, but no mutation identified. Neurodevelopmental deficits and behavioural disorders were detected in 6/32 (19%). Two have global developmental delay, three diagnosed with autistic spectrum disorders and one patient with GLUD1 mutation had absence seizures and learning difficulties.

Discussion

Patients with CHI were almost all born at term and AGA. Patients with genetic mutations were mainly males (66.6%) and born AGA (75%) compared to 50% males and 55% AGA in those with no gene mutation. There was no significant difference in birth weight between both groups. As K-ATP mutations are the main genetic abnormalities in our cohort, these results are in contrast to another study which highlighted statistically significant increased birth weight and younger age of presentation in K-ATP mutation-positive group as compared to K-ATP mutation-negative patients [11]. This may be related to K-ATP mutations being mainly homozygous or compound heterozygous in the Turkish cohort.

In the current study, 37.5% had a genetically confirmed diagnosis in comparison to 45.3%, 51.4%, 66.7% and 79% respectively in four different studies [8, 11,12,13]. Mutations in the K-ATP channel genes were the commonest (75%) and were detected mainly in patients with diazoxide-unresponsiveness/diazoxide-related adverse effects.

In the set of patients who were diazoxide-responsive, no identifiable mutations in the known genes were found in 68.7% of patients, in comparison to 28% in an Iranian cohort of CHI [14], and 80% in a larger Italian cohort [12].

18F-DOPA PET CT detected a focal lesion in all patients with paternally inherited ABCC8 mutation. A focal lesion was also identified in one patient who was diazoxide unresponsive in the absence of an identifiable genetic mutation.

The three patients with paternally inherited mutations in KCNJ11 had no focal lesions identified, however were still managed conservatively and off medication within the first year of life. This data supports previously reported data that a significant proportion, even as many as half, with paternal heterozygous K-ATP mutations may have diffuse CHI [15], and that resolution of CHI occurs in a significant proportion of those patients who were safely managed by conservative medical treatment [14, 16, 17].

Diazoxide-related circulatory side effects or cardiotoxicity were observed in 9.3% (3/32) of our patients in comparison to 41.9% in a recent report of Chinese patients (mainly oedema and fluid retention), suggesting the presence of different racial susceptibilities to diazoxide-related side effects [18].

Interestingly, 2/3 of paternally inherited KCNJ11 mutation carriers initially responded to diazoxide, but then developed diazoxide-related cardiotoxicity, enforcing the fact that in cardiomyocytes, Kir6.2 is integral in the make-up of myo-cellular K-ATP channel and that K-ATP channel malfunction has been implicated in the development and progression of heart disease [17]; however, with only 2 patients, we cannot make a definite conclusion.

Siblings with homozygous KCNJ11 mutation were diazoxide unresponsive and had no cardiovascular side effects.

The introduction of 18F-DOPA PET CT, together with advanced genetic studies, has been able to identify focal lesions more accurately, resulting in better surgical outcomes and should be also offered as early as indicated to avoid blind pancreatic resections and poor management outcomes [19].

Our results showed neurodevelopmental deficits in 19% of patients mainly autistic spectrum disorders and only one of them had an identified mutation (GLUD1, with associated absence seizures). There was wide variation in the reported neurodevelopmental deficits, from almost none in small series [20] to 44% in a larger cohort [21], and this may be related to sample size and phenotypical variability. Other studies reported the rate of psychomotor retardation to be more pronounced in neonatal onset versus infantile or later onset disease and also in the more severe forms of CHI [22, 23]. Most of our patients have no developmental delay, but those who do, learning difficulties are moderate to severe.

Conclusion

Although LGA and SGA are risk factors for CHI, most babies in our cohort were born AGA. Initial glucose requirement and insulin level at diagnosis do not influence the disease outcome. A genetic mutation does not exclude medical remission, with HH resolving in over 30% within the first 12 months of age; long-term conservative treatment of CHI is feasible as surgery sometimes does not guarantee complete remission, but it carries the risk of pancreatic dysfunction. The need for ongoing treatment in the absence of gene mutations suggests that there may be other novel genetic mechanisms involved in regulating insulin secretion. Early management of hypoglycaemia remains critical to prevent long-term neurological deficits.